Ultrafast Phenomena XIV Proceedings of the 14th International Conference Niigata Japan July 25--3

Springer Series in

CHEMICAL PHYSICS

79

Springer Series in

CHEMICAL PHYSICS Series Editors: A. W. Castleman, Jr. J. P. Toennies

W. Zinth

The purpose of this series is to provide comprehensive up-to-date monographs in both well established disciplines and emerging research areas within the broad fields of chemical physics and physical chemistry. The books deal with both fundamental science and applications, and may have either a theoretical or an experimental emphasis. They are aimed primarily at researchers and graduate students in chemical physics and related fields. 65 Fluorescence Correlation Spectroscopy Theory and Applications Editors: R. Rigler and E.S. Elson 66 Ultrafast Phenomena XII Editors: T. Elsaesser, S. Mukamel, M.M. Murnane, and N.R Scherer 67 Single Molecule Spectroscopy Nobel Conference Lectures Editors: R. Rigler, M. Orrit, T. Basche 68 Nonequilibrium Nondissipative Thermodynamics With Application to Low-Pressure Diamond Synthesis ByJ.-T.Wang 69 Selective Spectroscopy of Single Molecules By I.S. Osad'ko 70 Chemistry of Nanomolecular Systems Towards the Realization of Molecular Devices Editors: T. Nakamura, T. Matsumoto, H. Tada, K.-I. Sugiura 71 Ultrafast Phenomena XIII Editors: D. Miller, M.M. Murnane, N.R. Scherer, and A.M. Weiner 72 Physical Chemistry of Polymer Rheology By J. Furukawa

73 Organometallic Conjugation Structures, Reactions and Functions of d-d and d-TT Conjugated Systems Editors: A. Nakamura, N. Ueyama, and K. Yamaguchi 74 Surface and Interface Analysis An Electrochmists Toolbox By R. Holze 75 Basic Principles in Applied Catalysis By M. Baerns 76 The Chemical Bond A Fundamental Quantum-Mechanical Picture ByT. Shida 77 Heterogeneous Kinetics Theory of Ziegler-Natta-Kaminsky Polymerization ByT.Keii 78 Nuclear Fusion Research Understanding Plasma-Surface Interactions Editors: R.E.H. Clark and D.H. Reiter 79 Ultrafast Phenomena XIV Editors: T. Kobayashi, T. Okada, T. Kobayashi, K.A. Nelson, S. De Silvestri

Takayoshi Kobayashi Tadashi Okada Tetsuro Kobayashi Keith A. Nelson Sandro De Silvestri (Eds.)

Ultrafast Phenomena XIV Proceedings of the 14th International Conference, Niigata, Japan, July 25-30, 2004

With 577 Figures

Spri ringer

Professor Takayoshi Kobayashi

Professor Keith A. Nelson

University of Tokyo, Department of Physics Kongo 7-3-1, Bunkyo, Tokyo 113-0033, Japan

MIT Room 6-235 Massachusetts Avenue Cambridge, MA 02139, USA

-j-j

Professor Tadashi Okada Toyota Physical and Chemical Research Institute Nagakute, Aichi 480-1192, Japan

ProfeSSO Sandro De Silvestri Politecnico di Milano, Dipartimento di Fisica Piazza L. da Vinci 32,20133 Milano, Italy

Professor Tetsuro Kobayashi Osaka University, Engineering Science Machikaneyama-Cho 1-3, Toyonaka Osaka 560-8531, Japan

Series Editors: Professor A. W. Castleman, Jr. Department of Chemistry, The Pennsylvania State University 152 Davey Laboratory, University Park, PA 16802, USA

Professor J.P. Toennies Max-Planck-Institut fiir Stromungsforschung, Bunsenstrasse 10 37073 Gottingen, Germany

Professor W. Zinth Universitat Miinchen, Institut fiir Medizinische Optik Ottingerstr. d-j^ 80538 Miinchen, Germany

ISSN 0172-6218 ISBN 3-540- Springer Berlin Heidelberg New York Library of Congress Control Number: 2004116005 This work is subject to copyright. All rights are reserved, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilm or in any other way, and storage in data banks. DupUcation of this publication or parts thereof is permitted only under the provisions of the German Copyright Law of September 9,1965, in its current version, and permission for use must always be obtained from Springer. Violations are liable to prosecution under the German Copyright Law. Springer Berlin Heidelberg New York a member of BertelsmannSpringer Science-hBusiness Media GmbH springeronline.com © Springer-Verlag Berlin Heidelberg 2005 Printed in Germany The use of general descriptive names, registered names, trademarks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. Typesetting: Camera-ready copy by the authors Cover concept: eStudio Calamar Steinen Cover production: design & production GmbH, Heidelberg Printed on acid-free paper

SPIN: 11362715

57/3141/XO - 5 4 3 2 1 0

Preface

This volume is a collection of papers presented at the Fourteenth International Conference on Ultrafast Phenomena held in Niigata, Japan from July 25-30, 2004. The Ultrafast Phenomena Conferences are held every two years and provide a forum for discussion of the latest results in ultrafast optics and their applications in science and engineering. A total of more than 300 papers were presented, reporting the forefront of research in ultrashort pulse generation and characterization, including new techniques for shortening the duration of laser pulses, for stabilizing their absolute phase, and for improving tenability over broad wavelength ranges, output powers and peak intensities. Ultrafast spectroscopies, particularly time-resolved X-ray and electron diffraction and two-dimensional spectroscopy, continue to give new insights into fundamental processes in physics, chemistry and biology. Control and optimization of the outcome of ultrafast processes represent another important field of research. There are an increasing number of applications of ultrafast methodology in material diagnostics and processing, microscopy and medical imaging. The enthusiasm of the participants, the involvement of many students, the high quality of the papers in both oral and poster sessions made the conference very successful. Many people and organizations made invaluable contributions. The members of the international program committee reviewed the submissions and organized the program. The staff of the Optical Society of America deserves special thanks for making the meeting arrangements and running the meeting smoothly. We thank the Optical Society of America for sponsorship, and also acknowledge support from The Physical Society of Japan, Japan Society of Applied Physics, Niigata Prefecture, Niigata Visitors & Convention Bureau, Niigata University, Japan National Tourist Organization and Inoue Foundation for Science. Tokyo, Japan Osaka, Japan Osaka, Japan Cambridge, USA Milano, Italy September 2004

Takayoshi Kobayashi Tadashi Okada Tetsuro Kobayashi Keith A. Nelson Sandro De Silvestri

V

Preface by a chair of Local Organizing Committee

The Fourteenth International Conference on Ultrafast Phenomena was held during Jul. 25-30, 2004 in Niigata. Niigata is distinguished for being one of only five international ports opened in 1868 when Japan resumed contact with other countries after nearly 250 years of self-imposed isolation. Since that time, Niigata has developed into one of most important modern international ports in Japan. In this sense the city is appropriate for the International Conference on Ultrafast Phenomena to be held. Since this meeting is usually held twice in the USA and once outside of the country of three years even though there were exceptions, the style and organization of the conference were sometimes quite different from the ones held in the USA. Because of the time shift and also of differences in the way of organization of conference place, we has experienced some complicated difficulties several times but finally we could find to solve the problems after various efforts to disentangle such kinds of problems. This is a very good experience of having the conference outside of the USA and I hope it will be very useful to utilize the knowledge in future meetings. Personally I have attended all of the conferences of this series starting from the first Topical Meeting of Picosecond Phenomena organized by Drs. Charles V. Shank, Erich P. Ippen, and late Stanley L. Shapiro at Hilton Head in South Carolina in 1978, third Picosecond Phenomena Meeting at GarmischPartenkirchen Germany and the fourth Ultrafast Phenomena at Monterey in California, and 13^^ UP at Vancouver in 2002 including the conferences in between. Until the 13^^ UP conference, there were four people who had attended all of these conferences, Prof. R. M. Hochstrasser, Erich P. Ippen, Graham Fleming, and myself. In the last meeting held in Niigata the last three among the four have left being the regular visitors. It is amazing to see the development made in these 26 years namely more than a quarter of century shown in the meeting and it is surprising to see that this conference is still acting as the place of presentation of the frontier activities of the field. Even more delightful feature I found in the conference is that three were so many new young scientists attended and gave nice talks and presented excellent works at poster sessions. I am also extremely happy to be able to have many scientists both senior and young came to Japan from very far. I am really hoping to see even more developed and active research presentations in future meetings of the Ultrafast Phenomena. General Chair of Local Organizing Committee Takayoshi Kobayashi

VII

Contents

Part I Generation and Measurements Single-Cycle Optical Pulse Generation D.R. Walker, M. Shverdin, D. Yavuz, G.-Y. Yin, S.E. Harris

3

Toward a Terawatt Few-Optical-Cycle Driver Laser for Attosecond Spectroscopy N. Ishii, R. Butkus, A. Baltuska, E. Goulielmakis, M. Uiberacker, R. Kienberger, T. Fuji, V.S. Yakovlev, V. Smilgevicius, R. Danielius, A. Piskarskas, F. Krausz

8

2.8-fs Clean Single Transform-Limited Optical-Pulse Generation and Characterization K. Yamane, T. Kito, R. Morita, M, Yamashita

13

Coherent Amplification of Femtosecond Pulses with Passive Enhancement Cavities RJ. Jones, L.-S. Ma, J. Ye

16

Temporal and Spatial Pulse Compression in a Nonlinear Defocusing Material N. C. Nielsen, T. Honer zu Siederdissen, J. Kuhl, M. Schaarschmidt, J. Forstner, A. Knorr, S. W. Koch, H. Giessen

19

Intense CEO-Stabilized Few-Cycle Laser Pulses from Supercontinuum Generation in Filaments J. Biegert, C.P. Hauri, W. Komelis, A. Heinrich, F.W. Helbing, A. Couairon, A, Mysyrowicz, U. Keller

22

Generation of Ultra-Broadband High Energy Pulses without External Amplification A. Fuerbach, A. Fernandez G., T. Fuji, H. Mayer, P. Dombi, F. Krausz, A. Apolonski

25

Sub-10 fs Multi-mJ Ti:Sapphire Laser System with a Pressure-Gradient Hollow Fiber Y. Oishi, A. Suda, F. Kannari, K. Midorikawa

28

IX

Generation of 14-fs Ultrashort Pulse in All Fiber Scheme by Use of Highly Nonlinear Hybrid Fiber T. Hori, N. Nishizawa, T. Goto

31

High Peak Power Ultrashort Pulse Generation Using All-Fiber Chirped Pulse Amplification System with Small Core Multimode Fiber J. Takayanagi, N. Nishizawa, H. Nagai, M. Yoshida, T. Goto

34

Carrier-Envelope Phase Fluctuations of Amplified Laser Pulses Transmitted through Neon-Filled Hollow Fiber for Pulse Compression A. Ishizawa and H. Nakano

37

Temporal Self-Compression of Intense Femtosecond Pulses Propagating in Argon-Filled Hollow Waveguides N. Wagner, E.A. Gibson, S. Backus, M.M. Mumane, H.C. Kapteyn, LP. Christov

40

Stimulated Brillouin Scattering in Ultrahigh-Speed Femtosecond Soliton Pulse Compression with a DispersionDecreasing Fiber T. Hirooka, S. Ono, K.-L Hagiuda, M. Nakazawa

43

Control of the Spectral Broadening of Tens-Milli-Joules Laser Pulses in an Argon-Filled Hollow Fiber Using a Conjugate Pressure Gradient M. Nurhuda, A. Suda, K. Midorikawa

46

Microstructured Fiber Feedback Pulse Compression to Few Optical Cycles M. Adachi, K. Yamane, R. Morita, M. Yamashita

49

Spectral-Temporal Soliton Dynamics Analysis Near Second Zero-Dispersion Point in Photonic Crystal Fibers A, Efimov, A.J. Taylor, F.G. Omenetto, N. Joly, D.V. Skryabin, J.C. Knight, W.J. Wadsworth, P.S.J. Russell

52

Generation of Rotational Raman Emissions and SelfCompressed Femtosecond Pulses in a Hydrogen Gas S. Zaitsu, Y. Kida, T. Imasaka

55

Spectral Broadening of 50 Milli Joule Laser Pulses in a Neon-Filled Herriot Multiple-Pass Cell (MPC) M. Nurhuda, A. Suda, K. Midorikawa

58

X

CEO Phase Preservation in Chirped-Pulse Optical Parametric Amplification of 17.3-fs Pulses J. Biegert, C.P. Hauri, P. Schlup, W. Kornelis, F,W. Helhing, U. Keller, G. Arisholm

61

Long-Term Stabilization and Control of CEP of Idler from NOPA S. Adachi and T. Kobayashi

64

Experimental and Theoretical Study of a Visible Noncollinear Optical Parametric Amplified Pulse with 200 THz Bandwidth X Fang and T. Kobayashi

67

Tunable Wavelength Pulse Shaping of Visible N O P A Outputs with an Acousto-Optic Programmable Dispersive Filter D. Kaplan, P. Toumois, B. Chatel, A. Monmayrant

70

Broadband High Power Optical Chirped Pulse Amplification N. Ishii, R. Butkus, A. Baltuska, V. Smilgevicius, R, Danielius, A. Piskarskas, F. Krausz

73

Mid-Infrared Femtosecond Pulse Generation by Optical Parametric Amplification under Broadband Q P M Condition S. Ashihara, M. Ikeda, T. Shimura, K. Kuroda

76

Achromatic Second Harmonic Generation: Tunable Ultraviolet Pulses with Sub-10 fs Duration P. Baum, S. Lochbrunner, E. Riedle

79

Ultrabroad-Band Noncollinear Optical Parametric Amplification in Some N e w Nonlinear Optical Crystals P. Kumbhakar and T. Kobayashi

82

Design of Multilayer Mirrors for the Refiection of Sub-Femtosecond Pulses in the X U V Spectral Region A.S. Pirozhkov, H. Daido, S. V. Bulanov, E.N. Ragozin

85

Route to Design Electric Fields of Optical Pulses: A Combination of a Pulse Shaper and a Carrier-EnvelopePhase Stabilized Chirped-Pulse Amplifier System M. Kakehata, H. Takada, Y. Kobayashi, K. Torizuka, K. Nishijima, H. Takamiya, T. Homma, H. Takahashi

88

Towards Electric Field Reconstruction Using Coherent Transients in a Two-Level System A. Monmayrant, B. Chatel, B. Girard

91

XI

Spatiotemporal Determination of the Absolute Phase of Few-Cycle Laser Pulses F. Lindner, M. Schdtzel, G. Paulus, H. Walther, A. Baltuska, E. Goulielmakis, M. Lezius, F.c Krausz

94

Spatial Chirp and Pulse-Front Tilt in Ultrashort Laser Pulses and Their Measurement 5. Akturk, X. Gu, E. Zeek, R. Trebino

97

Principal Control Analysis: Gaining Insight from Feedback Learning Algorithms J.L. White, B.J. Pearson, P.H. Bucksbaum

100

Population-Split Genetic Algorithm for Phase Retrieval of Ultrafast Laser Pulses C.-W. Chen, S.-F. Shu, C.-K. Lee, C.-L. Pan

103

Eight-Frame Observation of Propagation Behavior of 0.49-mJ, 45-fs Optical Pulses Generated by a 1- kHz Laser System M. Fujimoto, S.-L Aoshima, Y. Tsuchiya

106

Self-Referenced Measurement of the Complete Electric Field of Ultrashort Pulses in Time and Space P. Gabolde, S. Akturk, R. Trebino

109

Pulse-Measurement Challenges at 1.5 Microns: Several-Cycle Pulses and Several-Element Devices S. Akturk, M. Kimmel, R. Trebino, S. Naumov, E. Sorokin, LT. Sorokina

112

Single-Shot Phase Measurement by Spectral Phase Interferometry Using a Streak Camera T. Akagawa, K. Misawa, R. Lang

115

FROG Measured 185 fs Pulses Generated by Down-Chirped Dispersion-Managed Breathing-Mode Semiconductor Laser B. Resan, L. Archundia, P.J. Delfyett

118

Direct Measurement of the Group Delay Dispersion of Ultrashort Pulses Utilizing Molecular Vibrations P.J. Rizo and T. Kobayashi

121

Spatially Encoded Spectral Interferometry for Complete Characterisation of Attosecond X U V Pulses E. Cormier, LA. Walmsley, E.M. Kosik, L. Corner, L.F. DiMauro. . . . 124

XII

Spatially Encoded Spectral Interferometry for Complete Characterization of Ultrashort Pulses E.M. Kosik, A.S. Radunsky, LA. Walmsley, C. Dorrer

, . . 127

Full Characterization of Ultraviolet and Visible 10-fs Pulses with Zero-Additional-Phase SPIDER P. Baum, S. Lochhrunner, E. Riedle

130

Direct Visualization of Transient Absorption by Real-Time Pump-Probe Imaging Spectroscopy A^. Furukawa, C.E. Mair, V.D. Kleiman, J. Takeda

133

Part II Strong Fields and High Order Harmonics Dynamic Molecular Imaging P.B, Corkum

139

Femtosecond Electron Diffraction: Towards Making the "Molecular Movie" J.R. Dwyer, R.E. Jordan, B.J. Siwick, C.T. Hebeisen, R.J. Dwayne Miller

144

Absolute Displacement Interferometry of Ultrafast Laser-Produced Plasma Expansion G. Rodriguez, S.A. Clarke, A.J. Taylor

149

Electron Acceleration through Spatiotemporal Shaping of Ultrashort Light Pulses D.H. Torchinsky, T. Feurer, K.A. Nelson

152

Mapping Attosecond Electron Wave Packet Motion H. Niikura, D.M. Villeneuve, P.B. Corkum

155

Quasi-Monoenergetic Electron Beam Generation in Laser-Driven Plasma Acceleration E. Miura, K. Koyama, M. Adachi, S. Kato, Y. Kwada, S. Masuda, T. Nakamura, N. Saito, M. Tanimoto

158

Control of Multiphoton Ionization Processes in Aligned 12 Molecules by Optimizing Time-Dependent Polarization of Femtosecond Pulses T. Suzuki, S. Minemoto, T. Kanai, H. Sakai

161

Tomographic Imaging of Molecular Orbital with High Harmonic Generation J. Ratani, J. Levesque, D. Zeidler, M. Spanner, P. B. Corkum, D.M. Villeneuve

164

XIII

Femtosecond Infrared Vibrational U p - P u m p i n g of Liquid Phase W(CO)6 T.L Witte, M.C. Motzkus, K.L. Kompa, J.S. Yeston, E.J. Heilweil

167

Ultrafast X-Ray Diffraction K. Sokolowski-Tinten, C. Blome, J. Blums, U. Shymanovich, M. Nicoul, A. Cavalleri, A. Tarasevitch, M. Horn-von Hoegen, M. Kammler, D. von der Linde

170

Quasi-Phase M a t c h i n g of High H a r m o n i c G e n e r a t i o n in t h e "water window" Soft X-Ray Region E.A. Gibson, A. Paul, S. Backus, R. Tohey, M.M. Murnane, H.C. Kapteyn, LP. Christov

175

A d a p t i v e Engineering of Coherent Soft X-Rays T. Pfeifer, D. Walter, C. Winterfeldt, C. Spielmann, G. Gerber

178

G e n e r a t i o n of Strong Soft X-Ray Field Based on H i g h - O r d e r Harmonics H. Mashiko, A. Suda, K. Midorikawa

181

G e n e r a t i o n of Sub-4-fs High H a r m o n i c Pulses a n d T h e i r Application t o t h e Above-Threshold Ionization T. Sekikawa, A. Kosuge, T. Kanai, S. Watanabe

184

Coherent Imaging of Laser-Plasma Interactions Using High-Harmonic E U V Light X. Zhang, D. Raymondson, A.R. Libertun, A.J. Paul, M.M. Murnane, H.C. Kapteyn, Y. Liu, D.T. Attwood

189

High-Order H a r m o n i c Generation from Argon Ions u p t o 250 eV E.A. Gibson, A. Paul, N. Wagner, S. Backus, M.M. Murnane, H.C. Kapteyn, LP. Christov

192

High-Order H a r m o n i c Generation from Femtosecond Laser-Aligned Molecules K. Miyazaki, M. Kaku, K. Masuda, G. Miyaji

195

Efficient G e n e r a t i o n of High-Order Sum a n d Difference Frequencies in t h e X U V Region by Combining a Weak, Longer-Wavelength Field Y. Nomura, T. Kanai, S. Minemoto, H. Sakai

198

Effects of Target Condition on Solid Surface H a r m o n i c s in t h e E x t r e m e Ultraviolet R a n g e T. Ozaki, J.-C. Kieffer, H. Nakano, A. Ishizawa

201

XIV

Control of t h e Frequency C h i r p R a t e of High H a r m o n i c Pulses J. Biegert, M. Bruck, C.P, Hauri, A. Heinrich, F.W. Helhing, W. Kornelis, P. Schlup, U. Keller, R. Lopez-Mart ens, J. Mauritsson, P. Johnsson, K. Varju, A. L'Huillier, M. Gaarde, K.J. Schafer

204

Wavefront Control in High Harmonics G e n e r a t i o n w i t h Fewa n d Many-Optical-Cycle Laser Pulses P. Villoresi, S. Bonora, M. Pascolini, L. Poletto, C. Vozzi, G. Sansone, S. Stagira, M. Nisoli

207

Phase-Driven Strong-Field Processes in t h e Multi-OpticalCycle Regime G. Sansone, S. Stagira, C. Vozzi, M. Pascolini, L. Poletto, P. Villoresi, G. Tondello, S. De Silvestri, M. Nisoli

210

Bright High-Order H a r m o n i c G e n e r a t i o n at 13 n m a n d Coherence M e a s u r e m e n t H.T. Kim, I.J. Kim, V. Tosa, Y.S. Lee, C.H. Nam

213

Attosecond P u l s e G e n e r a t i o n During t h e Laser Pulse Reflection at t h e P l a s m a - V a c u u m Interface A.S. Pirozhkov, H. Daido, S. V. Bulanov

216

Energetic P r o t o n a n d D e u t e r o n G e n e r a t i o n from a Microporous Polytetrafluoroethylene Film w i t h D e u t e r a t e d Polystyrene Using a 2.4-TW Table-Top Laser H. Takahashi, S. Okihara, S. Ohsuka, M. Fujimoto, S. Okazaki, T. Ito, S. Aoshima, Y. Tsuchiya

219

High-Energy P r o t o n s E m i t t e d from a P o l y m e r - C o a t e d M e t a l foil by 60-fs Laser Irradiation H. Kishimura, H. Morishita, Y.H. Okano, Y. Okano, Y. Hironaka, K.-L Kondo, Y. Oishi, K. Nemoto, K.G. Nakamura

222

E s t i m a t i o n of P r o t o n Source Size G e n e r a t e d by U l t r a i n t e n s e Laser Pulses Using a T h o m s o n Mass S p e c t r o m e t e r Y. Oishi, T. Nayuki, T. Fujii, Y.i Takizawa, X. Wang, T. Sekiya, A.A. Andreev, K. Horioka, T. Yamazaki, K. Nemoto

225

P a r t I I I Ultrafast Dynamics in Solid 1 ( P h o n o n a n d Exciton) Imaging N a n o s t r u c t u r e s with Picosecond Ultrasonic Pulses B.C. Daly, NCR. Holme, T. Buma, C Branciard, T.B. Norris, S. Pau, D.M. Tennant, J.A. Taylor, J.E. Bower

231

XV

U l t r a h i g h Frequency Acoustic P h o n o n Generation a n d Spectroscopy with D e a t h s t a r Pulse Shaping J.D. Beers, M. Yamaguchi, T. Feurer, B.J. Paxton, K. A. Nelson

236

P r o b i n g of Thermo-Acoustic Transients in Materials Using E U V Radiation R.I. Tohey, E.H. Gershgoren, M.E. Siemens, M.M. Murnane, H.C. Kapteyn, T. Feurer, K.A. Nelson

239

Ultrafast Dynamics of Coherent E l e c t r o n - P h o n o n I n t e r a c t i o n in Silicon M. Kitajima, M. Hase, A.M. Constantinescu, H. Petek

242

G e n e r a t i o n of Coherent Zone B o u n d a r y P h o n o n s by Impulsive Excitation of Molecules M. Giihr and N. Schwentner

245

A m p l i t u d e Collapse-Revival of C h i r p e d Coherent P h o n o n s u n d e r High-Density Optical Excitation K. Ishioka, O. V. Misochko, R. Lu, M. Hase, M. Kitajima

248

Intense Coherent Optical P h o n o n s Driven by Impulsive Excitonic Interference u n d e r Electric Fields O. Kojima, K. Mizoguchi, M. Nakayama

251

P h o n o n - P o l a r i t o n Based T H z Spectroscopy B.J. Paxton, M. Yamaguchi, K.A. Nelson

254

Excitonic Q u a n t u m B e a t s Dressed with C o h e r e n t P h o n o n s K. Mizoguchi, T. Furuichi, O. Kojima, M. Nakayama, K. Akahane, N. Yamamoto, N. Ohtani

257

Evidence of Higher-Order Nonlinearities in Excitonic F W M Signals in Microscopic T h e o r y a n d E x p e r i m e n t L. Wischmeier, M. Buck, S. Schumacher, G. Czycholl, F. Jahnke, I. Rilckmann, J. Gutowski

260

Ultrafast Anisotropic Processes of Exciton M a g n e t i c Polarons in C d T e / C d M n T e Q u a n t u m Wires R. Naganuma, T. Kita, S. Nagahara, 0. Wada, L. Marshal, H. Mariette263 Time-Resolved Mid-Infrared Spectroscopy of Excitons in Cu20 M. Kubouchi, R. Shimano, K. Yoshioka, A. Mysyrowicz, M. Kuwata-Gonokami

XVI

266

Exciton Dynamics in P e n t a c e n e a n d Tetracene Studied Using Optical P u m p - P r o b e Spectroscopy V,K. Thorsm0lle, R.D. Averitt, J. Demsar, X. Chi, D.L. Smith, A.P. Ramirez, A.J. Taylor

269

Dephasing Suppression of Excitons in Semiconductors T. Kishimoto, A. Hasegawa, Y. Mitsumori, M. Sasaki, F. Minami . . . . 272 Dynamical Stark Effect of Excitons in C u 2 0 by R e s o n a n t P u l s e d Excitation of t h e l s - 2 p Transition K. Yoshioka, M. Kuhouchi, R. Shimano, M. Kuwata-Gonokami

275

Ultrafast C h a r g e P h o t o g e n e r a t i o n a n d Exciton R e g e n e r a t i o n at Polymeric Semiconductor Heterojunctions A.C. Morteani, P. Sreearunothai, L.M. Herz, R.H. Friend, C. Silva . . . 278 Exciton Diffusion Dynamics in an Organic Semiconductor Nanostructure C. Daniel, L.M. Herz, S. Westenhoff, F. Makereel, D. Beljonne, F.J.M. Hoeben, P. Jonkheijm, A.P.H.J. Schenning, E.W. Meijer, C. Silva

281

P a r t I V Ultrafast Dynamics in Solid 2 Carrier-Envelope P h a s e Controlled Q u a n t u m Interference in a Semiconductor T.M. Fortier, P.A. Roos, D.J. Jones, S.T. Cundiff, R.D.R. Bhat, J.E. Sipe

287

Phase-Resolved Nonlinear Response of M o d u l a t i o n - D o p e d Q u a n t u m Wells u n d e r Femtosecond I n t e r s u b b a n d Excitation T. Shih, C.-W. Luo, K. Reimann, M. Woerner, T. Elsaesser, I. Waldmilller, A. Knorr, R. Hey, K.H. Ploog

292

Ultrafast I n t e r s u b b a n d Relaxation a n d Carrier Cooling in G a N / A l N Multiple Q u a n t u m Wells J. Hamazaki, H. Kunugita, K. Ema, S. Matsui, Y. Ishii, T. Morita, A. Kikuchi, K. Kishino

295

Polaritonics in Complex S t r u c t u r e s : Confinement, B a n d g a p Materials, a n d Coherent Control D.W. Ward, E.R. Statz, J.D. Beers, T. Feurer, J.D. Joannopoulos, R.M. Roth, R.M. Osgood, K.J. Webb, K.A. Nelson

298

XVII

Detection of Four-Wave Mixing Signal from Single Layer Quantum Dots M. Ikezawa, F. Suto, Y. Masumoto, H.-W. Ren

301

Wavepacket Interferometry and Wavepacket Dynamics in Condensed Phase M. Bargheer, M. Fushitani, M. Giihr, N. Schwentner

304

Femtosecond Wavepacket Dynamics of Potassium Adsorbate on P t ( l l l ) K. Watanabe, N. Takagi, Y. Matsumoto

307

Control of Tunnel Ionization in Molecules by Intense Femtosecond Laser Pulses With Time-Dependent Polarization T. Kanai, S. Minemoto, H. Sakai

310

Ultrafast Mid-Infrared Dynamics in the Colossal Magnetoresistance Pyrochlore T12Mn207 R.P. Prasankumar, A.J. Taylor, R.D. Averitt, H. Okamura, H. Imai, Y. Shimakawa, Y. Kubo

313

Femto-Magnetism Visualized in Three Dimensions J.-Y. Bigot, M. Vomir, L.H.F. Andrade, L. Guidoni, E. Beaurepaire, J. Arabski

316

Photo-Induced Demagnetization Observed by TimeResolved Mid-Infared Transmittance Spectroscopy in GaO.94MnO.06As, E. Kojima, J.B.t Heroux, R. Shimano, Y.i Hashimoto, S. Katsumoto, Y. lye, M. Kuwata-Gonokami

319

Optically Induced Magnetization and Ultrafast Spin Dynamics of Magnetic Ions in Ionic Crystals T. Kohmoto, K. Nakazono, S. Furue, M. Kunitomo, Y. Fukuda

322

Dynamic Coupling-Decoupling Crossover in the CurrentDriven Vortex-State in T12Ba2CaCu208 Studied Using Terahertz Time-Domain Spectroscopy V.K. Thorsm0lle, R.D. Averitt, I. Aranson, M.P. Maley, L.N. Bulaevskii, A.J. Taylor

325

Ultrafast Light Induced Charge Disordering Around Phase Transition Temperature in 2D Spin Ladder Compound NaV205 M. Aiba, M. Nakajima, M. Isobe, Y. Ueda, T. Suemoto

328

XVIII

Correlation of t h e Electronic Transitions in Semiconducting Single-Walled C a r b o n N a n o t u b e s Y.-Z. Ma, J. Stenger, S.L. Dexheimer, S.M. Bachilo, R.E. Smalley, R.B. Weisman, G.R. Fleming

331

Ultrafast Radial T r a n s p o r t in a Micron-Scale A l u m i n u m P l a s m a Excited at Relativistic Intensity B.T. Bowes, M.C. Downer, H. Langhoff, M. Wilcox, B. Hou, J. Nees, G, Mourou

334

C h i r p Control of Free Carrier Dynamics in G a A s T. Hattori, T. Yogi, Y. Hama, N. Watanabe

337

Ultrafast Insulator-to-Metal Switching by P h o t o i n d u c e d M o t t Transition S. Iwai, Y. Okimoto, M. Ono, H. Matsuzaki, A. Maeda, H. Kishida, H. Okamoto, Y, Tokura

340

Femtosecond N e a r Edge X-Ray Absorption M e a s u r e m e n t of t h e V 0 2 P h a s e Transition A. Cavalleri, H.H.W. Chong, S. Fourmaux, T.E. Glover, P.A. Heimannn, J.C. Kieffer, H.A. Padmore, R.W. Schoenlein

343

P h a s e Transition in Strongly-Correlated V 0 2 : T i m e - D o m a i n Assignment of Cause a n d Effect A. Cavalleri, T. Dekorsy, H.H. Chong, J.C Kieffer, R.W. Schoenlein . 346 Polarization-Dependent P h e n o m e n o n Induced by t h e Interaction between Focused Femtosecond Laser a n d T r a n s p a r e n t Materials Y. Shimotsuma, J. Qiu, P. G. Kazansky, K. Hirao

349

Investigation on t h e P a r a m e t e r s of Dense Electronic P l a s m a I n d u c e d by Femtosecond Laser in Fused Silica Q. Gong, Q. Sun, Y. Liu, Z. Wu, H. Yang, H. Jiang

354

Dynamical S y m m e t r y Breaking Induced by U l t r a s h o r t Laser Pulses in K T a 0 3 E. Matsubara, J.-I. Takahashi, K. Inoue, E. Hanamura

357

P a r t V Ultrafast Dynamics in Solution Sub-20-fs S t u d y of Energy Relaxation in carotenoids in solution a n d inside light harvesting complexes G. Cerullo, D. Polli, G. Lanzani, H. Hashimoto, R.J. Cogdell

363

XIX

Energy Flow in Carotenoids, Studied With Pump-DepleteProbe, Multiphoton and Coherent Control Spectroscopy T. Buckup, W. Wohlleben, J. Savolainen, B. Heinz, H. Hashimoto, R.J. Cogdell, J.L. Herek, M. Motzkus

368

Amplitude Spectra of Molecular Vibration Modes in Phthalocyanine: Comparison with Raman Excitation Profile T. Kobayashi, M. Hirasawa, Y. Sakazaki, H. Hane

371

Real Time Tracking of the Peaks in Transition Difference Spectra During Vibrational Periods in P D A Y. Yuasa, M. Ikuta, T. Kimura, H. Matsuda, T. Kobayashi

374

Time-Resolved CARS Studies of Vibrational Coherences in the Condensed Phase: 12 in Solid Krypton M. Karavitis, I. Goldschleger, V.A. Apkarian, T. Kumada

377

Measurement of Conical Intersection Dynamics by Impulsive Femtosecond Polarization Spectroscopy D.A. Farrow, W. Qian, E.R. Smith, D.M. Jonas

380

Vibrational Phase Characterization in Femtosecond-Pumped Molecules by Path-Length Modulation T. Taneichi, T. Fuji, Y.u Yuasa, T. Kobayashi

383

Vibrational Energy Relaxation in Water-Acetonitrile Mixtures D. Cringus, S. Yeremenko, M.S. Pshenichnikov, D.A. Wiersma

386

Cascaded Energy Redistribution upon O-H Stretching Excitation in an Intramolecular Hydrogen Bond K. Heyne, M. Petkovic, E.T.J. Nibbering, O. Kilhn, T. Elsaesser

389

Pure Intermolecular Energy Relaxation of the OH Bending Vibration of Water Molecules Dissolved in Organic Liquids G. Seifert, T. Patzlaff, K. Paradowska-Moszkowska, H. Graener

392

Time-Resolved Spectroscopy of an Azobenzene Derivative with a Small S1-S2 Energy Gap M. Hagiri, N. Ichinose, T. Nakayama, C. Zhao, H. Horiuchi, H. Hiratsuka

395

Photo-Thermalization Dynamics of Azulene in Supercritical Fluids Studied by the Transient Grating Method Y. Kimura, Y. Yamamoto, M. Terazima

398

Vibrational Self-Trapping in an a-Helix J. Edler, V. Pouthier, C. Falvo, R. Pfister, P. Hamm

401

XX

Infrared Photon-Echo Spectroscopy of Water: the ThermaHzation Effects M.S. Pshenichnikov, S. Yeremenko, D.A. Wiersma

404

Heterodyne 2D-IR Photon Echo Spectroscopy of Multi-Level OH Stretching Coherences in Hydrogen Bonds N. Huse, B.D. Bruner, M.L. Cowan, J. Dreyer, E.T.J. Nibhering, T. Elsaesser, R.J.D. Miller

407

A Unified Analysis of Ultrafast Vibrational and Orientational Dynamics of H O D in D 2 0 J.J. Loparo, C.J. Fecko, J.D. Eaves, S.T. Roberts, A. Tokmakoff

410

Time Resolved Direct Probing of the Change in the Local Solvent Response Following Excitation of a Solute D.F. Underwood and D.A. Blank

413

Surface Femtochemistry: Photocatalytic Reaction Dynamics of Methanol/Ti02(110) K. Onda, B. Li, H. Petek

416

Three Pulse Four Wave Mixing for the Study of Coherent Interactions, Nuclear Dynamics and Solvation Dynamics in Liquids J.-S. Park and T. Joo

419

Solvation Dynamics of N-Methylacetamide in D 2 0 , CDC13, and DMSO-d6 M.F. DeCamp, L.P. DeFlores, J.M. McCracken, A. Tokmakoff

422

Femtosecond Pump-Probe Measurements of Solvation Dynamics of Hydrogen-Bonding Complexes in NonAssociating Solvents D. Pines, E. Pines, Y.-Z. Ma, G.R. Fleming

425

Novel Time- and Frequency-Resolved Double P u m p Spectroscopy of Short-Lived Precursors: The Solvated Electron in Methanol A. Thaller, R. Laenen, A. Laubereau

428

Ultrafast IR Spectroscopy on Aqueous Reverse-Micellar Nano- D r oplet s D. Cringus, M.T.W. Milder, M.S. Pshenichnikov, D.A. Wiersma, J. Lindner, P. Vohringer

431

Pump-Probe Near-Field Optical Microscopy of Molecular Aggregates Using Supercontinuum T.o Nagahara, K. Imura, H. Okamoto

434

XXI

Vibrational and Rotational Relaxation Dynamics of Anions in Reverse Micelles by Ultrafast Infrared Spectroscopy J.C. Owrutsky, G.M. Sando, Q. Zhong, A.P. Baronavski

437

Part VI Reaction Dynamics in Solution Femtochemistry in the Electronic Groundstate? IR-Driven Cis-Trans Isomerization of HONO P. Hamm, R. Schanz, V. Botan

443

Bimodal Intermolecular Proton Transfer in Acid-Base Neutralization Reactions in Water O.F. Monhammed, M. Rini, J. Dreyer, B.-Z. Magnes, D. Pines, E.T.J. Nihhering, E. Pines

448

Ultrafast Excitation Energy Migration Processes in Various Porphyrin Arrays D. Kim

453

Energy Transfer in Phenylene Ethynylene Dendrimers E. Atas, C.E. Mair, J.S. Melinger, Z. Peng, V.D. Kleiman

456

A 40-fs Time-Resolved Absorption Study of Cis-Stilbene in Solution: Observation of Coherent Nuclear Wavepacket Motion in Reactive Excited State K. Ishii, S. Takeuchi, T. Tahara

459

Monitoring an Ultrafast Photo-Isomerization by Femtosecond Fluorescence, Absorption, and IR Spectroscopy P. Gilch, B. Schmidt, C. Sobotta, M. Braun, F. Roller, T. Schrader, A. Sieg, W. Schreier, W. Zinth

462

From Ultrafast Spectroscopy to Bidirectional Molecular Switches: D H A / V H F U. Schmidhammer, V. De Waele, G. Buntinx, E. Riedle

465

Ultrafast Intramolecular Electron Transfer of 9,9'-Bianthryl as Studied by Femtosecond Time-Resolved Near-Infrared Absorption and Anisotropy in the 950-1500 nm Region T. Takaya, K. Iwata, H. Hamaguchi, H. Kuroda

468

Real-Time Spectroscopy of Charge-Transfer Excitation in Phthalocyanine Tin Dichloride M. Hirasawa, Y. Sakazaki, H. Hane, T. Kobayashi

471

XXII

Coherent Nuclear Dynamics Coupled with Electron Transfer Reaction in Porphyrin-Ferrocence Dyads S. Nakashima, M. Kuho, M. Otani, M. Murakami, Y. Ishibashi, M. Yasuda, H. Miyasaka, Y. Mori, H. Imahori

474

Subpicosecond Pulse Radiolysis Study on Geminate Ion Recombination Process in n-Dodecane Y. Yoshida, A. Saeki, T. Kozawa, J.g Yang, S. Tagawa

479

Fast Spin Dynamics of Optically Induced Magnetization in Aqueous Solutions of Magnetic Ions S. Furue, T. Kohmoto, M. Kunitomo, Y. Fukuda

482

Vibrational Excitation and Energy Redistribution after Ultrafast Intramolecular Proton Transfer of T I N U V I N W. Werncke, V. Kozich, J. Dreyer

485

Coherent Nuclear Motion in Reacting Molecules: Ultrafast Pump-Probe Spectroscopy of Proton Transfer in Solution S. Takeuchi and T. Tahara

488

Ultrafast Double Proton Transfer: Symmetry Breaking Wavepacket Motion and Absence of Deuterium Isotope Effect S. Lochhrunner, K. Stock, C. Schriever, E. Riedle

491

Photodissociation Dynamics Studied via Time-Resolved Coincidence Imaging Spectroscopy O. Gefiner, E, Ter-Heersche Chrysostom, A.M.D. Lee, J.P. Shaffer, C.C. Hayden, A. Stolow

496

Femtosecond Photo-Induced Dissociation of the Trihalide Anions 13- and I2Br- in Solution P. Salen, M. Liu, P. van der Meulen

499

Coherent Control of Non-Radiative Transitions: Long-Range Electron Transfer B.D. Fainherg, V.A, Gorbunov, S.H. Lin

502

Teaching Lasers To Twist Molecules G. Vogt, G. Krampert, P. Niklaus, G. Gerber

505

Quantum Control of a Chiral Molecular Motor Driven by Linearly Polarized Laser Pulses M. Yamaki, K. Hoki, Y. Ohtsuki, H. Kono, Y. Fujimura

508

XXIII

Numerical Synthesis of Optimal Laser Pulses for Manipulating Dissociation Wave Packets of 12- in Water Y. Ohtsuki, Y. Nishiyama, T. Kato, H. Kono, Y. Fujimura

511

Molecular State Reconstruction by Nonlinear Wave Packet Interferometry T.S. Humble and J.A. Cina

514

Femtosecond Coherent Spectroscopic Study of Zn(II) Porphyrin by Chirping-Controlled Ultrashort Pulses M.-C. Yoon, S. Cho, D. Kim

517

Phase Analysis of Vibrational Wavepackets in the Ground and the Excited States in Polydiacetylene M. Ikuta, Y. Yuasa, T. Kimura, H. Matsuda, T. Kobayashi

520

Calculating Ultrafast Nonlinear Optical Signals from Molecules in Cryogenic Matrices M.A. Rohrdanz and J.A. Cina

523

Real-Time Observation of Phase-Controlled Vibrational Wave-Packets in Iodine Molecules Y. Sato, H. Chiba, M. Honda, Y. Hagihara, K. Fujiwara, K. Ohmori, K. Ueda

526

Single-Shot Transient Absorption of 13- in Solutions and Glasses P.R. Poulin and K.A. Nelson

529

Part VII Multiview and Multi-Dimensional Spectroscopy Dynamics of Hydrogen Bonds in Water: Vibrational Echoes and Two-Dimensional Infrared Spectroscopy C.J. Fecko, J.D. Eaves, J.J. Loparo, S.T. Roberts, A. Tokmakoff, P.L. Geissler

535

Dual-Frequency 2D IR Photon Echo of a Hydrogen Bond I. V. Rubtsov, K. Kumar, R.M. Hochstrasser

539

2D-IR Spectroscopy of Transient Species J. Bredenbeck, J. Helbing, P. Hamm

542

Resolving Conformations of Acetylproline-NH2 by Coherent 2D IR Spectroscopy D. Karaiskaj, S. Sul, Y. Jiang, N.-H. Ge

545

XXIV

Ultrafast Vibrational Dynamics of Rotaxanes O.F.A. Larsen, W.J. Buma, D.A. Leigh, S. Woutersen

548

Thermal Denaturing of Proteins: Equilibrium and Transient Studies Using Nonlinear Infrared Probes H.S. Chung, M. Khalil, A.W. Smith, Z. Ganim, A. Tokmakoff

551

Two-Dimensional Optical Heterodyne Spectroscopy of Molecular Complexes I. Stiopkin, T. Brixner, G.R. Fleming

554

Two-Dimensional Measurement of the Solvent Intermolecular Response in Solvation S. Park, J. Kim, N.F. Scherer

557

Two-Dimensional Femtosecond Coherent Anti-Stokes Raman Scattering Spectroscopy Using a Chirped Supercontinuum Generated from a Photonic Crystal Fiber H. Kano and H. Hamaguchi

560

Two-Dimensional Spectroscopy by Spectrally Resolved Real-Time Resonant Coherent Raman Scattering in Polydiacetylene N. Ishii, E. Tokunaga, S. Adachi, T. Kimura, H. Matsuda, T. Kobayashi563 Optical Two-Dimensional Fourier-Transform Spectroscopy of Semiconductor Quantum Wells C.N. Borca, T. Zhang, S.T. Cundiff

566

Degenerate Four-Wave Mixing Spectroscopy Based on Two Dimensional Pulse Shaping T. Hornung, J.C. Vaughan, T. Feurer, K.A. Nelson

569

Propagation and Detection Distortions of Four-Wave Mixing Signals: Application to 2D Spectroscopy A^. Belahas and D.M. Jonas

572

Fourier Transform Measurement of Two-Photon Excitation Spectra: Applications to Microscopy and Quantum Control K.J. Kuharych, J.P. Ogilvie, A. Alexandrou, M. Joffre

575

Part VIII Biology Watching Proteins Function with Picosecond Time-Resolved X-Ray Crystallography P.A. Anfinrud, F. Schotte, M. Wulff

581

XXV

E n e r g y Transfer P a t h w a y s in P h o t o s y s t e m I Studied by One a n d Two Color P h o t o n Echo spectroscopy H.M. Vaswani, J. Stenger, M. Yangz, P. Fromme, G.R. Fleming

586

Dynamics of Carotenoids P r o b e d by Femtosecond Absorption, Fluorescence, and R a m a n Spectroscopy M. Yoshizawa, D. Kosumi, M. Komukai, K. Yanagi, H. Hashimoto. . . . 589 M u l t i - P u l s e Transient Absorption a n d Carotenoid Excited-State Dynamics:/3-Carotene E. Papagiannakis, D.S. Larsen, M. Vengris, I.H.M. van Stokkum, R. van Grondelle

592

Observation and Control of All-Trans-?-Carotene Wavepacket M o t i o n Using P u m p - D e g e n e r a t e Four-Wave Mixing T. Hornung, H. Skenderovic, K.-L. Kompa, M. Motzkus

595

Vibrational a n d Electronic Coherence Observed in Two-Dimensional I n t e g r a t e d T h r e e - P u l s e P h o t o n Echo Y. Nagasawa, M. Ogasawara, Y. Nakagawa, Y. Mori, T. Okada, H. Miyasaka

598

P h o t o n Echo S t u d y of t h e E l e c t r o n - P h o n o n Coupling S t r e n g t h in Molecules and Molecular Aggregates V.I. Prokhorenko, R. van Grondelle, R.J. Dwayne Miller

601

T i m e a n d Frequency Domain Investigations on Ultrafast Photoisomerization Reaction Dynamics of P Y P H. Chosrowjan, S. Taniguchi, N. Malaga, N. Hamada, F. Tokunaga, M. Unno

604

E n t i r e View of Coherent Oscillations in Ultrafast Fluorescence for P h o t o a c t i v e Yellow P r o t e i n R. Nakamura, N. Hamada, H. Ichida, Y. Kanematsu, F. Tokunaga. . . . 607 Ultrafast Excited a n d G r o u n d - S t a t e Isomerization Dynamics of t h e G r e e n Fluorescent P r o t e i n C h r o m o p h o r e in Solution M. Vengris, I.H.M. van Stokkum, X. He, A.F. Bell, P.J. Tonge, R. van Grondelle, D.S. Larsen

610

O p t i m a l Control of Femtosecond Photoisomerization of Retinal in Rhodopsin: Effects of Conical Intersections M. Abe, Y. Ohtsuki, Y. Fujimura, W. Domcke

613

Ultrafast Polarization and Vibrational Motions in Bacteriorhodopsin Studied by Coherent Infrared Emission Spectroscopy

XXVI

H.A. Colonna, G.L Groma, J.-C. Lambry, M. Joffre, J.-L. Martin, M.H. Vos

616

Excited-State Dynamics of the IBu-f-, 3Ag—, and IBu— States in All-Trans-Spirilloxanthin as Revealed by Sub-5-fs Time-Resolved Absorption Spectroscopy T. Kobayashi, K. Nishimura, F.S. Rondonuwu, Y. Koyama

619

Ultrafast Relaxation Inside Proteins: Calculation and Measurement of Electron-Vibration Coupling in Enzymes B.M. Cho, R.a Walker, LP. Mercer, LR. Gould, D.R, Klug

622

Direct Observations of Ligand Rebinding Trajectories in Myoglobin by Femtosecond Mid-IR Spectroscopy S. Kim and M. Lim

625

Coherent Vibrational Climbing in Carboxy-Hemoglobin C. Ventalon, J.M, Eraser, M.H. Vos, A. Alexandrou, J.-L. Martin, M. Joffre

628

Evidence for Non-Separating Four-Point Correlation Functions From IR Pump-Probe Spectroscopy of CO in a Protein Internal Cavity J. Helbing, P. Hamm, K. Nienhaus, G. U. Nienhaus

631

The CO Oscillator as a Probe of Ligand Dissociation Dynamics in Myoglobin J.P. Ogilvie, T. Polack, S. Franzen, M.H. Vos, M. Joffre, J.-L. Martin, A. Alexandrou

634

A 2DIR Study of Backbone Structure and Dynamics of a Dipeptide in Membrane V. Volkov and P. Hamm

637

Engineering Cost Function for Optimizing Coherent Control between Processes with Different Nonlinearities J. Chen, H. Kawano, Y. Nabekawa, H. Mizuno, A. Miyawaki, T. Tanabe, F. Kannari, K. Midorikawa

640

Part IX Ultrafast Nanostructure Photonics and Plasmon Imaging of Localized Silver Plasmon Dynamics with Sub-fs Time and Nano-Meter Spatial Resolution A. Kubo, K. Onda, H. Petek, Z. Sun, Y.S. Jung, H.K. Kim

645

XXVII

Ultrafast Dynamics of Light Transmission t h r o u g h Plasmonic Crystals C. Ropers, R. Miiller, C. Lienau, G. Stibenz, G. Steinmeyer, D.-J. Park, Y.-C. Yoon, D.-S. Kim

650

Ultrafast Near-Field Microscope Imaging of Electron a n d P h o n o n Relaxation in Single Gold Nanoparticle K. Imura, T. Nagahara, H. Okamoto

655

P l a s m o n E n h a n c e d Ultrafast Optical Transmission in Metallic N a n o - A r r a y s A. Dechant and A. Y. Elezzabi

658

Excitation and P r o p a g a t i o n of Surface P l a s m o n Polaritons on Metallic Periodic S t r u c t u r e s G. Torosyan, C. Rau, B. Pradarutti, R. Beigang

661

Ultrafast Dynamics of Periodic Arrays of Holes in a Gold Film V. Halte, A. Benabbas, L. Guidoni, J.-Y. Bigot, A. Degiron, H.J. Lezec, T. W. Ebbesen, P. N. Saeta

664

Surface P l a s m o n Assisted 26 fs, 0.4 keV Electron P u l s e Generation S.E. Irvine and A. Y. Elezzabi

667

Space-Time Control in Ultrafast Nano-Optics T. Brixner, J. Schneider, W. Pfeiffer, F.J. Garcia de Abajo

670

Coherent Control of Ultrafast Linear and Nonlinear Optical P h e n o m e n a in N a n o s t r u c t u r e s M.I. Stockman, D.J. Bergman, T. Kobayashi

673

S P A S E R as Ultrafast Nanoscale P h e n o m e n o n and Device M.I. Stockman and D.J. Bergman

676

Ultrafast Quenching of t h e Ring Closure in Molecular Switches, Self-Assembled on Gold Nanoparticles R. Hania, A. Pugzlys, T. Kudernac, H.T. Jonkman, K. Duppen

679

P a r t X Terahertz Wave and Applications Temporal Spectroscopic Behavior of Terahertz Pulses T r a n s m i t t e d t h r o u g h M e t a l Hole A r r a y s F. Miyamaru and M. Hangyo

XXVIII

685

Surface-Plasraon-Polariton E n h a n c e d Tunneling of T H z R a d i a t i o n t h r o u g h Arrays of Sub-Wavelength A p e r t u r e s J.G. Rivas, C. Janke, P.H. Bolivar, H. Kurz

690

T e r a h e r t z Access t o t h e Nanoworld R. Kersting, H.-T. Chen, N. Karppwicz, G.C. Cho .

693

T e r a h e r t z Surface Plaslmon Polariton Coupling on Metallic Grating Structures J.F. O'Hara, R.D. Averitt, A.J. Taylor

696

Control of T H z Transmission t h r o u g h Two-Dimensional Metallic P h o t o n i c Crystals C.-L. Pan, C.-F. Hsieh, R.-P. Pan, M. Tanaka, F. Miyamaru, M. Tani, M. Hangyo

699

Teflon P h o t o n i c Crystal Fiber as Polarization-Preserving Waveguide in T H z Region M. Goto, A. Quema, H. Takahashi, S. Ono, N. Sarukura

702

G e n e r a t i o n of Coherent Tunable T H z Waves by Using Birefringent Crystal and G r a t i n g P a i r R. Yano, H. Gotoh, T. Hattori

705

U l t r a - W i d e B a n d w i d t h T H z Emission from a Semiconductor I r r a d i a t e d with Intense, Radially Polarized, Bessel-Gauss Pulses K.J. Chau and A. Y. Elezzabi

708

Mechanism Crossover of Terahertz Radiation from I n A s Surface Induced by a Magnetic Field at High Density Excitation M. Nakajima, Y. Oda, S. Saito, T. Suemoto

711

Transient G r a t i n g Generation a n d Waveform Shaping of Free-Space P r o p a g a t i n g , Picosecond, N a r r o w - B a n d T H z Radiation A.G. Stepanov, J. Hebling, J. Kuhl

714

T y p e s e t t i n g T H z Waveforms J.G. Vaughan, T. Feurer, T. Hornung, K.A. Nelson

717

Magnetically Induced Evolution of Terahetz R a d i a t i o n S p e c t r u m E m i t t e d from InAs u p t o 27T H. Takahashi, A. Quema, M. Goto, S. Ono, N. Sarukura, G. Nishijima, K. Watanahe

720

XXIX

A Liquid Crystal P h a s e Shifter with a Tuning R a n g e of Over 360 Degrees a r o u n d 1 T H z O.-Y. Chen, C.-F, Hsieh, R.-P. Chao, C.-L. Pan

723

Cooper Pair Breaking Dynamics in M g B 2 Using OpticalP u m p T e r a h e r t z - P r o b e Spectroscopy J. Demsar, R.D. Averitt, A.J. Taylor, V. V. Kabanov

726

Femtosecond Formation of P h o n o n - P l a s m o n Coupled M o d e s Studied by U l t r a b r o a d b a n d T H z Spectrocopy R. Huher, C. Kiibler, S. Tuhel, A. Brodschelm, F. Kohler, M.-C. Amann, A. Leitenstorfer

729

Solid-State P h a s e Transition Onset Detection in EstrogenLike Chemical via Terahertz Transmission Spectroscopy A.V. Quema, M. Goto, M. Sakai, G. Janairo, R. El Ouenzerfi, H. Takahashi, S. Ono, N. Sarukura

732

M a x i m u m E n t r o p y M e t h o d for Misplacement P h a s e E r r o r Correction in Terahertz Time-Domain Reflection Spectroscopy Y. Ino, R. Shimano, M. Kuwata-Gonokami, E.M. Vartiainen, Y.P. Svirko, K.E. Peiponen

735

P u l s e d T e r a h e r t z Spectroscopy and Imaging Applied t o Inspection of Explosives a n d Inflammable Liquids K. Yamamoto, M. Yamaguchi, F. Miyamaru, M. Tani, M. Hangyo, T. Ikeda, A. Matsushita, K. Koide, M. Tatsuno, Y. Minami

738

T e r a h e r t z T i m e - D o m a i n Spectroscopy of Surface P l a s m o n Polaritons on Semiconductor Surfaces J.G. Rivas, J. Saxler, M. Kuttge, P.H. Bolivar, H. Kurz

741

Evaluation of C o m p l e x Optical C o n s t a n t s of Semiconductor Wafers Using T e r a h e r t z Ellipsometry T. Nagashima and M. Hangyo

744

T e r a h e r t z Two-Dimensional Spectroscopic Imaging w i t h a High Speed C M O S C a m e r a H. Kitahara, T. Yonera, F. Miyamaru, M. Tani, M. Hangyo

747

Single-Shot T e r a h e r t z Imaging R. Rungsawang, A. Mochiduki, S.-I. Okuma, T. Hattori

750

U l t r a b r o a d b a n d Detection of M u l t i - T H z Field Transients w i t h GaSe Electro-Optic Sensors C. Kuhler, R. Huher, S. Tuhel, A. Leitenstorfer

753

XXX

T e t a h e r t z Field Detection beyond 30 T H z by P r o t o n Bombarded InP Photoconductive Antennas T.-A. Liu, M. Tani, M. Nakajima, M. Hangyo, K. Sakai, S.-I. Nakashima, C.-L. Pan

756

T H z Wave Near-Field Emission Microscope T. Yuan, H. Park, J, Xu, H. Han, X.-C. Zhang

759

P a r t XI Optoelectronics and O t h e r Applications Optimization of a 40 G H z Regeneratively a n d Harmonically Mode-Locked F i b e r Laser u n d e r P L L O p e r a t i o n a n d its Longitudinal M o d e Characteristics M. Yoshida, T. Yaguchi, S. Harada, M. Nakazawa

765

Femtosecond Synchronization of RF-Signals w i t h Optical P u l s e Trains J.-W. Kim, M.H, Perrott, F.X. KaeHner

768

Fast P h o t o - I n d u c e d P h a s e Switching in Organic C o n d u c t o r Crystal; ( E D O - T T F ) 2 P F 6 M.C. Chollet, L. Guerin, N. Uchida, S. Fukaya, T. Ishikawa, S.-Y. Koshihara, K. Matsuda, A. Ota, H. Yamochi, G. Saito

771

Molecular P h a s e - t o - A m p l i t u d e Converter Using Femtosecond Wave Packet Engineering /. Matsuda, K. Misawa, N.T. Hashimoto, R. Lang

774

Femtosecond Pulse Recoding and R e g e n e r a t i o n by a T w o - P h o t o n G a t e d Periodic Diffractive Optics H. Nishioka, H. Tomita, K.-L Ueda

777

Linewidth and R I N M e a s u r e m e n t s of Longitudinal M o d e s in Ultrahigh-Speed Mode-Locked Laser Diodes K. Haneda, H. Yokoyama, Y. Ogawa, M. Nakazawa

780

C h a r g e G e n e r a t i o n in I n o r g a n i c / O r g a n i c Photovoltaic Blends 5. Westenhoff, S. C, Hayes, N. C. Greenham, C. Silva

783

E n h a n c e d Polariton Decay in LiNb03 D u e t o Stimulated Emission of Acoustic P h o n o n s J. Hebling, A.G. Stepanov, G. Almdsi, J. Kuhl

786

Ultrafast Electro optic Deflector Using Quasi-VelocityMatching K. Shibuya, S. Hisatake, H. Kitano, T. Kobayashi

789

XXXI

Ultrafast Control of a Surface P l a s m o n Resonance via t h e Insulator t o M e t a l Transition in V 0 2 Nanoparticles M. Rini, A. Cavalleri, R. Lopez, L.A. Boatner, R.F. Haglund Jr., T.E. Haynes, L.C. Feldman, R. W. Schoenlein

792

E x t e r n a l G e n e r a t i o n of Flat Power-Envelope T H z M o d u l a t i o n Sidebands from a C W Laser Based on a n Electrooptic P h a s e Modulator S. Hisatake, Y. Nakase, K. Shihuya, M. Tobinaga, T. Kobayashi

795

P a r t X I I Microfabrication by Femtosecond Laser Pulses 3D P h o t o n i c Devices Fabricated in Glass by a Femtosecond Oscillator A.M. Kowalevicz, V. Sharma, E.P. Ippen, J.G. Fujimoto, K. MinoshimaSOl Writing of P h o t o n i c Devices and Waveguide Lasers by a D i o d e - P u m p e d Femtosecond Oscillator R. Osellame, N. Chiodo, G. Delia Voile, S. Taccheo, R. Ramponi, G. Cerullo, A. Killi, U. Morgner, M. Lederer, D. Kopf

804

Toward t h e Fabrication of H y b r i d P o l y m e r / M e t a l Three-Dimensional M i c r o s t r u c t u r e s T. Baldacchini, C.N. LaFratta, R.A. Farrer, A.C. Pons, J. Pons, M.J. Naughton, B.E.A. Saleh, M.C. Teich, J.T. Fourkas

807

P r o d u c t i o n of 3D, Dichroitic M i c r o s t r u c t u r e s in N a n o c o m p o s i t e Glasses by Femtosecond Laser Pulses G. Seifert, A.V. Podlipensky, A. Abdolvand, J. Lange, H. Graener . . . . 810 M i c r o m e t e r and Sub-Micrometer S t r u c t u r e s Fabrication a n d Analysis w i t h Femtosecond Laser Micro-Nanomachining System E. Vanagas, J. Kawai, Y. Zaparozhchanka, D. Tuzhilin, H. Musasa, P.I Rutkovski, I. Kudryashov, S. Suruga

813

Femtosecond Laser Effects on Osseous Tissues B. Girard, D. Yu, M.R. Armstrong, B.C. Wilson, CM.L. R.J.D. Miller

816

Clokie,

Femtosecond Laser Material Processing: How short is s h o r t ? Y. Prior, K. Zhang, V. Batenkov, Y. Paskover, I.S. Averbukh, F. Korte, C Fallnich

XXXII

819

D i o d e - P u m p e d C r 3 + : L i C A F Laser for Ultrahigh Resolution Optical Coherence Tomography P.C. Wagenhlast, T.H. Ko, V. Sharma, U. Morgner, J.G. Fujimoto, F.X. Kaertner

822

Time-Resolved Electron Imaging of Femtosecond Laser Ablation Y. Okano, Y. Hironaka, K.-L Kondo, K.G. Nakamura

825

P a r t X I I I Frequency Stabilization a n d Ultrawide Frequency C o m b A N e w U l t r a s t a b l e Cesium Optical Atomic Clock w i t h a 9.1926-GHz Regeneratively Mode-Locked Fiber Laser M. Yakabe, K. Nito, M. Yoshida, M. Nakazawa, Y. Koga, K. Hagimoto, T. Ikegami

831

Frequency Transfer of Optical S t a n d a r d s Through a F i b e r Network Using 1550-nm Mode-Locked Sources K. W. Holman, D.J. Jones, R.J. Jones, J. Ye

834

Femtosecond Laser Optical Frequency Synthesizers w i t h U n c e r t a i n t y at t h e 10-19 Level L.-S. Ma, Z. Bi, A. Bartels, L. Robertsson, M. Zucco, R. Windeler, G. Wilpers, C. Gates, L. Hollberg, S.A. Diddams

837

Femtosecond Laser Frequency Combs with Linewidths at t h e 1-Hz Level A.O. Bartels, S.A. Diddams, G.W. Gates, J.G. Bergquist, L. Hollberg . 840 Frequency Metrology w i t h a Turnkey All-Fiber System T.R. Schibli, K. Minoshima, F.-L. Hong, H. Inaba, A. Gnae, H. Matsumoto, I. Hartl, M.E. Fermann

843

Optical Frequency M e a s u r e m e n t Precision of Femtosecond Laser Optical C o m b System and t h e Stability of its H F Reference Frequency H. Ito, Y. Li, M. Fujieda, M. Imae, M. Hosokawa

846

Evaluation of Oscillation Frequency Stability of a Diode Laser Using a fs Laser Optical C o m b H. Kobayashi, T. Nimonji, A. Sawamura, T. Sato, M. Ghkawa, T. Maruyama, T. Yoshino, H. Kunimori, M. Hosokawa, H. Ro, Y. Li, S. Nagano, S. Kawamura

849

XXXIII

P a r t X I V Coherent Control and O t h e r Topics Coherent Cooling of Molecular Vibrational Motion w i t h Laser-Induced Dipole Forces H. Niikura, P.B. Corkum, DM. Villeneuve

855

Molecular Orientation of C H 3 F Induced by Phase-Controlled Lights H. Ohmura, F. Itoh, M. Tachiya

858

Observation a n d Manipulation of Q u a n t u m Interferences in Ladder Climbing B. Chatel, J. DegeH, S. Stock, B. Girard

861

A d a p t i v e Polarization Control of Molecular Dynamics T. Brixner, G. Krampert, T. Pfeifer, R. Selle, G. Gerher, M. Wollenhaupt, O. Graefe, C. Horn, D. Liese, T. Baumert

864

T w o - P h o t o n Absorption Imaging w i t h S h a p e d Femtosecond Laser Pulses W.S. Warren, A. Miller, W. Wagner, T. Ye, M. Fischer, G. Yurtsever 867 Selective T w o - P h o t o n Functional Imaging t h r o u g h Scattering M e d i a Based on Binary P h a s e Shaping I. Pastirk, J.M. Dela Cruz, M. Comstock, V.V. Lozovoy, M. Dantus. . . 870 P h o t o n N u m b e r Squeezing of U l t r a b r o a d b a n d Pulses G e n e r a t e d by M i c r o s t r u c t u r e Fibers H. Furumochi, A. Tada, K. Hirosawa, F. Kannari, M. Takeoka, M. Nakazawa

873

Real-Time, Ultrahigh-Resolution Optical Coherence Tomography a t 1.5 |im using a Femtosecond F i b e r Laser Continuum A^. Nishizawa, Y. Chen, P.-L. Hsiung, V. Sharma, T.H, Ko, E.P. Ippen, J. G. Fujimoto

876

Ultrafast Exciton T r a n s p o r t in Organic N a n o t u b e s A. Pugzlys, P.R. Hania, C. Didraga, V.A. Malyshev, J. Knoester, K. Duppen

879

Ultrafast Molecule t o Semiconductor Electron Transfer via Different Anchor G r o u p s in Ultra-High V a c u u m R. Ernstorfer, L. Gundlach, S. Felber, R. Eichberger, C Zimmermann, W. Storck, F. Willig

882

XXXIV

Controllability in Dissociative Ionization of Organic Molecules with Pulse-Shaped Intense Laser Fields H. Yazawa, T. Okamoto, T. Yamanaka, F. Kannari, R. Itakura, K. Yamanouchi

885

Laser Coulomb Explosion Imaging for Probing Molecular Structure and Dynamics F. Legare, K.F. Lee, I.V. Litvinyuk, P.W. Dooley, A.D. Bandrauk, D.M, Villeneuve, P.B. Corkum

888

Ultrafast Electron Transfer via a Bridge-Extended Donor Orbital R. Ernstorfer, L. Gundlach, S. Felber, W. Storck, R. Eichherger, C. Zimmermann, F. Willig

891

Multiple Ionization of Atoms by 25 and 7 fs Laser Pulses A. Rudenko, B. Feuerstein, K. Zrost, V.L.B. de Jesus, CD. Schroter, R. Moshammer, J. Ullrich

894

Time Resolved, Phase-Matched Harmonic Generation from Exploding Noble Gas Clusters B. Shim, G. Hays, M. Fomyts^kyi, A. Arefiev, B.s Breizman, T. Ditmire, M. C. Downer

897

Index of Contributors

901

XXXV

Part I

Generation and Measurements

Single-Cycle Optical Pulse Generation David R, Walker, Miroslav Shverdin, Deniz Yavuz, Guang-Yu Yin, and Stephen E. Harris Edward L. Ginzton Laboratory, Stanford University, Stanford, CA 94305, USA

Abstract. By electronically adjusting the phases of seven Raman sidebands which span 1.56 fim to 410 nm we generate a train of well-formed single-cycle optical pulses with a pulsewidth of 1.6 fs. We have recently attained a milestone result: the generation of a train of single-cycle optical pulses with a pulsewidth of 1.6 fs, a pulse spacing of 11 fs, and a peak power of f=5^: 1 MW. The paper describes the generation of these pulses and the first use of a single-cycle pulse for four-wave mixing to the ultraviolet. We use Raman generation at maximum coherence to produce a wide coUinear spectrum spanning 1.9 octaves from 1.56 /xm to 410 nm [1-3]. This generated spectrum is passed through a Uquid crystal spatial Hght modulator where the phases of seven sidebands are independently set to compensate for the dispersion of the following glass window and optics to form a mode-locked train of single-cycle pulses inside a cell of xenon. The synthesized temporal waveform is characterized by using nonresonant four-wave frequency mixing to the ultraviolet as a nonlinear detector. The Raman sidebands mix in xenon to produce six generated ultraviolet frequencies. By changing the relative phases of the incident Raman sidebands with the liquid crystal array, we coherently control the dipole moment of the medium and are able to change the ratio of the intensities of the generated ultraviolet wavelengths. By choosing phases that maximize the four-wave mixing signal at all ultraviolet frequencies, we form the shortest pulse that this spectrum may make* The pulsewidth and shape are determined by electronically synthesizing two pulses which are then cross-correlated. The experimental set-up is shown in Fig. 1. The Raman sidebands are produced by driving the x/" = 0, J'' = 0 —> i/' = 1, J ' = 0 vibrational transition of deuterium (D2) by two transform-limited laser pulses at 1064 nm and 807 nm, such that their frequency difference is slightly detuned from 2994 cm~"^, the (D2) transition frequency. The first laser is a Spectra Physics Quanta Ray GCR-290 Q-switched injection-seeded Nd:YAG, producing 70 mJ, 10 ns pulses at a 10 Hz repetition rate. The second laser is a homemade ringcavity Ti:Sapphire system pumped by the second harmonic of a separate Nd:YAG Spectra Physics Quanta Ray laser. This laser is injection-seeded by a diode laser and produces transform-limited 60 mJ, 15 ns pulses at the

Liquid crystal phase modulator \ ] ^

PMT

Y

^

1+ ^ 1

Xel

4 i4

D2cell

Xe cell Fig, 1. Experimental set-up for pulse generation and characterization. The Raman sidebands are laterally separated with a prism pair, allowing independent control of each sideband. After being recombined with another prism pair, the sidebands are focused into a Xe cell. Four-wave mixing serves as a phase-sensitive nonlinear detector. A soiar-bHnd photomultiplier tube measures the new frequencies produced in the xenon cell. seeding wavelength at 10 Hz. The seeding wavelength of the diode laser is tunable and monitored by a wavemeter. The two laser beams are combined and loosely focused into an 80 cm D2 cell, where the 1064 nm and §07 nm beams focus to spot sizes of 1.1 nam and 600 ^m, respectively. At the focus, the intensity of each laser is about 1 GW/cm^, and the Raman transition is driven about 700 MHz below resonance, allowing us to adiabatically prepare a uaolecular coherence with peak magnitude \pge[ ^ 0 . 1 . The deuterium cell is cooled with liquid N2 aud kept at a pressure of 60 torr and a temperature of 77 K. The cooling decreases the Doppler linewidth and increases the gi'ound state population, dramatically improving the generation efficiency. At the output of the D2 cell, we observe coUinear generation of 15 discrete vSidebands separated by 2994 cm""^ and extending from 1.56 jum to 207 nm. The generated beams are expanded and coUirnated by a pair of fused silica lenses. The beams pass through two interference filters which selectively attenuate the 1064 nm and 807 nm pump beams in order to avoid damaging the liquid crystal spatial light modulator (LCM). Next, the spectrum is dispersed with a pair of fused silica prisms and all sidebands with a wavelength shorter than 410 nm are blocked. The prisms are adjusted to make the dispersed beams parallel as they pass through the LCM. The LCM is a linear array of 640 pixels, each 97 /xm wide and separated by 3 /xm. The refractive index of each pixel is controlled by an applied voltage. This enables us to independently adjust the phase of each sideband. By using the full spectral range of the LCM, we have obtained phase control over seven of our Raman sidebands. After phase adjustment, these sidebands are spatially recombined with another prism pair and focused with an achromatic lens of focal length

20 cnij into an 8 cm long cell, containing xenon (Xe) vapor at a pressure of 100 torr. The four-wave mixing inside the Xe cell is used to characterize the generated pulses. The efftciency of conversion to the ultraviolet ranges, depending on the spectral component, from lO"*^^ to 10^^, These sidebands are detected by a solar-blind photomultiplier.

•

•

•

1

'

1

1

:

1

1

1

•

,•"

• -

I

'

cross-correlation data

c =5

4

4

s

t

la c

to E c d> cvi CO

-6

-4

1

1

-2

;• '

0 2 tirne delay, T (fs)

1

1

2 L

*

.COL

CO 1

e c to

-6

-4

-2

'

,

cro$s-Gorrelation data

-

4 i

1

1

*

4 1 /,

\J * •«

S

1

•

a

1

0 2 time delay, T (fs)

1 ^

• T i l l

4

*

•!

•-

6

Fig. 2. Correlation studies of the Raman spectrum. Shown is the four-wave mixing signal at 329 nm {top figure) and at 365 nm (bottom figure) versus the time delay between two phase-optimized pulses. The pulse consisting of the 1064 nm, 650 nm, and 468 nm sidebands is electronically delayed with respect to a pulse synthesized from the 1560 nm, 807 nm, 544 nm, and 410 nm sidebands. The resulting crosscorrelation data is represented by the solid circles. The dashed line represents the corresponding theoretically obtained cross-correlation. We achieve phase-locking inside the xenon cell by adjusting the phases of the sidebands to maximize generation at one of the ultraviolet frequencies. We

then perform a cross-correlation asfollows:We add to each even sideband a phase proportional to itsfrequencyand leave the phases of the odd sidebands unchanged. This effectively forms two pulses which have a v^lable time delay. By varying this delay, we obtain the cross-correlations of Fig. 2. The basis for the calculation is a formalism for optical frequency conversion with gaussian beams with different confocal parameters [4]. The formalism is modified to include the interference of all four-wave mixing contributions to the dipole moment at each generated ultraviolet frequency. Experiment (solid circles) and theory (dashed lines) are matched at their peak values for each of the two generated wavelengths which are shown. The agreement at all time delays is excellent. Repeated experimental tests have shown that changing the phase of any sideband by a fraction of a radian is sufficient to distort the subsequent correlation trace. Fig* 3 shows the calculated instantaneous electric field as obtained with the measured experimental intensities

-1$

-10

-5

0 time (fs)

5

2

0 time (fs)

2

Pig* 3. Theoretically calculated intensity waveform, showing the pulse train (left) and a single pulse (right). By adjusting the phases of the Raman sidebands, it is possible to produce an envelope with a full width at half maximum of L6 fs. and mode-locked phases. The excellent agreement of theory and experiment in Fig. 2, together with the sensitivity to deliberate phase distortion, substantiates our conclusion that we have obtained single-cycle pulses with a width of 1.6 fs. This work was supported by the U. S. Air Force Office of Scientific Research, the U. S. Army Research Office, and the U. S, Office of Naval Research, and the Fannie and John Hertz Foundation.

References 1. S, E. Harris and A. V. Sokolov: Phys. Rev. Lett. 81, 2894 (1998); A. V. Sokolov, D. R. Walker, D. D. Yavuz, G. Y. Yin, and S. E. Harris: Phys. Rev. Lett. 85, 562 (2000). 2. J, Q. Liang, M. Katsuragawa, F. Le Kien, and K. Hakuta: Phys. Rev. Lett. 85, 2474 (2000); M. Katsuragawa, J. Q. Liang, F. Le Kien, and K. Hakuta: Phys. Rev. A 65, 025801 (2002). 3. A. V. Sokolov, D. R. Walker, D. D. Yavuz, G. Y. Yin, and S. E. Harris: Phys. Rev. Lett. $7, 033402 (2001); D. D. Yavuz, D. R. Walker, M. Y. Shverdin, G. Y. Yin and S. E. Harris, Phys. Rev. Lett. 91, 233602 (2003). 4. G. Hilber, D. J. Brink, A. Lago and R. Wallenstein: Phys. Rev. A 38, 6231-6239 (1988).

Toward a terawatt few-optical-cycle driver laser for attosecond spectroscopy N. Ishii^'^ R. Butkus^ A. Baltuska^'^ E. Goulielmakis^'^ M. Uiberacker^'^ R. Kienberger^'^ T. Fuji\ V. S. Yakovlev\ V. Smilgevicius^ R. Danielius^ A. Piskarskas^, and F. Krausz^'^ ^ Institut fUr Photonik, TechnischeUniversitat Wien, Gusshausstrasse 27/387, A-1040 Vienna, Austria E-mail: [email protected] ^ Max-Planck-Institut fur Quantenoptik, Hans-Kopfermann-Strasse 1, D-85748 Garching, Germany ^ Laser Research Center, Vilnius University, Sauletekio ave. 10, LT-2040 Vilnius, Lithuania Abstract. We discuss routes towards developing an ultra-high peak power phase-stable source of few-cycle laser pulses suitable for driving a wide range of strong-field applications. Experiments with a phase-stable 0.1-TW 5-fs system based on a Ti:sapphire amplifier and the progress in construction of a 1-TW few-cycle optical parametric amplifier are presented. Generation of isolated attosecnd coherent pulses of XUV and soft-X-ray radiation [1-3] is opening, for the first time, an opportunity to perform a direct time- and frequency-domain investigation of atomic and molecular processes that are comparable in their duration to the electron orbiting time. A well-understood and currently favored method for producing such attosecond bursts is the higherorder harmonic generation in gases. This approach relies on a three-step interaction of an atom of gas with an intense driver laser field leading to ionization, electron acceleration by the electric field of the same light pulse, and electron recombination with the parent ion, upon which a high-energy quantum might be emitted [4]. To enable spectroscopy with isolated attosecond pulses, a dedicated driver laser has to be constructed according to a set of specific demands that determine the pulse duration, its intensity, and the position of optical field oscillations under the pulse envelope (referred to, for simplicity, as the carrierenvelope phase, or CEP). These demands can be summarized as follows: 1) the light-matter interaction leading to a non-recursive emission of a XUV-X-ray pulse has to be confined to a single (acting) optical cycle; 2) pre-ionization of the target gas by the optical cycles preceding this interaction should be low; 3) the peak intensity of the acting light cycle should be high; 4) the CEP should be controlled and stabilized; 5) the pulse-to-pulse amplitude fluctuations have to be minimized. Some applications may impose additional requirements, such as control over the time-dependent state of light polarization and a very high pulse contrast with respect to pre-pulses. Essentially, a laser delivering intense quasi-monocycle

CEP-controlled pulses is well suited for the role of an attosecond laboratory workhorse. Recently, we have succeeded in constructing such a laser system by merging the technology of a 5-fs amplifier with the CEP-stabilized Tiisapphire seed oscillator [5]. In this system, the slow CEP drift attributed to the multi-pass Ti:sapphire amplifier is monitored at the amplifier output and is pre-compensated in the CEP control loop of the seed oscillator. In this system, reproducible isolated attosecond pulses can be attained with this laser setup and maintained stable over a period up to several hours. Our setup for generation of isolated attosecond pulses driven by a phase controlled laser system is presented schematically in Fig.l. Ptimplas^r

AOM

Ti:Saosciliator

\^i.iZi"/!:S::.f9<

Sfreteher W

Quasi-mon.ocycIe cos driver laser field • Ih-pass

W pulse

Phase-tocking electronics

Fig. 1. Generic setup of an attosecond pulse source based on a quasi-monocycle carrierenvelope-phase-controlled laser. First, the higher-order harmonic radiation is emitted from a gas target that is exposed to a 5-fs pulse. The highest photon energy of this XUV radiation (also known as the cut-off region) scales proportionally to the intensity of the laser field. Consequently, for a driver pulse with a single most intense half-cycle (the so-called cos-likQ pulse, Fig.2a, top panel), the emission of the cut-off XUV radiation will not be repeated within the same laser pulse, which is indicated by a smooth cut-off spectrum (Fig.2b, top and bottom). Conversely, for a ^m-like driving field, this highest-frequency emission is recursive, which results in modulation of the cut-off region in the XUV spectrum (Fig.2b, center.) By reflecting the most energetic part of the spectrum off a bandpass dielectric mirror, we are able to generate an isolated attosecond pulse if a cos driver field is applied, whereas a series of two bursts appear in the case of a sin carrier wave. By using the technique of attosecond streaking [6,7], a single isolated as well as multiple attosecond pulses can be mapped, as is shown in Fig.2c. The presented technique of attosecond pulse generation is scalable. In particular, an increase of the cut-off frequency also corresponds to a broadened bandwidth of the non-recursively emitted X-rays. Therefore, substantially shorter isolated attosecond pulses should be possible at higher X-ray photon energies. However, the challenges of enhancing the XUV-X-ray photon flux and/or increasing the cut-off frequency demand greater laser intensities. Despite its robustness, the developed 5-fs 0.1-TW Ti:sapphire-based laser system cannot be easily up-scaled to meet the intensity demand. One very significant technical difficulty is the limitation of the traditional hollow-fiber compression technique producing the required spectral broadening at the expense of ionization losses, which are mounting with the increase of the pulse intensity. On the other hand,

the pulses emitted directly from a Ti:sapphire amplifier are not suitably short. Another significant technological difficulty in increasing the average output power of such a laser amplifier is the thermal management of the Ti:sapphire crystal which inevitably links the amplified pulse repetition rate and the pulse energy. Searching for a more economical approach, we have designed a laser system based on optical parametric chirped pulse amplification (OPCPA) [8,9].

-5

0

Time [fs]

5

10

110

120

130

Photon energy [eV]

140

50

60

70

80

90

Photoelectron energy [eV]

Fig. 2. Generation of isolated attosecond pulses from the spectral edge of the higher-order harmonic emission, (a) optical driver fields for to various settings of the carrier-envelope phase, (b) corresponding XUV spectra in the cut-off region, (c) streak images proving the presence of an isolated attosecond pulse (for "cos"- and "-cos"-like fields) and a double pulse (for the "sin" field). The dashed curve in (b) suggests a possible bandpass filter for isolating an attosecond pulse from a train of such pulses emitted at lower photon energies. Differently from the OPCPA schemes published in the literature, we are using '^60-ps-long pump pulses. In addition, the seed and pump pulses are derived from two independent lasers with no bandwidth overlap. In our scheme, the repetition rate of the pump laser is actively synchronized with that of a broadband Ti: sapphire seed oscillator. The factors that determine our choice of the pump source are the complexity, availability, and the maintenance cost of an appropriate picosecond laser; the parametric gain and amplification saturation for a given pump pulse duration and intensity; the stretching and recompression ratio of the signal pulse; optical damage of parametric crystals; and the precision of the seedpump pulse synchronization. The layout of our system is depicted in Fig.3 and the parameters of the amplified pulses are shown in Fig. 4. To provide an adequately broadband gain of parametric amplification, we have employed non-collinear phase matching in BBO and pumped the crystal with the second harmonic of a Nd:YAG amplifier. Whereas it is possible to generate 4-fs pulses using the pump wavelength around 400 nm [10], for the 532-nm pump, the gain bandwidth is significantly reduced and covers only about 125 THz. Pump pulses are obtained from a 20-Hz repetition-rate amplifier that is seeded by an independent picosecond seed oscillator. The repetition rates of the ps oscillator and of the fs Tiisapphire oscillator are actively locked to an external master clock with an estimated rms timing jitter below 2 ps. The broadband pulses from the fs oscillator are stretched in a negative dispersion stretcher to match the pump pulse duration. To pre-

10

compensate higher order dispersion of the bulk material in the amplifier chain, our stretcher incorporates specially designed micromachined optics and a deformable mirror, both used in the Fourier plane of a 4/grating stretcher. .-^^j£9J?j£.§yPJif2C9Dl?§l!£0. 76 MHz 1064 nm, 60 ps seed osctilator

SPIDER

Fig. 3. Overview of the two-stage, four-pass chirped pulse parametric amplifier. OPA 1 and 2, non-collinearly phase-matched Type I 4-mm BBO; HTG, holographic transmission diffraction grating; PL, parabolic lens; AP, micromachined aspheric fused silica plate; TDM, thermally activated deformable mirror; X/2, half wave plate @ 532 nm; TFP, thin film polarizer; CM, positive dispersion dielectric chirped mirror; SF57, Schott SF57 glass; FS, Suprasil synthetic quartz glass.

Fig. 4. Summary of OPCPA performance, (a) throughput of the negative dispersion stretcher. Solid curve, oscillator spectrum; dashed curve, spectrum transmitted through the stretcher; dash-dotted curve; theoretical transmission of the grating in 4 passes, (b) amplified pulse spectrum (shaded contour) and residual group delay (dashed curve), (c) recompressed pulse amplified to a 4-mJ energy(dash-dotted curve) and temporal phase (dashed curve). Dark contour in (c) shows the intensity profile of an ideally compressed pulse.

For adaptive phase correction of the amplified pulses, as well as for optimization of the spectral shape of the amplified pulse, we employ a 45-mmlong DAZZLER (Fastlite Ltd.), whose positive dispersion is also pre-compensated by the grating stretcher. The stretcher truncates the input seed spectrum and limits

11

its usable width to the 700-1040-nm spectral range (Fig.4a). The seed pulses are amplified in a two-stage, four-pass parametric amplifier to energies up to 8 mJ if the pump energy of the second stage is 40-mJ. The pump intensities of each stage are carefully chosen to reduce parametric gain and avoid competition between spontaneous emission (superfluorescence) and the external seed. Since the pump pulse duration exceeds the crystal length, it is also very important to minimize double reflections of the signal wave off the crystal faces to prevent repetitive OPA seeding. Because of these restrictions, the energy of reliably compressible amplified pulses is currently limited to 4-5 mJ. The duration of the partially compressed amplified pulse, obtained from a SPIDER measurement, is - 1 0 fs (Fig.4c), whereas the spectrum limited duration is <7 fs. A higher degree of pulse compression is expected upon improving the phase measurement setup and its calibration. The modulation seen in the amplified spectrum (Fig.4b) is currently unexplained and is probably associated with the use of the stretcher, described above. Such modulation was not observed in the pulses of the same energy which were amplified using an all-glass stretcher. Because of the negligible thermal load on the parametric crystals, we foresee that no changes in the developed amplifier design will be required to upgrade the repetition rate to a multi-kHz regime in the near future. Acknowledgements. This work was supported by the FWF (Austria, grants P15382, Z63, and F016) and by the European ATTO Network.

References 1. M. Hentschel, R. Kienberger, C. Spielmann, G. A. Reider, N. Milosevic, T. Brabec, P. Corkum, U. Heinzmann, M. Drescher, and F. Krausz, Nature, 414, 509, 2001. 2. M. Drescher, M. Hentschel, R. Kienberger, M. Uiberacker, V. Yakovlev, A. Scrinzi, T. Westerwalbesloh, U. Kleineberg, U. Heinzmann, and F. Krausz, Nature, 419, 803, 2002. 3. R. Kienberger, M. Hentschel, M. Uiberacker, C. Spielmann, M. Kitzler, A. Scrinzi, M. Wieland, T. Westerwalbesloh, U. Kleineberg, U. Heinzmann, M. Drescher, and F. Krausz, Science, 297, 1144, 2002. 4. P. B. Corkum, Phys. Rev. Lett., 71, 1994, 1993. 5. A. Baltuska, Th. Udem, M. Uiberacker, M. Hentschel, E. Goulielmakis, Ch. Gohle, R. Holzwarth, V. S. Yakovlev, A. Scrinzi, T. W. Hansch, and F. Krausz, Nature, 421, 611, 2003. 6. J. Itatani, F. Quere, G.L. Yudin, M.Yu. Ivanov., F. Krausz, P.B. Corkum, , Phys. Rev. Lett., 88, 173903, 2002. 7. R. Kienberger, E. Goulielmakis, M. Uiberacker, A. Baltuska, V. Yakovlev, F. Bammer, A. Scrinzi, Th. Westerwalbesloh, U. Kleineberg, U, Heinzmann, M. Drescher, and F. Krausz, Nature, 427, 817, 2004. 8. A. Dubietis, G. Jonusaskas, and A. Piskarskas, Opt. Commun., 88, 437, 1992. 9. L N. Ross, P. Matousek, M. Towrie, A. J. Langley, and J. L. Collier, Opt. Commun., 144, 125, 1997. 10. A. Baltuska, T. Fuji, and T. Kobayashi, Opt. Lett., 27, 306, 2002.

12

2.8-fs clean single transform-limited optical-pulse generation and characterization Keisaku Yamane, Toshihiko Kito, Ryuji Morita and Mikio Yamashita Department of Applied Physics, Hokkaido University, Kita-13, Nishi-8, Kita-ku, Sapporo, 060-8628 Japan A b s t r a c t . We compensated for chirp of ultrabroad-band pulses with an over-oneoctave bandwidth (460 - 1060 nm) using a feedback method. Consequently, 2.8-fs, 1.5-cycle transform-limited pulses were generated. In recent years, a lot of efforts for the ultrashort optical pulse generation are dedicated, and even sub-4-fs pulse generation was reported in some groups [1-3]. We already demonstrated pulse compression to 3.3 fs-^ for pulses with an over-one-octave bandwidth using a feedback phase compensation system [3] which consists of a modified S P I D E R (M-SPIDER) apparatus with a high sensitivity [4,5] and an active phase compensator with a large pixelnumber LC-SLM [4,6]. Recently, by further important improvements, we have succeeded to generate transform-limited (TL) optical pulses in the monocycle region. In this paper, we report the 2.8-fs, 1.5-cycle optical pulse generation. T h e experimental setup was almost the same as ref. [3]. We let femtosecond optical pulses (center wavelength: 790 nm, pulse energy: 133/iJ, pulse width: < 30 fs, repetition rate: I k H z ) propagate in a hollow fiber (length: / — 340 mm, core diameter: D = 0 . 1 mm) filled with 3.0-atm argon gas, and generated ultrabroad-band pulses by self phase modulation. Generated pulses were guided into the 4 - / phase compensator with the SLM. O u t p u t pulses ( ~ 5 0 0 n J ) from the active compensator were measured by the M-SPIDER apparatus (external reference pulses: <30fs, 790nm, 3.7/iJ). For further pulse compression, we paid much attention to the following important points: First, we optimized the spectral phase function to reduce variations of the applied spectral phase in each pixel of the SLM (load per pixel) by choosing the suitable combination of a constant phase and a constant group delay. Curves in Fig. 1(A) show spectral phase functions which have minimum phase values at different wavelengths (Amins) of (i) 500 nm, (ii) 515 nm, (iii) 600 nm, (iv) 700 nm and (v) 800 nm, and curves in Fig. 1(B) show the corresponding load per pixel as a function of the wavelength. For Amin = 600, 700 and 800 nm, the load exceeds TT in the region of < ~ 5 0 0 nm. On the other hand, for Amin = 500 nm the load has the value of ~ TT in ~ 6 3 0 ~ 7 7 0 nm. In the region where the load per pixel has such too large value, correct phase compensation is impossible. Therefore, we set Amin to 515 nm so t h a t we can perform as appropriate phase compensation as possible in ^ "3.4 fs" in ref. [3] was corrected to "3.3 fs" by precise recalibration of instruments

13

•2500

A,min - (i) 500 nm (ii) 515 nm (iii)600nm (iv) 700 nm (v)800nm V \ ^0\ \

§

0

(B)

^^^

"

vv..->J
•

600 700 800 Wavelength [nm]

500

"">*

•^iii

900 X

1000

Fig. 1. (a) Spectral phases at different An (b) A-dependency of phase modulation load per pixel at diflFerent Am the whole wavelength region. As a result, the structured spectral-amplitude modulation occurring sometimes owing to the 0-27r applied-phase j u m p in the short-wavelength region (, which hampers obtaining the robust S P I D E R signal in its wavelength region and hence makes reliable spectral-phase determination greatly difficult) was avoided completely. Second, the precise alignment of the 4 - / phase compensator was done. T h a t is, the negative 2ndorder wave (which is originally the negative Ist-order wave deflected by the input grating ( G l ) ) deflected by the output grating (G2) was interfered and overlapped angular-spatially with the negative 3rd-order wave (which is originally the negative 2nd-order wave deflected by G l ) deflected by G2 at the

(ii)

500

../•'"460 - 1060 nm (Av = 369 THz) (iii)

^-n

500

600

700 800 900 Wavelength [nm]

jz p^

1000

Fig. 2. Results of feedback phase compensation. (A) (i) Intensity spectrum. Spectral phases (ii) before, (iii) after 1st and (iv) after 2nd feedback chirp compensations. (B) detail of (iv).

14

spatially displaced position of the compensator output, by moving slightly b o t h distances of G2 and G l . Consequently, both positions of input 0 1 and output G2 gratings were adjusted to make the 4 - / configuration precisely dispersion-free. Figure 2 shows the spectral intensity and phase of pulses we generated. Pulses had an ultrabroad-band spectrum from 460 to 1060 nm which was broader and shorter t h a n our previous one [3], and the spectral phase profile over the whole wavelength range was reconstructed. We applied its negative phase to chirped pulses by the programmable compensator. After second feedback compensation, the spectral phase was almost completely flattened (Fig. 2(B)). T h e corresponding temporal intensity and phase were retrieved. We found t h a t side-lobes were almost eliminated and the duration was 2.8 fs (center wavelength: 580 nm) (see Fig. 3). In addition, the profile of generated pulses was almost the same as its transform-limited (TL) pulses (2.75fs). This is the shortest clean single TL pulse in the visible and near-infrared region.

0 Time [fs] Fig. 3. Retrieved temporal profiles. Pulse intensities (A) before and (B) (i) after 2nd feedback compensations and (ii) temporal phase after 2nd feedback compensation, (iii) Temporal intensity of Fourier transform limited pulses.

References 1. A. Baltuska, T. Fuji, and T. Kobayashi, Opt. Lett. 27, 306 (2002). 2. B. Schenkel, J. Biegert, U. Keller, C. Vozzi, M. Nisoli, G. Sansone, S. Stagira, S. De Silvestri, O. Svelto, Opt. Lett. 28, 1987 (2003). 3. K. Yamane, Z. Zhang, K. Oka, R. Morita, M. Yamashita, A. Suguro, Opt. Lett. 28, 2258 (2003). 4. R. Morita, M. Hirasawa, N. Karasawa, S. Kusaka, N. Nakagawa, K. Yamane, L. Li, A. Suguro, M. Yamashita, Meas. Sci. Technol. 13, 1710-1720 (2002). 5. M. Hirasawa, N. Nakagawa, K. Yamamoto, R. Morita, H. Shigekawa, M. Yamashita, Appl. Phys. B, 74, S291 (2002). 6. N. Karasawa, L. Li, A. Suguro, H. Shigekawa, R. Morita and M. Yamashita, J. Opt. Soc. Am. B 18, 1742 (2001).

15

Coherent amplification of femtosecond pulses with passive enhancement cavities R. Jason Jones\ Long-Sheng Ma^, and Jun Ye^ ^ JILA, University of Colorado and National InstittJte of Standards and Technology University of Colorado, Boulder CO 80309, USA E-mail: rjjones@jilaul .colorado.edu ^ Department of Physics, East China Normal University, Shanghai, China Abstract. We demonstrate a general technique for amplification of femtosecond pulses through coherent buildup in an external cavity. A net single-pulse energy gain of 42 to more than 70 times for 38 to 58 fs pulse durations, respectively, is achieved. This technique offers a flexible approach to enhance the energy of fs pulses while maintaining relatively high repetition rates.

1.

Introduction

Many applications require higher pulse energies than are typically available fi*om standard femtosecond laser oscillators. Recent advances permit lOO's of nJ pulse energies to be obtained directly from the laser [1,2]. There are numerous applications which can benefit fi-om simple, cost effective systems capable of producing pulse energies in the 100 nJ to 100 /iJ range. In previous work we proposed the use of high-fmesse enhancement cavities, specifically designed to compensate for linear dispersion, to coherently accumulate pulses until a fast switch was enabled to eject a single amplified pulse [3]. The stored intracavity field is the result of the coherent addition of pulses accumulated over the lifetime of the cavity. A fast acousto-optic modulator (AOM) can be used to switch out the stored pulse. Greater single-pulse amplifications are achieved using higher-finesse cavities (longer cavity lifetimes) at the expense of a reduction in the pulse repetition rate. Likewise, higher repetition rates can be maintained with more modest pulse amplifications. This concept was recently demonstrated with picosecond pulses [4] in which the limiting effects of cavity dispersion do not play a significant role. Here we demonstrate realization of this technique in the femtosecond regime for the first time using "fs enhancement cavities" to store and enhance pulses as short as 39 fs. This approach enables generation of lOO's of nJ pulse energies using only standard fs lasers or could potentially be used to compliment recent high-energy fs oscillators to produce pulse energies >10 /iJ.

2.

Experimental Methods

Figure 1 shows a schematic of the passive femtosecond pulse amplifier. The femtosecond enhancement cavity incorporates low loss negative group-delay

16

Fig. 1. Simplified schematic of pulse amplification with a fs enhancement cavity. Coherent accumulation and subsequent dumping of the passive cavity results in amplified pulse energies at repetition rates reduced "n" times. dispersion (GDD) mirrors to provide high cavity finesse and zero net cavity GDD. The cavity is placed in a vacuum chamber and pressure tuned for fine adjustment of the cavity dispersion. Error signals can be obtained for locking both degrees of fi-eedom of the pulse train to the cavity [5]. The first error signal (eO is sent to the enhancement cavity to keep the phase/frequency of the stored pulse resonant with the incident pulse train. The second error signal (e2) is fed back to the laser to lock the pulse repetition rate to the cavity. The overall gain achieved depends on intracavity losses, impedance/mode-matching, dumping efficiency of the intracavity AOM, and the pulse dumping rate for a given cavity finesse. With a 0.9 % input coupler at 800 nm and a pulse dumping efficiency of ~40%, the overall gain (G) obtained with the fs enhancement cavity varied from 42 times with 39 fs pulses to over 70 times with 52 fs pulses for dumping rates below 300 kHz. Measurements of fs pulse enhancement for this range are shown in Fig. 2. The decreased gain for larger incident bandwidths is consistent with our calculations showing spectral filtering by the cavity due to the third-order dispersion of the intracavity AOM (indicated by the gray lines in Fig. 2). The high intracavity peak intensity can result in a nonlinear phase shift acquired by the pulse. This deteriorates the degree of complete constructive interference maintained with the incident pulse train, thus limiting the intracavity buildup. Sufficient stretching of the incident pulses to minimize the peak intracavity intensity allows linear enhancement and recompression of the output pulse. With the pulses stretched to over 100 ps, the only limitation on the achieved energy per pulse was due to the inefficiency of the grating-based stretcher/compressor (--iS % each) and the limited power available firom the Ti: sapphire laser. For Figs. 2(b) and 2(d), the incident pulse energy of 2.5 nJ was increased 70 times to 175 nJ before recompression. When the laser was adjusted to produce 58 fs pulses, the incident pulse energy available went up to 3 nJ, resulting in enhanced pulse energies of 210 nJ fi-om the cavity before recompression back to 58 fs. A prismbased stretcher/compressor system provides much higher throughput and will make more efficient use of the power available from the laser, enabling pulse energies greater than 500 nJ starting with only conventional oscillators.

17

^j^<-~..^

(a) /

f /•/

/A -"'"•"•""" .. J 780

\

V

G=42 X \

\

\

Vu.,—

800 820 840 Wavelength (nm)

0 time (fs)

50

860

-50

0 time (fs)

50

Fig. 2. Pulse spectrum and recompressed pulse intensity and phase measured by FROG after cavity enhancement, (a) and (c) show the incident (dotted) and intracavity (soHd) pulse spectrum for two different incident pulse bandwidths. Solid gray line indicates the calculated spectral transferftmctionbased on the third-order dispersion of the intracavity fused silica, (b) and (d) show the corresponding pulse measurements of 39 fs and 52 fs pulses (FWHM), respectively.

4.

Conclusions

In conclusion, we have demonstrated the use of fs enhancement cavities for amplifying fs pulse energies through coherent addition of multiple pulses in a highfinesse cavity. We have identified limitations (solutions) to achievement of better performance in terms of both the peak power (pulse stretching and recompression) and intracavity dispersion (higher order compensation). Nonlinear interactions between fs pulses and intracavity samples can also be studied in this way. Even greater enhancements leading to much higher field strengths can be obtained when studying pulse interactions with more dilute intracavity samples.

References A. Fernandez, T. Fuji, A. Poppe, A. Furbach, F. Krausz, and A. Apolonski, Opt. Lett. 29, 1366(2004). A. M. Kowalevicz, Jr., A. T. Zare, F. X. Kartner, J. G. Fujimoto, S. Dewald, U. Morgner, V. Scheuer, and G. Angelow, Opt. Lett. 28, 1597 (2003). R. J. Jones and J. Ye, Opt. Lett. 27, 1848 (2002). E. O. Potma, C. Evans, X. S. Xie, R. J. Jones, and J. Ye, Opt. Lett. 28, 1835 (2003). R. J. Jones, L Thomann, and J. Ye, Phys. Rev. A 69, 051803 (2004).

18

Temporal and Spatial Pulse Compression in a Nonlinear Defocusing Material N. C. Nielsen^ T. Honer zu Siederdissen^ J. Kuhl\ M. Schaarschmidt^, J. Forstner^ A. Knorr^ S. W. Koch^ and H. Giessen"^ ^ Max-Planck-Institut flir Festkorperforschung, 70569 Stuttgart, Germany E-mail: [email protected] ^ Institut fur Theoretische Physik, Technische Universitat Berlin, 10623 Berlin, Germany ^ Department of Physics and Material Sciences Center, Philipps-Universitat, 35032 Marburg, Germany "^ Institute of Applied Physics, University of Bonn, 53115 Bonn, Germany Abstract. We investigate the spatiotemporal characteristics of subpicosecond pulse propagation in the nonlinear defocusing regime below the band edge of bulk GaAs. We observe temporal and spatial pulse compression and instabilities.

1.

Introduction

Temporal (dispersive) and transverse spatial (diffractive) effects become mutually coupled when an ultrashort pulse propagates through a dispersive bulk medium that exhibits a Kerr-type nonlinearity [1]. This nonlinear spatiotemporal coupling has been scarcely investigated so far. Much more effort was made to study propagation in fibers and one-dimensional waveguide structures, in which transverse confinement maintains the intensity over long propagation distances and allows to exclude diffractive effects. In these structures, the interplay of a focusing (n2 > 0) or defocusing (n2 < 0) nonlinearity with the appropriate anomalous (P2 < 0) or normal (P2 > 0) group-velocity dispersion leads to temporal pulse compression and the formation of soliton-like pulses [2]. Furthermore, this interplay causes breakup of a quasi-cw beam into a periodic train of pulses, a process referred to as modulational instability. The transverse confinement is removed for planar waveguide structures and bulk materials. Here, the limitation to a treatment in the time domain is not sufficient, especially if a defocusing nonlinearity without intrinsic self-guiding is considered. The spatiotemporal behavior of pulses propagating in a regime with real dielectric susceptibility can be described within the framework of the multidimensional nonlinear Schrodinger equation. Pulse compression in time and transverse space has been predicted for normally dispersive media with defocusing nonlinearity [1]. Spatiotemporal modulational instabilities developing a temporal pulse train and a ring pattern on the transverse intensity profile have also been considered [3]. The framework of the nonlinear Schrodinger equation does not hold for intense ultrashort pulses propagating spectrally close to a resonance. A more refined theory including the coherent dynamics of the optical polarization must therefore be adopted.

19

2. Experimental Technique Beam Splitter

Ti: Sapphire Oscillator

A Z7

Delay Stage Aperture

PM1 BBO

Objective I—1^ M = 7:l y Cryostat ^J^ Translation Stage x4

Fig. 1. Experimental setup. BBO: p-barium-borate, PMT: photomultiplier tube. To investigate the spatiotemporal evolution of subpicosecond pulses in the normally dispersive and nonlinear defocusing regime below the band edge of bulk GaAs, we performed a novel fast-scan cross-correlation experiment in which pulses are simultaneously resolved in time and transverse space. Figure 1 illustrates the experimental setup. We use 100 fs pulses at 836 nm from a Ti:Sapphire oscillator with a repetition rate of 76 MHz. The configuration involves splitting of the linearly polarized laser output into two portions: One part (33%) enters a variable delay line, while the second part (67%) passes through a pulse shaper to tailor 600 fs pulses with sech^ intensity profile. The shaped pulses are focused with a f = 25 mm microscope objective to a focal spot size of about 5 jim FWHM on ai600-|Lim-thick bulk GaAs sample that is kept at 9 K in a cold-finger cryostat. The virtual beam waist is imaged with a magnification of M = 7 : 1 onto a 15 jim precision pinhole mounted on a translation stage to provide the transverse spatial resolution. The transmitted pulses are time-resolved by cross correlation with the temporally delayed 100 fs pulses in a 300-|Lim-thick (3-barium-borate crystal cut for type-I phase matching. The intensity cross-correlation signal is detected in a photomultiplier tube. We employ a fast-scan sampling technique and average over many scans for low-noise pulse acquisition [4].

3. Results and Discussion Figure 2 shows measured spatiotemporal intensity distributions for 600 fs input pulses propagating at 836 nm through 600 jam of bulk GaAs at 9 K. We plotted the normalized cross-correlation signals as a function of time delay and pinhole displacement. Figure 2(a) represents the transmitted spatiotemporal pulse profile for an input intensity of 8 MW/cm^ (based on the Gaussian 1/e width WQ). In this linear propagation regime, the pulse duration is increased from initially 600 fs to 770 fs due to the normal material dispersion P2 > 0. Slight temporal wings occur on both sides of the main pulse. In contrast, the transverse spatial beam profile is nearly unaltered with respect to the input pulse. We measured a FWHM of 34.1 |im. Increasing the input intensity to 580 MW/cm^ [Fig. 2(b)] shows strong temporal pulse compression to 50% of the input pulse duration along with the formation of pronounced temporal wings. Due to the defocusing nonlinearity

20

n2<0, the pulse peak is shifted to earUer times by 160 fs. These temporal characteristics are associated with the interplay of the defocusing nonlinearity and normal dispersion, i.e., soliton-like behavior and modulational instability. They are accompanied by narrowing of the transverse beam width by 13% to 29.7 |Lim FWHM. Pronounced intensity modulations occur on the transverse spatial pulse profile. The spatial beam narrowing can be explained by the nonlinear defocusing Kerr lens excited in the sample and the subsequent imaging of the virtual beam waist. The beam distortions are caused by spherical and chromatic aberrations of the nonlinear Kerr lens. Thus, spatiotemporal coupling does not seem to be the origin of compression in the inspected parameter regime. (a) (b)

y-PosUion (|j.m)

Iimofpb)

y-Pu5iuuii (i-tni)

Time (ps)

Fig. 2. Measured spatiotemporal pulse profiles at (a) 8 MW/cm^ and (b) 580 MW/cm^. The temporal and spatial sections through the pulse peak are depicted on the back planes. To investigate the feasibility of spatiotemporal coupling in the semiconductor material, we performed numerical simulations based on pulse propagation in paraxial approximation and a near-resonant nonlinearity described by the lowest exciton resonance. For shorter propagation distances, higher intensity, and shorter pulse duration than in the experiment, the calculations indeed predict intrinsic spatiotemporal pulse compression connected to self-phase modulation in the adiabatic following limit [5] at the sample output facet. Spatiotemporal envelope distortions are observed as well. For longer propagation distances, the calculations show pulse splitting in time and broadening in transverse space, as actually expected for a self-defocusing material. We would like to thank the DFG (SPP 1113, SFB 296) for support.

References 1 2 3 4 5

A. T. Ryan and G. P. Agrawal, in J. Opt. Soc. Am. B, Vol. 12, 2382, 1995. P. Dumais et al., in Opt. Lett., Vol. 21, 260, 1996. L. W. Liou et al, in Phys. Rev. A, Vol. 46,4202, 1992. N. C. Nielsen et al., in Phys. Rev. B, Vol. 64, 245202, 2001. P. A. Harten et al., in Phys. Rev. Lett., Vol. 69, 852, 1992.

21

Intense CEO-stabilized few-cycle laser pulses from supercontinuum generation in filaments J. Biegert\ C. P. Hauri\ W. Kornelis\ A. Heiiirich\ F. W. Helbing\ A. Couairon^, A. Mysyrowicz^, U. Keller^ ^Swiss Federal Institute of Technology (ETH ZUrich), Physics Department, CH-8093 ZUrich, Switzerland, Phone: +41 1 633 65 30, fax: +41 1 633 10 59, email: biegert(a)phys.ethz.ch -Centre de Physique Theorique, Ecole Polytechnique, CNRS UMR 7644, F-91128 Palaiseau Cedex, France -Laboratoire d'Optique Appliquee, Ecole Polytechnique, F-91761 Palaiseau Cedex, France Abstract: A novel technique for the generation of intense few-cycle pulses is presented. Based on filamentation an intense CEO-stabilized 43-fs laser pulse yields a supercontinuum, supporting 1.8 fs pulses with a spectral range covering more than 600 THz. Chirped-mirror compression yields 5.7-fs pulses with 0.38 mJ energy in an excellent mode, with the CEO phase-lock being preserved.

Intense few-cycle laser pulses have become an important tool for many applications such as high-harmonic generation (HHG)[1,2] and single attosecond pulse generation [3]. Intense ultrashort laser pulses with only 2-3 optical cycles have been generated so far exclusively by use of hollow-fiber and subsequent chirped mirror compression technique [4]. In this paper we present a novel simple technique for the generation of intense few-cycle pulses based on self-filamentation. Self-filamentation provides a strong re-modulation of the driving laser field [5,6] with previously measured emerging pulse durations of longer than 64 fs [7,8]. In our experiment we succeeded in the generation of intense 5.7-fs pulses through self-filamentation caused by simple propagation of amplified 42-fs laser pulses at a repetition rate of 1 kHz in a noble gas atmosphere. - Laser Amplifier

Fig. 1. Experimental setup: Rl, R2: focusing mirrors (f=1000 mm) ; GC 1, GC 2 : gas cells filled with Ar; CM : chirped mirrors for GVD compensation. Fig. 1 shows our experimental setup. A 1-mJ laser pulse with a duration of 42 fs is focused into a 1.6-m-long gas cell filled with argon. The beam parameters and the gas pressure are chosen such that a 13-cm-long filament is generated inside the cell resulting in self-guiding and moderate self-phase modulation during propagation of the laser pulse. After leaving the gas cell the pulses are compressed by chirped mirrors and - afterwards - refocused into a second gas cell filled with argon. 22

250

400

550 700 Wavelength [nm]

850

1000

Fig. 2. a) Broadest spectrum obtained by filamentation with b) corresponding transform-limited pulse duration. In the second gas cell, the beam characteristics and gas pressure are tuned to facilitate filamentation with extreme spectral broadening. The generated spectrum (Fig. 2a) supports a pulse duration of 1.8 fs (Fig. 2b) with an energy of 0.42 mJ. The emerging supercontinuum is compressed by several reflections on conventional chirped mirrors and fine-tuning of the gas pressure resulting in 5.7-fs pulses being

-50

500

0 Time [fs]

600

700 800 Wavelength [nm]

900

1000

Fig. 3. SPIDER measurement of the emerging pulse: a) Temporal pulse shape and the spatial beam profile shown in the inset, b) Corresponding spectrum (solid) and spectral phase (dashed). characterized with SPIDER. Fig. 3a shows the temporal pulse shape together with the emerging excellent beam profile (inset). The associated pulse spectrum (solid) and spectral phase (dashed) are illustrated in Fig. 3b. In order to verify that this method allows generating CEO-phase-locked pulses, we monitored the interference pattern between the high frequency part of the emerging

2000 3000 Time (ms)

4000

5000

Fig. 4. Time-series of the f-2f interferometric signal indicating CEO phase locking. The oscillator is unlocked during the first 1.2 seconds, then the oscillator phase-lock is switched on. Persistence of the fringes indicates CEO phase preservation through the amplifier and the two filaments.

23

supercontinuum and the low frequency components frequency-doubled in a 250-^m thick BBO crystal [9,10]. Fig. 4 shows the temporal evolution of the interference signal over 5,000 consecutive pulses being recorded with a spectrometer equipped with a linear CCD array. With the phase stabilization of the oscillator switched off (0 to 1200 ms), no fringes are visible. After 1200 ms, the stabilization is switched on, and interference fringes are clearly resolved and stationary. To our knowledge, this is the first experimental verification of CEO phase preservation through filamentation. The obtained results of supercontinuum generation in two filaments and subsequent compression could be well reproduced by a code solving the 3Dnonlinear envelope equation describing the evolution of the field envelope[ll]. Fig. 5 shows the calculated spatial distribution of laser intensity in the second gas cell, which nicely illustrates the formation of a filament. X 10

200i

t 0

50

100

150

Fig. 5. Radial distribution of the laser intensity as function of propagation inside the second gas cell nicely confirms the formation of a filament. In conclusion, we have presented an simple method to produce CEO phaselocked, 5.7-fs, 380-^J pulses based on supercontinuum generation in a filament with the broadest spectrum supporting a pulse duration of 1.8 fs. The beauty of this setup is its simplicity, reliability, power scalability and it therefore provides an attractive alternative to the widely and exclusively used method of hollow-fibercompression to produce intense few-cycle laser pulses with an excellent intensity beam profile. This work was supported by the Swiss National Science Foundation (QP-NCCR) and by ETH Ziirich. We acknowledge the support of the EU FP6 program "Structuring the European Research Area", Marie Curie Research Training Network XTRA (Contract. No. FP6-505138). 1. A. McPherson, G. Gibson, H. Jara, U. Johann, T.S. Luk, I. Mclntyre, K. Boyer, C.K. Rhodes, J. Opt. Soc. Am. B 4, 595 (1987) 2. M. Ferray, A. L'Huillier, X.F. Li, L.A. Lompr, G. Mainfray, C. Manus, J. Phys. B: At. Mol. Opt. Phys. 21,L31 (1988) 3. R. Kienberger et. al., Nature 427, 817 (2004) 4. M. Nisoli, S. DeSilvestri, O. Svelto, R. Szipocs, K. Ferencz, C. Spielmann, S. Sartania, F. Krausz, Opt. Lett. 22, 522 (1997) 5. H.R. Lange et al., CLEO/IQEC, 3-8. May, 1998, San Francisco 6. M. Mlejnek et al., Opt. Lett. 23, 382 (1998) 7. D. Mikalauskas et al., Appl. Phys. B. 75, 899 (2002) 8. A Bernstein et al., Opt. Lett. 28, 2354 (2003) 9. H.R. Telle et al, Appl. Phys. B 69, 327 (1999) 10. M. Mehendale et al. Opt. Lett. 25 1672 (2000) 11. A Couairon et al., JOSA B 19,1117 (2002)

24

Generation of ultra-broadband high energy pulses without external amplification Alexander Fuerbach^, Alma Fernandez G.^, Takao Fuji^, Harald Mayer^, Peter Dombi^, Ferenc Krausz^ and Alexander Apolonski"^ ^ Femtolasers Produktions GmbH, Femkomgasse 10, A-1100 Vienna, Austria " Photonics Institute, Christian Doppler Laboratory, Vienna University of Technology, Gusshausstrasse 27, A-1040 Vienna, Austria " Photonics Institute, Christian Doppler Laboratory, Vienna University of Technology, Gusshausstrasse 27, A-1040 Vienna, Austria and Max-Planck-Institut fiir Quantenoptik, Hans-Kopfermann-Strasse 1, D-85748 Garching, Gennany "^ Photonics Institute, Christian Doppler Laboratory, Vienna University of Technology, Gusshausstrasse 27, A-1040 Vienna, Austria and Institute of Automation and Electrometry, Russian Academy of Sciences, 630090 Novosibirsk, Russia Abstract. We report on the generation of ultra-broadband radiation utilizing the Chirpedpulse oscillator (CPO) concept. With this technique, pulse energies in excess of 150 nJ could be obtained directly out of a Kerr-lens-mode-locked Tiisapphire oscillator. By spectral broadening of these pulses in a standard single-mode fiber, output spectra corresponding to a Fourier-limited pulse duration of 5,7 fs at an energy of 50 nJ have been achieved. Due to the corresponding high peak intensity approaching 10 MW, direct measurement of the carrierenvelope phase of the pulses is feasible.

1. Introduction The previously introduced technique of chirped-pulse oscillators (CPO) [1] allows the generation of high-energy femtosecond laser radiation without the need for an additional amplifier stage. The resulting pulses with a pulse duration as short as 26 fs have been proven to be ideally suited for micro- or even nanomachining of transparent dielectric materials [2]. However, other applications like high-harmonic generation, depend on even higher peak powers or shorter pulse durations, respectively. Recently it has been shown, that few-cycle laser pulses, with intensities slightly below the tunneling regime for helium atoms can be used, to directly measure and subsequently stabilize the carrier-envelope (CE) phase of the pulses [3]. Since the required intensities are in the order of 10^^ - 10^"^ W/cm^, chiiped pulse oscillators can theoretically be used for this purpose, provided that the pulses can be sufficiently shortened. As it is extremely difficult, to generate the few-cycle laser pulses directly out of a long-cavity oscillator, we have investigated different methods for extra cavity spectral broadening of the CPO pulses. In detail, hollow waveguides and standard telecom fibers have been considered tor this purpose.

25

2.

Experiment

The laser beam, out of the CPO was expanded with a spherical, convex dielectric mirror and was subsequently focused down into the different fiber types using an achromatic lens. By varying the distance between the optical elements, optimum matching of the free-space laser mode (TEMoo) to the lowest-order waveguide mode could be achieved. In addition, the focal spot was monitored using a CCDcamera. The prism-compressor, integrated into the CPO, was adjusted in order to obtain maximum broadening. 2.1. Hollow waveguides Spectral broadening in a hollow fiber filled with noble gas, has already been utilized to generate pulses below 5 fs in duration [4]. Due to the relatively large interaction area, laser pulses out of a kHz - amplifier system had to be used. By downscaling the bore diameter of the waveguide, we have investigated whether it is possible to adapt this technique for our purpose. As the waveguide losses are inversely proportional to the third power of the diameter, short fibers have to be used in combination with highly nonlinear gases under high pressure. We therefore have chosen Xenon as the nonlinear medium and have applied pressures as high as 40 bar. Using a fiber with a bore diameter of 30 ]xm and a length of 1 cm, a throughput of up to 40% was obtained. However, launching the full energy of around 150 nJ into the waveguide, only a weak broadening could be observed. We therefore conclude, that this energy level presents the lower limit for the applicability of this technique and that pulse energies in excess of 200 nJ are necessary to generate spectra wide enough to allow the generation of few-cycle laser pulses. 2.2. Fused silica fibers Several different types of fused silica based fibers have been investigated in order to broaden the spectrum generated by the CPO. In doing so we found out that with fibers, having a core diameter larger than approximately 5 pm, sufficient broadening is possible, however, the nonlinear process always leads to a coupling

1

i)^A

a)

0,1

1 1 V^

/ /

0,01 i

0,001

0,0001 500

1

•

1 1

1 1 / I

1 1

/ / /

/

600

1

/ 700

\

800

900

1000

1100

Wavelength [nm] Time [fs] Fig. 1. a) Spectral broadening in a standard flised silica fiber. Dotted line: Input spectrum, Full line: Output spectruni. b) Calculated Fourier-limited pulse.

26

into higher order spatial modes, making it impossible to combine the spectral components in a nice TEMoo beam. Using fibers with smaller core sizes, these problems can be circumvented. However, one has to take extra care not to damage the surface of the waveguide. The best results could be obtained using a standard single-mode fiber, intended for telecom applications (Coming HR 1060). Figure la) shows the corresponding output spectrum. When seeded with pulses having a spectral width of around 80 nm, a white light continuum spanning over almost 500 nm could be generated. Assuming a flat phase, pulses as short as 5,7 fs can potentially be obtained. To be sure, not to damage the fiber-end, a maximum energy of 100 nJ was launched into the fiber. At higher input powers, a degradation of the fiber end was observed after a certain time. Given a maximum total throughput of 50%, up to 50 nJ of ultra broadband radiation was generated. Taking the Fourier-limited pulse duration into account, a peak power of up to 8,77 MW can be achieved. As the generated white-light beam is a perfect TEMQO beam, a focus diameter of 2 um can easily be realized, resulting in a peak intensity of as high as 2,8*10^'* W/cm . Note that this value is already sufficient to enter the tunneling regime for a helium atom (single active electron approximation).

4.

Conclusions

We have demonstrated a way of generating high-intensity few-cycle laser pulses without the need for external amplification. The Chirped-pulse oscillator concept represents a scaleable technique for producing sub-30 fs laser pulses with high energies. Spectral broadening due to self-phase modulation in optical waveguides can be exploited to substanfially reduce the pulse duration with a corresponding increase in the peak intensity. For energies up to 100 nJ it has been shown, that standard fiised-silica single-mode fibers are well suited for this purpose. We think that limitations, resultingfi*omdamaging the fiber end, can be pushed to 200 nJ, if appropriate measures are taken. For even higher energies, which are accessible with the CPO technique, spectral broadening in high-pressure noble gases, enclosed in hollow dielectric waveguides is the method of choice. Acknowledgements. This research was supported in part by the Austrian Science ftind, by Femtolasers Produktions GmbH, and by the Christian Doppler Society.

References 1 A. Fernandez, T. Fuji, A. Poppe, A. Fuerbach, F. Krausz, A. Apolonski, Opt. Lett. 29,1366,2004. 2 M. Lenner, R. Graf, A. Fuerbach, F. Krausz, A. Apolonski, to be published. 3 S. Chelkowski, AD. Bandrauk, A Apolonski, Opt. Lett. 29, 1557, 2004. 4 M. Nisoli, S. De Silvestri, O. Svelto, R. Szipocs, K. Ferencz, Ch. Spielmann, S. Sartania, F. Krausz, Opt. Lett. 22, 522, 1997.

27

Sub-10 fs multi-mJ Ti:sapphire laser system with a pressure-gradient hollow fiber Yu Oishi'' ^ Akira Suda\ Fumihiko Kannari^, and Katsumi Midorikawa' ^ RIKEN, 2-1 Hirosawa, Wako-shi, Saitama 351-0198, Japan E-mail: [email protected] ^ Keio University, 3-14-1 Hiyoshi, Kohoku-ku, Yokohama 223-0061, Japan E-mail: [email protected] Abstract. The delivery of sub-10 fs multi-mJ pulses from a 1-kHz repetition-rate Ti:sapphire chirped-pulse amplification system with a pressure-gradient hollow fiber is reported. A novel technique ofhoUow fiber pulse compression allows the generation of short pulses of less than 10 fs in the multi-mJ energy level.

1.

Introduction

The generation of intense ultrashort laser pulses in the few-cycle regime is an important technology for studying ultrafast nonlinear phenomena such as attosecond pulse generation by high-harmonic generation. Such pulses are generated by means of conventional chirped-pulse amplification (CPA) and a hollow-fiber pulse-compression technique [1]. However, the compressed energy has been limited to several hundreds |LIJ. The problems associated with energy scaling using hollow fiber compression are mainly self focusing and/or ionization, both of which degrade the spatial and spectral qualities. We have proposed and demonstrated a novel technique using a pressure-gradient hollow fiber to suppress these unfavorable phenomena [2]. This straightforward technique can be applied to any Ti: sapphire CPA system without the use of complex spectral control in the amplifiers. In this article, we describe a 1-kHz repetition rate Ti:sapphire CPA system with a pressure-gradient hollow fiber to generate sub-10 fs, multi-mJ pulses.

2.

Ti:sapphire chirped pulse amplifier

A schematic of a 1-kHz repetition-rate Ti:sapphire CPA system is described as follows. The front end is a commercial Ti:sapphire oscillator (Femtolasers FemtoSource PRO), which generates 10 fs pulses at a center wavelength of 790 nm. After stretching the pulse to several hundreds ps by an OfiBier triplet stretcher [3] and passing it through a SFIO prism-pair compressor to pre-compensate the high-order dispersion in the system, a first stage multipass amplifier of threemirror 9-pass configuration [4] is used to boost the pulse energy to the mJ-level at 1 kHz repetition rate. A 5-mm-long Ti:sapphire crystal was pumped by half of the 20-mJ pulse energy from a frequency-doubled Nd: YLF laser (Quantronix 527DQ).

28

The output of the first amplifier was injected into the second stage of the 4-pass amplifier. The pump energy at the second stage was 35 mJ; with 25 mJ from another Nd:YLF laser (Quantronix 527DQE) and 10 mJ as residual energy from the first stage. The output energy from the second amplifier was 8.7 mJ. The amplified beam was expanded using a telescope to a 1/e diameter of 12 mm and then sent to a grating-pair compressor. The final output pulses of the CPA system were 23 fs long with an energy of 5.2 mJ. The output pulses from the CPA system were characterized using spectral phase interferometiy for direct electricfield reconstruction (SPIDER) [5].

3, Pulse compression using a pressure-gradient hollow fiber and chirped mirrors Figure 1 shows the experimental setup used for spectral broadening and pulse compression. A hollow fiber with an inner diameter of 500 jam and a length of 220 cm was placed on a straight-grooved metal block in a gas cell, which was separated into two sections. The exit side of the cell was filled with Ar gas and the entrance side was evacuated using a pump. The gas charge from the exit side flows down to the entrance side along the optical axis in the hollow fiber, which creates a pressure gradient. The gas pressure as a ftinction of axial distance is given by P-^(PH^-PL)ZIZO

(1)

+ PL^

where PH and PL are the gas pressure at the exit side and the entrance side respectively, zb is the fiber length, and z is the distance from the entrance. hollow fiber 0 = 500 Lim, I = 220 cm

23fs,5inJ

vacuum pump lens

lens

Ar gas| SPIDER

Chirped mirrors (-20 fsVreflection) I

Spectrometer

Fig. 1. Setup for pulse compression using a pressure-gradient hollow fiber. The amplified pulses were loosely focused by a lens (f=354 cm) into the hollow fiber. The beam diameter at the entrance of the hollow fiber was precisely adjusted to the fiber mode. At an input pulse energy of 4.9 mJ and an Ar gas pressure of 0.02 MPa (exit side), the output pulse had an energy of 3 mJ and a bandwidth of 160 nm in bandwidth (FWHM), which corresponds to a transformlimited pulse width of 7.4 fs. After being coUimated using another lens (f=150 cm), the broadened pulse was further compressed using a pair of chiiped mirrors (-

29

20 fs /reflection). We carefully adjusted the number of reflections on the mirrors based on the spectral phase from SPIDER. As a result, the pulses were adequately compressed after 16 reflections. Figures 2(a) and (b) show the measured spectral phase and the reconstructed temporal profile, respectively. The duration of the compressed pulses was 10 fs with an energy of 2.1 mJ. When the input energy was slightly reduced, the pulsewidth became 9 fs with an energy of 1.5 mJ. (a) 5

J

'S =! 3 x> ^ 2

[ ^ —/^ )Mk. /\/"^J

c3 x>

3

^

2

•/

>-.

/ 1 / I lOfs \^

1 1

co 1 c 0

00

800

Wavelength [nm]

^^-^ -^ -25

- ^ 0

25

50

Time [fs]

Fig. 2. (a) Spectral intensity and phase of the compressed pulse and (b) corresponding reconstructed temporal profile.

4.

Conclusions

We have developed a sub-10 fs multi-mJ laser system with a pressure-gradient hollow fiber. Further shortening of the pulsewidth is possible by increasing the input energy to the fiber and by optimizing the dispersion induced by the chirped mirrors. This straightforward pulse compression technique can be applied to terawatt-class CPA systems which will open a way to the study of ultrafast x-ray nonlinear optics. Acknowledgements. One of the authors (Oishi) was supported by the Junior Research Associate Program of RIKEN.

References 1 S. Sartania, Z. Cheng, M. Lenzner, G. Tempea, C. Spielmann, and F. Krausz, Opt. Lett.22,1562,1997. 2 A. Suda, M. Hatayama, K. Nagasaka, and K. Midorikawa, in Tech. Digest of CLEO2003, (OSA, Washington D.C., 2003) CThPDAl. 3 G. Cheriaux, P. Rousseau, F. Salin, J. P. Chambaret, B. Walker, and L. F. Dimauro, Opt. Lett. 21,414,1996. 4 S. Backus, J. Peatross, C. P. Huang, M. M. Murnane, and H. C. Kapteyn, Opt. Lett. 20,2000, 1995. 5 L. Gallmann, D. H. Sutter, N. Matuschek, G. Steinmeyer, U. Keller, C. laconis, and I. A Walmsley, Opt. Lett. 24, 1314, 1999.

30

Generation of 14-fs ultrashort pulse in all fiber scheme by use of highly nonlinear hybrid fiber Takashi Hori^ Norihiko Nishizawa\ and Toshio Goto^ ^ Department of Quantum Engineering, Nagoya University, Nagoya 464-8603, Japan E-mail: [email protected] Abstract. We present the all fiber pulse compression using the highly nonlinear hybrid fiber. The 100-fs pulse from the fiber laser is compressed into 14-fs pulse at the center wavelength of 1560 nm.

1.

Introduction

The generation of the ultrashort optical pulse provides a lot of applications such as the ultrafast spectroscopy, supercontinuum generation, optical sampling, and pump-probing analysis. The realization of all-fiber system in the femtosecond pulse generation is desirable in terms of stability, compactness, and practicality. So far, the generation of the 20-fs-class pulse by the higher-order soliton compression in the optical fibers have been reported [1,2]. Here, w^e present the all fiber sub20-fs pulse generation using the optimally designed highly nonlinear hybrid fiber.

2.

Experiment and results

(a)

Er-doped femtosecond fiber laser

c

SMF

i

Spectrometer

=^ 20 ^

HNL-DSF 3cm

-(b) ;:i-''-""'

Q.

BK7 (t)2mm PMF 17cm

40

E

SHG-FROG

__..^?

0

1-20 t/i

a -40

r

1

PMF

-•'^""-"" HNL-DSF1 Fiber laser

D

1400

1600

1800

2000

Wavelength (nm)

Fig. 1. (a) Experimental setup for the all fiber pulse compression and (b) dispersion curves of the fibers and BK7 glass used in the pulse compressor. Figure 1 shows the experimental setup. The light source is a passively modelocked erbium-doped fiber laser. It generates -100 fs sech^-like pulses at a repetition frequency of 48 MHz, the center wavelength being 1560 nm. The pulse compressor is constructed from the different kinds of fibers using the flision splicer. Figure 1 (b) shows the dispersion curves of these fibers. We determined the optimal length of each fiber from the results of the numerical calculation and the actual experiment. The compressed pulse is evaluated by the second-harmonic generation frequency-resolved optical gating (SHG-FROG) [3]. In the numerical 31

calculation, the generalized nonlinear Schrodinger equation is solved by using the split-step Fourier method [4], We considered the higher-order nonlinear effects, such as the self-steepening and stimulated Raman scattering and used the nonlinear response function of the fiber that both the electronic and vibrational Raman contributions were included. For the dispersion effects, we considered the terms up to the 6th order dispersion parameters of the fibers. ^950 S9OO (a) ^850 _! 800 ? 750

.(b)

. (c)

CO

^ 700

^0.2 J • I 0.0 — 1

(b 650 CO 600 -200

-100

0

100

200

-100

0

100

200

-100

0

100

200

Time (fs)

Fig. 2. Experimentally measured SHG-FROG traces at the output of the (a) fiber laser, (b) PMF, and (c) BK7 lens. Figure 2 shows the measured SHG-FROG traces at each point in the pulse compressor. The number of data points of the spectrograms used in the retrieval procedure are 256x256. The retrieved and numerically calculated temporal waveforms are shown in Fig 3. The retrieved error was less than 1%. The pulse from the fiber laser is firstly compressed by utilizing the effect of the higher-order soliton compression in the polarization maintaining fiber TPMF). The mode-field diameter and nonlinear coefficient are 5.84 jim and 4.8 W km"^ at the wavelength of 1.55 |im, respectively. As increasing the propagation length, the temporal width is compressed gradually until the optimal length. When the peak power of the pulse injected into the fiber was 3.7 kW (correspondmg to N-2 soltion), the 34-fs pulse was observed. (Fig. 3 (a)) Next the pulse is injected into the highly nonlinear dispersion shifted fiber (HNL-DSF) and its spectrum is broadened by the self-phase modulation (SPM). The ZDW of this fiber is 1545 nm and it is almost matched with the laser wavelength. The nonlinearity of this fiber is enhanced by doping the germanium into the silica core and by the small effective core area [5]. The mode-field diameter of the HNL-DSF is 3.7 |im and the nonlinear coefficient is as large as 21 W"^km"\ As increasing the fiber length of the HNL-DSF, the further spectral broadening occurs and the wideband supercontinuum is generated [6]. In the temporal domain, however, the oscillation structure appears on the waveform and the pulse break-up occurs due to the effect of the group delay dispersion (GVD) of the fiber [7]. The fiber length of the HNL-DSF is determined as the distortion of the temporal waveform can be ignored. In the final stage, the pulse is further compressed by compensating the chirp induced in the HNL-DSF. We used the BK7 collimator lens (^=2 mm) as the compressor. The positive chirp induced in the HNL-DSF is compensated by the negative chirp of this lens. According to the numerical calculation, the conventional single mode fiber (SMF) can be used in place of the lens. Finally, the

32

14-fs pulse was experimentally obtained. (Fig. 3 (b)) The output average power was 25 mW and the estimated peak power was -18 kW. Experiment

Numerical calculation

10

10

(a) 5

5 0

:---

.__^''

/\34fsFWHM 1 \

/ \

>> c

03 JO.

1

•(b)

)

-5 CD

10 i"


^ ^ / ' / l '

Z3

5 Q.

10 i

CO

c

__

,

J .L^^ 0

CD

1

x: ^

12.2 f s ; ' 0

r" -100

1

14 fs

-5

100

Tim e(fs)

-10 2C 0

0

-5 2

/ \

:.-""' 1 ../

-200

-100

-5 ^V,,.

0

5 0

—^1

1 /

-200

24.2 fs

_in

100

200

Time (fs)

Fig. 3. Retrieved and numerically calculated temporal waveforms at the output of the (a) PMF and (b) BK7 lens.

4.

Conclusions

We have demonstrated the all fiber pulse compression using the highly nonlinear hybrid fiber. The hybrid fiber was optimally designed so as to realize the effective compression. The 100-fs pulse fi-om the fiber laser was compressed into 14-fs pulse. The numerical analysis was used to simulate the pulse compression process and its results were in good agreement with the measured ones.

References Y. Matsui, M. D. Pelusi, and A. Suzuki, IEEE Photon. TechnoL Lett. 11, 1217, 1999. M. Tsuchiya, K. Igarashi, S. Saito, and M. Kishi, lEICE Trans. Electron. E85-C, 141, 2002.. R. Trebino, K. W. DeLong, D. N. Fittinghoff, J. N. Sweetser, M. A. Krumbugel, B. A. Richman, and D. J. Kane, Rev. Sci. Instrum. 68, 3277, 1997. G. P. Agrawal, Nonlinear Fiber Optics, 3rd ed.. Academic Press, San Diego, 2001. T. Okuno, M. Onishi, T. Kashiwada, S. Ishikawa, and M. Nishimura, IEEE J. Sel. Topics in Quantum Electron. 5, 1385, 1999. N. Nishizawa and T. Goto, Jpn. J. Apply. Phys. 40, L365, 2001. T. Hori, N. Nishizawa, T. Goto, and M. Yoshida, J. Opt. Soc. Am. B, in printing.

33

High peak power ultrashort pulse generation using all-fiber chirped pulse amplification system with small core multimode fiber Jun Takayanagi\ Norihiko Nishizawa^ Hiroyuki Nagai^, Makoto Yoshida^, and Toshio Goto^ ^ Department of Quantum Engineering, Nagoya University, Nagoya 464-8603, Japan E-mail: [email protected] ^ Business Planning Office Optics Annex, AISIN SEIKI Co. Ltd., Kariya 448-0003, Japan Abstract. We present an all-fiber chirped pulse amplification system based on a single mode Er-doped fiber and a multimode fiber with 25 |am core diameter. The peak power of the output pulses is up to 44 kW.

1» Introduction Ultrashort pulse lasers are very useful light sources for diverse applications. Especially passively mode-locked fiber lasers offer a number of practical advantages over bulk solid state lasers, including compact size, better stability, and freedom from misalignment. In fiber lasers, to overcome the inherent nonlinearity and increase the peak power, the chirped pulse amplification (CPA) can be a powerful technique. In the CPA system, a pair of diffraction gratings or prisms is often used as the pulse compressor [1-3]. But these mediums require the strict alignment and cause the degradation of pulse quality. On the other hand, all-fiber system can generate stable and high quality pulses without any maintenance because of its simple configuration. In this work we have demonstrated an all-fiber chirped pulse amplification system that is constructed by a WDM coupler, a single mode Erbium-doped fiber, and a small core multimode fiber. In this system, a small core multimode fiber is used as a pulse compressor. It is known that the single mode propagation is possible even in a multimode fiber [4]. This time we use a multimode fiber as a large mode field area single mode fiber to compress the pulses without the degradation by the nonlinear effect for the first fime. Each fiber is spliced in turn with low loss so that pulses can be amplified efficiently and the peak power increases up to tens of kW range by passing through only one optical fiber.

2.

Experimental Methods

The experimental setup is shown in Fig. 1. The passively mode-locked fiber laser oscillator provides 260 fs sech^-like pulses at repetifion frequency of 48 MHz. The

34

center wavelength is 1.56 jim and the average power is 8 mW. After power adjustment by the half-wave plate and the polarization beam splitter (PBS), the pulses are injected into the 5 m long WDM coupler. The power of the 1480 nm LD for EDF pumping is 560 mW. Subsequently the pulses are amplified and stretched in the normal dispersive EDF. The dispersion parameter, the mode-field diameter, and the length of the EDF are 6.18 psVkm, 8 jim, and 7.5 m, respectively. Then the amplified pulses are compressed by the small core multimode fiber. The core and the clad diameter of this fiber are 25 fim and 125 |Lim, respectively. By reducing the core diameter compared with standard multimode fibers, the mode-coupling coefficient can be made small and the pulses can maintain their fiindamental mode. We can confirm the single mode propagation of the light wave in this multimode fiber from far field pattern of the radiation. The dispersion parameter P2 measured by interferometer method is -30 ps^/km. This anomalous dispersion property enables to compress the linearly chirped pulses. Finally the wave plates and the PBS extract only linearly polarized component and the pump cut filter eliminates the residual pump light source. 25 p.m core MMF

Er-doped fiber laser oscillator

Fig. 1. All fiber chirped pulse amplification system using multimode fiber with the core of 25 |im diameter as the pulse compressor. Fig. 2 shows the amplified pulse waveform and phase in the (a) time and (b) spectral domain observed by SHG-FROG method. The pulsewidth is stretched to 1 ps without nonlinear pulse breaking and the pulses get linear chirp over the entire pulse width. The spectral bandwidth is broadened by the SPM from 10 nm (oscillator pulse) to 70 nm.

(a)

\

(b)

/ \

\

1.8 x:

d CO

>, w

c 0) c

/V\

*-/fi f

/

\ \ ^ \

^V

5 1.4 c 1.2

Time [ps]

3 CtJ

>* c
^ /1 ^

ft / llM/

1

\

^ ^

I

1.5 1.55 1.65 Wavelength [|am]

1.7

Fig. 2. Intensity (solid lines) and phase (broken lines) of the amplified chirped pulse in (a) time and (b) wavelength domain. Characteristics of the pulse are determined using SHG- FROG.

35

Fig. 3 shows the temporal and spectral waveforms of the most compressed output pulses with 70-cm long multimode fiber. The solid lines show the intensity and the broken lines show the phase. The pulsewidth of the central peak is compressed to be 80 fs. This quite small value compared with that of the oscillator pulses is attained by the spectral broadening by SPM in the EDF. Because the phases are almost flat in the both time and wavelength domain, the output pulses are compressed to be almost transform-limited one. This time, the average power of the output pulses is amplified to 200 mW. So considering the average power, the waveform, and the repetition frequency, the peak power of the output pulses is estimated to be up to 44 kW by this all-fiber CPA system. Considering the actual injected power to the optical fiber, the peak power is increased no less than 150 times compared with the initial state. 20

A

;(b) CD

10 V Q_

v/3-^

^

^ --

(

TO

/(

CD JZ

J

J 1

1.45 Time [ps]

15

DL

\ 5

1.5 1,55 1.6 1.65 1.7 Wavelength [|im]

Fig. 3. (a) Temporal and (b) spectral waveform and phase of the compressed pulse. Solid lines and broken lines are intensity and phase, respectively.

4.

Conclusions

In summary, we have demonstrated the all fiber chirped pulse amplification system using the multimode fiber as the pulse compressor for the first time. The generated pulses have the peak power of 44 kW and the pulsewidth of 80 fs. The low nonlinearity of the multimode fiber suppresses the pedestal component and the spectral distortion. To the best of our knowledge, the peak power is the highest one and the pulsewidth is the shortest one from the all fiber chirped pulse amplification system using only standard fibers.

References A. Galvanauskas, M. E. Fermann, and D. Harter, Opt. Lett., Vol. 19, 1201, 1994. F. O. Ilday, H. Lim, J. R. Buckley, and F. W. Wise, Opt. Lett., Vol. 28, 1362, 2003. F. Tauser and A. Leitenstorfer, Opt. Exp., Vol. 11, 594, 2003. Martin E. Fermann, Opt. Lett., Vol. 23, 52, 1998.

36

Carrier-envelope phasefluctuationsof amplifled laser pulses transmitted through neon^illed hollow fiber for pulse compression A. Ishizawa and H. Nakano NTT Basic Research Laboratories, NTT Corporation, 3-1 Morinosato Wakamiya, Atsugi, Kanagawa 243-0198, Japan E-mail: ishizawa®will.brl.ntt.co.jp Abstract. We measured the carrier-envelope phase drifts of spectrally broadened intense pulses that passed through a hollow fiber. An additional carrier-enveloped phase fluctuation of 0.47 rad was induced in white light generated in the hollow fiber due to the non-uniformity of self-^hase modulation in the fiber.

1.

Introduction

The oscillation of the laser field has a very large influence on the interaction of light with matter. In the interaction, the timing of oscillation cycles within a light pulse plays a role when the duration of the pulse becomes comparable to the light oscillation period. The processes driven by intense few-cycle laser pulses have been predicted to depend on the timing of the oscillation of the laser field. The timing can be quantified by the carrier-envelope phase (CEP) [1]. At present, white light generation through self-^hase modulation (SPM) in a gas-filled hollow fiber is indispensable. In our amplified laser system, few-cycle laser pulses with less than 7-fs pulse width have already been obtained by using that technology. Recently, it has been reported that the atomic processes driven by intense fewcycle laser pulses with a different CEP are very sensitive [2, 3]. To control the CEP of intense laser pulses, it is necessary to measure it directly after the seed laser pulse has been amplified. The CEP has been measured directly at the output of an amplifier with a broadband f-to-2f interferometer [3]. Actually, amplified laser pulses after a hollow fiber were used in that experiment. We are interested in whether the difference of the CEP drifts before and after a hollow fiber is negligible or not. In this paper, we report the CEP drifts of spectrally broadened intense pulses that passed through a hollow fiber.

2.

Experimental Setup

We used a multipass Ti: sapphire chirped^ulse amplifier. The CEP slip of the pulses delivered by the Ti: Sapphire oscillator was controlled. The amplifier delivers l^nJ energy in a 25-fs pulse. We measured the difference of the CEP drift before and after a hollow fiber using f-to-2f interferograms. Amplified laser pulses of about 8 jLiJ and 1 mJ were used in the measurements of the CEP drift. In the

37

measurement of the CEP drift before the hollow fiber, the laser pulse irradiated a 2-mm sapphire plate to generate white light whose spectrum covered more than one-octave bandwidth. We measured the interferograms using the second harmonics and blue components of the white light with a fiber-coupled spectrometer. On the other hand, in the measurements of the CEP drift after the hollow fiber, the amplified laser pulse was focused into a 1-m long hollow fiber placed in a chamber filled with 2-atm neon gas. Propagation in the hollow fiber broadens the spectrum more than one octave by self-^hase modulation (SPM). Thereafter, the measurement was the same as that before the hollow fiber.

3. Results and Discussion Figure 1(a) shows the CEP drift of the intense laser pulse delivered from the multi-pass amplifier. Even after the amplification, the CEP fluctuation was within 0.49 rad. Figure 1(b) shows the measured CEP drifts after the l^m hollow fiber.

(b)

(a) 7r/2

^

isW^!^^A

-vM


«

7r/2i

0

uj-7t/2\

^psii^

I

o -K

3 4 5 Time (s)

3 4 5 Time (s)

Fig. 1 CEP drifts of amplified laser pulses (a) before the hollow fiber and (b) after the hollow fiber. The standard deviation of the CEP drifts of the intense laser pulses after the hollow fiber was 0.96^:ad. Moreover, the center of the CEP was found to drift slowly on the time scale of a few seconds. Without controlling the CEP slip of the oscillator, the f^;o-2f interference fringe became blurred after a few hundred laser shots. In contrast, with the CEP slip control, the interference fringe did not become blurred after a few hundred laser shots. These results indicate that the CEP was stable after the hollow fiber. In this experiment, the difference of the standard deviation before and after the hollow fiber was 0.47 rad. There are three possible causes of the CEP fluctuations. These possible causes are related to each other and the complexity to the CEP fluctuations. The first is non-uniform gas pressure in the hollow fiber. It has been reported that if the laser intensity is high, the third order nonlinearity plays an important role and causes turbulence of the propagation dynamics [4]. If the gas pressure in the hollow fiber is not uniform, it will influence the output energy after the fiber. The CEP fluctuations are related to the difference of the output energy. The second is the non-uniform transverse mode of the laser pulses in the hollow fiber. Each element of the transverse mode of the laser pulse that pass through a hollow fiber has each CEP. The CEP that we can measure is the average of each

38

CEP for each element. So, if the transverse mode of the laser pulse changes in each laser shot, the CEP fluctuations become large. The single mode is best for minimizing the CEP fluctuations. The third is the gas pressure dependence of the CEP fluctuations in the hollow fiber. The CEP fluctuations can arise after propagation because of a differential phase shift between the carrier and envelope. A power fluctuation of AI will induce a change of the carrier-envelope phase of A(^ =

^AZ

(1)

c oco

where co is the frequency of the light, / the length of the hollow fiber, and n2 the nonlinear index. The dn2/dco part of Eq. (1) is proportional to the gas pressure. We measured the CEP fluctuations for different gas pressures in the hollow fiber and found that lower gas pressure in the hollow fiber makes Acj) small. When we want to obtain a short laser pulse, a wide spectral width is necessary. A wide spectral width can be obtained by two methods. One is to increase the laser energy. But the coupling efficiency is low when the transverse mode is single. Even if a wide spectral width is obtained by multi-mode, the CEP fluctuations become large. The other is to increase the gas pressure. However, the higher the gas pressure is, the larger the CEP fluctuations become. There is a trade-off between the CEP fluctuations and the spectral broadening.

4.

Conclusions

We have investigated CEP fluctuations of white light, which were obtained by transmitting an amplified CEP-controlled pulse through a neon-filled hollow fiber. We found that the CEP fluctuation after the fiber is larger by 0.47^ad after the laser pulses passed through the fiber. The main reason for this additional phase drift is non-uniformity of the SPM process in a fiber. By choosing proper parameters, such as the laser intensity and gas pressure, we can make the CEP drifts change slowly. Our experiment indicates that we should pay attention to CEP fluctuations not only before a hollow fiber but also after it to obtain CEP-stabilized few-cycle pulses.

References 1. 2. 3. 4.

T. Brabec and F. Krausz, Rev. Mod. Phys. 72, 545, 2001. G. G. Paulus, F. Grasbon, H. Walther, P. Villoresi, M. Nisoli, S. Stagira, E. Priori, and S. De Silvestri, Nature (London) 414, 182, 2001. A. Baltuska, Th. Udem, M. Uiberacker, M. Hentschel, E. Goulielmakis, Ch. Gohle, R. Holzwarth, V. S. Yakovlev, A. Scrinzi, T. W. Hansch, F. Krausz, Nature (London) 421, 611, 2003. M. Nurhuda, A. Suda, K. Midorikawa, M. Hatayama, and K Nagasaka,

39

Temporal Self-Compression of Intense Femtosecond Pulses Propagating in Argon-Filled Hollow Waveguides Nick Wagner\ Emily A. Gibson\ Sterling Backus\ Margaret M. Murnane\ Henry C. Kapteyn\ and Ivan P. Christov^ ^ Department of Physics and JILA, University of Colorado and NIST, Boulder, CO 80309-0440 E-mail: [email protected] ^ Department of Physics, Sofia University, Sofia, Bulgaria Abstract. We demonstrate temporal compression of intense femtosecond 800-nm pulses in a hollow-core fiber filled with low-pressure argon gas, without any subsequent dispersion compensation. We achieve a final compressed pulse width of- 13 fs

We show that intense laser pulses propagating at high intensity inside an argon-filled hollow waveguide can compress in time as they travel down the waveguide[l]. In this experiment, the peak intensity in the fiber is --10^^ W/cm^ so that the argon gas is fully ionized. Theoretical models show that the temporal compression is due to a spatial-temporal effect as the pulse interacts with the generated plasma and the waveguide. Unlike the method of Nisoli et al[2], where a gas filled fiber is used to broaden the spectrum of a light pulse using self-phase modulation, we do not need to use any subsequent dispersive optics (i.e. prisms, gratings, or chirped mirrors) to compress the pulse. In our experiment, we focus light from a Ti:sapphire amplifier producing -- 30 fs pulses with 2.5 mJ pulse energies at 1 kHz into a 2.5 cm long, 150 micron inner-diameter hollow glass waveguide filled with low pressure argon. The pulse is characterized before and after the fiber using second-harmonic Frequency-Resolved-Optical-Gating (SHG-FROG). The focusing region before and after the fiber is held under vacuum to prevent additional nonlinear effects. The intensity in the fiber is approximately 10^^ W/cm^, well above the saturation intensity for argon. Ionization of the argon gas as the pulse passes results in an index of refraction that decreases with time, causing a blue shift of the overall laser spectrum. Figure 1 shows the deconvolved FROG measurements of temporal pulse shape after the fiber for different argon pressures. With only 1 Torr of argon in the fiber, the initial pulse width of 32 fs has been drastically reduced. As the pressure is increased, the pulse width decreases to 15.5 fs at ~ 4 Torr. The power exiting the fiber is reduced by ~ 5 % at 4 Torr compared with the case of no gas in the waveguide, and scales linearly with pressure. Figure 2 shows the exit pulse width in the case of a slightly shorter 29 fs input pulse. In this case, the final pulse width is 13 fs. This pulse width is relatively constant up to 9 Torr, at which point the laser mode begins to break up[3]. Both the input pulse and the compressed final pulse have a near transform limited

40

time-bandwidth product, as shown in figure 3. U T o r r 22.3 fs 2.0Torr 19.2 fs 2.8 Torr 17.7 fs 3.8 Torr 15.5 fs No Gas - 32 fs

Time (fs)

Fig. 1. Measured temporal profile of pulse after fiber for different argon pressures with an input pulse width of 32 fs. The pulse width FWHM for each pressure is shown All FROG measurements were deconvolved using commercial software from Femtosoft Technologies (FROG Version 3.03) to an error of < 1%. The spectral marginals for each measurement agree with the experimentally measured spectra to within 5%. After exiting the fiber, the pulse passes through a 250 micron sapphire window and a 1 mm beam splitter in the FROG set-up. The total dispersion of the material is equivalent to a group velocity dispersion of 61 fs^ and a third-order dispersion of 40 fs^ at a center wavelength of 760 nm, not enough to significantly affect the pulse duration measurements (i.e. <3 fs). The pulse shape after the fiber without gas is the same as the input to within our measurement resolution. 3.9 Torr 5 Torr 6 Torr 7 Torr 8 Torr before fiber (29 fs)

1.0-

0.8-

• 0.6-

"13 3 f0.4-

0.2-

0.0- L,

,

r—^

,

,

^rx. •1

20

,

1

40

1

1

60

1

r-

80

Time (fs)

Fig. 2. Measured temporal profile of pulse after fiber for an input pulse width of 29 fs.

41

In conclusion, we demonstrate a new mechanism for compression of intense femtosecond pulses. This technique has the advantage of minimal distortion of the laser mode, low power loss, and no need for additional dispersion compensation. Compressing pulses directly in the fiber can also be an advantage for high-order harmonic generation, as using a fiber for the harmonic generation immediately after the compression fiber would eliminate the need to refocus the compressed light. — Input Pulse — Compressed Pulse

-60

-40

Time (fs)

Fig. 3. Temporal profile and phase of the input pulse and the final compressed pulse from Fig. 1. Acknowledgements. This work was supported by the Chemical Sciences, Geosciences and Biosciences Division of the Office of Basic Energy Sciences, U.S. Department of Energy, by the National Science Foundation, and was supported in part by the Engineering Research Centers Program of the National Science Foundation under NSF Award Number EEC-0310717.

References 1 2 3

42

N. L. Wagner, E. A. Gibson, T. Popmintchev, I. P. Christov, M. M. Murnane, and H. C. Kapteyn, "Self-compression of ultrashort pulses through ionization-induced spatiotemporal reshaping," Physical Review Letters, To be published, 2004. M. Nisoli et al., "Toward a terawatt-scale sub-10-fs laser technology," IEEE Journal of Selected Topics in Quantum Electronics 4, 414 (1998). S. C. Rae, "Ionization-induced defocusing of intense laser-pulses in high-pressure gases," Optics Communications 97, 25 (1993).

Stimulated Brillouin scattering in ultrahighspeed femtosecond soliton pulse compression with a dispersion-decreasing fiber Toshihiko Hirooka, Shinpei Ono, Ken-ichi Hagiuda, and Masataka Nakazawa Research Institute of Electrical Communication, Tohoku University 2-1-1 Katahira, Aoba-ku, Sendai, 980-8577, Japan E-mail: [email protected] Abstract. We report that stimulated Brillouin scattering in a dispersion-decreasing fiber plays a very important role in ultrahigh-speed femtosecond soliton compression that exceeds 40 GHz. A stable 40-GHz, 100-fs pulse train was successfully generated by suppressing SBS.

1.

Introduction

Adiabatic soliton compression with a dispersion-decreasing fiber (DDF) is useful for generating a femtosecond pulse train in the GHz region. In recent experiments, 10-GHz, 54-fs pulses and 50-GHz, 280-fs pulses have been successfully generated with a polarization-maintaining, dispersion-flattened DDF (PM-DF-DDF) [1,2]. As we increase the repetition rate, however, the compression becomes unstable because the optical power of each longitudinal mode easily exceeds the stimulated Brillouin scattering (SBS) threshold [3]. In this paper, we detail the SBS effect imposed on soliton compression with DDF for 40-GHz pulses. We also report the generation of 40-GHz, 100-fs pulses by compressing the output of a mode-locked fiber laser (MLFL) with a PM-DF-DDF, in which SBS is sufficiently suppressed by adding frequency modulation of 30 MHz to the source.

2.

SBS threshold in DDF soliton compression

The SBS threshold in DDF soliton compression is determined by the optical power of each longitudinal mode of a fundamental soliton. The average power of a fundamental soliton is given by X' ^A^-l P:;' = 0.776-^^

\D

BA,„

(1)

where D is the group-velocity dispersion (GVD), X is the wavelength, ^2 is the nonlinear refractive index, Ax is the FWHM, B is the repetition rate, and^eff is the effective area. By roughly assuming that the power is equally divided into M = Av/B longitudinal modes within the FWHM of the optical spectrum, Av, the power per mode, PQ, is proportional to |D(z)|5^. This indicates that the SBS has a

43

significant deteriorating effect as we increase the repetition rate. We also note that decrease in the GVD causes the power attenuation. Usually, PQ is attenuated by the fiber loss. However, in adiabatic soliton compression, PQ decreases more rapidly during propagation through DDF, as the spectrum is broadened and the energy is transferred to outer modes. Therefore the effective length in DDF can be written as ^eff = — ^ f' Poiz)dz = — ^ t ^(^)^^

'

(2)

The SBS threshold power in DDF compression is determined by redefining Lgff in Smith's formula [4]: gBPo(0)Leff /Aejf = 21, where gs is the Brillouin gain. In order to suppress SBS and stabilize the pulse compression, frequency modulation of the output signal from a pulse source must be employed so that gs is reduced [5].

3. Experimental results To examine the influence of SBS on femtosecond soliton pulse compression with DDF, we measured the optical spectrum of the backscattered light from a DDF. The experimental setup is shown in Fig. 1. A 40-GHz, 1.7-ps pulse train generated from MLFL, whose linewidth was ~1 kHz, was first frequency-modulated with an LN phase modulator to avoid SBS. The pulses were then amplified to the soliton power level with a high-power erbium-doped fiber amplifier (HP-EDFA). The amplified pulses were launched into a 950-m long PM-DM-DDF, whose GVD gradually decreased from 11 to 0.2 ps/nm/km and A^ff was 28 ^im^. The backscattered power was monitored through an optical circulator at the DDF input. Frequency Modulator

Optical Circulator

LN J

40 GHz, 1.7 ps Mode-locked Fiber Laser

* ^ ^

^ * ^

PM-DF-DDF 950 m

Auto-correlator

'imH

Pnlflri7ati Polarization Controller

Optical Spectrum Analyzer

Optical Spectrum Analyzer

Fig. 1. Experimental setup (Abbreviations are defined in the text). Figure 2 (a) shows the optical spectrum of the backscattered light without frequency modulation when N = {PJP^yj^^^y^ = 0.94. Five distinct Brillouin modes were clearly seen around the central wavelength and the power reached 7 dBm. Other longitudinal modes observed at a level lower than -15 dBm were the direct output of the source that leaked through the circulator. From the expanded view in the inset, the backscattered modes are shifted 0.09 nm (11 GHz) from the source toward a longer wavelength. This indicates that these reflected modes are caused by the SBS in the DDF. SBS occurs for modes with a high power level that exceeds the SBS threshold, namely around the center of the spectrum. In contrast, by frequency-modulating the source by 30 MHz, the SBS was reduced to a level as low as -32 dBm as shown in Fig. 2(b). Fig. 2(c) shows the transmitted and backscattered power versus input power. SBS was successfully suppressed by increasing the linewidth, Av^. Fig. 2(d) shows the autocorrelation waveforms at the

44

input and output of the DDF when Av^ = 30 MHz. The input power was Pin = 274 mW (N = 0.85). By eliminating SBS, we obtained a stable, high-quality femtosecond pulse with pedestals less than - 2 0 dB from the peak. The output durations were 104 fs assuming a sech shape and the spectral FWHM was 4.0 THz (32 nm), corresponding to TBP of 0.42. -10

20 r

H

10

-20 \

0 -10 -20 -30

1558

1559 1560 1561 Wavelength, nm

1562

1558

1559 1560 1561 Wavelength, nm

1562

(b)

(a) W/o Freq. Mod. Av_=:10MH2 20 MHz 30 MHz 50 MHz

1 0.8 0.6

0.5 ps/div

104fs / H 0.4 0.2 (•

.^-^

0 100

200

Launched power, mW

(c)

0

300 400

Time, ps

(d)

Fig. 2. (a), (b) Optical spectrum of the backscattered light without and with frequency modulation (Av^ = 30 MHz), respectively, (c) Transmitted and backscattered power, (d) Input (dotted) and output (solid) autocorrelation waveforms.

4.

Conclusions

We have described the effect of SBS on femtosecond soliton pulse compression with DDF at a high repetition rate. With the SBS suppression by adding 30-MHz frequency modulation to a mode-locked fiber laser, we successfully obtained a 40GHz, 100-fs pulse train.

References 1 2 3 4 5

K. R. Tamura and M. Nakazawa, Opt. Lett. 26, 762, 2001. K. R. Tamura and K. Sato, Opt. Lett. 27, 1268, 2002. E. P. Ippen and R. H. Stolen, Appl. Phys. Lett. 21, 539, 1972. R. G. Smith, Appl. Opt. 11, 2489, 1972. M. Denariez and G. Bret, Phys. Rev. 171, 160, 1968.

45

Control of the spectral broadening of tens-millijoules laser pulses in an argon-filled hollow fiber using a conjugate pressure gradient Muhammad Nurhuda\ Akira Suda^, and Katsumi Midorikawa^ ^ Physics Department, Brawijaya, Brawijaya University, Malang 65144, Indonesia E-mail: [email protected] ^RIKEN,2-1 Hirosawa, Wako-shi, Saitama 351-0198, Japan E-mail: [email protected] Abstract. A proposal for spectral broadening of tens milli Joule femtosecond laser pulses in an argon-filled hollow fiber is presented. The simulation results show that the method could be useful under conditions where ionization plays a greater role and that it can be used to control plasma-induced spectral broadening.

1. Introduction It is well known that the use of a hollow fiber filled with noble gas is an effective way for extending the interaction length between nonlinear optical material with high energy laser pulses [1]. Licreasing the pulse energy above the critical power raises the problem of self focusing, which is then followed by defocusing due to plasma generation and the saturation of high-order nonlinear susceptibility [2]. We have recently been able to formulate a general procedure to avoid/postpone self focusing using a so-called "pressure gradient method" [3]. This method could enable higher energy transmittance and better spatial phase of the transmitted laser field. However, if the laser power is several times higher than the critical power for self focusing, the interaction of the high intensity laser field with a nonlinear medium generates a plasma, which leads to the plasma blue-shifting of the transmitted laser field. The resulting pulses are therefore difficult to compress due to a lack of longer wavelength components. In this paper, we present a proposal for plasma-induced spectral broadening by controlling the density of the generated plasma. The proposed technique requires two pieces of hollow fiber that are connected to each other via a window through which the gas is allowed to flow. The free ends of each segment are maintained under vacuum. Accordingly, if the gas pressure varies such that it increases in the first segment, then it decreases in other segment. It is expected that the early self focusing can be avoided by using this strategy, while the occurence of selfphase modulation is maintained. Moreover, broadening towards shorter wavelength components is expected to occur due to plasma blueshifting. Next, as the pulse propagates in the hollow-fiber segment with decreasing gas pressure, we expect that i) the decreasing plasma density dN^izj/dz =fkp(z)/ck < 0, where p is the plasma density and p(z) is the gas pressure, could stimulate spectral

46

broadening of the longer wavelength components and ii) additional nonlinear processes, e.g. filamentation, self focusing and plasma defocusing can be prevented at the exit. The basic model for the pulse propagation is the extended nonlinear Schrodinger equation.

f

1 a2

(r,zj) =

r

dt This can be coupled with the plasma equation EQC

!=<-«

e'p

Ax(/)-

Kr.zj)

(1)

m8oCOo

P)r(/)

(2)

where / is the intensity, No is the initial density of the neutral atom and Tfl) is the ionization rate, calculated using Perelemov, Popov and Terent'ev (PPT) model [4]. The coupled equations (1) and (2) can be solved subject to the boundary condition using the method described in [3].

2. Results and Discussion

600

- .-\

1

1

1

,

1

1

1

z = 50 cm

-

700 800 900 Wavelaigth (nm)

1

A

. ^ Mv.r^, 1

700 800 900 Wavelength (nm)

1

1

-J

^

—J 600

'

900

-

•

•

600

'

800

z = 60 cm

-

'

700

A •

^. 1 1 ^ 700 800 900 Wavelength (nm) ^

600

Fig. 1. Evolution of the power spectrum inside the hollow fiber as a function of propagation distance. In Fig. 1, the power spectra evaluated at different propagation distances are displayed. As can be seen, the laser field clearly undergoes blue shifting during propagation along the first segment of the hollow fiber (from z =10 to 30 cm). This is due to the increasing of the plasma density. As the gas pressure decreases, the gradient of the generated plasma becomes negative and this can in

47

fact stimulate the growth of the longer wavelength components, as can be seen in Fig. 1 (for propagation from 30 to 60 cm) The energy transmitted at the end of the hollow fiber was 10.2 mJ. 4.4 mJ (43 %) of this was found in the fundamental mode, 22 % in the first excited mode, 14 % in the second excited mode, while the remainder was distributed among the higher-order modes. The losses due to leakage and ionization were 4.6 mJ and 0.2 mJ, respectively. For the purposes of phase compensation, however, only the low order modes are useful. The high order modes have to be separated from the low order ones by propagating the wave in free space (vacuum) to a distance of 80 cm. The power spectrum obtained by integrating the intensity in a circular area of 1 mm radius is plotted in Fig. 2 (left), and the corresponding spectral profile is shown in Fig. 2 (right). Finally, by performing the Fourier transform of the beam within the circular area of radius 1 mm, we obtained a single pulse with a duration of 5 fs and an energy of 5 mJ.

700 800 Wavelength (nm)

700 800 Wavelength (nm)

Fig. 2. Power spectrum (left) and spectral phase (right) in a circular area of 1 mm radius.

References 1. 2. 3. 4.

48

M. Nisoli, S. De Silvestri, and O. Svelto, Appl. Phys. Lett. 68,2793 (1966). G. Fibich and A. Gaeta, Opt. Lett. 25, 335 (2000). M. Nurhuda, A. Suda, M. Hatayama, K. Nagasaka, and K. Midorikawa, J. Opt. Soc.Am.B20,2002 (2003). A M. Perelemov, V. S. Popov, and M. V. Terent'ev, Sov. Phys. JETP 23, 924 (1966).

Microstructured fiber feedback pulse compression to few optical cycles Muneyuki Adachi^'^, Keisaku Yamane^, Ryuji Morita^ and Mikio Yamashita^ ^ Department of Applied Physics, Hokkaido University, Kita-13, Nishi-8, Kita-ku, Sapporo, 060-8628 Japan E-mail: [email protected] - NIDEK Co.,Ltd., 34-14 Maehama, Hiroishi, Gamagori, 443-0038 Japan Abstract. A feedback system that combined modified-SPIDER and 4-/ active-chirp compensation enabled us to compress photonic-crystal-fiber and tapered-fiber outputs to 6.6 and 8.4 fs, respectively. Compressed-pulse results well agree with fringe-resolved autocorrelation measurement ones.

Since the first demonstration of microstructured fibers (photonic crystal fiber (PCF) and tapered fiber (TF)) having unusual dispersion profiles [1,2], a variety of applications including pulse compression have been pointed out. It is essential for few-cycle pulse compression to overcome problems of the bandwidth limitation of conventional passive chirp compensators, insufficient compensation for the complicated spectral phase of their fiber outputs and the inaccurate pulse characterization, which caused the imperfect compression (for example, see [3]). In this paper, we demonstrate that the direct feedback technique of the measured spectral phase without using the Taylor expansion method enables us to solve those problems successfully. Photonic Crystal Fiber The employed experimental setup consists of a 12-fs Ti:sapphire laser, a 3-mm long PCF (2.6 ^im core diameter with 900 nm zero dispersion wavelength (ZDW)), and a computer-controlled feedback system that combines a 4-/ chirp compensator with a spatial light modulator (SLM) capable of an over-one-octave bandwidth and a modified spectral-phase interferometry for direct electric-field reconstruction (M-SPIDER) apparatus with a high sensitivity, which we have developed recently [4]. The PCF output spectrum with the 60 mW average power for the 450 mW input power was broadened from 600 to 945 nm. The intensity of the SPIDER interferogram utilizing directly the output from the oscillator as intensified reference chirped pulses was strong enough to characterize the spectral phase of the PCF output pulse over the whole spectral region. The reconstructed spectral phase before feedback is shown by a dashed-dotted line in Fig. 1(a). It indicates the complicated behavior in the 710 to 930 nm wavelength region and the variation over 30 rad. The temporal intensity profile of the PCF output broadens asymmetrically over 150 fs (Fig.l(c)). Dotted and solid lines in Fig.l(a) show the spectral phases after first and second feedbacks, respectively. The compensated spectral phase after second feedback was converged within 1.3 rad throughout the

49

PCF output

n r^Wj

(0

/

y'

/\''<'>

^^

WW700

^^-Z!i t?

800 wavelength (nm)

,(b)

900

Fig. 1. Results of PCF output pulse measurements, (a) Dashed-dotted (0), dotted (1) and solid lines (2) are reconstructed spectral phases before and after first and second feedbacks, respectively. A thin dashed line is SLM-output spectrum after second feedback, (b) Wrapped spectral phase applied by SLM at the second feedback time, (c)(e) Reconstructed temporal intensity profiles (solid lines) and phases (dotted lines) for (c) PCF output, after (d) first and (e) second feedbacks. A dashed-dotted line in (e) is the corresponding 6.3-fs transform limited pulse, (f) FRAC traces of direct measurement (solid line) and retrieved from M-SPIDER data (dotted line) pulse spectral region. The wrapped spectral phase applied by SLM at the second feedback is shown in Fig. 1(b). Corresponding reconstructed temporal intensity and phase profiles are shown in Fig.l(d)-(e). After first and second feedback compensations its pulse width was reduced to 7.1 fs (Fig. 1(d)) and 6.6 fs (Fig. 1(e)), respectively. This compressed pulse almost corresponds to the 6.3 fs transform-limited (TL) one. To the best of our knowledge, this is the first complete pulse compression of the PCF output in the two-cycle region. The compressed pulse after second feedback was also measured independently by the fringeresolved autocorrelation (FRAC) method (a solid line in Fig.l(f)). The FRAC trace retrieved from the M-SPIDER result is shown by a dotted line in Fig. 1(f). The agreement between their FRAC traces is excellent. This result suggests that the spectral-phase feedback technique without using any Taylor expansion method is significantly reliable and powerful even for complicated nonlinear-chirp compensation. Tapered Fiber The experimental system similar to the PCF's one was used. By heating and stretching a 10-mm silica fiber with a 125-|Lim outer diameter and a 9 ^im core diameter, a 84-mm totally long TF was produced with a 30-mm long waist of 2 ^im uniform diameter [5]. The TF output spectrum was broadened from 610 to 945 nm and had the 28-mW averaged power for the 185-mW input power. The reconstructed spectral phase had the large dispersion (GDD=+700 fs^ and TOD=+4500 fs at 800 nm), the complicated behavior in the 730-920 nm wavelength region and the variation over 200 rad (Fig.2(a)). Four-time feedback compensations were performed. The wrapped spectral phase applied by SLM at

50

700

M-SPIDER 1 FRAC . fl

I

0)

c

800 wavelength (nm)

900

'(h)

lift. 0 *w\A/VVVvU\^ \\WN.Vw^

AfcA ^ iM M l iUIAA A. I A A * i."

-20

0 time (fs)

Fig. 2. Results of TF output pulse measurements, (a) Dashed (0), dotted (1), Dasheddotted (2), 2-dotted-dashed (3) and solid lines (4) are reconstructed spectral phases before and after first, second, third and fourth feedbacks, respectively. A thin 2-dasheddotted line is SLM-output spectrum after fourth feedback, (b) Wrapped spectral phase applied by SLM at the fourth feedback time, (c)-(g) Reconstructed temporal intensity profiles (solid lines) and phases (dotted lines) for (c) TF output, (d) first, (e) second, (f) third and (g) fourth feedbacks. A dashed-dotted line in (g) is 6.4-fs transform limited pulse, (h) FRAC traces of direct measurement (solid line) and retrieved from M-SPIDER data (dotted line) the fourth feedback is shown in Fig.2(b). Corresponding reconstructed temporal intensity and phase profiles with the TF output are shown in Fig.2(c)-(g). The shortest pulse duration obtained after the fourth feedback compensation was 8.4 fs. To the best of our knowledge, this is the shortest pulse compression using TF. However, the generated pulse duration is somewhat longer than the TL pulse duration (6.4 fs). This may be due to the large GDD and higher order dispersion of the TF output (compare Fig.2(a) and (b) with Fig. 1(a) and (b)), which makes complete chirp compensation by the present 4-/ system difficult. The compressed pulse after fourth feedback was also measured independently by the FRAC method (a solid line in Fig.2(h)). The FRAC trace retrieved from the M-SPIDER result is shown by a dotted line in Fig.2(h). Comparison between them shows good agreement.

References 1 J. K. Ranka, R. S. Windeler and A. J. Stentz, Opt. Lett. 25, 25, 2000. 2 T. A. Birks, W. J. Wadsworth and P. St. J. Russell, Opt. Lett. 25, 1415, 2000. 3 S. Lako, J. Seres, P. Apai, J. Balazs, R. S. Windeler and R. Szipocs, Appl. Phys. B 76, 267, 2003. 4 K. Yamane, Z. Zhang, K. Oka, R. Morita, M. Yamashita, A. Suguro, Opt. Lett. 28, 2258, 2003. 5 M. Adachi, M. Hirasawa, A. Suguro, N. Karasawa, S. Kobayashi, R. Morita, M. Yamashita, Jpn. J. Appl. Phys. 42, L24, 2003.

51

Spectral-temporal soliton dynamics analysis near second zero-dispersion point in photonic crystal fibers Anatoly Efimov\ A. J. Taylor^ F. G. omenetto^ N. Joly^ D. V. Skryabin^ J. C. Knight^ W. J. Wadswo^th^ and P. St. J. RusselP ^ Materials Science and Technology Division, MS K764, Los Alamos National Laboratory, Los Alamos, NM 87545, USA. E-mail: [email protected] ^ Physics Division, MS H803, Los Alamos National Laboratory, Los Alamos, NM 87545. ^ Department of Physics, University of Bath, Bath, BA2 7AY, UK. Abstract. The propagation dynamics of an ultrashort optical pulse near the second zerodispersion point of a small-core high-delta photonic crystal fiber is investigated using cross-correlation frequency-resolved optical gating. Negative dispersion slope strongly influences the observed behavior.

1.

Introduction

Microstructured or photonic crystal fibers (PCF) can now be produced with engineered dispersion profiles by varying the structure and scale of the core and the air-filled cladding. In turn, dispersion properties strongly influence the nonlinear dynamics of the propagating ultrashort optical pulses [1]. Specifically, small-core high-delta cobweb structures can be engineered to posses a second zero-dispersion (ZD) point in the telecommunication wavelength region near X-l.S jLun, Fig. 1. Also, the small core size of such PCFs on the order of 1 |um provides for very high peak intensity of the propagating mode resulting in rich nonlinear dynamics. Uniquely, the negative slope of the dispersion curve at the

-w^ww,

Fig. 1. SEM of one of the small-core high-delta fibers used in the experiments (left). The size of the central guiding core determines the wavelength of the second ZD point. Experimentally measured dispersion curves for three core sizes (right) show the difference between A) 1.25 fim, B) 1.22 jam, and C) 1.20 \xm core diameter fibers.

52

second ZD point creates a regime completely unlike the one around ihQ first ZD in PCFs and regular fibers studied previously [2,3]-

2.

Experimental Methods

Experimentally, the most complete picture of an ultrashort pulse dynamics can be obtained using cross-correlationfirequency-resolvedoptical gating (X-FROG) [4]. Our experimental setup consists of a Millenia-Tsunami-Opal system from Spectra-Physics producing tunable femtosecond pulses of 105 fs duration. Average output powers in excess of 300 mW and wavelength tunability in the range of 1400-1600 nm are available at a repetition rate of 82 MHz. The input pulse train is split into reference and signal arms and the former propagates through a delay line. The signal pulse is launched into a fiber under test with efficiency of 20%. The signal and reference pulses are then mixed in a 200 jim thick BBO crystal to generate sum-frequency cross-correlation component which is spectrally resolved in a 0.4 nm-resolution spectrometer. Fundamental Wavelength (nm)

Sum-Frequency Wavelength (nm)

Fig. 2 X-FROG spectrograms at the output of a 130 cm-long PCF with ZD point at 1510 nm pumped at that wavelength. Formation of the soliton-hole pair is marked with an arrow at (c) and (d). Onset of the resonant energy transfer into the spectral band between the soliton and the cw is shown in (e). (f) demonstrates emission of the Cherenkov radiation (arrow). Vertical line shows the ZD point and separates regions of normal and anomalous dispersion.

3.

Results and Discussion

In a PCF with the second ZD at 1510 nm, such as fiber (C) in Fig. 1, the anomalous dispersion region is located at shorter wavelengths with respect to second ZD point. Fig. 2 shows the spectral-temporal dynamics of a 105 fs soliton launched into the fiber exactly at the ZD wavelength. Initially a broad spectrum is generated through self-phase modulation. One part of the spectrum resides in the normal dispersion region at longer wavelengths and forms a dispersive wave, AAiiile the other part forms a soliton on the anomalous side. Several resonant

53

interactions between the soliton and a cw wave can be seen in Fig. 2 (arrows). These resonances lead to either energy transfer from the cw to the soliton. Fig. 2(c,d) or to the generation of new spectral-temporal components. Fig. 2(e). At h i ^ input power the soliton returns to the ZD point and can transfer most of its energy across the ZD point through phase-matched Cherenkov continuum generation (see below) [5,6]. If the fiber is pumped on the anomalous dispersion side. Fig. 3, different dynamics occurs. With increasing input power soliton formation is observed followed by its self-frequency shift towards the ZD point. In the vicinity of the ZD point phase-matched Cherenkov continuum radiation becomes highly efficient [5] across the ZD point to the normal dispersion side. Fig. 3(b,c arrow). There is a continuous energy flow from the soliton to the continuum. This continuum generation results in the recoil force acting on the soliton's spectrum thus stabilizing it at the wavelength near the ZD point [6]. Further power increase leads to the second soliton formation v^ich interacts with the Cherenkov continuum generated by the first soliton. This resonant interaction leads to the generation of distinct features marked Avith the arrows in Fig. 3(e,f). Fundamental Wavelength (|ini) 1.7

1.4

1.S

1.6

1.7

1.4

Sum-Frequency W avelength (nm)

Fig 3 X-FROG spectrograms for 1430 nm pump wavelength. Linear propagation, (a), is followed by the formation of the primary soliton and Cherenkov radiation (b,c). (d) shows formation of the second weaker soliton. (e-f) show scattering of the Cherenkov radiation emitted by the first soliton fi-om the second solitcm. This scattering leads to the energy transfer into a spectral band located between the solitons and the Cherenkov radiaticm band.

References 1 W. H. Reeves, D. V. Skryabin, F. Biancalana, J. C. Knight, P. St. J. Russell, F. G. Omenetto, A. Efimov, A. J. Taylor, Nature 424, 511, 2003. 2 N. Nishizawa and T. Goto, Opt. Express 8, 328, 2001. 3 J. M. Dudley, X. Gu, L. Xu, M. Kimmel, E. Zeek, P. O'Shea, R. Trebino, S. Coen, and R. S. Windeler, Opt. Express 10, 1215, 2002. 4 S. Linden, J. Kuhl, and H. Giessen, C^t. Lett. 24, 569,1999. 5 N. Akhmediev, M. Karlsson, Phys. Rev. A 51, 2602, 1995. 6 D. V. Skryabin, F. Luan, J. C. Knight, P. St. J. Russell, Science 301, 1705,2003.

54

Generation of Rotational Raman Emissions and Self-compressed Femtosecond Pulses in a Hydrogen Gas Shinichi Zaitsu, Yuichiro Kida, and Totaro Imasaka Department of Applied Chemistry, Graduate School of Engineering, Kyushu University, 6-10-1 Hakozaki, Higashi-ku, Fukuoka, Japan E-mail: [email protected] Abstract We observed the characteristics of a femtosecond laser pulse after passing the beam through a hydrogen gas. The pulse compression accompanied by Stokes emissions implies the coherently spectral broadening based on a Raman rotational coherence.

1.

Introduction

Generation of a ultrashort pulse based on phase modulation using Raman coherence has received considerable attentions in a few years [1-7]. The phase modulation of a ultrashort pulse is caused by a time-varying refractive index of a media where Raman coherence is excited by two-color nanosecond lasers [1] or an intense uhrashort pulse [2,3] This modulation provides high-order stimulated Raman sidebands [2,4] and negatively chirped spectral broadening [3]; the correction of the phase relationship among the Raman sidebands [5] or the addition of a positive chirp [3,6,7] gives rise to extremely shortening of a pulse duration to the extent of a few fs. We have demonstrated the compression of a femtosecond laser pulse without phase compensation as a result of nonlinear propagation of a femtosecond laser pulse in a pressurized H2 gas [8]. The notable point in our results is that the self-compression was achieved without an initial preparation of a Raman active medium by other pump lasers. Therefore it has a possibility of producing an efficient compression method of a laser pulse only by passing it through a H2 gas. In this report, we present the temporal and spectral characteristics of femtosecond laser pulses modulated through a nonlinear propagation in a Raman active medium. We show here the behavior of the reshaping of a temporal shape was explained by the coherently phase-locked generation of the Stokes emissions of para- and ortho-H2^

2.

Experimental Setup

A Ti:sapphire laser regenerative amplifier system (Thales, Concert) was used to generate ultrashort laser pulses at a center wavelength of 785 nm with a 1-kHz repetition rate. The maximum output energy was 1.5 mJ, and it was attenuated using a polarizer and a rotating waveplate. A Raman cell was filled with a H2 gas consisting of 25% of pam-Hs and 75% of ortho-U^i ^t a pressure of 10 atm. The

55

polarization of an input pulse was varied between linear and circular by rotating a quarter-wave plate placed just before a 5-mm-thick input window of the Raman cell. The output pulses were collimated and measured after cutting off the center part of the beam using an aperture with a diameter of 3 mm. The spectrum was measured with a multichannel spectrometer (Ocean Optics, USB2000), and the pulse width was estimated from an autocorrelation trace measured with a noncollinear multi-shot second harmonic autocorrelator (APE, Pulse Check).

3.

Results and Discussions

We carried out simultaneous observation of the evolution of waveforms and spectra of output pulses, while we were continuously changing an input laser energy by rotating the waveplate of the attenuator. We use a negatively chirped 190-fs pulse with circular polarization as input pulses. As increasing the input energy, we observed the dynamic acumination of the temporal shape accompanied by the growth of rotational Stokes emissions para- and ortho-II2. Figure 1 shows the cooperative behavior of the reshaping of the temporal shape and the broadening of the spectrum. Below the input energy of 0.52 mJ, the temporal shape of an output pulse exhibits no substantial change compared with that of the input pulse although the spectrum was slightly broadened in the bottom part. The Stokes emission of para-H2 began to generate at the input energy of 0.52 mJ, producing a little bump on the top of the autocorrelation trace. In the limit of input energy from 0.52 mJ to 0.68 mJ, the bump was gradually sharpened with the increase of the intensity of the Stokes emission of para-R2. The Stokes emission of ortho-Yi2 appeared at the threshold of 0.68 mJ, the acumination of pulse shape dramatically progressed although the Stokes emission of para-H2 decreased. When the input energy reached to 0.81 mJ, the pulse shape showed the shortest duration with a full width at half maximum (FWHM) of 45 fs in the autocorrelation trace, which corresponds to 29 fs under the assumption of a sech^ pulse shape. Further

Autocorrelations

300fe -500 6 500~ Delay (fs)

750 800~ 850 Wavelength (nm)

Fig. 1 Autocorrelation traces and spectra of output pulses for the several values of input pulse energy.

56

increase of the input energy over 0.81 mJ resulted in the broadening of the pulse shape with the decrease of the Stokes emission of ortho-ll2. This observed results obviously shows that the change of spectral shape directly affected the temporal shape of the output pulse. A sech^ pulse with a FWHM of 30 fs needs to have a spectral width of a FWHM of 10.5 THz. The spectrum in Fig. 1 apparently did not cover such a spectral range without the Stokes components. In addition, our results shows that the pulse width was determined by the total width of spectrum, i.e. the drastically shortening of the pulse width occurred after the generation of the Stokes emission of ortho-ll2 that was 17.6 THz from the pump laser frequency. Therefore we conclude that the generation of Stokes emissions is a direct cause of the pulse compression. This conclusion implies that the Stokes components of ortho-ll2 will be generated in phase with the pump and the Stokes emission of para-ll2.

4.

Conclusions

In this report, we observed spectral broadening of a femtosecond laser pulse after passing the beam through a pressurized H2 gas. We demonstrated that the generation of the rotational Raman emissions provided the remarkable reshaping of the output pulse. It is noted that the pulse shape was acuminated with increasing the intensity of Stokes emissions of para- and ortho-H2\ the Stokes components crucially contribute to the shortening of the pulse duration. This indicates that the phase relationship between the input pulse and Stokes pulses was locked. The result presented here is the first observation of femtosecond-pulse self-shaping caused by a nonlinear polarization based on the excitation of Raman coherence. We believed that the present approach has a potential for development in a novel method for compressing a laser pulse.

References 1 S. E. Harris and A.V. Sokolov, in Phys. Rev. Lett. Vol.81, 2894, 1998. 2 A. Nazarkin, G. Korn, M. Wittmann, and T. Elsaesser, in Phys. Rev. Lett. Vol.83, 2560, 1999. 3 R. A. Bartels, T. C. Weinacht, N. Wagner, M. Baertschy, Chris H. Greene, M. M. Murnane, and H. C. Kapteyn, in Phys. Rev. Lett. Vol.88, 013903, 2002. 4 A. V. Sokolov, D.R. Walker, D. D. Yavus, G. Y. Yin, and S. E. Harris, in Phys. Rev. Lett. 85, 562, 2000. 5 A. V. Sokolov, D.R. Walker, D. D. Yavus, G. Y. Yin, and S. E. Harris, Phys. Rev. Lett. 87, 0334202, 2001. 6 M. Wittmann, A. Nazarkin, and G. Korn, in Opt. Lett. 26, 298, 2001. 7 N. Zhavoronkov and G. Korn, in Phys. Rev. Lett. 88, 203901, 2002. 8 H. Ohtsuka, T. Uchimura, and T. Imasaka, in Opt. Lett. 29, 400, 2004.

57

Spectral broadening of 50 milli joule laser pulses in a neon-filled Herriot multiple-pass cell (MPC) Muhammad Nurhuda\ Akira Suda^, and Katsumi Midorikawa^ ^ Physics Department, Brawijaya, Brawijaya University, Malang 65144, Indonesia E-mail: [email protected] ^RIKEN,2-1 Hirosawa, Wako-shi, Saitama 351-0198, Japan E-mail: [email protected] Abstract. The application of a Herriot multiple-pass cell (MPC) filled with neon for spectral broadening of a femtosecond, intense laser field has been studied theoretically. The simulation using 50 mJ, 60 fs laser pulses with a center wavelength of 780 nm has shown that relevant spectral broadening can be obtained after 10 passes of the pulse, yielding a compressed pulse with a width of 4.5 fs and an energy of 25 mJ.

1. Introduction Pulse compression is a useful technique for shortening the duration of pulses generated by oscillators and amplifiers. A popular pulse compression technique makes use of self-phase modulation (SPM) followed by successive phase compensation. Self-phase modulation can be obtained by propagating an electric field in a nonlinear medium, either in a hollow fiber [1] or in free space (unguided). For pulse propagation in gaseous medium with a constant gas pressure, the gain of the SPM is related to pL, where p is the gas pressure and L is the effective optical path of the pulse. However, the latter method may give rise to several unexpected problems, e.g. self focusing and filamentation [2], which leads to difficulty in controlling the SPM. On the other hand, if the pulse is propagated in free space, the effective optical path is too small to generate adequate SPM. hi order to extend the optical path, we propose the use of a Herriot multiplepass cell (MPC) filled with neon. The MPC consists of two mirrors with the same curvature arranged such that the pulses can be multiply-passed in the space between the two mirrors [3]. When a Gaussian input beam is properly adjusted to match the fundamental mode, all of the spot sizes of the beam on the mirrors become equal and the focused points are located at the center of each transit path. To show the effectiveness of the SPM in MPC, we have carried out simulation using a laser energy of 50 mJ and a pulse duration of 60 fs at full width at half maximum (FWHM). The laser wavelength used was 780 nm, i.e. the center wavelength of the Tiis^phire laser system. The MPC length, i.e. the distance between the two mirrors, was chosen to be 3 m, and with the radius of curvature of the mirror was 1.6 m. Neon was chosen as the gaseous medium at a pressure of 0.05 atm. The input beam was assumed to be the fundamental mode of a Laguerre-Gaussian fimction with a waist size of 0.32 mm and was located at the center of the MPC transit.

58

The simulation was based on the extended nonlinear Schrodinger equation, which includes diffraction, dispersion, Kerr self focusing, plasma defocusing and multiphoton absorption. We assumed that the beam is radially symmetric and the electric field is linearly polarized. The propagation is carried out recursively step-by-step using the split operator method, which has been described elsewhere [4]. In performing propagation with respect to diffraction, the wave is expanded in a set of Laguerre-Gaussian basis functions, and the associated coefficients of expansion are then multiplied with the corresponding propagation constants. The effect of nonlinearity is applied over time after reconstructing the spatial wave from the basis functions. The propagation step dz is treated as a variable, depending on the strength of the nonlinear interaction of the medium with the pulse. The number of LaguerreGaussian basis fiinctions used to expand the electric field is made to increasingly vary in each pass, guaranteeing that converged results could be obtained after each pass.

2.

Simulation Results and Discussion ,

'

I

'

I

'^ 1-5 h-

F

'

I

'

I

I

'

I '

I '

after 3 passes

after 2 passes

I after 1 pass

4

ip

% 0.5 o

800

900

T

600

700

800

Wavelength (nm)

Fig. 1.

500

600

'

600

700

800

Wavelength (nm)

700

I ' I

600

700

800

Wavelength (nm)

Development of the total power spectrum after each transit in the MPC.

hi Fig. (1), the spectra of the pulse after the n-th transit are displayed. It can be shown that the spectrum of the pulse becomes broader after the end of each transit. Moreover, due to the high plasma density, the broadening mostly occurs towards shorter wavelength. At the end of the 10th transit, the transmitted energy was 30 mJ. The total energy loss was found to be 20 mJ, of which 10

59

mJ was due to dispersion loss by the mirror and the rest was due to losses caused by ionization. The reflectance of the mirror in this case was assumed to be 98%. The power spectrum after 10 passes is shown in Fig. 2 (left). This power spectrum was obtained by integrating the beam inside a circular area of radius 4 mm. We have noticed that the beam profile after 10 passes remains almost Gaussian, and that the Section of the beam inside the target area is about 85%. For the purposes of phase compensation, it is necessary to check the spectral chirp, as shown in Fig. 2 (right). It can be seen that, in general, the spectral chirp is smooth and thus enables phase compensation to be performed. Finally, performing a Fourier transform of the wave within the wavelength range between 550 nm and 900 nm, and inside the central area of 4 mm radius, we obtained a single pulse with a duration of 4.5 fs (FWHM) and an energy of 24 mJ. H-.U

1

'

1 '

1 —^ 1 A

-j L_ -

2.0

. rsl 0.0

/

-2.0 -4.0 V 500

600 700 800 Wavelength (iim)

900

r 500

1

1 600

1 1 700

1

1

800

-

900

Wavelength (nm)

Fig. 2. The power spectrum in the circular area of 4 mm radius (left) and the spectral chirp at the center of the beam.

References 1. 2. 3. 4.

60

M. Nisoli, S. De Silvestri, and O. Svelto, Appl. Phys. Lett. 68, 2793 (1966). M. Nurhuda. A. Suda, M. Hatayama, K. Nagasaka, and K. Midorikawa, Phys. Rev. A. 66, 023811 (2002). T. Takasaki, A. Suda, K. Sato, K. Nagasaka, and H. Tashiro, Appl. Opt. 36, 3413 (1997). M. Nurhuda, A. Suda, M. Hatayama, K. Nagasaka, and K. Midorikawa, J. Opt. Soc. Am. B 20, 2002 (2003).

CEO phase preservation in chirped-pulse optical parametric amplification of 17.3-fs pulses Jens Biegert^ Christoph P, Hauri\ Philip Schlup^ Wouter Komelis^ Fiorian W. Helbing^ Ursula Keller^ and Gunnar Arisholm^ ^ Swiss Federal Institute of Technology (ETH), Physics Department, Zurich, Switzerland E-mail: [email protected] ^ Forsvarets forskningsinstitutt (Norwegian Defence Research Establishment), P.O. Box 25, NO-2027 Kjelier, Norway Abstract We demonstrate the preservation of the carrier-envelope offset (CEO) phase in a chirped-pulse optical parametric amplifier (CPOPA), which yields 85-/<J CEO phasestabilized pulses that are recompressed to a near-transform-limited duration of 17.3 fs. Chirped-pulse optical parametric amplification (CPOPA) [1-3] is an attractive alternative to conventional stimulated emission-based chirped pulse amplifier (CPA) systems for the amplification of carrier-envelope offset (CEO) phase stabilized, high intensity, ultrashort pulses, such as those required for high harmonic generation. Its advantages include large single-pass gains with virtually no thermal loading and large phase matching bandwidths that can be tailored by choice of crystal and interaction geometry. Previous ultrashort pulse OPA devices have focused on amplifying white-light continuum seed pulses at the well-known "magic" visible-wavelength broadband phase matching angle in BBO [4] to the few-microjoule level, and pulses as short as 4 fs have been demonstrated [5]. Here, we demonstrate direct amplification of Tiisapphire pulses to 85 piJ using CPOPA, and recompression to a nearly transform-limited duration of 17.3 fs, using the experimental configuration shown in Fig. 1. More significantly, we have demonstrated the phase preservation of CPOPA by amplifying the output of a phase-stabilized oscillator and measuring the CEO phase drift [6, 7] of the amplified pulses. Theory asserts that the phase of the amplified seed remains, aside from quantum noise, unaltered by the amplification with a non-stabilized pump, since the idler field dissipates the phase offset, but experimental verification of this has not, to our knowledge, been previously reported.

Fig 1. Experimental configuration. PLL, phase-locked loop to phase-stabilize the oscillator; SHG, second-harmonic generation crystal; OPA, 3-mm BBO for near-degenerate phasematched CPOPA.

61

The major limitation to CPOPA to date has been the availability of pump sources capable of delivering sufficiently short, high-energy pulses. Here, we chose a picosecond pump source since it represents an ideal compromise between pulses short enough to allow for bulk stretching and prism compression of the seed, avoiding potentially phase-disturbing diffraction gratings [8], but sufficiently long to alleviate the need for precise pulse-front matching [5]. We used a modified conunercial Ti:sapphire regenerative amplifier (Legend, Positive Light), and seeded it with part of the output from the seed oscillator to facilitate pulse synchronization. The pulses from the regenerative amplifier were frequencydoubled in BBO, yielding 2.4-ps pump pulses at 400 nm with a pulse energy of 0.9-mJ. The pump pulses were focused into the CPOPA crystal (3-mm BBO, 6 = 29.2°) to an intensity of 65 GW/cml The seed laser was a CEO-phase-stabilized, commercial Ti: sapphire oscillator, that delivered 700 mW of 12-fs pulses at a repetition rate of 76 MHz. After splitting off part of the beam for phase stabilization and seeding the pump, some 175 mW remained to seed the CPOPA. Seed pulses were selected at a 1-kHz repetition frequency and stretched in a DAZZLER before being amplified in the CPOPA and recompressed in a prism compressor. In practice, the DAZZLER simultaneously facilitated higher-order dispersion correction during the optimization of the pulse compression. The stretched 1-nJ seed pulses were loosely focused into the CPOPA crystal and temporally overlapped with the pump. We numerically modeled the CPOPA system using an oscillator pulse measured with SPIDER [9, 10], a full 3D simulation of the parametric amplification process [11], and ray-tracing of the prism compressor, and found that excellent passive compensation could be achieved using the DAZZLER as a stretcher and a suitably arranged prism compressor.

0 Time [Fs]

50 490

500

510 520 Wavelength [nm]

Fig 2. (a) Reconstructed pulse profile of the compressed, phase-stabilized pulses, and (inset) measured far-field spatial intensity distribution, (b) CEO phase interference fringes averaged over 10,000 shots for free-running (dashed line) and phase-stabilized (solid) seed pulses, demonstrating the phase preservation of the CPOPA. For a pump energy of 0.9 mJ, the 1-nJ seed was amplified to 85 piJ, corresponding to a single pass gain of 8.5 x 10"^. After amplification, the pulses were recompressed in a prism compressor, which reduced the energy available in the compressed pulses to 77 /^L The compressed pulses were characterized by SPIDER, and the reconstructed spectral phase variations were minimized using

62

the incident phase adjustment provided by the DAZZLER. The reconstructed, optimized temporal pulses shape, with a near-transform-limited duration of 17.3 fs, is shown in Fig. 2(a). The far-field transverse intensity profile, recorded with a high resolution CCD, is shown in the inset. In order to verify the phase preservation in our configuration of CPOPA, we measured the CEO phase of the amplified pulses using the so-called "/-to-2/' interferometer, where a spectral beat signal is recorded between the highfrequency and frequency-doubled low-frequency wings of a white-light spectrum, generated in a sapphire plate [12]. The CEO phase could be derived from the spectral location of the interference fringes, which were recorded with a linear CCD-equipped spectrometer. The dashed line in Fig. 2(b) shows averaging over 10,000 laser shots with the phase stabilization to the oscillator switched off, where no interference fringes are visible since successive pulses have random relative CEO phases. By contrast, the interference fringes for the CEO-phase-stabilized oscillator (solid line) are clearly resolved. Single-shot measurements of the CEO interference fringes indicated that the fringes remained resolved and stationary for in excess of 15,000 shots, apart from small drifts introduced by air currents and mechanical vibrations. In conclusion, we have demonstrated direct amplification of 1-nJ, phasestabilized oscillator pulses to 85 pi5 for a 0.9-mJ pump, and verified that the CEO phase is preserved during the CPOPA process. The amplified pulses were recompressed to a near-transform limited pulse duration of 17.3 fs. We anticipate that the CPOPA output will soon be suitable for our high-field physics experiments. The output energy could be directiy increased by multi-passing the CPOPA crystal, while our numerical modeling predicts further energy scaling with larger beam sizes.

References [I] A. Dubietis, G. Jonusauskas, A. Piskarskas, Opt. Commun, 88, A?»1-AAQ (1992). [2] T. Wilhelm, J. Pid, E. Riedle, Opt, Lett. 22, 1494^14% (1997). [3] I. N. Ross, P. Matousek, G. H. C. New, and K. Osvay, J. Opt. Soc. Am. B 19,29452956 (2002). [4] G.M. Gale, M. Cavallari, T.J. Driscoll, F. Hache, Opt. Lett. 20,1562-1564 (1995). [5] A. Baltuska, T. Fuji, T. Kobayashi, Opt. Lett. 27, 306-308 (2002). [6] H.R. Telle, G. Steinmeyer, A.E. Dunlop, J. Stenger, D.H. Sutter, U. Keller, Appl. Phys. 5 69,327-332(1999). [7] F. W. Helbing, G. Steinmeyer, U. Keller, IEEE J. ofSeL Top. in Quantum Electron. 9, 1030-1040 (2003). [8] F.W. Helbing, G. Steinmeyer, J. Stenger, H.R. Telle, U. Keller, Appl. Phys. B 74S, S35-S42 (2002). [9] C. laconis, I. A. Walmsley, Opt. Lett. 23,792-794 (1998). [10] W. Komelis, J. Biegert, J.W.G. Tisch, M. Nisoli, G. Sansone, C. Vozzi, S. De Silvestri, U. Keller, Opt. Lett. 28,281-283 (2003). [II] G. Arisholm, J. Opt. Soc. Am. B 16, 117-127 (1999). [12] M, Kakehata, H. Takada, Y. Kobayashi, K. Torizuka, Y. Fujihira, T. Homma, H. Takahashi, Opt. Lett. 26,1436-1438 (2001).

63

Long-term stabilization and control of CEP of idler from NOPA S. Adachi and T. Kobayashi Department of Physics, Graduate School of Science, University of Tokyo, Hongo 7-3-1, Bunkyo, Tokyo 113-0033, Japan E-mail: [email protected] Abstract. Compressed idler pulse from NOPA with a deformable mirror was characterized by SFM XFROG and long-term CEP of the idler was stabilized and controlled by f-to-2f interferometry scheme.

1.

Introduction

Broadband phase-matching property of p-BaB204 (BBO) crystal in a type-I noncollinear optical parametric amplifier (N0PA)[1] and the development of sophisticated pulse compression technique (a custom-designed ultra-broadband chirp mirror and an adaptive pulse shaper enabled to obtain as short as 4-fs pulse covermg nearly full visible spectral range and NIR was generated[l]. Furthermore, idler from the NOPA system[2], pumped with the second harmonic (SH) of the fundamental TiiSapphire radiation and seeded with the white light continuum produced by the same SH, showed very useful phenomenon of self-elimination of pulse-to-pulse carrier-envelop phase (CEP) slip. In this report, pulse characterization by SFM XFROG and compression with a deformable membrane mirror and long-term CEP stabilization and control by/-to2 / interferometry scheme is performed concerning the idler radiation from noncollinear OPA.

2.

Experimental Methods

Figure 1(a) shows the schematic of our setup of XFROG measurement with broadband type-I SFM in a BBO crystal. SH pulse and SH-originated supercontinuum pulse are introduced into the OPA BBO crystal with a noncollinear angle. The idler output from the NOPA has an angular dispersion that fulfills phase-matching condition among pump, signal, and idler. Generated idler pulse is injected to the BBO crystal for XFROG measurement after bounce of flexible mirror and transmission through 2-mm-thick BK7 glass plate. Idler radiation of NOPA is negatively chirped because the red part of the idler corresponds to the blue part of the positively chirped super-continuum seed and vice versa. This negative chirp has been substantially compensated by material dispersion of a BK7 glass and residual chup is eliminated with a deformable-membrane mirror (OKO technologies) with adaptive control of 19-pixels. An optical multi-channel analyzer

64

was used to detect the XFROG signal. The Ti:sapphire fundamental radiation (~ 790 nm, ~ 120 fs) was utilized as the reference pulse of XFROG. In the NOPA an octave-spanning idler spectrum can be generated directly as shown later in Fig. 2(a), moreover, frequency doubling of the idler can be attained in the same OPA crystal in a broad spectral range[2]. These properties enable us to employ/•to-2/interferometry for CEP drift measurements as shown in Fig. 1(b) without any additional spectral broadening. The 800-nm components of the idler and idler-SH emitted to the different directions are recombined and are transmitted through a 50-|xm pinhole, placed between a confocal lens pair, to ensure spatial coherence. Obtained phase signal is used to stabilize and to control the CEP of pump pulses by changing the path length inside the CaF2 as shown in Fig. 1(a). CEP of the idler from the NOPA system is consequently controlled because the CEP of the idler is determined by the difference between CEPs of the pump and the seed pulses. Regen. amp 120fs, SOOMJ 500 Hz, 790 nm

0

», variable delay line

Idler (f) & Idler SH (2f)

(b)

X ^ £.,»;S8» nm

E2f!400 nm

,

!

HWP@800 nm

I

1/ '

row horizontal polarization vertical polarization

idler & SH of idler

Fig.l. (a)Setup of CEP self-stabilizing NOPA. #1: ^72 wave plate; #2: variable neutral densityfilter;#3: 2-mm-thick CaF2 plate on vertical rotation stage to avoid damages due to high peak pulse energies; #4: 1-mm-thick CaF2 plate on horizontal rotation stage for CEP stabilization and control of the idler output. (b)Setup of f-to-2f interferometry for the CEP drift measurement.

3. Results and Discussion Figure 2(a) shows the idler spectrum from our NOPA system. TL pulse-width of the NOPA idler output is 4.0 fs, which corresponds to about 1.2 optical cycles of 3.3 fs with a center wavelength at 990 nm. Such a quasi-monocycle idler output with no CEP slip can be utilized to study CEP-sensitive phenomena. However, there is a difficulty m the characterization of idler pulses with more than an octave bandwidth utilizing conventional up- or down- conversion techniques and to achieve precise pulse compression. In order to overcome this difficulty in the characterization, we have employed a cross-correlation frequency-resolved optical gating (known as XFROG) method with broadband sum-frequency mixing (SFM)

65

taking advantage of the idler angular dispersion. Figure 2 shows (b) the pulse shape and (c) the spectrum along with time- and frequency-domain phases after retrieval, respectively. TL pulse-width that can be calculated from the retrieved spectrum is 4.2 fs, which is close to the calculated value from the NOP A idler output spectrum in Fig. 2(a). This result confirms that broadband SFM XFROG measurement was properly performed with fiill NOPA idler spectral bandwidth. Our retrieved pulse-width (4.3 fs) is also nearly equal to the TL pulse-width, which corresponds to ~ 1.3 optical cycles at the center wavelength of 970 nm. Therefore the chirp was successfully compensated with a deformable-membrane mirror with adaptive control.

800

1000

1200

1400

Wavelength (nm)

Time (fs)

lobo i2bo i4bo Wavelength (nm)

Fig. 2. (a)The spectrum of idler from the NOPA system. (b)The pulse shape and (c) the spectrum along with time- and frequency-domain phases after retrieval, respectively. Long-term relative CEP drift between idler and idler-SH of the NOPA is retrieved directly by Fourier Filtering method. The CEP of the idler is fau-ly well controlled and stabilized within a few feedback cycles. Residual phase instability is partially caused by the spectral instability of the idler. In the NOPA, spontaneous parametric fluorescence radiation is generated at the same time and it can interfere with the idler radiation around the 800-nm spectral range. This interference fringe is randomly fluctuating because spontaneous radiation has by no means a fixed phase relationship with the idler.

4.

Conclusion

The characterization of broadband idler output pulses from NOPA was performed with broadband SFM XFROG. By compensating the residual higher-order dispersion using adaptive control of deformable muror, quasi-monocycle pulses with 4.3 fs pulse duration were achieved. Besides, CEP control of the idler is successfiiUy performed by changing the path length of the pump of NOPA.

References 1 2

66

A. Baltuska, T. Fuji, and T. Kobayashi, " Visible pulse compression to 4 fs by optical parametric amplification and programmable dispersion control," Opt. Lett. 27, 306 (2002) A. Baltuska, T. Fuji, and T. Kobayashi, "Controlling the carrier-envelop phase of ultrashort light pulses with optical parametric amplifiers," Phys. Rev. Lett. 88, 133901 (2002)

Experimental and Theoretical Study of a Visible NoncoUinear Optical Parametric Amplified Pulse with 200 THz Bandwidth Xiaojun Fang and Takayoshi Kobayashi Department of Physics, Faculty of Science, University of Tokyo, 7-3-1 Hongo, Bunkyo, Tokyo 113-0033, Japan E-mail: [email protected] Abstract. We numerically simulated the operation of a visible y^-barium borate noncollinear phase-matching optical parametric amplifier (NOPA) pumped by a frequencydoubled 1-kHz Ti:sapphire amplifier. The theoretical results almost perfectly match the experimental results.

1.

Introduction

The optical parametric process is an important route to generate either few-cycle [ 1 , 2, 3, 4] or tunable optical pulses [5, 6], which are desirable for many experiments in physics, chemistry, and biology. Despite recent theoretical progress [ 7 , 8, 9], the mterpreting experimental observation and optimizing noncollinear optical parametric amplifier (NOPA) designs is still a hot research subjects because of the complicated interplay among material dispersions, secondand third- order optical nonlinearities of pump, signal, and idler pulses. Appropriate computer simulations can provide much insight into the mechanism of the processes involved and to exploit fully the potential of the design.

2.

Experimental setup and numerical simulation

The pump source used in our NOP A system is a 120-fs FWHM pulse of energy 1.5 mJ at 786 nm from a Ti:sapphire regenerative CPA (Thales Laser, Bright) with 1-kHz repetition rate. About 4% of energy is split off for supercontinuum (SC) generation, while the main beam with 96% pulse energy is telescoped to obtain high intensity for second-harmonic (SH) generation. After a beam splitter and a variable neutral-density (VND), SH pulse with 50 fiJ is focused to BBO crystal by a 400-mm fiised silica lens. To increase the overlapping of seed with the pump pulse in time domain, the pump pulse is elongated by 50-mm block of quartz, and the seed is also pre-compressed by prism pairs, in which a razor blade is inserted to cut off the fiindamental. Furthermore, the pump beam is transmitted through a 45° fiised-silica prism with an incident angle of 49° to overlap the pulse fronts of the pump and signal in BBO crystal and achieve the "pulse-front matching". Because the tilted-pump geometry is used for pulse-front matching, the electromagnetic fields were assumed to be plane wave, and accordingly the transverse variation of the field was ignored. Therefore, in the slowly varying envelope approximation, the Maxwell equations lead to the followmg nonlinear three-wave-interaction (TWI) equations:

67

^^

(

-4-

'

A \':^A

PdA^

\

Us COS a

dz ^ S

1

A

2

• ^ S

Up (2)

,

dt 4*

2

2 dcj' /-A I

^

«^c

f 1 1 ^ ;:5 ^ n a^, 1 ^+ Uj cos J3 u^ az 'pj dt A

' ^ 1

^,2

dt

. ^ ^ S

2/75C

QA

^ I

^'^h ^'^ i: d k^ d A^

iVi

A

A*

(3)

•

A

\ \ A

\2

2

V

i

.

|2

3

2

i

.

.:

^

|

.

3

,• ;:52/^

;::52 / d'ki d^Aj 2 do)^ dt / . A f

X

. ^<^/

(Vi

A \ ^ \

-~A,+i-!-xfiA,Asty.^{it^)^i—^XeffA\-\A,\

A

\2

\ A \2

+\A\

|2

+-\A,\

~&"~2 5«' a/' |2

2

WpC

2nj,c

\y

I ,

|2

3

Here, we used retarded time coordinates for the pump wave. With 3.7° between the direction of pump to that of the seed, the phase matching indicated an amplified visible pulses with a bandwidth of nearly 200 THz. Furthermore, the angular dispersion of the pump beam also enhances the phase-matching bandwidth [4], so we assume that the process is almost perfectly phase matched, a, p are the noncollinear angles, which the wavevectors of the signal and idler make with that of the pump, respectively. The refractive index and the group velocity are denoted nj and uj, where7 = 5, /, and P for signal, idler and pump, respectively. The c is the vacuum light speed, and x^^V^^^ X^^''e#are, respectively, the effective secondand third- order susceptibilities of the crystal. We simulated the nonlinear propagation with the algorithm combining split-step fast-Fourier-transform method and fourth-order Runge-Kutta method. The wavelength dependent phase-matching condition is also incorporated into equations [9]. It is found that the simulation generate significant errors based on the method in ref [10], when either the duration of pulse is shorter than 30 fs or bandwidth is more than 100 nm. In that algorithm, the group delay dispersions (GDD) are artificially separated from the nonlinear terms. The former terms are solved in frequency domain and the latter are solved in time domain. In our simulation, the algorithm is to (1) calculate every second-order nonlinearity using the convolution of the other two electronic fields in frequency domain, (2) to calculate the third-order nonlinearity in time domain, (3) to calculate the linear term in frequency domain, (4) to use the Runge-Kutta method to integrate the equation in frequency domain including all linear and nonlmear terms, (5) to inverse Fourier transform to yield the time envelopes of the fields.

3.

Results and Discussion

In our numerical simulations, the seed is a white-light continuum with spectrum extending from 450 to 800 nm and proper group delay. The latter is adjusted using the pre-compression prism pairs (GDD is set to 0 fs^ at 550 nm in our calculation for maximum amplification window time [4]). To compensate the group velocity mismatch (GVM) between the pump and signal, the seed is pre-delayed -500 fs with respect to the pump pulse. Figure 1 (a) is the experimental and theoretical spectra of signal after single-pass amplification. It is also found that the numerical

68

amplified spectral bandwidths are much narrower due to the effects of the dispersion of material and chirps of both seed and pump. The effects of self-phase and cross-phase modulation broaden the spectra of signal during amplification. 1.0-

(b)

. /r\

^^ 0 8c

///»/

A \ ^ Jr

/

\ 1

0.2-

*

A ^ /'\

\ \ \

/'

"eS 0.4-

signal

Calc/'

06-

Ti (U N

O

\

WExp. \

i 1 '

*

11/ ' j /

00500

550

600

650

700

Wavelength (nm) ^

/5U

SUU

5QQ

550

600

650

„ , i ^u / ^ Wavelength (nm)

Fig. 1. (a) The solid curve is the measured spectrum of amplified signal after single-pass amplification, while the dashed curve is the simulated spectrum, (b) The solid line is the experimental spectrum of the signal after double-pass amplification, while the dashed one is the simulated spectrum. To avoid the effect of temporal walk-off, the signal is double-pass amplified instead of passing in a thicker crystal. After balancing both pre-delays to enhance the red-shifted and blue-shifted amplification in single-pass and double-pass in BBO, we theoretically and experimentally got the maximum bandwidth operation. The Figure 1 (b) shows the spectra of both experimental and theoretical signal after double-pass amplification, which agrees fairly well with each other. Compensation of the dispersion of signal has induced the few-cycle Fourier transform limited visible pulses [4].

4.

Conclusions

In conclusion we simulated the dynamic processes in the visible noncollinear phase matching optical parametric amplifier pumped by a fi*equency-doubled 1kHz Ti:sapphire amplifier. This work offers much insight into the mechanism of the processes involved in NOP A and to exploit fiilly the potential of the design.

References 1 A. Shirakawa, I. Sakane, M. Takasaka, and T. Kobayashi, Appl. Phys. Lett. 74, 2268 (1999) 2 G. Cerullo, M. Nisoli, S. Stagira, S. De Silvestri, Opt. Lett. 23, 1283 (1998) 3 T. Kobayashi, and A. Shirakawa, Appl. Phys. B 70 [S], S239 (2000). 4 A. Baltuska, T. Kobayashi, Appl. Phys. B 75, 427(2002). 5 T. Wilhelm, J. Piel, E. Riedle, Opt. Lett. 22, 1494 (1997) 6 Akira Shirakawa, Isao Sakane, and Takayoshi Kobayashi, Opt. Lett. 23, 1292 (1998) 7 G. M. Gale, F. Hache, and M. Cavallari, IEEE J. Sel. Top. Quantum Electron. 4, 224(1998) 8 Stefan Reisner and Michael Gutmann, J. Opt. Soc. Am. B, 16, 1801 (1999) 9 Jiun-Cheng Wang, and Juen-Kai Wang, J. Opt. Soc. Am. B, 21, 45 (2004) 10 G. M. Gale, M. Cavallari, and F. Hache, J. Opt. Soc. Am. B, 15, 702 (1998)

69

Tunable wavelength pulse shaping of visible NOPA outputs with an Acousto-Optic Programmable Dispersive Filter D. Kaplan^ P. Toumois \ B. ChateP, and A. Monmayrant^ ' Fastlite, Campus de I'Ecole Polytechnique, 91128 Palaiseau, France E-mail: [email protected] - Lab. Collisions Agregats Reactivite, CNRS, IRSAMC-UPS, 118 route de Narbonne, 31062 Toulouse, France E-mail: [email protected] Abstract. An Acousto-Optic Programmable Dispersive Filter (AOPDF) is used to pulse shape a NOPA. We analyse limitations due to spectral dispersion, Kerr effect and acoustic attenuation. Better than 30 fs pulse duration is experimentally demonstrated. The recent introduction of the Acousto-Optic Programmable Dispersive Filter (AOPDF) has provided means to control spectral phase and amplitude of femtosecond pulses in a centimetre size solid state device. Most of the usages of the AOPDF have been to shape the seed oscillators of laser amplifiers where optical power and material dispersion are of no concern. Here, we report pulse shaping of NOPA outputs. Fig. 1 shows the basic experimental configuration. The tuning range is 460 to 770 nm in this experiment. optional glass stretchers NOPA

I—I

AOPDF Crystal

—J

Arbitrary Waveform RF Generator Fig, 1. General Set-up

Dispersion issues To generate a short pulse, the dispersion programming of the AOPDF must compensate for : a)the existing time dispersion of the NOPA source, b)the dispersion of the AOPDF material (Te02) and c)the dispersion of the optional glass stretchers. These stretchers are introduced to decrease optical intensity in the crystal. The maximum delay between any two frequencies that can be programmed is: Tmax ^ Ang cos^(9o) L/c where Ang is the group birefringence, OQ beam orientation with respect to the [110] crystal axis, L the crystal length and c the 70

speed of light. A part of this delay Tcomp will be used for the dispersion compensations referred to above. The difference Tghaping^^ T^ax - Tcomp constitutes the pulse shaping capability of the device around a given central wavelength.. As discussed in [1], one can choose the crystal orientation to increase Tcomp while losing some of the diffraction efficiency. Fig. 2 A shows the theoretical Tshaping (no NOPA dispersion) as a function of wavelength for the parameters used in our experimental study (0o=38.5°, L=25 mm, 100 fs dispersion from the glass stretchers). Pulses down to 20 fs can be handled over the whole wavelength range.

Acoustic Attenuation issues The acoustic attenuation for arbitrary orientation and frequencies in paratellurite had not been previously investigated. A 10 db maximum loss, at 480 nm, is estimated from our experiments on the crystal used in the present study (optimised for shortest pulse capability), resulting in correspondingly reduced diffraction efficiencies. Changing 0o, a better compromise between efficiency and pulse shaping capability will be realized in friture work (e.g. 30 % to 50 % efficiency, while maintaining pulse shaping capability in the 20 to 30 fs range).

0.55

0.6

0.55

0.65

0.6

0.65

0.75

lambda (microns)

lambda (microns)

Fig. 2. A) Pulse shaping capability Tshaping versus lambda for pulse durations 20,30,40 and 50 fs (increasing line thickness), B). Energy flux /B versus lambda for same pulse durations

Kerr effect issues A B-Integral limit of order 1 radian, or less, is required to avoid self focussing in the AOPDF. We take into account pulse duration changes as the pulse propagates in the crystal. To simplify the analysis, we consider the case of weak diffraction efficiency and compute the B-integral of the non diffracted beam. The local light intensity I(x) and B-Integral are given to a good approximation as: f(x) =

=— — = = = = - .,B B= — ^\+(l]„+7]x/Lf

X

\n2l(x).dx =

Z

.«2^.[arcsinh(7)-arcsinh(7(|)]

71

where IQ is the intensity at the input face, r| is the ratio of the material total dispersion time spread to the Fourier transform pulse duration limit tp of the pulse, r|o the ratio of dispersion time to tp at the input face and n2 is the non linear optical index. From results on sputtered Te02 films [2] and our own evaluation using a Zscan technique, the value of n2 is of order 1. 10"^"^ cmVw in our wavelength range. Fig. 2 B shows the laser input energy per unit area/B-integral value as a function of wavelength under the same conditions as fig. 2 A. The order of magnitude of the allowable intensity (B-Integral =1) is 300 |iJ/cm^ (30 |LIJ for 3 mm diameter beam).

Experimental results Experiments were performed with a double pass NOP A [3] similar to the one described in [4]. The beam diameter was around 1.5 mm. Using neutral densities, the energy in front of the AOPDF varies between 1 to 10 |LIJ corresponding to energy densities up to 600 |iJ/cm^. Fig. 3 shows the spectrum after the AOPDF for several wavelengths. In the insets the second harmonic autocorrelation in intensity, performed in a 20 jim BBO crystal is recorded. The FWHM is sub-30 fs assuming a sech" shape for the band between 520 nm and 630 nm. Around 500 nm, the pulse lengthened to 43 fs due to acoustic attenuation cutting the short wavelength edge of the pulse spectrum, as discussed above.

ST

52) 5« >.(nn)

ffiO

5«

SaO

580

eOO

62)

560

600

63D X(nnn)

&0

Fig. 3. Spectrum and intensity for various wavelengths In conclusion, the feasibility of using an AOPDF to pulse shape NOPA outputs in the 480 nm to 750 nm range has been analysed theoretically and confirmed experimentally. The constraints are compatible with typical NOPA performance.

References 1 2 3 4

72

D. Kaplan and P.Tournois, J. Phys. IV France, Vol.12 , Pr5-69, 2002. F.d'Amore et al., Journal of Applied Physics, Vol. 94, n^3, 1654, 2003. B. Chatel, J. Degert, S. Stock, and B. Girard, Phys. Rev. A 68, 041402R 2003. E. Riedle et al, Appl. Phys. B, Vol. 71, 457,2000.

Broadband high power optical chirped pulse amplification N. Ishii'", R. Butkus\ A. Baltuska'"^, V. SmiIgevicius^ R. Danielius\ A. Piskarskas^ and F. Krausz'" ' Max-Planck-Institut fur Quantenoptik, Hans-Kopfermann-Strasse 1, D-85748 Garching, Germany ~ Institut fiir Photonik, Technische Universitat Wien, Gusshausstrasse 27/387, A-1040 Wien, Austria E-maiI: [email protected] "^Quantum Electronics Department & Laser Research Center, Vilnius University, Sauletekio ave. 10, LT-2040 Vilnius, Lithuania Abstract. We report a 110-THz 8-mJ non-collinear optical parametric chirped pulse amplification in type I BBO. The seed laser was actively synchronized with a picosecond Nd:YAG pump laser. This compact system is projected to deliver a 1 TW peak power in a 6-fs pulse.

The concept of optical parametric chirped pulse amplification (OPCPA) [1,2] offers a promising route toward ultrahigh-peak-power laser systems. In comparison with power amplification based on gain media with population inversion, OPCPA has several clear advantages. The most important of them are: a typically very wide bandwidth of parametric gain, negligible thermal load of the transparent parametric crystal, and an extremely high single-pass gain. These prerequisites are ideally suited for designing high-repetition-rate laser systems generating high-power few-cycle laser pulses. Whereas high-intensity and highenergy sub-ps OPCPA pulses as well as sub-lGO-jLiJ few-cycle pulses have already been demonstrated [3-5], terawatt-class sub-10-fs parametric amplification is yet to be achieved. The overall performance of such OPCPA schemes and their suitability for replacing conventional laser amplifiers will depend strongly on a number of key characteristics, such as the stability of the spectral phase and amplitude and the complexity of the phase distortion imparted on the pulse through the process of amplification. These properties are in turn affected by the nonlinearity of the parametric crystal; the angular geometry and length of the parametric interaction; individual stability of the pump and the seed pulses; the quality of seed—pump pulse synchronization; and the overall mechanical stability that rapidly gains importance with the increase of the seed pulse stretching ratio. The setup, schematically drawn in Fig.l (left), consists of a femtosecond seed oscillator and an independent 60-ps pump laser, the repetition rates of which are actively synchronized to an external RF frequency clock. The broadband 2-nJ seed pulse (Fig. 1 (right), solid line) is stretched in bulk material to a nearly rectangular temporal profile with about 40-ps duration. This amount of seed stretching ensures that the intensity of the pump pulse is adequately high over the time window containing all spectral components of the seed pulse which can be simultaneously amplified in the parametric crystal. The OPCPA consists of two non-collinearly phase-matched stages: a 2-mm- and a 4-mm-thick type 1 BBO crystals, pumped

73

with -5 mJ and -45 mJ, respectively. The signal pulse is amplified to -250 juJ in the first stage and to 8 mJ in the second stage. This two stages strategy is adopted for angle dithering of the two crystals to control and enhance the spectral bandwidth [6,7] and to control the pump power for each stage to increase the contrast ratio of the powers between the amplified pulse and amplified superfluorescence. The spectrum of the amplified pulse (Fig. 1 (right), shaded contour) has a FWHM of about 110 THz and corroborates well with the cross section of the seed spectrum and the calculated gain bandwidth (dotted curve). In type I BBO, it is very difficult to amplify the blue wing of the seed spectrum below 670 nm because of a steeply mounting phase mismatch and the onset of idler absorption. Nevertheless, assuming ideal pulse compression, the designed amplifier supports 6-fs pulses (Fig.l (right), inset) with the corresponding peak intensity of over 1 TW. 532 mi50 lii 50|x 2C|-k (ShCcf l ( M n t i

YX;iccpi atiplifia' aitl pcwr aiplificr

475 450 425

375 350 325 300 275(THz)

1000 1100

Wavelength (nm)

Fig.l. Layout of the setup (left), seed spectrum (right), calculated gain of BBO in the nonlinear geometry (right), and OPCPA spectrum (right) : the solid curve, the shaded contour and the dashed curve show the seed pulse, the amplified spectra and the calculated gain of a 4-mm type I BBO, respectively. To investigate the phase reproducibility from shot to shot, which is a crucial problem for pulse compression and the stability of the pulse envelope, we have examined it after the second stage using single-shot spectral interference [8,9] between the amplified pulses in the first stage and the second stage. In Fig.2, the upper insets depict the spectra of the reference and the signal spectra (left) and a single-shot interferogram (right). With a standard Fourier analysis [9] applied to each interferogram, we have used an rms of intensity-weighted phase deviation for a quantitative assessment of the phase drift for the j th shot which is ploted in Fig.2 as solid dots. The mean of £• was found to be 0.093 rad for 450 shots. This indicates that, with an

74

appropriate compression system, stable 6-fs 1-TW laser pulses are expected to be obtained. We have showed various amplitude and phase properties of an OPCPA system that supports a 110-THz-wide spectrum around 800 nm and delivers 8-mJ pulses which could deliver a 6-fs 1-TW operation with a suitable compressor.

Single-shot interferogram

0

100

2a)

3(X)

400

Laser shot j Fig.2. Summary of phase drift characterization by spectral interferometry. See text for details.

References

4.

6 7.

8. 9.

A. Dubietis, G. Jonusaskas, and A. Piskarskas, Opt. Commiin., 88, 437 (1992). I. N. Ross, P. Matousek, M. Towrie, A. J. Langley, and J. L. Collier, Opt. Commun., 144, 125(1997). L. J. Waxer, V. Bagnoud, I. A. Begishev, M. J. Guardalben. J. Puth, and J. D. Zuegel, Opt. Lett.,2S, 1245(2003). C. P. Hauri, P. Schlup, G. Arisholm, J. Biegert, and U. Keller, Opt. Lett. 29, 1369 (2004). R. Biitkus, R. Danielius, A. Dubietis, and A. Piskarskas, Progress in Chirped Pulse Optical Parametric Amplifiers, in Ultmfast Optics IV, Vienna, Austria, June 29 - July 3, 2003, ed. by F. Krausz, G. Korn, P. B. Corkum, and I. A. Walmsley, Springer, p. 359, (2004) and references therein. T. Sosnowski, P B. Stephens, and T. B. Norris, Opt. Lett., 21, 140 (1996). E. Zeromskis, A. Dubietis, G. Tamosauskas, and A. Piskarskas, Opt. Commun., 203, 435 (2002). M. Takeda, H. Ina, and S. Kobayashi, J. Opt. Soc. Am.,12, 156 (1982). D. N. Fittinghoff, J. L. Bowie, J. N. Sweetser, R. T. Jennings, M. A. Krumbugel, K. W. DeLong, R. Trebino, and \. A. Walmsley, Opt. Lett.. 21, 884 (1996).

75

Mid-infrared femtosecond pulse generation by optical parametric amplification under broadband QPM condition Satoshi Ashihara, Manabu Ikeda, Tsutomu Shimura and Kazuo Kuroda Institute of Industrial Science, University of Tokyo 4-6-1 Komaba, Meguro-ku, Tokyo, 153-8505 Japan E-mail: [email protected] Abstract. We have generated femtosecond pulses of 700-nm bandwidths in a 3-4 [im spectral range by an optical parametric amplifier based on periodically-poled LiNb03. Numerical study indicates the possibility for generating -25-fs pulses by pre-chirp compensation.

1.

Introduction

Ultrashort optical pulses in the mid-infrared (MIR) range are useful for measuring and controlling the ultrafast dynamics of molecular vibrations. Recently, several studies have been made on generation and shaping of MIR pulses [1-4], but fiirther development on MIR source that can generate pulses of broader bandv^idth with wider wavelength tunability would be of particular importance. Optical parametric amplification (OPA) based on periodically-poled lithium niobate (PPLN) is promising for these ends, because the PPLN has large nonlinear coefficient, a possibility for engineering, and a MIR transparency up to 5.5 jiim. In this study, we have generated MIR pulses of as broad as 700 nm spectral bandwidths in a 3-4 |Lim spectral range by using the PPLN-based OP A. We have also conducted numerical calculations and investigated the possibility for obtaining shorter MIR pulses.

2.

PPLN-based OPA for MIR

Basic idea for increasing the bandwidth is to use the broadband quasi-phasematching (QPM) condition, which appears around the idler wavelength of 3600 nm in PPLN, pumped at 800 nm. Figure 1 shows the QPM period of PPLN-OPA for several pump wavelengths, calculated with the Sellmeier formula for congruent LiNbOs [5]. We can see that the broadband QPM condition varies in the idler wavelength range of 3-4 |Lim, while we tune the pumping wavelength within the tuning range of Ti:sapphire laser. Therefore, one can construct a broadband MIROPA system with wide wavelength tunability by using the multi-period PPLN devices (e.g. fanned-gratings) and the Ti:sapphire laser. The broadband QPM condition corresponds to the group-velocity matching between the signal and the idler pulses. Here we note that broadband QPM conditions under non-collinear geometry have been previously utilized in nanosecond regime [6].

76

28

1 i Congruent LiNbOg

26 h-

Temp.: 180 °C

24

-

22

-

.

1

" ^ --

Pump: 900 nm

,'j

~~'

1

850 nm

-

-•••••" 8 0 0 n m

_

Q_

i 20 J!{ l\f '\ o 18

16

V .

^^"^"""^ '

j1\

^^.--^'''^750 nm

1

1

I

2

3

4

Wavelength [nm] Fig. 1. Calculated QPM period of PPLN-OPA for different pump wavelengths. We have fabricated the PPLN device of 0.5-mm thickness and 21.2 jum reversal period by the electron beam lithography and the electric field poling (~21kV/mm) techniques. Then we constructed the double-stage OPA system. Pumping source is a regeneratively amplified Ti:sapphire laser pulse of 120 fs duration, 800 nm wavelength, 700 JLJ pulse energy, and 1 kHz repetition rate. The white light generated fi'om 1-mm-thick sapphire plate is used as a seed light and is preamplified in 2-mm-long PPLN device and then power-amplified in the 1-mm-long device. The temperatures of PPLN devices are controlled to be 180-260^C.

3.

Results and Discussion

Figure 2 shows the MIR spectra measured at 230°C. Three curves correspond to different time-delays between the pre-amplified signal and the power-amplifier pulse. Each spectrum has the bandwidth of -700 nm with the center wavelength of 3400, 3600, 3900 nm, respectively. This result confirms that the broadband QPM scheme is usefiil in the femtosecond regime. The pulse duration, measured by the intensity autocorrelator with 1-mm thick AgGaS2, changed around 65-120 fs depending on the temperature and the time delays. Figure 3 shows the intensity autocorrelation trace of the pulse with the center wavelength of 3400 nm. The pulse duration was ~66 fs in FWHM, which is broader than the transform-limited pulse of 25 fs, estimated fi'om the measured spectra. We numerically studied the pre-chirp dependences of the OPA processes and investigated the possibility for obtaining shorter MIR pulses. In fact the chirp of the seed-pulse can be adjusted by using the negatively chirped mirrors in experiments. Figure 3 shows (a) the temporal intensity and phase profiles of the generated MIR pulses without any chirp compensation of the seed-pulse, and (b) those with appropriate pre-chirp compensation. Here we introduced the groupdelay dispersion (GDD) of-3480 fs^ in the seed-pulse at a wavelength of 1020 nm. The result indicates that -25 fs pulses can be generated with appropriate prechirp compensation.

77

1.0

1 l^/"\ 1 / \ 3400 nm / ^ ^ 3600 nm ^ \ \ 3900 nm / , ' \ ^

0.8

(a)

// ^A\ \'' '

0.6

1 /

// / \ \ / ' ^** \ *^\ // /j' FWHM \\ \ \

0.4h 0.2

7" /

0,0 " ^ ^ s i : ^ ^ - ! 2500 3000

-700 nm

V

\

1

\

< -

>

3500

4000

4500

400 Time [fs]

Wavelength [nm]

Fig. 2. (a) Measured spectra for different time delays between pre-amplified signal pulse and the power-amplification pump pulse, (b) Measured SHG autocorrelation trace. 3

1.0 GDD: 0 fs Intensity Phase

0.8

1.0

/

2

0)

0 c^ -100 fs

(a)

0.0'— -400

-200

0 Time [fs]

200

-2 —'-3 400

GDD: -3480 fs^ Intensity Phase

-

2

^ 0 . 6 | --

^»

0

CO

Q)

-1 ^ 0.2

Aj

0.8

1 ?

SO.6

3

1

0.21 0.0 -400

Q3

1 -25 fs

I 0.4 h

! -200

-1

(b) " U 1 0 Time [fs]

w ^

1 200

^

-2

-3 400

Fig. 3. (a) Calculated temporal intensity and phase profiles of the idler pulses without any chirp compensation and (b) those with appropriate chirp compensation of the seed-pulse.

4.

Conclusions

In conclusion, we have generated femtosecond MIR pulses by utilizing the PPLNbased OPA under broadband QPM condition. The pulses with - 7 0 0 nm bandwidth and 65-120 fs duration in a 3-4 [xm spectral range were successfiilly generated. Numerical simulations showed that that it is possible to generate - 2 5 fs pulses by introducing pre-chirp compensation in the seed-pulse. Generation of such ultrashort pulses and their arbitrary shaping would be an important topic.

References 1 2 3 4 5 6

78

V. Petrov, F. Rotermund, and F. Noack, J. Opt. A: Pure Appl.Opt., 3, Rl (2001). F. Eichemeyer, R. A. Kaindel, M. Woemer, T. Elsaesser, and A. M. Weiner, Opt. Lett., 25, 1472 (2000). T. Witte, K. L. Kompa, and M. Motzkus, Appl. Phys. B, 76, 467 (2003). H. - S . Tan, E. Schreiber, and W. S. Warren, Opt. Lett. 27, 439 (2002). G. J. Edwards and M. Lawrence, Opt. Quantum Electron., 16, 373 (1984). C. -W. Hsu and C. -C. Yang, Opt. Lett., 26, 1412 (2001).

Achromatic second harmonic generation: tunable ultraviolet pulses with sub-10 fs duration p. Baum, S. Lochbrunner, and E. Riedle LS fiir BioMolekulare Optik, LMU Miinchen, Oettingenstr. 67, 80538 Mtinchen, Germany Abstract: Tunable ultraviolet (UV) pulses shorter than 10 fs are generated by achromatic frequency doubling the output of a noncollinear optical parametric amplifier. We achieve first and second order achromatic phase matching with a suitable two-prism sequence and thereby increase the natural bandwidth of the nonlinear crystal by a factor of 80. Extremely broad UV spectra with a Fourier limit of 2.9 fs are generated in a 360 |im thick BBO crystal at a conversion efficiency of 20 %. The angular dispersion and the first order chirp of the highly stable UV pulses is compensated with a second prism sequence. By controlling the higher order chirp with a deformable mirror we generate pulses as short as 7.1 fs.

1. Achromatic phase matching Tunable femtosecond pulses in the ultraviolet are essential for many experiments on ultrafast physical and chemical processes. Since a direct source of ultrashort UV pulses is not available, visible or near infrared pulses have to be transferred to the UV by suitable nonlinear processes. For UV pulses in the 5 fs regime the simultaneous phase matching of all spectral components v^ould require a frequency doubling crystal thiimer than 5 jim, precluding an efficient energy conversion. In this contribution we present a scheme to overcome this limitation [1]. It has been proposed that a broadband light pulse is angularly dispersed such that each frequency component propagates in the nonlinear crystal with its individual phase matching angle (achromatic phase matching) [2]. Several schemes were demonstrated to increase the efficiency [3,4] and to avoid adjusting the SHG crystal angle [5,6]. However, the ultimate potential of achromatic phase matching is to frequency double an extremely broadband spectrum at once in order to generate the shortest possible second harmonic pulses. Figure 1 shows an overview of the concept [6]. A broadband pulse is dispersed by a prism and the angular dispersion is converted to a lateral dispersion by a second prism. The resulting collimated beam of spatially displaced frequency components is focused to a common spot in a nonlinear crystal for efficient second harmonic generation. The first order of achromatic phase matching can be adjusted by the prism separation and the focal length. Furthermore, the second order of the angular dispersion at the crystal has the correct sign and magnitude and can be A ^^^ J^^^'^^fM^i^ ^^.r^j/U^^S!l!i:^^iiSff^ W^i^x^i •^^^k^S^T^*-^Tr

w

^^S- ^' Achromatic second harmonic generation scheme. Solid beam, low frequency components; dashed beam, high frequency components.

79

adjusted by choosing suitable prisms. For fused silica prisms and a focal length of 35 mm the residual phase mismatch in a 360-|Lim BBO crystal in the range from 520 to beyond 1000 nm is well within the main peak (±7i) of the sinc^ SHG efficiency function. To compensate for the spatial dispersion of the UV, a second two-prism sequence brings together all spectral components and yields a collimated output beam.

2. Experimental setup As a source for extremely broadband visible pulses we use a noncollinearly phase matched optical parametric amplifier (NOPA) [7]. The bandwidth of the NOPA output is strongly increased to about 200 THz by stretching the internal blue pump pulses in a 170 mm fused silica block. The resulting NOPA pulses cover most of the visible range (see Fig. 2b)) without a sophisticated precompression of the seed light. The pulses are dispersed by two fused silica Brewster prisms to a lateral width of about 25 mm. The polarization of the NOPA output is rotated by an achromatic half-wave plate to be perpendicular to the spectral distribution in order to fulfill the achromatic phase matching condition for type-I SHG. The spectrum is focused with an off-axis parabolic mirror (f = 35 mm) to a common spot in a 360-jLim thick type-I BBO crystal (Fig. 2a). With proper orientation of the crystal a spatially and spectrally broad UV spectrum is generated (Fig. 2c). Nearly the complete spectral width is transferred to the second harmonic. The spectral width of the UV spectrum exceeds 250 THz and its Fourier limit is 2.9 fs. Even without perfect compression, these extremely broad UV pulses are already an ideal source for broadband UV detection in pump-probe experiments. The frequency doubling efficiency at the crystal is about 20 % and the UV pulses have an energy of 250 nJ directly after the crystal and about 120 nJ after compression. The UV spectrum is recollimated with an off-axis parabolic mirror (f = 35 mm) and the spectral components are recombined by two calcium fluoride (CaF2) prisms cut for Brewster's angle. In addition the beam passes

|CPA f >

500

550

600

650

700

750

(nm)

250

275

300

325

350

375

(nm)

• ! ZAP-SPIDER

Fig. 2. (a) Experimental setup: X/2, achromatic half-wave plate; P1-P2, fused silica prisms; GAP, off-axis parabolic mirrors; BBO, frequency doubling crystal; P3-P5, calcium fluoride prisms; DM, deformable mirror, (b) NOPA spectrum and (c) ultraviolet spectrum generated by achromatic frequency doubling.

80

through a 4-f setup with a deformable mirror (DM) in the Fourier plane to adjust the higher order chirp [1]. The final UV beam propagates without noticeable angular dispersion or change in diameter over a distance of several meters. No lateral variation of the spectrum is found experimentally.

3. Characterization and pulse compression The presented scheme forms a prism compressor with a conversion of the visible spectral components to the UV in the middle. The overall compression can be varied without changing the achromatic behavior by scaling the prism separation together with the focal length of the off axis parabolic mirrors. The setup is designed to compensate for the initial chirp of the NOPA pulses and therefore the pulses are still slightly chirped at the point of second harmonic generation. This is advantageous to prevent sum frequency generation between separate parts of the spectrum. We fully characterize the temporal structure of the UV pulses with the zeroadditional-phase variant of SPIDER (ZAP-SPIDER) [8]. For the demonstration of tunable UV pulses the spectral width of the UV was restricted to a Fourier limit of about 8 to 10 fs by narrowing the NOPA spectrum. Highly stable (~ 3 % rms) UV pulses are generated in a wide tuning range from 275 to 335 nm (see Fig. 3a). Figure 3b) shows a spectrum around 290 nm with a Fourier limit of 6.4 fs and the corresponding pulse shape after adaptive compression with the deformable mirror (Fig. 3c)). The measured pulse duration of 7.1 fs is within 15% of the Fourier limit. To our knowledge these are the (^) -1/1 «rv^ 15 nm ._ «. 14 nm ^ ^ 17 nm 21 nm shortest tunable UV pulses re9.8 fs, ported so far.

260 - I — » - —I—"—1—»—r-i

r(b) 6.4 fs M -• • \^^J 1 1 1 „ L _ J 1 1260 280 300 (nm)

7.1 fs

-40-20 0

Fig. 3. (a) Spectra of tunable ultraviolet pulses with their widths and Fourier limits, (b) Spectrum and (c) temporal shape of a 7.1 fs pulse at 290 nm.

20 40 (fs)

References 1 2 3 4 5

p. Baum, S. Lochbrunner, and E. Riedle, Opt. Lett. 29, 1686 (2004). V. D. Volosov and E. V. Goryachkina, Sov. J. Quantum Electron. 6, 854 (1976). T. R. Zhang, H. R. Choo, and M. C. Downer, Appl. Opt. 29, 3927 (1990). T. Kanai, X. Zhou, T. Sekikawa, and S. Watanabe, Opt. Lett. 28, 1484 (2003). B. A. Richman, S. E. Bisson, R. Trebino, E. Sidick, and A. Jacobson, Opt. Lett. 23, 497 (1998). 6 R. A. Cheville, M. T. Reiten, and N. J. Halas, Opt. Lett. 17, 1343 (1992). 7 P. Baum, S. Lochbrunner, L. Gallmann, G. Steinmeyer, U. Keller, and E. Riedle, Appl. Phys. B 74 [Suppl.], S219 (2002). 8 P. Baum, S. Lochbrunner, and E. Riedle, Opt. Lett. 29, 210 (2004).

81

Ultrabroad-band noncollinear optical parametric amplification in some new nonlinear optical crystals Pathik Kumbhakar^'^, and Takayoshi Kobayashi^ ^ Department of Physics, Graduate School of Sciences, University of Tokyo, 7-3-1 Kongo, Bunkyo-ku, Tokyo 113-0033, Japan E-mail: [email protected] ^ Present address: Department of Physics, National Listitute of Technology-Durgapur (Deemed University), Durgapur-713209, India E-mail: [email protected] Abstract. Some new^ nonlinear optical crystals have been found to be suitable for generation of tunable visible ultrafast laser radiation by type-I noncollinear OPA. Moreover, the condition for ultrabroad-band parametric amplification has been derived straightforwardly.

1.

Introduction

Recently, generation of a-few optical cycle ultrashort visible and near-infrared optical pulses has become a routine technique by type-I noncollinear OPA (NOPA) in BBO crystal [1-3]. Growths of several new nonlinear optical (NLO) borate-group crystals, such as CLBO (CsLiBgOio), KABO (K2AI2B2O7), and LB4 (Li2B407) have also been reported recently. We have demonstrated for the first time the potentialities of these crystals for the generation of visible-near-infrared ultrashort laser radiation by type-I NOPA pumped by 395nm radiation [4,5]. The numerically estimated parametric bandwidths of type-I NOPA in CLBO, KABO, LB4, and BBO crystals are - 200, 184, 217, and 190 nm, respectively, for 1mm thick crystal, with the pump-seed noncollinear angle a= 3.0°, 3.4°, 2.9°, and 3.7°, respectively. A simple and straightforward method is also presented for the derivation of the condition of GV matching between the signal and the idler radiations of type-I NOPA [5].

2.

Condition for group-velocity matching and parametric bandwidths of type-I NOPA in BBO, CLBO, KABO, and LB4 crystals

A straightforward derivation of the condition required for achieving broad-band signal wavelength insensitive phase-matching in a type-I NOPA in a negative uniaxial crystal is presented below.

82

By solving the conditions of type-I noncollinear phase-matching in a negative uniaxial crystal, the following expression of 0 can be obtained. 6 = cos'

^[{(v/r^^)'-i}/{(v/r)'-i}

(1)

Where V = 27rWp7>^, V = ^nn^"'^. K = 2nn,VX,, k^ = Inn^lX,, ^"^^V[(OM^O^+2^sVcosi//], y/=a + p, p=sm\{Klkt)smal and a C ^ i s the noncollinear angle between the pump and signal (idler). The superscript (o/e) corresponds to the polarization. Now, in case of monochromatic pump ^in Eq. (1) will not be dependent (to the first order) on the frequency of die signal {co^ of NOP A, if dO/dco^ = 0. After some algebraic simplifications the condition becomes (cosi///vs - l/vi)/cos>^= 0, [5]. This is equivalent to. (2)

Vs = Vi cos (//,

where Vs and Vi are the group velocities of signal and idler, respectively. Equation (2) is called as the condition for GV matching between signal and idler of NOPA [2]. It is important for us to mention here that by taking the first derivative of the wave-vector mismatch {Ak) with respect to the signal frequency the above condition, Eq. (2) had been derived by others [2,6]. 1.0 1

^xxr

/ 0.8 1 i-i^ 11 V'r' ^ •

^

0.6H

\ r/

LB4 BBO - • • CLBO KABO

0.4 J I;/'

o

1 \'\ /;/ /;/'

0.2 11

\\ -A

"Lf*/

0.0 1

500

600

1•

1

700

'

—

1

800

Signal wavelength (nm) Fig. 1. Acceptable power spectrum of the type-I NOPA in LB4, BBO, CLBO, and KABO crystals with a =1.9°, 3.7°, 3.0°, and 3.4°, respectively.

For broadband amplification in LB4, CLBO, and KABO crystals, the optimum values of the pump-seed noncollinear angle (^opt) are 2.9°, 3.0°, and 3.4°, respectively [4,5]. The parametric gain of NOPA is proportional to sinc^ {LkLIT), the acceptable power spectrum [6], where M=k^(6)-k^zid%a-kiZQ^p, is the wavevector mismatch parallel to the pump (^^^), and L is the optical interaction length in the crystal. From Fig. 1 it is observed that the parametric bandwidths (FWHM) of NOPA in the considered noncollinear configuration in 1-mm thick BBO,

83

KABO, CLBO, and LB4 crystals are ~ 190, 184, 200, and 217 nm, respectively. The gain bandwidth of NOPA being inversely proportional to the effective inverse GV mismatch (GVMs.j) [2], the smaller value of GVMs.i is desirable. The calculation resuhs show that the maximum value of \GVMs.i\ is lower for CLBO and LB4 crystals than those of others in the wavelength region of consideration. 4.

Conclusions

We have found that in newly developed KABO, CLBO and LB4 crystals generation of few-cycle visible (around 630nm) laser radiation is possible by Ti:sapphire second harmonic (395nm) pumped type-I NOPA [4,5]. The numerically estimated bandwidth (~ 217nm) of type-I NOPA in 1-mm LB4 crystal is the widest. However, considering that the parametric gain (G) of the NOPA is proportional to cf^^ /p Z^, the damage threshold intensity of LB4 is - 3 times higher than that of BBO and d^a is ~ 1/10-th that of BBO; with the highest pump intensity (/p) the maximum value of G of an LB4 NOPA should be - 3/10^ times smaller than that of BBO, if the crystal thickness (Z) is the same. Taking a thicker crystal, say, 4.5-mm it may be possible to compensate the low parametric gain, however with reduced amplification bandwidth of-134 nm with L= 4.5-mm and a = 2.95°. The present results may bring interest to the researchers in this field to develop ultrafast laser radiation source using some of these new crystals having advantageous properties, in particular for easy growth and for handling in practical experiments. Acknowledgements. We are grateful to the Ministry of Education, Science and Culture, Government of Japan for the financial support (Grant No. 14002003).

References 1.

2.

3. 4.

5.

6.

84

F. Hache, M. Cavallari, and G. M. Gale, "Ultrafast visible optical parametric oscillators: a route to tunable sub-10-femtosecond pulse," Ultrafast Phenomena X, P. F. Barbara, J. G. Fujimoto, W. H. Knox, and W. Zinth, eds., Vol. 62 of Springer series in Chemical Physics (Berlin, Heidelberg, 1996), pp. 33. A. Shirakawa and T. Kobayashi, "Noncollinear phase- and group-velocity matching of optical parametric amplifier for ultrashort pulse generation," lEICE Trans. Electron. E81-C, 246-253 (1998). G. Cerullo and S. De Silvestri, "Ultrafast optical parametric amplifiers," Rev. Sci. Instrum. 74, 1-18 (2003). A. Baltuska and T. Kobayashi, "Adapting shaping of two-cycle visible pulses using a flexible mirror," Appl. Phys. B. Lasers Opt. B 75, 427- 443 (2002). P. Kumbhakar and T. Kobayashi, "Ultrabroad-band phase matching in two recently grown nonlinear optical crystals for the generation of tunable ultrafast laser radiation by type-I noncollinear optical parametric amplification," J. Appl. Phys. 94, 1329-1338 (2003). P. Kumbhakar and T. Kobayashi, "Nonlinear optical properties of Li2B407 (LB4) crystal for the generation of tunable ultra-fast laser radiation by optical parametric amplification," Appl. Phys. B. Lasers Opt. B 78, 165-170 (2004). Y. Nabekawa and K. Midorikawa, "Broadband sum frequency mixing using noncollinear angularly dispersed geometry for indirect phase control of sub-20-femtosecond UV pulses," Opt. Exp. 11, 324-338 (2003).

Design of Multilayer Mirrors for the Reflection of Sub-Femtosecond Pulses in the XUV Spectral Region Alexander S. Pirozhkov^ Hiroyuki Daido^ Sergei V. Bulanov^ Eugene N. Ragozin^ ^ Advanced Photon Research Center, Japan Atomic Energy Research Institute, 8-1 Umemidai, Kizu-cho, Soraku-gun, Kyoto 619-0215, Japan. E-mail: [email protected] ^ Division of Optics, P. N. Lebedev Physical histitute of the Russian Academy of Sciences, 53 Leninsky prospekt, 119991 Moscow, Russia Abstract. Theoretically demonstrated that aperiodic multilayer mirrors can compress negatively and positively chirped XUV pulses. In particular, temporal profiles of reflected trains of positively chirped high-order harmonics of a laser radiation were calculated.

Introduction. The recent progress in the generation and characterization of subfemtosecond pulses in the XUV spectral region [1,2] calls for the development of suitable optical elements. Among the most efficient XUV optical elements are multilayer mirrors. Up to now, only periodic multilayer mirrors were used for the reflection of sub-fs pulses. The shortest pulse generated in [1] had the duration of 0.25 fs which was almost limited by the bandwidth of the multilayer used (9 eV). This means that further decrease in the pulse duration is impossible without a broadband aperiodic multilayer mirrors. The first broadband aperiodic XUV multilayer mirrors for the 12.5-25 nm spectral range were successfully synthesized and used as focusing elements of the imaging XUV spectrograph in [3,4]. Evidently, for the reflection of ultrashort pulses an increased attention should be paid to dispersive (or phase) properties of multilayer structures. We extended the method of calculation of aperiodic multilayer mirrors for the reflection of ultrashort pulses and designed mirrors for the compression of chirped pulses and the simultaneous reflection of several high-order harmonics. Calculation of temporal profiles of reflected pulses. The inverse problem of XUV multilayer optics - that is, the problem of calculation of an aperiodic multilayer structure that gives an extremum of a prescribed merit function - was first stated in [5]. Later this problem was solved with different methods by many authors (e.g. [6-10]). We used a specially designed genetic algorithm for the optimization of aperiodic multilayer structures. Optimization parameters were thicknesses of all layers. In contrast to previous papers, the merit function was taken in the form ofl^r, where /max is the peak intensity of the reflected pulse, T = |/(Od///max is the effective duration of the reflected pulse, I{t) is the intensity envelope. FWHM durations of pulses presented here are somewhat shorter. In calculations, we assumed that multilayer structures consist of two alternating layers with ideal interfaces. We used the method of recurrent relations for the

85

calculation of the complex reflectivity amplitude of structures [11]. Optical constants of materials were taken from [12]. Compression of chirped XUV pulses. Intense chirped XUV pulses can be generated in the process of reflection and focusing of a source laser pulse by a flying plasma mirror [13]. The chirp of the XUV pulse can be controlled by the chirp of the source laser. Another possibility is to use a plasma with a density gradient as the speed of the flying plasma mirror depends on the density. Due to a small curvature radius of the flying mirror, the focal spot is situated inside the plasma region. The multilayer mirror can refocus the pulse to the desired point. hi the calculation, the incident pulse had a Gaussian form with the duration To = 2.5 fs and the chirp parameter b - -6.5 fs"^. The chirp was negative, so the leading part of the pulse had a higher frequency. This was consistent with the natural trend of a deeper penetration (hence, a longer delay) of the higherfrequency radiation into the mirror structure. The results of the calculation are shown in Fig. 1. The aperiodic mirror performed a 13-fold temporal compression with the energy reflectivity of 0.17. The peak intensity of the reflected pulse was more than twice the intensity of the incident pulse. ,

2

n 60

65

70

75

80 85 E,eV

90

95 100

I

,

5

.

.

.

.

J .

.

.

.

,

.

:

i

1

~

t"-

1

V"'

f-

|-J

i

i I

1.5 ~ 1, rel. u • 1 0.5

•

-4 !

^n . . . . t

"

i 1

1 ff

01 . 1 1r ^ , t,fs

1

" r

: -

.'VNX :::,r-r J

Fig. 1. Mirror for the compression of chirped pulses (Mo/Si, 80 monolayers, the incidence angle 5°). Left: the mirror reflectivity (thick line), the spectral phase (dashed line), the spectrum of the chirped incident pulse {hcoo - 86 eV, ro = 2.5 fs, b = -6.5 fs'^, thin line). Right: intensity envelopes of the incident (dashed line) and the reflected (solid line) pulses. T = 0.19 fs, the energy reflectivity 0.17

Simultaneous reflection of several high-order harmonics. It was recently shown that high-order harmonics of the laser radiation were not completely phased [2]. For harmonics generated by a TiiSapphire laser in Ar gas the emission time difference between two neighboring harmonics was measured to be A/ - te(q+2) 4(^) = 33 as. It is important to note that higher harmonics were emitted after lower ones. This means the positive chirp of the harmonics train, which is not consistent with the above-mentioned natural trend. To calculate mirrors for the simultaneous reflection of several harmonics, we set the incident train of pulses in the form Eo(t) = 2 ^ ^ cos(0)qt - OJ^At /4O)L) , where coq-qcoi^, so that the group delay was tg = cOgAt/(2coi) = qAt/2. We calculated a mirror for the reflection of odd harmonics from 25 to 69. Harmonic amplitudes a^ decreased exponentially with order from 1 to 0.1 (intensities from 1

86

to 0.01). The results of the calculation are shown in Fig. 2. The mirror performed a two-fold compression of pulses in the train. 0.5 I

....

0.4 0.3

az 0.1

A

i

30

40

:

;

4_

\

/

i

|rk

^ \

\

60

70 E,eV

80

: 20 15

\

Im K.n iIl.lM.lf AMt'.A II 50

r 25

:

J

i

90

, rel. u.

^ , rad 10

100

Fig. 2. Mirror for the simultaneous reflection of 25-to-69-order odd harmonics with an emission time difference of 33 as (Mo/Si, 80 monolayers, the incidence angle 5°). Left: the mirror reflectivity (thick line), the spectral phase (dashed line), the spectrum of the incident pulse (thin line). Right: intensity envelopes of the incident (thin line) and the reflected pulses (thick lines, scaled (x5) for convenience), TO = 0.27 fs, r = 0.13 fs, the energy reflectivity 0.11

Conclusion. We made a theoretical investigation of aperiodic multilayer mirrors for the reflection of ultrashort XUV pulses. We designed aperiodic structures for the 13-fold compression of negatively chirped XUV pulses and for the reflection of several high-order harmonics with different emission times. The use of aperiodic multilayer structures is indispensable for the further decrease in the pulse duration and the increase in the focused pulse intensity in the XUV spectral region. Acknowledgements. The work was supported by the Japan Society for the Promotion of Science (JSPS P-03543).

References 1 2 3 4 5 6 7 8 9 10 11 12

R. Kienberger et al. Nature 427, 817, 2004. Y. Mairesse et aL, Science 302, 1540, 2003. V. G. Kapralov et aL, Quantum Electron 32, 149, 2002. V. E. Levashov et aL, Flas. Phys. Reports 30,149,2004. J. F. Meekins et aL, Appl. Opt. 25,2757, 1986; ibid 26, 990, 1987. K. D. Joensen, Proc. SPIE 3113, 500, 1997. E. Ziegler et aL, Proc. SPIE 3737, 386, 1999. N. N. Kolachevsky et aL, Quantum Electron. 30, 428 2000. I. L. Beigman, A. S. Pirozhkov, E. N. Ragozin, JETP Lett. 74, 149,2001. I. L. Beigman et aL, J. Opt. A: Pure Appl. Opt, 4,433,2002. A. V. Vinogradov, ed., X-Ray Mirror Optics, Leningrad: Mashinostroenie, 1989. R. Soufli and E. M. Gullikson, Proc. SPIE 3113,222, 1997, http://cindv.Ibl. gov/optical constants/. 13 S. V. Bulanov, T. Esirkepov, T. Tajima, Phys. Rev. Lett. 91, 085001, 2003. 87

Route to design electric fields of optical pulses: A combination of a pulse shaper and a carrierenvelope-phase stabilized chirped-pulse amplifier system Masayuki Kakehata^ Hideyuki Takada\ Yohei Kobayashi^ and Kenji Torizuka^ Kazuki Nishijima^, Hiroaki Takamiya^, Tetsuya Homma^, and Hideo Takahashi^ ^ National Institute of Advanced Industrial Science and Technology (AIST), Tsukuba Central 2, 1-1-1 Umezono, Tsukuba 305-8568, Japan E-mail: [email protected] ^ Shibaura Institute of Technology, 3-9-14 Shibaura, Minato-ku, Tokyo 108-8548, Japan Abstract. We demonstrated active carrier-envelope-phase (CEP)-shift by utilizing a 4f-pulse-shaper in a CEP-stabilized chirped-pulse-amplification system. A combination of a pulse-shaper and CEP-stabilized pulses enables us to design electric fields, both the pulse envelope and the carrier-envelope relation.

1. Introduction The carrier-envelope phase (CEP) is the relative phase between the peak of the pulse envelope and the electric field; it is important in controlling the high-order harmonic generation and measuring attosecond phenomena. CEP-stabilized chirped-pulse amplification (CPA) systems are demonstrated [1, 2]. To control the CEP for pump-probe experiments, an active CEP shifter giving no effects on the pulse envelope and delay is desirable. We demonstrate an active CEP-shifting device employing a 4f-pulse shaper. A combination of the pulse shaper and the CEP-stabilized system will allow us to design the shape of the electric field (both the pulse shape and the carrier-envelope phase) of optical pulses.

2. Experiments Performance of a pulse shaper as a CEP shifter Figure 1 (a) depicts the spectral interferometry setup to test the CEP shifter. We measured the spectral interference (SI) between the reference pulse and the pulse passing through the pulse shaper. The CEP of the Ti:sapphire oscillator was not stabilized. The pulse shaper is composed of a liquid-crystal spatial light modulator (SLM) and the 4-f optical configuration. First, we measured the SI to obtain the phase difference ^i (a))=^rb,o (o))^ef (o) with giving a certain phase by SLM. Second, we applied periodic phase modulation generated by SLM and measured the phase difference

88

^ (o, t)f ^rb(o), t)-^ef. (©)• Then we calculated the phase shift ^s(o>) t)=^(o), t)^((o), then fit a linear function to ^sC®? t) in the co-if) domain and derived the time delay and the CEP shift as shown in Fig. 1(b) [3]. Pulse shaper (CEP shifter)

(b)

•f-

^

CEP shift

4

h^co) TO

^

^-2

S l o p e ; Delay shift % ''p^"

1x10 2x10 Angular frequency (rad/s)

3x10

Fig.l (a) Experimental setup to test the CEP shifter. CM: cylindrical mirror (^242mm). CEP of the input pulse {^) is not stabilized (b) Relation between the relative spectral phase and the CEP and the time delay Figure 2 shows the relative CEP-shift and delay shift fi)r (a) without phase modulation, (b) sinusoidal delay shift by a PZT, and (c) smusoidal CEP-shift by the pulse shaper. The delay-shift and the CEP-shift agreed with the applied shift within the experiment accuracy. Without modulation

(b)

(c)

./^Vy-N A

A A A

A

y V V V V Fig.2 Change of the CEP-shift and delay-shift. Measured (a) without modulation, (b)with sinusoidal delay modulated applied by the delay line in the interferometer, and (c) with sinusoidal CEP modulation by the pulse shaper. CEP control of chirped-pulse amplified pulses by the pulse shaper We installed the pulse shaper in a CEP-stabilized CPA system [2] and measured the CEP-shift by the self-referencing f-to-2f SI method [4]. Figure 3 illustrates the experimental setup. The amplified pulse is ImJ/pulse with the spectrum width of 20 nm FWHM, which corresponds to a 50 fs FWHM transform-limited pulse. The exposure time was 21msec, and the repetition rate of the amplifier was 555 Hz. Figure 4(a) was obtained by applying rectangular CEP-modulation at 0.5Hz; (b) was with a triangular CEP-modulation at 2Hz. The applied CEP modulations were clearly observed in the/^ro-2/SI, which confirms that the pulse shaper works as an active CEP shifter. Furthermore, because the pulse shaper can apply a desired relative phase and the desired CEP shift, a combination of a CEP stabilized system and a pulse-shaper can generate a designed electric field, a complicated envelope shape with controllable CEP.

89

: = 80 M H z —n 3 5 fs

T

i C E P shifts r

Pulse

selec

!_:!_^

3L

I H O HOW

(G fa t i n g s ) ! ^ > 220

fibe rj

ps

R e g e n . a m p Synch. I

I

R 9 a dy pjUse

i

|Do'ay|

D rv Id e r i CEO

stabilized

oscillator

^-j

T"

A m p l i f i e r s ta g e

Fig.3. Diagram of a CEP-stabilized, chirped-pulse amplification system with CEP sliifler. (a)

(b)

W a v e l e n g t h (nm )

Fig.4. Self-referencing SI fringe measured with applying CEP modulation by the pulse shaper. (s) Rectangular CEP modulation at 0.5Hz, (b) triangular CEP shift at 2 Hz.

3. Conclusion We demonstrated active carrier-envelope phase shifter using a 4f-pulse shaper. A combination of a pulse shaper and a CEP-stabilized CPA system will allow us to design electric fields of optical pulses. Acknowledgements. A part of this study was financially supported by the Budget for Nuclear Research of the Ministry of Education, Culture, Sports, Science and Technology, based on screening and counseling by the Atomic Energy Conamission, Japan.

References 1 A. Baltuska, Th. Udem, M. Ulberacker, M. Hentschel, E. Goullelmakis, Ch. Gohle, R. Holzwarth, V. S. Yakovlev, A. Scrinzi, T. W. HSnsch and F. Krausz, Nature. 421, 611-615 (2003). 2 M. Kakehata, H. Takada, Y. Kobayashi, K. Torizuka, H. Takamiya, K. Nishijima, T. Homma, H. Takahashi, K. Okubo, S. Nakamura, Y. Koyamada, Optics Express 12,2070-2080 (2004). 3 AW. Albrecht, J. D. Hybl, S. M. Gallagher Faeder, D. M. Jonas, J. Chem. Phys. 111,10934-10956(1999). 4 M. Kakehata, H. Takada, Y. Kobayashi, K, Torizuka, Y. Fujihira, T. Homma, and H. Takahashi, Opt. Lett. 26, 1436-1438 (2001).

90

Towards electric field reconstruction using coherent transients in a two-level system A. Monmayrant, B. Chatel and B. Girard Lab. Collisions Agregats Reactivite, CNRS UMR 5589, IRSAMC, UPS, 118 route de Narbonne. 31062 Toulouse, France E-mail: [email protected]

Abstract: Interaction between a two-level system and a weak chirped pulse leads to oscillations of the excited state amplitude, named "coherent transients". Their extreme sensitivity to the pulse shape provides a tool for electric field measurements.

1. Introduction One-photon transition in the weak field regime results in a linear interaction for which the final state populations can be entirely deduced from the power spectrum. However, the phase of the wave function is sensitive to the electric field. This can have important consequences for applications where a subsequent excitation is performed, in particular when coherent superpositions are involved. The transient evolution of excited state population is also strongly dependent on the detailed laser shape. As an intuitive illustration of this statement, the transient response to a non-resonant excitation follows the electric field pulse envelope, independently of its spectrum. For instance, simply changing the pulse duration will change this transient response. A resonant interaction leads to radically different behaviors. FT limited pulses produce a step-wise excitation in the weak-field, and Rabi oscillations in the intermediate and strong field regime. Chirped pulses produce a total population inversion in the strong field with a final state robust with respect to small variations of laser parameters. Chirped pulses in the weak field lead to Coherent Transients (CT). The laser frequency sweeps linearly with time and crosses the resonance. Most of the population transfer occurs at resonance. The small fraction of excited state amplitude transferred after resonance leads to strong oscillations due to interferences between the atomic dipole and the exciting field. On the other hand, interaction before resonance results in negligible effects [1]. Similarly, interferences between the field radiated by the atom and the incoming field leads to interferences which can be used to a partial analysis of the field [2]. The shape of Coherent Transients can be radically changed by shaping the pulse [3]. The high sensitivity of CT to slight modifications of the laser pulse [4] opens the possibility of characterization of the laser pulse itself. In a simple approach, if the general shape of the laser pulse is known and only few parameters need to be determined, one can use a simple adjustment of these parameters to fit the experimental curve with the predicted one. However, one would like to establish a general method able to determine any pulse shape. Using the CT needs to have a dominant quadratic spectral phase. It can be added to the pulse if

91

necessary. One difficulty is that only the part of the pulse after resonance leads to oscillations which can be used to determine its shape. Another difficulty is that the measured quantity is related to the excited state probability whereas the probability amplitude, proportional to the integral of the laser electric field, is not directly measured. However it is possible to overcome these difficulties by combining several CT measurements.

2. Experimental Set-up Atomic rubidium is used as a benchmark system. The 5s - 5p (P1/2) transition (at 795 nm) is resonantly excited by a sequence of a Fourier limited and a chirped pulse produced by a Regen amplifier. The transient excited state population is probed "in real time" on the (5p - ns, n'd) transitions with an ultrashort pulse (at 607 nm, 30 fs, Non-collinear Optical Parametric Amplifier (Nc-OFA)) compressed with chirped mirrors. The pump pulse sequence is produced by a programmable pulse-shaping device, recombined with the probe pulse and sent into a sealed rubidium cell. All the experiments are performed in the perturbative regime. The pump-probe signal is detected by monitoring the fluorescence at 420 nm due to the radiative cascade (ns,n'd) -> 6p -> 5s collected by a photomultiplier as a function of the pump-probe delay. The pulse shaping device is a 4f set-up composed of one pair each of reflective gratings and cylindrical mirrors. Its active elements -two 640 pixels liquid crystal mask- are installed in the common focal plane of both mirrors [5]. This allows for phase and amplitude shaping.

3. Results In order to observe oscillations on the whole pulse duration, the excited state must be initially populated so that interferences are present from the beginning until the end of the interaction between the chirped pulse and the atom. This is achieved with a sequence of two pulses with a well defined phase relationship. The high resolution of the pulse shaper allows generating these two pulses by applying a complex transmission H^^cd) in the spectral domain: H^{co) = \ + Qx^[i[d + (p\co-co^) + (l)\(o-co^y

/2)\

The first pulse is short (FT limited) and the second one is strongly chirped with f = 2.10'fs' in order to exhibit the CT, and delayed by / = 4ps. The first part of the second pulse (before resonance) produces interferences with the population excited by the first pulse as can be observed in Fig. 1. This scheme provides interferences on the whole duration of the second pulse. Two measurements are performed for ^ = 0 and 6 = njl. The combination of these two measurements allows one determining in-phase and in-quadrature contributions from the second pulse, so that the excited state probability amplitude produced by the second pulse can be deduced from this set of measurements. For a second pulse weaker than the first one, this inversion is linear with respect to the second pulse. For any relative intensities (between the first and second pulses), the inversion procedure is

92

nonlinear but still possible. The result of this inversion is shown in Fig. 2 which displays in the complex plane the reconstructed excited state probability amplitude a^{t) resulting from the second pulse.

Delay (ps)

Fig. 1: CT resulting from a FT limited pulse followed by a chirped pulse. In the second measurement, an extra phase shift of TT/l is applied to the second pulse. Solid line: theory, dots: experiment.

-0.2-0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 Re(a,(t)) (au)

Fig. 2: Reconstructed probability amplitude ^e{^) deduced from the combination of the two measurements presented in Fig. 1. The Cornu spiral appears clearly. Top: theory; Bottom: experimental result

References S. Zamith etaL, Phys. Rev. Lett. 87, 033001 (2001). J. E. Rothenberg and D. Grischkowsky, J. Opt. Soc. Am. B 2, 626 (1985); J. E. Rothenberg and D. Grischkowsky, J. Opt. Soc. Am. B 3, 1235 (1986); J. E. Rothenberg, IEEE J. Quant. Electronics QE-22,174 (1986). J. Degert etal, Phys. Rev. Lett. 89, 203003 (2002). W. Wohlleben etal, Appl. Phys. B 79, accepted (2004). A. Monmayrant and B. Chatel, Rev. Sci. Instr., accepted (2004).

93

Spatiotemporal determination of the absolute phase of few-cycle laser pulses Fabrizio Lindner\ Michael Schatzel\ Gerhard Paulus^'^' , Herbert Walther\ Andrius Baltuska^' , Eleftherios Goulielmakis^'*, Matthias Lezius ' , and Ferenc Krausz ' ^ Max-Planck-Institut fiir Quantenoptik, Hans-Kopfermann-Str. 1, 85748 Garching, Germany ^ Department of Physics, Texas A & M University, TX 77843-4242, College Station, Texas, USA ^ Sektion Physik, Ludwig-Maximilians-Universitat Munchen, Am Coulombwall 1, 85748 Garching, Germany "* Institut fiir Photonik, Technische Universitat Wien, Gusshausstr. 27, 1040 Wien, Austria ^ Institut fiir lonenphysik, Universitat Innsbruck, TechnikerstraBe 25, 6020 Innsbruck, Austria Abstract. We determined the carrier-envelope ("absolute") phase of linearly polarized fewcycle laser pulses by measuring the energy-resolved asymmetry of electron emission from noble gases. The Gouy phase shift in the laser focus is also measured, providing the first full spatiotemporal depiction of the electric field in the whole focal region.

Atomic processes induced by ultrashort laser pulses have recently attracted a lot of attention, in particular owing to their potential of generating isolated attosecond light pulses [1,2]. For these exciting prospects, few-cycle laser pulses, i.e. pulses consisting of merely a few electromagnetic oscillations, are necessary. The electric field of such laser pulses can be written as E(t) = e^' Eo(t) ' cos{cot -\-cp)

(1)

where e^ denotes the axis of polarization, EQ(t) the envelope of the pulse, co the carrier angular frequency of the laser, and cp the phase between the carrier and the envelope, often called carrier-envelope (CE) or, more directly, "absolute" phase. With respect to the latter, the convention in (1) implies that cp = 0 corresponds to a "cosine-like" pulse with the absolute maximum of the electric field pointing to the positive direction. Accordingly, cp = ± nil correspond to "sine-like" pulses. The absolute phase cp plays a major role in any atomic process driven by few-cycle laser pulses [3,4]; however, complete control over it was demonstrated only very recently [4-6]. Due to the influence of the absolute phase on the laser's electric field shape, nonlinear photoionization is a possible approach to determine its value. In our experiments noble gas atoms (typically Xe) were ionized with peak intensities of nearly 10^"^ W/cm^ by linearly polarized few-cycle pulses [6]. A detailed description of the experimental setup can be found in [5,7]. In short, electron

94

spectra originating from a well-confined focal area are simultaneously recorded in the two opposite polarization directions ("stereo-ATI") by means of time-of-flight spectroscopy. The energy-resolved left-right asymmetry can be used to retrieve the phase value. The phase-stabilized laser system [6] delivers pulses characterized by identical electric-field waveforms. Therefore, pulses with known phase differences Acp can be obtained by simply delaying the envelope with respect to the carrier in front of the experiment. Glass dispersion is an immediate way of achieving this: At a central wavelength of 760 nm adding 52 //m of fused silica change cp by 2jt without affecting the pulse duration distinctly. Thus, two glass wedges, which can be shifted with respect to each other, allow adjusting any phase. The electron spectra detected for different CE phases significantly differ in many aspects, the most striking being the high-energy part (ATI-plateau). The latter is shown in Fig. 1. for several absolute phases, separately for the detector located on the left and on the right side of the laser focus. Both semiclassical and quantum mechanical models [8] yield the basic result that the most energetic photoelectrons are mainly emitted in the direction of the peak of the electric field. Thus, the clear phase dependence of the cutoff position gives a direct read-out of the electric-field strength in the respective direction, i.e. it allows absolute phase determination. One can verify that Acp = n corresponds, as expected, to a change from left to right while Aq) = 2n exactly reproduces the spectra. Note that these results remove the jt ambiguity intrinsically connected with high-harmonic generation experiments [4]. An important issue of nonlinear processes driven with few-cycle laser pulses is that they typically take place in a laser focus. It is known that an electromagnetic beam propagating through a focus experience an additional JT phase shift with respect to a plane wave [9]. The effect of this so-called Gouy phase anomaly on the spatial dependence of the absolute phase is of critical importance for experiments taking place over an extended area of the focus, e.g. high-order harmonic generation and attosecond pulse generation. RIGHT

LEFT

10^

10^

0

Jt

absolute phase [rad]

2jt

0

Jt

2jt

MO^

absolute phase [rad]

Fig. 1. Electron count rate in the high-energy part of the ATI spectra as a function of electron energy and absolute phase for the two detectors

95

In order to analyze in detail the phase variation within the focal range, we acquired several electron spectra corresponding to different positions along the beam propagation direction. Two thin slits perpendicular to the beam axis and to the polarization axis allow selecting electrons originating from a well confined focal area. Then, the absolute phase value can be determined at each position by inspecting the corresponding left-right asymmetry in the recorded spectra. Figure 2. shows the retrieved absolute phase as a function of the propagation distance in the laser focus [7]. For comparison, the Gouy phase of a Gaussian beam with the nominal focusing parameters of our experiments is also shown. Note that the absolute phase shift is not expected to follow precisely the Gouy phase, the latter being a property of cw lasers. The pulses undergo the Ji phase shift within a few Rayleigh distances and, what is particularly important for experiments, the phase does not exhibit any wiggles or irregularities. These results constitute the first full characterization of few-cycle optical pulses in space and time, an essential step for any application of such laser systems.

-

^

CD Q.

B-' \ -2

0

^

: 2

propagation distance [mm] Fig. 2. Measured carrier-envelope phase variation as a function of the propagation distance in the focus due to the Gouy effect [5]. The solid line is the Gouy phase shift of a cw Gaussian beam, shown for comparison

Acknowledgements. This work has been supported by the Austrian Science Fund (Grants No. F016, No. Z63, and No. P15382) and by The Welch Foundation (Grant No. A-1562).

References 1 2 3 4 5 6

M. Hentschel et al, Nature 414, 509, 2001. R. Kienberger et al., Nature 427, 817, 2004. G. Paulus et al., Nature 414, 182, 2001. A. Baltuska et al.. Nature 421, 611, 2003. G. Paulus et al., Phys. Rev. Lett. 91, 253004, 2003. A. Baltuska et al., IEEE Journal of Selected Topics in Quantum Electronics 9, 972, 2003. 7 F. Lindner et al., Phys. Rev. Lett. 92, 113001, 2004. 8 D. Milosevic et al.. Optics Express 11, 1418, 2003. 9 L. G. Gouy, C. R. Acad. Sci. Paris 110, 1251, 1890.

96

Spatial chirp and pulse-front tilt in ultrashort laser pulses and their measurement Selcuk Akturk^ Xun Gu^ Erik Zeek* and Rick Trebino^ ^ School of Physics, Georgia Institute of Technology, Atlanta, Georgia 30332-0430, USA E-mail: [email protected] Abstract. We show that two physically different definitions of spatial chirp exist. We also show that pulse-front-tilt arises, not onlyfromangular dispersion, but alsofromspatial and temporal chirp. We verify these results using GRENOUILLE.

1.

Introduction

Spatio-temporal distortions are common in ultrafast optics. One such distortion is angular dispersion (AD), which is often introduced by using a dispersive element such as a prism or grating. Propagation of an angularly dispersed beam will induce spatial chirp (SC). Angular dispersion also causes pulse-front tilt (PFT) and indeed it has been previously shovm that PFT and AD are equivalent [1]. In this work, we show that AD and PFT are not equivalent, and we provide an additional source of PFT, in which no AD occurs. We also show that there are actually two different SC definitions. We then use our simple version of frequency-resolved optical gating (FROG), GRENOUILLE, to measure, not only the pulse intensity and phase, but also the spatial chirp and pulse-front tilt. Consider an initially transform-limited, but spatially chirped beam—^with no AD—passing through a dispersive medium (Fig. 1). Due to group-velocity dispersion in the medium, the redder side of the beam emerges from the medium earlier than the bluer side, resulting in PFT in the output beam. Because no angular dispersion occurs, this violates the well-known AD/PFT duality. Spatially chirped pulse with pulse-front tilt, but no angular dispersion Vg(red)>Vg(blue)

Fig. 1. Two sources of pulse-front tilt. Left: The well-known angular dispersion. Right: The combination of spatial and temporal chirp, which causes PFT without AD. In order to understand this effect, it is important to also understand spatial chirp, which we show can take two different forms. One involves writing the spatial chirp in terms of the variation of the beam center position (XQ) with

97

frequency a)\ ^ = 6XQIACO which we will call the spatial dispersion. Another involves writing the spatial chirp in terms of the variation of the pulse center frequency {COQ) with position x: V^ACOQI&K, which we call XhQ frequency gradient, A Gaussian beam with spatial dispersion ^ can be written: E (jc, (D) = E^ {(D) exp f- {x - C(^)^ 1^^ 1 (1) A beam v^th frequency gradient v, on the other hand, is of the form: \E{X,CO)\=f{x)\E,{a)-vx)\

(2)

If we also assume a Gaussian pulse in time or frequency, it is easy to find: E, {m) = E, exp(-fi>VoV4),

u = C/{C' + ^ \ ' A)

(3)

The two SC parameters do not correlate monotonically, and they have different physical meanings (Fig. 2). Spatial dispersion ^ arises directly from propagation with AD, and it describes globally how far different frequency components separate, while frequency gradient v describes the local variation of the spectrum.

2

3 4 Position [mm]

5

5

10 i:, [nm-fs/rad]

Fig. 2. (left) Experimental spectrum vs. position. The black and blue lines are x^ {X) and XQ{X) plots, respectively. Using the slopes of these plots, ^ is calculated to be 3.9x10^ nm-fs/rad, and v 3.3x10"* rad/nm-fs. With the measured w and TQ, Equation 2 is verified within 6% error, (b) Theoretical plot of ^ and v with the above w and r^. To see understand the role of SC in PFT, we also include angular dispersion {/}) and linear chirp ( ^ ) in the pulse/beam and inverse Fourier transform from the (x,(o) domain into the {xj) domain to obtain: ^ ( ^ , 0 = /Wexp[-(^"-^o)'A']exp[-i^i(^-/o)-i(<*2/2)(^-^o)']

(4)

(l>,^-vx, (l>, ^-(P21 {T,' lA + (p,'), T' ^T,' ^A(p^' IT,' , r,^ =ro'+4CVw^ (5) From (5), we see that SC in the {x,t) domain is characterized by frequency gradient v, /^ = t^ix) = px is the pulse-front arrival time at position x, and p is the total PFT. We find that the total pulse-front tilt,/?, is the sum of two terms: ;7 = dro/dA: = ; ? ^ + i?,hup'where p^-^kp,

98

p,^^=v(p^

(6)

The first term is due entirely to AD and the second is due to the combination of spatial and temporal chirp. While the first term is well known, the second is not. Nevertheless, the effect described by the second term is a very common effect. To verify our results, we used the GRENOUILLE device, which yields the pulse temporal/spectral intensity and phase [3], and SC and PFT as well [4,5]. SC will cause afi-equencysweep across the measured FROG trace, and PFT will induce a shift in the trace relative delay center. In our experiment, a matched prism pair introduced SC in the beam without residual AD. We measured the spatial dispersion to be 8.87x10"^ nm-fs. By translating one of the prisms, we varied the temporal chirp and observed changing shifts in the GRENOUILLE trace, i.e., different amounts of PFT. Fig. 3 (left) is a plot of the measured PFT vs. GDD. Its slope yields the frequency gradient, which is then converted to spatial dispersion using Eq. (2), resulting in a value of 8.78X 10"^ nm-fs, in excellent agreement with the measured spatial dispersion. This confirms the new (temporal and spatial) chirp-induced PFT effect.

—I 1 1 1 -1000 -800 -600 -4p0 Group-Delay Dispersion cp^ [fs /radj

0 50 100 Delay [Ts]

•100-50 0 50 100 Delay [fs]

Fig. 4. Left: Theory (line) and measurements (plus) of the GRENOUILLE-measured PFT vs. GDD for a pulse with a constant SC; Right: GRENOUILLE traces with different GDD (i.e., different temporal chirps) but the same SC. In conclusion, we have shown that two different types of spatial chirp exist, and we have found a new contribution to pulse-front tilt, arising from the combination of spatial and temporal chirps, which we have measured using GRENOUILLE.

References 1 Z. Bor, B. Racz, G. Szabo, M. Hilbert, and H. A. Hazim, "Femtosecond pulse front tilt caused by angular dispersion," Optical Engineering 32,2501-2504 (1993). 2 P. O'Shea, M. Kimmel, X. Gu, and R. Trebino, "Highly simplified device for ultrashort-pulse measurement," Opt. Lett. 26, 932-934 (2001). 3 S. Akturk, M. Kimmel, P. O'Shea, and R. Trebino, "Measuring spatial chirp in ultrashort pulses using single-shot Frequency-Resolved Optical Gating," Optics Express 11,68-78 (2003). 4 S. Akturk, M. Kimmel, P. O'Shea, and R. Trebino, "Measuring pulse-front tilt in ultrashort pulses using GRENOUILLE," Optics Express 11, 491-501 (2003).

99

Principal Control Analysis: Gaining Insight from Feedback Learning Algorithms J. L. White, B. J. Pearson, and P. H. Bucksbaum FOCUS Center, Physics Department, University of Michigan, Ann Arbor, MI 48109-1120 E-mail: [email protected] Abstract: Feedback learning algorithms are widely used to search for optical pulse shapes for quantum control. We propose a simple analysis of the pulse shapes to learn about the degrees of freedom in the effective control Hamiltonian. When applied to stimulated Raman scattering in methanol, the technique yields results consistent with a simple model of two-mode SRS in methanol.

1. Introduction We would like to find the optimal path to guide a quantum system from its initial state to some target final state [1,2]. Unfortunately, we often have only incomplete knowledge of the Hamiltonian. Theoretical methods can aid in the search [3,4], and there has also been considerable experimental success [5,6]. When the Hamiltonian is not known, feedback learning algorithms are often used. Feedback learning algorithms allow the physical system to explore its own quantum dynamics through an experimental search [7]. The pulse shapes are selected through a fitness-directed search protocol, as in the genetic algorithm [8]. The fitness is a measured quantity proportional to the projection of the final state onto the target state. The algorithms develop better solutions by combining elements of previous solutions [9]. Unfortunately, simply finding a good solution often provides little physical insight. The optimal pulse shape found by a learning algorithm, while sufficient to achieve control, is often complicated and may contain unnecessary features. In an effort to learn more about the system Hamiltonian, several modifications or extensions of learning feedback have been suggested [10-13]. Here we take a different approach based on an analysis of the ensemble of trial pulse shapes.

2. Principal Control Analysis We show how to find the essential degrees of freedom through covariance analysis of the pulse shapes evaluated during the search. Principal control analysis is the application of covariance techniques to a fitness-directed search [15]. In our learning algorithm, each pulse shape is encoded as a list of some 25 numbers (genes) encoding the phase of different segments of the optical spectrum. Linear combinations of genes contributing to high fitness should appear correlated in the pulse shapes developed by the fitness-driven algorithm. These correlated

100

linear combinations will correspond to important features of the control field that direct the quantum dynamics under investigation We begin the principal control analysis by calculating the covariance matrix for the ensemble of pulse shapes evaluated during the search. Diagonalizing the covariance matrix yields its eigenvectors uj and eigenvalues. The eigenvectors are uncorrelated control directions, while the eigenvalues are the variances along these directions. Next we represent each pulse shape as its projections rjj onto the eigenvectors wy, and calculate the correlations of the pulse shape fitnesses with the projections rjj. The eigenvectors that correlate most strongly with the fitness are called the principal controls, and are the most important control directions. Other eigenvectors can be ignored without losing substantial control. By projecting the trial solutions onto the principal controls, we reduce the dimension of the search space. Projecting the optimal pulse shape onto the principal controls produces the essential pulse. The essential pulse contains traits necessary to achieve the target, with minimal extraneous features.

3. Application of Principal Control Analysis to Liquid Methanol

3 4 Frequency [THz]

Fig. 1. Top: Magnitude of the Fourier transform of I(t) for optimal (dashed) and essential (solid) pulses for the symmetric mode. Bottom: Similar for the antisymmetric mode. We have applied this analysis to a feedback control experiment in methanol. The experiment has been described previously [16]. An intense shaped 800 nm ultrafast laser pulse is focused into a cell containing methanol. Either one of two Raman-active vibrational modes can be selectively excited by phase shaping of the pump pulse. A learning search was conducted for each mode.

101

The analysis for this problem began with a single covariance matrix for all the pulse shapes evaluated in the two learning searches. Two principal controls were selected for each mode, and the essential pulse shape for each mode calculated. Recent work demonstrated a control mechanism for SRS in methanol based on periodicity in I(t) [17]. The Fourier transform of/(r) reveals the important Raman coupling frequencies [18]. The Fourier transforms of I(t) for the optimal and essential pulses are shown in Fig. 1. Principal control analysis enhanced the magnitude of the Fourier transform of I(t) in the region around 3 THz, consistent with the model. Moreover, the phase difference between the Fourier transforms of the two essential pulses was 7r/2 in the region around 3 THz, again consistent with the model.

4. Conclusion We have shown how principal control analysis of a feedback learning algorithm can be used to learn about the number and character of the degrees of freedom responsible for control, and reveal essential features of the effective Hamiltonian. The technique should be applicable whenever learning algorithms have been used to study a system where the principal degrees of freedom can be described by linear combinations of the controls. Acknowledgements. This work was supported by NSF grant 9987916.

References 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

102

D. Tannor and S. Rice, Adv. Chem. Phys. 70, 441 (1988). P. Brumer and M. Shapiro, Ann. Rev. Phys. Chem. 43, 257 (1992). D. Tannor, R. Kosloff, and S. Rice, J. Chem. Phys. 85, 5805 (1986). A. Peirce, M. Dahleh, and H. Rabitz, Phys. Rev. A 37, 4950 (1988). A. Shnitman, et al., Phys. Rev. Lett. 76, 2886 (1996). L. Zhu, et al., Science 270, 77 (1995). R. Judson and H. Rabitz, Phys. Rev. Letters 68, 1500 (1992). J. Holland, Scientific American 267, 66 (1992). L. Davis, ed., Handbook of Genetic Algorithms (Van Norstrand Reinhold, New York, 1991). J. Geremia, E. Weiss, and H. Rabitz, Chemical Physics 267, 209 (2001). A. Mitra and H. Rabitz, Physical Review A 67, 33407 (2003). C. Daniel, et al., Science 299, 536 (2003). J. Geremia and H. Rabitz, Phys. Rev. Lett. 89, 263902 (2002). 1. Jolliffe, Principal Component Analysis (Springer Verlag, 2002), 2nd ed. J. L. White, B. J. Pearson, and P. H. Bucksbaum, lanl.arXiv.org (2004), quantph/0401018. B. J. Pearson, et al., Phys. Rev. A 63, 063412/1 (2001). B. J. Pearson and P.H. Bucksbaum, Phys. Rev. Lett. 92, 243003 (2004). D. Meshulach and Y. Silberberg, Natme 396, 239 (1998).

Population-split Genetic Algorithm for phase retrieval of ultrafast laser pulses Ching-Wei Chen \ Su-Frang Shu^, Chao-Kuei Lee^, and Ci-Ling Pan ^ ^ Institute of Electro-Optical Engineering, National Chiao Tung University, 1001, Ta-Hsueh R4 Hsinchu, Taiwan 300, R.O,C. E-mail: [email protected] ^ Department of Electronic Engineering, Ching Yun University, Taoyuan, Taiwan 320, R,O.C. E-mail: [email protected] ^ Institute of Electro-Optical Engineering, National Sun Yat-sen University, Kaohsiung, Taiwan 804, R.O.C. E-mail: [email protected] Abstract. We report a new population split genetic algorithm for phase retrieval of femtosecond pulses from int©rferometric autocorrelation traces. After 200 GA generations, pulse width, phase, and chirp could be determined with a convergence error of 6x 10"^.

1. Introduction The knowledge of both pulse amplitude and phase are required w^hen generating and utilizing the ultrafast light pulses. Several methods for acquiring this information have been developed and proposed [ M ] . Recently, v^e report a new population split genetic algorithm (PSGA) for retrieving the ultrafast laser parameters from FROG measurements [5]. In this work, PSGA combined with linear and second-order interferometric autocorrelation measurement is demonstrated to fiilly characterize the information of femtosecond pulses.

2, Phase Retrieval: Theory and Experiment The experiment was carried out on a femtosecond Kerr-lens-modelocked Ti: sapphire laser. The laser output is fed to a coUinear interferometric autocorrelator using GaAsP and Si photodiodes as the two-photon and one-photon detector, respectively. After the Fourier transform of the one/two photon autocorrelation traces, the spectrum

\E((&)\^ ,

the power spectrum |/(<»)r and the second-harmonic

spectrum \uim)l are obtained. The amplitude and phase of the pulses are then retrieved by the following steps shown in Fig, 1.

103

Jwf

Error Function

,4;' M

r ^ ,

r E^(p>) * ,

f +'^^

t

^»^(«>)

/^(o?)

4 »FT ^^(0

•L

.f'^T

/^.(O

t4,<0

;'t, M)f

(^«f ^(f

jLi

Fig» 1. Flow chart of the phase retrieval algorithm

3. Results and Discussions The retrieved amplitude and phase of the typical output of the sub-25fs laser are shown in Fig, 2. The convergence error by PSGA is about 6x 10'^. Numerically, the convergence error can be as low as lO'^-'lO"^ when retrieving some simulation data. We also compare this technique with thePICASO [3, 4]. This is shown in Fig. 3.

1

I

-2i

WVVic! — J

-150

/VAA^

T V - — | - . - . '-r--—J

-100

1

-50

1

r-1—1

0

60

1

f

100

1

1

150

ttme (fs)

Fig» 2. Amplitude and phase retrieval by PSGA of the output of a sub-25fs Ti:sapphire laser. Note that the pulse width retrieved by PICASO is somewhat narrower than that by PSGA. This is because that PICASO uses the interferometric cross correlation and spectrum to retrieve the pulse information. The spectrum is the necessary database to the retrieval process. Therefore, if the laser pulse is not transform-limited, the spectrum will be over-estimated. In our approach, the Fourier transform of the

104

linear autocorrelation trace is used. Although the PICASO solves problems of time ambiguity, this can be achieved by this technique in the same way. -PICASO retrieval r-PSGAHitricivai

a

4,0.6-

Tlwe delay (fs)

Fig. 3, The pulse retrieval results comparison between PSGA and PICASO in intensity i tiie phase

4

Conclusions

PSGA combined with linear/second-order interferometric autocorrelation traces is an accurate method for characterizing ultrashort pulses. The retrieval results are reliable and credible. This technique is well suited to a wide range of pulse widths and shapes and is easily implemented both experimentally and numerically. Meanwhile, this approach is believed to be widely applied in many applications including pulse shaping and ultrafast spectroscopy. Acknowledgements. This work was supported in part by the National Science Council of R.O.C. under Grants No, HSC 92-2215-E-009.030 and Program for the Pursuit of Academic Excellence of tlie Ministry of Education, R.O.C.

References L 2. 3. 4. 5.

K. W. Belong, R. Trebino, J. Hunter and W. E. White, X Opt Soa Am. B. 11, 2206 (1994). K Nagannma, K. Mogi, and H. Yamada, IEEE J, Quantum Electron. 25,1225 (1989). J. W. Nicholson, J. Jasapara, W. Rudolph, F. G. Omenetto, and A, J. Taylor, Opt Lett 24,1774(1999), J. W. Nicholson and W. Rudolph, J. Opt Soc, Am. B. 19,300 (2002). S. F. Shu, C. L Pan, and C. T. Sun, International Journal of Neural, Parallel d Scientific Computations. 11,207 (2003).

105

Eight-Frame Observation of Propagation Behavior of 0.49-mJ, 45-fs Optical Pulses Generated by a 1-kHz Laser System Masatoshi Fujimoto^'^, Shin-ichiro Aoshima^'^, and Yutaka Tsuchiya^'^ ^ Central Research Laboratory, Hamamatsu Photonics K.K., 5000 Hirakuchi, Hamakitacity 434-8601, Japan E-mail: fuj [email protected]. co.jp ^ CREATE Shizuoka of JST, REFOST, 5000 Hirakuchi, Hamakita-city 434-0041, Japan Abstract. We succeeded in observing instantaneous profiles of 0.49-mJ, 45-fs pulses propagating in an atmosphere. A laser system with 1-kHz repetition enabled us to measure them with a sufficient signal-to-noise ratio.

1.

Introduction

Recently, we have studied various propagation behaviors of femtosecond optical pulses via lateral image observation of instantaneous light profiles [1,2]. The observation method, which is called femtosecond time-resolved optical polarigraphy (FTOP), utilizes the birefringence induced by the electric field of an intense laser pulse to be observed, and captures instantaneous intensity distributions of the light pulse with the help of a femtosecond temporal window of an optical probe pulse. The previous observations were based on an experiment using a 10-Hz laser system, and therefore the integration was restricted to several shots. Hence, it was difficult to measure pulses in gaseous media with a sufficient signal-to-noise ratio when the pulse energ}^ was less than a millijoule [1]. On the other hand, since submillijoule pulses can be generated by more commonly used lasers and are widely applied, it is important to observe a submillijoule pulse in an atmosphere. In this report, we show^ a result of the large-number integration measurement of 0.49-mJ pulses propagating in air; the pulses w^ere produced using a laser system with a 1-kHz repetition. The obtained images will show that FTOP is effective even for submillijoule, femtosecond pulses propagating in gaseous media.

2.

Experimental Results

The experimental setup was similar to that reported before [1,2]. This time w^e used a high-repetition laser light source, i.e., an all-solid-state Ti:sapphire regenerative amplifier system, which generates linearly-polarized optical pulses of 810-nm central wavelength with a 1-kHz repetition rate. The temporal full width at half maximum (FWHM) of a pulse w^as estimated to be 45 fs with the

106

assumption of a Gaussian pulse shape. The output pulse energy of the laser system was -0.6 mJ. On the other hand, the object pulse energy was 0.49 mJ in front of the focusing lens, and therefore the peak power of the object pulse was 10 GW on the assumption of the same pulse shape as above. The cross section of the object pulse has an FWHM diameter of 5.7 mm. We used a n / = 200 mm achromatic lens to focus the object pulse, and observed the object pulse propagation in air around the focal point. Furthermore, instead of the quadruple-pulse generator [2], an octuple-pulse generator was installed in the probe path to convert an incident single pulse into an octuple pulse having definite time intervals. The energy ratio of the successive pulses in the probe light was 0.92:0.89:0.89:0.91:0.91:1.00:0.91:0.89 in temporal order. Since the probe light is composed of an octuple pulse train, the object pulse to be observed meets the pulses in the probe light eight times. As a result, the imaging system placed ^ e r the analyzer captures eight successive profiles, which represent eight instantaneous intensity distributions of the object pulse at eight different temporal points. Hence, an observed image shows the propagation dynamics of the object pulse. The width of the field of view can be adjusted from ~1 mm to ~6 mm by a change in the magnification of the imaging lens in front of the camera. The data shown here were taken under the condition of a 6.4-|Lim spatial resolution. Figure 1(a) shows an example of an eight-frame instantaneous FTOP image; this image was obtained with integration over 2,000 shots, i.e., during 2 s. Plasma, leaked-probe, and dark profiles were subtracted from an observed raw 0.24 FT)m (0,&ps)

(b)

9

P

Fig. 1. Lateral observation of 0.49-mJ, 45-fs object pulses propagating in air while they are being focused by an/= 200 mm lens. The arrow indicates the propagation direction of the object pulses. The time written in parentheses above the horizontal scale bar indicates how long light takes to travel through the scale-bar length, (a)-(d) Eight-frame observations of instantaneous profiles of object light every 0.4 ps; the different panels correspond to different observations under the same conditions. The brightness is nonnalized by its maximum in these four panels, (e) Profile of the plasma produced by object light. The brightness is normalized by its maximum in the panel.

107

image, where the plasma profile originates from a light emission of the plasma induced in air by the intense object pulse. The object pulse propagates in air toward the focal point of the / = 200 mm lens from left to right. The inter\^als between the peaks of the adjacent pulses in the probe light were fixed at 0.4 ps. As a resuh, the successive profiles of the object light were captured every 0.4 ps. Since the brightness is proportional to the probe pulse energy, the respective FTOP profiles have weights that correspond to the energ>^ ratio of the pulses in the probe light. From the image, we can see directly that the pulsed light eventually becomes focused. Nevertheless, the intensity of the pulse, i.e., the brightness of the profile in Fig. 1(a), gradually decreases with the object pulse propagation; this fact implies that before reaching the focal region, the object pulse loses a part of its energ}^ because of multiphoton absorption and/or ionization-induced refraction. For comparison, the plasma profile is also shown in Fig. 1(e), where the background is subtracted. Other observations under the same experimental conditions are shown in Figs. l(b)-(d). These profiles seem to be different from one another. (The difference can clearly be recognized by a false-color illustration.) Since the profiles were integrated over 2,000 shots, the shot-to-shot fluctuation of the laser light source is sufficiently canceled out. Hence, the difference of the profiles is probably due to atmospheric turbulence [3]. Here, we confirmed that the plasma profile was hardly changed according to the measurement.

3.

Conclusion

In conclusion, we succeeded in obser\dng instantaneous intensity distributions of 0.49-mJ, 45-fs (i.e., 10-GW peak power) optical pulses propagating in air by integrating multiple shots. Since the probe light was composed of an octuple pulse train, eight-frame images were captured at a stretch; the time resolutions of the respective frames were of a femtosecond scale, and the time intervals between adjacent frames were of a subpicosecond scale. The obtained eight-frame FTOP image showed not only the pulse Qncrgy reduction due to a laser-induced plasma, but also the propagation-profile variations due to the turbulence of the atmosphere. The image data clearly show that FTOP can be utilized for real-time monitoring of propagation of femtosecond optical pulses generated by a regenerative amplifier system. Acknowledgements. The authors would like to thank T. Hiruma, Y. Suzuki, and Professor S. Nakai for their encouragement.

References 1 For example, M. Fujimoto, S. Aoshima, and Y. Tsuchiya, Meas. Sci. Teclinol. 13, 1698,2002. 2 M. Fujimoto, S. Aoshima, and Y. Tsuchiya, Opt. Lett. 27, 309, 2002. 3 V. P. Kandidov, O. G. Kosareva, M. P. Tamarov, A. Brodeur, and S. L. Chin, Kvant. Elektron. 29, 73, 1999 [Sov. J. Quantum Electron. 29, 911, 1999].

108

Self-referenced measurement of the complete electric field of ultrashort pulses in time and space Pablo Gabolde, Selcuk Akturk, and Rick Trebino Georgia Institute of Technology, 837 State St NW, Atlanta GA 30332, USA E-mail: [email protected] Abstract. We propose and demonstrate a technique based on Fourier-synthesis digital holography andfrequency-resolvedoptical gating to measure the complete spatiotemporal behavior of potentially arbitrary ultrashort laser pulses.

1.

Introduction

Ultrashort-pulse measurement techniques usually ignore the pulse's spatial dependence. This representation conceals the possible coupling between space and frequency, or space and time, although such couplings are commonly used in pulse compressors, stretchers and shapers, and may subsequently appear as distortions in the pulse. Recently, we showed that frequency-resolved optical gating (FROG) can identify first-order spatiotemporal effects, but in only one spatial direction [1, 2], and occasionally ultrafast techniques, such as spectral interferometry, are extended to one transverse spatial dimension [3]. Currently, however, there does not exist a self-referenced technique that can measure the complete pulse electric field in space and time, E(x,y,z,t). Rather than extending time-domain techniques into the space domain, we will extend space-domain techniques into the time, or rather, frequency, domain. Intensity-and-phase spatial-beam measurements (e.g., holography and wave-front sensors [4]) have achieved great sophistication. Unfortunately, they either require an independent reference pulse [5] or are restricted to transform-limited pulses in time [6]. Here we combine Fourier-synthesis digital holography [7] with FROG to achieve true self-referenced four-dimensional measurements of the field of potentially arbitrary pulses, E(x, y, z, /).

2.

Method

First, because the electric field satisfies the wave equation, we note that it is sufficient to measure E(x,y,t) or Ej^x^y.o)) at a given position z = ZQ along the propagation axis. Then, to obtain E{x,y,co\ we perform a set of holographic measurements at discrete frequencies. A set of narrow band-pass filters is used to isolate the frequency components of the pulse. This yields the essentially monochromatic spatial complex field, E(x,y,a)) for each frequency cok, whose amplitude and phase must be measured. Different methods are available, but off-

109

axis digital holography [8] offers a convenient and inexpensive way to reconstruct the 2-D spatial amplitude and phase of monochromatic light from a single digital interferogram. We use the beam itself as the reference beam, which we spatially filter in order to obtain a spherical wave, whose spatial amplitude and phase are therefore known. The direct algorithm to extract the amplitude and phase of the object beam from the digital hologram is well established [9]. The relative phase of each frequency component is not determined, however, due to the self-referencing. Therefore we perform an additional FROG measurement over a small spatial portion of the beam, but without any band-pass filter, to correctly set the relative phase of each monochromatic field, E(x,y,C0f^), Then we fully reconstruct the complex field in the time domain using an inverse Fourier transform: 1 In

1

(1)

3. Coupling between space and frequency: spatial chirp Perhaps the coupling that occurs most often is the one between space and wavelength, which has been called spatial chirp. If one considers the functions XQ(X) and y^QC) to be the center position of the beam at a given wavelength, then spatial chirp can be described by the variations of JCQ and JO with l. We introduced spatial chirp with a pair of prisms arranged to cancel angular dispersion, then measured E(x,y,co) by the method presented in section 2, before computing xo(co) and yo(co). Our results are presented in Fig. 1. Our spatial chirp measurement (Sx:o/9A = 27|im/nm) agrees with the results obtained from a spatially resolved spectrum. As expected, we only found a significant spatial chirp along the jc-coordinate (parallel to the prisms base).

770

rao

7S0

800

810

Fig. 1. Spatial chirp measured for the same pulse according to two possible representations. Left: central position of the intensity of each spectral component. Right: Spatial profile of the pulse. The vertical axis corresponds to the intensity I(x^) of the pulse, and color to the local average wavelength AO(JCJ^).

110

4. Coupling between space and time: pulse-front tilt Gur technique also easily reveals pulse-front tilt, which arises from the coupling between space and time and must be computed from the intensity \E(x^,t)f in the time domain. Once \E(x,y,t)f has been computed, one can define tihe arrival time, to(x,y), of the maximum of |E(jc,;;,Of at a given position, so that pulse-front tilt along X is just dto/dx. Using our method, we measured the complete spatio-temporal field of pulses with varying amounts of pulse-front tilt (see Fig. 2).

X[tmil

Fig. 2. Pulses with minimal (left) and significant (right) pulse-front tilt. Arrival time to(xy) is plotted against the vertical axis, and color represents the instantaneous wavelength Ao(;cj;).

References 1.

2. 3.

4. 5.

6. 7.

8. 9.

S. Akturk, M. Kimmel, P. O'Shea, and R. Trebino, "Measuring spatial chirp in ultrashort pulses using single-shot Frequency-Resolved Optical Gating," Optics Express 11, 68-78 (2003). S. Akturk, M. Kimmel, P. O'Shea, and R. Trebino, "Measuring pulse-front tilt in ultrashort pulses using GRENOUILLE," Optics Express 11,491-501 (2003). L. Gallmann, G. Steinmeyer, D. H. Sutter, T. Rupp, C. laconis, I. A. Walmsley, and U. Keller, "Spatially resolved amplitude and phase characterization of femtosecond optical pulses," Optics Letters 26,96-98 (2001). B. C. Piatt and R. Shack, "History and Principles of Shack-Hartmann Wavefront Sensing," Journal of Refractive Surgery 17, S573-S577 (2001). T. Tanabe, H. Tanabe, Y. Teramura, and F. Karmari, "Spatiotemporal measurements based on spatial spectral interferometry for ultrashort optical pulses shaped by a Fourier pulse shaper," Journal of the Optical Society of America B 19,2795-2802 (2002). E. Arons, D. Dilworth, M. Shih, and P. C. Sun, "Use of Fourier synthesis holography to image through inhomogeneities," Optics Letters 18,1852-1854 (1993). E. Leith, C. Chen, Y. Chen, D. Dilworth, J. Lopez, J. Rudd, P. C. Sun, J. Valdmanis, and G. Vossler, "Imaging through scattering media with holography," Journal of the Optical Society of America A 9,1148-1153 (1992). S. Grilli, P. Ferraro, S. De Nicola, A. Finizo, G. Pierattini, and R. Meucci, "Whole optical wavefields reconstruction by Digital Holography," Optics Express 9,294-302 (2001). M. Takeda, H. Ina, and S. Kobayashi, "Fourier-transform method of fringe-pattem analysis for computer-based topography and interferometry," Journal of the Optical Society of America 72,156-160 (1982).

111

Pulse-measurement challenges at 1.5 microns: several-cycle pulses and several-element devices Selcuk Akturk^ Mark Kimmel^ and Rick Trebino\ Sergey Naumov^ Evgeni Sorokin^ and Irina T. Sorokina ^ School of Physics, Georgia Institute of Technology, Atlanta, Georgia 30332-0430, USA E-mail: [email protected] ^ Institut fur Photonik, TU Wien, Gusshausstr. 27/387, A-1040 Wien, Austria E-mail: [email protected] Abstract. We have demonstrated frequency-resolved optical gating (FROG) for measuring several-cycle 1.5-micron pulses using an angle-dithered-crystal geometry. We have also demonstrated experimentally a very simple GRENOUILLE device for measuring fewhundred-fs 1.5-micron pulses using the nonlinear-optical crystal Proustite.

1.

Introduction

Extremely broadband several-optical-cycle pulses near telecommunication wavelengths (-1.5 jLtm), are in high demand for numerous applications [1]. Mode locked fiber lasers can also generate moderately intense pulses (-lOOfs) useful for nonlinear optical applications besides telecommunications. Reliably generating and utilizing such NIR pulses requires reliable measurement techniques. In this work, we describe two different frequency-resolvedoptical-gating (FROG) [2] devices. The first utilizes an angle-dithered rather thick crystal, which avoids the phase-matching bandwidth problems, to measure several-cycle pulses at 1.5 |im. The second is an ultrasimple GRENOUILLE device, which uses a thick crystal's small phase-matching bandwidth to spectrally resolve the pulse. This is a challenge at 1.5-|Lun wavelengths, where dispersion is minimal. However, we have used a more dispersive nonlinear crystal, Proustite, which nicely yields pulse measurements from 100 fs to a few ps.

2. Several cycle NIR pulse measurements With FROG, it is possible to measure pulses over a wide range of wavelengths, pulse lengths, and complexities. However, FROG has not yet been used for several-cycle IR pulses. Thus, in this work, we engineer a FROG device, which combines several recent innovations, for measuring such pulses. All pulse-measurement techniques require the use of a nonlinear-optical process, phase-matched over the pulse bandwidth, requiring very thin crystals for broadband pulses. However, we have recently shown that angle-dithering a SHG crystal yields a significantly increased effective phase-matching bandwidth [3]. Because the SHG efficiency scales as the square of the crystal thickness, angle-

112

dithering also yields significantly greater signal strength. While angle-dithering avoids the phase-matching requirement, the SHG crystal cannot, however, be arbitrarily thick due to group-velocity dispersion (GVD). We find that 1-mm LiNbOg and LiI03 crystals have negligible GVD for several-optical-cycle pulses. We used a KLM Cr'^'^iYAG laser yielding pulses with - 110 nm of bandwidth near 1.55 juim [1] with 50 mW of average power at a 100 MHz repetition rate. In our FROG, we used a 1-mm LilOg crystal, mounted on a resonant scanner, oscillating at 30 Hz with an amplitude of 10°. The pulses were split and combined using a 177°-apex-angle Fresnel biprism, which automatically splits the pulse into two and ensures their temporal and spatial overlap in the crystal. A 10-mm cylindrical mirror yielded a line focus along the delay axis. The resulting SHG signal was imaged onto the slit of a spectrometer, the output of which was recorded by a CCD camera and the intensity and phase retrieved from the resulting FROG traces using the Femtosoft SHG FROG code. Figure 1 shows the measured and retrieved FROG traces, as well as the retrieved and independently measured spectra, all of which are in very good agreement with each other. Temporal Intensity and Phase

^

1-0

1-0.8 ^

0.8-

r-^-^ -

0.6-

Temporal / ^ Intensity / \ Temporal / \ Phase / \

--5.4 --5.6 -5.8 -6.0 "6.2 -6.4

1-°^ f 0.4 0 . I 0.2~

0.0

j/: -40

-"V 0

Tl zx a> Q.

40

Time [fs] Spectral Intensity and Phase

1300

1400

1500

1600

1700

Wavelength [nm]

Fig.l: FROG measurements of Cr4+:YAG laser. Retrieved pulse width is 37.1 fs (FWHM).

3. GRENOUILLE for longer 1.5-micron pulses Fiber lasers in general have much less bandwidth than the C/^:YAG lasers. Thus it should be easier to measure them. Indeed, we have designed a very simple GRENOUILLE [4], an experimentally simpler version of FROG to do so. This involves the use of a thick SHG crystal^—thick enough that its finite phasematching bandwidth is able to resolve the spectrum of the second-harmonic signal. We have found that, for fiber laser pulse measurements, an exotic crystal, "Proustite," matches GRENOUILLE's requirements perfectly. We built a GRENOUILLE using the conventional setup [4]. To expand the beam we used a 5x magnification telescope. The beam then passed through a 75mm focal length cylindrical lens. We used a l^O*^ apex-angle Fresnel-biprism, in order to generate large enough delay range. The signal was mapped onto the cam-

113

era by using a 50-mm cylindrical and 50-mm spherical lens doublet. The trace was recorded by a CCD camera and the intensity and phase retrieved from the resulting FROG traces using the Femtosoft SHG FROG code. In order to test our device, we used a Menlo Systems TC-1550-B fiber laser, operating at 1570 nm with an output power of 20.5 mW (25 MHz rep. rate). Figure 2 shows the measured and retrieved GRENOUILLE traces. For comparison, we also measured the same pulses using a conventional multi-shot FROG and we obtained excellent agreement between the two measurements (Fig.3).

-1QQD

T D 1000 De[ay [fs]

n—'—r 15S0 1500 1EQD Wavelength [nm]

-1D0D

0 1000 Delay [fsl Temporal livtei^sity ^\\<\ Piracy

-T500 Time ps]

Fig.2: GRENOUILLE measurements of fiber laser pulses: the retrieved pulse width is 779 fs and the bandwidth is 8.2 nm (FROG error 0.0055). Multi-shot FROG yielded a 765-fs pulse width and an 8.1-nm bandwidth.

4. Conclusion In conclusion, angle-dithered-crystal FROG is ideal for measuring several-cycle pulses at 1.5 jLim. And GRENOUILLE, using the SHG crystal, Proustite, is ideal for measuring longer pulses at this wavelength.

References 1 S. Naumov, E. Sorokin, V.L. Kalashnikov, G. Tempea, and I.T. Sorokina, "Selfstarting five optical cycle pulse generation in Cr'^':YAG laser," Appl. Phys. B 76, 1 (2003). 2 R. Trebino, Frequency-Resolved Optical Gating: The Measurement of Ultrashort Laser Pulses (Kluwer Academic Publishers, Boston, 2002). 3 P. O'Shea, M. Kimmel, X. Gu, and R. Trebino, "Increased-bandwidth in ultrashort-pulse measurement using an angle-dithered nonlinear-optical crystal," Opt. Express 7, 342-349 (2000). 4 P. O'Shea, M. Kimmel, X. Gu, and R. Trebino, "Highly simplified device for ultrashort-pulse measurement," Opt. Lett. 26, 932-934 (2001).

114

Single-shot phase measurement by spectral phase interferometry using a streak camera Takeshi .4kagawa^ ^, Kazuhiko Misawa^ ^, and Roy Lang^ ^ ^ Department of Applied Physics, Tokyo University of Agriculture and Technology ^ CREST, JST 2-24-16 Naka-cho, Koganei-shi, Tokyo, Japan E-mail: [email protected] Abstract We show a single-shot measurement of the nonlinear phase and amplitude changes given to a particular femtosecond pulse in a modulated pulse train using spectral phase interferometry and a streak camera.

1.

Introduction

Recently, progress in optoelectronics is remarkably rapid in the fields of teleconiniunication and storage. However, tlie present system mainly uses die intensity of light, but not the phase. In order to realize optical information processing in a more flexible way, we proposed a method for storing and processing the phase infonnation on the spectral components of femtosecond pulses [1]. A femtosecond pulse has a broad bandwidth as well as a short duration. It can be regarded as an optical wave packet, which is a hnear superposition of several longitudinal modes from the laser.

E{r, 0 = 2 \E{Q, I exp[i
(1)

k

WTiere E\Qj^) and 0{Qj^) are amplitude and phase of each mode i2^. , respectively. The phase of the spectral comix)nent is locked to each other within a single femtosecond pulse, hence the spectral phase can be used as the wavelengthdivision data set. From a viewpoint of application, single-shot phase measurement to determine individual phase information on each femtosecond pulse is required, because the each pulse should be a random packet in the time-division data stream. In the present paper, we demonstrate a single-shot measurement of the nonlinear* phase and amplitude changes given to a particular femtosecond pulse in a pulse train.

2.

Spectral Phase Interferometry

In order to measure the phase in a single pulse, si:)ectral phase interferometry (SPI) is required [2, 3]. Using SPI, the phase measurement is possible without scanning the delay time. SPI is based on a Michelson interferometer. One arm is die probe

115

pulse to be measured, and the other is the reference pulse. The probe and reference pulses with a delay time T are spectrally dispersed in a sj^ectrometer. The interference fringe between the probe and reference pulses aie measured in the spectral domain. By analyzing interference fringes, difference in the phase and ampUtude of the probe with respect to reference pulse can be obtained. The spectial interference fringe, DyQA. with nonhnear phase change is expressed as A0{Qj^) and nonlinear amplitude change JE\QJ^)

DiQ,) = \EiQ,f

+\E{Q,)+MiQ,f

+ 2\E{QJE(QJ+AE{Q,)}COS[TQ,

^^^ +M^J]-

To isolate nonlinear phase and amphtude changes, the sinusoidal component is extracted from the spectial interference fringe via Fourier filtering. The phase and amphtude difference can be calculated from the aigument and the absolute value of these complex terms. Especially, nonlinear phase and amplitude changes induced by photoexcitation can be obtained by deriving the difference between these values v^itliout and witli excitation.

3.

Single-shot Phase Measurement

We use laser pulses from a Ti: sapphire oscillator at a repetition frequency of 76 MHz as probe and reference pulses, and amplified pulses from a regenerative amplifier at 1 kHz seeded by tlie same oscillator as excitation pulse. The pulse duration and center wavelength is 50 fs and 800 nm, respectively. The phase of a particular single pulse in the pulse train is directiy modulated by photoexcitation of a nonlinear material. The sample as nonlinear optical phase and amplitude modulator is a cyanine dye, DTTC Iodide, which has relaxation time shorter enough than 13 ns repetition rate of the probe pulses. The excitation pulses induce refractive and index absolution change, which result in phase and amplitude changes in the probe pulse passing tliiough the sample just after excitation. The sample is put in the probe aim of the SPI. Arbitraiy delay time is set between the probe and reference pulses, and these pulses are ahgned as collinear as possible to observe the highest visibility of the fringe. The collinear and temporally displaced two pulses are spectrally dispersed in a spectrometer. To measure the spectral interference fringe we use a streak camera. The streak camera can capture a temporal image of the pulses. It's temporal resolution in 50 ps, which is faster than the pulse interval (13 ns) and slower than the pulse dulation (50 fs). As a result, the time-resolved spectral fringes can be captured and die nonlinear phase and amplitude changes of the modulated pulse can be measured. In the present setup, up to 10 GHz (1/100 ps) is possible for the repetition rate of the probe pulse. Figure 1 shows a two-dimensional streak image. The bright and dark interference fringes represent the time-resolved spectral fringes by a single-shot. The multi-channel plate (MCP) in the streak camera has a high gain enough for a single-shot measurement. Only the second pulse from the top in Fig. 1 is modulated by the excitation pulse. Figure 2 shows the nonlinear phase and amplitude changes by analyzing interference fringes extracted from Fig. 1. The thick and thin lines in Fig. 2 show the nonhnear phase and amplitude change.

116

respectively. In comparison, the linear absorption spectrum of DTTCI is shown as the dashed line. Although the laser is tuned on ilie lower edge of the lineai* absorption, the nonlinear amplitude change shows the peak ai'ound the center wavelength of the laser at 800 nm. This imphes transient hole-burning induced by the excitation. The dispersive structure of the nonlinear phase change is also observed.

Wavelength (nm) 700 750 800 850 900 >r

1 ^-i 'jrw

yoduteted pulses

100 ns

CD

c cc JZ

13 ns

o TO

T\

r 1

\ ''

r \\ Aji

\1 Time

Fig. 1. A two-dimensional streak image of modulated pulse train

4.

!

1 — !

^

r i I I[

[

1

( <*< \\

^ CO """"^sf

f

*-f«i!Nc**"

/ "^

o 03

"""^^^w^—^\K^

k^-'(' ^ \ 760 780 800 820 840 860 Wavelength (nm) 1

f:

^fgr•>X•r^
E

<

Fig. 2. The nonlinear phase and amplitude changes. Spectral interferograms with / without excitation are also depicted.

Conclusion

We succeeded in measuring die nonlinear phase and amplitude changes of, the paiticular pulse in a pulse train using time-resolved spectral inteiferometry. Tliis metliod enables us to realize a new tiansmission system by giving information on internal phase of broad spectral components in ultrashort optical pulses.

References K. Misawa, I. Matsuda, N. T. Hashimoto, and R. Lang, in J. Plod. Opt, accepted for publication. C. laconis, and L A. Walmsley, inlEEEJ, Quantum Electron., Vol. 35, 501 1999. L. Lepetit, G. Cheriaux, and M. Joffre, in J. Opt. Soc. Am. B, Vol. 12, 2467, 1995.

117

FROG measured 185 fs pulses generated by down-chirped dispersion-managed breathingmode semiconductor laser Bojan Resan^ Luis Archundia, and Peter I Delfyett College of Optics and Photonics/CREOL & FPCE, Univ. of Central Florida, Orlando, USA Phone: 407 823 6871, E-mail: [email protected], delfi^ett@creoLuc£edu Abstract: Linear down chirp compensation revealed 185 fs pulses measured by SHGFROG generated from a dispersion-managed semiconductor mode-locked laser. Up or down chirping allows broader mode-locked spectra depending on the temporal and spectral semiconductor gain dynamics.

Strong self*phase modulation (SPM) in a semiconductor gain media impresses a nonlinear chirp, limiting the attainable pulse duration from conventional semiconductor mode-locked lasers. A recently developed intracavity dispersion management concept decreases the SPM and, additionally, provides more efficient internal and external cavity pulse amplification [1]. The dispersion-managed laser operates in a breathing mode when the intracavity pulse is considerably stretched while passing through the cavity gain element and subsequently compressed before the saturable absorber. In the up chirping regime of the breathing-mode, the pulse propagating through the gain is stretched by normal group velocity dispersion (GVD) and in down chirping the pulse is stretched by anomalous GVD. SOA spontaneous

RFDC PBS

Fig. L Experimental setup pf (bottom) dispersion-managed breathing-mode semiconductor mode-locked sigma-ring cavity laser and (top) diagnostics. PBS: polarizing beam splitter, P: pellicle beam splitter. The inset graph (top right) isihows the spectra of the SOA spontaneous emission, the SA excitonic absorption band and the mode-locked laser

118

The setup for the laser and diagnostics is illustrated in Fig. 1. The dispersion elements 1 and 2 as well as the external compressor are dual pass grating compressors with internal telescopes, introducing tunable positive or negative GVD [2]. A typical SHG-FROG setup [3] is employed and pulse retrieval is performed with commercial software. The SOAs have a red shifted gain peak with respect to the saturable absorber (SA) excitonic absorption band as shown in the inset graph of Fig. 1. This wavelength situation leads to a stronger amplification of the red spectral part of the pulse. For up-chirped laser output pulses, gain depletion will cause strong amplification of the leading (red) part in the temporal domain. The combination of the temporal and spectral effects tends to amplify only the red portion of the pulse and therefore limits the mode-locked spectral width. For down-chirped pulses, the amplifications in the spectral and temporal domain are balanced. The red part of the pulse is stronger amplified due to the red shifted gain peak, but the blue part is first incident to the SOA and experiences greater gain before the SOA gain depletes. (nomi«l dicp«rsh>n}

-6-5 - 4 - 3 - 2 - 1 0 1 2 3 4 5 6 7 GVD introduced by element 2 (ps/nm)

Fig. 2. Externally compressed pulse autocorrelation and spectral FWHM versus introduced dispersion Fig. 2 displays the externally compressed laser output pulse autocorrelation and spectral FWHM with respect to the introduced GVD by the stretcher for the passively mode-locked laser. The spectra are broader and pulses are compressible to shorter durations for down chirping (left part of the Fig. 2) compared to the up chirping (right part). When there is no sufficient GVD introduced (middle part of the Fig. 2X the laser does not operate in the breathing mode. I.OT

K

,

-1.0-Cy.S-0.6-0,4-0.2 0,0 0,2 0.4 0,6 0.8 1.0

Time (ps)

1.0T

825

830

835 840 845 Wavelength (nm)

850

Fig. 3. (a) The comparison of the measured and calculated bandwidth limited pulse autocorrelation, (b) The measured spectrum with SHG-FROG retrieved spectral phase

119

The measured hybridly mode-locked laser output after external cavity amplification and compression is presented in Fig. 3. The broadened spectrum of 9 nm is obtained by use of down chirping compared to 3 nm with up chirping. 415^

f 417

I *^^ ^ 420] 421

-600 "400 "200 0 200 400 600 TJnne (fe)

-600-400'200 0 200 400 600 Time (fe)

Fig. 4. SHG-FROG (left) experimentally measured and (right) retrieved trace. The FROG QTtor was 0.006 with 128 x 128 grid size

1.0-

r6.0

FWHM=^185fsA

0.8-

V N

-5-5

V

i 0.6-

' P I ' 1

H'

4.6^

^ 0.2/^/

V^

-0 .8 -0.6 -0.4 -0.2 0.0 0.2

0.4

0.6 0 8

Time (ps)

Fig. 5. SHG-FROG retrieved pulse temporal intensity of FWHM= 185 fs and phase Fig. 4 demonstrates that the retrieved SHG-FROG trace recovers the salient features observed in the experimentally generated trace. The S H G - F R O G extracted temporal intensity profile and the corresponding phase are shown in Fig, 5. The pulse FWHM of 185 Is is achieved by linear chirp compensatiotl only. The spectral shaping by controlling the chirp and the semiconductor temporal and spectral gain dynamics provides broader mode-locked spectra. Linear down chirp compensated pulsesfix>mdispersion-managed mode-locked semiconductor laser are measured with SHG-FROG revealing duration as short as 185 fs.

References L B. Resati and P. J. Delfyett, IEEE J. Quantum Electron. 40,214,2004. 2. J. C. Diels and W. Rudolph, Ultrafast laser pulse phenomena:fimdamentals,techniques, and applications on a femtosecond time scale. Academic Press, San Diego, 1996, Chap. 2. 3. R. Trebino, Frequency^resolved optical gating: the measurement of ultrashort laser pw/i'^^', Kluwer Academic Publishers, 2000, Chap. 6.

120

Direct Measurement of the Group Delay Dispersion of Ultrashort pulses utilizing Molecular Vibrations Pedro Julian Rizo and Takayoshi Kobayashi Department of Physics, University of Tokyo, Kongo 7-3-1, Bunkyo-ku, Tokyo 113-0033, Japan E-mail: [email protected]

Abstract. An ensemble of coherently vibrating molecules can function as a periodic ultrafast laser modulator. Utilizing this laser modulator, an ultrashort pulse can be characterized even when the period of the modulations is several times longer than the width of the laser pulse. Furthermore, this method of characterization is not restricted to pulses with optical bandwidth narrower than one octave.

Accurate characterization of broadband ultrashort pulses is essential for pulse compression and for many experimental applications such as pump-probe spectroscopy and coherent control. For ultrafast spectroscopy it is desirable to have a precise and experimentally simple method to determine the phase [1,2] or relative group delay (RGD) [3,4] of ultrashort pulses. A novel method utilizing coherent intramolecular vibrations nonresonantly excited in common solvents to directly measure the group delay dispersion of ultrashort pulses utilizing a typical pump-probe configuration is presented in this contribution [5]. The method relies on determining the phase of slow periodic modulations produced by molecular vibrations on the pulse in order to determine its spectral RGD. A laser pulse can nonresonantly excite coherent molecular vibrations in diverse media by an impulsively stimulated Raman process [6]. These molecular vibrations modify the refractive index of the medium. The phase (p(t) of a delayed replica of the pump pulse passing through this medium is modulated by the change in refractive index An(t) depending on its delay time t after the pump: L

A(f>{t)=\-Anit)dl

(1)

0 ^

Here co is the field's angular frequency, c is the vacuum light speed and L is the length of sample. This phase modulation amounts to a spectral shift of the pulse: Ao} = - — A(/>{t) (2) dt Figure 1 shows the probe pulse modulations produced by the Raman-active modes of PCI3 impulsively excited by the pump. The difference transmittance (ATtrans) is

121

measured by simultaneous lock-in detection of spectral portions of the transmitted probe and plotted as a function of wavelength and delay time between the pulses. In such real-time experiments, the phase of the oscillations can be determined by the Fourier transform. The modulations shown in Fig. 1 can be utilized to measure the RGD of ultrashort pulses. The RGD is given by the terms in the derivative of the spectral phase (d(/)((o)/dco) that depend on co. Thus, this method can be used to determine the phase of a pulse up to an absolute phase and an overall time translation. The group delay dispersion and higher order dispersion terms are the coefficients of the Taylor series expansion of the measured RGD. The RGD of the laser pulse (in fs) is obtained by multiplying the phase versus wavelength curve for a given vibrational mode by T/2TZ, where T is the oscillation period of that particular mode. The phase of the molecular oscillation measured in each channel depends on the time delay with which the particular wavelength component reached the vibrationally excited volume of the sample. Thus, the uncompensated chirp in the laser pulse is the difference between the RGD of each wavelength expected for a transform-limited pulse and the RGD actually measured. The RGD expected if the pulse used were transform limited (TL) is given by the convolution:

Here 5"(co) is the derivative of the pulse spectral shape function and F/^X®) the detector's spectral response function. The relative delay expected for a TL pulse is shown in Fig. 2 along with the measured relative delay and their difference. The molecular vibration used to characterize the pulse was the 257 cm"^ symmetric deformation mode of PGI3. The curve for the uncompensated chirp has to be interpolated around the wavelengths where the derivative of the spectral shape function is close to zero, for our nearly Gaussian pulse, around the spectral peak at 785nm. The resulting curve shows the delay (in femtoseconds) with which the different wavelength components within the pulse bandwidth reach the sample.

^^^^^^^^^^1

240220200^lao^160-

Iff^'^^'iHIl V" • - .

^^^^^^1

|||||HH^^BKf .^ •^^jfaHn^i^

^SHH m^' "'-

0

>^^^^^^^^H ., .^i^SBU^^^^^H

iJK^^/VHIfl

a

m 140-

Si.

2 ^

'^^S^^^aH

120100-

eo£0-

..-^O^^^^IBH

H[|^^^^|b^ iiitfNIi^^Uw-^— 1 770

122

1

f ^ ^ ^ ^ ^ ^ ^ ^ ^

7S0 750 800 Wavelenglli (rm)

Fig. 1. A frequency dispersed pump-probe signal of PCI3 showing ATtrans of the pfobe pulse (gray scale, white and black correspond to positive and negative ATtrans respectively) as a function of the detected wavelength (horizontal) and the interpulse delay (vertical).

40-^ 30-|

20

^ 10

>.

•5 Q 0)

0-1

.>-10

a. -20 -30

Wavelength (nm)

Fig. 2. Experimentally measured RGD. The heavy-dotted curve shows the relative delay measured; the dashed curve is the relative delay expected if the pulse were TL and the continuous line is their difference. The light-dotted curve shows the interpolation of the data in the region where 5"(o)) is close to zero. The continuous curve along with the interpolation around 785nm gives the RGD of the laser pulse. In spectrally dispersed pump-probe experiments in solutions, measuring the RGD in the solvent used can serve to correct for both the uncompensated pulse chirp and the dispersion produced in the sample itself, thus, isolating the phase of the signal belonging exclusively to the response of the solute being studied. The method presented does not require fitting or finding the maxima of single peaks in the signal as occurs with other methods that directly measure the RGD of ultrashort laser pulses [3, 4]. The method presented is experimentally undemanding and can be used to characterize pulses with any bandwidth. The mam setback is the difficulty in determining the phase of the oscillatory signal at points where 5"(o)) is zero. If the Kerr-gate configuration is used to detect the signal, this problem can be avoided. In summary, a novel method for characterizing ultrashort laser pulses utilizing pump-probe setup and detection systems has been demonstrated. This method uses coherent molecular vibrations hnpulsively excited in transparent condensed media. For pump-probe experiments in solutions, this method has the capability of isolating the phase of the signal due to the solutefi-omthat due to the solvent and to uncompensated pulse chirp. This is the first demonstration of the use of slow periodic modulations of a laser pulse to measure its group delay dispersion.

References 1 2 3 4 5 6

D. J. Kane and R. Trebino, IEEE J. Quantum Electron., 29, 571, 1993. C. laconis and I. A. Walmsley, Opt. Lett., 23, 792, 1998. R. Fork, C. Shank, C. Hirlimann, R. Yen, and W. Tomlinson, Opt. Lett. 8, 1, 1983. J. L. A. Chilla and O. E. Martinez, Opt. Lett., 16, 39, 1991. P. J. Rizo and T. Kobayashi, Appl. Phys. Lett. 85, 28, 2004. Y-X. Yan, E. B. Gamble, Jr., and K. A. Nelson, J. Chem. Phys., 83, 5391, 1985.

123

Spatially encoded spectral interferometry for complete characterisation of attosecond XUV pulses Eric Cormier^ Ian A. Walmsley^, Ellen M. Kosik^'^, Laura Corner^ and Louis F. DiMauro"^ ^CELIA, University Bordeaux I, 351, Cours de la Liberation, 33405 TalenceCEDEX, France ^Clarendon Laboratory, University of Oxford, Parks Road, Oxford, 0X1 3PU, UK l.comerl @physics.ox.ac.uk ^The Institute of Optics, University of Rochester, Rochester, New York 14627, USA "^Brookhaven National Laboratory, Upton, New York 11973, USA Abstract: We propose a variant of the spectral phase interferometry for direct electric field reconstruction (SPIDER) technique for characterisation of attosecond pulses directly in the XUV by spatially encoding the phase information onto an interferogram. In this paper, we outline a technique for the complete characterisation of attosecond XUV pulses produced by high harmonic generation (HHG), based on spectral phase interferometry for direct electric field reconstruction (SPIDER) [1]. This self-referencing interferometric technique involves the mixing of two fields that are replicas of each other except that one of the replicas is spectrally shifted, or sheared, with respect to the other and delayed in time. The spectrally resolved interference of these pulses yields an interferogram from which the spectral phase can be extracted. The technique involves directly measuring XUV photons with a spectrometer and is thus more efficient and considerably less complex than methods which measure the photoelectron spectra emitted when the short pulse is mixed in a gas with longer optical pulses. It also has the advantage of a larger signal to noise ratio, and therefore the possibility of single-shot measurement, as well as higher accuracy from data accumulated over multiple shots. To create the sheared pulses we use the fact that the XUV radiation produced via HHG depends on the mean frequency of the driving pulse. Specifically, this means that a harmonic pulse train generated by a pulse of mean frequency co and one generated by a pulse of mean frequency co + 8co will be spectrally sheared with respect to one another by nSo) at the nth harmonic. To create an isolated attosecond pulse a section of the generated spectrum may be filtered out and overlapped in a XUV spectrometer to produce a SPIDER interferogram. Fig. la shows a simulation of the interferogram obtained in this way from a pair of harmonic pulses generated in argon by a pair of 30 fs pulses with mean wavelengths of 800 nm and 804 nm, temporally separated by 77 fs. The peak intensity of each pulse is 1.7 x 10^"^ W/cm^, which ensures that the ionization does not saturate. Total ionization yield is in this case below 10%. The interferogram can be inverted using

124

standard Fourier processing methods [1] to give the spectral phase of the X-ray pulse as shown in fig. lb. b) - ^ input phase reconstructed phase

30

31

32

33

34

35

36

37

38

30

31

HamxaiiccrdQ-

32

33

34

35

36

37

38

Harmonic order

Figure 1: a) Interferogram between harmonics generated by pump pulses at 800 and 804nm; b) Spectrum, original phase and phase reconstructedfromthe interferogram This XUV SPIDER scheme is applicable to pulses with almost arbitrarily short duration, as long as a shift in frequency of the pulse can be achieved by a shift in the pump pulse frequency. The requirements for implementing SPIDER in the conventional manner described above are not stringent. First, it is necessary to generate two replicas of the pump pulse that are identical except for a frequency shift. This is possible using conventional optical pulse shaping methods [2]. Secondly, the resolution of the XUV spectrometer must be sufficient to allow sampling of the interferogram fringes at the Nyquist limits. Although achievable, this requires a 0.01 nm resolution for an X-ray photon at 25 nm. For accurate pulse reconstruction it is also necessary that the time delay between the two pulses is measured accurately. Gas jet

Attosecond pulses Spherical Grating

frequency

Detector Figure 2. Schematic experimental arrangement for spatially encoded SPIDER characterisation of attosecond pulses produced by high harmonic generation A problem with the conventional SPIDER technique outlined above is that if the driving pulses have a high intensity (to produce the most intense XUV pulses), the first pulse will significantly ionise the HHG medium [3]. This issue

125

can be avoided by changing the geometry of the nonUnear interaction so that the interferogram has a spatial, rather than spectral, carrier. This is shown in fig. 2. The two driving pulses now generate two spectrally sheared harmonic pulses in spatially separated regions. The harmonic radiation then propagates as before to a single spectrometer, which records the spatial interference pattern [4] as a function of XUV wavelength. The interferogram measured in this Spatially Encoded Arrangement for SPIDER (SEA SPIDER) has the form of a spatially and spectrally resolved XUV pulse spectrum: . kx'^ . kx'x +1 L dx' S(x,co) = j [ E ( x » + E ( x > + Q ) e ^ ^ ' ] e 2L 1

(1)

Here cOo is the central frequency of the pulse where the phase is set to zero, K is the difference in the mean transverse wavevectors of the driving pulses, given by K = kX/L, where X is the lateral separation between the two driving pulses, L is the focal length of the focusing mirror, k is the mean wavevector of the XUV spectrum and the spectral shear Q is given by n5co. The spectral correlation term is extracted from the SEA SPIDER interferogram using the same algorithms as in conventional SPIDER, except that the Fourier transform is taken with respect to the spatial axis, so that the interferometric component is extracted in the kx domain. After the relevant interference term is selected, an inverse Fourier transform is taken, and the resulting function resolved with respect to the XUV beam transverse spatial coordinate x. Then the extracted phase is (p(x, OD - (Do)- (p(x, co - ODQ + i^). This is the same as the spectral phase difference returned in conventional SPIDER from which the spectral phase of the pulse can be easily extracted. The accuracy and precision of the reconstruction can be tested by setting the shear between the two driving pulses to zero. This guarantees that the harmonic shear Q is also zero, and therefore that the reconstructed phase is a constant. Any deviation from this represents a systematic error. SEA SPIDER enables the characterisation of individual attosecond X-ray pulses created by high harmonic generation. It may also be used to characterise attosecond pulse trains, with an additional measurement to obtain the phase between two adjacent harmonics. There are two important advantages to the SEA-SPIDER configuration. First, it avoids the problem of generating two harmonic pulses in the same region of the atomic gas, and secondly, the XUV spectrometer needs only sufficient spatial resolution to adequately sample the spectrum according to the Whittaker-Shannon limit for the attosecond pulse support. It is suitable for single-shot (i.e. non-pulse averaged) operation.

References [1] C. laconis and LA. Walmsley, IEEE QE 35, 501-509 (1999) [2] F. Verluise, V. Laude, Z. Cheng, C. Spielmann and P. Toumois, 2000, Opt. Letts. 25, 575. [3] M. Bellini, C. Lynga, A. Tozzi, M.B. Gaarde, T.W. Hansch, A. L'Huillier and C. G. Wahlstrom, Phys. Rev. Letts. 81,297 (1998). [4] P. Saileres, L. Le D^roff, T. Auguste, P. Monot, P. d'Oliveira, D.Campo, J.-F. Hergott, H. Merdji and B. Carr^, Phys. Rev. Letts. 83, 5483-5486 (1999)

126

Spatially encoded spectral interferometry for complete characterization of ultrashort pulses Ellen M. Kosik^'^, Aleksandr S. Radunsky^'^, Ian A. Walmsley^ and Christophe Dorrer^ ^ The Institute of Optics, University of Rochester, Rochester, New York 14627, USA E-mail: [email protected] ^ Clarendon Laboratory, University of Oxford, Parks Road, Oxford, 0X1 3PU, UK ^ Bell Laboratories, Lucent Technologies, 101 Crawfords Corner Rd, Holmdel, NJ 07733, USA Abstract. We present a modification of the SPIDER technique for characterizing ultrashort pulses with reduced spectrometer resolution and without the creation of identical replicas of the unknown pulse. This technique is ideal for ultra-broadband pulse characterization.

1. Introduction Techniques for characterizing ultrashort optical pulses typically incorporate measurements across the entire pulse energy spectrum [1-5]. Accommodating pulses of shorter duration and concomitant larger bandwidth with a spectrometer of fixed size necessarily degrades the available spectral resolution. The new technique, Spatially Encoded arrangement for SPIDER (SESPIDER), uses a spectrally resolved spatial and therefore has a much more relaxed spectral resolution requirement.

2. Experimental Methods In conventional SPIDER [1], two time-delayed replicas of the unknown pulse are mixed with a highly chirped pulse in a nonlinear crystal, resulting in two upconverted time-delayed pulses with a relative spectral shear between them. The delay between the pulses induces fringes on the measured spectrum, which allow the direct extraction of the spectral amplitude and phase using the Fourier transform-based algorithm. In SE-SPIDER, the two spectrally sheared pulses are created by mixing just one unknown pulse with two time-delayed chirped pulses in a nonlinear crystal. A schematic of this process is shown in figure 1. The three parallel, but laterally displaced pulses are focused into a nonlinear crystal. The resultant spectrally-sheared replicas have different transverse components to their propagation vectors. This relative wavefront tilt between the sheared replicas translates into a spatial interference pattern when the pulses are superposed, by focusing, onto the slit of a 2D imaging spectrometer.

127

^: ^

R(x,(i>ffi\)

Fig 1. Experimental set-up for SE-SPIDER. — zero-order X/2 waveplate.

TFP —thin film polarizer.

WP

The phase retrieval algorithm for SE-SPIDER is a slight modification on the conventional SPIDER algorithm, and similar to the method used for Space-Time SPIDER characterizations [6].

3. Results and Discussion By way of proving the principle, we have used the SE-SPIDER technique to measure 100 fs, 10 nJ pulses from a mode-locked Ti:Sapphire laser. However, it should be noted that 100 fs pulses are far from the limit of the shortest pulses which could be measured with this technique. We estimate that with our current setup (1.2 m imaging spectrometer with a 600 1pm grating and a 512 element CCD camera) we could easily characterize sub-15 fs pulses, and with a less dispersive spectrometer, even shorter pulses would be accessible. For example, a 0.5 m imaging spectrometer with a 6001pm grating and 2048 element array could accommodate the sub-6 fs pulses measured by Gallmann, et. al. [7] in a single shot. As a general check of the technique, we measured a pulse before and after passing through a 4 cm slab of SF8 glass. The dispersion of the glass block measured this way, 5600 fs^ is in good agreement with the known value of 5575 fs^, which demonstrates the accuracy of the technique. The measurements are also in excellent agreement with the pulse measured using conventional SPIDER. The spatial dimension of the SE-SPIDER interferogram can also be exploited to gain information about the spatial dependence of the pulse spectrum. In figure 2 we show the SE-SPIDER interferogram of a pulse, which was spatially chirped by passing through a prism pair. The spectral density and phase profiles at two different spatial points in the beam, also shown in figure 2, show the presence of spatial chirp. The measured value of this chirp, 109 THz/mm, is in good agreement with that calculated for our prism configuration, 112 THz/mm. If the mechanism of space-time coupling is known, as in this case, this measurement constitutes a complete characterization of the space-time field. Such assumptions have been used before to provide characterization when a functional form can be assumed [8, 9].

128

\

hi .^c'-" 790

//V

\ \

// / / \ // . \\ 795

^ \

•z'"

V^

/

\ X . .. 1 800

805

810

Fig 2. SE-SPIDER interferogram of a spatially-chirped pulse and the reconstructed fields for two different spatial slices. The spectral density and phase at spatial slice a are given by the solid and dotted lines respectively, while that for spatial slice b are given by the dash-dotted and dashed lines.

4. Conclusions We have demonstrated an interferometric technique for the complete characterization of ultrashort optical pulses with greatly reduced spectral resolution requirements which is therefore well suited for measuring very broadband fields. This technique, Spatially-Encoded SPIDER, is similar to the conventional SPIDER technique but employs a spatial, rather than spectral signal encoding and has an advantage of only requiring one replica of the pulse to be measured. The two-dimensional (space and frequency) trace allows reconstruction of the spatially-resolved temporal electric field, and can therefore be used to track chirp or more complex distortions.

References 1 C. laconis and LA. Walmsley, IEEE QE 35, 501-509 (1999). 2 R. Trebino, K. W. Belong, D. N. Fittinghoff, J. N. Sweetser, M. A. Krumbugel, B. A. Richman, and D. J. Kane, Rev. Sci. Instr. 68, 3277-3295 (1997). 3 J. Chilla and O. Martinez, IEEE J. Quant. Electron., 27,1228-1235 (1991). 4 K. C. Chu, J. P. Heritage, R. S. Grant, K. X. Liu, A. Dienes, W. E. White, and A. Sullivan, Opt. Lett., 20, 904-906 (1995). 5 D. T. Raid, IEEE J. of Quantum Electronic 35,1584-1589 (1999). 6 C. Dorrer, E. M. Kosik, and LA. Walmsley, Opt. Lett. 27, 548-550 (2002). 7 L. Gallmann, D. H. Sutter, N. Matuschek, G. Steinmeyer, U. Keller, C. laconis, and LA. Walmsley, Opt. Lett. 24,1314-1316 (1999). 8 C. Dorrer and I. A. Walmsley, Opt. Lett. 27,1947-1949 (2002). 9 S. Akturk, M. Kimmel, P. O'Shea, and R. Trebino, Opt. Exp. 11, 491-501 (2003).

129

Full characterization of ultraviolet and visible 10-fs pulses with zero-additional-phase SPIDER p. Baum, S. Lochbrunner, and E. Riedle LS fiir BioMolekulare Optik, LMU Mtinchen, Oettingenstr. 67, 80538 Mtinchen, Germany Abstract: Zero-additional-phase SPIDER is a novel spectral shearing interferometry setup capable to characterize the temporal amplitude and phase of ultrashort pulses over an extremely wide wavelength region. We demonstrate the full characterization of ultraviolet and visible pulses with durations in the 10-fs regime. The scheme does not alter the unknown pulses and yields the pulse shape at the interaction point of a spectroscopic experiment.

1. Zero-additional-phase SPIDER Tunable femtosecond pulses in various spectral regions are essential for experiments on ultrafast physical and chemical processes. For the management of higher order chirp a full characterization of amplitude and phase is essential. Due to dispersive optics like sample cells and the dispersion of air the pulses are only short at one point in the beam path. Conventional characterization schemes like FROG [1] and SPIDER [2] do not characterize the pulses at the experiment since they split the beam before the actual characterization and introduce additional phase. In this contribution we present an unambiguous and highly sensitive scheme to overcome these limitations and demonstrate the full characterization of extremely short ultraviolet and visible pulses directly at the experimental interaction point. In conventional SPIDER (Fig. 1 a) the unknown pulse is split into two delayed replica and mixed with a chirped pulse in a nonlinear crystal [2]. The stretched pulse is considered monochromatic during the interaction with each short pulse. Two identical but spectrally sheared and temporally delayed replica result which interfere in a spectrograph. The interferogram contains the differences in spectral phase between pairs of shifted frequencies and can be directly evaluated to yield the spectral phase of the original pulse. An additional measurement of the power spectrum gives a full reconstruction of the time dependent electric field [3]. The pulse is characterized in the nonlinear crystal, i.e. after the beam splitter. In the novel ZAP-SPIDER [4] (Fig. lb) the unknovm pulse is guided into the nonlinear crystal without manipulation. Two frequency shifted replica of the unknown pulse are generated by mixing with two strongly chirped auxiliary pulses C0„+f2 6),)

<^.l...^...J., n „. "XT '^-'™ unknown

delayed pulses

'V -A- X . A . 1 1 r-i pulse

_ / J > C _ S _ J\_

ZAP-SPIDER

Fig. 1. (a) (X-)SPIDER and (b) the novel ZAP-SPIDER. D(a)) denotes a spectrograph.

130

which are delayed with respect to each other and come from slightly different directions. The spectrally sheared replica propagating in different directions are generated at center frequencies COQ and COQ+Q. They are brought to interference in a spectrometer with a delay T chosen to give a suitable number of spectral fringes. The spectral phase of the unknown pulse is evaluated from the interferogram following the procedure of the conventional SPIDER [3]. The effective characterization point is the nonlinear crystal. To characterize a pulse at the interaction point of an experiment the sample can be replaced by the ZAP-SPIDER mixing crystal. No optical component has to be introduced before this point and the unknown pulse acquires zero additional phase.

2. Experimental Setup and Results As a source for extremely broadband visible pulses we use a NOP A [5-6]. The output or its second harmonic (Fig. 2a); solid line) is focused into a thin BBO crystal. Pulses split from the Ti:sapphire amplifier (dotted lines) are stretched to about 2 ps in a glass block (375 mm SF57), divided by a beam splitter and overlapped with the unkown pulses in the BBO crystal under small angles. The sheared replica (dashed lines) are imaged to a common focus at the entrance slit of the spectrograph. To interfere the two spatially separated beams the width of the slit is made smaller than the focus diameter. Diffraction ensures an interference pattern at the focal plane of the spectrograph despite the angle between the beams. Visible pulses at 600 nm are compressed in a prism sequence with a deformable end mirror [7] and the ZAP-SPIDER is equipped with a 25 |am BBO crystal aligned for type-I sum frequency generation. The interferogram is evaluated at 10 Hz update rate and the spectral phase information is used to optimize the deformable mirror. Fig. 2b) shows the NOPA spectrum and the residual spectral phase and Fig. 2c) the measured temporal intensity and phase. The pulse length of 10.0 fs is within 10 % of the Fourier limit. To validate the ZAP-SPIDER measure540

560

580

600 620 640 (nm)

_

(a) /I BBO

SF57 [J

i

4

775 nm unknown 150fs pulse

-20 0

20 (fs)

-20

0

20 (fs)

Fig. 2. (a) Experimental setup: SF57, dispersive glass; BS, beam splitter; Q, delay to adjust the spectral shear; BBO, mixing crystal; T, delay to adjust fringe spacing; SP, spectrograph with entrance slit, (b) Spectrum and spectral phase, and (c) temporal intensity and phase of a 10.0 fs visible pulse, (d) Comparison of the intensity autocorrelation calculated from the pulse shape determined by ZAP-SPIDER (solid trace) and the measured trace (dots).

131

ment an intensity autocorrelation is computed from the measured pulse shape and compared to an independently measured dispersion-free intensity autocorrelation (see Fig. 2d). The traces agree precisely and even the effects of the weak satellites of the pulse due to the flat top spectral profile are reproduced well. In a second experiment we generate the second harmonic of the NOPA at 292 nm in a 50 jiim thick BBO crystal and adapt the ZAP-SPIDER to type-I difference frequency mixing. The spectrally sheared pulses are generated in the visible with a 62 jLim thick BBO crystal. Fig. 3a) shows the interferogram and Fig. 3b) the spectrum and the evaluated spectral phase of the UV pulses. The residual third order phase is due to the imperfect compression. The temporal intensity and phase are shown in Fig. 3c). The pulse width of 18.7 fs is within 20% of the Fourier limit. As little as 10 nJ pulse energy in the UV is sufficient for the measurement. This is the first SPIDER characterization of sub-20 fs UV pulses. Recent results show that even 7 fs UV pulses can be effectively characterized with ZAPSPIDER.

.--(*******w. 460

470

480

(nm)

Fig. 3. (a) Difference frequency mixing ZAP-SPIDER interferogram. (b) Spectrum and spectral phase and (c) temporal intensity and phase of a UV pulse at 292 nm with 18.7 fs duration.

3. Conclusions 290 300 (nm) -30

0

30 60 (fs)

Zero-additional-phase SPIDER is a novel spectral shearing interferometry scheme to characterize the temporal amplitude and phase of ultrashort optical pulses over an extremely wide wavelength region. ZAP-SPIDER is essentially dispersion free and yields the pulse shape at the interaction point of a spectroscopic experiment. We fully characterize tunable 10-fs pulses in the visible and demonstrate for the first time an interferometric characterization of ultrashort UV pulses.

References 1 R. Trebino, K. W. DeLong, D. N. Fittinghoff, J. N. Sweeter, M. A. Krumbtigel, B. A. Richman, and D. J. Kane, Rev. Sci. Instrum. 68, 3277 (1997). 2 C. laconis and I. A. Walmsley, Opt. Lett. 23, 792 (1998). 3 C. laconis and I. A. Walmsley, IEEE J. Quantum. Electron. 35, 501 (1999). 4 P. Baum, S. Lochbrunner, and E. Riedle, Opt. Lett. 29, 210 (2004). 5 T. Wilhelm, J. Piel, and E. Riedle, Opt. Lett. 22, 1494 (1997). 6 E. Riedle, M. Beutter, S. Lochbrunner, J. Piel, S. Schenkl, S. Sporlein, W. Zinth, Appl. Phys. B 71, 457 (2000). 7 P. Baum, S. Lochbrunner, L. Gallmann, G. Steinmeyer, U. Keller, and E. Riedle, Appl. Phys. B 74 [SuppL], S219 (2002).

132

Direct visualization of transient absorption by real-time pump-probe imaging spectroscopy Naoki Furukawa^ Chad E. Mair^, Valeria D. Kleiman^ and Jun Takeda^ ^ Department of Physics, Graduate School of Engineering, Yokohama National University, Yokohama 240-8501, Japan E-mail: [email protected] ^ Department of Chemistry, University of Florida, Gainesville, Florida 32611, USA Abstract. We report a new method to visualize ultrafast transient absorption of materials. This method enables us to simultaneously map frequency- and time-resolved absorbance changes with femtosecond time resolution in real-time.

1.

Introduction

Observation of transient behavior at different frequencies is essential and of great importance in studying ultrafast photochemical and photophysical processes. To investigate such phenomena, femtosecond laser spectroscopy has been extensively utilized for the last tvs^o decades. Most ultrafast measurements are conducted w^ith an excitation pulse followed by a variably delayed probe pulse, a sequence repeated many times to cover the desired time and spectral region. A method that allows real-time visualization of ultrafast processes is therefore highly desired. With the purpose of monitoring the duration of pulses, spectroscopic tools such as a single-shot autocorrelator and frequency-resolved optical gatuig (FROG), both based on single-shot detection, have been developed. Single-shot pump-probe techniques have recently been used to investigate hreversible photochemical reactions in solids and viscous liquids [1,2]. Although using these techniques it is possible to collect multiple temporal data points at some frequency from a single probe pulse, however, it is not possible to take multiple frequency data points within a wide spectral region. Here we report a new scheme for taking pump-probe data, real-time pump-probe imaging spectroscopy, which enables us to simuhaneously visualize temporal and spectral data points with femtosecond tune resolution.

2.

Experimental Results and Discussion

The schematic diagram of the real-time imaging spectroscopy apparatus is shown m Fig. 1. The basic idea of this method is sunilar to that of the previous single-shot pump-probe techniques except that a white-light continuum generated by a self phase modulation is used as the probe. The collimated pump and probe beams derived from a regenerative Ti: sapphire laser system are separated by an angle 6 and cylindrically focused such that their focal lines are spatially coincident within

133

a sample. Since the probe beam reaches different portions of the sample at different times, the delay between the pump and probe beams is spatially encoded across the sample. The probe beam passing through the sample is recollimated with an appropriate magnification and then cylindrically focused onto the entrance slit of a monochromator coupled to a two-dimensional CCD detector. The temporal information of the probe beam is recorded along the direction parallel to the slit, while the spectral information is analyzed by the monochromator along the direction normal to the slit. The 2D mapping of the frequency- and time-resolved transient absorption is therefore obtained in real-time. Under our experimental conditions, wide temporal and spectral regions of ~ 5 ps and 430 - 620 nm are simultaneously mapped with 200 - 300 fs time resolution [3]. Monochromator witha2D-CCD

Pump beam

Delay time

Figure 1. Schematic diagram of the real-time pump-probe imaging spectroscopy As a first demonstration, using this technique, we measured ultrafast internal conversion processes of /^-carotene in benzene solution. Before analyzing the spectral information with the monochromator, to ensure that the probed passed through the sample, we inserted a mirror M in front of the monochromator and imaged the probe beam on a 1x1 cm^ screen. This image contains information of the transient absorption of y^-carotene. The image was taken by a digital camera with an exposure time of 4 sec. Figure 2 shows the images of the transmitted probe beam with the pump beam off (a) and the pump beam on with different delay times, 0 (b), 1 (c) and 2 ps (d). The delay between the pump and probe beams is tuned by a mechanical translation stage. In the image, the time-encoding direction is horizontal. When the pump beam is blocked, the image (a) shows a homogeneous color (orange) that reflects the transmission of the probe without the transient absorption of y^-carotene. When the pump beam is on, on the other hand, the images (b)-(d) have a dark area that moves along the time-encoding direction with the delay. This dark area corresponds to the transient absorption of y^-carotene, which attenuates the transmission intensity of the probe. This result shows that our technique successfully works to visualize the ultrafast transient absorption of materials. Next, we remove the mirror M from the optical path and the image is cylindrically focused onto an entrance slit of the monochromator to visualize the frequency- and time-resolved transient absorption of y^-carotene. The 2D map of the transient absorption is shown in Fig. 3. The absorbance change is indicated by contours. The correction of the group velocity dispersion of our system as well as the normalization of the transient absorption by the spatial profile of the pump beam is performed on a personal computer. To visualize the whole relaxation

134

processes of /^-carotene, several frames of the image having different delay times are tiled on the 2D map. This 2D map clearly shows the instantaneous absorption bleaching at 430 - 500 nm and the transient absorption at 500 - 620 nm with an apparent rise time of ~ 1 ps, which are observed in previous works [4, 5]. These results indicate that our method allows us to perform real-time mapping of frequency- and time-resolved absorbance change with femtosecond time resolution in a short acquisition time. (a)

t:.L..}

(b)

(c)

111!

Figure 2. Images of the probe beam transmitted through y^-carotene solution with the pump beam off (a) and the pump beam on with different delay times of 0 (b), 1 (c) and 2 ps (d)

20 S 15 S

ll

lofl

-o.sl

off450

500 550 oOO Wavelength / nm

Figure 3. 2D mapping of frequency- and time-resolved absorbance change in y^-carotene In summary, we can demonstrate direct visualization of frequency- and timeresolved transient absorption of materials using real-time pump-probe imaging spectroscopy. The acquisition time of this imaging technique is much shorter than the previous techniques because multiple temporal and spectral data points are collected in real-time, our technique is expected to be a powerftil tool to measure ultrafast photochemical and photophysical processes. Acknowledgements. This work was supported in part by The Ministry of Education, Culture, Sports, Science and Technology (No. 14654050), The Sumitomo Foundation, Yokohama Academic Foundation and Sasakawa Scientific Research Grant from The Japan Science Society (No. 15-064).

References 1 2 3 4 5

L. Dhar, J. T. Fourkas and K. A. Nelson, Opt. Lett. 19, 643, 1994. G. P. Wakeham and K. A. Nelson, Opt. Lett. 25, 505, 2000. N. Furukawa, C. E. Mair, V. D. Kleiman and J. Takeda, submitted. M. Yoshizawa, H. Aoki and H. Hashimoto, Phys. Rev. B 63, 180301(R), 2001. N. Furukawa and J. Takeda, Nonlinear Optics 29, 579, 2002.

135

Part II

Strong Fields and High Order Harmonics

Dynamic Molecular Imaging p. B. Corkum Steacie Institute for Molecular Sciences, National Research Council of Canada, 100 Sussex drive, Ottawa, Ontario KIA 0R6, Canada Abstract. Intense laser pulses allow new approaches to imaging the dynamics of molecules. Using laser Coulomb explosion as a probe, atomic positions can be measured with subAngstrom accuracy. The re-collision electron can also be used to probe of molecular structure. With re-collision we can excite (or diffract from) a molecule, even taking a holographic or tomographic image. All approaches are consistent with attosecond timeresolved measurements.

Molecules have 3(N+M)-3 spatial co-ordinates (where N and M are the number of electrons and atoms in the molecule) as well as time. Pump-probe molecular spectroscopy usually takes a low-dimensional projection of this inherently multidimensional problem. Hidden from view in most ultrafast molecular dynamics experiments are unobserved changes to the molecular structure. There are two approaches for measuring molecular structural changes during dynamics: (1) An ultrashort externally generated short wavelength probe pulse can diffract from the molecule revealing its structure. The probe pulse can be either a short X-ray or electron pulse. (2) An ultrashort internally generated short wavelength probe pulse can be used as if it were external. Since all molecules have ions and electrons they are the two choices. Two recent technical advances allow the second approach to be considered. They are intense few cycle laser pulses and molecular control, particularly control of molecular alignment [1]. Both technologies are now well developed.

Re-collision electrons for probing molecular dynamics We concentrate on dynamic molecular imaging using internally generated particles. Internally generated electrons can be produced by tunnel ionization. The wave packet is released near each field maximum with timing precision of ~ 200 attoseconds. As soon at the electron wave packet emerges from the ion it is caught in the strong laser field, pulled from the ion by the field, only to be driven back when the field reverses directions. The portion of the wave packet that tunnels from the ion during the VA cycle following the field maximum will be swept back

139

past the position of the parent ion (re-collide). Almost all electrons pass within 1 nm (assuming 800 nm light) in the lateral direction so their probability of inelastic scattering, or large angle elastic scattering is very large. The maximum energy of the re-collision electron is ~ 3.17 Up where Up=(qE)V(ma)^) and q, E, m and o) have their usual meaning. At an intensity of 2x10^^^ W/cm^, which is typical for ionization of molecules and >^=1.8^im, the maximum re-collision energy is ~200eV. Comparing an internally generated electron to an externally generated one, a recollision electron has a very high equivalent peak current density (Fig. 1). At

E a E 03

c o •o 0 3

8

o

10

Tim e / fs Figure 1. Equivalent current density seen by the Yii ion during the 10 femtoseconds following ionization of H2 by 800 nm light >.=800 nm, it reaches ~ 10^^ amps/cm^ [2] compared to about 10^ amps/cm^ for conventional electron sources. With wavelength shorter than the typical dimensions of a molecule, these recollision electrons can be used to probe the molecule from which they departed. In particular, they can probe the structure of a molecule at each step of a pumpprobe experiment.

Coulomb explosion for probing molecular dynamics. So far I have concentrated on electrons. However few cycle laser technology allows us to irradiate a molecule with very intense light. Multiple-ionization is easily achieved. Ionization is complete, and the electrons leave the vicinity of the ion (especially with circularly polarized light) before the laser pulse passes. Meanwhile, the ions are frozen by their inertia during the time that the electrons are stripped away.

140

Once the pulse is over, the fragments "explode". Of all ions are measured in coincidence, laser-induced Coulomb explosion imaging reveals the molecular structure at the time multiple-ionization occurred. A complete image is achieved for each molecule. Used as a probe in a pump-probe arrangement, Coulomb explosion can image ultrafast dynamics [1,4].

Imaging the fastest molecular motion. Figure 2 shows the evolution of a vibrational wave packet on the ground electronic state of 02"^. The wave packet is launched at t=0 by ionizing the neutral molecule. The upper curve of the left is the calculated position of the wave packet as a function of time. The lower curve shows the experimental measurement. Figure 2 shows that Coulomb explosion has the time and spatial resolution [4] to measure almost any photochemical process.

o< c o

1.2

CO

I 1.0 (0

^ 15

30

45

Delay (fs)

0.9-i 0.8 1 ,5 2.0

2.5

3.0

3.5

4.0

4.5

Time / fs

Figure 2. The first 75 femtoseconds of the life of a vibrational wave packet on 02"^. The left lower curve is measured using Coulomb explosion imaging. The right curve is measured using the re-collision electron. In another submission to these proceedings [5] we present how the re-collision electron can be used to take a three-dimensional image a molecular orbital — including both its amplitude and phase. The re-collision electron can also measure the position of the atoms in a molecule using laser induced electron diffraction [6]. However, for the remainder of the paper we concentrate on another unqiue property of high intensity technology. Internally generated particles always have a correlated partner. For example, any time that strong field ionization produces an ionized electron, it automatically forms a correlated ion. Exploiting correlation (or really entanglement) offers many options for enhanced measurement [3]. It allows us to measure attosecond electron [2] and ion [3] dynamics without attosecond pulses.

141

Imaging the fastest molecular motion. In the experiment illustrated in Fig. 3, strong field ionization produces correlated wave packets around each crest of the electric field: an electron wave packet that is directed by the laser field and a vibrational wave packet that moves on the ground electronic state of the molecular ion. Launching the correlated pair is the pump stage in our pump probe experiment based on correlation.

Probe step

D2^(X^2;,

Figure 3: Illustration of an ultrafast experiment based on correlation. The curve represents the potential energy of D2^, on which the vibrational wave packet moves. Ionization of D2 forms a vibrational wave packet on D2^. The electron is pulled away from the ion and then driven back after the laser electric field reverses direction. In the subsequent re-collision it probes the position of the wave packet by inelastic scattering. Measuring the fragment ion velocities measures nuclear positions. The delay between pump and probe is approximately 0.6 periods. One wave packet can be used to probe the other. Here we probe the vibrational wave packet motion by inelastic scattering of the electron wave packet. Changing the wavelength changes the pump-probe time delay. We observe nuclear motion on the D2"*" (X^^g^) electronic surfaces, whose half vibrational period is ~ 11 femtoseconds, by measuring the D"*" fragment kinetic energy where the fragmentation is caused by inelastic scattering. A form of Coulomb explosion, the fragment kinetic energy gives the interunclear separation. The time between when the correlated wave packets are formed and when they re-collide is controlled by the laser period. Our observation window is between 1.5 and 4.5 femtoseconds, determined by the tuning range from 800 nm to 1.85 microns of our ionizing laser beam. Figure 2, right panel, shows the experimentally determined time dependent position of the vibrational wave packet as, a function of time. The error bars measure the noise on our data. The calculated dynamics is shown by the solid curve. The agreement between the two implies a timing measurement accuracy of - 200 attoseconds and a spatial measurement accuracy of - 0.03A.

142

In conclusion, strong field molecular science is opening many new opportunities for molecular imaging — re-collision induced inelastic scattering [2], diffraction [6], holography [8], tomography [5], exploiting correlation for enhanced measurement [3,7] and laser Coulomb explosion imaging [1,4]. All are closely connected to attosecond pulse generation and measurement. Re-collision physics allows us to combined attosecond / Angstrom measurements — a whole new direction for molecular science.

References ^' 2

P. W. Dooley, I. V. Litvinyuk, K. F. Lee, D. M. Rayner, M. Spanner, D. M. Villeneuve and P. B. Corkum, "Direct Imaging of Rotational Wave Packet Dynamics of Diatomic Molecules", Phys. Rev. A. 68, 023406 (2003). H. Niikura, F. Legare, R. Hasbani, A. D. Bandrauk, M. Yu. Ivanov, D. M. Villeneuve and P. B. Corkum, "Sub-laser-cycle electron pulses for probing molecular dynamics", Nature, 417, 917, (2002).

3

H. Niikura F. Legare, R. Hasbani, M. Yu. Ivanov, D. M. Villeneuve and P. B. Corkum, "Using correlated pairs for sub-femtosecond-resolution wave packet measurements". Nature, 421, 826 (2003). 4

Francois Legare, Kevin F. Lee, I.V. Litvinyuk, P.W. Dooley, A.D. Bandrauk, D.M. Villeneuve, P.B. Corkum, "Laser Coulomb explosion imaging for probing molecular structure and dynamics", submitted to this conference. J. Itatani, J. Levesque, D. Zeidler, M. Spanner, P. B. Corkum, and D. M. Villeneuve "Tomographic Imaging of Molecular Orbital with High Harmonic Generation" submitted to this conference. T. Zuo, A. D. Bandrauk and P. B. Corkum, "Laser Induced Electron Diffraction: A New Tool for Probing Ultrafast Molecular Dynamics", Chem. Phys. Lett. 259, 313 (1996). 7

N. Milosevic, P. B. Corkum and T. Brabec, "Attosecond Laser ElectroNuclear Physics", Phys. Rev. Lett. 92, 013002, (2004); S. Chelkowski, A. D. Bandrauk and P. B. Corkum, "Control of Nuclear Processes with SuperIntense Laser Pulses", submitted to Phys. Rev. Lett. M. Spanner, O. Simova, P, B. Corkum and M. Yu Ivanov, "Reading diffraction images in strong field ionization of diatomic molecules" J. Phys. B 37, 243 (2004).

143

Femtosecond electron diffraction: making the "molecular movie^'

Towards

Jason R. Dwyer^ Robert E. Jordan^ Bradley J. Siwick^, Christoph T. Hebeisen^ and R.J. Dwayne Miller^ ^ Depts, of Chemistry and Physics, University of Toronto, Toronto ON Canada M5S 3H6 E-mail: [email protected] ^ Present address FOM Institute for Atomic and Molecular Physics (AMOLF), Kruislaan 407, 1098 SJ, Amsterdam, The Netherlands. Abstract: We use femtosecond electron diffraction to capture the disordering of the solid and emergence of liquid structure during strongly-driven laser-induced melting. We propose a method to measure femtosecond electron pulse durations and zero of time.

1.

Introduction

"Making the molecular movie" sums up the dream experiment of reaction dynamics: the direct experimental observation of the changing atomic positions during a chemical reaction. While femtosecond spectroscopy has made incredible gains in the temporal domain and can provide a mode-driven picture of dynamics, inverting the experimental measurement of the potential energy surface to an atom-centered picture is nontrivial and does not provide a unique solution in general. In contrast, time-resolved diffraction experiments directly probe the observable of interest, the nuclear coordinates: each time-delayed diffraction pattern can be simply inverted via Fourier transformation to provide a snapshot of the nuclear configuration of the sample as a function of time. We have recently made a significant advance in electron gun design that has permitted the generation of subpicosecond electron pulses of sufficient brightness for use in time-resolved electron diffraction experiments [1-3]. This development extends the original demonstrations by Mourou and Williamson in the solid-phase [4] and Ischenko and co-workers in the gas-phase [5] to the femtosecond time domain. The importance of time-resolved structural information as extracted from electron diffraction has been amply demonstrated by the picosecond electron diffraction studies by Zewail and co-workers [6] and Dudek and Weber [7] as applied to gas-phase reactions. In principle, one can obtain femtosecond electron pulses from previous generations of electron gun designs by using very low numbers of electrons per pulse [8]. In this case, the brightness of the electron source is not sufficient to fully resolve structural details under nonreversible conditions encountered for sample configurations that support femtosecond time resolution. Samples need to be thin enough to avoid geometrical time broadening from the velocity mismatch between the electron and laser pulses, and finite sample dimensions limit the total number of laser shots

144

{vide infra). The challenge is to come up with an electron pulse source that is bright enough to provide full structural details under near single shot conditions. The severe space-charge effects inherent to a high-brightness femtosecond electron pulse broaden the pulse as it propagates. For this reason, femtosecond electron sources cannot be designed independently of the experimental setup: ours is optimized to minimize the electron pulse propagation distance to confine the pulse to the femtosecond domain. As a result, we were able to follow, with sub-Angstrom spatial resolution and femtosecond temporal resolution, the atomic-level details of the laser-induced melting of a thin aluminum film. We are undertaking fiirther development of the electron source in order to decrease the electron pulse duration to the low hundreds of femtoseconds while maintaining the high brightness needed in these experiments. There are complementary efforts ongoing in short-pulse x-ray generation in which it is simpler to achieve subpicosecond pulses [9,10]. The difficulties in generating short electron pulses, however, are offset by the factor of 10^ higher electron scattering cross-section [11] and by the lower incidence, by a factor of 10^ [12], of inelastic scattering causing sample damage. The large cross section for electron scattering allows us to record a wide range of scattering vectors using the current source, leading to a diffraction pattern that can be Fourier-transformed to arrive at real-space atomic configurations.

2.

Experimental Results

For the first study with this new electron source, we chose the simplest possible structural transition: melting from an ordered solid to a disordered liquid. An intense laser pulse generates a hot electron gas that must re-equilibrate with the cold lattice, a process that is typically described within the framework of the twotemperature model (TTM) [131. Previous optical measurements have shown that under high fluence (70 mJ/cm ) the aluminum dielectric constant took on liquidlike values within 500 femtoseconds of excitation, leading to the suggestion that the aluminum had nonthermally melted [14]. Connecting changes of the dielectric constant to changes of the nuclear configuration of these highly nonequilibrium states is an extremely difficult task. It is, however, the nuclear configuration that provides the unambiguous distinction between the ordered solid and the disordered liquid. By using femtosecond electron diffraction, we are able to avoid this difficulty by directly probing this quantity of interest. Real-space structural information is provided by Fourier transformation of the experimental diffracted intensity, l{s\ to obtain the reduced density function, G(r), as shown in Equation (1). This function describes the deviation of the local density from the average atomic density as a function of the radial distance from an average atomic origin; such a description is the key to resolving the atomic motions involved in the phase transition.

G{r) = %7r\s ^ ^ ^

sm{27tsr)ds

(1)

0 ^J where N converts the observed intensity to absolute electron scattering units, /

145

is the atomic scattering factor, po is the average density and 5=2sin(0)/X. is the scattering vector with A. the electron wavelength. By using high brightness femtosecond electron pulses, we were able to watch as the lattice heated and then disordered in 3.5 picoseconds to the liquid. The observed dynamics, captured with atomic detail along the disordering coordinate, were consistent with a thermally propagated phase transition that could be described within the framework of a thermally activated barrier crossing [1,2]. We are currently in the process of investigating the detailed mechanism of strongly driven melting. The use of short pulsed lasers as a driving force offers the unique opportunity to investigate melting under extreme conditions of superheating. Melting of the solid via heterogeneous nucleation of surface liquid zones that then propagate at the speed of sound into the bulk is already wellknovm. Homogeneous nucleation is a more exotic phenomenon that occurs with greater likelihood the greater the superheating, giving a ftuence-dependent melting timescale [15] A third mechanism for melting—a nonthermal route—occurs due to destabilization of the lattice following excitation of a large number of the electrons in the sample. We have shown this mechanism not to be a factor in the melting of either gold or aluminum. Heterogeneous and homogeneous melting, however, are not exclusive of each other and the timescale we observed for melting in aluminum was consistent with either mechanism. As a consequence it is necessary to perform a fluence dependence to gain greater separation of the timescales. Many of the details of the melting process are strongly material dependent. In an attempt to more fully investigate the physics of melting we have undertaken to study the fluencedependence of the melting timescale in gold films. The heating rates in our aluminum experiments were consistent with the TTM calculations up until the sudden departure due to the onset of melting [1,2]. The lower rate of heating we see with gold is consistent with these observations. For the case of gold, the melting once more clearly proceeds via a thermal route—^as shown by the characteristic Debye-Waller damping of the diffi-action peaks (Fig. 1) and on a reasonable timescale. Our preliminary results indicate that gold melts within 12 picoseconds for an absorbed fluence of 14.9 mJ/cm"^. Both the onset time and the rate of melting are in good agreement with detailed molecular dynamics simulations of ultrafast melting of gold films which show homogeneous nucleation playing a strong role in the melting process [16]. Further work to quantitatively compare our experimental results to these simulations is ongoing.

3. Discussion These early experiments on fully nonreversible systems have demonstrated the promise of the technique, but they have also underscored the difficulties facing its application to molecular systems. Our experiments are carried out using transmission electron diffraction to prevent degradation of the experimental temporal resolution due to velocity mismatch over a large pumped volume [17]. This requires the use of thin (-100 nm) films, placing constraints on the maximum available free-standing sample area. Under the strongly driven conditions of the melting experiments, each laser pulse irreversibly damages the excited sample

146

area, which necessitates that the sample be translated to a fresh area between shots. Heat deposition from the exciting laser pulse in thin films of otherwise reversible systems may also lead to irreversible sample damage. The importance of the electron brightness under such conditions can be seen by examining Equation (1). Experimental parameters restrict the upper limit of integration to a value ^maxFor femtosecond electron diffraction, the value of ^^ax is tied to the electron source brightness and the effect of fewer scattering electrons is to prevent the emergence of the higher-5 peaks from the baseline noise, with consequent deleterious effects on G(r). Sample area limitations thus place the burden of obtaining the requisite signal on ftirther developments in higher brightness and sensitivity rather than accumulation over many shots.

"^^^

^^^SS^ZZl

<1

3500

S 3000

s (A')

s (A 1

Fig. 1. Diffraction patterns from 20 nm thick polycrystalline gold films absorbing 14.9 mJ/cm^ of 200 fs, 387 nm excitation to give a maximum achievable lattice temperature of 2.4 times the melting point. The legend shows the time delay in picoseconds and the effect of the low electron-phonon coupling constant is clear in the length of time the gold requires to melt. Fig. (a) shows the raw diffraction pattern while (b) is weighted by s^ in order to emphasize the Debye-Waller damping of the high-5 peaks during heating. The broadened 111 peak at longer delay time is characteristic of the formation of liquid gold. In addition to the challenge of developing higher brightness electron sources, new advances are required in order to measure the electron probe duration and time origin. The electron pulses used in Fig. 1 were determined to be 600 ± 100 fs through a combination of streak camera measurements and theoretical modeling [2,3]. The pulse durations were further confirmed through the experimentally discemable subpicosecond dynamics observed in the Al lattice dynamics [1,2]. Ideally the electron pulse duration would be determined by an independent and direct method at the sample position, but it is beyond the capability of streak camera technology alone to provide these measurements due to the short temporal duration and propagation characteristics of the high-brightness femtosecond electron pulses generated by our source. Further advances in pulse duration will only exacerbate the situation. In analogy with the technique of alloptical cross-correlation, we therefore propose to employ the ponderomotive potential of an intense, well-characterized optical pulse to measure both our electron pulse duration and zero of time [18]. The ponderomotive force,

F.

e A ^TT^ms^c'

I{r^t)

(2)

147

where / i s the intensity, X is the wavelength, c is the speed of light, and e and m are the electron charge and rest mass, can be used to alter the electron pulse profile [18-20]. By appropriate configuration of the interaction and detection schemes, scanning the delay between the two pulses effectively performs an electron-laser pulse cross-correlation. Numerical simulations have been performed that confirm the ability of this technique to measure electron pulse durations as short as 100 femtoseconds in the energy range 10-300 keV [18]. This method clearly holds promise as femtosecond electron optics for diffraction continues to develop. The above illustrates a new regime in electron pulse generation as well as steps towards developing an analogous set of diagnostics to those used in alloptical measurements for achieving pulse-width limited time resolution to dynamical events. The technology for capturing transition states—^making the so-called "molecular movie"—^based on time-resolved electron diffraction is now at hand.

References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22.

148

B.J. Siwick et al. Science 302, 1382 (2003). B.J. Siwick et al. Chemical Physics 299, 285 (2004). B.J. Siwick et al, J. AppL Phys. 92, 1643 (2002). G. Mourou and S. Williamson, Appl. Phys. Lett. 41, 44 (1984). A.A. Ischenko et al., Appl Phys. B 32, 161 (1983). R. Srinivasan et al, Helv. Chim. Acta 86 1763 (2003). R.C. Dudek and P.M. Weber, J. Phys. Chem. A 105, 4167 (2001). J. Cao et al, Appl. Phys. Lett, 83, 1044 (2003). A. Cavalleri et al, Phys. Rev. Lett. 87, 237401 (2001). K. Sokolowski-Tinten et al, Phys. Rev. Lett. 87, 225701 (2001). B.K. Vainshtein, Structure analysis by electron diffraction (Macmillan, New York, 1964). R. Henderson, Q. Rev. Biophys. 28, 171 (1995). M.L Kaganov, LM. Lifshitz and L.V. Tanatarov, Sov. Phys. JETP 4, 173 (1957). C. Guo et al, Phys. Rev. Lett. 84, 4493 (2000). B. Rethfeld, K. Sokolowski-Tinten and D. von der Linde, Phys. Rev. B 65, 092103 (2002). D.S. Ivanov and L.V. Zhigilei, Phys, Rev. B. 68, 064114 (2003) C.-Y. Ruan, et al, PNAS 101, 1123 (2004). B.J. Siwick et al. Opt. Lett., submitted. D.S. Ivanov and L.V. Zhigilei, Phys. Rev. B. 68, 064114 (2003) A.V. Gaponov and M.A. Miller, Sov. Phys. JETP 7, 168 (1958). T.W.B. Kibble, Phys. Rev. 150, 1060 (1966). P.H. Bucksbaum, M. Bashkansky and T.J. Mcllrath, Phys. Rev. Lett, 58, 349 (1987).

Absolute Displacement Interferometry of Ultrafast Laser-Produced Plasma Expansion George Rodriguez, Steven A. Clarke, and Antoinette J. Taylor Condensed Matter and Thermal Physics Group, Mail Stop K764, Los Alamos National Laboratory, Los Alamos NM 87545 E-mail: [email protected] Abstract. Microscopic interferometric measurement of plasma expansion and surface motion during the laser-heating period of a metal target using a single ultrafast pump pulse is reported. A spatial resolution of approximately a few nanometers is achieved.

1.

Introduction

In this paper we report on the development of a novel technique to measure the critical surface displacement in intense, ultrashort, laser-solid target experiments. Determination of the critical surface position is important for diagnosing short scale length {L « X) plasma expansion and hydrodynamic surface motion from short pulse laser-heated solid targets [1-4]. This technique directly measures surface displacement using a single ultrafast laser pump pulse without the need for a separate interacting probe pulse, and we highlight the ability of this technique to resolve surface displacement and expansion occurring during the laser pump pulse and with a capability to study the dynamic heat pumping process by varying the pump laser pulse width. By varying the pump laser pulse width (50 fs - 700 fs), we explore the dependence of the deposition process and the subsequently measured displacement over a range of plasma scale lengths (k/200
2.

Experimental

The technique, which we call absolute displacement interferometry (ADI), is based on the Fourier plane spatial imaging. The ADI operates by collecting and collimating focused light from a laser heating beam striking an ablation target and imaging the reflected beam through a double pinhole interferometer (see Figure 1). During ablation of a laser heated target, plasma generation and subsequent hydrodynamic expansion introduces a wavefront phase change at the input plane of double pinhole mask, and the ADI records the shifting of the diffraction and interference pattern to produce a "dynamic" fringe shift which is correlated with surface displacement. The experimental optical layout is shown in Figure 1. Pulses from an amplified 0.5 TW Ti:Al203 laser system are sent to a target chamber for production of a plasma off a solid target surface. The laser operates at 800 nm and produces 50-fs 25-mJ pulses at a 10 Hz repetition rate. An f/10 gold

149

parabolic concave mirror with a focal length of 25 mm is used to focus an approximately a 1-cm diameter laser beam down 60 |im in diameter to produce a maximum peak focal intensity of 10^^ W/cm^ on target. After collection and collimation, the reflected laser beam is sent to the double pinhole ADI outside the target chamber. A fringe pattern is generated after crossing the pinhole diffracted beams at the image plane and only after e a c h p a t h l e n e t h is adjusted for t e m p o r a l delay,

^^§- ^ • Target chamber and experimental layout of the absolute displacement interferometer (ADI).

Images are recorded using a shuttered 16-bit CCD camera. To make a displacement measurement, a reference low power fringe pattern must be recorded followed by a high power dynamic shot. In Figure 2 we demonstrate our fringe pattern image quality by showing the results for such a set of images taken for an aluminum target at two intensities -——WTW—--^^^ The fringe shift recorded between the ^ I ^ ^ H y m n l ^ l (**) low and high power shot is then computer processed to yield displacement as the surface expands. Static displacement to fringe ^^Bi]t|!tl!|i|!^^H \\ calibration is performed by manually ^ ^ H ^ j m i y ^ ^ l ffi^ translating the target surface and then _ _ „ , _ _ _ _ _ ^ ^ ^ _ measuring the frmge shift on the I D i S l ^ j S ^ ^ ^ H (^) camera. A typical static displacement calibration is 100 nm per fringe with a camera pixel calibration of-30 pixels per fringe. ^ ^ H l U l l l l l l t ^ ^ H :iii||l'^^

3.

Results

Using the above described procedure, we benchmark our technique using aluminum targets irradiated over a laser intensity range of 10^^ to 10^^ W/cm^ and apply a 1-D Lagrangian hydrodynamic model (HYADES) that incorporates a Heknholtz wave equation solver to describe the laser

150

Fig. 2. Output camera images and horizontal line out graphs of the absolute displacement interferometer for two successive 500-fs laser pulses. The camera images (a) low intensity (reference), 10^^ W/cm^ and at a (b) high intensity (dynamic) of 10^^ W/cm^ are for M metal. The high intensity image produces a fringe shift relative to the low intensity data. The corresponding line out graphs for each image (c) reference and (d) dynamic are taken from the middle row of each image.

interaction and subsequent motion of the expanding plasma created on the surface.

0

25

50

75

100

Time(fs)

125

150

175

200

10"

lO"

Intensity (W/cm")

Fig. 3. (a) Experimental displacement data (line + symbol) for a 100-fs pulse at 8x10^"^ W/cm^ intensity on an aluminum target. We also show HYADES hydrodynamic results for a 100-fs pulse at 8x10^"^ W/cm^ (thin line) and 1x10^^ W/cm^ (thick line), (b) Experimental measurements of the surface displacement data versus intensity for a series of pulsewidths: 50 fs (triangles), 100 fs (squares), and 700 fs (circles). The solid lines are corresponding HYADES calculated critical density position values versus intensity.

In addition to varying the laser pulse energy, variation of the laser heating period is also accomplished by adjusting the laser pulse width from 50 fs to 700 fs. Varying both, the laser pulse duration and pulse energy, explores the surface expansion dependence on energy deposited and peak intensity. It also gives insight on to the linearity of the initial expansion, smce a longer pulse records the expansion over a longer period of time. Figure 3(a) shows an example displacement plot versus time for a 100-fs pulse at 8x10^"^ W/cm^. For comparison with our hydrodynamic model, we also plot the computed critical density (Ucnt = 1.8x10^^ cm'^ at 800 nm) position versus time for a couple of intensities, 8x10^"^ W/cm^ and 1x10^^ W/cm^. We also measure the total displacement for a given intensity by taking the value obtained at the \le^ trailing edge time point in the laser mtensity temporal profile. This is indicated by an arrow in Figure 3(a) which shows data displacement of-21 nm for a 100-fs pulse at 8x10^^ W/cm^. Using this method, we plot in Figure 3(b) the ti/e2 displacement data versus intensity for a series of pulse widths: 50 fs, 100 fs, and 700 fs, and compare them with the hydro-model using the critical density position also at the ti/e2 displacement time position. Although there are slight deviations with the data, there is reasonable agreement between the data and model.

References 1 R. J. Kingham, P. Gibbon, W. Theobald, L. Veisz, and R. Sauerbrey, Phys. Rev. Lett. 86, 810,2001. 2 P. Blanc, P. Audebert, F. Fallies, J. P. Geindre, J.-C. Gauthier, A. Dos Santos, A. Mysyrowicz, and A. Antonetti, J. Opt. Soc. B 13, 118, 1996. 3 X Liu and D. Umstadter, Phys. Rev. Lett. 69, 1935, 1992. 4 K. Widmann, G. Guethlein, M. E. Foord, R. C. Cauble, F. G. Patterson, D. F. Price, F. J. Rogers, P. T. Springer, R. E. Stewart, A. Ng, T. Ao, and A. Forsman, Phys. Plasmas 8, 3869, 2001.

151

Electron acceleration through spatiotemporal shaping of ultrashort light pulses Darius H. Torchinsky\ T. Feurer^, and Keith A. Nelson^ ^Department of Physics, MIT, Cambridge, MA 02139 ^Department of Chemistry, MIT, Cambridge, MA 02139 Summary. We report on a novel scheme based on spatiotemporal pulse shaping to accelerate electrons in vacuum to high kinetic energies using the transverse component of the electromagnetic field. Relatively high energies on the order of 100 MeV can be reached with only modest laser intensities (^ 10^^ W/cm^).

1 Introduction High power lasers with focused intensities of 10^^ W/cm^ or even more are able to produce field strengths in excess of 10^^ V / m , thereby presenting themselves as interesting alternatives to present day accelerator devices which are limited to about 10^ V / m . Due to the enormous field strengths produced,.-such laser based accelerators can only operate in vacuum far from any mirrors, lenses, or other optical elements. Many suggestions have been published as to what extent vacuum acceleration of electrons is feasible. Most propos^*ls require a longitudinal component of the electric field, and the numerous g[eometries published in the literature seek to maximize the longitudinal component in a localized volume. Here we propose a scheme that, in contrast to previous efforts, utilizes predominantly the transverse component of the electric field and relies solely upon the technique of spatiotemporal pulse shapii^-g. [1]

2 Results and Discussion When a free electron interacts with a linearly polarized laser pulse, of modest intensity, it oscillates in the direction of polarization but returns tt^ rest after the laser pulse has left the interaction region - each successive half cycle of the laser's field exerts a force with alternating sign resulting in zero net work done on the electron. Now let us assume an electron initially at re??t with its initial position coincident with the maximum of the shaped electfomagnetic

152

field. The field is shaped such that ever3^where along its trajector,v the electron experiences the same maximum field strength and direction,
Fig, 1. Electron acceleration scheme. The electron's motion is locked to the shaped waveform, such that the electron surfs on a line of constant phase. Figure 1 illustrates the proposed scheme, showing the lines of constant phase of the ideal shaped laser pulse and the position of the electron at three different instants in time. It is possible to realize the condition B{x^z^t) = EQ at the instantaneous position of the electron by properly adjusting the combination of the group delay Ag{x) and the phase delay Ap{x). In order to compute the delays which render E{x^ z^ t) = EQ at the instantaneous location of the electron, we consider an electron propagating in a pair of crossed, static electric and magnetic fields E{x) and B{x) and then solve for the dHays A{x) that exactly realize this situation. In terms of the kinematic variables t{x) and z{x) the delay is A{x) = t{x) — z{x)/c. By solving the relativistic equations of motion for fields that only vary along the x direction with profik g{x)^ we find the electron's trajectory in the laboratory frame. The delay then is A{x)

(1) i

^(c/?0x70r.)2 + ^ 7 0 r . / ( 3 ; ' )

where mo is the electron's rest mass, e the elementary chargt?, c is the speed of light in vacuum and f{x) ^ JQ dx' g{x'). The solution incorporates initial velocities in both the x and the z direction. Although equation (1) allows one to calculate the necessary delay for any given field dependence, in the following we assume that it is constant on the interval x G [O.rcmax] and zero otherwise. This yields a relatively simple analytic solution foi 'A{x) and drastically reduces the computation time integrating the equations'^f motion.

153

To this point, we have been concerned with maintaining the ek^ctron precisely at the field extremum, which is a point of unstable equilibriuni. In a real acceleration experiment, the driving field suffers random fluctuations which may serve to severely reduce the efficacy of the scheme. The simulation show that the electron must be born with essentially ideal parameters or else it will not be accelerated. However, if the delay is calculated for a field I'^vel somewhat lower than the maximum field, then the acceleration is quite robust with respect to initial electron position or field fluctuations. The robustness improves further as the field strength EQ increases. We now have to show that one can actually generate a spatiot-emporally shaped waveform that is close to ideal, satisfies Maxwell's equations, and accelerates the electron to the desired energy. A very intuitive approach is to split the dispersed spectrum of the incoming beam into N equally wide parts, thus producing N pulselets which have different center frequencies and are A^ times as long as the original pulse. The 2D pulse shaping apparatus is used to adjust the spatial phase as well as the spectral phase of ea^'h pulselet separately. A linear phase sweep along the spatial dimension will displace the pulse vertically and a quadratic phase sweep will shift its focal position along the z direction. Additionally, a linear phase sweep along the spectral dimension will delay the pulselet in time. Each pulselet is described by a symmetrized Gaussian pulse correct to fifth order. [2] By adjusting the temporaWelays and the spatial position of all N pulselets appropriately, the desired waveform can be approximated.

3 Conclusions Two-dimensional pulse shaping has been proposed to effectively accelerate electrons to high kinetic energies. We have shown theoretically that this scheme is relatively robust and insensitive to fluctuations of the initial electron conditions and of the electric field itself. For realistic waveforms with approximately 10^^ W/cm^ field intensities, energies up to several 100 MeV may be achieved. Acknowledgements. We would like to acknowledge stimufeting and fruitful discussions with B. Graves, B. Zwiebach, and A. Torchinskj Financial support was provided through NSF Grant No. CHE-0212375.

References 1. T. Feurer, J.C. Vaughan, R.M. Koehl, K.A. Nelson, Opt. Lett. 77, 652-654 (2002). J.C. Vaughan, T. Feurer, K.A. Nelson, JOSA B 19, 2489 (2002). J.C. Vaughan, T. Feurer, K.A. Nelson, Opt. Lett. 28, 2408-2410 (20O4). 2. L.W. Davis, Phys. Rev. A 19, 1177-1179 (1979).

154

Mapping attosecond electron wave packet motion Hiromichi Niikura, D. M. Villeneuve, and P. B. Corkum National Research Council of Canada 100 Sussex Dr., Ottawa, Ontario Kl A 0R6 Canada Hiromichi.Niikura(a).nrc.ca

Abstract. We show, by numerical calculations, that attosecond bound electron wave packet dynamics can be measured from the spectrum of the attosecond optical pulses generated in the medium undergoing the dynamics. One of major aims for attosecond science is to observe electron dynamics. Recently, two types of electron wave packet motion have been observed with an attosecond time resolution. One is an Auger ionization dynamics excited by a single photon XUV laser pulse with ~ Ifs pulse duration [1]. The Auger electron is streaked by another delayed intense laser pulse that probes the time of ionization. The other is a re-collision electron dynamics during intense field ionization. The vibrational wave packet motion of H2^ produced simultaneously with the electron wave packet clocks the motion of the re-colliding electron[2]. However, in both cases, the electron that has been observed is the continuum wave packet. A method to observe the electron dynamics in a bound potential with an attosecond time scale has not been established.

We show a method of probing internal electron wave packet motion. Applying infrared, intense laser fields to the target atoms or molecules, we ionize the small part of the internal electron wave packet. When the fields change its sign, the ionized electron wave packet returns to the ion core with high kinetic energy. The dipole interaction between the re-collision electron wave packet and a remaining, internal electron wave packet generates the high harmonics whose spectrum contains information of the internal wave packet motion. We demonstrate this process by numerical calculations assuming a phase fixed few-cycle laser pulse. The initial wave packet {y/(t)) is prepared by a coherent sum of two electronic wave functions with an appropriate initial phase: y/(t) = y/o + y/i exp (-ixAE^t/h), where i//o^ / is a field-free wave function at the ground and first excited state, respectively, and AE is the energy separation between them.

155

The laser intensity is '-IxlO^'^W/cm^ and the pulse duration is '^6 fs. The carrier phase is fixed so that only one re-collision occurs, E(t)=sin(w(t-to))exp(-(ttof/2<^), where to^ 8fs. Solving a one-dimensional time-dependent Schrodinger equation for the electron under the laser fields, we calculate the high harmonic spectrum. Figure 1 plots the calculated high harmonic spectra with three different period of internal electron motion, (a) 290 as, (b) 330 as, and (c) 444 as. Compared to the spectrum when only the first excited state is initially populated (dotted line), the spectrum has a periodic intensity modulation.

(a)290 as

20

40 60 80 Photon energy / eV

100

Figure 1 plots the calculated high harmonic spectra when the bound electron motion has a period of (a) 290 as, (b) 330 as, and (c) 440 as. The kinetic energy of the re-collision electron, thus radiation photon energy, is uniquely related with time when only one re-collision occurs during the laser pulse. Thus, from the modulation period, we know the period of the electron motion. From the relation between radiation energy and time, we found that

156

radiation intensity is suppressed when the electron wave packet counterpropagates to the direction of returning electron wave packet. The period of the electron wave packet can be also measured by a conventional pump-probe way. Figure 2 plots the intensity map of the spectrum as a function of pump-probe delay for the same condition in Fig. 1 (b), but using 800 nm, 3fs. Although only few intensity dips are found, the period is measured by the intensity modulation as delay changes.

I I 1E-19 0

10

20

30

40

50

Photon energy / eV Figure 2 plots the intensity distribution of high harmonic spectrum as a function of the pump-probe delay. As the delay changes, the modulation dip shifts periodically. In conclusion, the re-collision electron is a powerful tool for observing the dynamics not only the vibrational motion but also the electron motion with an attosecond time resolution because of its coherence. We can detect the electron motion in a sub-laser-cycle time scale. References [1] M. Dresher et aL, Nature (London) 419, 803 (2002) [2] H. Niikura et aL, Nature 417, 917 (2002)

157

Quasi-monoenergetic electron beam generation in laser-driven plasma acceleration E. Miura\ K. Koyama\ M. Adachi', S. Kato\ Y. Kawada', S. Masuda^ T. Nakamura"^, N. Saito^ and M. Tanimoto^ National Institute of Advanced Industrial Science and Technology (AIST), Tsukuba Central 2, 1-1-1 Umezono, Tsukuba, Ibaraki 305-8568, Japan E-mail: [email protected] 1

3

" Hiroshima University, 4

University of Tsukuba, 5

National Institute of Radiological Sciences, Meisei University Abstract. The generation of a quasi-monoenergetic electron beam with an energy peak at 7 MeV was demonstrated in laser-driven plasma acceleration.

1. Introduction Laser-driven plasma acceleration has been intensively studied[ll since the proposal of the fundamental concept[2], because the realization of a compact accelerator is expected due to a very large acceleration field. In addition, the generation of an ultrashort electron bunch in femtosecond region is also expected Although the generation of energetic electrons over several hundreds MeV has been observed[3], the energy spectra of accelerated electrons have so far been Boltzmann-like or power-law distributions and the energy spread has been wide. The key issue in realizing an advanced compact accelerator based on laser-driven plasma acceleration is the generation of a monoenergetic beam. In this paper, we report the generation of a quasi-monoenergetic electron beam from a plasma produced by an intense Ti:sapphire laser pulse.

2. Experimental results and discussion A 2-TW Ti:sapphire laser pulse (center wavelength:800 nm) with 50-fs duration was focused on a supersonic nitrogen gas jet. The laser intensity was 5x10^^ W/cm-. The molecular density at the gas jet center was 1.3x10^^ cm'^. The electron density of the plasma was estimated to be 1.3x10"° cm'^ according to the barrier suppression ionization model [4]. An energy spectrum of an electron beam emitted in the forward direction of the laser propagation was observed. The observed energy range was from 0.2 to 30 MeV. To investigate the acceleration mechanism, a spectrum of laser light transmitted through a plasma and a sidescattered light image were observed. Figure 1(a) shows an energy-resolved electron image obtained by the accumulation of 90 laser shots. A small spot around 7 MeV shows the emission

158

of a quasi-monoenergetic beam with small divergence. Figure 1(b) shows the electron energy spectrum. The energy spread of the quasi-monoenergetic beam at 7 MeV was 30%. The observed energy spread was determined by the low energy resolution of the electron spectrometer. The energy spread of the quasimonoenergetic beam may be much narrower. The divergence of the quasimonoenergetic beam was estimated to be 2.4° from the vertical size of the small spot in the electron image shown in Fig. 1(a).

5 10 15 20 25 Electron Energy [MeV]

30

F i g . 1. (a) Energy-resolved electron image, (b) Electron energy spectrum with a quasi-monoenegetic component.

Figure 2(a) shows a spectrum of laser light transmitted through a plasma The first Stokes sideband of stimulated Raman forward scattering was observed at 1030 nm. This is the evidence for the excitation of a plasma wave. From the wavelength of the sideband, the electron density of the plasma wave excitation region was estimated to be 1.3x10^° cm"^ which corresponds to the electron density at the gas jet center. When the Stokes sideband appeared, an intense sidescattered light image with a fishbone structure was observed around the gas jet center as shown in Fig. 2(b). The wavelength and the polarization of the sidescattered light were the same as those of the incident laser light. It is thought that the side-scattered light is due to the coherent Thomson scattering induced by a plasma wave. The length of the plasma wave excitation region was estimated to be 500 pim from the length of the image with a fishbone structure Here, we discuss the mechanism of the quasi-monoenergetic beam generation. As described above, it is obvious that the electron acceleration by a plasma wave plays a main role. The phase velocity of the plasma wave is slower than the velocity of accelerated electrons. If the plasma wave excitation region is sufficiently long, trapped and accelerated electrons in the plasma wave outrun the acceleration phase and move into the deceleration phase. That is dephasing. At the electron density of 1.3x10^° cm'^, the dephasing length was estimated to be 68 pim, which is much shorter than the observed length of the plasma wave excitation region of 500 ptm. In such case, the acceleration and the deceleration can be repeated On the other hand, a plasma wave induces both longitudinal and radial fields. The radial field can push out trapped electrons from the region where the plasma wave was excited especially in the deceleration phase. The loss of trapped electrons in the deceleration phase has been suggested by the simulation reported

159

in Ref. [5]. The loss of trapped electrons can be repeated, because the length of the plasma wave excitation region is far beyond the dephasing length. The repeat of the loss of trapped electrons may limit the energy spread and spatial spread. 10000 b

^

•

.

•

I

•

.

•

I

•

I

1000

10

600 700

800 900 1000 1100 1200 Wavelength [nm]

Nozzle Exit'

F i g . 2 . (a) Spectrum of laser light transmitted through a plasma, (b) Side-scattered light image observed through an interference filter of 800 nm.

3. Conclusions We demonstrated the generation of a cpaasi-monoenergetic electron beam with a peak at 7 MeV from a plasma produced by an intense Ti:sapphire laser pulse. The beam divergence was 2.4°. The quasi-monoenergetic beam generation may be due to the loss of electrons trapped in a plasma wave, which was brought about by the excitation of the plasma wave far beyond the dephasing length. Acknowledgements. A part of this work was financially supported by the Budget for Nuclear Research of the MEXT and the Advanced Compact Accelerator Development Program of the MEXT.

References 1 D. Umstadter, J. Phys. D 3 6 , R151, 2003, R. Bingham, J. T. Mendonga, and P. K. Shukla, Plasma Phys. Control. Fusion 4 6 , R l , 2004, and references therein. 2 T. Tajima and J. M. Dawson, Phys. Rev. Lett. 4 3, 267, 1979. 3 V. Malka, S. Fritzler, E. Lefebvre, M. M. Aleonard, F. Burgy, J. P. Chambaret, J. F. Chemin, K. Krushelnick, G. Malka, S. P. D Mangles, Z Najmidon, M. Pittman, J. P. Rousseau, J. N. Scheurer, B. Walton, andA. E. Dangor, Science 2 9 8, 1596, 2002. 4 S. Augst, D Strickland, D. D Meyerhofer, S. L. Chin, and J. H. Eberly, Phys. Rev. Lett. 6 3,2212, 1989. 5 S. Y. Chen, M. Krishnan, A. Maksimchuk, R. Wagner, and D Umstadter, Phys. Plasma 6, 4739, 1999.

160

Control of multiphoton ionization processes in aligned I^ molecules by optimizing timedependent polarization of femtosecond pulses Takayuki Suzuki, Shinichirou Minemoto, Tsuneto Kanai, and Hirofumi Sakai Department of Physics, Graduate School of Science, The University of Tokyo, 7-3-1, Hongo, Bunkyo-ku, Tokyo 113-0033, Japan E-mail: [email protected] Abstract. Multiphoton ionization processes in aligned I2 molecules are actively controlled by optimizing time-dependent polarization of femtosecond pulses. Thereby, both external and internal degrees of freedom in molecules are simultaneously controlled for the first time. Since its first proposal of feedback quantum control [1], many demonstrative experiments have been performed for molecular systems. However, in experiments performed in gas-phase molecules, they are randomly oriented. This limits the outcome of optimal control experiments. Here, we demonstrate optimal control experiment where a sample of aligned molecules [2,3] is employed for the first time to our knowledge. Another important feature of the present experiment is that a time-dependent polarization pulse is used as a new control parameter. Here, we pay close attention to the multiphoton ionization processes in l^ molecules. The molecules can ionize sequentially or nonsequentially [4]. The nonsequential double ionization is caused by recollision between a parent ion and an electron produced by tunnel ionization. For the recollision to occur, the electron has to be driven by the laser field along the linear polarization and gain sufficient kinetic energy to knock another electron in the parent ion. If the polarization deviates very much from linear, the electron misses the parent ion. As a corollary of nonsequential double ionization, we can expect that it contributes more to the production of evenly-charged molecular ions such as l/^ than to that of oddly-charged molecular ions such as (1/ and) l/^. It means that we can expect a correlation between the polarization and the production efficiency of evenly- or oddly-charged molecular ions. The experimental setup is shown in Fig. 1 and fuller descriptions can be found in our recent papers [5,6]. A pulsed supersonic beam of \ molecules is introduced into a time-of-flight (TOF) mass spectrometer. In order to align l^ molecules along the TOF axis, we use pulses from an injection-seeded Nd:YAG laser. Intense femtosecond Ti:sapphire laser pulses are used to ionize the I^ molecules. The fragment ions produced by photodissociation and Coulomb explosion are accelerated by a static electric field toward a microchannel plate detector positioned on-axis with the TOF axis. We adopt a fitness F defined by F = /(l,l)/(/(l,0)+/(l,2)+/(2,l)), where /(m,n) stands for the integrated V^ signals of the fragmentation channel i^^^+f ^ produced from I^^™^"^"^. We measure the ellipticity dependence of the fitness F with the combination of a A/4 plate and a A/2 plate. The major axis of the elliptical

161

polarization is kept parallel to the TOF axis, i.e., the molecular axis of aligned I^ molecules (See Fig. 1 and the orientation angle ^ = 0 in the present case.) [ Pulse shaper ]

Ti sapphire(probe' wavelength A ~ 80C nin pulse width T~ 4f ft peak intensity / = 2 > lO''' W/cm^

NdYAG( alignment; wavelength / = 1064 nm pulse width T~ 12 ns peak intensity 7 = 2 > lO'^ W/cm'

• TOF axis

[ Optimization

- - - « . . . . . , , ^ _ T ^ U Moiecmar Deam Detector

[ TOF]

n i^"i^ti'i V ®@5^

Electrodes spectrometerj

[^"JF^"'|

Fig. 1. An experimental setup. EUipticity f and the orientation angle ^are also illustrated. The ellipticity e varies from £"=0 (linear polarization) to ^ =1 (circular polarization). The fitness Fis 0.215 for linear polarization and gradually decreases down to 0.195 for circular polarization. It means that the relative production efficiency of evenly-charged molecular ions (l/^) is high for linear polarization and that of oddly-charged molecular ions (1/ and l/"") is high for circular polarization, which is qualitatively consistent with our hypothesis. The contrast defined by (the maximum i^/(the minimum /^ is -- 1.1. Then we optimize the time-dependent polarization of the ionization pulse to maximize or minimize the fitness F with the homemade learning-loop optimal control system where a time-dependent polarization pulse can be generated and controlled. We use a typical 4-/configuration pulse shaper but remove polarizers to generate a time-dependent polarization pulse. We employ the genetic algorithm (GA) as an optimization algorithm and search for an optimal time-dependent polarization pulse. We use the ellipticity for each frequency component as the only parameter in GA. We maximize and minimize the fitness F. The maximum (minimum) fitness obtained is 0.34 (0.14) and the contrast is ~ 2.5 which is much larger than that obtained with a fixed ellipticity (1.1). Our results show much better ability to control the charged state of molecular ions with time-dependent polarization pulses than with elliptically polarized pulses having a fixed ellipticity. We characterize the optimal pulses with the technique known as POLLIWOG [7]. The results are shown in Fig. 2. When the fitness /^ is maximized, i.e., the production efficiency of l/^ is enhanced with that of oddly-charged molecular ions suppressed, the pulse becomes linear polarization parallel to the TOF axis (^=0) at around the peak of the pulse. Such a polarization corresponds to the one which we consider to be a favorable situation for the production of the (1,1) channel due to recoUision. On the other hand, when the fitness F i s minimized, i.e., the production efficiency of oddly-charged molecular ions is enhanced with that of l/^ suppressed, the ellipticity stays small {e- 0.1) from -60 fs to 4-30 fs, i.e., over the main part of the pulse. Here, we note that the orientation angle changes more rapidly than in the maximized case and it goes through zero just as the case with the fitness F

162

maximized. Qualitative features of the optimized polarizations are in consistency with our expectations. However, the ellipticity of ~ 0.1 over the main part of the pulse optimized to minimize the fitness Fis rather small compared to the results obtained with wave plates. The results suggest the importance of the time evolution of the potential energy surface associated with the increase of the internuclear distance as well as the electrons' dynamics explained in the introduction.

•TOFaxis -100

C

Time [fs^

100

100

C

100

Time (fs^

Fig. 2. The time-dependent polarizations and the intensity profiles of the optimized pulses. In the upper panels, both ellipticity e (solid curve, left ordinate) and the orientation angle 0 (dashed curve, right ordinate) are presented. The sign of £ is related to the helicity of the elliptical polarization. The lower panels show the instantaneous intensity profiles. The left and right panels correspond to the optimized pulses which give the maximum and the minimum F, respectively. We find a correlation between a femtosecond time-dependent polarization pulse and the production efficiency of evenly- or oddly-charged molecular ions using a sample of aligned molecules. Much better controllability of the ionization processes is achieved with a time-dependent polarization pulse than with a pulse having a fixed ellipticity. Using a sample of aligned molecules and a timedependent polarization pulse, both external and internal degrees of freedom in molecules are simultaneously controlled [8].

References 1. 2. 3. 4. 5. 6. 7. 8.

R. S. Judson and H. Rabitz, Phys. Rev. Lett. 68, 1500, 1992. H. Sakai et al, J. Chem. Phys. 110, 10235, 1999. J. J. Larsen et al, J. Chem. Phys. Ill, 7774, 1999. H. Sakai et al, Phys. Rev. A 67, 063404, 2003 and references therein. H. Sakai et al, Phys. Rev. Lett. 90, 083001, 2003. S. Minemoto et al, J. Chem. Phys. 118, 4052, 2003. W. J. Walecki et al, Opt. Lett. 22, 81, 1997. T. Suzuki et al, Phys. Rev. Lett. 92, 133005, 2004.

163

Tomographic Imaging of Molecular Orbital with High Harmonic Generation J. Itatani^'^ J. Levesque^'^ D. Zeidler\ M. Spanner^'^ P. B. Corkum^ and D. M. Villeneuve^ ^ National Research Council of Canada, 100 Sussex Drive, Ottawa, Ontario KIA 0R6, Canada ^ University of Ottawa, 150 Louis Pasteur, Ottawa, Ontario KIN 6N5, Canada ^ INRS-Energie et Materiaux, Varennes, Quebec J3X 1S2, Canada "^ University of Waterloo, 200 University Avenue W., Waterloo, Ontario N2L 3G1, Canada Abstract. High harmonics produced in aligned molecules contain structural infomiation of bound-state electronic states. We have produced high hamionics from N2 molecules aligned in two orthogonal directions. The projected images of the highest molecular orbital (HOMO) are successfully reconstructed using an algorithm of computed tomography using the observed hamionic spectra.

The essence of high harmonic generation is the fast-moving electron that is produced by tunnel ionization, accelerated by the intense laser field, and that re-collides with the parent atom or molecule on the attosecond time scale [1]. At the time of re-collision, the kinetic energy of the electron can exceed -100 eV. This energy corresponds to a de Broglie wavelength of ~ 1 A - thus, atomic-scale dynamic imaging should be possible. We show that the boundstate electron wavefunction can be reconstructed from a series of harmonic spectra produced from aligned molecules. High Harmonic Generation from Aligned Molecules Molecular alignment is achieved via the rotational wavepacket technique: First, an intense but non-ionizing laser pulse (60 fs, -10^^ W/cm^, "pump") is focused into a supersonic N2 gas jet. This laser field gives a kick to the molecules towards the laser polarization direction to create rotational wavepackets. Another time-delayed intense laser pulse (30 fs, 3x10^"^ W/cm^, "probe") is then focused into the gas jet to produce harmonics. Both pulses are linearly polarized in the same direction. Figure 1 shows the intensities of 19* harmonic (left axis, solid line) as a function of the pump-probe delay. The variation of harmonic intensities (left axis) agrees well to the degree of molecular alignment (right axis, simulated). All orders of harmonics show similar revivals. This result indicates that harmonics are enhanced when molecules are aligned parallel to

164

the field, and suppressed when perpendicular. The best aUgnment along the polarization of the pump pulse is achieved at the half revival at a delay of 4.1 ps„ followed by the best perpendicular alignment at 4.3 ps. We use these two ahgnments to partially reconstruct the molecular orbital.

3

4

5

6

Delay time (ps) Figure 1. Revival of rotational wavepackets in N2 probed by high harmonics Reconstruction of Molecular Orhitals Under our experimental conditions, the alignment dependence of tunnel ionization is small (angular variation -25%) and approximately canceled by the spatial spread of the electron wavepacket in the continuum. We therefore conclude that the alignment dependence of harmonic intensities is dominated by the recombination process. When the wavepacket \|/c re-coUides with N2, its lateral spread reaches '-^9 A by quantum diffusion. This size is considerably larger than the size of molecules. The electron wavepacket seen by the molecule can thus be approximated by a chirped plane wave, i.e., i|/c=2^pA(p)exp[ipx-iKt], where p is the momentum of the electron and K(= p^/2) is the kinetic energy. The induced dipole is then given by d(co=K)~A(p)<\|/g|r|exp(ipx)>. This expression is the spatial Fourier transform of r^g projected to the perpendicular direction to the laser polarization. The harmonic spectrum is related to this oscillating dipole as S(a))~co'^|d(co)p. Once we know the harmonic spectrum including the spectral phase and polarization, we can deduce the projected wavefunction by inverse Fourier transforming the harmonic spectrum. By controlling the direction of molecular alignment with respect to the probe pulse, we can measiire a set of harmonic spectra. This is equivalent to measuring the projected images of the wavefunction ^g in different directions. We can therefore reconstruct the image of the boundstate wavefunction by applying the algorithm of computed tomography. Although our present experimental setup does not allow us to control the molecular alignment to arbitral directions, we can partly retrieve the wavefunction from harmonic spectra produced from two orthogonal

165

alignments. In the reconstructing procedure, we used the harmonic spectra of Argon as a reference to cahbrate the A(p) term, hi addition, we assume the spectral phase and polarization that are predicted numerically. Figure 2 shows the molecular orbital of N2 (filled circles) retrieved from harmonics and the one obtained by the ab initio calculation for the HOMO of N2 (ag, solid lines). Both curves, including the sign of the wave fiinction, agree quantitatively well inside the molecular orbital. Further improvement should be possible by aligning molecules at a number of angles. Since this method can selectively image the HOMO, it will open the way to directly observe the ultrafast electronic processes of small molecules that undergo a chemical reaction with sub-A and sub-femtosecond precision. The clear relation between the molecular orbital and harmonics will enable the shaping of attosecond XUV pulses by selecting and controlling molecules.

p^i

(b) -3

-2

-1

0 Distance (A)

1

2

3

Figure 2. Retrieved and exact molecular orbitals of N2. (a) Projection of the molecular orbital onto the plane parallel to the molecular axis, (b) projection onto the plane perpendicular to the molecular axis.. Acknowledgements. The authors acknowledge financial support from Photonic Research Ontario, Canadian Institute for Photonic Innovation, and the Alexander von Humboldt-Stiftung.

References 1 P. B. Corkum, Phys. Rev. Lett. 71, 1994 (1993).

166

Femtosecond Infrared Vibrational Up-Pumping of Liquid Phase W(CO)6 Thomas Witte\ Marcus Motzkus\ Karl Kompa\ Jake Yeston^, and Edwin Heilweil^

2

Max-Planck-Institiit fiir Quantenoptik, Hans-Kopfermann-Str. 1, 85748 Garching, Germany E-mail: klk(a)mpq.mpg.dc Optical Technology Division, National Institute of Standards and Technology, 100 Bureau Drive, Gaithersburg, Maryland 20899-8443 E-mail: edwin.heil\[email protected]

Abstract. CO-stretch excitation of W(C0)6 in room temperature n-hexane using 5 ^im femtosecond pulses transfers vibrational population to v>5, as measured by transient midinfrared spectroscopy and compared to Bloch model calculations. These results constitute significant steps towards controlling molecular ground state populations and reactions.

1. Introduction Efficient preparation of highly excited vibrational states in condensed-phase ground electronic state molecular systems remains a significant chemical objective. Beyond interpretation of vibrational dynamics, relaxation, and evolution of microscopic structure in complex molecules, one may use tailored excitation pulses to control energy dynamics and reaction processes via excitation of vibrational modes. Such efforts are particularly important if vibrational excitation along a reaction coordinate leads to more efficient reaction [1]. Femtosecond mid-IR (MIR) pulses are v^ell-suited for vibrational mode overtone excitation to dissociate molecular bonds. Generating femtosecond laser pulses with tailored amplitude and phase is now possible, conceptually enabling more extensive studies and control of complex molecular dynamics. We recently demonstrated gas-phase MIR induced dissociation of metal carbonyl compounds [2,3] and diazomethane [4]. In this work, we excite with broadband femtosecond IR pulses and apply transient absorption MIR spectroscopy to probe condensedphase vibrational overtone excitation and population distributions. Following earlier work by Heilweil, et al. [5-7], tungsten hexacarbonyl, W(C0)6, was selected because the Tiu CO-stretching fundamental has a large 0-^1 transition dipole moment («1 Debye) and its overtone anharmonic shifts are well-resolved. In addition to controlling population distributions, we measure spectroscopic anharmonic shifts and bandwidths for the higher lying overtone transitions and compare our results to Bloch model calculations for the excitation process.

167

2. Experimental Method Mid-infrared «12 /iJ, 160 fs pulses in a 160 cm'^ FWHM envelope near 5 ^im (2000 cm'^) were generated by difference frequency mixing (in AgGaS2) signal and idler pulses from a tunable optical parametric amplifier pumped by 1.5 mJ, 110 fs, 800 nm output pulses from a 1 kHz regeneratively amplified Ti:Sapphire laser. The experimental apparatus was a modified MIR autocorrelator using a wedged CaF2 window beamsplitter. Pump and probe beams were focused into a static 0.9 mm sample cell (3 mm CaF2 windows) by a 10 cm focal length goldcoated spherical mirror. Commercial W(C0)6 was dissolved in room temperature w-hexane to yield optical density (OD) « 0.7 at the 1980 cm'^Tiu COstretching fundamental. Further experimental and data acquisition details are in ref. [8].

3. Results and Discussion The main result of this work is the transient absorbance spectrum shown below:

1^cm-'i907cm',332^.,

pump energy — 8.5 MJ At-1ps 1980 cm2000

1980

1960

1940

1920

1900

1880

wavenumber (cm'^)

Fig 1. Differential absorption (AOD) for excited W(C0)6 in hexane as a function of probe frequency showing parent bleach at 1980 cm"^ and overtone absorptions. The pump-probe pulse delay used (1 ps) separated the pulses in time to avoid coherence artifacts and reduce effects from relaxation. The strong fundamental bleach at 1980 cm'^ is accompanied by five new absorption bands from the first five overtone transitions (cm"^): 1-^2 at 1966, 2-^3 at 1951, 3-*4 at 1930, 4->5 at 1907, and 5-^6 at -1882. By fitting each feature to a Gaussian function, the FWHM widths (cm'^) for the fundamental through fourth overtone are: 5.6 (± 0.1), 7.0 (± 0.3), 7.1 (± 0.3), 8.8 (± 0.5), 11.1 (± 0.8) (type A errors, k = 1, la). Poor signal-to-noise precluded measurement of the fifth overtone width. Note that observing absorption for the 5->6 transition implies some population transfers to V = 6. Since the pump has «20 times higher energy at this frequency, it must 168

create nonzero population in v = 6. A principal limitation for ladder climbing is the MIR laser pulse spectral bandwidth. The current bandwidth should permit v=6-*7 excitation, but this absorption is not observed. The excited state population distribution was modeled by numerical solution of Bloch equations via the density matrix formalism [5,6]. We used a Gaussian spatial intensity profile of the pump beam discretized at the focus into a series of concentric annuli to estimate Rabi frequencies. Excellent agreement with experiment occurred with a focal diameter of 400 jxm leading to the following theoretical population profile: 74 % in v=0,14 % in v=l, 6.1 % in v=2, 3.2 % in v=3, 1.5 % in v=4, and 0.5 % in v=5 [8].

4.

Conclusions

Generation of vibrational population up to v = 6 in the Tiu W(C0)6 CO-stretching mode was achieved by femtosecond mid-IR excitation and ladder climbing. The anharmonic transitions, center frequencies and spectral widths were determined up to V = 5-^6 by MIR transient absorption spectroscopy. With a maximum energy of nearly 12,000 cm"^ deposited in a single mode, the molecule is excited to within the dissociation energy of a W-C bond. Further experiments may induce ground state molecular dissociation in the condensed phase and detect dissociation products [9]. Extensions of this coherent control study in the liquid phase may apply amplitude and phase modulated MIR laser pulses [10,11] with feedback.

References 1 2 3 4 5 6 7 8 9 10 11

J. C. Polanyi, Ace. Chem. Research 5, 161,1972. L. Windhorn, T. Witte, J. Yeston, D. Proch, M. Motzkus, K. L. Kompa, and W. Fuss, Chem. Phys. LeU. 357, 85, 2002. T. Witte, T. Hornung, L. Windhorn, D. Proch, R. de Vivie-Riedle, M. Motzkus, K. L. Kompa, J. Chem. Phys. 118, 2021, 2003. L. Windhorn, J. Yeston, T. Witte, W. Fuss, M. Motzkus, D. Proch, K.L. Kompa, and C.B. Moore, J. Chem. Phys. 119, 641, 2003. V. D. Kleiman, J. S. Melinger, E. J. Heilweil, in Femtochemistry and Femtobiology: Ultrafast Dynamics in Molecular Science, World Scientific, New Jersey, 409, 2002. V. D. Kleiman, S. M. Arrivo, J. S. Melinger, and E. J. Heilweil, Chem. Phys. 233, 207, 1998. S. M. Arrivo, T. P. Dougherty, W. T. Grubbs, and E. J. Heilweil, Chem. Phys. Lett. 235, 247,1995. T. Witte, J. S. Yeston, M. Motzkus, E. J. Heilweil, and K-L. Kompa, Chem. Phys. Lett., 392,156, 2004. U. Liebl, G. Lipowski, M. Negrerie, L. C. Lambry, J. L. Martin and M. H. Vos, Nature 401, 181,1999. T. Witte, D. Zeidler, D. Proch, K. L. Kompa, and M. Motzkus, Optics Lett. 27, 131, 2002. T. Witte, K. L. Kompa, and M. Motzkus, Appl. Phys. B 76, 476, 2003.

169

Ultrafast X-ray diffraction K. Sokolowski-Tinten^ C. Blome^'^ J. Blums^'^ U. Shymanovich^ M. Nicoul\ A. Cavalleri'^, A. Tarasevitch\ M. Hom-von Hoegen^ M. Kammler^ and D. von der Linde' ' Institut fiir Experimentelle Physik, Universitat Duisburg-Essen, 45117 Essen, Germany E-mail: [email protected] ^ HASYLAB at DESY, Notkestr. 85, 22603 Hamburg, Germany ^ Riga Technical University, Tech. Physics Institute, I Kalku str., Riga, LV-1658, Latvia '^ Materials Science Div., Lawrence Berkeley Ntl. Lab., Berkeley, California 94720, USA ^ Institut fiir Halbleitertechnologie, Universitat Hannover, 30167 Hannover, Germany Abstract. In the last few years the generation of femtosecond pulses in the X-ray regime has become possible. These ultrashort X-ray pulses have enabled femtosecond timeresolution to be extended to X-rays.

1. Introduction Optical pump-probe spectroscopy has been used to study ultrafast phenomena for more than three decades. A short laser pulse is used to excite the process of interest, and a similarly short, delayed laser probe pulse interrogates the subsequent evolution of the system. The time resolution is essentially determined by the duration of the probe pulse. Today, time resolution of 10"^"^ s or below is almost routinely achieved. The laser probe pulse typically monitors the pump-induced changes in the optical properties of the system. Because optical techniques lack the necessary spatial resolution, it was previously not possible to directly observe ultrafast transients in the atomic structure. During the last years, this situation has changed rapidly. New types of radiation sources have been developed or are being developed, which provide ultrashort pulses of photons or electrons with sufficiently short wavelength to enable the resolution of the atomic structure. Pump-probe experiments employing ultrashort X-ray or electron probe pulses combine atomic scale spatial resolution with femtosecond time resolution and enable direct measurements of ultrafast transients in the atomic structure.

2. Laboratory femtosecond kilovolt X-ray source Extremely powerful X-ray pulses from free electron lasers and other X-ray sources relying on electron beams will become available, probably in a couple of years. While these future sources require large-scale electron accelerator facilities, small-scale laser-driven X-ray sources already exist and provide femtosecond kilovolt X-ray pulses, albeit at relatively low X-ray flux. X-rays are generated by

170

focusing femtosecond laser pulses onto the surface of a solid or a liquid target to produce a microplasma. The microplasma emits a short burst of incoherent X-rays into the full solid angle. A portion of these X-rays can be collected and focused onto a sample using a suitable X-ray mirror. We use thin, toroidally bent semiconductor crystals as focusing elements which are configured to select and focus the characteristic K^ line radiation of the respective target atoms. The laser pulses are generated by a Titanium sapphire CPA laser system (120 fs, 800 nm). A small fraction of the laser output serves as an excitation pulse, while the main portion is used for the generation of the X-rays. With a suitable optical delay line the timing of the laser pump pulse and the X-ray probe pulse can be accurately controlled.

3. Femtosecond laser-induced structural phase transitions It is well known that in semiconductors such as Si, Ge, GaAs etc. a structural phase transition from the crystalline to the liquid state can be brought about by strong electronic excitation. For a sufficiently high density of excited conduction electrons (typically 10 percent of the density of valence electrons) the crystal lattice becomes unstable and a transition to the liquid phase takes place. Compelling evidence has previously been obtained from ultrafast optical spectroscopy to show that this solid-liquid phase transition occurs in less than one picosecond [1]. However, direct evidence of the ultrafast structural change was obtained only recently from sub-picoseond time-resolved X-ray diffraction [2-4]. We have studied ultrafast, electronically induced melting in thin crystalline layers of Germanium using an (optical pump)-(X-ray probe) scheme in which the Bragg diffraction of the probe pulse from (111) lattice planes was used to monitor the long range crystalline order of the Ge layer. Figure 1 shows the measured Xray diffraction signal as a function of the delay time between pump and probe for two different laser fluences. ^ > ^ 1.0

1

} ^ T

-

300 fs

1

d)

\\ 1 1

0)

1 : 1 ' 1 1 1 1

1S 0.8 Q •J

C

1

•

0.2 J/cm^

•

OAJIcm^

I '•^

A /' / A / /• / / / 1' /

1*0.6 •

._ _|11

1

.

1

A

1

0 1 Delay time [ps]

1

2

•

1

1

00

Fig. 1. Ultrafast laser-induced melting in Germanium: X-ray diffraction signal of the (111) Bragg diffraction order as a function of delay time between the laser and the X-ray pulse for two different pump fluences.

171

The signals for large negative delay represent the diffraction from the intact Ge film. It can be seen that the interaction with the pump pulse leads to an initial decrease in the diffraction signals within 300 fs, followed by a further reduction over the next few picoseconds. The diffraction signal does not vanish completely because only a fraction of the layer (whose thickness depends on the pump fluence) is actually molten and disordered. The observed decrease in the X-ray diffraction is consistent with the conclusion derived from optical experiments that subpicosecond (partial) melting of the Ge layer takes place. The X-ray diffraction experiments also put an upper limit of about 300 fs on the X-ray pulse duration.

4. Coherent optical phonons With a short enough X-ray pulse it is possible to freeze the rapid vibrational motion of the atoms and to take snapshots of the instantaneous atomic configuration. In crystals one can use X-ray diffraction to recover the vibrational dynamics of the lattice from the sequence of snapshots of diffraction patterns. In the semimetal Bismuth it is possible to photoexcite coherent lattice vibrations corresponding to a zone center optical phonon mode by means of a mechanism called displacive excitation. This mode corresponds to an internal relative motion of the two Bi basis atoms that form the primitive unit cell of crystalline Bi. One can track this atomic motion using X-ray diffraction by taking advantage of the changes in the geometrical structure factor associated with the changes in the position of the Bi atoms [5]. Figure 2 shows a result of a timeresolved X-ray diffraction experiment in which the Bragg diffraction from (222) lattice planes of Bi was measured. o c

a* u 0)

73 (U

c

0

1 2 Delay tinne [ps]

3

Fig. 2. Coherent optical phonons in Bismuth: X-ray diffraction from (222) lattice planes versus delay between the laser pulse and the X-ray pulse.

172

Photoexcitation of the Bi crystal by a femtosecond laser pump pulse at 800 nm leads to an increase in the diffraction signal of a few percent followed by distinct oscillations over several pisoseconds. The oscillation period of 347 ps deduced from the sequence of diffraction patterns corresponds to a frequency of 2.12 THz. This value is close to, but significantly less than the well-known frequency of the totally symmetric optical phonon mode of Bi, 2.92 THz. The shift to lower frequencies depends on the pump fluence and can be attributed to mode softening and anharmonicity of the strongly driven optical phonon mode.

5. Lattice heating When light interacts with matter the optical energy is initially deposited in the electronic states of the system. A complex sequence of relaxation processes then redistributes the energy over all the other degrees of freedom of the system until thermal equilibrium is reached. An important step in the relaxation chain is the transfer of energy from the electronic states to the phonon modes of the lattice. The heating of the lattice resulting from the electron-phonon interaction is accompanied by an increase of the random motion of the lattice atoms. In X-ray diffraction this random motion leads to a reduction of the intensity of the Bragg diffraction by an amount given by the Debye-Waller factor. We have studied lattice heating in Germanium following intense interband photoexcitation using time-resolved X-ray diffraction. We were able to observe the decrease in the X-ray diffraction signal caused by the Debye-Waller effect for three different diffraction orders ((111), (311) & (400)). Figure 3 shows time dependencies of the diffraction signal measured on a 170 nm thick Ge film for an excitation fluence of 35 mJ/cm^. 170 nm Ge on Si, 35 mJ/cm^

0

2 4 Delay Time [ps]

6

Fig. 2. Transient Debye-Waller-effect in Germanium: temporal evolution of the (111)(triangles), (311)- (squares) and (400)- (circles) diffraction order after excitation with a 120 fs, 35 mJ/cm^ laser pulse. '

173

All three traces exhibit an approximately exponential decrease with a time constant of 1.1 ps which represents the electron-lattice energy relaxation time. Moreover, the relative decrease for the different diffraction orders is consistent with the Debye-Waller factor. For a fluence of 35 mJ/cm^ the final lattice temperature reached after a few picoseconds is 550 K.

6.

Conclusions

We have shown that ultrafast X-ray diffraction can reveal not only substantial changes in the atomic structure such as structural phase transitions but also much more subtle transient changes in the atomic configuration associated with coherent or random lattice vibrations. An attractive feature of laser-driven X-ray sources is that they enable relatively small-scale, almost table-top types of experiments. On the other hand, large-scale facilities for the generation of very powerful ultrashort X-ray pulses will be available in the future and will dramatically widen the scope of ultrafast X-ray science. As a counterpart of the development in the X-ray field impressive progress is also being made in the generation of femtosecond electron pulses and ultrafast electron spectroscopies. At the present time, two highly developed disciplines of science are about to be brought together, namely the field of structural investigations and that of ultrafast time-resolved spectroscopy. The combination of high time-resolution with atornicscale spatial resolution will eventually enable researchers to record detailed, quasi-instantaneous pictures of complex atomic structures. Acknowledgements. Financial support by the Deutsche Forschungsgemeinschaft and the European Community (RT-network XPOSE) is gratefully acknowledged.

References 1 C.V. Shank et al., , Phys. Rev. Lett. 50, 454 (1983); K. Sokolowski-Tinten et al, Phys. Rev. B 51, 14186 (1995); L. Huang et al., Phys. Rev. Lett. 80, 185 (1998). 2 C.W. Siders et al., Science 286, 1340 (1999) 3 A. Rousse et al., Nature 410, 65 (2001). 4 K. Sokolowski-Tinten et al., Phys. Rev. Lett. 87, 225701 (2001). 5 K. Sokolowski-Tinten et al., Nature 422, 287 (2003).

174

Quasi-phase matching of high harmonic generation in the "water window'' soft x-ray region Emily A. Gibson\ Ariel Paul\ Sterling Backus^ Ra'anan Tobey\ Margaret M. Murnane^ Henry C. Kapteyn^' and Ivan P. Christov^ ^JILA and Department of Physics, University of Colorado and NIST, Boulder, CO 80309-0440 E-mail: [email protected] ^Department of Physics, Sofia University, Sofia, Bulgaria Abstract. We demonstrate that high-order harmonic generation can be phase-matched in the "water window" region of the spectrum around 300eV using hollow fibers with a periodically modulated diameter.

High harmonic generation (HHG) is a useful source of coherent light in the extreme ultraviolet (EUV) region of the spectrum. In HHG, an intense femtosecond laser is focused into a gas, generating high harmonics that emerge as a coherent, low-divergence beam. However, both the conversion efficiency and the highest achievable photon energy have been limited to-date by the inability to phase-match the frequency conversion process at high ionization levels, where plasma-induced dispersion prevents the laser and the EUV light from propagating at the same phase velocity. Since the higher harmonics are generated at higher laser intensities after much of the gas has ionized, this limits efficient HHG to energies < lOOeV. Overcoming ionization-induced phase-mismatch has thus been a critical challenge to the further development of coherent EUV sources at wavelengths of interest to lithography, high-resolution imaging, site-specific spectroscopy and bio-microscopy. In previous work, we demonstrated quasi-phase matching (QPM) of HHG up to energies of ISOeV, corresponding to ionization levels < 5%.[1] QPM is implemented by focusing the driving laser into a gas-filled hollow-core glass fiber with a periodically modulated inner diameter. The modulations increase and decrease the laser intensity along the length of the fiber. Since the HHG process is highly sensitive to the intensity of the driving light, the harmonic signal is amplitude and phase-modulated. When the period of modulation matches the coherence length of the process, the HHG signal can build-up coherently. An enhancement in harmonic signal can also occur from higher-order QPM where the period of modulation is an odd integer multiple of the coherence length.[2,3] In this work, we show enhancement of HHG flux in the "water window" region of the soft x-ray spectrum (> 284 eV) by using higher laser intensities combined with shorter modulation periods.[4,5] The application of nonlinear optical techniques to HHG promises a practical, coherent, "water-window" light source for biomicroscopy. This work is also first demonstration of HHG at these energies

175

using neon. Previously, it was only possible to generate water window light using helium, which has a much smaller nonlinear susceptibility.

Fig. 1. Optical microscope image of a modulated hollow-core waveguide with a period of 0.25mm and an inner diameter of ISOjum. The modulations were produced using glass-blowing techniques, where a straight 150mm inner diameter hollow-core fiber was pressurized and heated. From Ref. 4. In our experiment, we focus 22 fs duration, 800 nm laser pulses, with energies of 3 mJ, into a 150 jum diameter, 2.5 cm long, hollow glass fiber filled with neon. The fiber, shown in Figure 1, is modulated with period 0.25mm. Figure 2 shows the harmonic emission using a straight fiber (black) and a modulated fiber (red). For the modulated fiber, the carbon edge is clearly visible, while the highest photon energy observed from the straight fiber is - 225 eV. Fig. 3 shows the harmonic emission from the modulated fiber through Boron (black) and Carbon (red) filters. The presence of sharp absorption edges confirms the accuracy of our energy calibration. The C edge at 284eV represents the highest harmonics ever observed in neon gas and the highest harmonics that have been quasi-phase-matched.

Fig. 2. Harmonic emission from neon using a straight fiber (black) and a 0.25mm-period modulated fiber (red), taken with the same exposure time and filters. From Ref. 4.

176

These results demonstrate that QPM can be applied to HHG in highly-ionized gasses, and that phase-matching issues are not a fundamental limitation for HHG. Since the scaling of the high-harmonic "cutoff photon energy is linear with the intensity of the driving laser, these techniques will be useful for extending HHG to > keV. Moreover, using tighter modulation periods and longer fibers, the HHG flux can be further enhanced.

•S

1000.

Vi%Hi*£^^ 100

150

200

250

300

350

400

450

Energy (eV)

Fig. 3. Harmonic emission from 9 Torr of Neon from a modulated fiber. Red curve, right axis: Ag filters and C, showing C edge at 284eV edge. Black curve, left axis: Ag and B filters to show Boron edge at 188eV. From Ref 4. Acknowledgements. This work was supported by the National Science Foundation and the Chemical Sciences, Geosciences and Biosciences Division of the Office of Basic Energy Sciences, U.S. Department of Energy, and made use of Engineering Research Centers Shared Facilities supported by the National Science Foundation under award number EEC-0310717.

References 1. 2. 3. 4. 5.

A. Paul et al., "Quasi-phase-matched generation of coherent extreme-ultraviolet light," Nature 421, 51(2003). M. M. Fejer et al., "Quasi-phase-matched 2nd harmonic-generation — Tuning and tolerances," IEEE J. Quant. Electron. 28, 2631 (1992). I. P. Christov et al., "Quasi-phase matching of high-harmonics and attosecond pulses in modulated waveguides," Opt. Express 7, 362 (2000). E. A. Gibson et al., "Coherent soft x-ray generation in the water window with quasi-phase matching," Science 302, 95 (2003). E. A. Gibson et al., "Extreme Nonlinear Optics: Attosecond Photonics at Short Wavelengths", invited paper to be published in JSTQE, (2005).

177

Adaptive engineering of coherent soft x-rays T. Pfeifer, D. Walter, C. Winterfeldt, C. Spielmann, and G. Gerber Physikalisches Institut, Universitat Wiirzburg, 97074 Wiirzburg, Germany Email: [email protected] Abstract. We demonstrate the generation of arbitrarily shaped spectra of coherent soft xrays by adaptive control of the driving laser pulses. These are the first steps towards coherent control in the soft-x-ray region and attosecond-pulse shaping. With coherent-control techniques it is possible to control the behavior of quantum systems on their natural time- and length scale. This is achieved by applying ultrashort coherent light fields, which can be variably shaped in time and space. We are nowadays able to directly steer chemical reactions to evolve in a certain way, leading to specific user-defined target states [1,2]. The duration of the laser pulses used in these experiments has to be short compared to the characteristic time scale of the process. For molecular systems this time scale is given by the vibrational periods and is therefore on the order of femtoseconds. Since the next step in understanding the inner dynamics of atoms and molecules is to directly measure and control the electronic motion, we need to find a suitable coherent light source operating in the x-ray region, delivering user-defined pulses with durations in the attosecond range. 1.0

A

(11), 21

(I)

^•

g 0.8 • B

23

^°-«

1 1

1 0.4-

fi

^0? 0.0-

1 19

J\J\J^

.*^fKJ\

22 24 26 28 30 32 34 36 38 40 42 Wavelength [nm]

Fig. 1. Selective enhancement of high-harmonic generation. (A) Optimization of the low frequency spectral part while simultaneously reducing the high frequency part and vice versa (B). Our work shows that we made the first important step to this new field of quantum control of electron dynamics. By applying femtosecond laser pulse shaping techniques to the process of high-order harmonic generation, we are able to generate arbitrarily shaped coherent soft x-ray spectra spanning almost an octave. The degree of control includes the selective generation of groups of harmonic orders at different frequencies, isolating single harmonics at different orders and creating holes in extended harmonic plateau emission, which has never been observed in the literature. Previous experiments to control harmonic emission focused on the enhancement of harmonic yield, line width [4], and position [5] . Demonstration to vary the overall qualitative spectral shape with high fidelity opens the door to a variety of applications devoted to the monitoring and steering of the time-dependent electronic wavefunction on the attosecond time-scale. In addition, our experimental results are the first demonstration of an 178

attosecond pulse shaper. Mairesse et al. [6] recently showed that different harmonics in the plateau region exhibit a fixed phase relationship. The control over which group of harmonics (Fig. 1) or how many harmonics (Fig. 2) contribute to the spectrum directly determines the duration and shape of the emerging attosecond pulses. This fact contains important consequences to the developing field of ultrafast x-ray science. A

^21 25 1 23 M 2927 H 1

31 IWI ^ i W

\j\

.jAk.

24 26 28 30 32 34 36 38 40 Wavelength [nm]

24 26 28 30 32 34 36 38 40 Wavelength [nm]

24 26 28 30 32 34 36 38 40 -4 - 3 - 2 - 1 0 1 2 Wavelength [nm] Time [fs]

Fig. 2. Suppression of harmonics while generating adjacent ones at about ten times higher efficiency. (A) One suppressed harmonic order, (B) two suppressed harmonic orders, (C) typical harmonic emission in the plateau region. The high degree of controllability of softx-ray spectral shape over a large range of photon energies implies major modifications of the corresponding temporal shape on a sub-femtosecond time scale, as can be seen by looking at the Fourier transformed spectral amplitudes (D) (solid and dotted lines correspond to spectrum (B) and (C), respectively. The dashed line indicates the fundamental laser electric field). Our experimental apparatus consists of the following: A Ti:sapphire regeneratively amplified femtosecond laser system delivers pulses (1 kHz, 800 /xJ, 80 fs, 800 nm) which are initially compressed to approx.-^—20 fs using self-phase modulation in a hollow-fiber setup with a prism compressor. The retroreflector in the compressor is a deformable membrane mirror [7,8]. Since the laser pulse spectrum is spatially separated at that point, this device constitutes a phase-only pulse shaper. After passing through the prism compressor, the shaped pulses are focused into an argon-filled hollow capillary to generate high-harmonic radiation [9]. The produced soft x-rays are separated from the generating laser pulse with a 300 nm aluminum filter and analyzed by a grazing incidence spectrometer equipped with a backside-illuminated charge-coupled-device (CCD) camera. A computer is used to control the deformable mirror and read the spectrometer. An evolutionary algorithm [10] performs an optimization of the soft x-ray spectral shape with regard to a certain shape criterion. This approach allows us to analyze which spectral modifications are possible. We obtained positive results for the basic operations of soft-x-ray spectral control: a) selective generation at defined photon energies with suppression of the plateau spectrum (Fig. 1), b) selective suppression (spectral holes) at single photon energies in a surrounding plateau spectrum (Fig. 2), and c) selective generation of groups of harmonics at different central wavelengths. A simulation

179

of the single-atom response of the process (Fig. 3) yields two main results: First, it is evident that the origin of harmonic control is associated with spatial and propagation effects as also pointed out by Reitze et al. [5] and cannot only be attributed to the single-atom response as claimed by Christov et al. [11], thus answering a long-standing debate in the field. Second, we see that very minor changes to the IR-pulse shape result in substantial changes to the shape of the attosecond pulses produced in high-harmonic generation. This is contrary to the intuitive consideration that sub-optical-cycle changes to the driver pulse are required in order to alter the attosecond pulse shape. 30

1 /\ A

20

^10 3

-Bo

40.00

IItP1 1 1 \ w

•

•j-0.05

"11 VIU U

time [fs]

Fig. 3. Simulation of controlled high-harmonic generation for (A) maximization and (B) minimization of the 23rd harmonic order (spectra on the left-hand side). The temporal structure of the soft-x-ray emission (solid line) changes dramatically for these two scenarios while the fundamental laser electric field (dotted line) is only slightly different. In conclusion our results provide an open road towards the future field of coherent control of electron dynamics. With the possibility to arbitrarily engineer the spectrum of coherent soft x-rays, we made the first step by creating a versatile light source for these experiments. At the moment, we are performing experiments where these shaped spectra are applied to control molecules with coherent soft xray light. We expect new discoveries and much improved knowledge from the application of these new techniques.

References 1 2 3 4 5 6 7 8 9 10 11

180

C. J. Bardeen et al, Chem.Phys.Lett. 280, 151-158, 1997. A. Assion et al. Science 282, 919, 1998. P. B. Corkum, Phys. Rev. Lett. 71, 1994, 1993. R. Bartels et al. Nature 406, 164, 2000. D. H. Reitze et al, Opt. Lett. 29, 86, 2004. Y. Mairesse et al. Science 302, 1540, 2003. E. Zeek et al. Opt. LeU. 24, 493, 1999. T. Pfeifer, U. Weichmann, S. Zipfel, G. Gerber, J. Mod. Opt. 50, 705, 2003. A. Rundquist et al. Science 280, 1412, 1998. T. Baumert et al, Appl. Phys. B 65, 779, 1997. LP. Christov et al, Phys. Rev. Lett. 86, 5458, 2001.

Generation of strong soft x-ray field based on high-order harmonics Hiroki Mashiko^'^, Akira Suda\ and Katsumi Midorikawa^^ ^ RIKEN, 2-1 Hirosawa, Wako-shi, Saitama 351-0198, Japan E-mail: [email protected] ^ Graduate school of science and engineering, Saitama University, 255 Okubo, Saitamaa, Saitama 338-8570, Japan E-mail: [email protected] Abstract. We have generated a strong soft x-ray field based on high-order harmonics. The 27th harmonic wave was selected and focused to a spot size of less than 1 jim using a Si beam splitter and an off-axis parabolic mirror. The highest intensity achieved at 29.6 nm was 1 x 10^"^ W/cm^.

1.

Introduction

The high-order harmonic wave is a high-brightness coherent light source in the XUY and soft x-ray wavelength regions. The energy of high-order harmonics has reached the microjoule level by accurately controlling the phase-matching conditions between the ftmdamental and the harmonic waves [1, 2]. Furthermore, excellent beam quality and spatial coherence show the possibility of focusing down to a micron-scale spot size [3,4]. We demonstrate a technique for focusing intense soft x-ray high-order harmonics to a 1 |im spot size with a peak intensity as high as 10^"* W/cm^. The 27th harmonic wave at 29.6 nm is focused by an off-axis parabolic mirror with a SiC/Mg multilayer coating, and focal spot images are observed via visible fluorescence fi-om a Ce:YAG scintillator. We also observe ablation patterns on a metal target induced by a strong field of the 27th harmonic wave.

2.

Experimental Methods

We used a Ti: sapphire chirped-pulse amplification system with an output energy of 14 mJ at 10 Hz. The pulsewidth was 22 fs and the center wavelength was 785 nm. The output pulse was loosely focused into an interaction cell filled with Ar at 9.2 Torr to stimulate harmonic generation. The generated harmonics were sent to a Si beam splitter. This beam splitter has a reflectivity of 50% at 29.6 nm, which also has an extinction ratio of 10" for the fiindamental wave [5]. A 200-nm-thick Al filter was used to block the fiindamental beam and the loworder harmonics, with the transmittance of 14% at 29.6 nm. The harmonic spectrum was monitored using a grazing-incidence spectrometer with a micro channel plate. A CCD camera detected the two-dimensional images of the spectrally-resolved far-field profiles. We derived the relative intensity fi*om the

181

spectrograph. The absolute energy of the high-order harmonics was measured using a calibrated silicon XUV photodiode. The total energy measured with an Al filter was 75 nJ. The 27th harmonic energy wasderived from the total energy and the relative intensity (30%). Consequently, typical measured energies for the 27th harmonic wave with and without an Al filter were 23 nJ and 160 nJ respectively. We used an imaging system as shown in Fig. 1 to measure the focal spot of the soft x-ray beam. The 27th harmonic wave was selected and focused by an offaxis parabolic mirror (f=6 cm, 6=24°) with a SiC/Mg multilayer coating. The reflectivity of the multilayer-coated mirror was 40% at 29.6 nm with a FWHM bandwidth of 2 nm. The focal spot image of the 27th harmonic was observed from the visible fluorescence induced by the soft x-ray photons on a CeiYAG scintillator [4, 6]. A set of image relay optics were constructed with an objective lens (NA=0.65, f=5 mm) and a convex lens (f=500 mm), that transferred the visible light image onto an image-intensified CCD camera. We have also observed the spot size from an ablation pattem on an Au target placed at the focal position. Si beam splitter

Si(/Ms: mirror

(•^5 mm

Ce:YAG intillator

r=500 mm

Fig. 1. Experimental setup for imaging the focused profile.

3.

Results and Discussion

Figure 2(a) shows an image observed using the image relay system with the CeiYAG scintillator. The spot size of ~1 |im corresponds to 1.7 times the diffraction-limited value. We estimated the focused intensity of the 27th harmonic wave from the energy, the spot size, and the pulsewidth. The energies at the target with and without the Al filter were 9 nJ and 64 nJ, respectively. Assuming that the pulsewidth of the 27th harmonic wave is the same as that of the fiindamental wave, the focused intensities with and without the Al filter were 1.4 X 10^^ W/cm^ and 1.0 x 10^^ W/cm^ respectively. Figure 2(b) shows an optical microscope image of an ablation spot on the Au target. The diameter of the circular hole is approximately 2 jim, which agrees well with the Ce:YAG scintillator experiment. Figure 2(c) shows an example of

182

the ablation patterns taken in a single shot and observed using an Atomic Force Microscope (AFM). (a) (b) (c)

Fig. 2. (a) Typical focal spot image observed by an imaging system using a Ce:YAG scintillator, (b) Optical microscope image for an ablation spot on an Au target, (c) AFM image for an ablation spot on an Au target.

4.

Conclusions

Li conclusion, we have generated a strong soft x-ray field based on 27th harmonic generation. The 27th harmonic wave was selected and focused using a Si beam splitter and an ofF-axis parabolic mirror. The focal spot images were observed after visualization with a CeiYAG scintillator. The spot radius was less than 1 iim, resulting in a focused peak intensity of 1.0 x 10^ W/cm^. We have also observed the spot size fi-om an ablation pattem on an Au target placed at the focal position. The diameter of the circular hole of the ablation pattem observed by AFM was approximately 2 |im, which agrees well with the CeiYAG scintillator experiment. Acknowledgements. H. M. acknowledges support fi-om the Junior Research Associate Program of RIKEN. References 1 E. Takahashi, Y. Nabekawa, T. Otsuka, M. Obara, and K. Midorikawa, Phys. Rev. A66,021802 (2002). 2 J.-F. Hergott, M. Kovacev, H. Merdji, C. Hubert, Y. Mairesse, E. Jean, P. Breger, P. Agostini, B. Carree, and P. Salieres, Phys. Rev. A 6 6,021801 (2002). 3 D. Yoshitomi, T. Shimizu, T. Sekikawa, and S. Watanabe, Opt. Lett. 27, 2170 (2002). 4 C. Valentin, D. Douillet, S. Kazamias, Th. Lefrou, G. Grillon, F. Auge, G. MuUot, Ph. Balcou, P. Mercere, and Ph. Zeitoun, Opt. Lett. 28,1049 (2003). 5 E. Takahashi, H. Hasegawa, Y. Nabekawa, and K. Midorikawa, Opt. Lett. 29, 507 (2004). 6 M. Moszynski, T. Ludziejewski, D. Wolski, W. Klamra, and L. O. Norlin, Nucl. Instrum. and Meth. A345,461 (1994).

183

Generation of sub-4-fs high harmonic pulses and their appHcation to the above-threshold ionization Taro Sekikawa, Atsushi Kosuge, Teruto Kanai, and Shuntaro Watanabe Institute for Solid State Physics, University of Tokyo, 5-1-5 Kashiwanoha, Kashiwa 2778581,Japan E-mail: [email protected] Abstract. Sub-4-fs high harmonic pulses with photon energy of 27.9 eV were generated by using the 20-fs second harmonic pulses of Ti:sapphire laser, produced by the broadbandfrequency conversion. The 27.9-eV pulses were fully characterized by the cross-correlation frequency resolved optical gating technique. A nonlinear optical process in the extreme ultraviolet (XUV) region, Le. above-threshold ionization of helium atoms, by the high harmonic pulses was observed, demonstrating the high harmonic generation with high spatial and temporal coherence. This enables the autocorrelation measurement of attosecond pulses.

1.

Introduction

In the last decade, high harmonic generation has been investigated extensively. One feature of high harmonics is the short pulse duration, compared with synchrotron orbital radiation (SOR), which is the representative light source in the extreme ultraviolet (XUV) and soft x-ray regions. Thus, one promising research field for high harmonics is time-resolved spectroscopy to trace ultrafast processes in inner-shell excited states in atoms, molecules, and solid-state materials, surface states, electron motion, and so forth [1-3]. For such applications, it is necessary to characterize the high harmonic pulses to determine the temporal resolution. In addition, the phase of the pulse is necessary for further application of high harmonic pulses to so-called "coherent control". It is, however, not easy to characterize them by the autocorrelation, which is the common method for the pulse measurement in the visible region, for lack of nonlinear optical processes with high nonlinear susceptibility in the XUV and soft x-ray regions. The only nonlinear optical process demonstrated so far was two-photon ionization of rare gases. The harmonic order, whose autocorrelation measurement was demonstrated, was less than 13th of the Ti:sapphire laser [4, 5]. So, the pulse durations of higher harmonic pulses have been characterized by the crosscorrelation with the strong fundamental laser pulses, enabling to obtain relatively strong nonlinear signals [6, 7]. One drawback of the crosscorrelation measurement is the temporal resolution limited by the pulse duration of the fundamental pulses. To overcome this drawback, recently, we have proposed the extension of cross-correlation frequency-resolved optical gating (XFROG) technique, which is a well-established 184

characterization method in visible, to the extreme ultraviolet (XUV) region and demonstrated a 10-fs high harmonic pulse generation [8]. Since a XFROG trace contains both temporal and frequency information, temporally longer pulses with a narrower bandwidth can be used as sharp knives to slice unknown shorter pulses with wider bandwidths in the frequency domain. It is almost equivalent to measuring the group delay. But, a XFROG trace contains the spectra. Then, the unknown shorter pulses are retrieved accurately from the XFROG trace by using the relatively longer reference pulses in contrast with the traditional crosscorrelation. In this paper, by shortening the pulse duration of the driving laser pulse further, sub-4-fs high harmonic pulses with photon energy of 27.9 eV were generated and fully characterized by XFROG. In addition, two-photon above-threshold ionization (ATI) of helium atoms was observed by using the high harmonic pulses.

2. Experimental Methods The 27.9-eV high harmonics were generated by using the second harmonic (SH) pulses of a Ti:sapphire laser system in the scheme of broadband frequency conversion [9], which was modified to employ a telescope system for spectral dispersionto correct the spatial pulse front distortion and the spatial mismatch of the phase matching angle. The optical arrangement is shown in Fig. 1. Since the beam diameter of the TiS laser pulses was 2 cm, the spot beam approximation was no longer valid. Special care was taken to align the mirrors symmetrically with respect to the BBO crystal to avoid the pulse front distortion. The pulse duration and energy in the present experiment were 20 fs and 700 |LJ, respectively, at a repetition rate of 1 kHz. One important merit of the frequency converter is that the pulse duration can be shortened in high efficiency, because the frequency bandwidth is broadened by doubling all the spectral components. Grating 150 Iine/mm

Grating 300 I ine/nin

Fig. 1: Broadband frequency converter. "R" indicates the radius of curvature of a mirror. Blue laser is the second harmonic of TiS laser. The SH beam was spatially divided into an annular part for high harmonic generation in argon gas and a central part for probe. The ninth harmonic of the SH pulse was selected and focused into helium gas by a Sc/Si multilayer mirror. The

185

mixing of the ninth harmonic with the SH pulse at the focus produces sidebands at an interval of the photon energy of the SH pulses in a primary photoelectron (PE) spectrum resulting from XUV pulses, corresponding to the absorption of a XUV photon together with the absorption and emission of a fundamental photon. The spectra of the neighboring sidebands are identical to those of the sum or difference frequency generation. As a consequence, a spectrogram called a FROG trace can be created by measuring the PE spectra of a sideband as a function of the delay time between the SH and XUV pulses. Here, we extracted the sum frequency process to avoid the contribution of the seventh harmonic in the difference frequency process. The ejected photoelectrons were energy-resolved by a magnetic bottle time-of-flight spectrometer and detected by a multichannel plate. Spectrum resolution was improved by applying a retardation voltage. Since two beams propagated collinearly, the pulse front mismatch does not degrade the time resolution of this system.

Aperture

Fig. 2: Experimental setup for XFROG measurement. The ejected photoelectrons were energy-resolved by a time-of-flight (TOP) photoelectron spectrometer. 2co is the second harmonic pulses of the TiS laser pulses. Sc/Si is a multilayer mirror.

3. Results and Discussion Figures 3 shows the 128 x 128 XFROG trace and the retrieved intensity and the phase of the harmonic pulse. The FROG error was 0.033. The retrieved pulse duration was 3.9 fs and the pulse was almost Fourier-transform limited. The process of the short pulse generation is as follows: The leading edge of the high harmonic pulse evolves with the nonlinear response of the dipole moment, while the trailing edge is formed by the exhaustion of neutral atoms caused by the tunneling ionization under the optical electric field [2]. So, to obtain shorter pulses in this scheme, especially attosecond pulses, the driving laser with shorter pulse duration is required. The pulse energy of the high harmonic pulses at focus was 0.9 nJ. In spite of small pulse energy, it is possible to achieve the peak intensity of 10 TW/cm at the focus by virtue of the short pulse duration, which is high enough to cause nonlinear optical phenomena in the XUV region. One interesting XUV nonlinear optical process is the above-threshold ionization (ATI), in which an atom absorbs more

186

photons necessary for ionization and the photoelectrons are emitted. To observe ATI, the high harmonic pulses were focused into helium gas. Photoelectrons ejected by the two-photon ATI process were successfully detected by using a magnetic bottle photoelectron spectrometer and an electron counter. The photoelectron spectrum with a retardation voltage of 10 V is shown in Fig. 3. This result shows that the generated high harmonic pulses were well focused to cause ATI. Thus, the high harmonic pulses are not only temporally but also spatially coherent. Coherent high harmonic pulses open the way to XUY nonlinear optics. j

^

1.U

—

b) ^0.8

f\ A ,

n

1

1

,

J"

-50

0

50

100

;

mV -

£•04

-100

/

on

2"^

1T 0^

K»i^ii:tt.iiifa.

0

-20

20

Time (fe)

Delay Time (fs)

Fig. 3: a) XFROG trace and b) retrieved pulse shape and phase of a 27.9-eV harmonic pulse.

r

140

Arllthl

120h (0

100

"I 80 o O

He 9th X 2

60

He 13th

ArOtri

40|-

2ol0

100

200

300

400

500

Tinfie of Flight (ns)

Fig. 4: Photoelectron spectrum by high harmonics. "He 9thx2" indicates the ATI photoelectrons from helium atoms by the ninth harmonic. Other notations indicate the photoelectrons ejected from parent atoms by one-photon absorption of high harmonics.

187

4. Conclusions In summary, we have generated 20-fs SH pulses of TiS laser by broadband frequency doubling, in which the telescope optical system was employed to correct the spatial pulse front distortion. 3.9-fs harmonics pulses with photon energy of 27.9 eV were generated by using the SH pulses and characterized by XFROG technique. For further pulse shortening, especially for attosecond pulse generation, we are planning to generate sub-10-fs SH pulses by shortening the TiS laser pulses to 20 fs. Attosecond pulses will be characterized by the autocorrelation method by observing the two-photon ATI process.

References [1] L. Nugent-Glandorf, M. Scherer, D. A. Samuels, V. M. Bierbaum, and S. R. Leone, J. Chem.Phys. 117,6108(2002). [2] T. Shimizu, T. Sekikawa, T. Kanai, S. Watanabe, and M. Itoh, Phys. Rev. Lett. 91, 017401 (2003). [3] R. Kienberger, E. Goulielmakis, M. Uiberacker, A. Baltuska, V. Yakovlev, F. Bammer, A. Scrinzi, T. Westerwalbesloh, U. Kleineberg, U. Heinzmann, M. Drescher, and F. Krausz, Nature 427, 817 (2004). [4] Y. Kobayashi, T. Sekikawa, Y. Nabekawa, and S. Watanabe, Opt. Le«. 23, 64 (1998). [5] N. A. Papadogiannis, L. A. A. Nikolopoulos, D. Charalambidis, G. D. Tsakiris, P. Tzallas, and K. Witte, Phys. Rev. LeU. 90, 133902 (2003). [6] T. E. Glover, R. W. Schoenlein, A. H. Chin, and C. V. Shank, Phys. Rev. LeU. 76, 2468 (1996). [7] J. M. Schins, P. Breger, P. Agostini, R. C. Constantinescu, H. G. Muller, A. Bouhal, G. Grillon, A. Antonetti, and A. Mysyrowicz, J. Opt. Soc. Am. B 13, 197 (1996). [8] T. Sekikawa, T. Kanai, and S. Watanabe, Phys. Rev. LeU. 91, 103902 (2003). [9] T. Kanai, X. Zhou, T. Sekikawa, S. Watanabe, and T. Togashi, Opt. LeU. 28, 1484 (2003).

188

Coherent imaging of laser-plasma interactions using high-harmonic EUV Light X. Zhang\ D. Ravmondson^ A. R. Libertun\ A. Paul\ M. M. Murnane\ H. C. Kapteyn , Y. Liu^ and D. T. Attwood^ ^ JILA and Department of Physics, University of Colorado and NIST, Boulder, Colorado 80309-0440, USA E-mail: [email protected] ^ CXRO, LBNL and Applied Science and Technology Graduate Group, University of California, Berkeley, CA 94720, USA Abstract. We demonstrate that light generated using high-harmonic conversion in waveguides has very high spatial coherence at 30 nm and 13nm. Using this light source, we demonstrate EUV images of the explosion of a micron-size water droplet illuminated by an intense femtosecond laser using an all-reflective, double-multilayer mirror setup and a CCD camera as an image recording device.

The technique of high-harmonic generation is attractive as a table-top source of coherent light with femtosecond or shorter pulse duration. One possible use of this radiation source is for time-resolved imaging of dense plasmas. The short-wavelength light can penetrate to very high free-electron densities with minimal refraction, making it possible to probe dynamics of high-density plasmas with high (micron or better) spatial resolution. In the first part of this work, we characterized the coherence of the HHG light source, at 30 nm, and at 13 nm. We used phase matched high harmonic generation (HHG) in waveguides and periodically-modulated waveguides, and measured essentially full spatial coherence.[l-4] Fig. 1 shows double-slit coherence measurements taken at wavelength of 13nm; previous measurements were done at 30 nm. Waveguides shorter than -6 cm exhibit reduced coherence. The use of HHG radiation for static imaging in the EUV at 13nm has recently been demonstrated;[5] the objective of the work described below is to extend these imaging techniques to time-resolved studies. We have succeeded in an initial demonstration of this use using EUV light at wavelength of 30 nm. The experimental setup uses a pair of curved multilayer mirrors that act to magnify and spectrally filter the image. A CCD camera, shielded from laser light and visible plasma fluorescence with aluminum filters is used to record the images. Multilayer mirrors, rather than zone plates were used to obtain a large working distance for imaging of plasmas without contamination of the optics. The two mirrors produce a 50 X magnified image of the object at the image plane. The CCD camera has a pixel size of 13.5 jum. The overall resolution of the system is limited at present by the NA and astigmatism of the imaging system to -1 jum. Initially, a metal mesh with lOjum width wires was used as a static test object. This image is shown in Fig. 2(a). The overall resolution is better than 3 jum. Next,

189

we used a water droplet jet, similar to those used for EUV lithography studies,[6] to supply a stable but constantly-renewed object. The droplet driver is synchronized with the 2 kHz laser pulses. Fig. 2(b) shows a stroboscopic image of the water drops at 50 X magnification. The drops are -20 jum diameter. Figure 2(c) shows the case where the drop near the top of the image was pumped using pulses from the same laser system. The size of the laser focus is about 20 jum, with a peak power of 6.5xlO^^W/cm^andthe pump-probe delay was 1.7 ns. Figs. 2(b) and (c) were taken using 3 minutes exposure time; i.e. using 360,000 laser pulses. Figure 2(d) shows the evolution of the exploding droplets obtained by varying the relative timing of the ionizing laser pulse and the EUV illumination.

Fig. 1: Typical images (upper) and intensity profiles (lower) of the interference patterns of 13 nm EUV light generated using modulated waveguides filled with He at 150 torr. (a) and (b): using SOfim pinholes, placed 150 jum and 250^am apart respectively, with acquisition times of 240 seconds; (c) and (d) using 20 /^m diameter pinholes, placed 300//m and 450 ^m apart respectively, with acquisition times of 1800 sec. The pinholes are centered on the EUV, to ensure equal illumination. From Ref. 4.

In conclusion, these experimental results prove the feasibility of time-resolved plasma imaging using high repetition-rate lasers in a stroboscopic imaging mode. This demonstration required solving a number of experimental issues, such as isolating the CCD camera from plasma debris and light from the illuminating laser, preventing the buildup of ice stalagmites resulting from the droplet jet, and optimizing the imaging geometry. Further improvements will allow us to increase the illumination intensity by at least lOOx, allowing us to increase the image resolution. This setup is suitable for doing time-resolved movies of plasma dynamics, with femtosecond time resolution and -100 nm spatial resolution. Acknowledgements. This work was supported by the U.S. Department of Energy Stockpile Stewardship Academic Alliances Grant No. 190

DE-FG03-02NA00063, and by the National Science Foundation.

i;^:'M*#

-'&^^^. 4-f

Fig.2. EUV images using high harmonic generation at 30nm. (a). 50x magnified image of a static object, (b). 50x magnified image of water droplets, (c). 50x magnified image of water droplets with the second drop from the top illuminated by 800 nm, 22 fs laser pulses at a peak intensity of 6.5 x 10^"^ W/cm^. (d) Series of images with 0-2 nsec pump-probe time delay showing the expansion of a water droplet after being hit by a laser pulse.

References 1.

2. 3.

4.

5. 6.

R. A. Bartels, A. Paul, H. Green, H. C. Kapteyn, M. M. Murnane, S, Backus, I. P. Christov, Y. W. Liu, D. Attwood, and C. Jacobsen, in Science, vol. 297, pp. 376-378, 2002. R. A. Bartels, A. Paul, M. M. Murnane, H. C. Kapteyn, S. Backus, Y. Liu, and D. T. Attwood, in Optics Letters, vol. 27, pp. 707-709, 2002. A. R. Libertun, X. Zhang, A. Paul, E. Gagnon, T. Popmintchev, S. Backus, M. M. Murnane, H. C. Kapteyn, and L P. Christov, in Applied Physics Letters, vol. 84, pp. 3903-3905, 2004. X. Zhang, A. R. Libertun, A. Paul, E. Gagnon, S. Backus, I. P. Christov, M. M. Murnane, H. C. Kapteyn, R. A. Bartels, Y. Liu, and D. T. Attwood, in Optics Letters, vol. 29, pp. 1357-1359, 2004. M. Wieland, R. Frueke, T. Wilhein, C. Spielmann, M. Pohl, and U. Kleineberg, in Applied Physics Letters, vol. 81, pp. 2520-2522, 2002. L. Malmqvist, L. Rymell, M. Berglund, and H. M. Hertz, in Review of Scientific Instruments, vol. 67, pp. 4150-4153,1996.

191

High-Order Harmonic Generation from Argon Ions up to 250 eV Emily A. Gibson^Ariel Paul\ Nick Wagner\ , Sterling Backus^ Margaret M. Murnane\ Henry C. Kapteyn\ and Ivan P. Christov^ ^ Department of Physics and JILA, University of Colorado and NIST, Boulder, CO 80309-0440 E-mail: [email protected] ^ Department of Physics, Sofia University, Sofia, Bulgaria Abstract. We demonstrate that harmonic generation from ions can extend to significantly higher energies than emission from neutrals. We extend the highest cutoff observed in argon using 800 nm light from 100 eV to 250 eV.

High-order harmonic generation (HHG) provides a useful source of coherent, ultrafast light in the extreme ultraviolet (EUV) region of the spectrum, with applications in lithography, high-resolution imaging, site-specific spectroscopy and bio-microscopy. In HHG, an intense laser pulse is focused into a medium, typically a noble gas. The highly nonlinear interaction between the laser light and the atoms creates higher-order harmonics in a coherent, low-divergence beam The highest possible photon energy from HHG is described by the cutoff rule: hv^ax = Ip + 3.17Up, where Ip is the ionization potential of the gas and Up oc I I X-^ is the ponderomotive energy for a laser beam of intensity I I and wavelength X. Thus, increasing the laser intensity should result in a linear increase in the highest observed harmonic for any gas species. However, in practice, above the saturation intensity where 98% of the gas is ionized, the large amount of plasma prevents HHG at higher photon energies. This is due to two effects: plasma induced defocusing of the laser, and reduced conversion efficiency due to the large phase mismatch between the laser and EUV light. Typically, the highest harmonic energies have been observed using atoms with a high Ip such as He, because the amount of ionization is lower for a given intensity. However, using larger atoms such as Ne and Ar could potentially increase the flux of harmonic emission due to their larger effective nonlinearity. Previously, the highest harmonic energy observed from Ar using short duration laser pulses, longer driving wavelengths, or high laser energies was only 150 eV. In our experiment,[l] we guide the driving laser beam in a hollow-core glass fiber filled with low pressure argon. At high intensities, the ionization of argon creates a spatially varying electron density that defocuses the laser beam. [2] The laser can nevertheless remain guided as a result of glancing incidence reflections from the walls, allowing higher laser intensities in a fully ionized medium than would otherwise be possible. This method allows us to observe harmonic emission from Ar up to 250 eV, or 100 eV higher than previous results using any other approach. At the intensities necessary to generate the harmonics, we are above the saturation intensity of neutral Ar -

192

therefore we attribute these harmonics to HHG from ions. In our experiment, we focus 3 mJ, 22 fs, pulses from a 1 kHz Ti: sapphire laser system into a 150 jum diameter, 2.5 cm long, hollow-core glass fiber filled with Ar. Figure 1 shows the harmonic emission from a straight fiber filled with 7 Torr of Ar at a peak laser intensity of ~ 9 x 10^"^ Wcm'^. The harmonic emission extends to ~ 180 eV.

1000

100 150

100

250

200

Energy (eV)

Fig. 1. Harmonic emission from a straight 150 pim inner diameter, 2.5 cm long fiber filled with low-pressure Ar (7 Torr), with a peak laser intensity of 9 x 10^"^ Wcm'^. A zirconium filter was used to reject the laser light. From Ref 1.

10000

150

200

250

300

350

Energy (eV)

Fig. 2. Harmonic emission at a higher peak laser intensity of 1.3 x 10^^ Wcm"^, using a 0.25 mm period modulated fiber. The fall off at lower energies is due to the silver filter used to reject the laser light. From Ref. 1. In Figure 2, we show the harmonic emission from a 0.25 mm-period modulated

193

fiber filled with 5 Torr of Ar at a higher laser intensity of - 1.3 x 10^^ Wcm'^. In this case, we observe harmonic emission up to the argon L-edge at 250 eV. The modulated fiber enhances the signal by high-order quasi-phase matching.[3,4] Using ADK ionization rates, we calculate that at the laser intensity required to generate photons above -160 eV, argon is fully ionized. Therefore, the emission is from argon ions. In conclusion, we demonstrate that HHG from ions results in a dramatic extension of the cutoff harmonic energy. We show it is possible to overcome the problem of plasma-induced defocusing of the laser by using hollow-core fibers, allowing us to achieve high laser intensities in an ionized gas. Using the same fiber, we can enhance the harmonic signal by quasi-phase matching even at high ionization levels. This method can in principle be scaled to higher laser intensities, allowing the possibility of generation of keV energy harmonics. Acknowledgements. This work was supported by the Chemical Sciences, Geosciences and Biosciences Division of the Office of Basic Energy Sciences, U.S. Department of Energy, by the National Science Foundation, and made use of facilities supported by the Engineering Research Centers Program of the National Science Foundation under NSF Award Number EEC-0310717.

References 1 2 3 4

194

E..A.Gibson et al., "High-order harmonic generation up to 250 eV from highly ionized argon", Physical Review Letters 92, 033001 (2004). S. C. Rae, "lonization-induced defocusing of intense laser-pulses in high-pressure gases," Optics Communications 97, 25 (1993). A. Paul et al., "Quasi-phase-matched generation of coherent extreme-ultraviolet light", Nature 421, 51 (2003). E..A.Gibson et al., "Coherent soft x-ray generation in the water window with quasi-phase matching", Science 302, 95 (2003).

High-Order Harmonic Generation from Femtosecond Laser-Aligned Molecules Kenzo Miyazaki, Masanori Kaku, Keita Masuda, and Godai Miyaji Institute of Advanced Energy, Kyoto University, Gokasho, Uji, Kyoto 611-0011, Japan E-mail: [email protected] Abstract. We report a sensitive method using high-order harmonic generation to detect revival structures in fs-laser-induced, field free alignment of a molecular rotational wave packet. The revival structure is directiy observed on the time-dependent harmonic signal produced with the pump-probe technique. The harmonic signals for N2 and O2 represent the characteristic behaviors depending on the molecular structure. 1. Introduction Laser-induced spatial alignment of molecules suggests a new possiblity of strong laser filed to control molecules and its applications [1,2]. Recently, present authors [3] and Zeidler et al. [4] have reported that a time-dependent revival structure of femtosecond-laser-induced molecular alignment can be detected using high-order harmonic generation (HHG). In our experiment, a nonresonant 40-fs laser pulse was focused in a N2 gas jet to form a ground-state rotational wave packet, and the temporal evolution of wave packet was detected by observing the HHG. Since the harmonic yield is sensitive to the molecular orientation with respect to the laser field [5], the time-dependent harmonic signal represents the revival structure in the field-free alignment of molecules. The revival structures in the ps- and fs-laser-induced alignment of molecules have been observed so far using Coulomb explosion of molecules [6-8]. Here we present the characteristic results observed with the HHG for N2 and O2. 2. Experimental The experiment was performed using fs laser pulses (1 TW, 40 fs, 800 nm, 10 Hz) from a Tiisapphire chirped-pulse amplification system [2,3]. The linearly polarized output was split in two to produce a pair of pump and probe pulses with a variable time delay A^. The pulse energy was controlled by a set of thin-film polarizer and wave plate. The pump and probe pulses were recombined collinearly and focused by a 50-cm focal length lens into a pulsed supersonic molecular gas jet of 1-mm in diameter. The focused laser intensity was in a range of (0.5 - 2.5) x 10^"^ W/cm". The pump pulse energy was kept low enough to produce neither ion nor harmonic signal. Then, the pump pulse would induce a ground-state rotational wave packet of molecules, and the delayed probe pulse could generate high-order harmonic radiation from aligning molecules. The harmonic radiation was detected by an electron multiplier mounted on a VUV monochromator. The signal was processed by a boxcar averager, converted to the digital signal with an A/D converter and stored on a personal computer. 3. Resutls and discussion Figure 1(a) shows a typical example of the 19-th harmonic (k -^ 42.1 nm) signal observed for N2 as a function of the dme delay Ar. The harmonic signal

195

10 TIME DELAY (ps)

20

30

40

50

60

70

FREQUENCY in Be

Fig.l. (a) Time dependent harmonic signal for N2 and (b) its Fourier transform. Pump and probe pulse intensities are 0.8 and 1.7 x 10^"* W/cm^, respectively. represents the typical revival structure of a rotational wave packet formed by the pump pulse. At Ar - 0, the pump pulse exerts a torque on molecules to align along the laser polarization direction, while coherently exciting a number of rotational states to form a rotational wave packet in the ground state N2. The harmonic signal at A^ < 0 is produced from the non-aligned molecules. When the probe pulse is overlapped with the pump at Ar '^ 0, the harmonic signal is minimized due most likely to strong and rapid ionization induced by the superimposed intense fs pulses. The first large enhancement in the signal appears at Ar ~ 0.2 ps, after the pump pulse interaction is over. This first peak indicates the onset of dynamic alignment of a rotational wave packet. The peaked harmonic signal rapidly decreases, but the signal is kept at a level higher than that for Ar < 0 due to the effect of alignment. A full revival of the alignment takes place at A^ -^ 8.4 ps, corresponding to the revival time T= l/(2Bc), where the rotational constant B = 2.0 cm'^ for N2. The rapid signal change at the full revival comes from the fact that the rotational wave packet aligns in the direction perpendicular and parallel to the laser polarization axis in ~ 0.2 ps. In Fig.l (a), the strong enhancement of the harmonic signal is also observed at At ~ 4.2 ps or the half revival time 7/2 = l/(4Bc). Here, the wave packet is also aligned but 180° out of phase with that at the full revival. Figure 1 (b) represents the Fourier transform of the time-dependent harmonic signal shown in (a). It is clearly demonstrated that the harmonic signal is dominated by beats between rotational states populated through the Raman transitions (AJ = ± 2), being composed of beat frequencies Bc(4J+6). The frequency distribution also shows the intensity alternation between spectral lines that is a well-known influence due to the nuclear spin of N2. In Fig.l (a) the harmonic signal suggests a partial revival of the rotational wave packet at (1/4)7 and (3/4)7. The Fourier analysis has shown that the small revival signal is due to contributions in opposite phase from the wave packets formed by even- and odd-7 states of N2, while the wave packets formed by both J states contribute in phase to the revival structures at 7/2 and 7. As shown in Fig.l (a), the alignment of the rotational wave packet in N2 is formed after the pump pulse interaction. This ensures that the probe fs pulse itself provides no effect on the alignment dynamics observed. The result also supports our recent observation that the neutral N2 molecules are never aligned during the ultrashort interaction with the 40-fs pulses [2].

196

ii|iiii|iiU|iiii|iiii|mi|M

jlf^i^*-^^ l.l,...I.M.I....I....I....I....I....I....I....I,...l....l....l.. 6 8 10 TIME DELAY (ps)

Fig.2. Time-dependent harmonic signal for O2. Pump and probe intensities are 0.5 and 1.2 x 10'"* W/cm^, respectively.

ANGLE (degree)

Fig.3. Harmonic signal as a function of angle between field polarization and aligned molecular axis.

The time-dependent harmonic signal was also measured for O2, and a typical result is shown in Fig.2. A full revival of the wave packet is observed at T = 11.6 ps, corresponding to B = 1.4 cm'^ for O2. In contrast to the result for N2, the revival signals at TI4, Til, and 37/4 represent almost the same amplitude as that at T. This is due to the fact that only odd-7 states are populated in O2 molecules having no nuclear spin. This has been confirmed by the Fourier transform of the harmonic signal. The beat frequencies between J and J±2 states consist of spectral lines separated by ^Bc, since only odd-J states contribute to the spectrum. The analysis has also demonstrated that the small revival signal observed at r/8, 3r/8, 57/8, and 77/8 arises from beats between (7+2) and (7-2) states. Figure 3 shows the 19th harmonic signal measured as a function of angle 9 between the aligned molecular axis and the polarization direction of the probe laser filed. In the measurements, A? is fixed on the signal peak at the full revival time 7 = 8.4 ps for N2 and on the minimum at 7 = 11.6 ps for O2, while the probepulse polarization is rotated. The harmonic signal for N2 decreases with an increase in 9, while the signal for O2 is enhanced at 9 = 45 -^ 90°. The characteristic behaviors of angle-dependent harmonic signal are due most likely to the different orbital symmetry in the ground states of N2 and O2 [9]. Acknowledgements. The authors would like to thank F.H.M.Faisal for his suggestions. References 1 H.Stapelfeld and T.Seideman, Rev. Mod. Phys. 75, 543 (2003). 2 K.Miyazaki, T.Shimizu and D.Normand, J. Phys. B 37, 753 (2004). 3 M.Kaku, K.Masuda, and K.Miyazaki, Jpn. J. Appl. Phys. 43, 4B, L591 (2004). 4 D.Zeidler et a/., in Ultrafast Optics IV, Edited by F.Krausz, G.Korn, P.Corkum, LA.Walmsley (Springer-Verlag, New York, 2004) p.247. 5 N.Hay, R.Vellota, M.Lein, R.Nalda, E.Heesel, M.Castillejo and J.P.Maranos, Phys. Rev. A 65, 53805 (2002). 6 Rosca-Pruna and M.J.J.Vrakking, J. Chem. Phys. 116, 6567 (2002); V.Litvinyuk, K.F.Lee, RW.Dooley, D.M.Rayner, D.M.Villeneuve and RB.Corkum, Phys. Rev. Lett. 90, 233003 (2003); RW.Dooley et a/., Phys. Rev. A 68, 23406 (2003). 7 A.J.Becker, A.Becker, and FH.M.Faisal, Phys. Rev. A 69, 23410 (2004).

197

Efficient generation of high-order sum and difference frequencies in the XUV region by combining a weak, longer-wavelength field Yutaka Nomura, Tsuneto Kanai, Shinichirou Minemoto, and Hirofumi Sakai Department of Physics, Graduate School of Science, The University of Tokyo, 7-3-1, Kongo, Bunkyo-ku, Tokyo 113-0033, Japan E-mail: [email protected] Abstract. We demonstrate the efficient generation of sum and difference frequencies in the extreme-ultraviolet wavelength region both experimentally and theoretically. The importance of the wavelength of the weak combined field is emphasized.

1. Introduction High-order harmonic generation (HHG) has attracted a significant attention because of its potential as a coherent extreme-ultraviolet (XUV) radiation source. In pursuing high efficiency and tunability, using a two-color laser field is one of the promising approaches [1-4]. Sum- and difference-frequency mixing enables us to generate wavelengths not obtainable by the "pure", i.e., usual odd-order harmonics. However, the intensities of the sum and difference frequencies reported so far are not high enough, compared to the pure harmonics. Here, we demonstrate the efficient generation of high-order sum and difference frequencies in the XUV region by combining a weak, longer-wavelength field both experimentally and theoretically. The importance of the combined-field wavelength is pointed out for the first time to our knowledge.

2. Experimental setup In the experiment, the sum- and difference-frequency radiations are generated in a pulsed supersonic gas jet of N2 and Ar. As the fundamental field, pulses from a Tiisapphire system (A ~ 800 nm, T ~ 50 fs) are used. We combine fundamentals (A = 1064 nm, r ~ 12 ns) of a Nd:YAG laser. The Tiisapphire and the YAG beams are spatially overlapped by a dichroic mirror and are focused with a lens (f= 300 mm) into the gas jet. Intensities of the Tiisapphire and the YAG pulses at the focus are 2x10"^ and 6x10'' W/cm^ respectively. Harmonics are not generated by only the YAG field. Generated pure harmonics, sum frequencies, and difference frequencies are spectrally dispersed using a 1-m grazing-incidence monochromator and detected by an electron multiplier (EM). The signals from the EM are accumulated by a digital oscilloscope and processed in a PC.

198

3. Results and discussions Figure 1 shows HHG spectra observed in N2. Without the YAG field [(a)], the plateau extends to the 21st harmonic. When the YAG field is applied with the polarization parallel to that of the Ti: sapphire field [(b)], two peaks appear in the lower- and the higher-energy sides of pure harmonic peaks. The two lower-energy side peaks are shifted by 0.4 and 0.8 eV from the pure harmonic peaks. By considering the energy difference (~ 0.38 eV) between the Ti:sapphire and the YAG photons, these peaks are reasonably attributed to the sum frequencies of the Tirsapphire photons and one and two YAG photons, respectively. In the same way, the peaks on the higher energy side are attributed to one- and two-YAGphoton emitting difference frequencies. We refer to peaks corresponding to oneand two-YAG-photon absorbing (emitting) processes as +1 and +2 (-1 and -2) peaks. As shown in Fig. 1(b), the intensities of the +1 and -1 peaks increase as the photon energy increases, while those of the pure harmonic peaks decrease. Of particular note is that the intensities of the +1 and -1 peaks are comparable to or even larger than those of pure harmonic peaks for photon energies higher than ~ 25 eV, though the intensity of the YAG field is two orders of magnitude lower than that of the Tirsapphire field. On the other hand, when the polarization of the YAG field is perpendicular to that of the Tirsapphire field [Fig. 1(c)], the oneYAG-photon processes are suppressed due to the parity conservation law and the low intensity of the YAG field [5]. Similar results are observed also in Ar. ^ 0.5-1 - J jiou. 'c

Harmonic order 15th 17th 119th 121st

O1M_JLJLJJ!:I

13 J^

0

^

0.0

2 0.5 H (C)

~" 0.0-H20

ill 25

30

35

Photon energy (eV)

Fig. 1. Harmonic spectra observed in N^ (a) without and (b, c) with YAG pulses. The polarization of the YAG pulses is (b) parallel and (c) perpendicular to that of Tirsapphire pulses. The spectra are corrected for monochromator and EM efficiencies. Theoretical calculations support the experimental results. We extended the Lewenstein model [6] to study HHG processes using two-color fields. At the plateau region, the calculated intensities of the sum and difference frequencies are approximately one order of magnitude lower than those of the pure harmonic. On the other hand, near the cutoff, they are comparable with those of the pure harmonics. These results are consistent with our experimental observations. Furthermore, theoretical calculations also show the importance of the wavelength of the combined field. Figure 2 shows the calculated intensities of +1 and -1 peaks for the (a) 15th (plateau) and (b) 23rd (cutoff) nonlinear orders as a function of the wavelength of the weak combined field. The wavelength of the

199

fundamental is fixed to 800 nm. One can easily recognize that the sum and difference frequencies are efficiently generated when the wavelength of the combined field is close to or longer than that of the fundamental field (0.8 |Lim). In Fig. 2(a), the oscillatory structures from 0.8 to 10 |Lim are due to quantum interference effects and vary significantly when the intensities of the two fields are changed. In Fig. 2(b), the intensities of the sum and difference frequency converge to a certain value as the wavelength increases, which corresponds to the value when a static field is combined with the fundamental field (the longwavelength limit). 1.5

(a)

c 0.0

1

u

1.0 •

1

Wavelength (}am)

10

1

10

Wavelength (jum)

Fig. 2. The intensities of (a) 15th- and (b) 23rd-order sum- (solid curves) and differencefrequencies (dashed curves) as a function of the wavelengths of the weak combined field. Relative intensities used in the experiments for Ar are assumed.

4.

Summary

In summary, we demonstrate the efficient generation of the sum and difference frequencies by combining a weak laser field with a wavelength longer than that of the fundamental field. We emphasize that the wavelength of the combined field is of crucial importance for the efficient sum and difference frequency generation.

References 1. M. D. Perry and J. K. Crane, Phys. Rev. A, 48, R4051, 1993. 2. H. Eichmann, S. Meyer, K. Riepl, C. Momma, and B. Wellegehausen, Phys. Rev. A, 50, R2834, 1994. 3. M. B. Gaarde, P. Antoine, A. Persson, B. Carre, A. L'Huillier, and C.-G. Wahlstrom, J. Phys. B, 29, L163, 1996. 4. R. Hasbani, E. Cormier, and H. Bachau, J. Opt. Soc. Am. B, 16, 1880, 1999. 5. S. Long, W. Becker, and J. K. Mclver, Phys. Rev. A, 52, 2262, 1995. 6. M. Lewenstein, P. Balcou, M. Y. Ivanov, A. L'Huillier, and P. B. Corkum, Phys. Rev. A, 49, 2114, 1994.

200

Effects of Target Condition on Solid Surface Harmonics in the Extreme Ultraviolet Range Tsuneyuki Ozaki^ Jean-Claude Kieffer\ Hidetoshi Nakano^, and Atsushi Ishizawa" ' INRS, Universite du Quebec, 1650 Lionel-Boulet, Varennes Quebec J3X 1S2 Canada E-mail: [email protected] ^ NTT Basic Research Laboratories, 3-1 Morinosato-Wakamiya ,Atsugi, Kanagawa 2430198 Japan E-mail: [email protected] Abstract. Solid surface harmonics, a potential source for attosecond pulses, is investigated in the extreme ultraviolet spectral range. High-order harmonics up to the 16-th order are observed using silicon wafer targets. A significant difference is observed in the harmonic intensity for different target material. From systematic investigations with visible harmonics, we find that the preplasma scale length that is produced on the target surface has a large effect on solid surface harmonic generation.

1. Introduction Solid surface harmonics is an alternative method of generating high-order harmonics in the extreme ultraviolet (xuv) and soft x-ray spectral range. In this scheme, a p-polarizedrelativistic intensity femtosecond laser pulse is focused onto a solid target, producing a thin layer of high-density plasma on the surface. The component of the laser's p-polarized electric field normal to the target forces the surface plasma to oscillate in the direction perpendicular to the target surface. This large-amplitude "oscillating mirror" modulates the pump laser itself, producing harmonics in the specular direction. One intriguing by-product of solid surface harmonics is the possibility of generating high energy, single, isolated attosecond pulses, by focusing the pump beam to a spot size of one wavelength of the pump laser [1]. However, due to its physical nature, high efficiency with this scheme can only be realized by optimizing factors such as the pedestal and prepulse of the pump laser, as well as the target material. The main goal of the present paper is to investigate this latter effect of target condition on the conversion efficiency of surface harmonic generation. We shall first demonstrate the generation of xuv harmonics for various targets. The effects of target condition are further investigated using low-order harmonics in the visible spectrum.

2. Extreme Ultraviolet Harmonics In these experiments, outputs from a high intensity Ti:sapphire laser system (35 mJ, 55 fs, 790 nm center wavelength) are focused onto a solid target using an offaxis parabolic mirror. The typical peak intensity of the pump pulse is 5x10^^ W 201

cm"^. To observe the spectrum of the xuv harmonics in the specular direction, we developed a flat-field spectrometer using the type 639 grating (Hitachi), with a central groove density of 600 lines/mm. The high intensity pump laser reflected from the target in the specular direction is attenuated by an anti-reflection coated cylindrical glass window. The glass window is also used to collimate the highly divergent harmonics. A microchannel plate is placed at the focal plane 1 ' » i' ' M of the spectrometer to observe the J 60 harmonic spectrum. We also use a In i 1 1. commercial spectrometer (Ocean Optics S 50 . 1 ^ s USB2000) to observe the low-order ^ 40 1 11 1 I 1 1visible harmonics. 'r S: 1 i .1 In Fig. 1 we show the xuv spectrum 1 30 1 using a silicon wafer target. The pump 0 i laser energy for this shot is 26 mJ, and 20 the peak intensity is 4x10^^ W cm'^. J . 1 . . 1 1...,.,l,. 1 i l l 45 60 50 55 The time-integrated nature results in a Wavelength [nm] relatively strong background continuum originating from ionic emission of the F i g . l . Spectrum of xuv harmonics plasma However, a trace of the between 43 and 60 nm. The numbers spectrum and wavelength calibration correspond to the harmonic order. reveals clear harmonics, corresponding to the 14-th to 16-th harmonic of the 790 nm central wavelength pump laser. We can also perform interesting "tricks" by actively controlling the spectrum of the pump laser. The typical spectrum of the amplified pulse is centered at 790 nm, with a bandwidth of 35 nm FWHM. We generate in this spectrum a high spectral intensity narrow component, by finely adjusting the alignment of the oscillator. The narrow spectrum component has a peak intensity that is three-times larger than that of the broad component, and its bandwidth is below the 1 nm spectral resolution of the visible spectrometer. 50 55 The xuv spectrum between 43 and 60 nm Wavelength [nm] observed using this amplified pulse is F i g . 2 , Spectrum of xuv harmonics shown in Fig.2. We see that the using pump laser with a narrow harmonics generated using a pump laser spectral component. with a narrow spectral component also has a narrow component. This is the first demonstration showing that the shape of the solid-surface harmonic spectrum can be controlled by varying the spectrum of the pump laser.

1 '

1V

202

iL

3. Target Material Effects XUV harmonic experiments were also attempted using aluminum and silver targets coated on BK7 glass substrates. However, clear indications of harmonics were not observed In order to investigate the reason for this target dependence, we investigated the harmonics in the visible spectral range, where the harmonic intensities were much stronger. The results for aluminum and non-coated BK7 glass targets are shown in Fig.3. For aluminum targets, the results for both p- (O) and s-polarized pump lasers (T) are t

T^r

t

1

.

iQ

1000 c

1

6 c shown. We see that harmonics are •? j ^"^ 3 4 stronger for p-polarized pumps, which .Qk. 2 is consistent with the polarization re 100 selection rule of solid surface ^* (fi 6 harmonics. We also see that the co 4 c second harmonic emission from BK7 2 glass targets (open circles) is an order O X 10 (0 of magnitude higher than that for aluminum targets, for pump " ' *" "i6 " 2. " intensities of 1 - 2x 10'^ W c m l ^^m intensity [10 Wcm"! This observation can be explained F i g . 3 . SHG intensity generated by the difference in the ionization using BK7 target and p-polarized threshold of the two targets, and due to pump (circle), Al target and pthe resulting different scale length of polarized pump (triangle), and Al the preplasma produced on the target target with s-polarized pump (upside surface at the time when the main down triangle), high-intensity pump laser irradiates the target. Metadlic targets begin to ionize at pump intensities of 100 mJ cm"^, which is much lower than the lO-'ZO J cm'^ threshold of BK7 glass. As a result, preplasma on metallic target is generated and starts to expand at a time earlier than that for BK7 glass, resulting in a preplasma with a longer scale. For second harmonic generation, PIC simulations have shown that maximum conversion efficiency is obtained for a preplasma scale length of 0.59 K [2], where X is the wavelength of the pump laser. &

•

#

-

4. Conclusions We have demonstrated solid surface harmonics up to the 16-th order (49.4 nm) using a modest-sized Tirsapphire laser system. Significant difference in harmonic intensity is observed for metallic-coated and bulk glass targets. Additional experiments suggest the effects of preplasma scale length as the major cause for this phenomenon.

References 1 N. M. Naumova et al, Phys. Rev. Lett. 9 2, 063902 (2004). 2 M. Zepf ^^ a/., Phys.Rev.E, 5 8, R5253 (1998).

203

Control of the frequency chirp rate of high harmonic pulses Jens Biegert\ Mathis Bmck^ Christoph P. Hauri\ Arne Heinrich\ Florian W. Helbing\ Wouter Kornelis^ Philip Schlup\ Ursula Keller^ Rodrigo LopezMartens^, Johan Mauritsson^, Per Johnsson^, Katalin Varju^, Anne L'Huillier^, Mette Gaarde^, and Ken J. Schafer^ ^ Swiss Federal Institute of Technology (ETH), Physics Department, Zurich, Switzerland E-mail: [email protected] ^ Department of Physics, Lund Institute of Technology, Lund, Sweden ^ Department of Physics and Astronomy, Louisiana State University, Baton Rouge, USA Abstract. We measured the frequency chirp rate of harmonics 13 to 23 in argon by crosscorrelation with a short femtosecond probe pulse. We directly measured the negative chirp due to the atomic dipole phase and demonstrated that an additional chirp on the pump pulse was transferred to the ^th harmonic as q times the fundamental chirp. High harmonic generation is a widely used method to produce ultrashort extreme ultraviolet (XUV) pulses [1]. Recently, methods have been developed to measure the time-dependent frequency of individual harmonics [2-4], which is a direct measure of their coherence properties. An important application of these measurements is in the characterization of the relative coherence between successive harmonics, since this determines the temporal properties of their superposition. The time-dependent frequency of the emitted radiation is strongly related to the dynamics of the generation process and each harmonic contains contributions from several space-time quantum orbits [5-7], which have different phase characteristics dependent on laser intensity and harmonic order. We measured and manipulated the time-dependent frequency of harmonics 13 through 23 generated in argon with laser pulses centered around 815 nm. Minimizing ionization effects and keeping the harmonic flux constant, we performed a first direct measurement of the order-dependent phase coefficient a^ (describing the harmonic dipole phase) by varying the spectral phase of the driving field. By varying the fundamental chirp we show that the driving laser pulse phase is transferred to the ^th harmonic as q times the fundamental phase [8]. Detailed simulations show that our results are consistent with the interpretation that the imposed chirp simply adds to the intrinsic chirp. The time-frequency structure of our harmonic pulses is characterized via crosscorrelation frequency resolved optical gating (XFROG) in a magnetic bottle electron spectrometer [2]. We use transform-limited 12-fs pulses to probe the higher harmonics of the fundamental pulses on which different chirps are imposed and whose duration thus varies between 35 and 90 fs. The pump and probe pulses are both continuously monitored using two online spectral phase interferometry for direct electric field reconstruction (SPIDER) devices [9-11].

204

We carried out numerical simulations [8,12,13] which were designed to closely match the experimental conditions. In particular, we used as input fields the fundamental and probe pulses retrieved from the SPIDER measurements. A typical result is shown in Fig. 1 where experimental (upper row) and theoretical (lower row) data are compared for sideband 18, obtained using fundamental pulses with different spectral phases. We found that, in both the experiment and the simulation, the slope of the sideband, which corresponds to the chirp of the harmonics, had the same sign as the chirp of the fundamental pulse and also increased with its magnitude. We note that when the fundamental pulse is transform limited (case (c)), the harmonics exhibit a negative chirp.

27.5 28.3 27,5 28.3 27.5 28.3 27.5 28.3 27.5 28.3 Energy (eV)

Fig.l: Photoelectron signal corresponding to sideband 18 for negative (a), zero (c) and positive (e) fundamental chirp rates To investigate the origin of the harmonic chirp rates, we studied their orderdependence for the same fundamental chirp rates as in Fig. 1. Figure 2 shows a comparison between experimental and theoretical results for these five cases. The agreement between the experiment and theory is very good for the cases (a), (c), (d), and (e). For our experimental conditions the harmonics are made close to the peak of the driving pulse and thus the dipole chirp contribution is linear. Furthermore there is a contribution from the spectral phase of the fundamental, which in these cases is close to a linear chirp. There is no such contribution for a transform-limited pump pulse (case (c), in Fig. 2) which allowed us to extract the phase coefficient a . We find that the magnitude of a increases with harmonic order, which is characteristic of the shortest quantum oroit (black line in Fig. 2). The measured a , together with the prescription that the fundamental chirp rate was transferred to the ^th harmonic as q times, was used to predict the chirp

205

rates for the non transform limited cases in Fig. 2. The predictions are shown as lines and their agreement with both the experimental and the simulated results shows the robustness of the transfer of the fundamental chirp to the harmonics. x10 e)

A

A

A A

I

A

£\

A J

1

d)'"

S- 0

i § -1 •

" " ^ " - - - « ^

•"

E

•

^7

= -2 b)

12

14

16 18 20 Harmonic order

^ 22

: 24

Fig.2: Measured (closed symbols) and simulated (open symbols) chirp rates as function of harmonic order for same pump chirp rates as in Fig. 1. The lines are explained in the text. In conclusion, measuring the harmonic chirp rates as a function of order allowed us to obtain information on how the single atom attosecond dynamics evolve during the driving laser pulse. We have shown that the laser chirp is transferred to the harmonics according to the semiclassical model of harmonic generation. Furthermore, by manipulating the spectral phase and the duration of the driving laser pulse, we can exert control over the temporal properties of the harmonics and of attosecond pulse trains.

References [I] M. Ferray et al., J. Phys. B 21, L31 (1988). [2] J. Norin et al, Phys. Rev. Lett. 88,193901 (2002). [3] T. Sekikawa et al., Phys. Rev. Lett. 88,193902 (2002). [4] T. Sekikawa, T. Kanai, and S. Watanabe, Phys. Rev. Lett. 91,103902 (2003). [5] M. Lewenstein et al., Phys. Rev. A 52, 4747 (1995). [6] M. Bellini et al., Phys. Rev. Lett. 81, 297 (1998). [7] P. Salieres et al.. Science 292, 902 (2001). [8] J. Mauritson et.al., Phys. Rev. A 70, 021801(R) (2004) [9] C. laconis and L A. Walmsley,Opt. Lett. 23, 792 (1998). [10] C. laconis et al., IEEE J. Quant. Electr. 35, 501 (1999). [II] W. Kornelis et al.. Opt. Le«., 28, 281, (2003) [12] A. L'Huillier et al., Phys. Rev. A 46, 2778 (1992). [13] K. J. Schafer and K. C. Kulander, Phys. Rev. Le«. 78, 638 (1997).

206

Wavefront Control in High Harmonics Generation with Few- and Many-optical-cycle Laser Pulses p. Villoresi^ S. Bonora\ M. Pascolini\ L. Poletto\ C. Vozzi^ G. Sansone^ Stagira^ M. Nisoli^

S.

' INFM Laboratory for Ultraviolet and X-ray Optical Research - D.E.I. - Universita degli Studi di Padova, v. Gradenigo 6, 35131 Padova, Italy E-mail: [email protected] ^ INFM - Dipartimento di Fisica, Universita degli Studi, Milano, Italy ^ INFM National Laboratory for Ultrafast and Ultraintense Optical Studies- Dipartimento di Fisica, Politecnico, Piazza L. da Vinci 32, 20133 Milano, Italy Abstract. The wavefront of a ultrafast pulse is modified with a deformable-mirror. The generation of high-order harmonics (HHs) with both few- and many-optical-cycle pulses is controlled and optimised with a closed loop that controls the wavefront shape.

The purpose of the present v^ork is to investigate the role of the laser wavefront in the HHs generation. This is obtained with an optical system, using a deformable mirror (DM), that may induce modifications on the wavefront of the pulse that generates the HHs. Cciietic iilgcriihin

1

JL^^^

i^t\.%m^ihmmmt%

\s^^4''*''*'''****^\

• •

% 1Hf am)\n'i^S'^m'

t

1

'^^v

*

HHG

KpeclfiMiJCte : » ^: *

* *|^^Jj|lJ

»

'"^ '

Hsap Fig. 1 Experimental setup used for the closed-loop control of the HHs generation with a modified laser wavefront. The meaning of the acronyms is provided in the text.

Since the wavefront influences both the intensity and phase distribution in the focal region, where occurs the interaction with the gas, its control is crucial. We realized a closed loop on the control of the deformable mirror, that takes the intensity of the HHs in a selected spectral portion and use as merit function for an genetic algorithm that drives the mirrors searching the maximum intensity. Figure 1 sketches the closed loop used in the experiment. The laser pulse is generated with a CPA system, and is then injected in a hollow fibre (HF)

207

compressor [1]. The laser pulse is then directed on the deformable mirror (DM), for the modification of the wavefront and to a focusing optics, realized by a spherical mirror of 250 mm focal length. The interaction medium is a steady gas jet about the laser focus, realized by a rectangular nozzle of 150 |im of width and with a length of 4 mm along the laser propagation axis. The HHs beam is analysed by a grazing incidence spectrometer using a combination cf two toroidal mirror and a plane varied line-spacing (VLS) grating [2]. A portion of the spectrum is selected by a slit and measured by a photomoultiplier, PD, which can slide along the focalplane image. The PD signal is then acquired by a computer as a merit function for the genetic algorithm that controls the DM. The closed-loop approach here described was applied on the HHs generation from both the pulses with about two-optical-cycle duration, typically 6 fs, and these with many-cycle, with about 20 fs duration. The merit function in the former case was set in the upper plateau region of the HHs spectrum, in order to enhance the number of harmonic order. The result of the application of the loop is compared with the harmonic spectrum obtained with the deformable mirror in a flat configuration, so to realize the generation without the wavefront control. flat mirror optimum case

Wavelength (nm)

Fig. 2. Harmonic spectra in neon (pulse duration 6 fs; pulse energy 400 jiJ; gas jet thickness 4 mm). The duration of the optimisation is of the order of 4 minutes.

Figure 2 shows the comparison of the flat-mirror and the optimal configuration obtained with neon gas as target, and 400 |iJ of input energy. A very evident enhancement of the harmonic signal was obtained, together with the extension of the spectrum cutoff of above 16 nonlinear orders (8 peaks). The optimum spectrum has the top-plateau portion with irregular shape and the cutoff portion with weakly resolved harmonic peaks, in agreement with the model based on the effects of the carrier-envelope-phase, recently presented by our group. Figure 3 shows the result of the closed loop optimisation using many-optical-cycle pulses on argon gas. The target here chosen was the plateau intensity enhancement. The lineshape is asymmetrical as resulted from the propagation effects in the extended interacting medium. Also in this regime, the harmonics in the

208

plateau after the optimisation are much stronger with respect of these in the flat mirror case. The interpretation of the effect of the optimisation algorithm on the mirror actuators and on the pulse wavefront has been carried out by a Hartmann sensor. From this latter data, the calculated laser field in the interaction region can be obtained via the diffraction integral. The phase front acquired in the flat mirror case resulted affected by nonlinear effects as well as residual optical aberrations of the collimation and focusing optical systems. This leads to an non-symmetrical intensity distribution, with an intensity peak, in the case of the Fig. 2 spectrum, of Iflat =2 10^"^ W/cm^. In the case of the optimal wavefront, the intensity distribution is very symmetric and peak intensity results of lopt =6 10^"* W/cm^ [3].

800

-flat mirror -optimum case

600

200

25

30

35

Wavelength (nm)

Fig. 3. Harmonic spectra in argon (pulse duration 20 fs; pulse energy 190 |iJ; gas jet thickness 4 mm). In conclusion, we have shown that the control of the laser wavefront is very effective in the enhancement of the HHs intensity. This approach is complementary with respect to the optimisation of the pulse temporal structure, that was already carried out with success by other groups. The findings point out that the laser wavefront for the flat-mirror case, for both pulse lengths, may differ significantly from being optimal, and the corrections introduced on the wavefront are very effective.

References 1. M. NisoH etal, Appl. Phys. Lett. 68, 2793-2796 (1996) 2. L. Poletto , G. Tondello and P. Villoresi, Appl. Opt. 42, 6367-6373 (2003) 3. P. Villoresi etal, Opt. Lett. 29, 207-209 (2004)

209

Phase-driven strong-field processes in the multioptical-cycle regime Giuseppe Sansone\ Salvatore Stagira\ Caterina Vozzi^, Michele Pascolini^, Luca Poletto^, Paolo Villoresi^, Giuseppe Tondello^, Sandro De Silvestri\ and Mauro Nisoli^ ^ INFM - Dipartimento di Fisica, Politecnico, Piazza L. da Vinci 32, 20133 Milano, Italy E-mail: [email protected] ^ INFM - Dipartimento di Fisica, Universita degli Studi, Milano, Italy ^ INFM - DEI- Universita di Padova, Padova, Italy Abstract. We show, both experimentally and theoretically, that the strong-field processes involved in high-order harmonic generation are significantly affected by the pulse carrierenvelope phase even in the multi-optical-cycle regime. Experimental evidences of the role of the phase offset between the carrier and the envelope [carrier-envelope phase (CEP)] of few-optical-cycle light pulses have been obtained in strong-field photo-ionization [1], and in the process of high-order harmonic generation [2,3]- In this work we demonstrate, both experimentally and theoretically, that the CEP of 20-fs pulses (multi-cycle regime) induces clear signatures in the spectral characteristics of high-order harmonic radiation [4]. Moreover, we demonstrate that the short and long electron quantum paths contributing to harmonic generation are influenced in a different way by the CEP of a light pulse [5]. In particular, clear phase effects are visible on the long quantum paths even in the multi-optical-cycle regime, whilst the short quantum paths are significantly influenced by the CEP only in the few-opticalcycle regime. This is the first clear evidence that the experimental access to the carrier-envelope phase is fundamental for a complete investigation of light-matter interaction processes in the femtosecond regime. This observation is consistent with results reported by Christov et ah, demonstrating that coherent control of electronic processes in the strong-field regime can be achieved by shaping a laser pulse on an attosecond time scale [6]. In the experiments the harmonic beam is produced by focusing 20-fs light pulses, generated by a Tiisapphire laser system, into an argon jet. To observe the harmonic radiation on a single-shot basis a high-resolution flat-field spectrometer and a high-resolution CCD detector have been used. The single-shot spectra reported in this work correspond to consecutive light pulses with almost identical temporal and spatial intensity profile, spectral amplitude and phase. Figure I shows two single-shot harmonic spectra measured when the gas jet was located around the laser focus. Besides the odd harmonic peaks, discrete and well resolved peaks are clearly visible in between, whose spectral position is pulse dependent. Such peaks disappear locating the gas jet after the laser focus. The experimental results can be understood in the framework of the strong-field approximation (SFA) [7].

210

20 21 Harmonic order Fig. 1. Measured single-shot spectra in argon generated by two 20-fs driving pulses (gas jet located around the laser focus); the inset shows an enlarged view of the spectra. Using a nonadiabatic three-dimensional numerical propagation model [8] it is possible to demonstrate that small variations of either the peak intensity or the duration of the driving pulses cannot induced the observed spectral shift. Moreover such calculations strongly support the conclusion that the physical mechanisms, which determine the spectral behavior shown in Fig. 1, are based on the interaction of the driving pulse with a single atom. For this reason we have calculated the nonadiabatic single-atom response of the nonlinear medium , using the saddle-point approximation (SPA) [7], generalized to account for nonadiabatic effect. The nonadiabatic SPA is the ideal tool for the investigation of the effects of CEP. In fact, in the framework of the SPA, it is possible to identify the contributions to the harmonic emission from the different quantum paths. We point out that, from an experimental point of view, efficient selection of the short quantum paths can be achieved when the gas jet is placed after the laser focus; whereas, when the gas jet is located before the laser focus, the contribution of the long quantum paths increases [9]. Using the nonadiabatic saddle-point method it is possible to show that CEP significantly influences the short quantum paths only in the few-optical-cycle regime. In fact, the XUV spectrum calculated as coherent superposition of the short path contributions is characterized by sharp peaks, corresponding to the odd harmonics of the ftmdamental radiation. Such spectrum is not affected by the CEP of the driving pulse. Figure 2 shows the spectra calculated as coherent superposition of short and long paths, for two CEP values. The CEP dependent peaks located between the odd harmonics of the fundamental radiation are due to the long quantum paths. Therefore, we can conclude that the long quantum paths are more sensitive to even small changes of the electric field induced by different CEPs, as a consequence of the longer time spent by the electrons in the continuum after ionization. Since the calculated spectral features are in good agreement with the experimental results, we can conclude that the physical mechanisms, which

211

determine the spectral behavior shown in Fig.l, are based on the nonadiabatic interaction of the driving pulse with a single atom. 1

10 8

•

6

-

'

1

'

1 —

~-

13

(|) = 571/8 A

4

6 = 0 A/

/ \

\ Mb .•

1\

•(/)

c

CD

2 n

^ fti

-J L A I'^LJ \

Yf V

19

20 21 Harmonic order

T'^

J

22

Fig. 2. Single-atom emission spectra calculated using the saddle-point method, for two different CEPs.

The reported results lead to the general conclusion that phase-driven strong-field processes are feasible even in the multi-cycle regime. We expect that our observations, together with the use of the single-shot measurement technique, will promote new developments in the field of coherent phase-controlled phenomena in a novel and easily accessible temporal regime.

References 1 G.G. Paulas, F. Grasbon, H. Walther, P. Villoresi, M. Nisoli, S. Stagira, E. Priori, and S. De Silvestri, Nature (London) 414, 182, 2001. 2 A. Baltuska, Th. Udem, M. Uiberacker, M. Hentschel, E. Goulielmakis, Ch. Gohle, R. Holzwarth, V.S. Yakovlev, A. Scrinzi, T.W. Hansch, and F. Krausz, Nature (London) 421, 611, 2003. 3 M. Nisoli, G. Sansone, S. Stagira, S. De Silvestri, C. Vozzi, M. Pascolini, L. Poletto, P. Villoresi, and G. Tondello, Phys. Rev. Lett. 91, 213905, 2003. 4 G. Sansone, C. Vozzi, S. Stagira, M. Pascolini, L. Poletto, P. Villoresi, G. Tondello, S. De Silvestri, and M. Nisoli, Phys. Rev. LeU. 92, 113904, 2004. 5 G. Sansone, C. Vozzi, S. Stagira, and M. Nisoli, Phys. Rev. A, in press, 2004. 6 LP. Christov, R. Bartels, H.C. Kapteyn, and M.M. Mumane, Phys. Rev. LeU. 86, 5458,2001. 7 M. Lewenstein, Ph. Balcou, M.Y. Ivanov, A. L'Huillier, and P.B. Corkum, Phys. Rev. A 49, 2117, 1994. 8 E. Priori, G. Cerullo, M. Nisoli, S. Stagira, S. De Silvestri, P. Villoresi, L. Poletto, P. Ceccherini, C. Altucci, R. Bruzzese, and C. de Lisio, Phys. Rev. A 61, 63801, 2000. 9 P. Salieres, B. Carre, L. Le Deroff, F. Grasbon, G.G. Paulus, H. Walther, R. Kopold, W. Becker, D.B. Milosevic, A. Sanpera, and M. Lewenstein, Science 292, 902,2001.

212

Bright high-order harmonic generation at 13 nm and coherence measurement H. T. Kim, I. J. Kim, V. Tosa, Y. S. Lee, and C. H. Nam Dept. of Physics, Coherent X-ray Research Center, KAIST, Yuseong-gu, Daejeon 305-701, Korea htkim@kaist. ac.kr

Abstract: Bright high-order harmonic generation at 13 nm was achieved by utihzing selfguided and chirped laser pulses in a 9-mm long Ne gas jet. The spatial coherence of the bright harmonics was measured by double-pinhole interference. 1. Introduction Bright harmonic at 13 nm from a 9-mm long Ne gas jet was generated by using selfguided and chirped laser pulses. This long and dense medium must properly interact with intense laser pulse for the bright harmonic generation. Hence, the evolution of laser pulses in the ionizing medium should be controlled in both space and time domains. We obtained the conditions for self-guided propagation and laser chirp control to generate bright high-order harmonics [1]. With self-guided negatively chirped laser pulses, bright 61st harmonic at 13 nm was generated with low beam divergence and narrow bandwidth. For the examination of spatial coherence, we performed double-pinhole interference experiments using high-order harmonics from the long Ne gas jet.

2. Bright harmonic generation from a long Ne gas jet The experiment was performed with femtosecond laser pulses of 5-mJ energy and 827-nm center wavelength. The laser pulses were focused using a spherical mirror (f=l .2 m) onto a long neon gas jet with a 9-mm slit nozzle. The measured beam waist at the laser focus was 72 / / m in full width at half maximum (FWHM) and peak density of the gas jet was 40 Torr. When the gas jet center coincides with laser focus, the gas jet position is defined as z=0. To investigate the laser pulse propagation in an ionizing medium, we monitored the visible plasma image and laser beam profile at the end of medium. At z=0, we observed that the plasma image was bright only for the first 2-mm section of the medium due to the plasma defocusing. As the gas jet position moved away from the laser focus, the bright part of plasma image was extended and a nearly uniform plasma image over the entire gas jet was formed at z=-18 mm (the minus sign means a gas jet position before the laser focus). At z=-18 mm, the laser profile at the end of medium

213

was flattened and still intense, while the laser profile was smeared out at the gas jet position of z=0. Hence, the profile flattening and self-guiding of laser pulses can be achieved by proper selection of gas jet position. The self-guiding and profile-flattening can provide an ideal condition for well phase-matched strong harmonic generation in a large volume. In addition, the laser chirp should be controlled to obtain sharp and bright harmonic spectrum [2]. In ionizing medium, self-phase modulation (SPM) of laser chirp induces positive chirp of laser at the leading edge. Therefore, the negatively chirped laser pulses should be applied to compensate SPM effects on high-order harmonic generation when the harmonics are mainly generated in the leading edge. We achieved the optimization of harmonic brightness by the self-guiding and laser chirp control. Figure 1 shows the spectral brightness of 6r* harmonic with various gas jet positions and laser chirp. The brightness of 6r^ harmonic was optimized at the gas jet position of z=-18 mm and negatively chirped 42-fs laser pulses. At the gas jet position of z=-18 mm, the self-guiding and profile flattening of laser pulses forms an ideal condition for harmonic generation due to its uniform field distribution. In addition, profile flattening of laser pulses reduced the divergence of 6r* harmonic to 0.5 mrad by enlarging the phase-matched harmonic generation cross section. The positive chiq) of laser pulses, induced by SPM, was compensated by using negative chirped laser pulses. The bandwidth of 6V^ harmonic decreased to 0.07 nm with negatively chirped 42-fs laser pulses. Consequently, the self-guided and chirped laser pulses generated high-brightness 61 ^* harmonic with narrow bandwidth.

Fig. 1. Spectral brightness of the 6V^ harmonic with respect to gas jet position and laser chirp. The sign of pulse duration indicates the sign of laser chirp.

3. Double-pinhole interference of bright high-order harmonics For the examination of spatial coherence of bright harmonics from a 9-mm long Ne gas jet, we performed interference experiments using a double-pinhole plate [3].

214

The harmonics were generated by using negatively chirped 48-fs laser pulses at the optimized conditions for the 6V^ harmonic as indicated in Fig. 1. A thick Al foil (18jum thickness) with two pinholes was placed 25 cm after the gas medium. Each pinhole had 10-//m diameter and the separation of two pinholes was 60 jum. The harmonic beam size was about 100 jum in FWHM on the double-pinhole plate. The interference pattern was recorded on an X-ray CCD located 80 cm from the doublepinhole plate. Figure 2 shows the fringe pattern and its horizontal intensity profile. The fringe visibility was 0.7 at the center of interference pattern. Even though the bright harmonics from a 9-mm Ne gas jet did not exhibit full spatial coherence due to the broad spectral width of full spectrum from 12 nm to 20 nm, it still had a quite good coherence. With two pinholes with 100- jum separation, we observed a fringe visibility of 0.5. Thus, the bright harmonics from the 9-mm Ne gas jet possessed a good spatial coherence. 300

,^ B 250

1

c 3

Xi

1

w

1

i ^^° (0

g

^ \h

200

/I

100

/

_c 50 0- Ks^ftr

/

AJ ,

\nL MJ

^ ^1 ,

,

,

^U , T^.p.A,

Fig. 2. Double-pinhole interferogram of bright harmonics from a 9-mm Ne gas jet with 60-mm pinhole separation: (a) jfringe pattern, (b) horizontal intensity profile of the interferogram.

4. Summary We demonstrated the optimization of harmonic brightness using self-guided and chirped laser pulses in a 9-mm long Ne gas jet. Bright 61'* harmonic was generated with low beam divergence of 0.5 mrad and narrow bandwidth of 0.07 nm. The spatial coherence of the bright harmonics from the long Ne gas jet was examined by a doublepinhole interferogram. It showed a good spatial coherence of harmonics in most of the beam cross section.

References 1. H. T. Kim, et al., Phys. Rev. A 69 031805 (2004). 2. D. G. Lee, et al., Phys. Rev. Lett. 87,243902(2001). 3. D. G. Lee, et al., Opt. Lett. 28,480-482 (2003).

215

Attosecond Pulse Generation During the Laser Pulse Reflection at the Plasma-Vacuum Interface Alexander S. Pirozhkov, Hiroyuki Daido, Sergei V. Bulanov Advanced Photon Research Center, Japan Atomic Energy Research Institute, 8-1 Umemidai, Kizu-cho, Soraku-gun, Kyoto 619-0215, Japan. E-mail: [email protected] Abstract. We demonstrate the dependence of high-order harmonic generation and attosecond pulse formation during the laser - overdense plasma interaction on the carrier-envelope phase.

Introduction. Recently [1-4], the method of attosecond pulse generation using several high-order harmonics of a laser radiation [5] was experimentally fulfilled using noble gases as generating media. The energy of harmonics generated in gases was rather lov^. The energy of harmonics of a relativistic laser pulse reflected fi*om a solid target, in the contrast, can be as high as several mJ in the 10-30 nm region (2K sr) [6]. For this reason, use of high-order harmonics fi-om solid targets for the generation of attosecond pulses is of considerable interest. One of the models accounting for the emission of high-order harmonics during the reflection of a relativistic laser pulsefi*oma solid target or a thin foil is a model of oscillating mirror [7]. In this model, the target is represented by a thin layer of electrons, which oscillates as a whole under the influence of the incident laser pulse. Such collective behaviour of electrons is justified by PIC simulations [7, 8]. Due to the relativistic intensity of the laser pulse, electrons oscillate not only in the plane of the layer, but also along the direction of the wave vector [9]. This oscillation leads to a Doppler up- and dovm shift of the fi'equency of a reflected electromagnetic wave. This model is applicable when the influence of ions and the back reaction of the reflected radiation are negligible, so the normalized plasma density ^o = cc^ll{2(0oc) «1. Here, COQ is the laser fi-equency, c is the speed of light, cWp - (Anne^lmf'^ is the plasmafi*equency,n is the electron density, e is the electron charge, m is the electron mass, and / is the thickness of the plasma layer. The generation of attosecond pulses in the case of an underdense plasma with 8{)<\ was considered in [10]. We consider the reverse case of an overdense plasma with Nonlinear boundary conditions for the laser pulse reflection at the thin plasma layer. We adopt the model of the electron layer with a constant density. In the case of an overdense plasma considered here, the electron layer cannot be separated fi-om the immobile ion layer, so electrons cannot move in the direction perpendicular to the plasma layer. However, electrons can move along the layer. We consider a plane electromagnetic wave with a wave vector ^0 = {(Oolc cosO, COQ/C sinO, 0} obliquely incident on the foil located at x = 0. Here, 6 is the incidence angle. We perform the Lorentz transform fi-om the laboratory

216

reference frame into the reference frame moving along the y axis with a velocity V=c sinO. In the boosted reference frame, the field amplitude is EQ - £'ocos^, the frequency is COQ - COQCOSO, the wave vector is ^o' - {COQ/C COSO, 0, 0} and therefore the wave propagates along the normal to the foil [11]. A prime denotes quantities in the boosted reference frame. The dimensionless vector potential a'(0 taken at the foil position x~0 can be found from equation [8]

Here, ao*(0 is the vector potential of the incident pulse at the foil position, Po - {0, - tan^, 0} is the electron momentum in the absence of the electromagnetic field in the boosted reference frame, and f is measured in units of I/COQ. We use dimensionless units, where velocities are normalized to c, momenta to mc, fields to mccoo/e, and vector potentials to mc^/e. The electric field of the reflected wave is ^^P' -po E^{xJ^) = eo\ [po + aif+x)ll [ l l + [po + a'(r'+x)r J

(2)

Generation of high-order harmonics. Equation (1) was numerically integrated for the case of a p-polarized laser pulse with its amplitude a =eEo/(mccoo) - 8.8 and duration To=15/cyo (we use FWHM pulse durations), the incidence angle 0=2S.6\ If ^ = 800 nm, this corresponds to the intensity of about 1.6-10^^ W/cm^ and to the pulse duration of 6.3 fs. Reflected pulse spectra calculated in the laboratory reference frame are shown in the left part of Fig. 1. The normalized plasma density in the layer is ^ o - 8.8 (curves 1 and 2) and 13.2 (curve 3). The high-order harmonic intensity decreases with plasma density. Note that in the high-frequency part spectra do not longer consist of individual harmonics. They rather resemble a descending supercontinuum. Such behavior is characteristic for short pulses. Longer pulses produce spectra with a pronounced harmonic structure. Furthermore, in the case of a short pulse the shape of the spectrum of the reflected pulse depends on the carrier-envelope phase, so that changing this phase it is possible to obtain spectra with or without harmonic structure (curves 1 and 2). The similar effect in the completely different regime of not relativistic harmonics generation was experimentally observed in [12, 13], where high-order harmonics from a gas target were generated by using 5-6 fs laser pulses focused to the intensity below 10^^ W/cm^. Generation of attosecond pulses. In order to obtain the attosecond pulse from the reflected electromagnetic wave, we need to select a certain frequency band [1-5]. The right part of Fig. 2 shows intensity envelopes of reflected pulses that pass through a bandpass spectral filter IICOQ- 47COO (curve 1: cosine-like incident pulse, curve 2: sine-like pulse). The carrier-envelope phase influences the form of the reflected attosecond pulse. In the case of a cosine-like laser pulse, the reflected pulse is shorter, has higher energy and no satellite pulses. The reflected pulse (curve 1) has a duration of 0.41/a;o (0.17 fs for io=800nm), the conversion efficiency is 1.4* 10'^.

217

Fig. 1. Left: spectra of reflected pulses. Curve 1: ^o = 8.8, the cosine-like pulse. Curve 2: 6^0 = 8.8, the sine-like pulse. Curve 3: eo-13.2, the cosine-like pulse. Right: intensity envelopes of the incident pulse (curve 3) and spectrallyfilteredreflected pulses (curves 1 and 2). Curve 1: the cosine-like incident pulse, curve 2: the sine-like pulse. The spectral filter transmits radiation withfrequenciesfromllcooto 47COQ. Durations offilteredreflected pulses are r - 0.41/c9o (cosine-like incident pulse) and QAAICOQ (sine-like pulse).

Conclusion. In the model v^here the overdense plasma was represented by the thin electron layer, the motion of electrons in the direction perpendicular to the plasma surface was negligible. Using this model, we calculated high-order harmonics generated during the reflection of the high intensity laser pulse. We demonstrated that these high-order harmonics were phase-locked and formed attosecond pulses if one sums several harmonics. We showed that cosine-like laser pulses generated attosecond pulses with higher energy and shorter duration than sine-like ones. Acknowledgements. The work was supported by the Japan Society for the Promotion of Science (JSPS P-03543).

References

9 10 11 12 13

218

R. Kienberger et aL, Nature 427, 817,2004. Y. Mairesse et aL, Science 302,1540,2003. S. A. Aseyev et aL, Phys. Rev. Lett. 91,223902,2003. P. Tzallas et aL, Nature 426,267,2003. Gy. Farkas and Cs. Toth, Phys. Lett. A168,447,1992. M. Zepf et aL, presented on Jan. 6, 2004 at the REEFS workshop, APRC-JAERI, Kyoto, Japan. S. V. Bulanov, N. M. Naumova, F. Pegoraro, Phys. Plasmas 1, 745,1994. V. A YshivkoY etaL, Vhys. Plasmas 5,2727, 1998. L.D.Landau, E.M.Lifshitz, The Classical Theory of Fields, Pergamon, Oxford, 1983. N. M. Naumova et aL, Phys. Rev. Lett. 92, 063902,2004. A. Bourdier, Phys. Fluids 26, 1804, 1983. A Baltuska et aL, Nature 421, 611,2003. M. NisoH et aL, Phys. Rev. Lett. 91,213905,2003.

Energetic Proton and Deuteron Generation from a Microporous Polytetrafluoroethylene Film with Deuterated Polystyrene using a 2.4-TW Table-Top Laser H. Takahashi^'^ S. Okihara^ S. Ohsuka^ ^ M. Fujimoto^ ^ S. Okazaki\ T. Ito\ S. Aoshima^'^, and Y. Tsuchiya^'^ ^ Central Research Laboratory', Hamamatsu Photonics K.K., 5000 Hirakuchi, Hamakitacity 434-8601, Japan E-mail: [email protected]. co.jp ^ CREATE Shizuoka of JST, REFOST, 5000 Hirakuchi, Hamakita-city 434-0041, Japan Abstract. Protons and deuterons over 1 MeV in energy were generated by 2.4-TW, 50-fs, 10-Hz laser pulses interacting with a film target, hi particular, microporous polytetrafluoroethylene film infiltrated with deuterated polystyrene was effective for this target.

1.

Introduction

It has been demonstrated that multi-MeV particles, such as electrons, protons and hQ2i\j ions, are produced when an intense laser pulse interacts with a target of a gas, and thin film. This new technolog}^ has the potential to be used as an ion beam source alternative to a cyclotron [1,2]. The technolog}^ could also be applied to positron emission tomography (PET) for cancer screening by developing a compact production system of short-lived positron emitters, which can be generated by a nuclear reaction using an MeV-order ion beam. Such radioactive isotopes were produced at facilities with large-scale laser systems, wdiich can generate only single-shot pulses [2,3]. Positron emitters are usually produced by an accelerator via (p,n) and (d,n) nuclear reactions. Here, the (d,n) reactions are much more important than the (p,n) because of their threshold energies. Although generation of energetic deuterons using a table-top laser is believed to be difficult, there appear to be many w^ays to improve performance. In this report, w^e discuss the experiments carried out with a 2.4-TW table-top laser system using two types of laserirradiation target, both of which contain deuterated polystyrene (CgDg: d-PSt), and compare the energy distributions of generated protons and deuterons. From the results, we can clearly see a strong dependence of the energy distribution on the construction of the target.

2.

Experimental Setup

The experiments w^ere carried out using a table-top Ti:Al203 laser system. The

219

system delivered on a thin film target pulses of energ}^ 120 mJ, pulse duration 50 fs, and central wavelength 815 nm. The p-polarized laser beam was focused at normal incidence onto the target surface with an off-axis parabolic mirror. The maximum intensit}^ at the focal point was estimated to be 3x10^^ W/cm^. The target film was set on a rotating holder in a vacuum chamber maintained at 2x10"^ Pa, and was irradiated for about 120 seconds with 10-Hz repetition. The energ}^ and species of ions emitted from the target were measured using a Thomson parabola ion spectrometer with CR-39. The ion detector was placed in an almost completely forw^ard direction. We estimated the energ>^ and number of ions from the corresponding traces recorded on CR-39. In our experiment, we selected two t}^es of film and put d-PSt on them to make the laser-irradiation target. One of the selections was a Mylar film. We produced a target by coating a Mylar film of 6-|im thickness with a d-PSt layer of 3-|im thickness (DMylar). The other was a polytetrafluoroethylene (PTFE) microporous membrane filter. This selection is because PTFE is composed of only carbon and fluorine; making the added deuterons the lightest ions in the target. The PTFE membrane filter was a microporous plastic film of ~3-|Lim pore diameter, 83% porosity, 75-)Lim thickness, and 2.7-mg/cm^ density. Figure 1 shows a scanning electron micrograph (SEM) image of the PTFE substrate. From Fig. 1, we can see that the PTFE membrane filter is constructed of many fibers, the diameters of which are about 200 nm. We added d-PSt to PTFE, where the total amount of the absorbed d-PSt was estimated to be equivalent to a dense layer of 7-juim thickness. Since d-PSt was dispersed over the whole PTFE substrate, the produced DPTFE also had a microporous structure.

3.

Results and Discussion

Figure 2 shows the energy distributions of generated protons and deuterons. The distributions were calculated from the Thomson parabola traces obtained. Here, when DMylar was used as the target, the d-PSt layer was on the laser irradiation side. (When the direction of DMylar was reverse, energetic deuterons were too few to be detected.) From Fig. 2, we can clearly see that energetic protons and • • D O

Proton (D_PTFE) Deuteron (D_PTFE) Proton (D_Mylar) Deuteron (D_Mylar)

D

Energy [MeV]

Fig. 1. SEM image of the PTFE membrane filter used as the substrate of the D_PTFE target.

220

Fig. 2. Energy distributions of protons and deuterons emitted from a laser-irradiation target of D_MYlar and D_PTFE films.

deuterons were emitted from the DPTFE target. In particular, the maximum energies of protons and deuterons observed were 2.0 MeV and 1.6 MeV, respectively. Although the ideal DPTFE has no light hydrogen, our DPTFE target included protons; these protons were a contamination of the target probably from environmental water. Nevertheless, the number of protons emitted from DPTFE was similar to that from D_Mylar over the whole energ}^ range. By considering the mechanism of production of energetic ions, we can explain the experimental result that deuterons emitted from DMylar were fewer than those from DPTFE. Energetic ions are produced by energ}^ transfer from a hotelectron population to ions, in the target. Since DMylar contains a vast amount of light hydrogen, almost all the transferred energy is absorbed by protons; as a result, only smaller energy can be transferred to deuterons. In addition, figure 2 shows that the energies of protons and deuterons exhibit nearly-Boltzmann-like distributions. This result implies that generation of these energetic ions should be based on the mechanism of electron-induced electrostatic field acceleration. In the case of fon^^ard-emitted ions, however, this mechanism is believed to be effective only for a target, the thickness of which is less than about the Debye length of a laser-induced plasma, i.e., several micrometers under the present condition [4]. Furthermore, MeV ions w^ere obtained using the tabletop TW laser system, although the thickness of the DPTFE target used was greater than the reported optimum thickness, i.e., 20-40 fim [5]. Probably, these peculiar properties come from the internal construction of the target. That is, the microporous structure of PTFE can enhance the transmission of laser light in the target and expand the ionization area up to the rear surface. 4.

Conclusion

In conclusion, w^e succeeded in generating protons and deuterons over 1 MeV in energy by focusing 2.4-TW, 50-fs, 10-Hz laser pulses onto a film target. We selected two types of film substrates, i.e., Mylar and PTFE, and put d-PSt on them to make the target. Here, PTFE has a microporous structure, which is considered to be advantageous to particle acceleration. From the experimental results, the tncrgy and number of generated ions were found to be strongly dependent on the construction of the target. Acknowledgements. The authors would like to thank T. Hiruma, Y. Suzuki, and Professor S. Nakai for their encouragement. Professor S. Sakabe for his lending of a parabolic mirror, and M. Matsuda for his help with SEM obser\/ations.

References 1 2 3 4 5

K. W. D. Ledingham, P. McKemia, and R. P. Singhal, Science 300, 1107, 2003. I. Spencer et ai, Nucl. Instrum. & Methods in Phys. Res. B 183, 449, 2001. K. Nemoto et ai, Appl. Phys. Lett. 78, 595, 2001. S. P. Hatchett et ai, Phys. Plasmas 7, 2076, 2000. I. Spencer et ai, Phys. Rev. E. 67, 046402, 2003.

221

High-energy protons emitted from a polymercoated metal foil by 60-fs laser irradiation Hiroaki Kishimura^ Hiroto Morishita\ Yasuhisa H. Okano\ Yasuaki Okano\ Yoichiro Hironaka\ Ken-ichi Kondo\ Yuji Oishi^, and Koshichi Nemoto^, Kazutaka G. Nakamura* ^ Materials and Structures Laboratory, Tokyo Institute of Technology, 4259, Nagatsuta, Midori, Yokohama 226-8503, Japan ^ Central Research Institute of Electric Power Industry, 2-11-1, Iwado Kita,, Komae-shi, Tokyo 201-8511, Japan Abstract. High-energy protons are generated by a 60-fs-laser irradiation at 1.5 x 10^^ Wcm'^ on a polymer-coated thin metal foil. The intensity of protons from the polymercoated target is enlarged about 80-fold higher than that from the uncoated target. The maximum energy of protonfromthe polymer-coated target is twofold higher than that from the uncoated target.

1.

Introduction

An uiteraction between an ultrafast laser pulse and materials has been widely studied. A short-pulse high-intensity laser makes it possible to generate and accelerate ions to multi-MeV energies [1]. The force for proton acceleration is considered to originate from the electrostatic field in the plasma sheath on the target surface, which is sustained by high-energy electrons produced by an intense laser pulse. Plastic and metal targets are often used for proton acceleration experiments. In the case of metal targets, the origin of the proton is considered to be contaminations on the surface such as hydrocarbons and water. Recently, a double-layer target, which consists of metal and polymer layers, has been proposed to improve the efficiency of the ion acceleration [2, 3] and increases in proton energy and intensity have been demonstrated [4]. Here, we report the generation of fast protons from a polymer-coated copper foil by femtosecond laser irradiation at 1.5 X 10^^ Wcm'^. Quantitative comparisons of proton energies and intensities were performed between the polymer-coated and uncoated targets.

2.

Experimental Methods

The laser system used was a Ti:sapphire laser (B. M. Industries Co. Ltd., alOus), which can produce pulses having energies of up to 600 mJ, a pulse width of 60 fs at a wavelength of 790 nm, and a repetition rate of 10 Hz. The prepulse was present from 8 ns before the main pulse at an intensity contrast ratio of 10"^: 1. The pulse energy of 180 mJ was focused onto the target by an off-axis parabola mirror to p polarization at an incident angle of 30 degree with respect to the surface

222

normal in a vacuum chamber (10"^ Pa). The focal spot diameter was 50 fim and the maximum intensity on the target was calculated to be 1.5 x 10^^ Wcm'^. The targets for the experiment were copper foils of 5 |jm thickness. An additional layer (-^ a few microns) of pol)^inylmethylether (PVME), was coated dn the rear side of the ipopper foil by a solvent*c^st process. The target was moimted cm a motorized stage and moved for e^cb l$se^ shot. Ions emitted ^om the rear side of the target were cbllimate^ by a pinhdle (500 j^m diametefr, solid angle -^ 8.0 x 10"^ sr) and passed through a Thomson mass spectrometer (B=0.35 T, B-200 kV/in). The distance between the target and the pinhole was 157 mm. The ioil3 were recorded using a CR-39 plastic nuclear track detector^, which is sensitive to ions with energies above 100 keV. Tbe number of protons is obtained from that of pits. Data for 50 shots were accumulated for each target. The experimental i^etup is shown in Fig. 1. liisappliirej laser 60 fs

Pjlii bale

spectronsBter

1 Ptrabftla

g^ f o j | PVME layer

^^^

Wffrcif

Fig, 1. Experimental arrangement for ion emission measurement.

3. Results and Discussion Figure 2(a) shows a typical example of the ion track obtained for the PVMEcoated target along with a theoretical Thomson-parabola curve for protons. Thp experiinent^l curve is very close to the theoretical curves for protpiis. Figure 2(b) shows the prpton energy di3tribution obtained from th^ PVME-coated and the uncoated copper foil. F6r the lincoated target, the maxnmim ener^ of the protons is only ?S0 keV and the number of protons is l^ss than 10^ protons/keV/sr/sJiot at 100 keV. On the other band, the proton^ emitted from the coaled target were accelerated up to 570 keV. The number Of protons reaches up to 4 x 10^ protoi^s/keV/sr/shot at 160 keV and is about 80-fold higher than that from the uncojated copper target at 200 keVi The total numlier of prptons is estimatfed to be 7.2 X 10^ for the coated target and 1.4 x 10^ for the uncoated target Because contaminants ^ifli a couple of nanometers thickness c ^ provide e n o u ^ protons (approximately 10^^ protons), it is not clear that tiie CH layer is directly responsible for tjie increased yield. The drivi&g force of proton acceleration is thougl^t to be due to electrostatic field in flie plasiiia sheatili on the rear surface, which is formed by high-energy

223

electrons produced by an intense laser [5]. T(ie plasma conditions at the front Surface of the target are same CM* botli the PVME-cp^ted apfi iincoated cppp^r target. Those eleclrostatic fi^ld3 in the plasma steath at the rear surface may be different, which affects on the proton acceleration^ but evid^ce for determiiiing the drigin of protons has not bpen foimd yet Examination of the detailed i^echanism and determinaticMi tbe origin of protons would reqyire the use of a deuteratedPVME. . coaiingi • - " * -WjtnOUt L

'—^-^ PVME coating j

N.

\

§ 10

S

0^

3

10

„

"^"N.

% '**%

% % %

\\ \

\

„

\

\

rfl is.i'] I ' M J J I ' ^ i B,t,1,lMt [; I i vC\

100 200 300 400 500 600 Proton energy (keV) Fig. 2. Proton ttaces from a polyvinylmethylether -coated target, (a) The ion track obtained for the pVME-coated target along with a theoretical Tliomson-parabola curve for protons, (b) Comparison of proton spectra from the coated target (solid line) m^ uncpated target (dashed line).

Ackndwiedgements. This wcH-k was pjartially supported by Qrants~in-Aid for the JISPS fellowship and Scientific Research No. 14077209 (Scientific Research on Priority Areas) from Ministry of Educatibn, Culture, Sports, Science and Technology. The authors thank M. Hasegawa for h^s help in constructing the experimental setup.

References 1. T. Fujii, Y. Oishi, T. Nayuki, Y. Takizawa, K. Nemoto, T. Kayoiji, K. Horiok^ Y. Okano, Y. Hironaka, K. G. Nakamur^ and K. Kondo, Appl. Phys. Lett. 83, 1524, 2003. 2. S. V. Bulanov and V. S. Khoroshkov,, Plasma Phys. Rep. 28, 453,2002. 3. T. Zh. Esirkepov, S. V. Bulanov, K. Nishihai^a, T. Tajima, F. Pegor^o, V. S. Khoroshkov, K. Mima, H. Daido, Y. Kato, Y. Kitagawa, K. N^gM, ^ d S. Sakabe^ Phys. Rev. Lett. 89,175003,2002. 4. J. Badziak, E. Woiyna, P. Parys, K. Yu. Platonov, S. Jaboski, L. Ry, A. B. Vaiikov, and J. Woowski, Phys. Rev. L^tt. 87,215001,2001. 5. 3 P Hatchett, C. G. Brown, T. E. Cowan, E. A. Hemy, J. S. Johnson, M. H. Key, L A. Koch, A. Bmce JLangdon, B. F. Lasinski, R. W. Lee, A. J. Mackinnon, D. M. Ppnnjngton, M. D. POTy, T. W. Phillips, M. koth, T. Craig Sangster, M. S. Sii^gh, R. A Snavely, M. A. Stoyer, S. C. Wilks, aod K. Yasuike,Phys. Plasmas 7,2076,2000,

224

Estimation of proton source size generated by ultraintense laser pulses using a Thomson mass spectrometer Y, Oishi^ T. Nayuki^ T. Fujii\ Y. Takizawa\ X. Wang^ T. Sekiya^ A.A. Andreev^, K. Horioka^, T. Yamazaki^, and K. Nemoto^ ^ Central Research Institute of Electric Power Industry, 2-11-1, Iwado kita, Komae-shi, Tokyo 201-8511, Japan E-mail: [email protected] ^ Interdisciplinary Graduate School of Science and Engineering, Tokyo Institute of Technology, 4259, Nagatsuta-cho, Midori-ku, Yokohama 226-8502, Japan ^ Research Institute for Laser Physics, 12, Birzhevaya Line, St. Petersburg, 199034, Russia Abstract. Source sizes of proton beams generated by irradiation of 55 fs, 6.6 xio^^ W/cm^ laser pulses on a 5- /z m-thick copper tape target were estimated using a Thomson mass spectrometer. Typical beam sizes were 190 /z m at E=400 keV and 100 /x m at £"=1.1 MeV.

1.

Introduction

The measurement of the proton profile on the target surface can be very useful for studies on the acceleration mechanism of protons and leads to engineering applications, such as an increase in ion yield and improved directivity of ion beams. Recently, some groups have reported the proton beams profiles using penumbral imaging technique [1], direct surface imaging [2] and periodically structured test objects [3]. Here, we propose another method of estimation of the proton source size on the rear surface of a target using a Thomson mass spectrometer and report experimental results using this method. The experiment was conducted by irradiation of 55 fs, 6.6x10^^ W/cm^ laser pulses on a 5- ^tmthick copper tape target.

2.

Theory and Experiments

The key point is that we use a pinhole in front of the Thomson mass spectrometer to provide a spatial profile with energy resolution. However, the pinhole diameter is usually comparable to or larger than the ion source size. Thus, a numerical procedure is necessary to retrieve the ion source profile from the ion pattern obtained on the observing plane. Figure 1 shows a schematic drawing of the coordinates for calculation. f(x,y,E), g(X,Y,E), and g' (Y',E) are proton density profiles on the rear surface of the target, on the detector without the Thomson mass spectrometer, and on the detector with the Thomson mass spectrometer, respectively. Here, we consider one-dimensional profiles f(y,E), g(Y,E), and g'

225

(Y',E) for simplicity. /; is the distance from the target to the pinhole, I2 is that from the pinhole to the detector./(3^,£^ and g(Y,E) satisfy g{Y,E) = l f{y,E)h{yJ) -dy

(1)

2 ( / i + / 2 ) tan i9.

The denominator is derived from particle number conservation, when it is assumed that the protons are emitted homogeneously in a cone angle of ^^ (half angle). h(y,Y) determines whether protons at y can reach Y on the detector. In our experiment, these values were //= 17cm, /2 = 13 cm, and 0c = 20°for E >100 keV. We assume the dependence of 0c on E SLS ^c ^ E'^'^ based on ref. [4]. Next, the transfer from g(X E) to g'(Y', E) by the Thomson mass spectrometer should be considered. Protons with energy Eo are transferred at approximately (XEO, YEO)^( kx //~Eo ),kY / Eo ) , where kx, ky are the Thomson mass spectrometer constants. The I2 in eq. (1) should be replaced by I2+ V^(XEO^+YEO^) because the distance between the pinhole and transferred point depends on E. Finally, we take a Gaussian profile, such disf(y,E) = a(E) expf^-^ /b(E) ) , and find parameters a(E), b(E) which best fit the experimental result g'exp(Y', E) by calculating eq. (1). Thomson mass spectrometer

CR39

Target

f(x,y,E) f(y,E)

Profile on the target

g(X,Y,E) g(Y,E)

g'(Y',E) g'(Y\E)

Profile on CR39 without Thomson spectrometer

Profile on CR39 with Thomson spectrometer

Fig. 1. Schematic drawing of the coordinates for theory The experiments were carried out using a Ti:Al203 laser (THALES LASER, TheAlpha 10/US-20TW). The delivered energy to the experimental chamber was 220 mJ, the pulse duration was 55 fs. The p-polarized laser beam was focused using a fll> off-axis parabolic mirror, and was incident at an angle of 45°to the target surface. The focused spot size on the target was 4x11 ^tm^ in frill width at half maximum (FWHM),which included approximately 55% of the total energy. Therefore, the laser intensity was estimated to be 6.6x10^^ W/cm^. A thin copper tape of 5 /z m thickness was used as the target[5], and was translated for every shot so that a fresh surface was irradiated. The energy spectra of protons ejected perpendicularly from the target surface at the opposite side of the laser irradiation were measured using the Thomson mass spectrometer. The electric and magnetic fields of the spectrometer were 3.3x10^ V/m and 5.1 k Gauss, respectively. The deflection length by the fields was 4.5 cm. A 50 ^i m pinhole was located in front of the Thomson mass spectrometer. A nuclear track plate CR39 located 10 cm from the center of the deflector was used as an ion detector.

226

3.

Results and Discussions

Figure 2 shows the dependences of source size of proton beams in FWHM. We should pay attention to the fact that the size is not constant, such as 190 // m at E= 400 keV, 120 /xm at £" = 630 keV, and lOO^m at £" = 1.1 MeV. The slope change at approximately E = 400 keV is particularly large. Two or more temperatures in proton spectra observed thus far [6] might be related with the slope change. Proton source siz is large more than 10 times of the laser spot size, which is close to ref [2,3]. The force for proton acceleration is thought to be produced by the plasma sheath at the surface of the target and during the acceleration processes, the regime of proton acceleration should be spread. ,300

700

1200

E [keV]

Fig. 2. Dependences of source size of proton beams on proton energy £".

4.

Summary

In summary, we proposed and conducted the measurement of the energy-resolved source profile of proton beams on the rear surface of a target using a Thomson mass spectrometer. The source size of the proton beams on the target was found to be energy dependent: it clearly changes at approximately E = 400 keV. Typical beam sizes in FWHM were 190 M m at J? = 400 keV and 100 /z m at ^ = 1.1 MeV.

References 1 2 3 4 5 6

M. Zepf, et al., in Phys. Plasmas Vol 8, 2323, 2001. M. Roth, et al., in Plasma Phys. Controlled Fusion Vol. 44, B99, 2002. M. Borghesi, et al, in Phys. Rev. Lett., Vol. 92, 055003-1, 2004. Y. Murakami, et al, in Phys. Plasmas Vol. 8, 4138, 2001. T. Nayuki, et al, in Rev. Sci. Instrum. Vol. 74, 3293, 2003. I. Spencer, et al, in Phys. Rev. E. Vol. 67, 046402-1, 2003.; T. Fujii, et al, in Appl. Phys. Lett. Vol. 83, 1524, 2003.

227

Part III

Ultrafast Dynamics in Solid 1 (Phonon and Exciton)

Imaging Nanostructures with Picosecond Ultrasonic Pulses Brian C. Daly^ Niels C.R. Holme\ Takashi Buma\ Cyril Branciard^ Theodore B. Norris\ S. Pau^ D.M. Tennant^, J. A. Taylor^, and J.E. Bower^. ^ Center for Ultrafast Optical Science aad EECS Department, University of Michigan, Ann Arbor, MI 48109-2099, USA. E-mail: [email protected] ^ Bell Laboratories, Lucent Technologies, Murray Hill, NJ 07974 Abstract. We describe a novel imaging technique that employs coherent acoustic phonon pulses which are generated and detected by ultrafast optical methods. Sub-micron resolution images of Al patterns lithographically etched on a Si substrate are shown.

Progress in nanoscience is facilitated by new approaches to nanoscale imaging. In this paper, we describe an ultrafast opto-acoustic technique that enables highresolution (potentially ~45 nm) acoustic microscopy to be performed on nanostructures on a semiconductor substrate. State-of-the-art acoustic microscopes have a maximum resolution in the range of 1-10 microns, and use sound waves with frequencies as high as 1 GHz. We propose to improve this resolution dramatically by using considerably higher frequency (> 100 GHz) acoustic waves which are generated and detected by means of an ultrafast pump and probe technique known as picosecond ultrasonics. In picosecond ultrasonics, an ultrashort (typically 100-fs) optical pump pulse impulsively heats a thin (typically lO-nm-thick) metal film, and the resulting thermal expansion launches a roughly single-cycle coherent acoustic phonon pulse (a longitudinal strain pulse) into the substrate. If the acoustic pulse encounters a buried interface, the reflected component will return to the free surface where the strain-induced reflectivity changes can be detected by means of a time-delayed ultrashort optical probe pulse. This technique has become a powerful tool for nanometer-scale thin-film metrology and semiconductor process control [1,2]. We emphasize here, however, that picosecond ultrasonics is capable of imaging 2-D lateral nano-structures and potentially 3-D nano-structures as well. To accomplish this, high-resolution optical measurement of the transverse spatial variation of the acoustic wave is required. We can achieve this resolution by allowing the scattered (diffracted) acoustic wave to propagate a sufficient distance such that the spatial variation becomes broad enough to be measured by a 1 micron-diameter optical probe beam. Unfortunately, the ballistic propagation length of terahertz-bandwidth acoustic phonons is typically only a few microns, which is roughly two orders of magnitude shorter than the diffraction length required for our method. However, it has been shown recently that coherent phonon pulses can propagate for several mm in crystalline substrates such as Si at cryogenic temperatures [3].

231

The concept underlying our imaging method is illustrated in Fig. 1. Input pulses are scattered from an object plane I and time domain measurements of the scattered field amplitude w(Po,0 are made at a number of different detector positions PQ on a surface S' in the far field. If the field on the surface Z' is then time-reversed, it will "back-propagate" to the object plane due to the timereversal synunetry of the wave equation. The sum of these numerically backpropagated waves yields a reconstruction of the input field at the object plane, and therefore of the object itself. Thus the object can be reconstructed by using the measuredfieldas the input to the "time-reversed" diffraction integral [4],

d ( -w • dt V

r ^ 0?*

da'

(1)

^ ;

where c is the wave velocity, /lo is the normal to the far-field surface, and roi is the vector between PQ a point Pi in the object plane. We refer to this method as "time-reversal imaging," and have previously demonstrated its applicability to both 2-D [5] and 3-D [6] imaging with single-cycle terahertz electromagnetic pulses. Scattered Fields

Detection Plane Object Plane

Fig. 1. Illustration of the basic time reversal imaging concept. The scattered fields at the detection plane D' are used as the input to Eq. 1 in order to numerically backpropagate the waves to the input aperture plane Z.

A simple illustration of the picosecond ultrasonic imaging scheme in our experiment is shown in Fig. 2. Optical pump pulses derived from a mode-locked Ti:sapphire laser are focused (radius -3 [im) onto a patterned 15-nm Alfilmon a

232

high-purity (111) Si substrate that is 0.5 mm thick. The sample is kept at a temperature below 20 K in order to minimize attenuation due to phonon-phonon scattering in the Si. We used e-beam lithography to etch the Al film with lines or grids that are 1 \xm in width. Note that since the Al lines are illuminated by the optical piunp beam, the object to be imaged is actually the source of our detected waves. The optically induced thermal expansion of the Al film produces single cycle strain pulses composed of longitudinal acoustic phonons with peak frequency and bandwidth of -100 GHz, and amplitude of the order of IxlO'"^. The peak frequency corresponds to an acoustic wavelength in (111) Si of-90 nm; thus we ultimately expect this system to achieve a diffraction limited resolution of -45 nm. The acoustic pulses propagate to a second Al transducer on the other side of the Si substrate. There, a slight change in reflectivity induced by the time varying strain field can be detected at a series of positions by an optical probe beam (radius -0.5 p.m). The amplitude of the strain field at each position can therefore be measured in the time domain simply by varying the time delay between the optical pump and probe pulses. We have shown previously that for this sample, the experiment is capable of measuring phonon pulses in the far field of the acoustic source[7].

(111)SI Probe

(-0.2 nJ)

Acoustic Pulse 15 nm 0.5 mm

15 nm

UHrafast OptiosM ^ Pulses --200 fe, '^SIO nm Fig. 2. Schematic diagram of the experiment. Measurement of the spatio-temporal diffracted acoustic wave field allows us to employ the reconstruction algorithm based on time-reversal described above. It is important to note, however, that since we are not using light waves in a vacuum but rather sound waves in a crystalline solid, we are forced to account for elastic anisotropy [8]. In other words, to correctly perform the reconstruction algorithm, the term c in Eq. 1 must be replaced with a phonon group velocity, CQ that varies with propagation direction. In Fig. 3 we show the anisotropy-corrected acoustic image for the Al lines along with an SEM of the Al pattern. In Fig. 4 we show the same results for a structure with 2-D periodicity.

233

a)

b)

1^1

\0mm Fig. 3. a) SEM image of Al lines, b) Acoustic phonon image of Al lines. The white scale bar on both plots represents a length of 5 jLim (Note: the scale of a) and b) are not equal).

Fig. 4. a) SEM image of Al grid, b) Acoustic phonon image of Al grid. The white scale bar on both plots represents a length of 5 [xm (Note: the scale of a) and b) are not equal).

The results are an encouraging proof of concept for this technique. An analysis of the image indicates a resolution of-- 700 nm. Technical improvements should lead to a resolution much closer to the theoretical limit of- 45 nm. Although the initial experiment described here imaged an object grown on a substrate surface, this technique should be able to image buried structures as well. Since conventional metrology of buried structures is typically destructive, this method of high frequency acoustic microscopy could potentially provide novel information on such nanostructures. A second possibility is the imaging of biological samples bonded to a wafer surface.

234

References 1 2 3 4 5 6 7 8

C. Thomsen et al., Phys. Rev. B 34, 4129, 1986. H.J. Maris, Scientific American 278, 86, 1998. H.Y. Hao and H.J. Maris, Phys. Rev. Lett. 84, 5556, 2000. A.B. Ruffin et. a l . Opt. Lett. 26, 681,2001. A.B. Ruffin et. al., IEEE J. Quant. Electron. 38, 1110,2002. T. Buma and T.B. Norris, Appl. Phys. Lett. 84,2196,2004. N.C.R. Hohne et. al. Appl. Phys. Lett. 83,292, 2003. J.P. Wolfe, Imaging Phonons (Cambridge University Press, 1998).

235

Ultrahigh frequency acoustic phonon generation and spectroscopy with Deathstar pulse shaping Jaime D. Beers, Masashi Yamaguchi, Thomas Feurer, Benjamin J. Paxton, and Keith A. Nelson Department of Chemistry, Massachusetts Institute of Technology, Cambridge, MA 02139, USA E-mail: jdbeers@mit,edu Abstract. A novel pulse shaping technique is used to generate tunable multiplecycle acoustic waves with 2-500 GHz frequencies. These are used for study of nanoscale inhomogeneities and dynamics in amorphous materials.

1. Introduction Photoacoustic methods have permitted generation of broadband acoustic pulses of picosecond duration, with frequency content up to roughly 500 GHz and wavelength down to several nanometers [1]. Typically, femtosecond pulsed irradiation of a thin metal film gives rise to sudden thermal expansion and broadband acoustic wavepacket generation [2]. The acoustic response can be coupled into a material of interest, and its spectral modifications owing to dynamical and structural interactions may be determined through analysis of the "echoes" which return to the excitation surface after partial reflection from buried interfaces. [3-5]. However, signal/noise levels sufficient for reliable spectral decomposition at the highest frequencies of the reflected acoustic pulses often present challenges. The use of narrowband acoustic waves that are tunable throughout this high-frequency range would facilitate numerous spectroscopic applications, but generation of such waves has so far required fixed material structures such as multiple-quantum wells with specified spatial period [6] or films with specified thickness and multiple reflections at boundaries [7], with a specially fabricated sample to study each desired frequency. An alternative approach using femtosecond pulse shaping for generation of a narrowband excitation suggests itself. Here we introduce a novel pulse shaping method, called Deathstar, which permits facile generation of optical pulse sequences between 2-2500 GHz., a range far exceeding that allowed by traditional pulseshaping methods [8]. We demonstrate generation of tunable ultrahighfrequency phonons and their use for spectroscopy of amorphous Si02, whose lowtemperature thermal properties are attributed largely to the disorder-induced scattering of high-wavevector acoustic phonons [9].

236

2. Results and Discussion In the Deathstar pulse shaper, shown in Fig. 1, an incident pulse makes seven round trips in the multiple-retroreflector cavity. Partial transmission through a custom-designed reflector results in a Gaussian envelope for the resulting pulse train, whose frequency is tuned through motion of a single translational stage. The pulse train is used to irradiate a thin (15 nm) aluminum film that acts as a transducer and is coupled to the material of interest, here a glassy Si02 layer of thickness 100-1000 nm. The transmitted acoustic response is monitored interferometrically [10] at the back side of the sample by a delayed probe pulse.

Tpulse? '

Fig. 1. Schematic illustration of Deathstar pulse shaper and interferometric probe. The response of a bare aluminum transducer to multiple-pulse excitation across a wide range of frequencies is shown in Fig. 2. Figs. 2a and 2b correspond to excitation of the film at at its resonant frequency times V^ or 1, respectively. The highest frequency excitation, shown in 2c, shows an excellent signal/noise ratio; following broadband excitation the spectral content and signal/noise at this frequency would be very small. The gradual rise in signal is attributed to thermal diffusion from the back to the front of the film.

:i(bi

Fig. 2. Narrowband phonons generated with Deathstar pulse shaper. pulse trains of frequency (a) 100, (b) 200, and (c) 415 GHz.

:: m

17 nm Al film excited with

Coupling the phonons into a material sample.allows for acoustic spectroscopy across a wide frequency range. As the wave propagates through the material, it interacts with dynamical fluctuations on the timescale of the acoustic period, and with structural heterogeneities on the lengthscale of the acoustic wavelength. The damping is determined here by taking the derivative of the displacement, on the order of 0.4 nm, yielding the strain, on the order of 0.1%. Fig. 3 illustrates the strong damping that an acoustic wave at 150 GHz experiences on transmission

237

through two different thicknesses of Si02 at 290 K. By comparing the intensity of the transmitted wave through different propagation depths, the temperature and frequency dependent phonon mean free path may be determined, as given in Fig.

4. 2.5x10'-| • • A T

2.0x10*-

a. J

20 K Experimental 290 K Experimental 30 K Model 300 K Model

1.0x10*-

§ 5.0x10'-

'*•• v^.. . .

0.0. 150

200

250

300

350

400

450

500

Frequency (GHz)

-20

3.

0

20

Fig. 3 (left). Transmission through S i 0 2 at 290 K. Fig. 4 (above) Phonon mean free path at 290 K, 20 K, and calculated from tunneling model.[12]

Conclusions

A novel pulse shaper has permitted generation of tunable optical pulse trains from 2-2500 GHz and acoustic waves up to 500 GHz. These have been used to characterize amorphous material acoustic responses. The application of ultrahighfrequency acoustic spectroscopy to amorphous solids, liquids, and other complex materials promises to reveal important details about nanoscale structure and picosecond structural relaxation dynamics. Acknowledgements. This work was supported in part by the Department of Energy (grant no. DE-FG02-00ER15087).

References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10.

T. Saito, O. Matsuda, O.B. Wright, Phys. Rev. B 67, 20541 (2003). G.L. Eesiey, B.M. Clemens, C.A. Paddock, App. Phys. Lett. 50, 717-719 (1987). O.B. Wright, J. App. Phys. 71, 1617-1629 (1992). B. Bonello, B. Perrin, E. Romatet, J.C. Jeannet, Ultrasonics 35 223-231 (1997). T.F. Crimmins, A.A. Maznev, K.A. Nelson, App. Phys. Lett. 74, 1344-1346 (1999). t). Ozgur, C.W. Lee, H.O. Everitt, Phys. Rev. Lett. 86, 5604-5607 (2001). T.C. Zhu, H.J. Maris, J. Tauc, Phys. Rev. B 44, 4281-4289 (1991). A.M. Weiner, Prog. Quant. Elec. 19, 161-237 (1995). M.P. Zaitlin, A.C. Anderson, Phys. Rev. B. 12, 4475-4487 (1975) C. Glorieux, J.D. Beers, E.H. Bentefour, K. Van de Rostyne, K.A. Nelson, Rev. Sci. Inst., in press (2004). 11. R.M. Slayton, K.A. Nelson, A.A. Maznev, J. App. Phys. 90, 4392-4402 (2001). 12. A.K. Raychaudhuri, Phys. Rev. B 39, 1927-1931 (1989).

238

Probing of Thermo-Acoustic Transients in Materials Using EUV Radiation Ra'anan Tobey\ Erez Gershgoren\ Mark Siemens^ Margaret M. Murnane\ Henry C. Kapteyn^ Thomas Feurer and Keith A. Nelson^ ^ Department of Physics and JILA, University of Colorado and NIST, Boulder, Colorado 80309-0440 E-mail: [email protected] ^ Department of Chemistry, Massachusetts Institute of Technology, 77 Massachusetts Avenue, Cambridge, MA 02139 Abstract: The first application of EUV high-harmonic light to probe the nonlinear optical photoacoustic response of a sample is reported. This will enable measurements of thermal and acoustic transients in materials with sub-lOOnm resolution and wavelength.

Nondestructive acoustic methods for material analysis have found increasing utility, both in fundamental science and in industry. As an example, microelectronic manufacturers can implement nondestructive analysis to characterize films properties, interfaces and defects in semiconductor and metal structures. In the last decade, optical techniques have been found to work well for generating and probing both longitudinal and surface acoustic response in a variety of materials and structures [1,2]. An ultrashort light pulse incident on a surface interacts first with the electron gas; this electron heating then transfers to the lattice though electron-phonon interactions. In photoacoustic probing, the thermal and acoustic response of the lattice is then monitored in real time using another ultrashort pulse in either a zero background diffractive geometry, or a reflective geometry. Longitudinal wave generation is largely dependant upon the electron-phonon scattering length and the optical absorption depth. Surface acoustic phenomenon, while dependent on the above factors, is also dependent on the excitation geometry. Optical excitation of the sample in a transient grating geometry can generate a spatially-periodic impulsive surface excitation with a minimum wavelength that corresponds to the wavelength of the incident light; i.e. about 500nm. Furthermore, transient grating probing of the material response is limited to wavelengths longer than that of the probe laser. To characterize surface acoustic modes at very high frequencies, one must make use of smaller spatial periods. One obvious way is to shorten the excitation wavelength into the ultraviolet, extreme-UV, or soft x-ray regions of the spectrum. In this paper, we report on the first experiments to use ultrashort-pulse EUV light pulses created using high harmonic generation to detect photoacoustic material transients. [3] The experimental apparatus makes use of a 2 kHz repetition-rate Titanium-doped-sapphire chirped pulse amplifier [4]. Pulses of energy 1.75mJ and

239

pulse duration 27fs are separated into pump and probe beams. The probe beam is upconverted into the EUV using a phase-matched geometry in an Argon-filled glass capillary, producing a comb of harmonics centered at 30nm with about 20/iW of power [5,6]. Both pump and probe are then spatially and temporally overlapped on our sample. The sample in this experiment consists of copper stripes on an oxidized silicon substrate (damascene) forming both amplitude and phase grating. Optical pumping of the structure differentially excites the copper and silicon, and the time delayed diffraction intensity of the EUV is monitored. Fig. 1. shows a typical scan where thermal expansion and subsequent surface acoustic oscillations can be easily seen. Absolute changes in diffraction efficiency are very large and approach 10" . Grating periods down to 1 micron have been investigated to date, and sub-micron periods will be studied in the near future. 0.00

-0.01

Q

-0.02

-0.03

2000

4000

6000

8000

Time Delay (ps)

Fig. 1. Typical photoacoustic data of damascene probed by EUV light. The data shows the first order diffraction from .Smicron copper stripes in a 1 micron period grating. From Ref. 3. Probing with EUV light has several distinct advantages over its optical counterpart. First, as can be seen from the scan above, the signal levels are very high. In the particular scan above, using moderate pumping energy, we show a signal level of 3%. This signal level already exceeds the analogous optical experiment by an order of magnitude. Depending on the pumping fluence, we have observed scans where signal levels exceed 10"\ surpassing their corresponding optical detection by up to 2 orders of magnitude. This is expected, considering that the thermal/acoustic excitation of the surface, usually measured as fractions of a nanometer, is a larger fraction of the EUV wavelength than of the optical wavelength for the two different probing conditions. Figure 2 shows the relationship between signal level and pump intensity; by comparing the data with calculations of expected diffraction efficiency, we demonstrate that we can reliably see surface displacements as small as .02nm, or X/1500 at 30nm.

240

Surface Displacement (nm) 0.1

0.2

0.3

0.4

0.5

Pump Energy (mJ/cm )

Fig 2. Pump dependence data demonstrating increased sensitivity down to .02nm (points, bottom axis) with linear fit (solid line). Simulations(dashed line, top axis) show excellent agreement. From Ref. 3. In conclusion, we have demonstrated that we can adapt standard photoacoustic and photothermal techniques to shorter EUV wavelengths. Large signal levels and good signal to noise look promising for future, more elaborate, all-EUV, transient grating experiments to excite and detect transient material dynamics on the sub-lOOnm spatial scale. Furthermore, using the full bandwidth of our source, we will have the ability to make these measurements energy selective. Acknowledgements. This work was supported by the Chemical Sciences, Geosciences and Biosciences Division of the Office of Basic Energy Sciences, U.S. Department of Energy, and was supported in part by the Engineering Research Centers Program of the National Science Foundation under NSF Award Number EEC-0310717.

References 1. 2. 3. 4. 5. 6.

J.ARogers, et.al., Ann. Rev. Mat. Sci., 30: 117 (2000). W.S. Capinski and H.J. Marris, Rev. Sci. Inst., 67 (8): 2720 (1996). R. I. Tobey, E. H. Gershgoren, M. E. Siemens, M. M. Murnane, H. C. Kapteyn, T. Feurer, and K. A. Nelson, Applied Physics Letters, vol. 85, pp. 564, 2004. S. Backus, et.al.. Opt. Lett. 26 (7): 465 (2001). C.G. Durfee, et.al.,Phys. Rev. Lett., 83 (11): 2187 (1999). A. Rundquist, et. al.,Science, 280 (5368): 1412 (1998).

241

Ultrafast dynamics of coherent electron-phonon interaction in silicon Masahiro Kitajima^'^, Muneaki Hase^'^, Anca Monia Constantinescu^, Hrvoje Petek^

and

'National Institute for Materials Science, Tsukuba 305-0047, Japan E-mail: [email protected] ^Department of Physics and Astronomy, University of Pittsburgh, Pittsburgh, PA 15260, USA ^Grad ^Graduate School of Pure and Applied Sciences, University of Tsukuba, Tsukuba 305-8577, Japan Abstract: A coherent response of silicon to excitation with a 10-femtosecond laser pulse is reported. Transforming the transient reflectivity signal into frequency-time space reveals the carrier-phonon interactions leading to the coherent phonon generation.

1.

Introduction

The interaction between electrons and phonons via the Coulomb force defines many of the physical properties of Si, e. g., the electrical conductivity: real and virtual scattering processes transfer carriers between different momentum states and renormalize their mass [1]. Much of our knowledge of the electron-phonon interaction in Si is limited to transport measurements at near equilibrium conditions. Measurements of the carrier relaxation in silicon by optical techniques have been restricted mainly to transient changes of index of refraction following excitation by intense laser pulses [2,3], because the indirect band gap of Si obviates more direct probes. In this paper we report on the coherent response of the coupled Si carrier-lattice system to excitation by a 10 fs, 406 nm laser pulse. The transient anisotropic reflectivity signal is measured and transformed into timefrequency space revealing the dynamics of formation of the coherent longitudinal optical (LO) phonon at 15.3 THz [4], and its subsequent dressing by interaction with the photogenerated carrier plasma.

2.

Experimental

The anisotropic transient reflectivity of n-doped Si(OOl) (-1x10^^ cm'^) under ambient conditions was measured by the electro-optic sampling[5]. Nearly collinear, pump and probe pulses (10 fs pulse duration; 406 nm (3.05 eV) wavelength; 80 MHz repetition rate) are focused to overlap at a 10 |Lim spot on the sample. The pump pulse with 40 mW average power generates a carrier density of 5x10^^ cm'^. After reflecting from the sample, the probe beam was analyzed into polarization components parallel and perpendicular to that of the pump, A/?// and

242

A/?i, and detected with photodiodes. The respective photocurrents are subtracted, and their difference A/?eo(T)""A/?//(T)-A7?i(T) as the coherent response is recorded as a fianction of the pump-probe delay r. The transient reflectivity was measured both with the [110] crystalline axis of Si oriented parallel and at 45° to the pump polarization to explore the r25' and ri2 symmetry response, respectively

3.

Results and Discussions

The ARQO(T)/RO measurement for the r25' geometry in Fig.la is composed of an aperiodic component near zero delay followed by a coherent LO phonon oscillation. The latter component can be fit by AQ exp(-T/1.30) cos[27r(15.24T+0.016x^+0.064)], where the parameters in the order of appearance are the dephasing time in picoseconds, the oscillation frequency in THz, chirp, and phase shift. The departure fi-om the LO phonon dephasing time of 3.5 ps and fi*equency of 15.60 THz of intrinsic Si reflects the coherent phonon self-energy, which depended on the excitation density. By contrast, the Tu response consisted of only a half cycle transient reminiscence of the aperiodic component in the r25' response [4]. The 3.05 eV light is near resonant with the EQ' and the more intense Ei transitions at 3.320 eV and 3.396 eV, which correspond to the excitation e-h pairs at the r point and that for a range of momenta along the A line, respectively[5]. The predominant excitation of electrons to 4 out-of 8 equivalent L-valleys and holes along A lines creates a reciprocal space anisotropy. Moreover, for the alignment of the pump polarization in the r25' geometry (Fig. lb insert) only two out-of-four tetrahedral bonds of each Si atom are excited creating a real space anisotropy. The anisotropic carrier distribution generated in this manner exert a force on the lattice through an imbalance of deformation and of chemical potentials between different nonthermal carrier populations. These two carrierlattice interactions exert a step function electrostrictive force on the lattice driving the coherent LO phonon oscillation polarized in the [001] direction[6]. By contrast, equivalent excitation of each bond in the F ^ geometry (not shown) balances the forces to zero precluding the LO phonon excitation. Figure lb shows the ARQO(T)/RO signals converted into time-dependent spectral amplitudes by continuous wavelet transform (CWT). The CWT decomposes the signal into an ultra-broadband (>70 THz) response straddling zero delay in both geometries, and the LO phonon response at 15.3THz present only in the r25' geometry. The fast response is the coherent electronic coupling of the pump and probe fields via the nonlinear susceptibility of the sample. The real and virtual processes associated with near-resonant photoexcitation across the direct band gap of Si, both modulate the anisotropic reflectivity and generate the driving force for the coherent phonon excitation. The data in Fig. 1 resolve the buildup of the electronic force and the ensuing response of the lattice that constitute the electronphonon interaction in solid.

243

0.2

0.4 0.6 0.8 Time delay (ps)

Fig. 1 Transient electro-optic reflectivity signal for Si(OOl) in r25' geometry (a) and its continuous wavelet transform(CWT) (b). The electronic response is broad in frequency straddling zero delay, but decays rapidly, while the phonons persist over the duration of the experiment at 15.3 THz. Insert in (b) defines the polarization of the laser beams relative to the crystalline axes in the r25' geometry. The signal in a) is derived from the difference in Ai?// and AR ^, the orthogonal polarization components of the probe beam.

4.

Conclusions

The most intriguing aspect of the coherent response of Si is the anti-resonance (dip) at -20 fs and 15.3 THz. We attribute this dip to the coupling of the coherent LO phonon and electron-hole pair continuum via impulsive Raman scattering process, as well as by direct electron-phonon interaction, leading to the Fano interference [7,8]. Thus, transforming the transient reflectivity measurement into fi*equency-time space reveals coherent phonon-electron-hole-pair continuum correlation effects involved in the coherent phonon generation and subsequent dressing by electron-hole pair excitations. Furthermore, we demonstrate that EO sampling measurements with 10 fs pulse excitation are a powerful way to probe electron-phonon interaction in solid-state materials.

References 1 2 3 4 5 6 7 8

244

D. Pines, and P. Nozieres, in Theory of Quantum Liquids'', 1966. T. Sjodin, H. Petek, & H. -L. Dai, Phys. Rev. Lett. 81, 5664, 1998. J. Sabbah, and D. Riffe, Phys. Rev. B 66, 165217, 2002. M. Hase et al., 51, Nature 426, 51 , 2003. P. Lautenschlager et al., Phys. Rev. B 36, 4821, 1987.. T.E. Stevens, J. Kuhl,and R. Merlin, Phys. Rev. B 65, 144304, 2002. F. Cerdeira, T. A. Fjeldly,and M. Cardona, Phys. Rev. B 8, 4734, 1973. U. Fano, Phys. Rev. 124, 1866,1961.

Generation of Coherent Zone Boundary Phonons by Impulsive Excitation of Molecules M. Giihr and N. Schwentner Institut fiiir Experimentalphysik, Freie Universitat Berlin, Amimallee 14, 14195 Berlin, Gennany E-mail: Markus. Guehr@physik. fu-berlin. Abstract. Coherent zone boundary phonons in rare gas solids are observed in ultrafast pump-probe spectroscopy of molecular dopants. The excitation is impulsive during the pump pulse and phonons are decoupled from the molecular vibrations.

1.

Introduction

Coherent phonons are an active research field in X-ray [1] and optical spectroscopy [2]. We investigate the generation and interrogation of coherent Zone Boundary Phonons (ZBP) in rare gas crystals doped with halogen molecules. These guest molecules occupy a double substitutional site in the rare gas cage and show no rotation. The electronic excitation of a guest molecule embedded in a crystal will enforce an expansion of the moecular electronic wave function and therefore change the equilibrium position of host atoms in the vicinity of the excited guest. After excitation, the host atoms can be expected to oscillate around their new equilibrium position. This mechanism is known as Displacive Excitation of Coherent Phonons (DECP) [3]. We observe coherent phonons in ultrafast pump-probe spectra on the system l2:Kr [4] and in very new spectra of Br2:Ar.

2.

Experimental Methods

Our crack free samples of IiiKr and Br2:Ar are grown and investigated under UHV conditions at 20 K. The concentrations are 1:1000 for h'.Ki and 1:500 for Br2:Ar respectively. We use two NOP As (NoncoUinear Optical Parametric Amplifiers) to generate light pulses with duration below^ 40 fs, which are tuneable between 470 and 750 nm. In addition we utilize the doubled fundamental of our pump system (Clark-MXR CPA 2001) at 387.5 nm with duration of 120 fs. A first ultrashort laser pulse (pump pulse) creates a molecular \ibrational wave packet on the excited covalent states (B or A) of the halogen molecule (see fig. la). A second ultrashort pulse (probe pulse) interrogates the wave packet motion on the covalent state via a transition to a charge transfer state (E or p infig.la). We record the laser-induced fluorescence (LIF) of the charge transfer states versus the delay between pump and probe pulse. The charge transfer states are redshifted a few^ thousand wavenmnbers due to the solvation energy they

245

experience in a polarizable rare gas matrix. The intemuclear position of the probe process is given by the energy difference between the covalent state and the appropriate charge transfer state, which has to match the photon energy.

3.

Results and Discussion

Some resuhs are shown in fig. Ib-d. Since the molecules suffer collisions with the surrounding rare gas atoms, the molecular wave packet relaxes below the probe Franck-Condon range. The original spectra therefore show a general decay on a timescale of a few ps, which was eliminated in fig. Ib-d by normalization on the mean value of one period.

0,3

04 R.Jnm]

0.5

3

4

5

Fig. 1. a: Potential energy scheme of tiKr. b-d: pump-probe spectra with phonon dynamics.

The pump-probe spectra show the dynamics of intramolecular vibrational wave packets in the first 2 ps (Br2:Ar in fig. lb) to 4 ps (l2:Kr in fig. Ic and d). The initial molecular oscillation period is 310 fs for Br2:Ar in fig. lb and 420 fs for l2:Kr (fig. Ic). After the decay of the intramolecular wave packet oscillations due to relaxation, dispersion and dephasing, a new dynamics with a 500 fs (Br2:Ar) or 650 fs (l2:Kr) period appears. This modulation happens to be phase stable with respect to the pump pulse upon excitation of different electronic states of the molecule (compare fig. Ic and d by dashed line) and different excitation energy in the anharmonic well of the electronic state (not shown here). Intramolecular dynamics in another state is excluded by different pump-probe conditions and fluorescence selection, therefore we attribute this feature to coherent host

246

dynamics. It shows up independent of the molecular oscillation period and the phase stability indicates that it cannot be a host mode driven by the molecular vibration. The excitation of the host dynamics is impulsive during the pump process at t = 0 fs. The pump pulse transfers the molecule from its electronic ground state X (^E) to a covalent state like B CUQ) or A (^no. Thereby the electronic cloud blows up resulting in a new equilibrium position of the neighbouring host atoms. The host atoms start to oscillate around their new equilibrium and phonons are excited. By comparison of the Fourier spectra of fig. lb and c with the phonon dispersion relations of Ar and Kr one realizes, that the host induced dynamics with narrow resonances at 1.5 THz for l2:Kr and 2 THz for Br2:Ar exactly matches the frequency of the ZBP in Ar and Kr. The linewidth of the phonon in the Fourier spectra is given by the observation time window in the pump-probe measurement. The detection mechanism is based on a combination of phonon and intramolecular dynamics: The pump pulse impulsively creates phonons and an intramolecular wave packet. The wave packet relaxes below the probe window after a few ps and spreads over a large intemuclear distance because of dispersion and dephasing. The ZBP stay in the vicinity of the molecule, because of their vanishing group velocity. The ZBP modulate the host density in the vicinity of the chromophore. Due to this periodic density modulation, the solvation energy of the molecular charge transfer states oscillates up and down with the phonon frequency. If the phonon reduces the density, the charge transfer states are shifted upwards. The probe window (probe in fig. la) follows the horizontal arrow in fig. 1 to the right in keeping the resonance condition. Therefore it shifts away from the relaxed intramolecular wave packet resulting in a weaker probe signal. If on the other side, the phonon increases the local density, the charge transfer states shift downwards, and the probe window moves towards the center of the molecular wave packet.

4.

Conclusions

The observed matrix induced dynamics can be consistently explained in an electronic excitation and solvation shift probe scheme. Acknowledgements. We gratefully acknowledge support by the Deutsche Forschungsgeitieinschaft (DFG) and the very stimulating discussions with Prof V. A. Apkarian, Dr. M. Bargheer, Prof R.B. Gerber and Prof J. Manz.

References 1 2 3 4 5 6

K. Sokolowski-Tinten et al. Nature 422, 287, 2003. O.V. Misochko, M. Hase, K. Ishioka, M. Kitajima, Phys. Lett. A 321, 381, 2004 HJ. Zeiger et al., Phys. Rev. B 45, 768,1992. M. Giihr, M. Bargheer, N. Schwentner, Phys. Rev. Lett. 91,085504, 2003 V.S. Batista and D.F. Coker (private communication) M.L. Klein and J.A. Venables, Rare Gas Solids, Academic Press, New York, 1977

247

Amplitude Collapse -Revival of Chirped Coherent Phonons under High-density Optical Excitation Kunie Ishioka^ Oleg V. Misochko^, Rong Lu\ Muneaki Hase\ and Kitajima^

Masahiro

^ National Institute for Materials Science, Tsukuba, 305-0047 Japan E-mail: ishioka.kunie @nims .go .jp ^ Institute of Solid State Physics, Russian Academy of Sciences, 142432 Chernogolovka, Russia E-mail: [email protected] Abstract. Collapse and revival on picosecond timescale is observed for the chirped phonons of bismuth and graphite under intense photo-excitation at lOK. This phenomenon has no classical analogue and is explained by the quantum interference of wave packets in anharmonic potential.

1.

Introduction

Coherent excitation of optical phonons occurs whenever ultrashort laser pulses interact with a crystalline solid with Raman-active phonons with frequencies smaller than the inverse of the laser pulse duration. Under high-density optical excitation, the coherent phonons show different characteristics from those under low-density excitation. Above a well-defined threshold, the amplitude of the coherent oscillations is no longer linear to the excitation density, and the frequency becomes a function of optical density. Although classical description is adequate in most situations involving coherent phonons, quantum fluctuations become dominant at very low temperatures or at short enough timescale. This quantum behavior has been successfiilly demonstrated by the observation of squeezed coherent phonons. In the present paper we report that the coherent phonons in bismuth and graphite are chirped and, above a critical optical density, they exhibit the collapse and revival phenomenon, testifying to the non-classical dynamics.

2. Experimental Methods Samples were a single crystal of Bi and highly oriented pyrolytic graphite. Pumpprobe reflectivity measurements were performed using an 800 nm output of a 100 kHz regenerative amplifier of a Tiisapphire oscillator with 130fs pulse duration. Isotropic and anisotropic reflectivity was measured to detect the Aig phonon of Bi and E2gi phonon of graphite, respectively. All experiments were performed at optical densities lower than the threshold for permanent damage.

248

3. Results and Discussion The transient reflectivity change of Bi consists of non-oscillatory and oscillatory components, corresponding to the photo-excited carriers and coherent Ajg phonon, respectively. As the optical density increases, the oscillation deviates from a single damped-harmonic, and it requires to introduce the chirp a of the coherent phonon to reproduce the oscillatory part for all the time delays [1]. f(t) =A exp i-t/r) sin {{2jiv-^at)t+8),

(1)

Further increasing the laser leads the oscillation once dies out at x^ but revives some time later at x^, instead of decaying monotonically, as shown in Fig. 1(a). We can gain a deeper insight into the physics of collapse and revival when we Fourier-transform the oscillation for different time spans, as shown in Fig. 1(b). For short time delays (KT^), the phonon lineshape is broad, red-shifted and strongly asymmetric, while it is extremely narrow and has a slightly blue-shifted frequency for long time delays (t>rr). The lifetime of the phonon after the revival is much longer than that before the collapse and that of thermal phonons observed by spontaneous Raman scattering [2]. These observation indicates that we are not looking at the coupling between two classical oscillators, but quantum interference among different number states in anharmonic vibrational potential [3,4]. Similar revival was observed also for the coherent Ej^i phonon of graphite under intense optical excitation, as shown in Fig. 2(a). The result suggests that the quantum interference among vibrational states is universal in solids under highdensity optical excitation. There are, however, several differences between the experimental observations for graphite and Bi. Firstly, the chirp a for the Ejgi mode of graphite is one order of magnitude smaller than Ajg of Bi. Secondly, r^ and Tr were almost independent of the optical density for graphite, while they were decreased monotonically with optical density for Bi. Thirdly, the beating of OxlO"-M

1. til t

8-J

14.3mJ/cm^

t

t

t

t 4-

9.1mJ/cm^

IIPI" 2-

2.5 3.0 3.5 Frequency (THz)

(a) 1

1

4

6

1

i

1

10 12 8 Delay time (ps)

1

1

14

1(

Fig. 1. (a) The reflectivity change of Bi under different optical densities at lOK. Black and grey arrows indicate collapse (xj and revival (x^) times for the phonon amplitude, (b) Fourier transform power spectra for temporally sliced oscillations under excitation of 14.3 mJ/cm^ at 10 K.

249

0x10

mUlHllllAAA

46.0mJ/cm ^

IftllHHllMAAAAA

39^J/cm^

50-

40ninyilllftAAAAAA A

31.8mJ/cm"'

^30-

lIlllllllllAAhAAAAAAA 24.8mjW 20-

10-

0-

III 11...

17.7mJ/cm

n llljllij|ll»»»««

10.6mJ/cm

|nliili

0 ~'|""i i" I I I I I I I I I [ I I I I I I I I I I I i'

0.8

(3)

' ' ' ' 1 ' ' ' ' 1 ' ' '' 1 ' ' ' • 1 ' • ' ' f

5

10 15 Time delay(ps)

20

1.0

1.2

1.4 1.6 THz

1.8 2.0

25

Fig. 2. (a) Oscillatory part of the anisotropic reflectivity change, and (b) its Fourier transforms, of graphite under different optical densities at lOK. graphite is reproduced by two damped-harmonic oscillators with comparable frequencies, as shown in the Fourier spectra in Fig. 2(b). The differences are attributed to the different magnitudes of vibrational anharmonicity and of electronic softening under photo-excitation between two phonon modes in question. Comparison with molecular systems [4] suggests that the collapse-reviv^ of the graphite interlayer mode is dominated by wave packet motion in a moderately anharmonic vibrational pontential [5]. The strong chirp and the power dependence of Tc and Tf of Bi Ajg phonon, on the other hand, suggest large anharmonicity and significant softening of the vibrational potential during photo-excitation [2].

4.

Conclusions

The collapse and revival of a collective lattice mode provides convincing evidence for non-classical state of crystal lattice created by the ultrafast laser pulses. We believe that our work will trigger further investigations of coherent lattice dynamics resulting in the creation of new non-classical lattice states.

References 1 2 3 4

Hase, Kitajima, Nakashima, Mizoguchi, Phys. Rev, Lett.SH, 067401 (2002). Hase, Mizoguchi, Harima, and Nakashima, Phys. Rev. B 3 8, 5448 (1998). Misochko, Hase, Ishioka, and Kitajima, Phys. Rev. Lett. 9 2, 197401 (2004). Gruebele, Roberts, Dantus, Bowman, and Zewail, Chem. Phys. Lett. 166, 459 (1990). 5 C. S. GCousins andM.L Heggie, Phys. Rev. B 6 5, 027109 (2003).

250

Intense coherent optical phonons driven by impulsive excitonic interference under electric fields Osamu Kojima, Kohji Mizoguchi, and Masaaki Nakayama Department of Applied Physics, Graduate School of Engineering, Osaka City University, 3-3-138 Sugimoto, Sumiyoshi-ku, Osaka, 558-8585, Japan E-mail: [email protected]

Abstract. We report on the generation of the intense coherent longitudinal optical phonons driven by the impulsive interference between the excitons with higher subbands in a GaAs/AlAs multiple quantum well under applied electric fields. Recently, we reported the enhancement of the coherent longitudinal optical (LO) phonons in multiple quantum wells (MQW's) due to the coupling with quantum beats of heavy-hole (HH) and light-hole (LH) [1,2]. In the case that the energy of the excitonic quantum beat is almost equal to the LO phonon energy (^Lo)? the excitonic quantum beat acts as a driving force for the coherent LO phonon. It is well known that the subband energies in a MQW can be controlled by an applied electric field, which is called a quantum confined Stark effect (QCSE) [3], as well as by a quantum size effect. Hence, it is possible to tune the intersubband energies to £LO- In the present work, we show the generation of the intense coherent GaAs-like LO phonon in a GaAs/AlAs MQW by tuning the intersubband energy to £LO with use of the QCSE. We used a GaAs/AlAs MQW embedded in a p-i-n structure on a (001) «^-GaAs substrate grown by molecular-beam epitaxy, where the intrinsic layer corresponds to the MQW. The MQW consists of 20 periods of GaAs and AlAs layers whose thicknesses are 15.3 and 4.5 nm, respectively. The coherent LO phonon was measured by a reflection-type pump-probe technique at 10 K. The laser source was a mode-locked Ti:sapphire pulse laser delivering 100-fs pulse with repetition of 82 MHz. The pump and probe beams were orthogonally polarized to each other in order to eliminate the pump-beam contribution to the probe beam. The pump density was kept at about 35 nJ/cm^. Assuming that the excitonic absorption coefficient of the fundamental transition energy under the no electric field is about Ixio"^ cm"^ [4], the photoexcited carrier density is estimated to be about 1x10^^ cm"^ We also performed photocurrent (PC) measurements at 10 K in order to evaluate the exciton energies using a 32-cm single monochromator with a resolution of 0.3 nm. The electric field strength F is estimated from F'={Vf-V)IL, where Vb is built-in voltage of the p-^ junction, V is applied bias, and L is the total length of the MQW. Figure 1 shows the PC spectra at various electric fields in the GaAs(15.3 nm)/AlAs(4.5 nm) MQW. Many excitonic absorption peaks are observed. The origins of these PC peaks are assigned by comparison with the transition energies calculated by transfer matrix method [5]. The energy shifts of the PC peaks are due

251

to the QCSE. Since the energy difference between the E2HH1 and E2HH2 excitons at 130 kV/cm is equal to £LO of GaAs as shown by the broken lines, we tuned the pump energy to 1.571 eV that is around the center energy between the E2HH1 and E2HH2 excitons. -T" 777 "X" GaAs(15.3 nm)/AlAs (4.5 nm) MQW

10 K

E1HH2 E2HH2 E2HH1| E2LH2 ElLHl ElHHl

1.50

1.55

1.60

1.65

Photon Energy (eV)

Fig. 1. PC spectra of GaAs(15.3 nm)/ AlAs(4.5 nm) MQW at 10 K at various electric field strengths. The broken lines indicate the E2HH1 and E2HH2 exciton energies at 130 kV/cm. The dotted curves are guides for the eyes to see the change of the peak energies of E2HH1 and E2HH2 excitons at different electric fields. Figure 2(a) shows the time-resolved reflectivity-changes at various electric fields. The time-domain signals consist of the large reflectivity change around 0 ps and a long-lived oscillation. The initial part of the signal arises from a change of the carrier density. The oscillatory structures corresponding to the quantum beat between the E2HH1 and E2HH2 excitons are not observed in the time-domain signal. This disappearance is attributed to ultrashort dephasing of the excitons at the higher subbands. The oscillatory structures with the period of 113 fs in each signal lasts over 4.0 ps and that is assigned to the coherent GaAs-like LO phonon. The amplitude of the coherent LO phonon changes with the electric field. We note that the amplitude is remarkably enhanced around 155 kV/cm. In order to estimate the amplitude of the coherent LO phonon, we analyzed the time-domain signals in the time range of 1.0 - 4.0 ps by fitting with a damped harmonic oscillation. In Fig. 2(b), the amplitude of the coherent LO phonon and the energy difference between the E2HH1 and E2HH2 excitons (AEHHI-HH2) estimated from the PC spectra are plotted as a function of electric field, where the closed and open circles indicate A£HHI-HH2 and the amplitude of the coherent LO phonon, respectively. The amplitude of the coherent LO phonon at 155 kV/cm is about 20 times larger than the lowest amplitude. Since the laser pulse has the spectral width broader than AEHHI-HH2, the E2HH1 and E2HH2 excitons are simultaneously generated and the impulsive longitudinal polarization will be generated through the interference between the E2HH1 and E2HH2 excitons. Since A£HHI-HH2 around 155 kV/cm is almost equal to £LO of GaAs, the longitudinal polarization induced by the interference between these excitons resonantly interacts with the LO phonon. Therefore, this enhancement arises from the coupling between the coherent LO

252

phonon and the impulsive interference between the E2HH1 and E2HH2 excitons via the longitudinal polarization. Although the electric field for the resonance between ^LO and the HH2-HH1 intersubband energy is estimated to be 130 kV/cm from PC, the electric field seems to be slightly shifted by Coulomb screening due to photoexcited carriers in the pump-probe measurement. 11111111111111111111111

(a)

lOK Pump Energy = 1.571 eVi xlO 170kV/cml

vVV^AA/S^«^^^/vvvvVVVV^AAAAAlVwy

xlO 165kV/cm"1 \/W\AAAAAWVWW\/VWVWW>| 155kV/cm xlO 145kV/cm^ /\/WV\AAAAA/WlWwVWWWVd x30

0.0

lOOkV/cmj

1.0 2.0 3.0 Time Delay (ps)

4.0

40

80 120 160 Electric Field (kV/cm)

Fig. 2 (a) Time-resolved reflectivity-changes at various electric fields, (b) Amplitude of the coherent LO phonon and AEHHI.P^HZ? which are indicated by open and closed circles respectively, as a function of electric field. The broken line indicates £LO of GaAs. In summary, we have investigated the coherent LO phonon in a GaAs/AlAs MQW by the reflection-type pump-probe technique at various electric field strengths. We have found that the coherent LO phonon is markedly enhanced under the condition that the intersubband energy is tuned to ^LG by the applied electric field. This fact indicates that the impulsive excitonic interference can drive the coherent LO phonon by tuning the intersubband energy to J^LOAcknowledgements. This work was partially supported by a Grant-in-Aid for the Scientific Research (No. 15340102) from the Ministry of Education, Culture, Sports, Science, and Technology of Japan.

References 1 O. Kojima, K. Mizoguchi, and M. Nakayama, Phys. Rev. B 68, 155325, 2003. 2 O. Kojima, K. Mizoguchi, and M. Nakayama, J. Lumin. 108, 195, 2004. 3 D. A. B. Miller, D. S. Chemla, T. C. Damen, A. C. Gossard, W. Wiegmann, T. H. Wood, and C. A. Burrus, Phys. Rev. B 32, 1043, 1985. 4 G. D. Sanders and K. K. Bajaj, Phys. Rev. B 35, 2308, 1987. 5 I. Tanaka, M. Nakayama, H. Nishimura, K. Kawashima, K. Fujiwara, Phys. Rev. B 46,7656, 1992.

253

Phonon-polariton Based THz Spectroscopy Benjamin J. Paxton, Masashi Yamaguchi, and Keith A. Nelson Massachusetts Institute of Technology, Cambridge MA 02139, USA E-mail: [email protected] Abstract. We demonstrate a THz spectrometer based on grating interferometric measurement of phonon-polariton propagation before and after interaction with a sample. The temperature dependent dielectric response of the relaxor ferroelectric KTN was measured.

1.

Introduction

The development of pulsed THz radiation sources and time-domain detection methods has permitted spectroscopic study and THz imaging of a variety of important systems [1,2]. The use of phonon-polaritons generated in an ionic crystal through impulsive stimulated Raman scattering as a source for THz spectroscopy suggests itself because of the versatility of polariton waveform shaping and detection. [3] In particular, generation of single-cycle, broadband polariton wavepackets, of tunable, narrowband waves, and of arbitrarily shaped waveforms may be achieved in the same crystal (typically LiNbOs or LiTaOs), and polariton detection through measurement of induced birefringence, second harmonic generation, spatiotemporal images, or other observables is generally straightforward. [4] Here we present a novel spectroscopic implementation in which the electro-optic refractive index modulation induced by the polaritons is detected through the use of a compact, grating-based interferometer whose two arms pass on either side of a sample that is sandwiched between two LiTaOs crystals, permitting measurement of the phase and amplitude of the polariton electric field before and after interaction with the sample. LiTaOa LiTaOs

Probe 2 Probe 1 Pump Fig. 1. Schematic illustration polariton-based THz spectroscopy cell. The LiTa03 crystals are cut to compensate for the forward polariton wavevector component.

254

Measurement of the transmitted and reflected polaritons allows determination of the complex dielectric constant. The forward wavevector component of the polaritons is compensated for by cutting the ionic crystal edges such that polariton transmission and reflection occur at approximately normal incidence. The polariton-based interferometric system has been applied toward measurement of the complex dielectric response of the relaxor ferroelectric KTa]. ^NKOs x=0.18 (KTN) from 300 to 5 K. In this system, the KTaOs host crystal is paraelectric and cubic, with Ta"^^ ions at the unit cell center on average, while the Nb^^ impurity ions move off-center, creating a strong local polarization that draws Ta ions in neighboring unit cells off-center in the same direction. The randomly oriented polar "nanoregions" thus formed grow in correlation length as the temperature is reduced, and eventually they begin to interact. At this point the polar nanoregions may become aligned to form a ferroelectric phase with longrange dipole ordering. Alternatively, if T is too low to permit facile reorientation, the system may become frozen into a dipole glass state. In either case, marked polarization fluctuations occur over a wide range of time scales and with a strong T-dependence. Polarization dynamics in the THz range have not been well characterized [5] and are poorly understood.

2.

Experimental Methods

The present experiments were conducted using single-cycle, broadband (roughly 0.05 to 0.5 THz) polariton wavepackets that were generated by femtosecond pulses cylindrically focused to a "line" inside a LiTaOs crystal. The output of an amplified Ti: Sapphire laser system (Coherent Reg A 9000) was used to excite the phonon-polaritons. The crystals used were typically LiTaOs, LiNbOs, or MgO doped LiNbOs, x-cut and approximately 2 mm thick in the direction of excitation pulse propagation. The excitation spot size was chosen to maximize the polariton frequency content while still ensuring that the depth of focus exceeds the crystal thickness. A narrowband scheme of crossed excitation pulses can generate far higher polariton frequencies. The interferometer uses transmissive gratings (i.e. binary phase mask patterns) and common lenses for generation, imaging, and recombination of the dual probe arms. The use of all common path optics provides excellent phase stability without the need for active feedback loops. This interferometer design also allows for facile changing of the separation between the probe arms by changing the phase mask patterns. The distance between the probe beams and the excitation beam was measured by imaging onto a CCD.

3.

Results and Discussion

Figure 2 shows the measured refractive index of KTN as a function of temperature, determined from both transmission and reflection experiments. The weak LiTaOs temperature dependence was taken into account in the data analysis.

255

Reflection spectrosocpy •{ Transmisison spectroscopy

70 60

Ic 50 0 40

Iso 1 20

^•^-^^

10 0

50

100 150 200 250 300 Temperature [K]

Fig. 2. The temperature dependent refractive index of KTN.

The results show a peak in the refractive index at low temperature in accord with the ferroelectric phase transition temperature rC=35K. At low temperatures, reflection of the polariton is high due to the large index mismatch between KTN and LiTaOs-

4.

Conclusions

We have demonstrated a compact THz spectrometer based on grating interferometer detection of phonon polaritons and used for preliminary measurements of the complex T-dependent dielectric response of the relaxor ferroelectric KTN. Using polaritons for a THz sources offers several practical advantages, including avoidance of far-IR detectors, optics, and free-space propagation, control over the THz bandwidth and waveform, compact cell assembly, and detection of both transmitted and reflected THz fields. Acknowledgements. Thanks to Prof. Jean Toulouse for KTN samples and valuable discussions. Funding NSF CHE-0212375.

References 1 M.C. Beard, G.M. Turner, and CA. Schmuttenmaer, 'Terahertz Spectroscopy," J. Phys. Chem. B. 106, 7146-7159 (2002). 2 B. Ferguson, X.C Zhang, "Materials for terahertz science and technology" Nature Matt. 1, 26-33 (2002). 3 T. F. Crimmins, N. S. Stoyanov, and K. A. Nelson, "Heterodyned impulsive stimulated Raman scattering of phonon-polaritons in LiTaOs and LiNbOs," J. Chem. Phys. 117, 2882 (2002). 4 T. Feurer, J.C. Vaughan, and K.A. Nelson, "Spatiotemporal coherent control of lattice vibrational waves," Science 299, 374-377 (2003). 5 G.A. Samara, "The relaxational properties of compositionally disordered ABO3 perovskites" J. Phys. Cond. Matt. 15, R367-R411 (2003). 6 J. Toulouse and R. Pattnaik, "Collective Behaviors in the Disordered Ferroelectrics KLT and KTN" J. Korean Phys. Soc. 32, S942-S946 (1998).

256

Excitonic quantum beats dressed with coherent phonons K. Mizoguchi^ T. Furuichi', O. Kojima^ M. Nakayama', K. Akahane^, N. Yamamoto^, andN. Ohtani^ Department of Applied Physics, Osaka City University, 3-3-138 Sugimoto, Sumiyoshi-ku, Osaka 558-8585, JAPAN E-mail: [email protected] 2

Communications Research Laboratory, 4-2-1 Nukui-Kitamachi, Koganei-shi, Tokyo 184-8795, JAPAN Abstract. We report on excitonic quantum beats dressed with coherent phonons in GaAs/AlAs multiple quantum wells. We discuss the dispersion relation of the dressed quantum beat vs. the splitting energy of the heavy-hole and light-hole excitons. The stucty on the dynamical properties of the coupling of coherent phonons and photo-generated carriers is one of interesting subjects in ultrafast phenomena [1, 2] For the typical behavior of the coupled mode it is well known that its frequency markedly deviates from each frequency of individual modes when two oscillation modes are resonant in frequency, so called an anticrossing behavior. Recently, it is reported that in GaAs/AlGaAs superlattices the coupling of the longitudinal optical (LO) phonon and Bloch oscillation observed only in the time domain resonantly enhances the LO phonon. [2] In the coupling of the Bloch oscillation and coherent LO phonon, however, the dispersion relation does not indicate an anticrossing behavior. One question may arise as whether other coherent oscillations of photo-generated carriers peculiar to the time domain, e.g., quantum beats (QB's) of excitons, [3] make a coupled mode with the LO phonon or not. In the present work, we have focused on the coupling of excitonic QB and the coherent LO phonon in GaAs/AlAs multiple quantum wells (MQW's) with different splitting energies of the heavy-hole (HH) and light-hole (LH) excitons. The samples used are (GaAsy(AlAs),„ MQW's with 50 periods (m=l8, 19, 20, 30, and 35) on a (001) GaAs substrate grown by molecular beam epitaxy, where the subscript m denotes the number of the monolayers in the constituent layer. We will call these samples \m, m) MQW', hereafter. The HH and LH exciton energies in the GaAs/AlAs MQW's were determined by using photoluminescence excitation spectroscopy. The HH-LH splitting energy, A^HH-LH^ of each sample is shown in Fig. 1(a). Femtosecond time-domain measurements were carried out at 10 K with a reflective electro-optic sampling technique. The energy of the laser pulse was tuned to the central position between the HH and LH exciton energies of each MQW. Figure 1(a) shows the time-resolved oscillatory traces in the GaAs/AlAs MQW's observed at 10 K, where each signal is normalized by the first peak of the oscillation. In the time range before '-1 ps, the strong oscillations with the dephasing time shorter than 0.5 ps are observed Figure 1(b) shows the Fourier transform (FT) spectra of the time-domain signals in the MQW's. As AEHHLH is

257

increased, the peak frequency is increased In the (30, 30) and (35, 35) MQW's, the peak frequencies are in agreement with the frequencies of the QB's estimated from A^HH-LH that are marked with the arrows in Fig. 1(b). On the other hand, when A^HH-LH becomes close to the LO phonon energy (EI^Q=36.S meV), the peak frequency deviates from the frequency of the QB. We note that the FT spectra in the (18, 18), (19, 19) and (20, 20) MQW's have the doublet peak profile, where the major and side peaks are marked with asterisks and daggers, respectively. In Fig. 1(c), the peak frequencies observed in FT spectra are plotted as a function of AEHH-LH- The broken lines indicate the frequencies of the GaAs-like LO phonon, /LO, and the excitonic QB, /QB, where the value of/QB is estimated from A£^HH-LHWhen A£^HH-LH ^ ^LO' an anticrossing behavior is clearly observed This anticrossing behavior suggests the existence of the coupled mode of the excitonic QB and GaAs-like LO phonon. The coupling leads to the enhancement of the long-lived coherent LO phonon.[4] Although there is no applied electric field, a surface electric field, which usually occurs in a semiconductor, is estimated to be about 10 kV/cm from Franz-Kelcfysh oscillations of the band edge of the GaAs substrate observed in photoreflectance signals. Under the presence of the surface electric field along the growth direction of the MQW, the symmetry of the electron, HH and LH envelope functions is broken, and the energy levels of electrons, HH and LH are slightly shifted, which is well known as the quantum-confined Stark effect.[5] Owing to the symmetry breaking of the electron, HH, and LH envelope functions, the excitonic QB leads to the oscillation of the longitudinal polarization along the growth direction of the

0.0

0.5

1.0

1.5

2.0

Time Delay (ps)

10

2.5

Frequency (THz)

20

30

40

50

Splitting Energy (meV)

F i g . 1. (a) Transient reflectivity changes in the GaAs/AlAs MQW's observed at 10 K. A£HH-LH indicates the splitting energy of the HH and LH excitons. (b) Fourier transformed spectra of the time-domain signals. The dashed line represents the frequency of the GaAs-like LO phonon. The arrow indicates the frequency of the quantum beat estimated from the HH-LH splitting energy in each MQW. (c) Peak frequencies observed in Fig. 1(b) plotted as a function of the HH-LH splitting energy. The solid curves indicate the calculated dispersion relation of the excitonic QB dressed with the LO phonon.

258

MQW.[3] Since the GaAs-like LO phonon in the MQW has also the longitudinal polarization, the excitonic QB and the GaAs-like LO phonon will couple with each other through the longitudinal polarization under the condition of A£HH-LH ~ Ei^Q. When the longitudinal polarizations due to the excitonic QB and the GaAslike LO phonon linearly interact with each other, the characteristic frequencies of the coupled oscillation, / , in the classical dynamics are simply represented by

f =l\fQB^ fLO±^l{f^^^^^ 2

(1)

where C is the coupling constant used as a fitting parameter. The solid curves in Fig. 1(c) indicate the dispersion relation of the coupled mode fitted by using Eq. (1), where the value of the coupling constant C is 0.14. The fitted dispersion relation is in good agreement with the experimental results. Assuming that the coupled mode generates the polaron which consists of the HH and LH excitons with the LO phonon clouds, we can evaluate the coupling constant C from the Frohlich coupling constants of HH and LH polarons. Gerlach and Luckzak calculated the coupling strengths of HH and LH polarons which are 0.177 and 0.075, respectively.[6] The obtained coupling constant C is almost consistent with the average value of the calculated coupling constants of HH and LH polarons. This result demonstrates that the strong oscillation observed in the GaAs/AlAs MQW's under the condition of AE^H-LH ^ ^LO is assigned to the excitonic QB dressed with the coherent GaAs-like LO phonon. In summary, we have investigated the excitonic QB dressed with the coherent GaAs-like LO phonon in the GaAs/AlAs MQW's with the HH-LH splitting energy close to the LO phonon energy. The dependence of the frequency of the excitonic QB dressed with the LO phonon on the HH-LH splitting energy exhibits an anticrossing behavior peculiar to the coupling between elementary excitations. A c k n o w l e d g e m e n t s . This work was partially supported by a Grant-in-Aid for the Scientific Research from the Ministry of Education, Culture, Sports, Science, and Technology of Japan.

References 1 J. Shah, Ultmfast Spectroscopy of Semiconductors and Semiconductor Nanostructures, edited by M. Cardona, (Springer, Berlin, 1996), Chap. 2. 2 T. Dekorsy, A. Bartels, H. Kurz, K. Kohler, R. Hey, and K. Ploog, Phys. Rev. Lett. 8 5, 1080-1083(2000). 3 M. S. C Luo, S. L Chuang, P. C M. Planken, I. Brener, H. G. Roskos, and M. C. Nuss, lEEEJ. Quantum Electron. 3 0, 1478-1488(1994). 4 O. Kojima, K. Mizoguchi, andM. Nakayama, Phys. Rev. B6 8, 155325 (2003). 5 D. A. B. Miller, D. S. Chemla, S. Schmitt-Rink, Phys. Rev. B 3 3 , 6976-6982 (1986). 6 B. Gerlach, and F. Luczak, Phys. Rev. B5 7, 1814-1819 (1998).

259

Evidence of higher-order nonlinearities in excitonic FWM signals in microscopic theory and experiment L. Wischmeier\ M. Buck^, S. Schumacher^, G. Czycholl^, F. Jahnke^, I. Ruckmann^ and J. Gutowski^ ^ Institut fur Festkorperphysik, Universitat Bremen, P.O. Box 330440, D-28334 Bremen E-mail: [email protected] ^ Institut fur Theoretische Physik, Universitat Bremen, P.O. Box 330440, D-28334 Bremen Abstract. The intensity dependence and the polarization state of the four-wave-mixing signal at the excitonic resonance of a ZnSe single-quantum well is studied and compared with a microscopic model covering coherent higher-order optical nonlinearities. The ultrafast coherent dynamics of excitons and biexcitons in semiconductor nanostructures provides a wide field of applications. Coherent control of biexcitons allows to realize an all-optical quantum gate for quantum computation. The coherent manipulation of optical excitations in semiconductors critically depends on their natural many-particle interactions. The nonlinear optical response is strongly influenced by the Coulomb interaction of carriers which leads to excitonic and biexcitonic resonances as well as to decoherence, dephasing and relaxation of excitations on a very short time scale (= 10'^^ s) [1]. We have studied the polarization state and intensity dependence of the detected four-wave-mixing (FWM) signals in ZnSe quantum wells. The results are analyzed in terms of a microscopic theory of coherent excitonic and biexcitonic nonlinearities. Polarization-dependent effects at higher intensities, which cannot be explained by a phenomenological model based on extended optical Bloch equations (OBE) [2], are successfully explained. A frequency doubled Ti:sapphire laser is used as excitation source for the timeintegrated FWM experiments. The laser emits 110 fs pulses at a repetition rate of 82 MHz. The FWM signal is generated by two pulses which are focused and spatially overlapped onto the sample from directions ki and kz . The delay time t^ei between the incidence of the pulses can be varied. Both pulses are linearly polarized. The angle ^poi between the polarizations of the pulses and the intensity of each pulse can be varied. The polarizations, created in the sample by the kj and k2 pulses interfere generating a polarization grating and resulting in selfdiffraction. One of these diffraction FWM signals is emitted in direction 2^2 - ki and is recorded spectrally resolved by a combination of a spectrometer and a charge-coupled device (CCD) camera. The central wavelength of the pulses is set 5 meV below the biexcitonic resonance to excite exclusively resonantly the heavyhole exciton-biexciton system. The used 10 nm ZnSe/ZnSo.ovSeo.gs single-quantum well was grown by molecular-beam epitaxy on a GaAs substrate. The sample has been removed from the substrate and mounted onto a glass plate to carry out experiments in transmission geometry. The sample is kept at 4 K in a cryostat.

260

The propagating fields of the FWM experiment are described by a solution of Maxwell s equations in terms of the induced coherent quantum-well polarization which is given by the probability amplitude for electron-hole-pair transitions, ••{elK)

(1)

and the dipole coupling. On the level of coherent optical nonlinearities up to third order in the optical field [3, 4], P^^ only couples to the biexcitonic transition amplitude,

<^X..=««'^0-(«>(«>+«^0«C)

(2)

The latter describes correlated four-particle transitions beyond the Hartree-Fock (HF) level. Here ^^and /i^are creation operators with momentum k for electrons and holes, respectively. The coupled equations of motion for P and B describe coherent optical nonlinearities including phase-space filling (due to transient population effects). Coulomb exchange contributions, and biexcitonic effects due to correlated four-particle interactions. For our numerical calculations, P and B are expanded in terms of excitonic eigenfunctions (restricted to Is contributions to reduce the numerical effort). For the description of FWM experiments, a Fourier decomposition for the signal components in various directions {ki, kz, 2ki - kz and 2k2 - kj) is used. While the theory systematically includes all effects up to third order in the optical field E, a restricted class of higher-order effects can be obtained from solving the equations of motion self-consistently up to arbitrary order in E. E^.^ = 12.2pJ

excHon resonance

„

10000 ;. a>^ = 75° ///•y '. , . N .' i: \ ;. /

f B

.1:

1000

y^ >»—'

3

"^iv. -

I

-.'% ' x^.'... '• \ .

N»« _

/

•*/•

1 •

SLx

.

^rr\'

,^

,"'/! :t - . /

2

-" /' \ \

'"" /

y^-

J-.'

^««i 12.2pJ 8.5pJ -••• 4.8pJ 1.2pJ

7

d. -

1 u__— 1 2 -

*

1

X

0

1, .I!3!3k._..-i

1

2

3

Ui [Ps)

Fig. 1. FWM signal vs. delay time for various normalized intensities of the ki pulse obtained from numerical calculations (left). The lowest intensity corresponds to a Rabi energy ofdE/E^ = 2.2 • 10'^ in units of the 3d excitonic Rydberg energy EB- Corresponding experimental results are shown on the right. In Figure 1 the intensity dependence of the FWM signal at the exciton resonance is shown as a function of delay-time for an angle ^poi of 75 ° between the linear polarization orientations of both incoming beams. For this angle, the contribution to the 2k2 - ki excitonic diffraction signal of every included many-particle effect is pronounced in the phenomenological model [2]. The intensity of the kj pulse was varied and the intensity of the pulse in kz direction was kept constant. For low intensities of the kj pulse the observed oscillations of the excitonic FWM signal 261

due to simultaneous excitation of excitonic and biexcitonic states appear only for negative delay times tdei- In this regime results are consistent with the model based on extended OBE [2]. However, at higher intensities these oscillations also emerge for positive delay. While this effect is not included in [2], the above discussed theory attributes these oscillations to higher-order (beyond x^) nonlinearities. Furthermore, both in theory and experiment, oscillations for detection at the biexciton are absent (not shown). Considering only the dominant Is-exciton contributions to correlation effects (to facilitate the calculations) the biexciton binding energy is slightly underestimated. Hence maintaining the observed oscillation period to decay time ratio leads to a slightly different time scale in the calculations. Results for the variation of the angle ^^^z between linear polarization directions of the two pulses between 90° (cross-linear excitation) and 0° (co-linear excitation) at fixed Epuise = 8.5 pJ and equal intensity of both pulses are displayed in Figure 2. Experiment and theory exhibit the same behavior: for negative delay time the exciton-biexciton oscillations disappear when increasing the angle (ppoi from 0° to 90°, while for positive delay time the oscillations emerge with increasing (fipoifordecreasing overall dephasing times. Only slight oscillations are obtained for detection at the biexciton in the co-linear case (not shown). exciton resonance 10000 ' %>s.r^.^^^-5pj ,-**',

I

// v •

1000

** ' ' f

r/. ;• f •** 7\ . ft •' ^\i

\

^ .—i. •

.• "*

\ X \ "\ .-^ - \ "^

^ - " ^ •* '^. \ V^.^'-N \ «-

1

td.! 'del [PS]

'^^ = -...0 — "•••30°:

\

t..

Fig. 2. FWM signal vs. delay time for equal intensities at various angles between the linear polarization vectors of the two pulses. The numerical results are shown on the left for dE/E^ = 1 • 10"^. Corresponding experimental results are shown on the right. The light polarization and intensity dependence of time-integrated FWM signal has been successfully analyzed in terms of a microscopic theory. Higher order nonlinearities lead to exciton-biexciton oscillations in a regime where former models are not applicable.

References 1 D. S. Chemla and J. Shah, in PNAS, Vol.97, 2437, 2000. 2 A. L. Smirl, in Semiconductor quantum optoelectronics. Edited by A. Miller, M Ebrahimzadeh, and D. M. Finlayson, Institute of Physics Publishing, 1999. 3 M. Lindberg, Y. Z. Hu, R. Binder, and S. W. Koch, Phys. Rev. B Vol.50, 18060, 1994. 4 V. M. Axt and A. Stahl, Z Physik B Vol.93, 195, 1994.

262

Ultrafast Anisotropic Processes of Exciton Magnetic Polarons in CdTe/CdMnTe Quantum Wires R. Naganuma\ T. Kita\ S. Nagahara\ O. Wada^ L. MarsaP and H. Mariette^ ^ Department of Electronics and Electrical Engineering, Kobe University, Rokkodai 1-1, Nada, Kobe 657-8501, Japan E-mail: [email protected] ^ Laboratoire de Spectrometrie Physique, Universite J. Fourier, Grenoble I, CNRS (UMR 5588), Boite Postal 87, F-38402 Saint Martin d'Heres Cedex, France Abstract. We have studied dynamics of exciton magnetic polarons in CdTe/CdMnTe quantum wires. Anisotropic ultrafast evolution of the exciton magnetic polaron formation process has been found for magnetic fields parallel and perpendicular to the wire direction.

1.

Introduction

Recently, the possibility to control both the charge and spin of electrons has attracted a great attention for discovery of new functional devices. Diluted magnetic semiconductors (DMSs), especially II-VI[1] and III-V[2] semiconductors, look promising to realize them. Low-dimensional DMS heterostructures are expected to realize further new magneto-optical phenomena by the quantum confined excitonic states. [3] Optical and spin properties of the excitons separated spatially from the magnetic ions are expected to be controlled by controlling the wave function of the exciton. Especially, excitons in lower-dimensional quantum structures can interact sensitively with the DMS surrounding them. In this paper, we have studied magneto-photoluminescence (PL) from (CdTe)o.5(Cdo.75Mno.25Te)o.5 tilted superlattices (TSLs) grown on a vicinal surface of Cdo.74Mgo.26Te (001). We have investigated the anisotropic exchange interactions in magnetic field by observing dynamics of exciton magnetic polarons (EMPs) in the wire.

2.

Sample growth and experiments

CdTe/CdMnTe wire structures have been fabricated by a self-regulated method for the growth of TSLs using molecular beam epitaxy. Cdo.96Zno.o4Te(001) substrate misoriented T toward the [110] direction was used in this study. [4] This substrate has steps aligned along the [1-10] direction. After growing a Cdo.74Mgo.26Te buffer layer, (CdTe)o.5(Cdo.75Mno.25Te)o.5 TSLs were fabricated by repeated deposition of 1/2 ML of CdTe followed by 1/2 ML of Cdo.75Mno.25Te.

263

This cycle was repeated 30 times. As a consequence, square wires with a crosssection of 9.3x9.7 nm^ were formed. This size is comparable with the bulk exciton Bohr radius (-6.5 nm) of CdTe. We have investigated the wire structure by transmission-microscope (TEM). TEM images show an apparent compositional modulation in the (001) plane. PL experiments were carried out under the excitation of the 488 nm line of an Ar-ion laser. Magneto-PL was measured at 1.5K in an optical cryostat with a superconducting split-coil magnet up to 5 T. In time-resolved PL measurements, the second harmonic light pulses (387 nm, 200 fs) of a Ti:sapphire laser excited by an Ar-ion laser were used as the excitation. The excitation density is 0.27 |iJ/cm^.

Results and Discussion Figure 1 shows a polarized PL spectrum of the sample. As shown in Fig. 1, the PL intensity of the CdTe-rich wire shows maxima at polarizations parallel to the wires. This result demonstrates the lateral quantum confinement in the wire, in which the quantization direction is perpendicular to the wire direction. 4.0CdTe-rich wire CdTe-rich wire

X^ 3.0^

§

ENERGY (eV)

1.04

(Cd,Mg)Te barrier

X 10 r

90 180 270 360 POLARIZATION ANGLE (degree)

Fig. 1. PL spectrum and polarization angle dependence of the PL intensity. A remarkable red shift is found at the low temperature less than lOK. The spectral line width shows an increase as well. The observed behavior results in exciton localization through the formation of BMP, in which the p-d and s-d exchange interactions are responsible for the organization of the paramagnetic local spins within the exciton waveftmction. Here, it is noted that the red shift includes potential fluctuations caused by the BMP, alloy effects, and the size fluctuation. The BMP formation depends on the interaction between excitons and localized Mn ions. In other words, the exciton states confined in the wires can be controlled in magnetic field by utilizing the interaction. To investigate the dynamic process, we have performed time-resolved PL measurements. The PL-peak energy shows a remarkable shift with increasing the delay time. This is a direct evidence of exciton localization caused by the BMP formation. The peak shift shows

264

characteristic behavior in magnetic field. Figure 3 summarizes the results. The magnetic field perpendicular to the wire direction increases the localization time, in contrast to a reduction in the parallel magnetic field. Since the EMP formation time is considered to be short enough, the change of the localization time comes from a change of the mean exciton-diffusion length. The observed anisotropy is attributable to spin reorientation of holes driven by Mn ions. 75 ps

OT 5T

-^-^CTo^J^-aaa' X

> o •10 a:

93 f ^

LU

z 100 200 300 400 500 TIME (ps)

-15

100 200 300 400 500 TIME (ps)

Fig. 2. Temporal evolution of EMP in perpendicular and parallel magnetic fields.

4.

Conclusions

EMP formation in CdTe/CdMnTe quantum wires has been investigated. Since the wire size is close to bulk exciton Bohr radius of CdTe, the exciton states confined in the wires can be controlled in magnetic field by utilizing the exchange interactions. We have found that the EMP formation shows an anisotropic character in the dynamic exciton localization process.

References 1 J. K. Furdyna, J Appl. Phys. 64, R29, 1988. 2 H. Ohno, Science 281, 951, 1998. 3 Y. Oka, J. Shen, K. Takabayashi, N. Takahashi, H. Mitsu, I. Souma, and R. Pittine, J. Lumin. 83/84, 83, 1999. 4 S. Nagahara, T. Kita, O. Wada, L. Marsal, and H. Mariette, Phys. Rev. B69, 233308, 2004.

265

Time-resolved mid-infrared spectroscopy of excitons in CU2O M. Kubouchi\ R. Shimano^*, K. Yoshioka\ A. Mysyrowicz^, and M. Kuwata-Gonokami^ ^ Department of Applied Physics, the University of Tokyo, and Solution Oriented Research for Science and Technology (SORST), JST, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-8656, Japan E-mail: [email protected] ^ Laboratoire d'Optique Appliquee, ENSTA, Ecole Polytechnique, Palaiseau, France Present address: Department of Physics, the University of Tokyo, 7-3-1 Hongo, Bunkyoku, Tokyo 113-8656, Japan Abstract. Formation and thermalization dynamics of the yellow series excitons in CU2O are observed by monitoring the Is to 2p Lyman transition with femtosecond mid-infrared transient absorption spectroscopy. We discover fast and efficient generation of paraexcitons.

1.

Introduction

The Is states of the yellow series of CU2O consist of electric quadrupole allowed orthoexcitons (r5"^) and optically forbidden paraexcitons (r2'^). Both ortho- and paraexcitons have long radiative lifetimes because their direct radiative recombinations in the dipole approximation are forbidden. Accordingly, the yellow series excitons have been intensively investigated as candidates of excitonic BoseEinstein condensation (BEC) [1,2]. However, the interpretation of luminescence measurements is still unclear [3]. In addition it has been difficult to detect optically inactive paraexcitons sensitively, so the dynamics needed for the discussion about the possibility of BEC is not well known

2.

Experimental method

We demonstrate a new spectroscopic method to detect the temporal evolution of excitons based on the induced absorption measurement associated with the 1 s to 2p mid-infrared transition of excitons. In the induced absorption spectra the signal from ortho- and paraexcitons can be differentiated. While the energy splitting of 2p excitons are almost zero due to relatively large radii, the energy of orthoexciton is 12 meV higher than paraexciton due to the large electron-hole exchange interaction associated with the small Bohr radius close to the lattice constant. Beside, we can obtain information on the distribution functions of excitons from

266

the line shape analysis of the induced absorption spectra. Large central cell correction due to the small Bohr radius makes 1 s exciton mass heavier than 2p exciton mass expected to be almost the same as the sum of the individual conduction electron and valence hole masses, so the transition energy depends on exciton momenta, Thus, by measuring 1 s to 2p transition with time-resolved midinfrared pump-probe spectroscopy, we can obtain information on the spatiotemporal evolution of distribution functions and spin-dependent population [4]. The experimental setup is shown in Fig. 1(b). The pulse from a TiiSapphire based regenerative amplifier is converted to a pump (probe) pulse around 600 nm (10 |Lim) wavelength by second harmonic generation (difference frequency generation) in each optical parametric amplifier system, respectively. The pump wavelength mainly corresponds to T^^ phonon emitting absorption band of the orthoexciton. A 170 jim thick single crystal sample cut along the [100] plane is immersed in gas helium, and temperature is controlled by contact with liquid helium and a heater. The delayed probe pulse traverses the pumped region of the sample under a small incidence angle with respect to the pump beam. It is then analyzed by a monochromator (resolution ~ 0.1 meV) followed by a HgCdTe detector. (b)

(a)

Ti:Sapphire Regen, (1 kHz, 150fs)

HgCdTe d e t e c t o r / ^

Fig. 1. (a): Energy diagram for the Is and 2p exciton states in CU2O. Ascending arrows correspond to the measured absorption lines, (b): Experimental setup for induced absorption measurement.

3.

Results

The induced spectra shown in Fig. 2(a) in the aftermath of excitation are very broad, and excitons seem to have a large kinetic energy and not to be thermalized. Then broad and high-energy-skewed two induced spectra reflecting the exciton thermal distribution appear around 117 meV and 130 meV corresponding to Is to 2p transitions of ortho- and paraexcitons respectively. They gradually become narrower and less asymmetric exhibiting a red shift of their peak towards about 116 meV and 128 meV for ortho- and paraexcitons respectively with their thermal relaxation. Exciton temperature approaches lattice temperature after 200 ps (Fig. 2(b)). The induced absorption of paraexciton is as large as orthoexciton at a delay of a few tens picoseconds and that indicates that ortho- to paraexciton conversion occurs rapidly and efficiently. The Induced absorption of orthoexcitons decays

267

more rapidly than paraexcitons reflecting the difference of the lifetime due to their symmetry.

11.0

wavelength (iim) 10.5 10.0 9.5

800 ps 200 ps 60 ps 40 ps 20 ps

0.0-

115

120 125 energy (meV)

130

(b)

'^

X para o ortho

0100.

^ ' 6-^0 CC CD Q.

E 0)

42-

7.6 ps

c o o 10-

Ops

0

86-

X

o

^

1^

s 0

o

o

X

X

400 800 time (ps)

Fig. 2. (a): Differential absorption spectra obtained under one-photon excitation. The sample temperature is 4.2 K and the pump intensity is 170 mJ/cm^. Dashed lines are obtained spectra and their characteristic fringe patterns are due to multiple reflection of probe pulse in the sample. Solid lines are the spectra after removing the interference by Fourier transform filtering, (b): Exciton temperature obtained from a fitting of experimental spectra with a Maxwell-Boltzmann distribution using a Lorentzian with a 1 meV energy for the 2p state.

4.

Conclusions

By femtosecond mid-infrared pump-probe spectroscopy, we have succeeded in observing Is to 2p transitions of excitons in CU2O. Paraexcitons whose dynamics have not been well known are detected sensitively. We have measured the temporal dynamics of excitons and it has been found that paraexcitons are efficiently converted from orthoexcitons..

References 1 2 3 4

268

D. W. Snoke, J. P. Wolfe, and A. Mysyrowicz, Phys. Rev. 6 41,11171, 1990. N. Naka and N. Nagasawa, Phys. Status Solidi B 238, 379, 2003. K. E. O'Hara and J. P. Wolfe, Phys. Rev. B 62, 12909, 2000. M. Kuwata-Gonokami, M. Kubouchi, R. Shimano, and A. Mysyrowicz, J. Phys. Soc. Jpn. 73, 1065,2004.

Exciton dynamics in pentacene and tetracene studied using optical pump-probe spectroscopy V. K. Thorsm0lle\ R. D. Averitt^ J. Demsar\ X. Chi^ D. L. Smith\ A. P. Ramirez^ and A. J. Taylor^ ^Los Alamos National Lab, Los Alamos, New Mexico 87545, USA ^Columbia University, New York, New York, 10027, USA ^Lucent Technologies, 600 Mountain Avenue, Murray Hill, New Jersey 07974, USA Abstract: We present room temperature photoinduced reflection and transmission measurements in pentacene and tetracene single crystals using optical pump-probe spectroscopy. Singlet exciton recombination, singlet-triplet fission, excited singlet, and triplet state absorption is observed.

1.

Introduction

Organic semiconductors are attracting much interest due to their strong potential for use in technological applications [1]. Charge carriers play a key role in such applications, but their nature (molecular excitons or semiconducting band carriers) is still not completely understood and remains controversial [2,3]. In the molecular exciton model the excited states are localized and the primary photoexcitations are excitons, which can dissociate into free polarons, while in the semiconductor band model, they are delocalized, and mobile polarons are created directly from free electron-hole pairs by the absorption of light. Ultrafast optical measurements are important in this regard as the dynamics of the photoexcitations can be temporally resolved, and pump-probe experiments have been widely used to help elucidate the excited state dynamics of singlets and triplets [4,5]. Here we present optical pump-probe measurements of photoinduced (PI) changes in the reflectivity (AR/R) and transmissivity (AT/T) of tetracene and pentacene single crystals, respectively. Tetracene and pentacene belong to the polyacene series of organic crystals which, in order of increasing molecular size, includes, napthalene, anthracene, tetracene, and pentacene. The energy difference between the lowest singlet exciton and two lowest triplet excitons E{Si)-2E(Ti) is -1.3 eV in napthalene, -0.55 eV in anthracene, -0.21 eV in tetracene, and 0.11 eV in pentacene [4]. Therefore, in pentacene, the excitonic fission process from the lowest single exciton to a pair of lowest triplet excitons Si -^ 2Ti is energetically allowed, while in tetracene the Ti level can only be populated by thermally activated fission with a thermal excitation energy greater than 0.21 eV. This process is strongly suppressed in napthalene and anthracene. In these experiments we studied the photoexcited carrier relaxation dynamics in tetracene and pentacene as a function of probe photon energy aiming to elucidate the electronic structure and carrier dynamics in the singlet and triplet manifolds. In our data we see a strong PI absorption peak in pentacene, which is much weaker in tetracene

269

in agreement with the physics of the excitonic fission process. In addition, we infer from the broadness of the PI absorption band in both pentacene and tetracene that the final state in the triplet manifold is a band-like state. High quality single crystals were grown in a flow of inert gas. The experiments utilized a commercial-based regeneratively amplified Ti:Al203 laser system operating at 250 KHz producing nominally 10 jiJ, sub-50 fs pulses at 1.5 eV. The samples were excited at 3.0 eV with an excitation fluence of -100 jiJ/cm^, and the PI changes in reflectivity (transmissivity) were measured over the range of probe photon energies from 0.6-2.5 eV using an optical parametric amplifier. The photon energy of the pump pulse was above the apsorption band of -1.9 eV in pentacene, and -2.4 eV in tetracene. The approximate energy level diagrams for tetracene and pentacene are shown as insets to Fig. 1. Negative AT/T is associated with PI absorption, and positive AT/T is associated with PI bleaching. The thickness and the lower gap of the pentacene crystals complicated transmission measurements and the pentacene measurements were therefore obtained in reflection. It was however verified that AR/R had the same sign as AT/T.

2.

Results and Discussion

Fig. 1 shows the dynamics of PI changes in the (a) transmissivity of tetracene and (b) reflectivity of pentacene as a function of time for several probe photon energies. In both compounds the dynamics consists of two contributions: an exponential decay component with decay time x - 100±20 ps for tetracene and x 0.7±0.2 ps for pentacene, and a much longer lived contribution, manifested as an offset at times t » x. In tetracene, as Fig. 1(a) shows, at low and high photon energies the recovery is almost completed on a 100 ps timescale. This PI absorption can be associated with Si -^ Sn transitions and indicates a large singlet exciton population that recombines to the ground state. However, the long offset that is observed in the 1.33, 1.67, and 1.85 eV probe scans shows that some of the l.67eV

.^.^^vT+T ^ 3.1 eV 1 g (0.21 eV ^

N 2.2 eV

1.07 eV

X 0.86 eV

^^g^g^s^^^^^

400

600

Time [ps]

800

1200

100

200

300

400

500

Time [ps]

Fig. 1. Time-resolved PI change in tetracene single crystals versus pump-probe delay time at varies probe photon energies for (a) tetracene in transmission and (b) pentacene in reflection. The thickness and the lower gap of the pentacene crystals complicated transmission measurements.

270

population is transferred to the triplet state. As Fig. 1(b) shows, in pentacene the dynamics are substantially different. In this case the Si -> 2Ti fission process occurs on a 1 ps timescale, and the majority of the Si exciton population is rapidly transferred to the triplet manifold consistent with 2E(Ti) < E(Si). Fig. 2 shows the PI spectrum of the (a) transmissivity of tetracene and (b) reflectivity of pentacene at different time delays after photoexcitation. Both compounds display a broad long-lived ( » 1 ns) photoinduced absorption. In tetracene this feature is centered at approximately -1.7 eV, while in pentacene it is centered around -1.4 eV and is very pronounced. The long relaxation time suggest that the state being probed is the triplet state Ti (i.e. Ti -> Tn). The width of the PI absorption band suggests that the final state is a band-like state at 1.7 eV (1.4 eV) above the triplet state in tetracene (pentacene). This observation is consistent with the semiconductor band model [2,3]. Interestingly, we also note that the 1.7 eV (1.4 eV) PI absorption is very close to the energy difference ^

^°T^'

s^^s^ "^'yY^i o

-2

r' r

1 A//

s, — s ,

v^

- • — 1 ps ^:^10pS -*'—lOOps -^^-- 1 ns

(a) —.

1

1

1

1

r

1.0 1.5 2.0 Probe Energy [eV]

2.5

1.0 1.5 2.0 Probe Energy [eV]

Fig. 2. (a) Transient AT/T spectra at 1 ps, 10 ps, 100 ps, and 1 ns delay in tetracene. (b) Transient /IR/R spectra at 1 ps, 10 ps, and 100 ps delay in pentacene.

between the Ti energy level and Sn of 1.85 eV (1.34 eV) for tetracene and (pentacene) as shown in the approximate energy level diagrams in Fig. 1. Further investigations into the origin of the observed features are currently being pursued.

References 1 2 3 4 5

M.A. Baldo, M.E. Thompson, and S.R. Forrest, Nature 403, 750 (2000). F.A. Hegmann, Physics in Canada 59, 127 (2003). V.K. Thorsm0lle, R.D. Averitt, X. Chi, D.J. Hilton, D.L. Smith, A.P. Ramirez, and A.J. Taylor, Appl. Phys. Lett. 84, 891 (2004). C. Jundt, G. Klein, B. Sipp, J.Le Moigne, M. Joucla, and A.A. Villaeys, Chem. Phys. Lett. 241, 84 (1995). C. Frolov, Ch. Klog, J.H. Schon, and B. Badogg, Chem. Phys. Lett. 334, 65 (2001).

271

Dephasing suppression of excitons in semiconductors Tadashi Kishimoto , Atsushi Hasegawa , Yasuyoshi Mitsumori , Masahide Sasaki and Fujio Minami^ ^ Department of Physics, Tokyo Institute of Technology, Meguro-ku, Tokyo, 152-8551, Japan ^Communications Research Laboratory, Koganei-shi, Tokyo 184-8795, Japan 3i Research Institute of Electrical Communication, Tohoku University, Aoba-ku, Sendai 980-8577,Japan Abstract. The dephasing control was performed for excitons in GaSe by using successive three femtosecond pulses. By changing the pulse interval conditions, we confirmed for the first time the suppression of the dephasing of the excitons by n pulse irradiation.

1.

Introduction

Decoherence is one of the most important processes which gives information of the interaction between the electron system and the thermal reservoir. It also attracts much attention in device physics attempting to implement quantum information processing devices. A central issue is how to control and suppress the decoherence of well coded quantum states. Two well known methods to suppress decoherence are quantum error correcting codes and encoding with decoherence free subspace, which are still very challenging in practice. The other, and only plausible, method with present technologies might be the active dephasing control by using external pulsed-field sequences. The key idea is to reverse the time evolution of non-Markov dynamics by using optical n pulses [1-3]. Due to this time reversibility in the non-Markov regime, the dephasing time can be extended up to the energy relaxation time Ti by irradiating a sequence of TT pulses. Here we present the first observation, to our knowledge, of the suppression of exciton dephasing in a semiconductor by optical pulse irradiation. The experiment was performed for the excitons in GaSe by using three successive short pulses, i.e., sixwave-mixing (SWM) configuration. We compared the signal profile between the SWM and four-wave-mixing (FWM) signals, and confirmed that the exciton dephasing is suppressed by an additional n pulse.

2.

Experimental Methods

The FWM and SWM experiments were performed for the IS exciton (2.11 eV) in layered semiconductor GaSe by using an optical parametric oscillator (OPO), with a repetition rate of 76 MHz and a pulse duration of -200 fs, pumped synchronously by a mode-locked Ti:sapphire laser.

272

The excitation light was divided into three beams, and sent into the sample from three different directions with wave vectors ki, kj, and k^. In this article, the pulse with the ki vector will be denoted as #i, and the relative delay time between #i and #j is written as Ty. For the SWM in the inhomogeneous broadening case, the echo signal appears 2T23 away from the arriving time of the #1 pulse [4]. In the SWM experiments, therefore, T23 was scanned with a fixed value of T12 and the signal in the ki-2k2+2k3 direction was observed. In this phase-matching direction, we can detect only the signals generated by the process subjected to an additional 7c-pulse irradiation, as compared to the FWM case. For the FWM, the signal intensity in the -ki+2k3 direction was measured as a function of T13. It is possible to select the SWM and FWM processes only by choosing the observation direction without changing any other conditions. Therefore, the pump intensity and measured temperature (5K) are the same in both experiments.

3.

Results and Discussion

Figure 1 shows the signal profiles of the SWM as a fimction of T23 obtained for several values of T12. It can be seen from the figure that the decay profile shifts toward longer delays with the delay of the incident timing of the #2 pulse. On the other hand, the FWM profile was unaffected by the temporal position of the #2 pulse. The decay profile of the SWM for Ti2=0 ps is almost the same as that of the FWM. The dephasing of excitons arises from the random motion of the thermal reservoir, which rapidly modulates the exciton transition frequency and therefore causes degradation of the phase coherence. When the observation time scale is much shorter than the reservoir correlation time TC (non-Markovian limit), however, the modulation by the reservoir cannot be regarded as a fully random process. In this case, the slow frequency modulation due to the reservoir plays the role similar to that of the inhomogeneous broadening [5]. In the non-Markovian regime, therefore, the #2 pulse causes some of excitons to precess back towards their initial state when the #1 pulse was encountered, even if the excitons experience a pure dephasing caused by the exciton-reservoir interaction. This nonMarkovian behavior explains perfectly what we have observed [6]. To our knowledge, this is the first observation of the dephasing-suppression of the excitonic state. This shows that the coherence time can be lengthened to the theoretical lifetime limit of the exciton by irradiating a sequence of 71 pulses.

273

e e" a

e

*S

e

T T - T —- — T — • - - T

——

^=Ops 2=0.1ps ^=0.2ps 2=0.3ps 2=0.5ps

-1.0

2.5

Delay Time T^^ (ps) Fig. 1. The T^ dependence of the T23 scanned SWM profile

4.

Conclusions

We succeed the dephasing suppression of excitons in semiconductor by using the SWM method. By comparing the signal profile between FWM and SWM, the dephasing time is lengthened significantly with an additional pulse irradiation. This success shows the coherence time can be lengthened to the lifetime of exciton theoretically and opens the future of the application of excitons for quantum information processing.

References 1 2 3 4

M. Ban, J. Mod. Opt. 45, 2315 (1998). L. Viola and S. Lloyd, Phys. Rev. A58, 2733 (1998). C. Uchiyama and M. Aihara, Phys. Rev. A66, 032313 (2002). A. Hasegawa, T. Kishimoto, A. Hasegawa, M. Sasaki, and F. Minami, J. Lumin. 108,211 (2004). 5 M. Aihara, Phys. Rev. B25, 53 (1982). 6 M. Sasaki, A. Hasegawa, A. Hasegawa, F. Minami, J. Lumin. 108, 215 (2004).

274

Dynamical Stark effect of excitons in CU2O by resonant pulsed excitation of the ls-2/7 transition Kosuke Yoshioka, Motoyoshi Kubouchi, Ryo Shimano and Makoto KuwataGonokami Department of Applied Physics, Faculty of Engineering, University of Tokyo, and SORST (JST), 7-3-1 Kongo, Bunkyo-ku, Tokyo 113-8656, Japan Email: [email protected] Abstract. We examine the perturbed free induction decay of \s orthoexcitons in CU2O under resonant pumping of the \s-2p transition. Fourier transform pump-probe spectra indicate deep Stark potential of 2.7 meV with a pump intensity of 480 MW/cm^.

1.

Introduction

The manipulation of material states by coherent light-matter interactions is a challenge. In semiconductors, strong excitation-induced dephasing often prevents the use of this effect. However, the AC Stark effect [1] may be used for the optical manipulation of excitons or carriers if there is an efficient coupling between the radiation field and excitonic systems far from the band edge. A transition between excitons and biexcitons is an example of such a case [2,3]. Another way is with exciton internal transitions. In bulk CU2O, excitons from the yellow series have hydrogen-like Rydberg energy levels with a 150 meV total binding energy. The 1^ state is split into threefold degenerate, electric quadrupole allowed T^^ orthoexciton states and a non-degenerate, spin-forbidden r2^ paraexciton state due to the electron-hole exchange interaction. An inter-exciton transition such as \s-2p affects only the exciton envelope function and leaves the Bloch wave functions unperturbed. Thus we expect large transition dipole moments, which are favorable to enhance nonlinear optical phenomena. The AC Stark effect of orthoexcitons in CU2O under resonant pumping of \s-2p transition using nanosecond CO2 laser pulses has been demonstrated [4,5]. In this paper, we revisit this problem using femtosecond time-domain spectroscopy.

2.

Experimental Set-up

The sample is a naturally grown, 220 jim thick single crystal and is cooled down to 10 K. The beam of a 1 kHz regenerative amplifier system centered at 775nm with a pulse duration of approximately 200 fs is divided into two components. One is converted to a mid-infrared pump beam centered around 10 ^m using optical parametric amplification and difference-frequency generation processes. The other component is converted to a visible probe beam centered around 610 nm using optical parametric amplification and sum-frequency generation processes. Both

275

beams are focused on the (100) surface of the sample and have an angle around 10° with each other. Differential transmission spectra of the probe pulse are analyzed using a 50 cm focal length spectrometer with a liquid nitrogen cooled charge-coupled device (CCD).

3.

Experimental Results and discussions

Figure 1 shows differential transmission spectra obtained in the vicinity of the 1 s orthoexciton resonance with varying negative probe delays (i.e., probe preceding the pump pulse). No signal was observed with positive delays. When the pump frequency is almost in resonance with the \s-2p transition (Fig. la), the asymmetrical patterns for short time delays suggest a redshift of the orthoexciton level. On the contrary, when the pump frequency is above the \s-2p transition (Fig. lb), the asymmetrical patterns are reversed due to the blueshift of the Is energy level. The amplitude of the signal decreases as pump detuning increases. The oscillatory signal persists for negative delays longer than 50 ps. This indicates a long dephasing time for 1^ orthoexcitons in CU2O. 0 ps

2.030 2.032 2.034 2.036 Photon Energy (eV)

2.030 2.032 2,034 2.036 Photon Energy (eV)

Fig. 1. Differential transmission spectra around the Is orthoexciton resonance at 10 K. The pumpprobe time delay varies by 1 ps increment between each curve. The central pump frequency is: (a) 115 meV; (b) 134 meV. The \s-2p transition frequency is 116 meV. The pump energy is 95 |xJ/cm^ in both cases. The spectra shown at the bottom of each figure are the linear absorption \s orthoexcitons.

The observed spectral oscillatory structures can be interpreted as a free induction decay of the coherent quadrupole polarization of the l^- orthoexcitons perturbed by a nearly instantaneous Stark shift. We analyzed the differential transmission spectra in the limit of adiabatic dynamic regime, following the procedure used by Joffre and co-workers [6]. The extracted time evolutions of the Stark shifts are shown in Fig. 2. When the pump is almost resonant with the ls-2p transition, a large 2.7 meV Stark redshift is extracted from the spectrum shown in Fig. 1(a), while a 0.33 meV blueshift is extracted from the data shown in Fig. 1(b). Assuming a 200 fs pump pulse duration, we obtain a transition dipole moment M]s-2p - 1-5 eA from the energy shift shown in Fig. 2(b). Thus in order to form an optical potential for trapping quantum degenerate l^- excitons (paraexcitons as well as orthoexcitons) for a time as long as 10 ns with a far (5 meV), off-resonance mid-infrared pulse at a 2 K lattice temperature (0.2 meV), we need a tightly focused beam of approximately 5 |iJ per pulse, a value that can be reached with present-day technology.

276

3.0

3.5 4.0 4.5 Time (ps)

5.0

0.0

0.5 1.0 1.5 Time (ps)

2.0

Fig. 2. Time evolution of the Stark shift of the Is orthoexciton level (following the method reported in Ref [6]) extracted from the spectrum shown in Fig 2: (a) at a -4 ps time delay; (b) at a -1 ps time delay. The discrepancy between the experimental time delay and the extracted time evolution in (b) is due to ambiguity in determining the zero delay time.

4.

Conclusion

We observed the spectral oscillations of the 1^ orthoexciton level in CU2O caused by mid-infrared pulsed excitation nearly resonant with the \s-2p transition. This observation is explained by the perturbed free induction decay of coherent quadrupole polarization caused by the optical Stark effect. With a Fourier transform analysis, we extracted the temporal evolution of the exciton energy shift and estimated the ls-2p transition dipole moment to be 1.5 eA. This value is important for quantitative studies of exciton density using induced absorption experiment [7] and also for the design of optical traps to realize an exciton condensate. Further experiments using a lower energy detuning are in progress. Acknowledgement. The authors are grateful to Y.P. Svirko, A. Mysyrowicz and J. B. Heroux for fruitful discussions. * Present address: Department of Physics, Graduate School of Science, University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-8656, Japan.

References 1 A. Mysyrowicz, D. Hulin, A. Antonetti, A. Migus, W. T. Masselink and H. Morkog, Phys. Rev. Lett. 56, 2748, 1986. 2 R. Shimano and M. Kuwata-Gonokami, Phys. Rev. Lett. 72, 530, 1994. 3 S. Chesi, M. Artoni, G. C. La Rocca, F. Bassani, and A. Mysyrowicz, Phys. Rev. Lett. 91, 057402, 2003. 4 D. FrdhHch, A. Nothe, and K. Reimann, Phys. Rev. Lett. 55, 1335, 1985. 5 D. Frohlich, C. Neumann, B. Uebbing, R. Wille, Phys. Stat. Sol. (b) 159, 297, 1990. 6 M. Joffre, D. Hulin, J.-P. Foing, J.-P. Chambaret, A. Migus, and A. Antonetti, IEEE J. Quantum Electron. 25, 2505, 1989. 7 M. Kuwata-Gonokami, M. Kubouchi, R. Shimano, and A. Mysyrowicz, J. Phys. Soc. Jpn.,73, 1065,2004.

277

Ultrafast Charge Photogeneration and Exciton Regeneration at Polymeric Semiconductor Heterojunctions A.C. Morteani, P. Sreearunothai, L.M. Herz, R.H. Friend, and C. Silva Cavendish Laboratory, Cambridge University, Cambridge CBS OHE, UK Organic photovoltaic devices usually employ type II heterojunctions (see band diagram in Fig. 1(a)), in order to achieve efficient charge generation [1]. Typical devices involve thin films of blends of hole- and electron-accepting polymers sandwiched between two electrodes. Bulk excitons show relatively strong Coulombic binding (~ 0.5 eV), and can be trapped at the heterojunction, acquiring some charge-transfer character. Such excitations are termed exciplexes when seen in isolated donor-acceptor systems and are characterized by red-shifted emission spectra and long radiative decay times. We have shown that exciplexes form in blends of poly(9,9'-dioctylfluorene-co-benzothiadiazole), F8BT, with poly(9,9'-dioctylfluorene-co-bis-N,N'-(4-butylphenyl)-bis-N,N'-phenyl-l,4 phenylenediamine), P F B , and that these exciplex states can undergo endothermic energy transfer to form bulk F8BT excitons (EA ~ 200 meV [2]). Here we investigate thin films of PFB:F8BT blends which contain a high density of heterojunction sites throughout the film. We have applied femtosecond transient absorption (FTA) spectroscopy, timecorrelated single photon counting (TCSPC), and photoluminescence upconversion (PLUG) spectroscopy, as described elsewhere [2-4] to investigate the photoexcitation dynamics in PFBiFBBT. The excitation energies were 3.100, 3.046, and 3.061 eV and the time resolution was 200 fs, 100 ps, and 300 fs, respectively. The photoluminescence (PL) measurements were done on diode structures to investigate the electric field effects on exciton and exciplex dynamics. Polymer blends (mass ratio 1:1) were spin-coated from common chloroform solution onto indium-tin-oxide (ITO) substrates to form 170 nm films. Ca electrodes (60 nm) were deposited by thermal evaporation and encapsulated by an Al layer. Devices were fabricated in N2. The electric field was applied by reverse-biasing the diodes to prevent charge injection. For FTA, similar films were spin-coated onto Spectrosil-B substrates. Fig. 1(a) displays the FTA spectrum of a PFBiFSBT film at a pump-probe delay of 1 ps. At this short delay, the spectrum consists of a positive feature centered at 2.7eV, assigned to the ground-state photobleach (PB) of F8BT due to the similarity with its ground-state absorption. To the red of 2.5 eV, the spectrum is dominated by a broad photoinduced absorption (PA) that extends into the nearinfrared region of the spectrum. This has been assigned previously to absorption of photogenerated polarons in F8BT and related polyfluorene materials [5]. Fig. 1(b) demonstrates that both P B and PA display instrument-limited growth, indicating that charge separation occurs on a < 200 fs timescale. In this donor-

278

acceptor blend, the length scale of phase segregation is small enough [6] such that static quenching at heterojunctions produces the rapid growth of chargeinduced absorption. The transient spectrum decays uniformly on a timescale of ~ 5ps, followed by a much slower decay resulting in a nearly constant offset in the 1-ns experimental window. We interpret the rapid component as radiationless recombination of geminate pairs, with a significant charge population surviving on the nanosecond timescale. We now turn to the dynamics of neutral

(a)

Energy

-\

t:

PFB TFB

1 1 1

< 1

)

1 1 1 1 1 1 1

2.0

•

1 1 . 1

2.4

Photon Energy (eV)

10

1000

Time (ps)

Fig. 1. (a) Femtosecond transient absorption spectrum from PFB:F8BT at a pumpprobe delay of 1 ps. The inset shows the band offsets at a type II heterojunction (see also [6]). (b) Absorption transients at the probe photon energies indicated in the figure

excitations (bulk excitons and exciplexes). We demonstrated previously that the PL of PFB:F8BT is susceptible to electric-field-induced quenching even at low fields [6]. The electric field mostly quenches the exciplex contribution in the red part of the spectrum (> 50% quenching for photon energies < 1.94eV at -10 V bias). If the PL quenching arises from field-assisted dissociation of an emissive state, its PL decay rate should be field-dependent. Fig. 2(a) shows T C S P C measurements at 1.937eV in a PFB:F8BT diode with different applied voltages. All curves consist of an instrument-limited decay, and a slow, mono-exponential decay with 40 ± 5 ns decay constant. The two components are assigned to the bulk exciton and the exciplex state, respectively [2]. Exciplex generation occurs within ~ 1 ns and its generation efficiency is strongly field-dependent, while its decay constant shows no significant variation. Therefore, an exciplex precursor must be quenched by the field. To investigate the field dependence on the bulk exciton decay rate, we have performed field-dependent PLUG measurements (Fig. 2(b)). The exciton decay dynamics are not field dependent. Therefore, a dark intermediate state must be dissociated by the field. We postulate that this state is an interfacial geminate polaron pair that follows charge transfer from the bulk exciton and survives radiationless geminate recombination. We propose the following picture. Photogenerated, "primary" excitons chargetransfer at the heterojunction to form geminate polaron pairs which either dissociate fully or relax into the luminescent exciplex. The branching ratio between free polarons and exciplexes is field-dependent and hence also the PL emission yield and spectrum. The exciplexes either decay or endothermically back-transfer

279

4

6 8 Time (ns)

40 60 Time (ps)

100

Fig. 2. (a) PL decay measured using TCSPC (< 4nJ/cm^ excitation fluence, detection at 1.937 eV) from a PFB:F8BT device at room temperature under 0 V (continuous line), 13 V (circles) and 30 V (triangles) applied reverse biases, (b) PLUG measurements (42nJ/cm^ excitation fluence, detection at 2.255 eV) from a similar device at 0 V (continuous line), 5V (squares) and 12.5 V (triangles) reverse bias. For comparison, data for a device with pure F8BT at 0 V (continuous line) and 12 V (circles) are also plotted

to bulk excitons ("secondary" excitons) [6]. Since the exciplex density is reduced by the electric field, there is less secondary exciton generation, and hence the observed field-induced PL quenching spectrum [6] contains an excitonic contribution. The ratio of secondary excitons to exciplexes is thermally activated with EA = 200 d= 50 meV. We note that the regeneration of the exciton via the thermally-driven circular process (exciton-^geminate pair^exciplex^exciton) means that even though charge transfer occurs, the excitation energy might eventually still be emitted in the form of bulk exciton luminescence. The PFB:F8BT blend is an important example for efficient charge generation. Given that excitedstate electronic dimers are commonly observed in polymeric semiconductors [7], we consider exciplex formation and exciton regeneration to also be general phenomena at type-II heterojunctions. In photovoltaic diodes, the presence of the exciplex provides an unwanted loss channel, and exciplex stabilization could be inhibited by increasing intermolecular distances and decreasing configurational relaxation.

References 1. J. J. M. Halls, C. A. Walsh, N. C. Greenham, E. A. Marseglia, R. H. Friend, S. C. Moratti, A. B. Holmes, Nature 376, 498, 1995. 2. A. C. Morteani, A. S. Dhoot, J.-S. Kim, C. Silva, N. C. Greenham, R. H. Friend, C. Murphy, E. Moons, S. Cina, J. Burroughes, Adv. Mat. 15, 1708, 2003. 3. C. Daniel, L. M. Herz, C. Silva, F. J. M. Hoeben, A. P. H. J. Schenning, E. W. Meijer, Phys. Rev. B 68, 235212, 2003. 4. L. M. Herz, C. Daniel, C. Silva, F. J. M. Hoeben, A. P. H. Schenning, E. W. Meijer, R. H. Friend, R. T. Phillips, Phys. Rev. B 68, 045203, 2003. 5. M. A. Stevens, C. Silva, D. M. Russell, R. H. Friend, Phys. Rev. B 63, 165213, 2001. 6. A. C. Morteani, P. Sreearunothai, L. M. Herz, R. H. Friend, C. Silva: Phys. Rev. Lett. 92, 247402 (2004) 7. B. J. Schwartz, Annu. Rev. Phys. Chem. 54, 141, 2003 and references therein.

280

Exciton diffusion dynamics in an organic semiconductor nanostructure Clement Daniel^, Laura M. Herz^, Sebastian Westenhoff^, Frangois Makereel^, David Beljonne^, Freek J. M. Hoeben^, Pascal Jonkheijm^, Albertus P. H. J. Schenning^, E. W. Meijer^, and Carlos Silva^ ^ Cavendish Laboratory, University of Cambridge, Madingley Road, Cambridge CBS OHE, United Kingdom ^ Chemistry of Novel Materials, University of Mons-Hainaut, Place du Pare 20, B-7000 Mons, Belgium ^ Laboratory of Macromolecular and Organic Chemistry, Eindhoven University of Technology, P.O. Box 513, 5600 MB Eindhoven, The Netherlands

1

Introduction

Polymeric semiconductors are now finding commercial applications in optoelectronic devices such as light-emitting diodes and photovoltaic diodes. Here, we investigate exciton dynamics in chiral helices of an oligo-p-phenylenevinylene derivative monofunctionalized with a ureido-s-triazine (MOPV, see Fig. 1(a) for the molecular structure). The latter functional group leads to oligomer dimerization by hydrogen bonding in dodecane solution, resulting in reversible supramolecular assembly by solvophobic and TT-TT interactions [1]. MOPV is a model system to study exciton transfer dynamics as it is possible to investigate the effect of intermolecular interactions by comparing optical properties in the dissolved phase (above the transition temperature) and the supramolecular assemblies.

2

Experimental

The synthesis of MOPV has been described in detail elsewhere [1]. The material was dissolved in anhydrous dodecane at a concentration of 1.6 xlO~^M and then kept under inert atmosphere except during absorption (Abs) measurements. Measurements were carried out using a temperature-controlled Abs cuvette. The transient photoluminescence (PL) and Abs apparatus have been described in detail previously [2,3].

3

Results and Discussion

We have undertaken transient PL and Abs studies of MOPV in dodecane solution at sufficiently low concentration to explore exciton dynamics within isolated supramolecular assemblies. Fig. 1(b) plots the PL intensity as a function of time at various solution temperatures, at the peak of the PL spectrum. Below the transition temperature for supramolecular assembly (65°C), nonexponential decay is

281

observed, which we have assigned to diffusion-assisted exciton quenching at defect sites [2]. This is accompanied by a dynamic red shift of the PL spectrum and by PL depolarization, both assigned to the exciton diffusion process [2]. Fig. 1(c) displays photoinduced Abs transients at a photon energy that probes Abs of the lowest exciton, with higher excitation fluence than used in the transient PL measurements. An intensity-dependent picosecond component is assigned to exciton bimolecular annihilation dynamics [3]. Both of these sets of results indicate that exciton diffusion in MOPV stacks in solution occurs on picosecond timescales. We have developed a simple Monte-Carlo simulation (MC) in order to obtain

(a)

.y ^

IT

%^^M 100 200 300 400 500 Time (ps)

Fig. 1. (a) Chemical structure of MOPV and schematic representation of a MOPV supramolecular stack, (b) PL decay kinetics at 2.226 eV photon energy at various temperatures, (c) Abs transients with 1.46 eV probe photon energy at 14° C and at various pump fluences. a microscopic understanding of these dynamics. The model generates randomly orientated helicoidal packs of oligomers with uncorrelated random energies (following a Gaussian law). The excitons are created at random positions and can then diffuse on the supramolecular nanostructures. At any time step, any exciton can transfer to another unoccupied site, annihilate with any other exciton or decay with a constant probability (the inverse of the lifetime). The hopping and the annihilation events are treated as resonance energy transfer events and modelled with a Forster-type probability law using a line-dipole approximation. The latter is a refinement of the point-dipole approximation and is needed as the stacks are very closely packed. In this scheme, each oligomer is divided into small segments used to calculate the electronic coupling using a point dipole approximation (see [4] for a discussion). The transfer rate includes an overlap integral of the homogeneous PL and Abs spectra which depends on the difference in site energies. Several parameters were deduced independently from previous measurements. The PL, Abs and CD spectra were fitted with two vibronic progressions and a Gaussian broadening to obtain the homogeneous spectra (the

282

transfer probabilities follow a detailed balance principle). The model allows us to extract the decays of the exciton population (see Fig. 2(a)), of the average exciton energy (see Fig. 2(b)) and of the PL polarisation anisotropy (see Fig. 2(c)).

1.0 c 0 Q

0.5 0.0

Q.

2 "5 "c < CD

C LU

Experimental X Low Fluence O High Fluence Simulation

-4::

-+-

1—

0.4 0.2

0.0 2.175

O MOPV Simulation

(c)

2.150 2.125

MOPV O Simulation J I L

0.01 0.1 1 10 Time (ps)

100

Fig. 2. (a) The symbols are the normalised exciton population decays at high and low excitation fluences. The curves are the decays from the MC simulation, (b) The circles are the polarisation anisotropy of the exciton Abs, while the curve is from the MC simulation, (c) The circles are the PL average energy measured with PL up-conversion, while the curve is from the MC simulation.

4

Conclusion

The model allows to extract microscopic informations such as the average exciton hopping annihilation rates. We believe that this model allows us to describe quantitatively the exciton dynamics on supramolecular architectures from subpicosecond to nanosecond timescales.

References 1. P. Jonkheijm, F. J. M. Hoeben, R. Kleppinger, J. van Herrikhuyzen, A. P. H. J. Schenning and E. W. Meijer, J. Am. Chem. Soc. 125, 15941, 2003. 2. L. M. Herz, C. Daniel, C. Silva, F. J. M. Hoeben, A. P. H. J. Schenning, E. W. Meijer, R. H. Friend and R. T. Phillips, Phys. Rev. B 68, 045203, 2003. 3. C. Daniel, L. M. Herz, C. Silva, F. J. M. Hoeben, P. Jonkheijm, A. P H. J. Schenning and E. W. Meijer, Phys. Rev. B 68, 235212, 2003. 4. W. J. D. Beenken and T. Pullerits, J. Chem. Phys. 120, 2490, 2004.

283

Part IV

Ultrafast Dynamics in Solid 2

Carrier-envelope phase controlled quantum interference in a semiconductor T. M. Fortier^ P. A. Roos^ David J. Jones^ S. T. Cundiff, R. D. R. Bhat^ and J. E. Sipe^ ^ JILA/ National Institute of Standards and Technology and the University of Colorado, Boulder CO 80309-0440 E-mail: [email protected] ^ Department of Physics, University of Toronto, Toronto ON M5S 1A7, Canada Abstract. We demonstrate quantum interference of injected photocurrents in a semiconductor using a phase stabilized modelocked Ti:sapphire laser. Using this technique we detect the laser offsetfrequencywith a 40 dB signal to noise ratio in a 10 Hz resolution bandwidth.

1.

Introduction

One recent development in optics is the pov^erful ability to precisely control the optical phase of femtosecond (fs) light pulses. Optical manipulation at this level permits investigation into coherent control experiments that influence quantum reaction channels in atomic, molecular or solid-state systems. So far, experiments of this type have only been observed in the high-field regime [1, 2]. We demonstrate coherent control in the perturbative regime, whereby quantum interference (QI) in a semiconductor is used to produce an injected photocurrent whose direction and magnitude is sensitive to phase of the electric field of a fs laser pulse train [3]. The presented technique allows for realization of a solid-state optical phase detector for direct measurement of the laser offset frequency. One means for exploring QI in a semiconductor is obtained by interfering oneand two- photon pathways that connect the continuum levels of the valence and conduction bands. In the presence of a two-color light field of frequencies, v and 2 v, interference between the one- and two- photon pathways may occur, whereby the excited carrier population to the conduction band depends on the phase parameter, A(fr=2(fiy-(/)2y.

(1)

UQYQ, ^y ((f>2y) is the phase of the v (2 v) light field. For GaAs when this phase parameter is such to produce constructive interference at k, it produces destructive interference at -k [see Figure. 1 (a)], thereby resulting in a rate of population transfer that is odd in k. The resulting asymmetry in electron population results in a net injected current flow J, with a dependence on the phase parameter.

287

Fig.l. (a) Schematic depicting quantum interference between one- and two- photon pathways that connect the valence and conductions bands of LT-GaAs. Control of injected photocurrents is obtained by manipulating the carrier-envelope phase of light pulses, (b) The carrier-envelope phase between adjacent pulses in the pulse train evolves at a defined rate,/o = 1/T.

dJ (A^)/dt = C {EyfE2y sin(zl^) - J/r,

(2)

Here Ey and E2v are the electric fields corresponding to the optical frequencies v and 2v, respectively. C is a constant that includes the semiconductor nonlinear susceptibility and r is a phenomelogical decay time due to electron momentum relaxation (-200 fs). The first observations of QI in GaAs where presented by Hache et al. using fundamental light from an optical parametric oscillator and its generated second harmonic [4]. In their experiments, directional control of the injected photocurrent was obtained by varying the relative optical path lengths in a two-color interferometer. In contrast, the measurement presented here controls the inject photocurrent in LT-GaAs by directly manipulating the carrier phase of the laser pulse train of a stabilized Ti:sapphire laser. This technique enables single pulse control of the generated photocurrents. Pulse formation in a mode-locked laser results from the coherent addition of resonant longitudinal cavity modes that satisfy the condition v„ = « frep when intracavity dispersion is negelected, where frep is the laser repetition rate and n is a large integer (~10^). In a Tiisapphire (Ti:S) laser, Kerr-Lens modelocking forces coherence between the individual electric fields, E(Vr), such that each mode shares a

288

ja.5!K...\_'_r_.h*j;_v-..

Fig. 2. Experimental setup used for quantum control of injected photocurrents in LT-GaAs. The laser offsetfrequencyis measured using a v-to-2v interferometer and is stabilized via negative feedback to the laser end mirror (EM). The laser output is broadened using microstructured (MS) fiber, both for QI and for stabilization. Because of the resulting fiber dispersion, a prism sequence is used for time delay compensation, and a split mirror is used for spectral filtering. Light V and 2v are p-polarized such that the electric field oscillates transversely along the axis between the two gold electrodes on the QI sample (inset).

common phase, (/>cE(t)' Thus,:E(v„)= Eyn exp {-/ ^vw } = ^v« exp {-/ [ 2 Tcnfrep t + ^c£;(0]}-A non-zero phase shift, d^cEWdt 9^ 0, is the result of intracavity material dispersion, which alters the vacuum cavity resonance condition such that v„ ^ nf^ep . The net result of dispersion is a rigid shift of the optical comb by a common offset frequQncy f0=1/(2TT) d(^cE/dt, such that v„ = n frep + fo [5], Control of the carrierenvelope phase evolution of the laser pulse train is thus obtained via active stabilization offo. The offset frequency is measured using one- and two-photon optical interference using a v-to-2v interferometer [see Fig. 2] (for details concerning the interferometer and stabilization, please consult ref. 6) Stabilization of/o is then obtained via negative feedback to the laser end mirror for fine control of the laser intracavity dispersion. Stabilization of the pulse train carrier-envelope phase using this technique may result in phase coherence times of several minutes [6]. We induce injected photocurrents in LT-GaAs using interference between the spectral extremes of an octave spanning spectrum generated by coherently broadening Ti:S laser light in microstructured fiber [7]. The broadened light is focused onto a LTGaAs sample, between two gold electrodes separated by 10 jiim [see Fig. 2 inset]

289

2000

2200

n 2400

frequency (Hz)

1

r 2600

400

600

800

1000

1200

wavelength (nm)

Fig. 3 (a) Spectrum of the measured QI signal (10 Hz bandwidth, 100 averages). The linewidth (resolution limited) of the measured QI signal indicates the stability of the laser offset frequency. The dotted and the solid grey lines show the QI signal corresponding to the optical spectra in (b). The solid line indicates the background when the light is blocked to the sample, (b) Sidebands on the 2.38 kHz signal are the result of unsuppressed optical noise. The dip at 500 nm is the result of unavoidable filtering due to the small gap between the split mirrors in Fig. 2. (for details about sample preparation and the sample itself, please consult ref. 8). Interaction of the system with coherently related light at v and 2 v results in an injected photocurrent whose sign and magnitude are determined by Eqn. 2. Here the phase parameter is zl^ =2 ^^ - ^^v = (I>CE (0 ^ '^^fo t + ^o. where ^o is a constant phase offset. Figure 3(a) shows the frequency spectrum of the induced photocurrent in LTGaAs for a carrier-envelope phase evolution corresponding to stabilizing/^ to 2.38 kHz. We show the photocurrent for two optical spectra, one filtered and one unfiltered since the inclusion of above gap carriers (>874 nm), not contributing to the interference, in some instance can reduce the detected photocurrent signal. This is not the case since QI in the semiconductor is observed to behave as a current source (the injected photo-carriers are collected directly at the electrodes) rather than a voltage source (the injected photo-carriers result in a charge separation in the sample, measured as a potential difference across the electrodes). The observed photocurrent signal is 130 pA with a 40 dB SNR in a 10 Hz bandwidth, with 100 averages. The same signal, measured using lock-in detection was observed to be greater than 20 |LIV (using a 100 kn load resistor) for a pulse repetition rate of 93 MHz and with an average power (spot size) at vand2vof 11.9 mW (10.2 ± 0.7 jim) and 1.53 mW(11.5 ± 0.7 jLim), respectively. An additional degree of freedom that we manipulate is the overall phase of the carrier-electric field ^^J which is controlled by passing the laser beam through a thin (176 \xm) piece of glass. By rotating the angle of the glass plate in the path of the beam, the effective thickness of the plate is changed, thereby dispersively altering the

290

(b) 1.5

(a) 1.5

CO

1.0

0) Q.

0.0 40

60

time (s)

I

I

I

10 20 30 rotation angle (deg)

40

Fig. 4. (a) Phase of the 2.38 kHz QI signal measured using lock-in detection. The phase of the signal is measured relative to the reference used to stabilize the laser, A time record of the phase of the QI signal (100 ms time constant) shows phase jumps associated with rotation of the glass plate from 0 degrees to ^and back again for 8 different rotation angles, (b) Comparison between the phase and the calculated change in carrier-envelope phase versus rotation angle, 6. Deviation of experiment from calculation at larger rotation angles results from a slight misalignment of the beam into the v-to-2v interferometer.

pulse train electric field. Changes in the phase of the pulse train are measured as changes in phase of the photocurrent signal as presented in Fig. 4. In summary, we have demonstrated quantum interference control of directional photocurrents in a semiconductor using the carrier-envelope phase of a stabilized Ti:S laser pulse train. Additionally, this technique demonstrates a convenient means for solid-state detection of the laser offset frequency and for measurement of small changes in the pulse train carrier-envelope phase. Acknowledgements. Funding for this project was provided by the NSF, NIST and ONR. T.M.F. thanks NSERC for personal funding.

References A. Baltuska, et al. Nature 421, 611 (2003) G. G. Paulus, et al., Nature 414, 182 (2001) T.M. Fortier, et al, Phys. Rev. Lett. 92, 147403 (2004). A. Hache, et al, Phys. Rev. Lett. 78,306 (1997) H. Telle, et al, Appl. Phys. B 69,327 (1999) D.J Jones, et al. Science 288, 5466 (2000); T. M. Fortier, et al, IEEE J. Sel. Topics in Quantum Electr., 9,1002 (2003) J. K. Ranka, R. S. Windeler and A. J. Stentz, Opt. Lett. 25, 25 (2002) P. A. Roos, et al. Opt. Exp. 11, 2081 (2003) 9. P.A. Roos, et al, to appear in J. Opt. Soc. Am. B (2004).

291

Phase-resolved nonlinear response of modulation-doped quantum wells under femtosecond intersubband excitation T. Shih^ C. W. Luo^ K. Reimann^ M. Woemer^ T. Elsaesser^ I. Waldmullei^, A. Knorr^, R. Hey^ and K. H. Ploog^ ^ Max-Bom-Institut fur Nichtlineare Optik und Kurzzeitspektroskopie, 12489 Berlin, Germany, e-mail: [email protected] ^ Institut fur Theoretische Physik, Technische Universitat Berlin, 10623 Berlin, Germany ^ Paul-Drude-Institut fur Festkorperelektronik, 10117 Berlin, Germany Summary. Coherent nonlinear propagation of ultrafast electric field transients through intersubband resonances of GaAs/AlGaAs quantum wells shows Rabi oscillations for low electron densities, while radiative coupling between quantum wells dominates the nonlinearity at high densities. The linear and nonlinear optical properties of semiconductor nanostructures like quantum dots, wires, and wells are typically dominated by the linear and nonlinear susceptibilities of the individual quantum object. Even if these objects are electronically uncoupled, they can still be coupled by radiative coupling. This effect has often been neglected, especially for intersubband transitions with their long transition wavelengths. Here we present linear and nonlinear transmission measurements of the intersubband resonance on n-type modulation-doped GaAs/AlGaAs multiple quantum wells, which clearly show that for high electron densities radiative coupling becomes the dominant nonlinearity. In contrast, at low electron densities the quantum wells act almost as uncoupled two-level systems, allowing to observe coherent Rabi oscillations. For the experiments we use two n-type modulation-doped GaAs/AlGaAs multiple quantum well (MQW) samples, grown by molecular beam epitaxy, which differ in their electron density. One sample, sample L, has a low electron density of 5 X 10^^ cm"-^ per quantum well, the other, sample H, has a high electron density of 1.2 X 10^^ cm~^. Otherwise the samples are identical, both consisting of 51 GaAs wells of 10-nm thickness separated by 20-nm AIQ 35GaQ ^^As barriers, thick enough to prevent an electronic coupling between quantum wells. Because of the prism shape of the samples [1], the incident beam hits the quantum wells at an angle of 60° to achieve a strong coupling of the p-polarized incident beam with the intersubband dipoles, which are oriented perpendicular to the layers. After passing the quantum wells, the beam travels through a spacer layer, is totally reflected from the sample surface, and hits the quantum wells a second time. The thickness of the spacer layer is chosen so that the quantum wells are in the antinode of the resulting standing wave.

292

Low electron density

High electron density

(c) Experinnent

1.0 h

(b) Theory

110 120 90 Photon energy (meV)

Fig. 1. Measured [(a), (c)] and calculated [(b), (d)] linear intersubband absorption spectra of two n-type modulationdoped GaAs/AlGaAs samples [sample L: (a) and (b), sample H: (c) and(d)].

Linear absorption measurements show for both samples an intersubband absorption line around 100 meV (Fig. 1). Surprisingly, the linewidth of this line is about the same in both samples. Even more surprising, the absorption strength of both samples is equal, although sample H has 24-times as many electrons as sample L. The nonlinear optical properties are investigated in coherent nonlinear propagation experiments similar to [1] using, however, much higher electric field amplitudes [2, 3]. Nonlinear transmission measurements are performed with 200-fs pulses resonant to the intersubband transition. After passing through the sample, the electric field of the transmitted pulse is time-resolved by electro-optic sampling. At amplitudes of the electric field higher than 10 kV/cm, sample L shows clear Rabi oscillations, seen here as a periodic switching from absorption to gain (Fig. 2). Despite the similarity of the absorption measurements in samples L and H nonlinear transmission measurements on sample H give completely different results.

- II

Incident

"""'^i

Re-emitted

- . . 1 11 .

-

Fig. 2. Results for sample L (electron density 5 x 10^" cm""^ per quantum well) for three different amplitudes of the incident electricfield.The thin lines are the incident electric field transients, the thick lines the transients of the re-emittedfield,obtained as the difference of incident and transmittedfield.For low amplitudes (a) the re-emitted field shows the typical free-induction decay. For higher amplitudes one observes Rabi oscillafions, e.g., in (c) the re-emittedfieldisfirstout of phase with the incidentfield,showing absorption, then in phase (gain), and then again out of phase (absorption).

293

1.0

0.8

§ 0.6 ,00

^ 0.4 1 . 2 x 1 0 ' ^ cm 0.2

0.0

10

100

Electric field amplitude (kV/cm)

Fig. 3. Nonlinear transmission r of a 200-femtosecond infrared pulse resonant to the intersubband transition as a function of the amplitude of the incident electric field amplitude for samples L (dots) and H (triangles). The lines are theoretical results.

The most striking difference (Fig. 3) between samples H and L is seen in the nonlinear transmission T defined by J = jEQ^^{t)^dt/jE^^{tydt. For sample L one finds the expected result, namely an increase of the transmission (bleaching) with increasing pulse energy according to the coherent driving of Rabi oscillations [2]. In contrast, for sample H the transmission first decreases with increasing amplitude (induced absorption), reaches a minimum, and then increases again. These surprising results can be explained by the radiative coupling between the individual quantum wells in the multiple quantum well sample, which leads to the formation of a new correlated excitation involving all quantum wells. For the linear response (Fig. 1), we developed a many-body theory by combining radiative coupling with a microscopic material theory (including carrier-carrier and carrier-phonon correlations) [4]. Here, we extent this theory for the description of time-dependent nonlinear phenomena. Because the full calculation of the dephasing terms is numerically very demanding, time-dependent relaxation processes are approximated by collision terms in the relaxation rate approximation. The corresponding constant scattering rates are determined from the microscopic calculations for the linear spectra. Already this approach is in good agreement with the experiment (see lines in Fig. 3). Calculations including the full, time-dependent carrier-carrier and carrier-phonon correlations are under way.

References 1. 2. 3. 4.

F. Eickemeyer, et al, Appl. Phys. Lett. 79, 165, 2001. C. W. Luo, et al., Phys. Rev. Lett. 92, 047402, 2004. K. Reimann, et al. Opt. Lett. 28, 471, 2003. L Waldmiiller, et al, Phys. Rev. B 69, 205307, 2004.

294

Ultrafast Intersubband Relaxation and Carrier Cooling in GaN/AlN multiple quantum wells Junichi Hamazaki\ Hideyuki Kunugita\ Kazuhiro Ema\ Satoshi Matsui^, Yohei Ishii^, Takayuki Morita^, Akihiko Kikuchi^ and Katsumi Kishino^ ^ Department of physics, Sophia University, 7-1 Kioi-cho Chiyoda-ku, Tokyo, Japan E-mail: [email protected] ^ Department of Electric and Electronics Engineering, Sophia University, 7-1 Kioi-cho Chiyoda-ku, Tokyo, Japan Abstract. We have investigated intersubband relaxation dynamics in GaN/AlN multiple quantum wells by the two-color pump-probe technique. We have clarified the ultrafast relaxation scenario which includes thermalization and cooling processes, and found that relaxation is influenced by a hot-phonon effect.

1. Introduction Since the intersubband transition (ISBT) energy in GaN/AlGaN-based heterostructures covers the optical communication wavelength range (1.2-1.6 jum (0.8-1 eV)) [1-3], the ISBT has attracted much attention as a good candidate for ultrafast opto-electronic devices. In our previous investigation, we showed that the ISBT relaxation dynamics in GaN/AlN multiple quantum wells (MQW) has two decay constants (ti ~ 140 fs and t2 ~ 1.3 ps) which were tentatively attribute to inter-subband and intra-subband relaxation times [4]. However the detailed mechanism of the ultrafast relaxation dynamics has not been clarified so far. In this paper we report experimental results of one-color and two-color pumpprobe measurements and discuss the ISBT relaxation dynamics from a viewpoint of electron thermodynamics. Our results elucidates the relaxation scenario including the role of intersubband relaxation, thermalization and cooling processes.

2. Experimental set up The investigated GaN/AlN MQW sample was grown on a (0001) sapphire substrate by improved rf-molecular beam epitaxy [5]. In this sample the well width is 11 A and the absorption peak due to the ISBT is observed at 1.56 jum. The ISBT relaxation time was measured using one-color and two-color timeresolved pump-probe techniques. The pump and probe pulses were produced by an optical parametric amplifier (OPA) excited by an amplified mode-locked Ti:Al203 laser at a repetition rate of 100 kHz. The temporal width of the optical pulses was -100 fs. We set the wavelengths of the signal and idler pulses from the OPA to 1560 nm (coi) and 1650 nm (CO2), respectively. We performed three types of pump-

295

probe experiments: [A] a one-color degenerate type using (Oi, [B] a two-color type with the pump pulse at (O2 and the probe pulse at coi, [C] a two-color type with the pump pulse at coi and the probe pulse at C02- The intensity of probe pulse was set to be I/IO^*" of that of the pump. In order to ensure sufficient interaction with the ISBT dipole moment, the incident angle of the pump beam was set equal to the Brewster angle for the sample. The pump beam was chopped at a frequency of 1840 Hz for lock-in detection.

3. Result and discussion Fig. 1 shows the temporal traces of the change in the transmittance of the probe pulses induced by the pump pulse for [A], [B] and [C] experiments. In all experiments we observe transient increases of transmission, which rise within the time resolution and decay non single-exponentially. The decay dynamics of [B] is almost identical to that of [A], while the initial decay of [C] is clearly slower than those of [A] and [B]. The initial fast decay of [A] and [B] is corresponding to the fast components of 140 fs in Ref [4]. In experiments [A] and [B], the coi pulse probes states at the bottom of the n=2 band (lower energy part). On the other hand, in experiment [C], the CO2 pulse probes states in the higher energy part. The intersubband energy spacing becomes smaller for larger wave-numbers due to the nonparabolicity of the subbands. Therefore the origin of the increase in the transmittance at t =0 of [A] and [B] is a phase space filling at the bottom of the n=2 band. This indicates that the fast component of 140 fs is the intersubband relaxation time. There is no phase space filling effect in experiment [C], because the CO2 pulse probes the large wave-number states which have no electron population. Fig. 2(a) shows the electron distribution of the n=l band at room temperature before the pump excitation. After the excitation, the holes in the n=l band recover immediately due to carrier-carrier scatterings and a quasi-equilibrium distribution with a lower Fermi energy is formed within the time resolution of the experiments (Fig. 2(b)). As shown in Fig. 2, there is little difference between the transient electron distribution (b) and the initial distribution (a) at the coi-probe energy position, while there is a substantial decrease of electron population at the CO2probe energy position. Therefore the origin of the transmittance increase in experiment [C] is the decrease of the electron population in the n=l band. We suppose that the initial decay of curve [C] is corresponding to the recovery of the electron population. The electrons transferred by intersubband relaxation to highlying n=l states (see Fig. 2(c)) return to the bottom of the band. The population recovery time is longer than the intersubband relaxation time, because it needs to emit 6-7 LO-phonons to recover the initial Fermi energy (process of Fig. 2(c) => Fig. 2(d)). After a delay time of 500 fs, the dynamics seems to be the same in all the experiments. The relaxation time is of the order of the pico-second. We suppose that it corresponds to the carrier cooling process in the n=l band (process of Fig. 2(d) => Fig. 2(a)). After the initial relaxation process is finished, the electron

296

distribution can be described by a hot Fermi distribution characterized by an elevated electron temperature (Fig. 2(d)). Carrier cooling is the process of decreasing the temperature by LO phonon emissions. However, the intra-band electron-LO-phonon scattering time is calculated to be -10 fs by Suzuki et-al [6], which is much shorter than the observed decay time. We suppose that this difference is caused by reabsorption of emitted LO-phonons, i.e. a hot-phonon effect. In addition we found that the slower component decrease follows a powerlaw decay rather than an exponential decay. We suggest that the power-law decay is caused by LO-phonon momentum space diffusion. \

-

[A] — [B] - - - - [C]

^ ^ 13 cd ^ T ^ ^ •' •

'I'l^W^

B

0

1000 2000 3000 4000 5000

Delaytime (fs) Fig. 1. Temporal trace of the change in the transmittance of [A] (dashed line), [B] (solid line) and [C] (doted line) in semi-log scale.

4.

.cooling ¥ power-law decay ^ ~ps

i/^^^^y ^Oii

Ki^^ -1000

]

(b)

Electron Biergy (lu.)

(d)

V^^JMoj

eH

<1

pump excitation & instantaneous carriercarrier scattering

(c)

liintersubband i relaxation --140 fs

<^

2 Electron Energy (a.u.)

thermalization 350 ~ 400 fs

' Electron Energy (a.u.)

Fig. 2. Schematic of the electron distribution dynamics in the n=l band, (a) Before ISBT excitation, (b) Just after ISBT excitation, the total electron number is decreased. The initial electron distribution before excitation is shown as a dashed line, (c) After intersubband relaxation to higher-energy parts of the n=l band by LO-phonon scattering, (d) Hot Fermi distribution with an elevated temperature after the thermalization process.

Conclusions

We have clarified the ultrafast ISBT relaxation dynamics in GaN/AlN MQW and found that the origin of the slower component is carrier-cooling process influenced by a hot-phonon effect. Consequently the ultrafast performance of ISBT-based switching device is determined by the cooling process.

References 1 2 3 4 5 6

K. Kishino, A. Kikuchi, H. Kanazawa, and T. Tachibana, Appl. Phys. Lett. 8 1 , 1234(2002) N. lizuka, K. Kaneko, and N. Suzuki, Appl. Phys. Lett. 8 1 , 1803 (2002) C. Gmchl, H. M. Ng, G. Chu, and A. Y. Cho, Appl. Phys. Lett. 77,1322 (2000) J. Hamazaki, S. Matsui, H. Kunugita, K. Ema, H. Kanazawa, T. Tachibana, A. Kikuchi, and K. Kishino, Appl. Phys. Lett. 8 4 , 1 1 0 2 (2004) Kishino and A. Kikuchi, phys. Stat. sol. (a) 190, 23 (2002) N. Suzuki and N lizuka, Jpn. L Appl. Phys. Part 2 37, L369 (1998)

297

Polaritonics in complex structures: Confinement, bandgap materials, and coherent control David W. Ward^ Eric R. Statz\ Jaime D. Beers\ T. Feiirer\ John D. Joannopoul Joannopoulos', Ryan M. Roth^, Richard M. Osgood^, Kevin J. Webb"^, and Keith A. Nelson^ ^ Department of Chemistry, Massachusetts Institute of Technology, Cambridge, MA 02139, USA E-mail: [email protected] ^ Center for Materials Science and Engineering, Department of Physics, Massachusetts Institute of Technology, Cambridge, MA 02139, USA ^ Microelectronics Sciences Laboratories, Department of Physics, Columbia University, New York, NY 10027, USA ^ School of Engineering and Computer Engineering, Purdue University, West Lafayette, IN 47907, USA Abstract. We report on the design, fabrication, and testing of ferroelectric patterned materials in the guided-wave and polaritonic regime. We demonstrate their functionality and exploit polariton confinement for ampHfication and coherent control using temporal pulse shaping.

1.

Introduction

Polaritonics is defined in an intermediate regime between electronics and photonics, roughly in the band above 100 GHz and below 10 THz. An ultrafast optical pulse focused into a ferroelectric crystal hke LiNb03 (LN) or LiTaOa (LT) generates phonon-polaritons, admixtures of electromagnetic and TO phonon responses henceforth called polaritons, in this frequency range through impulsive stimulated Raman scattering (ISRS). The polariton response is a traveling wave, so its propagation through the host crystal can be controlled using the techniques of guided wave, diffractive, and dispersive optics [1]. This THz generation and guidance scheme offers a fully integrated platform that supports generation and manipulation in a single patterned material. Here we extend the polaritonics platform to include single and coupled resonance structures. The resonant modes within the structures are illustrated and exploited for coherent control and ampHfication through temporal pulse shaping. We also characterize a planar waveguide from which polariton signal images can be recorded even though the sample thickness is only 10 |Lim.

298

Polaritonfo i d n d g a p Cylfudrlcai Material Resonators

Sliver Enhanced Resonators

Mode Convertor

Fig. 1. Femtosecond laser machined structures including a 1-D polaritonic bandgap material, 2-D cylindrical resonators, silver coated resonators, and a mode converter structure.

2. Results and Discussion Patterned materials fabricated in our group include resonators, waveguides, photonic bandgap materials, mode converters, as well as THz integrated platforms such as Mach-Zehnder interferometers, prisms, gratings, and waveguide couplers. Son^e of these structures are illustrated in Fig. 1. In the 'polaritonic' bandgap regime, when the bandgap due to the spatial period overlaps spectrally with the intrinsic polariton bandgap, extreme localization of the electromagnetic energy and frequency dependent relocation may result [2]. Aperiodic structures also offer a means of sensitive frequency-dependent mode selection [3]. To demonstrate the planar waveguide dispersion properties of a 10 jLim LN film made with crystal ion sHcing [4], we have generated narrowband polaritons through ISRS using crossed beam excitation and monitored their temporal and spatial evolution using polariton imaging [5]. Deviations of the dispersion relation from that in bulk occur when the polariton wavelength is longer than the slab thickness. Fig. 2. <^) . - l O u f H

> - 56 \\\\\

f-^1

b)7

urn

1 ' "" "TW(Fh"

'

m ..

^!i^:!^':

10

Jim Film

20 30 40 VVavetength (jim)

50

60

Fig. 2. Narrowband polaritons are generated using crossed excitation beams in 10 |im thick LN, (film shown in part c). Their spatial and temporal evolution are monitored simultaneously by polariton imaging, and the polariton dispersion curve (b) results from extracting frequency and wavevector information from each scan.

299

a) I

b),'

>, 320

Q 200? 2:0 40 TImeCps)

60

80

100

40Q

600

800

F r e q u e n c y (GHz)

Fig. 3. Illustration of selective driving of a THz resonator in the (a) time and (b) frequency domain. Polariton signals are strongest and longest in persistence when the cavity modes are driven on resonance (240 Ghz and 320 GHz). Coherent control over polaritons ordinarily requires spatial as well as temporal pulse shaping in order to enable manipulation following propagation away from the excitation region. Confinement within a resonator, however, causes polaritons to return repeatedly to the excitation region where they may be manipulated by successive excitation pulses that arrive there with specified timing. We have recently reported coherent control in a THz resonator and amplification by a factor of ten in energy [6,7]. This is illustrated in Fig. 3.

3.

Conclusions

In conclusion, we have extended the polaritonics platform to include single and coupled resonance structures and planar waveguides. The characteristic polariton responses in these structures have been characterized, and resonance has been exploited for coherent control by properly timed pulse sequences. Acknowledgements. This work was supported in part by the Army Research Office (grant no. DAAD10-01-1-0674), and the National Science Foundation (grant no. CHE-0212375).

References 1. 2. 3. 4. 5. 6. 7.

300

N.S. Stoyanov, D.W. Ward, T. Feurer, and K.A. Nelson, Nature Materials 1, 95-98 (2002). K.C. Huang, P. Bienstman, J.D. Joannopoulos, K.A. Nelson, and S. Fan, Phys. Rev. Lett. 90,196402(2003). M.C. Yang, J.H. Li, and K.J. Webb, App. Phys. Lett. 83, 2736-2738 (2003). M. Levy, R.M. Osgood, Jr., R. Liu, E. Cross, G.S. Cargill III, A. Kumar and H. Bakhru, Appl. Phys. Lett. 73, 2293-2295 (1998). D.W. Ward, E.R. Statz, R.M. Roth, R.M. Osgood, and K.A. Nelson, Submitted to Appl. Phys. Lett. N.S. Stoyanov, T. Feurer, D.W. Ward, E.R. Statz, and K.A. Nelson, Opt. Express, 12, 2387-2396 (2004). D.W. Ward, J.D. Beers, T. Feurer, N.S. Stoyanov, and K.A. Nelson, Submitted to Optics Letters.

Detection of four-wave mixing signal from single layer quantum dots Michio Ikezawa, Fumitaka Suto, Yasuaki Masumoto, and Hong-Wen Ren* Institute of Tsukuba, University of Tsukuba, Tsukuba 305-8571, Japan E-mail: [email protected] Abstract. The authors study four-wave mixing of two kinds of single layer quantum dots, that is strain-induced GaAs quantum dots and InP quantum dots equipped with a transparent electrode, by using highly-sensitive heterodyne method. A pronounced beat due to biexcitons is observed in GaAs quantum dots, and a quite strong bias dependence of the four-wave mixing signal is observed in InP quantum dots.

1.

Introduction

In recent years, semiconductor quantum dots (QDs) attract a considerable interest, and various kinds of high quality self-assembled QDs are fabricated. Although a large number of studies have been made on the self-assembled QDs, little is known about their coherent nature because of lack of experimental data in time domain like a four-w^ave mixing (FWM). Since FWM measurement in self-assembled QDs is rather difficult due to small number density of QDs, this technique is not widely used yet. Recently, we developed a highly-sensitive heterodyne system purposing the detection of FWM signal from single layer QDs. In this paper, we will present the results of degenerate FWM of two kinds of self-assembled single layer QDs, that is GaAs strain-induced dots (SID) and InP QDs.

2.

Experimental Methods

Our detection system is based on the works of [1,2], but it is modified to allow FWM measurement in reflection geometry for the application to opaque substrates. The excitation laser is a mode-locked Ti:Sapphire oscillator which operates at 80 MHz. Typical output pulse width is 80 fs. The output beam is divided into three beams named Pump, Probe and Reference. Probe and Reference were frequencyshifted by acousto optic modulators to (v + llOMHz) and (v + lllMHz), respectively, where v is the optical frequency of Pump beam. Very weak FWM signal whose frequency is 2 (v + 1 lOMHz) - v = v + 220MHz induced by Pump and Probe pulse is overlapped with strong Reference (v + lllMHz) beam by a half mirror, and detected by a PIN-photodiode. Only the interference component (v +220 MHz) - (v +111 MHz) =109 MHz is selected by a spectrum analyzer. Probe beam is intensity modulated by a mechanical chopper and output signal from the spectrum analyzer is accumulated by a lock-in amplifier. This system allows us to measure FWM signal even from single layer QDs.

301

3.

Results and Discussion

SID is formed in a single quantum well (QW) by introducing additional lateral confmement in the plane of the QW by stressors grown on the top surface of the QW [3]. In this kind of QDs, almost parabolic confmement potential is realized, providing unique properties such as equally-spaced energy levels. However, quite high sensitivity is indispensable for FWM measurement in SID, because it is difficult to stack a lot of SID layers to increase the number density due to the relaxation of the strain field. The sample used here is a GaAs-SID formed in 3.8 nm GaAs single QW by InP stressors whose diameter is about 90 nm. The areal density of the stressor is 3*10^ cm'^.

Time Delay (ps)

Time Delay (ps)

Fig. 1. (a) Time-integrated FWM signal from GaAs-QW and GaAs-SID. Inset shows a PL spectrum of the sample at 2 K. (b) Initial part of FWM signal from SID. Biexcitonic beat with beating period of 1 ps is clearly observed, (c) Initial part of FWM signal from QW. The photoluminescence spectrum of the sample at 2 K is shown in the inset of Fig.l (a). There are two peaks at 753 nm and at 775 nm, which come from GaAs QW and GaAs SID, respectively. The center wavelength of the excitation laser in the FWM measurements was tuned for these peaks. FWM signals obtained by our system are shown in Fig. 1(a). The signal-to-noise ratio of the curve marked by SID is quite high considering that it comes from single layer QDs. The decay curve consists of a single exponential decay with a decay time of 12 ps and a clear damped oscillation with a period of 1 ps when two excitation pulses have co-linear polarization. The beat is explained by the exciton-biexciton beat in SID. This assignment was confirmed by measuring polarization dependence of the beat signal. When SID was excited by two pulses of the same circular polarization, the beat disappeared as shown by dots in Fig. 1(b). This fact strongly supports our assignment, because biexciton should be excited by two photon of the opposite circular polarization. The similar oscillating signal was observed if we tune the excitation laser for QW (see Fig.l (a)), but the beating period was 1.5 times longer than that of SID (see Fig. 1(c)). From these results, we can conclude that the

302

biexciton binding energy is enhanced 1.5 times by the additional two-dimensional lateral confinement in the SID.

J

.

1

,

\

,

20

L_

40

Time Delay (ps) Fig. 2. FWM signal from self assembled InP QDs at various bias voltages. Another example of single layer QDs is self-assembled InP QDs. They are grown on an n^-GaAs substrate with an areal density of 10^^cm"^. It is known that InP QDs studied here are non-intentionally n-doped and that the InP QDs can be neutralized under the negative electric bias [4]. The FWM signal from the InP QDs showed two dramatic changes depending on the electric bias. When the electric bias is changed from 0 V to -1.0 V, the InP QDs are changed to be neutral and the FWM signal increases by 2 orders of magnitudes. When the negative bias voltage is increased frirther from -1.0 V to -2.5 V, the decay time of the FWM signal decreases drastically as shown in Fig. 2. This bias dependence may be explained by the change of the hole tunneling rate. We note that the decay profile of the FWM is not a single exponential but Gaussian-like function, indicating nonMarkovian dephasing process [5]. The details will be published elsewhere.

References * Present address: Applied Optronics Incorporation, 13111 Jess Pirtle Boulevard, Sugar Land, Texas 77478, USA 1 M. Hofmann, S. D. Brorson, and J. M0rk, Appl. Phys. Lett. 68, 3236, 1996. 2 P. Borri, W. Langbein, S. Schneider, and U. Woggon, Phys. Rev. Lett. 87, 157401, 2001. 3 H. Lipsanen, M. Sopanen, and J. Ahopelto, Phys. Rev. B 51, 13868, 1995. 4 I. E. Kozin, V.G. Davydov, I.V. Ignatiev, AV. Kavokin, K.V. Kavokin, G. Malpuech, H.W. Ren, M. Sugisaki, S. Sugou, Y. Masumoto, Phys. Rev. B 65, 241312,2002. 5 M. Aihara, Phys. Rev. B 25, 53, 1982.

303

Wavepacket Interferometry and Wavepacket Dynamics in Condensed Phase Matias Bargheer\ Mizuho Fushitani^, Markus Giihr^, and Nikolaus Schwentner^ ^ Max-Bom Institut, Max-Bom-Str. 2a, 12489 Berlin, Germany E-mail: [email protected] ^ Experimentalphysik, Freie Universitat Berlin, Amimallee 14, 14195 Berlin, Germany Abstract. The diatomic CI2 in solid Ar is examined to discuss the information obtained by wavepacket interferometry in condensed phase and to contrast it with femtosecond pumpprobe and linear absorption spectra.

1.

Introduction

Scherer et al. used femtosecond phase-locked pulse pairs (PLPP) to coherently excite vibrational w^avepackets in electronically excited states of the free I2 molecule. [1] In a second paper the exact shape of the experimental signal could be simulated w^ith great accuracy, accounting for rotation, vibrational temperature etc. [2]. The shape of this fluorescence interferogram v^as derived by a density matrix description of linear response theory. The extensions of the method to systems with dephasing processes is sketched already in the end of the second article, however, it was never exploited for wavepacket interferometry in dissipative systems. The PLPP have been successfully applied to atoms, in photoemission from metal surfaces, excitons in semiconductors and molecular crystals. The present experiment is the first clear example of transferring the technique of PLPP to the ultrafast vibration of molecules in a dissipative environment with strong coupling evidenced by phonon sidebands. The linear excitation spectrum [3] of CyAr (Fig. 1) shows zero phonon lines, as well as phonon sidebands.

2. Experimental Methods A 50 fs pulse tuned to the B-state resonance of CI2 at 515 nm is split up in a Michelson interferometer to produce a pair of collinear, phase-locked pulses. The pulse pair excites a thin film of Cl2/Ar solid, kept at a temperature of 5 K, and the laser induced fluorescence (LIF) from the excited state is measured versus time delay and relative phase. The first pulse prepares a coherent superposition of the molecule in its ground and excited state. The second pulse interacts with the same molecule in the ground state again, preparing a second excited-state wavepacket which may interfere constructively or destructively with the first packet. As a consequence, the population in the exited state can be coherently controlled.

304

a)_

E o 0.5

m

piniltiniiii T = 260 fs

/^ 525

520

515

510

?./nm

Fig. 1. Linear excitation spectrum of CyAr (solid line, according to [3]) with sharp zero phonon lines and phonon sidebands. The dotted line indicates the spectral interference of the two phase-locked pulses delayed by t == 260 fs with a relative phase leading to constructive interference of the molecular wavepacket.

' ' ' / \ 'b)A . \ r = 260 fs^ ^/

0.4

0 X) 0.5

t/ps 1.0

1.5

^ :

f^^ww.^^'^'^vv

100

200 t/ps

300

400

Fig. 2. a) Wavepacket interference as measured by LIF (open circles) and calculated using spectrale interferences (solid circles), b) Pump-probe signal. For explanation see text.

The LIF is proportional to the excited state population after the interaction with both light fields. Fig. 2a shows the LIF measured after excitation of the system with a phase locked pulse pair (PLPP). At each time delay the LIF is measured for 16 different phase values (open circles). In addition, the spectral interferences of the PLPP are recorded on a fiber-optic spectrometer, and the result for a selected phase at a pulse-delay of 260 fs is plotted in Fig. 1. The method bears a close similarity to the COIN method. [4,5]

3.

Results and Discussion

The envelope of the open circles in Fig. 2a shows a wavepacket recurrence at 260 fs, which corresponds to the round-trip time of the CI2 wavepacket excited at 515 nm. For a simple and intuitive argument consider the frequency domain. The spacing of the zero phonon lines in Fig. 1 corresponds to the vibrational round trip time T=260 fs of the CI2 wavepacket. For constructive interference the maxima of the interference pattern (dotted line in Fig. 1) observed in the spectrograph match the zero-phonon lines. The destructive interference at 260 fs is observed for the phase value, which produces spectral interferences with minima at the zero phonon lines (not shown). For time delays different from the round trip time T = 260 fs, the spectral interferences do not coincide with the peaks of the linear absorption spectrum, and thus the modulation of the LIF signal as a function of the phase is smaller. The solid circles in Fig. 2a indicate the LIF calculated from a multiplication of the absorption spectrum (Fig. 1) with the measured spectral interferences for each time delay and each phase. The recurrence at T=260 fs in this signal corroborates the simple argument given above.

305

Analysing the result of excitation by two phase locked pulses in this case, it can be stated that the constructive interference for the zero phonon lines enhances excitation of freely oscillating molecules and suppresses excitation of molecules coupled to the phonon bath. If the minima of the spectral interference coincide with the zero phonon lines, we preferentially excite molecules coupled dissipatively to the phonon bath. From a theoretical viewpoint, the relevant processes are described in the framework of linear response theory. The wavepacket interferograms can be exploited to dissect broad and overlapping contributions to linear absorption spectra. If the width of a contributing line corresponds to a decay time shorter than the vibrational period, it will be damped out and invisible in the PLPP spectrum. Thus broad lines (such as phonon sidebands) only contribute to the initial fast decay and buried narrower lines can be seen in the oscillation. To contrast this peculiar method of wavepacket interferometry with a better known technique, we show in Fig. 2b the wavepacket dynamics recorded in a pump-probe experiment with ?tpump= 520 nm and ^^probe"^ 283 nm. Here the pump pulse excites the same wavepacket dynamics of Cl2/Ar in the electronic B-state as they are observed in Fig 2a. However, now the wavepacket population is probed by projecting the wavepacket into the ion-pair manifold (E-state) with the timedelayed probe pulse and recording the corresponding LIF. The first round-trip consistently shows the 260 fs period. Then the oscillations speed up to 200 fs due to the anharmonicity, when the wavepacket quickly relaxes to lower vibrational levels. This method is also detecting the wavepacket irrespective of relaxation and dephasing, as long as its energy is above the probe window.

4.

Conclusions

We show coherent wave packet interferometry and classical pump-probe data for a conceptionally simple model system as an educative example of ultrafast spectroscopies of molecules in condensed phase. The concepts will be used in pulse-shaping experiments controlling molecular dynamics in dissipative environments. Acknowledgements. Founding by the German Science Foundation via SfB450.

References 1 N.F. Scherer, R.J. Carlson, A. Matro, M. Du, A.J. Ruggiero, V. Romero-Rochin, J.A. Cina, G.R. Fleming and SA. Rice, J. Chem. Phys., 95,1487-1511 (1991) 2 N.F. Scherer, A. Matro, L.D. Ziegler, R.J. Carlson, J.A. Cina and G.R. Fleming, J. Chem. Phys., 96,4180-4194 (1991). 3 V. E. Bondybey and C. Fletcher,, J. Chem. Phys., 64, 3615-3620 (1976). 4 O. Kinrot, I. Sh. Averbukh and Y. Prior, Phys. Rev. Lett., 75, 3822-3825 (1995) 5 A. Tortschanoff, K. Brunner, Ch. Warmuth, and H.F. Kauffmann, J. Chem. Phys., 110,4493-4504(1999)

306

Femtosecond wavepacket dynamics of potassium adsorbate on Pt(lll) Kazuya Watanabe^ Noriaki Takagi^ and Yoshiyasu Matsumoto^'^ ^ Department of Photoscience, School of Advanced Sciences, The Graduate University for Advanced Studies (SOKENDAI), Hayama, Kanagawa 240-0193, Japan ^ Institute for Molecular Science, Okazaki, Aichi 444-8585, Japan E-mail: [email protected] Abstract. Femtosecond wavepacket dynamics of potassium adsorbed on Pt(lll) is investigated by femtosecond time-resolved second harmonic generation under an ultra-high vacuum condition. Detailed potassium coverage dependence is explored.

1.

Introduction

To understand the mechanism of surface photochemistry, it is vital to know^ hov^ photoinduced electronic excitation couples to adsorbate nuclear motions that ultimately lead to the chemical transformations. In spite of the significant progress in understanding ultrafast processes at metal surfaces, direct time domain observations of primary photochemical events are still scarce. Recently, it WSLS shown that the time-resolved second harmonic generation (TRSHG) is a powerful tool to investigate the dynamics of vibrational coherence at solid surfaces [1-3]. We have applied this technique to cesium atoms adsorbed on Pt(lll), and successfully observed the vibrational coherence of the Cs-Pt stretching mode in time domain [4,5]. In this paper, we extended this approach to potassium adsorbate on Pt(l 11). We explored a detailed potassium coverage (0) dependence of TRSHG signals and found that the oscillatory features are strongly affected by a slight change in the potassium coverage.

2.

Experimental Methods

The experiments were carried out in an ultrahigh vacuum chamber [4]. The clean Pt(lll) at 110 K was dosed with potassium atoms from a well-degassed alkali dispenser (SAES getters). The alkali coverage was determined by Auger electron spectroscopy (1 ML = 1.5 x 10^^ cm"^). The apparatus for the TRSHG measurement is basically the same as previously reported [4], except for that we used a home-built noncollinear optical parametric amplifier (NOPA) as a light source. In this work, the previously developed Tiisapphire based NOPA [6] (1 kHz repetition rate) is modified to have two independent outputs by splitting the 400-nm OPA pump beam into two and pumping two BBO crystals independently. In this work, we performed pump and probe measurements by fixing the center

307

wavelengths of the two NOP A outputs at 580 nm. The p-polarized pump and probe beams were focused on a sample at an incidence angle of -70°. The ppolarized component of the SH of the probe pulse was detected by a photomultiplier. The intensity modulation of the probe SH induced by the pump pulse was detected by a lock-in amplifier as a function of the pump-probe delay, t. From the cross correlation profile of the pump and probe pulses at the sample surface, the pulse durations were estimated to be 25 fs. The transient change of the SH intensity, ASH(0, is defined as A SH(0 = (SH(0-SHs)/SHs, where SH(0 and SHs is the SH intensity with and without the pump pulse, respectively. Sample temperature was kept at 110 K during the measurements.

3.

Results and Discussion

Fig. 1(a) shows TRSHG traces from potassium-covered Pt(l 11) as a function of the potassium coverage. At 9 > 0.24 ML, clear oscillatory signals are observed. The time domain data can be fitted well by the linear combinations of damped sinusoids. Fig. 1(b) shows the Fourier spectra of the oscillatory part in the time domain data in Fig. 1(a), obtained after subtracting low frequency (< 1 THz) background components. The spectra show strong peaks at 4.5 ~ 5.2 THz (150 -173 cm"^), depending upon the potassium coverage. These frequencies are consistent with those for a K-Pt stretching mode observed by a high-resolution electron energy loss spectroscopy (HREELS) [7]. Thus, the oscillatory modulation of the SH intensity is ascribed to the nuclear wavepacket motion of K atoms on Pt(l 11) as in the case of Cs/Pt(l 11) [4]. At 0.10 ML O < 0.36 ML, potassium is known to form several ordered structures depending on the coverage [8]. At 0.20 < 6 < 0.24 ML, a 2x2 structure is formed, which is replaced by a (V3xV3)R30° domain with further increasing the coverage. At 9 ~ 0.33 ML, the (V3xV3)R30° structure dominates. We tentatively assign the peaks at 5.2 THz and at 4.5 THz for 0.24 < 9 < 0.31 ML to the K-Pt stretching modes in the 2x2 domain and to those in the (V3xV3)R30° domain, respectively. I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

(a)

HWAA-

5:^^^ 0.31 ML 0.28 ML

B

0.24 ML

< \o

I I I I I I I I I I I I I • I I I I I I • I I

0

1 2 3 Delay time (ps)

4

1 2 3 4 5 6 7 8 9 Frequency (THz)

10

Fig. 1. (a) TRSHG traces taken from K-covered Pt(lll) surfaces at various K coverages and (b) Fourier spectra of their oscillatory components.

308

At 0 > 0.35 ML, a strong peak appears at 4.7 THz and weak peaks are found at 2.6, 3.7, and 6.2 THz. In this coverage range, a pottassium overlayer is continuously compressed, resulting in rotational epitaxial or aligned structure [8]. The 3.8 and 4.7 THz components are tentatively assigned to the K-Pt stretching modes in this multiphase region. The 2.6 THz component is too low to be ascribed to the K-Pt stretching mode. The frequency is very close to that of the surface Rayleigh phonon mode on a clean Pt surface at K zone boundary point (2.7 THz) [9], and would be ascribed to the Pt Rayleigh mode. The oscillatory signal appears at 6 > 0.20 ML, and its amplitude is enhanced at 0 ~ 0.30 ML. Thus the modulation amplitude is not proportional to the potassium coverage. In general, with the alkali metal adsorption, the energy levels of the image states and alkali induced unoccupied state are lowered with respect to the Fermi level [10]. As for K/Pt(lll), the employed photon energy (2.1 eV) would be close to the electronic resonance between the potassium induced surface electronic states at 0 - 0.3 ML, so that the vibrational coherence in the electronic ground state is induced by the resonance enhanced impulsive Raman scattering. In Fig. 1(a), the observed K-Pt wavepacket motion typically possesses the dephasing times of ~ 1 ps, which is much longer than the known lifetime of the electronic excitation of the alkali adsorbates [11]. Thus, the contribution of the nuclear wavepacket motion in the electronic excited states is not significant under the current condition. Acknowledgements. This work was supported in part by Grants-in-Aid for Young Scientists (A) (14703009), Scientific Research for Priority Area (417), Creative Scientific Research Collaboratory on Electron Correlation-Toward a New Research Network between Physics and Chemistry (13NP0201) from MEXT of Japan, and KAKENHI (14340176) from JSPS.

References 1 Y. M. Chang, L. Xu, and H. W. K. Tom, Phys. Rev. Lett. 78, 4649, 1997. 2 Y. M. Chang, L. Xu, and H. W. K. Tom, Chem. Phys. 251, 283, 2000. 3 K. Watanabe, D. T. Dimitrov, N. Takagi, and Y. Matsumoto, Phys. Rev. B 65, 235328, 2002. 4 K. Watanabe, N. Takagi, and Y. Matsumoto, Chem. Phys. Lett. 366, 606, 2002. 5 K. Watanabe, N. Takagi, and Y. Matsumoto, Phys. Rev. Lett. 92, 057401, 2004. 6 D. Ino, K. Watanabe, N. Takagi, and Y. Matsumoto, Chem. Phys. Lett. 383, 261, 2004. 7 C. Kliinker, C. Steimer, J. B. Hannon, M. Giesen, and H. Ibach, Surf Sci. 420, 25, 1999. 8 G. Pirug and H. P. Bonzel, Surf Sci. 194, 159, 1988. 9 U. Harten, J. P. Toennies, C W611, and G. Zhang, Phys. Rev. Lett. 55, 2308, 1985. 10 N. Fisher, S. Schuppler, R. Fischer, Th. Fauster, and W. Steinmann, Phys. Rev. B 47,4705,1993. 11 H. Petek, M. J. Weida, H. Nagano, and S. Ogawa, Science 288, 1402, 2000.

309

Control of tunnel ionization in molecules by intense femtosecond laser pulses with timedependent polarization Tsuneto Kanai, Shinichirou Minemoto, and Hirofumi Sakai Department of Physics, Graduate School of Science, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033, Japan E-mail: [email protected] Abstract. We propose a new way of controlling tunnel ionization in molecules by timedependent polarization pulses. The model consists of the successful combination of the molecular ADK theory reflecting the symmetry of molecular orbitals and Landau-Zener resonances. Tunnel ionization of molecules with an intense laser field is a fundamental process in many interesting phenomena such as production of multiply-charged molecular ions [1,2] and high-order harmonic generation from molecules [3] and therefore has been extensively investigated over decades. The recent progress in molecular alignment [4] and orientation [5] has opened a way to study the correlation between the molecular axis and the laser polarization in above-mentioned phenomena. Combining another new technique of generating and controlling timedependent polarization pulses, we have succeeded in optimally controlling multiphoton ionization processes in aligned I2 molecules [6]. The results suggest the existence of an unknown tunnel ionization mechanism which is characteristic of time-dependent polarization pulses and inspire us to develop a new theoretical model to predict tunnel ionization probabilities in molecules with time-dependent polarization pulses. The so-called ADK [7] and KFR [8] theories successfully predict the ionization rates of atoms. Unfortunately, we cannot apply these theories directly to molecules. In fact, these theories fail to explain suppressed ionization observed for O2 molecules [9]. This puzzling observation was recently explained by the generalized many-body S-matrix theory [10] and modified ADK theory (molecular ADK [11]), both of which take account of the symmetry of valence orbitals. Since these theories do not cover dynamical effects such as enhanced ionization [12], ionization rates of molecules in time-dependent polarization pulses cannot be calculated. In this paper, we successfully combine the molecular ADK theory and Landau-Zener transitions in an intense femtosecond laser field. Using our model, we demonstrate that ionization probabilities of H2 molecules are controlled with time-dependent polarization pulses. Ionization of atoms and molecules in an intense laser field is a transition from quasi-bound states to continuum states. When the ponderomotive energy Up is large enough, the continuum states can be treated as free states and we expand a time-dependent wave function as,

310

| ^ ( ^ ) ) = ^ a , ( t ) | r ) + /d^.;6(^^0h.) i=o

(1)

J

where fl/(r)'s are the amplitudes of the quasi-bound states and biy, ^)'s are those of the corresponding free states. For « = 0, only the ground state is considered as the quasi-bound state. This assumption is used in the Lewenstein model [13], which is known as a standard model of high-order harmonic generation. The Lewenstein model also derives ionization rates predicted by the ADK model. To express ionization rates from a molecular orbital, instead of free states, we introduce the non-Hermitian Hamiltonian Hdecay? which gives angular dependent ionization rates. In the real molecules, there are more than two electronic states that contribute to the dynamics of the electrons, which lead to "enhanced ionization", as investigated by both theoretically [14,15] and experimentally [12]. In order to describe the anisotropic ionization processes, we adopt the molecular orbital method and calculate the population of each electronic state by taking account of the level crossing between the quasi-bound states. Following the molecular ADK theory, the ionization rate of the electronic state in the electric field F is given by

where Z^ is the effective Coulomb charge, /r is a parameter determined by the ionization potential, w ' is a magnetic quantum number along the molecular axis, and B{m \ R) is a constants that is determined by the shape of the valence orbital of molecules and Euler angles R between the molecular axis and the field direction. Note that B{m \ R) reflects the symmetry of the molecular orbital and therefore gives angular-dependent ionization rates. We calculate the ionization probability as a function of time by integrating the ionization rate Wstati^^ R) during the pulse. Here, polarization of the laser pulse affects the anisotropic ionization rates and also the dynamic effects including quantum interference during the repetitive Landau-Zener transitions. Figure 1 shows the ionization probability of H2 by a 7-fs Gaussian pulse with a peak intensity of 2 x lO^'* W/cm^ as a function of time during the pulse. The ground X^2g and excited B^2u states are considered in the calculation. When the polarization is circular [Fig. 1(a)], the ionization probability is minimum. In the case of a lineally-polarized pulses with the orientation angle aligned along the molecular axis [(b)], the probability is much larger than the circular case as predicted by the molecular ADK theory. By appropriately modulating the polarization during the pulse [Fig. 1 (c)], the ionization probability can be further enhanced. The enhancement is achieved by the following processes. First, the excited state is effectively populated through repetitive Landau-Zener transitions. At around the peak of the pulse, the population can be increased by taking advantage of the quantum interference among excitation paths of the LandauZener transitions. Then, when the population of the excited state is maximized, the orientation angle of the pulse is suddenly changed to a direction along which the excited molecules are favorably ionized. By this pulse, we achieve the ionization probability of 0.7, which is 1.4 times as high as that obtained with a linearlypolarized pulse.

311

Intensity Profile

K,7fs /=2x10'*W/cm!'

(c) (b)

(a)

Time (fs)

Fig. 1. Ionization probabilities of H2 in a (a) circular, (b) linear, and (c) time-dependent polarization pulse with a peak intensity of 2 x 10^"^ W/cm^.

In summary, we have developed a new model to calculate tunnel ionization probabilities of molecules with time-dependent polarization pulses. The model is based on the combination of the molecular ADK theory and Landau-Zener transitions. When the laser intensity is strong enough, the time-dependent laser field mixes the electronic states of molecules and induces repetitive Landau-Zener resonances. Due to the quantum interference among the excitation paths in the resonances, the excited state is effectively populated and the ionization rate is enhanced. Furthermore, by using the anisotropy in ionization rates, one can control ionization probability by appropriately designed time-dependent polarization pulses. Shaping time-dependent polarization pulses is feasible with the state-of-the art femtosecond laser technologies [6].

References

9 10 11 12 13 14 15

312

H. Sakai et al., Phys. Rev. Lett. 81, 2217,1998. H. Sakai et al., Phys. Rev. A 67, 063404, 2003. H. Sakai and K. Miyazaki, Appl. Phys. B, 61, 493, 1995. H. Sakai etal., J. Chem. Phys. 110, 10235,1999. H. Sakai et al., Phys. Rev. Lett. 90, 083001, 2003. T. Suzuki et al, Phys. Rev. Lett. 92, 133005, 2004. M. V. Ammosov, N. B. Delone, and V. P. Krainov, Sov. Phys. JETP 64, 1191, 1986. L. V. Keldysh, Sov. Phys. JETP 20, 1307, 1965; F. H. M. Faisal, J. Phys. B 6, L89,1973; H. R. Reiss, Phys. Rev. A 22, 1786, 1980. J. Muth-Bohm, A. Becker, and F. H. M. Faisal, Phys. Rev. Lett. 85, 2280, 2000. A. Talebpour, C.-Y. Chien, and S. L Chin, J. Phys. B 29, L677, 1996. X. M. Tong, Z. X. Zhao, and C. D. Lin, Phys. Rev. A 66, 033402, 2002. E. Constant, H. Stapelfeldt, and P. B. Corkum, Phys. Rev. Lett. 76, 4140, 1996. M. Lewenstein et al, Phys. Rev. A 49, 2117, 1994. K. Harumiya et al, Phys. Rev. A 66, 043403, 2002. T. Seidemann, M.Y. Ivanov, and P. B. Corkum, Phys. Rev. Lett. 75, 2819, 1995.

Ultrafast Mid-Infrared Dynamics in the Colossal Magnetoresistance Pyrochlore TliMnjO? R. P. Prasankumar^ A. J. Taylor\ R. D. Averitt\ H. Okamura^ H. Imai^ Y. Shimakawa^, and Y. Kubo^ ^MST-10, Condensed Matter and Thermal Physics, Los Alamos National Laboratory; Los Alamos, NM 87545 Tel: 505-665-2993, Fax: 505-665-7652 eniail: rpprasan@laiil. gov ^Graduate School of Science and Technology, Kobe University, Kobe 657-8501, Japan ^Fundamental Research Laboratories, NEC Corporation, Tsukuba 305-8501, Japan Abstract: An optical pump, mid-infrared (IR) probe system is used to investigate ultrafast temperature-dependent dynamics of the colossal magnetoresistance pyrochlore Tl2Mn207. The dynamics change appreciably near the Curie temperature (Tc), indicating a dependence on ferromagnetic ordering.

Colossal magnetoresistance (CMR) materials, in which the electrical resistance changes drastically with application of a magnetic field near the Curie temperature, have received a great deal of attention in recent years, both due to fundamental interest in this phenomenon as well as potential applications in areas such as magnetic sensing and recording. The best known example of these materials are the ferromagnetic perovskite manganites such as Lai-xSrxMnOs [1], in which the CMR mechanism likely originatesfi*omthe combined effects of double exchange and Jalm-Teller distortions. Ultrafast optical spectroscopy has contributed to an understanding of the manganites by differentiating between the relative contributions of the lattice and spin degrees of freedom [2]. The pyrochlore TlsMusO? also exhibits CMR below Tc~120 K, with ferromagnetic ordering due to Mn quasi-localized spins, but the CMR mechanism is expected to be very different since double exchange and Jahn-Teller effects are negligible in these compounds due to tlie low carrier density (0.7-5x10^^ cm'^) and absence of the Jahn-Teller ion Mn^^ [3]. Optical conductivity measurements revealed a transition fi-om an insulator-like to a metallic structure as the temperature was tuned below T^ with a corresponding increase in the carrier density and change of the reflectivity with temperature near the plasma edge (-0.1 eV-see inset to Figure 1(a)) [4]. In this work, we use an optical pump, mid-IR probe system to investigate ultrafast dynamics in Tl2Mn207 near the plasma edge of the reflectivity as afimctionof temperature. We observe distinct changes as the temperature is tuned through Tc, in particular a large increase in the peak amplitude and offset of the signal, corresponding to an increase in the fi:actional carrier density near Tc that may be attributed to photoinduced enhancement of ferromagnetic ordering.

313

|0..5.

II H i i i n I I I n i l Mil II n i l 11 I'm m i I I I I I I I I ! 200 300 400 500 Time (ps)

(b)

Fig. 1. Temperature-dependent measurements on Tl2Mn207 with an 800 nm pump and 10 Jim probe, (a) Dynamics for TTc. The output of a 1 kHz regenerative amplifier producing 2 mJ, 60 fs pulses at 800 nm is split into two beams to excite the sample and pump an optical parametric amplifier (TOPAS, Light Conversion Systems). The si^ial and idler beams from the TOPAS are mixed in a nonlinear crystal to generate die tunable 320 jxm mid-IR probe. The sample preparation is described in reference [3]. The sample was pumped in reflection at 800 nm (1.55 eV) with a fluence of 460 jiJ/cm^, exciting a carrier density of 2.65x10^^ cm"^. The pump only excites electrons in the minority spin band (the energy gap in the majority spin band is ~2 eV), which are believed responsible for CMR in Tl2Mn207. Temperaturedependent measurements with a 10 ^un (0.125 eV) probe are shown in Figure 1.

Peak signal

100 160 200 Temperature (degrees Kelvin)

250

Fig. 2. Peak amplitude and offset from zero of the pump-probe data sho\^Ti ia figure 1 as a ftinction of tempemture. There is a large reflectivity enhancement near Tc, indicating that a greater fraction of the photoexcited carriers contribute to the charge transport (i.e., Dmde response). The peak AR occurs below Tc due to heating effects at this pump fluence. The initial change in reflectivity at all temperatures is due to a shift of the plasma edge resulting from the large excited carrier density; measurements performed at other pump fluences (not shown) as well as simulations based on a Drude model for the reflectivity support this argument The fast initial relaxation in AR/R is due to a decrease in carriers that contribute to the Drude-like response. However, near T^ the peak amplitude and offset from zero of the signal at long time dela3^ increase significantly. This dependence is shown in Figure 2. This suggests that the photoexcited carriers interact with the quasi-localized Mn moments, possibly enhancing the alignment of Mn spins. 314

This can be seen more clearly in Figure 3, where we have plotted the photoinduced carrier density at long tunes and compared it to the maximum possible change in carrier densit>^ (given by n(lO K)-^(T), with n the carrier density extrapolated from the data in ref. [4]). At low temperatures the magnetization, M, is saturated; therefore the excess carrier density cannot enhance the aUgnment of Mn spins and the carriers rapidly decay. Above Tc, thermal fluctuations prevent the excess carriers from interacting with the Mn moments, and again the carriers rapidly relax, shifting the plasma edge back towards its equilibrium value. In the vicinity of Tc, however, it appears that a fraction of the initially excited carrier density enhances the ferromagnetic ordering.

-r 100

150

200

Temperature (degrees Kelvin)

Fig. 3. Photo-induced carrier density at long time delays compared with the maximum possible carrier density change extrapolated from the data in ref [4]. The negative values at low temperatures (<80 K) in the experimental plot are due to lattice heating effects, and the magmtude is multiplied by 28 for comparison with the calculated plot.

In conclusion, we have investigated the ultrafast quasiparticle dynamics of the pyrochlore TlsMnsO? with an optical pump, mid-IR probe system. We find that with 800 nm excitation, a large increase in the peak amplitude and offset of the signal is observed near Tc, corresponding to a long-lived carrier density increase which may be due to reduced spin scattering resulting from dynamically enhanced ferromagnetic ordering. Future work will include time-resolved magneto-optical Kerr measurements to unambiguously resolve the magnetization dynamics. Jvcicrcnccs 1. S. Jin, T. H. Tiefel, M. McCormack, R. A. Fastnacht, R. Ramesh and L. H. Chen, "Thousandfold change in resistivity in magnetoresistive La-Ca-Mn-O fihns," Science 264, 413-415 (1994). 2. R. D. Averitt and A. J. Taylor, "Ultrafast optical and far-infrared quasiparticle dynamics in correlated electron materials," J. Phvs. Condens, Matter 14, R1357Rl 390 (2002). 3. Y. Shimakawa, Y. Kubo and T. Manako, "Giant magnetoresistance in TIjMnjOT with the pyrochlore structure," Nature 379, 53-55 (1996). Sliimakawa and Y. Kubo, "Charge dynamics in the colossal magnetoresistance pyrochlore Tl2Mn207," Phys, Rev, B 64,180409-1-180409-4 (2001).

315

Femto-magnetism visualized in three dimensions J.-Y. Bigot, M. Vomir, L.H.F. Andrade, M. Albrecht, J. Arabski, E. Beaurepaire Institut de Physique et Chimie des Materiaux de Strasbourg, Unite Mixte 7504 CNRS, Universite Louis Pasteur, BP. 43, 23 rue du Loess, 67034 Strasbourg Cedex, France E-mail: [email protected] Abstract. We measured the three-dimensional real-space trajectory of the magnetization in ferromagnetic metallic films perturbed by femtosecond optical pulses. It reveals the dynamics of the initial ultrafast demagnetisation followed by the spin precession and damping.

1. Introduction Femto-magnetism consists in using femtosecond optical pulses to perturb the magnetization of ferromagnetic metallic materials [1]. Indeed, several studies made on various materials and performed with different techniques, lead to the conclusion that one can take advantage of optics to induce ultrafast magnetic changes in the femtosecond timescale. Such method of inducing and observing time dependent magnetic phenomena with optical pulses presents several advantages as compared to using pulsed magnetic fields. First, the temporal resolution is much better since it is in principle limited by the duration of the optical pulse. Second, a large perturbation of the ferromagnet can be made locally, within the focused beam, with a small total energy deposited in the material therefore avoiding large permanent heating. In this paper, we have investigated the magnetization dynamics in the three dimensions of the real space. This experiment allows to visualize the magnetization trajectory after perturbation by the pump pulse. It clearly shows how the initial demagnetization induced by the pump pulse is further transferred to a precession and damping of the spins when an additional static field is used to observe the ferromagnetic resonance.

2. Experimental configuration: three dimensional probing of the magnetization dynamics in Cobalt films The experimental configuration is designed in order to retrieve the variation of the magnetization of a ferromagnetic Cobalt thin film in the three directions of space. To do so, we measured simultaneously the time resolved polar, longitudinal and transverse magneto-optical Kerr signals induced by femtosecond pump pulses. The laser source is an amplified Titanium:sapphire centered at 790 nm, operating at 2.5 kHz and with a pulse duration of 120 fs. Part of the beam, the pump, is frequency doubled at 395 nm and focused onto the sample within a spot of

316

-100 jLim diameter with a maximum fiuence of 2 mJcm" . The other part of the laser beam, the probe, is focused onto the sample within a spot size of -30 ^im diameter. The angle of incidence of the probe beam is set at 52° and contains the information both on the polar and longitudinal signals. The incident polarization state of the probe can be adjusted to s or p. The polarization rotation e or ellipticity T) of the probe, as well as its reflectivity (R) and transmission (T) are monitored simultaneously. Therefore we have access to the four differential signals: (AT/T), (AR/R), (Ae/e), (Ar|/r|) as a function of the pump-probe delay x. The sample is a 16.5 nm thick Cobalt film grown by molecular beam epitaxy on a sapphire substrate, with an in-plane spontaneous magnetization. A uniform static magnetic field HQ of 3000 Oe is applied to the sample and points in a direction determined by the plane of incidence of the probe and the angle ([) with respect to the normal of the sample (0 < ([)< 180°). Each of the differential signals defined above, is measured for the two angles ^ and 7i - (|) named (j)"^ and (f hereafter. The quantities of interest are the rotations and ellipticities in the polar (Pol), longitudinal (Long) and transverse (Trans) magneto-optical configurations. These quantities are proportional to the projections of the magnetization on the three orthogonal coordinate axis: e^ (normal to the surface of the sample), 5 and e^ (in the plane of the sample) e^ being perpendicular to the plane of incidence of the probe. When the probe is p polarized these projections are obtained respectively by the quantities: APol e (x) = V2[(Ae/epoi)((t)"^) - [(Aei/epoi)((t)~)]; ALong e (x) = i/2[(Ae/eiong)((l)"') + [(Ae/eio„g)((|)")] and ATrans e (x) = i/2[(AR/eiong)((l)'') - [(AR/eiong)((t)")]. The same definitions hold for the polar, longitudinal and transverse components of the ellipticities replacing e by r).

3, Magnetization trajectory induced by femtosecond pulses Figures la) and lb) show APol e (x) for the two angles (t) = 0 and 15°. The insets show their short delay behaviour. The initial fast signal corresponds to the ultrafast demagnetization. The damped oscillatory behaviour is due to the precession of the spins around the effective field resulting from the applied field Ho as well as the shape and magneto-crystalline anisotropy fields. This dynamics is the analogous of the ferromagnetic resonance. It can be induced ail-optically as reported in an exchange-coupled NiFe/NiO bi-layer [2] or in Ni and permalloy NigoFeio films [3]. 0.00 -0.02 Q.

<

--|

a)(j) = 15°

0

10 . 2 0 f »

-0.04 -0.06

V

0.00 -0.04 -0.08 0.0

0 0.2

0.4

. 10 . 20PS 0.6 0.8 1.0

Delay (ns)

0.0

0.2

0.4

0.6

0.8

1.0

Delay (ns)

Fig.l: a) and b): Spin precession and ultrafast demagnetisation (inset) of Co thin film, c) and d): corresponding differential transmission and reflectivity.

317

Figures Ic) and Id) represent AT/T(T) and AR/R(T) of the probe for the long and short delays (inset). The relaxation occurs in two steps. The first one corresponds to a rapid cooling of the electrons to the lattice, within 0.9 ps, via the electronphonon interaction. Second, the heat diffusion occurs within the next 200 ps. In figure 2 we have represented the variations of APol 8 (x), ALong e (x) and ATrans e (x) in figures 2a), 2b) and 2c) respectively ((t)"" = 15° and ([)" = 165°). The short delay behaviours are displayed in the insets. The corresponding trajectories in the XY, XZ and YZ planes are shown in the figures 2d), 2e) and 2f). Figure 2 clearly shows that, for short delays, it is the modulus of the magnetization which diminishes since the corresponding signals APol e (x) and ALong e (x) along the X and Y axis decrease simultaneously. In addition there is no contribution along the Z axis showing that the magnetization remains in the XY plane during the demagnetization process. The next step is a recovery of the magnetization (the modulus re-increases) as well as a rotation in the XY plane. This overall behaviour results from the demagnetization induced by the pump pulse, followed by an angular variation of the magnetization related to a decrease of the anisotropy. The following dynamics is then driven by the precession.

ip-^

-r:::^^

^^^^==^ -10

P \/\A/\v/^>^«vXvyv*vv^w*^^

-5

5

= ^ ^ ^

'

-10

0

d) Y/10"'

Zm

-5

0

5

e) Y/10"

] ^ ^ ^ t)

Fig. 2: Polar (a), longitudinal (b) and transverse (c) magneto-optical signals from the Co film. Real space trajectory of the magnetization in the XY (d), ZY (e) and ZX (f) planes. In conclusion, we studied the real space trajectory of the magnetization dynamics induced by femtosecond optical pulses in Co thin films. The magnitude of the magnetization first decreases during the first hundreds of femtoseconds. Simultaneously, the magnetization rotates in the plane of incidence. Finally, it undergoes a motion of precession which is observed in the XYZ referential.

References 1 see for example: J.-Y. Bigot, Femtosecond magneto-optical processes in metals, Comptes Rendus de TAcademie des Sciences, serie IV, no 10, 1483 (2001) 2 Ganping Ju, Lu Chen, A.V. Nurmikko, R.F.C. Farrow, R.F. Marks, M. J. Carey, B.A. Gumey, Coherent magnetization rotation induced by optical modulation in ferromagnetic/antiferromagnetic exchange-coupled bilayers, Phys. Rev. B, 62, 1171(2000). 3 M. van Kampen, C. Jozsa, J.T. Kohlhepp, P. LeClair, L. Lagae, W.J.M. de Jonge, B. Koopmans, All-Optical Probe of Coherent Spin Waves, Phys. Rev. Lett., 88, 227201 (2002).

318

Photo-induced demagnetization observed by time-resolved mid-infrared transmittance spectroscopy in Gao.94Mno.06As E. Kojima,^ J.B. Heroux,^ R.Shimano/ Y. Hashimoto,^ S. Katsumoto,^ Y. lye^ and M. Kuwata-Gonokami^ * Department of Applied Physics, Faculty of Engineering, University of Tokyo, and SORST (JST), 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033, Japan Email: [email protected] ^ Institute for Solid State Physics, The University of Tokyo, 6-3-7 Kashiwanoha, Kashiwa, Chiba 277-8581, Japan Abstract. The spin response of a dilute magnetic GaowMno.oeAs semiconductor sample subject to femtosecond photoexcitation is observed by mid-infrared differential transmittance. A slow rise-time in the hundreds of picoseconds timescale is found.

1. Introduction The control and manipulation of spins by ultrafast optical excitation causing photo-induced demagnetization and magnetic phase transition has been achieved in ferromagnetic metals[l], doped semiconductors[2] and II-VI dilute magnetic semiconductors[3]. Ultrafast optical excitation is a powerful characterization technique to investigate spin dynamics because the spin, lattice and electron temperature rise-times due to an ultrashort light pulse are decoupled and may be observed independently. In this work, spin dynamics in a III-V dilute magnetic semiconductor, Gao.94Mno.06As, is investigated by photo-induced demagnetization. Experimental results obtained using two different time-resolved probe techniques - mid-infrared (MIR) transmittance and visible/near-infrared magneto-optical Kerr effect (TRMOKE) - are compared. The 1.05 |Lim thick Gao94Mno.o6As layer was grown by molecular beam epitaxy on an undoped (001) GaAs substrate, annealed at 280 °C for 15 minutes and has a Curie temperature of 110 K. The source of the pump and probe pulses is an amplified mode-locked Ti-sapphire laser emitting 1.55 eV light pulses with a 150 fs duration at a 1 kHz repetition rate. For the pump, the pulses are frequency-doubled to obtain a 3.1 eV photon energy with a 660 |iJ/cm^ intensity. The probe pulses for the MIR transmittance measurements are obtained from the difference frequency generation of an AgGaSz crystal using the signal and idler beams of an optical parametric amplifier and have energies varying between 100 and 400 meV.

319

2. Results Figure 1 shows the differential transmittance obtained at temperatures of 54, 70 and 120 K using a 250 meV probe wavelength. An ultrafast decrease of the transmitted light intensity due to a change of the dielectric constant induced by photo-mjected carriers is initially observed. The subsequent rapid increase of AT indicates a fast non-radiative recombination of the photo-generated carriers and an increase of the lattice temperature. A shoulder is clearly visible around 3 ps, after which a transition to a much slower rise of the differential transmittance is seen for the 54 and 70 K curves up to 1500 ps approximately. The horizontal lines indicate the magnitude of this amplitude, labeled ATmag/T. It can be clearly observed, first, that ATmag/T decreases as the temperature increases and, second, that it disappears altogether above the Curie temperature. This provides a strong indication that this slow component of the differential transmittance is related to demagnetization and associated with a slow response of the spins to the excitation pulse. Measurements performed at different probe wavelengths exhibit a similar behavior, confirming the generality of the resuhs. L 250meV

AT^agJ 54K 70K

<

120K

'v^ 1

ff^i'^^A'^'^^^^'^isi^igifif^'r*'^

1 1 1 1 1 1 III

L 1 1 1 1 1 III

10

100

1

1 1 1 1 1 III

B^%tfaP>

1 1—LJ

1000

Time(ps) Fig. 1. Mid-infrared differential transmittance with a 250 meV probe energy at 54, 70 and 120 K. The demagnetization curve obtained by time-resolved magneto-optical Keneffect spectroscopy[4] and pump-probe transmittance data points obtained with a 190 meV probe energy are plotted in figure 2. A clear correlation is observed. It should be noted that the demagnetization risetime around 500 ps is atypically long compared with values reported for example for Ni[l] or InMnAs[5] and indicates a spin-dependent band structure (spin polarization of hole charge carriers) [6]. If acceptor-boimd magnetic polarons are formed in GaMnAs[7], the detection of spin-related properties by mid-infrared transmittance could be explained by a variation of the polaron binding energy. The experimental results shown in figure 2 clearly indicate that MIR transmittance measurements are related to ferromagnetism in GaMnAs. Within a few picoseconds, the photo-excitation of holes induces a rapid non-radiative

320

recombination and a temperature increase of the holes and lattice phonons. Due to the spin polarized state of the holes, an increase of their kinetic energy does not directly affect their spin configuration. After a few hundred picoseconds, the spin temperature gradually increases to reach quasi-thermal equilibrium with the hole and lattice systems. Finally, above 1500 ps approximately, relaxation to the initial state begins. 1

1

1

1

1

1

4.5 h

_l| o

r

o^„^0^^'ff.1^ - - ^ ^

4.0 L_

j^^^^o

J

°o Q}

3<

3.5 h 3.0

L07o

0

^^4^ ° ° P

O AT/T at 190meV J -AM(t) from TR-MOKE 1

1

1

1

1

1

-iJ

100

200

300

400

500

600

Time(ps) Fig. 2. Magnitude of the photo-induced demagnetization measured by time-resolved magneto-optical Kerr effect (right axis) and of AT/T (left axis).

4. Conclusions The spin temporal response of a Gao.94Mno.06As sample subject to a 3.1 eV femtosecond laser pulse was investigated. We found a slow response of the spin system to photo-excitation, which indicates a spin-dependent hole band structure in this material. Mid-infrared differential transmittance spectra are related to ferromagnetic order in this material, as shown by the temperature dependence of the temporal response obtained in the hundreds of picosecond timescale.

References 1 E. Beaurepaire, Phys. Rev. Lett. 76, 4250-4253 (1996). 2 J. M. Kikkawa and D. D. Awschalom, Phys. Rev. Lett. 80 4313-4315(1998). 3 See, e.g., C. Buss, R. Pankoke, P. Leisching, J. Cibert, R. Frey, and C. Flytzanis Phys. Rev. Lett. 78 4123-4125 (1997). 4 E. Kojima, R. Shimano, Y. Hashimoto, S. Katsumoto, Y. lye, and M. KuwataGonokami, Phys. Rev. B 68, 193203 (2003). 5 J. Wang, G.A. Khodaparast, J. Kono, T. Slupinski, A. Oiwa and H. Munekata, Physica E 20, 412-418 (2004). 6 T. Kise, T. Ogasawara, M. Ashida, Y. Tomiaka, Y. Tokura and M. KuwataGonokami, Phys. Rev. Lett. 85, 1986-1988 (2000). 7 A. Kaminski and S. Das Sarma, Phys. Rev. Lett. 88, 247202 (2002).

321

Optically Induced Magnetization and Ultrafast Spin Dynamics of Magnetic Ions in Ionic Crystals T. Kohmoto^ K. Nakazono^, S. Furue^, M. Kunitomo^ and Y. Fukuda^ ^ Department of Physics, Faculty of Science, Kobe University, Kobe, 657-8501, Japan E-mail: [email protected] ^ Graduate School of Science and Technology, Kobe University, Kobe, 657-8501, Japan Abstract. Optically induced magnetizations in Tm^^:CaF2 and Cr'^^:Al203 are observed by using the polarization spectroscopy. Quantum-beat free-induction decay and ultrafast spinlattice relaxation near room temperature in the ground states of the magnetic ions are investigated..

1.

Introduction

Optically induced magnetization and ultrafast spin dynamics in the ground state of magnetic-ion doped crystals are studied by the polarization spectroscopy. Quantum-beat free-induction decay in subnanosecond region and ultrafast spinlattice relaxation in picosecond region near room temperature were observed for the thulium ions in calcium fluoride (Tm^^:CaF2) and chromium ions in sapphire (ruby, Cr^^:Al203). Such ultrafast spin dynamics cannot be observed by the conventional electron spin resonance (ESR), whose time resolution is usually of the order of microseconds. The first study on optically induced magnetization and spui dynamics in crystals was reported for transition-metal ions at room temperature [1-3]. In these experiments, the magnetization was mduced by circularly polarized light pulses, and the time derivative of the optically induced magnetization was monitored using pickup coils. However, the time resolution of the detection system in such experiments is of the order of nanoseconds at the highest. If the induced magnetization is monitored by using optical pulses, the time resolution can be significantly improved because the time resolution is limited only by the temporal width of the optical pulses. Thus, it is possible to observe the ultrafast spin dynamics in the picosecond or femtosecond region. In order to study the ultrafast spin dynamics of magnetic ions in ionic crystals near room temperature, we applied polarization spectroscopy using the pump-probe technique for the induction and detection of magnetization. Circular dichroism of the optical transition is responsible for the creation and the detection of the magnetization. The magnetization is created in the ground state by a ch-cularly polarized pump pulse (400 nm, 0.2 ps, ~10 jiJ), and its time evolution is detected by a quarter-wave plate and a polarimeter as the change of the polarization of a linearly polarized probe pulse. The time separation between the pump and probe pulses is swept by an optical delay line.

322

2. Thulium Ions in Calcium Fluoride Optically induced magnetization for 0.02%Tm^^:CaF2 in the magnetic field of 1.0 kOe perpendicular to the pump beam observed at 60 K is shown in Fig. 1(a), where the Zeeman coherences among the ground-state sublevels (S=l/2, 7=1/2) are created by the pump pulse, and the free precession of the magnetization is detected by the probe pulse (560 nm). This type of signal is referred to as the quantum-beat free-induction-decay signal. The beat signal consists of two components of oscillation whose periods are 180 ps and 230 ps (5.5 GHz and 4.3 GHz). The Fourier transform of the observed signal gives the ESR spectrum. This all-optical method may be called optically induced Fourier-transform (FT) ESR spectroscopy. The decay curve of the magnetization in the magnetic field of 0.2 kOe parallel to the pump beam observed at room temperature is shown in Fig. 1(b), where the spin-lattice relaxation time is 5 ps. The observed temperature dependence of the spin-lattice relaxation rate l/Ti near room temperature is deviated from the T^ dependence, which is expected from the well-known Raman process of phonons. Considering the effect of the Debye temperature (^=513 K) of the host crystal on the Raman process, the observed temperature dependence of l/Ti can be explained well. (a) Tm'^:CaF2 (1.0kOe,60K)

2 4 Delay Time (ns)

10 Delay Time (ps) Fig. 1. (a) Quantum-beatfree-induction-decaysignal in Tm^^:CaF2 observed at 60 K in the transverse magnetic field of 1.0 kOe. The beat signal consists of two components of oscillation whose periods are 180 ps and 230 ps (5.5 GHz and 4.3 GHz), (b) Decay curve of the magnetization in Tm^^:CaF2 observed at room temperature. The spin-lattice relaxation time is 5 ps.

323

3.

Chromium Ions in Sapphire

Optically induced magnetization for 0.2%Cr^^:Al2O3 in the magnetic field of 2.9 kOe perpendicular to the pump beam observed at room temperature is shown in Fig.2(a), where the wavelength of the probe pulse is 694 nm (R transition) and the pump and probe pulses are incident along the c axis of the crystal. The oscillating magnetization signal, the quantum-beat free-induction-decay signal, appears in the transverse magnetic field. The relaxation time of the magnetization in ruby is longer and in the nanoseconds region even at room temperature. Fourier transform of the observed quantum-beat free-induction-decay signal in Fig.2(a) is shown in Fig.2(b), where two main peaks around 4 GHz and 11.5 GHz appear as shown by the two arrows. The energy levels in the ground state of the Cr^^ ion in the transverse magnetic field is shown in the inset of Fig.2(b). The observed oscillation frequencies are explained by the energy splittings in the ground state. The all-optical technique in the present work can be a powerfril tool to study the ultrafast spin dynamics in condensed matters, since this method can create a large population difference even at room temperature and give a high time resolution, which cannot be achieved by the conventional magnetic-resonance methods.

Delay Tune (ns) (b) Fourier spectrum

cd CD OH

CO

10 20 Frequency (GHz) Fig. 2. (a) Quantum-beat free-induction-decay signal in Cr^'^:Al203 observed at room temperature in the transverse magnetic field of 2.9 kOe. (b) Fourier transform of the observed quantum-beat free-induction-decay signal in (a). The inset shows the energy levels in the ground state of the Cr^^ ion in the transverse magnetic field.

References 1 G. F. Hull Jr., J. T. Smith and A. F. Quesada, Appl. Optics 4, 1117 (1965). 2 J. P. van der Zeil and N. Bloembergen, Phys. Rev. 138, A1287 (1965). 3 Y. Takagi, Y. Fukuda and T. Hashi, Opt. Commun. 55, 115 (1985). 324

Dynamic coupling-decoupling crossover in the current-driven vortex-state in Tl2Ba2CaCu208 studied using terahertz time-domain spectroscopy V. K. Thorsm0lle^ R. D. Averitt\ I. Aranson^ M. P. Maley\ L. N. Bulaevskii\ and A. J. Taylor^ ^Los Alamos National Lab, Los Alamos, New Mexico 87545, USA ^Argonne National Lab, 9700 South Cass Ave., Argonne, Illinois 60439, USA Abstract: Employing terahertz time-domain spectroscopy in transmission, we have measured the Josephson plasma resonance in Tl2Ba2CaCu208 high-T^ thin films, and studied the current-driven coupling-decoupling crossover in the driven vortex lattice.

1.

Introduction

The properties of driven periodic structures subject to quenched disorder, including charge-density waves, Wigner crystals and vortex lattices, have become one of the central issues in the phenomenology of nonequilibrium statistical mechanics [1-4]. In the context of the vortex lattice, Koshelev and Vinokur predicted the driven system to undergo a dynamic phase transition at some threshold current between the fluid-like and crystal-like moving states [2]. In other words, an applied current puts forth a scenario where a pinned vortex lattice flows plastically at first as some vortices are depinned, and then becomes more ordered at higher applied currents as more vortices are depinned, possibly forming a moving ordered vortex lattice. Thus, beyond some critical value a dynamic phase transition may occur to a more ordered state, characterized by a change from incoherent to coherent vortex motion. The c-axis correlations of pancake vortices in a highly anisotropic high-r^ superconductor are directly related to the interlayer phase coherence, ^ in the vortex state [5]. (pn,n+i(Y,B) is the phase difference between layer n and n+1, r is the in-plane coordinate, and B is the magnetic field. When the pancake vortices form straight lines perpendicular to the layers, (pn,n+i(^,B) vanishes and the average of the cosine of the phase difference = 1. However, when the pancake vortices are misaligned along the direction perpendicular to the layers, a nonzero phase difference is induced, which results in the reduction of from unity. Interlayer correlations of pancakes in a driven vortex system in highly anisotropic layered superconductors were studied by Aranson et al by numerical simulations of the time-dependent Ginzburg-Landau-Lawrence-Doniach equations in a model of two coupled layers [1,2]. The authors calculated the average of the cosine of the phase difference, Wc = , as a function of the electric field E

325

along the layers produced by the intralayer current. The authors found a decrease of Wc at low electric fields (currents) in the plastic flow regime accompanied by an increase of Wc at higher electric fields (currents) as smectic flow replaces plastic flow. The rate of increase becomes stronger as the anisotropy decreases because the vortex system approaches the case of an isotropic superconductor where a dynamic transition should occur at some critical value of the current. The authors concluded that a dynamic coupling-decoupling crossover takes place regarding the interlayer pancake ordering, while a change of intralayer ordering with current can be described as a dynamic melting transition. Here we use the c-axis Josephson plasma resonance (JPR) [6], which is one of the most powerful experimental probes for probing the interlayer phase coherence and c-axis ordering of pancake vortices in highly anisotropic layered superconductors. The JPR is a Cooper pair charge oscillation mode perpendicular to the CUO2 layers. In zero magnetic field the JPR is a direct probe of the Josephson coupling between the layers. In this case the JPR frequency is given as C0Q{T) = CI[\{T)^]

= CI[Y\^{T)^].

Here, A, and Xat are the London

penetration depths along the c-axis and (2Z?-plane, respectively, y is the anisotropy parameter, c is the speed of light, and e^ is the high-frequency dielectric constant along the c-axis. In the presence of a c-axis magnetic field B, the JPR can be written as [5] COI{BJ)

= (OI{T)W^,

W;=(cos[^„„,,(r,5)]).

(1)

Here is the local thermal and disorder average of the cosine of the gauge-invariant phase difference between adjacent layers n and n+l. Thus, the JPR probes the value of W^ which is directly related to the correlations of pancake vortices along the c-axis.

2. Results and Discussion Here we present the first studies of the current-driven vortex state in a high-r^ superconductor, Tl2Ba2CaCu208+6 (Tl-2212), by direcdy probing the interlayer phase coherence with the JPR, and measured employing THz-TDS in transmission. The growth process of Tl-2212, the experimental setup are discussed in Ref. 6. All experiments were performed in field-cooled mode. factor, In Fig. 1 we show the interlayer phase coherence W^ =CO][B

= 2.5kG,T,l)/a)l

(0,10K,0), as a function of current in the ab-

plane at 10, 60, 80, and 90 K in a 2.5 kG c-axis appUed magnetic field. We see three different types of behavior of Wdl) with increasing current: a) at low currents Wc decreases with /, i.e. plastic flow regime; b) rapid increase of Wc with / as smectic flow is established above the threshold current; and c) slower increase of Wc with /. Next we compare our experimental data with theoretical simulations for Wc at 80 and 90 K. We note that in a magnetic field of 2.5 kG and at temperatures 80 and 90 K the current-driven vortex state corresponds to the liquid

326

vortex state [7]. We associate the pronounced increase in at 80 and 90 K with a

[mA]

Frequency (THz]

Fig. 1. Interlayer phase coherence factor versus apphed aZ?-plane current in Tl2212 in 2.5 kG c-axis magnetic field. The solid hues are guides for the eye. The arrows at 0 mA indicates the value of for each temperature after the current has been switched off. The shown fits for 80 and 90 K are obtained from simulations using 4 = 24 mA. The two right figures illustrates the effect of the current on the JPR.

dynamic phase transition as predicted by Koshelev and Vinokur [2], where a crossover from coupled to decoupled liquid vortex state occurs. We conclude that there is qualitative good agreement between the experimental and theoretical results. The minor discrepancy between the experimental and theoretical results in the regime of small currents (from 0 to 20 mA) can be attributed to the fact that thermal fluctuations were neglected in the simulations. These fluctuations will possibly enhance the vertical alignment in the small current regime.

References 1. 2. 3. 4. 5. 6. 7.

L Aranson, A.E. Koshelev, and V.M. Vinokur, Phys Rev. B 56, 5136 (1997). A.E. Koshelev and V.M. Vinokur, Phys. Rev. Lett. 73, 3580 (1994). U. Yaron, P.L. Gammel, D.A. Huse, R.N. Kleiman, C.S. Oglesby, E. Bucher, B. Batlogg, and D.J. Bishop, Phys. Rev. Lett. 73, 2748 (1994). J.A. Fendrich, U. Welp, W.K. Kwok, A.E. Koshelev, G.W. Crabtree, and B.W. Veal, Phys. Rev. Lett. 77, 2073 (1996). A.E. Koshelev, L.N. Bulaevskii, and M.P. Maley, Phys. Rev. B 62, 14403 (2000). V.K. Thorsm0lle, R.D. Averitt, M.P. Maley, L.N. Bulaevskii, C. Helm, and A.J. Taylor, Opt. Lett. 26, 292 (2001). V.K. Thorsm0lle, R.D. Averitt, M.P. Maley, M.F. Hundley, A.E. Koshelev, L.N. Bulaevskii, and A.J. Taylor, Phys. Rev. B 66, 012519 (2002).

327

Ultrafast light induced charge disordering around phase transition temperature in 2D spin ladder compound NaV205 Motoo Aiba, Makoto Nakajima, Masahiko Isobe, Yutaka Ueda, and Tohru Suemoto The Institute for Solid State Physics, The University of Tokyo,5-l-5 Kashiwanoha, Kashiwa, Chiba 277-8581,Japan E-mail: [email protected] Abstract. The light induced phenomena on charge-ordered phase in NaV205 has been investigated by femtosecond time-resolved reflection spectroscopy. A large reflectivity change with a rise time of 10 ps and a decay time of several hundreds picoseconds has been observed and ascribed to a photo-induced phase transition accompanying a destruction of the charge order.

1. Introduction There has been an increasing interest in the light induced phenomena leading to macroscopic change of the order in solids [1]. However, only few materials have been subjected to research of the light induced phase change. Especially, attempts to the destruction of the charge order have been limited only for some perovskite oxides [2]. In this paper, we report the light induced phenomena on the chargeordered phase in NaV205. The structure of a'-NaV205 consists of two-leg ladders formed by comersharing distorted V2O5 pyramids running along the b axis, with their rungs along the a axis. The rungs are formed by two V ions bridged by an O ion. Neighboring ladders are linked via common edges of the pyramids to form ab layers and Na atoms lie between the layers. It is known to show a phase transition at Tc=34K from the measurements of the magnetic susceptibility[3]. This phase transition is basically a charge ordering accompanied by a formation of spin singlets. In the high temperature phase (HP), all V ions are charged equally (V^^^), while in the low temperature phase (LP), charge separation occurs, resulting in V"^^ and V^^ which form zigzag double chains along the b axis [4]. The two V"^^ ions between the neighboring ladders form a spin singlet with a lattice distortion which fornis the 2ax2bx4c superlattice structure [5].

2. Experimental The time evolution of the reflectivity change of NaV205 has been investigated by femtosecond pump-probe measurements, using a Ti: sapphire laser (central wave length 785nm, pulse width lOfs, repetition rate 75MHz). The intensity of the 328

probe pulse reflected from the sample surface was measured as a function of the delay time relative to the pump pulse. The pump beam was chopped at a frequency 1.02kHz and the signal was taken with a lock-in amplifier. These two beams were focused to the sample surface by an achromatic lens (f = 100 mm). The sample which had been grown from the melt, using the self-flux method was set in vacuum space of a He-flow cryostat. The measurements were performed on the a-b plane of the sample.

3.

Results and Discussion

The time-resolved reflectivity change is shown in Fig. 1(a) and (b). Here shown are the net change of the reflectivity subtracting the background signal, which does not depend on the delay. The polarization configuration is Epump //a and Eprobe //b. As shown in Fig. 1 (a), the response at room temperature (RT) shows a decrease with a short lifetime (-500 fs) and a small amplitude on the order of 10"^. Similar behavior is found in whole temperature range from RT down to 60K. In contrast, below the phase transition temperature, we found a large and slow reflectivity change after photo-excitation as shown in Fig. 1 (b). The reflectivity decreases within 10 ps and recovers in several hundreds picoseconds at 4 K. Aside from this time resolved measurements, we performed reflectivity spectrum measurements on the same sample around the phase transition temperature by using a conventional micro-spectrometry setup. In increasing temperature, we found a stepwise decrease of about 1 % in the reflectivity at Tc. This indicates that the transient decrease of reflectivity observed in the time-resoled measurement corresponds to the phase change from the LP to the HP.

xlO

0.0 h-

1 fci K^J

1 ^

-0.2 hIcio" 5

-0.4 L-0.6 [•

1

40 K 30 K

_

0

''""''--''^----v^Z.L^

-5

-

-oA~-10

RT

-l.oL~-15 -I.2L

\

• I I I

0.0 1

-10

(b)'

• Transient xlO' D Background

B <-0.5k

4K _ ^ H 0.0

>

2.5 1

xlO"

20 K 1

0 10 Delay time (ps)

-

1

20

10 20 30 Temperature (K)

Fig. 1. (a) Transient reflectance change at lower temperatures and at RT (inset), (b) Temperature dependence of the amplitude of the transient response and the background components. The background component is plotted as a fimction of temperature in Fig.l (c) by open squares. This component appears only when the pump and probe beams spatially overlap each other exactly, but does not depend on the delay time. This

329

component increases at a temperature slightly below Tc. This indicates that the average temperature of the sample surface has increased due to the laser heating and the phase has been changed from the LP to the HP when the chopper is open. In contrast, the transient response shown by solid circles in Fig. 1 is seen at the lowest temperature as well, where more heat is required to induce the phase transition. Moreover, it shows no increase around the phase transition temperature. These facts suggest that the transient decrease of the reflectance should not be ascribed to a thermally induced phase transition. In order to confirm this interpretation, the amplitude ratio with several polarization configuration was examined. The reflectivity change is larger for Eprobe //b when Epump//a, where E's denote the electric vector of the corresponding beam. On the other hand, when Epump//b, it is larger for Eprobe //a. It indicates that the polarization character of the generated state is dependent on the direction of the excitation light. If the effect of the pump pulse were only to put heat on the sample, the character of the generated state should not depend on the polarization of the pump pulse. Thus these facts support our interpretation that the observed phenomenon is not a thermally induced, but a photo-induced phase transition. In the LP, the nearest ions of V"^^ are always V^^ because of its zigzag configuration. Therefore, the charge transfer transition i.e., an electron transfer from V"^^ ion sites to V^^ ion sites, occurs efficiently by light irradiation. Through this electron transfer, the electronic structure in NaV205 dismisses its charge ordering and changes to the charge disordered phase similar to the HP. Since the sites neighboring the V"^^ ions are not always V ^ in the disordered state, the charge transfer transition becomes less and the reflectivity decreases. The long lifetime exceeding 100 ps will correspond to the time required to reconstruct the zig-zag configuration In summary, we have studied the ultrafast dynamics of the photoinduced effect on the charge-ordered phase in NaV205 and conclude that the charge order in the LP is collapsed within 10 ps by the photo-excitation and that the charge order recovers within several hundreds picoseconds. Acknowledgements. This work has been supported by the Grant-in-Aid for Scientific Research (A) from the Ministry of Education, Culture, Sports, Science and Technology of Japan.

References 1 K. Nasu, in Relaxation of excited states and photo-induced structural phase transition, Springer, 1997. 2 T. Ogasawara, T. Kimura, T. Ishikawa, M. Kuwata-Gonokami, and Y. Tokura, Phys. Rev. B 63, 113105,2001. 3 M. Isobe and Y. Ueda, J. Phys. See. Jpn. 65, 1178, 1996. 4 T. Ohama, H. Yasuoka, M. Isobe, and Y. Ueda, Phys. Rev. B 59, 3299, 1999. 5 Y. Fujii, H. Nakao, T. Yoshima, M. Nishi, K. Nakajima, K. Kakurai, M. Isobe, Y. Ueda and H. Sawa, J. Phys. See. Jpn. 66, 326, 1997.

330

Correlation of the Electronic Transitions in Semiconducting Single-walled Carbon Nanotubes Y.-Z. Ma\ J. Stenger^ S.L. Dexheimer\ S.M. Bachilo^ R.E. Smalley^ R.B. Weisman^, and G.R. Fleming^ ^ Department of Chemistry, University of California, Berkeley and Physical Biosciences Division, Lawrence Berkeley National Laboratory, Berkeley, California 94720-1460, USA E-mail: [email protected] ^ Department of Chemistry, Center for Nanoscale Science and Technology, and Center for Biological and Environmental Nanotechnology, Rice University, 6100 Main Street, Houston, Texas 77005 Abstract. Frequency-resolved femtosecond transient absorption was applied to study the spectral properties and excited-state dynamics in semiconducting single-walled carbon nanotubes. We find that the electronic transitions are intrinsically correlated, revealing the excitonic nature of the states.

Semiconducting single-walled carbon nanotubes (SWNT) are a novel class of onedimensional, nano-scale materials characterized by discrete energy levels and diameter-tunable transition energies [1]. These unique electronic properties make SWNT an important candidate for novel nanoscale opto-electronic and photonic applications. While significant progress has been made on both experimental and theoretical aspects [2-4], the fundamental nature of the optical excitations in individual nanotube types remains poorly understood. Our previous time-resolved studies of these materials revealed rapid dynamics consistent with exciton annihilation [4]. In this work, we present studies that further address the nature of the optical excitations in semiconducting SWNT. Transient absorption (TA) measurements with sub-100 fs time resolution are carried out using a 250-kHz regeneratively amplified Ti:sapphire system. An optical parametric amplifier is used to generate pump pulses, and a white-light continuum serves as a broadband probe. Pump wavelengths are chosen to selectively excite the first (Ej) and second (Ej) optical transitions of a particular tube type, and the photo-induced absorption changes in the visible and near IR regions are spectrally resolved with a CCD camera at a series of time delays between the pump and probe pulses (r^). Fig. 1 shows the TA spectra recorded in the visible and NIR regions at Td = 50 fs and TJ = 1 ps, respectively. Excitation of the Ej transition of the (8,3) tube at 660 nm leads to a broad induced transmission (IT) band peaking at ~660 nm as the result of photobleaching (PB) of the selected transition and concurrently excited, energetically similar E2 transitions of other nanotube types, namely (7,5), (7,6) and (9,5) [3]. In addition, a second IT band at 731 nm and an

331

induced absorption (lA) feature, located between the two IT bands, are also found. These IT bands match the peaks in the linear absorption spectrum well. In the NIR region (Fig. lb), the spectrum is characterized by two distinct IT bands peaked at 954 and 1026 nm owing to the PB of the relevant transitions and the related stimulated emission [3,4]. These two IT bands can be assigned to the Ej transitions of the (8,3) and (7,5) nanotubes [3]. Direct excitation of the Ej transition at 953 nm results in two resolvable IT bands at r^> 1 ps, which peak at 956 and 975 nm, respectively (Fig. lb). These IT bands correspond to the Ej transitions of nanotubes (8,3) and (6,5) [3], which are simultaneously excited owing to their similar transition energies. In the visible region, two distinct IT bands observed upon 953 nm excitation are located at essentially the same positions as the IT bands observed upon excitation at 660 nm. Their spectral widths and amplitudes, however, are significantly different, and for the 660 nm band a reduction of amplitude by -15 times is seen. The instantaneous response of the (8,3) tube's E2 transition to the excitation of its EJ transition is striking. In order to discern the origin of this response, we further examine the temporal evolution of the TA probed at 660 and 953 nm upon excitation of the Ej and E2 transitions. As shown in Fig. 2a, nearly identical decay is observed after normalization at the maximum amplitude. The similarity illustrated in Fig. 2a strongly indicates that IT bands peaked at 660 and 953 nm (Fig. 1) are indeed associated with the Ei and Ei transitions of the (8,3) nanotube.

a

T^ = 50 fs aJ^ 0.03 an L .ft^W^

0,02

0,01

\ V

f - J .

J

A\ f

\J\

/

onn Wavelength (nm)

Wavelength (nm)

Fig. 1. TA spectra measured in the visible (a) and NIR (b) regions upon excitation at 660 (filled circles) and 953 nm (open squares). The solid lines are the linear absorption spectrum (scaled for ease of visualization) and the dotted lines are the spectra of the pump pulses. The TA spectra are vertically offset for clarity, and the thin dashed lines are the baselines.

These observations are inconsistent with the simple electronic band description based on tight-binding calculations [1]. While excitation of the ^^ transition would be expected to rapidly populate the states associated with the E] transition via intraband relaxation, no signal should be detected in the reverse scheme, namely pumping at Ei and probing at E2. Interpretation of our data is possible by considering the excitonic nature of the optical excitations in semiconducting SWNT as suggested in a recent theoretical study [4]. Accordingly, the E] and E2 transitions correspond to transitions to the first and second excitonic states, respectively, which are derived primarily from the electron and hole states of their corresponding quasiparticle bands. However, the Coulomb interaction leads to mixed characteristics of the resultant exciton state so that a given exciton state is formed not only from the electron and hole states of its own band but also from the states of other bands [5]. Consequently, the Ej and E2 transitions are intrinsically

332

correlated and will respond to optical excitations resonant with one of the transitions in a simultaneous manner. We note that accessibility of the E2 transition via two-photon absorption can be excluded based on the linear dependence shown in Fig. 2b.

-

• •

50 fs 170fs

-f'

:M : -

*.i-*-

-*-

15

20

•,'*:»••-•• 20 40 60 80100 Delay Time (ps)

10

b 25

30

Pump Fluence (^J/cm^)

Fig. 2. (a) NormaHzed temporal profiles probed at 660 and 953 nm upon resonant excitation of the Ej (open circles) and E2 (soHd lines) transitions of the (8,3) nanotube type, (b) Plots of the absolute amplitude of the TA profiles probed at 660 nm at r,/= 50 fs (circles) and r,/ = 170 fs (squares). Fits to the signal amplitudes (dashed lines) show a linear dependence on pump fluence, with a small offset resulting from superposition of an induced absorption signal.

In conclusion, we found that the two lowest electronic transitions of individual semiconducting nanotubes are intrinsically correlated, giving rise to simultaneous excitation of the relevant transitions. This correlation effect provides direct evidence for the excitonic nature of the optical excitation in semiconducting SWNT. Acknowledgements. This work was supported by National Science Foundation. J.S. thanks the German Academic Exchange Service (DAAD) for generous support.

References 1 2 3 4 5

S. Saito, G. Dresselhaus, and M. S. Dresselhaus, Physical Properties of Carbon Nanotubes (Imperial College Press, London, 1998). C. D. Spataru, S. Ismail-Beigi, L. X. Benedict, and S. G. Louie, Phys. Rev. Lett. 92, 077402, 2004. S. M. Bachilo, M. S. Strano, C. Kittrell, R. H. Hauge, R. E. Smalley, and R. B. Weisman, Science 298, 2361, 2002. Y.-Z. Ma, J. Stenger, J. Zimmermann, S. M. Bachilo, R. E. Smalley, R. B. Weisman and G. R. Fleming, J. Chem. Phys. 120, 3368, 2004. S. Schmitt-Rink, D. S. Chemla, and D. A. B. Miller, Adv. Phys. 38, 89, 1988.

333

Ultrafast radial transport in a micron-scale aluminum plasma excited at relativistic intensity B.T. Bowes*'^ M. C. Downer^ ^ H. Langhoff^ M. Wilcox'^ B. Hou''^ J. Nees^'^^andG. Mourou''* ^ FOCUS Center ^ Department of Physics, University of Texas at Austin, Austin, Texas, USA 78712 ^ Physikalisches Institut der Universitat Wurzburg, 97074 Wurzburg, Germany ^ Center for Ultrafast Optical Science, University of Michigan, Ann Arbor, Michigan, USA 48109 Abstract. Using femtosecond microscopy, we observe a thermal/ionization front expand radially at ~10^cm/s from a A,^-size spot of an aluminum target excited at >10^^W/cm^. Numerical modeling shows transport is predominantly radiative and may be initially nonlocal. Intense, high contrast femtosecond laser pulses deposit energy into the electrons of a solid faster than it escapes from the initially-excited volume and much faster than the target surface expands hydrodynamically. When initial electron temperature kTe exceeds several hundred eV radiative heat transport begins to dominate over collisional transport [1]. Past experiments in this regime [1] used loosely focused, ~1J, --Ips pump pulses and probed the target transversely in transmission and thus were restricted to observing late stages of ID radiative transport in an optically transparent material on a time scale of tens of picoseconds. We present new measurements using ImJ, 24fs pump pulses focused to a diffraction-limited X^-size spot (1.5|im diam.) to excite a metal target surface at relativistic intensity. We probe the target in reflection through microscope optics. This geometry enables us to observe the earliest stages of radiative transport in 2D on any target material on a sub-picosecond time scale. New features of radiative transport are expected on this space-time scale. For example, simple transport models in the diffusive limit (i.e. Rosseland radiation mean free path XK « heated spot size X) predict very different temporal evolution of the thermal/ionization front in 2D vs. ID [3]. Moreover, since our focal spot size is on the order of A.R, our experiment may open up fs studies of nonlocal radiative transport. Pump pulses, focused with an f/1 off-axis paraboloid coupled to adaptive optics [2], excite an Al target at IkHz repetition rate. At the highest focused pump intensity, Ka and continuum bremsstrahlung x-rays are clearly observed. The target is translated so each laser shot interacts with clean target material. Surface wobble is limited to
334

excitation at 1.8xlO^^W/cm^ obtained by the measuring the center of the imaged pump-excited region. AR is neghgible for an S-polarized probe showing that the drop of the P-polarized reflectivity is caused by resonance absorption (RA) [4] at the expanding critical surface. The critical surface expands vertically at the ion acoustic velocity and max. RA occurs when it expands to VzAtmax^^probe/SwSOnm. [4]. From Fig. lb, Atmax«0.3ps, implying Vz«2xl0 cm/s. This Vz is consistent with ionization state Z«10 and kTe« several keV. The slight delay between ionizing and heating a point on the surface to (Z,kTe) and the appearance of maximum RA must be taken into account in modeling the radial expansion data. Since vertical expansion VzAt remains much less than the radial spot size during the time interval (0
.*.7 * * « ' " ^ V V / c m *

i Si-<•.»-

^

^

-H \r^m:Xt^-H K-

"^V

t„^

^

'N

^' , , , .

,^ / .^ ,

•

i*^

r

- K .

, , . . - -

.!»««*l,.jU>S*i..,„.-,

-.,......-,.„.-,..........-.......-„.........-.....-..„,..

Fig. 1. a) Lineouts from probe images at two different pump laser intensities. Large radial expansion of the laser excited region is clearly observed at the highest pump intensity (left) but is nearly absent when the pump laser intensity is reduced to 3.7xlO^'^W/cm^. b) Normalized S- and P-polarized probe beam reflectivity ARprobe(At)/Ro of the center of the laser excited region for pump intensity 1.8xlO^^W/cm^. Dashed curve: fit of P-polarized data to a ID RA model [4] assuming constant vertical expansion velocity Vz=2xl0^cm/s. The data deviate from the model for At>lps because the critical surface expansion becomes 3D and kTe has cooled. We model the evolution of the electron temperature profile Te(r, z<0, t) at constant solid density p by numerically solving the nonlinear diffusion equation dTJdi = V»(xVTe) with a Te-dependent thermal diffusivity Y = (KSH + KR)/PCV that included collisional (Spitzer-Harm) conductivity KSH ~ (kTe) /(Z+1) and radiative conductivity KR = 16asBTe^>iR/3 [3], where >cR[cm] = (9 x 10^)TerK]VZne[cm-^] is a simplified radiative mean free path for hydrogenic ions [1,3]. The initial condition was defined by partitioning absorbed pump energy (~lmJ, Gaussian radial profile) between electron thermal energy kTg and ionization Z(kTe) assuming Saha equilibrium. The choice of initial absorption depth Zabs was not critical because Te(r,z<0,t) quickly evolved to a nearly hemispherical profile which became the effective initial condition. Our code reproduces the ID transport results of Ref [1]. We complete the model by coupling the calculated Te(r,z<0,t) to the ID hydrodynamic model assuming Tc(r,z>0,t)=Te(r,0,t). This approximation is justified by the low heat capacity and high thermal conductivity of the expanding coronal plasma.

335

Fig.2a shows the calculated profile of the critical surface S(r,z) at several At for excitation at 1.8xlO^^W/cm^. During 0Lprobc/8 (max. RA). The resulting time evolution rc(At) is plotted in Fig. 2b and compared with the measured HWHM rdark(At) of the dark spot.

f ^ « l « l l u « tV^*^•*)

Fig. 2. a) Calculations of expanding critical surface of the laser-excited target from combining 2D radiative transport model with ID hydrodynamic model at different times At. b) Data points: Measured radius rdark(At) of darkened pump-excited plasma region at pump intensities 1.8xlO'^W/cm^ (circles) and 3.7xlO'^W/cm^ (triangles). Curves: Calculated radius rc(At) of critical surface highest intensity (solid), lower intensity (dotted) and neglecting radiative transport (dashed).

Both curves reproduce the measured spatial extent of radial expansion quite well. The main discrepancy between measured rdark(At) and calculated rc(At) (inc. KR) is the faster initial evolution of the former. We believe that this is caused in part by the initially nonlocal character of the radiative transport, which cannot be described by a diffusion equation. This belief is based on an estimated >-R~l^m, comparable to the pump focal spot size, for our estimated initial conditions. Additional complications may be the dependence of >.R on material opacity, which has been greatly simplified in our model, and the transport of hot electrons by ExB drift. [6] The latter effect pulls electrons out of the corona into vacuum, moves them radially away from the plasma region then drives them back into cool target material. 1. T. Ditmire et ai, "Supersonic ionization wave driven by radiation transport in a shortpulse laser-produced plasma," Phys. Rev. Lett. 77, 498 (1996). 2. O. Albert et al, "Generation of relativistic intensity pulses at a kilohertz repetition rate," Opt. Lett. 25, 1125(2000). 3. Y.B. Zerdovich and Y.P. Raizer, Physics of Shock Waves and High-Temperature Hydrodynamic Phenomena, edited by W.D. Hayes and R.F. Probstein (Dover, New York, 2002) 4. W. L. Kruer, The Physics of Laser-Plasma Interactions (Addison-Wesley, 1988). 5. J.P. Christiansen et a/., "MEDUSA a one-dimensional laser fusion code," Comp. Phys. Commun.7,271(1974). 6. Y. Sentoku et. al, "Laser light and hot electron micro focusing using a conical target," Phys. Plasmas 11, 3083 (2004).

336

Chirp Control of Free Carrier Dynamics in GaAs Toshiaki Hattori, Takeshi Yogi, Yoshikazu Hama, and Naoki Watanabe Institute of Applied Physics, University of Tsukuba, 1-1-1 Tennodai, Tsukuba, Ibaraki, 305-8573 Japan E-mail: [email protected] Abstract. The dynamics of free carriers in bufk GaAs was controlled by changing the chirp of the excitation light pulses having a duration in the 10 fs regime. Pump-probe measurements showed that the transmittance increases for negatively chirped pump pulses, which is in contrast to the trend observed with other materials. The result is explained by a combination of a pump-dump process and bandgap renormalization, and shows the possibility of a new way to control nonlinear optical response of semiconductors. Coherent control of dynamical properties of semiconductor materials is of great interest since it can open a new way to create electron states unachievable by other methods and to obtam new nonlinear optical properties of materials. Chup is the most important factor that describes the coherent properties of ultrashort optical pulses, and chirp dependences of ultrafast nonlinear optical response of organic molecules have been observed [1,2]. In these studies, the excited-state population of the molecules was measured using a pump-probe method while changing the chirp of the pump pulses, and it was observed that the excited-state population created by the pump pulses decreases for negatively chirped pump pulses. This observation has been explained based on the chirp dependence of the efficiency of an intrapulse pump-dump process. Kunde et at. observed chirp-controlled nonlinear optical response of AlGaAs at a carrier density of 3 x 10 ^ cm~^, and observed chirp-dependent pump-probe response in the delay time region where the pump and the probe pulses are overlapped [3]. The observed results were explained based on the chirp-dependence overlap of different frequency components of the pump and probe pulses. They did not observed, on the other hand, any dependence of the response at large delay times, where there is no overlapping between the pump and the probe pulses. We measured the chirp dependence of the transmission-type pump-probe signal intensity of a bulk GaAs sample having a thickness of about 10 jam at room temperature. At carrier densities 3 x 10^^ cm~^ and higher, we observed that the transmittance of the sample at large delay times increases for negatively chirped pump pulses, which is in the opposite direction to the trend observed with other materials [1,2]. The pump and probe pulses had a central wavelength of 790 nm and a spectral width of 100 nm. The pulses had a temporal width of 13 fs when unchirped, and the chirp of the pulses, which is characterized by the second-order phase, (j)2, was controlled using a fused silica prism pair. Both the pump and the probe pulses had the same chirp in the experiments. In Fig. 1 are shovm the pumpprobe signal intensities obtained using negatively chirped {^2 ^ -71 fs^), almost

337

200 400 Delay (fs)

600

Fig. 1. Delay-time dependence of transmittance change observed by pump-probe measurements of a 10-|am-thick GaAs sample at a carrier density of 5.9 x 10^^ cm"^. Results obtained using three values of second-order phase, (|)2, of the pump and probe pulses are plotted.

1

0.9

-

0.8

-

0.7

'

1

•

1

•

•

1.0

\

\

\

0.6 1.

-100

-50

1

1

50

•

•

. •

100

150

200

Fig. 2. Chirp dependence of the transmittance change of the pump-probe measurements at large delays, obtained at a carrier density of 5.9 x 10^^ cm~^. transform-limited ((^2 = 6 fs^), and positively chirped ((j)2 = 159 fs^) pulses. The pump pulse energy was kept constant, and the density of the created carriers was estunated as 5.9 x 10^^ cm" . In the figure, it is clearly seen that two features depend on the chirp; (i) the temporal profiles between -200 to 200 fs, and (ii) the signal level at large (> 300 fs) delays. The first feature was also observed by Kunde et al [3], and is explauied by taking into account the chirp-dependent temporal overlapping between the pump and probe pulses. The second feature, on the other hand, was not observed by Kunde et al. [3], where the signal level at large delays was independent of chirp. Since the carrier density is higher in the present study, we can attribute the difference to higher-order effects. We plot ui Fig. 2 the chirp dependence of the

338

pump-probe signal intensity at large delays obtained at the same carrier density. The observed dependence shows larger bleaching at negative chirp. In experiments using organic molecules, opposite chirp dependence of pump-probe signal was observed [1,2]. Those results were attributed to an intrapulse pump-dump process which exists only when the pump pulse is negatively chirped. Since the chirp dependence of the temporal signal profile around zero delay, as seen in Fig. 1, indicates that a pump-diunp process surely exists between the pump and the probe pulses, an intrapulse pump-dump process is expected to occur also in the present measurements. We have found that the cause of the observed non-intuitive chirp dependence of the pump-probe signal is a peculiar carrier-density dependence of the pump-probe signal intensity obtained with the present experimental configuration. From the measurement of the carrier-density dependence of the pump-probe signal intensity at large delays using unchirped pump pulses, we observed that the signal intensities increase linearly to the carrier density for smaller values (< 3 x 10^^ cm"^) of carrier density, and decrease for larger values. Since the present sample had a thickness of about 10 |^m, which is much larger than the penetration length of resonant light, only the spectral portion of the probe pulse below the bandgap was detected. Pump-probe measurements conducted under this condition probe the absorbance change in the Urbach tail spectral region of the sample. Most of the pump pulse energy is also absorbed by the sample of this thickness regardless of the chirp, and only the spatial distribution of carrier density depends on the chirp. When the carrier density is small, the measurement can detect the bleaching that is proportional to the carrier density. When the carrier density becomes larger, however, bandgap renormalization occurs, which will raise absorbance in the Urbach tail, leading to decrease in transmitted light intensity. The carrier density region where chirp dependence similar to that shown in Fig. 2 was observed almost agrees with that where the pump-probe signal shows negative carrierdensity dependence. In this carrier density region, a pump-dump process occurs for negatively chirped pump, which leads to more uniform distribution of carriers. Decrease of signal intensity at (j)2 == 0 suggests that there is a contribution from instantaneous two-photon absorption, which is dependent only on the pump pulse width. In conclusion, we have observed chirp-controlled carrier creation in GaAs. We observed a chirp dependence of pump-probe signal intensities of a thick GaAs sample which is opposite to that observed previously with organic molecules. This result can be explained by considering a pump-dump process and bandgap renormalization in a thick sample.

References 1 G. Cerullo, C. J. Bardeen, Q. Wang, and C. V. Shank, Chem. Phys. Lett. 262, 362, 1996. 2 K. Misawa and T. Kobayashi, J. Chem. Phys. 113, 7546 2000. 3 J. Kunde, U. Siegner, S. Arlt, F. Morier-Genoud, and U. Keller, Appl. Phys. Lett. 73, 3025 1998.

339

Ultrafast insulator-to-metal switching by photoinduced Mott transition S. Iwai^'^ Y. Okimoto^M. Ono^ H. Matsuzaki^ A. Maeda^ H. Kishida^ ^ H. Okamoto^ ^ Y. Tokura^'^ ^ Tohoku University, Sendai, 980-8578, Japan, e-mail [email protected] ^PRESTO-JST,Kawaguchi, 332-0012 Japan ^Correlated Electron Research Center (CERC), AIST,Tsukuba, 305-0046, Japan "^ University of Tokyo, Tokyo, 113-0033, Japan Abstract. We demonstrated the ultrafast photoinduced Mott transition from a charge trmisfer (CT) insulator to a metal in a halogen-bridged Ni chain compound by visible-mid IR pump-probe reflection spectroscopy. Upon the excitation of the CT band^ the spectral weight of the gap transition is transferred to the inner-gap region. When the excitation density exceeds 0.1 per Ni site, the Drude-like high-reflection band appears in the infrared region^ indicating the formation of a metallic state. The photogeneration of the metallic state and the subsequent recovery of the original insulating state occur within a few picoseconds.

1.

Introduction

Since the discovery of high-Tc superconductivity in perovskite cuprates and colossal magnetoresistance in manganites, the doping-induced insulator-metal (IM) transition in Mott insulators (or equivalently, the Mott transition) has attracted much attention[l]. The ultrafast photoinduced Mott transition offers a new approach to high-speed optoelectronic devices for Tb/s operation. From the viewpoint of basic research for correlated electron materials, photodoping can change a filling of 3 ^ band without changmg the other parameters. The observation of a hidden quasistable phase by a transient measurement will promote the understanding of the correlated electron system near the I-M phase boundary. The halogen-bridged Ni compound |Ni(ctDai)2Br]Br2 is well knovm as a typical half-filled Mott(CT) insulator. The Ni ^ and Br" ions line up alternately along the b-axis[2]. The four N atoms of the two ligand units (chxn: cyclohexanediamine) coordinating a Ni ion produce such a strong ligand field that the Ni^^ ion is in a low spin state {(£ :S=l/2) with an unpaired electron in the d^ orbital. The ID electronic state is formed via the hybridization of the Ni d^ and Br/?^ orbitals. The large on-site Coulomb energy U of Ni 3 ^ electrons opens a gap between the Ni 3 J upper Hubbard (UH) band and the lower Hubbard (LH) band. The occupied Br 4/? band positions itself within the Mott-Hubbard gap, and hence the lowest-energy electronic excitation is a charge transfer (CT) transition from the Br 4/? valence band to the Ni 3c/UH band [3-5].

340

2. Results and Discussion In Fig. 1(a), the polarized reflectivity spectrum is represented by a gray line. A sharp peak evident at 1.3 eV is due to the CT gap transition. The transient reflectivity (TR) spectra observed at delay time td after the photouradiation are represented by dots and lines in the same figure [6]. The intensity of the irradiated

^p*=0-5

t^

(a)

^^=0.12

200h

0.2 0.4 1 Photon Eiiergy[ey|

Fig. 1. (a) Transient reflectivity and (b) optical conductivity a(o)) spectra at room temperature.

6.2X10^ (x5)

^^i=0.12(x2) rr-°^^t^

/^=o.lps-

I I /5

0.5 1 Photon Enei^[ey]

Fig. 2 a(o)) afl:er the photoexcitation {td =0.1) for various excitation densities Xph .

light was 3.6 mJ/cm^. Under this condition, the average excitation density Xph of the absorbed photon is 0.5 per Ni site within the absorption depth (460 A), as evaluated by taking account of the reflection loss (30 %) and the imit cell volume (8.68x10'^^ cm'^). Immediately afl;er the photoirradiation {td = 0 . 1 ps), the reflectivity in the mid-IR region increases markedly, being reminiscent of the Drude-like response. The optical conductivity a(o)) was obtained by applying the Kramers-Kronig transformation (KKT) to the reflectivity spectrum. a(a)) at ^^=0.1 ps monotonically increases upon lowering the probe photon energy to 0.12 eV, suggesting the closure of the optical gap. To clarify the photoinduced change of the electronic state in detail, we have investigated the excitation density Xph dependence. For the weak excitation oixph= 6.2x10"'^, a midgap absorption \vith a peak around 0.5 eV is observed in the a(ci)) spectrum as shown in Fig.2. As Xph increases, the low-energy part of a(o)) below 0.2 eV increases significantly, and for jc^^> 0.1, the optical gap seems to disappear. To investigate the evolution of the photoinduced Mott transition as well as to ascertain the validity of the overall analysis of the TR spectra using KKT, it is usefiil to examine the transfer of the spectral weight fi*om the CT gap region to the inner-gap region. The spectral weight can be quantitatively analyzed in terms of the effective number of electrons Neff((o), defined by equation (1). Here, mo is the fi*ee electron mass and N the number of Ni atoms per unit volume [1]. Neffio) is the measure for the kinetic energy of electrons on an energy scale of electrons on an energy scale of ho).

341

2m,

^A-)-^l<^i-'y^-'-

(1)

Nevertheless, the observed systematic changes of ANeffiai) and the approximate holding of the sum rule over the wide range of Xph ensure that the analysis using KKT presented here clearly reflects the photoinduced changes of the electronic state. In Fig.3, we plot ANeff(l eV) at td= 0.1 ps as a function of x^/,. ANeffil.O eV), which represents the total spectral weight transferred from the CT pon, increases with a slope i=d{AN^^)/dx ;,) larger than 1. ANeff with Xph has been observed for the chemical doping Dnsidered to be a characteristic feature of the system with >n[l,7]. hmg from an insulator to a metal has been successfiilly 'T insulator of the halogen-bridged Ni-chain compound, tation density changes the photoproduct from the midgap e. This is the first observation of the photoinduced Mott Drrelated electron system. We also succeeded in realizeing nsition in perovskite cobalt oxide LaCoOs. An ultrafast Fig.3 ANeffil.O QW) at td=0.l rise and sub-ps decay ) is sunilarly observed in LaCoOs. ps as a function of the ^^ transition in 3d metal oxides has been extensively photodopihg concentration ^^..^^j.^j^ ^^^.^^^ ^j^^ enormous amount of spectral data avaiiaoie is useiui lur uiterpreting the photoinduced spectral change. Furthermore, various attempts to achieve the fabrication of thin films and super structures have been made, with the aim of realizing correlated electron devices. The photoinduced Mott transition in 3d metal oxides will offers new possibilities for ultrafast I-M switching devices. References [1] M. Imada, A. Fujimori, Y. Tokura,"Metal-insulator transitions," Rev. Mod. Phys. 70,1039(1998). [2] M. Yamashita, T. Manabe, K. Inoue, T. Kawashima, H. Okamoto, H. Kitagawa, T. Mitani, K. Toriumi, H. Miyamae, R. Ikeda, Inog.Chem. ,38; 1894(1999). [3] K. Toriumi, Y. Wada, T. Mitani, S. Bandow, M. Yamashita, and Y. Fujii, J. Am. Chem. Soc. I l l , 2341 (1989). [4] H. Okamoto, Y. Shimada,Y. Oka, A. Chainani, T. Takahashi, H. Kitagawa, T. Mitani, K. Toriumi, K. Inoue, T. Manabe, M. Yamashita, Phys. Rev. B54, 8483(1996). [5] H. Kishida, H. Matsuzaki, H. Okamoto, T. Manabe, M. Yamashita, Y. Taguchi, Y. Tokura, Nature 405,929 (2000). [6] S.Iwai, M. Ono, A. Maeda, H. Matsuzaki, H. Kishida, H. Okamoto,Y. Tokura, Phys. Rev. Lett. 91,057401(2003). [7] H. Eskes, A. M. Oles, M. B. J. Meinders, W. Stephan, Phys. Rev. B. 50,17980(1994).

342

Femtosecond Near Edge X-ray Absorption Measurement of the VO2 Phase Transition. A. Cavalleri\ H.H.W. Chong^^^ S. Fourmaux^'^ T.E. Glover^^^ P.A.Heimannn^^\ J.C. Kieffer^'^ H.A. Padmore^^\ R.W. Schoenlein^^\ Abstract: We measure the insulator-to-metal transition in VO2 using femtosecond Near-Edge X-ray Absorption. Sliced pulses of synchrotron radiation are used to detect the photo-induced dynamics at the 516-eV Vanadium Lgedge. Email: [email protected] Time-resolved spectroscopy can probe new physical pathways of phase transitions and reveal fundamental aspects that are hidden in time-integrated measurements. However, limited quantitative information about electronic structure can be extracted from measurements at visible wavelengths, motivating interest in shortpulse x-ray spectroscopy.

V 2p3/2 E c c a>

01s

/

. 400 K -300K

/ y

p.

400 K 300 K

£ o O c o

o

0.02

514

516

530

532

0.01

in

< 0.00

/ 515

V 2p3/2

\f 2pi;2

525 Energy (eV)

0 1s

1 535

Figure 1, Static, VO2 x-ray absorption spectra in the region around the V L23 and the O K edge. Spectral changes at the V Ledge and the O K edge observed above 340 K

Near-edge X-ray absorption spectroscopy techniques probe unoccupied valence states by measuring transitions from core levels, rather than from extended occupied valence states as in visible spectroscopy. Element specificity, symmetry selection rules and linear/circular dichroic effects are some of its most powerful aspects. Due to the stringent tunability requirements on the source, time-resolved spectroscopy with soft x-rays is ideally performed using synchrotron radiation. Novel laser e-electron beam interaction schemes provide schemes to access the subpicosecond time domain. Here, we report on picosecond and femtosecond soft x-ray absorption measurements of the insulator-to-metal transition in photo-excited VO2. a non-magnetic compound that undergoes a transition between a monoclinic

343

insulator and a rutile metal when heated above 340 K. Previous ultrafast optical and x-ray diffraction experiments [3] on the photo-induced transition in VO2 show that changes in both atomic and electronic structure occur on the sub-picosecond timescale, where their detailed relationship is yet to be fully understood. The experiments discussed here are performed either with the full 70-ps pulse of the Advanced Light Source [1] or exploiting laser modulation of electron bunches [2]. Picosecond XAS measurements: Time-resolved XAS experiments were performed using bend magnet radiation at beamline 5.3.1 of the Advanced Light Source. 70-ps x-ray pulses were radiated by a bend magnet and focused onto the VO2 sample using a toroidally bent silicon mirror, which imaged the e-beam into the x-ray hutch. A fraction of the x-ray pulses, radiated once every 656 ns (roundtrip time of the storage ring), were used for our experiments at 1-KHz repetition rate. A flat-field imaging spectrometer was used to disperse the transmitted soft x-rays after the sample, generating spectra in the range between 100 eV and 800 eV, with a resolution of approximately 4 eV at 500 eV.

-0.04i -200

0

200

400

600

Delay (ps) Figure 2: Time-resolved transmission changes at the Vanadium L3 edge.

Differential transmission measurements at the Vanadium L3 edge show the expected decrease in the transmission of the sample, due to the photo-induced collapse of the bandgap. The data in figure 2 are fit with a fast response corresponding to DT/Ts:l%, followed by a slower decrease at a rate of approximately 1%/ns. Since the thickness of the transformed layer scales linearly with the signal, we estimate that a fast transformation over approximately the first 50-nm occurs within the xray probe pulse, followed by slower thermal growth at 40 m/sec. Femtosecond XAS measurements: Femtosecond XAS was performed using the same apparatus described above, combined with laser modulation of electron bunches in the wiggler of straight section 5 of the storage ring. A pair of slits was placed in the x-ray hutch at the image plane of the storage ring, immediately before the VO2 windows, effectively isolating the image of the femtosecond x-ray source in the spatial wings of the beam. The flux in the femtosecond x-rays was 344

approximately four orders of magnitude lower than in the picosecond pulse, amounting to a few thousand photons/(sec 1% BW) at 500 eV in a 1-KHz train. Sample transmission (30%), spectrometer efficiency (6%) and detector efficiency (50%) resulted in about 10 photons/sec in the 0.5% bandwidth where the experiment was performed.

delay (ps) Figure 3: Femtosecond dynamics of the as measured at the V L3 edge. The experiments probe unoccupied d states after excitation, resulting in increased absorption.

The femtosecond, time-resolved response is reported in figure 3, where a prompt drop in the transmission of the sample can be observed immediately after laser excitation. This drop is interpreted as the effect of holes created in the valence bands of d symmetry, while electrons are promoted to the conduction band, formed by orbitals of mixed p-d character and therefore less evident at the V L edges. Thus Ledge measurements are more sensitive to the holes of purely d symmetry (causing over absorption) than to the electrons in the conduction band. Future measurements will probe also the conduction band at the Oxygen Is edge. In summary, we report the first soft X-ray absorption measurement with femtosecond resolution, demonstrated for the photo-induced insulator-to-metal transition in the strongly correlated compound VO2. In the picosecond measurements we detect a shift of the Vanadium L3 edge, associated with the collapse of the bandgap. We also report on the first femtosecond soft XAS measurement, which reveal the early photo-doping process through a drop in transmission at the VL3 edge. This work was supported by the Director, Office of Science, Office of Basic Energy Sciences, Division of Materials Sciences, of the U.S. Department of Energy under Contract No. DE-AC03-76SF00098. [1] A. Cavalleri et al. Phys Rev. B 69, 153106 (2004). [2] R.W. Schoenlein et al. Science 287, 2237 (2000) [3] A. Cavalleri et al. Phys. Rev. Lett. 87, 237401 (2001).

345

Phase Transition in strongly-correlated VO2: Time-domain Assignment of Cause and Effect. A.Cavalleri^^^*, Th. Dekorsy^^\ H.H. Chong^^^ J.C. Kieffer^^\ R.W. Schoenlein'(1) ^^^Materials Sciences Division, Lawrence Berkeley National Laboratory. ^^^Forschungszentrum Rossendorf, Dresden Germany. ^^^Universite du Quebec, INRS energie et materiaux, Varennes, Quebec. E-mail: [email protected] Abstract. We establish time-domain hierarchy between structural and electronic effects in the strongly correlated system VOj. The insulator-to-metal transition is driven directly by structural change rather than by electron-electron correlations. We study VO2, a strongly-correlated compound that exhibits cell doubling in "concomitance" with a metal-insulator transition below 340 K [1]. Previous ultrafast optical [2] and x-ray diffraction [3] experiments demonstrated that hole photo-doping into the correlated valence band of the low-T insulator causes an ultrafast transition in both electronic properties and atomic structural arrangement.

0.50 'ins

c o '55 tn

0.48 ^ ^

'met

0.46

0.190

E

ui c (0

H

^met

0.180 0

1000

2000

3000

Delay (fs) Figure 1: Time-resolved reflectivity/transmission evolution of VO2 / Si3N4 structures after 100-fs excitation pulses at 790-nm wavelength and 50 mJ/cm^.

346

We now report evidence of a 75-fs, bottleneck timescale for the formation of the metallic phase, evidence for a structurally driven mechanism and a reverse-Peierls transition. Time-resolved optical spectroscopy was performed at several wavelengths in the visible. An insulator-to-metal transition is evidenced by increase of the reflectivity and decrease in the transmission (figure 1). The fundamental transition time was then measured as a function of pulse duration in the range between 1.5 ps and 15 fs. A bottleneck timescale of 75 fs appeared, even in the cases where excitation was complete on the 10-fs timescale (figure 2).

-0.1 0 5001000 Delay (fs)

0 200 Delay (fs)

J\

^

'o)

103

•^


E •!-• c o

X •^ 102

*

: I

:T


- » >»

s

=

•

jut

1

Slruclutal Bollleneck

^ 1

X

• •

1ini

Z

10^

1

•

Electronic response time •

102

•

103

FWHM pulse width (fs) Figure 2: Pump-probe reflectivity experiments of the phase transition for different pulse durations.

The nature of the bottleneck timescale was further investigated using Raman and coherent phonon spectroscopy. These experiments revealed that photo-excitation results in coherent structural motion involving modes of Ag symmetry. Remarkably, the bottleneck timescale for the phase transition observed at higher fluence corresponds approximately to half period of the two coherent modes. Analysis of the real-space motion of atoms reveals that the coherently excited modes project the atomic arrangements of the two crystallographic phases onto one another. These modes have been identified as the order parameter for the transition.

347

X

tr cc < 2 4 delay (ps)

4 6 Frequency (THz)

^^

A

o 1 ^

X

-1

cc cc

\ \ \ \ x-"'"'^^^'^^^

3 -3 1

1..

—

—

1

—

Delay (ps) Figure 3: Time-resolved evolution of the reflectivity in the low-T phase. The coherently excited modes are symmetric Ag modes with displacements that project onto the high-T crystallographic phase, as revealed by comparison with the CW Raman spectrum.

In summary, we have shown that ultrafast spectroscopy on the sub-vibrational timescale can be applied to resolve ambiguous cause-effect assignments across phase transitions in strongly correlated electron systems. Based on the ultrafast response of to photo-excitation, we conclude that that the atomic arrangement of the high-T unit cell is necessary for the formation of the metallic phase of VO2, even if the correlated d band is highly depleted (hole-doped). This result is suggestive of significant Band-like character for the low-T insulator. A highly controversial issue on the nature of the insulating phase in this prototypical, correlated system is settled by means of time-resolved spectroscopy. This work was supported by the Director, Office of Science, Office of Basic Energy Sciences, Division of Materials Sciences, of the U.S. Department of Energy under Contract No. DE-AC03-76SF00098 [1] F.J. Morin Phys. Rev. Lett. 3, 34 (1959). [2] M.F. Becker et al. AppL Phys. Lett. 65, 1507 (1994). [3]A. Cavalleri et al. Phys. Rev. Lett. 87, 237401 (2001).

348

Polarization-dependent phenomenon induced by the interaction between focused femtosecond laser and transparent materials Yasuhiko Shimotsuma\ Jiarong Qiu^, Peter G. Kazansky^, and Kazuyuki Hirao^ ^ Department of Material Chemistry, Graduate School of Engineering, Kyoto University, Kyotodaigaku-Katsura, Nishikyo-ku, 615-8510, Japan E-mail: 3iMunKMlJ^yai3i^^ and hnio(igteccLUlkliy<^ ^ Photon Craft Project, Japan Science and Technology Agency, Keihanna-Plaza, Kyoto 619-0237, Japan E-mail: iru(^|li)llotonjSL^^ ^ Optoelectronics Research Centre, University of Southampton, SO 17 IB J, United Kingdom E-mail: i)gtoim:c,sgtoxijc,u^ Abstract. Polarization-dependent phenomenon is observed inside transparent materials after irradiation by a focused single femtosecond laser beam. The phenomenon is interrupted in terms of the interaction of interference between incident light field and electron plasma wave.

1.

Introduction

Interaction of the high powder ultrashort pulse laser with matter attracts a great deal of interest in physics and technology. In particular, since the discovery of the laser, many studies have been carried out on the interaction of intense ultrashort pulse laser radiation with transparent materials, which is a key material in optical communication technology [1-6]. The ultrashort pulse light-matter interaction gives rise to the various new phenomena, such as high-order nonlinear absorption, structural change, and permanent refractive index modification. Although the process of energy absorption is now well understood [7], little known fact is that the actual structural change in focal spot [8-11]. Here we report the first observation of the emergence of periodic structurally changed regions of nanometer size mside silica glass after irradiation by just single femtosecond laser beam. This phenomenon is interpreted in terms of interference between the incident light field and the electric field of bulk electron plasma wave, resulting in periodic modulation of electron plasma concentration and permanent structural changes in glass.

2.

Experimental Methods

A regeneratively amplified mode-locked Ti: sapphire laser (150 fs pulse duration, 200 kHz repetition rate) operating at X = 800 nm was focused via lOOx (numerical

349

aperture == 0.95) microscope objective into a silica glass samples placed on the XYZ piezo-translation stage. The silica glass samples were commercially available synthetic silica glass, ED-H (Tosoh Quartz Corp., brands of wet silica with OH ~ 50 ppm) of 10 mm x 10 mm x 2 mm size. The laser beam was focused at ~ 100 ^m below the surface and the beam waist diameter was estimated at ~ 1 |xm. The laser irradiation parameters were controlled by an electronic shutter, a variable neutral density filter, and a half-wave plate placed in an optical path of the laser. After laser irradiation, the sample was observed optical microscope. The optical micrograph indicates by a void formation or refractive index change in an irradiated focal spot. Then we analyzed the surface which was polished to the focal spot location by scanning electron microscope and Auger electron spectroscope.

3.

Results and Discussion

Fig. 1 shows the comparison between secondary electron images (SEI) and backscattermg electron images (BEI) of the same surface. The SEI of the polished surface indicates that the morphology of an irradiated focal spot regions almost dose not change, namely, a void dose not exist. On the other hand, BEI reveal a periodic structure of stripe-like dark regions with low density of material and of ~ 20 nm width which are aligned perpendicular to the writing laser polarization direction.

5 ^i'^.^^^^'rix.

.

Fig. 1 Secondary electron images (SEI) and backscattering electron images (BEI) of silica glass surface polished close to the depth of focal spot. Furthermore, we carried out Auger spectra mapping of silicon and oxygen on the same surface (Fig. 2). The Auger signal of the oxygen in the region corresponding to dark domains in the BEI is lower compared to other regions. This indicates low oxygen concentration in these domains (Fig. 2a). On the other hand, the signal intensity of the silicon is the same in the whole imaged region (Fig. 2b).These results indicate that the periodic nanostructure consists of the periodic

350

arrangement of oxygen defect region (Si02.jc) was formed within the focal spot. The value of x was estimated at about 0.4 from the Auger signal intensity. We observed the decrease of the grating period with an increase of the number of light pulses for fixed pulse energy of 1 jiJ. This indicates a logarithmic dependence of the grating period A on the number of pulses. The dependence of the periodic nanostructures on pulse energy for fixed the number of pulses was also investigated. An increase of the period with the pulse energy was observed (Fig. 3).

Fig. 2 Auger spectra maps of oxygen (a) and silicon (b) on the same polished surface of silica glass. We proposed as follows the mechanism of oxygen defect periodic nanostructure formation by using a just single beam of ultrashort pulse laser. Once high free electron plasma is produced by multiphoton ionization, the laser energy will be absorbed by the electron plasma. Since the light absorption electron plasma is generated inside bulk, it will excite longitudinal waves. The electric field of a longitudinal plasma wave is parallel to the propagation direction. Such electron plasma wave could interact with the incident light wave only if it propagates in the plane of light polarization. The interaction creates a periodic pattern of electron plasma concentration by the interference between the incident light field and the electric field of the electron plasma wave, consequently the periodic modulation of oxygen defects will be induced within the glass samples. We tried to verify the periodic nanostructure formation mechanism by the interference according to theoretical calculation. The period of the nanostructure is defined by the following momentum conservation condition, k "^ grating

=k

—k

'^plasma

** photon

where A:gi.ating i=2K/A) is the modulation vector of the grating period, A:photon (^<^photon^/<^: n is the refractive index, and c is speed of light) is the wave vector of mcident light, and ^piasma is the wave vector of the electron plasma wave. The dispersion relation of the electron plasma wave follows,

^plasma

"*"

'^plasma

351

where n^ is the electron density, e is the electron charge, /Wg is the electron mass, TQ is the electron temperature, and KB is Boltzmann constant. From above equation, the period of nanostructure (A) could be expressed as follows. 2

A

0

y

'^photon

m^Q),photon UTT

KT,

This equation shows that the period of nanostructure depends on the electron density and the electron temperature, the nano-grating period increases with the increase of electron density and electron temperature. We compared the theory to experiment by this dependence (Fig. 3), resulting the theoretical calculation and the experimental results were relatively well in agreement. Pulse energy (|LIJ) 1 2

3

400

300

200

73 O

100

Fig. 3 The dependence of the period of nanostructure on electron temperature (T^) for electron density n^ = 1.7 X 10^^ cm"^. The comparison of theoretical calculation and experimental data is also shown.

4.

Conclusions

In conclusion, we have observed phenomenon as the first direct evidence of interference between light and electron plasma waves. Furthermore, the observed nanostructures are the smallest embedded nano-grating fabricated by using laser light.

352

References 1 K. M. Davis, K. Miura, N. Sugimoto, and K. Hirao, "Writing waveguides in glass with a femtosecond laser". Opt. Lett., 21, 1729-1731 (1996). 2 K. Miura, J. Qiu, H. Inoue, T. Mitsuyu, and K. Hirao, "Photowritten optical waveguides in various glasses with ultrashort pulse laser", Appl. Phys. Lett., 71, 3329-3331(1997). 3 Y. Kondo, K. Nouchi, T. Mitsuyu, M. Watanabe, P. G. Kazansky, and K. Hirao, "Fabrication of long-period fiber gratings by focused irradiation of infrared femtosecond laser pulses". Opt. Lett., 24, 646-648 (1999). 4 E. Fertein, C. Przygodzki, H. Delbarre, A. Hidayat, M. Douay, and P. Niay, "Refractive-index changes of standard telecommunication fiber through exposure to femtosecond laser pulses at 810 cm", Appl. Opt, 40, 3506-3508 (2001). 5 H. - B . Sun, S. Matsuo, and H. Misawa, "Microfabrication and characteristics of two-dimensional photonic crystal structures in vitreous silica". Opt. Rev., 6, 396398(1999). 6 H. - B . Sun, Y. Xu, S. Juodkazis, K. Sun, M. Watanabe, S. Matsuo, H. Misawa, and J. Nishii, "Arbitrary-lattice photonic crystals created by muhiphoton microfabrication", Opt. Lett., 26, 325-327 (2001). 7 C. B. Schaffer, A. Brodeur, and E. Mazur, "Laser-induced breakdown and damage in bulk transparent materials induced by tightly focused femtosecond laser pulses", Meas. Sci. TechnoL, 12, 1784-1794 (2001). 8 P. G. Kazansky, H. Inoue, T. Mitsuyu, K. Miura, J. Qiu, K. Hirao, and F. Starrost, "Anomalous Anisotropic Light Scattering in Ge-Doped Silica Glass", Phys. Rev. Lett., 82, 2199-2202 (1999). 9 J. Qiu, P. G. Kazansky, J. Si, K. Miura, T. Mitsuyu, K. Hirao, and A. Gaeta, "Memorized polarization-dependent light scattering in rare-earth-ion-doped glass", Appl. Phys. Lett., 77, 1940-1942 (2000). 10 L. Sudrie, M. Franco, B. Prade, and A. Mysyrowicz, "Writing of permanent birefringent microlayers in bulk fused silica with femtosecond laser pulses". Opt. Commun., 171, 279-284 (1999). 11 J. D. Mills, P. G. Kazansky, E. Bricchi, and J. J. Baumberg, "Embedded anisotropic microreflectors by femtosecond-laser nanomachining", Appl. Phys. Lett., 81, 196-198 (2002).

353

Investigation on the parameters of dense electronic plasma induced by femtosecond laser in fused silica Qihuang Gong, Quan Sun, Yi Liu, Zhaoxin Wu, Hong Yang and Hongbing Jiang State Key Laboratory for Mesoscopic Physics & Department of physics, Peking University, Beijing, 100871, P.R.China E-mail: [email protected] Abstract. Electron plasma induced by femtosecond laser in fused silica was investigated by pump-probe time-resolved shadow and interferometric imaging. Electron collision time is obtained as 1.7 fs at density of 4.2x1 O^^cm'^ and the plasma lifetime is 170fs.

The interaction and propagation of a focused femtosecond laser in transparent dielectric material have attracted great attention recently. Many interesting phenomena, such as pulse splitting [1], supercontinum generation [2], soliton generation [3], and pulse compression [4] were found to be associated with the femtosecond laser induced electron plasma, which is highly depended on the nonlinear ionization process [5]. The electronic collision time T in conduct band is an important parameter for exploration the underlying mechanisms. Amod et.al pointed out theoretically that the electron momentum relaxation rates in Si02 thin film is energy dependent. The corresponding collision time T varied from 10'^"^ to 10"^^ s [6,7]. While, this electron collision time T has not been measured in experiment using ultrafast laser pulse. Furthermore, different value of x was employed in different calculation, such as 23.3 fs by Sudrie et al [8] and 0.2 fs by Mao et al [9]. Meanwhile, the density of electron plasma (iie) induced by femtosecond laser pulse is also another essential issue to explore the mechanism at the fundamental level. Ml Z .

Delay line Polarizer

Sample

WoUaston Analyzer CCD prism

Fig. 1 experiment setup for time-resolved shadowgraph measurement.

354

In this report, we combined the shadow measurement with the interferometric fringe measurement [10] to diagnose the generated electronic plasma. The time-resolved shadowgraph measurement diagram is shown in Fig.l. A shutter was used to obtain a single pulse. The laser beam was split into pump and probe pulses by a beam splitter. The objective A (Olympus, 4x, 12-mm working distance) with a numerical aperture (NA) of 0.16 was used to focus the pump laser beam. The sample could be transferred perpendicular to the pump beam and probe beam to avoid multi-interaction. The interaction region was inside the sample about several hundred |Lim from the front surface. The probe beam passed through an optical delay stage and a polarizer to illuminate the interaction region perpendicularly to the pump beam, and was imaged onto a charge-couple-device (CCD) by the objective B (20x). For the interferometric fringe measurement, a Wollaston prism with its optical axes 45^ to the polarization direction of incident probe beam is utilized after the object B to split the probe beam into two beams with 2 degrees separation. Then the two beams pass an analyzer with its polarization direction parallel to the polarizer to form the interference fringes. 2mts 4mfs

Fig.2 Time-resolved shadowgraphs for femtosecond laser-induced plasma in fused silica. The intensity is close to 7x lO^^W/cm^ in focal region.

Amis

Fig.3 Interferometric image for femtosecond laser-induced plasma in fused silica at 400fs delay time. Figure 2 show four shadowgraphs measured at different delay time, with the laser intensity is close to 7xlO^^W/cm^. Fig.3 shows the interferometric fringe image at the delay time of 400fs. Then, using the probe beam intensity transmittance Ipo/Ipd ( Ipo and Ipd represent the probe laser intensities before and

355

after passing through the plasma region) obtained jfrom the shadowgraphs and the stripe shift number D from the interferometric fringe image at the same delay time and the same position, the electron collision time T is determined by l n ( / , o / / , . ) 1 + {wry

^ ^ ^

^j^

And the the electron plasma density Ug is estimated by ^ ^ MIpp/^p,)

(2)

^ (TI Finally, the electron collision time is observed to be density dependence due to different electron distribution in conduct band. Four T values were estimated to be 1.3fs, 1.4fs, 1.7fs, 2.5fs at different time delay at z = 90|Lim, corresponding to the electron density of 6.2 x 10^^ cm"^ 4.6 x lO^^cm"^ 4.2 x 10^^ cm•^ and 2.8 x 10^^ cm'^ respectively. In addition, the electron plasma life-time Xg was also measured to be IVOfs in fused silica. It is in good agreement with the measurement performed by Audebert et al [11], which has been cited widely.

Reference: 1 2 3

J. K. Ranka, R. W. Schirmer and A. L. Gaeta, Phys. Rev. Lett. 77, 3783, 1996. A. Brodeur and S L Chin, Phys. Rev Lett. 80,4406,1998. I. G. Koprinkov, A. Auda, R Wang and K Midorikawa, Phys. Rev Lett. 84, 3847, 2000. 4 Helene Ward, and Luc Berge, Phys. Rev Lett. 90, 053901, 2003. 5 C. B. Schaffer, A. Brodeur, and E. Mazur, Meas. Sci.and Technol. 12, 1784, 2001. 6 D. Amod, E. Cartier, and D. J. DiMariza, Phys Rev B, 45, 1477, 1992. 7 D. Amod, E. Cartier, and D. J. DiMariza, Phys Rev B, 49, 10278, 1994. 8 L. Sudrie, A. Couairon, M. Franco, B. Lamouroux, B. Prade, S. Tzortzakis, A. Mysyrowicz, Phys. Rev Lett. 89, 186601, 2002. 9 X. Mao, S. Mao, R. Russo, Appl. Phys. Lett. 82, 697, 2003. 10 H. Yang, J. Zhang, and Y. Li, et. al, Phys. Rev E. 66, 016406, 2002. 11 R Audebert, Ph. Daguzan, and A. Santos, et al., Phys. Rev Lett. 73, 1990, 1994.

356

Dynamical Symmetry Breaking induced by Ultrashort Laser Pulses in KTaOs Eiichi Matsubara\ Jun-ichi Takahashi\ Kuon Inoue^'^, and Eiichi Hanamura^'^ ' Japan Science and Technology Agency(CREST) ~ Chitose Institute of Science and Technology, 758-65, Bibi, Chitose, Hokkaido, JAPAN E-mail: [email protected] Abstract. Raman selection rule is observed to be violated under resonant and coherent pumping of Raman-active two-phonons at the Brilloiun zone (BZ) edge in KTaOs by two near-infrared femtosecond laser pulses. Here the multistep coherent Anti-Stokes Raman scattering signals due to originally Raman-inactive single phonons at the BZ edge become observable. The model of BZ folding is presented to explain all these phenomena. The crystal KTaOs has the highest symmetry O^ and all F-point phonons are of odd mode so that they are all Raman-inactive. However, two-phonon Raman spectrum has been observed to show sharp structure because the joint density-of-states for this process diverges at the BZ-edge points. A (001 )-oriented KTaOs single crystal is irradiated by two near-infrared femtosecond laser pulses with frequencies coi and oh obtained from an Optical Parametric Amplifier (OPA) pumped by a regeneratively amplified mode-locked Ti: sapphire laser. When the frequency difference Aco ^ 0)1-0)2'^^ nearly equal to the energy of Raman-allowed two-phonon excitation, the conventional Anti-Stokes Raman Scattering (CARS) is observed at coi + Aco. If we increase the power of coi beyond a critical value, the Raman selection rule in O^ symmetry is observed to be temporally violated as shown in Fig. 1, Fig. 2, and Fig. 3. The angular distributions of these signals also give important information on the present multi-photon processes. Let the two incident pulses with the frequency coj and CO2 have the wavevector k\ and kj, respectively. Then the «-th step CARS signal should be observed in the direction kx + n {kx - k'^. In these figures, however, this relation is also violated. Under nearly resonant pumping of the two-phonon combination of TO2 (JG, 211 cm~^) at the -X point and TO4 {X5, 527 cm~^) at the X-point and vice versa, i.e., by using 770 cm" , the mulfi-step CARS signals with the frequency spacing of 527 cm~^ have been also observed in the direction of larger angles in the visible region as shown in Fig. 2. Under pumping of 2LO2 mode at the X-points by using A(o= 900 cm~\ both the multi-step CARS signals with 527 cm~^ spacing of JG (TO4) phonon and those with 443 cm"^ spacing ofXi (LO2) phonon are observed as shown in Fig. 3. The selection rule of Raman scattering is clearly broken because X-point phonons have become observable.

357

1st CARS+T0,

0 7000

7500

1+2TO,

8000

Wavenumber [cm

8500

]

Fig. 1. The angular dependence of the first CARS and the single-phonon signals under nearly resonant pumping of two-phonons TO4 (527 cm"\ X5) and TO2 (211 cm"\ X5) at the X-point by using Aco scoj- 0)2= 749 cm"^ at room temperature with coj = 6649 cm"^ (1504 nm, 5.0 ^J/pulse) and 0)2= 5900 cm"^(1695 nm, 4.8 jxJ/pulse). The angle between the CO]- and the CO2- beams is 7.0 degrees in the air.

Under nearly resonant pumping of two-phonon combination modes which consist of 2LO2 (^1) or TO4 + TO2 (both X5) at the X-point, two-mode phonons at the X- and - X point are coherently driven and the dynamical grating of these phonons is formed. As a consequence, the unit-cell size is doubled in the X-direction and the X-point in the BZ is folded onto the F-point. Therefore, Xi and X5 phonons at the X-point are reduced to Ai^ and E^ modes, respectively, at the F-point of Z)^/, symmetry. As a result, both modes become Raman-active and the n-th step CARS signal cOn is inelastically scattered by these dynamical gratings, i.e., by absorbing and emitting these single phonons which have been originally Raman-forbidden. These scattered signals are observed with the frequency spacing of the originally Raman-inactive single phonon as shown in Fig. 1, Fig. 2 and Fig. 3. The angular and frequency distribution of these peak signals of single-phonon CARS spectra can be well explained by the present model of combining the conventional CARS and scattering processes by the dynamical gratings of phonons. Here these phonons have been Raman-activated by the dynamical symmetry breaking of the crystal due to the resonant and coherent excitation of the BZ edge phonons by two incident ultrashort laser pulses. In addition, we have confirmed the existence of quasi-thereshold of pumping power Pi with the fixed P2 for this dynamical symmetry breaking.

358

45 40 '8' 35

I 30 < 20 15

52:

1

S rn^^H wlvM—J

^^^m i—i

10 12000

14000

16000

18000

Wavenumber [cm'^]

Fig. 2. The angular dependence of the multi-step CARS spectra due to the originally Raman forbidden Xs (TO4) phonon with the frequency spacing 527 cm~^ by using Aco =770cm~^ .The angle between the coj- and the (02- beams is 5.5 degrees in the air.

35

1

527 cm~^

jj 443 cm "J

•

2 30

I^PM

o

"N

l^"

S 25 L 20

V

15 12000

/f wJf^

#

\sL '

•

14000

16000

18000

Wavenumber [cm"^]

Fig. 3. The multi-step CARS spectra and their angular dependence under nearly resonant pumping of the two-phonon mode 2LO2 (443 cm'\ Xi) at the X-point by using Aco = 900 cm"\ Note here the coexistence of two series of multi-step CARS signals with the equal frequency spacing 443 cm"^ of Xi and 527 cm"^ of A^5. The angle between the coj- and the C02- beams is 6.0 degrees in the air.

References 1 W. G. Nilsen and J. G. Skinner, J. Chem. Phys. 47, 1413 (1967).

359

Part V

Ultrafast Dynamics in Solution

Sub-20-fs study of energy relaxation in carotenoids in solution and inside light harvesting complexes Giulio Cerullo , Dario Polli , Guglielmo Lanzani , Hideki Hashimoto , and Richard CogdelP ' National Laboratory for Ultrafast and Ultraintense Optical Science - INFM, Dipartimento di Fisica, Politecnico di Milano , Piazza L. da Vinci 32, 20133 Milano, Italy E-mail: [email protected] ~ "Light and Control", PRESTO/JST, Department of Physics, Osaka City University, Osaka 558-8585, Japan ^ Division of Biochemistry and Molecular Biology, IBLS, University of Glasgow, Glasgow, UK Abstract. Using sub-20-fs pulses, tunable from the UV to the IR, we study the 82^81 internal conversion dynamics in several carotenoids. We resolve an intermediate excited state and directly determine the energy transfer efficiency to Bacteriochlorophylls.

1.

Introduction

Carotenoids are ubiquitous pigments in photosynthetic organisms, performing the essential functions of photoprotection and light harvesting [1]. They protect the photosynthetic apparatus by quenching both triplet excited (bacterio)chlorophyll ((B)Chl) and singlet oxygen. In addition, they serve as accessory light-harvesting pigments: light absorbed by carotenoids in the blue, green and yellow regions of the spectrum is transferred to (B)Chl, thereby making it available to drive photosynthesis. Traditionally, excited states dynamics in carotenoids have been described in terms of two low-lying excited singlet states called Si and S2 [2]. The transition from So (I'Ag") to S] (2^Ag") is one-photon forbidden for reasons of symmetry, so that the one-photon-allowed optical transition goes from So to S2 (I'BU^). After photoexcitation, S2 undergoes a rapid internal conversion (IC) process to Si, completed in a few hundreds fs [3]. An additional IC process from S| back to So occurs on the ps timescale. Study of the IC process in carotenoids is a prerequisite for the understanding of the primary events in photosynthesis, since energy transfer to (B)Chl occurs both from S2 and Si states. In this work, we study the S2->Si IC dynamics in carotenoids with conjugation length (number of conjugated double bonds) n ranging from n=5 to n=\5, both in solution and inside the light harvesting complex 2 (LH2) of purple photosynthetic bacteria. To this purpose, we use a specially developed spectroscopic system combining very high temporal resolution (pulse width less than 20 fs) and broad spectral tunability, from the visible to the near infrared (NIR) wavelength range. This allows us to simultaneously probe the dynamics of S2 and Si states and thus to obtain a complete picture of the IC process. The main results of these studies are

363

the detection of an intermediate state mediating the S2->Si IC process and the direct measurement of the energy transfer efficiency to the BChl.

2.

Experimental Methods

The setup developed for these experiments is shown in Fig. 1 and consists of three independent non-collinear optical parametric amplifiers (NOPAs) [4] and one frequency up-conversion stage. These devices are pumped by an amplified Ti:sapphire system (500-p,J, 150 fs, 1 kHz). The first parametric amplifier, NOPAl, generates 15-20 fs pulses with -30 THz bandwidth tunable in the visible wavelength range, which are used to excite the carotenoids in resonance. The pump pulses are tuned to the red So^'S2 absorption edge, in order to minimize vibronic relaxation effects. In order to pump shorter carotenoids, absorbing in the UV, the NOPAl pulses are frequency up-converted with 790 nm pulses in a thin BBO crystal, thus generating -25 fs pulses tunable in the 310-360 nm wavelength range. The second optical parametric amplifier, N0PA2, generates sub-10-fs pulses with spectrum spanning the 500-720 nm wavelength range, while the third, N0PA3, generates -15 fs pulses in the 830-1050 nm region. Pulse compression is achieved by chirped mirrors for NOPAl and N0PA2 [5] and by a fused-silica prism pair for N0PA3 and the UV pulses. The pulses from these sources are synchronized by delay lines and are used to probe the differential transmission change (AT/T) following excitation. 150 fs 790 nm 500 j.iJ IkHz

CPATiiSa system

Photodiode Interference Filter

BS

NOPAl 15-20 fs 520 nm

PMT <10fs 500-720 nm

'-15fs 830-1050 nm^'^

25 fs 310-360 nm

Up-conv Fig. 1. Scheme of the set-up used for the experiments. CPA = chirped pulse amplification; BS = beam splitter; Up-conv = frequency up-conversion stage; DL = delay line; PMT = photomultiplier; SFG = sum frequency generation crystal.

364

3.

Results and Discussion

We first study all-rra^^'-p-carotene (/7=11) in cyclohexane solution. Fig. 2a shows -AT/T spectra at three pump-probe delays, following band-edge excitation by a 15-fs pulse at 510 nm [6]. We observe the prompt appearance of a broad photoinduced absorption band (PA2), extending from 600 to 950 nm. This band quickly decays and is replaced, within -50 fs, by an other band (PAx) peaking at 980 nm; PAx decays within -500 fs with a kinetic matching the formation of a new band (PAi) at 565 nm. PAi is a well-known feature of carotenoids, assigned to Si->Sn absorption, while PA2 stems from the initially photoexcited S2 state. These data clearly indicate the presence of an extra excited state (which we call Sx state) between S2 and Si. We assign PA2 and PAx to transient absorption from the S2 and Sx states, respectively, and the rapid spectral evolution in the first 50 fs to the S2->Sx transition, followed by the Sx->Si transition within a few hundred femtoseconds. The presence of intermediate state of B^' symmetry had already been predicted theoretically [7] and indirectly confirmed experimentally: our studies allow the direct dynamic resolution of this state. Wavelength (nm) 1000 900 800

700

600

(a)

y PA V |2%

\

M^A2

yN

\ 1.4

1.6

1.8

2.0

2.2

0

100

200

300

Probe Delay (fs)

-75

0

75

150 225 300

Probe Delay (fs)

0

200

400

600

Probe Delay (fs)

2.4

Enerqv (eV)

Fig. 2. (a) -AT/T spectra in all-/ra«^-p-carotene at different time delays following photoexcitation by a 15-fs pulse, (b)-(d) -AT/T dynamics (triangles) of all-/ra/75-p-carotene (b), M15 (c) and neurosporene (d) at different selected probe wavelengths on a short timescale. Solid lines are fits obtained using the four-state model described in the text.

This scenario is confirmed by -AT/T dynamics at different probe wavelengths (Fig. 2b). The instantaneous rise of PA2 and its subsequent fast decay is seen at the wavelength of 850 nm; at 980 nm we see the delayed PAx absorption, with a risetime matching the PA2 decay. Finally, PAx decay at 980 and 850 nm corresponds to PAi risetime at 560 nm. PA] does not rise instantaneously after photoexcitation, but is delayed by « 50 fs, in agreement with our kinetic model. The presence of the intermediate state is a common feature of several carotenoids. In Fig. 2(c) and 2(d) we show -AT/T dynamics of Ml 5 (a longer (3carotene analogue with ^=15) and of neurosporene in=9), respectively. In both these carotenoids we once again observe in the NIR region the instantaneous rise of the PA2 band, followed by the delayed formation of the PAx signal (upper and

365

middle traces of Fig. 2), which subsequently turns into the delayed PAi band in the visible (lower traces in Fig. 2). These results confirm that the S2^'Si transition is a two-step process in carotenoids with ri>9. Solid lines in Fig. 2 are exponential fittings of the data points using a four-level scheme and convolution with the instrumental response of the system, given by the pump-probe cross-correlation. This enables us to determine the rate constants for all the S2->Sx->S|->So IC processes. Fig. 3(a) compares the overall 82^^81 IC rate constants measured for the three carotenoids studied so far and two other (3carotene analogues: Ml3 (n=l3) and m5 (^==5). While for Ml3 we still observe the presence of the intermediate state, for m5 a three-level model is able to satisfactorily describe the IC dynamics. The plot in Fig. 3(a) is drawn as a function of l/(2n+l), as it has been demonstrated [8] that this is the scaling law of the S2-S] energy gap. The non-monotonic behavior of the measured k2i rate constant observed across n=9 can be explained by considering that, according to the theoretical model [7, 8], the 1 ^B^ level enters the S2-S] gap for n=9. 15 11 9

0.04

7

0.06

0.08

1/(2n+1)

0.10

-100 0

100 200 300 400

Probe Delay (fs)

0

100 200 300 400 500

Probe Delay (fs)

Fig. 3. (a) S2->Si rate constants for the IC to Si as a function of conjugation length; dashed curve is a guide to the eye. (b) -AT/T dynamics of rhodopin-glucoside in acetone solution (upper panel) and the LH2 complex of Rhodopseudomonas Acidophila (lower panel) with exponential fits; (c) -AT/T dynamics of okenone in cyclohexane solution (upper panel) and the LH2 complex of Chromatium Purpuratum (lower panel) with exponential fits.

Whitin the LH2 complexes a new relaxation channel, in addition to IC to Si, opens up: energy transfer to the BChls. To investigate this we compare the IC dynamics of carotenoids in solution and within the LH2 complexes. Figures 3(b) and 3(c) (thick solid lines) show -AT/T dynamics of rhodopin glucoside and okenone, respectively. Upper traces refer to the carotenoid in solution, while lower ones have been measured inside the respective LH2 complex. In both the carotenoids we observe the formation of the PA] signal corresponding to the Si->Sn absorption. Pump-probe traces are taken at the wavelength corresponding to the peak of the PA] signal (560 nm for rhodopin and 600 nm for okenone). In both the carotenoids we observe a shortening of the PAi risetime upon switching from the solution to the LH2 complex. Single exponential fittings (thin solid lines in Fig. 3b and 3c) enable us to estimate the S2->BChl energy transfer (ET) efficiency. In the Rhodopseudomonas Acidophila complex the Rhodopin

366

Glucoside risetime shortens from ii = 145 fs to ii == 61 fs, with a consequent ET efficiency of -58%. In the Chromatium Purpuratum complex the Okenone risetime shortens from ii = 130 fs to TI = 80 fs, with a consequent ET efficiency of -38%. Not shown here, the time constant for the decay of Si to So is nearly unchanged inside and outside the LH2 complex, indicating that there is negligible energy transfer from Si to the BChls. A still open question is whether ET to the BChl takes place from S2 or Sx.

4.

Conclusions

In conclusion, carotenoids show a very rich dynamics on the ultrafast timescale, which is directly related to their function in photosynthetic light harvesting. Its complete investigation requires sub-20-fs pulses broadly tunable from the UV to the IR spectral ranges. Acknowledgements. RJC acknowledges support from the European Community Access to Research Infrastructure action of the improving Human Potential Programme, contract N. HPRI-CT-2001-00148 (Center For Ultrafast Science and Biomedical Optics, CUSBO).

References 1 H. A. Frank and R.J. Cogdell, Photochem. Photobiol. 63, 257, 1996. 2 B. Hudson and B. Kohler, Chem. Phys. Lett. 14, 299, 1972; K. Schulten and M. Karplus, Chem. Phys. Lett. 14, 305, 1972. 3 H. Kandori, H. Sasabe and M. J. Mimuro, J. Am. Chem. Soc. 116, 2671, 1994. 4 A. Shirakawa, 1. Sakane, M. Takasaka, and T. Kobayashi, Appl. Phys. Lett. 74, 2268, 1999. 5 M. Zavelani-Rossi, G. Cerullo, S. De Silvestri, L. Gallmann, N. Matuschek, G. Steinmeyer, U. Keller, G. Angelow, V. Scheuer and T. Tschudi, Opt. Lett. 26, 1155,2001. 6 G. Cerullo, D. Polli, G. Lanzani, S. De Silvestri, H. Hashimoto, R. J. Cogdell, Science 298, 2395, 2002. 7 P. Tavan and K. Schulten, Phys. Rev. B. 36, 4337, 1987. 8 K. Furuichi, T. Sashima, Y. Koyama, Chem. Phys. Lett. 356, 547, 2002.

367

Energy flow in carotenoids, studied with pumpdeplete-probe, multiphoton and coherent control spectroscopy T. Buckup\ W. Wohlleben^'^ J. Savolainen^ B. Heinz\ H. Hashimoto^ RJ. Cogdell', J.L. Herek^ and M. Motzkus^'^ ^ Max-Planck-Institut fur Quantenoptik, D-85748 Garching, Germany ^ Philipps-Universitat Marburg, Physikalische Chemie, D-35032 Marburg E-mail: [email protected]. de ^ FOM-Institute for Atomic and Molecular Physics, 1098 SJ Amsterdam, The Netherlands ^ Light and control', Dpt. of Physics, Osaka City University, Osaka 558-8585, Japan ^ IBLS, University of Glasgow, Glasgow G12 8QQ, United Kingdom Abstract. We introduce alternative femtosecond spectroscopy techniques to study unattributed carotenoid deactivation signals in broadband transient absorption. These are shaped VIS excitation, nonlinear IR excitation, or insertion of an additional IR depletion pulse to manipulate excited state population.

1. Introduction. The ultrafast energy flow in carotenoids has mostly been studied with linear spectroscopy, where a visible (VIS) pump pulse excites the S0-S2 transition and a probe pulse records the time resolved absorption and/or fluorescence of the deactivation process. One of the prominant signals is the VIS absorption band Si-Sn of the singlet dark state Si, traditionally seen as the only channel in a sequential deactivation scheme S2-S1-S0. However, with decreasing temporal resolution and spectrally broadband detection numerous spectroscopic signals were found that are seemingly not in accord with the mentioned threestate model [1]. One feature is the very fast spectral dynamics in absorption [2] and fluorescence [3], often interpreted as evidence for an intermediate electronic state. Another feature is the S* signal, an absorption band at the lower-frequency side of the ground state absorption, attributed to either vibrationally hot ground state [4] or to an alternative deactivation channel that coexists with Si [5, 6]. The symmetry deformation of a carotenoid that is incorporated into light-harvesting antennae has been proposed to switch the deactivation behaviour [4, 7]. 2. VIS pump - VIS probe for N=9-15 and LH2. First we characterise the spectral profile and lifetime of the S* and Si. In solution, we observe that the lifetime of the signal S*, in the following S*soi, does not depend on conjugation length A^, but instead is constant 6.2±0.4 ps for carotenoids with A^= 11-19 [4]. This observation contravenes an identification of S*soi with any excited state, but it is in favor of an attribution to vibrationally hot ground state hotSo (Fig. la). The charm of this attribution is that it naturally rationalizes the spectral position of the S*soi signal at the lower-frequency wing of the cold So absorption. Contrary to these findings, VIS transient absorption in LH2 from Rps. acidophila shows ultrafast carotenoid triplet population and an additional 10 ps

368

timescale for excitation energy transfer [5-7]. These observations favor the assumption of an aUernative deactivation channel S*T, being an electronically excited singlet state, possibly of IB^" symmetry. Deactivation from S2 into S*T would then be activated by the external symmetry deformation due to incorporation into the LH assembly (Fig. la). At this point, further conventional pump - probe studies can hardly lift the apparent contradiction. 3. VIS pump - IR deplete - VIS probe for N=9-15 and LH2. Inspired by the Tannor-Rice scheme of coherent control [8], we now use an additional 'depletion' pulse to manipulate the excited state population [4]. The depletion pulse can be tuned in delay and wavelength to the Franck-Condon window of a specific absorption/emission band, thus intercepting the natural photoreaction. The correlation of the changes that are induced on the different signals helps for their attribution within a model: If we tune the depletion pulse into the infrared (IR), where the S2 state absorbs, every state that is populated by deactivation of S2, i.e. every electronically excited state, should show diminished population. This approach was used on lycopene, zeaxanthin and A^=9,ll,13 and 15 |3-carotene in solution to show that the S*soi signal has no correlation with 82 population, because it is not depleted [4]. Hence it is due to a vibrationally hot ground state (Fig. la), generated by impulsive Raman scattering of the VIS pump pulse. Note that carotenoids are known to be among the most efficient Raman scatterers. But, if the model of two coexisting deactivation channels in LH2 (Fig. la) holds true, one expects to observe a correlation of the S*T signal with S2 population: S*T should decrease with S2 depletion. Furthermore, when the pumpdeplete delay T (Fig. lb) is tuned such that S2 has decayed, but that the population in Si can be re-excited to S2, from where it re-decays into both Si and S*T, one expects to see an increased S*T signal. Indeed, the experiment verifies these predictions and produces both depletion and re-population of the S*T a) / in LH \ / environment\ ' S%(fB,

T,(fB. /\-;—hotSo(S*3,)

^

L_ 500

550 600 650 Probe wavelength, nm

Fig. 1. a) Model of carotenoid energy flow. Levels with thick lines contribute to excitation energy transfer to BChl. Energy flow into S*T is actived by external symmetry deformation, b) Pump - deplete - probe spectra for rhodopin glucoside in LH2. Difference spectra in grey-shades indicate the change induced by the depletion pulse. Depending on VIS pump - IR deplete delay T, we observe depletion (7=^0, dashed AOD, dark grey AAOD) or re-population of the S*T signal (7=150 fs, dotted AOD, light grey AAOD). The negative signal around 590 nm is the B800 Qx bleach generated by the IR pulse alone.

369

signal, depending on delay T (Fig. lb). This ultimate test is strong support for the existence of S*T (IBu) in LH2 and the 'symmetry switched deactivation' model. 4. Shaped VIS pump - VIS probe for LH2. In the next step, we excite S2 in LH2 with shaped VIS pulses and optimise the excitation shape for maximum ratio of internal conversion compared to excitation energy transfer [9]. From the characteristic modulation of the optimal multipulse excitation one can extract a frequency of 160±25cm'^ that can be identified with the bu mode that promotes S2 - Si internal conversion [10]. This measurement is impossible with conventional spectroscopy. Detailed studies of the performance of the optimised pulse shape show that we achieve robust and coherent control of the energy flow. Selective Raman scattering induces the promoting bu mode [10]. On top of the controlled energy flow, saturation changes the integrated excitation per irradiated photon. 5. Multiphoton IR pump - VIS/IR probe for N=ll,13. Finally, we follow Walla et al. to measure transient absorption upon nonlinear IR excitation [11], but now with broadband VIS and IR detection in the region of the hypothetical new states. By the characteristic spectral profile and lifetime we identify the anticipated Si excitation (2 IR photons), and additionally the dublet radical cation Car^ (5 photons) and of the triplet Ti, populated from Tk (4 photons). The highly nonlinear excitation pathways contribute nonetheless with high signal amplitude due to the intermediate electronic resonances of the Si and S2 states. Population of S*soi or S*T can be excluded, in accord with the symmetries and energies of the current model (Fig. la). 6. Summary. Innovative techniques that go beyond linear pump - probe experiments aquire the 'missing-link' of spectroscopic knowledge that complements conventional spectroscopy. Here we applied pump - deplete probe, shaped pump - probe and nonlinear pump - probe in order to propose a consistent model for the ultrafast energy flow in carotenoids.

References 1 2 3 4 5 6 7 8 9 10

T. Polivka and V. Sundstrom, Chem. Rev. 104, 2021-2071, 2004. G. Cerullo, et al., Science 298, 2395-2398, 2002. R. Fujii, et al, Chem. Phys. Lett. 384, 9-15, 2004. W. Wohlleben, et al., J. Phys. Chem. B 108, 3320-3325, 2004. C. C. Gradinaru, et al, Proc. Nafl. Acad. Sci. U.S.A. 98, 2364-2369, 2001. W. Wohlleben, et al, Biophys. J. 85, 442-450, 2003. E. Papagiannakis, et al, J. Phys. Chem. B 107, 5642-5649, 2003. D. J. Tannor, et al, J. Chem. Phys. 85, 5805-5820, 1986. J. L. Herek, et a l . Nature 417, 533-535, 2002. W. Wohlleben, et al, in Femtochemistry andfemtobiology: Ultrafast events in molecular science, 91-95, Edited by J. L. Hynes and M. M. Martin, Elsevier, 2004. 11 P. J. Walla, et al, J. Phys. Chem. A 106, 1909-1916, 2002.

370

Amplitude spectra of molecular vibration modes in phthalocyanine: comparison with Raman excitation profile Takayoshi Kobayashi, Masakatsu Hirasawa, Yuzo Sakazaki, Hiroki Hane Department of Physics, Graduate School of Science, The University of Tokyo, 7-3-1 Kongo, Bunkyo, Tokyo 113-0033, Japan E-mail: [email protected] Abstract. Ultrashort pulse lasers with 6-fs and 20-fs durations were utilized for a phthalocyanine tin (IV) dichloride thin film to induce several vibrational modes and vibration amplitude spectra were determined by multiwavelength measurement technique. From the spectra we could identify the electronic states, which couple to two vibrational modes with frequencies of 670 and 750 cm'\ It was shown that the vibrational amplitude profile obtained by the method can be used for providing information for the assignment of the vibrational mode.

1.

Introduction

Dynamical processes of phthalocyanine (Pc) including vibronic coupling is highly relevant with optical nonlinearity because it changes the electronic spectra substantially in such highly symmetric system as Pes. Raman spectroscopy is useful to clarify such systems since it is a valuable tool for the identification of the relevant states to the nonlinear processes and the relaxation processes. The Raman spectra obtained from Pes are relatively sharp, molecularly specific, and can be resonance enhanced, allowing clear identification of the species. Pes are characterized by two broad strong electronic absorption bands in the UV-visible region [1,2], chemical and thermal stability [3], relatively easy of thinfilm fabrication [4], and a high laser damage threshold [3]. In this letter we applied a new method of real-time spectroscopy to a sample of Pc thin film to clarify the mechanism of vibronic coupling which are relevant to chemical reactions and carrier transport.

2.

Experimental

A thin film sample of phthalocyanine tin (IV) dichloride (SnPc) prepared by evaporation in a vacuum on a glass substrate was used as a sample. The setup for femtosecond time-resolved absorption spectroscopy was described previously [58].

371

3.

Results and Discussion

Fig. 1 shows the stationary absorption spectrum of the SnPc film and spectra of the 6-fs and 20-fs pulsed lasers, and the time-resolved difference absorption spectra extending from 11500 to 19000 cm'^ obtained by the 6-fs and 20-fs lasers at several delay times. There are very broad and intense bleaching in the range of 15700-19500 cm"^ and an absorption peak in the range of 14000-14900 cm"^ Even though the stationary absorption spectrum shown in Fig. 1 is inhomogeneously broadened because of amorphous character, the peaks at 13300 and 14600 cm"' are well established to be assigned as Qy and Qx bands [1,2]. The difference between the sign-reversed ground-state absorption spectrum and the transient spectrum is caused by the existence of the excited-state absorption spectrum which is known to have peaks around 14800 cm"' and at lower than 18000 cm"'. In the 11600-13200 cm"' range pump-probed by the 20-fs pulse laser, bleaching is observed.

11

12

13

14

15

16

17

18

19

Fig. 1. (a)Stationary abso avenum er x cm ^^ ^^^ j^^^^ intensity (dotted curve; 20-fs laser, dashed curve; 6-fs laser), (b)-(c) time-resolved difference absorption spectra at several probe delay times. In the real-time transmittance change, the oscillating components due to molecular vibrations with mode frequencies of 590, 670, 750, and 1340 cm"' are observed. They were assigned as Aig C-N-C bending. Big macrocycle deformation, B2g C-N-C bending, and the first over-tone (the second harmonic) of the Big macrocycle deformation, respectively [9]. In order to study detailed probe frequency dependence of several modes, the Fourier amplitude is plotted for 670 cm" and 750 cm"' modes against probe photon energy in Figs. 2(a) and 2(b), respectively. The former figure shows that the vibrational amplitude of the mode of 670 cm"' fits well with the absorption spectrum composed of both Qx and Qy transitions. Since the absorption cross section of CT transition is small, it is difficult to tell whether the CT absorption ranging in the 16000—18000 cm"'

372

region is or is not contributing to the 670 cm" mode. The latter figure shows that the mode of 750 cm"^ has only the contribution from QY transition. Hereafter more general discussion or the difference between REP and vibrational amplitude profile (VAP) is to be made. Both of them are related to the Raman processes, the former is related to the spontaneous Raman and the latter is induced by the stimulated Raman process. Therefore there is a possibility of difference induced by the mode density at the frequencies of the fields. By combining the two methods we can obtain even more detailed insight into the vibronic coupling, molecular vibration, and electronic state characterization. In conclusion, we for the first time compared vibrational amplitude profile (VAP) with the Raman excitation profile (REP), and focused the difference between them. The VAP obtained by ultrashort pulse was shown to be one of a useful tool for the mode assignment of those molecular vibrations which commonly appear in the Raman and real-time snectra. (a)

^

i(b)

670 CI Q'+Q'+CT

•••.v. V#.

>/vSS^ 12

13

14

15

16

17

18 12

13

14

15

16

17

Wave number( x 10^cm')

Fig. 2. (a) The Fourier amplitude of the 670-cm-l mode (circles and squares) and absorbance of Q-band and CT-band. (b) The Fourier amplitude of the 750-cm-l mode (circles and squares) and absorbance of Qy-band and Q-band. The circles and squares were data obtained by using 6-fs and 20-fs lasers, respectively.

References 1 2 3 4 5 6 7 8 9

A.J. Bovill, A.A. McConnell, J.A. Nimmo, W.E. Smith, J. Phys. Chem. 90, 569, 1986. H. Yoshida, Y. Tokura, T. Koda, Chem. Phys. 109, 375, 1986. G. Ricciardi, A. Rosa, E.J. Baerends, J. Phys. Chem A 105, 5242, 2001. C. Jennings, R. Aroca, A. Hor, R.O. Loutfy, Spectrochim. Acta 41, 1095, 1985. T. Kobayashi, T. Saito, H. Ohtani, Nature 414, 531, 2001. A. Baltuska, T. Fuji, T. Kobayashi, Opt. Lett. 27, 306, 2002. T. K obayashi, A. Shirakawa, H. Matsuzawa, H. Nakanishi, Chem. Phys. Lett. 321, 385, 2000. A. Shirakawa, I. Sakane, T. Kobayashi, Opt. Lett. 23, 1292, 1998. T.V. Basova, B.A. Kolesov, J. Struct. Chem. 41, 770, 2000.

373

Real-time tracking of the peaks in transition difference spectra during vibrational periods in PDA Yoshiharu Yuasa\ Mitsuhiro Ikuta\ Tatsumi Kimura^, Hiroo Matsuda^, and Takayoshi Kobayashi* ^ Department of physics, Graduate School of Science, University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo, 113-0033, Japan E-mail: [email protected] ^ National Institute of Advanced Industrial Science and Technology, Tsukuba Central 5, 11-1 Higashi, Tsukuba, Ibaraki, 305-8565, Japan E-mail: [email protected] Abstract. Molecular dynamics are investigated using spectrally resolved pumpprobe measurement. Stokes shift associated with geometrical relaxation and electronic spectral change caused by lattice vibration are directly observed in time by a novel peak-energy tracing method.

1.

Introduction

Optical and electrical properties of conjugated polymers have attracted attention because of their unique properties as model compounds of one-dimensional electronic systems. Among conjugated polymers polydiacetylenes (PDA's) have special interests because of their large nonlinearity [1]. The ultrafast optical responses in PDA's have been intensively investigated using femtosecond absorption spectroscopy. Blue-phase PDAs have much smaller fluorescence yield than red-phase PDA. The mechanism was explained by a model proposed by Kobayashi et al. as follows [2]. First, photoexcited ^Bu free exciton (FE) relaxes to the nonfluorescent ^Ag state, which lies lower than the ^Bu exciton. The nonthermal ^Ag state relaxes to the bottom of potential curve of the self-trapped exciton (STE) state and then thermalizes. The time constants of the relaxation and thermalization are about 60 fs and 1 ps, respectively. Finally, ^Ag relaxes to the ground state with a decay time of l-2ps.

2.

Experimental Methods

The sample used in this experiment is a cast film of blue-phase polydiacetylene-3butoxycarbonylmethylurethane (PDA-3BCMU), poly [4, 6-docadiyn-l, lO-diolbis (n-butoxy -carbonylmethylurethane)] on a grass substrate. Output pulses from a noncollinear optical parametric amplification (NOPA) with prism pare and chirp

374

mirrors seeded by a white-light continuum which have 4.7 fs width at 5-kHz repetition rate were used as both pump and probe pulses. The spectra of the pulses covered from 520 nm to 750 nm with a nearly constant phase. The energies of the pump and probe pulses were about 35 and 5 nJ, respectively.

3.

Results and Discussion

Real-time evolution of the energies and intensities of features in difference transmission spectra (DTS) was investigated. Especially several positive and negative peaks, and the position of the largest slope on each side of the peaks were investigated. In the normalized difference transmission (AT/T) spectra, two components with different lifetimes were separated by the singular value decomposition (SVD) method. One is short-lived component of about 60 fs and the other is long-lived component of about 0.9 ps. In the separated spectrum of short-lived component, five positive peaks were found which are assigned to mduced emission from the excited-state (ES) vibrational levels to the ground-state (GS) vibrational levels. In the latter no population exists because of much larger energies of the levels than the thermal energy. The time trace of energy of a short-lived peak near 1.88 eV, which disappears within 100 fs, is shown in Fig. 1 with an exponential fitting curve. The transition energy from ES to GS is dbectly observed and it is found to shift to lower associated with the decay process from l^Bu FE to STE and geometrically relaxed 2^Ag state after photo-excitation. The time constant of the Stokes shift agrees fairly well with decay time (-50 fs) of the induced emission. This is quite consistent with a smaller quantum efficiency of emission than 10"^. Stokes shift has been esthnated to be about 50 to 90 meV by stationary fluorescence spectra [3-5]. These results are consistent with the model of relaxation dynamics in PDA proposed by Kobayashi and others [2]. •

'

1

I

'

l

l

1.90 s^

^ *

1

1.86

(0 4)

1.84 C)

20 40 60 80 Pump-probe delay time (fs)

Fig. 1. Oscillating shift of an induced-emission peak energy due to molecular vibration during the geometrical relaxation from l^Bu FE to 2^Ag.

375

There are two species in the long-lived (0.9 ps) component of the AA spectrum. One is photobleaching due to depopulation of GS and the other is induced absorption (IA) from 2 Ag to n^Bu. Several peaks are found in DTS. One negative peak at 1.8 eV is considered to be due to lA from 2^Ag to n^Bu. On the other hand, in the bleaching region, three interesting features are found and they can be discussed by comparing with the stationary absorption spectrum. First, boundary region between positive and negative AA is located at almost the same energy as the edge of stationary absorption spectrum. Second, the lowest energy peak at 1.99 eV in bleaching region corresponds to the 0-0 transition from the ground state. Third, clear vibrational structure, considered to be phonon side band, is observed. Laser spectrum has a sharp-peak structure at 2.27 eV with broad pedestal and several fine structures. This resulted in a kind of "hole burning" in the difference absorption (AA) spectrum of the PDA, which have a large absorbance at the photon energy. On the other hand, homogenous spectrum has a vibrational structure determined by the vibronic coupling, which is also observed in the bleaching spectrum. This means the bleaching spectrum is less inhomogeneously broadened than the stationary absorption spectrum. The tracking of vibronic transition peaks in DTS revealed the dynamic behavior of DTS modulated by molecular vibrations.

4.

Conclusions

In conclusion, real-time tracking of the transition energies and real-time monitoring of intensities of vibronic transition peaks modulated by molecular vibration were performed for the first time. This method does not have a problem of artificial interference as in spectrogram analysis. Stokes shift due to geometrical relaxation is directly time-resolved in real-time. Oscillation of transition energy caused by motion of wave packet through vibronic coupling is detected. Decomposition of the real-time spectral change into energy change and change in vibronic transition probability was achieved and their photon energy dependence was investigated.

References 1 T. Kobayashi, M. Yoshizawa, U. Stamm, M. Taiji and M. Hasegawa, J. Opt. Soc. Am.B7, 1558, 1990. 2 T. Kobayashi, M. Yasuda, S. Okada, H. Matsuda and H. Nakanishi, Chem. Phys. Letters 261, ^12, 1997. 3 M. Yoshizawa, A. Kubo and S. Saikan, Phys. Rev. B 60, 15632, 1999. 4 K. Miyano and T. Maeda, Phys. Rev. B 33, 4386, 1986. 5 A. Yasuda, M. Yoshizawa and T. Kobayashi, Chem. Phys. Letters 209, 281, 1993.

376

Time-resolved CARS studies of vibrational coherences in the condensed phase: I2 in solid krypton Michael Karavitis^ Ilya Goldschleger^ V. Ara Apkarian^ and Takayuki Kumada^ ^ Department of Chemistry, University of CaHfomia, Irvine, CA 92697-2025, USA E-mail: [email protected] ^ Advanced Science Research Center, Japan Atomic Energy Research Institute, Tokai, Ibaraki 319-1195, Japan E-mail: [email protected] Abstract Vibrational dephasing in the prototypical system of I2 isolated in solid Kr is studied using TRCARS. Decay |(?>
1. Introduction Time-resolved, electronically resonant, parametric four-wave mixing processes allow detailed preparation, manipulation, and interrogation of molecular rovibronic coherences. We implement this tool in the present to carry out systematic studies of vibrational dephasing of molecular iodine isolated in solid Kr. Using ultrashort laser pulses with adjustable time-bandwidth profiles, we extract vdependent rates of dephasing and dissipation for v = 7 - 79, as a function of temperature for T = 6 - 45 K.

2. Experimental The experiments are carried out on thin films of 12/Kr, using three 25-100 fs laser pulses from two home-built non-collinear optical parametric amplifiers (NOPA) pumped by a IkHz regeneratively amplified TiiSapphire laser. Intensities of time-resolved coherent anti-stokes Raman scattering (TR-CARS) from the sample are measured for coincident pump (P) and Stokes (S) beams, as a function ofprobe(P')time.

3. Results and Discussion The measurements are restricted to the P^^'^^ component of the third order polarization, in which the three input fields act on the state ket (bra) while the bra (ket) evolves field free [1]. With coincident pump (P) and Stokes (S) pulses, the

377

prepared vibrational coherence corresponds to the Raman packet is created on the ket:

cp''\k, - ^,;0X^^1 = e^''^-'^''Y.cr''''^^Me

(1)

The time-delayed probe pulse (P') interrogates the vibrational coherence, with the anti-Stokes radiation along kAs = kp — ks + kp' serving as signal. The coherence will decay by coupling to the lattice phonons, by pure dephasing r?* and dissipation Tj processes. The dephasing of a vibrational wave packet composed of an TV-state superposition will be described by A^-dephasing rate constants /v, describing the loss of amplitude correlation between state v and the ground vibrational level (v = 0). After tracing over the bath, the decay of the reduced system density can be described as: A,v = E^^v^^ cos(6>,^/+
(2)

v,V

TR-CARS signals obtained with a Stokes shift centered on v = 17 and 18 are shown in Fig. 1. The extracted T-dependent values of y(v = 17, J8) are also plotted, y is independent of T below 10 K, but it steeply increases with T above. Fig. 2 shows Y as a ftmction of v. The v-dependence is essentially linear at 7 K, whereas it can be fit with a quadratic dependence on v at 34 K. The entire data set can be well represented by the ftinctional form:

25

30

Fig. 1. Temperature dependence of vibrational dephasing rate for a wavepacket centered at V = 17, 18. The data points are extracted from the decay of D.C. component of TRCARS signal. The solid line represents the best fit to experimental data using Eq. (3). The inset shows two representative TRCARS signals obtained at 7 and 34 K. The dephasing rates of the two signals vary by more than factor of three.

378

Fig. 2. Dephasing rate as a function of vibrational quantum number measured at 7 and 34 K. At 7 K, dephasing rate displays a nearly linear dependence on v, while at 34 K, this dependence nominally quadratic. The inset illustrates a time domain signal containing a superposition of vibrational states v = 2 -4. This signal exhibits nodes separated by 26 ps, corresponding to the anharmonic shift between adjacent vibrational levels.

V

V T^e"'-\

(3)

where ry = 355 ± 10 ps, rj = 560 ± 70 ps, and ^ = 54 ± 4 K, to be compared with the Debye temperature of 72 K of solid Kr. For T < 10 K, when the high frequency phonons are not populated, the decay becomes T-independent, suggesting dissipation by the spontaneous creation of phonons. Note that to bridge the vibrational energy gap of the molecule, the dissipation must involve the simultaneous creation of four phonons. On the other hand, for T > 10 K, pure dephasing is activated by the thermally populated phonons. A density of states analysis for phonon projections on the molecular coordinate is used to identify the nature of the pseudo local phonons responsible for the observed dephasing. Given the relatively well known interaction potentials between guest and host, quantitative treatments of this system can be expected.

References 1 M. Karavitis, R. Zadoyan, V. A. Apkarian, in Journal of Chemical Physics, Vol.114, 4131, 2001. 2 M. Karavitis, V. A. Apkarian, "Vibrational Coherence of I2 in Solid Kr, in Journal of Chemical Physics, Vol.120, 292, 2004. 379

Measurement of Conical Intersection Dynamics by Impulsive Femtosecond Polarization Spectroscopy Darcie Farrow\ Wei Qian^, Eric R. Smith' and David Jonas' ' Department of Chemistry and Biochemistry, University of Colorado, Boulder, CO 803090215 ^ School of Chemistry, Georgia Institute of Technology, Atlanta, GA 30332-0400 E-mail: [email protected] Abstract. The -100 fs anisotropy decay from 0.4 toward 0.1 after excitation of a degenerate transition in a square molecule can be predicted from quantum beats of asymmetric vibrations and curve crossing at a conical intersection.

The breakdown of the Born-Oppenheimer adiabatic separation between fast electronic and slow nuclear motions is important throughout chemistry. In the adiabatic picture, slow nuclei move on a potential energy surface given by the nuclear coordinate dependent energy of the electronic quantum state. This adiabatic picture breaks down when the electronic motions are slowed by near electronic degeneracy. With several vibrational degrees of freedom, crossing between potential curves of the same symmetry cannot be avoided: the geometry of the crossing region resembles the vertex of a right circular cone, where two conical surfaces smoothly connect through a point.[l] As a path between electronic states, these "conical intersections" have been implicated in many photochemical reactions (e.g. the primary photo-isomerization in vision). The experiments described here impulsively excite electronic-vibrational wavepackets directly onto a symmetry required conical intersection. Since fourfold symmetry requires a 90° rotation of the electronic wavefunction between degenerate surfaces, femtosecond polarization spectroscopy can measure conical intersection dynamics in square symmetric molecules by probing both electronic reorientation and the vibrations driven by the conical intersection. We carried out pump-probe anisotropy measurements using near transformlimited 26 fs pulses that covered the doubly degenerate Q(0-0) electronic transition of silicon 2,3-naphthalocyanine bis(trihexylsilyloxide) in benzonitrile solution. Naphthalocyanine has the same point group symmetry as a square, and the Q(0-0) transition is analogous to the transition from the ground state {nx=\, ny=\) to the degenerate excited states (nx=2, ny=l) and (rix^'l, ny=2) for a particle in a square box. Signals with the probe polarization parallel, perpendicular, and at the magic angle to the pump polarization were measured (see Fig. 1). The magic angle signal measures isotropic dynamics, while the pump-probe polarization anisotropy (inset to Fig. 1), r = (S\\ - S_J/(S\\ + 25'j^), reflects rotational motion.

380

Within -100 fs, the pumpprobe anisotropy in Fig. 1 decays from an initial value r(0)~0.4, which is characteristic of dipole 2,0 excitation, toward r{oo)=0.100, which is characteristic of dipole 1,5 derealization within a plane. Molecular rotation takes -450 ps, and the magic angle signal 1,0 indicates no population dynamics beyond weak quantum beats, so 0,5 the anisotropy decay in Fig. 1 0.01. must arise from some kind of 0 200 400 6O0 300 1000 0,0 dfslay r (fs) rotation of the electronic wavefunction within the plane of 500 0 1000 1500 2G00 square symmetry. [2] The key to delay T (fs) interpretation lies in the Fig. 1. Experimental pump-probe transients for polarization dependence of the 3 relative polarizations and the polarization quantum beats. Upon electronic anisotropy (inset). excitation to a degenerate state, a square molecule can lower its total energy by exciting asymmetric vibrations which break the square symmetry (Jahn-Teller effect). Totally symmetric vibrations can also be excited by the usual Franck-Condon mechanism. In offresonant femtosecond Raman experiments, ground state vibrations of different symmetry have different polarization signatures.[3] The amplitudes, phases, and anisotropics for each vibrational quantum beat were calculated from damped cosine fits to the pump-probe signals. The two highest frequency vibrations have anisotropics equal to 1/10 within error and appear in all three signals with the same phase. In contrast, the three lowest frequency vibrations do not appear in the magic angle signal and exhibit a n phase difference between parallel and perpendicular signals, with anisotropics that could be infinite (due to a vanishing denominator) within experimental error. In the ground electronic state, an anisotropy of 1/10 indicates a totally symmetric vibration (^ig symmetry), while an infinite anisotropy indicates a Jahn-Teller active asymmetric vibration (^ig or Z?2g symmetry). The anisotropy for vibrations on the excited electronic state matches that for a ground state vibration only after both electronic dephasing and non-adiabatic population transfer between the two components of the doubly degenerate excited electronic state are complete. Based on ref. [4], we developed a Brownian oscillator type model for the asymmetric vibrations on the four electronic states analogous to (1,1), (2,1), (1,2), and (2,2) for the particle in a 2D box. If the Jahn-Teller distortions were all of the same vibrational symmetry (all b\g or all b2g), this adiabatic model would work because the two degenerate electronic states would have different symmetries in the lower symmetry point group. The model predicts an impulsive anisotropy of 2,5

u

381

K n = (l/10) + (3/10)exp(-4Re[g(r)])

(1)

where g(T) is the Brownian oscillator lineshape function for the asymmetric vibrations in the linear absorption spectrum. The model yields Jahn-Teller stabilization energies of 5 cm'\ 7 cm'^ and 17 cm"^ for the three asymmetric vibrations. Inverting Eq. (1) in the high temperature approximation yields d' Re[g(/)] / dt^ = 2{Dco)(kJ I h)M{t).

(2)

Both the Jahn-Teller stabilization energy {Da^ and correlation function M{t) of the asymmetric motions responsible for the initial anisotropy decay can be recovered from Eq. (1) and (2). (Dew) is only about 5 CWL\ while M{f) has an initially underdamped character, oscillating to negative values but suddenly and permanently dropping to nearly zero at around 150 fs. Inserting the vibrational frequencies and Jahn-Teller stabilization energies into Eq. (1) reproduces the initial anisotropy decay without any adjustable parameters. However, Eq. (1) predicts a subsequent anisotropy recurrence that is not seen experimentally. The "sudden death" of the correlation function occurs just when vibrational wavepackets would return to the conical intersection at zero displacement. To test the hypothesis that population transfer at the conical intersection is the key missing ingredient in the above theory, quantum mechanical calculations with linear Jahn-Teller coupling along both b\^ and big coordinates were carried out. The largest effect arises from the 140 cm"^ and 306 cm'^ vibrations, which reduce the anisotropy recurrence two-fold. Calculations show that overdamped low frequency asymmetric motions might also help prevent recurrence. However, further evidence for a conical intersection comes from the anisotropy of the totally symmetric modes, which calculations predict to exceed 0.25 unless the nonadiabatic population transfer at the conical intersection reduces it to 0.1. The speed of electronic motion at the conical intersection increases with stabilization [Eq. (1)] and lOOOx larger stabilizations occur in many chemical reactions. Although the semi-classical Eq. (2) does not apply, the full quantum theory predicts electronic "jumps" as fast as - 2 fs at conical intersections. Acknowledgement. This work was supported by the U.S. National Science Foundation. D.A.F and E.R.S. are National Science Foundation Optical Science and Engineering Program graduate fellows.

References 1. 2. 3. 4.

382

D. R. Yarkony, Reviews of Modern Physics 68, 985(1996). A. Albrecht Ferro, D. M. Jonas, Journal of Chemical Physics 115, 6281 (2001). M. Khalil, O. Golonzka, N. Demirdoven, C. J. Fecko, A. Tokmakoff, Chemical Physics Letters 321, 231 (2000). J. Sung, R. J. Silbey, Journal of Chemical Physics 115, 9266 (2001).

Vibrational Phase Characterization in Femtosecond-pumped Molecules by Path-length Modulation Takachi Taneichi, Takao Fuji, Yoshiharu Yuasa, and Takayoshi Kobayashi Department of Physics, Graduate School of Science, University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033, Japan E-mail: [email protected] Abstract. The ultrafast dynamics of a cyanine dye is studied by the pump-probe spectroscopy using 20 fs pulses establishing a high sensitivity on the molecular vibration. The simulation reproduces the observed phase dependence on the mode frequencies.

1.

Introduction

The observation of molecular vibration dynamics has been realized by means of the femtosecond pump-probe type spectroscopy. Coherent vibrational modes of molecules are well formulated by time-dependent wavepacket description, by which one can provide theoretical bases to the experimental results and predict novel phenomena. The most popular way to observe the wavepacket dynamics is pump-probe experiment with resonant femtosecond pulses. Since usual molecules have nanosecond population decay time, which is much longer than period of molecular vibration and decoherence time of the vibration, large background signal is always observed at the pump-probe experiment and it limits dynamic range of the vibrational signal. In this paper, we apply a pathlength modulation technique to the pump-probe spectroscopy for investigation of molecular vibrational dynamics. The technique basically provide the time-derivative of the transmittance-change signal. It reduces substantially slow population decay components at the measurement process, as a result, much higher dynamics range for molecular vibrational signal can be achieved.

2.

Experimental Methods

Pump-probe spectroscopy based on a 20-fs Ti: sapphire laser is used in the present measurement. The sample is ethanol solution of cyanine dye, l,r,3,3,3',3'hexamethyl-4,4',5,5'-dibenzo-2,2'-indotricarbocyanine (HDITC). The wavepacket dynamics in the excited state of HDITC was recently observed in our group [1]. Path-length modulation technique is used to implement an improved sensitivity for

383

the oscillatory signals compared to the ordinary amplitude modulation technique. In order to discuss properly the probe-wavelength dependence of the time-shift in the pump-probe signal, we characterized the pulses by means of a phase sensitive method [2,3], the FROG (Frequency Resolved Optical Gating) trace, which optimizes the time-resolution to implement a precise measurement. These experimental techniques enable one to detect fine oscillational structures in the measured signals, which are the consequences of molecular vibration. From the pump-probe signals obtained with these techniques, one can obtain the information about a vibrational phase (Figure 1.).

L20

1.25

1.30

Probe Frequency (104cm-l) Fig. 1. Observed and calculated plots of the phase dependence on the probefrequencyfor three modes are shown. Solid dots and solid lines show the observed and calculated phases, respectively, (a) ^o=140 cm'\ (b) a^o=440 cm'^ and (c) coo=130 cm"' With a pump pulse of which width is shorter than the molecular vibration period, the molecule is vibronically excited coherently resulting in molecular vibrations of several modes in general with definite phases.

384

3.

Theory

With the help of the calculation of wavepacket motion, one can investigate the variation of molecular vibrational scheme due to difference in pulse duration, that is, molecular vibration caused by a ultrashort pulse can be described by a welllocalized wavepacket. Effects of the oscillation scheme on the pump-probe signal were investigated theoretically. Calculation was carried out using the effective linear response approach [4], which describes the pump-probe third-order nonlinear process as a linear interaction process of a probe-pulse with a nonstationary molecular system induced by the pump-pulse interaction. This approach shortens substantially the calculation time. The formulation is based on the sequential well-separated pulse limit in which the time delay between the two pulses of pump and probe is assumed to be much larger than the pulse duration, enforcing a sequential pump-first probe-second time ordering. From the calculated signal, the phases of the molecular oscillation are obtained (Figure 1.) and compared with the measured data. It is shown that the change of the phase is a consequence of the wavepacket formation scheme. The wavepacket motion, which is affected by the number of excited vibrational levels in the electronic excited state, reflects the pump-probe signal which in tern modulate the oscillation phase.

4.

Conclusions

The phase-sensitive technique is established by introducing the path-length modulation to the pump-pulse path, and is applied to a measurement of a molecular vibration by the pump-probe spectroscopy. The measured pump-probe signals have pronounced oscillatory features of which phases depend on the mode frequencies. Characteristic feature of the probe-frequency dependence of the phase is resolved and analyzed by means of the pump-probe signal simulation based on the effective linear response approach, for the first time. It is shown that the change of the phase is a consequence of the wavepacket formation scheme. The wavepacket motion, which is affected by the number of excited vibrational levels in the electronic excited state, reflects the pump-probe signal which in tern modulate the oscillation phase.

References 1 T. Fuji, H. J. Ong, and T. Kobayashi, Chemical Physics Letters, Vol.380, 135, 2003. 2 M. Ziolek, M. Lorenc, and R. Naskrecki, Applied Physics B, Vol. 72, 843, 2001. 3 S. Yeremenko, A. Baltuska, F. de Hann, M. S. Pshenichnikov, and D. A. Wiersma, Optics Letters, Vol. 27, 1171, 2002. 4 A. T. N. Kumar, F. Rosa, A. Widom, and P. M. Champion, Journal of Chemical Physics, Vol 114,701,2001.

385

Vibrational Energy Relaxation in Water-Acetonitrile Mixtures Dan Cringus, Sergey Yeremenko, Maxim S. Pshenichnikov, and Douwe A. Wiersma Ultrafast Laser and Spectroscopy Laboratory, Materials Science Centre Department of Chemical Physics, University of Groningen, Nijenborgh 4, 9747 AG Groningen, The Netherlands E-mail: [email protected] Abstract. IR pump-probe spectroscopy is used to study the effect of hydrogen bonding on the vibrational energy relaxation pathways. Hydrogen bonding accelerates the population relaxation from 12ps in diluted acetonitrile solution to 700fs in bulk water.

1. Introduction Water is a unique substance by being the medium in which most chemical reactions that sustain life occur. One of the important issues concerning water dynamics is the question how excess vibrational energy is disposed of. Vibrational energy relaxation (VER) is mainly a radiation-less process, ultimately leading to excitation of the low-frequency thermal motions of molecules. Unveiling the relaxation mechanism implies finding the intermediate steps and routes of the energy transfer process. Water is a very "dynamical" fluid in the sense that the rearrangements of the local structure surrounding a water molecule, occur on a sub-picosecond timescale [1-3]. Vibrational relaxation dynamics in water is also very fast. For instance, the population relaxation lifetime of the OH-stretch mode of HDO molecules in liquid D2O is about 740 fs [4], while in pure H2O it reduces to 250 fs [5]. In contrast, the relaxation lifetime is substantially longer (tens of picoseconds) when water molecules are isolated from each other ui an inert liquid solvent matrix [6]. Hence, vibrational population (energy) relaxation in water molecules seems to be strongly affected by hydrogen bonding. From another hand, the apparent connection between the VER rate and the presence of the hydrogen bonding in water paves the way for studying the phenomenon of hydrogen bonding itself. In this contribution we present a study of the effect of hydrogen bonding on VER of the OH-stretch mode of HDO molecules in the liquid phase. To control the extent of the hydrogen bond network, water is dissolved in an inert solvent (acetonitrile) capable of forming water solutions of any concentration. We demonstrate that IR frequency-resolved pump-probe experiments with a 70-fs temporal resolution allow the determination of the solution composition of water and acetonitrile. On the basis of the experimental results, a model is developed that describes the effect of the hydrogen bonding on the process of VER in liquid water.

386

2. Results and discussion The result of a combined time-frequency scan for a 50% solution is presented in the Fig. 1. It is clear that the signal decay rate is wavelength dependent. Within the induced bleach (induced absorption) band the signal decays slower at the shorter wavelengths than at longer ones (solid curve). Such dependence of the population lifetime on the wavelength is determined by the effect of diverse solvent environments for different OH-bonds. The formation of a hydrogen bond between water molecules leads to a red shift of the absorption band of the OH-stretch vibration. The OH-oscillators that form a hydrogen bond with acetonitrile or are non-hydrogen-bonded, have higher frequencies (-3550 cm"^) and, as Fig.l shows, longer relaxation times. In contrast, the OH-oscillators that form a hydrogen bond to other water molecules have lower frequencies (-3400 cm'^) and much shorter relaxation times. Decomposition of the total absorption line into contributions from acetonitrilebonded and water-bonded water molecules is based on the assumption that there is a clear separation of time scales between the dynamics of water-bonded and nonwater-bonded oscillators [7]. At any given wavelength, the pump-probe signal can be represented as a linear combination of these two contributions. We have performed a global fit of the experimental data, i.e. the complete time-frequency scan (like the one depicted in Fig.l) is simulated for each concentration in a single fitting session. The concentration dependencies of the population lifetimes of non-water-bonded and water-bonded OH-oscillators, derived from the global fit to the experunental data, are shown in the Fig.2. Our analysis shows that the main pathway for the vibrational relaxation of the OH-stretching mode in pure water

^

0

water-bonded

J

^^HMI

\

V

^\i\. non water-bonded

3100

3200

3300 3400 3500 Wavenumbers (cm"')

3600

Fig.l. A typical time-frequency pump-probe scan (50% of water in acetonitrile). Thick solid line in the 2D plot depicts the level at which the signal decays by a factor of 1/e. Triangles at the top plot depict the linear absorption spectrum for this concentration, while the open squares and circles show the spectral components corresponding to water-bonded and non water-bonded OH-oscillators, respectively. A schematic illustration of the microscopic structure is depicted at the right side.

387

n O

water-bonded non water-bonded

E (cm')

•^'^ M

3550-, 3400. 2900-

Water-bonded

Non water-bonded y t ^ l ps

i 0.7-3 P^ ,,'-'' 12ps Bending overtone

>' '

0.2

0.4 0.6 Molar fraction of water

0.8

Overa

1 cooling

Fig.2. Left: concentration dependence of population lifetimes for non-water-bonded (circles) and water-bonded (squares) OH-oscillators. The solid curves represent the results of simulations. Right: schematic representation of the energy relaxation pathways for the OH-stretch vibrational mode of water molecules in acetonitrile solution. involves the overtone of the bending mode [8,9]. Hydrogen bonding is found to accelerate the population relaxation from 3 ps in dilute solutions to 700 fs in neat v^ater, since the energy overlap between the donor and acceptor modes increases. For the OH oscillators that have not initially been water-bonded, there are two relaxation pathways. First, the oscillator can relax directly to the first overtone of the bending mode with a 12-ps lifetime. Second, the water molecule first rotates (-2 ps as derived from rotational anisotropy experunents), then forms a hydrogen bond to a neighboring water molecule, becomes hydrogen-bonded, and finally relaxes via resonance with the bending mode overtone. While the second path is much faster that the first one, it requkes another water molecule in close proximity and therefore becomes only efficient at high concentrations. The rate-limitmg step ui this process is molecular reorientation that occurs at a 2 ps time scale.

References 1 S. Yeremenko, M. S. Pshenichnikov, and D. A. Wiersma, Chem. Phys. Lett. 369, 107 (2003). 2 J. Stenger, D. Madsen, P. Hamm, E. T. J. Nibbering, and T. Elsaesser, J. Phys. Chem. A106, 2341 (2002). 3 C. J. Fecko, J. D. Eaves, J. J. Loparo, A. Tokmakoff, and P. L. Geissler, Science 301,1698(2003). 4 H.-K. Nienhuys, S. Woutersen, R. A. van Santen, and H. J. Bakker, J. Chem. Phys. Ill, 1494(1999). 5 A. J. Lock and H. J. Bakker, J. Chem. Phys. 117, 1708 (2002). 6 H. Graener, G. Seifert, and A. Laubereau, Chem. Phys. 175, 193 (1993). 7 D. Cringus, S. Yeremenko, M.S. Pshenichnikov, and D.A. Wiersma, J. Phys. Chem. BIOS, 10376 (2004). 8 D. D. Dlott, Chem. Phys. 266, 149 (2001). 9 C. P. Lawrence and J. L. Skinner, J. Chem. Phys. 117, 5827 (2002). 388

Cascaded energy redistribution upon O-H stretching excitation in an intramolecular hydrogen bond Karsten Heyne\ Milena Petkovic^, Erik T.J. Nibbering\ Oliver K i i W and Thomas Elsaesser' ^ Max Bom Institut fiir Nichtlineare Optik und Kurzzeitspektroskopie, Max Born Strasse 2A, D-12489 Berlin, Germany, E-mail: [email protected] ^ Freie Universitat Berlin, Institut fiir Chemie, Physikalische und Theoretische Chemie, Takustr. 3, D-12195 Berlin, Germany, E-mail: ok@ chemie.fu-berlin.de Abstract. We demonstrate in a combined two-color pump-probe and quantum dynamical study that population of the 0-H stretching oscillator of a medium-strong intramolecular hydrogen bond is redistributed along the 0-H bending vibration. Nonequilibrium vibrational excitations of molecules in the condensed phase decay by intramolecular vibrational redistribution (IVR) and subsequent energy dissipation to the surrounding solvent (VER). Understanding the pathways of ultrafast IVR is relevant for elucidating both vibrational couplings and energy exchange in chemical reactions. Here, we report a femtosecond two-color infrared pump-probe study of 0 - H stretching (VQH) and 0-H bending (5OH) excitations of phthalic acid monomethyl ester (PMME) in nonpolar solution. We demonstrate that the intramolecular vibrational energy redistribution pathway from the VQH to the 5oH oscillator is an efficient relaxation channel for this intramolecular mediumstrong hydrogen bonded system. Moreover, we perform quantum chemical and quantum dynamical calculations to analyze the cascaded energy redistribution pathways using a relaxation mechanism involving simultaneous energy release into intramolecular and solvent modes Ultrafast vibrational dynamics were studied in pump-probe experiments with two independently tunable 100 fs pulses at a repetition rate of 1 kHz. Different pump-probe schemes were applied: Excitation of the VOH? C = 0 stretching (vco) and 5oH oscillator and probing the VQH and 5OH oscillator. Results of measurements performed in the range of the VQH band are presented in Fig. la. The spectrally integrated change of absorbance is plotted as a function of time delay between pump and probe pulses centered at 3000 and 2200 c m ' \ respectively. At this spectral position the contribution of the VQH absorption is negligible. Since spectral diffusion is negligible, the observed increased absorption is due to the 0-H excited state absorption which displays an instantaneous rise and a fast decay superposed by oscillatory signals. The data are well reproduced by a kinetic model consisting of an instantaneous rise and an exponential decay with a time constant of 220 ± 80 fs, convoluted with the cross correlation of pump and probe. On top of the incoherent dynamics oscillatory signals with a frequency of

389

(d)

(e) (1,0.0,0)

4

H-Q

^^H

"^

H

• 1C~H

(0.1,2,0) (0.1,1.1) (0,2,0,0) (0,1,0.2)

C—C

\ 0-H"

O

^"3

(0,1,1,0) (0,1,0,1)

(0 (0,0,2,0) (0.1,0,0) : — ioioii.i) ""f •!«««.»«- (0.0,0.2)

0

(0.0,2.0) : (0.0,1.1) ' (0,1,0.0) (0,0,0,2)

•• (0,0,1.0) (0,0,0,1)

^ (0,0,1,0) ' (0,0,0,1)

. (0,0,0,0)

. (0,0,0,0)

1000 2000 Delay Time (fs)

Fig. 1. (a) solid line: Transient of the v=l state of the VQH at 2200 cm'^ ; (b) Fourier intensity of the oscillatory signal; (c) Transients probing the 5OH upon excitation of: VQHJ solid lines; 5OH? circles; Vco? dashed lines; transients are scaled at the slow dynamics from 10 to 40 ps. (d) schematic of PMME (e) Diabatic level scheme and cascaded energy flow after VQH excitation, (f) Diabatic level scheme and energy flow after 5OH excitation. 100 cm"^ (Fig. lb) are detected, indicating coherent wavepacket motions [1]. In Fig. Ic we present transients measured in the spectral range of the 5OH absorbance upon excitation of the VQH mode, Vco mode and 5OH mode. One finds transient bleaching signals of the steady state 5OH absorption at 1415 cm"^ plotted in Fig. Ic (negative signals) and enhanced absorption signals at lower frequencies with a maximum at 1390 cm'^ (positive signals in Fig. Ic). After the overlap of pumpprobe pulses a rate-like increase and decrease of absorption is well reproduced by an exponential decay with time constants of 800 ± 100 fs and 7 ± 1 ps (Fig. Ic). In contrast, upon excitation of the Vco mode, the 5OH only shows slow dynamics. Our data demonstrate that energy relaxes efficiently from the V(VOH)=1 to the V(5OH)=1 vibration. Direct excitation of the V(5OH)=0 to 1 transition shows a simultaneous rise of bleaching (1415 cm'^) and enhanced absorption (1390 cm'^) and a fast decay to 30% of its maximal amplitude, followed by a slow decay (Fig. Ic). We attribute the fast 800 fs decay of the signals to stimulated emission (at 1415 cm-^) and red shifted excited state absorption (at 1390 cm"^), as a consequence of the anharmonicity. The subsequent decay of 7 ps originates from absorption of the 5OH fundamental transition transiently red-shifted by anharmonic coupling to other modes, that are populated through 5OH relaxation. Transients measured upon

390

excitation of the v(vco)=0 to 1 transition exhibit only the slow dynamics of the SQH (Fig. Ic). Thus, in the latter case the 5OH vibration remains unexcited. In contrast, excitation of the V(VOH)^0 to 1 transition results in very similar transients at 1415 cm'^ and 1390 cm'^ as observed upon 5OH excitation. The pump-probe signals clearly show the 800 fs dynamics on top of the slow dynamics indicating that the 5oH vibration is excited [2]. Comparison of the signal amplitudes with the number of molecules excited lead to an estimation of more than 30 % of VQH excitation energy relaxes over the 5OH mode. The vibrational energy relaxation in solution was modelled using the systembath approach of condensed phase quantum dynamics [3]. Geometry optimization and normal modes for representing the Hamiltonian have been calculated using the DFT/B3LYP (6-31+G(d,p)) method. The potentials including mode correlations up to the fourth order were obtained from single point calculations on a grid supplemented by anharmonic force fields [2]. We observed two modes, YOHI and yoH2 between 700 and 800 cm'^ with combination transitions close to resonance with the 5oH vibration and large third order force constants. Combinations of these two modes with the 5OH are almost resonant with the VQH fundamental. These four modes and a low-frequency hydrogen bond mode comprise our 5D model for vibrational energy flow in PMME. A diabatic representation is used with respect to the four fast modes where the diabatic states are characterized by the quantum numbers: (V(VOH), V(5OH)5 v(yoHi), v(yoH2))- The diabatic level scheme is illustrated in Fig. le,f The vibrational energy relaxes from the VQH excitation (1,0,0,0) over (0,1,1,1), (0,1,2,0) and (0,1,0,2) to (0,1,1,0) and (0,1,0,1) to the 5OH mode (0,1,0,0). The latter relaxes via modes YOHI and yoH2 to the ground state (Fig. If).This cascaded model reproduces with identical parameters the measured decays of the VQH and 5OH- We can exclude the alternative relaxation mechanism from the VQH to the low-frequency modes, because of the small overlap of the wavefunctions. In conclusion we have demonstrated that the fast population relaxation of the VQH vibration in a medium-strong hydrogen bond originates from a cascaded energy redistribution over the SQH and other accepting modes between 700 and 800 cm"', that couple strongly to the VQH vibration.

References 1 D. Madsen, J. Stenger, J. Dreyer, P. Hamm, ETJ. Nibbering and T. Elsaesser, Bull Chem. Soc. Jpn. 75, 909, 2002. 2 K. Heyne, E.T.J. Nibbering, T. Elsaesser, M. Petkovic and O. Ktihn, J. Phys. Chem. A 108,6083,2004. 3 V. May and O. Kiihn, Charge and Energy Transfer Dynamics in Molecular Systems, 2""^ Revised and Enlarged Edition, Wiley-VCH, Weinheim, 2004.

391

Pure intermolecular energy relaxation of the OH bending vibration of water molecules dissolved in organic liquids Gerhard Seifert, Toralf Patzlaff, Katarzyna Paradowska-Moszkowska, and Heinrich Graener Martin-Luther-University Halle, Physics Institute, Hoher Weg 8, D-06099 Halle, Germany E-mail: [email protected] Abstract. A strong solvent dependence has been found investigating the intermolecular OH bending relaxation of water molecules dissolved in various organic liquids. From these resuhs, the vibrational relaxation of monomeric water molecules in the liquid phase can be explained comprehensively.

Water molecules have only three vibrational fundamental modes, and provide a unique opportunity to study the vibrational relaxation of a high-frequency "gate mode": V2 is the lowest vibrational mode of H2O, but has a frequency of Eyib « 1600 cm"\ which is approximately a factor of 8 larger than thermal energy at ambient temperature (kfiT corresponds to 208 cm'^ at 300 K). So, looking at binary mixtures of water in relatively nonpolar solvents, it is very improbable that the OH bending vibration transfers its energy directly into non-vibrational, lowfrequency modes of the surrounding liquid. Instead, an intermolecular process populatmg suitable vibrations of the solvent molecules can be assumed as the main relaxation process, which in turn lets one expect a considerable dependence of the water OH bend relaxation on the resonance structure of the solvent molecules. The situation is different in neat liquid water, the relaxation dynamics of which was studied extensively in the last years by different methods with fs time resolution: in neat water (or isotopic mixtures) the complex hydrogen bond network provides a large density of states of low-frequency modes, which can accept the relaxed energy, giving rise to population lifetimes of the order of a picosecond or faster [1,2]. It is generally accepted in this context that the relaxation of OH stretching quanta involves the bending mode V2, as was concluded previously from experiments on neat water [2], monomeric water molecules in non-polar solvents [3] and theoretical calculations [4]. Here we report a series of experiments on the OH bending relaxation of (monomeric) water molecules in non-polar solvents using time-resolved picosecond IR pump-probe spectroscopy. We investigated H2O diluted in various halogenated and/or deuterated methane and ethane derivates. All these organic liquids act as relatively non-polar solvents, where the water molecules are dissolved dominatuigly in monomeric form, often accompanied by a quite low total solubility. The experimental setup used for our study is an IR pump-probe

392

experiment applying two separately tunable pulses of 2.5 ps duration. The tuning range for pump and probe pulses is ^ 1200 - 4000 cm"\ With this setup, excess vibrational population can be created on both OH stretching and bending modes of H2O; the following redistribution of vibrational energy within the molecule and to the solvent is then monitored by the delayed probe pulses via transient absorbance changes, which are directly proportional to population differences between two vibrational levels [5]. Using this setup, it could be shown that the vibrational energy lifetime of the OH bending mode of H2O in CHCI3 and CDCI3 occurs via an intermolecular energy transfer process to the solvent [6], but with very different relaxation rate constants (decay times of 28 and 8.5 ps, respectively) despite the obvious similarity of the solvent molecules. This similarity could be shown also by the second order rotational correlation times T^^\ which were identical within experimental accuracy for both chloroform isotopes: x^^^ = (2.7±0.7) ps was

100-

'

1

'

I

'

1

'

r

1

^

1

1

1

1

1

'

1

GO

- • - T ^ ^ ^ (OH Stretch)

_^o 6

//'

a.

'"^\% ^

0 0 Q

1

'

1

-^y

'%

320

o»

\i ^v^ • ^

'"^r.

'

i% i

'%

-

300

]

\ \ \

^,''^' ^^ ^.'"'' ^^ ^^/''"' ^^ ^ ^^^

hro 10-

J

0'"

- 0 - T2o(OHbend)

E

H

0.

1

340

'

1

360

'

T "— 1

380

1

400

<

1

1

420

T

440

'

460

-1

Frequency Difference (v -2v ) / cm

Fig. 1: Measured vibrational lifetimes of the OH stretch ensemble (circles) and the OH bending mode (square) of H2O in various solvents, plotted as a function of the frequency difference Vi - 2v9; see text for explanation. determined from exponential fits of the induced dichroism [7], demonstrating that not only the interaction potential, but also the interaction probability between water and the two different solvents are fairly identical. This result already shows clearly that the vibrational resonance structure of the acceptor molecules plays a decisive role for the energy transfer. However, a close inspection of the acceptor

393

modes with low total number of vibrational quanta (which is a prerequisite for considerable transition probability) showed that no simple assignment of, e.g., one dominating accepting mode is possible. The situation becomes even more complicated taking into account additional results for relaxation times (T20) obtained on various binary mixtures of H2O in organic liquids, as shown in Fig. 1 (solid squares). The values are ordered along the frequency difference Av = (vi - 2V2) between the water symmetric stretching vibration and the overtone of the bending mode. Av is a very reliable and sensitive probe for the interaction strength between solute and solvent [3], as can be seen looking at the vibrational effective lifetimes of the OH stretch ensemble (given as open circles in Fig. 1), which exhibit a very clear, nearly monotonous decrease with decreasing Av (increasing interaction strength). In contrast to that, the relaxation rate of the OH bending mode shows no correlation at all with Av, but varies significantly from solvent to solvent including the already mentioned, quite unexpected difference of a factor of 3 between the chloroform isotopes CHCI3 and CDCI3. We are already preparing series of comparable experiments on the D2O and HDO bending modes, which should allow us to gain more insight into the accepting modes most probably involved in this particular intermolecular energy transfer, and to promote the theoretical understanding of the process in general. Apart from the problem that a detailed understanding of the characteristic solvent dependence observed is still lacking, the results of our study can be used to explain comprehensively the vibrational energy relaxation scheme of monomeric H2O molecules. The picture used previously for both neat and monomeric water after OH stretch excitation (relaxation via bending overtone and fimdamental) is strongly supported. Furthermore we have used an approach of populational 2D-IR spectroscopy applying pump and probe pulses in quite different spectral regions; this yields the anharmonic constants of water molecules which turn out to be very close to the values known from the gas phase in all of our solvents. The results will be presented in detail in a forthcoming publication. The 2D-IR-spectroscopy using vibrational population will be of particular interest for future structural studies on larger (biological) molecules.

References 1 A. J. Lock, S. Woutersen, and H. J. Bakker, J. Phys. Chem. A 105, 1238, 2001. 2 A. Pakoulev, Z. Wang, Y. Pang, and D. D. Dlott, Chem. Phys. Lett. 380, 404, 2003. 3 H. Graener, G. Seifert, and A. Laubereau, Chem. Phys. 175, 193, 1993. 4 R. Rey and J. T. Hynes, J. Chem. Phys. 104, 2356, 1996. 5 G. Seifert, T. Patzlaff, and H. Graener, Phys. Rev. Lett. 88, 147402, 2002. 6 G. Seifert, T. Patzlaff, and H. Graener, J. Chem. Phys. 120, 8866, 2004. 7 G. Seifert, T. Patzlaff, and H. Graener, Vib. Spectrosc. 23, 219 2000.

394

Time-resolved spectroscopy of an azobenzene derivative with a small S-S^ energy gap Masahide Hagiri^ Nobuyuki Ichinose^ Toshihiro Nakayama\ Changli Zhao^, Hiroaki Horiuchi^ and Hiroshi Hiratsuka^ ^ Department of Chemistry and Materials Technology, Faculty of Engineering and Design, Kyoto Institute of Technology, Matsugasaki Gosho-Kaidocho, Sakyo-ku, Kyoto 6068585,Japan E-mail: [email protected] ^ Department of Chemistry, Gunma University, Tenjin-cho, Kiiyu, Gunma 376-8515, Japan Abstract. Femtosecond time-resolved absorption spectroscopy on the relaxation dynamics of rrart5-(4-methoxyphenylazo)-4'-nitrobenzene has indicated the rapid internal conversion of the S2 state which is facilitated by the small energy gap between the 82(71,71*) and Si(n,7r*) states. 1, Introduction Azobenzene (AB) and its derivatives have been interested in terms of photochemical cis-trans isomerization [1-4]. Recently it is concentrated whether the rotational motion around N=N double bond in the 82(71,71*) state is existed [2].

.-lOJ O-^

O2N

MNAB

+

OMe

^OMe

O2N

^ ^

^;\^OMe

O—N

Q-

Fig. 1. We have introduced a pair of redox-active substituents at the 4,4'-positions of AB, which may cause charge-transfer (CT) character in the excited states, resulting in reduction of the N=N bond order. A quinoid structure with a N-N single bond character would appear as a resonant form of the CT state of/ra«5-(4methoxyphenylazo)-4'-nitrobenzene (MNAB), which is expected in the 82 state (Fig. 1). This paper deals with the photophysic of MNAB, especially in the 82 and 81 states, using femtosecond laser photolysis. 2. Results and Discussion Experimental details for femtosecond laser photolysis are described previously [4]. The pump pulse (400 nm) excites MNAB to the 82 state. Time resolved absorption spectra of MNAB in THF are shown in Fig. 2(a). The spectrum unmediately after the excitation exhibited a broad absorption band around 630 nm, which is gradually blue-shifted to 575 nm. It is also clearly found that the rise time at 575 nm is longer than the instrumental response time (250 fs). Furthermore, after the build-up of this absorption band decayed exponentially with a mean lifetime of 2.0 ± 0.2 ps, as shown in Fig. 2(b) (o).

395

-

(b)

- ^

I^

0

^.^ 5o

i

400

450

500

550

600

650

Wavelength / nm

700

-2.0 0.0

1

1

.1

2.0 4.0 6.0 Time / ps

1

8.0 10.0

Fig. 2. (a) Femtosecond transient absorption spectra of MNAB in THF upon 400nm excitation, (b) Decay curves of transient absorption monitored at 575 nm upon 400nm (o) and 266 nm (A) excitation. The obtained results are almost similar to those reported for the Sn^^Si absorption of substituted azobenzenes [1,3]. Thus, it is reasonable ascribed the broad absorption band to the Sn^<—Si absorption of MNAB. However, the decay time observed at the peak (570 nm) is different from that observed at 700 nm ( t = 0.5-1.0 ps). It is suggested that absorption of S2 state is ovelapped m the region of 700 nm. Recently Tahara at al.[2] reported a broad (offset-like) absorption immediately after the S2<—So excitation of AB, which is assignable to absorption of S2 state. In contrast to their observations, it seems unlikely to assign the broad absorption of <600 nm to the Sn^<—S2 absorption of MNAB. Considring the low fluorescence quantimi yield (^f < 1x10"^) even in 77 K MTHF matrix and the radiative rate constant (2.8 x 10^ s'^), the lifetime of the S2 state is reasonably calculated to be shorter than 35 fs. The decay time constant (0.5-1.0 ps) at 700 nm could be too slow to assign it to the Sn^^S2 absorption. The rise time of 0.5 ps for the Si state would be considerably longer than that expected for the small Si$2 energy gap (4500 cm"^) in comparison with that of AB (10000 cm'^). From these results, it is safely concluded that the <600 nm band is assigned to the vibrationally excited Si state and the origin of the blue-shift of the Sn^*—Si absorption band is attributed to vibrational relaxation in the Si state after fast S2-^Si internal conversion. Such a blue-shift has been also observed during the relaxation of the Si state for carotenoids after S2 excitation, whose S1-S2 energy gap (6000 cm'^) is comparable with that of MNAB. For carotenoids, it is reported that intramolecular vibrational redistribution in the Si state takes place with a time constant of 600 fs [5]. Since the lifetime of the Si state for MNAB is very short, it is noteworthy that the excess vibrational energy will not be transferred to solvent molecules within the lifetime of the Si state for MNAB. As clearly shown in Fig. 2(b), fiirthermore, no excitation energy dependence of the decay kinetics is observed. From these results, our observation strongly suggests that the S2 state of MNAB has a shorter lifetime in comparison with that of AB. The quantum yield

396

for trans-^cis photoisomerization of MNAB upon the S2 excitation was, however, comparable to that of AB. This would be indicated that there is no effect on change of efficiency for N=N bond rotation caused by the decrease of bond order of the N=N bond as expected for the contribution of the CT character for MNAB. However, the CT character for MNAB in the excited state seems to lower the S2 level, resulting in the decrease in the S1-S2 energy gap. The spectrum at 2.0 ps delay after the excitation exhibited two absorption bands peaked at 450 and 575 nm. Compared with previous results, during disappearance of Si state the 450-nm band obtained here could be due to vibrationally excited MNAB in the So state and is decayed with 7.0 ps for the vibrational cooling time [1-3]. Table 1. Solvent effect on the decay kinetics of MNAB Solvent «-Hexane THF Acetonitrile

S1-S2 energy gap /cm'^ 5000 ±200 4800 ± 600 4900 ± 600

S2-^Si internal conversion / ps -0.8 -0.8 -1.0

Si lifetime / ps 2.5 ± 0.3 2.0 ± 0.2 1.2 ±0.1

So cooling /ps 8.3 ± 0.8 7.0 ±0.7 2.8 ±0.3

We also observed these kinetics in various solvents, whose time constants of Si state and vibrational excited So state of MNAB are summarized in Table 1. Since Si lifetime in acetonitrile is shorter than that in THF, it may be suggested that the CT character for MNAB is participated in Si state, which reduces the SoSi energy gap. Furthermore, the time constants for the So cooling in THF are significantly different from those obtained in acetonitrile. This observation could be explained by the low So-Si gap caused by the CT character from introduction of substituents in AB or by change in numbers of accepting mode of the solvents. 3. Conclusions Time-resolved absorption spectroscopy of an azobenzene derivative with push-pull type substituents has revealed that the S2 state undergoes rapid internal conversion to give vibrationally excited Si state. This behavior is attributed to the small energy gap between the Si and S2 states, to which the CT character plays an indirect role rather than direct facilitation of the cis-trans isomerization. These results are indicative of the absence of the rotational isomerization mechanism in the S2 state as has been accepted for azobenzene derivatives.

References 1 2 3 4

J. Azuma, N. Tamai, A. Shishido, T. Ikeda, Chem. Phys. Lett. 288, 77 (1998). T. Fujino, S. Y. Arzhantsev, T. Tahara, Bull. Chem. Soc. Jpn. 75, 1031 (2002). Y. Hirose, H. Yui, T. Sawada, J. Phys. Chem. A 106, 3067 (2002). M. Hagiri, N. Ichinose, C. Zhao, H. Horiuchi, H. Hiratsuka, T. Nakayama, Chem. Phys. Lett. 391, 297 (2004), and reference cited therein 5 G. Cerullo, G. Lanzani, M. Zavelani-Rossi, S. De Silvestri, Phys. Rev. B 63, 241104(2001).

397

Photo-thermalization dynamics of azulene in supercritical fluids studied by the transient grating method Yoshifumi Kimura\ Yoshinori Yamamoto^, and Masahide Terazima^ ^ Division of Research Initiatives, International Innovation Center, Kyoto University, Kyoto 606-8501, Japan E-mail: [email protected] ^ Department of Chemistry, Graduate School of Science, Kyoto University, Kyoto, 6068502,Japan Abstract. Transient grating spectroscopy has been applied to the study on the vibrational energy relaxation of azulene in supercritical xenon and ethane. The roles of the V-T and V-V energy transfers are discussed.

1.

Introduction

The vibrational energy relaxation (VER) process of photo-excited molecules in solution have been studied by using various kinds of ultrafast spectroscopic methods. Previous studies, however, mostly monitored the decay of the excess vibrational energy of solute molecules. There are few studies which monitor the energy of the solvent molecule that is the energy acceptor, although recent development of IR-Raman technique revealed interesting features of the mode selective energy flows in pure solvents[l]. In the present study, we have applied the transient grating (TG) spectroscopy to the monitor of the temperature increase of the solvent after the energy relaxation of the vibrationally excited azulene in supercritical xenon and ethane for the first time. The buildup rate of the translational kinetic energy of the solvent molecules was measured from the temporal profile of the acoustic wave[2,3]. Supercritical fluids are quite useftil because important factors which affect the VER process such as the coUisional frequency can be largely varied by the fluid density. By comparing our results with previously reported cooling rates of the vibrational energy of azulene from the transient absorption (TA) measurements [4-6], we found that the vibrationalvibrational (V-V) energy transfer from azulene to the solvent molecules and the energy transfer between solvent molecules are important for the thermalization process in solution of polyatomic molecules.

2.

Experimental

The experimental setup is shown in Fig. 1. An output pulse (570 nm, 4 |iJ) from a home-built optical parametric amplifier (OPA), which was pumped by a doubled output (388 nm) of an amplifier integrated fiber laser (Clark-MXR, CPA-2001, 1

398

Fig. 1. Schematic drawings of the optical ahgnment and the high pressure optical cell. ND: Neutral density filter. P: Polarizer. X/2: Half wave plate. L: Lens. PM: Photo-multiplier.

kHz, 775 nm, 0.7 mJ), was divided into two pulses by a beam splitter. One of them was temporally delayed by an optical delay line, and was used as a probe pulse. The rest pulse was further divided into two pulses with equal energy and used as pump pulses. We used counter-propagating alignment of the pump pulses in order to obtain the largest grating wavenumber (q=2n n sin(^/2)//l; X\ pump pulse wavelength, 6: crossing angle, n: refractive index), since the time resolution of the system is dominantly determined by the sound frequency {co) given by qc (c: sound velocity of the fluid). The probe pulse was incident on the sample at an angle of ca. 3 degrees to the pump pulse. The diffracted signal was detected by a photo-multiplier and averaged by a boxcar. The inset figure in Fig. 1 illustrates a high pressure optical cell used for measurement[7].

3.

Results and Discussion

Figures 2(a) and (b) show typical examples of the TG signal of azulene in xenon(a) and in ethane(b). After the photo-excitation to the Si state, a very fast internal conversion (within ca. 1 ps) occurs and a vibrationally hot ground state azulene molecule is produced. This hot azulene molecule dissipates its excess vibrational energy to the solvent, which produces the TG acoustic signal(Fig. 2). In order to extract the temperature rise time from this signal, we assume that the energy dissipation process to the solvent translational energy is described by a single exponential fiinction with the decay constant rtemp"^ The time profile of the acoustic TG signal /TG(0 is given as follows[2]: ^TG(0 ^ t((^thlco)sm{(Dt)-cos{(ot))x exp(-dj)+ exp(-k,j)]l{k,^ + o)^) - [((V^^temp)sinM-cosM)>< exp(-<0+exp(-^/r,,„jy^^^^ where A^h = Dx\,q^ (Ah-the thermal diffusivity) and d^ is the acoustic dumping constant, respectively. The fittings of the signal to Eq. (1) are shown by the solid lines in Fig. 2, which simulate the signal very well. The uncertainty of rtemp was estimated as ±15% of its value. In Fig. 2(c), the temperature rise rates obtained in this study (rtemp'^) are compared with the energy dissipation rates obtained from the TA measurements (^TA^)[6]. In the case of the xenon, the values of rtemp" ^ are close to TTA'^ estimated from the interpolation of the density dependence. On the other hand, in ethane there is a significant difference between them: i.e., rtemp"^ is about half of r^^^ compared at the same reduced density.

399

J

•

TG(C"H;» 2

1^ o

6

O TAtC'^H^) - n TA(Xe) 1 • TGCXc) 0 04 0 0.03

O

. (c)

•• • •

*••

• 400

800

1200

time / ps Fig. 2 (a)(b) Typical examples of the TG acoustic signals (circles) (a) in Xe at 383 K and 44.7 MPa (p, = 1.78) and (b) in ethane at 383 K and 17.5 MPa (p^ = 1.27). The solid lines show the fit to Eq. (1) (rtemp = (a)134 ps, and (b)38 ps, respectively.) The broken lines show the trial fit under the assumption of Ttemp = TJA ((a) 125 ps and (b) 20 ps[6], respectively), (c) Solvent density dependence of the temperature rise rate (xtemp"^) obtained from the TG measurements (filled symbols) and the reported VER rates (TJA'^) obtained from the TA measurements (open symbols) [6] in Xe and in C2H6, respectively. We consider that this difference comes from the different energy transfer channels from azulene to the solvent molecules. Since xenon is a monatomic molecule, only the V-T energy transfer is possible. Therefore the dissipated energy fi-om azulene immediately becomes the translational energy of the solvent, which can be detected by the TG acoustic method. On the other hand, ethane has rotational and vibrational degrees offi-eedom,both of which can accept the excess energy of azulene. If substantial amount of the excess energy of azulene is transferred to the vibrational energy of the solvent, the increase of the translational energy may be delayed due to the V-T energy transfer process between "solvent" molecules. We consider that the energy transfer process in vibrationalfi*eedomis the reason for the difference between rtemp and TJAAcknowledgements. This work is supported by the Grant for Basic Science Research Projectsfi-omthe Sumitomo Foundation.

References D. D. Dlott, Chem. Phys. 266, 149, 2001. L. Genberg, Q. Bao, S. Gracewski and R. J. D. Miller, Chem. Phys. 131, 81, 1989. M. Terazima, M. Takezaki, S. Yamaguchi, and N. Hirota, J. Chem. Phys. 109, 603, 1998. L U. Sukowski, A. Seilmeier, T. Elsaesser , and S. F. Fischer, J. Chem. Phys. 93, 4094, 1990. K. E. Schultz, D. J. Russell, and C. B. Harris, J. Chem. Phys. 97, 5431, 1992. D. Schwarzer, J. Troe, M. Votsmeier, and M. Zerezke, J. Chem. Phys. 105, 3121 1996. Y. Kimura, N.Saga, M. Terazima, and N. Hirota, Anal. Sci. 17, s234, 2001.

400

Vibrational Self-Trapping in an a-Helix Julian Edler^ Vincent Pouthier^, Cyril Falvo^, Rolf Pfister^ Peter Hamm^ ^ Universitat Zurich, Physikalisch Chemisches Institut, Winterthurerstrasse 190, 8057 Zurich, Switzerland ^ Laboratoire de Physique Moleculaire, UMR CNRS 6624, Faculte des Sciences, La Boulie, Universite de Franche-Comte, 25030 Besangon Cedex, France Abstract: Infrared pump probe experiments on the NH mode of an a-helix reveal two excited state absorption signals, which are an explicit indication of self-localized vibrational excitons.

The formation of localized states in solids, caused by the interaction of electrons or excitons with lattice vibrations, can result in exciting energy transport properties and novel optical characteristics. Such localized states are generally known as polarons and the phenomenon is commonly called self-trapping or selflocalization. Self-trapping has been observed in various materials. In particular quasi-one-dimensional structures, such as halogen-bridged metal complexes or hydrogen bonded molecular crystals, are ideal systems to study self trapping. In the first case the excitations are electronic, whereas they are vibrational in molecular crystals. In proteins vibrational energy is transported with an astonishing efficiency over remarkable long distances. It has been suggested that vibrational self-trapping in a-helices, the most common secondary structure of proteins, plays an important role in the energy transport and storage [1]. An a-helix consists of a helical chain of peptide units, which is stabilized by quasi-one-dimensional chains of hydrogen bonds. Since this system is translational invariant, one describes vibrations of the individual peptide units along the hydrogen bonded chain as lattice phonons. Vibrational self-trapping in an a-helix is caused by two different coupling mechanisms: excitonic coupling and vibronphonon coupling [2]. Excitonic coupling is caused by the electrostatic interactions between the individual molecular oscillators, and leads to the delocalization of a vibrational excitation. This state, called vibron, is coupled, in turn, to lattice phonons through an anharmonic (nonlinear) term, which is mediated by the hydrogen bonds. As a consequence, the initially delocalized vibron collapses to form a self-localized state. Vibrational self-trapping has been investigated in detail for hydrogen bonded molecular crystals such as acetanilide, which is considered to be a model system for a-helices. [2,3,4,5]. However, so far no clear evidence for self-trapping in real a-helices has been found. We decided to study a-helical Polyy-benzyl-L-glutamate (PBLG, MW: -22000 g/mol), because this polypeptide forms extremely stable, long a-helices in both helicogenic solvents and films grovm from these solvents [6]. We performed femtosecond infrared pump probe experiments with pulses centered at 3200 cm'^ (spectral width 200 cm'^ pulse duration 130 fs FWHM) to excite the NH stretching mode in PBLG. Figure lb shows the absorbance change

401

Simulated Nonlinear Response

3000 3200 , Wavenumber [cm

3400

3000 3200 3400 Wavenumber [cm"^]

3000 3200 . Wavenumber [cm ]

Fig. 1. (a) Absorption spectra of a PBLG film and (b) corresponding nonlinear response for 600 fs delay time at 293 K and 18 K. (c) Absorption spectra of PBLG dissolved in Chloroform-D (with 3% TFA) at 293 K (helical conformation) and at 260 K (random coil) and (d) corresponding nonlinear signal for 700 fs delay time, (e) Simulated pump probe response using the theory of Re£ [8]. Insert: Schematic of energy levels. The arrows indicate the interactions with the probe pulse.

of a PBLG film, for pump probe delay times of 600 fs. Normally one expects to observe in a pump probe experiment a negative band, associated with bleach and stimulated emission and a positive band for the excited state absorbtion. We observe indeed a negative signal at the frequency of the NH fundamental (3280 cm"^). However we see not one, but two positive bands: at ~ 3160 cm'^ and 3005 cm'\ The two positive peaks have the same intensity and polarization dependence and decay both on a 1.6 ps time scale. When one lowers the temperature of a solution of PBLG in chloroform and trifluoroacetic acid (TFA), the conformation of PBLG changes from an a-helix to a random coil (cold denaturation) [7] (Figure 1 c,d). In case of the helical structure one observes the same spectral signature as in the PBLG fihn: one negative peak and two positive bands at lower frequencies. The nonlinear response of the randomly coiled PBLG is completely different. Here we observe the typical signal of a vibrational mode: one negative peak and one positive peak. The negative peak in the nonlinear signal and the NH band in the absorbance spectrum are both blue shifted with respect to the helical structure, reflecting the change in the hydrogen bond strength. We also performed temperature dependent pump probe experiments on PBLG films. In this case the molecule remains always in the helical conformation and the two positive peaks are still observed at temperatures as low as 18 K (Fig. lb). Thus the variation of the nonlinear signal with the helix-coil transition reflects the structural change, which is triggered by the temperature change. Hence the unusual signal with the two positive bands is a property of the helical structure. The helical structure is stabilized by chains of hydrogen bonds that mediate a nonlinear coupling between the NH vibration and (acoustic) lattice phonons. In other words the unusual signal is only observed when a couplmg between the NH mode and the lattice phonons is possible, i.e. a vibron-phonon coupling. To confirm that the signal is really induced by such a coupling we have excluded other, more traditional explanations. Multiphoton excitation was

402

excluded by performing experiments with varied pump pulse intensities. Narrow band pump (spectral width -30 cm"^) probe experiments on fully C-deuterated PBLG were used to prove that the signal is only associated with the NH mode. In order to exclude Fermi resonances, we have performed theoretical studies and temperature dependent two color pump probe experiments. The effect of the vibron-phonon coupling on the nonlinear signal of a-helical PBLG can be described by a model, which was recently developed by Pouthier and coworkers [8]. The theory is based on the Davydov model [1] and considers the excitonic coupling, the vibron-phonon coupling, and the intrinsic anharmonicity of the vibrational modes. Both the intramolecular anharmonicity and the phonon coupling favor the formation of two-vibron bound states [8]. The two-vibron energy spectrum exhibits three types of states (Fig. le): (i) two-vibron free states (TVFS) belonging to an energy continuum which correspond to two non-interacting vibrons (ii) two-vibron bound states I (TVBS-I) which refer to the trapping of two vibrons at the same site, and (iii) two-vibron bound states II (TVBS-II) which characterize vibrons trapped at nearest neighbor sites. The model is used to compute the pump-probe signal, which reproduces the experimental data very well (Fig. le). The calculated spectrum shows a single negative peak located at 3280 cm"^ and two positive peaks at 3005 cm'^ and 3160 cm"^ The negative peak refers to the bleach and stimulated emission between the ground state and the zero wave vector single-vibron state. This process overcompensates the absorption from the single-vibron state to the zero wave vector TVFS. In contrast, the positive peaks correspond to the excited state absorption from the zero wave vector singlevibron state to the zero wave vector TVBS-I and TVBS-II. Both the intensity ratios and the temperature dependence of the three bands reproduce the experimental data. In conclusion the temperature dependent pump probe experiments show unambiguously that the two positive peaks are a signature of the helical conformation. In the helical conformation the individual NH vibrations are correlated by acoustic phonons, enabling the formation of polaron states, i.e. two kinds of two-vibron bound states. When the helical structure is destroyed, the correlation vanishes and the two bound states disappear. As we have successftilly excluded all alternative explanations, we can conclude that we have observed a signature of vibrational self-trapping. To our knowledge this study is the first explicit experimental evidence for vibrational, self-trapped states in an a-helix.

References 1 A. S. Davydov, J. Theor. Biol. 66, 377 (1977). 2 A. C. Scott, Phys. Rep. 217, 1 (1992) 1 G. Careri, U. Buontempo, F. Galluzzi, A.C. Scott, E. Gratton, E. Shyamsunder, Phys. Rev. B 30, 4689 (1984). 2 J. Edler, P. Hamm, A. C. Scott, Phys. Rev. Lett. 88, 067403 (2002). 3 J. Edler, P. Hamm, J. Chem. Phys. 117, 2415 (2002). 4 P. Doty, J.H. Bradbury, A. M. Holtzer, J. Am. Chem. Soc. 78, 947 (1956). 5 J. A. Ferretti, B. W. Ninham, Macromolecules 3, 30 (1970). 8 V. Pouthier, Phys. Rev. E 68, 021909 (2003).

403

Infrared Photon-Echo Spectroscopy of Water: The Thermalization Effects Maxim S. Pshenichnikov, Sergey Yeremenko, and Douwe A. Wiersma Ultrafast Laser and Spectroscopy Laboratory, Department of Physicsl Chemistry, Materials Science Centre University of Groningen, Nijenborgh 4, 9747 AG Groningen, The Netherlands E-mail: [email protected] Abstract. The larger part of the nonlinear response in IR photon-echo and transient-grating spectroscopy on HDO-D2O mixtures at >l-ps delays is found to originate from the D2O refractive index modulation due to local volume thermalization.

1. Introduction The complex and even peculiar properties of liquid water are largely determined by the presence of a network of hydrogen bonds. The hydrogen bonds undergo continuous transformations that occur on sub-ps timescales [1-6]. The molecular vibrations are especially sensitive to the presence of the hydrogen-bond network. The microscopic structure and the dynamics of water manifest themselves in the IR vibrational spectrum, and, therefore, can be studied by methods of ultrafast IR spectroscopy. The technique of IR photon echo [3-7] allows obtaining the most direct information about the systems dynamics. It is well known that the hydrogen bonding in water is extremely sensitive to temperature. Excitation of a vibrational mode, followed by ultrafast population relaxation and energy thermalization, ultimately leads to a local increase of the temperature. Therefore, the delayed probe pulse mterrogates a system whose spectral properties have been modified by the absorbed energy of the first pulse(s), causing unpedunents ui the analysis and interpretation of the experimental results. In this contribution a comprehensive study of thermal effects in IR spectroscopy of liquid water, employing heterodyne-detected photon echo and transient grating (TO) techniques, is presented. We directly demonstrate that the main part of the thermal contribution to the TO signal originates from a refractive index modulation of the solvent (D2O). The model that includes energy absorption by the OH stretching mode, local temperature increase after population relaxation, and subsequent heat diffusion, perfectly describes the experunental data.

2. Results and discussion The TO signal of the OH-stretch vibrational mode of HDO molecules dissolved in D2O is shown in Fig. la by a solid line. The initial part of the signal decays with a time constant of -700 fs that is consistent wi th the population lifetime [8].

404

However, after reaching a minimum around 2 ps the TG signal begins to grow again and finally levels-off at -10 ps. Heterodyne-detected transients are shown in Fig. lb. At short delays (t23=500 fs), the response is mainly determined by the HDO molecules. This conclusion is made on the basis of signal position, shape, and decay constant: the contour is shifted from zero delay to positive values and its decay rate equals the OH-stretching vibration population relaxation. Also, the spectrum of the response coincides well with the pulse spectrum convoluted with the absorption spectrum (Fig.lc). In contrast, at long delays fe^lO ps) the characteristics of the response change drastically: it narrows in time, its phase shifts by --n (Fig.la, inset), and its spectrum becomes identical to the spectrum of the probe pulse (Fig.lc) which points at the instantaneous origin of the system response. This makes us conclude that the TG signal at long delays results from the grating imprinted in the solvent, that is, D2O. A schematic representation of the energy equilibration is depicted in Fig.2a. The population (absorption) spatial grating originally imprinted in HDO by the excitation pulses, relaxes with a lifetime of-700 fs (Fig. lb, open circles). The energy, stored in the grating, is released by the chromophore to the surrounding D2O molecules, giving rise to an increase in local temperature. This temperature raise leads to modifications of the shape and amplitude of the D2O stretching mode absorption band at 4.5 j^m and, hence, both absorption and refractive index spatial gratings are imprinted in the solvent. The former is of a little importance for the TG signal at 3 \xm because its amplitude scales with the inverse squared of the detuning (-1000 cm"^). However, the spatial modulations of the D2O refractive index decrease much slower, as the inverse detuning from the resonant frequency. Therefore, the probe pulse is scattered of the off-resonance grating written in the solvent refractive index (Fig. la, solid squares). This is reflected in the phase of the

\2h^

3(PS)

100 0 100 200 Delay t ( f s )

3000

3200

3400

Wavenumbers (cm"')

Fig.l. (a): Experimental TG signal (solid line) and amplitudes of its additive components: chromophore response (open circles) and solvent response (solid squares) as found from the analysis of heterodyne-detected TG data. Inset: the phase of the TG signal at t34=0. (b): Amplitudes (solid dots) and phases (open squares) of heterodynedetected TG signals. Results of simulations are depicted as shaded contours (amplitudes) and solid lines (phases), (c): Spectra (shaded contours) and phases (lines) of the corresponding signals as calculated by Fourier transformation of the experimental data. Closed and open circles show the pulse spectrum and its convolution with the system response, respectively.

405

Excited OH-oscillators

- "•- • t^^ scanned - O • t|3 scanned M(t)

1

2 3 Delay t,3 or 1^3 (ps)

Fig.2. (a): The energy equilibration scheme for photon echo experiments on HDO molecules dissolved in D2O. (b): Experimental echo-peak shift data for the fixed delays ti3 (solid circles) and t23 (empty circles), and theoretical simulations (dashed lines). Note that the EPS functions acquired while keeping the delays t^ and t23 fixed, are shifted along the vertical axis as a consequence of the relatively short excited-state lifetime (700 fs). heterodyned signal that is almost Tu-shifted compared to the signal phase at short delays (Fig. la, inset). The temporal build-up of the refractive index grating is well described by the macroscopic heat diffiision equation (Fig.la, dashed line) with a time constant of-1.5 ps. We applied the developed model to the analysis of the echo-peak shift experiments (Fig.2b) in order to extract mformation about the water dynamics. The peak in the EPS function around ~2 ps is explained as arising from interference between the chromophore and solvent responses. The delicate balance between the phases of genuinely nonlinear and thermal contributions as the delay t^ between the two excitation pulses is increased, leads to the enhancement of the integrated signal that is measured in the EPS experiment.

References 1 G.M. Gale, G. Gallot, F. Hache, N. Lascoux, S. Bratos, J.-C. Leicknam, Phys. Rev. Lett. 82, 1068 (1999). 2 A.J. Lock, S. Woutersen, H.J. Bakker, J. Phys. Chem. A 105, 1238 (2001). 3 J. Stenger, D. Madsen, P. Hamm, E.T.J. Nibbering, T. Elsaesser, Phys. Rev. Lett. 87, 027401 (2001). 4 J. Stenger, D. Madsen, P. Hamm, E.T.J. Nibbering, T. Elsaesser, J. Phys. Chem. A 106, 2341 (2002). 5 S. Yeremenko, M.S. Pshenichnikov, D.A. Wiersma, Chem. Phys. Lett. 369, 10713 (2003). 6 C. J. Fecko, J. D. Eaves, J. J. Loparo, A.Tokmakoff, P. L. Geissler, Science 301, 1698-1702(2003). 7 C.P. Lawrence, J.L. Skinner, Chem. Phys. Lett. 369, 472 (2003). 8 S. Woutersen, U. Emmerichs, H.-K. Nienhuys, H.J. Bakker, Phys. Rev. Lett. 81, 1106(1998).

406

Heterodyne 2D-IR Photon Echo Spectroscopy of Multi-Level OH Stretching Coherences in Hydrogen Bonds N. Huse\ B. D. Bruner^ M. L. Cowan^ J. Dreyer\ E. T. J. Nibbering^ T. Elsaesser^ and R. J. D. Miller^ ^ Max Bom Institut flier Nichtlineare Optik und Kurzzeitspektroskopie, Max Bom Strasse 2A, D-12489 Berlin, Germany E-mail: [email protected] ^ Departments of Physics and Chemistry, University of Toronto, 80 St. George St., Toronto, ON, Canada M5S3H6 Abstract. We explore the multilevel stmcture and vibrational couplings of hydrogen bonded 0-H stretching transitions in acetic acid dimer using femtosecond 2D-IR photon echo spectroscopy. Anharmonic coupling with low-frequency modes and Fermi resonances with overtone/combination levels dominate the complex ultrafast dynamics. Hydrogen bonding, i.e., the formation of a wQak bond between a hydrogen donor (an 0-H group) and a hydrogen acceptor group plays a key role in the microscopic structure of protic liquids, proteins and DNA. Liquids such as water form a multitude of hydrogen bonding geometries and coupling strengths, which display complex dynamics of dephasing and spectral diffusion that underlie the strongly broadened 0-H stretching bands [1]. For well-defmed hydrogen bonding geometries, as in cyclic dimers of carboxylic acids, the 0-H stretching bands show a complex envelope with a pronounced spectral substructure [2,3]. Although extensive theoretical work has been performed on resonant (excitonic) coupling of 0-H stretching oscillators, anharmonic coupling to low-frequency modes, and Fermi resonances to overtone/combination levels, a quantitative understanding and a separation of such contributions has not been possible. In particular, vibrational dephasing of 0-H stretching excitations in the presence of these couplings has remained rather unexplored. We have studied the ultrafast dephasing of coherent 0-H stretching excitations in cyclic dimers of acetic acid in CCI4 solution. These dimers contain two intermolecular hydrogen bonds, which represent an important model system for studying basic vibrational couplings. With homodyne detected two pulse photon echo measurements we have shown the rapid dephasing of the 0-H stretching oscillator and quantum beats [4] originating from anharmonic coupling with lowfrequency modes that strongly modulate the hydrogen bond distance [5]. We use a recently developed diffractive optic (DO) technique [6] to measure both the absorptive and dispersive components of the two-dimensional-IR spectra. This technique uses a DO to generate two passively phase-locked beam pairs: the two pump pulses (ki, k2) and the probe and local oscillator (k^, RLO), delayed with respect to each other by a time ^13. After the DO, the coherence time r is

407

generated by delaying k2 with respect to pulse ki, and the local oscillator with respect to pulse ka. Since the same retroreflective delay line is used to produce both delays, the phase noise correlations set up by the DO are preserved. Locking of the relative phases of the beams was observed to be better than /1/150. Furthermore, the local oscillator generated by the DO is automatically spatially overlapped with the echo signal for heterodyne detection, which we implement using spectral interferometry [7]. The spectral fringes are detected in a monochronometer using a HgCdTe array, directly giving the echo as a function of the detected frequency V3. The coherence time r is then scanned at constant population time T, and the signal is Fourier transformed along the r dimension to produce the excitation frequency dimension Vi. The heterodyne detected echo signal was measured using 85 fs mid-IR pulses tuned to a centre wavelength of 2950 cm'^ Figure 1 displays the signal detected on the array as a function of the delay r between the two pump pulses. Ridges are visible at frequency positions corresponding to prominent spectral features in the steady-state spectrum, indicating the frequency components responsible for the recurrences in the homodyne photon echo signal.

0

200 400 600 Coherence Time (fs)

0.2 0.4 Abs. (OD)

Fig. 1. Spectrally-resolved echo signal as function of r (for 7=0). The main features in the linear acetic acid spectrum (shown on the right) appear as ridges along the r direction. The solid line in the upper panel is the cross section through 2800 cm'^ For comparison the previously reported homodyne echo signal [4], is shown as dotted line. The amplitude of the measured 2D spectrum is presented in Fig. 2 (left panel) as a function of the two frequency variables Vi and V3. The most prominent features are the diagonal peaks corresponding to transitions near 2920 cm"^ and

408

2990 cm" along with their corresponding crosspeaks. Peaks due to the red-shifted excited state absorption also appear. While our homodyne echo measurements suggest the importance of Franck-Condon progressions due to anharmonic coupling with low-frequency modes [4], the observation of pronounced crosspeaks leads to the conclusion that Fermi resonance interactions of the 0-H stretching vibration with overtone/combination levels of fingerprint vibrations are also important to the substructure of the 0-H stretching band. This is confirmed by ab initio simulations revealing that the anharmonic couplings of the lowfrequency hydrogen bond modes are on the same order of magnitude as the Fermi resonances (Fig. 2, right panel).

2600

2800

3000

V3 (cm'^)

3200

2600

2800

3000

3200

V3 (cm"')

Fig. 2. Measured (left) and ab initio calculated (right) amplitude of the 2D IR spectra.

References 1 2 3 4

E.T.J. Nibbering and T. Elsaesser, Chem. Rev. 104, 1887, 2004. Y. Marechal, J. Chem. Phys. 87, 6344, 1987. C. Emmeluth, M.A. Suhm and D.M. Luckhaus, J. Chem. Phys. 118, 2242, 2003. N. Huse, K. Heyne, J. Dreyer, E. T. J. Nibbering and T. Elsaesser, Phys. Rev. Lett. 91,197401,2003. 5 K. Heyne, N. Huse, J. Dreyer, E. T. J. Nibbering, T. Elsaesser and S. Mukamel, J. Chem. Phys. 121, 902, 2004. 6 M. L. Cowan, J. P. Ogilvie and R. J. D. Miller, Chem. Phys. Lett. 386, 184, 2004. 7 J.-P. Likforman, M. Joffre, V. Thierry-Mieg, Opt. Lett., 22, 1104, 1997

409

A unified analysis of ultrafast vibrational and orientational dynamics of HOD in D2O J.J. Loparo, C.J. Fecko, J.D. Eaves, S.T. Roberts, A. Tokmakoff Department of Chemistry and George R. Harrison Spectroscopy Laboratory, Massachusetts Institute of Technology, Cambridge MA 02139, USA E-mail: [email protected] Abstract. Vibrational echo peak shift and polarization-selective pump-probe measurements with broadband 45 femtosecond infrared pulses have been used to study the OH stretching vibration of HOD in D2O. A unified analysis of all experiments allows a characterization of the ensemble-averaged vibrational dephasing, lifetime, and orientational dynamics. The OH stretch of HOD in D2O is often used as a model system to study hydrogen-bond dynamics of water, because the OH frequency is sensitive to hydrogen bonding to the proton. Ultrafast infrared spectroscopy can follow the changes of OH vibrational frequency with time which reflect the underlying dynamics of water's hydrogen bonding network. A number of studies have built on this idea by using transient IR hole burning to characterize the vibrational dynamics of HOD in D2O (1-3). In these experiments a narrow pump pulse in frequency is used to create a spectral hole, and a time-delayed probe pulse follows the subsequent spectral diffusion as this sub-ensemble equilibrates. Though much has been learned with these narrow band experiments, the relatively long pulses required have limited observations to times scales longer than 150 fs, even though other intermolecular motions in water are faster. Analysis of any experiment in water is also complicated by population and orientational relaxation processes which occur on similar time scales. We have performed a set of ultrafast IR experiments that address these difficulties. Femtosecond vibrational echo and polarization-selective pump-probe experiments with broadband excitation of the entire OH absorption line are used to characterize the vibrational and orientational dynamics of the OH stretch vibration averaged over all environments. Assuming Gaussian fluctuations, it is possible to use these experiments to extract correlation functions describing vibrational dephasing, lifetime and reorientation, yielding an equilibrium ensemble averaged description of the vibrational dynamics. Experiments were performed with 45 fs IR pulses at 3300 cm"^ generated with a dual stage BBO/KNbOs OPA. Second harmonic generation FROG showed that the pulses had relatively flat spectral phase over their 400 cm'^ of bandwidth, yielding enough coherent bandwidth to span the entire fundamental OH stretch absorption lineshape of HOD and most of its anharmonically shifted v=l-2 transition. In vibrational echo peak shift (PS) measurements tliree pulses were crossed in the sample to generate a background free signal. The signal intensity as a function of the first two pulses X] was measured for several fixed waiting times between the second and third pulse T2. The X\ delay of maximum signal Xj is the peak shift. Pump-probe transient absorption measurements were made in a two-

410

pulse geometry with a delay T2, with the polarizations controlled by wire-grid polarizers. The sample of HOD in D2O was flowed as a 50 \xm jet, which was necessary to eliminate the strong nonresonant signal from sample cell windows. PS measurements of TI as a function of I2 are primarily sensitive to the timescales of vibrational dephasing, closely tracking the vibrational frequency correlation function. Figure l a shows the PS data for HOD in D2O. It shows a rapid 60 fs decay and an underdamped oscillation at 160 fs, with a long time decay of 1.2 ps. This oscillation largely reflects the underdamped intermolecular displacement of the hydrogen bond (4).

500

1000

X2 (fs)

Fig. 1. (a) The vibrational echo peak shift, (b) the pump-probe in parallel polarization geometry, and (c) the pump-probe anisotropy decay.

In contrast to the PS, pump-probe measurements are most sensitive to vibrational relaxation, but also contain contributions from dephasing and reorientation. Polarized detection methods can be used to separate these relaxation processes. The signal collected in the magic angle geometry, which is free of the effects of reorientation yielded an asymptotic exponential decay due to the OH lifetime with Ti= 700 fs (Fig. lb). The reorientational motion of the OH dipole is isolated from the parallel and perpendicular pump-probe nieasurements through a determination of the anisotropy. We found the anisotropy decay to be bimodal with 50 fs and 3 ps components (Fig. Ic). The fast 50 fs component has been predicted by molecular dynamics simulations and originates in the water librational motion (5). The effects of vibrational dephasing on the pump-probe transient are apparent by subtracting the effects of reorientation and population relaxation, as shown for the parallel pump-probe measurement in the inset to Fig. lb. This residual shows a rapid decay followed by a weak recurrence at -180 fs before decaying. As expected the timescale for the beat is nearly identical to the one observed in the PS measurement. On timescales long compared to T] we observe a 3% percent offset in the PP transient that has previously been ascribed to relaxation induced changes in the population of bath states (6, 7). This offset appears to remain constant for more than 50 ps, which is consistent with equilibration due to thermal diffusion. The PS and PP measurements can be described self-consistently with a formalism that contains only three dynamical quantities: the OH frequency correlation function C(T) = (8CO(T) 5CO(0)), the vibrational lifetime T], and an orientational correlation function for the OH vector p2(T) = P2(U(T) U(0)). The formalism is based on a multilevel nonlinear response function. Vibrational

411

lifetimes are added phenomenologically. To reduce the number of free parameters, we evoke a number of harmonic scahng relationships that relate the dynamics of the v=0-l transition to energy gaps involving the v=2 state. The experimental signals are obtained by convolving the third-order response function with the electric field of the pulses. The solid lines plotted along with the data in Fig. 1 reflect the best simultaneous fits to the PS and the PP data. The resulting underdamped frequency correlation function, vibrational lifetime and orientational correlation function (Fig. 2) provide an equilibrium ensemble-averaged description of the vibrational dynamics of HOD in D2O over all the relevant time scales of water's hydrogen bonding network. An overall picture emerges in which multiple intermolecular dynamics influence the OH infrared spectroscopy in a number of ways. Consistent with molecular dynamics, underdamped intermolecular motions of the hydrogen bond, with translational and hindered rotational character modulate the OH frequency prior to more permanent H-bond breaking on >300 fs time scales. Collective reorientational motions from librations and diffusive reorientation also induce changes in OH orientation. The combination of forces from these intermolecular fluctuations, largely of an electrostatic nature, also act on the OH coordinate to lead to fast vibrational relaxation.

0.75-

0

a

0)

•D 3

0.50-

•Q.

E (0

b

0.25-

c C)

500

1000

time (fs)

Fig. 2. Plots of the normalized (a) OH orientational correlation function, (b) OH vibrational lifetime decay and (c) the OH frequency correlation function.

References R. Laenen, C. Rauscher, A. Laubereau, J. Phys. Chem. B 102, 9304 (1998). S. Woutersen, H. J. Bakker, Phys. Rev. LeU. 83, 2077 (1999). G.M. Gale, G. Gallot, F. Hache, N. Lascoux, S. Bratos, J.-C. Leicknam, Phys. Rev. Lett. 82 (1999) 1068. C. J. Fecko, J. D. Eaves, J. J. Loparo, A. Tokmakoff, P. L. Geissler, Science 301, 1698(2003). C. P. Lawrence, J. L. Skinner, J. Chem. Phys. 118, 264-272 (2003). H. K. Nienhuys, S. Woutersen, R. A. v. Santen, H. J. Bakker, J. Chem. Phys. I l l , 1494-1500(1999). J. Stenger, D. Madsen, P. Hamm, E. T. J. Nibbering, T. Elsaesser, J. Phys. Chem. A 106,2341-2350(2002).

412

Time Resolved Direct Probing of the Change in the Local Solvent Response Following Excitation of a Solute David F. Underwood and David A. Blank Department of Chemistry, University of Minnesota, 207 Pleasant ST SE, Minneapolis, Minnesota 55455, USA Phone: (612) 624-0571, Fax:(612) 626-6570, e-mail: [email protected] Abstract. The change in the low frequency (0.1 - 500 cm~^) non-resonant Raman response following excitation of a dye molecule in solution has been measured as a function of time after resonant excitation of a solute using a two-dimensional mixed resonant, non-resonant time domain spectroscopy. This method provides a direct measurement of the change in the spectrum of the local environment as a dynamic event proceeds in a condensed phase. Recently we reported the first results from a new experimental technique involving a resonant excitation of a chromophore followed by a third-order Raman probe of the solvent (RaPTORS, ResonAnt-Pump Third Order Raman Spectroscopy) [1, 2]. This two-dimensional method allows direct observation of the solvent participation in response to a dynamic event, such as the creation of an excited state in a chromophore or a charge transfer reaction. Here, our efforts are focused on directly measuring the change in the complete spectrum of the low-frequency ( 0 . 1 - 500 cm~^) molecular solvent response as a function of time from photo-excitation of a solute. The experiment involves an initial pulse that is resonant with an electronic transition in the solute, E^. A time t later, a set of three electronically non-resonant pulses, Enri-Enr3, follow to probc the third order Raman response of the solvent as a function of the intrinsic time delay r between time coincident pair £^nri/^nr2 and ^nr3- The two dimensional signal is detected along the same direction, ks' = /Cf —/Cr+ZCnrl —fcnr2+ ^nr3 ^S t h e nOU-rCSOnant t h i r d o r d e r s i g n a l ks = ^nrl — ^nr2 + ^nr3-

The result is heterodyne detection of the two dimensional response with the nonresonant bulk solvent signal providing the local oscillator. The two dimensional response can be interpreted as a small change in the TOR probe measured as the cross term [2], /RaProRs(t,r) ^ ( ^ ) Re

[E^2Ur)AE('Ht^r)

(1)

We have employed a forward convolution approach to extract the change in the bare third-order non-resonant response function, which is proportional to the change in the third-order signal field, AE^^\t,r)cxAR^^\t,T).

(2)

We first measure the heterodyne detected TOR spectrum of the sample [3, 4, 5] to 413

obtain the bare third-order response, which is dominated by the bulk solvent [6], i?f^) solvent (r)

h

{[a(r),a(0)]).

(3)

Slices from the two-dimensional data set at fixed values of t are then simulated with iterative optimization of the change in the third order response, Ai?(^)(r; t). /RaProRs(r;t)

oc

H I,,,it') Jo

( T \Jo

R^,,^,{t")IUt'-t"+

r)dt"^ J

^ r AR^^\t''; t)hS' - ^" + ^W') dt' (4) The third order response is modelled as a collection of Brownian oscillators [6] and the change in the third order response is obtained by making small alterations in ^solvent ('^) ^^^ subtracting the result from the originally determined insolvent ('^)The left hand side of Fig. 1 shows the raw data and fit for a slice along r for t = 1 ps after resonant excitation of Coumarin 102, C102, in acetonitrile. The right hand side of Fig. 1 presents the change in the low frequency Raman response of the local solvent at 1 ps and 300 ps after excitation of CI02. The increase in the intermolecular interaction upon excitation of CI02, see calculations below, results in a tightening of local intermolecular potential. Analogous to an increase in density, the result is that more localized motions, for example librations, shift to higher frequencies and the collective motions shift to lower frequencies. After C102 excitation there is an increase at the low and high ends of the intermolecular response and a concomitant loss in the middle frequencies (10-100 cm~^), as shown in Fig. 1. In addition to the intermolecular motions, we also see the change in the 388 cm~^ intramolecular bending vibration in the local acetonitrile solvent molecules. The frequency appears to increase slightly over time in response to the increased local dipolar interaction with the CI02 solute. xlO

100

200

300

400

CO / cm"^

Fig. 1. The left plot is the raw RaPTORS response for a delay /; = 1 ps after resonant excitation of CI02 in acetonitrile. This is a direct probe of the change in the solvent response. The right plot shows the fit to the measured response for delays of t = 1 ps and t = 300 ps demonstrating the change in the solvent response and its evolution over time.

414

The measured change in the intermolecular solvent response is considered in the context of their intermolecular interaction energies defined by

r6 { 3kT(47reo)^ ^

(47reo)2

^ 2(47reo)2 (£;f + £;|)

where the first, second, and third terms represent the electrostatic, induction, and dispersion energies respectively [7, 8, 9]. Optimized gas-phase structures were calculated in Gaussian 98 at the B3LYP/Midi! level, and these were used to determine CM2 dipole moments, //, in acetonitrile in both the ground and excited states (SM5.42R/INDO/S2, VEM42/INDO/S2 calculations respectively) using the ZINDOMN [10] program. Polarizabilities, a, and ionizations potentials, E^ were extracted from these calculations. Solvent parameters for the solvation calculations were taken from the Minnesota Solvent Descriptor Database [11]. We have demonstrated the ability to recover the change in the complete low frequency non-resonant Raman response as a function of time after photo-initiation of a dynamic event in solution. Direct probing of the local environment is a powerful new tool for the investigation of the dynamic participation of the local environment when ultrafast events take place in condensed media. This work was carried out in part using computing resources at the University of Minnesota Supercomputing Institute. References 1. D. R Underwood and D. A. Blank, Investigation of low-frequency, condensed phase solvation dynamics using a novel two-dimensional, resonant-pump, third-order Raman-probe technique. Vol. 72 of Trends in Optics and Photonics (Thirteenth International Conference on Ultrafast Phenomena) (2002). 2. D. F. Underwood and D. A. Blank, J. Phys. Chem. A 107, 956-961 (2003). 3. A. A. Maznev, K. A. Nelson, and A. Rogers, Opt. Lett. 23, 1319-1321 (1998). 4. G. D. Goodno, G. Dadusc, and R. J. D. Miller, J. Opt. Soc. Am. B 15,1791-1794 (1998). 5. M. Khalil, N. Demirdoven, O. Golonzka, C. J. Fecko, and A. Tokmakoff, J. Phys. Chem. A 104, 5711-5715(2000). 6. S. Mukamel, Principles of nonlinear optical spectroscopy (Oxford University Press, Inc., New York, 1995). 7. S. Park, B. N. Flanders, X. Shang, R. A. Westervelt, J. Kim, and N. F Scherer, J. Chem. Phys. 118, 3917-3920 (2003). 8. P. L. Huyskens, W. A. P. Luck, and T. Zeegers-Huyskens, Intermolecular Forces: An Introduction to Modern Methods and Results (1991). 9. G. C. Maitland, M. Rigby, E. B. Smith, and W. A. Wakeham, Intermolecular Forces. Their Origin and Determination (1987). 10. M. Zerner, J. Ridley, A. Bacon, J. Head, W. Edwards, J. Head, J. McKelvey, J. Cuberson, P. Knappe, M. Cory, B. Weiner, J. Baker, W. Parkinson, D. Kannis, J. Yu, N. Roesch, M. Kotzian, T. Tamm, M. Karelson, X. Zheng, G. Pearl, A. Broo, K. Albert, J. Cullen, J. Li, G. Hawkins, J. Thompson, D. Liotard, C. Cramer, and Truhlar, D. G., Quantum Theory and Project, University of Florida, Gainesville, and Department of Chemistry, University of and Minnesota, Minneapolis, "ZINDOMN, version 1.1," (2002). 11. P. Winget, D. M. Dolney, D. J. Giesen, C. J. Cramer, and D. G. Truhlar, "Minnesota Solvent Descriptor Database," (1999).

415

Surface Femtochemistry: Photocatalytic Reaction Dynamics of Methanol / TiOzCllO) Ken Onda, Bin Li, and Hrvoje Petek Department of Physics and Astronomy, University of Pittsburgh, Pittsburgh, PA15260, USA E-mail: [email protected] Abstract. We observe a strong charge-transfer resonance for the methanol/TiO2(110) surface by time-resolved two-photon photoemission spectroscopy. The resonance decays in a two-component process corresponding to the charge transfer induced wave packet motion followed by solvation by the surrounding molecules.

1.

Introduction

Photocatalytic reactions on titanium dioxide (Ti02) have a broad range of application in decontamination, as self-cleaning super-hydrophilic surfaces, and for solar cells [1, 2]. Because photocatalysis is of fundamental interest and potentially of vast, practical importance, the mechanism of photocatalytic reactions has been studied by ultrafast techniques [3]. Hov^ever, such studies have been performed on nanocrystalline samples, which provide high surface area, but are otherwise poorly characterized [3]. An alternative way to study surface electronic structure and ultrafast dynamics of interest to photocatalysis is time-resolved twophoton photoemission (TR-2PP) [4]. We have used TR-2PP to study the electronic structure and dynamics of Ti02(l 10) surfaces modified by defects and atmospheric molecules [5, 6]. For H20/Ti02 we discovered a charge transfer resonance 2.45 eV above the Fermi level with a lifetime of -10 fs, which is observed when both surface hydroxyls and molecularly adsorbed water exist on Ti02. Here we investigate the excited state electronic structure and charge-transfer dynamics of CH30H/Ti02 surface at sub-monolayer to multilayer coverages by TR-2PP.

2.

Experimental

All measurements are carried out in a UHV chamber. After preparation of nearly perfect rutile TiO2(110)-(l x 1) surface, oxygen vacancy defects are created by annealing for 30 min at 900 K in vacuum. Sample temperature is kept at 100 K during measurement. Ultrashort pulses (3.05 eV energy, -10 fs duration) are obtained by frequency doubling a self-made Ti:Sapphire oscillator. The pulse is split equally into pulse pairs in a Mach-Zehnder interferometer. The pulse delay is generated with a piezo-driven stage. 2PP spectra are measured with single pulses, while interferometric two-pulse correlations (I2PC) for specific photoelectron energies are measured by scanning the delay [4]. Electron energy analyzer records two-photon photoemission current with energy referenced to the Fermi level.

416

1.0

Intermediate state energy (eV) 1.4 1.8 2.2 2.6 methoxy bridging oxygen

five coordinated Ti"** 4.4 4.8 5.2 Final state energy (eV)

5.6

Fig. 1. (a) 2PP spectra of bare TiOa surface and surfaces after 1 ML methanol adsorption when measured with p- and s- polarized light, (b) Ball and stick model of dissociatively adsorbed methanol on the TiOzCl lO)-(lxl) surface.

3. Results and Discussion Figure la shows typical 2PP spectra of the bare TiO2(110) surface with ppolarization and p- and s-polarized spectra after adsorption of 1 monolayer (ML) of methanol. The strong resonance peak at 2.3 eV is observed only with ppolarization, indicating that the transition moment is perpendicular to the surface. Since the energy and coverage dependence of the methanol resonance is very similar to that observed for H2O/TiO2(110) [6], we attribute it to the charge transfer involving the structure in Fig. lb. Upon formation of the bridging hydroxyl group, an electron is transferred to the most acidic site, i.e. fivecoordinate Ti"*^. The presence of electron donating species such as methanol or methoxy at thus reduced five-coordinate Ti^^ destabilizes the ground state, thereby making the charge transfer excitation from Ti^^ back to the bridged hydroxyl species accessible with 3.05 eV photons. In Fig. 2a we present the phase-averaged envelopes of the I2PC measurements. The I2PCs for the bare surface, and for surfaces with 1, 2, and 3 ML coverage show dramatic change in the dynamics with methanol coverage. On the bare surface, the phase averaged I2PC is essentially identical to the pulse autocorrelation indicating that the lifetime of the intermediate state with 2.3 eV energy is extremely short (< 5 fs). At 1 ML, the lifetime becomes significantly longer than for the bare surface and can be modeled with a single exponential with a time constant of 30 fs. The lower resonance energy and longer lifetime indicate that the methanol adsorption at the five-coordinated Ti"^^ site stabilizes the excited state with respect to water [6]. This we attribute to methanol and methoxy being both more polarizable and basic than water. At 2 and 3 ML coverage, the decay profiles clearly exhibit nonexponential behavior with the excited state lifetimes extending to picosecond time scale. The I2PC measurements for >1 ML coverage exhibit highly complex dynamics that strongly depend on the observation energy. To elucidate the excited state dynamics, we measure I2PC scans in 0.1 eV intervals and plot the intensity as a function of energy and time to obtain 3D contour plots in Fig. 2b. The resonance peak position at different delay times is represented by the dashed line.

417

50

100 Delay (fs)

150

Fig. 2. (a) Coverage dependent phase-averaged I2PC traces at 2.0 eV on the TiO2(110) surface, (b) 3D contour plot of phase-averaged I2PC for the 2 ML CH3OH adsorbed on Ti02(l 10) surface. Dashed line represents peak position at each energy. As this plot clearly shows, the resonance peak located at 2.3 eV at 0 fs shifts to 2.1 eV within 30 fs followed by a much slower shift to 2.0 eV over the next 250 fs. This result indicates that at least two processes are involved in stabilization of the excited state. The complex relaxation dynamics of methanol-induced resonance are characteristic of a charge transfer process [7]. The excited state is prepared far from the equilibrium geometry leading to the initial nuclear wave packet motion. The initial relaxation occurs on the characteristic time scale of Ti-0 stretching vibration (-800 cm'^). After the ballistic motion along the relaxation coordinate brings the resonance near the excited state equilibrium, the dynamics proceeds on a longer time scale. We expect that the subsequent relaxation involves the reorganization of the surrounding methanol molecules, which solvate the redistributed charge density in the excited state. The I2PC trace in Fig. 2a and the contour diagram in Fig. 2b for 2 ML coverage suggest that the dynamics may involve more than one excited state. This can be seen by the delayed secondary maxima at -35 fs delay in Figs. 2a and Fig. 2b, which suggest that the initial wave packet motion leads to a non-adiabatic surface crossing to another state. A candidate is suggested in Fig. lb, by the implied ground state hydrogen bonding between the hydroxy and methoxy species, as predicted by theory [8]. It is possible that the charge-transfer excitation sufficiently weakens the 0-H bond to transfer the proton to reform methanol. Experiments and theory to test this excited-state tautomerization hypothesis are in progress.

References 1 A. L. Linsebiglar, G. Lu, and J. T. Yates, Jr., in Chem. Rev. Vol.95, 735, 1995. 2 U. Diebold, in Surf. Sci. Rep. Vol. 48, 53, 2003. 3 for example, R. Huber, J.-E. Moser, M. Gratzel, and J. Wachtveitl, in J. Phys. Chem. B, Vol. 106, 6494, 2002. 4 H. Petek and S. Ogawa, in Prog. Surf. Sci. Vol. 56, 239, 1997. 5 K. Onda, B. Li, and H. Petek, in Phys. Rev. B (in press). 6 K. Onda, B. Li, and H. Petek, (submitted). 7 P. J. Rossky and J. D. Simon, in Nature, Vol. 370, 263, 1994. 8 S. P. Bates, M. J. Gillan, G. Kresse, in J. Phys. Chem. B, Vol. 102, 2017, 1998.

418

Three pulse four wave mixing for the study of coherent interactions, nuclear dynamics and solvation dynamics in liquids June-sik Park and Taiha Joo Division of Molecular and Life Sciences, Department of Chemistry, Pohang University of Science and Technology, Pohang, 790-784, Korea Abstract. By use of precise controlled time delay between two pump pulses, four wave mixing experiments reveal some unknown features of ultrafast dynamics in solution. We investigate origin of coherent spike and nuclear wave packet dynamics and solvation dynamics which are very different between similar dye molecule used in this experiment.

1. Introduction Femtosecond four wave mixing (FWM) has been used extensively for the study of dynamics in condensed phases. There are many variants of the FWM techniques in time domain that may probe electronic coherence, vibrational and rotational coherences, and population dynamics. In its most general form, a FWM technique uses three pulses with two independent time delay controls to probe various dynamics in condensed phases depending on the resonances between the light pulses and molecular system.[l] We report a electronically resonant FWM experiment with accurate electronic coherence time delay control with attosecond precision to reveal new features of the light-matter interaction in third order nonlinear spectroscopies. For molecular systems in liquids, certain dynamics in a FWM signal can be either enhanced or attenuated rather sensitively by the control of the electronic coherence time of a few femtoseconds. This feature facilitates the measurement of a dynamics of interest. In addition, when the electronic coherence time is varied, different probe molecules show quite different behaviors in vibrational wave packet dynamics.

2. Experimental Methods Femtosecond pulses generated from a home built Ti:sapphire laser were split into two by a cube beam splitter. Both beams were focused on a fused silica diffractive beam splitter by a 35 cm focal length lens. Two first order (±1) diffractions of one beam served as pump pulses, ki and k2, while one of the first order diffractions of the other beam was used as a probe pulse, ks. The k2 beam passes through a pair of BK7 wedge prisms with wedge angle 1.93°. One of the wedge prisms was mounted on a motorized translational stage to control the amount of glass in the beam path, which controls the optical delay between ki and k2. This allows time resolution of - 3 attoseconds with stability better than 0.5 fs for several hours. A

419

FWM signal into the -ki+k2+k3 phase matching direction was detected spectrally integrated by using a photodiode and a lock-in amplifier.

3. Results and Discussion The two experimentally controllable time delays in pulse pairs, (ki k2) and (ki ks), are labeled as x and T. By this definition, x and T- x represent electronic coherence and population periods, respectively. The three pulse FWM signals into -ki+k2+k3 phase matching direction for IR125 and IR144 samples at several different T values are shown in Fig. 1. The x delay is defined to be positive when ki reaches the sample prior to k2.

'>f- U)\i I'ViO

Fig. 1. Three pulse four wave mixing signal into -ki+k2+k3 phase matching direction for IR125 (a) and IR144 (b) methanol solution for several different x values. From bottom to top, X is set to -12, -8, -4, 0, 4, 8,12,16, 20, 24 fs. Immediately apparent in the figure is that the signal changes sensitively even with a few femtosecond change in x. The FWM signals consist of four components; a coherent spike (CS) near time zero, oscillations due to the vibrational wave packet motion, exponential decay components due to solvation processes, and a large offset, which eventually decays by the population relaxation. The relative amplitude of the CS increases as x increases, whereas the CS disappears almost completely for negative x. The position of CS also changes rather sensitively on x. These features can be accounted for by realizing that for x > 0, the signal decays by the electronic dephasing time as T is scanned from 0 to x. Both samples show the same behavior in this regard. Close inspection reveals that the relative contribution of the solvation component in the FWM signals increases as x increases in both samples. This can be understood by the consideration of the third order response functions. The response function with rephasing gives sensitivity on solvation dynamics. Because the rephasing occurs between the two electronic coherence periods, it comes about only when x 7^ 0 fs. Thus, as x increases, the response function becomes more sensitive to the solvation process. In Fig. 2, differences between the signals with the same coherence time but the pump pulse sequence is opposite i. e. for example,

420

the signal when x = 24 fs minus x= -24 fs. Then, we can remove the effect of nonrephasing signal and the remaining signal is very sensitive to solvation dynamics.

I 1

500 1000 1500 Time Delay (fs)

2000

0

500 1000 1500 Time Delay (fs)

2000

Fig. 2. Differences between the four wave mixing signals having the same coherence periods, (a) IR125 (b) IR144 in methanol. From bottom to top |x| is set to 4, 8, 12, 16, 20, 24.

The FWM signals also show oscillations due to intramolecular vibrational wave packet dynamics. In IR125, the amplitude of the oscillation increases as x increases. This is analogous to the wave packet control experiment by controlling the chirp of the pump pulses [2]. In IR144, however, it behaves oppositely; the contribution increases as x decreases. This can be accounted by assuming that the potential surfaces of the states involved in the probe step are displaced oppositely in the two molecules. That is, the excited state potential surface of IR144 has larger equilibrium bond length or the probe step in IR144 is realized by the resonance between the Si and a higher excited state.

4.

Conclusions

By the control of the electronic coherence time, the contribution of the various components in a FWM signal can be controlled. This allows the enhancement or attenuation of dynamical components to improve the sensitivity of the FWM signal to the dynamics of interest. The time delay, however, must be controlled precisely within 1 fs to fully realize the potential.

References 1 J, S. Park and T. Joo, J. Chem. Phys. 120, 5269 (2004). 2 M.-C. Yoon, D. H. Jeong, S. Cho, D. Kim, H. Rhee, and T. Joo, J. Chem. Phys. 116, 10801 (2002).

421

Solvation dynamics of N-methylacetamide in D2O, CDCI3 and DMSO-d6 M. F. DeCamp, L. P. DeFlores, J. M. McCracken, and A. Tokmakoff Chemistry Department, Massachusetts Institute of Technology, USA E-mail: [email protected] Abstract. We study amide I vibrational solvation dynamics of N-methylacetamide in D2O, CDCI3, and DMSO-dfi. Three pulse photon echo and 2D-IR measurements characterize the frequency correlation function, which reflects fluctuating electric fields acting on the vibration.

1.

Introduction

The spectroscopy of the amide I vibration band forms the basis for understanding much of the infrared spectroscopy of proteins and peptides. In particular, the amide I band is sensitive to vibrational couplings and hydrogen bonding to the peptide group [1-3], Since the influence of hydrogen bonding and other solvent dynamics plays a fundamental role in describing protein and peptide vibrational spectroscopy in solution, we have studied the role of solvation on the frequency fluctuations of the amide I coordinate. N-methylacetamide (NMA) has a single peptide unit making it an ideal system for the study the solvent dependent dynamics. Recent experimental investigations have made significant progress towards extracting a frequency-frequency correlation function of solvated NMA in D2O [4], Molecular dynamics simulations have indicated that translational and librational modes of the solvent play important roles in the solvation dynamics [5]. Here we expand on this work by performing stimulated integrated three-pulse photon echo (3PE) and time-resolved two-dimensional Fourier transform infrared spectroscopy (2D-IR) measurements to extract the frequency-frequency correlation function of NMA in D2O, CDCI3, and DMSO-dfi. We beheve that these three solvents represent model interactions with the amide I vibration. D2O is both a strong hydrogen bond donor and acceptor for the peptide group, whereas CDCI3 and DMSO-dfi interact very weakly with the peptide unit [5].

2.

Experimental Methods

An ultrafast 800 nm, 1 kHz laser system pumps a commercial OPA which produced ~1 |LiJ, 90 fs, 6 |Lim pulses. Three identical IR pulses are focused in the box-car geometry resulting in an echo signal emitted in the phase-matched direction. The delays Ti, T2, and T3 represent the time separations between the excitation pulses as well as the emitted echo signal. For the 3PE experiments the

422

Xi Dday [fs]

x, Delay [fs]

x^ Delay [fs]

Figure 1. Time-resolved integrated plioton eciio of NMA in (a) D2O, (b) DMSO-d^, and (c) CDCI3. The solid (dashed) lines are the data (fit). Waiting times of 150, 300, 700, 1000, 1500, 2000 fs are shown on each plot from top to bottom. A constant offset is added to each curve for clarity. integrated echo was measured as a function of Xi and X2. For the 2D-IR experiments, the echo was heterodyned detected by overlapping with a local oscillator pulse and spectrally dispersed in a monochromator. The time dependent signal was Fourier transformed along Ti giving a 2D frequency-frequency surface. The shape, position, and time dependence of the 3PE and 2D-IR have distinctive signs of the solvation dynamics. In particular, the temporal position and asymmetry of the 3PE signal reflects the effect of the inhomogeneous broadening. This same phenomena is reflected in the 2D-IR experiment by the shape and relative slope of the 2D line shapes [7]. To retrieve the frequency-frequency correlation function, the 3PE data and the IR absorption line shape were simultaneously fit to a theoretical nonlinear response [6]. Qualitative comparisons were made between the measured and calculated 2D-IR spectra. The vibrational lifetime was incorporated such that multiple echoes could fit simultaneously.

3. Results Figure 1 shows the 3PE for NMA in D2O, CDCI3, and DMSO-dr, for a series of waiting times T2. For all tliree solvents, the 3PE shows clear signs of inhomogeneous broadening (e.g. the temporal asymmetry and peak shift). Fitting the 3PE data D2O solvation with a bi-exponential correlation function, decay constants of ~20fs and -llOOfs were retrieved. This is consistent with the 2D lineshapes where the transition remains inhomogenously broadened until a waiting time of ~1.5ps (see Figure 2a). In contrast, the 3PE and 2D-IR data of NMA in CDCI3 and DMSO-dr, clearly shows an inhomogenously broadened transition even at long waiting times. Fits to the 3PE and 2D line shape reveal that the correlation function has a primary decay at least 2 times slower than that of the D2O solution while also having a quasistatic decay of slower than 5ps. The retrieved timescales follow the expected strength of hydrogen bonding or electrostatic interactions of the solvent. Drawing on recent models of vibrational frequency shifts in hydrogen bonding liquids, we assign the solvation dynamics to

423

Figure 2. 2D-IR lineshapes of NMA in (a) D2O, (b) DMSO-dfi, and (c) CDCI3 all at a waiting time of 2ps. fluctuating electric fields from the solvent acting on the amide I coordinate. These originate largely from librational, translational, and orientational dynamics of the solvent. The dominant source of these frequency fluctuations will be the closest charges, presumably the hydrogen bound protons (deuterons) of the solvent, but ultimately, collective dielectric fluctuations will also contribute both to fast and slow time scales. Comparisons to results provided by molecular dynamics simulations indicate that electric field fluctuations accurately reproduce the observed frequency fluctuations [5,9]. In summary, using several spectroscopic techniques we have extracted the frequency-frequency correlation function of NMA in several solvents. We observe solvent dependent differences, which are attributed to solvent induced dynamics. Acknowledgements. We would like to thank Min Cho for giving us access to the results of his molecular dynamics simulations. Funding provided by the NSF, DOE, and Packard Foundation.

References p. Hamm, M. Lim, W. F. DeGrado, and R. Hochstrasser, Proc. Nat. Acad. Sci. 96, 2036,1999. S. Woutersen and P. Hamm, /. Chem. Phys., 114, 2727, 2001. J. Bredenbeck and P. Hamm, / Chem. Phys., 119, 1569 2003. M. T. Zanni, M. C. Asplund, and R. M. Hochstrasser, /. Chem. Phys., 114, 4579 2001. K. Kwac and M. Cho, J. Chem. Phys., 119, 2247, 2003. K. Kwac and M. Cho, J. Chem. Phys., 119, 2256, 2003. ]. Sung and R. J. Silbey, J. Chem. Phys., 115, 9266, 2001. A. Tokmakoff, J. Phys. Chem. A, 104, 4247, 2000. S. Woutersen et al., J. Chem. Phys., Ill, 6833, 2002. M. Cho, personal communication, 2004.

424

Femtosecond Pump-probe Measurements of Solvation Dynamics of Hydrogen-bonding Complexes in Non-associating Solvents D. Finest E. Pines', Y-Z Ma' and G. R. Fleming' 'chemistry Department, Ben-Gurion University of the Negev, P.O.B 84105, Beer-Sheva, Israel E-mail: [email protected] ^ Department of Chemisry, University of California Berkeley, Berkeley, and Physical Biosciences Division, Lawrence Berkeley National Laboratory, Berkeley, California 94720-1460, USA E-mail: [email protected] Abstract. We present an ultrafast pump-probe study of solvation dynamics of hydrogenbonding complexes in non-associating solvents. Additional blue-shift of sub-100 fs duration was observed in the transient absorption of HPTA-DMSO complexes over the corresponding transient spectra of uncomplexed HPTA and that of its methoxy-derivative.

1.

Introduction

The overall objective of this study is to show that it is possible to follow transient changes in the hydrogen-bonding interaction of a dye molecule by monitoring its transient optical Stokes Shift. The acidity of a strong photoacid forming well-defined hydrogen-bonding complexes in solution was optically switched by fs-pulse excitation. We carried out a double-control (optical) pump-probe experiment in order to verify the effect of the hydrogen-bonding interaction on the transient stokes shift of the complexed photoacid. The solvation response of the hydrogen-bonded complex and that of uncomplexed photoacid and its methoxy-derivative were compared to reveal the transient changes in the hydrogen-bonding interaction upon the formation of the excited state of the photoacid.

2. A sub-100 fs solvation-like response in the hydrogen-bonding complex is initiated by the induced photoactivity We have used a novel strong photoacid, 8-hydroxypyrene 1,3,6 trisdimethylsulfonamide (HPTA) complexed by hydrogen bonding to mild oxygen bases in polar non-associating solvents [1,2]. By optically exciting the photoacid, we have switched the acidity of the OH group by about 7 pKa units, the photoacid being much stronger acid in the excited-state. The enhanced acidity has caused a sudden increase in the hydrogen-bonding interactions of the photoacid of the type

425

O-H-'-O. Hydrogen-bonding complexes of HPTA were prepared in the groundstate of the photoacid by adding small amounts of dimethylsulfoxide (DMSO) to solutions of HPTA in dichloroethane (DCE) and dichloromethane (DCM). Figure 1 (a) shows the large effect in the absorption spectra of HPTA when adding small amount (1:1000) of DMSO to solutions of HPTA in DCM. As a control, similar concentrations of DMSO were added to solutions of the methoxyderivative of HPTA, the MPTA molecule. No change in the absorption spectra of MPTA was observed. The observed absorption spectrum of HPTA in presence of DMSO was fully reconstructed assuming it was made of two populations: the uncomplexed and complexed HPTA. The relative amplitudes of the two populations used to reconstruct the experimental data conforms to 1:1 complex between HPTA and DMSO with a complexation constant of 360 M'' [2]. Utilizing a pump-probe set-up described elsewhere [3], with 400 nm excitation, the dynamic Stokes shift of HPTA was analyzed with about 50 fs time-resolution. Figure 1 (b) compares the transient absorption kinetics of HPTA in pure DCM at 590 nm and in a 10"' M solution of DMSO in DCM at 570 nm after photoexcitation at 400nm. The two wavelengths have been chosen so as to compensate for the spectral shift in the ground state absorption spectra of the hydrogen-bonded complexes. Analyzing the transient absorption spectrum of uncomplexed and complexed HPTA revealed faster decay components in the complexed HPTA spectrum. Comparison at identical wavelengths resulted in even larger change between the kinetics of the uncomplexed and complexed HPTA. In the control experiment similar time resolved measurements were made in solutions of MPTA and revealed identical behavior of MPTA in presence of and without DMSO in DCM solutions. -

- -HPTA in DCM MPTA in DCM

fft \

1.5

W
0

MPTAinDCM+DMSO HPTA-DMSO in DCM

1.0^ \V/\N 0.5-^

0.0

ill

HPTA-DMSO complex in DCM X , = 570 nm

(a)^~

\B^^M^

22

24

26

28

V (10^ cm^)

30

0

500

1000

1500

2000

Delay time (fs)

Fig.l. Steady state (a) and transient absorption (b) spectra of HPTA and MPTA Single wavelength measurements were used to find the solvation correlation response of the absorption spectra. Figure 2 shows the effect of creating hydrogenbonding complexes between HPTA and DMSO on the solvation correlation

426

function of HPTA, C(t) in DCM and DCE. The C(t) shows the dynamics of hydrogen-bonded HPTA having an additional sub-100 fs component over the C(t) of uncomplexed HPTA and the C(t) of its methoxy-derivative. We assign this transient solvation-like effect to the transient strengthening of the hydrogenbonding interaction following the induced photoacidity [4].

200

3000

100

200

300

400

500

Delay time (fs)

Fig.2. Solvation correlation function of the excited state absorption of HPTA, solid circles - uncomlexed HPTA, open circles - HPTA-DMSO hydrogen bonded complex. The solid lines represent the fits of C(t) (a) in DCM: the C(t) of uncomplexed HPTA is fitted with T ,= 0.26 ps (40%) and T2 = 2.3 ps (60%), the C(t) for 80% complexation is fitted with Ti - 0.26 ps (29%), 12 - 2.3 ps (43%) and T3 = 56 fs (28%); (b) in DCE: the C(t) of uncomplexed HPTA is fitted with T, = 0.36 ps (34%) and T2 = 3.5 ps (66%), the C(t) for 80%) complexation is fitted with i, = 0.36 ps (26%), 12 = 3.5 ps (51%) and T3 = 42 fs (23%)

Acknowledgements. GRF acknowledges financial support of the National Science Foundation.

References 1 E. Pines, D. Pines, Y-Z Ma and G. R. Fleming, in Femtochemistry and Femtobiology: Ultrafast Molecular Events in Molecular science. Edited by J. T. Hynes, M. Martin, Elsevier, Amsterdam, 2004. 2 E. Pines, D. Pines, Y-Z Ma and G. R. Fleming, ChemPhysChem, 2004 (in press). 3 Q.-H. Xu, Y.-Z. Ma, G. R. Fleming, Chem. Phys. Lett. 338, 254, 2001. 4 B.-Z. Magnes, D. Pines, N. Strashnikova and E. Pines, Solid Statelonics, 168, 225, 2004.

427

Novel Time- and Frequency-resolved Double Pump Spectroscopy of Short-lived Precursors: The Solvated Electron in Methanol. A. Thaller, R. Laenen and A. Laubereau TechnischeUniversitatMunchen, Physik-DepartmentEll, James-Franck-Strasse, 85748 Garching, Germany; E-mail: [email protected] Abstract: A combined investigation of the generation process and relaxation dynamics after reexcitation of intermediates of solvated electrons is presented. The experimental technique provides a much more detailed picture of charge separation in methanol.

More than a century after the first report on electron detachment in liquids [1], there is still a keen interest in the solvated electron. From a spectroscopic point of view, the solvated electron can be utilized as a probe for the structural dynamics of the solvent. The properties of the electron are strongly influenced by the solvent molecules [2] manifesting in a change of the optical density of the sample [3]. The generation process of the solvated electrons occurs on a fs to ps timescale in various Hquids, e.g. water and alcohols [4,5], and was studied with ultrafast pump-probe spectroscopy techniques. In previous experiments on the solvated electron, various important questions have been left open, especially regarding the nature of the intermediate states following the detachment process. In this paper we demonstrate a combined investigation of both the generation process of solvated electrons and of pump-probe-spectroscopy of short-lived intermediate states. a)

a o

OM 0>02

om b)

- - - p"

}^,si 790 nm L

30

10

4D

50

0.003 H 0.002 h 0<001 0.000 -0.001 h 0

5

10

AU (ps) Fig. la: Change in optical density at 790 nm due to the generation of electrons at t=0 (filled dots). The gray line was calculated using a rate equation system; the stippled and dashed lines represent contributions to the signal of the individual species p', p", s and s involved in the relaxation process, b: transient change in optical density at 1283 nm due to reexcitation of the detached electrons at different delay times after generation (0.5ps, 5ps and 60ps; denoted by the vertical dashed lines in la)

428

For our experiments, we use a pump - delayed repump - probe setup with a state-of-the-art Ti:Sapphire laser system [6]. The third harmonic at 273 nm (4.55 eV) is used for ionization via two-photon absorption in a sample jet of neat methanol at room temperature. To reexcite the photo-detached electrons, a part of the original laser pulse at 820 nm is apphed with variable delay time (termed preparation time) after the UV pump pulse. The detection of the probe transmission in the range of 450 nm to 5.5 \xm is polarisation-resolved. The transient absorption data are treated in a self-consistent, global analysis: Signal transients computed from one single rate equation system with discrete set of initial, intermediate and final states are fitted to the measured data. The individual states contribute to the overall probe signal via wavelength-dependent absorption coefficients that provide the spectral signature of the level and allow to identify its nature. Fig.2a shows a sketch of the generation process (horizontal) with subsequent reexcitation of some of the detached electrons 500 fs after the preparation pulse (vertical). Exemplarily, Fig.2c shows an energy level scheme illustrating that part of the rate equation system that is used to describe the dynamics following the reexcitation. Corresponding schemes are used for relaxation channels following reexcitation of other species. S20 tm ^\ 272 nm

V^""*"^^*::*^ Cp 0^ U^ /c$) C4> f ^) CnE> {^j ¥k

Pmimp

'P^hi

^f-oth>

%<&}<><

CKOH

b)

\i W
^mi^'i \0\ ^r> tt) T ^^'T'

1 %

x>^>

1<0

^'"^"""•|""""""'

^ ' J ^:^ ^

•

Cv*
^

2
* '^<

^.

\ 1

rt

i

V,

¥ 1

*"*1

e

><*'

* ^

-:i!-J *.

/^ ///

b

•<..• p . ^ «• j * * «

"" il " J

«i

*j^ <.*,4^

§ m It

i

Fig. 2. a: Sketch of the generation process and the relaxation following reexcitation of the pVp" states of the electron as obtained from the global analysis of all data (see also Figure la), b: transient spectra of the individual species taking part in the generation process, c: energy level scheme that underlies the rate equation model describing the relaxation dynamics after reexcitation of p' and p".

After ionizing a methonol molecule with two-photon absorption of the UV preparation pulse, the emerging electron e'free shows a broad IR absorption centered around 2 \xm (gray dotted line in Fig.2b). Within 125 fs after excitation, the quasifree electron appears to be trapped in a suitable configuration of methanol molecules [5]. The accompanying spectral signature is composed of a broad absorption band covering the whole probing range with a clear dip at around 1.2 jam (line denoted as p' in Fig.2b), proposed to originate from stimulated emission out of

429

the intermediate state p' to a corresponding, unoccupied lower state. Therefore, the electron appears to be initially trapped in an excited, p-like state. Within 0.6 ps, the spectrum shifts towards shorter wavelengths ascribed to a cooling of the local environment of the electron that is still in an excited state with measurable anisotropy (p" in Fig. 2b). The absorption of the subsequent species evolving with a time constant of 3.5 ps shows a completely different signature (Shot in Fig. 2b). Its position and shape are consistent with a ground state absorption of the solvated electron in a hot local environment [7]. The following relaxation on a time scale of 7.9 ps leading to Seq is correspondingly interpreted as ground state cooling. To further investigate the intermediate states, we apply an 820 nm pump pulse to the sample with preparation times of 0.5 ps, 5 ps and 60 ps, in order to selectively excite the precursors p7p". Shot and Seq. Fig.2c shows the considered relaxation channel after reexcitation of p7p" with 0.5 ps delay. It is important to note that the rate equation system models all four experiments, with and without reexcitation pulse, self consistently. The relaxation rates after reexcitation, however, are modified. The results suggest that shortly after the reexcitation of Shot and Seq the electron is in an excited state p'* with a lifetime of 1.14 ps and 0.6 ps, respectively. After relaxation to a hot ground state Shot*, ground state cooling seems to occur with a time constant of 6.2 ps and 5.3 ps, respectively. The reexcitation of the early pVp" states 500 fs after the generation pulse, however, represents a different case. Here, the data support the picture that the electron energetically but not necessarily locally escapes its potential well. Within 140 fs it is retrapped in a p-like state. Changing the polarisation of the reexcitation pulse with respect to the generation pulse yields data that strongly point towards retrapping of the electron in the original solvation shell (data not shown). The subsequent dynamics again shows conversion to a hot ground state followed by ground state cooling. Comparing the time constants for the three relaxation situations following reexcitation of the respective intermediate states, a significant slowing-down of the relaxation steps for shorter preparation time and thus higher temperature of the local environment of the electron is found. To understand the observed correlation it is interesting to recall that with increasing temperature the H-bonding between methanol molecules is weakened. This coiTesponds to an increase in molecular mobility obviously allowing the solvation shell to adapt faster to the pump-induced p-like electronic charge distribution. The faster this adaptation takes place the better the p-like state may be supported by the solvation shell leading to an enlarged excited state lifetime of the electron. Also, with the H-bonds acting as the main coupling channel for energy transport to the environment, cooling is suggested to take longer for a decreased efficiency of this channel at higher temperature, as observed experimentally.

References 1 2 3 4 5 6 7

W. Weyl, APC 123, 350-367 (1864). E.J. Hart and M. Anbar, The hydrated electron, (Wiley, New York, 1970). S. Bratos, J.-Cl. Leicknam, D. Borgis und A. Staib, PR E 55, 7217-7227 (1997). B. Bagchi, ARPC 40, 115-141 (1989), and references therein. T. Scheidt and R. Laenen, GPL 371, 445-450 (2003). R. Laenen, T. Roth and A. Laubereau, PRL 85, 50-53 (2000). V. Herrmann and P. Krebs, JPG 99, 6794-6800 (1995).

430

Ultrafast IR Spectroscopy on Aqueous Reverse-Micellar Nano-Droplets Dan Cringus^ Maaike T.W. Milder^ Maxim S. Pshenichnikov^ Douwe A. Wiersma , Jorg Lindner^, and Peter Vohringer^ Department of Physical Chemistry, Materials Science Centre, University of Groningen, Nijenborgh 4, 9747 AG, Groningen, The Netherlands E-mail: [email protected] Universite Louis Pasteur, Faculte de Chimie, 4 Rue Blaise Pascal, 67000 Strasbourg, France E-mail: [email protected] Abstract. The ultrafast dynamics of water nano-droplets (1-10 nm size) of the L2-phase of the ternary mixture H2O-AOT-CCI4 have been studied using frequency-resolved midinfrared pump-probe spectroscopy in the spectral region of the OH-stretching vibration.

1. Introduction The geometric confinement of liquid water is a central issue for a wide variety of research areas ranging from materials to life sciences (for example, see [1] and references therein). Aqueous droplets embedded in nano-porous hosts serve as perfectly size-adapted reaction media for the heterogeneous synthesis of novel semi-conducting colloidal materials with promising applications in optoelectronics. Nanometer-dimensioned water inclusions such as water wires and pockets are key ingredients that determine the tertiary structure and, consequently, the function of proteins. In the past, the direct observation of molecular dynamics in aqueous bulk phases has been under extensive experimental scrutiny through a variety of femtosecond time-resolved spectroscopies. Despite its tremendous relevance, experimental information regarding dynamical properties of water under geometrically confined conditions is still rather sparse. Therefore, aqueous nanodroplets confined to reverse micelles of the type oil-surfactant-water have recently raised considerable interest. Studies on molecular dynamical processes in aqueous reverse-micellar nano-droplets included solvation dynamics of water-soluble chromophores [2], and vibrational energy relaxation of small inorganic anions [3]. Furthermore, the dynamics of thermal cooling of reverse micellar nano-droplets has been measured, following an initial deposition of vibrational energy by pumping the intramolecular OH-stretching vibration of water [4]. Here, we report on time and frequency-resolved femtosecond experiments on vibrational energy relaxation of aqueous reverse micellar nanodroplets following the intramolecular OH-stretching excitation of the liquid ternary mixture CCI4: Aerosol 0T:H20. The experiments are performed with either 150-fs tunable or 70-fs IR pulses on a sample contained in a rotating sample-cell or a 100-|Lim thick freestanding jet.

431

•

H^O in acetonitrile, 3375 cm'^

A d = 1 nm, 3250 cm"^ o d= 10nm, 3500cm'^

Q

O

<

1 2 Delay (ps)

Fig.l. Transients for three types of water nano-droplets and bulk water. The solid curves represent the results of simulations.

Fig.2. Rotational anisotropics for H2O in small (d = 1 nm), large (d = 10 nm) micelles, and acetonitrile matrix.

2. Results and discussion Typical pump-probe transients for three representative micro-emulsions are compared in Fig. 1 to equivalent data on bulk water. For micelles with diameter of d=l nm, the amount of free "bulk-like" water is vanishing, for d=3 nm an approximately equal number of bound and free water is present into the nanodroplet while for d=10 nm most of the water molecules belong to the aqueous core. In the case of bulk water the signal remains nearly constant [5] for pump-probe time delays of up to 100 ps. Contrary to the bulk phase, for the water nanodroplets entrapped in reverse micelles, the transient bleach partially recovers within our observation window (Fig.l). Figure 2 presents the measured rotational anisotropy for two types of micelles. For the large micelles, where the bulk water properties are dominant, the anisotropy decays with a time constant of 200±50 fs, which is in line with previous measurements on bulk water [6]. In the smallest micelles (d=l nm), most of the water molecules are embedded in the membrane and are isolated from each other. The anisotropy displays a bi-exponential behavior, with fast component of 200±50 fs and slow one of 3±1 ps. The latter value is similar to the orientational diffusion time found for HOD [7]. For comparison, we measured the anisotropy decay of a water molecule isolated in an acetonitrile matrix (Fig. 2, filled circles), which turns out to be similar to the case of the smallest micelles. This strongly suggests that the 200-fs anisotropy decay is mainly due to an (anharmonic) coupling between symmetric and antisymmetric modes of the H2O molecule. The analysis of the data of aqueous nano-droplets is built on the generalized kinetic model depicted schematically in Fig.3. The model combines the previously reported mechanism for energy equilibration in the aqueous bulk [4], with a heat conduction description for subsequent cooling of the droplets, (i) At time t = 0, the pump-pulse (dashed arrow) excites a fraction of the entrapped water molecules from the ground state to the first excited state of the OH stretching

432

ao2(a)) CT*oi(o))

1

I aoi(co I

5a (co) "^cooling

0

r J ^

'te-^M##PH^ia

1

Fig.3. Energy level diagram for the kinetic model and the relaxation schematics. vibration, (ii) The absorbed photon energy is then transferred with a time constant Ti, into vibrational states of the same molecule and/or to vibrational quanta of neighboring particles, (iii) The excited vibrational states of these accepting molecules are depopulated with a rate constant, l/7eq, resulting in a complete and statistical redistribution of the pump energy such that the droplet temperature is raised, (iv) Finally, thermal relaxation due to heat diffusion from the entrapped water pools through the surfactant layer into the surrounding oil phase results in a decay of the local temperature. The proposed model therefore describes a lighttriggered temperature jump experiment, which takes into account a delayed perturbation resulting from the intermediate energy transfer (ii) and redistribution (iii) events. Representative fits with this model are depicted in Figs. 1,2 as solid curves. It can be concluded that by ~l-ps time delay, the interior micelles volume is canonically heated such that a temperature can be assigned. The times Ti and Tgq for the microemulsions obtained from the transients are for d=l nm (10 nm) req = 3 ps (--0.5 ps) with Ti decreasing from 800 fs for d=l rmi to 250 fs for d=10 nm. The latter value is consistent with the population lifetime for bulk water [8]. The longer time of ~8 ps (d=4 nm) and 30 ps (d=10 nm) (Fig.l) is assigned to heat diffusion out of the micelles [4].

References 1 G.-G. Chang, T.-M. Huang, and H.-C. Hung, Proc. Natl. Sci. Counc. ROC(B) 24, 89(2000). 2 D. M. Willard, R. E. Riter, and N. E. Levinger, J. Am. Chem. Soc. 120, 4151 (1998). 3 Q. Zhong, P. Baronavski, and J. C. Owrustsky, J. Chem. Phys. 118, 7074 (2003). 4 G. Seifet, T. Patzlaff, and H. Graener, Phys. Rev. Lett. 88, 147402-1 (2002). 5 A. J. Lock, S. Woutersen, and H. J. Bakker, J. Phys. Chem. A105, 1238 (2001). 6 S. Woutersen and H.J. Bakker, Nature 402, 507(1999). 7 H. J. Bakker, S. Woutersen, and H.-K. Nienhuys, Chem. Phys. 258, 233 (2000). 8 A. J. Lock and H. J. Bakker, J. Chem. Phys. 117, 1708 (2002). 9 A. Pakoulev, Z. Wang, and D. D. Dlott, Chem. Phys. Lett. 371, 594 (2003).

433

Pump-probe near-field optical microscopy of molecular aggregates using supercontinuum Tetsuhiko Nagahara^ Kohei Imura^'^, and Hiromi Okamoto^'^ ^ Institute for Molecular Science, Okazaki 444-8585, JAPAN E-mail: [email protected] ^ The Graduate University for Advanced Studies, Okazaki 444-8585, JAPAN Abstract. A novel apparatus for femtosecond pump-probe near-field optical microscopy is described. Probe pulses in visible to near-infrared regions are generated by focusing laser pulses in microstructure fiber. Excited-state dynamics of porphyrin J-aggregates are presented.

1.

Introduction

The use of scanning near-field optical microscope (SNOM), which enables spatial resolution beyond the diffraction limit, in combination w^ith ultrafast spectroscopy has been recognized as a promising technique that offers the direct information of ultrafast dynamics in mesoscopic systems [1-3]. This combined technique shed light on the relationship between dynamics and nanometric structure. Up to now, we have explored quasi-one-dimensional systems by one-color equal-pulse correlation technique [4-6]. In this contribution, we describe a SNOM apparatus for time- and space-resolved experiments in two-color pump-probe scheme equipped with supercontinuum (SC) light source. SC generation by amplified laser pulses in bulk materials has been widely used as broadband light sources outside the tuning range of the laser. Nevertheless, there are only very few report dealing with its application to the near-field microscopy primarily because of high-intensity of SC pulses which may damage the near-field tip. However, the spectral broadening of unamplified laser pulses in the microstructure fibers was demonstrated recently [7], and has been successfully applied to various spectroscopies [8-10]. The relatively low-intensity/highrepetition-rate continuum generated in this way is an ideal broadband/tunable light source for SNOM experiments.

2.

Experimental IMethods

Schematic drawings of the apparatus are shown in Fig. 1. Our homemade SNOM consists of a closed-loop piezo-electric x-y-z stage, microscope objective, and detection systems. The near-field apertured probe (aperture -100 nm) used was made of a single-mode optical fiber. The samples were illuminated through the aperture. Light sources were coupled to the other end of the fiber. The distance between the tip and the sample surface was regulated by shear force feedback.

434

Pulses from a Ti:Sapphire laser (< 100 fs, 780 nm, 80 MHz) were used as the light source. The beam was split into two parts, pump and probe. In one-color experiment, both beams ("A" and "B") were collinearly coupled to the near-field probe after attenuation and GDD compensation, while in two-color experiments SC was generated by focusing one of the split beams ("B", --250 mW, 780 nm) into the microstructure fiber (1 m, zero-dispersion at 780 nm). The continuum was coUimated and used either as it was (after attenuated to '- 1 mW) or as a pseudo-monochoromatic light sources (< 0.1 mW) by passing through interference bandpass filters (10 nm FWHM). near-field

Fig. 1. Schematic illustrations of pump-probe SNOM.

HM: half-mirror. MO: microscope objective. Si photodiode

3.

Results and Discussion

Time-integrated spectrum of SC pulses after passing through the SNOM apparatus was measured by a CCD spectrometer. We have observed broadband SC from 560 nm to more than 1 |.im. The characteristics of the spectrum were similar to those found in other experiments [7,10]. The peaks in the spectrum fluctuate in intensity and position due to the fluctuation of the incident laser. In time-resolved experiments, it is essential to characterize temporal profiles of the light pulses. By the use of two-photon-induced photocurrent in photodiodes (SiC and GaAsP), cross-correlations have been measured at several wavelengths in the near-field. As shown in Fig. 2, the correlation width obtained are dependent on probe wavelength: 1-2 ps for near-IR and ^5 ps for visible (570 nm). The observed group delay agreed well with sum of group delay for the microstructure fiber and that for fused silica fiber used in the near-field probe. 830 nm

1.0

Time / ps

Tlme/ps

Time / ps

0.0

1.0

Time / ps

Fig. 2. Cross-correlation traces obtained at various probing wavelengths

435

To illustrate performance of the apparatus in two-color pump-probe scheme, excited-state dynamics of thin film of tetrakis(4-methoxyphenyl)porphyrin (TMeOPP) J-aggregate have been measured. Recently we investigated the sample by near-field equal-pulse transmission correlation (EPC) technique, and obtained site-dependent time constants in several tens of ps region [4,5]. The results are probably due to the site-specificity of excited-state dynamics, but it was not very clear because of the experimental uncertainties. In one-color EPC measurements, the time-resolved signal is expressed as a convolution of the instrumental response function and a symmetric double-sided decaying function, where the decay is attributed to the excited-state population decrease [1], This is because pump pulse is identical with probe pulse when it is time reversed. Consequently, we encounter difficulties in curve fitting since the background level and/or contributions from the long-lived species are not known. On the other hand, site-specificities in excited-state dynamics were revealed much more clearly (shown in Fig. 3) in the present experiment compared to our previous one-color results, since the baseline was directly determined from intensities at negative delay times. Sjinni (a) (b) (c) T = (79*6)ps

100 200 Time/ps

300

Time / ps

Fig. 3. (b),(c) Pump-probe signals obtained at position indicated in topograph (a).

4. Conclusions A pump-probe SNOM apparatus utilizing SC as a light source has been developed. It has been shown that the broadband continuum can be used as a wavelength-tunable light source for probing absorption of excited states at arbitrary wavelengths in SNOM. Using two-color pump-probe scheme, sitespecificities in dynamics of excited TMeOPP J-aggregates were revealed clearly.

References 1 2 3 4 5 6 7 8 9 10

436

S. Smith et al., Ultramicroscopy 71, 213, 1998. T. Guenther et al., Appl. Phys. Lett. 75, 3500, 1999. H. Kawashima et al, Appl Phys Lett 77, 1283, 2000. T. Nagahara et al, Chem. Phys. Lett. 381, 3,68, 2003. T. Nagahara et al., Scanning, in press. K. Imura et al., J. Phys. Chem. B., in press, 2004. J.K. Ranka et al., Opt. Lett. 25, 25, 2000. I. Hartl et al., Opt. Lett. 26, 608, 2001. H.N. Paulsen et al. Opt. Lett. 28, 1123, 2003. V. Nagarajan et al. Rev. Sci. Instrum. 73, 4145, 2002.

Vibrational and Rotational Relaxation Dynamics of Anions in Reverse Micelles by Ultrafast Infrared Spectroscopy J.C. Owrutsky, G.M. Sando, Q. Zhong, and A.P. Baronavski Chemistry Division, US Naval Research Laboratory, Washington DC 20375-5342 Abstract. Vibrational energy and rotational relaxation times for triatomic anions in the water pools of reverse micelles (RM) vs^ere measured by ultrafast infrared spectroscopy. The dynamics are slower than in bulk water due to confinement effects. Studies of vibrational and rotational dynamics of small ions in reverse micelles (RM) are used to investigate confinement effects on ion solvation. RMs are composed of nanosize water droplets stabilized in bulk organic phase by surfactants. The water droplet size increases with the water content {co = [H20]/[surfactant]). Due to the immobilized water molecules that hydrate the surfactant headgroups, the water inside the RMs has less hydrogen bonding and lower polarity than bulk water as shown in previous studies [1,2]. A complication in using solute probes to investigate properties of the water core in RMs is the ambiguity of the solute location. Small molecular ions are very hydrophilic due to their size and charge, so they are convenient solutes for probing the RM water core interior via infrared spectroscopy. The vibrational frequency of the antisymmetric stretching band of azide (N3*) in RMs depends on the surfactant charge [3]. Compared to bulk water, this band near 2000 cm'^ is red-shifted in RMs composed of nonionic surfactants (alkyl-polyoxyethylenes, NP and Brij30), more so in cationic cetyltrimethylammonium bromide (CTAB) RMs, and is slightly blue-shifted in anionic sodium bis(2-ethylhexyl) sulfosuccinate (AOT) RMs. Similar results were found for cyanate (NCO). As seen in many previous studies of spectral and dynamical processess in RMs, the effects are most pronounced for the smallest RMs (lowest 00) and tend toward the bulk water value as the RM water content and size are increased. We report ultrafast infrared studies of the vibrational energy (TO and rotational reorientation (T^) times for Na" NCO", and NCS" in NP RMs [4,5], as well as for N3" in AOT and CTAB RMs. In bulk solution of polar solvents, the vibrational energy relaxation (VER) of these ions is known to be rapid due to Coulombic interactions [6], providing a context for interpreting confinement effects of RMs on ion solvation. Polarization-resolved infrared pump-probe transient absorption measurements are performed on static solutions of ions dissolved in RMs composed of different surfactants [3]. Infrared pulses near 2000 cm"^ (FWHM of -250 fs and 150 cm'^) are generated via a 1 kHz Ti:Sapphire regenerative amplifier/OPA difference frequency generation scheme and split into pump and probe beams. Pump-induced transmission changes are monitored on the probe beam, which is frequency and polarization resolved after the sample. Ti times are measured at the magic angle (54.7°) and Tr times are determined from the anisotropy decay.

437

Fig. 1 shows transient absorption decays measured for N3" in NP RMs for several values of co. Compared to the bulk water value (0.8 ps), the VER decay times are about three times longer (2.5 ps) for the smallest RM studied (co=l) and become shorter and approach the bulk water value with increasing RM water content. The results for all three triatomic anions in NP RMs, which are listed in Table 1, indicate similar confinement effects in the RMs. This behavior is consistent with the previously established trend for dynamics in RMs that as the water pool size gets larger, the influence of the surfactant headgroups on the water interior is reduced. The same trend is observed in the static IR spectra.

(A)

(B)

0.4

i

IA\ '^'^'^

^"'^oovv.cw^ 2.5±0.2 ps, co=1 0.1 §

lfy\

^°V«55.^5^3_^^^

1.2±0.2ps, 0=10

'-""--"^f ^ X.,,^^^^^

.

1

2

co = 1 CO = 1 0

%«vs^^

^^^"^T"""""^

1.7±0.2ps, 0 = 4

bulk water

0.8±0.1 ps, bulkHjO

3

2

4

3 Delay Time / ps

Delay Time / ps

Fig. 1. Transient absorption (A) and anisotropy decay (B) curves for azide ion in NP reverse micelles.

The VER rate is only moderately affected in RMs compared to the more dramatic rate reduction reported for other dynamical processes, such as solvation dynamics [1,2]. The vibrational solvent shift and VER rate both reflect the strength of the solute-solvent interaction strength, so that they depend on co in a similar way. This spectral-lifetime correlation is also observed for N3" and NCO" in bulk solution for different solvents. Table 1. Vibrational energy relaxation times (TO for N3 , NCO , and NCS in NP reverse micelles and in bulk water.

N3-

NCS" NCO"

co=l 2.5 (0.2)

co=2

2.0 (0.2) 10(2) 2.2 (0.2)

co=6 1.4(0.2) 6.4(1) 1.5(0.2)

co=10 1.2(0.1) 1.2(0.2)

H2O

0.8(0.1) 2.7 (0.3) 0.9(0.1)

Panel (B) in Fig. 1 shows anisotropy decay curves measured for azide in NP RMs for low and high co as well as in bulk water. The reorientation times are lengthened more than the Ti times compared to bulk water. Reorientation occurs more rapidly for the larger RMs, but it is still substantially longer than in bulk water. This provides evidence that the ion mobility in RMs is reduced more than the solute-solvent interaction that mediates VER. The decrease in Tr with CO

438

suggests that the ion might reside inside the water interior. If the ion were attached to a headgroup at the surfactant wall, the Tr might increase with the RM size. Additional information regarding the dynamics and solute location in the RMs was obtained by similar studies of azide in AOT and CTAB RMs. The rotational and vibrational dynamics of azide in AOT are indistinguishable from those in bulk water, suggesting that the anion is located in the water core region. For cationic CTAB, the VER rate is about three times slower than bulk water for the smallest RM, similar to nonionic NP RMs, but the VER rate for CTAB RMs levels off at co = 5. This is consistent with the anion being attached to or strongly interacting with the micelle interface. The results for nonionic Brij-30 RMs are similar to those for NP RMs. For CTAB and AOT RMs, the relative rates and spectral shifts are similar in D2O and H2O. Measurements were also performed in mixtures of water and tri(ethyleneglycol) monomethyl ether (TGE) monomer, in which the latter resembles the hydrophilic portion of the nonionic surfactants. The results for the spectra and VER times as a function of water content strongly resemble those for the nonionic RMs. Water and azide appear to interact with the poly-oxo chains for small co, especially for the longer chained NP surfactant. As co increases and the water pool becomes better defined, azide begins to prefer a location within the water pool, allowing bulk-like behavior to be approached for large RMs. The dependence of the spectra and dynamics on RM charge is attributed to the ion location in the RMs. For anionic AOT, azide appears to prefer the center of the water pool for all values of co, resulting in bulk-like dynamics. The behavior in CTAB is consistent with a strong interaction between the cationic surfactant and anionic azide, which draws the anion toward the outer water layers. The results for nonionic RMs and PEO/water mixtures are similar and suggest that water penetrates into and hydrates the poly-oxo chains before a water pool is formed. The interfacial region in nonionic RMs include the poly-oxo chains, which appear to be hydrated, so that the boundary between the interface and the water core is less clearly defined than for the ionic RMs. The results of this work demonstrate the effects of RM confinement on spectral and dynamical vibrational properties of small ions. The results depend on the surfactant charge in a manner that is consistent with the ion location being determined by Coulombic interactions with the surfactant headgroup. Acknowledgements. This work was supported by the Office of Naval Research at the Naval Research Laboratory. G.M.S. acknowledges the NRL-ASEE Postdoctoral Fellowship program. Q.Z. acknowledges the NRL-NRC Research Associateship program.

References 1. N. E. Levinger, Current Opinion in Colloid and Interface Science 5, 118 (2000). 2. N. Nandi, K. Bhattacharyya, and B. Bagchi, Chem. Rev. 100, 2013 (2000). 3. Q. Zhong, D. A. Steinhurst, E. E. Carpenter, and J. C. Owrutsky, Lang. 18, 7401 (2002). 4. Q. Zhong; A. P. Baronavski, and J. C. Owrutsky, J. Chem. Phys. 118, 7074 (2003). 5. Q. Zhong, A.P. Baronavski, and J. C. Owrutsky, J. Chem. Phys. 119, 9171 (2003). 6. M. Li, J. C. Owrutsky, M. Sarisky, J. P. Culver, A. Yodh, and R. M. Hochstrasser, J. Chem. Phys. 98, 5499 (1993).

439

Part VI

Reaction Dynamics in Solution

Femtochemistry in the electronic groundstate? IR-driven cis-trans isomerization of HONO Peter Hamm, Roland Schanz, Virgiliu Botan Physikalisch Chemisches Institut, Universitat Zurich, Winterthurerstr. 190, 8057 Zurich, Switzerland E-mail: [email protected] Abstract. We investigate the dynamics and mechanism of the IR-driven cis-transisomerization of nitrous acid (HONO) in a low-temperature krypton matrix applying ultrafast timeresolved IR spectroscopy. Cis HONO isomerizes with 10 % quantum yield on a 20 ps time scale to trans HONO. We developed a 4D model of the system, which includes the three proton intramolecular degrees of freedom of HONO fully quantummechanically and one intermolecular translational degree of freedom of the molecule in the crystal cage. This allows us to suggest a possible reaction pathway.

1. Introduction It is now well established that electronically driven photochemistry is ultrafast in general. However, most (not all) chemistry is happening on electronic ground state potential energy surfaces, rather than through electronically excited states. From the experimental point of view, very little is known about the dynamics of groimd state chemical reactions in the condensed phase on a microscopic, atomic level. Time resolved studies with high time resolution require a sharp trigger, which, in the case of electronically driven photochemistry, is a short pump laser pulse in the visible or UV spectral range. The same has become possible only recently in the IR spectral region, i.e. on the electronic ground state surface, with the recent advances in femtosecond IR technology. We present ultrafast time resolved spectroscopy of the IR-driven cis-trans isomerization of nitrous acid (HONO) in a low-temperature Kr-matrix. The molecule is one of the few known examples which undergo a conformational transition upon excitation of one quantum of a vibrational mode (i.e. a one-phonon-one-photon-process) [I]. The energy of the c/^'-configuration of HONO lies 100-200 cm"^ above that of the trans configuration [2]. The transition state is located along the torsional coordinate at (|)=86° with an energy of 4000±500 cm"\ Hence, the energy of one quantum of the OH stretch vibration («3500 cm-1) is of the same order as the transition state. In fact, the cis-trans quantum yield after excitation of the cis OH stretching mode has been reported to approach 100% [1,3]; a puzzling result, given that it is not the reaction coordinate which is initially excited. Apparently, the OH-stretching coordinate is efficiently coupled to the torsional coordinate. The quantum yield of the trans-cis back reaction is «14 %. Experimental information about the photo-isomerization of HONO stems either from stationary spectroscopy, revealing for example the relative energetics of the cis and the trans

443

CD O

0.2

H

c CO

3402 cm"'

ON,

O 3552 cm"' CIS

8 0.1

H

trans ^

0.0 3400

3450 3500 Wavenumber [cm'^] Fig. 1. Absorption spectrum of HONO in solid Kr

3550

state through thermodynamic measurements, or from kinetic measurements, revealing the quantum efficiency of the reaction. However, the mechanism of the photo-isomerization, or merely the order of magnitude of its timescale (whether it is femtoseconds or milliseconds) is completely unknown. It is not known whether coherent wavepacket motion or tunneling is involved, or whether it is a quasiclassical random walk which gives rise to the cis-trans isomerization. A summary of experimental and theoretical work on HONO is given in Ref. [2].

2.

Experimental Results

The steady state absorption spectrum of HONO in solid krypton at 30K is depicted in Fig. 1. In the spectral region shown it consists of two main bands, the OHstretching band of the cis isomer at 3402 cm"^ and that of the trans isomer at 3552 cm"^ respectively. Fig. 2 (a) shows transient difference spectra after selective excitation of the OH stretching band of the trans isomer. The dominating bands are the trans bleach/ stimulated emission band at 3552 cm"^ and the trans excited state absorption band at 3390 cm"^ Both decay bi-exponentially on a 8 ps and 260 ps timescale. Simultaneously with this decay, a third signal is growing in between 3515-3550 cm"^ which we refer to as 'dark states'. It stems from trans molecules that have relaxed from the originally excited OH-stretching state, the bright state, to lower lying vibrational modes which are not observed in the steady state absorption spectrum and thus called dark states. These states can be overtones and/or combination modes, of which many exist in the molecule. As a consequence of anharmonic couplings between these dark states and the bright OH-stretching mode, the later absorbs slightly red-shifted with respect to its original frequency, once one of the dark state is populated. Fig. 2 (b) shows transient difference spectra after selective excitation of the OH stretching band of the cis isomer. This time, only the strong cis bleach/stimulated emission signal at 3402 cm"^ is observed, while the cis excited state absorption is outside the spectral window of this measurement. The cis excited state is depopulated on a faster 20 ps timescale. Most important is a small, broad signal in

444

cmr^^^-^

^ c 1 ^

0 D) C CD

o 0

o c

M

CD

o

cis pump>^ f/'a^s

CO

<

3400 3450 3500 3550

1

.

11 1

500 ps 3500 ji

1

1 3520 1

1

3540 1

J

3400 3450 3500 3550

Wavenumber [cm" ]

Fig. 2 (a) Transient response after selective excitation of trans HONO revealing a trans bleach/stimulated emission signal at 3552 cm'^ a trans excited state absorption signal at 3390 cm'^ and a trans 'dark state signal between 3515-3550 cm'^ (see arrow), (b) Transient response after selective excitation of cis HONO revealing a cis bleach/stimulated emission signal at 3402 cm'^ and a trans (!) dark state signal between 3515-3550 cm'^ (see arrow). The cis excited state absorption is outside the spectral window of this measurement. The signal is superimposed on a broad background which is due to a water film that is growing on the sample. the region between 3515 cm'^ and 3550 cm'^ (see arrow) which we attribute to trans (!) dark states. This signal directly reflects the cis-trans isomerization. The inset of Fig. 2 (b) compares the dark state signal after pumping trans HONO directly (gray line), with that after pumping cis HONO (black line). As both are essentially the same, we conclude that the same trans overtones and combination modes are populated after isomerization. Fig. 3 compares the appearance of the trans hot states after direct pumping of trans HONO (gray line) with that after pumping of cis HONO with subsequent isomerization (black line). The latter signal rises on a 20 ps time scale which directly reflects the isomerization rate. From the amplitudes of the signals we can estimate for the cis-trans isomerization quantum yield 10% within the first 500 ps. We presently cannot exclude additional slower reaction channels.

445

c

• 1 •• • '-1 cis pump

CD

D) C

CO -C

O 0 o c

*

• # # ^

-

•

*

j

^

i.'^^"*'^^ trans pump

CO

o CO

<

•

• • ' ' ''

*

•

•

1

1 1 t 1 1 1

10

100

1

I

1 1

500

Time [ps] Fig. 3 Appearance of the trans hot states after direct pumping of trans HONO (gray line, bi-exponential fit with 8 ps and 260 ps time constants) and after pumping of cis HONO with subsequent isomerization (black line, fit with 20 ps time constant).

3. Discussion Recent high level quantum dynamic calculations suggest that the molecule would not isomerize in the gas phase since the density of states is too small to efficiently couple both isomers [4, 5]. In the gas phase, the proton states are discrete and are stationary eigenstates. This changes dramatically when the molecule is brought into the condensed phase where it is coupled to a quasi-continuum of states. We have developed a 4E) model with three HONO intramolecular degrees of freedom and one translational degree of freedom of the HONO molecule in a rigid matrix cage [6]. We assumed that the -ONO body is stiff and considered as intramolecular modes only the three hydrogen degrees of freedom with respect to the -ONO body. We used an empirical proton potential and modeled the interaction with the matrix using Lennard Jones parameters. We find that the HONO molecule occupies an one-atom substitutional site. Not surprisingly, however, we find that the minimum positions of the cis and trans isomers are slightly different, leading to the important conclusion that the cis-trans isomerization of HONO is necessarily accompanied by a translational motion of the molecule as a whole. It is this coupling between intramolecular and intermolecular degree of freedom which is responsible for efficient isomerization pathways. Quantum mechanical calculations as function of the position of the molecule in the matrix cage reveal that the proton eigenstates a tuned with respect to each other. Effective transfer between two states is possible only when donor and acceptor states are in close resonance, and translational coordinates in the matrix

446

cage may fine-tune this resonance. In the condensed phase, resonance is no longer a matter of an unlikely coincidence, but will almost necessarily occur at a particular position of the molecule in the matrix cage. Within a reasonable set of parameters for the proton potential, we may suggest the following reaction pathway for cis-trans isomerization [6]: After excitation from the c/s-ground state, the cis OH stretch excited state (Vstretch,cis) is populated. Direct coupling to the trans OH-stretch excited state (Vstretch,trans) will be extremely weak, because of the energy mismatch and the vanishing spatial overlap of the two wave functions. Accordingly, we do not observe any excited state absorption from Vstretch,trans ^ftcr the decay of the cis OH-stretching state. However, we do find a curve crossing between Vstretch,cis and the 8* overtone of the torsional mode, 8vtors,cisj at a particular position of the molecule in the cage. The coupling between both states is weak (0.4 cm"^), but nonzero. In the avoided crossing region, the system might hop from one diabatic surface to the other, and thereby transfer population from the stretching mode to a high torsional mode. The coupling between 8vtors,cis and 8vtors,trans is large (110 cm"^) because of a large spatial overlap of the wave fimctions. The system therefore will stay on the adiabatic potential energy surface and transfer population to the trans side of the molecule. The timelimiting step of this reaction pathway is the transfer of population from Vstretch,cis into a high torsional mode. The subsequent transfer to the trans side will be almost instantaneous since it proceeds on one adiabatic potential energy surface. Accordingly, we observe the appearance of vibrationally excited trans species on the same timescale as the disappearance of the cis OH-stretch excited state. The model we developed in Ref [6] is not quantitative. However, we believe it gives insights into the types of couplings that might occur and it does in fact explain qualitatively all observations we make in the experiment.

References 1 R. Hall and G. Pimentel, J. Chem. Phys. 38, 1889 (1963) 2 G. DeMare and Y. Moussaoui, Int. Rev. Phys. Chem. 18, 91 (1999). 3 L. Khriachtchev, J. Lundell, E. Isoniemi, and M. Rasanen, J. Chem. Phys. 113, 4265 (2000). 4 F. Richter, M. Hochlaf, P. Rosmus, F. Gatti and H-D. Meyer, J. Chem. Phys. 120, 1306 (2004) 5 D. Luckhaus, J. Chem. Phys. 118, 8797 (2003) 6 R. Schanz, V. Botan, P. Hamm, J. Chem. Phys. submitted

447

Bimodal Intermolecular Proton Transfer in Acid-Base Neutralization Reactions in Water O. F. Mohammed^ M. Rini\ J. Dreyer\ B.-Z. Magnes^ D. Pines^ E. T. J. Nibbering^ and E. Pines^ ' Max Bom Institut fuer Nichtlineare Optik und Kurzzeitspektroskopie, Max Bom Strasse 2A, D-12489 Berlin, Gemiany E-mail: [email protected] ^ Department of Chemistry, Ben-Gurion University of the Negev, P.O.B. 653, Beer-Sheva 84105, Israel E-mail: [email protected] Abstract. We present an ultrafast mid-infrared study of excited-state intermolecular proton transfer in photoacid-base pairs in water. Observation of bimodal reaction dynamics demand refinement of the established Eigen-Weller mechanism.

Exploring the mechanism of proton transfer between acid and bases is of fundamental importance for the understanding of e.g. the autoionization of water and the associated von Grotthuss mechanism, acid-base neutralizations, enzyme catalysis, and proton pumps through membranes. In the pioneering work of Weller and Eigen [1,2] intermolecular proton transfer between acid and bases is understood to be controlled by mutual diffusion forming an encounter pair, with an apparent reaction radius (contact separation) of typically 6 to 8 A. Within the Eigen-Weller model of acid-base reactions, the lower limit for on-contact proton exchange rate is typically (10 ps)'^ M'^ for the overall reaction rate to be diffusion limited. Direct access to the actual proton transfer remains problematic due to the slower diffusion process dominating the overall dynamics. Our approach to decipher the mechanisms of intrinsic proton transfer is to monitor specific vibrational marker modes of acid and base after initiation of the proton transfer reaction. Photoacids can be used as a means of light triggered proton transfer, thus enabling time-resolved studies [3]. Photoacidity of aromatic alcohols, with typically a decrease in pK^ of 5-10 units upon excitation, is a research subject on its ovm, where traditionally a charge transfer from the oxygen atom of the OH-group to the aromatic ring is thought to be responsible for the increased acidity of the photoacid. A more detailed, yet to be verified, explanation invokes the fact that for aromatic systems electronic states excited along the short and long axes are often energetically close, with a possible occurrence of state inversion due to polar interactions with the solvent [4]. Yet another explanation involves a level crossing to a state with the net effect of transfer of a hydrogen atom [5]. We use a photoacid, the widely used dye stain pyranine (8-hydroxy-1,3,6trisulfonate-pyrene, HPTS), to trigger the proton transfer reaction. We have

448

studied pyranine with femtosecond mid-infrared spectroscopy in the fingerprint region in H2O (between 850-1350 cm''), where vibrational bands of the sulfonategroups are located, and D2O, CD3OD and dimethylsulfoxide-de (1250-1800 cm"') where aromatic ring and C-0 modes have their resonances. When exciting around 400 nm in the electronic origin transition, the photoacid vibrational bands in the SpState appear within time resolution (150 fs), after which no changes in intensity have been observed in the time range up to 10 ps (see Fig. 1). At longer delay times, after correction for rotational diffusion effects, the decay of these bands, (and the rise of bands associated with the conjugated photobase) is explained by the proton (deuteron) transfer to the solvent with a time constant of 90 (250) ps (Fig. 1). An identical band structure is observed when exciting with excess energy at 349 nm. In the latter case the photoacid bands appear slightly red-shifted, due to anharmonic coupling with highly excited Raman-active modes [6]. Subsequent blue-shifting is then the consequence of intramolecular vibrational redistribution and cooling, taking place on a time scale up to 20 ps, similar to observations in excited state intramolecular hydrogen transfer. The transient spectra of pyranine dissolved in deuterated dimethylsulfoxide and deuterated methanol on the other hand show a distinctly different vibrational mode pattern. By comparison of experimental and calculated vibrational spectra of pyranine the electronic ground state we derive that the electronic state reached in 150 fs after excitation in water is clearly distinct from the state observed in methanol or dimethylsulfoxide. Femtosecond infrared spectroscopy has the potential to give direct insight into an acid-base neutralization reaction in a site-specific way: dynamics of vibrational

1600

1500

1400

1300

Wavenumber (cm*^) Fig. 1. Transient IR spectra of pyranine recorded 0.5 ps after excitation at 400 nm in three deuterated solvents dimethylsulfoxide, methanol and water, showing a marked solvent dependent vibrational pattern.

449

bands of the photoacid or conjugated photobase indicate when the deuteron leaves the photoacid, whereas a marker mode of the acid formed is indicative of the arrival of the deuteron at the accepting base. We have studied the acid-base neutralization reaction of pyranine with several bases: acetate [7,8], chloroacetate, dichloroacetate and trichloroacetate. When exciting pyranine dissolved at a fixed concentration of 20 mM in D2O, with varying amounts of the base acetate added (ranging from 0.25 M - 4 M), the deuteron transfer reaction can be followed in real time by inspection of the decay of the 1486 cm"^ marker mode of the photoacid, the rise of signal at the 1503 cm'^ transition of the conjugated photobase, and the rise of the C=0 band of acetic acid at 1720 cm"^ (Fig. 2). At low base concentrations (0.25 - 0.5 M) deuteron transfer to the solvent, followed by deuteron pick-up by acetate, dominates the dynamics, as can be learnt from a faster rise of the photobase signal than of the acetic acid band. This feature cannot be observed in time-resolved fluorescence or UV-pump/VIS-probe measurements where it was indirectly studied by monitoring the proton-scavenging effect on the geminate recombination reaction of the photoacid [9,10]. At high base concentrations (> 1 M) practically identical rise times for photobase and acetic acid are observed indicating a dominant direct deuteron scavenging mechanism from the photoacid by the base. At these high concentrations the observed reaction dynamics are bimodal. The two contributions to the signals can be ascribed to pyranine-acetate complexes with a pre-formed hydrogen bond along the reaction coordinate, and initially uncomplexed pyranine, that first has to form an encounter pair with

c

C

20

40

60

80

100

Pulse Delay (ps)

40

60

80

100

Pulse Delay (ps)

Fig. 2. Rise of the carbonyl stretching marker mode of acetic acid at 1720 cm" for different concentrations of acetate (dots). The solid lines denote calculated signals including contributions by "tight" complexes and "loose" complexes (a), or by "tight" complexes and "loose" complexes with addition of a static component (b).

450

acetate before a reaction can proceed. The pre-formed complex shows deuteron transfer faster than 150 fs. In contrast, for the fraction of initially uncomplexed pyranine, where the reaction coordinate is established by the diffusion of the reactants and by solvent fluctuations, a much slower bimolecular reaction rate on contact (a = 6.3 A) of (30 ps)"^ M"^ is found from our data analysis using theory of diffusion-controlled bimolecular reaction dynamics as given by von Smoluchowski with Collins-Kimball radiative boundary condition (SCK) [11] assuming fully screened potential at the high base concentrations used (Fig.2, left). A better fit is achieved when a static reaction component (with a time constant of 6 ps) is added to the SCK model (Fig. 2, right), describing a fraction of reactive pairs already at close range (but not directly complexed to) each other at the time of initiation of the reaction, and thus not delayed by diffusion. The fact that the deuteron transfer reaction in the pre-formed "tight" pyranine-acetate complex is at least 2 orders of magnitude faster than the "loose" pyranine" acetate encounter complex, leads to the important finding that the deuteron transfer mechanism as initially suggested by Eigen and Weller has to be refined. One explanation follows the line of argument that the acid and base in the encounter complex can only react after substantial rearrangement of water molecules in the solvation shells before acid and base reach direct contact, and the slower reaction rate of the encounter complex is due to this bottleneck solvent rearrangement dynamics. An alternative explanation lies in the possibility of a von Grotthuss-type hopping of the proton from the acid to the base via solvation shell water molecules, in which case acid and base never reach direct contact and the overall concerted proton transfer reaction is considerably slower than the hopping time of the proton along a single hydrogen-bond. The latter suggestion can be related to 1

•

-

1

1

,

1

1

KAr

CO

1

1

1

1

1

1

1

•^fji^'''l^i!sCx

•y*'"\r •

c

"oi c

1

•.

•

2l-^^ j^^i^^^j^x**"^ ' ' ^.^t^^"*^'^ ^* >* Jc^y^o\ko w^^^f^**" **

't/T

0)

1

•JT «

*A X

AJ

m'

\<£'\J\o J
i'jf ^'^

r V^^

1

•

*!s

*JSF

' t% S

y>^

Iff ^' ^ tlf 1A J^

•

*•

*>^*

~

KCIAc

•

-

7K*

•

(a)' Bsr . . 1 . . . 1 . . . 1 . . .

200

400

600

Pulse Delay (ps)

. ..

800

(b) 200

400

600

800

1000

Pulse Delay (ps)

Fig. 3. Rise of the carbonyl stretching marker modes of acetic acid, monochloro-, dichloro- and trichloroacetic acid at 1720 cm'' for 1 M of base (a) and 3 M of base (b). The solid lines denote calculated signals.

451

recent numerical studies of the von Grotthuss mechanism of proton conductivity [12] and HF ionization and ion pair formation in water [13]. Comparison of the rise of the C=0 stretching band of the conjugated acids for the bases acetate, monochloro-, dichloro-, and trichloro-acetate, reveals that the contribution of the "tight" complex appears clearly within time resolution for the first two bases, whereas for latter two the data suggest a finite rise on a time scale of about 200-300 fs. The rise time of the fraction due to the diffusion controlled reaction dynamics on the other hand depends mainly on the base used. The reaction rate on contact, ko, found in our preliminary kinetic study is found to gradually decrease when moving from acetate to trichloroacetate. The bimolecular proton transfer rate constant on contact, ko, from the photoacid to the carboxylic bases is: (12ps)-^M-\ (100ps)-^M-\ (450ps)-^M-\ (2500ps)-^M"\ for the acetate, monochloro-, dichloro- and trichloro-acetate bases, respectively. The results reflect the gradual decrease in basicity of the carboxylic bases. Such a behaviour is expected when taking into account the increasing acidity of the conjugated carboxylic acid series, pKa = 4.7 (acetic acid), 2.9 (monochloroacetic acid), 1.4 (dichloroacetic acid), ~0 (trichloroacetic acid). A c k n o w l e d g e m e n t s . This work has been supported by the German-Israeli Foundation for Scientific Research and Development (GIF 722/01), the LIMANS Cluster of Large Scale Laser Facilities (MBI000228), the James Franck GermanIsrael Binational Program in Laser-Matter Interaction (EP) and a long term mission fellowship of the Egyptian government (OFM).

References 1 A. Weller, Prog. React. Kin. 1, 187, 1961. 2 M. Eigen, Angew. Chem. Int. Ed. 3, 1, 1964. 3 E. Pines and D. Pines, in Ultrafast Hydrogen Bonding Dynamics and Proton Transfer Processes in the Condensed Phase, Edited by T. Elsaesser and H.J. Bakker, Kluwer, Dordrecht, the Netherlands, 155, 2002. 4 T.-H. Tran-Thi, C. Prayer, Ph. Millie, P. Uznanski and J.T. Hynes, J. Phys. Chem. A 106, 2244, 2002. 5 W. Domcke and A.L. Sobolewski, Science 302, 1693, 2003. 6 M. Rini, A. Kummrow, J. Dreyer, E. T. J. Nibbering and T. Elsaesser, Faraday Discuss. 122, 27, 2003. 7 M. Rini, B.-Z. Magnes, E. Pines and E.T.J. Nibbering, Science 301, 349, 2003. 8 M. Rini, B.-Z. Magnes, D. Pines, E. Pines and E.T.J. Nibbering, J. Chem. Phys., in press, 2004. 9 E. Pines, B.-Z. Magnes, M.J. Lang and G.R. Fleming, Chem. Phys. Lett. 281, 413, 1997. 10 L. Genosar, B. Cohen and D. Huppert, J. Phys. Chem. A 104, 6689, 2000. 11 A. Szabo, J. Phys. Chem. 93, 6929, 1989. 12 D. Marx, M.E. Tuckerman, J. Hutter and M. Parrinello, Nature 397, 601, 1999. 13 K. Ando and J.T. Hynes, J. Phys. Chem. A 103, 10398, 1999.

452

Ultrafast Excitation Energy Migration Processes in Various Porphyrin Arrays. Dongho Kim Department of Chemistry, Yonsei University, Seoul 120-749, Korea E-mail: [email protected] Abstract We have investigated excitation energy migration processes in various forms of porphyrin arrays (linear, cyclic, box) by time-resolved spectroscopic techniques in tenns of exciton coupling between adjacent porphyrin units in these arrays. A major source of mspiration for the design and synthesis of optical nanostmctures comes from the light-harvesting antenna complexes of natural photosynthetic systems. Continuing efforts to realize the mimicry of solar energy harvesting complexes have enabled the design and synthesis of various types of covalently linked porphyrin arrays w^ith the goal of applying these arrays to molecular photonic devices and artificial light-harvesting array systems. The recent success in elaborating various porphyrin architectures using several types of linkers via meso position attachment has brought up the issues on the electronic coupling, of v^hich the extent is determined by the interconnection length and relative orientation between the adjacent porphyrin moieties. By exploring the excitation energy transfer (EET) processes and exciton coupling dynamics, we can gain fiirther insight into the relationship of the EET phenomena with the interconnection length and relative configuration between adjacent porphyrin pigments [1]. Thus, a fine control of molecular structures is highly desired in regular and well-arranged architectures with the improved lightharvesting ability. In this context, an investigation on the photophysics of the newly designed strongly coupled porphyrin arrays in relation to the EET processes would be pertinent to the improvement of the EET efficiency between the chromophores.

Fig. 1, Directly linked porphyrin array (left) and ZnA system (right) First, the EET processes occurring in directly linked linear porphyrin arrays (Zn, n=l,2,3,4...) has been investigated by various time-resolved spectroscopic measurements [2]. The exciton coherent length of Nc=4.5 units is evaluated based on a semiemprirical equation and also confirmed by the SI state lifetime dependence on the number of porphyrin units in Zn. The exciton coupling

453

dynamics between the monomer units in Zn was observed to be less than 200 fs by femtosecond transient absorption anisotropy decay. We also investigated the EET process as well as its efficiency by attaching porphyrin donor moiety via phenylene linkage to one end of Zn with a change in the number of porphyrin units in Zn (ZnA) ,The EET rate for ZnA can be measured by monitoring the transient absorption spectral changes after selective photoexcitation of Zn moieties in ZnA system. The overall EET time to acceptor A in ZnA becomes slower with an increase in the number of porphyrin units in energy donor array Zn due to the increased excitation energy migration time within Zn prior to EET to A The estimated EET rates based on Fermi golden rule can be well correlated by the exciton coherence length Nc=4.5 units in Zn. On the other hand, as energy donor Zn array becomes longer in ZnA, the overall EET efficiency decreases drastically presumably due to the increased conformational heterogeneities in longer Zn, The conformational heterogeneities especially in longer Zn are believed to provide nonradiative deactivation channels acting as fluorescence quenchers. The time scales for solvation as well as conformational dynamics can also be retrieved from various excitation wavelength and solvent dependent time-resolved transient absotption spectra.

Fig* 2. Cyclic porphyrin arrays : c-{Z2)6 (left), s-(Z2)5 (center), and s-(Z2)6 (right) Second, to mimic the wheel-like giant architecture of photosynthetic pigments (LH2 and LHl), the construction of cyclic porphyrin arrays has been attempted, which may aid the understanding of the ftmdamental mechanisms of EET in the natural photosynthetic antenna or find new applications as optoelectronic material. For this purpose, cyclic porphyrin wheels would be good candidates as a lightharvesting photosynthetic antenna. As one candidate, the cyclic porphyrin array where six Z2 units are covalently coupled via 1,3-phenylene linkage to form cyclic porphyrin arrays c-(Z2)6 has been prepared. Based on our time-resolved spectroscopic data, we obtained the EET time of 3,6 ps between the adjacent Z2 units in addition to <200 fs EET coupling time within Z2 unit as confirmed by the exciton coupling dynamics in Zn. As a supplementary route in the preparation of porphyrin arrays, a synthetic strategy utilizing a supramolecular chemistry has been envisaged, since it provides versatility in molecular networking. In this regard, the self-assembled cyclic porphyrin decamer s-(Z2)5 and dodecamers s-(Z2)6 where self-coordinated zinc(II)porphyrin dimers are connected together to form cyclic arrays. We also carried out time-resolved spectroscopic measurements such as fluorescence polarization anisotropy dynamics and exciton-exciton annihilation

454

processes to reveal the EET processes occurring in these two cyclic arrays. For this purpose; the exciton coupling dynamics in model compounds such as selfcoordinated dimer units has provided the basis for furthering the EET processes occurring in cyclic arrays where the self-assembled porphyrin dimer units serves as a building block element.

Fig, 3. Box porphyrin arrays: Bl (left), B2 (center), and B3 (right) To extend the idea of self-assembly into three-dimensional space to increase the efficiency in light-harvesting and exciton coupling, the porphyrin boxes with various sizes can be prepared by intermolecular coordination of the orthogonallylinked porphyrin dimer. Excitation energy migration processes within threedimensional zinc(II) porphyrin boxes with different sizes (Bl, 82, and B3) and their corresponding constituent units (orthogonally linked zinc(n) porphyrin dimers) are comparatively investigated by steady-state and time-resolved spectroscopic methods in conjunction with polarization anisotropy measurements. The steady-state absorption, fluorescence excitation anisotropy, aind time-resolved fluorescence anisotropy decay profile concomitantly reflect the EET process within the 2inc(II) porphyrin boxes (Bn). Both the pump-power dependence on the femtosecond transient absorption and the transient absorption anisotropy decay profiles are directly associated with the excitation energy migration process within the Bn boxes, where the exciton-exciton annihilation time and the polarization anisotropy rise time are well described in terms of the Forster-type incoherent energy hoppmg model by assuming N = 4 cyclic array that has the exciton coherence length of L ^ 2. Consequently, the excitation energy hopping times of 48, 96-100, and 356-365 ps are obtained for Bl, B2, and B3 that have different inter-chromophoric distances between energy donor and acceptor. Conclusively, the EET processes occurring in multiporphyrin arrays in various forms (linear, cyclic, box etc.) revealed by various time-resolved spectroscopic techniques are expected to give further insight into the EET phenomena in LHl and LH2 complexes and the application in molecular photonic wires and artificial light-harvesting apparatus.

References 1 D Kim and A. Oswka, Ace, Chem. Res. (2004), in press. 2 D. Kim and A. Osuka, J. Phys. Chem. A (a feature article) 107,8789-8816 (2003).

455

Energy transfer in phenylene ethynylene dendrimers Evrim Atas\ Chad Mair\ Joseph S. Melinger^, Zhonghua Peng^, and Valeria D. Kleiman^* ^Department of Chemistry, University of Florida, Gainesville, Florida 32611-7200 [email protected]. ^Naval Research Laboratory, Washington, DC 20375, ^University of Missouri-Kansas City, Missouri, 64110 Abstract. Ultrafast energy transfer in phenylene ethynylene dendrons has been investigated using both transient absorption and fluorescence up-conversion. The nature of electronic excitations and the mechanisms directing the excitation energy to the trap are discussed.

1. Introduction The ability to synthesize large conjugated macromolecules with light-harvesting properties and defined architectures allows a comprehensive investigation of electronic coupling [1], exciton formation [2, 3] and energy transfer. Previous experimental [3, 4] and theoretical studies [2] in multichromophore phenylene ethynylene (PE) based dendrimers indicate that electronic excitations in metaconjugated dendrimers are localized on the linear segments between branching points. Recently, Bardeen and Martinez postulated that while PE units are weakly coupled in their ground state, they can be strongly coupled in their relaxed geometry on the excited state [1]. Peng and co-workers have synthesized PE dendrimers including both para- and ortho-substitutions to create unsymmetrical geometries [5]. These dendrimers have rapidly growing conjugation lengths with increased generation providing broad absorption spectra, large molar absorptivities and high fluorescent quantum yields [6]. We present here ultrafast time-resolved spectroscopic results from an unsymmetric second generation didendron coupled in meta position to a phenylethynylperylene (PEP) unit (Fig. 1). The PEP unit, serving as an energy trap, is attached to the focal point of the dendrons to quantitatively probe energy transfer and excited state dynamics. pig, X. 2G2mPer dendrimer

2. Experiment Femtosecond up-conversion and transient absorption (TA) spectroscopy were employed to explore excited state dynamics and energy transfer. Experimental setups are described elsewhere [7]. Briefly, excitation pulses are generated in an

456

OP A, pumped by an amplified Ti-Sapphire laser (Spitfire, Spectra Physics, 1 kHz). Signal or idler OP A output are used to generate tunable excitation pulses in the 320-470 nm range. A small portion of the amplified beam (-30 |LJ) is spatially and temporally overlapped with the collected fluorescence in a P-BBO (0.5 mm) crystal generating a nonlinear response signal in the UV. Signal detected at X =302 nm corresponds to PEP fluorescence at 490 nm. TA experiments utilize the same excitation pulse, and a tunable probe generated by a second OPA. Pump fluences are kept below 15 jiJ/cm^ for -100 fs (FWHM) pulses to guarantee linear response. All measurements are done in rotating cells with 1mm optical path. The sample is dissolved in CH2CI2 and has an optical density of 0.3.

3. Results and Discussion The steady state absorption spectrum of 2G2mPer (Fig.2) reproduces the spectroscopic features of its individual components, 2G2mOH and PEP, suggesting only weak coupling between the backbone and the PEP trap. The broad absorption (300430 nm) in 2G2mPer corresponds to the backbone while absorption features at X >410 nm are assigned to the PEP. Emission from 2G2mPer arises almost entirely from the PEP t r a p w i t h OEnergy Transfer > 0 . 9 0 [ 5 ] .

Femtosecond fluorescence experiments follow the energy migration from the initially excited state to the PEP. Fig 3 shows the temporal evolution of the fluorescence as a function of excitation wavelength. The fits correspond to the convolution of exponential functions with the IRF. The top panel shows the experimental detection limit ( T < 1 5 0 fs) measured from the fluorescence response following direct excitation of the PEP trap at 465 nm. Longer rise times are attributed to excited states dynamics in the backbone, and energy transfer to the PEP. Excitation at 420 nm shows contribution from backbone-to-

0.0

500 400 X (nm)

300

600

Fig. 2. Abs. of 2G2mOH (-—), PEP ( ), and 2G2mPer ( and emission from 2G2mPer 0

1

2

3

rfilJ'j>ft^^!(i8igU!^48s3a''''^l»i»s]

(a). J^Kiai^iyiari^MftWd

ibL ^^

mtJ»i»if!Pfmii»fii?0'

i£L *»w^"i^

M.

P E P e n e r g y t r a n s f e r ( 1 5 % , Xgnergy transfer " 680

Fig. 3. Fluorescence

signal for

fs) as well as direct excitation of the PEP excitations at X (a) 465 nm, (b) 420 (85%, T-150 fs). Excitation at 380 nm and nm, (c) 380 nm, and (d) 340 nm. 340 nm show only energy transfer, since direct excitation of PEP is negligible. In this region, there is evidence of two channels (T 1 - 680 fs, 65% and x 2 ~ 300 fs, 35 %). The whole energy transfer process is completed in less than 2 ps, which makes this molecule one of the fastest light 457

harvesters. TA spectroscopy provides additional insight into the primary transfer mechanisms. Excitation-induced changes in the transmission allow simultaneous monitoring of the excited state population of the PE backbone and PEP moieties. Figure 4 shows the changes in probe transmission detected at 520 nm after excitation at 330 nm. The initial positive TA signal was found to be strongly dependent on ° ^ "^ ^ ® the relative pump/probe polarizations and it is p. ^ ^ ^ signal X ^pump i-^nr.L i l ps is assigned to stimulated emission from the PEP trap excited state since neither backbone nor PEP absorb at this X. The temporal behavior of the SE reflects the backbone-to-PEP energy transfer. The time constants obtained are consistent with the up-conversion experimental results. Excitation dependent dynamics can be attributed to at least two relaxation processes initiated in localized states within the PE backbone. The simplest interpretation of energy transfer is based on the very weak coupling limit. It assumes a pure Coulombic interaction and, for donor-acceptor separations much larger than the dipole sizes, the energy transfer can be evaluated within the pointdipole approximation (Forster). Under this model, the Xenergy transfer constants presented here correspond to very short donor-acceptor distances, comparable to the molecular dimensions. In addition, we estimate the donor-acceptor interactionenergy to be ~ 112 cm"^ which is at the upper limit of very weak coupling [8]. Thus, the validity of the Forster formalism for this system has to be questioned.

4. Conclusions Ultrafast dynamics of energy transfer in dendritic structures were studied revealing the presence of two localized states in the dendrimer backbone. From these states the excitation energy is transferred to the PEP trap in less than 2 ps, making this molecule one of the fastest light harvesters. Acknowledgements This work is supported by the NSF, CHE:0239120.

References 1 2 3 4 5 6 7 8

A. L. Thompson et al., J. Phys. Chem. A 108, 671-682 (2004). S. Tretiak et al., J. Phys. Chem. B 102, 3310-3315 (1998). R. Kopelman et al., Phys. Rev. Lett. 78, 1239-1242 (1997). V. D. Kleiman et al., J. Phys. Chem. B 105, 5595-5598 (2001). Y. C. Pan et al., J. Org. Chem. 68, 6952-6958 (2003). J. S. Melinger et al., J. Am. Chem. Soc. 124, 12002-12012 (2002). E. Atas et al., in preparation. B. Valeur, Molecular Fluorescence (Wiley-VCH 2002, Chap. 4)

458

A 40-fs Time-Resolved Absorption Study of CisStilbene in Solution: Observation of Coherent Nuclear Wavepacket Motion in Reactive Excited State Kimihiko Ishii, Satoshi Takeuchi and Tahei Tahara Molecular Spectroscopy Laboratory, RIKEN (The Institute of Physical and Chemical Research), 2-1 Hirosawa, Wako, Saitama 351-0198, Japan E-mail: [email protected] Abstract. Ultrafast photoisomerization reaction of c/5-stilbene in solution was studied by time-resolved absorption spectroscopy with 40-fs resolution. Rapidly damping coherence of a -220 cm'^ vibrational mode was observed in the reactive excited state. The obtained data showed that the initial wavepacket motion rapidly dephases owing to the significant anharmonicity of the Si potential surface, and the isomerization proceeds after intramolecular vibrational energy redistribution.

1.

Introduction

Cis-trans photoisomerization of cw-stilbene (Fig. 1) at the central C=C double bond is one of the most fundamental photochemical reactions. Since this reaction proceeds in a nearly barrierless way, c/^-stilbene shows very short Si lifetime (~1 ps[l]). This is m clear contrast to the reverse trans-cis photoisomerization (-100 ps). It is crucial to know the nuclear structure and its dynamics in the reactive excited state to elucidate the mechanism of such an ultrafast reaction. In order to elucidate the mechanism of such an ultrafast reaction, we studied this reaction by the time-resolved pump-probe absorption method using ultrashort light pulses.

V/ \ / Fig. 1. Chemical structure of cz^-stilbene

459

2.

Experimental Methods

An apparatus for UV-pump and visible-probe time-resolved absorption measurements was developed in our group and described in detail elsewhere[2,3]. Ultrashort light pulses for creating excited state molecules and for probing transient absorption changes were obtained from two home-made NOP As (noncoUinear optical parametric amplifiers) excited by a commercial Ti:sapphire regenerative amplifier system. The center wavelengths of the generated pulses were 315 nm (pump) and 660 nm (probe). The time resolution of the transient absorption experiment was evaluated to be about 40 fs by a cross-correlation measurement.

3.

Results and Discussion

We pumped c/^-stilbene at the red edge of the absorption spectrum (315 nm), and generated excited molecules in the Si state that is the precursor of the isomerization reaction. It is well known that the Si state of cz5'-stilbene shows a strong Sn <- Si transient absorption band in the 500 - 700 nm region. Fig. 2(a) depicts a time-resolved absorption signal at 660 nm obtained from a cyclohexane solution. In the picosecond time scale, the transient absorption shows a single exponential decay ( r = 1.25 ps) that corresponds to the population decrease of the Si state due to the isomerization reaction. 100 h (a)

%

(b)

8 c

U 50 1

V.

1

0.0 •

1

1

1

1

1

1

1

1

0.5

^

o < / ' ' I ' ' » « I « ' ' ' 1 ' ' ' ' I ' ' ' ' i « ' ' ' ' ' I ' ' 1 ' « ' « I ' '

0

1

2

3

Delay Time / ps Fig. 2. (a) Time-resolved absorption signal of c/5-stilbene in cyclohexane (b) Oscillatory component of the time-resolved absorption signal

460

In addition to this population component, we clearly observed a beating feature in the sub-picosecond region, which reflects a nuclear wavepacket motion in the Si state generated by the ultrashort pump pulse. Figure 2(b) shows the oscillatory component obtained from the time-resolved signal by subtracting the population component. A Fourier transform analysis exhibits that a broad spectral peak at -220 cm"^ predominantly contributes to the oscillatory signal. The peak frequency of this low-frequency band as well as its broad bandwidth is consistent with reported spontaneous Raman spectra of Si c/5-stilbene[4,5]. According to a vibrational analysis based on a CIS calculation of the Si state, this vibration has been assigned to a vibrational mode that involves the C=C torsion, C-phenyl torsion and C-phenyl in-plane bending motions[5,6]. The oscillatory component in Fig. 2(b) shows a quite fast dephasing ( r = 0.21 ps). This dephasing is much faster than typical vibrational dephasing time in solution at room-temperature. More importantly, it is significantly shorter than the time constant of the photoisomerization (1.25 ps). This implies that a fast vibrational dephasing takes place in Si cz5-stilbene, before isomerization proceeds. The dephasing and isomerization times show different solvent dependence: The dephasing time does not change significantly between the cyclohexane and methanol solutions, whereas the isomerization time becomes much shorter in methanol (0.46 ps) compared to the cyclohexane solution (1.25 ps). It shows that the observed wavepacket motion is not directly coupled with the reaction coordinate. The fast dephasing is probably caused by the efficient intramolecular vibrational energy redistribution (IVR) over other low-frequency vibrational modes. We concluded that the observed coherent nuclear motion of -220 cm"^ mode dephases rapidly, reflecting the anharmonicity of the Si potential surface, and that the cis-trans isomerization proceeds from a state after (partial) intramolecular vibrational energy redistribution.

References 1 2 3 4

D. H. Waldeck, Chem. Rev. 91, 415, 1991. S. Takeuchi and T. Tahara, Chem. Phys. Lett. 326, 430, 2000. S. Takeuchi and T. Tahara, J. Chem. Phys. 120, 4768, 2004. P. Matousek, A. W. Parker, D. Phillips, G. D. Scholes, W. T. Toner, and M. Towrie, Chem. Phys. Lett. 278, 56, 1997. 5 W. M. Kwok, C. Ma, D. Phillips, A. Beeby, T. B. Marder, R. L. Thomas, C. Tschuschke, G. Baranovic, P. Matousek, M. Towrie, and A. W. Parker, J. Raman Spectrosc. 34, 886, 2003. 6 G. Baranovic, J. Raman Spectrosc. 32, 293, 2001.

461

Monitoring an Ultrafast Photo-Isomerization by Femtosecond Fluorescence, Absorption, and I R Spectroscopy p . Gilch, B. Schmidt, C. Sobotta, M. Braun, F. Roller, T. Schrader, A. Sieg, W. Schreier, and W. Zinth Department fiir Physik, Ludwig-Maximilians-Universitat Miinchen, Oettingenstr. 67, D-80538 Miinchen, Germany E-Mail: [email protected] A b s t r a c t . Four distinct stages (motion on an excited state surface, internal conversion, vibrational cooling and slow structural re-arrangements) of the photoisomerization of an azobenzene are identified by means of femtosecond fluorescence, absorption, and IR techniques.

Photo-isomerizations are of both great practical and scientific importance (for a recent review see [1]). From a practical point of view molecules undergoing photo-isomerization allow for "on demand" switching of material properties. AppHcations range from optical storage devices to photo-chromic sun glasses. From the view point of molecular science photo-isomerizations are indispensable when it comes to the real time study of chemical processes. As these isomerizations can be triggered by a femtosecond laser pulse kinetic experiments on that time scale are feasible. The study (and understanding) of such processes is challenging since one deals with multi-dimensional motions on at least two potential energy surfaces (that of the ground and of one or more electronically excited states). Here, we demonstrate t h a t by combining diflPerent femtosecond techniques distinct stages of a photo-isomerization can be clearly identified. The experiments were performed on an azobenzene derivative, 4-nitro-4'-(dimethylamino)-azobenzene (NA, for structure see Figure 1) dissolved in toluene. NA as the parent compound azobenzene undergoes isomerization transforming the thermally stable trans isomer to the meta stable cis form [2]. In comparison to azobenzene NA has a higher extinction coefficient in the visible and contains a nitro group. The former property allows for a very sensitive detection of the processes on the excited state surface and a nitro group is a good IR marker for vibrational excitation [3]. The time dependent fluorescence emission of NA after excitation with NOPA pulses (480 nm) tuned to its absorption maximum was detected using a novel Kerr based fluorescence spectrometer [4]. Immediately after the excitation a spectrally narrow and intense emission is observed (Figure l a ) . This emission decays within ~ 100 fs and gives way to a weaker and much broader

462

Fig. 1. Spectral changes in the visible occurring during the photoisomerization of NA. (a) Kerr gated fluorescence spectra, (b) Transient absorption spectra. emission in turn decaying in ~ 1 ps. Transient absorption spectra recorded under virtually identical excitation conditions (Figure lb) at the first glance lack the 100 fs contribution. The transient spectra consist of a ground state bleach contribution around 500 nm and an excited state absorption extending to the NIR. The dominant decay time of both features is ~ 1 ps, in addition a weaker contribution with a longer characteristic time and an offset reflecting the formation of the cis isomer is observed. A global independent analysis of the data reveals that in both experiments processes with a 100 fs and a 0.8 ps time constant contribute to the signal. (That bi-phasic behavior is similar to that observed for azobenzene [5].) In the absorption experiment the 100 fs process is associated with a delayed rise. An additional time constant of 4 ps only present in the absorption experiment carries a sigmoidal spectral signature. This comparison of fluorescence and absorption experiments allows to disentangle processes occurring on the excited state surface from those on the ground state surface. The 100 fs process does not deplete the excited state and can be associated with a motion on the excited state surface out of the Franck Condon region. The 0.8 ps terminates the emission and recovers a greater part of the ground state bleach. It can, therefore, be safely assigned to the internal conversion between the two states. Finally, the molecule transfers the vibrational excitation gained in the internal conversion to its solvent surrounding, a process that this responsible for the 4 ps feature. As the cooling occurs in the ground state, this feature is absent in the fluorescence experiment. More detailed information on the dynamics in the ground state can be derived from femtosecond IR experiments. The response of the IR spectrum on the optical excitation (407 nm) was recorded in three spectral regions. Here, we will focus on the behavior of symmetric NO2 stretch vibration located at 1340 cm"-^ (Figure 2). For positive delay times a strong bleach of the NO2 stretch resonance is observed. This bleach is accompanied by an induced absorption adjacent to the low frequency side of the bleach. The decay of the induced absorption goes along with a partial recovery of the bleach and has

463

1300 " 1350 S1400 2 0

10 20 30 40 50 Delay time [ps]

Fig. 2. Spectral changes in the mid IR occurring during the photo-isomerization of NA. Lower IR absorption (stronger bleach) is presented by darker shading. a characteristic time of '^ 5 ps. An additional longer component of ^ 50 ps recovers a substantial part of the bleach. Two of the IR features are easily assigned. The residual bleach at long delay times is due to the cis formation — in the cis form the oscillator strength of the NO2 mode is smaller than in the trans from. The induced IR absorption relaxing with a characteristic time of 5 ps is caused by a red-shift of the NO2 vibration. This shift is induced by excited low frequency vibrations which couple anharmonically to the NO2 mode. As these vibrations cool down the shift relaxes. The time scale of this process is completely in line with the transient absorption experiment in the visible range. Astonishingly, for later times there are still strong dynamics in the IR. The spectral signature associated with these dynamics is very similar to the IR difference spectrum of the two isomers. Thus, it looks as if a species with an IR spectrum close to that of the cis form decays within ~ 50 ps partially forming the cis isomer and partially replenishing the trans from. The underlying process might be a slow structural re-arrangement required to transform the geometry of NA after internal conversion to that of the two stable isomers. Whatever the origin of these slow IR transients is, these experiments clearly demonstrate that the story of the photo-isomerization does not stop at the time visible spectroscopy tells us that it does.

References 1. 2. 3. 4. 5.

N. Tamai and H. Miyasaka, in Chem. Rev., Vol.lOO, 1875, 2000. K. Gille, H. Knoll, and K. Quitzsch, in Int. J. Chem. Kinet, Vol.31, 337, 1999. T. Schrader et al, in Chem. Phys. Lett, Vol.392, 358, 2004. B. Schmidt et al., in Appl. Phys. B, Vol.76, 809, 2003. Y.-C. Lu, C.W. Chang, and E. W.-G. Diau, in J. Chin. Chem. Soc. Taip., VoL49, 693, 2002.

464

From ultrafast spectroscopy to bidirectional molecular switches: DHA / VHF U. Schmidhammer^ V. De Waele^ G. Buntinx^ and E. Riedle^ ^LS fur BioMolekulare Optik, LMU Munchen, Oettingenstr. 67, 80538 Mtinchen, Germany ^LASIR (UMR 8516), CERLA (FR 2416), Bat C5, 59655 Villeneuve d'Ascq Cedex, France Abstract: The photoconversion from dihydroazulene (DHA) to vinylheptafulvene (VHF) is governed by two mechanisms: The ring opening proceeds on the excited energy surface on the picosecond time scale. It is followed by an internal conversion to the VHF ground state that is accelerated by the presence of a conical intersection in the case of cyclopenta-DHA. This conical intersection hinders the photoinduced back reaction from the final VHF products. However, we successfully photoconverted the cyanophenyl-VHF-cis back to the DHA with a second delayed pulse. This opens the route to the development of bistable DHAs.

1. Investigations on the photochromic reaction paths The realization of logic functions at the molecular level is an exciting and active area. Tow^ard this end, the photochromic compounds constitute an important link to convert light into chemical energy or information. The dihydroazulenes (DHA) are very established compounds for molecular sv^itching [1]. They shov^ good efficiency of the photoconversion and fatigue resistance. On the other hand, the switching of DHA can only be photoinduced in one direction: Under irradiation in the near UV the molecule undergoes an efficient ring opening reaction based on a IOTC electrocyclization and is converted to the vinylheptafulvene isomer (VHF). The back-reaction proceeds by thermal activation. This behavior can restrain the application of DHA and it is important to elucidate the parameters that govern the photochromism and are responsible for the absence of optical bistability. -CN-DHA

CN-DHA

CN-VHF-cis

CN-VHF-trans

300

400

500 ^(nm) 600

700

Fig. 1. Species involved in the photochromism of CN-DHA and corresponding spectra. The photochemical process was triggered with laser pulses at 375 nm.

465

We study the DHAA^HF photoconversion by broadband transient absorption with 100 fs pulses and two color experiments with sub-30 fs pulses. The former provides an identification of the transient states, while the latter allows us to determine the kinetics. Through the analysis of the coherent signal observed in addition we are able to identify the structural evolution of DHA directly after the application of the ultrashort pump pulse. We compare the dynamics of the l,l(8aH)-azulendicarbonitrile,2-(4-cyanophenyl) derivative (CN-DHA/CN-VHF) with l,2,3,8a,9-pentahydro-cyclopent[a]azulene-9,9-dicarbonitril (CP-DHA). The final product after excitation of CN-DHA to the Si state is the CN-VHF-trans isomer. The complete process involves a cistrans isomerization besides the ring opening (Fig. 1). The transient spectra reveal that an absorption band around 510 nm is formed from an excited state within the first 15 ps. This ground state absorption is red shifted by 30 nm compared to CNVHF-trans and is attributed to the So state of CN-VHF-cis. The final trans conformer is formed within 10 jis as shown by a flash photolysis experiment. The ring opening itself is an ultrafast process in the Si state separated from the cis-trans isomerization [2]. The kinetics are the same for all probe wavelengths (485 to 690 nm): A structural relaxation within 100 fs is followed by the 1.2 ps ring opening. Strong signal oscillations appear immediately after the pulse excitation which we assign to a coherent vibronic wavepacket oscillating around the DHA Si state minimum. Four normal modes at 150, 190, 330 and 500 cm"^ were identified which contribute towards the planarisation of the azulene and thus support the ring opening. The conversion of CP-DHA to CP-VHF is described theoretically [3] with the same two reactive coordinates as found by our investigations of CN-DHA: ring opening and internal conversion. Nevertheless, the behavior of CP-DHA is found to be distinctly different (Fig. 2, [4]). The complete process proceeds via a single exponential decay of 600 fs. Despite identical experimental conditions as for CNDHA, the ring opening dynamics can not be distinguished from the internal conversion and moreover no wavepacket oscillations are observed. This reveals the direct formation of the CP-VHF Si state on the time scale of the pump pulse. CN-DHA -VHF-cis*

time (ps)

Fig. 2. Primary dynamics of CN-DHA and CP-DHA excited at 350 nm and probed at 630 nm and the deduced model of the Si energy surfaces.

466

The wavepacket excited at the CP-DHA Franck-Condon point, which is strongly different from the CP-VHF one, dephases immediately. The ultrafast internal conversion to the CP-VHF ground state is governed by a conical intersection as predicted by [3].

2. Photoinduced ring closure from the transient CN-VHF-cis So The strong interaction between the excited state and the ground state via the conical intersection hinders the optical bistability of CP-DHA. In contrast, CN-DHA shows a much smaller efficiency of the internal conversion indicating that no conical intersection is involved. We can therefore attempt to photo-induce a backreaction from the transient CN-VHF-cis ground state. We irradiated two identical samples of CN-DHA [5]: In an one-pulseexperiment the photoconversion was triggered under similar conditions as in the two color measurements. The irradiation time was chosen such that a significant amount of CN-VHF-trans was obtained. In a two-pulse-experiment an additional pulse at 530 nm with a 25 ps delay excited the CN-VHF-cis. After the irradiation much less CN-VHF-trans is found in the two-pulse-experiment than in the onepulse-experiment (Fig. 3). We conclude that a significant amount of CN-DHA was regenerated from the transient species by the second pulse. This experiment demonstrates that when manipulated by femtosecond pulses, the CN-DHA possesses the remarkable property of multimode switching: a thermal switch between the VHF-cis and DHA via the VHF-trans conformation and an ultrafast photoreversible switch between the DHA and VHF-cis conformers. 2 - pulse experiment

- pulse experiment

Q O

/ \ ^

0.5-1 /

0.4

\

unexposed

unexposed

0.5 0.4 0.3-1 0.2 0.1-1 0.0

340nm

0.3 0.2 0.1-1 0.0 200

300

400

500

600 >.(nm)

200

530nm

300

400

500

600

700

Fig. 3. Cw absorption spectra recorded before and after the one pulse experiment (left hand side) respectively the two pulse experiment (right hand side).

References 1 T. Mrozek, H. Gomer and J. Daub, Chem. Eur. J. 7, 1028 (2001). 2 V. De Waele, M. Beutter, U. Schmidhammer, E. Riedle, and J. Daub, CPL 390, 328 (2004). 3 M. Baggio-Pasqua, et al., J. Am. Chem. Soc. 124, 1456 (2002). 4 J. Em, et al., Chem. Phys. 259, 331 (2000). 5 V. De Waele, U. Schmidhammer, T. Mrozek, J. Daub and E. Riedle, J. Am. Chem. Soc. 124, 2438 (2002).

467

Ultrafast Intramolecular Electron Transfer of 9^9'-Bianthryl as Studied by Femtosecond TimeResolved Near-Infrared Absorption and Anisotropy in the 950-1500 nm Region Tomohisa Takaya\ Koichi Iwata^, Hiro-o Hamaguchi\ and Haruo Kuroda^ ^ Department of Chemistry and ^ Research Centre for Spectrochemistry, School of Science, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033, Japan E-mail: [email protected] ^ IR-FEL Research Center, Research Institute for Science and Technology, Tokyo University of Science, Yamazaki 2641, Noda, Chiba 278-8510, Japan Abstract. Femtosecond near-infrared absorption and its anisotropy of 9,9'-bianthryl in the 950-1500 nm region reveal that the locally-excited (LE) state is directly converted to the charge-transfer (CT) state in 0.3 ps.

1.

Introduction

Photoinduced intramolecular electron transfer reactions are strongly affected by the solvent. In the case of 9,9'-bianthryl (BA), the intramolecular electron transfer takes place in polar solvents when the molecule is photoexcited with the UV light, while such a reaction does not proceed in non-polar solvents. The reaction dynamics has been studied by time-resolved fluorescence spectroscopy [1,2] and by time-resolved UV/visible spectroscopy [3,4]. We develop a femtosecond near-infrared absorption spectrometer as a powerful tool for investigating the photoinduced electron transfer process. In the nearinfrared region, loosely bound electrons in CT complexes and ion pairs show broad and structureless absorption bands. When time-resolved near-infrared spectroscopy is applied to BA, its CT band is expected to be observed in this spectral region. In addition, its LE band is expected to be observed as well, because anthracene has a transient absorption band in the near-infrared.

2.

Results and Discussion

Time-resolved near-infrared absorption spectra. Time-resolved near-infrared absorption spectra of photoexcited BA were measured in a non-polar solvent heptane. The results are shown in Fig. la. An absorption band that has a peak at 1030 nm appears within the instrumental response. Its position and spectral shape resembles those of the transient near-infrared absorption band of Si anthracene,

468

which forms a monomer unit of BA. The BA band is therefore assigned to a locally excited (LE) state band.

1000

1100

1200 1300 1400 Wavelength / nm

1500

1000

1100

1200 1300 1400 Wavelength / nm

1500

Fig. 1. Time-resolved near-infrared absorption spectra of 9,9'-bianthryl in heptane (a) and in acetonitrile (b). Concentration of each solution is 5 x lO'"^ mol dm'"^. The spectral shape of the LE band changes slightly between 0 ps and 2 ps. The absorption decreases in 950-1150 nm, whereas the absorption increases in 12501500 nm. The decay and the rise take place simultaneously with a temporal isosbestic point at 1180 nm. We suggest that the spectral change in heptane is caused by internal rotation around the central C-C bond of B A in the LE state. Time-resolved near-infrared absorption spectra measured in a polar solvent acetonitrile reveal different dynamics. The results are shown in Fig. lb. A transient absorption band centered at 1020 nm is observed at 0 ps. It is assigned to the LE band, because the band is almost identical with the LE band observed in heptane. At time delays later than 1 ps, the LE band disappears and a structureless absorption band centered at around 1250 nm appears. It is possible to assign this band as a charge transfer (CT) state band. The rise behavior of the band, however, is not clear because of the spectral overlap with the LE band over the whole wavelength region from 950 to 1500 nm. We removed the contribution of the LE band by using the transient absorption spectrum in heptane at the time delay of 0 ps, which is considered as the pure LE band. After the subtraction, the broad and structureless CT band is successfully retrieved at early time delays at which the LE band is originally observed. By fitting with a single exponential function, the decay constant of the LE band is determined to be (0.38±0.01) ps, while the rise constant of the CT band is determined to be (0.3±0.1) ps. The two time constants are well in agreement with each other. We conclude that no reaction intermediate that has a sub-picosecond or longer lifetime exists in the electron transfer process of B A in acetonitrile. Time-resolved near-infrared absorption anisotropy. Time dependence of the absorption anisotropy was measured for the spectral region of 950-1400 nm. The obtained "time-resolved near-infrared absorption anisotropy spectra" are shown in Fig. 2. In heptane (Fig. 2a), the anisotropy value at 0 ps is -0.2 around the peak of the LE band. It increases to -0.1 in 3 ps. The anisotropy change shows a similar kinetics with the spectral change shown in Fig. la. The anisotropy change can be therefore interpreted as the conformational change. In acetonitrile (Fig. 2b), the anisotropy value changes from -0.2 to 0.3 in 3 ps at around the peak of the LE band. The change proceeds slower than the charge transfer.

469

1000

1100 1200 1300 Wavelength / nm

1400

1000

1100 1200 1300 Wavelength / nm

1400

Fig. 2. Time-resolved near-infrared absorption anisotropy spectra of 9,9'-bianthryl in heptane (a) and in acetonitrile (b). Anisotropy r is described by r = (3cos^ 0-\)l5 , where 6 represents the angle between the transition dipole moments of the photoexcitation and the transient absorption. At 0 ps, 6 is 90^ both in heptane and in acetonitrile because r is -0.2 for both of them. This result is consistent with the electronic structure of anthracene. Its UV absorption at 370 nm is along the short axis of the molecule, while its transient absorption in the near-infrared is along the long axis. It is clear that only one of the anthracene moieties of BA is photoexcited. Because there is no inter-ring interaction observed, two anthracene rings should be perpendicular to each other both in the ground state and in the LE state at 0 ps. In heptane, 0 becomes 66^ at 3 ps. The decrease of 0 indicates that a transition dipole component along the central C-C bond of BA appears due to the interaction between the two anthracene rings. The internal rotation makes the interaction possible. In acetonitrile, 6 becomes 24° at 3 ps. The transition dipole moment at 3 ps has a direction close to the central C-C bond. This result suggests the presence of strong inter-ring interaction in the CT state. The transition is associated with the electron rearrangement between the anthracene rings. Acknowledgements. This work is supported by Grant-in-Aid's for Creative Scientific Research (No. IINPOIOI) and Scientific Research (B) (No. 15350005) from Japan Society for the Promotion of Science, and Grant-in-Aid for Scientific Research on Priority Areas (Area 417, No. 15033219) from the Ministry of Education, Culture, Sports, Science and Technology (MEXT) of the Japanese Government. KI is a recipient of research grants from The Morino Foundation and The Kurata Memorial Hitachi Science and Technology Foundation.

References 1 T. J. Kang, M. A. Kahlow, D. Giser, S. Swallen, V. Nagarajan, W. Jarzeba, and P. F. Barbara, J. Phys. Chem. 92, 6800, 1988. 2 P. F. Barbara and W. Jarzeba, Advan. Photochem. 15, 1, 1990. 3 N. Nakashima, M. Murakawa, and N. Mataga, Bull. Chem. Soc. Jpn. 49, 854, 1976. 4 N. Mataga, S. Nishikawa, and T. Okada, Chem. Phys. Lett. 257, 327, 1996.

470

Real-time spectroscopy of charge-transfer excitation in phthalocyanine tin dichloride Masakatsu Hirasawa, Yuzo Sakazaki, Hiroki Hane, and Takayoshi Kobayashi Department of Physics, Graduate School of Science, The University of Tok>o, 7-3-1 Hongo, Bunkyo, Tokyo 113-0033, Japan E-mail: [email protected] Abstract* Charge-transfer (CT)-excited state in phthalocyanine tin dichloride is observed by real-time vibrational spectroscopy with a 6-fs pulse laser. The main absorption (Q-band) has no obvious CT character in the primary stage of the photocarrier generation.

!• Introduction Phthalocyanines (Pc's) are important material for photo-detectors in industrial. The mechanism of carrier generation in Pc's is related to charge transfer (CT)excited states, A model of photocarrier generation in Pc was proposed in the late 1970s [1], The model called "two-step process" explains the mechanism: 1) A singlet ii-^*-excited state in a molecule is generated by light. 2) The excited state relaxes non-radiatively to the CT-excited state. The CT-excited state is then ionized by an applied electric field to generate carriers. Recently, a novel model [2] has been proposed fi-om electro-absorption experiments, which claims that the directly photo-induced CT-excited state plays a main role in generating photocarriers. Although electro-absorption shows an intense signal in the lowest excited state (Q-band), it does not necessarily mean that the photocarrier yield is efficient there. To clarify this point, we studied a CT-excited state by real-time spectroscopy of vibration.

2. Experimental A thin film sample of phthalocyanine tin (IV) dichloride (SnPc) was prepared by evaporation in a vacuum (1.6 x 10'^ Torr) on a glass substrate. The pump-probe measurements were carried out by a 6-fs non-coUinear optical parametric amplifier (pump intensity: 1.6 x 10^'^ photons/cm^) and a combined system of a spectrometer and a 128-channel DSP lock-in amplifier with 3-nm resolution.

3* Results and Discussion The absorption spectrum in Fig. 1 shows two remarkable peaks, which are referred to as the Qx and Qy bands at 15000 and 13200 cm'\ respectively,

471

originating from the singlet TC-TC* transition in the Pc macro-cycle. To deconvolute the absorption bands, nonlinear least-squares method is applied assuming singlemode Fanck-Condon line-shapes. The absorption could be well described with a wave number of the 0-0 line being located at 13200 cm"^ an inhomogeneous width of 870 cm'^ (FWHM), and the Huang-Rhys factor of L40 for the Qy band, (15000 cm"^ 780 cm'^and L54 for the Qx band). Here we used a common vibrational mode of 660 cm"\ which is close to the value reported as a helper mode (670 cm"^) in the Raman excitation profile (REP) of CuPc. The Huang-Rhys factors estimated from the fitting are also consistent with the REP [3]. The remainder of the spectrum is plotted with a thin solid curve, which can be attributed to the CTexcited state. The lower absorption peak of 16700 cm'^ for CT excitation in Fig. 1 agrees well with the previous report [4]. The higher peak at 18200 cm"^ can be attributed to a 0-1 side band of the intramolecular vibration. Figure 2(a) shows contour maps of the Fourier amplitudes of normalized transmittance change. Two spectral regions are shown to be divided clearly by the Fourier amplitude pattern; the high-energy region A ranges from 560 to 600 nm and the low-energy region B from 600 to 660 nm. For region B only two modes of 670 and 1340 cm"^ are observed, while strong low-frequency modes appear in region A at 120, 190, 390 cm"^ and a few more. The two regions A and B are well overlapped with the CT band and the singlet 7t-;r* excitation (Q-band), respectively. The 190- and 670-cm"^ modes have been assigned using the polarization dependence of Raman spectroscopy to the lattice vibration of the molecular crystal [5] and the deformation mode of the macro-cycle in a Pc molecule, respectively. The difference in the Fourier amplitude in regions A and B can be understood as follows: CT excitation between a pair of neighboring molecules displaces them from the equilibrium positions perpendicular to the molecular plane. The displacement stimulates the lattice vibration of the one-dimensional Pc columns. The Q-band excitation modifies only the charge distribution around the macrocycle in a molecule and causes the deformation of the macro-cycle. Thus, the CT character of the electronic excitation states in region A are well confirmed. For region B, however, the Q band shows no CT character in the vibrational spectrum obtained, while electro-reflection or electro-absorption spectroscopy has demonstrated large signals in previous report [2]. Our results do not support the direct excitation model but the two-step process for photocarrier generation in the Q-band absorption. To describe the excitation mechanism in detail, we illustrate the Fourier amplitude for the 190- and 670-cm'^ modes in Fig. 2(b). The deconvoluted absorption spectra for the CT-band and the Qx band in Fig. 1 are also shown. For 190-cm'^ modes, the spectral dependence of the amplitude agrees well with the CT absorption spectrum. The result suggests that the mode is an electronic groundstate vibration of lattice caused by the CT-band resonance. The amplitude of the 670 cm"^ mode also corresponds well to the absorption of the Qx band, which means the ground-state molecular vibration. In conclusion we have manifested the difference of the electronic excitation between the CT band and the Q band by real-time vibrational spectroscopy. For the CT band, intense low-frequency modes appear in 120, 190, and 390 cm"^ and intramolecular modes are very week, while for the Q band only an intramolecular

472

mode of 670 cm'^and its overtone show up. This finding demonstrates that the CT character of the electronic excitation causes the intermolecular vibration. The finding also shows that the electronic excitation for the Q band has no CT character. Our experimental results suggest that the two-step process is dominant for photocarrier generation in the region of Q-band absorption.

wave number {X10* cm'^)

Fig, L Absorption spectrum and deconvoluted spectra plotted on a logarithmic scale, QY (dashed curves) and Qx bands (dotted ciives) arefittedwith a linear combination of die Gaussian functions, respectively Wave number (X icf cm'^) 15J

16

IBS

17

17,S 1$

1^,0

le.S

17.0

17.5

Wavenumber (x 10* cm') WavelengJft
Fig. 2. (a) Contour map of the Fourier amplitude of vibration in the nomialized transmittance difference AT/T. The Fourier amplitude is indicated on the logarithmic scale witli a step of two. (b) Fourier amplitude spectra of the vibrational modes of 190 (solid squares) and 670 cm"^ (open circles).

References Z,D. Popovic, Chem. Phys. 86, 311-32 L 1984. T. Saito, W. Sisk, T. Kobayashi, S. Suzuki, T. Iwayanagi, J. Phys. Chem. 97 , 8026, 1993. K. Yamasaki, O. Okada, K. Inami, K. Oka, M. Kotani, H. Yamada, J. Phys. Chem.

B 101, 13, 1997. A.J. Bovill, A.A. McConnell J.A. Nimmo, W.E. Smith, J. Phys. Chem. 90, 569575, 1986 H. Yoshida, Y. Tokura, T. Koda, Chem. Phys. 109, 375,1986. T.V. Basova, B.A. Kolesov, J. Struct Chem. 41, 770-777, 2000.

473

Coherent nuclear dynamics coupled with electron transfer reaction in porphyrin-ferrocence dyads S. Nakashima, M. Kubo, M. Otani, M. Murakami, Y. Ishibashi, M. Yasuda and H. Miyasaka Department of Materials Engineering Science, Graduate School of Engineering Science, Osaka University, Toyonaka 560-8531, Japan satoru@chem. es. osaka-u. ac.jp

Y. Mori, H. Imahori Department of Molecular Engineering, Graduate School of Engineering Science, Kyoto University, Kyoto 615-8510, Japan

Abstract: We report coherent nuclear dynamics coupled with photoinduced electron transfer using a newly designed donor-acceptor system with a strong electronic coupling. Our results by femtosecond spectroscopy show that the ultrafast electron transfer (~ 110 fs from Si and -60 fs from S2) was independent of solvent properties. Moreover, the electron transfer was accompanied by temporal phase shift of the selective porphyrin vibrational modes. The reaction coordinate for ultrafast electron transfer was shown to involve coherent intramolecular motions of the selective modes rather than stochastic solvation dynamics. 1. Introduction Photoinduced electron transfer (ET) reactions have been extensively studied to elucidate the nature of the intrinsic factors that determine chemical reaction rates.[l] It was revealed that photoinduced ET reactions are regulated by various parameters such as the free energy gap between the initial and final states, the magnitude of the electronic coupling (F^/) between the electron donor (D) and acceptor (A), the solvation dynamics around the system, and the intramolecular vibrational modes of the D and A. When the V^i is small, the nonadiabatic mechanism can be applied to explain the reaction rate constant using well-understood conventional theories. Recently, intramolecualr high-frequency vibrations are understood to be essential factors in the ultrafast electron transfer reaction.[2-4] Such quantum effects may be more dominant for ultrafast ET reactions in D-A liked molecules with stronger interactions.[5] However, the difficult combination of the well-designed synthesis of D-A linked molecules and availability of ultrashort 474

laser pulse system has precluded experimental exploitation of ultrafast ET reactions faster than a time constant of ca. 100 fs Here, we report the photoinduced ET of several dyads where porphyrin linked with ferrocene (XP-Fc, X=H2 or Zn Fig. 1) observed by fluorescence up-conversion (UC) and pump-probe (PP) measurements. Photoexcitation of this dyad into the porphyrin S2 state leads to ET from ferrocene to excited porphyrin. In those cases, ET took place ~ 110 fs from Si and -60 fs from S2. Furthermore, the phase shift of the selective porphyrin vibrational modes accompanied with the ET. The reaction coordinate for ultrafast electron transfer will be shown to involve coherent intramolecular motions of the selective modes.

Fig.l Compounds used in the experiments. BPZnP-Fc (M=Zn, Ar = 3,5-di-tert-butylphenyl) FPZnP-Fc (M=Zn, Ar = pentafluorophenyl) BPH2P-FC (M=H2, Ar = 3,5-di-tert-butylphenyl) FPH2P-FC (M=H2, Ar = pentafluorophenyl)

2. Experimental The details of the laser system and UC measurements are described elsewhere.[7,8] Briefly, for PP measurements the output of the second harmonics of the cavity-dumped TiiSapphire laser, centered at 424 nm and having a 200 kHz repetition rate, was divided into pump and probe beams. Pulse energy for pump and probe beams were 600 pJ and 70 pJ, respectively. The pump pulse passes through the delay line and both beams were focused into the sample using a 10-cm focal lens. The autocorrelation trace gave -35 fs fwhm at the sample point. The sample was placed into the custom-made rotational cuvette. This cuvette has a columned structure having a 1 mm path length of the sample sandwiched between two 25 mm diameter windows of 1 mm in width. The cuvette rotated at 2000 rpm in order to avoid photodamage to the sample. The probe beam was detected directly by a Si photodiode, and the signal was analyzed using a lock-in amplifier.

3. Results and Discussion The porphyrin Soret band of XP-Fc (its shape and relative intensity to the Q-band) was similar to that of the reference XP, which indicates that the porphyrin S2 character is little affected by ferrocene. On the other hand, the peak of the Q-band

475

was red-shifted; and the shape of it was broadened by ferrocene. The time-resolved fluorescence emission was measured in benzene by UC method. For example, the S2 lifetimes of the reference compounds, FPZnP, were 460 fs, and the S2 lifetimes of FPZnP-Fc was reduced to 82 fs by opening the path to charge separate (CS) state. We also examined time-resolved transient absorption spectra of FPZnP-Fc after excitation at 532 nm. The spectra showed an intense band in the region 430-454 nm and a weak, broad band at 705 nm, which are characteristic bands of zinc porphyrin anion radicals. Therefore, ET could be confirmed to take place fi'om porphyrin Si and S2 excited state to ferrocene cation. The dynamics was also measured in tetrahydrofran (THF) and benzene by PP method (Fig. 2). The PP signals of XP-Fc in THF showed the common dynamics:

I 0.0

I 0.5

I I I 1.0 1.5 2.0 delay time / ps

1.0 -

iIV

T 2.5

3.0

1

0.8-

\

0.6-

0.0

^w^M^

»iwr

1 . 1 1 1 3.0 1.0 2.0

04 — 0.2 0 0-

J

1

1

0.0

0.5

1

1

1

1

1

1.0 1.5 2.0 2.5 3.0 delay time / ps Fig. 2 Pump-probe signal of (a) reference BPZn and (b) BPZnP-Fc. Insets are their oscillatory components. a rise component with a time constant of ~ 60-70 fs, a decay component with a time constant o f - 160 fs, a decay component with a time constant of several picoseconds, and a slower component. The results were consistent with those of the fluorescence data. From these result ET rates were estimated and The Marcus parabola was simulated by the semi-quantum Marcus theory. (Fig. 3) Although the 476

ET rates were similar among the three dyads, the energy gaps were different by - 0 . 4 eV. As seen in Figure 3, the ET rates could not be fitted by the semi-quantum Marcus theory even if an electronic coupling was assumed to be considerably large (~ 60 meV), which is by far out of nonadiabatic approximation. Taken together with the results showing that the ET rates were independent of the solvent, we conclude that the energy-gap law is not valid any more for ultrafast photoinduced ET of porphyrin-ferrocene dyads with the strong electronic coupling. 14-i

I2J

In I0J

^ sJ O

eJ 4J T

\

-0.5 0.0

1

1

1

0.5 1.0 1.5 -AGcs/eV

1

\

2.0

2.5

Fig. 3 Marcus plots of various dyads Oscillatory components were clearly seen in the PP signals of FPH2P and FPH2P-FC. To analyze the vibrational coherency dynamics, we made Fourier transform of the oscillatory component data of FPH2P-FC during shifting of the data starting points from 50 fs to 1 ps. In the Fourier transform spectra, four band peaks appeared at 140, 200, 250, and 330 c m \ We here found that the phase of the three modes below 300 cm'^ shifted by ~ 2.57c with respect to that of the 330 cm'^ mode during ET. The phase shift between the modes during ET reflects the difference in overlapping between their vibrational coordinates and the reaction coordinate. ET requires shortening of the porphyrin-ferrocene distance and rotation of ferrocene relative to the porphyrin plane to maximize the orbital overlap between porphyrin and ferrocene. Therefore, the ferrocene translational and pyrrole-ring swivel modes are reasonable candidates for ET-promoting modes, rather than the pyrrole-ring breathing mode.

4. Summary In the present study, we designed directly linked porphyrin-ferrocene dyads and monitored their photoinduced ET by femtosecond spectroscopy. The ET rates of these dyads were much faster than the solvation time, and were independent of the

477

solvent properties. Because of the strong interaction between D and A the energy-gap law failed to describe the ultrafast adiabatic ET of porphyrin-ferrocene dyads. Such strongly adiabatic ET was shown to coupled with coherent vibrational motions of the selective porphyrin modes. Phase dynamics of vibrational coherency can be used to discriminate the reaction-promoting modes from others. These finding will give us new insight into the fundamental concept of the chemical reaction. 5. References [1] Marcus, R. A., Sutin, N. "Electron transfers in chemistry and biology." Biochim. Biophys. Acta 811,265-322(1985). [2] Sumi, H., Marcus, R. A. "Dynamical effects in electron transfer reactions." J. Chem. Phys. 84, 4894-4914(1986). [3] Bagchi, B., Gayathri, N. "Interplay between ultrafast polar solvation and vibrational dynamics in electron transfer reactions: role of high-frequency vibrational modes." Adv. Chem. Phys. 107, 1-80 (1999). [4] Mataga, N. et al. "First unequivocal observation of the whole bell-shaped energy gap law in intramolecular charge separation from S2 excited state of directly linked porphyrin-imide dyads and its solvent-polarity dependencies." J. Am. Chem. Soc. 123, 12422-12423 (2001) [5] Michael Thoss , Wolfgang Domcke, Haobin Wang "Theoretical study of vibrational wave-packet dynamics in electron-transfer systems" Chemical Physics 296 217-229 (2004) [6] Nakashima, S. et al. "Coherent dynamics in ultrafast charge-transfer reaction of plastocyanin" Chem. Phys. Lett. 331, 396-402 (2000). [7] Bhasikuttan, A. C , Suzuki, M., Nakashima, S., Okada, T. "Ultrafast fluorescence detection in Tris(2,2'-bipyridine)ruthenium(II) complex in solution: relaxation dynamics involving higher excited states." J. Am. Chem. Soc. 124, 8398-8405 (2002). [8] Kubo, M. et al "Ultrafast photoinduced adiabatic electron transfer of porphyrin-ferrocene dyads" submitted.

478

Subpicosecond Pulse Radiolysis Study on Geminate Ion Recombination Process in n-Dodecane Yoichi Yoshida, Akinori Saeki, Takahiro Kozawa, Jinfeng Yang, Seiichi Tagawa

The Institute of Scientific and Industrial Research, Osaka University, Mihogaoka 8-1, Ibaraki, Osaka 567-0047, Japan E"mail* [email protected] Abstract. Subpicosecond pulse radiolysis system was developed by using subpicosecond electron pulse from L-band linear accelerator and femtosecond laser pulse. Timing-jitter compensation system and double beam method were used to get high time-resolution and high reliability. The geminate ion recombination in n-dodecane was investigated by using the subpicosecond pulse radiolysis. The geminate decay of cation radical after 50 ps can be explained by the diffusion theory. However, the difference between the experimental result and the theory was observed within 50 ps.

1.

Introduction

The geminate ion recombination [1-3] has been studied as an important reaction in the primary process of radiation chemistry. The initial distributions and geminate decay were observed by using picosecond pulse radiolysis, and the recombination process was analyzed by the Smoluchowski equation base on the diffusion theory. However, only tail part of the geminate decay could be observed in the picosecond pulse radiolysis, because its time-scale is several picoseconds. Picosecond pulse radiolysis is one of the promising methods to measure such a fast reaction and has been developed all over the world. Previously, the measurement of reactions that occur within 30 ps had been difficult because of a low time resolution for a few decades. Recently, a new picosecond pulse radiolysis system, in which a femtosecond laser was used, was proposed. The synchronization of the femtosecond laser with the subpicosecond electron single pulse was succeeded at the Institute of Scientific and Industrial Research, Osaka University [4-5]. Very recently, the new system [6] was improved to the higher resolution pulse radiolysis system by using a subpicosecond electron pulse and a jitter compensation system. We reexamined the geminate process in very short time region by using the subpicosecond pulse radiolysis.

479

2.

Subpicosecond Pulse Radiolysis

Figure 1 shows the subpicosecond pulse radiolysis system. The system consists of a subpicosecond electron linac as an irradiation source, a femtosecond laser as an analyzing light, and a jitter compensation system. A sample was irradiated by a subpicosecond electron single pulse. The subpicosecond electron single pulse was obtained by compressing a 30 ps electron single pulse from the ISIR L-band linac with a magnetic pulse compressor. This system can compress the width of electron pulse to approximately 125 fs (rms) [6]. The time-resolved optical absorption was detected with a femtosecond laser. A modelocked ultra fast Ti: Sapphire laser (Tsunami, Spectra-Physics Lasers, Inc.) was synchronized to the ISIR L-band Linac using a commercially available phase lock loop. The width of the laser pulse was 60 fs (FWHM). The intensity of the laser pulse was measured by a Si photodiode. The timing between the electron pulse and the laser pulse was controlled by radio frequency (RF) system. In order to avoid effects of the jitter on the time resolution, a jitter compensation system was designed. The time interval between the electron pulse (Cherenkov light) and the laser pulse was measured by the streak camera at every shot. The Cherenkov radiation was emitted by the electron pulse in air at the end of the beam line. The laser pulse was separated from the analyzing light by a half mirror. The precious time interval could be obtained by the analysis of the streak image. All equipment described above was controlled by a personal computer.

I TImlng-Jitter System

LigM j

; System

i Trigger !~

Fig. 1. Block diagram of subpicosecond pulse radiolysis system.

3.

Geminate Ion Recombination in n-Dodecane

A thermalized electron and a parent positive ion (geminate ion pair) in irradiated non-polar liquid diffuse under the Coulomb attracting potential. Most of the geminate

480

Time [ps)

Fig. 2. Time-dependent behavior of radical cation obtained in the subpicosecond pulse radiolysis of n-dodecane monitored at 790 nm. ion pairs recombine inhomogeneously in the picosecond order. Theoretically, the geminate ion recombination can be described by the Smoluchowski equation. Figure 2 shows the time-dependent behavior of radical cation obtained in the subpicosecond pulse radiolysis of n-dodecane monitored at 790 nm. The solid line shows the theoretical decay based on the Smoluchowski equation. The sum of the diffusion coefficient of 6 x lO""^ cm^/s and the experimental distribution with the initial separation of 6.6 nm were used [6]. The geminate decay of cation radical after 50 ps can be explained by the diffusion theory. However, the difference between the experimental result and the theory was observed within 50 ps. The difference in the short time region should be caused by several reasons as follows. 1) Overlap of the optical absorption by another short-lived species 2) Effect of the multi ion spur, 3) High mobile precursor of the radical cation, 4) Diffusion theory. To solve the problem, we consider the very fast process such as thermalization process in femtosecond and subfemtosecond region.

References 1 Y. Yoshida, Y. Mizutani, T. Kozawa, A. Saeki, S. Seki, S. Tagawa and K. Ushida, Radit. Phys. Chem., 60, 313, 2001. 2 K. Kozawa, Y. Mizutani, M. Miki, T. Yamamoto, S. Suemine Y. Yoshida and S. Tagawa, Nucl. Instrum. Meth., A440, 251, 2000. 3 K. Kozawa, A. Saeki, Y. Yoshida and S. Tagawa, Jpn. J. Appl. Phys. Pt. 1, 41, 4208, 2002. 4 A. Saeki, T. Kozawa, Y. Yoshida and S. Tagawa, Radiat. Phys. Chem., 62, 319-322, 2001. 5 A. Saeki, T. Kozawa, Y, Yoshida and S. Tagawa, J. Phys. Chem., A108,1475, 2003. 6 Y. Yoshida, T. Ueda, T. Kobayashi, H. Shibata and S. Tagawa, Nucl. Instrum. Meth., A327,41, 1993.

481

Fast Spin Dynamics of Optically Induced Magnetization in Aqueous Solutions of Magnetic Ions Shigenori Fume\ Toshiro Kohmoto^, Masakazu Kunitomo^, and Yukio Fukuda^ ^ Graduate School of Science and Technology, Kobe University, Kobe, 657-8501, Japan E-mail: [email protected] ^ Department of Physics, Faculty of Science, Kobe University, Kobe, 657-8501, Japan E-mail: [email protected] Abstract. Fast spin dynamics of optically induced magnetization in aqueous solutions of transition-metal ions is studied by the polarization spectroscopy. Concentration dependence suggests the decay of the magnetization is caused by the spin-spin interaction.

1. Introduction Optical pumping with a short laser pulse is very useful to study ultrafast spin dynamics in condensed matter at room temperature. The observations of optically induced magnetization and the spin dynamics of the transition-metal ions in crystals or aqueous solutions at room temperature have been reported [1-3]. In these experiments the magnetization was created by circularly polarized laser pulses, and the time derivative of the optically induced magnetization was monitored by pickup coils. Thus the time resolution of such a detection system is of the order of nanoseconds at the highest. If the induced magnetization is monitored by optical pulses, the time resolution can be remarkably improved because the time resolution is limited only by the temporal width of the light pulses. Then it is possible to observe the ultrafast spin dynamics in the picosecond or femtosecond region. In the present work, we apply the polarization spectroscopy with the optical pump-probe technique to investigate the fast spin dynamics in aqueous solutions of transition-metal ions at room temperature. The magnetization in the ground state of the copper ions (electron spin S = 111) in aqueous solutions of copper sulfate (CUSO4) is created by a circularly polarized pump pulse, which can create instantaneously a large population difference even at high temperatures and in low magnetic fields. The decay of the magnetization in the region of subnanoseconds is monitored through the change of the linear polarization of the probe pulse due to circular dichroism.

2. Experimental Result The pump and probe pulses are provided by optical parametric amplifiers, where their wavelengths are 1.3 jxm and 650 nm, respectively. The repetition rate of the laser pulses is 1 kHz, the pulse width is 0.2 ps, and the pulse energies of the pump and probe pulses are respectively about 10 |iJ and 1 ^iJ in front of the sample cell. The change of the polarization of the probe pulse is monitored by using crossed polarizers and a photomultiplier. To attain higher signal-to-noise

482

ratio, a photoelastic modulator for the pump pulse, which alters the sense of the circular polarization for every pulse, and a lock-in amplifier for the photomultiplier output are used. The observed decay curve of the induced magnetization in a saturated solution of CUSO4 (1.4 mol// at 25 °C) with no external magnetic field is shown in Fig. 1(a). The decay time is in the region of lO'^^-lO'^ second. Such direct observation of the fast decay of the magnetization in the time domain can be realized only by our all-optical method. Figures 1(b) and 1(c) show the decay curves for 70% and 50% diluted solutions of CUSO4. For the diluted samples, the time constant of the decay becomes larger. saturated solution

1.4 mol / / //=0kOe R.T.

7,= 270ps

70% diluted solution

o.98 mol / / //=0kOe RT.

7g=420ps

50% diluted solution

0.7 mol / / // = 0 kOe R.T.

7g = 560 ps

Delay Time t ( ns ) Fig. 1. Observed decay curves of the optically induced magnetization (a) in a saturated (1.4 mol//) and (b), (c) in diluted (0.98 and 0.7 mol//) aqueous solutions of CUSO4 in no external magnetic field. The solid curves are the best fits of Eq. (1) to the observed data.

3. Discussion The observed concentration dependence of the magnetization signals suggests that the decay of the magnetization is caused by the spin-spin interaction betw^een the randomly-distributed copper ions. In this case, the decay curve does not become the usual single exponential function because of the distribution of the distance between the randomly distributed spins. According to a theory of the spin cross relaxation [4], the relaxation curve has a shape of exp[-(y/)^^^] with the spin cross relaxation rate y. We assume the shape of exp[-(70^^^] for the spin cross relaxation and analyze the observed decay curves by using the following decay function S{i), which are shown as the solid curves in Figs. 1(a) - 1(c);

483

S(t) oc exp(- y ^ jexp(- t/T^)

(1)

where Ti is the spin-lattice relaxation time and its value is L3 ns. The spinlattice relaxation time is measured in the external magnetic field of 4 kOe parallel to the pump beam, where the effect of the spin cross relaxation can be neglected. When the external magnetic field is higher enough than the internal magnetic field due to the surrounding spins, the spin-phonon interaction becomes dominant in the decay of the magnetization because the magnetization is not changed by the spin-spin interaction. The obtained relaxation times l/y^, which are defined as the time of 1/e point, are 270 ps, 420 ps, and 560 ps for the saturated and diluted solutions. The concentration dependence of the spin cross relaxation rate for the copper ions together with that for the manganese ions is shown in Fig.2, where the relaxation rate y is the contribution from the spin cross relaxation in Eq. (1) and does not contain that from the spin-lattice relaxation. When the concentration of the copper ions is decreased, the relaxation rate y due to the spin-spin interaction becomes smaller as is proportional to the second power of the concentration, which is expected from the theory [4]. This result shows that the observed decay of the magnetization is caused by the spin cross relaxation due to the magnetic dipole-dipole interaction between the copper ions. Such direct observation of the fast decay of the magnetization due to the spin cross relaxation can be realized by our all-optical method.

•

/T

Cu^*

/

A Mn'"

/

M

o

^ 0

0.2

0.4

0.6

0.8

1

Normalized Density p Fig. 2. Concentration dependence of the spin cross relaxation rate for the copper (solid circles) and the manganese (solid triangles) ions in aqueous solutions. The spin density p is normalized by that for the saturated solutions. The solid lines show the quadratic curves expected from the theory of cross relaxation [4].

References 1 2 3 4

484

G. F. Hull Jr., J. T. Smith, and A. F. Quesada, Appl. Opt. 4, 1117, 1965. J. P. van der Zeil and N. Bloembergen, Phys. Rev. B 138, A1287 1965. Y. Takagi, Opt. Commun. 59, 122, 1986. T. Endo, and T. Muramoto, Phys. Rev. B 29, 6043, 1984.

Vibrational excitation and energy redistribution after ultrafast intramolecular proton transfer of TINUVIN W. Werncke, V. Kozich, and J. Dreyer Max-Bom-Institut, Max-Bom-Strasse 2A, D-12489 Berlin, Germany E-mail: [email protected]

Abstract: Vibrational excitation and energy redistribution after ultrafast intramolecular proton transfer of TINUVIN is investigated by picosecond resonance Raman spectroscopy. It is demonstrated that a low-frequency proton transfer promoting mode serves as the major accepting mode.

1.

Introduction

2- (2' -hydroxy-5' -methy 1-phenyl)benzotriazole (TINUVIN) is a prototypical example undergoing a complete reaction cycle of intramolecular photo-induced proton transfer within about 1 ps [1]. It is still unknown which mode(s) accept(s) the large excess energy of the process. A prominent role of a low-frequency mode at 469 cm"^ is already indicated by coherent oscillations during and after the reaction cycle as well as by resonance Raman data. This mode which contains a coordinate changing the 0...N separation in the intramolecular hydrogen bond has been identified as a proton transfer promoting mode [1]. To identify the accepting modes we studied vibrational excitation and energy redistribution of TINUVIN after a proton transfer cycle applying picosecond resonance Raman spectroscopy.

2. Experimental Methods For time-resolved resonance Raman spectroscopy we used a two-color 1 kHz Tisapphire laser system delivering picosecond pulses at 310-325 and 340-360 nm [2]. After frequency doubling of the fundamental radiation at 790 nm these pulses were generated using an optical parametic generator/amplifier followed either by frequency doubling or sum frequency generation with the 790 nm radiation. A typical cross correlation width of the near UV pulses was 1.5 ps. Pulses around 350 nm, i.e. near to the 0-0 electronic transition of TINUVIN were applied for electronic excitation, whereas the pulses around 320 nm served for anti-Stokes and Stokes resonance Raman probing, respectively.

485

3. Results and Discussion The kinetics of Stokes resonance Raman intensities of three prominent Raman lines (469, 685 and the spectrally unresolved doublet at 1435 cm"^) is shown in figure lA. Immediately after excitation we observed an about 5-fold decrease followed by a slow recovery within about 30 ps. Time-resolved anti-Stokes resonance Raman intensities of the three modes are depicted in figure IB. The 1435 cm'^ doublet shows a fast rise of intensity close to our temporal resolution being in contrast to the much slower rises of the 469 and 685 cm'^ modes approaching their maxima at about 7 and 10 ps delay times, respectively. In addition to the changes of intensities for all modes, we observed frequency downshifts after excitation on a sub-picosecond time scale which recover to their frequency positions at ambient temperature within 8-10 ps. _j 8001 5 700-

,

A

1

• J!--*^

1435cm''

^^'^

_g;600^500-

X"'

C 400QC 300-

y/m

•

0

.^^-TT-o

685cm"'

^ ^ ^

W 200O 100-

^.^^,-A-*-^-*-^ 469cm"' 10

20 30 40 Time delay, ps

50

469cm'

10 Time delay, ps

20

30

Time delay, ps

Fig. 1. Time-dependence of resonance Raman intensities. A: Stokes, B: Anti-Stokes, C: Anti-Stokes/Stokes intensity ratios ("mode temperatures" are indicated) As the proton transfer cycle is complete within about 1 ps the much slower recovery of Stokes Raman intensities shown in figure lA must be explained by time-dependent resonance Raman scattering cross sections due to vibrational excitation. Our simulations of Stokes Raman resonance intensities which are based on transform theory [3] and on the time-dependent absorption line shapes reported in ref. 4 predict changes of comparable magnitude as well as similar kinetics as observed experimentally. Furthermore, anti-Stokes resonance Raman intensities of low-frequency modes can even increase although the vibrational

486

energy of the molecule is already decreasing. Consequently, relating anti-Stokes resonance Raman intensities directly to vibrational excess populations can be completely misleading. To avoid the influence of time-dependent resonance Raman cross sections, instead of using anti-Stokes Raman intensities, we characterized vibrational populations by the ratios of time-dependent anti-Stokes and Stokes resonance Raman intensities [5]. They are presented in figure IC. Excitations of the modes are expressed by "mode temperatures" which follow from the ratios. At earlier times the ratios indicate pronounced non-thermal vibrational population distributions. For instance, 5 ps after excitation the anti-Stokes intensity ratios of the 1435 and 685 cm'^ modes may still correspond to the same temperature of about 1050 K. In contrast, the "mode temperature" of 2300 K of the 469 cm"^ mode is significantly higher than the highest possible equilibrium temperature of TINUVIN of 1200 K [4]. These findings indicate a strong excess population of the 469 cm'^ mode, i.e. this mode acts as an accepting mode. Thermal equilibrium among the vibrations occurs on a 10-15 ps time scale.

4.

Conclusions

Evidence of strongly time-dependent resonance Raman cross sections after the ultrafast proton reaction cycle of TINUVIN has been obtained. They prevent from extracting valid kinetic data directly from time-resolved resonance Raman measurements of vibrationally excited molecules. We demonstrate, however, that information on vibrational temperatures or more in general, vibrational excess populations can be derived from combined anti-Stokes and Stokes resonance Raman intensity measurements. We show that a low-frequency mode modulating the O.. .N separation in the intramolecular hydrogen bond is not only coherently excited on a femtosecond time scale but also absorbs a major amount of excess energy. Acknowledgements. This Forschungsgemeinschaft.

research

was

supported

by

the

Deutsche

References T. Elsaesser, in Ultrafast Hydrogen Bonding Dynamics and Proton Transfer Processes in the Condensed Phase, Edited by T. Elsaesser and H. Bakker, Kluwer Academic Publishers, Dordrecht, 119, 2002. V. Kozich, W. Wemcke, A.I. Vodchits, and J. Dreyer, J. Chem. Phys. 118, 1808, 2003. X. Ye, A. Demidov, F. Rosea, W. Wang, A. Kumar, D. lonascu, L. Zhu, D. Barrick, d. Wharton, and P.M. Champion, J. Phys. Chem. A 107, 8157, 2003. K. Lenz, M. Pfeiffer, A. Lau, and T. Elsaesser, Chem. Phys. Lett. 229, 340, 1994. K.T. Schomacker, and P.M. Champion, J. Chem. Phys. 90, 5982, 1989.

487

Coherent Nuclear Motion in Reacting Molecules: Ultrafast Pump-Probe Spectroscopy of Proton Transfer in Solution Satoshi Takeuchi and Tahei Tahara Molecular Spectroscopy Laboratory, RIKEN (The Institute of Physical and Chemical Research), 2-1 Hirosawa, Wako 351-0198, Japan E-mail: [email protected] and [email protected] Abstract. Time-resolved stimulated emission of 10-hydroxybenzoquinoline was measured in solution with 27-fs resolution. It clearly showed beating feature due to the excited-state wavepacket motion during the proton transfer reaction. We discuss the coherent nuclear motion in the reactive excited state and its relation to the reaction coordinate.

1.

Introduction

Ultrashort pulses have a bandw^idth broad enough to excite a number of vibrational eigen states simultaneously and generate their coherent superposition. The time evolution of this coherent state (the nuclear vs^avepacket motion) often gives rise to beating feature in time-resolved signals, which corresponds to 'realtime' observation of the motion of nuclei. This time-domain measurement of the nuclear motion has been receiving a great deal of attention in advanced femtosecond spectroscopy [1, 2]. Especially, the measurements for ultrafast reactive systems give us an opportunity to observe the coherent nuclear dynamics in the early stage of chemical reactions [3, 4]. It is very crucial to know^ the initial motion of nuclei and its relevance to the reaction coordinate to elucidate the mechanism of chemical reactions. Unlike the case of simple diatomic molecules, the reaction coordinate in polyatomic molecules does not simply correspond to the change of a particular chemical bond. Therefore, it is not yet clear for polyatomic molecules how^ the observed wavepacket motion is related to the reaction coordinate. With this complexity of polyatomic molecules in mind, we have studied several ultrafast reactive systems in solution, such as intramolecular proton transfer, by using two-

hv enol form

keto form

Fig. 1. Intramolecular proton transfer reaction of 10-hydroxybenzoquinoline. The enol form is converted to the keto form in the photoexcited state

488

color pump-probe spectroscopy with tunable 10-fs pulses. Here, we discuss the coherent nuclear motion in the excited state of 10-hydroxybenzoquinoline and its relation to the reaction coordinate.

2.

Experimental

Pump-probe absorption experiments were carried out by using tunable 10-fs pulses generated from noncollinear optical parametric amplifiers (NOPAs). For independent tuning of the pump and probe wavelengths, we employed two NOPAs that were synchronously driven by amplified Tiisapphire laser pulses (800 nm, 1 mJ, 100 fs, 1 kHz). The output of the first NOPA was frequency-doubled, and the generated uv pulse was recompressed to a sub-20 fs regime by a prism pair. This uv pulse was used as a pump pulse for photoexcitation of the sample. The output of the second NOPA in the visible region was used as a probe and a reference pulse for monitoring the time-resolved absorption signal. Both the pump (0.1 |LJ) and probe (< 1 nJ) pulses were focused on a 50-|im thick jet stream of the sample solution. The intensities of the probe and reference pulses were detected by photodiodes and the signals were processed on a shot-to-shot basis for the evaluation of pump-induced absorbance change. The time resolution of this pump-probe measurement was typically 27 fs.

3.

Results and Discussion

10-Hydroxybenzoquinoline (10-HBQ) undergoes intramolecular proton transfer reaction in the photoexcited state, and is converted from the enol form to the keto form (Figure 1). It was reported that this proton transfer proceeds in less than 100 fs in nonpolar solvents [5]. Transient absorption spectrum that we measured at 0.5 ps after photoexcitation indicated that the reacting excited state shows

cross correlation (27-fs fwhm)

oscillatory feature due t o . coherent nuclear motion

Fig. 2. Time-resolved stimulated emission signal of 10-hydroxybenzoquinoline in cyclohexane (30 mM) measured with 27-fs resolution. The pump and probe wavelengths are 360 and 620 nm, respectively. The oscillatory feature observed over the entire time region reflects the nuclear wavepacket motion in the excited state

489

absorption in the wavelength region shorter than 600 nm and stimulated emission in the longer wavelength region. Based on this spectral information, we carried out two-color pump-probe measurements using the NOPA system in order to examine the coherent nuclear dynamics in this ultrafast proton transfer. We excited this molecule by ultrashort uv pulses at 360 nm, and measured timeresolved absorption at 560 and 620 nm. At both probe wavelengths, we clearly observed beating feature that reflects the nuclear wavepacket motion in the reactive excited state. As shown in Figure 2, the observed stimulated emission signal showed complicated oscillatory feature until a delay time of 4 ps that is much longer than the reaction time. This indicates that the molecule exhibits a coherent nuclear motion during, and even after, the reaction. A Fourier transform analysis indicated that four vibrational modes in the 200 - 700 cm'^ region contribute to the observed oscillation. Especially, the lowest frequency mode at 242 cm"^ showed a faster vibrational dephasing compared to the others, suggesting the importance of this vibrational mode in the reaction. We assigned this lowest frequency mode on the basis of Raman data as well as results of the density functional calculations. As shown in Figure 3, this mode involves a large displacement of the OH group and an in-plane skeletal deformation that seems to assist the motion of the OH group. These motions significantly modulate the -OH N- hydrogen-bond length, which is closely related to the translocation of the proton. We concluded that the coherent nuclear motion observed in the excited state of 10-HBQ is substantially correlated with the reaction coordinate of the proton transfer of this molecule.

Fig. 3. Nuclear motion of the (ground-state) vibration at 243 cm"^ that corresponds to the lowest-frequency wavepacket motion observed in the reacting excited state

References 1 2 3 4 5

490

L. Dhar, J. A. Rogers, K. A. Nelson, Chem. Rev. 94, 157, 1994. S. Takeuchi and T. Tahara, Chem. Phys. Lett. 326, 430, 2000. S. Lochbrunner, A. J. Wurzer, and E. Riedle, J. Phys. Chem. A 107, 10580, 2003. S. Takeuchi and T. Tahara, J. Chem. Phys. 120, 4768, 2004. P. T. Chou, Y. C. Chen, W. S. Yu, Y. H. Chou, C. Y. Wei, and Y. M. Cheng, J. Phys. Chem. A 105, 1731 2001.

Ultrafast double proton transfer: symmetry breaking wavepacket motion and absence of deuterium isotope effect S. Lochbrunner, K. Stock, C. Schriever, and E. Riedle LS far BioMolekulare Optik, LMU Miinchen, Oettingenstr. 67, 80538 Miinchen, Germany Abstract: The double proton transfer of [2,2'-Bipyridyl]-3,3'-diol is investigated by UVvisible pump-probe spectroscopy with 30 fs time resolution. The first step of the sequential transfer proceeds with a 40 fs delay and the second step with a time constant of 10 ps. The concerted transfer of the two protons takes 50 fs. Both transfer paths are accompanied by characteristic wavepacket motions, a bending motion for the sequential transfer and symmetric stretching motions for the concerted. The nontotally-symmetric bending vibration cannot be excited by the optical transition. This demonstrates that the reactive process itself and not only the optical excitation drives the vibrational motions. We show by absence of a deuterium isotope effect on the proton transfer that the ESIPT dynamics is entirely determined by the skeletal modes and that it should not be described by tunneling of the protons.

!• Introduction We recently performed UV-visible pump-probe measurements v^ith 30 fs time resolution on various molecules exhibiting excited state intramolecular proton transfer (ESIPT) [1-5]. The results led to a very precise model of the reaction dynamics. The ESIPT proceeds as a ballistic v^avepacket motion along skeletal coordinates from the Franck-Condon region to the product w^ell. The initial deformation of the molecule is driven by the gradient of the excited potential energy surface and reduces the proton donor/acceptor distance by more than 0.2 A. This leads to a change of the electronic configuration and the bonds alter. Subsequently the nuclei move towards their nev^ equilibrium positions. A few participating normal modes with a significant alternation of the donor/acceptor distance are coherently excited by the process leading to a characteristic ringing of the molecule and to strong oscillations of the pump-probe signal. The oscillations are damped on a time scale of about 1 ps. Two consequences arise from this model: First, the coherent excitation of normal modes can result from the reactive process and need not be caused by the optical excitation. Optically inactive modes can be excited if they contribute to the reaction path. Second, the proton is passively shifted by the skeletal movements and stays all the time near its local potential well. A substitution of the proton by a deuteron should therefore not influence the observable dynamics. In contrast strong variations are expected, if turmeling of the proton is the central step. For both predictions experimental verifications with sufficient time resolution are still missing.

491

time (fs)

2000

3000

4000

I I I I I I I I

0.0

-1.0

^o..,^n= pump 375 nm X^,^u= probe 630 nm

500 time(fs) 1500

2000

^pump= 350 nm X^^^^^= 630 nm

O time (fs) 1000

1500

2000

Fig. 1. Transient transmission change of BP(0H)2 at different excitation and probe wavelengths. Insets show the Fourier transformations of the oscillatory contributions detected at 630 nm for different excitation wavelengths. For a rigorous test of the described model of the ESIFT and its predictions we investigate the double proton transfer system [2,2'-Bipyridyl]-3,3'-diol (BP(0H)2) dissolved in cyclohexane. The molecule contains two H-chelate rings and exhibits inversion symmetry in the electronic ground state (see Fig. 1) and at the FranckCondon point. Glasbeek and coworkers found that after optical excitation the molecule can undergo double proton transfer in a concerted or sequential fashion leading to the diketo form (both H-Atoms transferred) respective to a monoketo intermediate which does not show inversion symmetry [6]. The monoketo yield increases with shorter excitation wavelengths [7]. The monoketo intermediate transforms in a second step to the final diketo form within 10 ps. The initial steps of the proton transfer have not been observed in real time up to now. If our understanding of ESIPT is correct and generally applicable, it should also allow the description of the double proton transfer, even so it was derived from experiments and calculations on single proton transfer.

2. Symmetry breaking wavepacket dynamics We performed transient absorption measurements on BP(0H)2 with a spectrometer based on two noncollinearly phase matched optical parametric amplifiers (NOPAs) pumped by a homebuilt regenerative Ti:sapphire laser system [1,8]. The tunable UV pump pulses are generated by frequency doubling the output of one of the NOP As. The other NOPA provides the visible probe pulses. The cross correlation between pump and probe pulses has a typical width (FWHM) of 40 fs. The

492

sample is a cyclohexane solution of BP(0H)2 pumped through a flow cell with a 120 iim thick chamiel. Fig. 1 shows measurements with excitation at 375 and 350 nm and probe wavelengths of 485 and 630 nm. The change of the sample transmission is depicted in dependence on the delay time. Strong oscillations are observed in all traces in addition to the slower changes in the signal. In the spectral region of the product fluorescence at ^probe = 630 nm we find an emission rise after a small but significant delay with respect to the pump pulse. The delay is interpreted as the time the optically prepared wavepacket needs to travel from the Franck-Condon region to the product minimum [2,4,5]. The initial motion is driven by the change in the electronic configuration due to the optical excitation. A further configuration change occurs shortly before the product minimum, namely the change from the enol to the keto form. It is this latter variation that gives rise to the strongly Stokes shifted emission and the observed ultrafast signal. Since dominantly monoketo is observed at 630 nm [6], the 40 fs delay is interpreted as the time for the first step of the sequential proton transfer. The oscillating signal components originate from coherently excited vibrational modes in the product form of the proton transfer. Dominating contributions are identified at 196 and 295 cm"^ a weaker one at 331 cm"^ is seen in addition in some measurements. This is demonstrated by the Fourier transformations of the oscillatory contributions shown in Fig. 1. At 485 nm the transient spectrum of BP(OH)2 is found to change from stimulated emission to transient absorption with a steep slope [9]. As a consequence very pronounced oscillations are found and only a weak average transmission change. This overall signal slowly changes and thereby indicates the second step in the sequential proton transfer. The very pronounced initial decrease in transmission is due to the appearance of transient absorption that occurs instantaneously with the optical excitation and is almost canceled 50 fs later by the beginning stimulated emission of the diketo form. The spiked feature together with the accurate calibration of zero delay time allows for the precise determination of the 50 fs proton transfer time for the concerted pathway.

Calc.

206 cm"^

,

318 cm-^

iJLo Exp.

196 cm-^

295 cm-^

f

Fig. 2. Normal modes of BP(0H)2 determined from DFT calculations. A close agreement between the calculated and measured frequencies is found. We performed DFT calculations of BP(0H)2 in the electronic ground state and found two vibrational normal modes fitting nicely to the observed frequencies (see Fig. 2). The low frequency mode at 196 cm'^ is attributed to a nontotally-

493

symmetric in-plane bending vibration. This motion alternately reduces the distance between the proton donors and acceptors in the two H-chelate rings. Therefore we believe that it is associated with the single proton transfer. If the molecule is deformed by chance asymmetrically, e.g. due to thermal excitation of this bending mode, the electronic configuration change will happen in that chelate ring where the first contraction occurs. After the ESIPT the molecule exhibits a pronounced ringing in this mode. The 295 and 331 cm"^ modes are symmetric inplane stretching vibrations (compare Fig. 2). The corresponding contraction of the molecule reduces the donor acceptor distances in both chelate rings simultaneously and initiates the electronic configuration change of the concerted ESIPT. The concerted double proton transfer leads therefore to a multidimensional ringing of the molecule in these modes. This assignment is further corroborated by comparing measurements at different excitation wavelengths (see Fig. 1). Changing the excitation from 375 nm to 350 nm increases the yield for the monoketo intermediate by a factor of 2 from 16% to 30% [6]. Correspondingly, the amplitude of the signal oscillations at 196 cm'^ is much smaller for pumping at 375 nm than at 350 nm. Obviously, the 196 cm'^ mode is more strongly excited if more monoketo is generated, demonstrating that this mode contributes in fact to the single proton transfer. The oscillatory contribution at 295 cm"^ is of comparable strength for both excitation wavelengths. Since the yield for the concerted transfer decreases only by a factor of 0.82 the strength of the 295 cm"^ mode is much less affected by a change in the excitation wavelength. The 196 cm'^ mode is nontotally-symmetric and thereby optically inactive in the electronic excitation and does not appear in the Raman spectrum. Since a direct optical excitation of the fundamental and odd overtones is excluded by symmetry selection rules, no vibronic wavepacket oscillating at the fundamental frequency can be generated. Since such a wavepacket is however observed, we conclude that it is solely excited by the single proton transfer which breaks the symmetry. This demonstrates for the first time unambiguously that the coherent excitation of a vibrational mode results exclusively from an ultrafast reactive process.

3. Absence of deuterium isotope effect The protons of the hydroxy groups were deuterated by dissolving BP(0H)2 in cyclohexane and shaking the solution with deuterated water for several hours. After decantation pump-probe measurements of BP(OD)2 in cyclohexane were recorded and are compared to BP(0H)2 in Fig. 3. Both samples were excited at 350 nm and probed at 505 nm. The delay of the emission rise of about 50 fs is equal in both cases and the coherent excitation of the vibrations is identical with respect to frequencies, phases and amplitudes. The ESIPT dynamics is obviously not altered by the deuteration and the mass of the proton has no influence on the transfer speed. A tunneling model would lead to an increase of the transfer time from the 50 fs to approximately 150 fs. Even for a deuteration of only 50 % such a change would easily be observable in our experiment. The degree of deuteration was checked for similar systems by NMR to be at least 70%. We can therefore exclude that tunneling of the proton determines the speed of the transfer. The meas-

494

urements provide direct proof for a passive behavior of the protons in the ESIPT. They are carried along with the changing skeletal geometry and once the donor/acceptor distance is decreased sufficiently, the bonds in the chelate ring change. deuterated BP(OH). BP(OD),

X

= 350 nm

pump

X , = 505 nm probe

tjme (fs) I I I t I 111 I

-100

100

200

300

400

500

600

Fig. 3. Comparison of the transient transmission of BP(0H)2 (triangles) and BP(0D)2 (dots) excited at 350 nm and probed at 505 nm. No differences in the delay of the emission rise and the oscillation pattern are observed.

4. Conclusions We observe the coherent excitation of an optically inactive mode proving that the reactive process itself and not only the optical excitation drives the observed vibrational motions. We also demonstrate that during the ESIPT the proton is adiabatically shifted from one site to the other and tunneling of the proton does not determine the speed of the process. The dynamics is entirely controlled by the skeletal modes.

References 1 2 3 4 5 6 7 8 9

A.J. Wurzer, S. Lochbrunner, and E. Riedle, Appl. Phys. B 71,405 (2000). S. Lochbrunner, A.J. Wurzer, and E. Riedle, J. Chem. Phys. 112, 10699 (2000). K. Stock, T. Bizjak, and S. Lochbrunner, Chem. Phys. Lett. 354, 409 (2002). S. Lochbrunner, A.J. Wurzer, and E. Riedle, J. Phys. Chem. A 107, 10580 (2003). R. de Vivie-Riedle, V. de Waele, L. Kurtz, and E. Riedle, J. Phys. Chem. A 107, 10591 (2003). H. Zhang, P. van der Meulen and M. Glasbeek, Chem. Phys. Lett. 253, 97 (1996). D. Marks, P. Prosposito, H. Zhang, and M. Glasbeek, Chem. Phys. Lett. 289, 535 (1998). E. Riedle, M. Beutter, S. Lochbrunner, J. Piel, S. Schenkl, S. Sporlein, and W. Zinth, Appl. Phys. B 71, 457 (2000). K. Stock, C. Schriever, S. Lochbrunner, and E. Riedle, Chem. Phys. Lett., submitted for publication

495

Photodissociation dynamics studied via TimeResolved Coincidence Imaging Spectroscopy O. GeBner\ E.t-H. Chrysostom^ A.M.D. Lee^ ^ J.P. Shaffer^ C.C. Hayden^ and A. Stolow^'^ ^ Steacie Institute for Molecular Sciences, National Research Council, 100 Sussex Drive, Ottawa ON Canada KIA 0R6, E-mail: [email protected] ^ Combustion Research Facility, Sandia National Laboratories, Livermore CA 94551 USA, E-mail: [email protected] ^ Department of Chemistry, Queen's University, Kingston ON K7L 3N6 Canada "^ Department of Physics and Astronomy, University of Oklahoma, Norman, OK 73019 USA Abstract. Femtosecond time resolved photoelectron-photoion Coincidence Imaging Spectroscopy was used to study the non-adiabatic photodissociation dynamics of the NO dimer at 209 nm. Correlated photoelectron-photofragment energy and angular distributions reveal new details of the dissociation dynamics.

1.

Introduction

Femtosecond time resolved Coincidence Imaging Spectroscopy (CIS) is a new technique for studying complex reaction dynamics in a highly differential manner. It combines femtosecond Time-Resolved Photoelectron Spectroscopy (TRPES) [1] with a kinematically complete analysis of all coincident particles (photoelectron and ionized photofragment) [2]. Typical representations of the results include time dependent correlated energy distributions between photoelectrons and photofragments and their time dependent correlated angular distributions [2]. These can be used to disentangle complex reaction pathways which may be undiscernable via more conventional methods.

2. Experiment and results The CIS spectrometer consists of coaxial electron and ion time-of-flight (TOP) spectrometers placed on opposite sides of the interaction region. Each spectrometer is equipped with a time- and hit position-sensitive delay line detector. From the anode hit positions and the TOE, one can derive the complete 3dimensional emission momentum vectors of both the photoelectrons and photofragments in coincidence. Thus, for each pump-probe event one gets kinematically complete information on the time-resolved correlated emission of a photoelectron and its mass resolved photoion [2]. In Fig.l we show 2D contour maps of the energy-energy correlation between photoelectrons and NO^ photofragment ions upon 209 nm excitation and 278 nm probing (ionization) of

496

the NO dimer photodissociation, at two time delays: 130 fs (left) and 1000 fs (right). Integration of a contour map within the different electron kinetic energy ranges shown leads to the ID cartesian plots below each map.

1000 fs

130 fs

range 2

0.4

0.8

1.2

fragment total KER / eV

0

0.4

0.8

1.2

fragment total KER / eV

Fig. 1. Photoelectron-photoion kinetic energy correlations From TRPES [3] and photofragment imaging spectroscopy [4,5] it is known that the initially excited neutral dimer state decays either directly or via an intermediate state into a neutral dissociation channel which leads to an excited NO(A^I'^) and a ground state NO(X ^n) fragment. The pronounced feature in range 1 of Fig. 1 b) is the signal from the NO(A^S"^) fragments which are ionized by the probe pulse. Integration over all electron energies within this range and subtraction of the dissociative ionization background leads to the kinetic energy release (KER) spectrum of ionic fragments seen in Fig.l d). It is seen to differ significantly from the ion KER spectrum in Fig.l f) which corresponds to the dissociative ionization of the dimer into an excited dimer cation state. While the KER of the NO(A^E'^) fragments is more or less evenly spread over the whole range allowed by energy conservation, most of the fragments from the dissociative ionization channel are very slow and their KER spectrum does virtually not change with the pump-probe delay (Figs.l e)+f)). The electron and ion kinetic energy distributions for the dissociative ionization channel peak towards zero in coincidence (Figs.l a)+b)). In other words, the available energy of the highly excited ionic states is deposited mainly into internal energy of the monomer fragments, independently of the pump-probe delay. By contrast, the KER of the NO(A ^S"^) fragments indicates a more limited internal energy distribution in the NO(A) fi-agment. Only with correlated measurements like CIS is this differentiation into the dissociative ionization and the neutral dissociation channel possible for all pump-probe delays.

497

Another sensitive probe available through CIS is access to state selective angular distributions of both photoelectrons and photo fragments. Fig.2 shows the laboratory frame angular distributions of ionic fragments at a pump-probe delay of 1 ps from two different channels: (1) the dissociative ionization channel (Fig.2 a)); (2) the neutral dissociation channel followed by ionization of the NO(A ^I^) fragment (Fig.2 b), (including an underlying contribution from the dissociative ionization channel). The light polarization vector lies vertical as indicated.

Fig. 2. Laboratory frame NO"^ photoion angular distributions at a pump-probe delay of 1 ps All fragments are emitted preferentially along the laser polarization vector, indicating an initial excitation transition moment parallel to the N-N bond in agreement with previous measurements [4,5]. The ion emission anisotropy in the dissociative ionization channel is significantly smaller than in the neutral dissociation channel although still quite pronounced. Without further illustration, we note that the photoelectron lab frame angular distributions for the different reaction channels differ dramatically and that the fully correlated angular distributions also permit transformation of the photoelectron angular distribution into the photofragment recoil frame (rather than the lab frame). This is possible only via CIS and could reveal details of the time-resolved angular momentum polarization of the photofragments and on the nature of the excited dimer state itself

References A. Stolow, Annu. Rev. Phys. Chem., 54, 89 (2003); C.C. Hayden & A. Stolow, in Advanced Physical Chemistry Vol. 10, C.-Y. Ng, Ed. (World Scientific, Singapore, 2000). J.A. Davies, J. E. LeClaire, R. E. Continetti, & C.C. Hayden, J. Chem. Phys. Ill, 1 (1999); J.A. Davies , R.E. Continetti, D.W. Chandler & C.C. Hayden, Phys. Rev. Lett. 84, 5983 (2000). V. Blanchet & A. Stolow, J. Chem. Phys. 108, 4371 (1998). A.B. Potter, V. Dribrinski, A.V. Demyanenko, and H. Reisler, J. Chem. Phys. 119, 7197(2003). M. Tsubouchi, C.A. de Lange, T. Suzuki, J. Chem. Phys. 119, 11728 (2003).

498

Femtosecond Photo-induced Dissociation of the Trihalide Anions I3" and IiBr" in Solution Peter Salen\ Ming Liu^, and Peter van der Meulen^'^ ^ Stockholm University, Albanova Nova University Center, Department of Physics, SE10691 Stockholm, Sweden E-mail: [email protected] ^ KTH, Albanova Nova University Center, Department of Chemical Physics, SE-10691 Stockholm, Sweden Abstract. The photo-induced dissociation of the trihalide anions I3' and l2Br' in methanol and acetonitrile solution is investigated using magic angle and polarization sensitive femtosecond transient absorption spectroscopy. The dissociation of l2Br' and I3' in acetonitrile are found to be very similar, except at the earliest times. A noticeable solvent dependence for the dissociation process is seen, e.g. in the degree of vibrational coherence and the rotational temperature of the !{fragmentand in the amount of recombination.

1.

Introduction

Tri-atomic species form ideal model systems for the study of dissociation dynamics in solution [1]. In this contribution we will use femtosecond transient absorption spectroscopy to investigate the photo-induced dissociation of the triatomic anions I3' and l2Br": I ~+hv->I^"+I

l2Br~+hv->I^~+Br

(1)

By comparing the dissociation of a symmetric (I3") and an asymmetric (l2Br") halide in two different polar solvents (methanol and acetonitrile; l2Br" proved to be unstable in methanol) with regard to vibrational coherences, dephasing and relaxation, rotational excitation of the I2' photoproduct, geminate recombination and solvent caging one may anticipate to obtain an improved understanding of the dissociation process on a molecular level. Compared to the symmetric halide I3', the permanent dipole moment of the asymmetric halide l2Br' will probably lead to a larger interaction with the solvent surrounding it. This may perhaps affect its initial geometry as well as the outcome of the reaction. The photo-dissociation reaction (1) is induced by a 390 nm pump pulse generated by frequency doubling the output of an integrated amplified Ti: Sapphire femtosecond laser system (Clark MXR CPA2001) producing pulses of 150 fs FWHM. The white light continuum probe was dispersed after the sample by means of a polychromator and detected by a CCD. The sample consisted of a 250 |Lim thick flow cell. All measurements have been shifted to a common time origin.

499

2.

Results and Discussion Ij" in methanol

0.04

Ij' in methanol

in acetonitrile

I^' in acetonitrile

LBr" in acetonitrile

0.03

Bf in acetonitrile

8

,

< 0.02-j

I I o.oi. 0

2

1

o c

3

0

-25 0 0.00'

0.04-1 (c)

Delay

0.04n (d)

0.03-

— 2ps - "' ' 3 ps

0.03-

0.02-

<

»M

0.00-

5ps 7ps ——9ps ——lips 15 ps 21 ps

21 ps

-0.01-

0.02-

— 4 p s

P

0.01-

Q O

250

500

delay time (fs)

delay time (ps)

Delay

vf

^^CVSs^llps

0.01Q O 0.00-

1=;

lips

<

-0.01-0.02-

-0.02-

-0.03400

500

600

wavelength (nm)

700

800

400

500

600

700

800

wavelength (nm)

Fig. 1. Normalized magic angle transient absorption curves for a 600 nm probe covering the initial 3 ps (a) and 500 fs (b). Magic angle spectra at different delays for l{ in methanol (c) and acetonitrile (d). From figure la it can be seen that the signals belonging to the I2' fragments produced from either !{ or l2Br" are essentially identical on a longer time scale (>0.5 ps). This is contrary to earlier measurements on l2Br" [2] v^ith a 310 nm pump but is consistent with molecular dynamics simulations [3] demonstrating that the interaction between the photofragments vanishes after 400 fs. Furthermore, the large I2' oscillations found in methanol are almost absent in acetonitrile. Our results indicate that this can not be simply ascribed to asymmetry in the parent triatomic anion [2] and the detailed origin of this solvent effect is currently unknovm. In the early time behavior of the photodissociation process shown in figure lb, a shoulder following the initial peak is visible. The initial peak corresponds to the transition state [1] and the shoulder represents the first sign of the I2" fragment. In acetonitrile the shoulder seems to appear slightly earlier for l2Br" than for I3" where the separation between the first peak and the shoulder is more prominent. This may indicate that the dissociation is faster for the asymmetric l2Br" compared to the symmetric I3" due to an earlier choice of the exit charmel for the wave packet initially travelling down the symmetric stretch (in I3"). In methanol the shoulder appears even earlier and is also much larger, suggesting that I3" in methanol is indeed also asymmetric and that the dissociation is extremely rapid. This would be consistent with a higher degree of preserved coherence during the dissociation process. Figures Ic and Id show the vibrational cooling of the hot I2" fragments and a decrease of the signal with time resulting from recombination. The decrease is

500

more pronounced and continues for longer times in methanol than in acetonitrile where for both I3' and l2Br' spectral evolution terminates at ca. 7 ps demonstrating a lower degree of recombination. The higher level of recombination in methanol is possibly due to the hydrogen bonds formed between the solvent molecules thereby hindering the fragments to separate. The ratio between the intensity of the left I Br in acetonitrile and right !{ bands in methanol is I ' in acetonitrile smaller than in acetonitrile indicating a 1/ in methanol larger percentage of three-body dissociation. The time dependence of the rotational anisotropy r(T) of the I2" fragment is presented in Figure 2. The initial slope of this curve depends only on the rotational temperature of the !{ ions formed in the photodissociation. Clearly, from Figure 2 one can deduce Fig. 2. Fit of the initial decay of the that the I2" fragment is formed with a anisotropy r for a 600 nm probe plotted higher degree of rotational excitation in as a function of the square of the delay methanol than in acetonitrile and that after reaching its maximum, see [4], there is little difference in the rotational temperature of the I2' photoproduct obtained from 13" or l2Br". These results might be rationalized by assuming a bent structure for the Is" anion in methanol, as has been suggested before [4], and a linear geometry for I3" and l2Br" in acetonitrile.

3.

Conclusions

Our results show that for a 390 nm pump pulse the photo-induced dissociation of l2Br' and I3" in acetonitrile are very similar. Only at the very earliest times can a difference be discerned, in accordance with molecular dynamics simulations. The effect of the solvent is illustrated by our measurements in methanol which indicate a faster dissociation, a higher degree of vibrational coherence, a higher rotational temperature and higher levels of three-body dissociation and recombination. Acknowledgements. We are indebted to Django Andrews (University of Colorado) for experimental assistance during the early stages of this work.

References 1 T. Kiihne, R. Kiister, and P. Vohringer, Chemical Physics 233, 161, 1998. 2 E. Gershgoren, E. Gordon, and S. Ruhman, Journal of Chemical Physics 106, 4806, 1997. 3 I. Benjamin, U. Banin, and S. Ruhman, Journal of Chemical Physics 98, 8337, 1993 4 H. Bursting, J. Lindner, S. Hess, and P. Vohringer, Applied Physics B 71, 411, 2000.

501

Coherent control of non-radiative transitions: long-range electron transfer B.D. Fainberg^'^ V.A. Gorbunov \ and S.H. Lin^ ^ Department of Sciences, Holon Academic Institute of Technology, 52 Golomb St., Hoion 58102, Israel ^ Raymond and Beverly Sackler Faculty of Exact Sciences, School of Chemistry, Tel-Aviv University, Tel-Aviv 69978, Israel E-mail: [email protected] ^ Institute of Atomic and Molecular Science, Academia Sinica, P.O. Box 23-166,Taipei, Taiwan 106, R.O.C. Abstract. We have extended the concepts and ideas of optical population transfer to radiationless transitions controlled with strong electromagnetic field. It provides a possibility to realize "non-radiative" analogies to pump-dump process and adiabatic rapid passage.

1.

Introduction

The possibility of the optical control of molecular dynamics using properly tailored pulses has been the subject of intensive studies in the last few years [1-3]. Chirped pulses can selectively excite coherent wave packet motion either on the ground electronic potential energy surface of a molecule or on the excited electronic potential energy surface due to the intrapulse pump-dump process [1]. In addition, they are very efficient for achieving a total population transfer between molecular electronic states by adiabatic rapid passage (ARP) [2]. In [3] we have studied ARP in molecules in solution. We have shown that relaxation does not hinder a coherent population transfer for positive chirped pulses and moderate detuning of the central pulse frequency with respect to the frequency of Franck-Condon transition. In the present paper we extend the concepts and ideas of population transfer developed for optical transitions to non-radiative transitions controlled with strong electromagnetic field. As an example we will consider the long-range electron transfer (ET) that was the topic of active recent research [4]. For mixed-valence transition metal ET complexes the difference AD in permanent dipole moments between donor and acceptor electronic states can be very large (-70D). Interaction of strong electromagnetic field with such a system leads to modulation of its energetic spectrum by the field frequency co [5] and rearranges the configurational surfaces corresponding to different electronic states. This leads to essential change in the ET rate due to its strong dependence on the difference in the state energies. The efficiency of the process depends on parameter z{t) = ADE{t) /(hco), where E(t) is the amplitude of electromagnetic field. Due to large AD, parameter z can exceed 1

502

for the fields E ~10^-10^ V/cm which are smaller than breakdown thresholds of typical solvents in the short pulse regime. In this work we intend to clarify the following issues: Is it possible to realize analogies to pump-dump process and ARP for the long-range ET controlled with strong electromagnetic field? Can one achieve a total population transfer to the acceptor state, using the analogy to ARP? If yes, what are appropriate parameters of such a field? The objective of this paper is to answer all these questions.

2. Extending concepts and ideas of optical population transfer to long-range electron transfer systems Our starting point is an analogy in description of a donor/acceptor system with two electronic states and direct optical transition in two-state system excited by a phase-modulated pulse. Here we only trace out the derivation, details can be found in [7]. Relaxation in the system is treated as a diffusion on electronic potential energy surfaces [3] which is characterized by the second moment 02^ and correlation time TS of stochastic disturbances in nuclear motion. Using the interaction picture [5] and imposing the A^-photon resonant condition, i.e. only Nco be close to the frequency of the transition 1 ^^ 2 (A=l,2,...), we have obtained equations for the components of the pseudospin vector. In these equations in generalization of the Rabi frequency Q = D^2^(^)/^ for an optical transition we have introduced the generalized Rabi frequency Q. j^ {t) ^-(21 fi)J ^ {z{t))R[2^ where J^ is the Ath-order Bessel function, ^j2 ^V\2-fiNcoD^2 I^^ is ^^ effective operator for the interaction between electronic states 1 and 2, Vu and D12 are static transfer coupling and dipole transition moment, respectively, between localized diabatic states 1 and 2. This operator describes both the radiationless (Fi2^0) and radiation (Di27^0) transitions 1 ^ 2 , and the interference between them. With Q in place of Cl^{t), the equations for the components of the pseudospin vector obtained here coincide with Eqs. (9) of [3] for A=l. Adiabatic rapid passage. For a strongly chirped pulse with duration much larger than that of the transform limited one [3, using the generalized Rabi frequency, one can extend the ARP criteria for a two-level system [3,6] to the following

T~^, \do^(t)/dt\«\n^(tf

(1)

Here co(t) is the instantaneous pulse frequency, T' = {'cja2^^'^ is the irreversible dephasing time of the electronic transition [3], and doo/dt^\/0'\co\ where 0"(co)= 0"(v)/(47rf is the chirp rate in the frequency domain [1,3]. From these inequalities one can obtain approximate estimates for ARP: 7' >>50 fs and 0'\v) »8-10^ fs^ for Vu =100 cm\Du=0, and N=\, bearing in mind that Jj max=0.5815. Pump-dump process. An effective intrapulse pump-dump process takes place at comparable populations of both electronic states. It gives the following estimate: 1(2/^)^12 ^A^(z(0)l^=l/|"(^)l whenDi2=0. Therefore, for F^^lOOcm"' and A/^=l an intrapulse pump-dump process can be realized when 0"(v)>8-10"* fs^.

503

3.

Results and Discussion

Fig.l shows the calculation results of the acceptor state population «2 as a function of 0"(v) after the completion of pulse action for different values of relaxation time Is, and different relaxation models (see [3]). The figure relates to one-photon (7V=1) light-induced radiationless (Z)i2=0) transition. The values of parameters were the followings: ^12=100 cm"\ AD=70D, (72s^^^= 560 c m \ parameters of the initial transform-limited (non-chirped) pulse /po=10 fs, E^ =7.5-10^V/cm. i.a.

O.ffl"

n. o.d ' 0.4^ 0.2]

o.oL 1.0i.. O.ffl

o.d 0.4J'' 0.2J 1.0-j 0.8^ 0.6^ 0.4-]

0.24 0.0^ -4.0x10-2.0x10'

2.0x10' 4.0x10'

(()"(v), f s '

Fig. 1. Acceptor state population n2 after the completion of the pulse action as a function of 0"(v) for the total (solid lines) partial relaxation (dashed lines), and relaxation-free (dotted lines) models. The correlation time T^ =lps (a), lOps (b), and lOOps (c) As is seen from Figs.lb,lc for moderately large |0"(v)|~( 1.2-2.0)-10^ fs^, the acceptor state population «2 reaches the value, which is close to its maximum that stands out above 0.9 for the relaxation-free model. Such a behavior corresponds to ARP criteria (1). The further increase in |<^"(v)| causes the value of «2 to decrease slightly for models with relaxation, since one of criteria (1): T'<\(l/h)\V\2'J^{z(t))t is satisfied not so well. When Ts=lps (r-45fs) the last criterion ceases to be ftilfilled, therefore Fig.la shows a situation corresponding to the intrapulse pumpdump process.

References 1 2 3 4 5 6 7

504

G. Gerullo, C.J. Harden, Q.Wang et al., Chem. Phys. Lett, Vol. 262, 362, 1996 N.V.Vitanov, THalfmann, B.W.Shore et al, Annu.Rev.Phys.Chem.,yol52,763, 2001 B.D. Fainberg and V.A. Gorbunov, J. Chem. Phys. Vol. 117, 7222, 2002 Y. Dakhnovskii and R.D. Coalson, J. Chem. Phys. Vol. 103, 2908, 1995 B.D. Fainberg, Opt. Spectrosc. Vol. 41, 558, 1976 B.D. Fainberg and V.A. Gorbunov, in Femtochemistry and Femtobiology: Ultrafast Events in Molecular Science, Edited by M. Martin and J. Hynes, Elsevier, p. 131, 2004 B.D. Fainberg, V.A. Gorbunov, and S.H.Lin, Chem. Phys., 2004, in press

Teaching lasers to twist molecules G. Vogt, G. Krampert, P. Niklaus, F. Santoro, G. Gerber Physikalisches Institut, Universitat Wurzburg, D-97074 Wurzburg, Germany Email: [email protected] Abstract. We report on optimal control of the photoisomerization reaction of 3,3-diethyl2,2-thiacyanine iodide (NK88) dissolved in methanol. Enhancement as well as reduction of therelative yield of cis- to trans-isomexs are achieved.

1.

Introduction

Over the last decade remarkable theoretical and experimental progress was achieved in the field of optimal control of chemical reactions [1-5]. In this multiparameter control scenario, coherent electric fields best suited for solving the control task are found by the quantum system itself in an automated learning loop. The cis-trans isomerization has attracted much attention because of its importance in chemistry and biology (e.g. primary step of vision). An intensely investigated class of molecules that exhibit cis-trans isomerization are symmetrical cyanines (see reference [7] for a summary). The cyanine molecule NK88 exists in a transand in a cw-configuration with absorption maxima at 420nm and 454nm respectively. Quantum-chemical calculations in our group of the absorption maxima for the two isomers in comparison to the experimentally observed ground-state spectrum show that the thermodynamic stable isomer of these molecules has a fm/z5-geometry, same to similar cyanines. Under roomtemperature conditions the concentration of the unstable cw-isomer is negligible. This can be seen by comparing our measured ground-state absorption spectrum of the dissolved NK88 with the available literature-spectra for both isomers [7].

C5xi? reaction coordinate

Fig. 1. (a) Molecular structure of the two isomers of NK88. (b) Simplified potential energy A simplified scheme for the photoisomerization process of the short-chain, symmetrical cyanines assumes only one reaction coordinate, namely the twist about the C=C double bond (s. Fig 1). This simple model views the shape of the ground-state energy surface as a double-minimum potential and the first excited

505

state surface to be barrierless [8]. The absorption of a 400nm photon transfers the stable trans-isomoT from the ground state to the first excited state s^. From there, it reaches a twisted molecular configuration. Through a conical intersection it can either relax back to the trans- or reach the cis-gronnd state. We excite the molecule with frequency doubled spectrally phase-shaped 800nm femtosecond laser pulses, that are thus shaped in spectral phase and amplitude. To probe the reaction fs-whitelight radiation is used, that is centered around 400nm.

2. Results and Discussion The pump-probe transients as a function of time and relative transmission (AT/T), recorded at 400nm and 460nm (s. Fig. 2a) are for large delay times mainly affected by one of the two isomers. The pump-probe signal at 400nm for larger delay-times is proportional to the amount of ^raw5-molecules, that were initially excited, while the curve at 460nm measures the fraction of molecules that actually undergo trans-cis isomerization. The ratio of the probe signals therefore reflects directly the isomerization efficiency (i.e., quantum yield). The more complicated dynamics at early pump-probe delay times originate from stimulated emission and excited-state absorption of the probe-laser pulse to higher lying electronic states. -o- unmodulated pulse 3 1 .b 4-»-shaped _». shapedpulsePuj?|/Si,*v^

400nm

5

10 15 time delay [ps]

5 10 15 20 25 30 generation

Fig. 2. (a) Transients recorded for 400nm pump and 460nm as well as 400nm probe, (b) Maximization and (c) minimization of the cis/trans ratio as a function of generation. Filled dots represent results obtained with shaped laser pulses, open dots show the reference signal obtained with an unshaped laser pulse. On the y-axis the achievable ratio changes for intensity (dark gray) and quadratic spectral phase (light gray) variation are shown. The goal of our experiment was to demonstrate that adaptive femtosecond pulse shaping is able to control the cis-trans isomerization of a complex molecule in the liquid phase. Therefore the pump pulse is sent through a femtosecond pulse shaper capable of producing complex laser pulse shapes. An automated "learning loop" is then employed, wherein an experimental feedback signal from the physical control object itself guides the evolutionary algorithm to find laser fields optimized for the control task. In our case, we have chosen as feedback signal the ratio of ci^-isomers in its ground state after the photoisomerization process to the amount of initially excited trans-isomtxs (i.e., the relative reaction yield). To 506

determine these quantities, we recorded the transient absorption signal at the two wavelengths of 400nm and 460nm at a given specific pump-probe delay time. The pump-probe delay time during the optimization experiment was set to 20ps, because there both transients show only small changes. The ratio between the two signals can only negligibly be changed by the time-shift of the tailored pump pulse. Additionally, an appreciable influence of stimulated emission or excited state absorption is limited to very early pump-probe delay-times and therefore is negligible at a delay time of 20ps. In order to avoid very small signals with low signal-to-noise ratio to enter into the fitness function and thereby to cause physically meaningless high values of the cis/trans ratio a suitable discriminator is used as a lower threshold. This discriminator is given by the lowest measurable denominator for our signal-to-noise ratio. By employing the automated learning loop, we find laser pulse shapes, which enhance or reduce the isomerization reaction. The effect of the optimized pulses on the ratio is much higher than the additionally performed single parameter experiments of chirp and pulse energy variation (s. Fig. 2b/c). The evolution of the cis/trans ratio as a function of generation is shown in Fig. 2b/2c for the two cases of enhancement and suppression of isomerization. In order to monitor the stability of the experimental conditions we applied an unshaped laser pulse after each generation. The cis/trans ratio for this is essentially constant over the time of the optimization (open circles), while it changes significantly for the two cases of minimization and maximization. Hence our experimental results clearly demonstrate that optimized laser pulses can steer the isomerization reaction in both directions. With tailored fs-laser pulses preselected conformational changes of large molecules in the liquid phase can be driven.

3.

Conclusions

In summary, we have demonstrated that adaptive femtosecond pulse shaping is able to control isomerization reactions of a complex molecular system in the liquid phase. By optimizing the ratio of the two isomers in an optimal-control experiment, we demonstrate that isomerization reactions can either be enhanced or reduced. The obtained optimization results show that adaptive femtosecond pulse shaping can by applied to many challenges in chemical, biological or medical research where isomerization reactions are of vital importance.

References 1. 2. 3. 4. 5. 6. 7. 8.

R. S. Judson and H. Rabitz, Phys. Rev. Lett. 68, 1500, 1992. C. J. Bardeen et al.., Chem. Phys. Lett. 280, 151, 1997. A. Assion et al.. Science 282, 919, 1998. T. Brixner, N. H. Damrauer, P. Niklaus, and G. Gerber, Nature 414, 57, 2001. J. L. Herek et al.. Nature 417, 533 2002. N. Dudovich, D. Oron, and Y. Silberberg, Nature 418, 512, 2002. Y. H. Meyer, M. Pittman, and P. Plaza, J. Photochem. Photobiol. A 114, 1,1998. U. Aberg et al., Chem. Phys. 183, 269 1994. 507

Quantum control of a chiral molecular motor driven by linearly polarized laser pulses Masahiro, Yamaki^ Kunihito, Hoki^, Yukiyoshi Ohtsuki^ Hirohiko Kono^ and Yuichi Fujimura^ ^Department of Chemistry, Graduate School of Science, Tohoku University, Sendai 980-8578, Japan E-mail: [email protected] ^Department of Chemistry, University of Toronto, Toronto, Ontario M5S 3H6, Canada Abstract. We present a theoretical study on quantum control of a chiral molecular motor driven by linearly polarized ultrashort laser pulses. Electricfieldsof laser pulses to drive the motor in desired directions are designed using a quantum control method.

1. Introduction Design of molecular motors is one of the hot subjects in molecular nanoscience [1]. Molecular motors driven by thermal, chemical or electromagnetic forces have been proposed. Laser is an effective driving force because their control can be carried out within ultrashort time regimes. We have recently proposed models of chiral molecular motors that can generate unidirectional internal rotations ignited by linearly polarized laser pulses [2, 3]. The molecules consist of two parts, fixed scaffolding and moving part that has dipole moment enough to interact with electric fields. Chirality of molecules whose potential energy surface (PES) of internal rotation has asymmetric character makes motors rotate unidirectionally. Rotational wave packets go up to the PES of the gentle slope. In this paper we report a theoretical treatment of unidirectional motions of a chiral molecular motor on the basis of a quantum control theory, and clarify how intuitive or unintuitive rotation can be induced using linearly polarized laser pulses.

2. Control of unidirectional rotations of molecular motor We set the initial state to the ground state \n = 0) of a molecular motor, where {!«)} are eigenstates of molecular Hamiltonian HM = - " 1 / ^ + ^(^)» and {Sn} are eigenvalues. Our purpose is to control the direction of the internal rotation a. So we consider signs of expectation values of angular momentum operator, defined as 4 = ~^^J^' Eigenvalues of ^o for quantum number ±m are degenerated corresponding to left / right handed rotation except for m = 0. On the other hand, eigenvalues of molecular Hamiltonian, e„, is not degenerated because of the existence of potential energy V{a). If we choose a pair of eigenstates, \2n-l) and 12n), having higher energy than that of the potential threshold, the difference of the eigenvalues of the pair becomes negligible, and the angular momentum eigenstates

508

I m) can be written in terms of a linear combination of a quasi-degenerated pair of molecular eigenstates {\n)}. The direction of rotation is controlled by the phase difference of the superposition. We now design electricfieldE{t) to transfer a rotational wave packet from the initial state 10) to the target state \m) using a locally optimized control method. In a local control method, E{t) is determined from E{t) = -2Alm{iif{t)\Wtl\xir{t))

(1)

where A is a regulation parameter of the laser intensity, and fl{a) is the dipole moment of motor. W is a target operator whose expectation value gives the maximum value at afinaltime, v^(^) is state vector that is obtained by solving time-dependent Schrodingerequation, ih-§^\i/{t) = [HM-tl{(x)-E{t)]\i/{t),

3. Results and discussion We choose a molecule, 2-oxabicyclo[2.2.1]heptan-7-ol as a simple model system of a real quantum molecular motor shown in Figure.la. This molecule is assumed to befixedon a surface. The 0^-C^ bond is the axis for internal rotation, and rotation of hydroxyl group -O^H^ around it is the motor action whose coordinate denoted by a. Figure.lb shows the potential energy function V{a), and Figure. Ic shows (jc, y, z) components of the dipole moment function |i (a). We set both angular momentum with plus sign and that with minus sign as targets. The former corresponds (a) to the intuitive rotation, and the latter to the unintuitive rotation. Figure.2a(b) shows the results of the quantum control under the target condition of m = 9 (-9). Upper panels in Figure.2 show expectation values of angular momentum operator £{t) = n-, 400 0 (V/'(r)|4lV^(0)' ^^d lower panels 1 show the locally controlled fields 1^ -400 obtained. In this calculation, a lin-800 early polarized electric field is assumed. Notice that the time lag between pulse 1 and pulse 2 in a 2.0 Figure.2a is about 19 ps, which is 0.0 Mice) >"(«), longer than that of in Figure.2b. -2.0 H This difference in time delay is 0.0 2.0 0.5 1.5 1.0 expressed as ^fi where Ae cora [Tcrad] responds to energy difference between eigenstates, n = l3 and 14, and Ae = e^ - e^ = 0.90 cm~^ Fig. 1: (a) A model of molecular motor (b) PoBefore pulse 2 comes, the angular tential energy function (c) Dipole moment funcmomentum £{t) in Figure.2a turns tions

509

10

20 t[ps]

(a)|+m) as a target

10

20

40

tips]

(b) \—m) as a target

Fig. 2: Expectation values of angular momentum operator i{t) (upper panels) in the locally controlled electricfieldsE{t) (lower panels). sign from minus to plus. This means that the direction of internal rotation can be controlled by changing the time delay between pulse 1 and pulse 2. This control mechanism can be explained using a four-state model in which two lower states are quasi-degenerated, and two upper states are degenerated.

4. Conclusion Optimal electric fields of linearly polarized laser pulses to drive a chiral molecular motor in desired directions are designed using a local control method. Rotational wave packets created in the ground electronic state of a chiral molecule were controlled in picosecond time scale. Another method for control of rotational wave packets via an electronic excited state can ignite molecular motors within femtosecond time scale [3]. Acknowledgments. This work was partly supported by a Grant-in-Aid for Scientific Research (No. 15550002) from the Ministry of Education, Culture, Sports, Science and Technology of Japan.

References 1 J. -P. Sauvage, editor. Molecular machines and motors. Springer, Berlin (2001); N. Koumura, R.W.J. Zijlstra, R.A. van Delden, N. Harada, and B.L. Feringa, Nature 401, 152 (1999); J. Vacek and J. Michl, Proc. Natl. Acad. Sci. USA, 98, 5481 (2001). 2 K. Hoki, M. Yamaki, S. Koseki, and Y. Fujimura, / Chem. Phys. 118, 497-504 (2003); K. Hoki, M. Yamaki, S. Koseki, and Y. Fujimura, J. Chem. Phys. 119, 12393-12398 (2003). 3 K. Hoki, M. Sato, M. Yamaki, R. Sahnoun, L. Gonzdlez, S. Koseki, and Y. Fujimura, J. Phys. Chem. B. 386, 248-253, (2004).

510

Numerical Synthesis of Optimal Laser Pulses for Manipulating Dissociation Wave Packets of I2" in Water Yukiyoshi Ohtsuki^ Yoshikazu Nishiyama, Tsuyoshi Kato , Hirohiko Kono and Yuichi Fujimura Department of Chemistry, Graduate School of Science, Tohoku University Sendai 980-8578 Japan ' E-mail: [email protected] ^ E-mail: kato@ mcl.chem.tohoku.ac.jp Abstract. A linearized optimal control method in combination with mixed quantum/classical molecular dynamics (MD) simulation is used for numerically investigating the possibility of controlling dissociation wave packets of l2' in water.

1.

Introduction

The number of successful quantum control experiments, many of which use optimal laser pulses designed by learning algorithms, has been increasing steadily. In condensed phases, however, it is still unclear how efficiently the dynamics can be coherently controlled because of the presence of relaxation [1]. We thus conduct a case study of the manipulation of dissociation wave packets of I2' in water based on optimal control theory as the strong solute-solvent interaction corresponds to the most unfavorable condition [2]. Assuming weak field regimes, we have a linearized pulse design equation that is expressed in terms of a molecular response function. This simplification enables us to employ realistic treatments of condensed-phase dynamics, for example, mixed quantum/classical MD simulations. Using this approach, we design optimal laser pulses, aiming at the creation of localized dissociation wave packets of I2'. The control mechanisms are examined using a statistically averaged effective dissociation potential.

2.

Theory

We define an optimal pulse so that it maximizes the expectation value of a target operator, W , that specifies a physical objective subject to minimal pulse fluence. Restricting ourselves to a weak field regime in which the optical processes are approximated by the linear response term, we obtain a linearized pulse design equation expressed in terms of a molecular response function. The time evolution

511

is calculated by means of the mixed quantum/classical MD simulation, in which the vibrational motion of I2" is treated quantum mechanically, while the other degrees of freedom are treated classical mechanically.

3. Results and Discussion In our mixed MD simulation [2], we employ a cubic main cell that contains one I2", a counter cation, Na^, and 254 water molecules modeled by SPC/E. The density and the temperature are set to 1.023 g/cm^ and 300 K, respectively. The photdissociation dynamics of I2' is approximately described by a one-dimensional, two-electronic state model. Because of the low dissociation quantum yield of I2' in water -0.12, we first collect dissociation samples using classical MD simulations according to the algorithm shown in Fig. 1. initial configuration thermostat (T=300 K) construction of initial canonical ensemble

classical MD

time evolution of the total system using classical MD every 1 ps i NO

YES

mixed MD

time evolution of the total system using the mixed quantum/classical MD M^ store sample next trial (after 5 ps)

Fig. 1. Schematic illustration of the algorithm for collecting dissociation samples

At every 1 ps, we check whether the configuration is dissociative or not. To see this, we replace the ground-state potential of !{ with the excited-state potential and calculate the time evolution for 350 fs. If the intemuclear distance of I2' exceed 6.0 A, the sample is regarded as a dissociation sample. Once we find a dissociation sample, the whole system evolves for 5 ps, during which we do not collect dissociation samples, in order to remove statistical correlation between the dissociation samples. We collect 30 dissociation samples, of which dissociation quantum yields are 8.8%. Using dissociation samples, we design optimal laser pulses, aiming at the creation of spatially localized wave packets. Before discussing numerical results, we introduce a statistically averaged dissociation potential of !{ (effective potential) that consists of the dissociation potential of the excited I2' and the statistical average of the mean field potential originating from the heat bath. The effective potential is expected to extract the characteristic properties of the ensemble. We also design an optimal pulse using the effective potential, which is called an effective optimal pulse.

512

0.03 H

100

200

300

time (fs)

Fig. 2. (a) Optimal laser pulse, (b) effective optimal laser pulse, and (c) time evolution of the normalized target population calculated by (a) (thick solid line), by (b) (thin solid line) and by the Franck-Condon wave packet (dashed line). Fig. 2 shows the results when the target region is chosen as [5.0 A, 5.2 A]. This intemuclear distance is greater than the charge separation length of 5 A. Comparing the peak values of the control achievements, the designed optimal pulse lead to nearly two times greater control achievement than that obtained by the Franck-Condon wave packet. The effective optimal has much simpler structure than the corresponding optimal pulse; however, the effective optimal pulse leads to almost the same control achievement as that calculated by the optimal pulse.

4.

Conclusions

We have applied a linearized optimal control method for controlling the photodissociation dynamics of I2" in water, the time evolution of which is calculated by a mixed quantum/classical MD simulation. Optimal laser pulses are designed using an ensemble of dissociation samples. Their control mechanisms can be interpreted using a statistically averaged effective potential. Acknowledgements. This work was partly supported by a Grant-in-Aid for Scientific Research on Priority Areas, "Control of Molecules in Intense Laser Fields" from the METX of the Japanese Government.

References 1 Y. Ohtsuki, J. Chem. Phys. 119, 661, 2003 and references therein. 2 Y. Nishiyama, T. Kato, Y. Ohtsuki and Y. Fujimura, J. Chem. Phys. in press, 2004.

513

Molecular State Reconstruction by Nonlinear Wave Packet Interferometry Travis S. Humble and Jeffrey A. Cina Department of Chemistry, University of Oregon, Eugene OR 97403, USA E-mail: [email protected] Abstract. We investigate the reconstruction of optically prepared vibrational wave packets using nonlinear wave packet interferometry. Simulated results for a model photo-dissociative diatomic demonstrate the technique's effectiveness in identifying dynamics induced by shaped laser pulses.

1.

Introduction

Recent achievements in the control of chemical reactions using adaptive laser-pulse shaping strategies [1] pose the challenge of how to identify the ultrafast photoinduced molecular dynamics [2]. Optimization of an incident waveform neither directly elucidates the light-induced reaction mechanism nor identifies the optically prepared initiating state, especially when the state propagates under a poorly characterized Hamiltonian. One means for characterizing an optically prepared target state in the absence of Hamiltonian information is reconstruction of the time-dependent probability amplitude (as opposed to probability density), which provides a complete picture of the molecule's photoinduced dynamics [3-5]. Nonlinear wave packet interferometry (WPI) [4,5] uses a pair of phase-locked pulse-pairs to create a linear superposition of nuclear wave packets in an excited electronic state (/). In a two-color experiment the first pulse-pair accesses transitions between the ground state (g) and an intermediate electronic state (e), while the second pulse-pair d r i v e s / ^ g transitions. The overlap (ref42i Itar^) between a target state shaped by the third pulse, |t^,^^^^^(M<)-:.)^--A3/r^^^/.|^^^^^

(1)

and a reference state created by the first, second, and fourth pulses, \vQf,,,)

= -e^^'^^'''^^-'^^^P/'e-'"^^^^^^

(2)

is isolated by combining measurements of the total /-state population taken at different values of the phase-locking angles 02i and 0^3 [5,6]. The transfer of nuclear amplitude between electronic states a and b via t h e / pulse in the rotating wave approximation is denoted by the pulse propagator P"'' [7], (l).{(x))-ojt. is the spectral phase of t h e / pulse, and t ^^ = t^ -t,^.

514

Measurements of {vQf^^^ |tar3) made with fixed ^43 and varied ^21, ^32 yield a set of simultaneous equations z=Rt. The vector z stores the overlaps, and the reference matrix R, whose rows comprise the conjugate wave functions of (2) at different delays, is calculated from prior knowledge of the g and e electronic states. The unknown target wave function t, which propagates solely on the/state, is determined using singular value decomposition of R and construction of its pseudo-inverse. The fidelity of the reconstructed vector r, which minimizes the norm and residual of all possible solutions [5], is quantified as / = |t^T|/(||t|H|r||), and lies between 1 (good) and 0 (bad). For a thermally populated mixture, a weighted sum of overlaps arising from different initial states is measured. Reconstruction proceeds as before but with the reconstructed vector partitioned to characterize the different target states, and the cumulative fidelity given as a weighted sum of the individual fidelities.

2.

Simulations

To demonstrate reconstruction we simulated the nonlinear WPI signal for a model nonrotating photodissociative diatomic, using equal-frequency displaced harmonic oscillators for the ground and intermediate states (frequency 2JTC(250 cm''), mass 63.5 amu, and displacement 0.0614 A) and an exponentially decaying final state with Franck-Condon energy /zc(1000 cm"') and length scale 0.1096 A. The first, second, and fourth pulses are transform-limited Gaussians (5 fs FWHM intensity), vertically resonant with their respective electronic transitions. The third pulse is a vertically resonant Gaussian (20 fs FWHM) with a linear frequency chirp of -144 fs^. The delays ^21 and ^32 are scanned through a ground-state vibrational period (Xg = 133.4 fs) while ^43 = 25.4 fs. At 270 K, the first five vibrational levels account for over 99% of the total population and the resulting interferogram, calculated using grid-based propagation techniques, is shown in Fig. 1.

Re 0.75h

0.5 0.25

Fig. 1. Calculated interferogram for the model dissociative system. Solid (dashed) contours are separated by positive (negative) increments of 1/10 the interferogram maximum.

515

We apply our reconstruction procedure using the calculated interferogram with 5% uncorrected Gaussian noise added. The first three reconstructed states and their associated target states are shown in Fig. 3. The total fidelity is 0.922. Note that individual-state reconstruction fidelity decreases with initial population;

0 ,2 position (A) Fig. 2. Target and reconstructed states obtained with the noisy signal. Solid (dashed) lines are amplitude (phase). The target (reconstructed) state is represented by light (heavy) lines. The initial populations and individual fidelities are stated for the (left to right) ground, first, and second excited initial states.

accurate reconstruction becomes impossible for states with populations much less than the noise level. Reconstruction is also limited by the finite bandwidth of the first, second and fourth pulses, which confines the spatial range over which reference states can be effectively prepared. State reconstruction for systems with multiple degrees of freedom, including rotations, is currently being studied [8].

References R.J. Levis, G.M. Menkir, and H. Rabitz, Science, 292, 709-713 (2001); M. Bergt, T. Brixner, B. Keifer, M. Strehle, and G. Gerber, J. Phys. Chem. A, 103, 10381-10387 (1999); J.L. Herek, W. Wohlleben, R.J. Cogdell, D. Zeidler, and M. Motzkus, Nature, 417, 533-535 (2002) C. Daniel, J. Full, L. Gonzalez, C. Lupulescu, J. Manz, A. Merli, S. Vajda, and L. Woste, Science, 299, 536-539 (2003) A. Zucchetti, W. Vogel, D.-G. Welsch, and I.A. Walmsley, Phys. Rev. A, 60, 2716-2725 (1999); I.Sh. Averbukh, M. Shapiro, C. Leichtle, and W.P. Schleich, Phys. Rev. A, 59, 2163-2173 (1999); H. Stapelfeldt, E. Constant, H. Sakai, and P.B. Corkum, Phys. Rev. A, 58, 426-433 (1998) J.A. Cina, J. Chem. Phys., 113, 9488-9496 (2000) T.S. Humble and J.A. Cina, Phys. Rev. Lett, (in press) P. Tian, D. Keusters, Y. Suzaki, and W.S. Warren, Science, 300, 1553-1555 (2003). Y.-C. Shen and J.A. Cina, J. Chem. Phys., 110, 9793-9806 (1999) This work was supported by the US-NSF and a fellowship (J.A.C.) from the J.S. Guggenheim Foundation

516

Femtosecond coherent spectroscopic study of Zn(II) porphyrin by chirping-controlled ultrashort pulses Min-Chul Yoon, Sung Cho, and Dongho Kim Center for Ultrafast Optical Characteristics Control and Department of Chemistry, Yonsei University, Seoul 120-749, Korea E-mail: [email protected] Abstract. Impulsively photo-induced coherent vibrational wave packet dynamics of Zn(II) porphyrin in the electronic ground and excited states have been investigated. Chirped ultrashort pulses enable us to control generation and propagation of wave packet on individual surface. The femtosecond coherent vibrational spectroscopy is based on the ultrashort excitation pulse which leads to a generation of a wave packet. The oscillatory behavior of the wave packet is attributed to vibrational coherence (wave packet motion), whose kinetics is usually observed by fluorescence up-conversion and pump-probe transient absorption signal. The oscillatory feature reflects the coherence among the vibrational states induced by the ultrashort broadband excitation, indicating that the vibrational coherence both in the ground and excited state contributes to the oscillatory component [1-4]. Temporal profiles of one-color pump-probe transient absorption (TA) and fluorescence up-conversion signal of Zn"-5,15-diphenylporphyrin (Zn DPP) in toluene using transform-limited (TL) pulses show exponential decay due to S2-S1 internal conversion with a time constant of 1.6 ps and superimposed prominent oscillatory components induced by coherent wave packet motions, which damp within ~2 ps (not shown). Using fast Fourier-transform (FFT) and linear prediction of singular value decomposition (LPSVD) analyses, frequency domain spectra are retrieved with dephasing time and the relative phase. It is noted that two remarkable vibrational bands are observed in the low frequency region, which have the center frequencies of 314 and 394 cm"^ with the dephasing time constants of 1.8 and 1.1 ps, respectively (Fig. 1). In ground state resonance Raman spectrum with 441.6 nm excitation, two strong vibrational bands are appeared at 314 and 392 cm with almost equal intensities which belong to the totally symmetric Ag vibrational modes activated via the Albrecht ^-term scattering. With a help of semi-empirical PM3 calculation and normal mode analyses, the vibrational mode at 314 cm"^ can be assigned to the (t)io mode which is mainly contributed by phenyl group translational vibration as well as a small amount of pyrrole rotational local vibration. The other vibrational mode at 392 cm"^ can be assigned to the Vg mode which is associated with Zn-porphyrin stretching without peripheral phenyl vibrations, which is called porphyrin ring breathing mode. This band is always strongly activated with B-band excitation in the resonance Raman spectrum due to the 71-71* transition of porphyrin macrocycle (See Fig. 2 inset). Temporal profiles of TA signals of Zn^DPP in toluene by controlling the chirp

517

i O

100

200

300

400

500

200

300

400

500

200

300

400

500

Frequency (cm"^)

Fig. 1. Fourier-transformed power spectra from the oscillation components of transient absorption signals followed by control of the chirp with excitation at (a) 405, (b) 413 and (c) 420 nm.

and excitation wavelength femtosecond pulses are obtained and their frequency spectra are retrieved by using FFT technique in Fig. 1. It is observed that the amplitude of each decay component depends on degree of the chirp and the pump/probe wavelength (not shown). As the pump/probe wavelength becomes longer, the relative amplitude for the fast decay component with the time constant of 1.6 ps increases, since the stimulated emission contribution to the overall TA signal becomes larger as the probe wavelength is closer to the S2 emission band. The vibrational coherence in the electronic ground state is believed to be generated through the mechanism similar to the impulsive stimulated Raman scattering process [2]. The observation that, for a given pulse width, positively chirped (PC) and negatively chirped (NC) pulses yield very different oscillation signals, hence the different FFT spectra, indicates that these effects are not just due to the temporal broadening of the excitation pulses. Fig. 2 (a) and (b) show the intensity dependence of two normal modes as a function of degree of chirping with three different excitation wavelengths. The two vibrational modes appearing at 314 and 394 cm'^ exhibit different dependence on the chirping and energy of the excitation pulses. The (j) 10 mode at 314 cm"^ is slightly more enhanced by the NC pulse excitation at 413 and 420 nm compared with that at 405 nm which shows negligible dependence on the chirping control. In general, the intensity of vibrational mode in the FFT spectrum is correlated with the wave packet oscillations on the PES. It has been reported that the effective oscillation is due to the localized wave packets by the NC pulse excitation in other work [2, 4]. In this regard, the enhanced 314 cm' mode by the NC pulse excitation at 413 and 420 nm can be explained by molecular characteristics such as a displacement between the S2 and So PES. In the slightly NC pulse excitation, the most localized wave packets are created in the ground state due to a small displacement between the S2 and So potential energy surfaces (PES), which results in the enhanced 314 cm"^ mode in the FFT spectrum. On the other hand, the dependence on the chirping control is not so manifest for the Vg mode at 394 cm'^

518

402

- 405 nm -413nm - 420 nm

(a)

400

j

^

i

(c)

398 396 394 392 i 390 ] 388 386 320

(b)

1 j

\ . _ _; A-A ;

3

•

. p.- Q

...

^ "^^ (d)

318 316

,n—U—D--

314

1

....Q?^^_^^^o*-^o---

312

''n

310

^x

Xb—o

-800 -600 -400 -200

0

200 400 600 800

Quadratic Phase Term
308

i A-A

306 304 -800-600-400-200 0

200 400 600 800

Quadratic Phase Term
Fig. 2. The variation of peak intensities of (a) vg mode, (b) (j)io mode and center frequencies of (c) vs mode, (d) ^\o mode with different photoexcitation wavelength as a function of the chirp degree. Horizontal dashed lines represent the vibrational frequencies in the resonance Raman spectrum.

presumably due to a negligible displacement between the S2 and So PESs as well as its weakness in intensity, which originates from the fast damping process (1.1 ps) and the limited time-resolution of our present set-up. The frequency changes for these two vibrational modes are presented in Fig. 2(c) and 2(d) as a function of wavelength and chirping of the excitation pulses. Upon photoexcitation at 420 nm these two modes are always observed at lower frequencies and slightly shifted to even lower fi-equencies as the excitation pulses are more positively chirped. This trend indicates that the excited state contribution to the wave packet dynamics is enhanced in the PC pulse excitation at the red-side of the B-band of Zn DPP. Since the probe wavelength of 420 nm is close to the S2 emission maximum, the SE contribution is larger than the other wavelengths of 405 and 413 nm, which increases the contribution of the excited state to the overall TA signals. Thus, the control of chirping in the excitation pulses leads to a different degree of the wave packet enhancement created either in the ground or in the excited state, probing the structural changes occurring in the excited state. This illustrates the feasibility of the active control of the wave packet dynamics by the chirping of optical pulses.

References 1 2 3 4

K. Misawa and T. Kobayashi, J. Chem. Phys. 113, 7546-7553 (2000). C. J. Bardeen, Q. Wang and C. V. Shank, Phys. Rev. Lett. 75, 3410-3413 (1995). F. Rosea, A. T. N. Kumar, X. Ye, T. Sjodin, A. A. Demidov and P. M. Champion, J. Phys. Chem. A 104, 4280-4290 (2000). M.-C. Yoon, D. H. Jeong, S. Cho, D. Kim, H. Rhee and T. Joo, J. Chem. Phys. 118, 164-171 (2003).

519

Phase analysis of vibrational wavepackets in the ground and the excited states in polydiacetylene Mitsuhiro Ikuta^ Yoshiharu Yuasa\ Tatsumi Kimura^, Hiroo Matsuda^ and Takayoshi Kobayashi^ ^ Department of physics, Graduate School of Science, University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo, 113-0033, Japan E-mail: [email protected] ^ National Institute of Advanced Industrial Science and Technology, Tsukuba Central 5, 11-1 Higashi, Tsukuba, Ibaraki, 305-8565, Japan Abstract. We have observed the real-time vibration signals due to the wavepacket motions in the ground state and the excited state by sub-5fs spectroscopy. All the three modes in the self-trapped states were observed just after excitation indicating that initial relaxed state is not a butatriene-like but has an acetylene-like structure.

1.

Introduction

The polymers are of interest not only in technological application but also in fundamental physics because they have large optical nonlinearities due to the formation of localized nonlinear excitations such as solitons, polarons, and a selftrapped exciton (STE) formed via an exciton-phonon coupling in a 1-dimensional system. From previous extensive studies [1-3], the initial spectral changes and their kinetics after photoexcitation of polydiacetylene (PDA) are explained by the relaxation of a free exciton (FE) to a STE within lOOfs [7,8].

2.

Experimental IMethods

The sample used in the present study is a cast film of blue-phase PDA-3BCMU (poly [4,6-docadiyn-l, 10-diolbis (n-butoxycarbonylmethylurethane)]) on a glass substrate. The pulse-front-matched NOPA (4.7-fs, 5-[xJ pulses at 5kHz) was used for both pump and probe [4-6]. The spectrum of the pulses covers from 520 to 730 nm with a nearly constant phase. Laser pulse energies of the pump and probe pulses are about 35 and 5 nJ, respectively. Difference absorbance (AA(t)) was measured at a pump-probe delay-time (t) from -100 to 1200 fs with every 1-fs step from 540 to 740 nm using a 300 grooves/mum grating monochromator (spectral resolution of about 3.6 nm). All the measurements were performed at 295K.

520

3. Results and Discussion In order to investigate the vibrational mode, we analyzed the pump-probe signal by LP-SVD (Linear prediction by singular value decomposition). By LP-SVD, single or multiple mode(s) of damped oscillation are extracted from the data such as y{t) = Aexp(-t/T)cos(cot-\-6), where A is an initial amplitude, T is a decay time, co is the mode frequency, and 0 is the initial phase. Each vibrational mode of C-C, C=C, and C=C was extracted by LP-SVD from the 200-900 fs data after box frequency filtering of the Fourier power spectrum (1150-1290, 1395-1535, and 2010-2150 cm"^). The amplitude at 300 fs and the absolute initial phase of each mode are shown in Figs. 1(a) and (b), respectively. In Fig. 1(b), the phases from about 1.9 to 2.0 eV and those from about 2.0 to 2.1 eV have opposite signs for all of the three stretching modes. In the case of C=C the phases at 1.95 and 2.05 eV are close to n/2 and -n/2, respectively, corresponding to the peaks in Fig. 1(a). The midpoint around 2.0 eV of the two energies (1.95 and 2.05 eV), the value of the phase is zero. In the cases of C-C and C=C, the phases at 1.95 and 2.05 eV are slightly smaller than n/2, but their frequency dependence have similar features to those of C=C stretching, i. e. the sign of the phase at 1.95 eV is opposite to that at 2.05 eV. The phase spectrum of C=C is positive in the range from 1.99 to 2.01 eV, though in the case of the other two modes the phase at 2.0 eV is close to zero. This energy (2.0 eV) corresponds to the photon energy shifted by the vibrational energy of C=C from 2.2eV of the peak of the probe which is expected to give n/2 phase. Therefore it can be concluded that the phases around 2.0 eV in C=C are shifted to positive due to the contribution of the Raman gain signal. From the fact that the phases at 1.95 and 2.05 eV have opposite sign to each other, it can be concluded that the vibrational-amplitude peaks at 1.95 and 2.05 eV are due to the modulation of the intensity of 1 ^Bu-FE absorption (resonant at 2.0 eV) by the motion of the vibrational wavepacket induced by the stimulated Raman scattering (SRS) in the ground state. The coincidence between the peak position of energy-derivative of AA(t) and the vibration-amplitude peak also supports this conclusion. In addition, the phases of C-C and C=C have opposite sign to that of C=C in the probe-photon energy from 1.95 to 2.05 eV. This means that the vibrational wavepackets of C-C and C=C stretching modes in the ground state start to vibrate to the opposite direction to that of C=C. Therefore the transition energy can be concluded for the first time to start to decrease in C-C and C=C, and it starts to increase in C=C just after excitation by SRS. The peak at 1.78 eV in Fig. 1(a) is considered to be due to the motions of the wavepackets along the excited-state potential surface of the geometrically relaxed 2^Ag state because there is no ground-state absorption. The peak at 1.78 eV in Fig. 1(a) is coincident with the first energy-derivative of AA(t). Therefore this result indicates that not only C-C and C=C but also C=C bonds exist in the excited state [9]. For the differences in the frequencies determined by LP-SVD the three modes of stretching between the ground state and the STE state were found to be less than 10 cm^

521

3

(a)

0.0015-

1 2

to 0.0010-

•S

1

0.0005-

"^^^^^^:^^

0.00001.75

1.80

1.85

^

1.90

1.95

ZOO

2.05

Probe photon energy (eV)

2.10

2.15 Probe photon energy (eV)

Fig. 1. (a) Vibrational amplitude at 300 fs of each mode of normalized transmittance change determined by LP-SVD;1, C-C; 2, C=C; 3, C=C. (b) Initial vibrational phase of each mode of normalized transmittance change determined by LP-SVD; 1, C-C; 2, C=C; 3, C=C.

4.

Conclusions

In conclusion we have separated the ground-state and the excited-state contributions in the real-time vibration signals due to the wavepacket motions from the transmittance change spectra obtained by sub-5fs spectroscopy. The wavepackets of C-C and C=C stretching modes in the ground state were found first time to start to oscillate to the opposite direction to the case of C=C. Also the vibration of C=C stretching mode in the geometrically relaxed 2^Ag state is observed as well as C-C and C=C even after the FE is converted to the geometrically relaxed state within 100 fs. This clearly demonstrates that the full geometrically relaxed butatriene-like structure is not formed, but it can still have configuration with the acetylene-like structure.

References T. Kobayashi, M. Yoshizawa, U. Stamm, M. Taiji, and M. Hasegawa, J. Opt. Soc. Am. B7, 1558, 1990. M. Yoshizawa, Y. Hattori, and T. Kobayashi, Phys. Rev. B. 49, 13259, 1994. T. Kobayashi, A. Shirakawa, H. Matsuzawa, and H. Nakanishi, Chem. Phys. Lett. 321, 385, 2000. A. Shirakawa, I. Sakane, M. Takasaka, and T. Kobayashi, Appl. Phys. Lett. 74, 2268, 1999. A. Baltuska and T. Kobayashi, Appl. Phys. B. 75, 427, 2002. T. Kobayashi and A. Shirakawa, Appl. Phys. B. 70, S239, 2000. M. Yoshizawa, Y. Hattori, and T. Kobayashi, Phys. Rev. B. 47, 3882,1993. M. Yoshizawa, A. Kubo, and S. Saikan, Phys. Rev. B. 60, 15632, 1999. T. A. Pham, A. Daunois, J. C. Merle, J. Le Moigne, and J. Y. Bigot, Phys. Rev. Lett. 74, 904, 1995.

522

Calculating Ultrafast Nonlinear Optical Signals from Molecules in Cryogenic Matrices Mary A. Rohrdanz ^'^, and Jeffrey A. Cina ^'^ ^Department of Chemistry, University of Oregon, Eugene OR 97403, USA ^E-mail: [email protected] ^E-mail: [email protected] Abstract. We are pursuing the calculation of time-resolved coherent anti-Stokes Raman scattering signals from I2 molecules in cryogenic argon using semi-classical Gaussian wave packet dynamics. Evolving Wigner functions are used to interpret the dynamics underlying the nonlinear optical signal.

1. Introduction We are developing techniques for simulating and interpreting ultrafast nonlinear optical signals from molecules in condensed media, starting from a many-body internuclear potential. Our current focus is on time-resolved coherent anti-Stokes Raman scattering (tr-CARS) from l2-doped cryogenic Ar matrices. These experiments prepare and probe vibrational coherences on the ground electronic surface. By comparison between theory and experiment, we hope to gain insight on the effects of a surrounding medium on chromophore dynamics. The tr-CARS experiments most relevant to our work are on gas-phase I2 [1] and I2 in cryogenic Ar and Kr matrices [2]. Tr-CARS experiments involve 3 ultrashort laser pulses resonant between 2 electronic states. The V^ pulse pumps amplitude from the ground electronic state to an excited electronic state. This P^-order wave packet evolves for a fixed time delay (10 to 20 fs) before a 2"^^ pulse dumps amplitude to the ground state. The 2"'^-order wave packet evolves for a variable delay until the 3'"^ pulse re-excites it to the excited electronic state, thereby inducing a 3''^-order molecular dipole. The signal is the time-integrated intensity of the 3''^-order electric field, which is emitted in the phase-matched direction.

2. Theory and Calculations Third-order time-dependent perturbation theory produces an expression for the trCARS signal based on the induced 3''^-order molecular dipole of the chromophore. The expression for the relevant portion of the dipole can be written as the overlap between a 3''^-order ket and a 0^^-order bra,

d,,{t) = -i^imlip^ \u^{-t)u^{t - t,)p,u^{t, - t,)p;ush - OP^u^itAip)

(1)

523

where \yj j is the initial nuclear wave function, (j ^^Ar) governs time evolution on the ground (excited) electronic state, the P. s are pulse propagators [3] describing the effect of the f^ pulse on the nuclear wave function, and t. is the temporal center of the i^^ pulse. The homodyne-detected signal is calculated as signal oc I dridQjT

(2)

J t-.

These calculations are to be performed on I2 substituted into an array of 254 Ar atoms, with pair-wise additive interactions. The wave function is modeled as a 768-dimensional "thawed" Gaussian with cross-terms, and is propagated via Heller's locally quadratic Hamiltonian approximation [4]. This propagation is only valid for short times; however it has favorable scaling properties allowing for calculations on condensed-phase systems. The pulse electric field envelopes are Gaussian and the propagators are evaluated using the semi-classical Franck-Condon approximation: in

\.

s

/

J

(3)

where a is the temporal standard deviation of the electric field, Q is the center frequency, and q)is the optical phase of the pulse. V^g is the coordinate-dependent multi-dimensional difference potential.

3. Results from Model 1- and 2-Dimensional Systems To compare with the Ar-matrix system, calculations have been performed on nonrotating I2 and a non-rotating, T-shaped, 2-dimensional I2-Ar van der Waals complex. Fig. 1 shows calculated VAr signals for arbitrarily short and finiteduration ( a = 5 fs) pulses, with 5 fs between the V^ and 2*"^ pulses and the finiteduration pulses electronically resonant at the ground state minimum. ,n.

I ^ 'i

arbitrar lly s h o r t pulses finite-d u ration pulses

j1 0

200

400

600

800

10 0 0

t i m e d elay. f s

Fig. 1. Tr-CARS signal from I^-Ar The calculations for bare I2 are nearly identical to Fig. 1. Since there is little energy in the I2-Ar mode after the 2"'^ pulse, this mode does not contribute strongly to the signals. We focus solely on the bare I2 calculations in the analysis below.

524

These signals can be analyzed via Wigner distribution plots (Fig. 2). The light dashed curves are phase space potential energy contours for the excited (B) state of I2, and the thin solid curves are the Wigner function for the 0*-order wave function. In the left panel (arbitrarily short pulses), the dashed wave packet is the Wigner function for the S'^^-order wave packet just after the 3'^* pulse propagator, for a 76 fs time delay (a maximum in the signal). This packet has been launched on a trajectory that will send it directly through the 0^^-order wave packet, providing a large time-integrated overlap and resulting in a large signal. The first minimum in the signal is at the 147 fs time delay, and the relevant 3"^-order wave packet is represented by thick solid curves. This 147 fs packet has a positive momentum and will not pass through the 0^^-order wave packet, yielding a smaller signal. Similar analyses can be performed for other delays. Finite-duration pulses change the moments of the S'^^'-order wave packet. The right panel of Fig. 2 shows the 119 fs-delay wave packet just before (dashed) and just after (thick solid) the 3"^ pulse propagator. The pulse propagator shifts the wave packet toward the resonance line; since much of the 2"'^-order wave packet is far from the resonant location, the norm of the 3'^'^-order packet is reduced. This loss of amplitude accounts for the diminished amplitude of the finite-pulse signal. ////////////

mum mmi

///

liiilillllil/////////////// 'lllimillll!!//////////;

imiiii^mmHJ//

iiiiiiiiiii

\\\\\\\\\\\\\\\\\\\^K<^^>>

;MW\\\\\\\\mmxvA\\\\ ;\\\\\\\\\\\\\\\\\\\\\\\\\\\^'^\\;x\

\\\\\\\\\\\\\\\\\\\\Ute'^v^\\ "'

WWWWWWWWWWWWWW.W- \

Position (A)

Position (A) Fig. 2. Wave packets made by arbitrarily short (left), and finite-duration pulses (right)

The next step in this work is to perform calculations on the 768-dimensional system described above. We also plan to develop a reduced-dynamical description based on Redfield theory.

References M. Schmitt, G. Knopp, A. Materny, and W. Kiefer, Chem. Phys. Lett. 270, 9, 1997. M. Karavitis, R. Zadoyan, and V. Ara Apkarian, J. Chem. Phys. 114, 4131, 2001; M. Karavitis and V. A. Apkarian, ibid. 120,292,2004. Yu-Chen Shen and Jeffrey A. Cina, J. Chem. Phys. 110,9793, 1999. Eric J. Heller, J. Chem. Phys. 62, 154^, 1975. This work was supported by the US-NSF and a fellowship (JAC) from the J. S. Guggenheim Memorial Foundation.

525

Real-Time Observation of Phase-Controlled Vibrational Wave-packets in Iodine Molecules Yukinori Sato*'-^, Hisashi Chiba^'^, Masahiro Honda^'-^, Yusuke Hagihara*'^, Katsuhiro Fujiwara^'-^, Kenji Ohmori^'^, and Kiyoshi Ueda^'^ ' IMRAM, Tohoku University, Katahira, Aobaku,Sendai 980-8577, Japan E-mail: [email protected], [email protected] ^ Institute for Molecular Science, Okazaki 444-8585, Japan E-mail: [email protected] ^ CREST-FEMD, JST (Japan Science and Technology Agency) Abstract. A double-pulse sequence of femtosecond laser pulse (100 fs, 616 nm) is used to pump and control the vibrational wave packet on the B state of the molecular iodine. The inter-pulse-delay (control delay) is tuned with about 10 attoseconds accuracy. Time evolution of the wave packet is monitored by a probe pulse (100 fs, 389 nm). The pump and control evolution of the B state wave packet is monitored in real time for the control delay tuned around the second recurrence time of the wave packet.

1.

Introduction

Wave packet (WP) interferometry has been performed for Rydberg WP's in atoms [1,2] and vibrational WP's in molecules [3-5]. The WP interferometry is viewed as the coherent interaction of two WP's formed by each of the optical double pulses (time-domain picture), or as the interference of the two optical pulses filtered by the atomic or molecular transitions (frequency-domain picture). When the two pulses are identical, the time delay between the pulses is a control parameter that determines the phase characteristic of the double-pulse light source. Phase control of a WP is thus possible by transferring the controlled optical phase to the WP. We have developed an attosecond phase modulator "APM" for tuning the time delay Control delay) with an accuracy of about 10 attoseconds and applied it to observe high precision WP interferometry in HgAr [5]. Double-pulse-control of vibrational WP on the B state of gas phase molecular iodine has been investigated by the phase-locked-double-pulse excitation and fluorescence detection method by Scherer et at [3]. We present here a different approach to demonstrate time-domain pictures of the double-pulse pump and control sequence for the vibrational WP formed by the 5 - X transition of the molecular iodine.

2.

Experimental IMethods

An output pulse of a Ti:Sapphire laser (800 nm, 100 fs, 2.2 W, repetition rate 1 kHz,) was divided and directed into two OPGA systems to obtain a pump pulse around 616 nm and a probe pulse around 389nm. The pump pulse was further

526

divided into a double-pulse sequence by the APM [5], which has a Michelson-type double-arm optical arrangement with a mechanical delay stage in one arm and a gas cell placed in another arm. The APM generates a pair of time-separated identical 100 fs pulses at a repetition rate of 1 kHz. The time delay, control delay, between the pulse pair was tuned coarsely by the mechanical delay stage and finely by varying pressure of Ar in the APM gas cell. The pulse pair is directed into a vacuum chamber to cross an effusive beam of molecular iodine carried by Ar gas. The pulse pair was used for pump/control operation on the B - X transition of the molecular iodine. The center wavelength of the pulse pair (616 nm) was selected so that the v' = 3, 4 and 5 vibrational eigenstates of the B state were mainly involved in the B - ^transition. The probe pulse (389 nm, 100 fs) was used to monitor the pump/control time evolution of the B state vibrational WP. The center wavelength of the probe pulse was selected to induce the E<-B transition of the molecule with the Franck-Condon region located around the outer turning point of the B state potential. Every time the WP comes to the time window set by the probe pulse, its probability amplitude was partially transferred to the E state to give the probe-pulse-induced fluorescence. The fluorescence intensity was recorded as a function of thQ probe delay with respect to the timing of the first pump -pulse. The probe delay was mechanically tuned with an optical delay stage.

3. Results and Discussion

*^

^

r ""'%'

""f""

¥

T

i

ti'u.x {HI

Fig. 1. Dependence of thefluorescenceintensity on the probe delay for control delays at 2Tyi\y. Timings of the pump and control pulses are shown by a pair of arrows. Difference in the control delay between (a) and (b) is 1.0 fs which is equal to a half of the optical cycle. Wave packet simulations are also shown corresponding to the observations in (a) and (b). Figure 1 shows examples of the probe-pulse-induced fluorescence intensity plotted against the probe delay. A periodic modulation of the fluorescence signal is seen with a cycle of about 280 fs, which is equal to the recurrence time of the B

527

state WP, i.e., the vibrational period Tvib ? corresponding to the vibrational level spacing of about 119 cnf^ for the v = 3 - 5 levels [6]. The center time of the pump or control pulse is shown by arrows. The delay of the control pulse with respect to the pump pulse, control delay, is about 27Vib (560 fs) in both (a) and (b), but a difference of 1.03 fs was set between (a) and (b). The diffeience is a half of the optical cycle of the pump laser field, i.e., the phaseTT difference. The phase 7t difference gives constructive and destructive WP interaction (Ramsay interference). Figure 2 shows clearly a constructive interaction in (a) and a destructive one in (b). Numerical WP calculations are performed to understand time and space evolution of the double-pulse controlled WP. Examples of the calculations are shown also in Fig. 1 for control delays at 562.899 fs and 563.927 fs.

4. Conclusions We have performed pump -control-probe experiments of the vibrational WP in the molecular iodine. Employing the method with two variable time delays, control delay diVid probe delay, time evolution of the double-pulse-controlled WP is demonstrated in real-time for the sequence of pump and control. Numerical WP calculation is of help to understand the observed time evolution of the double-pulse-controlled WP. Acknowledgements. The authors acknowledge Prof. R. Lang, Prof. K. Misawa, and Dr. N. Hashimoto for helpful discussions and technical advices on this work. K.O. thanks Prof. K. Yamanouchi for his support and encouragement. This work was supported in part by Grants-in Aid for Scientific Research (Priority Areas: "Control of Molecules in Intense Laser Fields" and "Molecular Physical Chemistry", No. 13440171, No. 13440120, No. 11640383) from MEXT of Japan.

References 1 R. R. Jones, D. W. Schumacher, T. F. Gallagher, and P. H. Bucksbaum, J. Phys. B: At. Mol. Phys. Vol. 28, L405, 1995 2 T. C. Weinacht, J. Ahn, and P. H. Bucksbaum, Phys. Rev. Lett.Vol. 80, 5508, 1998. 3 N. F. Scherer, R. J. Carlson, A. Matro, M. Du, A. J. Ruggiero, V. Romero-Rochin, J. A. Cina, R Fleming, and S. A. Rice, J. Chem. Phys. Vol. 95, 1487, 1991; ibid. Vol. 96,4180, 1992. 4 V. Blanchet, M. A. Bouchene, and B. Girard, J. Chem. Phys. Vol. 108, 4862, 1998. 5 K. Ohmori, Y. Sato, E. E. Nikitin, and S. A. Rice, Phys. Rev. Letters, 91, 243003, 2003 6 R. F. Barrow and K. K. Yee, J. Chem. Soc. Faraday Trans. 2 69, 684 (1973)

528

Single-shot Transient Absorption of I3" in Solutions and Glasses Peter R. Poulin and Keith A. Nelson Department of Chemistry, Massachusetts Institute of Technology, 77 Massachusetts Ave., Cambridge, MA 02139 USA E-mail: [email protected] Abstract. A new single-shot femtosecond spectroscopy method has been applied to the study of triiodide photochemistry in solution and in room temperature glasses.

1.

Introduction

Liquid-state photochemical reaction dynamics have been extensively studied via femtosecond spectroscopy in recent years, but irreversible solid-state chemistry, wherein the local environmental effects of primary interest are expected to be considerably stronger, is less v^ell explored due to the practical problems of sample quality, reaction product buildup and structural deterioration after repeated irradiation. In order to overcome these limitations, we have developed a new technique for pump-probe spectroscopy which is designed to measure the entire temporal response of a sample in a single laser shot. The triiodide system has been the subject of recent experimental and theoretical interest [1-4]- We have studied thje transient absorption of I3" in a variety of environments in order to assess the role that local structural constraints play in determining chemical reactivity.

2.

Experimental Methods

An illustration of the experiment is shown in Figure 1(a). The output of a 10-Hz Ti:sapphire amplifier (50 fs pulse duration, 815 nm center wavelength, and 4 mJ probe beam

crossed echelons TO

>. C 0)

^

/w\^A/^^vvwvvvVvwvvv

c •0 (U

F

CO

(b)

CD

0

2000 4000 6000 Delay Time (fe)

8000

Fig. 1. (a) Schematic of single-shot femtosecond spectroscopy setup, (b) ISRS data from bismuth germanate.

529

pulse energy) is regulated with a shutter to allow single pulses to exit the system. The 5 mm diameter probe beam is expanded to more than 15 cm, and a 2.5 cm aperture selects the spatial center region, resulting in a nearly uniform crosssectional intensity distribution. The beam is collimated and then passed through two orthogonal echelons, which are right-angled prisms with coarse rectangular gratings inscribed along the hypotenuse. [5] Each echelon has 20 step surfaces, and so the probe beam is divided into a 20x20 array of smaller beams. Step thicknesses are such that each subsequent square in the 2D grid pattern thus formed is delayed by an additional 25 fs relative to its preceding square, yielding a total temporal window of 10 ps. All beams are focused through the excitation region, imaged onto a CCD camera, and the raw data deconvolved to extract a time-resolved signal.

3.

Results and Discussion

As a test of the sensitivity of the technique, we measured impulsive stimulated Raman scattering (ISRS) from a commonly used x-ray scintillator crystal, bismuth germanate Bi4Ge30i2 (BGO). The signal was detected by employing a narrow bandpass filter with central frequency to one side of the maximum in the probe pulse spectrum. Measurement of probe transmission through the filter reveals time-dependent oscillations due to vibrationally-induced shifts of the probe spectrum. Figure 1(b) shows the single pulse-derived signal. The oscillating response persists for times longer than temporal window of our instrument, and the 90 cm"^ optic phonon frequency [6] is easily recovered. The triiodide ion is a 22-electron, singly-charged anion with absorption maxima at roughly 290 nm and 360 nm in ethanol. Single-photon absorption from either band leads to a directly dissociative excited state, which eventually results in the formation of weakly bound diiodide ion, I2', and I radicals. Subsequent absorption from the ground state of I2" also leads to one of two closely-spaced dissociative states, which generate iodide ions, I", and I radicals in one of two spin states. The very broad absorption band of I2" is centered at 740 nm and is well isolated spectrally from the I3" bands. In order to realize the wavelength tunability required for this system, two home-built broadband noncollinear optical parametric amplifiers (NOPAs) [7] were utilized to generate the pump and probe pulses. Both NOPAs produce 25 fs pulses tunable throughout the visible region. Single-shot pump-probe scans of triiodide in ethanol solution are presented in Figure 2(a). The second broad peak is the spectral signature of diiodide ions. Near the center wavelength of the I2' band, strong absorption occurs throughout the probe temporal window. Toward the wings of the band, the absorption signal decays with a time constant of about 4.5 ps. This leads to an overall narrowing of the absorption band as excess vibrational energy is dissipated, presumably due to both strong dynamical interactions with solvent molecules and the inhomogeneous distribution of initial coherent states. In several scans, I2' vibrational oscillations are also observed [8]. To gain insight into the constraints imposed by the solid phase and the implications of such restrictions with regard to photoreactivity, we present in Figure 2(b) transient absorption pump-probe scans at 300 nm of triiodide ions in

530

2000

4000

D e l a y T i m e (fs)

6000

2000 4000 Delay Time (fs)

6000

8000

Figure 2. (a) Transient absorption in the visible region to monitor dynamic evolution of the I2" photoproduct. (b) Transient absorption at 300 nm ofl{ in glassy CDE solutions, as a function of temperature. cresolphthaleindimethylether (CDE), a glass-forming liquid with Tg = 310 K. As the temperature is low^ered, the solution increasingly takes on the character of a disordered solid. Oscillatory behavior (I3' symmetric stretch) is observed and the periodic intensity modulations persist for longer times in the glass, likely due to relatively less efficient solvent-induced dephasing. The bleach decay rate increases markedly, reflecting more rapid I3' regeneration on the ground state surface.

4.

Conclusions

A new technique for single-shot spectroscopy has been applied to the study of triiodide photochemistry. The restrictions imposed by the increasing rigidity of the solution environment effectively hinder the otherwise facile photodissociation reaction. Simulations are underway in an effort to provide detailed analysis of the interactions responsible for the observed dynamics. Acknowledgements. This work is supported by the Office of Naval Research grant number NOGO14-01-1 -0802.

References 1. 2. 3. 4. 5. 6. 7. 8.

U. Banin, A. Bartana et al, J. Chem. Phys. 101(10), 8461-8481 (1994). E. Gershgoren, J. Vala et«/., J. Phys. Chem A 105, 5081-5095 (2001). S. Hess, H. Busing and P. Vohringer, J. Chem. Phys. 111(12), 5461-5473 (1999). H. Choi, R.T. Bise et ai, J. Chem. Phys. 113(6), 2255-2262 (2000). G.P. Wakeham and K.A. Nelson, Opt. Lett. 25(7), 505-507 (2000). B. Mihailova, D. Toncheva et ai, Solid State Commun. 112, 11-15 (1999). T. Kobayashi and A. Shirakawa, Appl. Phys. B 70, S239-S246 (2000). D.A. Kliner, J.C. Alfano and P.F. Barbara, J.Chem. Phys. 98(7), 5375-5389 (1993).

531

Part VII

Multiview and Multi-Dimensional Spectroscopy

Dynamics of hydrogen bonds in water: Vibrational echoes and two-dimensional infrared spectroscopy C J . Fecko\ J.D. Eaves\ J J . Loparo\ S T . Roberts\ A. Tokmakoff\ P.L. Geissler^ ^ Department of Chemistry and George R. Harrison Spectroscopy Laboratory, Massachusetts Institute of Technology, Cambridge MA 02139, USA E-mail: [email protected] ^ Department of Chemistry, University of California, Berkeley, CA 94720 Abstract. Water hydrogen bond dynamics are studied by interpreting the OH frequency correlation function of HOD in D2O obtained from a vibrational echo peak shift measurement with molecular dynaiTiics simulations. Heterogeneous hydrogen bond dynainics are investigated using 2D IR spectroscopy.

1.

Introduction

The physical and chemical properties of water are dictated by hydrogen bond dynamics, the reconfiguration of the structure of water during the breaking and forming of hydrogen bonds. Water is a remarkably structured hquid, considering that at any instant it has 80-90% of the hydrogen bonds found in the tetrahedral structure of ice. But the structure imposed by these highly directional hydrogen bonds evolves very quickly, with large scale fluctuations on femtosecond time scales and subsequent more permanent hydrogen bond breaking and forming on roughly 1 ps time scales. Presently, our understanding of hydrogen bond dynamics draws largely from classical molecular dynamics computer simulations, but ultrafast infrared spectroscopy is rapidly opening new windows into the evolution of intermolecular structure in water. We have investigated the hydrogen bond dynamics of water with a combined experimental and theoretical study of the OH stretch spectroscopy of HOD in D2O. This widely studied model system can be used to characterize the vibrational dynamics of an isolated OH stretch vibration within D2O molecules, whose hydrogen bond dynamics closely mirror those of H2O [1,2]. The OH stretch frequency co is particularly sensitive to the hydrogen bonding environment leading to a broad absorption line in the mid-infrared (3 |Lim). Femtosecond infrared spectroscopy can be used to characterize spectral diffusion within the OH absorption line, which in turn is determined by the hydrogen bond dynamics. To ensure that dynamics on all possible time-scales are observed, we use 45 fs pulses with enough bandwidth to span the entire absorption line. Infrared vibrational echo peak shift measurements were performed to characterize spectral diffusion through the OH frequency correlation function. 2D IR experiments are used to probe variation in the femtosecond relaxation dynamics for initially hydrogen-bound and

535

non-hydrogen bound species. These vibrational dynamics are related to the underlying hydrogen-bond dynamics by developing frequency-structure correlations using a model for OH frequency shifts that draws on molecular dynamics simulations.

2.

Infrared Vibrational Echo Spectroscopy

We have used IR vibrational echo peak shift (PS) spectroscopy to characterize the time-scales for the OH frequency fluctuations [2]. The shift of the peak echo signal as a function of waiting period is closely related to the OH frequency correlation function, C(T) = (5CO(T) 5CO(0)). The PS data shows three features: a rapid sub-100 fs decay, a beat at 150 fs, and a long time exponential decay. The C(T) obtained by modeling the experiment with the pulse amplitude and phase (Fig. la) is found to have several characteristic features: a 60 fs decay, a pronounced oscillation with a 180 fs period, and a 1.4 ps long time decay. The beat corresponds closely to the time-scale associated with the intermolecular (hydrogen bond) stretch in femtosecond optical Kerr effect spectroscopy and in spontaneous Raman spectroscopy. The correlation function implies that the OH frequency is modulated by underdamped intermolecular displacement of the hydrogen bond, which precedes more permanent hydrogen bond breaking on time scales greater than 300 fs.

200

400

t(fs)

600

800

1000

0

200

400

600

800

1000

t (fs)

Fig. 1. (a) C(T) derived from vibrational echo peak shift, (b) C(T) obtained from molecular dynamics based modeling [2].

3.

Modeling Spectroscopy with Classical Molecular Dynamics

A model for the OH frequency shifts for HOD in D2O, which accounts for intermolecular interactions using classical molecular dynamics simulations, has been used to investigate the relationship between vibrational frequency and hydrogen bonding configurations about the HOD molecule [2]. The approach is similar to more elaborate methods used by others for OH frequency shifts in this system [3]. We use a model in which the quantum mechanical OH coordinate, represented with a gas phase Morse potential, is perturbed by the time-dependent interactions experienced by an OH coordinate in a molecular dynamics simulation of one HOD molecule in 107 D2O molecules. The OH frequency distribution taken

536

from static configurations is 260 cm'^ wide, with strongly hydrogen bound species on the red side of the Une and weak or broken hydrogen bonds on the blue side of the line. The C(T) calculated from this model (Fig. lb) show the same three qualitative features as the experiment. The model predicts a reasonably strong correlation between OH frequency and the 0H---0 hydrogen bond length to the nearest neighbor (p=0.77), and a weaker correlation with hydrogen bonding angle (p=0.49). Poor correlation exists between OH frequency and the tetrahedrality of the first solvent shell. We find that no simple structural or geometric order parameter captures all of the observed dynamics. However, the OH frequency correlates strongly (p=0.99) with the molecular electric field experienced by the OH coordinate due to the changing configurations of all D2O molecules. Although the configuration of the hydrogen bonding partner plays the largest role in the OH frequency shift, there are both local and collective contributions to the short time dynamics and the origin of the 150 fs beat.

4.

Heterogeneous Hydrogen Bond Dynamics

The correlation function forms an ensemble-averaged measure of the amplitude and time-scales of the frequency fluctuations for tliis system; however, we also observe that the OH frequency discriminates between hydrogen bonded and nonhydrogen bonded species by absorption frequency. We expect that the underlying dynamics and frequency fluctuations experienced by these sub-ensembles should be somewhat different on times <300 fs. 2D IR spectroscopy has been used to study heterogeneous dynamics within the OH lineshape. Our model has been used to reveal how 2D IR spectroscopy reveals different fast time dynamics for hydrogen-bound and non-hydrogen bound species as well as the kinetics of hydrogen bond breaking and forming. 2D IR spectroscopy extends three-pulse vibrational echo spectroscopy, correlating how molecules initially at one frequency (coi) evolve to a final frequency (CO3) during the course of a waiting period (T2). 2D IR is a femtosecond Fourier-transform four-wave-mixing experiment in which the electric field radiated by the sample is characterized by heterodyne detection with a local oscillator pulse. The Fourier transform of data acquired as a function of an excitation pulse pair delay (TI) and a detection period (T3) is used to reveal a 2D IR correlation spectrum. The shape of the spectrum evolves from a diagonally elongated (quasiinhomogeneous) line toward a symmetric (homogeneous) lineshape as the waiting time is increased. The overall ellipticity in the lineshape can be directly related to the peak shift measurement and correlation function, whereas asymmetries in the lineshape are related to heterogeneity in the underlying dynamics. Our 2D IR spectra of the OH transition reveals some dynamical heterogeneity, in which the blue side of the 2D IR lineshape broadens faster with waiting time than the red side. This behavior is consistent with simulations of the 2D IR experiment drawn from our model, which predict that initially hydrogen bound molecules on the red side of the line are subject to strong intermolecular oscillations whereas those initially on the blue side return rapidly to line center without pronounced beats. On longer time scales, heterogeneity reflects the kinetic rates of hydrogen bond breaking and forming.

537

MH)fS

MJ ' ,.- 3600

.-. 36CC o 34CG

32CC

320Q

3400

3600

200

3400

3BC0

Fig. 2. 2D IR correlation spectra of the OH stretch of HOD in D2O at waiting times of 80 and 500 fs.

Acknowledgement. This work was supported by the U.S. Department of Energy.

References R. Laenen, C. Rauscher, and A. Laubereau, J. Phys. Chem. B 102, 9304 (1998); S. Woutersen, U. Emmerichs, and H.J. Bakker, Science 278, 658 (1997); G. M. Gale, G. Gallot, F. Hache, N. Lascoux, S. Bratos, and J-Cl. Leicknam, Phys. Rev. Lett., 1068 (1999); J. Stenger, D. Madsen, P. Hamm, E.T.J. Nibbering, and T. Elsaesser, J. Phys. Chem. A 106 (10), 2341 (2002); S. Yeremenko, M. S. Pshenichnikov, D. A. Wiersma, Chem. Phys. Lett. 369, 107 (2003); J. B. Asbury, T. Steinel, C. Stromberg, S.A. Corcelli, C.P. Lawrence, J.L. Skinner, and M. D. Payer, J. Phys. Chem. A (2004). 2. C.J. Fecko, J.D. Eaves, J.J. Loparo, A. Tokmakoff, and P.L. Geissler, Science 301, 1698(2003). R. Rey, K. B. Moller, and J. T. Hynes, J. Phys. Chem. A 106 (50), 11993 (2002); C. P. Lawrence and J. L. Skinner, J. Chem. Phys. 118, 264 (2003).

538

Dual-Frequency 2D IR Photon Echo of a Hydrogen Bond Igor V, Rubtsov, Keshav Kumar, Robin M. Hochstrasser Department of Chemistry, University ofPennsylvania, Philadelphia, PA 19104-6323, USA Abstract. Dual-frequency 2D IR heterodyned photon echo experiments with independently tunable mid-IR pulses are presented. Intermolecular hydrogen bonding between NH and CO groups is characterized via observation of interaction of the NH and CO vibrational modes using a sequence of 3, 3, and 6 |Lim pulses. Structural and dynamical parameters determined include the mode coupling, angular constraints, and a correlation factor for the frequency distributions. 2D IR analogs of 2D NMR techniques have recently been emerged as powerful methods of structural recognition in solution under ambient conditions.^' ^ Observation of interaction of more than one vibrational mode can greatly enhance the scope of the 2D IR methods. The strength of the dual-frequency heterodyne photon echo (HPE) technique predicted theoretically"^ has been recently demonstrated experimentally through the signal from the phase-locked 6 and 7 |im and IR pulses."^ Here we present new 2D IR HPE experiments with IR pulses independently tunable in a wide spectral range are not required to be phased locked. Using these pulses 2D photon echo spectra were acquired for vibrational modes with very different frequencies, such as NH (~3 jam) and CO (~6 \xm) stretch vibrations, which allowed characterization of an intermolecular N-H*"0=0 hydrogen bonding through observation of the NH and CO modes interaction. The signal and idler pulses from two optical parametric amplifiers produce midIR pulses at - 3 |^m (130 fs, 0.8 |iJ) and ~6 |^m (0.4 \xi) by difference frequency generation. The 3 )Lim beam was split into two equal parts and interacted with the sample via k\ and ^2 directions (Fig. 1 A). The 6 |^m beam provides the third beam, ^3, and a local oscillator (LO) beam. An MCT detector recorded the sum of the field generated by the sample in a -A:i+^2+^3 direction and the LO field delayed by delay time /. Two-dimensional (x, t) data sets were double Fourier transformed to obtain 2D (co^, co^) spectra. 2D spectra where recorded in rephasing and nonrephasing experimental conditions,"^ which differ by the time ordering of the first and second pulses. Liouville pathways of the rephasing conditions, responsible for the cross peak in the impulsive limit are shown in Fig. IB. The sample was a solution of ca. 330 mM pyrrole and ca. 60 mM N,Ndimethylacetamide (DMA) in benzene (Fig. IC). One-to-one complex formation was confirmed by a titration. The NH frequency of pyrrole shifts from 3460 to -3300 cm"^ and broadens upon hydrogen bond formation (Fig. ID). The spectra of the IR pulses were tuned to cover the corresponding IR bands of the sample: ki and k2 accessed the NH transition of pyrrole and k^ covered the CO mode of DMA. Interestingly, the field generated by the pyrrole/DMA sample in the dual-frequency experiment demonstrates clear "photon-echo" type behavior: a rephasing effect is observed. Figure 2A shows two signals measured in rephasing

539

Me

Me

1650 3200 Frequency / cm''

3300

Fig.L A. The spatial arrangement of the beams interacting with the sample. B. Liouville pathways contributing to the cross peak in the rephasing pulse sequence with the 3 jim pulse first. C. Schematics of the DMA-pyrrole H-bond formation. D. Difference linear absorption spectra: the spectrum of DMA/pyrrole in benzene minus that of pyrrole in benzene (black line) and the spectrum of DMA in benzene minus benzene spectrum (gray line). experimental conditions with different coherence times (x). The signal maximum shows a clear trend: the signal peaks at larger time delays / for larger coherence times. To have such effect in the dual-frequency experiment the NH coherence created by interaction with the first IR pulse has to be preserved in CO frequency distribution after interaction with all three IR pulses. Such coherence transfer is possible only if the NH and CO fi-equency distributions are correlated. Since the rephasing effect is observed in the "rephasing" experimental conditions, the correlation factor has to be positive."^ A precise value of the correlation factor can be obtained from the 2D IR spectrum. Figure 2C shows a broad view absolute value spectrum of the sample. Since there is no spectral overlap of the k\ and k2, laser pulses, only the cross peaks are observed. An absorptive spectrum, which is a sum of the real spectra measured in the rephasing and nonrephasing conditions is shown in Fig. 2D. A relative phase in the real spectra was tuned to match the co^ projection of the 2D spectrum and a dual-frequency pump/probe spectrum. The cross peaks in the absorptive spectrum are tilted in the diagonal direction (Fig. 2E). jCTco» where So is the frequency The correlation ooQmoiQnt f {f = {Sco^^dcDro)I
540

B

-400 -300 -200 -100

0 100 200 300 400 t/fS

'-•xO , 3350

^ ^

>~"-<<5

3300

_.,^ V " '•

M'^l''^''

3250 3200

0 ^ 3150 1800 (0, / c m

2000

' \

;Sv i • 1580 1600 1620 1640 1660 CO. / c m

Fig. 2. A. The field generated by the sample in rephasing experimental conditions with x = 0 (Upper) and x = 280 fs (Lower). B. Integrated dual-frequency photon echo of the sample at r=300 fs. C. Rephasing absolute 2D spectrum, r = 300 fs. D. Absorptive 2D spectrum, T = 300 fs. E. Simulated absorptive 2D IR spectrum with/= 0.63, T= 300 fs. variation of the tilt angle and the width of the spectrum, is observed experimentally. This dependence is observed over several ps time window, which demonstrates the ability of HPE spectroscopy to follow the H-bond dynamics over a time range which is much larger than the dephasing time in the system. An example of the integrated photon echo trace measured at T = 300 fs is shown in Fig. 2B. 2D anisotropy spectra of the cross peak have been measured. Experimental anisotropy values of the cross peak (+0.25) predict the mean angle between the NH and CO transition moments to be 38°. Taking the direction of the CO (amide I) dipole at ca. 15° to the C=0 bond and giving equal weight for all possible NHOC torsion angles, the resulting mean angle is calculated to be ca. 36°. In conclusion, the first dual-frequency 2D IR heterodyned photon echo experiments with independently tunable mid-IR pulses are presented. Polarized, absorptive, and integrated photon echo spectra with two widely separated frequencies, which are not phased locked, have been acquired. Implementation of this technique to NH and CO modes of the groups forming a hydrogen bond allows detailed insights into hydrogen bond arrangement.

References p. Hamm, M. Lim, and R. M. Hochstrasser, J. Phys. Chem. B 102, 6123 (1998). P. Hamm and R. M. Hochstrasser, in Ultrafast Infrared and Raman Spectroscopy, edited by M. D. Payer (Marcel Dekker Inc., New York, 2000), p. 273. C. Scheurer and S. Mukamel, J. Chem. Phys. 116, 6803 (2002). I. Rubtsov, J. Wang, and R. Hochstrasser, Proc. Natl. Acad. Sci. USA 100, 5601 (2003). 541

2D-IR spectroscopy of transient species Jens Bredenbeck, Jan Helbing, Peter Hamm Physikalisch-Chemisches Institut, Universitat Zurich, 8057 Zurich, Switzerland Abstract. Transient two-dimensional infrared spectroscopy (T2D-IR) extends 2D-IR spectroscopy to the non-equilibrium regime. We demonstrate different types of T2D-IR experiments for a charge transfer state of [Re(CH3-bpy(CO)3Cl], including the vibrational analogue of NMR-exchange spectroscopy.

T2D-IR has been demonstrated recently for the picosecond conformational transition of a photoswitchable cyclic peptide [1]. The transition was triggered by an UV laser pulse preceding the 2D-IR part of the experiment (sequence Fig. la). Here, we discuss two novel T2D-IR experiments: Sequence lb constitutes a dynamic 2D hole burning experiment on the non-equilibrium ensemble created by the UVpump pulse. Sequence Ic is referred to as "band labeling spectroscopy", as it allows to label a specific vibration by the narrowband IRpump pulse before the UVpump pulse is applied and follows this label in the course of the subsequent photo-reaction. The experiments are demonstrated for photo-triggered metal-toligand charge transfer (MLCT) in [Re(CH3-bpy(CO)3Cl]. MLCT induces a strong blue shift of the CO vibrations of the complex due to the change in electronic structure and subsequent solvation [2]. Fig. 2a shows the bands of the CO modes and their shift upon MLCT. a

W

Vix

b

^x

tvar

C

i^.i^L

i /n,. i X a , MX,

^pump

'^mp'^Vobe

'-'^pump ''>ump

'^obe

'Rpurrp^pump^^obe

Fig. 1. T2D-IR experiments: a - regular T2D-IR spectroscopy, b - hole burning T2D-IR spectroscopy, c - band labeling T2D-IR spectroscopy. Band-labeling T2D-IR can be used to assign the CO vibrations in the MLCT state. Fig. 2a shows the TID-IR spectrum at 20 ps UVpump-IRprobe delay. The spectrum does not reveal which vibration shifts where upon UV excitation. The intuitive assignment used in the literature corresponds to the dotted arrows in Fig. 2 [2]. However, theoretical investigations suggest an assignment according to the solid arrows. Such problems can be addressed by band labeling T2D-IR spectroscopy in a very direct manner: the narrowband IRpump pulse precedes the UVpump pulse and 'labels' a vibration by transferring population fi-om the v=0 to the v=l state. The UV-pulse then excites the molecule to the MLCT state and induces the frequency shift. The IRprobe pulse records the frequency of the labeled and shifted vibration. In the labeling T2D-IR spectrum (Fig. 2c), the vibrational shift from ground to MLCT state occurs parallel to the x-axis. The y-position

542

1880 1920 1960 2000 2040 2080 2120 frequency [cm'']

1880 1920 1960 2000 2040 probe frequency [cm" ]

1880 1920 1960 2000 probefrequency[cm" ]

Fig. 2. (a) Rhenium complex, absorption spectrum with assigned CO vibrations, transient spectrum after 20 ps. Arrows mark the shift of the a'(2) and a" bands upon electronic excitation. Dotted arrows: old assignment for the transient bands, solid arrows: band shift as revealed by labeling T2D-IR. (b) TID-IR spectrum after 2 ps. (c) Labeling T2D-IR spectrum, UVpump-IRpump delay 2 ps, IRpump-IRprobe delay 2 ps. Empty circles mark diagonal peaks of the electronic ground state, filled circles mark shifted peaks of the MLCT state, (d) Labeling T2D-IR spectrum employing special polarization conditions to suppress ground state contributions. The solid arrows correspond to the solid arrows in (a).

(IRpump frequency) of the peak does not change upon electronic excitation, since vibrational excitation took place in the electronic ground state. Thus excited state and ground state vibrations are correlated by the labeling pulse sequence and their respective frequencies are given by the x and y positions of the MLCT peaks. This can be nicely seen for the a'(l) mode in the upper right comer of Fig. 2c (Pos. 1). For the a" and a'(2) modes (Fig. 2c, Pos. 2 and 3) the effect is not that obvious. It can be enhanced using special polarization conditions [3] that suppress disturbing contributions from the electronic ground state (Fig. 2d), giving direct evidence for the new assignment as indicated by the solid arrows in Fig. 2a. The labeling experiment constitutes the vibrational analog of NMR-exchange spectroscopy in that it generates off-diagonal peaks connecting frequencies of the initial and of the final state of a reaction. Hole-burning T2D-IR. T2D-IR spectroscopy is a 5*-order experiment, where the UVpump pulse creates a non-equilibrium ensemble on which a 2D-IR measurement is carried out. In [Re(CH3-bpy(CO)3Cl], this non-equilibrium state is the MLCT state directly after UV excitation. T2D-IR can be used to test whether the subsequent solvation dynamics occurs within the regime of linear response. Solvation is monitored by the shift of the a'(l) CO vibration. Fig. 3a shows the TID-IR and T2D-IR spectra after 100 ps, when solvation is completed. The T2D-IR spectrum shows a characteristic tilt, which is a clear signature of inhomogeneity on the picosecond timescale. The tilt as a function of IRpump-IRprobe delay is a direct measure of the equilibrium frequency fluctuation correlation function c(t) [4]. Within linear response theory, the relaxation kinetics of the mean of a non-equilibrium ensemble ls.o5{t) ? as monitored by the overall shift of the band, should be the same as that of c(t), as monitored by the tilt.

543

2020 2040 2060

2020 2040 20602020 2040 20602020 2040 20602020 2040 20602020 2040 2060 Probe Frequency [cm'^]

Fig. 3 (a) T2D-IR spectrum of the a'(l) band after equilibration in the MLCT state, (b) T2D-IR spectra during the solvation shift. The decay of the signal due to vibrational relaxation is normalized out. The TID-IR spectra are measured at corresponding times with the IRpump pulse blocked, (c+d) corresponding data from a linear response model calculation. Fig. 3b shows TlD-IR and T2D-IR spectra of the excited state a'(l) band with an UVpump-IRpump delay time of 1 ps and an IRpump-IRprobe delay varied between 1 ps and 6.5 ps. At first sight, the T2D-IR spectra evolve, as one intuitively expects: First, the T2D-IR spectrum is strongly tilted, reflecting strong inhomogeneity. With increasing IRpump-IRprobe delay, the band orients vertically as a result of the loss of memory of the initial pump frequency. At the same time, the 2D-IR band center shifts towards the right, away fi^om the diagonal and in parallel with the solvation shift in the TlD-IR spectra. However, at a time of 6.5 ps, where the shift of the band is not finished, the tilt has completely disappeared, in contrast to what is expected from linear response. For comparison we simulated the experiment within the framework of linear response. The results are shown in Fig. 3c and d. They reproduce the general trend of a shifting and tilting band, however, they cannot yield a relaxation of the tilt that is faster then the overall shift. As expected, there is still an appreciable tilt after 6.5 ps in the simulation. The present results demonstrate the capability of T2D-IR to monitor spectral diffusion within the shifting vibrational band of a non-equilibrium system. It provides previously inaccessible information and shows in the present case, that a description of solvation within linear response theory is not sufficient.

References 1 J. Bredenbeck, J. Helbing, R. Behrendt, C. Renner, L. Moroder, J. Wachtveitl, P. Hamm, J. Phys. Chem. B. 107, 8654, 2003 2 J. Bredenbeck, J. Helbing, P. Hamm, J. Am. Chem. Soc. 126, 990, 2004 3 J. Bredenbeck, J.Helbing, P. Hamm, J. Chem. Phys., in press 4 K. Kwac, M. Cho, J. Chem. Phys. 119, 2256, 2003

544

Resolving Conformations of Acetylproline-NH2 by Coherent 2D IR Spectroscopy Denis Karaiskaj, Soohwan Sul, Ying Jiang, and Nien-Hui Ge Department of Chemistry, University of California, Irvine, CA 92697-2025, USA Email: [email protected] Abstract. Coherent 2D IR techniques with various pulse sequences and polarization conditions reveal new spectral features associated with multiple conformations of acetylproline-NH2 in CDCI3. Introduction. Detailed understanding of protein structure and dynamics is crucial for essentially all biologically relevant reactions such as protein folding or electron transfer. Much theoretical and experimental effort has been focused on the study of small peptides as paradigms for larger systems. However, it is often difficult to characterize the structure of small peptides in solution because their conformations fluctuate on time scales which are rapid on the 2D NMR time scales. Two-dimensional infrared (2D IR) spectroscopy recently emerged as a new method to obtain structural features of complex molecular systems [I]. To fully develop 2D IR spectroscopy as a structural probe, it is important to explore and understand the effects of different experimental tools, such as using different pulse ordering and phase matching [2,3], under specific polarization conditions [4]. In this paper, we demonstrate the ability of 2D IR to resolve new spectral features most likely associated with multiple conformations of acetoproline-NH2. Experimental Methods. Experimentally, femtosecond IR pulses (100 fs, 1 }iJ, 6 |im) were generated by mixing the signal and idler beams of an 800 nm pumped OP A in an AgGaS2 crystal. The output was split into three beams (wavevectors k^, k\y, and kc, each 300 nJ) and a local oscillator beam (^LO^ 25 nJ) which was combined with the signal field in the phase-matched direction k^ = -k^ + A:b + ^c? and focused into a single channel HgCdTe detector. Figure 1 illustrates the three pulse sequences used in these experiments: rephasing (RP), non-rephasing (NR), and reverse photon echo (RPE). The time interval between the first and second pulses, between the second and third pulses, and between the third pulse and A:LO is t\, ti, and /3, respectively. The predicted real part of 2D IR spectra for the case of two weakly coupled vibrators, / and y, are schematically shown in Fig. 1. In contrast to the RP and NR spectra, peaks in the RPE spectra appear at overtone and combination band frequencies in the oyi direction, allowing a direct determination of diagonal and off-diagonal anharmonicities. Also, the diagonal peaks are separated farther apart in this case, potentially making the observation of the cross peaks easier. Results and Discussion. The acetylproline-NH2 has been studied previously using different techniques, such as NMR and optical spectroscopy. The most recent 2D IR studies using the RP sequence alone [5] suggested that in CDCI3 two

545

(a)

-/ca

'^A

^Lo

(b) k^

-^-f\

kLO

(c)

A ^

M.-

2(0,

A, (0;4-<0y

2«)C

(0/

W/

coy

/CLO

-/fa

A/y

T

/A/y

^11

...^ ej®

1A,-

(0/

COy

Fig. 1. Three pulse sequences used in experiments and the predicted 2D spectra for two weakly coupled vibrators (a) RP; (b) NR; (c) RPE. The large circles represent the diagonal peaks separated by the diagonal anharmonicity A/ or A,, while the small circles represent the cross peaks separated by the off-diagonal anharmonicity A,y. conformations of this peptide may exist. We expanded these investigations by using two additional sequences, NR and RPE, at various polarization conditions. Figure 2 shows our results. The peak labeled A in the spectra is the amide II band at the amino end of the peptide. The peaks labeled B and C are the amide I bands at the acetyl and amino ends, respectively. In agreement with the previous study, the diagonal peaks in the RP spectrum are not resolved. They become resolved, however, in the NR and RPE spectra. The RPE sequence exhibits enhanced resolving power for closely spaced spectral features because it shares similar nonrephasing properties as the NR sequence [2]. A "diagonal cut" through the RPE spectra shown in Fig. 3 reveals doublets in the amide I peaks indicating the existence of two different conformations. By line fitting this diagonal profile and by assuming that both species have the same extinction coefficient, we estimate a ratio of about 77% and 23% between the conformations, ignoring the interference effect due to fitting the spectra in the absolute magnitude instead of the real part.

1550

1600

1650

co3{cm-i>

1700

1750

1550

1600 1650 1700 1750

<%(cm-i)

1550

1650

1750

ro3(cm-i)

Fig. 2. Experimental absolute value 2D IR spectra for 150 mM acetylproline-NH2 in CDCI3 acquired using three techniques: (a) RP; (b) NP; (c) RPE at (0°,0°,0^0°) polarization.

546

Structural information is obtained by acquiring 2D IR spectra at two other polarizations, (0°,90°,90°,0°) and (90°, 90°, 0°, 0°) [corresponding to {K,K,kc,k^\ From the polarization dependence of cross peak intensities, we estimate the angles between the transition dipoles of two coupled vibrators to be 140° and 22° for cross peaks G and D, respectively. Since the cross peaks are not resolved, 1600 1800 these values are weighted the G). (cm'^) averages of both conformations. We compare these values with calculated Fig. 3. Diagonal cut through the RPE spectra. transition dipole angles. For the trans The amide I peaks show doublet structure C7 structure of acetylproline-NH2, labeled as B, B' and C, C . The peak A known as the global minimum, we originates from the amide II band. obtained transition dipole angles of 145° and 25°, assuming that the amide I mode is 20° away from the CO bond, and the amide II mode is parallel to the CN bond [6]. We also significantly enhance the cross peaks for all three sequences by using the polarization condition (45°, -45°, 90°, 0°) to remove the diagonal peaks. Our results show that acetylproline-NH2 in CDCI3 has most likely two conformations. The dominating conformation appears to be the trans C7 that exhibits an intramolecular hydrogen bond between the acetyl and amino ends of the peptide. Acquiring 2D IR spectra using all three sequences can enhance our ability to extract structural information. Acknowledgements. We are grateful to the Donors of the American Chemical Society Petroleum Research Fund for financial support.

References 1 M.C. Asplund, M.T. Zanni, R.M. Hochstrasser, Proc. Natl Acad. ScL U. S. A. 97, 8219 (2000). O. Golonzka, M. Khalil, N. Demirdoven, A. Tokmakoff, Phys. Rev. Lett. 86, 2154 (2001). P. Hamm, M. H. Lim, R. M. Hochstrasser, J. Phys. Chem. B 102,6123(1998). 2 N.-H. Ge, M. T. Zanni, R.M. Hochstrasser, J. Phys. Chem. A 106, 962 (2002). 3 W.M. Zhang, V. Chemyak, S. Mukamel, J. Chem. Phys. 110, 5011 (1999). B.C. Fulmer, P. Mukherjee, A.T. Krummel, M.T. Zanni, J. Chem. Phys. 120, 80 (2004). 4 Zanni, M.T.; Ge, N.-H.; Kim, Y.S.; Hochstrasser, R. M. Proc. Natl. Acad. Sci. USA 9S, 11265(2001). 5 M.T. Zanni, S. Gnanakaran, J. Stenger, R.M. Hochstrasser, J. Phys. Chem. B 105, 6520(2001). 6 M. Ramek, A.-M. Kelterer, B.J. Teppen, L. Schafer, J. Mol. Struct. 352/353, 59 (1995). S.-H. Lee, S. Krimm, Chem. Phys. 230, 277 (1998).

547

Ultrafast vibrational dynamics of rotaxanes Olaf F.A. Larsen\ Wybren J. Buma^, David A. Leigh^, and Sander Woutersen^ ^ FOM Institute for Atomic and Molecular Physics [AMOLF], Kruislaan 407, 1098 SJ Amsterdam, The Netherlands E-mail: [email protected] ^ Molecular Excited States group. University of Amsterdam, Nieuwe Achtergracht 127-129, 1018 WS Amsterdam, The Netherlands ^ School of Chemistry, University of Edinburgh, United Kingdom Abstract. Time-resolved infrared spectroscopy on a rotaxane consisting of a macrocycle and a short linear thread reveals coupling between the C=0 stretching modes of macrocycle and thread, as well as spectral relaxation on a picosecond time scale.

1.

Introduction

Rotaxanes constitute a class of mechanically interlocked molecules, in which a macrocycle is non-covalently attached to a linear thread. Rotaxanes are known to operate like molecular shuttles [1] and spinning motors [2] upon application of external stimuli like light. In the systems we study, the macrocycle is attached to the thread by means of hydrogen bonds. It is our aim to monitor the making-andbreaking of the hydrogen bonds caused by shuttling and spinning reactions in real time using time-resolved 2D-vibrational spectroscopy. As a first step in this direction, a rotaxane with a short thread has been studied. Its three-dimensional structure and ultrafast structural fluctuations have been investigated by measuring the coupling between and spectral dynamics of the C = 0 stretching vibrations.

1600

1620

1640

1660

1680

Frequency (cm ^) Fig.l. Linear infrared absorption spectrum (C=0 stretching region) of the model compound, together with its chemical structure

548

2. Results and Discussion Polarized 2D-vibrational spectra have been measured at a fixed delay time of 1 ps. The 2D-spectrum shows a clear bleaching (negative signal) and excited-state absorption (positive signal) around the pump frequency (peaks on the diagonal). A coupling between the carbonyls of the thread and the macrocycle can be observed: when exciting the 1610 cm"^ band, a signal around 1660 cm"^ is also oberved, and vice versa (the cross peaks). The intensity of the cross peaks is a direct measure of the coupling between the modes, and depends on the relative distance and orientation between the different transition dipoles. The anisotropy of the cross peaks is determined only by the angle between the transition dipoles.

1580

1600

1620

1640

1660

1680

^eOO

1620

Probe frequency (cm"^)

Fig. 2. Left: 2D-vibrationaI spectrum measured at 1 ps delay time. Perpendicular polarization between pump and probe beam. Right: Cross section of the 2D-vibrational spectra at a pump frequency of 1610 cm'^ (arrow). Dotted: linear-absorption spectrum. Both signals are normalized on the excited-state absorption. Squares: parallel polarization between pump and probe beam. Stars: perpendicular polarization between pump and probe beam. The cross peaks can be more clearly seen in cross sections of the 2D-spectra (Figure 2, right). The different polarization dependence of the cross peaks as compared to the diagonal peaks demonstrates that the cross peaks are not the effect of a simultaneous excitation of both absorption bands, but reflect a coupling between the different transition dipoles of both bands. Ultrafast fluctuations of the conformational degrees of freedom of the macrocycle, representing the elementary steps of macrocycle pirouetting and shuttling, have been studied by performing spectral relaxation measurements. Transient absorption spectra are measured for several delay times using different pump-frequencies within the inhomogeneously broadened 1610 cm'^ band. Since the instantaneous stretching frequency is a measure of the strength of the H-bond to the C=0 group [3], the spectral dynamics directly reflect the fluctuations of the hydrogen bonds between the thread and macrocycle.

549

Pump frequency • • *

1618

(cm'^) 1628 1610 1593

1617 1616 ^1615

<

J 1614 H1613 1612 1611 1610 1609 1580

1600

1620

1640

1660

Probe frequency (cm"^)

1680

2.0

2.5

3.0

3.5

4.0

Delay time (ps)

Fig. 3. Left: Transient-absorption spectra as a function of delay time between pump and probe pulses. For this pump frequency (1628 cm'\ arrow), a small fraction of the macrocycle carbonyls (absorbing around 1660 cm"^) is also excited, resulting in a small bleaching and excited-state absorption signal. Dotted: linear-absorption spectrum. Right: First moment of the bleaching of the 1610 cm'^ band as a function of delay time for different pump frequencies

3.

Conclusions

In principle, it is possible to determine the ring-thread conformation from the 2DIR spectrum. At the moment this is still an ongoing process, in which we use ab initio and MD calculations, the latter in cooperation with F. Zerbetto (University of Bologna, Italy). The spectral relaxation measurements show that part of the macrocycle fluctuations takes place on a timescale of a few picoseconds. Acknowledgements. This work is part of the research program of the "Stichting voor Fundamenteel Onderzoek der Materie (FOM)", which is financially supported by the "Nederlandse organisatie voor Wetenschappelijk Onderzoek (NWO)". The authors gratefully acknowledge financial support from the EC funded RTN Network "Exploitation Mechanical Motion of Molecular Architectures" (EMMMA, HPRN-CT-2002-00168).

References 1 A. M. Brouwer et al.. Science 291, 2124, 2001. 2 V. Bermudez, N. Capron, T. Gase, F. G. Gatti, F. Kajzar, D. A. Leigh, F. Zerbetto, and S. Zhang, Nature 406, 608, 2000. 3 H. Torii, T. Tatsumi, T. Kanazawa, and M. Tasumi, J. Phys. Chem. B 102, 309, 1998.

550

Thermal denaturing of proteins: Equilibrium and transient studies using nonlinear infrared probes H. S. Chung, M. Khalil, A. W. Smith, Z. Ganim and A. Tokmakoff Department of Chemistry, Massachusetts Institute of Technology, Cambridge MA 02139 E-mail: [email protected] Abstract. Thermal unfolding of |3-sheets in ribonuclease A and ubiquitin is revealed by disappearance of cross peaks in 2D IR spectra. Transient unfolding probed with vibrational echoes following a temperature jump reveals nanosecond to millisecond dynamics.

1.

Introduction

Two-dimensional infrared (2D IR) spectroscopy can be used to describe transient molecular structure in solution. The position, shape, and intensity of the peaks in the 2D IR spectrum reveal information on vibrational couplings and dipole orientations, which can be modeled in terms of molecular structure and interactions [1]. These characteristics can be used to develop sensitive tools of biomolecular structure, dynamics, and kinetics. 2D IR spectroscopy has already been used to study the structure and dynamics of peptides [2,3]. We have investigated the sensitivity of vibrational echoes and 2D IR spectroscopy of protein amide I transitions to (i-sheet secondary structure [4], and applied these techniques to follow structural changes during thermal denaturing. Ideal antiparallel (AP) P-sheets have two IR active amide I transitions, an intense low frequency mode a - and a weak high frequency mode a+ that result from electrostatic couplings between amide vibrations in the sheet [5]. 2D IR spectra reveal a characteristic cross peak between the a+ and a-. 2D IR spectra as a function of temperature reveal the melting of P-sheets on denaturing through the disappearance of this cross-peak. These signatures can also be seen in a frequency dispersed vibrational echo (DYE) spectrum, which is related to projection of complex 2D IR spectrum. Transient probing of thermal denaturing with DVE spectra following a nanosecond temperature jump (T-jump), reveals structural changes on nanosecond to millisecond time scales.

2.

Experimental

Third-order nonlinear signals are generated by a sequence of three 90 fs 6 //m pulses, separated by time periods i] and T2, "and followed by a detection period T3. In the 2D IR experiment, a heterodyne-detected signal is obtained by overlapping the signal with a local oscillator field and dispersing them onto an array detector to obtain the CO3 dimension. Xi is varied with a linear stage and numerically Fourier

551

transformed to obtain the coi dimension. For the DVE measurement, all three pulses are overlapped (TI=X2=0)J and the signal field is dispersed onto the array detector. For transient experiments, thermal unfolding is induced by a nanosecond T-jump and the structural change is monitored by a DVE probe.

3. 2D IR Studies of Equilibrium Thermal Denaturing Figure 1 shows the 2D IR spectra of RNase A and ubiquitin in the native and thermally denatured states. In the native states, the diagonal a - peak is the most intense and is broadened in the diagonal direction. This is consistent with the calculations showing that the a - mode frequency is sensitive to the size and structural inhomogeneity of the (J-sheet [5]. The diagonal a+ peak does not appear because of its low transition dipole strength, but is revealed through the cross peak. Since the a+ frequency is not sensitive to the structural variation of p-sheet, the cross peak in the upper left region is elongated along the cOi axis, forming a ridge. Constructive interferences between negative features stretch out the other cross peak, a-helices and disordered regions contribute to the diagonally elongated features in the 1650 cm"^ region of the spectrum. All of these cross peak contributions, in addition to the diagonal elongation, lead to a characteristic "Z"shape in the contour profiles.

1600

1650

1700

(O^/2;LC (cm"'')

1600

1650

1700

1600

1650

1700

1600

1650

1700

(u,,/2/ic (cm''')

Fig. 1. 2D IR spectra of the native (25°C) and thermally denatured (72°C) states of (a) RNase A and (b) Ubiquitin. In the thermally denatured spectra (72°C), the diagonal peaks blue-shift and become symmetric. Although the small ridge in the cross peak region indicates some residual secondary structure at tliis temperature, the loss of "Z"-shape affirms that the nearest-neighbor vibrational couplings characteristic of the P-sheet in the native structure of proteins are disrupted. Figure 2a shows the thermal denaturing of ubiquitin probed by DVE spectra. The DVE spectrum can be related to a projection of the complex 2D IR correlation spectrum onto 0)3. The two pronounced peaks in the DVE spectrum reflect the projection of the (3-sheet cross peak ridges in the 2D IR spectrum. The loss of this two-peak structure as the protein is thermally denatured is also apparent in the DVE spectra.

552

4. Transient Studies of Protein Denaturing Kinetics DYE spectra are a background free probe of protein conformation that can be used to follow the dynamics of protein denaturing following a T-jump. For monitoring spectral changes as a function of the delay following the initiation, the difference between the DYE spectrum at temperature T at delay T and the equilibrium spectrum at To is plotted in Fig. 2b. An instantaneous increase of intensity on the T-jump timescale reflects changes faster than the T-jump pulse duration. A decrease of the 1621 cm"^ region corresponding to the a - transition is observed within 10"^ s, which is followed by changes on the lO'"^ to 10'^ s timescale in other regions including the a-h transition (1677 cm'^), and the a-hehcal/random coil region (-1650 cm'^). The maximum observed change (at 9 ms) is just 10% of the equilibrium difference spectrum between 58°C and 68 °C, implying that many of the kinetics occur on longer time-scales. The time-dependence of several frequency channels is plotted in Fig. 2c. All traces are highly non-exponential, showing that the unfolding does not follow a simple two state kinetics. The most striking feature is the multiple time-scale kinetics on )LIS and ms time scales for the a - transition of 1621 cm'^ The changes in this frequency show that a small fraction of (3-sheets denature with a range of fast microsecond rates, whereas most of the protein unfolding kinetics are observed on time scales greater than 10 ms. (a) (b) (c) 400 / \

1.0

0.04

4B.'/'C

0.5

13

<

.. 0.02 200

0

CO

lio \^


\

0

;' \ f

-0.02 1600 1650 1700

1550

-0.5 . 9^s

1650 6)3/2710 (cm-

175C)

)

-1.0 y6 1

0162lGm' 01658 cm'

^ h

r::?')? 1677 cm' -— Temperature ^Q-6

^Q-4

\

J B ^\ IxM ^Q-2

T (S)

Fig. 2. (a) Thermal unfolding of ubiquitin probed by DYE spectroscopy (b) Difference DYE spectra of ubiquitin at different time delays after T-jump (c) Responses of three frequency channels with temperature profile.

References 1 M. Khalil, N. Demirdoven and A. Tokmakoff, J. Phys. Chem. A 107, 5258, 2003. 2 M. T. Zanni and R. M. Hochstrasser, Curr. Opin. Struct. Biol. 11, 516, 2001. 3 S. Woutersen, Y. Mu, G. Stock and P. Hamm, Proc. Nad. Acad. Sci. 98, 11254, 2001. 4 N. Demirdoven, C. M. Cheatum, H. S. Chung, M. Khalil, J. Knoester and A. Tokmakoff, J. Am. Chem. Soc. 126, 7981, 2004. 5 C. M. Cheatum, A. Tokmakoff and J. Knoester, J. Chem. Phys. 120, 8201, 2004. *

553

Two-Dimensional Optical Heterodyne Spectroscopy of Molecular Complexes Igor Stiopkin, Tobias Brixner and Graham R. Fleming Department of Chemistry and QB3 histitute University of Cahfomia, Berkeley and Physical Biosciences Division, Lawrence Berkeley National Laboratory, Berkeley, California 94720 E-mail: [email protected] Abstract Optical heterodyne two-dimensional (2D) photon echoesfromroom-temperature BIC J-aggregates were recorded. Analysis of the time-dependent 2D spectrum reveals the frequency dependence of the exciton relaxation, and discriminates diagonal and offdiagonal exciton phonon coupling.

Frequency-dependent dynamics is expected to be ubiquitous in systems with band structures. Quantum dynamical models of relaxation in such systems generally predict heterogeneous dynamical behavior, but most experiments provide information that is averaged over the distribution of dynamics. In this paper, we show that two-dimensional (2D) optical heterodyne photon echo spectroscopy [1] can reveal the frequency dependence of exciton relaxation in room-temperature BIC J-aggregates in aqueous solution. Our method also allows discrimination of the influence of diagonal and off-diagonal exciton-phonon coupling. We have constructed a 2D Fourier-transform spectrometer which is shown in Figure 1. coh. time

pop. ame

...5pGclraH m niGlei

signal rime

A A'A'A^A. 3

5ig

dfffratUve DpllD 2f

delay 1

Fig. 1. Experimental setup. Two timedelayed parallel beams are focused on a diffractive optics (DO) by a 20-cm lens. The beams from the first diffraction orders provide the excitation pulses (1-3) and the local oscillator (4 = LO). They hit a spherical mirror (2/ = 5 0 cm) and are focused into the sample via a plane folding mirror. Time delay T between pulses 1 and 2 is introduced with interferometric precision by movable glass wedges.

The apparatus shown in Figure 1 [2, 3] has a number of significant features: (1) The heterodyne detection allows recording of very weak signals ( < 100 aJ in the signal pulse), (2) the small ( r ) wedge angles (delay 1, delay 2 in Figure 1) provide highly precise and repeatable delays (± 5 as ), (3) the phase stability is very high and fringe patterns do not change over many tens of minutes (4)

554

scattering is subtracted by an automated shutter. We wish to detect only E4 -Estg. This is achieved by taking the following subtraction \Ej+E2-^Es+Esigf'\Es-^E4\ \Ei+E2+Esigf', here £;, E2 and Es represent scattered electric field of pump pulses 1,2 and 3 in the direction of the collinear reference and the signal beams with electric fields E4 and Egtg respectively. This allows us to take 2D spectra with background of <2% from quite highly scattering samples, (5) we determined the absolute phase of the 2D spectrum via comparison with a separately measured frequency resolved pump-probe signal. The experiments were carried out with 40 fs near bandwidth limited 596 nm pulses at 3 kHz repetition rate. The signal field is recovered via spectral interferometry [4, 5,6], while the coherence period, x, and p opulation p eriod, 7, d elays were s canned in t he t ime d omain. Pump p ulse intensities below 10^^ photons/(pulse cm^) were used to minimize the effect of exciton-exciton annihilation. Figure 2a shows the real part of the 2D spectrum for population times 7=0, 50 and 500 fs. The signal consists of two parts: a positive region (upper) representing bleaching and stimulated emission in the ground to 1-exciton band, and a negative region representing absorption from the 1-exciton to the 2-exciton band. Several features are apparent in the evolution of the real 2D spectra. The slope of the nodal line, which corresponds to the memory in the system, has considerably decreased

-17100 -16900 -16700 -16500 co^(cm"^)

-17100 -16900 -16700 -16500 (o^(cm''')

-17100 -16900 -16700 -16500

Fig. 2(a). Real part of the 2D spectrum from BIC Jaggregates at (from left to right) population periods T of 0 fs, 50 fs and 500 fs. Dashed (solid) lines represent negative (positive) lines. (b) Absolute value 2D spectra at population times T (left to right) 0 fs, 50 fs and 500 fs. The ratio of the maximum values of the 0, 50 and 500 fs plots is 2:1.6:1.

by 500 fs as a result of spectral diffusion from exciton population relaxation. Second, the negative feature in the spectra decreases in intensity with respect to the positive feature as population time increases. Modeling shows that this results mostly from exciton relaxation. Figure 2b shows the absolute value 2D spectra at r=0, 50 and 500 fs. The 2D spectra in Figures 2a and 2b can be calculated from the full third-order response function R^^^(T,T, t). For the remainder of this paper we will focus on the absolute value spectra in Figure 2b. In the 2D frequency plane (0)^ ^cojesich contribution to the signal is centered on its emission (cojsind absorption (cOj) frequency points with a width related to the decay of coherences during r and t.

555

Population transfer during the population period, T, broadens the anti-diagonal width of the 2D spectrum as the initial and final transitions occur at different frequencies. Thus measuring the anti-diagonal width of the absolute value spectrum as a function of frequency reveals the frequency dependence of the relaxation. Figure 3a shows the experimental anti-diagonal width as a function of frequency. The width in the high frequency region increases substantially as T increases, as a result of the exciton population relaxation. Calculations based on modified Redfield theory and allowing for exciton relaxation induced dephasing during all time periods as well as the population transfer in the second time period (described with the master equation) [7, 8] are shown in Figure 3b. Figure 3c shows simulations without the exciton relaxation. Figures 3b shows that exciton relaxation is required to reproduce the experimental trend. SmulationswithoU Exciton Relaxertion

T=50fs T=500fs

Position along the diagonal (cm*^) Smuiationsvuth Exciton Relaxation

-T=50fs T=5(X)fs| )

^ioo

'

5

'

ioo

2

Position along the diagonal (cm'^)

16600

16800

17000

Fig. 3. Anti-diagonal width of the absolute-value 2D spectrum as a function of frequency relative to the 2D peak value for population times r = 50 fs (solid line) and T = 500 fs (dashed line), (a) Experiment, (b) calculation including exciton relaxation, (c) calculation without exciton relaxation and (d) simulated absorption spectrum with frequency dependent relaxation (depopulation) rate superimposed.

Frequency (cm"^)

In summary, our two-dimensional photon echo studies of BIC J-aggregates demonstrate the frequency dependent dynamical behavior across the exciton band (Fig. 3d) explicitly. The method also retrieves renormalized values of the static disorder and the electron-phonon coupling strength. The method can directly be applied to biological aggregates and solid state systems. Acknowledgements. This work was supported by the National Science Foundation

References 1. D.M. Jonas, Ann. Rev. Phys. Chem. 54 425 (2003). 2. T. Brixner, I.V.Stiopkin, and G.R.Fleming C^t. Lett. 29, 884, 2004. 3. T. Brixner, T. Mancal, I. V. Stiopkin, and G. R. Fleming, J.Chem. Phys., in press. 4. J. P. Ogilvie, et al., in Ultrafast Phenomena XIII, edited by R. J. D. Miller, et al. Springer, 2003, p. 571, M. L. Cowan, et al, Chem. Phys. Lett. 386, 184, 2004. 5. C.Dorrer, et al., J. Opt. Soc. Am. B 17, 1795, 2000. 6. J.D. Hybl, A. A. Ferro, and D.M. Jonas, J. Chem. Phys. 115, 6606, 2001. 7. K. Ohta, M.Yang, and G.R.Fleming , J. Chem. Phys. 115, 7609, 2001. 8. M.Yang, and G.R.Fleming, Chem. Phys. 275, 355, 2002.

556

2-Dimensional Measurement of the Solvent Intermolecular Response in Solvation Sungnam Park, Jeongho Kim, and Norbert F. Scherer Department of Chemistry, the Institute for Biophysical Dynamics and the James Franck Institute, University of Chicago, Chicago, IL 60637 E-mail: [email protected] Abstract. The solvent intermolecular response in solvation is studied by 2-D polarizability response spectroscopy. The isotropic and anisotropic solvent response spectra of Coumarin 153 in CH3CN, measured during solvation, reveal comparable translational vs. orientational dynamics.

Although the collective chemical intuition about the affect of solvent properties (e.g. polarity, dielectric constant) on chemical reactivity is encyclopedic, our understanding of solvent dynamics during the reaction is less comprehensive. Time-resolved fluorescence Stokes-shift and photon echo spectroscopies^'^ have been able to extract the time-integrated solvent spectral density in reactive processes, particularly solvation. However, what is really desired is a measurement of the instantaneous dvnamics during the reaction. This paper describes such a method with a C153CH3CN system. See refs.^'"* for background. As shown in Fig. 1, 2-dimensional polarizability response spectroscopy (2D-P0RS) is performed in an optical heterodyne detected-transient grating (OHD-TG) geometry; a pump field that is resonant only with a solute is added. The local oscillator field allows selectively measuring the real part of the third-order polarization. A nonequilibrium solutesolvent state is created by electronic excitation of a solute with a resonant-pump pulse at r=0 ps. After a time delay of T, relaxation of the entire system is monitored by OHD-TG spectroscopy scanning two pairs of pulses over time "/". Fig. 1. Experimental scheme and beam geometry. En.pu is the resonant pump field, and Epu, Epr, and ELO are the off-resonant TG pump, probe, and local oscillator fields, respectively, for OHD-TG spectroscopy.

557

This approach enables observing the relaxation of the nonequilibrium solute-solvent state to a new equilibrium state as the solvent undergoes a time-dependent reorganization around solute. The OHD-TG spectrometer employed a diffractive optic to arrange the 800 nm beams in a boxcar geometry.^ The second-harmonic (resonant pump) pulse, generated in a BBO crystal, was directed to the center of the square (see Fig. 1). The 2D-P0RS signal, measured with lock-in detection by chopping the resonant pump, was 90° out-of-phase relative to the LO field. Thus, the 2D-P0RS experiment measures the /-(time)dependent change in the index of refraction of the system induced by the resonant pump at successive delay times, T\ i.e. S{Tj) • The 2D iso- and anisotropic responses of CI 53 in CH3CN are calculated from zzzz and yyzz tensor elements. The anisotropic response selectively measures orientational relaxation, of both solute and solvent, with respect to a solute transition dipole. The isotropic response is sensitive to the isotropic change in the system (e.g. local solvent density around solute induced by translational interactions). In both responses, a drastic change occurs at T< 1.0 ps and an asymptotic behavior is observed with increasing T, At early time (in a few picoseconds), nonequilibrium solvent dynamics are observed in both responses. On a longer time scale, the isotropic response at each T contains a constant offset while the anisotropic response contains a long decay component. The signal measured in 2D-P0RS is the molecular response that is induced by the resonant-pump (i.e., pump-on - pump-off) and contains a component that is the difference in local solvent molecular structure (configuration) around the excited vs. ground state solute in addition to a change in solute (e.g. solute reorientation). Therefore, the signal at a given T can be described by the contributions from the solvent around solute (solvation) and the solute SiD-poRsi^'^T^) = S'''''"'''"{t\T)-¥S'"'"'\t'J) (1) These two different dynamics are separable in time because the nonequilibrium solvation dynamics occur on a picosecond time scale while the solute dynamics are on an order of magnitude longer time scale. The solute reorientation that appears in the anisotropic response is well described by a single exponential and its time constant agrees well with fluorescence Stokes shift measurements.^ The solute response in the isotropic response can be described as a change in either the solute size or the solvent cavity around solute; e.g., as in the viscoelastic model for nonpolar solvation.^ The amplitudes of longer time components in both responses converge with increasing T\ i.e. as the solvation progresses. The plot of amplitudes of longer time components against T gives the timescale of solvation. The solute contribution on the longer time scale can be properly removed from the 2D-P0RS signal based on Eq.(l) to yield nonequilibrium solvent dynamics in early times. Fourier deconvolution of nonequilibrium solvent responses in the time domain enables calculation of susceptibilities in the frequency domain.'* Fig. 2 shows the evolution of solvent

558

molecular spectra along the solvation time T. The isotropic response is of a little bit greater magnitude than the anisotropic response indicating that solvent translational and orientational motions have comparable involvement in solvation. The details of the spectral evolution with time differ between the iso- and anisotropic responses. However, even so, the trend is not simple and may reflect the frequency dependent coupling previously reported."* Higher frequency responses of the solvent are dominant in the isotropic response. Instantaneous susceptibilities (hence spectral densities at a given T) can be obtained by differencing of adjacent T-slices. For now, the modest S/N ratio doesn't allow obtaining a quality representation of the instantaneous susceptibility at each value of T, This response will have comprehensive information on solvent dynamics in solvation. The improvement of the S/N ratio of the 2D-P0RS measurement is in progress.

0

20 40 60 80 100 120 140 160 180 0

col cm'

20 40 60 80 100 120 140 160 180

a;/ cm'

Fig. 2. 2-D (frequency vs. T) contour representation of the isotropic (A) and anisotropic (B) PORS responses of solvent molecules around solute. Solid line is positive and dotted line is negative amplitude signal.

References (1) (2) (3) (4) (5) (6) (7)

Homg, M. L.; Gardecki, J. A.; Maroncelli, M. J. Phys. Chem. 1995, 99, 17311. Fleming, G. R.; Cho, M. Annu. Rev. Phys. Chem. 1996, 47, 109. Underwood, D. F.; Blank, D. A. J. Phys. Chem. A 2003,107, 956. Park, S.; Flanders, B. N.; Shang, X.; Westervelt, R. A.; Kim, J.; Scherer, N. F. J. Chem. Phys. 2003,118, 3917. Goodno, G. D.; Dadusc, G.; Miller, R. J. D. J. Opt. Soc. Am. B 1998,15, 1791. Homg, M. L.; Gardecki, J. A.; Maroncelli, M. J. Phys. Chem. A 1997,101, 1030. Berg, M. J. Phys. Chem. A 1998,102, 17.

559

Two-dimensional femtosecond coherent antiStokes Raman scattering spectroscopy using a chirped supercontinuum generated from a photonic crystal fiber Hideaki Kano and Hiro-o Hamaguchi Department of Chemistry, School of Science, The University of Tokyo, Kongo 7-3-1, Bunkyo, Tokyo 113-0033, JAPAN E-mail: [email protected] Abstract. Two-dimensional femtosecond CARS spectroscopy is performed using a Ti:sapphire oscillator. The CARS signal of cyclohexane exhibits well-defined beats with a period of 430fs, which agrees well with a frequency difference between two C-H stretching modes.

1. Introduction Femtosecond time-resolved coherent anti-Stokes Raman scattering (fs-CARS) attracts much attention because vibrational dynamics can be directly investigated [1-3]. In the present study, we explored a new type of the fs-CARS setup using the chirped Stokes pulse in order to study a vibrational mode-mode coupling mechanism. It is based on a two-dimensional CARS spectroscopy. This technique can be exploited to several applications. First, a doubly Raman excitation with different delay times can be performed. It corresponds to a six-wave mixing process. If there is a coupling between the doubly Raman excited vibrational modes, the signal can be enhanced. Second, vibrational coherence transfer can also be investigated [4]. If the vibrational coherence is transferred from a particular vibrational mode I^AIO another vibrational mode I^B, the CARS signal can be observed not only at the Raman resonance v^ but also at v^ in different delay time. For the Stokes pulse, we use a supercontinuum generated from a photonic crystal fiber (PCF). By injecting readily available low-power ultrashort pulses into the PCF, an ultrabroadband supercontinuum spanning more than one octave has been achieved [5]. Because of the easy handling and compactness, this technique is being extensively used in various applications [6, 7].

2. Experimental Methods A schematic overview of the experimental setup is shown in Fig. 1 (a). The light source is an unamplified Ti:sapphire laser (Coherent, Vitesse-800). The laser pulses are divided into the pump and seed pulses for the supercontinuum generation with the PCF (Crystal Fibre, NL-1.7-690). In the present study, the

560

fundamental of the Tiisapphire laser (co^) and the supercontinuum (co^) are used as the pump and Stokes pulses, respectively. As shown in the energy diagram indicated in Fig. 1 (b), three laser pulses, namely pumpl, pump2, and Stokes pulses, contribute to the fs-CARS process. Figure 1 (c) shows the time ordering of the three laser pulses. The Stokes pulse defines an arbitrary zero point. The delay times Tj (delay time between pumpl and Stokes pulses) and r2 (delay time between pump2 and Stokes pulses) can be varied in the measurement. In the present fs-CARS experiment, two pump pulses have the same wavelength (w^), and the Stokes pulse (co^) is set to a longer wavelength in such a way that the difference between co^ and co^ is resonant with a vibrational Raman transition of the sample. The sample is liquid contained in a 1-mm thick quartz cell.

Fig. 1. (a) experimental setup of femtosecond-CARS spectroscopy: BS, beam splitter; PMT, photomultiplier tube, (b) energy diagram, and (c) time ordering of three laser pulses. Note that the Stokes pulse is temporally chirped.

3. Results and Discussion Figure 2 (a) shows a two-dimensional intensity log plot of the fs-CARS signal of cyclohexane as functions of the two delay times, r^ and r^. The detected wavelength is 652.5 nm (2841 cm"^), which corresponds to the CH-stretching region. The CARS signal is observed along rj = 0 and T2 = 0. This is due to the fact that the CARS signal is generated in the electronically off-resonance condition. As expected, the signal is symmetric for r^^r^. The instantaneous signal measured at r^=T^ =0 is attributed to the nonresonant electronic response. A subsequent beating signal with a frequency of -^1/430 fs'^ are clearly observed in Fig. 2 (b). Such a beating signal is often observed in fs-CARS signal, and is ascribed to be interference between the CARS polarization fields generated from different vibrational modes. The Fourier power spectrum of the oscillatory component clearly shows a peak at 75cm"'. The value 75cm"' agrees well with the energy separation of the symmetric and anti-symmetric CH2-stretching motion (79cm"'). It means that the two vibrational modes are coherently excited at the same time.

561

(a)

75cm

—^— -~i0.0 -0.5

1 0.5

r~ 1.0

— j —

1.5

0.0

-r 0.5 Delay I / p s

1.0

1.5

Fig. 2 (a) Two-dimensional intensity log plot of the fs-CARS signal of cyclohexane. (b) Delay-time dependence of the fs-CARS signal. Inset: Fourier power spectrum of the oscillating component.

4. Conclusions In conclusion, two-dimensional femtosecond CARS spectroscopy is demonstrated using a Ti:sapphire oscillator and a PCF. If the vibrational coherence transfer occurs from a particular vibrational mode to another vibrational mode, an additional CARS signal can be observed along r^ ^0 and T2 ^^ 0. It is one of the advantages in the two-dimensional measurement of the fs-CARS signal using the chirped Stokes laser pulse. Although the signal due to such a mode-mode coupling is not observed in the present study, the two-dimensional CARS setup enables us to open a new possibility in time-resolved vibrational spectroscopy. Acknowledgements. H. K. is supported by a Grant-in-Aid for Young Scientists (B) (No. 15750005) from Japan Society for the Promotion of Science, and research grants from The Kurata Memorial Hitachi Science and Technology Foundation, and Iketani Science and Technology Foundation.

References 1. R. Leonhardt, W. Holzapfel, W. Zinth, and W. Kaiser, Chem. Phys. Lett. 133, 373 (1987). 2. H. Okamoto and K. Yoshihara, J. Opt. Soc. Am. B 7, 1702 (1990). 3. W. Kiefer, A. Matemy, M. Schmitt, Naturwissenschaften 89, 250 (2002). 4. M. Hayashi, Y. Nomura, and Y. Fujimura, J. Chem. Phys. 89, 34 (1988). 5. J. Ranka, R. Windeler, and A. Stentz, Opt. Lett. 25, 25 (2000). 6. H. Kano and H. Hamaguchi, Opt. Lett. 28, 2860 (2003). 7. H. N. Paulsen, K. M. Hilligsoe, J. Thogersen, S. R. Keiding, and J. J. Larsen, Opt. Le^^. 28, 1123(2003).

562

Two-dimensional spectroscopy by spectrally resolved real-time resonant coherent Raman scattering in polydiacetylene N. Ishii , E. Tokunaga , S. Adachi , T. Kimura , H. Matsuda , and T. Kobayashi 'Department of Physics, Graduate School of Science, University of Tokyo, 7-3-1 Kongo, Bunkyo-ku, Tokyo, 113-0033, Japan E-mail: [email protected] ^Department of Physics, Graduate School of Science, Tokyo University of Science, 1-3 Kagurazaka, Shinjuku-ku, Tokyo, 162-8601, Japan ^National Institute of Advanced Industrial Science and Technology, 1-1-1 Higashi, Tsukuba, Ibaragi, 305-8561, Japan Abstract. Spectrally resolved real-time coherent resonant Raman scattering was observed by a 4-fs ultrashort pulse with a spectrally resolving multi-channel lock-in amplifier and was identified by a theoretical calculation. An introduced optical frequency- and vibrational time-resolved two-dimensional spectrum reveals the electronic ground-state dynamics below the absorption edge.

1. Introduction We report the observation of spectrally resolved real-time coherent resonant Raman scattering induced by a 4-fs ultrashort-pulse laser [1] with a multi-channel lock-in amplifier, which makes it possible to detect small signals at multi-probed-wavelengths. Fig. 1 shows a schematic of a nonlinear process which is mainly used in this experiment to investigate the ground-state dynamics without coherent excitation of the excited-state dynamics. A theoretically calculated nonlinear polarization which takes the ground-state wavepackets into account reproduced fairly well the observed dynamics. It is confirmed that the ground-state vibrations play a significant role in the dynamics of the wavepacket even far below the absorption edge as well as the excited-state vibrations. In this configuration, contribution of coherent excitation of the excited-state dynamics is depressed and the ground-state dynamics will mainly contribute to observed signals.

2. Theory and Experimental A transmittance change of the probe pulse as a function of a delay-time T between the pump and the probe and the probed optical frequency CO is calculated by a nonlinear spectroscopy calculation [2-4] for the nonlinear process in Fig. 1 (right) as - Q sin( (O^T) + y^ cos( AT{6?,T)/ P{a)) o^ (co^ - co^)cxp( -r IT^) ^\ Q + ^2 W\i\\Q. = co-(o^+co^,

Y2=\IT2,

hCD^=E^ and ho)^ = E^ (1).

(O^T)

P{co),

563

Ti, and 7'4, which are a pulse energy of the incident probe, an electronic transverse dephasing time constant, and an vibrational transverse dephasing time constant between vibrational states of the ground-states, respectively. 1

11

^nrrc\ e03rm

j^ljt J^^M^VV^AAM/v^^^A^»vvv•

^

Stationary absorption TraisniLssicn aoo^ ctage by ncnlinear 5io: polarizatioi ao5: LEiser spectnin

-100 0 100 200 300 400 500

Intensity

RBCVBTy(aTi^)

DelaytiiTB(fs)

Fig. 1. Schematic diagram of the 3rd-order nonlinear processes used in this experiment (left), Real-time transmission changes (center) at 6 probed wavelengths, and FFT power spectra (right) of periodic signals of the corresponding real-time changes. P^^^: 3rd-prder nonlinear polarization. Eg: absorption energy of 0-0 transition, and Ey: vibrational energy. From the equation, the modulated pump-probe signals by different vibrations appear at the different probed wavelengths, because a vibration with its angular frequency (o^ modulates the pump-probe signal most at C0 = CO^— 0)^ of the probed wavelengths, which are different for different vibrations and can be located far below the absorption edge (0-0 transition in this case). Experimentally the transmittance changes were measured by a usual pump-probe measurement with the 4-fs ultrashort-pulse laser and lock-in amplified by the multi-channel lock-in amplifier at 128 wavelengths simultaneously.

3.

Result and Discussion

In Fig. 1 (center and left) typical real-time transmittance changes and corresponding Fourier transform power spectra at 6 wavelengths are selected. In the real-time transmission changes, positive transmittance changes found at 571 and 603 nm can be attributed to the bleaching of the ground-state population, which is not included in the caluclation. At the other wavelengths, where neither bleaching nor photo-induced absorption was found, periodic modulations around Til = 0 are observed as predicted by the Eq. Two prominent vibrational modes are found in the FT power spectra around 1500 and 2100 c m ' , which can be identified with the stretching modes of carbon-carbon double- and triple-bond in the PDA main chain [5]. The relative intensities of C=C and C=C stretching are changing gradually with the probe wavelength from 603 to 682 nm. To investigate more detailed tendency, FT power spectra at all wavelengths are shown in Fig. 2.

564

Power (arb. units) 2750 -

:r

2500 2250 -

2100 c m '

iip**-

1250 1000 -

,.^^,^.,., 1500 c m ' '

1750 -

P:

750 -

500 4"

560

580

600

620

640

660

68 0

700

720

74

W a v e l e n g t h (nm

Fig. 2. FT power spectra of the periodic components of the pump-probe signal over the all probed wavelength. X, y, and z axes correspond to the probed wavelength (nm), FT frequency (cm"^), and FT power (arb. units), respectively. In Fig. 2, two broad peaks along the probed wavelength are found around 570 and 630 nm for 1516 cm'^ mode, and the two more signals are found around 570 and 670 nm for 2089 cm'^ mode. The energy separation between each peaks for the 1516 cm"^ mode and the 2089 cm"^ mode corresponds well to each vibrational energy. This fact is clearly indicated in the Eq. as the periodically modulated signal appears at 0) = 0)^—0)^. As predicted in the calculation, this result clearly shows the spectrally resolved real-time impulsive resonant Raman scattering of the ground-state vibrations, which had not been achieved by this time. The theoretical calculation and experiment have revealed that a pump-probe signal is affected by the ground-state dynamics at the energy far below the absorption energy. From a different point of view, in this two-dimensional diagram of Fig. 2, the spectral width of the signal (nm, horizontal axis) is related to the electronic dephasing time T2 and the signal spectral width as a function of frequency (cm"^ vertical axis) independently reflects the vibrational dephasing time T4. This is one of the advantages of this two-dimensional diagram, in which an electronic and vibrational dephasing times are clearly separated along the two different axes. This can be applied to a direct way to measure the transverse dephasing time between the vibrational excited-state of the electronic excited-state and the vibrational ground-state of the electronic ground-state. Whereas a stationary absorption shows the complex and broad structures by many vibration modes and broadenings, the two-dimensional diagram has resolved the overlapped vibrational spectra to each vibrational mode,

References 1 A. Baltuska, T. Fuji, and T. Kobayashi, Opt. Lett. 27, 306, 2002. 2 S. Mukamel, Principles of Nonlinear Optical Spectroscopy, Oxford University Press, New York, 1995. 3 Y. J. Yan and S. Mukamel, Phys. Rev. A 41, 6485, 1990. 4 A. T. N. Kumar, F. Rosea, A. Widom, and P. M. Champion, J. Chem. Phys. 114,701,2001. 5 W. F. Lewis and D. N. Batchelder, Chem. Phys. Lett. 60, 232, 1979.

565

Optical Two-dimensional Fourier-transform Spectroscopy of Semiconductor Quantum Wells C. N. Borca, T. Zhang, and S. T. Cundiff JILA, University of Colorado and National Institute of Standards and Technology, Boulder, CO 80309-0440, USA E-mail: [email protected] Abstract. We demonstrate optical two-dimensional Fourier transform spectroscopy of heavy- and light-hole excitons in GaAs quantum wells. This is enabled by active interferometric stabilization of the excitation pulse separation and of a reference pulse. Time domain optical spectroscopy has tended to recapitulate nuclear magnetic resonance (NMR). Recently, there has been much interest in optical implementations of multi-dimensional Fourier transform spectroscopy, an NMR technique that was developed in the 1970s [1]. Molecular vibrations have been probed by implementing two-dimensional Fourier transform spectroscopy in the infrared [2] and electronic excitations in molecules probed in the near-infrared [3]. We present two-dimensional Fourier transform spectroscopy (2DFTS) of excitons in semiconductors. Two-dimensional spectroscopy is an enhanced version of transient four-wave-mixing (TFWM), which has proven powerful for studying many-body interactions in semiconductors [4], and thus two-dimensional spectroscopy may provide further insight. Two-dimensional spectroscopy is particularly promising for looking at the effects of disorder in semiconductors because it excels at determining whether or not resonances are coupled. Indeed, there is an unresolved question as to whether or not excitons in quantum wells localized by width fluctuations interact with one another, which is phrased as the question of quantum beats versus polarization interference [5]. Measurements based on various non-Fourier transform two-dimensional techniques yield contradictory answers [6, 7]. Two-dimensional Fourier-transform technique promises to resolve these issues. Two-dimensional Fourier-transform spectroscopy is based on TFWM, the key enhancement being that full amplitude and phase information of the emitted signal is obtained as a function of delay between the first two excitation pulses, which is controlled with interferometric accuracy. The beam geometry is shown in Figure 1. By taking interferometrically accurate steps between excitation pulses, it is then possible to take the Fourier transform with respect to the time between excitations pulses (x). We actively stabilize the delay between the pulses, as opposed to simply measuring the delay and resampling, as was done for electronic excitations in molecules [3]. This is achieved by monitoring the interference fringes from a He-Ne laser that co-propagates with the femtosecond pulses. The fringes produce

566

H«) ^ / ''

1^ Spectrometer detects S((o)

Fig. 1. Schematic of the beam geometry used for optical two-dimensional Fourier transform spectroscopy. The three pulse sequence creates a third order coherence that radiates, generating E(co), and can be spatially separated from the other pulses. The amplitude and phase of the nonlinear signal radiated by the sample while scanning delay x is recovered using a collinear reference pulse E^{(o) and spectral interferometry. Fourier transforming the recovered signal SipS) along the time axis x results in a two-dimensional spectrum with axes cOt and CD,. an error signal for a feedback loop that adjusts the path length. The feedback is disabled while the delay is changed. In addition, the change in the overall phase of the emitted field must be measured, not just the relative phases across the spectrum. This requires that the spectral interferometry reference pulse also have a stable phase [8]. This is also achieved by active stabilization. For the first 2DFTS measurement of semiconductors, we choose to study the heavy-hole (hh) and light-hole (Ih) exciton resonance in a GaAs/AlGaAs multiquantum well sample with 10 periods of 10 nm wells and barriers. Bulk GaAs has two degenerate valence bands (hh and Ih bands), quantum confinement breaks the degeneracy so that two distinct bands and associated excitons can be observed. The amplitude, real and imaginary parts of the two-dimensional spectra for linearly polarized excitation is shown in Figure 2. Diagonal peaks correspond to the hh and Ih excitations, off-diagonal peaks indicate that they are coherently coupled to each other, i.e., there is an excitation sequence where the first pulse excites the hh exiton, but the corresponding coherent emission occurs at the Ih exciton, and vice-versa. For linearly polarized excitation, this is expected since each can couple a given sub-state in conduction band to either a hh or Ih valence band state. Based on this simple level diagram, it would be expected that the resonances are uncoupled for co-circularly polarized excitation and thus should show no off-diagonal peaks in two-dimensional spectra. However prior work using time-resolved TFWM [5] has shown that the hh and Ih excitons still appear to be coupled in this case [9]. 2DTFS measurement using co-circularly polarized pulses is in progress to see if the same conclusion is reached.

567

Magnitude

Imaginary

Real

Emission Frequency (THz)

Fig. 2. Two-dimensional spectrum for a GaAs/AlGaAs quantum well structure, separated in amplitude, real and imaginary parts. The diagonals are guides for the eye, and contour spacing is set to ~10%. Dotted and solid contours in the real and imaginary spectra indicate negative and positive values, respectively.

In summary, we have presented the first optical two-dimensional Fouriertransform spectroscopy of exciton resonance in semiconductors. This technique provides important insight into exciton dynamics in semiconductors. Of particular interest is localized excitons, either due to disorder or purposeful confinement in quantum dots, where interaction between excitons is not well understood.

References 1 R. R. Ernst, G Bodenhausen, and A. Wokaun, in Principles of Nuclear Magnetic Resonance in One and Two Dimensions (Oxford University Press, Oxford, 1989) 2 O. Golonzka, M. Khalil, N. Demirdoven, and A. Tokmakoff, Phys. Rev. Lett. 86, 2154, 2001 3 J. D. Hybl, A. W. Albrecht, S. M. G. Faeder, and D. M. Jonas, Chem. Phys. Lett. 297, 307, 1998 4 D. S. Chemla and J. Shah, Nature 411,549,2001 5 M. Koch, J. Feldmann, G von Plessen, E. O. Gobel, R Thomas, and K. Kohler, Phys. Rev. Lett. 69, 3631,1992 6 M. Koch, J. Feldmann, E. O. Gobel, R Thomas, J. Shah, and K. Kohler, Phys. Rev. B 48, 11480,1993 7 A. Euteneuer, E. Finger, M. Hofmann, W. Stolz, T Meier, R Thomas, S. W. Koch, W. W. Rtihle, R. Hey, and K. Ploog, Phys. Rev. Lett. 83, 2073, 1999 8 L.Lepetit, G Cheriaux, and M. Joflfre, J. Opt. Soc. Am. B 12, 2467, 1995 9 A. L. Smirl, M. J. Stevens, X. Chen, and O. Buccafusca, Phys. Rev. B 60, 8267, 1999

568

Degenerate four-wave mixing spectroscopy based on two dimensional pulse shaping Thomas Homung, Joshua C. Vaughan, T. Feurer, and Keith A. Nelson Massachusetts Institute of Technology, Cambridge, MA 02139, USA E-mail: [email protected] Abstract. Degenerate, noncollineai*, optically heterodyned, spectrally resolved, multidimensional femtosecond y^^^ measurements are executed with waveforms, delays, and spectral phases and amplitudes of all fields controlled by a single active device in a fully phase matched beam geometry.

Femtosecond multidimensional nonlinear spectroscopy is a powerful tool for investigating the evolution of microscopic structure [1]. Among its recent achievements are separation of homogeneous vs. inhomogeneous time scales, and probing of the couplings between and relative orientations of different molecular modes [1,2]. Interesting new coherent control phenomena are expected to occur if interference among Liouville space pathways is exploited through control over the temporal shape of each interacting pulse. For instance, phase cycling can be used to select nonlinear signals of interest [3], and control the interferences among third order photon echo, virtual photon echo and fifth order echo contributions. More advanced spectral phase shaping has made single pulse femtosecond CARS microscopy possible [4]. Despite all its capabiHties, femtosecond multidimensional phase-coherent spectroscopy remains a technically demanding speciahzation in the optical regime. Here we present a method for performing phase-coherent multiple-pulse spectroscopy based on a single active element, with no delay lines, phase-locked loops, beamsplitters, or even reflectors for the separate beams. It is a twodimensional (2D) extension [5] of femtosecond pulse shaping, in which use of a 2D spatial light modulator (SLM) at the spectral plane of a zero-dispersion compressor allows a single input pulse to be divided into many slices which may be independently shaped or delayed. The method is adapted to multiple-pulse spectroscopy by using phase patterns with a vertical phase tilt in addition to the spectral phase [6]. The vertical phase tilt leads to diffraction of the beams of interest into first order (see Fig. 1 left). Diffraction allows for arbitrary amplitude and phase shaped pulses. A lens is used to focus the pulses from selected regions of the beam onto a sample. The noncolhnear phase matched geometry allows the generated signal to be detected in a unique direction. Furthermore, since all of the pulses travel through common optics, and no delay stages or beam splitters are used, the device possesses excellent phase stability. Here we demonstrate the capabilities of this method to perform frequency, timefrequency and pure time-domain multidimensional four-wave mixing (FWM) spectroscopy for samples in the gas, hquid and solid phase.

569

As a test system in the solid phase we choose LiNb03, due to its strong and well-characterized Raman response. The generation of phonon-polaritons in LiNbOs occurs via impulsive stimulated Raman scattering (ISRS). Since ISRS is a non-resonant process, only when the pump pulses are time-coincident is an excitation grating created. The probe pulse is coherently scattered by the generated phonon-polaritons, resulting in a third order signal in the direction ks=-ka+kb+kc. It was spatially filtered and imaged to a spectrometer for detection together with a local oscillator pulse, activated through a fourth phase region on the SLM. Here the excitation pulses where fixed in time at t=0 (pulse a and c) and t=1.5 ps for pulse b, where each was adjusted to be at the maximum of their carrier oscillation. The local oscillator was scanned in time over the duration of the probe pulse b, thus obtaining the spectral interferogram shown in Fig. 1 (a).

0.5

1

1.5

2

2.6

0.5

1

1.5

2

2.5

delay 2 (ps)

Fig. 1. Left: Schematic illustration of the experimental setup. Right: Frequency resolved heterodyne detection of the DFWM response as a function of the reference (local oscillator) delay, (a) Probe has phase 0. (b) Pump 1 with phase n. Simulation of heterodyne detected signal between a reference field and a second shaped electric field that is (c) Till phase shifted, and (d) 3 nil phase shifted. The shape of the spectral interferogram in Fig. 1 (a) reveals that the emitted signal is nil out of phase with each of the interacting pulses. Given the full phase stability of the device, one can now apply a n phase to each of the three interacting pulses with each extra n phase being directly reflected in the emitted polarization. For comparison. Fig. 1 (b) shows exemplarily the heterodyned polarization for an additional TC phase shift to pulse a. The emitted polarization in Fig. 1 (b) thus flips phase as compared to Fig. 1 (a). This result is expected for a third order perturbative interaction, where each pulse contributes one photon. The spectral interferogram simulation of the data above are shown in Figs. 1 (c) and (d). Next we show time-frequency domain FWM results obtained from resonant excitation of Rubidium atoms in the gas phase at a temperature of 390 K. The signal is homodyne-detected with a photodiode in the direction ks= ka-kb+kc. Beams a and b are spectrally amphtude modulated to each consist of two frequency components centered at the wavelengths 780 nm (5s-5p3/2) and 795 nm (5s-5pi/2), while pulse c is Fourier-limited and scanned in time. The observed 130 fs oscillations, due to coherent excitation of the 5p levels, are out of phase with each other if the phase of the 780 nm frequency component in beams a and b are toggled between 0 and 7i, respectively (Fig. 2 (a)).

570

Finally we present data with non-resonant DFWM spectroscopy on diiodomethane performed in a phase-matched BOXCARS geometry. This molecule in the liquid phase has a slow mode at 123 cm-1 (270 fs) (ICI bending mode) and a nearly four times higher mode at 488 cm-1 (68 fs) (symmetric C-I stretch). Here, the DFWM signal after excitation with the same phase related pulse pair in beams a and b, is recorded as a function of the simultaneous delay of probe and local oscillator (delay T). The different transients obtained as a function of the intrapulse separation of the pulse pairs (delay A) versus delay T, are shown in Fig. 2 (b). Fourier transform of the data reveals, that the Raman mode that has a frequency that matches an integer multiple of the current intrapulse separation is most efficiently excited. For instance at A=410 fs corresponding to 6 times the oscillation period of the high frequency mode and 1.5 times of the slow mode, the Fourier spectrum shows that only the high frequency mode is active (Fig. 2 (b)).

0

0.5

delay c (ps)

123

488

v-p(cm-'')

Fig. 2. (a) Resonant time-frequency FWM spectroscopy of Rubidium. Pump and stokes are shaped into coherent frequency pairs (780 and 795 nm). The time resolved FWM signals are out of phase with each other, depending on the relative phase within the frequency pair, (b) Non-resonant DFWM spectroscopy and control of diiodomethane. The intrapulse separation of the pulse pairs used for excitation controls which Raman mode is activated. In summary, we have presented four-wave mixing measurements using a novel spectroscopic method based on 2D pulse shaping. A simple change of the appHed phase pattern allows for control of the arrival times and temporal phase and amplitude of the interacting pulses with full phase stabiHty. Funding from NSF grant no. CHE-0212375, ARO grant no. DAAD10-01-10674 and DFG (Forschungsstipendium of T.H.) is acknowledged.

References 1. 2. 3. 4. 5. 6.

S. Mukamel, Annu. Rev. Phys. Chem. 51, 691 (2000), and references therein M. Khalil, N. Demirdoven, A. Tokmakoff, Phys. Rev. Lett. 90, 047401 (2003) P.F. Tian, D. Keusters, Y. Suzaki, and W.S. Warren, Science 300, 1553 (2003) D. Oron, N. Dudovich, and Y. Silberberg, J. Chem. Phys. 118, 9208 (2003) J.C. Vaughan, T. Feurer, and K.A. Nelson , Opt. Lett. 28, 2408 (2003) T. Hornung, J.C. Vaughan, T. Feurer, and K.A. Nelson , Opt. Lett.(2004) in print

571

Propagation and detection distortions of fourwave mixing signals: application to 2D spectroscopy Nadia Belabas and David M. Jonas Department of Chemistry and Biochemistry, University of Colorado, Boulder, CO 803090215 telephone (303)492-3818; fax (303)492-5894; e-mail: [email protected] Abstract. We present a computationally efficient three dimensional Fourier transform algorithm for four-wave mixing signal calculation in optically dense samples of arbitrary nonlinear response. Memory effects and lineshape distortion on integrated and bidimensional signals are demonstrated. In studies of microscopic dynamics of condensed phase systems via femtosecond four-wave mixing (4WM) experiments, the third order nonlinear response of the system is complicated by memory effects (because of the ultrafast timescale) and thus limited to a weak field regime where the maximum signal occurs at an optical density (OD) of 1. We present an exact solution of Maxwell equations for the 4WM signal field applicable to complicated response functions (in contrast with Maxwell-Bloch methods [1]), arbitrary OD (in contrast with perturbative methods [2]) and computationally efficient (in contrast with multiple convolutions [3]). After a brief description of the strategy behind the three-dimensional Fourier transform (3DFT) algorithm, we show the importance of modeling propagation effects using a simple integrated 2 pulse photon echo (2PE) from systems with frequency memory. A level of complexity is added as we illustrate how phase matching geometry and crossing-angle smearing can distort real and imaginary 2D spectra (two-dimensional signal dispersed along an excitation and a detection frequency). As 2DFT spectroscopy explores different local environments (solvent coordinates) via the measurement of the time-evolution of lineshapes of real and imaginary two-dimensional spectra, accurate and complete modeling is again crucial to disentangle real microscopic and molecular effects from signal distortions inherent to the experimental conditions of ultrafast 4WM and specifically 2D experiments. The 3DFT algorithm starts with an arbitrary third order time-domain nonlinear response and does an inverse 3DFT to get the third order susceptibility, from which the 3D frequency domain signal can be deduced for a monochromatic wave triplet [4]. Provided broadband excitation pulses at the entrance of the sample are known, another 3DFT yields the 3D time-domain field for all pulse delays between excitation pulses [5] Algorithmically, the impulse response in only evaluated once on each point of a grid, which is essential when evaluation of the impulse response itself is computationally costly (for example microscopic simulation, breakdown of the rotating wave approximation).

572

We validated the algorithm by calculating propagation effects on integrated impulsive 2PE signals (detection of the energy of the 4WM signal emitted after a sequence of two coUinear delta pulses separated by a variable time delay x) for an optical Bloch model, and then explored several two-level systems with frequency memory characterized by Kubo's correlation function (^da)(0)Sco(t)) = A^ exp(-r/T^) . At ODs where perturbation theory is no longer valid, a slowing of the positive x 2PE decay with increasing OD was found for systems with memory.[5] This qualitatively new propagation distortion is not particular to the stochastic model and is also encountered in other two-level systems with frequency memory. The exact solution of Maxwell's equations for TEM waves can be generalized to non-collinear beams, finite beam cross-sections and pulse durations by tri-linearly superposing signal fields from a distribution of excitation wave vectors and incorporation the spectral filtering induced by the pulse spectra. The principal complication arises because electromagnetic waves in absorbing samples are not transverse for oblique incidence. For a non-collinear excitation by the pulses a, b and c, an interferometric detection with a field E^ and assuming well collimated beams and a slowly varying propagation matrix II^"^^ [5], the total interference signal in the 3D frequency domain can be written S{oj„co^,co„oJ,)^^[U''\klk!,k:,d) 2-„c

(1)

where the propagation matrix 11^^^ which connects the signal field to the third order polarization depends only on the linear optical properties of the sample and windows. II^"^^ incorporates all departures from the radiation by an infinite plane of nonlinearly polarized material in vacuum. The directional filter O^^^ is the result of a quadruple integration over wave-vector distributions for pulses a, b, c and d (centered on k^^a,b,c,d )• It is a function only of the frequencies and depends parametrically on the phase-matching geometry. It corresponds to the spectral filtering originating from the crossing angle between the excitation pulses. The algorithm's simultaneous calculation of all signal fields (for all interpulse delays) makes it well suited to calculation of multidimensional signals. To illustrate this, the various distorting effects accounted for by equation (1) on the theoretical 2D spectra of a dye molecule (multi-mode Brownian oscillator model [6] representing solute, solvent, and mixed solute/solvent motions for HDITCP) have been calculated.

573

lilliig.

Me^il

'

'

•

^

'

•

-

litem V X''"'

Plane vmv^ rXmm^

M^Nhm^ II-' V

X m

-J2

In summary the 3DFT algorithm enables efficient calculation of FWM signals, including propagation distortions, for samples with a complicated nonlinear response. The dynamic range is sufficient even for frequency dispersed signal fields and multidimensional spectra over the entire experimentally important range of optical densities. The only restrictions on the nonlinear impulse response are imposed by the grid of the calctilation.

References R. W. Olson, H. W. H. Lee, F. G. Patterson and M. D. Payer, J. Chem. Phys. 76, 31 (1982). M. Bonn, S. Woutersen and H. Bakker, Opt. Commun. 147, 138-142 (1998). N. F. Scherer and J. A. Griietzmacher, Opt. Lett. 28, 573 (2003). D. Keusters and W. S. Warren, J. Chem. Phys. 119 (8), 4478-4489 (2003). O. Kinrot and Y. Prior, Phys. Rev. A 51, 4996-5007 (1995). M. N. Belov, E. A. Manykin and M. A. Selifanov, Opt. Commun. 99,101-104 (1993). S. Yeremenko, M. S. Pshenichnikov and D. A. Wiersma, Chem. Phys. LeU. 369, 107(2003). N. Bloembergen and P. S. Pershan, Phys. Rev. 128, 606 (1962). N. Belabas and D. M. Jonas, Opt. Lett, accepted (2004), N. Belabas and D. M. Jonas, "3D view of signal propagation in femtosecond four-wave mixing with application to the boxcars geometry"submitted (2004). John D. Hybl, Anchi Yu, Darcie A. Farrow and David M. Jonas, J. Phys. Chem. 106, 7651-7654 (2002)

574

Fourier Transform Measurement of Two-Photon Excitation Spectra: Applications to Microscopy and Quantum Control Kevin J. Kubarych, Jennifer P. Ogilvie, Antigoni Alexandrou, and Manuel Joffre Laboratoire d'Optique et Biosciences, UMR CNRS 7654 — INSERM U451 Ecole Polytechnique — ENSTA F-91128 Palaiseau, France E-mail: [email protected] Abstract. We report a novel Fourier transform method of measuring two-photon excitation spectra. We demonstrate this method using a simple dye system and discuss its applications in two-photon fluorescence microscopy and quantum control.

The use of nonlinear light-matter interactions is experiencing explosive growth in several new areas including such diverse applications as microscopy[l-5] and quantum control.[6,7] Two-photon fluorescence microscopyjl] for example, exploits the enhanced spatial resolution due to the nonlinearity of the excitation, while using near-IR excitation that is less strongly scattered than conventional visible sources. Nonlinear absorption has also been used to preferentially excite multiple dye molecules[2,3] as well as to demonstrate learning-loop-based quantum control. [6,7] While microscopy and control may seem to be unrelated, in each of these various contexts, the physically relevant parameter is the electric field at the appropriate order of interaction. For example, the two-photon excitation of a dye molecule depends on the second harmonic (SH) spectrum of the laser, which is intimately determined by the phase of the fundamental. We present a Fourier transform method for obtaining two-photon excitation (TPE) spectra, and we use phase-only pulse shaping to produce tunable narrow-band SH fields. Using the TPE spectra thus obtained, we show that a simple quantum control experiment[7] can be understood by comparing the SH spectrum of the shaped laser pulse with the TPE spectrum of the medium. The SH spectrum of a laser pulse can be determined through the Fourier transform of a second-order interferometric autocorrelation (IAC).[8] Similarly, the Fourier transform of the I AC signal arising from the fluorescence of a dye is the product of the laser's SH spectrum and the TPE spectrum of the dye. Thus simultaneous I AC measurements of the fluorescence and SH laser spectrum permit the determination of the TPE spectrum of the dye. To perform the lAC measurements, a sub-20 fs Ti:sapphire oscillator was sent into either a Michelson interferometer or a Dazzler acousto-optic pulse shaper,[9] which can act as a Michelson by creating two pulse replicas at variable delays. 60 percent of the output of the interferometer was focused into the dye sample, and the fluorescence signal was detected with a photomultiplier tube. The remaining 40% was focused into a GaAsP two-photon photodiode to characterize the laser's SH spectrum. The analog signals were digitized at 5 MHz. The same setup was used for the quantum control experiment discussed below.

575

Two dyes, coumarin 460 and 540, were studied using the Michel son interferometer setup. Figure lA show^s their measured TPE spectra, along with their one photon absorption spectra. In molecules of higher symmetry, one might expect a difference between the direct and nonlinear absorption.[10,ll] In the present examples, however, we find very clear agreement over our spectral range. The nonlinear response characteristics of many dye molecules have been studied carefully by Webb et aL, [1,10,11] and while the two-photon excitation spectra have been measured, they were obtained using a tunable narrow-band source. Our goal here is to employ a single source to measure the spectrum over the relevant bandwidth, while demonstrating the capability of determining the TPE spectrum by use of an lAC. 0.6-

3

A

oin

3

.--«*-s:

0.5-

.*.**^^®^°° sSD "(C 0.4i t e e ^ ^fli^r^^a rnrninlP*' c V r, ^ ^J ^ O) T CO 0.3LU * ° ^ * ^ ^ CL 0.2H

(0

o "o

* ^

DO

•

0.1-

• ••

I

0.0-

T

1

T"

"T

Wavelength (nm)

r

*~r~

390

Wavelength (nm)

Fig. 1. (A) TPE spectra of coumaiin 460 (filled circles) and 540 (empty squares) along with linear absolution spectra (solid and dashed lines, respectively). (B) Two-photon spectra of phase-only shaped pulses for different phase offsets. It was shown recently that a sinusoidal phase modulation can be used to tune the SH spectrum of a laser pulse.[2,3] To demonstrate the Dazzler's ability to tune the SH laser spectrum in a similar manner, the full laser power was sent into the Dazzler and the output was focused into the two-photon photodiode. Fig. IB shows SH laser spectra obtained by the I AC measurement with the Dazzler for various sinusoidal phase profiles. The phase modulation—periodic in wavelength—had a frequency of 0.04 nm'\ an amplitude of 7JI, and a phase offset that was varied to tune the central wavelength of the narrow second harmonic band. The central wavelength of each ^5 nm wide band depends essentially linearly on the phase offset. The ability to characterize both the TPE spectrum of a dye and the SH spectrum of an arbitrarily phase-shaped pulse provides insight into a simple quantum control experiment. The experiment consists of maximizing the ratio of fluorescence emission of coumarin 460, to SH pulse intensity measured with a two-photon diode. To perform the experiment, we used the Dazzler pulse-shaper controlled by a genetic search (GA) algorithm,[12] applying phase-shaped pulses to optimize the ratio. The two photon excitation maximum of coumarin 460 lies to the blue of the SH laser spectrum (see Fig 1.), while the response of the two photon diode is spectrally flat. Thus we expect that optimum solutions found by the GA search should be characterized by two photon spectra that are tuned to the blue edge of the unshaped SH laser spectrum. This expectation is verified in Fig. 2, where we

576

CD v ^ 2

3 CD

j ^0.6-

XO.4-

0.2-

740

760

780

800

Wavelength (nm) Wavelength (nm) Fig. 2. GA-optimized measuied two-photon spectrum (dots), calculated (solid) shown with the transform-limited spectmm (dashed) for a polynomial basis. (B) Fundamental spectral intensit}^ (solid) and applied phase (dashed). show the measured SH spectrum of a representative optimum pulse shape found for phase-shaping with a 12^^ order polynomial basis. The maximum ratio found by the GA was 1.3 (where a ratio of 1 corresponds to a transform-limited pulse). In summary, we have demononstrated that phase-only pulse shaping with the Dazzler can efficiently tune the SH spectrum of a laser. As a pulse-shaping device, the Dazzler has the advantage of being quantitative, with no calibration required. Moreover, it is faster than spatial light modulator based methods, making it possible to alternate between two pulse shapes at a 10 kHz frequency. This provides the ability to switch between pulse shapes in microscopy applications. We have also demonstrated a method for measuring TPE spectra. This measurement should have applications in microscopy, and in the interpretation of simple quantum control experiments such as the one discussed above. In any nonlinear absorption-based control experiment, it is crucial to examine the effect of spectral overlap. In this simple example, a measurement of the two photon excitation spectrum of the dye, and the SH spectrum of the optimum pulse provided an intuitive understanding of the result of the control experiment. 1. C. Xu, W. Zipfel, J. B. Shear, R. M. Wilhams, W. Webb. Proc. Natl Acad. ScL USA, 93, 10763-8, (1996). 2. K. A. Walowicz, I. Pastirk, V. V. Lozovoy, M. Dantus. J. Phys. Chem. A, 106, 93713. (2003). 3. V. V. Lozovoy, I. Pastirk, K. A. Walowicz, M. Dantus. J. Chem. Phys., 118, 318796, (2003). 4. J. X. Cheng, X. S. Xie. J. Phys. Chem. B, 108, 827-40, (2004). 5. N. Dudovich, D. Oron, Y. Silberberg. Nature, 418, 512-4, (2002). 6. T. Brixner, N. H. Damrauer, P. Niklaus, G. Gerber. Nature, 414, 57-60, (2001). 7. T. Brixner, N. H. Damrauer, B. Keifer, G. Gerber. J. Chem. Phys., 118, 3692-701, (2003). 8. K. Naganuma, K. Mogi, H. Yamada. IEEE J. Quantum Electron., 25, 1225-33,(1989). 9. P. Tournois. Opt. Commun., 140, 245-9, (1997). 10. C. Xu, J. Guild, W. Webb, W. Denk. Opt. Lett., 20, 2372-4, (1995). 11. C. Xu, W. Webb. J. Opt. Soc. Am. B, 13, 481-91, (1996). 12. D. Zeidler, S. Frey, K.-L. Kompa, M. MoXzkm Phys. Rev. A, 64, 023420, (2002).

577

Part VIII

Biology

Watching proteins function with picosecond time-resolved X-ray crystallography Philip Anfinrud^ Friedrich Schotte^ and Michael Wulff^ ^ Laboratory of Chemical Physics, NIDDK, National Institutes of Health, Bethesda, MD 20892, USA E-mail: [email protected]; [email protected] ^ European Synchrotron Radiation Facility, Grenoble Cedex 38043, FRANCE E-mail: wulff@esrf fr Abstract. We have developed the method of picosecond time-resolved crystallography and used this technique to investigate structural dynamics in biological macromolecules at ambient temperature. Time-resolved snapshots of myoglobin following flash photolysis of the CO adduct were determined with 150 ps time resolution and < 2 A spatial resolution. The structures reveal numerous sites in which CO becomes transiently trapped, as well as correlated motion of the protein side chains. When a single point mutation was introduced in a position near the binding site (L29F), the departure of CO from the primary docking site was significantly accelerated. The dramatic differences in the correlated protein displacements provide a structural explanation for these kinetic differences.

1. Introduction Ultrafast phenomena in systems ranging from small molecules in the gas phase to proteins in the condensed phase have been characterized spectroscopically with numerous pump-probe methods. The transitions probed, whether rotational, vibrational, or electronic, can reveal much about the dynamics of the phenomenon being studied, but generally provide rather limited information about structural evolution. For crystalline samples. X-ray diffraction provides an extremely powerful structural tool, with A spatial resolution attainable with proteins containing thousands of atoms. We have recently extended the technique of nanosecond X-ray crystallography [1] to the picosecond time domain, and used this method to probe structural changes in myoglobin and its L29F mutant. Myoglobin is a heme protein that reversibly binds small ligands such as O2, CO, and NO. The photosensitivity of the ligand bond [2] and the reversibility of ligand binding allow structural changes associated with ligand migration to be determined with picosecond time-resolved crystallography. By comparing the structural evolution of Mb with one of its mutants, we aim to assess the functional role of highly conserved side chains in the vicinity of the active binding site. When Leu (L) in the 29 position is replaced by Phe (F), i.e., L29F MbCO, the oxygen binding affinity is enhanced by a factor of 10 [3] and the ligand escape dynamics are dramatically altered [4]. Here, we report 150-ps time-resolved structures of both wild-type MbCO and L29F MbCO. Their structural differences on the picosecond time scale are far greater than those observed on the ns time scale.

581

2.

Experimental Methods Laser _ beam

The experiments reported here were conducted on the time-resolved TT^nQTi ID09B beam hne at the European Synchrotron and Radiation FaciHty (ESRF) in Grenoble, France. Diffraction images were acquired using the pump-probe method: a ^

>=>

^

r- r^

•

'

•

'

J r

\

X-ray beam

picosecond laser pulse (2-100 ps; ~ 580 nm) triggered ligand dissociation in a ~ 250 micron P6 MbCO crystal and a variably delayed X-ray pulse (150 ps; 0.79 „

t

X

'opS^ Beamstop ^ ? capmarywith

\

f

7

'"'' colimator

I| ^

^-^" f^J^^ i ,, A-ray aetector Goniometer spindle

A with a 3.5% bandwidth) probed its * structure (see Figure 1). Synchrotron Fig. 1. Schematic diagram of the pumpRadiation consists of X-ray bursts of ps probe geometry used to acquire timeduration, but at repetition rate far too high resolved X-ray diffraction images, for pump-probe experiments. At the ESRF, this rate is 355 kHz or 1.42 MHz, depending on whether one or four electron bunches are stored in the synchrotron ring. Therefore we employed a highspeed mechanical chopper to reduce the rate to 1 KHz and a millisecond shutter to select single X-ray pulses on demand [5]. Synchronization between the laser and X-ray pulses was accomplished with ps precision using a Synchro-Lock-equipped Tirsapphire laser (Coherent Mira). The low energy oscillator pulses were boosted to ~ 1 mJ in a regenerative amplifier (Spectra Physics Hurricane) and used to pump an OFA (Quantronix Topas), whose tunable output was stretched to the ps regime in a 3-m fiber prior to delivery to the protein crystal. X-ray pulses of -10^^ photons/shot are generated when relativistic (6 GeV) electron bunches pass through the magnetic field of a 2-m long undulator. The mildly divergent X-ray radiation is focused by a toroidal mirror down to - 100 jim at a distance approximately 60 m from the source. The crystalline sample is positioned so that the X-rays pass through it upper edge, where the level of photolysis is greatest. The diffracted X-ray photons are detected with high quantum efficiency on a MAR CCD X-ray detector. Reconstruction of the protein structure with atomic resolution requires diffraction images from different crystal orientations spanning 60 degrees (owing to the six-fold symmetry of the crystal). With undulator radiation, the X-ray bandwidth is sufficient to obtain approximately 5-fold redundant data with images collected every 2 degrees. To obtain high dynamic range diffraction images with the available X-ray flux, approximately 8-32 X-ray shots were integrated on the MAR CCD before image readout. Because the protein crystal requires sufficient time to recover between photolysis pulses, which are intense enough to excite a significant fi-action of the protein molecules, the maximum repetition rate used was 3.3 Hz. Diffraction images were accumulated with and without photolysis to generate accurate differences between the two diffraction data sets.

582

3.

Results and Discussion

The electron density of the protein is constructed by Fourier transforming appropriately scaled diffraction spot intensities (structure factors). The electron density for the unphotolyzed crystal is shown in Figure 2a. Because the photolysis is incomplete, the electron density of the photolyzed crystal, shown in Figure 2b, is a mixture of photolyzed and unphotolyzed states. To generate the electron density map for the photolyzed state, shown in Figure 2c, the fraction of photolysis was estimated, and the partially photolyzed electron density map was extrapolated to complete photolysis.

. not photalyzed r

^^MW^^^^B

, partially p h o t o t y z e d , ^ B i l ^ ^ ^ B B

. extrapolated to 1QQ% phP

Fig. 2. (a) Electron density map of MbCO before photolysis, (b) Electron density of MbCO after about 25% photolysis (at 100 ps). Note the partial occupancy of CO in both binding and primary docking sites, (c) Electron density of the photolyzed state obtained after extrapolating the datafromthe middle image to complete photolysis. High-resolution time-resolved electron density maps of wild-type and L29F MbCO are shown in Figures 3 and 4, respectively. These images were rendered using a novel color-coded method for visualizing structural changes. The images reveal, with atomic resolution, the order of events that accompany ligand translocation. Numerous features are observed in both wild-type and L29F MbCO, including the displacement of the heme iron toward the proximal histidine, tilting of the heme, docking of CO in a site near the heme iron, and the correlated motion of several protein side chains. However, dramatic differences are observed on the distal side of the heme, in particular the motion of the residues in the 29 and 64 positions. In wild-type MbCO, Leu29 moves upward and His64 shifts toward the site once occupied by CO, raising the barrier to geminate recombination. In L29F MbCO, Phe29 is pushed toward H64, which shifts down and away from the heme iron, and displaces a water molecule on the surface of the protein. Only when the CO departs from the primary docking site does the His64 side chain assume a position similar to that found with wild-type Mb. On the nanosecond time scale, CO slips around to the other side of the heme and is found in the so-called Xel docking site. The CO escapes from this site into the surrounding solvent on the microsecond time scale. On the millisecond time scale, CO rebinds with the heme and completes the photocycle. The cycle can be repeated thousands of times without appreciable damage to the protein crystal.

583

Fig. 3. Electron density maps of wild-type MbCO determined before and at various times after photolysis at 10 °C. The electron density of the unphotolyzed protein is colored black while that of the photolyzed protein is colored white. Where both densities overlap, the two shades blend to gray. The direction of molecular motion follows the black to white gradient. The white stick models correspond to the static crystal structures, and are included to guide the eye. The solid circles denote occupied CO sites and the dotted circles denote evacuated CO sites. The photolyzed CO is initially trapped in the primary docking site about 2 A from the binding site, but subsequently migrates to the Xel site on the opposite side of the heme.

Fig. 4. Electron density maps of L29F MbCO determined before and at various times after photolysis at 10 °C. Soon after photolysis, CO migrates to sites labeled Phe29 and Xe4. The CO continues its migration with a significant fraction of the CO accumulating in Xel and possibly Xe2. The most dramatic differences in the side chain motion of wild-type and L29F Mb are manifested in the sub-ns maps, demonstrating the need to probe protein dynamics on a sub-ns time scale.

584

4.

Conclusions

The mechanism for excreting toxic CO in the wild-type and mutant MbCO is different. In wild-type protein, His64 rapidly relaxes toward the heme iron, which protects it from CO rebinding. In L29F MbCO, the strain induced on Phe29 and His64 by docked CO is rapidly relieved when Phe29 "sweeps" the CO away from the primary docking site. Though the mechanism of excretion is different, both are effective. The structural changes that accompany ligand translocation, as illustrated here, help explain how the protein is able to excrete toxic CO with high efficiency, even though the CO is temporarily located so close to the active binding site. Clearly, time-resolved crystallographic studies can unveil at a high level of structural detail the conformational changes that accompany protein function in this and other protein systems. The fact that global conformational changes are apparent at 100 ps, the earliest time point recorded, demonstrates that much higher time resolution will be required to follow the protein quake that ensues after ligand photolysis. This phase of the conformational motion will be accessible to 4th generation X-ray sources, which promise to deliver intense, sub 100-fs X-ray pulses [6,7]. Acknowledgements. We thank Prof. J.S. Olson and Dr. J. Soman for supplying the P6 MbCO crystals used in this study. We thank G. Hummer for helpful comments and D. Bourgeois for sharing his expertise in the analysis of Laue diffraction data.

References 1 2 3 4 5 6 7

V. Srajer et al., Science 274, 1726-9., 1996. Q. H. Gibson, S. Ainsworth, Nature 180, 1416-7, 1957. T. E. Carver et al., J. Biol. Chem. 267, 14443-14450, 1992. F. Schotte et al. Science 300, 1944-7, 2003. F. Schotte et al., in Third-Generation Hard X-ray Synchrotron Sources D. M. Mills, Edited by D. M. Mills, Wiley and Sons, New York, 2002 A. Cho, Science 296, 1008-10., 2002. M. Abd-Elmeguid et al., in TESLA: The Superconducting Electron-Positron Linear Collider with Integrated X-ray Laser Laboratory - Technical Design Report G. Materlik, T. Tschentscher, Edited by G. Materlik, T. Tschentscher, DESY, Hamburg, 2001

585

Energy transfer pathways in Photosystem I studied by one and two color photon echo spectroscopy H. M. Vaswani\ J. Stenger\ M. Y a n g \ P. Fromme^ G. R. Fleming^ ^ Department of Chemistry, University of California, Berkeley and Physical Biosciences Division, Lawrence Berkeley National Laboratory, Berkeley, California, 94720, USA E-mail: [email protected] ^ Department of Chemistry and Biochemistry, Arizona State University, Tempe, Arizona 85287, USA Abstract. One and two color three pulse photon echo spectroscopy is used to probe ultrafast energy transfer, energetic disorder and correlation between the 96 non-equivalent chlorophylls in the photosynthetic light-harvesting complex Photosystem I.

In an energy transfer system with as much spatial, energetic and coupling disorder as the 96-Chlorophyll Photosystem I complex (PSI), traditional ultrafast methods that only probe population dynamics do not provide the complete picture. The population dynamics of PSI has been studied extensively in the past [1-3] yet little is known about the actual pathway the excitation follows through the complex landscape of PSI. The three pulse photon echo peak shift (3PEPS) method is an incisive tool in the study of energetically disordered systems, exploiting the disorder to determine energy transfer timescales, electron-phonon coupling and the degree to which the disorder is correlated within individual complexes. We use one-color and the newly developed two-color 3PEPS method to study the pathways of excitation in PSI. One-color 3PEPS provides the disorder of the diagonal elements of the electronic Hamiltonian of the system of interest, however, two-dimensional electronic spectroscopy is required to obtain information about the off-diagonal coupling terms. Our group has recently developed a two-dimensional extension of traditional 3PEPS, two-color 3PEPS (2C3PEPS), to determine electronic mixing between pairs of excitonically coupled molecules, without prior knowledge of the uncoupled site energies. This new technique provides previously immeasurable information about the degree of correlation and excitonic coupling between the initial and final states [4,5].

... I k

(a)

700 nm Q

^

;^—--~-_______690nm

__ 690 nm 675 nm

1

713 nm 0.5 Population Time [ps]

586

(b)

^'^v*^^^^^^^ 1.0

°

675 nm

•

o

1

2

3

Population Time [ps]

Fig. 1, Wavelengthdependent one color 3PEPS data. The spectral widths of the laser pulses are a) -25 nm. b) -10 nm.

Experimentally, two-color 3PEPS is very similar to one-color 3PEPS. Both methods use homodyne detection of the echo signal and, unlike the more general heterodyne-detected two-dimensional electronic spectroscopy, 3PEPS does not require interferometrically precise control of the optical delays. In addition, 2C3PEPS does not require the pulses to span the full spectrum of interest, making 2C3PEPS experimentally more feasible. 2C3PEPS has been successfully tested on a phythalocyanine dimer [6]. Here we present the first application of 2C3PEPS to an energy transfer system, one that is crucial to oxygenic photosynthesis performed by cyanobactiera, algae and plants, Photosystem I. In one-color 3PEPS we probe the decay of phase memory as a function of population time, T, within the Qy absorption band of PSI for different center wavelength and spectral widths of the employed laser pulses. The decay of the peak shift is strongly dependent on the spectral position. Fig. la illustrates the broadband (-25 nm FWHM) 3PEPS measurements. Peak shift traces recorded near the absorption peak (675, 690, and 700 nm) do not decay completely within the detection time window. At longer wavelengths the peak shift decays slower; at 700 nm, the peak shift is essentially constant in the time frame of our experiment. This behavior is also seen when narrow bandwidth pulses (-10 nm FWHM) are used (Fig. lb). In fact, the peak shift dynamics are very similar to the broadband measurements after the initial period when pulse duration strongly influences the signal. For population times up to 4 ps a large part of the initial peak shift has still not decayed. This shows substantial inhomogeneity in these spectral ranges is preserved on this time scale and that a complete sampling of all chlorophylls (Chls) with excitation energies inside the laser window takes place on a much longer time scale. At 713 nm the trend toward slower decays with increasing wavelength is broken and a fast 100 fs decay in the peak shift is followed by a component on the order of 800 fs. After the first picosecond the peak shift is essentially zero implying rapid sampling of the red absorbing Chls. This suggests the red Chls are in close proximity to one other. Two-color 3PEPS measurements on PSI are illustrated in Fig. 2. With the first two pulses the system is excited at 675 nm with a bandwidth of 10 nm. After a period called the population time the phase memory is read out with a pulse of different color centered at 705 nm. We call this the 'downhill' peak shift as the excitation pulses are at higher energies than the probe pulse. In order to extract the phase memory, we measure Type I (rephasing) and Type II (non-rephasing) peak shifts (Fig. 2a). The difference between Type I and Type II peak shifts is the most informative quantity of the two-color experiment: the difference peak shift (DPS). The DPS provides quantitative insight into the correlation of the two different spectral regions. In previous 2C3PEPS experiments a rise in the DPS is observed [5,6]. However, in our experiment the DPS starts at ~5 fs and decays monotonically to zero on a time scale of 2 ps. The non-vanishing value of the DPS at zero population time hints at fast energy transfer and/or excitonic coupling between pigments at 675 nm to those at 705 nm. Through the strong coupling the phase memory is transferred within the pulse width and thus an initial DPS is generated. In previous work we have developed a model for PSI based on site energies calculated quantum mechanically (QM) from the crystal structure, spectral densities obtained from hole-burning data, and modified Redfield theory (MRT) [7]. We used the model to compute the 587

inhomogeneous contribution to the ensemble averaged frequency correlation function, I(t). I(t) fits the experimental DPS well (Fig. 2b). If we use Forster theory (FT) instead of MRT, I(t) decays more slowly. In an artificially flat energy landscape, where all but the chlorophyll trap have the same site energy and lineshape, I(t) decays more rapidly. We compute I(t) for various permutations of the QM site energies and for random orientations of the pigments. The results show that 2C3PEPS is sensitive to both the exact distribution of energies (not just the gross distribution) and the coupling generated by the crystal structure orientations (data not shown). 101 ,

(a)

0.8-

0

W 5 SO.6Q ffj» S t « • •

° 0

• • • •

DPS

(b)

[

h:. QM energies

3O.4-

°*»'^*;°Type°ll

0.5 1.0 1.5 Population Time [ps]

'

1.0

0.22.0

r\n. 0.0

Flat energy landscape

0.5

1.0

1.5

Population Time [ps]

'

2.0

Fig. 2. 2C3PEPS data of PSI. a) Type I and Type II peak shifts, b) Normalized DPS and I(t) for the QM energies using MRT (solid line) and FT (dotted line). The dashed line shows I(t) (using MRT) in a flat energy landscape.

In summary, the combination of one and two color photon echo spectroscopy together with master equation calculations can decipher energy and coupling information in extended many pigment systems with large heterogeneity in site energies, couplings and energy transfer rates. Our one and two color photon echo study gives information on sampling time scales for different spectral regions of the Qy absorption band and correlations between different spectral regions. In principle, further analysis will provide the full electronic Hamiltonian of PSI. Acknowledgements. This work used resources provided by the National Energy Research Scientific Computing Center and was supported by the Director, Office of Science, Office of Basic Energy Sciences, Chemical Sciences Division of the U.S. Department of Energy under Contract No. DE-AC03-76SF00098. J. S. thanks the German Academic Exchange Service (DAAD) for generous support.

References 1 2 3 4 5 6 7

588

J. T. M. Kennis, ... G. R. Fleming, J. Phys. Chem. B 105, 4485, 2001. B. Gobets,... R. van Grondelle, Biophys. J. 85, 3883, 2003. M. G. MuUer, J. Niklas, W. Lubitz, A. R. Holzwarth, Biophys. J. 85, 3899, 2003. M. Yang and G. R. Fleming, J. Chem. Phys., 110, 2983, 1999. R. Agarwal, ... G.R. Fleming, J. Chem. Phys. 116, 6243, 2002. B. Prall, ... G.R. Fleming, J. Chem. Phys. 120, 2537, 2004. M. Yang, ... G.R. Fleming, Biophys. J. 85, 140, 2003.

Dynamics of Carotenoids Probed by Femtosecond Absorption, Fluorescence, and Raman Spectroscopy M. Yoshizawa', D. Kosumi', M. Komukai', K. Yanagi^ and H. Hashimoto^ ' Department of Physics, Graduate School of Science, Tohoku University, Aramali-aza-aoba, Aoba-ku, Sendai 980-8578, Japan Email: [email protected] ^ "Light and Control", PRESTO/JST, Department of Physics, Graduate School of Science, Osaka City University. 3-3-138 Sugimoto, Sumiyoshi-ku, Osaka 558-8585, Japan Abstract. Ultrafast optical responses in /:^ -carotene and lycopene depend on the pump wavelength. Excess vibrational energy induced by the photoexcitation remains longer than several picoseconds in the excited states and slows down the relaxation kinetics.

1.

Introduction

Carotenoids play important roles in light-harvesting function in bacterial photosynthesis. In antenna complexes of the photosynthesis, the light energy is absorbed by all-rra^5-carotenoid and transferred to bacteriochlorophyll with high efficiency. The relaxation kinetics of ^ -carotene after photoexcitation has been proposed as lBu'^lBu'->2Ag"-»lAg" (ground state) [1,2]. A lifetime of the 18^^ state has been estimated to be about 100 fs by fluorescence spectroscopy [3], but Cerullo et al. have assigned absorbance changes with time constants of 10 fs and 150 fs to the IBu^ and IBu" states, respectively [4]. The initial relaxation kinetics of carotenoids has not been well-understood. In this study, the electronic and vibrational relaxations in carotenoids have been investigated using tunable femtosecond pump pulses.

2.

Experimental

Experimental setup of the femtosecond absorption and stimulated Raman spectroscopy was described elsewhere [5]. An amplified femtosecond pulse (794 nm, <150 fs, -700 jiJ) was separated to three pulses and used as the first pump, Raman pump, and probe pulses. The time-resolved fluorescence induced by the first pump pulse was measured by Kerr-shutter technique. All-/r^/75^-P-carotene was purchased fi-om Wako Pure Chemical Industries Ltd. and recrystallized. AW-tramAycopene was extracted fi-om tomato puree and purified. The sample solutions in benzene, cyclohexane, and ^-hexane were circulated using a 1 mm flow cell.

589

3.

Results and Discussion

Figure 1 shows photoinduced absorbance changes of all-/ra^5-P-carotene in benzene. The transient absorption around 1.2 eV appears just after the photoexcitation and disappears within 1 ps. Since temporal response of the 1.2 eV signal is equal to that of the fluorescence as shown in Fig.2, it is assigned to the optically allowed 18^^ state. Signal at 2.18 eV is assigned to the 2Ag' state. The formation of the 2Ag' state is equal to the decay of the IBu^ state. The IBu" state cannot be time-resolved. The lifetime of the IBu' state is estimated to be shorter than 100 fs. The photoinduced absorbance changes show dependence on the pump photon energy. The 2Ag" state induced by the 3.1 eV pump has broader absorption band at 1.0 ps. The small difference around 2.3 eV remains longer than 10 ps. The decay of the IBu' state is faster after the 2.5 eV pump as shown in Fig.2. The lifetime of the 2Ag' state also shows the difference. It is 8.4 ps after the 2.5 eV pump and 9.1 ps after the 3.1 eV pump. The differences are due to the excess vibrational energy induced by the 3.1 eV pump.

1.2

1.4 Photon Energy (eV)

Fig.l. Absorbance changes of p-carotene in benzene after 2.5 eV pump (solid curves) and 3.1 eV pump (dashed curves). A dash-dotted curve shows stationary absorption. '

^

^(b)

: 1 0 1 Delay Time (ps)

'

b T = 0.25+0.05 p s ;

^ ^^JX-QL^Cl

^OrvSvS^ 1

1

1 0 1 Delay Time(ps) t

^T

Fig.2. Temporal responses of fluorescence at 2.25 eV (open circles) and absorption at 1.28 eV (closed circles) of p-carotene in benzene, (a) 2.5 eV pump and (b) 3.1 eV pump.

590

Figure 3 shows the time-resolved Raman signal of P-carotene in cyclohexane. The photoinduced changes at 1520 cm"' are negative because of the depletion of the ground state. The signals due to the excited states appear around 1800 cm"'. The positive and negative signals are interpreted in terms of the /=1 vibrational excited level (vi mode) of the 2Ag" state [1,2]. The Raman signals around 1800 cm"' induced by the 2.5 eV pump are smaller than those induced by the 3.1 eV pump. This shows that the /=! and /=0 levels of the 2Ag" state have nearly equal populations (N\-No) after the 2.5 eV pump. The similar dynamics has been observed in all-^ra/75-lycopene. The lifetimes of the IBii" and 2Ag" states in benzene solution are, respectively, 0.14 ps and 4.4 ps after the 2.5 eV pump and 0.27 ps and 4.9 ps after the 3.1 eV pump. _ ^, '

1

'

1

'

_

(b)

/ \ ,^, \ \^^l^

1600 1800 Raman Shift (cm' )

i -1

2000 B 1400

1

1

i

l

1600 1800 Raman Shift (cm" )

l

"

2000

Fig.3. Photoinduced changes of Raman signals of p-carotene in cyclohexane at 3.Ops. Dashed curves show signals of the ground state, (a) 2.5 eV pump and (b) 3.1 eV pump.

4.

Conclusions

The relaxation kinetics of carotenoids depends on the excess vibrational energy induced by the photoexcitation. The relaxation fi*om the IBu' state to the 2Ag' state occurs with the time constant of 0.1-0.3 ps passing through the IBu" state. The redistribution and dissipation of the vibrational energy are expected to take place simultaneously with the internal conversions. However, the special vibrational modes such as the Vi mode have long lifetimes. The excess vibrational energy remains longer than several picoseconds in the 2Ag" state. The vibrational energy affect the relaxation kinetics and slows down the internal conversions.

References 1 M. Yoshizawa, H. Aoki, and H. Hashimoto, Phys. Rev. B 63. 18030K 2001. 2 M. Yoshizawa, H. Aoki, M. lie, and H. Hashimoto, Phys. Rev. B 67, 174302, 2003. 3 S. Akimoto. I. Yamazaki, S. Takaichi. and M. Mimuro. Chem. Phys. Lett. 313, 63. 1999. 4 G. Cerullo, D. Polli. G. Lanzani, S. De Silvestri. H. Hashimoto, and R. J. Cogdell. Science 298, 2395, 2002. 5 M. Yoshizawa and M. Kurosawa. Phys. Rev. A 61, 013808, 2000.

591

Multi-Pulse Transient Absorption and Carotenoid Excited-State Dynamics: /?-Carotene Emmanouil Papagiannakis, Delmar S. Larsen, Mikas Vengris, Ivo H.M. van Stokkum, Rienk van Grondelle Physics Department, Vrije Universiteit, De Boelelaan 1081,1081HV, Amsterdam, the Netherlands E-mail: papagian @ nat. vu. nl Abstract. Pump-dump-probe and pump-repump-probe transient absorption experiments with broad-band detection have allowed the identification of a hitherto unknown relaxation pathway in P-carotene which opens after excitation to the S2 state with high-energy photons.

Introduction Carotenoids are vital for photosynthetic organisms due to their light-harvesting and photoprotective functions, which are governed by their conjugated 7r-electron backbone. Solar light promotes carotenoids to the S2 state; one-photon transitions betv^een So and Si are symmetry-forbidden. Si is populated by intemal conversion from S2 within -150 fs and decays within picoseconds. Several more "dark" excited states have been included in describing the excited state manifold of carotenoids [1]. The correct description of their excited-state properties now demands the development of experimental tools capable of dissecting coexisting and interacting excited species. We have extended pump-probe (PP) by adding to it a third pulse (Fig. 1), timed and tuned to selectively interact with either a stimulated emission (SE) or an excited state absorption (ESA). This pulse transfers excited-state populations either to a higher state (pump-repump-probe, PrPP) or to the ground state (pump-Qump-probe, PDP) [2-4]. Probing across a broad spectral region (400-700 nm) we observe the effect of the added pulse on bands attributed to different states; we can accordingly characterize the effect and moreover separate the dynamics and trace the connectivity of the involved states. Here we present a study of the "model" ail-trans P-carotene (purchased from Fluka and dissolved in n-hexane), illustrating how the aforementioned spectroscopies can be effectively used in studying the excited state dynamics of carotenoids.

A^

^-carotene

dump/repump

Action pump

AA^ probe

Fig. l.Three-pulse measuring schemes and the molecular structure of p-carotene

592

P-carotene excited state dynamics Comparing the PP spectra of P-carotene measured 3 ps after excitation at 400 and 500 nm shows that high-energy excitation results into a spectrum with a pronounced shoulder on the blue side (500-525 nm) of the strong Si ESA which peaks at -550 nm (Fig. 2A). The decay of this feature has a distinct long-living component (-60 ps), which is absent in the decay of the Si ESA which has a 10-ps lifetime. To examine this additional feature, we performed a series of "surgical" multi-pulse experiments. We use the double-difference in absorption (AAOD) to monitor the induced changes: AAOD(/l,r,0 = PDP(A,r,0-PP(/l,0

(1)

(X: probe wavelength, t: probe delay, r: dump time). The AAOD visualizes the induced effects as it is non-zero only when the additional pulse has an effect on the system. In the first, PrPP experiment, an 800-nm pulse repumped the Si state to a higher excited state, 1 ps after 500 and 400-nm excitation. Figures 2B,C show that the effect of the 800-nm pulse is markedly different; after 500-nm excitation the AAOD spectrum is similar to the respective PP spectrum, whereas after 400-nm excitation the AAOD spectrum lacks the shoulder observed in the PP spectrum. The observation that the different bands have different response to the 800-nm pulse shows that the ESA shoulder originates from a species other than the Si state, which we refer to as S*. 30 S" 20 O ^ 1 0 Q

-10

A _ PP

-S,/^'

400-nm excitation 500-nm excitation -

500 550 600 650 wavelength (nm)

500 550 wavelength (nm)

500 550 wavelength (nm)

Fig. 2. p-carotene transient absorption measurements. A. pump-probe; B. 500-nm pump, 800-nm repumpprobe; C. 400-nm pump, 800-nm repump-probe. Repump delay: 1 ps, probe delay: 3 ps.

The second experiment involves a 530-nm pulse added to the PP scheme. This pulse is resonant with the SE of S2 and the ESA of Si and therefore it will dump or repump depending on the timing. We measured an "action trace" which probes the changes on the 3-ps PP spectrum as the delay of the 530-nm pulse varies (Fig. 1). Figure 3A contains the action trace at 450 and 550 nm (SQ bleach and Si ESA peak), where two timescales can be distinguished. At 450 nm and early times the AAOD signal corresponds to a bleach loss, resulting from the dumping of S2 to SQ. At 550 nm, the effect is initially similar; the Si population decreases with the S2 dump. The additional, longer, timescale present at 550 nm, describes the removal of Si population by ESA repump. By global analysis, we estimated the corresponding timescales and spectra

593

(Fig. 3B). A fast, 300 fs, component describes the full loss of the ESA band (shown for comparison), i.e. both the Si and the S" bands, whereas the slow, 10-ps, component, describes the preferential loss of the Si ESA.

• pump-probe AAOD by dump of S^ AAOD by repump of S,

0.0

0.5 1.0 1.5 530-nm pulse Delay (ps)

550 600 500 wavelength (nm)

Fig. 3. Action trace measured on p-carotene after 400-nm excitation. A. Action kinetics measured at the bleach region (top) and at the Si ESA peak (bottom); B. The Action spectra corresponding to the two timescales present in the Action kinetics

Concluding Remarks We used multi-pulse transient absorption to investigate the coexistence of different excited states in P-carotene after excitation to the S2 state with high-energy photons. The loss of the 500-525 nm ESA shoulder, which is only generated by 400-nm excitation, after dumping S2 shows that it corresponds to an excited state (which we term S*) formed via S2, whereas the two repump experiments (at 800 nm and 530 nm) preferentially deplete Si without affecting S*, showing that they are separate excited states. We have thus illustrated that high-energy photons open a secondary relaxation pathway which involves a hitherto unknown singlet excited state, S", which is disconnected from Si. The properties of S* require further investigation.

References 1 2

3

4

594

T. Polivka, V. Sundstrom, "Ultrafast dynamics of carotenoid excited states. From solution to natural and artificial systems", Chemical Reviews. 104(4), 2021-71 (2003) F. Gai, J.C. McDonald, P.A. Anfmrud, "Pump-dump-probe spectroscopy of bacteriorhodosin: Evidence for a near-IR excited state absorbance", Journal of the American Chemical Society. 119, 6201-6202 (1997) S.L. Logunov, V.V. Volkov, M. Braun, M.A. El-Sayed, "The relaxation dynamics of the excited electronic states of retinal in bacteriorhodopsin by two-pump-probe femtosecond studies", Proceedings of the National Academy of Sciences of the United States of America. 98, 8475-8479 (2001) D.S. Larsen, E. Papagiannakis, I.H.M. van Stokkum, M. Vengris, J.T.M. Kennis, R. van Grondelle, "Excited state dynamics of P-carotene explored with dispersed multipulse transient absorption". Chemical Physics Letters. 381, 733-742 (2003)

Observation and control of all-^mws-Pcarotene wavepacket motion using pumpdegenerate four-wave mixing Thomas Homung^'^\ Hrvoje Skenderovic^-, Karl-Ludwig Kompa^^ and Marcus Motzkus^'"^^ 1) Max-Planck-Institut fur Quantenoptik, Hans-Kopfermarm-Str. 1, D-85748 Garching, Germany 2) Massachusetts Institute of Technology, 77 Massachusetts Ave., Cambridge MA 02139-4307, USA 3) Philipps-Universitat, Hans-Meerwein-Str., D-35032 Marburg, Germany E-mail:[email protected], [email protected] Abstract. Wavepacket dynamics on the ground and optically dark, first electronic state of all-trans-p-carotene are studied with 16 fs time resolution using pump-degenerate fourwave mixing spectroscopy. Moreover control over the vibrational ground state modes is shown.

Carotenoids, being part of light-harvesting complexes (LH) play an important role in photosynthesis. They absorb light in blue-green and efficiently transfer energy to chlorophylls. The electronic states of interest to this study are the electronic ground 1 Ag" (So), the first optically allowed l^B/ (S2) and the optically dark state 2^Ag' (Si) [1]. The long time kinetics of P-carotene have been studied extensively [1,2] and only very recent work has been successful in monitoring the coherent dynamics of the very high frequency modes of So using transient absorption [3]. To observe the wavepacket dynamics in Si poses an extra challenge besides the short pulse duration required, since this state is not directly optically accessible from the ground state, but only via the population transfer from S2 to Si. This pathway seems to be however incoherent [3], thus making simple pump probe measurements impossible. Recently it has been realized that by a pulse sequence that incorporates stimulated emission pumping (SEP) coherent motion in dark states can be observed [4]. FSRS is able to collect vibrational spectra of Si with high temporal and spectral resolution, but it is not sensitive to the relative phase between modes, thus a wavepacket motion cannot be reconstructed from this data. In this work we apply degenerate four-wave mixing spectroscopy (DFWM) combined with narrowband detection (1 nm spectral resolution) to probe and control the wavepacket dynamics in SQ. In combination with a pump pulse (pumpDFWM) we are able to monitor coherent dynamics of the dark Si state. The vibrational spectrum of So and Si has been observed in several picosecond resonance Raman studies and has been recently studied with <100 fs time resolution with femtosecond-stimulated Raman spectroscopy (FSRS) [5], Compressed output pulses of 10 |uJ from a NOP A are frequency tunable from 480-700nm and have a typical pulse duration of 16 fs. The output of one NOPA is

595

split into the 3 DFWM beams (ks=ki-k2+k3). The pulses within two of the beams are passed through an all-reflective pulse shaper to control their temporal profile. The second NOPA delivers the pump pulse for the pump-DFWM experiments used for probing Si dynamics. A 300 jLim sample holds all-trans-P-carotene dissolved in cyclohexane. The data obtained for resonant DFWM on the S0-S2 transition, first without an additional pump pulse, show pronounced wavepacket oscillations for red shifted small bandwidth detection at 546 nm (Fig. 1 (a)). Fourier transformation reveals that the wavepacket is due to coherent SQ mode dynamics (Fig. 1 (b)). The low frequency peak at -370 cm"^ originates from beatings between the strongest P-carotene modes. No dynamics from S2 can be observed, possibly due to its fast decay. The So C=C stretch mode requires a more red or blue shifted detection compared to the lower modes. Non-resonant DFWM (center wavelength 555 nm) at higher pulse energies excites and probes a ground state wavepacket via Raman excitation. Again wavepacket motion is only observed for red and blue shifted small bandwidth detection (Fig. 1 (a)). Shapiug the time coincident DFWM pump and Stokes into a pulse train allowed the selective excitation of the C-C stretch ground state mode (Fig. 1 (a)). Ofi

(a)

resonant {\^^ = 546 nm)

non-resonant {X

(b)

= 617 nm)

LL LALiJAJ non-resonant control

0.6

0.8

delay t„(ps)

1.0

0.0 0.5 1.0 1.5 2.C

v(10^cm"')

Fig. 1. Resonant and non-resonant DFWM data of the ground state So- Also shown is the control of the SQ wavepacket using pulse train excitation, (b) FFT of transients in (a). We proceed with the first time observation of coherent motion in the Si / hot Si state of all-^ran5-P-carotene. The Si state can not be directly excited from SQ, and thus can only be populated via transfer through the conical intersection between S2 and Si- Since strong evidence exists that this population transfer occurs incoherently, it is necessary to create coherent motion in Si by a second optical excitation process in order to be able to probe the coherent time dynamics of Si modes. Therefore in our experiments we use a pump pulse at 490 nm previous to the DFWM sequence at 560 nm (the center wavelength of the Si-Sn absorption band) to transfer population from SQ to S2 (see schematic in Fig. 2). This technique is known in literature as pump-DFWM [6]. The DFWM sequence at 560 nm is used to probe and excite the incoherent population originating from the population pumped into S2 and transferred via conical intersection to Si. In order to avoid non-resonant excitation of the ground state the pump and DFWM pulse intensities were decreased. A phase-locked chopper was used to periodically block the S0-S2

596

resonant pump. In all transients the DFWM pump and Stokes were time coincident (ti2=0) and the third DFWM beam was delayed fe). No signal is observed when the pump is in its off state, while for "pump on" Si wavepacket dynamics could be observed when detecting at 610 nm (Fig. 2 (a)). The oscillatory motion is located on top of an exponential decaying signal, with a decay constant of T~ 112-160 fs depending on the pump delay. The wavepacket motion is strongest if the pump precedes the DFWM sequence by 4 ps. Evidence for the Si nature of the observed modes is provided by a multimode fit model that fits all transients and spectra with good accuracy.

delay t^g (fs)

v(10^cm"'')

Fig. 2. Wavepacket dynamics of the optically dark Si state of all-trans-(3-carotene probed with pump-DFWM. (a) DFWM signal with "pump on" detected at 610 nm, for different delays between pump and DFWM. Without pump no signal is observed. The oscillations are on top of a signal, which decays exponentially with rate T. Scaling factor x of the data, with respect to the transient recorded at a pump delay of 4 ps. (b) FFT of transients in (a).

In summary we have shown that pump-DFWM in combination with spectral detection of 1 nm resolution can be successfully applied to the study of complex molecular systems, where optically dark electronic states are common.

References 1. T. Polivka and V. Sundstrom, Chem. Rev. 104, 2021 (2004) and ref therein 2. T. Siebert, V. Engel, A. Matemy, W. Kiefer, and M. Schmitt, J. Phys. Chem. A 107,8355(2003) 3. G. CeruUo, G. Lanzani, M. Zavelani-Rossi, and S. D. Silvestri, Phys. Rev. B 63, 2411041(2001) 4. G. Cerullo, L. Luer, C. Manzoni, S. de Silvestri, O. Shoshana, and S. Ruhman, J. Phys.Chem. A 107, 8339 (2003) 5. D. McCamant, P. Kukura, and R. A. Mathies, J. Phys. Chem. A 107, 8208 (2003) 6. M. Motzkus, S. Pedersen, and A. H. Zewail, J. Phys. Chem. 100, 5620 (1996)

597

Vibrational and Electronic Coherence Observed in Two-Dimensional Integrated Three-Pulse Photon Echo Yutaka Nagasawa, Mayu Ogasawara, Yukako Nakagawa, Yoshio Mori, Tadashi Okada, and Hiroshi Miyasaka Division of Frontier Materials Science, Graduate School of Engineering Science and Research Center for Materials Science at Extreme Conditions, Osaka University, Toyonaka, Osaka 560-8531, Japan E-mail: [email protected] Abstract. Two-dimensional integrated three-pulse photon echo measurement was carried out on saccharide glasses and in poly vinylalcohol (PVA). The echo signals were modulated by coherent molecular vibrations which depended on the delay between the first and the second pulse. For saccharide glasses, a low-frequency critically damped oscillation at -40cm'^, which was assigned to the glass phonon mode, was also observed.

1. Introduction Three-pulse photon echo (3PPE) is essentially a three-dimensional coherent spectroscopy with three time periods to control. We have been measuring twodimensional 3PPE signals of low temperature glassy solids with controlling the first (tj2) and the second time periods (t23 or tjs) by integrating the echo intensity over the third one.[l] When femtosecond laser pulses are utilized, low-frequency molecular vibrations are also coherently induced and oscillations can be observed in the echo signal. To investigate the effect of vibrational coherence on dephasing and rephasing of the electronic coherence, we have carried out 2D-3PPE measurements of PVA, and saccharide glasses. Saccharides, such as trehalose and glucose, are known to function as a protectant against dehydration and freezing in many microorganisms, insects, reptiles, amphibians, and plants.[2,3] The saccharides undergo glass transition upon freezing or dehydration and protect the organic tissue from damaging.

2. Experimental Methods The second harmonic of a femtosecond cavity-dumped chromium: forsterite laser centered at 635 nm with pulse duration of '-26fs (fwhm) was utilized as the light source. The second harmonic beam was split into three beams with similar intensity ('-300pJ) for the measurement. A photodiode and a lock-in amplifier were utilized for the signal detection. Glass films of trehalose, glucose, and PVA were produced with an organic dye, oxazine 4 (0x4), doped as the probe molecule. The optical densities of the samples were set to -1.0 at the absorption maximum. The

598

sample thickness was -O.lmm for PVA and 0.2-0,3mm for the saccharides. 3PPE measurements were carried out with the samples set in a closed-cycled helium gas cryostat.

3. Results and Discussion When the temperature was >200 K, all of the echo signals decayed very rapidly within the pulse duration. At 10 K, although the signal completely decayed after 200 fs for PVA, there was a much slower decay component for trehalose and glucose (Fig.l(a)). The energy barrier for the dephasing was calculated to be 64cal/mol (23cm'^) and 120cal/mol (42cm'^) for PVA and trehalose, respectively, from the Arrhenius plot. The echo signals were modulated by the intramolecular vibrational mode of 0x4 with a frequency of -590cm'\ The echo signals of the saccharides exhibited a recurrence around 0.9 ps, indicating a critically damped oscillation induced by the glass phonon mode. The Fourier transformed spectra of the signals had a peak around 40cm'^ for both saccharides (Fig. 1(b)). It can be concluded that PVA has larger free volume than the saccharides and give rise to lower frequency phonon modes. The phonon mode of PVA was reported to peak at ~26cm'^ which agrees with the value of the energy barrier obtained from our temperature dependent experiment, i.e. 23cm"\[4] 20

1.0-f

1 <»

(a)

^

0.5 H

10 K — Trehalose •• Glucose .. PVA

1^

(b)

Trehalose Glucose PVA

•F 16H^

o ^

10-

in

\V

5H X.

0.0-

T

1

200 ti2/PS

1

1

400 600 Frequency / cm

1 800

—r 1000

Fig, 1. (a) 3PPE signals from 0x4 doped in saccharides and PVA glasses at lOK. The echo intensity was measured by scanning tj2 with (23fixedat 387fs. (b) The real part of the Fourier transformed spectra of the echo signals. The coherent intramolecular oscillations had an interesting dependence on tjj as can be seen in Fig.2. At ^i5=0fs, the strongest oscillation observed by scanning t22 had a frequency of 590cm'\ When tjs was extended to 53fs, the modes around 310cm'^ became stronger, and at ^75=107fs, the mode around 310cm'^ disappeared although the one at 590cm"^ remained and the phonon mode at 40cm'^ appeared. This observation was reproducible by a computer simulation assuming harmonic linear coupling. The modes at 590cm'^ and 40cm'^ are still observable even when t23 was extended to >100ps. Since the phonon mode is a critically damped oscillation, there should be no vibrational coherence remained in this time region. We concluded that the molecular oscillations are memorized in the media as a

599

hologram. When the signal intensity was plotted 2-dimesionally for the time region of 0-2.5ps, it was found that a gorge ran along ti2=ti3, which corresponds to 2pulse echo. The recurrence caused by the phonon mode runs along tu^l^s or ^i2=lps. These results demonstrate that the echo decay is not a simple exponential and careful two-dimensional analysis is necessary.

r 2 1^2/ps

->

•55

T 200

1—'—r~*~~" 400 600 800 Wavenumber / cm"^

1000

11 Sip

0.1

0,01 H

^

^

^

^

0.001 H

^ 1 t^^/ps

mS^mBSBBBB^^

1

2

ti2/PS

Fig. 2, (a) 3PPE signals of Ox4/glucose with t^ fixed at Ofs, 53fs, and 107fs, and (b) the real parts of the Fourier transformed spectra of the signals, (c) 3PPE signals of Ox4/trehalose measured at various values of %. (d) 2D-3PPE signal of Ox4/trehalose, The signal intensity is strong at bright parts and weak at dark parts.

References 1 2 3 4

600

Y. Nagasawa et al, J. Phys. Chem. A 107, 2431, 2003. J. H. Crowe and A. F. Cooper Jr., Scientific American IIS, 30, 1971. K. B. Storey and J. M. Storey, Scientific American 263, 62, 1990. R. Yano et al, J, Phys. Soc. Jpn, 58, 3814,1989.

Photon Echo Study of the Electron-Phonon Coupling Strength in Molecules and Molecular Aggregates Valentin I. Prokhorenko^'^, Rienk van Grondelle^ and R.J. Dwayne Miller^ ^ Department of Biophysics, Faculty of Science, Vrije Universiteit Amsterdam, De Boeielaan 1081,1081 HV Amsterdam, The Netherlands E-mail: [email protected] ^ Departments of Chemistry and Physics, University of Toronto, 80 St. George Street, Toronto, Ontario, Canada, M5S 3H6, E-mail: [email protected] Abstract. A method for direct measurement of the system-bath coupling strength in molecules and molecular aggregates, based on the three pulse photon echo technique with non-trivial time ordering of interacting laser pulses, is proposed and realized. A surprisingly low magnitude of the Huang-Rhys factor (~ 1) was measured for the natural molecular aggregates (chlorosomes from the Cf. aurantiacus bacteria) at room temperature.

1. Introduction The magnitude of the electron-phonon coupling strength in molecules can be characterized by the Huang-Rhys factor (HRF),5 - - In[ZPL/(ZPL + PW^)], where ZPL corresponds to the area of the so-called zero-phonon line, and PW to the phonon vAxvg. The HRF can be measured using methods such as site-selective fluorescence spectroscopy or hole-burning spectroscopy techniques but only at lov^ temperatures. We have found an alternative approach to measure the HRF at room temperature using a photon echo phenomenon. The method is based on the threepulse PE with non-trivial time ordering of the interacting light pulses. In the conventional 3PEPS the time evolution of the PE signal is observed in the ^2 +^3-^1 direction and investigated by changing the delay between the T^ and 2"^ pulses ("dephasing period" i'), whereas the time delay between 2""^ and 3'"^ pulses is positive ("waiting period" i) and fixed. Since the PE is invariant to the permutation of the time ordering of pulses 2 and 3, we can change the delay between these pulses to negative values (with respect to pulse #2), and perform the measurement by variation of delay i at fixed delay i'. In that case the maximal amplitude of the PE-signal, generated in the same direction, will be observed by time overlapping of pulses 1 and 3. However, by time overlapping of the 2"^ and 3*^^ pulses, a hole in the PE-trace will also be observed. Figure la illustrates this effect, calculated for a model line-shape function corresponding to a HRF of 1.2 (impulsive limit; "dephasing period" - 200 fs). Calculations were performed in the framework of the conventional PE-theory [3] for the case of homodyne detection. It can be shown that the ratio between amplitudes of PE-signal and hole depth is related to the magnitude of HRF S as PE^^ IPE^^^^ « exp(-45) - exp(-6S).

601

2.

Experimental Methods

Chlorosomes from the green bacteria Cf. aurantiacus where prepared and isolated from the base-plate as described in ref [4]. Measurements were carried out under aerobic conditions, room temperature, using a flow cell with 0.2 mm path length. Typical optical densities of the chlorosome samples, diluted in a KPh-buffer (pH=7) were - 0.3. The samples were excited by the laser pulses of 35 fs duration and an excitation energy density of -10^'^photons/cm^, centered at - 745 nm (corresponding to the absorption maximum of the Qy-band); repetition rate 100 kHz. The PE-measurements were performed using a boxcar beam geometry with homodyne detection. The measurement of the non-aggregated bacteriochlorophyll (BChl) molecules dissolved in MeOH (extracted from the same chlorosomes and purified accordingly), was performed using the same experimental conditions except for the excitation wavelength that was centered at 668 nm (maximum of the Qy-band of BChl c). To enhance the dynamic range of the measured signals and to prevent a possible contribution of the "pump-probe" signal to the measured PEkinetics, a double lock-in modulation technique was applied. A signal-to-noise ratio of ~ 1000:1 was typically achieved.

3,

Results and Discussion

Figure lb shows the PE-trace for the chlorosomes, recorded by variation of the (A)

(B)

Q. CO 0 T3

-200

T'

E

< -200

-100

0

100

T[fS]

200

300

200

400

600

800

= - 200 fs

-i-i, -600

-300

0

-400

-200

0

200

400

600

800

400

T[fs]

Fig. 1. (A) Modeled 2PE-trace for the timing/wave-vector diagram discussed in text. (B) PE-trace recorded for 'dephasing period" r', fixed at - 200 fs. "waiting period" rand the "dephasing period"r' fixed at - 200 fs. The "hole" in the PE-trace around zero delay is clearly resolved (see also the inset). From the ratio between the maximal PE-amplitude and the hole depth (0.008 ± 20%) the magnitude of the HRF can be estimated (using Eq.l) to be S = 1 ± 15%. This value reflects the effective HRF, since we cannot excite a single exciton transition [4] using short laser pulses having approximately 20 niii spectral width. The measurements were repeated for the "dephasing period" r' fixed at - 300 fs (data

602

not shown). In that case the hole around zero delay was also observed, and the magnitude of HRF was 5 = 1.1 ± 30%. The observed decrease in the accuracy of the HRF-magnitude is due to the smaller magnitude of the PE-signal that is related to rapid energy transfer processes taking place in the chlorosomes (ca. 600 fs decay time; measured by the time-resolved transient absorption spectroscopy in the same geometry and under the same experimental conditions). For BChl/MeOH we did not observe a hole around zero delay in the PE-traces, recorded under the same experimental conditions. Taking into account the dynamic range for the PEmeasurements we can conclude that the HRF of the BChl-monomers in MeOH is at least greater than 2.5. For measurements of the electron-phonon coupling strength with a high value of the HRF (>2,) a heterodyne detection of the PEsignals should be applied [5].

4. Conclusions The HRF of the natural molecular BChl c aggregates (chlorosomes) has been measured using a new experimental method, proposed and realized for the first time. The magnitude of the HRF at room temperature is surprisingly low (~ 1) as compared to the HRF of the BChl-monomers in methanol (HRF > 2.5). The maximum magnitude of the HRF which can be measured under reasonable experimental conditions depends on the method of PE-signal detection. For homodyne detection it can be estimated to be ~ 2.5, whereas for heterodyne detection the sensitivity in the HRF measurements can be enhanced by a factor ~ 2. Acknowledgements. V.I.P. thanks for financial support the Laser Centre of Vrije Universiteit Amsterdam, the Netherlands (grant # 375) and the Netherlands Organization for Scientific Research, grant # B 81-729. Authors thank M. Reus (MPI fuer Strahlenchemie, Germany) for the preparation of chlorosomes.

References 1 T. Joo and AC. Albrecht, "Electronic dephasing studies of molecules in solution at room temperature by femtosecond degenerate 4-wave-mixing", Chem. Phys. 176, 233-247 (1993) 2 M. Cho, J.Y. Yu, T. Joo, Y. Nagasawa, S.A. Passino, and G.R. Fleming, "The integrated photon echo and solvation dynamics", J. Phys. Chem. 100,11944-11953 (1996) 3 S. Mukamel, Principles of nonlinear optical spectroscopy, Oxford University Press, New York (1995) 4 V.I. Prokhorenko, D.B. Steensgaard, and A.R. Holzwarth, "Exciton theory for the supramolecular chlorosomal aggregates: 1. Aggregate size dependence of the linear spectra", Biophys. J., 85, 3173-3186 (2003) 5 M. Cowan, J.P. Ogilvie, and R.J.D. Miller, "Two-dimensional spectroscopy using diffractive optics based phased-locked photon echoes", Chem. Phys. Lett., 386, 184-189 (2004)

603

Time and Frequency Domain Investigations on Ultrafast Photoisomerization Reaction Dynamics of PYP Haik Chosrowjan , Seiji Taniguchi, Noboru Mataga , Norio Hamada , Fumio Tokunaga^, and Masashi Unno^ ^ Institute for Laser Technology, c/o Technical research Center, Kansai Electric Power Company 3-11-20, Nakoji, Amagasaki, Hyogo 661-0974, JAPAN E-mail: [email protected] ^ Department of Earth and Space Science, Graduate School of Science, Osaka University, Toyonaka, Osaka 560-0043, JAPAN ^ Institute for Multidisciplinary Research for Advanced Materials, Tohoku University, Sendai, Miyagi 980-8577, JAPAN Abstract. Low frequency modes of PYP (photoactive yellow protein), its several mutant and analogue systems have been investigated by fluorescence up-conversion and resonance Raman scattering techniques, complemented by DFT and ab initio MO calculations. Assignments of the oscillatory components to particular vibrations are proposed. Their role in photoisomerization reaction is discussed.

1.

Introduction

PYP functions as a blue light photoreceptor for the negative phototaxis of the purple sulfiir bacterium Ectothiorhodospira halophila, Photoexcitation of PYP starts a photocycle that involves several intermediate states. Photosomerization of its chromophore (4-hydroxycinnamic acid (4HCA), referred also as p-coumaric acid) has been identified as the primary photoreaction triggering the photocycle. The ultrafast time scale of this initial event and involvement of coherently coupled vibrational modes in the reaction have been directly demonstrated by timeresolved fluorescence studies [1]. Although the observed oscillations in the fluorescence dynamics are an interesting physical phenomenon, it still remained to be demonstrated that they are important for understanding the early events in PYP's photoreaction. Following simple considerations emphasize their potential importance. A flipping, twisting or complete isomerization of the chromophore is presumably unidirectional structural deformation occurring on a characteristic time scale. Hence, specific low-fi-equency intra-chromophore modes (constituting periodic structural deformations) with periods on the same or comparable time scale could be effectively coupled to isomerization and play a crucial role in triggering the later reaction. In the case of PYP, the fastest flipping reaction occurs on -300 fs time scale; thus, vibrational modes with periods in the 100 fs - 1 ps time region (33 - 330 cm"^) could be of special importance. To our best knowledge, there are no studies on PYP or related systems to identify and assign the low-fi'equency vibrations involved in the primary photoisomerization process.

604

2.

Results and Discussion

Fluorescence dynamics of native PYP and related systems have been measured at 10 nm equidistant steps in a broad spectral region (450 - 650 nm) covering whole fluorescence spectra. The observed dynamics for native PYP and seven mutants (E46Q, T50V, R52Q, P68A, W119G, E46Q/T50V and E46Q/R52Q) were qualitatively rather similar. Data analysis showed that the best fits could be reached by double exponential decays superimposed with exponentially damping two oscillatory modes. The under-damped mode (-135 cm"^) does not change dramatically due to the mutation, an indication that it is of an intrachromophore origin. On the other hand, for the over-damped mode (-50 cm"^) the relative spread of frequencies depending on mutation was comparatively larger, indicating that it possibly "feels" the weak chromophore-protein interactions. It is worth to mention that recent experiments with 30 fs time resolution do not reveal any additional modes coupled to the spontaneous fluorescence dynamics. Primary dynamics of PYP analogues with modified chromophores qualitatively differ from those of native and mutant PYPs. Fluorescence dynamics of ferulic acid, dmac acid and caffeic acid analogues show a familiar picture of solvation dynamics qualitatively resembling that of the denatured PYP, whereas the oscillations are almost negligible. In contrast to denatured PYP's case, however, it has been shown that the observed phenomena in analogues are of intraprotein origin. On the other hand, dynamics of PYP analogue with locked chromophore is of primary interest. In contrast to other systems, no solvation effect in its primary dynamics was observed. Furthermore, in this analogue, where the twisting motion around the double bond is hampered, fluorescence decay becomes very long (-60 ps) indicating that the fast photoisomerization does not occur at all. Additionally, oscillatory features on both short- and long-wavelength regions could be observed. It was not possible to reproduce the decay profiles with two oscillatory components in analogy with those of native PYP, however, a model function with three oscillatory components could nicely fit the data A decay example together with the dynamics of the native PYP is shown in Fig 1. (a)

(b)

y^r~{ I /.-..

X

--ci^- ^ F i g . 1. {right) Decay examples for native PYP (a) and locked chromophore analogue (b) at the blue edge of the corresponding spectra, {left) A sketch of the optimized structures for trans-WCKY (a) and HCCET (b), and some selected low-frequency modes in the ground state. Arrows in each sketch show corresponding atomic displacement vectors.

605

The fits predict frequencies of coupled modes for the locked analogue to be at -60, 115 aad 165 cm'\ Comparative analysis of low-frequency RR spectra of native PYP and locked analogue could shed more light on the nature and role of coherently coupled vibrational modes in photoisomerization reaction dynamics of PYP. In native PYP four Raman bands at 104, 153, 183 and 202 cm"^ could be observed. These bands represent skeletal vibrations of the chromophore as a whole. Optimized structures and calculated atomic displacements for selected normal modes of HCET and HCCET are presented in Fig. 1. The RR bands in native PYP's spectrum at 183 and 202 cm"^ are assigned to in-plane skeleton-bending modes V42 and V41, while the bands at 104 and 153 nm"^ are assigned to out-ofplane modes 7 n and 716, respectively. Furthermore, a close examination of the computed frequencies in the ground and excited states reveals a slight downshift tendency for the out-of-plane (7) modes in the Si state whereas in-plane (v) modes are almost unchanged in both So and Si states. This result supports the assignment of the 135 cm"^ mode observed in time-domain fluorescence experiments to 716 (130 cm"^) because the RR spectrum of the native PYP shows a band at 153 cm"\ Conceming the -50 cm"^ mode in time-domain experiments, the calculations predict two possibilities, V43 (69 cm'^) or 7 is (63 cm'^). In the PYP analogue with locked chromophore the 60 cm"^ mode could be either 7 is or 719 (61 and 59 cm"^ in CIS calculations for HCCET), the 115 cm"^ mode is the in-plane V43 (106 cm'^) and the 165 cm"^ - V42 (141 cm'^). As expected, in this system the main contributions come from the totally symmetric A' (inplane) modes with no role in photoisomerization reaction. Based on obtained results and analysis, here we present a possible photoisomerization model for PYP. After a blue-light absorption, the system is elevated into the FC excited state. The FC to Fl state conversion takes place rapidly accompanied with ultrafast FVR leading to the Fl state with selectively excited coherent modes. One of these modes, identified as 716, manifests itself as oscillations in the fluorescence decay, flips the C8 carbon and corresponding hydrogen atoms of the chromophore (see corresponding atomic displacements of the 716 mode in Fig. 1), and effectively triggers the isomerization. Though the estimated energy of this mode is pretty low (150 cm"^ - 0.5 kcal/mol), it still may drive a portion of excited chromophores into the twisted state because the fastest process in PYP has been shown to be of a barrierless origin [2]. Because the mutation effect in time-resolved experiments almost does not affect this mode, we suppose that it only triggers the reaction without controlling its rate. On the other hand, for the lowest mode (-50 cm"^), the relative spread of frequencies depending on mutation was remarkably larger. Hence, we speculate that this mode could "guide" the twisting and probably control also the reaction rate. Acknowledgement This work was supported by Grant-in-Aid for Sci. Research (No.: 16350016) from the Japanese Ministry of Education, Science and Culture.

References 1 N. Malaga, H. Chosrowjan, et al., Chem. Phys. Lett. 352, 220, 2002. 2 N. Mataga, H. Chosrowjan, et al., J. Phys. Chem. B 104, 5191, 2000.

606

Entire view of coherent oscillations in ultrafast fluorescence for photoactive yellow protein R. Nakamura^ N. Hamada^ H. Ichida^ Y. Kanematsu^ and F. Tokunaga^ ^ JST-CREST, Venture Business Laboratory, Osaka University, Yamadaoka 2-1, Suita, Osaka, 565-0871, Japan E-mail: [email protected] ^ JST-CREST, Graduate School of Science, Osaka University, Machikaneyama 1-1, Toyonaka, Osaka, 560-0043, Japan Abstract. Remarkable oscillatory conponents are observed in the 2-dimentional timewavelength map of ultrafast fluorescence for photoactive yellow protein, by using the optical Kerr gating system with 180-fs time resolution and 5-nm spectral resolution.

1.

Introduction

Ultrafast spectroscopies have revealed that the coherence generated by photo excitation is conserved for some picoseconds in several kinds of proteins, including photosynthetic proteins, heme proteins, retinal proteins, and other proteins [1]. Such coherence has been observed as oscillations in the time profiles of transient absorption, stimulated emission, and also spontaneous emission. The time-resolved fluorescence spectroscopy is straightforward in comparison with transient absorption spectroscopy, since the fluorescence process is only related to the photoemissionfi-omthe excited state. Recently, the time-resolved fluorescence for photoactive yellow protein (PYP) has been measured by using the upconversion technique and coherent oscillations were observed in the time profile [2,3]. hi their study, it is recognized that the coherent oscillations observed at red and blue edges of the broad fluorescence spectrum have opposite phases. It is of considerable interest whether or not the oscillation continuously covers the entire energy region of the fluorescence spectrum with large amplitude of wave packet oscillation, like strongly coupled electron-phonon systems, hi order to directly observe the wave packet motion in the electronic excited state, time- and wavelength-resolved fluorescence for PYP has been measured by employing the optical Kerr gate technique.

2.

Experimental

The time-resolved fluorescence spectroscopy has been performed at room temperature by using a measurement system based on the optical Kerr-gate technique [4]. The time resolution, the spectral resolution, and the Kerr-gated transmittance of our system are 180 fs, 5 nm, and 5 %, respectively.

607

3. Results and Discussion Figure 1(a) shows the time-resolved fluorescence spectra normalized by the peak intensity at each wavelength. This mapping helps clear view of the phase relation among time profiles at different wavelengths. On the upper horizontal axis, we indicate ^ o - ^ to give a guide for evaluation of the energy difference and the energy location of the emitted photon. Here, OJ^Q indicates the difference energy of potential minima between electronic ground and excited states as illustrated in Fig. 1(c), and ^ 0 can be estimated as 21367 cm~^ for PYP from absorption and fluorescence spectra. When emission occurs from the bottom of the potential surface in the electronic excited state, cc^^rOh corresponds to the excited vibrational energy of intramolecular and/or intermolecular motions in the ground state. On the other hand, the region of the minus values of Oh^rCOi corresponds dominantly to the relaxation process in the electronic excited state. The most striking feature in the contour map of Fig. 1(a) is that four or more components with oscillation exist at mean energy of cc^^-Oh =850, 1250, 1600, and 1850 cm~^ as indicated by solid lines. The oscillatory components are respectively packed into narrow energy sections, which can be seen as ridges, as indicated by dotted-tracing arrows, in the landscape on the map. The ridges are winding and descending with time. The period of the oscillation is about 0.5 ps and the width of each energy section is about 200-500 cm"^ The period of the sub-picosecond

1000

2000

200M

iStOCJ

UiMO

HMO

Excited State

(c)

4e0

500

540

Vibronlij State

WAVELENGTH (nm)

Ground Stale

Fig. 1. (a) Contour representation of time-resolved fluorescence spectra normalized by the peak intensity at each wavelength. Excitation wavelength is 400 nm. (b) Contour representation of model calculations. The parameters used in the calculations: <2;i =20000 cm~^ £'R=250 cm~\ and fiill width at half maximum of excitation pulse is 150 fs. (c) Schematic potential energy diagram.

608

oscillation is consistent with that observed in the time profile measured by upconversion technique [2,3]. However, it has been recognized, for the first time, that the oscillation does not continuously cover the entire energy region of the fluorescence spectrum but is restricted within the narrow energy section. The width of energy section of about 200-500 cm"^ suggests that the interaction between the chromophore and the surrounding protein is not very strong. In order to analyze the above feature, model calculations are performed under the assumption that two-level electronic states linearly coupled with harmonic oscillations [5]. To simplify the calculations, we take into account WDOS (density of vibronic states weighted by the coupling strength) which consists of a Gaussian distribution function with the peak energy of 30 cm~^ for low-frequency modes, and a Lorenz function with the peak energy of 1500 cm~^ for a high-frequency mode. As shown in Fig. 1(b), some oscillatory components are recognized. The oscillation period of 1.5 ps corresponds to the peak energy of the Gaussian distribution function for low-frequency modes. The oscillatory components located at mean energies of ^o-<^ =250 (=£'R) and 1750 cm~^ (=£'R+1500) are correspond to photo emission accompanied with excitation of low-frequency modes, and low- and high-frequency modes in the electronic ground state, respectively. The other components located in the higher energy region of OJt^Q-coi are related to overtones of high-frequency mode. From these calculations, the mean energies of each components observed in the experiment are possibly assigned to the combination of excited vibrational energy of low-frequency modes and the fundamental tones of intramolecular vibrational modes in the electronic ground state in the chromophore. By assuming that the E^ is 350 cm~^ in PYP, intramolecular vibrational energies are estimated as 500, 900, 1250, 1500 cm~^ from the observed mean energies. This result is consistent with resonance Raman scattering experiments [3,6]. hi summary, the early stage in the photo-induced dynamics of PYP has been investigated at room temperature by the femtosecond time-resolved fluorescence spectroscopy using an optical Kerr gate technique. Remarkable oscillatory components in time and energy have been directly observed, the period of which is about 0.5 ps. The oscillatory components can be attributed to the combination of the intramolecular vibration of the chromophore and the surrounding protein motions.

References 1 M. H. Vos and J.-L. Martin, in Biochim. Biophys. Acta, Vol. 1411, 1, 1999. 2 N. Mataga, H. Chosrwjan, S. Taniguchi, N. Hamada, F. Tokunaga, Y. Imamoto, and M. Kataoka, inPhys. Chem. Chem. Phys., Vol. 5, 2454, 2003. 3 H. Chosrowjan, S. Taniguchi, N. Mataga, M. Unno, S. Yamauchi, N. Hamada, M. Kumauchi, and F. Tokunaga, in J. Phys. Chem. B, Vol. 108, 2686, 2004. 4 R. Nakamura and Y. Kanematsu, in Rev. Set Inst., Vol. 75, 636, 2004. 5 M. Hama and M. Aihara, in Phys. Rev. A, Vol. 35, 4842, 1987. 6 M. Unno, M. Kumauchi, J. Sasaki, F. Tokunaga, and S. Yamauchi, in Biochemistry, Vol. 41,5668,2002.

609

Ultrafast excited and ground-state isomerization dynamics of the Green Fluorescent Protein chromophore in solution Mikas Vengris^ Ivo H.M. van Stokkum^ Xiang He^, Alasdair F. BelP, Peter J. Tonge^, Rienk van Grondelle^ Delmar S. Larsen^ 1 Faculty of Sciences, Vrije Universiteit Amsterdam, De Boelelaan 1081, 1081 HV Amsterdam, The Netherlands E-mail: [email protected] 2 Department of Chemistry, SUNY at Stony Brook, Stony Brook, New York 11794-3400, The USA Abstract. Ultrafast dispersed pump-dump-probe spectroscopy was applied to a model Green Fluorescent Protein chromophore in solution. Sub-ps photodynamics in the excited and ground state has been observed that is ascribed to a hula-twist isomerization mechanism. In the recent years, the Green Fluorescent Protein (GFP) first isolated from the jellyfish Aequorea victoria has become a widely used fluorescent probe in molecular biology. In contrast to potentially toxic fluorescent dyes, GFP can be fused to other proteins at the genomic level, enabling experimentalists to easily monitor their distribution and movement in cells. GFP has the distinctive property of forming the intrinsic fluorescent chromophore autocatalytically out of a sequence of three amino acid residues. The observed photodynamics of the synthetic GFP chromophore in solution (4'-hydroxybenzylidene-2,3-dimethylimidazolinone, HBDI [1]) is strikingly different from that of the GFP protein: whilst the excited state lifetime of GFP protein is 3.3 ns, the isolated GFP chromophore is virtually non-fluorescent in room temperature liquids with a ~1 ps excited state lifetime [2]. The popular explanation for this marked difference is that in the absence of the protein, the chromophore will undergo excited state isomerization about the bridging bond between the phenolic and imidazolinone rings (Fig. lA, inset). Isomerization enhances internal conversion which considerably shortens fluorescence lifetime. We have performed dispersed pump-dump-probe (PDP) experiments on HBDI in aqueous solutions at different pH's and also in a 66% glycerol/water solution. These multi-pulse experiments are insightful for studying the systems that exhibit complex light-induced dynamics involving multiple excited and ground state evolution pathways with spectrally and temporally overlapping bands [3]. First, an ultrafast laser pulse brings the sample into an excited state, after which, a second pulse resonant with the stimulated emission (SE) is applied which demotes the excited chromophores back to the ground electronic state. This technique allows a greater spectroscopic insight into the dynamics of the chromophore on the ground and excited state potential surfaces. Figures 1B,C show the dispersed pump-probe (PP) results on HBDI at pH=10. The pump-probe spectrum measured 200 fs after the excitation shows distinct

610

ground state bleach (GSB) peaking at 425 nm, stimulated emission (SE) at 510 nm and some induced absoqDtion (lA) around 350 nm. Within several ps, the SE band decays and a new lA band rises at 420 nm which lives longer than the SE band (Fig. IB). The PP signals do not decay to zero but to a terminal value that stays around for longer than 5 ns. This terminal spectrum (which may be incorrectly ascribed to an isomerized HBDI product) exhibits GSB and two IA bands: one broad, peaking above 650 nm, and a narrower one, peaking around 470 nm. The measured peculiar pseudo-linear excitation power dependence of the terminal spectrum (not shown) allows us to ascribe these bands to the formation of solvated electrons (the broad lA band in the red) and HBDI radicals (narrow lA band around 470 nm) via a resonantly enhanced two photon absorption mechanism [4]. pH=5

1.0 0.8 0.6

0.4 0.2

/, -^ / - ^' / X)

350

pH=1

—

' A

pH=10.5

* *

» \ '*

*,

K:v.\

400

450

VV^velen^h(nm)

500

i-

5e

5350

400

450

500

550

600

650

V\^avelength (nm)

Fig. 1. A: The structure and absorption spectra of HBDI at different pH values. The arrows indicate the wavelengths of pump and dump pulses used in the experiments. B: pumpprobe traces on HBDI in water at pH=10. C: time-gated pump-probe spectra. Wavelengths for the traces and delay times for the spectra are shown in the legends. To further investigate the underlying pathways of the observed dynamics we performed PDP experiments, which allow the controlled manipulation of excited and ground state populations, and provide better insight into the origin of the PP bands [3]. In the PDP experiment, the dumping effect on the SE is clearly seen in the kinetic trace in Fig. 2D. Around 30% of SE is lost upon dumping. The induced absorption at 362 nm (Fig. 2A) is also lost instantaneously, indicating that it is due to excited-state absorption (ESA). The signal at the GSB wavelengths (Fig. 2B) does not recover instantaneously. In fact, it initially increases (probably due to the overlapping ESA from the band at 350 nm which is dumped instantaneously) and then slowly drops on the timescale of several hundred fs. The most peculiar behaviour is observed at the lA band at 470 nm (Fig. 2C). In contrast to lA band at 350 nm (Fig. 2A), the difference absorption here does not decrease, but rather increases after the dump. This indicates that this band is due to an unrelaxed ground state that is formed when a twisted chromophore dissipates its excited state. It can be produced via internal conversion (hence it also appears in a PP experiment) or via dumping. The terminal 10 ps signal (hydrated electron and radical - Fig. IC) is not affected by the presence of the dump pulse - PP and PDP signals are identical at this delay time. This is in line with the two-photon ionization mechanism suggested above.

611

1jb 2 3 4 5 6

Time (ps)

Time (ps)

Time(ps)

Time(ps)

Fig. 2. Pump-probe (open squares), pump-dump-probe (filled circles) and the difference between the two (open triangles) measured in HBDI in water at pH=10. The probe wavelengths (compare with the bands in pump-probe spectra in Fig. IC) are indicated on the graphs. Solid lines show global analysis fit to the data. The global analysis of the data revealed that the observed rates of ground state evolution are slower than those of the excited state. This is slightly surprising because one would expect the excited state potential to be flatter than the one in the ground state: the excitation modifies the conjugation of the ;7r-electron system, weakening the double bonds and making the isomerization energetically favorable. Thus the observed difference in the rates implies that either 1) the ground-state potential may have an appreciable energy barrier or 2) the evolution on the two electronic states proceeds along different reaction coordinates. In both cases, multi-exponential dynamics (as observed) would be expected. The groundstate barrier has not been predicted by the ab-initio calculations [7]. The second explanation requires structural evolution along more than one rotational degree of freedom (which is compatible with a proposed hula-twist isomerization mechanism [6]). In that case, it is impossible to properly define a one dimensional reaction coordinate. The experimental observations highlight the complexity of the HBDI photo-induced dynamics and poses a challenge for the theoretical attempts to understand it.

References 1 A.F. Bell, X. He, R.M. Wachter, P.J. Tonge, Biochemistry. 39, 4423-4431 (2000). 2 K.L. Litvinenko, N.M. Webber, S.R. Meech, Chemical Physics Letters. 346, 47-53 (2001). 3 F. Gai, J.C. McDonald, P. Anfinrud, Journal of American Chemical Society. 119, 6201-6202(1997). 4 M. Vengris, I.H.M. van Stokkum, X. He, A. Bell, P.J. Tonge, R. van Grondelle, D.S. Larsen, Journal of Physical Chemistry A. 108, 4587-4598 (2004). 5 A.R. Holzwarth in Biophysical Techniques in Photosynthesis Edited by J. Amesz, A.J. Hoff, Kluwer, Dordrecht, The Netherlands, 1996. 6 R.S.H. Liu, Accounts of Chemical Research. 34, 555-562 (2001). 7 A. Toniolo, S. Olsen, L. Manohar, T.J. Martinez, Faraday Discussions, in press, (2004).

612

Optimal Control of Femtosecond Photoisomerization of Retinal in Rhodopsin: Effects of Conical Intersections Mayumi Abe\ Yukiyoshi Ohtsuki*\ Yuichi Fujimura\ and Wolfgang Domcke^ ^ Department of Chemistry, Graduate School of Science, Tohoku University Sendai 980-8578 Japan *E-mail: [email protected] ^ Institute of Physical and Theoretical Chemistry, Technical University of Munich D-85747 Garching, Germany Abstract. Femtosecond laser pulses that control the photoisomerization of retinal in rhodopsin are numerically designed using quantum optimal control theory. Preparation of squeezed reactant wave packets through multiple electronic transitions is essential for creating localized product wave packets.

1. Introduction The cis-trans photoisomerization of the retinal chromophore in rhodospin is the initial event of vision, and has been extensively studied both experimentally and theoretically [1-3]. This photoisomerization process may be modeled by lowdimensional potentials as the high quantum yield of isomerization is attributed to partial dynamic guidance provided by surrounding media. For example, Hahn and Stock [2] proposed coupled two-dimensional, two-electronic potentials with conical intersections, which are shown in Fig. 1. This model can reproduce the transient absorption spectra observed by Shank's group [1], in which quantum beat signals are assigned to nonadiabatic packet motion along the reaction coordinate.

'^ct,.

Fig. 1. Model potential energy surfaces.

613

In the light of the fundamental aspects as well as the possible application to nanodevices such as laser-driven photoswitches and memory devices, we will address the coherent control of wave packet motion associated with the cis-trans photoisomerization of the retinal using quantum optimal control simulations.

2.

Results and Discussion

To examine the characteristics of the model potential, we calculate the time evolutions of the Franck-Condon packet and that of the displaced Franck-Condon packet (Fig. 2(Left)). The latter wave packet has the same form as the FranckCondon packet, the central position of which is shifted by 0.5;^ along the effective coupling coordinate. This displacement leads to ca. two-quantum excitation of the coupling mode. In Fig. 2(Left) (a), we can see modulation with a period of -500 fs (-60 cm"^), in good agreement with that of the experimentally observed beat signal [1,2]. In Fig. 2(Left) (b), on the other hand, we see virtually no modulation after the rapid population exchange processes (<300 fs). (a)

0.8O C 0.6|o.40.200 10

0.8-

o

(b)

\ i»/\jt'*"\Nn

1-0.40.20.0-

1000

time (fs) 0.471

0.57t

O.fk

0.771

reaction coordinate

Fig. 2. (Left) (a) Time-dependent populations '^-'^"——••-- -^^ ground (solid line) and excited (dotted line) electronic states with the initial condition of (a) the FranckCondon packet and (b) the displaced Franck-Condon packet (see text). (Right) Effective transition regions (for definition, see text) associated with (a) the Franck-Condon packet and with (b) the displaced Franck-Condon packet. The dashed lines illustrate classical trajectories. The considerable difference in the population modulation patterns can be interpreted by introducing an "effective transition region," as shown in Fig. 2 (Right). The effective transition region is a region in which the potential energy differences between excited and ground states are smaller than the kinetic energy of the wave packet estimated by a classical approximation. Note that the region size is quite sensitive to the kinetic energy particularly when nonadiabatic transitions occur around a conical intersection.

614

According to an optimal control procedure [4], we design optimal pulses that create a localized ground-state wave packet associated with the trans isomer. Figure 3(a) shows the calculated optimal laser pulse. In Fig. 3(b), the solid line shows the time evolution of the target population. Comparing that for the FranckCondon packet (thin solid line), the controlled wave packet is localized in a specified region with much higher probability than the Franck-Condon case, which clearly shows the effectiveness of coherent control.

;l''O 2-

(a)

, 1

li mH Inr f iLikjM

^^•**'\'

o -2-

'

•§ -3-

T

f t

1.0-

(b) /"v

o ^

D. O^-

;'

\• /

0.6-

\ . r\'' "•'•

V

0.2- i\ 1 1 \ / / 0.0100

v/^;^v___

200

300

/

• ^» / • 'i

/^ '^ 1 AA^» /

^ - ^ N

400

time (fs)

Fig. 3. (a) Optimal laser pulse and (b) time evolutions of the target population (solid line), the ground-state population (dotted line) and the excited-state population (dashed line).

3.

Conclusions

We showed that shaping the reactant packet by appropriately designed laser pulses can efficiently manipulate the product packet in the photoisomerization of retinal in rhodopsin. Acknowledgements. This work was partly supported by a Grant-in-Aid for Scientific Research on Priority Areas, "Control of Molecules in Intense Laser Fields" from the METX of the Japanese Government.

References 1 Q. Wang, R. W. Schoenlein, L. A. Peteanu, R. A. Mathies and C. V. Shank, Science 266, 422, 1994. 2 S. Hahn and G. Stock, J. Phys. Chem. B 104, 1146, 2000. 3 A. L. Sobolewski and W. Domcke, J. Phys. Chem. A 105, 9275, 2001. 4 Y. Ohtsuki, K. Ohara, M. Abe, K. Nakagami and Y. Fujimura, Chem. Phys. Lett. 369,525,2003. 615

Ultrafast polarization and vibrational motions in bacteriorhodopsin studied by coherent infrared emission spectroscopy A. Colonna^ G. I. Groma^, J.-C. Lambry\ M. Joffre\ J.-L. Martin^ and M. H. Vos^ ^Laboratory for Optical Biosciences, INSERM U451, CNRS UMR 7645, Ecole Polytechnique, 91128 Palaiseau cedex, France Email: [email protected] 'Institute of Biophysics, Biological Research Re ^Institute Centre of the Hungarian Academy of Sciences, Szeged, H-6726, Hungary Abstract: The primary events in bacteriorhodopsin are investigated by coherent infrared emission spectroscopy of oriented purple membranes. Long-lived vibrational motions involving charge displacements are observed following sudden (<11 fs) macroscopic membrane polarization appearing upon visible excitation.

1.

Introduction

Bacteriorhodopsin is a membrane protein acting as a light-driven proton pump in the purple membrane of bacterium Halobactehum salinarum. Following absorption of a photon by the retinal cofactor, bacteriorhodopsin goes through a photocycle during which protons are pumped across the membrane. Its capacity to withstand a high photon flux and its simple photosynthetic system make it a convenient model for understanding light-driven processes involving retinal proteins, such as vision and energy transduction. It is generally accepted that the conversion of light energy in bacteriorhodopsin involves both isomerization of the retinal chromophore and charge separation, but their respective roles are subject of intense debate. Our present approach makes use of the nonlinear properties of bacteriorhodopsin, to investigate both the electronic polarization and the infrared active vibrational response of the retinal/protein system. Second order nonlinear properties of bacteriorhodopsin have previously been characterized by frequency upconversion (second harmonic generation) upon off-resonance nanosecond excitation [1]. Here we use 11 fs pulses in resonance with the optical transition, and frequency downconversion (optical rectification). Analysis methods and experimental protocols are designed to get access to the electronic and vibrational response of the protein and eventually to their interplay.

616

2.

Material and methods

Purple membrane containing bacteriorhodopsin were isolated from Halobacterium salinarum strain S9. Our X(2) method requires macroscopically oriented samples. Oriented bacteriorhodopsin films can be obtained by electrophoretic deposition [2]. In our case the infrared transparent material germanium was used as electrode/substrate during the deposition. A non-collinear optical parametric amplifier is tuned to deliver 11-fs visible excitation pulses centered at 568 nm. The beam is focused on the rotating sample that radiates in the infrared region. This optical rectification radiation is interferometrically detected using a reference emission from a type >< 0 1 1 GaAs sample [3]. After passing through the bacteriorhodopsin sample, both infrared beams are focused on a HgCdTe detector. The time delay between both infrared beams is varied up to 5000 fs (fig.l). T i :S.a

Fig.l. Experimental setup

3.

Results and discussion

As in optical rectification processes in semiconductors, but unlike those in myoglobm crystals [4], a strong broadband THz radiation is emitted aroimd t=0 (fig.2). This implies that ultrafast (<11 fs) membrane polarization occurs, as discussed in detail elsewhere [5]. The symmetric part of the ringing is due to the low-frequency cut-off of the detector. In addition, vibrational motion of the retinalprotem system gives rise to asymmetric oscillations (t>0 only) whose coherence persists for several picoseconds, i.e. beyond retinal isomerization.

Time(fs)

-500

0

500 Time(fs)

1000

1500

Fig.2. A. Interferometric detection of the infrared emission of Br. Arrows indicate the part used to calculate the spectrum in figure 3. B. Zoom around t=0 Inset: interferogram with a GaAs crystal at the position of the sample. 617

Two quite distinct time regions can be discerned in this vibrational region: at t<300fs the signal is dominated by strong oscillations with frequencies at 8001050cm-i consistent with hydrogen out of plane vibrational modes [6]. These modes diminish in a rather sudden way at t~300fs, roughly the time required for retinal isomerization. The remaining very long-lived oscillations contain sharp peaks (fig.3). Some but not all are well-known from resonance Raman studies and can be identified as bacteriorhodopsin ground state modes [7]. To discriminate between different origins of the emission, these data are analysed by Fourier transformation and a combination of sliding window analysis and frill simulation of the data [4]. In particular, analysis with a sliding Harming window shows time-dependant shifts in frequency corresponding to the hydrogen out of plane mode. The results will be discussed in terms of recent models of charge transfer and retinal isomerization [8, 9].

Frequency (cm-1)

Fig.3. Power spectrum of the vibrational part of the interferogram (t>300fs)

References 1 J. Huang, A. Lewis and T. Rasing, in Journal of Physical Chemistry, Vol. 92, 1756, 1988. 2 G. Varo and L. Kesthelyi, in BiophysicalJournal, Vol. 43, 47, 1983. 3 A. Bonvalet, J. Nagle, V. Berger, A. Migus, J.-L. Martin and M. Joffre, in Physical Review Letters, Vol. 76, 4392, 1996. 4 M.-L. Groot, M.H. Vos, I. Schlichting, F. Van Mourik, M. Joffre, J.-C. Lambry and J.-L. Martin, in Proceedings of the National Academy, of Science of U.S.A., Vol.99, 1323,2002. 5 G. L Groma, J. Hebling, C. Ludwig and J. Kuhl, in Biophysical Journal, Vol.69, 2060, 1995. 6 T. Kobayashi, T. Saito and H. Ohtani, in Nature, Vol. 414, 531, 2001. 7 B. Meyers, R. A.Harris and R. A., Mathies, in Journal of Chemical Physics, Vol.79, 603, 1983. 8 R. Gonzalez-Luque, M. Garavelli, F. Bemardi, M. Merchan, M.A. Robb, and M. Olivucci, in Proceedings of the National Academy of Science of. U.S.A., Vol. 97, 2000. 9 D. Xu, C. Martin and K. Schulten, in Biophysical Journal, Vol.70, 453, 1996.

618

Excited-state dynamics of the lBu% 3Ag", and IBu" states in all-^ra/is-spirilloxanthin as revealed by sub-5-fs time-resolved absorption spectroscopy Takayoshi Kobayashi^ Kumiko Nishimura\ Ferdy S. Rondonuwu^, and Yasushi Koyama^ ^ Department of Physics, Faculty of Science, University of Tokyo, 7-3-1 Hongo, Bunkyoku, Tokyo 113-0033, Japan. E-mail: [email protected] ^ Faculty of Science and Technology, Kwansei Gakuin University, 2-1 Gakuen, Sanda 6691337,Japan E-mail: [email protected] Abstract. The difference absorption spectra and the lifetimes of the four singlet states in all-trans-spirilloxanthin (number of double bonds, n = \3) involved were separately determined by the singular-value decomposition and global fitting, using the title sequential model. The lifetimes of the IB^^, 3Ag', and IBu" states identified together for the first time were determined to be ~ 10, 25 ± 2, and 140 ± 30 fs, respectively.

1.

Introduction

The Pariser-Parr-Pople calculations by the multi-reference method including singly- and doubly-excited configurational interactions (PPP-MR-SDCI) of shorter polyenes predicted the presence of the low-lymg lBu~, 3Ag~, IBu"^, and 2Ag" states [1,2]. The energies of these excited states decrease as linear functions of 1 / (2« + 1), where n is the number of conjugated double bonds in the polyene molecule. In the case of carotenoids havmg the number of conjugated double bond n= \l - 13, the state ordering of singlet states is IBu^ > 3Ag" > lBu~ > 2Ag~ > lAg~ (ground state). In these carotenoids, the 3Ag~ state could not be time-resolved by the use of subpicosecond pulses, and the spectral pattern of the lBu~ and the hot 2Ag~ states sometimes mixed [3-5]. In the present investigation, we utilized sub-5fs pulses, in order to identify the 3Ag" and lBu~ states, and to determine the intrinsic relaxation dynamics of spirilloxanthin (n = 13).

2.

Experimental Methods

All-/ra«5'-spirilloxanthin (n = 13) was prepared as reported [5]. In the timeresolved difference transmission measurement, spirilloxanthin (1 x 10"^ M) in tetrahydrofiiran (THF) was used as a sample.

619

The setup for femtosecond time-resolved absorption spectroscopy was described elsewhere [6-8]. A sub-5-fs 1 kHz pulse train was generated from the noncollinear optical parametric amplifier (NOPA) [9] in the range of 500 - 750 nm. Energy of the pump pulse at the sample was « 40 nJ and that of the probe pulse was about a quarter of the pump pulse. The normalized transmittance changes were measured in the pump-probe delay-time ranging -30 - 1000 fs with a 5-fs or 2-fs interval, and in the spectral region of 520 - 700 nm.

3.

Results and Discussion

The singular-value decomposition (SVD) and subsequent global fitting [10-12], to a sequential internal-conversion of IBu^ -^ 3Ag~ -^ lBu~ -^ 2Ag~ -^ lAg~ (ground), were applied to a time-resolved data matrix, in the 520 - 700 nm spectral- and 15 - 280 fs time-regions, consisting of 117 x 53 data points. Figure 1 depicts the species-associated difference spectra (SADS) that are assigned to the IBu^ 3Ag", IBu", and 2Ag"* states. The SADS of the IBu^ state reproduces the doubly-peaked transient absorption and a pair of IBu^ -^ lAg" stimulated-emission peaks found m the measured timeresolved spectra at 15 and 20 fs. The SADS of the 3Ag" state reproduces fairly well the vibrational progression of the IBu^ -^ lAg" stimulated emission and the transient absorption around 1.85 eV that are seen in the observed spectrum at 35 fs. The SADS of the IBu" state reproduces a pan* of negative peaks ascribable to the vibrational progression of the IBu^ -> lAg' stimulated emission and the double-peaked transient absorption that are clearly seen in the time-resolved spectra in the 50 - 65 fs region. The SADS of the 2Ag"* state exhibits bleaching of the IBu^
<

Photon energy (eV)

Fig. 1. SADS (species-associated difference spectra) Results of SVD and GFA analysis using a sequential model.

620

The difference spectrum agrees well with the observed one in the time-resolved spectra around 220 fs at which the main contribution is of the 2Ag'* state. The lifetimes of the I B / , 3Ag~, and lBu~ states were determined to be ~ 10, 25 ± 2, and 140 ± 30 fs, respectively by the analysis, in which the lifetime of the longlived species, i.e., the 2Ag" state, was fixed to 1.4 ps [10].

4.

Conclusions

The internal-conversion processes in the sequence of IBu^ -^ 3Ag~ -> lBu~ -> 2Ag~ -> 1 Ag" (ground) have been clearly identified for the first time by the present two-dimensional time-resolved absorption spectroscopy using sub-5-fs pump-andprobe pulses. In particular, the 3Ag~ state, which has been theoretically predicted and verified [1-5], is now conclusively confirmed.

References 1 2 3 4 5 6 7 8 9 10 11 12

P. Tavan, and K. Schulten, J. Chem. Phys. 85, 6602, 1986. P. Tavan, and K. Schulten, Phys. Rev. B 36, 4337, 1987. K. Furuichi, T. Sashima, and Y. Koyama, Chem. Phys. Lett. 356, 547, 2002. T. Sashima, H. Nagae, M. Kuki, and Y. Koyama, Chem. Phys. Lett. 299, 187, 1999. T. Sashima, Y. Koyama, T. Yamada, and H. Hashimoto, J. Phys. Chem. B 104, 5011,2000. A. Shirakawa, and T. Kobayashi, Appl. Phys. Lett. 72, 147, 1998. A. Shirakawa, I. Sakane, M. Takasaka, and T. Kobayashi, Appl. Phys. Lett. 74, 2268, 1999. T. Kobayashi, T. Saito, and H. Ohtani, Nature 414, 531, 2001. A. Baltuska, T. Fuji, and T. Kobayashi, Opt. Lett. 27, 306, 2002. F. S. Rondonuwu, Y. Watanabe, R. Fujii, and Y. Koyama, Chem. Phys. Lett. 376, 292, 2003. J.-P. Zhang, R. Fujii, P. Qian, T. Inaba, T. Mizoguchi, Y. Koyama, K. Onaka, Y. Watanabe, and H. Nagae, J. Phys. Chem. B 104, 3683, 2000. H. Nagae, F. S. Rondonuwu, Y. Koyama, and R. J. Cogdell, Biophysical Journal, submitted.

621

Ultrafast relaxation inside proteins: Calculation and measurement of electron-vibration coupling in enzymes B. M. Cho^ R. C, Walker^ L P. Mercer^ I R. Gouldl, D. R. Klug^ ^Biophysics and Biological Chemistry Group and Molecular Dynamics Group, Department of Chemistry, Imperial College London, UK E-mail: [email protected] ^The Scripps Research Institute, La Jolia CA 92037, USA E-mail: [email protected] Abstract: Electron-vibration coupling in alcohol dehydrogenase and zinc substituted myoglobin was calculated using a quantum mechanics/molecular mechanics method. Good agreement with experimental measurements demonstrates the viability of the method.

1. Introduction Central to understanding chemical dynamics in condensed phases is the understanding of how the energy fluctuation of a solvated macromolecule is influenced by its environment. For example, in non-adiabatic electron-transfer theory, the curvature and the origin of the quadratic free energy surfaces are due to equilibrium fluctuations of structure that bring about the energy fluctuation mentioned above [1]. This is an effect of electron-vibration coupling. To this effect, we have applied a recently established quantum mechanics/molecular mechanics (QM/MM) method to calculate the electronvibration coupling in proteins [2, 3]. In our work, we show that the methodology can be applied to proteins with a small energy relaxation, (i.e., Stokes shift) zinc substituted myoglobin, ZnP-MB (0.007eV--0.3kT) and very large relaxation, liver alcohol dehydrogenase, LADH (0.87eV--35kT), which tests the assumption of linear response as the system greatly deviates from equilibrium. In order to validate our methodology, we compare the calculated optical observables with those measured via experiment.

2.Materials and methods The quantum mechanical/molecular mechanical calculation protocol to calculate the electron-vibration coupling in alcohol dehydrogenase and zinc substituted myoglobin is detailed elsewhere [4, 5]. The details of biochemical sample preparations for spectroscopic measurements are found in ref [5, 6]. We performed classical molecular dynamic calculations that sample the phasespace on both proteins, i.e., LADH inhibited with n-cyclohexyl formamide and ZnP-MB, using the AMBER 6.0 suite of molecular dynamics programs. We subsequently used these trajectories to calculate fluctuating ground- to excited-

622

state energy gaps for the NADH co-factor and ZnP. From the time-dependent energy gap, we determine the auto-correlation function of thefluctuatingenergy gap which is the central quantity that allows us to calculate the optical absorption and emission spectra, via optical response theory [7]. The methodology was tested by a direct comparison of the calculated spectra with those experimentally measured. The steady state spectra can only provide an ensemble-averaged picture and hence cannot reveal explicitly the time-scales of the individual coupled interactions. However, because the time ordering is maintained during the sampling of the phase-space, the time-dependent information is contained within thefluctuatingenergy gap trace. This therefore enables the calculation of time-varying optical observable, such as the three-pulsephoton-echo-peak-shift (3PEPS) function via the autocorrelation function and reveals the time-scales of the interactions [8], Simulation of the 3PEPS function can therefore test whether the QM/MM methodology can correctly account for the dynamics. ZnP-MB conveniently absorbs in the visible region. This made it possible to carry out a 3PEPS measurement on ZnP-MB with a home-built non-coUinear optical parametric amplifier (NOPA) which converts the beam of 800nm wavelength, 120fs pulse duration at 4jJ pulse energy to 20fs long, 50nJ pulses, tunable in the visible [6]. The design of the 3PEPS experiment is largely based on those which are widely reported in the literature [9, 10]. Further details are found in Ref [6].

3.Result$ The main results are summarised in Figure 1 and Figure 2.

S

1.8

Wavetength (nfr») S 8

2.0 Z2 2.4 2.6 2.8 3.0 3.2 3.4 3.0 3.8 Energy (eV) Figure 1 Calculated (solid) and nrieasured (dotted) steady state spectra of ZnP-MB and LADH

4.0

4.2

time/fe Figure 2 Calculated (solid line) and measured (circle) 3PEPSof2nP.MB.

Figure 1 shows the calculated absorption and emission spectra (solid lines) of LADH and ZnP-MB overlaid with that of measured (dotted). Good qualitative agreement shows that the electron-vibration coupling strength and dynamics are well represented in the QM/MM calculation for proteins with small and large

623

relaxation. In fact even with such a large Stokes shift seen in LADH, the Stokes shift and the width are recovered within 20% accuracy. Figure 2 compares the measured and calculated 3PEPS of ZnP-MB. The initial rapid decay in both traces agrees very well. However, the long-time offset and the oscillatory features are not very well reproduced in the QM/MM calculation. We are currently investigating the source of the offset in the 3PEPS data which probably comes from having two nearly degenerate electronic states (that are simultaneously excited in the 3PEPS experiment) that are adiabatically coupled.

4«Condusion QM/MM calculations were performed on proteins with small and large energy relaxation. The steady state spectra were well reproduced even in the case of a large Stokes shift (0.87eV). This demonstrates the versatility of the methodology in calculating the electron-vibration coupling in proteins. The ultrafast initial dynamics (the first 50fs) are well reproduced in the QM/MM calculation. The discrepancies in the calculated and measured 3PEPS functions of ZnP-MB may be due to adiabatic effects in the electronic states.

Reference 1. Warshel, A. and W. W. Parson, Computer-Simulations of Electron-Transfer Reactions in Solution and in Photosynthetic Reaction Centers, Annual Review of Physical Chemistry, 1991. 42: p. 279-309. 2. Mercer, LP., I.R. Gould, and D.R. Klug, A quantum mechanical/molecular mechanical approach to relaxation dynamics: Calculation of the optical properties of solvated bacteriochlorophyll-a, Joumal of Physical Chemistry B, 1999. 103(36): p. 77207727, 3. Mercer, LP., LR. Gould, and D.R. Klug, Optical properties of solvated molecules calculated by a QMMM method - Chlorophyll a and bacteriochlorophyll a. Faraday Discussions, 1997(108): p. 51-62. 4. Walker, R.C., The Development of a QM/MM Based Linear Response Method and its Application to Proteins. PhD Thesis 2003, Imperial College London, 5. Walker, R.C., et al.. Large andfast relaxations inside a protein: Calculation and measurement of reorganization energies in mcohol dehydrogenase. Joumal of Physical Chemistry B, 2002, 106(44): p. 11658-11665. 6. Cho, B.M., Protein Dynamics Measured By Non4inear Spectroscopy. PhD Thesis 2003, Department of Chemistry, Imperial College: London. 7. Mukamel, S., Principles of Optical Nonlinear Spectroscopy. 1st ed. Oxford Series in Optical and Imaging Science, ed. M. Lapp, et al. 1995, Oxford, New York: Oxford University Press. 543. 8. Cho, M.H., et al., The integratedphoton echo and solvation dynamics. Joumal of Physical Chemistry, 1996. 100(29): p. 11944-11953. 9. Joo, T.H., et al., Third-order nonlinear time domain probes of solvation dynamics. Joumal of Chemical Physics, 1996. 104(16): p. 6089-6108. 10. deBoeij, W.P., M.S. Pshenichnikov, and D.A. Wiersma, System-bath correlation junction probed by conventional and time-gated stimulated photon echo, Joumal of Physical Chemistry, 1996. 100(29): p. 11806-11823.

624

Direct observations of ligand rebinding trajectories in myoglobin by femtosecond midIR spectroscopy Seongheun Kim and Manho Lim Department of Chemistry, Pusan National University, Busan 609-735 Korea E-mail: [email protected]. The rebinding dynamics of NO in photolyzed MbNO are investigated by femtosecond mid-IR spectroscopy. The spectra with conformer-specific kinetics reveal the details of ligand binding trajectories and suggest that the conformational relaxation controls ligand binding.

1. Introduction Detailed dynamics of ligand motion in proteins are necessary to understand complex biochemical reactions. Myoglobin (Mb), a ligand-binding heme protein, has long served as a model system for investigating ligand transport and binding in proteins. Ultrafast spectroscopic techniques, in particular femtosecond IR spectroscopy of CO after photolysis of MbCO, have contributed much to understanding ligand dissociation dynamics [1]. Hov^ever much less is known about the trajectories of ligand binding from the heme pocket in biologically relevant condition. Vibrational spectrum of ligand, very sensitive to its environment, is advantageous to directly follov^ the ligand motion. In contrast to MbCO, most of the photolyzed NO from MbNO are geminately recombined in picoseconds at room temperature and its rebinding kinetics is substantially nonexponential [2]. Thus photolyzed NO from MbNO provides means to study the trajectories of the ligand binding process, which cannot be addressed in widely studied CO experiment. The study of geminate rebinding (GR) can also probe the coupling of protein motion to the rebinding process, which can address the nature of the underlying protein dynamics that control ligand binding. Microperoxidase (MPll), an excellent model system for the active sites of heme proteins, was also probed to learn intrinsic properties of the heme affecting ligand-binding dynamics free from protein environment.

2. Materials and Methods Two identical home-built optical parametric amplifiers, pumped by a commercial Ti:sapphire amplifier, are used to generate a visible pump pulse and a mid-IR probe pulse. The polarization of the pump pulse was set at the magic angle relative to the probe pulse to recover the isotropic absorption spectrum. The broadband

625

transmitted probe pulse is detected with a 64-elements N2(/)-cooled HgCdTe array detector. The instrument response function is typically 180 fs. All samples (12 mM MbNO, 12 mM MpNO) were prepared in a pD 7.4 phosphate buffer. Throughout the experiments the integrity and concentration of sample was routinely checked using uv-vis and FT-IR spectroscopy. During data collection the sample cell was rotated or flowed sufficiently fast so that each photolyzing laser pulse illuminated a fresh volume of the sample. The temperature of the sample cell was kept at 283 ± IK.

3.

Results and Discussion (a)

MpNO

(b)

MbNO

^v^f 1720

1690

1660

1630

1600

1650

1620

1560

Fig. 1. (a) Representative time resolved IR spectra of photolyzed MbNO and MpNO in D2O (mOD = 10'^ optical density), (b) Kinetics of GR of NO to Mb and Mp. Fig. 1(a) shows time-resolved mid-IR absorption spectra of MbNO and MpNO photolyzed at 580 nm. The negative going features, arising from the loss of bound NO, can be modeled with a sum of two Gaussians for MbNO and a single Gaussian for MpNO. The two Gaussians imply that MbNO exists in at least two distinct conformational substates. The relative amplitude of the two Gaussians of MbNO is time independent, suggesting that two conformers have the same GR kinetics. The interconversion rate between two conformers with different rebinding barriers is likely too slow to account for the same GR kinetics. The same GR kinetics is observed because a significant portion of the ligands rebind before each substate relaxes into its own distribution of rebinding barrier. As shown in fig. 1(b), rebinding of NO to Mp is ultrafast (5.6 ps time constant) with near unity geminate yield, suggesting that intrinsic rebinding of NO to the active heme free from the protein environment is almost a barrierless process. The kinetics of NO rebinding to Mb is nonexponential, which is well described by the time-dependent rebinding rate incorporating a time-dependent barrier [2]. The initial rebinding rate, (7.7 ps)"\ similar to the intrinsic rebinding rate to heme, (5.6 ps)'\ gradually decreases as the barrier grows toward its equilibrium height with protein relaxation. The rate constant for the variation in the barrier, found to be (16.5 ps)'\

626

compares well with the effective rate for conformational relaxation (CR) of Mb, (18 p s ) \ These results clearly show that CR plays a dominant role in the nonexponential NO rebinding to Mb and static inhomogeneity has little influence on the dynamics of NO rebinding to Mb under physiological condition. Fig. 2(a) shows transient IR spectra of photolyzed NO at various delays. The transient spectra of photoproducts, appearing ultrafast timescale, can be decomposed into evolving three vibrational bands {BQ, BJ and B2) plus their redshifted replicas arising from vibrationally hot band. They are slightly shifted from the gas state value, 1876 cm'^ and similar to that observed at 7 K [3], indicating that they are arising from NO in the heme pocket. Spectral evolutions shown in fig 2(b), also observed in photolyzed CO spectra of MbCO, arise from CR of protein after photolysis. Initial increase of the total integrated area is attributed to structural rearrangement of the residues surrounding the heme pocket to fill empty binding site [1]. NO rebinding, transitions between 5-states and vibrational relaxation (VR) of NO can affect the integrated absorbance of 5-states. We have considered various models to describe the kinetics of NO rebinding including transitions between 5-states and VR of NO but have found only one acceptable one. We assume that there are three j8-states that can interconvert and NO rebinds only from Bj with a time-dependent rate constant. The model suggests that photolysis leads from A to BQ and Bj and geminate recombination occurs via thermal transitions from BQ to Bj, between Bj and B2, and Bj to A. It is consistent with faster rebinding ofB] at cryogenic temperature [3]. 3.2 ps 5.6 ps 10 ps 18 ps 32 ps 56 ps

(a)

(b)

3

"E

B-

0) O)

1

Q 1860

[^

B

^

=M J

Bi

2

ro

1

\

-e

3L

"v

s 1

0

20

40

60

80 100

1

1 B,

I

1

A>-~U - - ^

1 [ \ ^ CO

§

, A-,,

1910 1890 1870 1850 1830 1810

c

BJ

o- 1870

2

B.

t \

1880

JAM Tioops -

*^B7

(c)

'c

f^.': r-rrrr\ 0

20

40

60

80 100

Fig. 2. (a) Representative time resolved spectra of photolyzed NO from Mb. FTIR difference (photolyzed minus unphotolyzed) spectrum of ferrous MbNO at 7 K is obtained from reference 3. (b) The fitted parameters for Gaussians. Filled (open) symbols represent centerfrequencies(FWHM's) of Gaussians (c) Integrated absorbance changes. Acknowledgements. This work was supported by grants from the Korea Science and Engineering Foundation through the CMS (POSTECH).

References 1 M. Lim, T. A. Jackson, and P. A. Anfmrud, Nature Struct. Biol, 4, 209, 1997. 2 J. W. Petrich et al. Biochemistry, 30, 3975, 1991. 3 L. M. Miller, A. J. Pedreza and M. R. Chance, Biochemistry, 36, 12199, 1997.

627

Coherent vibrational climbing in carboxyhemoglobin Cathie Ventalon, James M. Fraser, Marten H. Vos, Antigoni Alexandrou, JeanLouis Martin and Manuel Joffre Laboratoire d'Optique et Biosciences, INSERM U451, CNRS UMR 7645, Ecole Poly technique, 91128 Palaiseau cedex, France E-mail: [email protected] Abstract. We demonstrate vibrational climbing up to level v=6 in carboxyhemoglobin by use of intense negatively-chirped infrared pulses. This technique gave new spectroscopic insight into carboxy-hemoglobin, such as transition frequencies and dephasing times up to the V = 6 to V = 7 vibrational transition.

1. Introduction Coherent vibrational ladder climbing using infrared ultrashort laser pulses is an attractive approach for direct control of the nuclear motion of molecules. In particular, it could be a powerful tool for studying proteins since it allows one to explore the potential energy surface far from the harmonic region. This technique relies on the use of infrared pulses which need to be negatively chirped in order to take into account the anharmonicity of the addressed molecular vibration [1]. The rising edge of the pulse is thus resonant with the lower transitions of the ladder while the trailing edge is resonant with the upper transitions. Vibrational climbing has been demonstrated previously in small molecules such as NO [2], W(CO)6[3], Cr(C0)6 [4] and CH2N2 [5] with evidence of molecular dissociation in the last two cases. We report here on the first observation of coherent vibrational climbing in a biological system, carboxy-hemoglobin (HbCO) [6].

2. Experimental Methods The CO vibration is selectively addressed using mid-infrared pulses of energy 2.2 ju] centered at 1900 cm'^ with a spectral width of 170 cm'^ (repetition rate 1 kHz). These pulses are easily chirped using transmission through CaF2 (cp <0) or germanium (qp >0) plates. The sample is a 50-jum thick cell of HbCO in D2O buffer at a heme concentration of roughly 20 mM. To detect the presence of a population in the excited vibrational states, we record the differential spectrum of a broadband infrared probe after transmission through the sample, using a Fouriertransform spectrometer. The induced absorption at frequency (Oy+i^ v is proportional to the population difference between levels v and v+1.

628

3. Results and Discussion Fig. 1. shows differential frequency-resolved pump-probe spectra recorded for different values of the chirp and the pump-probe time delay. Spectra (a), (b) and (c) are obtained for a pump-probe time delay of 16 ps and for three different values of the chirp. For each differential spectrum, we record a positive peak at frequency coio which results both from the decrease of absorption between levels v=0 and v=l and from stimulated emission from v=l to v=0. We also record negative peaks at frequency a)v+i,v with v>l, which are associated with the induced absorption from level v to level v+1. These absorptions occur at a lower frequency due to the vibrational anharmonicity. The fact that all of these peaks have negative sign indicates that the population is decreasing as we climb the ladder. By comparing curves (a) and (c), we confirm that vibrational climbing is much more efficient for a strong negative chirp than for a positive chirp. More specifically, with qp =-32000 fs^ (c) we observe the presence of a negative peak at frequency a>76 indicating that level v=6 has been populated. In contrast, with cp =+6000 fs^ only the first two excited states are significantly populated (curve (a)).

0.6 U

1750

1800

1900 1950 1850 Wavenumber [cm ]

2000

Fig. 1. Differential spectra -Aa(co)L measured in HbCO for a pump-probe time delay x and a chirp qp (qp is the second order derivative of the spectral phase), (a) x = 16ps and qp =6000 fs^; (b) X = 16ps and cp =-6000 fs^;. (c) x = 16ps and cp =-32000 fs^;. (d) x = 7ps and cp =32000 fsl Curve (d) shows a differential spectrum measured for an earlier time delay (7ps). We then observe a positive peak at frequency (1)54 demonstrating that population inversion has been achieved, i.e. population in level 5 is larger than

629

population in level 4. Moreover, the presence of a small negative peak at frequency cogv suggests that level v=7 has been populated. In Fig. 1, we also observe that the absorption lines up to the 6—>7 transition are regularly spaced: C0v+2,v+i -Wv+i,v = 25.0 ± 0.1 cm"\ This is in agreement with the transition frequencies expected for a Morse oscillator. In addition, it appears that the spectral widths of the absorption lines are independent of the considered transition, with a value of 6.5 cm'\ corresponding to dephasing times of roughly 1.6 ps for each transition. Such values of the dephasing times ensure that the excitation takes place in the coherent regime even for the longest pulses used in this work (duration 1.4 ps). The last point we want to address is the value of the integral of the measured differential spectra ( Cls.a((o)L d(0). For a weakly anharmonic oscillator, this integral should vanish. Fig 1 (d) shows that it is not the case. A possible interpretation for this behavior is that the transition dipoles of the CO vibration strongly deviate from those of a harmonic oscillator.

4. Conclusions We performed the first demonstration of coherent vibrational climbing in a biological system, carboxy-hemoglobin (HbCO). This study shows that a specific vibrational transition of a ligand in a protein can be efficiently excited via coherent excitation in the infrared domain. Indeed, we were able to deposit in HbCO an energy equivalent to 6 or 7 quanta of the CO vibration, which means that we can explore the potential energy surface far from the harmonic region. We also measured important information such as vibrational frequencies and dephasing times which could help in improving our knowledge of the protein force field.

References 1 S. Chelkowski, A.D. Bandrauk, and P.B. Corkum, Phys. Rev. Lett 65, 2355-2358, 1990. 2 D.J. Maas, D.I. Duncan, R.B. Vrijen, W.J. van der Zande, and L.D. Noordam, Chem. Phys. Lett, 290, 75-80,1998 3 S. M. Arrivo, T. P. Dougherty, W. T. Grubbs, and E. J. Heilweil, Chem. Phys. Lett. 235, 247,1995. 4 T. Witte, T. Hornung, L. Windhorn, D. Proch, R. de Vivie-Riedle, M. Motzkus, and K. L. Kompa, J. Chem. Phys. 118, 202, 2003. 5 L. Windhorn, J. S. Yeston, T. Witte, W. FuB, M. Motzkus, D. Proch, K.-L. Kompa, and C. B. Moore, J. Chem. Phys. 119, 641, 2003. 6 C. Ventalon, J. M. Fraser, M. H. Vos, A. Alexandrou, J.-L. Martin, M. Joffre, to appear in PNAS.

630

Evidence for non-separating four-point correlation functions from IR pump-probe spectroscopy of CO in a protein internal cavity Jan Helbing^ Peter Hamm\ Karin Nienhaus^, and G. Ulrich Nienhaus^'^ ^ Universitat Zurich, Physikalisch-Chemisches Institut, Winterthurerstrasse 190, CH-8057 Zurich, Switzerland E-mail: [email protected] ^Universitat Ulm, Abteilung Biophysik, Albert-Einstein-Allee 11, D-89081 Ulm, Germany ^ University of Illinois at Urbana-Champaign, 1110 W. Green Street, Urbana IL 61801, USA Abstract. Polarization-dependent transient infrared spectroscopy probes the restricted ultrafast orientational motion of CO inside the Xe4 cavity of myoglobin. The four-point correlation function \/i(0)^(0)^(T)^(T+t)) does not factorize, only its fiill evaluation reproduces the observed anisotropics.

1. Introduction The orientation dependence of third order response functions is usually evaluated by factorizing the ensemble average of the full four-point correlation function into products of two-point correlation functions, WO)M/3(T)/X/X+T)^^T+T4-t)> oc{jUjiO)lUp{T)) . This separation is exact for isotropic distributions and unrestricted rotational diffusion, but is not necessarily correct in the case of restricted orientational motion. Still, in third order spectroscopies such as fluorescence and pump-probe measurements (where x = 0), only the central correlation function (ju/(0);u/(7)) is normally considered when studying orientational effects [1]. For spherical difPusers or rod-like molecules that are part of randomly oriented macromolecules, the signal anisotropy is then given by [2]: ^/5.nt,

(2)

where P2 is the second Legendre polynomial and flXO) and fl\T) are the dipole moment orientations of the initially excited transition i and the transition j probed after time T. The average Oint is taken over the orientational distribution of these dipole moments within each macromolecule. In the Xe4 binding pocket of myoglobin, CO molecules can get trapped in two opposite directions. Due to a strong electric field inside the pocket, the two orientations give rise to two separate peaks in the IR absorption spectrum (Fig. 631

(1)

la), which spHt further apart and become more prominent as the temperature is lowered. These observations are well reproduced with a double-well potential V(0) =Vo cos^e, where 9 is the angle between the CO molecular axis and a protein-fixed axis, and VQ is a barrier of approximately 170 cm"^ [3]. The spectral changes as a function of temperature indicate that ultrafast librations of the CO molecules are the dominant contributions to the fast decay of the transition dipole autocorrelation function, Ci(t) =(|i(0)-|/(t)) [3]. The higher the temperature, the less restricted is the motion, Ci(t) quickly decays to zero and the absorption spectrum is very broad. At cryogenic temperatures, restricted rotational motion results in a biphasic decay of ^i(t) [4], with an additional slow component governed by pure dephasing and exchange between the two potential wells. This leads to the narrow peaks. Here we investigate this motion directly by ultrafast IR spectroscopy.

2. Results and Discussion I3^I8Q j^olecules trapped in the Xe4 binding pocket of the myoglobin mutant L29W-S108L {= 20 mM in a 3:1 (vol/vol) glycerol/phosphate buffer mixture) were excited with a broad-band IR pulse (FWHM = 200 cm' \ 100 fs), centered at 2020 cm"^ This transfers population to the v = 1 vibrational levels of CO in the two trapping sites and gives rise to the v = 1 -^ v = 2 excited state absorption signals at 2002 cm" ^ and 2015 cm" ^ (Fig. la). Only the narrow components of the bands are seen in the pump-probe spectra, the broad components escape detection. 0.2

I'' ' H

3 0.151V

I' I ' ' ' I

MO)n(0)n(T)n(T+t)>^

0.1 t0.05 h

2000 0.4

S:o.3 2 S 0.2 E-

0.1

0

2020 , 2040 Energy [cm ]

• • * *" 20 K ' ' j

0.4 tT F

20 40 60 0 20 40 60 T [arb. units] t [arb. units] • • • I ' ' ' I ' ' ' I ' • ' I ' ' ' M

4 0.3

D [J

50 K

U

n

: •

° ^4 0.2

lb 1 10 100 Waiting Time T [ps]

0.1

20

30 40 50 60 Temperature [Kelvin]

70

Fig. 1. (a) FTIR and pump-probe spectrum of ^^C^^O in the Xe4 pocket of myoglobin at 30 K, recorded 50 ps after excitation with a broad band IR-pulse. (b) Anisotropy decay for the excited state absorption signal at 2002 cm"^ at 20 K and 50 K. The solid lines are exponentials with time constants of 1 ps and 650 fs, which correspond to ballistic motion, (c) and (d) Cuts through the four-point correlation functions probed by pump-probe spectroscopy with parallel and perpendicular polarizations, (e) Experimental (solid circles) and calculated anisotropics (solid line: eq. 2, triangles: MonteCarlo analysis of full fourpoint correlation functions). 632

The ultrafast libration of the CO molecules inside the protein cavity is directly observable by the anisotropy decay of the 2002 cm" ^ band, as shown for two different temperatures in Fig, lb. At 20 K, orientational motion is strongly hindered. The fast anisotropy decay at 50 K is consistent with ballistic motion inside the cavity with a time constant {I/ksT)'^'^ = 650 fs, governed by the moment of inertia /^SxlO""^^ kg m^ [3]. The filled circles in Fig. le show the anisotropy at different temperatures after reorientation of CO within the pocket is completed. They are significantly larger than the anisotropy expected from eq. 2 using the best-fit potential V(6) from linear spectroscopy (thick solid line). We can resolve this discrepancy by evaluating the full four-point correlation functions ^(0)M.(0)M.(T)M.(T+t)) (II signal) and {^MnMl^ymHy(T+t)) ( 1 signal), without the separation of eq. 1. Cuts through these correlation functions (from a MonteCarlo simulation) are shown in Fig. Ic and Id. After equilibration within each potential well, the correlation fimctions are constant as a function of waiting time T. As a function of t (for T longer than this equilibration time) a biphasic decay causes narrow and broad features in the pump-probe spectrum, just as the biphasic decay of C](t) gives rise to the broad and narrow components in the linear spectrum. In the four-point correlation ftinctions, however, the t-decay is polarization-dependent, and the contribution of the fast component is much larger for the correlation fiinction belonging to the perpendicular signal. Since the pumpprobe spectra are given by the t-Fourier transform of the four-point correlation functions, the relative contribution of the narrow component in the perpendicular spectra is therefore smaller. We are thus confronted with the unusual situation that the spectral shape of the pump-probe signal depends on polarization, i.e. the signal does not factorize into a polarization-independent spectrum and a polarizationdependent intensity as implied by eq. 1. The intensity of the narrow spectral components is proportional to the weight of the slow t-decay component in the four-point correlation ftinction. Taking this into account, MonteCarlo calculations based on the potential V(0) that fits the linear spectroscopy indeed yield anisotropics that are in very good agreement with the experimental data (open triangles in Fig. le). In summary, direct observation of the ballistic libration of CO in a protein internal cavity shows that orientational motion is much faster than vibrational dephasing and determines the fast decay of the dipole correlation ftinctions. Apparent discrepancies between the results of linear and transient spectroscopy were resolved by showing that the ensemble average of the third-order response does not factorize, as is usually the case, into a product of pair-correlation ftinctions. This results in a polarization-dependent spectral shape of the pump-probe signal.

References 1 G. Lipari and A. Szabo, J. Am. Chem. Soc, 104, 4546-4559 (1982). 2 A. Szabo, J. Chem. Phys. 81, 150-167 (1984). 3 J. M. Kriegl, K. Nienhaus, P. Deng, J. Fuchs, and G. U. Nienhaus, Proc. Natl. Acad. Sci. U.S.A. 100, 7069-7074 (2003). 4 M. Lim, T. A. Jackson, and P. A. Anfinrud, J. Chem. Phys. 102, 4355-4366 (1995). 633

The CO oscillator as a probe of ligand dissociation dynamics in myoglobin Jennifer P. Ogilvie^ Thomas Polack^ Stefan Franzen^' Martin H. Vos\ Manuel Joffre\ Jean-Louis Martin\ Antigoni Alexandrou^ ' Laboratoire d'Optique et Biosciences, UMR CNRS 7645 - INSERM U451 - Ecole Polytechnique - ENSTA, F-91128 Palaiseau, France E-mail: anti goni. alexandrou @ poly technique, fr ' Department of Chemistry, North Carolina State University, Rayleigh, NC 27695 USA We report spectrally-integrated, visible-pump, mid-infrared probe studies of the CO ligand in myoglobin. Supported by densit>' functional calculations, we fmd that the CO oscillator strength and frequency changes occui' on disparate timescales following dissociation. Heme proteins accomplish a wide variety of functions including oxygen storage and transport, as well as more recently discovered roles as gas sensing molecules. Understanding how the different heme proteins discriminate between ligands such as O2, CO and NO requires detailed knowledge of the ligand's interaction with the heme binding site and the surrounding residues. In this study, we report ultrafast visible-pump mid-infrared probe studies of carboxymyoglobin (MbCO) to characterize the dynamics of CO oscillator strength and frequency changes as it explores the heme environment following photodissociation. The majority of spectroscopic studies of heme proteins have used visible pump-probe methods and are thus sensitive to the heme electronic transition changes that accompany ligand dissociation. These experiments have established that CO photodissociation occurs on a timescale of <50 fs. Experiments with a mid-infrared probe offer a different perspective since they are directly sensitive to the CO ligand.^'^ The CO stretching frequency and oscillator strength are sensitive probes of the local environment of the CO oscillator. Previous studies of bound CO have revealed a main vibrational line, and two weaker absorption bands arising from different protein substates or conformations."^ Spectrally-resolved mid-infrared probe experiments have shown that upon photodissociation, the CO enters a "docking site", which displays two vibrational peaks corresponding to different orientations of the CO inside the docking site.^'^ As an alternative approach, we spectrally integrate the differential transmission (DT) of the probe. This method has the advantage that it is not subject to perturbed polarization effects that occur in spectrally resolved measurements at negative time delays."^ The laser source for the experiment consists of a 100 kHz Ti:sapphire amplified laser (Coherent RegA 9000), part of which generates a continuum to seed a 800 nm-pumped optical parametric amplifier (OPA). We produce the midinfrared probe via difference frequency in a 1 mm type II AgGaSi crystal (probe bandwidth 100 c m ) . The remaining 800 nm light from the OPA is frequency doubled to provide the pump. The pump and probe geometry is collinear, with the pump polarization at the magic angle (54.7^) with respect to the probe. The

634

MbCO samples (10 mM) are prepared as described previously.^ The high repetition rate requires sample spinning and translation. We chop the pump beam at 50 kHz, and lock-in detect the difference between the transmitted IR intensity and a reference beam. We determine the zero pump-probe delay via a crosscorrelation in GaAs, yielding pump and probe durations of'-250 fs. Fig. 1 (left) shows the DT curves for various probe wavelengths.^ The form of the 1945 cm'^ probe signal shows the expected absorption decrease as the CO band moves out of the probe spectrum after dissociation. The other probes that are centered away from the bound CO frequency display an unexpected early rise that can only be explained if the absorption strength of the CO remains strong at early times. We verified that these signals come from the transient CO state via control experiments in deoxyMb and buffer solution which yielded no signal under identical experimental conditions. In Fig. 1 (right) we show the results of a phenomenological model for the pump-probe signal that reproduces the general features of the data. This model includes a time varying CO frequency, oscillator strength and angular orientation with respect to the heme plane. We modelled the probe spectra as gaussians with spectral widths in accordance with the experimental data. To reduce the number of parameters, we kept the CO absorption width constant, and modelled it as a gaussian of 12 cm' width, centered initially at the dominant CO frequency of 1945 cm"\ and evolving to a single docking site frequency at 2130 cm'\ We considered many functional forms for both the oscillator strength change and frequency change dynamics between the known bound and docking site values. The dashed lines show that a quasi-instantaneous change in oscillator strength cannot explain the data. We achieved the best qualitative fit to the data for a quasiinstantaneous (within 50 fs) change in CO frequency, and a slower 400 fs exponential decay in oscillator strength.

] 2064 cm"'

Delay (fs) Delay (fs) Fig. 1 Differential transmission in MbCO for various probes central wavelengths. Left: experimental results. Right: calculated results using a phenomenological model that assumes a quasi-instantaneous frequency change, accompanied by a progressive oscillator strength change (solid) or a quasi-instantaneous one (dashed). To further our understanding of the factors determining the CO frequency and oscillator strength we performed density functional (DFT) calculations for a model system consisting of an iron porphine with a bound CO, and two imidazoles to

635

mimic the proximal and distal histidines/ The distal imidizole was hydrogen bonded to the CO ligand, with a Fe-O...H angle of 110'' and a Fe-O...H separation of 1.92 A. Beginning with the geometry optimized structure, the Fe-C distance was increased and the CO bond length was reoptimized while other coordinates were held fixed. Fig. 2 shows the resulting CO vibrational frequency and absorption intensity as a function of Fe-C distance. In qualitative agreement with our data, the CO oscillator strength does not decrease as quickly as the frequency increases. The DFT calculations show a charge displacement from the Fe to the CO that affects the strength of the CO absorption in a manner that is not linearly coupled to the CO frequency. The delayed CO oscillator strength decay was not observed in previous spectrally resolved experiments, where it would have been difficult to discern due to a number of broadening mechanisms: broadening induced by the rotational freedom of the CO, the inhomogeneity of different protein conformations and trajectories to the docking site, and inherent broadening due to the decrease in oscillator strength and rotation of the CO. Spectrally integrated experiments are not sensitive to these mechanisms as long as the CO absorption line remains narrower than the probe spectrum, making them ideal for a measure of total

2.0

2.5

3.0

3.5

Fe-C distance (A) Fig. 2 DFT of the vibrational CO absolution intensity (solid) and frequency oscillator strength. In conclusion, we have observed that the CO oscillator exhibits a gradual reduction in oscillator strength (400 fs), accompanied by a faster change in frequency as it travels to the docking site. The DFT calculations support these observations. To our knowledge this is the first observation of a progressive oscillator strength change following dissociation, and may be a general feature related to bond breaking in ligated metals. In the case of heme proteins, this implies that the CO absorption strength can be used as a probe of ligand dynamics following dissociation. 1. Petrich, J. W.; Poyart, C ; Martin, J. L. Biochem. 1988, 27, 4049-4060. 2. Lim, M.; Jackson, T. A.; Anfinrud, P. A. Science 1995, 269, 962-966. 3. Lim, M.; Jackson, T. A.; Anfinrud, P. A. J. Chem. Phys. 1995,102, 4355-4366. 4. Frauenfelder, H.; Sligar, S. G.; Wolynes, P. G. Science 1991, 254, 1598-1603. 5. Joffre, M.; Hulin, D.; Migus, A.; Antonetti, A.; Benoit a la Guillaume, C ; Peyghambai-ian, N.; Lindberg, M.; Koch, S. W. Opt. Lett. 1988, 13, 276. 6. Polack, T.; Ogilvie, J. P.; Franzen, S.; Vos, M. H.; Joffre, M.; Martin, J.-L.; Alexandrou, A. Phys. Rev. Lett. 2004, 93, 018102. 7. Franzen, S. J. Am. Chem. Soc. 2002, 124, 13271-13281.

636

A 2DIR study of backbone structure and dynamics of a dipeptide in membrane Victor Volkov, and Peter Hamm Physical Chemistry Institute, University Zuerich, Winterthurerstrasse 190, 8057 Zuerich, Switzerland E-mail: [email protected] Abstract. Spectral and temporal analysis of diagonal and cross peaks in two-dimensional infrared response from Trp-Ala-Alkyl didpeptide indicates a change in structure and in ultra-fast backbone dynamics when the molecule inserts into a membrane.

1.

Introduction

Two-dimensional infrared (2D-IR) experiment [1-3] provides information on structural molecular properties on every time scale of interest even when in a heterogeneous environment. Here we report on time resolved structural properties of a didpeptide in surfactant and lipid membranes, probed in nonlinear ultra-fast IR experiments. The description on both instrumentation and methodology of 2D-IR experiment has been provided elsewhere [4]. Figure 1 represents the molecular components PLPC

SDS

WAC

Fig. 1. The structures of the WAC didpeptide, SDS and its micelle, PLPC phospholipids and its liposome studied in the ultrafast IR experiment. Tryptophan-alanine (WAC) didpeptide was the molecular probe. The carbonyl of Alanine was ^^C labelled. We prepared both micelles of Sodium Dodecyl Sulfate (SDS) and bilayer liposomes of a Phosphatidylcholine (PLPC) as membrane media. FTIR spectra of the prepared samples showed resonances of the carbonyls of alanine at 1600 cm'\ of tryptophan at 1675 cm*', and of PLPC at 1732 cm"'(Figure 2).

637

2.

Results and Discussion

2D-IR spectra of WAC (Figure 2) demonstrate that the diagonal peaks of the carbonyl resonances of alanine is governed by homogeneous broadenning in D2O and by inhomogeneous broadening in membrane. The response of tryptophan is dominated by homogeneous broadenning in all media. It is important to notice that only multidimensional spectroscopy may separate the broadening contributions which are hidden in the linear responses. 1600

1650

1700

1750

1600

1750

1600

1650

1700

1750

1600

1650

1700

1750

1600 1650 1700 1750 Probe Frequency (cm'^)

1600

1650

1700

1750

1650

1700

Fig. 2. The FTIR and the 2D-IR spectra of WAC molecule in D2O, SDS and PLPC (as indicated) for perpendicular polarizations of the pump and the probe pulse. 2D spectra show explicitly both inter- and intra-molecular cross-peaks. This is another imique and powerful aspect of multidimensional spectroscopy. Using the observed intra-molecular cross-peak anisotropics, and the relative intensities of the 1600

1650

1700

1750

I Choline

17504

i 17004

1600

1650 1700 Probe Frequency (cm"*)

1750

Fig. 3. Left side: difference between the perpendicular and the parallel signals for the didpeptide in PLPC liposomes. The markers indicate the cross-peaks between alanine (A), tryptophan (W), and phospholipids (P). Right side: structural model.

638

diagonal and cross-peaks we calculated G - the angles between the amide I dipole moments and the coupling constant (3 between the site oscillators [1]. Matching these structural parameters with the coupling map [5] we propose possible dihedral ((|),\|/) angles of the central peptide bond. Two regions in the Ramachandran plot are consistent with the 2D-IR results: in the area of the unfolded right-handed helix and/or in the region of a strongly folded beta segment. In the first case we obtain ((|),\|/) ~ (-55^ 64'') in D2O and ((|),\|/) « (-70°, 40"") in membrane. In the second case we estimate ((t),\|/) «(-120',-80') in D2O and ((t),\|/) « (-140°,-50') in membrane. This is a very tentative assignment. A careful evaluation of structural distributions in the samples under studies would require results of molecular dynamic studies as it was done for tri-alanine molecule recently [3]. In Figure 2 (left side) we plot the difference between the pump-probe signals under different polarisation conditions to contrast the inter-molecular cross-peaks. Presence of the inter-molecular cross-peaks provides necessary measures to estimate the localization of dipeptide moieties in phospholipid membrane interface (Figure 2). Specifically we propose that the carbonyls of alanine and PLPC are very proximal and the carbonyl of tryptophan is in the outer compartment of the membrane interface. In result we assign the dominant homogeneity of tryptophan response in 2D spectra (Figure 2) to its exposure to ambient water. Correspondingly, the inhomogeneity of the resonance of alanine is due to the rigidity of its local environment which is in the area near to the carbonyl moiety of lipid. It is consistent with the fact that water does not penetrate the membrane interface further than phosphate group [6]. The results of our studies on the rate of cross-relaxation between the two amide I modes of WAC [4] provide an additional support to this conclusion. More specifically we observed the threefold cross relaxation rate decrease when the didpeptide inserts into membrane. The decrease in the cross relaxation indicates to the slower backbone conformational fluctuations when dipeptide partitions into a membrane [2]. In conclusion our results demonstrate the resolving power of 2D-IR spectroscopy in probing structural and dynamical properties of molecules in heterogeneous environment. 2D-IR spectroscopy may play an exclusive role in investigating fluctuating structural properties of a variety of relatively small polypeptides, toxins, amphipathic segments, associated with membranes in a dynamic manner.

References 1 P. Hamm, M. Lim, W. DeGrado, and R. Hochstrasser, Proc. Natl. Acad. Sci., Vol. 96,2036,1999. 2 S. Woutersen, Y, Mu, G. Stock, and P. Hamm, Proc. Natl. Acad. Sci., Vol. 98, 11254,2001. 3 S. Woutersen, R. Pfister, P. Hamm, Y. Mu, D. Kosov, and G. Stock, J. Chem. Phys.,Vol. 117,6833. 4 V.Volkov, and P. Hamm, Biophys. J. to be published. 5 Torii, H., and M. Tasumi, J. Raman. Spectrosc, Vol. 29, 81, 1998. 6 Grossfield, A., and T.B. Woolf, Langmuir, Vol. 18, 198, 2002.

639

Engineering cost function for optimizing coherent control between processes with different nonlinearities Jianfang Chen^'^, Hiroyuki Kawano^, Yasuo Nabekawa^ Hideaki Mizuno^, Atsushi Miyawaki^, Takasumi Tanabe"^, Fumihiko Kannari"^, and Katsumi Midorikawa^'^ ^ Laser Technology Laboratory, RIKEN, 2-1 Hirosawa, Wako, Saitama 351-0198, Japan E-mail: [email protected] ^ Graduate School of Science and Engineering, Saitama University, Shimo-Okubo 255, Saitama City, Saitama 338-8570, Japan ^ Laboratory for Cell Function Dynamics, Brain Science Institute, RIKEN, 2-1 Hirosawa, Wako, Saitama 351-0198, Japan ^ Keio University, 3-14-1 Hiyoshi Kouhoku-ku, Yokohama 223-8522, Japan Abstract. It has been known that three-photon absorption of DNA or protein is one of the causes of phototoxicity. Here, we apply coherent control to enhance the contrast between the two-photon absorption of fluorescent label and the three-photon absorption of biomaterial with engineered cost function, leading to effectively reduced three-photon absorption with few loss of valuable two-photon fluorescence.

1.

Introduction

Using fs laser as light source to excite fluorescent labels, both tw^o-photon (2P) and three-photon (3P) fluorescence microscopies have been demonstrated in 3D imaging of biosamples [1]. However, the ultrahigh peak intensity of fs laser pulse also induces many side effects, like photobleaching and phototoxicity, w^hich deform, dissociate, and ionize either fluorescent labels or biosamples. By use of coherent control technique, optimal pulse shapes for reducing the photobleaching of green fluorescence protein (GFP) and its variants have been achieved recently, resulting in fourfold-enhanced fluorescence lifetime [2]. In this work, we attempt to reduce the phototoxicity by enhancing the ratio between the 2P fluorescence from the label and the 3P fluorescence from the biosample using coherent control with engineered cost frinctions.

2.

Experimental

We used a Ti:Sapphire oscillator (Femtosource, Femtolasers Produktions GmbH.) as the excitation source, which generates laser pulses with a central wavelength of 785rmi, a pulse duration of 12fs, a repetition rate of 75MHz, and an average power of 400mW. The spectral phase of the laser pulse was modulated using a liquid crystal spatial light modulator (SLM-128, Cambridge Research & Instrumentation, Inc.). The shaped pulses at an average power of 90mW were then directed into a

640

long working distance objective lens (40x, OLYMPUS) with numerical aperture (NA) of 0.6, and focused into a quartz cuvette that contained a mixed solution of enhanced green fluorescent protein (EGFP, 0.15mg/mL) as 2P fluorescent label and L-Tryptophan (55mM) as 3P fluorescence biosample in Hepes (N-2hydroxyethylpiperazine-N'-2-ethanesulfonic acid) NaOH buffer (20mM, pH 7.4). 2P fluorescence (peak wavelength, 510nm) and 3P fluorescence (peak wavelength, 350nm) signals were then detected by two photomultiplier tubes (PMT, H6780-06, Hamamatsu) separately installed on the two sides of the cuvette. The fluorescence analog signals were converted to digital signals using a digital oscilloscope (LeCroy 9350A) and transferred to a PC for generating a cost function in the feedback coherent control.

3.

Results

First, we simply set the cost function with a form C=F2p/Fsp, where F2P and FSP are the 2P and 3P fluorescence signals. The experiment was started from transformlimited pulse shape, with initial value of the cost function adjusted to 1. After a self-learning based on simulated annealing algorithm, the value of the cost function increased to 27, while F2P was reduced by 4.8 times. The significant reduction of F2P may be an important issue for practical application. Thus, we designed a new group of cost fiinctions with a form C=(F2pf/Fsp, where X is a new control parameter. We changed x from 1 to 10, and found that the loss of 2P fluorescence could be mitigated by increasing the value of x. For example, when we chose x=1.5, the intensity of 2P fluorescence signal was reduced by 4 times, and the 3P fluorescence signal was reduced by about 40 times. The temporal profile of the shaped pulse for jc=1.5 is shown in Fig. 1(a). Roughly speaking, the pulse was split into a train of sub-pulses and the pulse width was significantly broadened, leading to enhanced fluorescence ratio. However, when we chose jc=6, the intensity of 2P fluorescence was almost unchanged but the intensity of 3P fluorescence was almost halved. The temporal profile is shown in Fig. 1(b), which exhibits a near-transform-limited pulse with broadened rising edge.

1.0-

(a) • 'Uy 0.8 -

n

<

0 6-

%

0-4-

—

0.2-

0.0-

y,: 1i . f/CI Time(fs)

Time(fs)

Fig. 1, Second-harmonic-generation frequency-resolved optical grating (SHG-FROG) constructed temporal profiles of shaped pulses for (a) jc=1.5 and (b) x=6.

641

Figure 2. shows the ratio between the 2P and 3P fluorescence signals as well as the intensity of the 2P fluorescence as functions of x. It shows that the 2? fluorescence signal ascends as the value of x increases, in contrast, the ratio of 2P and 3? signals descends with increased x. After x exceeds about 4.5, the intensity of 2P fluorescence could be maintained at almost its initial level. When we put the value larger than 6 into x, we could not find significant change of F2P and C=F2p/Fsp from those obtained with the transform-limited pulse.

Fig. 2. Ratio between the 2? and 3P fluorescence signals (dashed Hne) and the 2P fluorescence signal itself (solid line) as a function of x. The fluorescence signals F2P and Fsp defined in the text are normalized by those excited by transform-limited pulses and notified with df(GFP) and df(Tryptophan), in this figure, respectively.

4.

Discussion and Conclusions

For the coherent control with low x values, since 3P process is more strongly dependent on the peak intensity than 2P process, the high contrast between 2P and 3P fluorescence always can be achieved by lowering the peak intensity of the pulses, or in our case, broadening the pulse width by regrouping the phases of the frequency components. This strategy would cause over-reduced 2P excitation efficiency which could be an issue for practical applications. However, if high x values are used, it can be regarded as that the nonliearity of the 2P process is "artificially" increased by jc times, so that the over-reduction of 2P fluorescence can be suppressed. By properly choosing an x value, a trade-off between the fluorescence contrast and the fluorescence efficiency can be achieved.

References 1 W. W. Webb, Science 248, 73, 1990. 2 H. Kawano, Y. Nabekawa, A. Suda, Y. Oishi, H. Mizuno, A. Miyawaki, K. Midorikawa, Biophys. Biochem. Res. Commun. 311, 592, 2003.

642

Part IX

Ultrafast Nanostructure Photonics and Plasmon

Imaging of localized silver plasmon dynamics with sub-fs time and nano-meter spatial resolution Atsushi Kubo \ Ken Onda \ and Hrvoje Petek ^ Zhijun Sun ^, Yun Suk Jung ^, and Hong Koo Kim^ ' Department of Physics and Astronomy and Institute of NanoScience and Engineering, University of Pittsburgh, Pittsburgh, PA 15260, USA E-mail: [email protected] ^ Department of Electrical Engineering and Institute of NanoScience and Engineering, University of Pittsburgh, Pittsburgh, PA 15261, USA Abstract. Images of local plasmons on a silver grating excited by 400-nm, 10-fs pulses are obtained with 50-as time and 50-nm spatial resolution. The technique is based on interferometric time-resolved two-photon photoemission and photoelectron emission microscopy.

1.

Introduction

Light interacting with nano-structured metals or rough metal surfaces can interact with free electrons to excite collective density excitations known as surface and bulk plasmons. Recently, surface plasmons (SPs) have attracted much interest in subwavelength optics where they facilitate applications in data storage, light generation and filtering, microscopy, and bio-photonics [1]. Frequencies and dynamical properties of SPs are determined by the dielectric properties of metal interfaces, the shapes of metal structures, and the inter-structure coupling. SPs are confined to surface structures with dimensions of subwavelength order [2,3], and the dynamical energy transfer between the material polarization and propagating electromagnetic fields takes place at the surface plasma frequency. We combine the interferometric time-resolved two-photon photoemission (TR2PP) technique [4] with the photoelectron emission microscopy (PEEM) [5] to image the spatio-temporal dynamics of localized surface plasmons. Employing this time-resolved PEEM (TR-PEEM) method, we generate movies and measure interferometric two-pulse correlations of individual plasmon resonances with <50as time and <50-nm spatial resolution.

2.

Experimental Methods

The sample investigated is a silver grating (nanophotonic structure) with a period of 780-nm and film thickness of 400-nm [6]. One-dimensional (ID) array of mesa structures is formed on a quartz substrate by holographic lithography, and

645

silver grating with 100-nm slit width is formed by off-normal evaporation of Ag onto the patterned substrate. The experimental setup for TR-PEEM is as follows: 400-nm light (pulse duration 10-fs, average power 90-mW, repetition frequency 90-MHz) is generated by frequency doubling a Ti:sapphire laser oscillator in an 80-jLim thick BBO crystal. After compensation for the group velocity dispersion and adjustment of the polarization, the 400-nm light is focused onto a 200-jLim spot on the sample at the image plane of a PEEM microscope contained in a UHV chamber. The incidence angle is 65° from the surface normal and the optical plane is perpendicular to the grating. Two-photon absorption of 400-nm light (hv= 3.1 eV) excites electrons above the work function of silver (~4.3 eV), and the photoemitted electrons are imaged with the PEEM electron optics with <50-nm resolution. One-photon photoemission IPP-PEEM images are taken with a conventional mercury lamp (k = 254 nm, hv= 4.9 eV) to contrast with the twophoton photoemission 2PP-PEEM images. To image the dynamics of the localized SP, TR-PEEM images of the grating structure are recorded with phase-locked pump-probe pulse pair (PPP) excitation. PPP are generated by splitting the 400-nm pulses with a Mach-Zehnder interferometer (MZI) to generate identical pulse-pairs with a time delay (Zd) controlled to an accuracy of <50-as [4]. The images are advanced with 1/4-cycle or 0.33 fs time intervals.

Results and Discussion Figure 1(a) shows a topographic image of the sample obtained by a scanning electron microscope (SEM). The grating structure corresponding to the SEM image can be clearly discerned in the IPP-PEEM image of Fig. 1(b) taken with the Hg lamp. The 254-nm light exceeds the energies of plasmon resonances in Ag, thus the IPPE-PEEM image in Fig. 1(b) reflects the probability of exciting singleparticle excitations in the Ag array. By contrast, the 2PP-PEEM image excited with the p-polarized (TM) ~400-nm fs-laser light in Fig. 1(c) shows much richer structure: the 2PP-PEEM image is dominated by grainy structure (hot spots) formed on top of the ID array. The intensity ratio of the brightest hot spots to other parts of the array is about 20. A strong dependence of the photoelectron emission on the laser polarization is observed with the yield enhanced by >50 for the j:?-polarization as compared with ^-polarization (TE). The energy of surface plasmons on a flat silver surface is 3.63-eV, but this

I Fig. 1. Micrograms of the silver grating: (a) SEM, (b) PEEM with 254-nm light excitation (IPP-PEEM), (c) 2PP-PEEM of the same area as (b) with 400-nm fslaser excitation. The grating period is 780-nm. 646

J230 nm -1/4 15 (19.62fs)

+1/4 +1/2 +3/4 16 P h a s e d e l a y (1>{X2%)

+1/4

+1/2 (21.95fs)

Fig. 2. TR-PEEM micrograms of a hot spot on the silver grating (indicated by the rectangle in Fig. 1(c)) recorded at a rate of 0.33-fs/frame. The range of the pump-probe delay times, Td, is 19.62 - 21.95-fs corresponding to phase delay ^of 14%-2;r-> 16)^-2;r mode cannot be excited by light propagating from vacuum, because momentum conservation is not satisfied. Light can couple to the localized or propagating SP modes, respectively, on surfaces with nanometer scale roughness or periodic shape modulation. The resonant excitation of these modes extends into the visible spectrum depending on the feature size and morphology of the Ag film. The existence of strong hot spots in Fig. 1(c) suggests that the local SP modes are excited at random surface roughness features on <10-nm scale that is visible in Fig. 1(a) [7,8]. The coupling of the 400-imi light at an incident angle of 65^ to the propagating SP modes requires a grating period of 1,950-nm. Fourier transforms of the PEEM images (not shown) suggest that the images are dominated by excitation of the localized SP modes. Shalaev and Stockman and coworkers [2,3,7] have predicted the localization of light in such nanostructures leading to enhancement of non-linear optical processes such as 2PP. The SP resonance frequency in random nanostructured films depends on the local morphology and coupling among individual particles. Thus, the 2PP-PEEM images map out the distribution of SP polarization fields excited in the sample. Figure 2 shows a detail of a series of TR-PEEM images taken with Zd in the range between 19.62 - 21.95-fs corresponding to phase delay (j) of IAYA-IK ^>\6y2'2n:. The images are advanced with a time interval of 0.33 fs corresponding to 1/4-cycle of the carrier wave. In this interval, the phasedependent change in the intensity of the particle-pair (dipole) is summarized as follows: (a) For T^= In- K (in-phase excitation), both spots are intense; (b) for z^ = {In^X)' n (Tc-out-of-phase excitation) both spots are weak; (c) for r^ = (2f7 + X) • ^ the upper spot is intense; and (d) for z^ = (2w -Y-i)- n the lower spot is intense. The SP oscillations in the dipole can be understood by considering the interparticle coupling and the delay-dependent Fourier spectra of the excitation PPP. Considering a typical SP propagation length of 850-nm [9], and the spatial separation of the hot spots in Fig. 2 of 230 nm, we can expect that the particlelocalized modes couple to hybridize into a pair of symmetric and anti-symmetric excitations delocalized on the particle pair [10]. Because of the hybridization, the symmetric eigenmode of the pair will be stabilized and the anti-symmetric mode will be destabilized by an equal amount with respect to the parent modes that are localized on each particle. Excitation fields can be selected to excite exclusively one or the other eigenmode, or to create a coherent superposition of the eigenmodes that can localize the excitation in one or the other particle. 647

I

I

I

L_J

'

'

'

'

L-J

•

'

'

•

'

'

I

I

L—I

I

L_

Delay: 15-25 X27t

'-'•^

I

0

I

m

I

r~i

10

I m

i

i i

i i i i i

20 30 Time delay [fs]

T

40

50

Fig. 3. Microscopic I2PC spectrum from an individual hot spot on the sample (indicated by the circle in Fig. 1(c)). The anti-phase oscillations in the dipole intensity in Fig. 2 are indicative of creation of a coherent superposition of eigenmodes of the system. The Fourier transforms (spectra) of the PPP exhibit modulation (Ramsay fringes) due to the constructive and destructive interference at frequencies where the delay r^ corresponds to 2n • TT or (2n + 1) • ;T optical cycles, respectively [11]. Therefore, if the frequencies of both the eigenmodes were to be excited by the field of the pulse pair, there will be a delay where the spectral intensity will be maximum for one mode and zero for the other. At this delay only, the pump pulse will excite both modes, but the probe pulse will interfere constructively at one eigenfrequency and destructively at the other, leaving the system in a pure eigenmode. At other delays, the PPP excitation will prepare a superposition of the eigenmodes, which with the appropriate phase and amplitude can localize the excitation on either particle. The delay dependent spatial oscillation of the intensity in Fig. 2 is indicative of preparation of such coherent superposition states, and constitutes a demonstration of quantum control of SP fields in a photonic nanostructure. We can study the polarization dynamics more quantitatively by recording interferometric two-pulse correlation (12PC) measurements for individual particles. I2PCs from different microscopic regions provide information on the resonant frequency and dephasing of localized plasmons. Figure 3 shows an I2PC scan from a single isolated hot spot on the sample (circled in Fig. 1(c)). The PEEM signal intensity from the hot spot, which is proportional to the photoyield, is recorded as a frmction of the pump-prove delay Td. Each hot spot is driven at the carrier frequency, but after the excitation is over, it oscillates at its own resonance frequency, which may be different. This leads to a beating pattern that is observed for delays corresponding to 1 5 - 2 5 cycles, which arises from the interference between the pump and probe induced polarizations excited in the particle. Such beating patterns are specific for each hot spot because each has its own resonant frequency as well as couplings to the environment, which lead to the dephasing. In case of the dipole in Fig. 2, the coupling is mainly between two

648

particles; other particles may interact more weakly with several neighboring particles, as well as with the propagating modes of the photonic structure. Modeling of the I2PC for the hot spot gives a dephasing (T2) and population decay time (Ti) of about 6-fs and 65-fs, respectively. The plasmons decay into single particle excitations (hot electrons), which are responsible for the slow population decay. Quantitative analysis of I2PC scans will yield the resonant frequencies and dephasing times for the individual plasmon excitations.

4.

Conclusions

We have demonstrated the two-dimensional imaging and quantum control of localized silver plasmon dynamics with <50-as time and <50-nm spatial resolution using I2PC and TR-PEEM technique. Coupled plasmon resonances show antiphase oscillation at the pump-probe delay of ~20-fs (~15*271), which we attribute to the interparticle coupling and interference between pump and probe induced polarization waves. By synthesis of specific excitation fields, we can control the SP field distribution excited in the sample. The TR-PEEM technique will allow us to make quantitative measurements of localized and propagating plasmon excitations metallic nanostructure arrays. Acknowledgements. The authors thank A. P. Heberle for discussions. This research has been supported by the NSF, the University of Pittsburgh. A.K. acknowledges fellowship support from the Yamada Science Foundation of Japan.

References 1 2 3 4 5 6 7 8 9

W. L. Barnes, A. Dereux and T. W. Ebbesen, Nature 424, 824, 2003. S. Gresillon, et aL, Phys. Rev. Lett. 82, 4520, 1999 M. I. Stockman, et aL, Phys. Rev. Lett. 92, 057402, 2004 H. Petek, and S. Ogawa, Prof Surf Sci. 56, 239, 1997 H. H. Rotermund, Surf Sci. Rep. 29, 265, 1997. Z. Sun, Y. S. Jung, and H. K. Kim, Appl. Phys. Lett. 83, 3021, 2003. V. M. Shalaev, et aL, Phys. Rev. B 53, 11193, 1996. P. Monchicourt, et aL, J. Phys. Condens. Matter 9 5765, 1997 The propagation length of the surface plasmon, ^p, is inversely proportional to the imaginary part (k^p ) of the complex SP wave vector, k^p = kj + \kj', which is us: 1 1 given by, Ssp= - = — 2ks;

,' /sj

where ko is the free space wave

^0 ^

number, £v the dielectric function of vacuum (unity), and s^ and s^ the real and imaginary parts of the dielectric function of the silver. Using this equation, SSP of silver surface plasmon at the silver-vacuum interface is estimated to be about 850nm. 10 P. Nordlander, et aL, Nano Lett. 4, 899, 2004 11 H. Petek, et aL, Phys. Rev. Lett. 79, 4649, 1997

649

Ultrafast Dynamics of Light Transmission Through Plasmonic Crystals Claus Ropers \ Roland Muller\ Christoph Lienau\ Gero Stibenz\ Giinter Steinmeyer\ Doo-Jae Park^, Yeo-Chan Yoon^, and Dai-Sik Kim^ ^ Max-Bom-Institut, Max-Bom-Str. 2A, 12489 Berlin, Germany E-mail: [email protected] ^ School of Physics, Seoul National University, Seoul 151-742, Korea E-mail: [email protected] Abstract. We study with 10-fs light pulses the ultrafast dynamics of light transmission through plasmonic 2D- and ID-nanocrystals. Two interfering transmission channels are identified: ultrafast non-resonant transmission due to photon tunneling, and time-delayed resonant re-emission of surface plasmons.

1.

Introduction

The linear and nonlinear optical properties of plasmonic nanostructures currently attract much attention, as such structures are promising candidates for novel applications in nano-lensing [1], perfect lensing [2], field localization, nanoscale w^ave-guiding [3], nano-lasing [4], and ultrafast switching. A prominent example for the unusual optical effects associated with plasmonic nanocrystals is the enhanced transmission of light through periodic nano-hole arrays in metal films [5]. It is now well understood that the optical properties of such nanostructures are governed by short-lived surface plasmon polariton (SPP) excitations with life times in the 3 - 100 fs range. Yet, the ultrafast dynamics of these excitations have largely remained elusive. Previous studies of ultrafast pulse propagation through nano-hole arrays reported a 10-fs delay in transmission [6,7]. The physical interpretation of this delay, assigned to either the finite transit time through the nanoholes [6] or to the SPP lifetime [7], however, remains controversial. Here we report the first experimental study of ultrafast light propagation through plasmonic nano-crystals using light pulses much shorter than the SPP damping time. Phase-resolved measurements of the time structure of the transmitted light allow to clearly distinguish two different contributions to enhanced transmission: non-resonant tunneling and SPP re-radiation. We demonstrate that the optical spectra of plasmonic crystals are governed by Fanolike interferences between these channels.

2.

Experimental methods

We investigate 150-300 nm thick gold films deposited onto a sapphire substrate and perforated with periodic square arrays of holes with a radius of 125 nm and

650

periods of 700 and 800 nm [7], or with a linear array of 50 nm wide nano-slits with periods of 600 and 700 nm. The samples are illuminated at an angle 9 with weakly collimated linearly polarized 10-fs light pulses from a Ti:sapphire oscillator (Fig. 1(a)). The time structure of the transmitted light is studied using either interferometric autocorrelation (lAC) or spectral interferometry (SI).

3.

Results and Discussion

In Fig. 1, autocorrelation traces of the incident light (b) and the light transmitted through a linear slit array (600 nm periodicity, 50 nm slit width) at two different angles are displayed (c,d). Linear transmission spectra are shown in the insets. In Fig. 1(c), the laser spectrum overlaps only weakly with SPP resonances. The transmitted light consists of a strong initial peak due to non-resonant transmission through the slits and a second long-lived but weak contribution due to resonant excitation and re-radiation of SPP at the sapphire/metal (SM) interface [6,7].

I

' I

'

I

80 -60 -40 -20

20

0

40

60

80

At [fs]

I

'

I

'

I

•

-80 -60 -40 -20

I

0

'

i

20

At [fs]

'

I

40

I

'

I

'

I

60 -40 -20

'

I

0

'

I

20

'

I

40

•

I

60

'

I

80

At [fs]

Fig. 1. (a) Schematic of the experimental setup, (b) Interferometric autocorrelation (lAC) of the incident 10-fs light pulse. (c,d) lAC of the light transmitted through a 600-nm period array of 50-nm wide nano-slits in a 150 nm thick gold film at incidence angles of 0 = 28° (c) and 35° (d). The laser spectrum (b) and linear transmission spectra (c,d) are shown in the insets. When the overlap between SPP resonances and laser spectrum is optimized by angle tuning (Fig. 1(d)), the SPP contribution is strongly enhanced, and the time profile of the transmitted light is dominated by the pronounced polarization interference between two SPP resonances persisting for more than 80 fs [8]. The lAC measurements show directly that the time structure of the transmitted field

651

E^ = E^^ + E^pp is given as a superposition between the short burst due to nonresonant transmission through the nanosHts E^^ and the weakly damped emission from different SPP resonances E^pp =^ E^-exp[-io)J-yj). The SPP dispersion relation for a flat metal film predicts SPP resonance frequencies ^n AM ^^^ ^n SM ^^^ ^^^ air-metal and sapphire-metal interface, respectively. In the spectral domain this gives rise to asymmetric Fano-like lineshapes l{co) =

(-)-Z.

(1)

as recently proposed [8,9]. Similar experimental results as those shown in Fig. 1 are also obtained for two-dimensional nanohole arrays.

Fig. 2. Spectral intensity (a) and spectral phase (b) of the light transmitted through a 600nm period array of 50-nm wide nano-slits in a gold film. The phase data are obtained by spectral interferometry with 10-fs light pulses as a function of incidence angle 9. Different SPP transmission resonances on the air/metal (AM) and the sapphire/metal (SM) interfaces are resolved, showing distinctly different phase signatures. The interferometric autocorrelation measurements alone do not allow for a quantitative analysis of the different interfering contributions or for a study of the effects of coherent coupling between SPP resonances and the formation of bandgaps in these plasmonic nanocrystals. We thus use spectral interferometry to directly measure the phase of the transmitted field by interfering it with a replica of the 10-fs incident laser. Results of the angle-dependent measurements are summarized in Fig. 2 for the 600-nm period slit array. Different SPP resonances at either the air/metal (AM) or sapphire/metal (SM) interface are clearly resolved. The resonances are each characterized by an asymmetric line shape (Fig. 3(a)) and

652

a pronounced phase variation near the resonance (Fig. 3(b)). The phase signatures for AM and SM resonances are distinctly different. The Hneshape model introduced above allows for a quantitative description of the amplitude and phase variations. ^

b)

a)

13 CD

03

>^

i\

c

2 0-

»>W

E o c

2

0) (0

p«ininfflirtr,

1.5

1

1.6 E(eV)

,

1

1.7

"^

1

1.5

*"^ '

1

1.6 E(eV)

'

1

1.

Fig. 3. Experimental transmission spectrum (a) and spectral phase (b) of the Hght transmitted through a 700-nm period array of 50-nm wide nano-sHts in a gold film (open circles) and comparison to the lineshape model introduced in the text (solid lines). The asymmetric shape of the transmission spectrum arises from the Fano interference between non-resonant transmission and SPP re-radiation. Important insight into the physics of enhanced light transmission through metal films perforated with periodic nanohole arrays is gained from this analysis. First, our results give clear evidence for two distinct transmission channels. A fraction of the incident light directly tunnels through the nanoholes and is rescattered into the far-field. Second, SPP modes can be excited at both the AM and the SM interface by the grating coupling effect at specific resonance energies. The lifetime of these SPP modes is much longer than the non-resonant tunneling time and can exceed 100 fs. It is limited by a Rayleigh-like scattering of these evanescent modes into progating far-field radiation, with the nanoholes acting as scattering centers [7]. The linear far-field transmission spectra result from the interference of these two transmission channels, and thus SPP excitation may both enhance and decrease the far-field transmission, depending on the phase delay between the two contributions. A generalized Fano model quantitatively describes the far-field transmission spectra. The model allows to derive the dispersion relation of this plasmonic crystal with the energy positions o)^ and the damping rates y^ of the SPP resonances. Direct evidence for the formation of plasmonic bandgaps comes from the observation of a 40-meV splitting at the crossing of the SM(1) and SM(2) resonances. An even more striking consequence of the coherent coupling between different SPP eigenmodes is the pronounced variation of the damping rates ;r„ near the crossing region. The coupling leads to the formation of new SPP eigenmodes having different spatial overlap with the nano-slits. As the nano-slits

653

variation of the spatial mode structure gives rise to a change in the radiative damping of the SPP excitation and thus of y^. This effect, which is analogous to the Dicke sub-/superradiance in coupled atomic systems, is directly observed from the reduction in line width and amplitude of the SM(1) resonance near the crossing region (Fig. 2). At certain angles, we observe ultranarrow transmission resonances with linewidths corresponding to SPP lifetimes of over 150 fs, which is more than 15 times longer than reported in previous time-resolved transmission experiments [6,7]. Spatially resolved near-field SPP spectra, reported elsewhere, give evidence for the different symmetry of the sub- and superradiant SPP modes. In summary, we have presented the first experimental study of ultrafast light propagation through plasmonic nanocrystals with a time resolution beyond the damping times of the relevant surface plasmon polaritons. The dynamics of two distinct transmission channels - non-resonant tunneling and SPP re-radiation - are directly resolved, and it is shown that the linear transmission spectra of such nanostructures are governed by interferences between both contributions. Coherent couplings between different SPP resonances, Le,, formation of plasmonic bandgaps and sub-/superradiant damping are quantitatively probed. We believe that similar studies are highly relevant for a detailed microscopic understanding of the linear and nonlinear optical properties of plasmonic and photonic crystals. Acknowledgements. We gratefully acknowledge financial support of the work in Germany by the DFG (SFB296) and the European Union (SQID) and that in Korea by MOST (NRL and Nano-Photonics program) and KOSEF (SRC program).

References 1 H. J. Lezec, A. Degion, E. Devaux, R.A. Linke, L. Martin-Moreno,F. J. GarciaVidal, T. W. Ebbesen, in Science, Vol. 297, 820,2002. 2 J, B. Pendry, in Physical Review Letters, Vol. 85, 3966,2000. 3 S. A. Maier, P. G. Kik, H. A. Atwater, S. Meltzer, E. Habel, B. E. Koel, A. A. G Requicha,'mNature Material Vol. 2,229, 2003. 4 D. J. Bergman and M, I. Stockman, in Physical Review Letters, Vol. 90, 027402 2003. 5 T. W. Ebbesen, H. J. Lezec, H. F. Ghaemi, T. Thio, P. A. Wolff, in Nature, Vol. 391,667,1998. 6 A. Dogariu, T. Thio, L. J. Wang, T. W. Ebbesen, H. J. Lezec, in Optics Letters, Vol. 26,450,2001. 7 D. S. Kim, S. C. Hohng, V. Malyarchuk, Y. C. Yoon, Y. H. Ahn, K. J. Yee, J. W. Park, J. Kim, Q. H. Park, and C. Lienau, in Physical Review Letters, Vol. 91, 143901,2003. 8 R. Mueller, V. Malyarchuk, and C. Lienau, in Physical Review B, Vol. 68, 205413, 2003. 9 C. Genet, M. P. van Exeter, and J. P. Woerdman, ia Optics Communications, Vol. 225,331,2003.

654

Ultrafast Near-Field Microscope Imaging of Electron and Phonon Relaxation in Single Gold Nanoparticle K. Imura, T. Nagahara, and H. Okamoto Institute for Molecular Science, Okazaki 444-8585, Japan, E-mail: [email protected] Abstract. We investigated near-field optical properties of single gold nanorods. Transmission images observed near the surface plasmon resonances agree with calculated local density of states. Ultrafast temporal responses of single rods have been imaged by combining near-field microscope with time-resolved technique. It has been found that dynamic behavior in the center of the rod is different from that in the both ends.

!•

Introduction

It is of fundamental importance to know how electron-electron (e-e) and electron-phonon (e-ph) scattering processes after photoexcitation depend upon size and shape of gold nanoparticles and how they proceed inside the particle [13]. Dynamical spectroscopy of single particles with high temporal and spatial resolution must be informative for this purpose. We have performed ultrafast near-field pump-probe imaging of single gold nanorods (NRs) to investigate dynamic behavior of energy dissipation processes.

2* Experimental Methods Gold NRs (diameter 15-40 nm, length 100-500 nm) were synthesized chemically in solutions and spin-coated on a cover-slip. An apertured (diameter ca. 50-100 nm) near-field microscope was used under ambient condition. A Ti:sapphire laser (X = 780 nm, <100 fs, 80 MHz) was used for time-resolved pump-probe measurements [4,5]. The pump and probe beams were mechanically chopped at different frequencies, and collinearly coupled to the other end of the near-field fiber probe after pre-compensation of group velocity dispersion due to the optical fiber. The signal of interest was recovered through phase-sensitive detection at the difference of the chopping frequencies.

3.

Results and Discussion

Shear-force topographic image of a single NR is shown in Fig. 1(a). In the near-field transmission spectrum, the NR showed two distinct extinction bands at

655

Fig. 1. (a) Topography of a gold nanorod (NR) (length: 180 nm, diameter: 30 nm). (b),(c) Observed transmission images of the NR measured around 530 and 780 nm, respectively. (d),(e) Calculated LDOS images of a gold NR corresponding to (b) and (c), respectively. Image size: 480 nm x 480 nm. ca. 530 and 850 nm. From polarization dependent measurements, these bands are assigned to transverse and longitudinal surface plasmon (SP) resonances, respectively. Figures l(b,c) shows experimentally obtained optical near-field images at these SP resonance frequencies. Bright parts correspond to reduction of transmitted light. To analyze the observed optical images of NR, we calculated electromagnetic local density of states (LDOS) around the NR following the Green dyadic method [6,7]. Figures l(d,e) are calculated LDOS images, where bright parts correspond to higher densities. Height and length of the NR were obtained from the topography image. In Figs. l(c,e), the images show a characteristic oscillating pattern and they agree qualitatively well with each other. The images observed are reflected by characters of specific resonant SP-mode eigenfunction. Now we observe temporal LDOS behavior of NRs by ultrafast measurements combined with near-field microscope. Near-infrared photons from the Ti:sapphire laser are used for perturbing the energy distribution of the conduction electrons, and also for probing the energy dissipation processes. The electron- and phononrelaxation processes can be observed by changing pump-probe delay time. The pump and probe fields are polarized along the long axis of the NR. Figure 2(b) shows time responses at positions A to E indicated in Fig. 2(a). Temporal response of central part of the NR is different from those of the both ends. Bright and dark parts correspond to bleached absorption and induced absorption, respectively. A fast rise and a slow decay are seen for positions A, C, D and E, while the slower decay component is hardly recognized at position B. These temporal responses can be fit to a single or double exponential function. The time constant for the faster component was 0.6±0.1 ps and the slower one (2.8-1.5)±0.3 ps (2.8 ps for positions A and C, L8 ps for D, and 1.5 ps for E). The 0.6-ps decay at position B as well as positions A, C, D and E may be attributed to e-e process, since similar time response (0.5 ps) has been reported for a 20-nm thick gold film [9]. The slower lifetime is close to those for e-ph relaxation obtained by ensemble measurements [3] and is assigned to e-ph process. Position dependent e-ph time-responses (positions C-E) near the end part of NR may be partly related to the different excitation efficiency, because higher electronic temperature gives longer electron-phonon relaxation time [3,8]. However, it is not obvious if this effect sufficiently explains the whole observations. Other effects, such as inelastic electron-surface scattering, may contribute. It has been claimed, however, such a process plays only a minor role for gold spheres [3,8]. Further discussion is needed on this point.

656

KO)

C CO

--g 2-

^D 1 •

•

a>

-

1:

_

=

====»===

1—-"V"

0

2

4

6

8

10

12

14

16

Delay time / ps Fig. 2. (a) A transient transmission image of a gold NR taken at 0.4-ps pump-probe delay time. Image size: 600 nm x 600 nm. (b) Transient transmission changes as functions of delay time. Positions A-E are specified in (a). Solid curves indicate double (for positions A, C, D, and E) and single (for position B) exponential fits, respectively.

4.

Conclusions

We report the static and time-resolved spatial characteristics of LDOS in single gold NRs. Static images show characteristic spatial features of SP modes and are in good agreement with the calculated LDOS. Temporal responses in the NR show apparent position dependence. The e-ph relaxation becomes faster toward the end edge of the NR. The present observation demonstrates that local excitation can spatially as well as temporarily modify the SP state of NR. Acknowledgements. This study has been partially supported by Grants in Aid for Scientific Research (16350015, 16750017) from Japan Society for the Promotion of Science.

References 1 A. Arbouet, C. Voisin, D. Christifilos, P. Langot, N. Del Fatti, F. Vallee, J. Lerme, G. Celep, E. Cottancin, M. Gaudry, M. Pellarin, M. Broyer, M. Maillard, M.P. Pileni, M. Treguer, Phys. Rev. Lett. 90, 177401 (2003). 2 M. Hu and G.V. Hartland, J. Phys. Chem. B, 106, 7029-7033 (2002). 3 S. Link and M.A. El-Sayed, L Phys. Chem. B, 103, 8410-8426 (1999). 4 T. Nagahara, K. Imura, H. Okamoto, Chem. Phys. Lett. 381, 368 (2003). 5 K. Imura, T. Nagahara, H. Okamoto, J. Phys. Chem. B. 108 (2004) in press. 6 C. Girard and A. Dereux, Rep. Prog. Phys. 59, 657 (1996). 7 J.-J. Greffet and R. Carminati, Prog. Sur. Sci. 56, 133 (1997). 8 J.H. Hodak, A. Henglein, G.V. Hartland, J. Phys. Chem. B. 104, 9954 (2000). 9 C.K. Sun, F. Valee, L.H. Acioli, E.P. Ippen, J.G. Fujimoto, Phys. Rev. B 50, 15337 (1994).

657

Plasmon enhanced ultrafast optical transmission in metallic nano-arrays A. Dechant and A. Y. Elezzabi Ultrafast Photonics and Nano-Optics Laboratory, Department of Electrical and Computer Engineering, University of Alberta, Edmonton, Alberta, Canada T6G 2V4 Abstract: Periodic, sub-wavelength silver slit arrays have a significant effect on the propagation of ultrashort optical pulses. Due to surface plasmon effects, enhanced transmission, significant pulse train re-radiation, and super-luminal lightfloware observed.

1.

Introduction

Recently, several types of nano-structures have been shown to have both a significant transmission enhancement [1,2,3], as well as a noticeable effect on the temporal characteristics of ultrashort pulses [4]. In these experiments, surface plasmons (SPs) are optically excited at a metal-dielectric interface [5], and by coupling to the opposite surface via nano-apertures, the SPs are eventually reemitted as far-field light. We demonstrate that coupling of few-cycle optical pulses with SPs results in several distinct re-radiating modes, and a modification of the group velocity of the incident pulse leading to, in some cases, superluminal pulse propagation.

2.

Methods

Using the finite difference time domain (FDTD) method, the spectral shaping of the incident ultrashort pulse, and the SP-stimulated optical enhancement are determined. Consideration of SP coupling in the metal nano-structure is accounted for via the Drude model S(0)) = 6^£^+

^ , icov-co

(1)

where co^ is the plasma frequency, v is the damping frequency, ar^^is the dc dielectric constant, and 6*, is the permittivity of free space [6]. Incorporation ofe((o) is achieved by employing the auxiliary differential equation formalism in which D = s(o))E is recast in the time-domain by Fourier transform techniques to yield a supplementary equation for the displacement, D, and electric field, E, vectors [7]. This equation is found to be dD

d^b

2 f;

dE

d^E

v - ^ + — — = coeoE + v€^€o-— + €^eo — r - .

at

dr

dt

,^. (2)

dr

Eq. 2, in combination with Maxwell's equations are evaluated for £ , D and H, respectively, using a central differencing scheme [8].

658

3. Results and Discussion The nano-structure consists of an array of 300 nm wide slits ingrained in a 300 nm thin silver film. Here, co^and v are taken to be 5.66x10^^ rad/s and 2.244x10^^ rad/s, respectively. SP waves are excited by a normally incident, p-polarized femtosecond pulse with a central wavelength of 800 nm corresponding to that of a Ti:Sapphire laser. The temporal evolution of the pulse can be seen in Figure 1. At t = 18 fs, coherent SP re-radiation from the nano-slit array is evident, whereas incoherent re-radiation is dominant at later times. The nano-array with a^ = 870 nm exhibits a transmission maximum of 90% at 2 = 800 greatly exceeding the predictions of classical diffraction theory.

|lil|:||li|^SR|l^^

Fig. 1. Evolution of a 10 fs pulse impingent upon a slit array with 660 nm period. Superluminal pulse velocity (Vg > c) is in evidence in Figure 2 where the peak pulse delay is charted as a function of the incident pulse width. Arrays with a, = 750 and a^ = 870 nm reveal a positive pulse delay for all incident pulse widths. Furthermore, as the pulse widths increase towards the continuous wave regime (>0.5 ps), the optical delays induced by each nano-array approach a steady state. However, for extremely short pulses, less than 20 fs, the transmitted pulse is actually advanced in time for the a^ = 660 nm array. Our results reveal that superluminal propagation does not violate the theory of relativity since the energy flow velocity of the pulse is less than or equal to the speed of light.

a,, = 870 nm

100

200

300

400

500

Pulse Width (fs)

Fig. 2. Pulse delay as a function of pulse width for the three nano-arrays.

659

0

200

400

600 0

200

400

600

800

0

20

40

60

80 100

Time (fs)

Fig. 3. Transmitted (black) and incident (grey) pulse envelopes for a„ = 660 nm. Figure 5 depicts the temporal evolution of three representative ultrashort pulses of durations 10, 50 and 100 fs for a^ = 660 nm. As expected, the 100 fs pulse is broadened to 111 fs. In addition, the peak of the pulse shows a transit time delay of 19 fs, corresponding to an effective group velocity of Vg = 0.45c. Similar pulse delays have previously been reported for metallic aperture arrays [9]. When pulses shorter than the SP's lifetime (-30 fs) are used to excite the metallic nano-array extraordinary pulse shaping is apparent. The 10 fs pulse breaks into two separate 5 fs high amplitude pulses followed by a low amplitude pulse train of 10 to 15 fs pulses. On time scales larger than the plasmon lifetime (>30 fs), the SPs are incoherently coupled, and their re-radiation pattern is due to collective interference as shown in Figure 5 for the 50 fs incident pulse.

4. Conclusion In conclusion, we have demonstrated the pulse reshaping characteristics of subwavelength metallic slit arrays. Geometrical factors play a large role in the reshaping process, as they affect the SP resonance, propagation and re-radiation conditions. Any change in the array's periodicity has a sizeable effect on both the transmission properties of the array as well as its dispersive characteristics, and in several cases superluminal propagation through these nano-arrays is observed.

References 1 2 3 4 5 6 7 8 9

660

T. W. Ebbesen, H. J. Lezec, H. F. Ghaemi, T. Thio and P. A. Wolff, Nature 391 667 (1998) H. F. Ghaemi, T. Thio, D. E. Grupp, T. W. Ebbesen, and H. J. Lezec, Phys. Rev. B 58 6779 (1998). Grupp D E, Lezec H J, Thio T and Ebbesen T W 1999 Adv. Mater. 11 860 A. Dogariu, T. Thio, L. J. Wang, T. W. Ebbesen, and H. J. Lezec, Opt. Lett. 26 450(2001). Raether H 1988 Surface Plasmons on Smooth and Rough Surfaces and on Gratings (Berlin: Springer) N. W. Ashcroft, and N. D. Mermin, Solid State Physics, (Saunders College Publishing, New York, 1976), pp 1-28. T. Kashiwa and I. Fukai, Microw. Opt. Techn. Let. 3 203 (1990). K. S. Yee, IEEE T. Antenn. Propag. 14 302 (1966). A. Dogariu, A. Nahata, R. A. Linke, L.J. Wang, and R. Trebino, Appl. Phys. BLasers O 74 s69 (2002).

Excitation and propagation of surface plasmon polaritons on metallic periodic structures G. Torosyan, C. Rau, B. Pradamtti, and R. Beigang Fachbereich Physik, Technische Universitat Kaiserslautem, Erwin Schrodinger Str., 67663 Kaiserslautem, Germany E-mail: [email protected] Abstract. Surface plasmon polaritons are excited on a periodic structure of metallic cylinders with ultrashort THz pulses. Propagation and damping of the surface plasmons as well as reemitted THz radiation is investigated experimentally.

1.

Introduction

Surface plasmons have been studied in thin metallic films extensively over the past years. Their properties with respect to parameters as the material constants and film thicknesses are well understood [1]. Over the past few decades, interest has grown in enhanced transmission through periodic metallic samples, such as hole arrays and deep metallic gratings (see e. g. [2-4]). In this contribution we present results for the excitation and propagation of SPPs in periodic structures of metallic cylinders. Both transmission and propagation properties of surface plasmon polaritons excited with ultrashort THz pulses are investigated for different periodic structures. Studying the resonant transmission and propagation processes at THz frequencies allows for ease of fabrication of samples with different periods and surface geometries.

2.

Experimental Methods

A standard THz time domain spectroscopic system was used for the experiments with an InAs surface emitter and a Silicon-on-sapphire photoconductive switch as the detector. The pulse length of the single cycle THz pulses was in the order of 1 ps. For measurements extending only over a small spectral range optical rectification in periodically poled lithium niobate (PPLN) [5] has been applied for narrowband generation. Excitation and propagation of SPPs were measured in the collimated part of the THz beam. A typical structure consists of a series of parallel metallic rods (length 60 mm) with diameters between 500 jim and 2.5 mm which are in close contact to each other. The aperture of the structure is 60x60 mm^. The THz pulse is polarized perpendicular to the metallic cylinders and the angle of incidence can be changed from -40° to +40°. The transmission of the THz pulse was measured behind the structure using time domain spectroscopy.

661

3. Results and Discussion Typical transmission spectra are shown in Fig. 1 (upper trace) for a periodic structure of 40 stainless steel cylinders with a diameter of d = 1.5 mm. A total transmission in the order of 10 % was observed with regularly spaced minima at frequencies corresponding to Wood's anomalies: v^ =m-—{l + sma) d where c,a and m are the speed of light, the angle of incidence and an integer number, respectively. Similar frequency spectra have been obtained for other diameters. At these frequencies where the diffracted beam is parallel to the surface of the cylindrical structure surface plasmon polaritons are excited. The k-vector of the SPPs is given by the sum of the parallel component of the THz wave vector and an integer multiple of the inverse grating vector '•kjfj^ sina±m'G ••• ^THz sin cir ± m • - d. The excited SPPs propagate perpendicular to the axis of the cylinders along the surface and are diffracted from the periodic structure. ?

'

'

1

Fig. 1: a) Transmission spectra through a one dimensional array of metallic cylinders.

Absorption

2 '1

3 /

^ 0 ''«-^*^

Li£_l

4

/ / \

5 Emission

b) Emission spectra from the same array of metallic cylinders measured 2 ps after the exciting broadband THz pulse.

1

0.4 0.8 Frequency V [THz]

The emission of the SPPs can be observed directly by measuring the transmitted THz pulse with a certain time delay after the exciting pulse. A typical emission spectrum with a time delay of 2 ps can be seen in the lower trace of Fig. 1 for SPPs at frequencies v^ with m = 1, 2, 3, 4 and 5. The positions of the maxima in emitted THz radiation correspond to the positions of minima in the transmitted radiation. Increasing the time delay results in a decrease of the amplitude of the SPPs due to their limited lifetime. In order to determine the propagation speed of the SPPs the periodic structure is excited on one end of the structure over a limited spatial area only. The emission is observed at the other end of the structure and the time delay between excitation and emission is measured. A typical result is shown in Fig. 2. In this case the excitation was done by a narrowband (Av = 100 GHz) THz pulse from optical rectification in PPLN in order to excite only one SPP frequency (third order m = 3). The propagation length was 15 mm (corresponding to 10 cylinders) and a

662

propagation time of -50 ps assuming a propagation speed close to the vacuum speed of light was expected. From the observed time delay (-65 ps) it is obvious that the SPPs do not propagate along the complete surface of the cylinders (this would correspond to a minimum time delay of approximately 78 ps). The effective path length is considerably shortened as there is already a strong coupling between 1

'

1

'

1

'

1

1

'

'

' • 1

AT -f2 Z>

<

-

0.5

•D

0 LL

O

0.0

4= 111

ix ^•ii*

-0.5

-1.0

1

1

L

t"—

THz in 1

(

20

1

1

il SI

^

THz out

999%

-

1

1

40

Fig. 2: Measurement of the propagation speed of surface plasmon polaritons on a periodic structure of metallic cylinders. The insert shows schematically the excitation and detection geometry.

60

1

•

80

Time Delay [ps]

the cylinders at distances within fractions of the wavelength. The effective path length depends on the distance between the cylinders and their shape.

4.

Conclusions

Surface plasmon polaritons were excited in a periodic structure of cylindrical rods with dimensions in the order of the wavelength. Excitation occurs at frequencies which correspond to the frequencies of Wood's anomalies identical to multiples of the inverse grating vector. Emission of THz radiation caused by the propagating surface plasmon polaritons was detected directly. The propagation of the SPPs along the surface was resolved in time and the velocity of propagation was determined.

References H. Raether, in Surface Plasmons on Smooth and Rough Surfaces and on Gratings, Springer Tracts in Modem Physics Vol. Ill (Springer-Verlag, Berlin, Heidelberg, 1988). T. Thio, H. F. Ghaemi, H. J. Lezec, P. A. Wolff, and T. W. Ebbesen, J. Opt. Soc. Am. B 16, 1743 (1999) H. F. Ghaemi, T. Thio, D. E. Grupp, T. W. Ebbesen, and H. J. Lezec, Phys. Rev. B 58, 6779 (1998) J. A. Porto, F. J. Garcia-Vidal, and J. B. Pendry, Phys. Rev. Lett. 83, 2845 (1999) C.Weiss, et al., Opt. Lett. 26, 563 (2001)

663

Ultrafast dynamics of periodic arrays of holes in a gold film V. Halted A. Benabbas\ L. Guidoni\ J.-Y. Bigot\ A. Degi^on^ H. J. Lezec^ T. W. Ebbesen^ P. N. Saeta^ ^IPCMS-GONLO ,23 rue du Loess, F-67034 Strasbourg FRANCE ^ISIS, Louis Pasteur University, 67000 Strasbourg FRANCE •^Department of Physics, Harvey Mudd College, Claremont, CA, US Abstract: We show that the enhanced transmission of gold nanostructures made of arrays of holes exhibits two kinds of resonances. Their spectral position and line-width strongly depend on the electron dynamics induced by femtosecond optical pulses.

1. Introduction Recently it has been demonstrated that a proper design of nanometric apertures in opaque metallic films results in an enhanced light transmission [1] much more efficient than the classical transmission through sub-wavelength apertures where non-propagating modes and important diffraction losses take place. Most of theoretical and experimental works that have investigated the mechanisms responsible for the enhanced transmission of the arrays of holes [2,3] are based on the resolution of Maxwell equations taking into account adequate boundary conditions. In addition, the excitation of plasmons is thought to play an important role as a mechanism to explain the enhanced linear transmission. In this paper, we focus on the different qualitative static and dynamical behaviors observed in two types of resonances present in gold nanostructures. One of them can be assigned to the individual response of the holes whereas the other kind results from the diffractive character of the array of holes. In both cases, plasmons may be involved with different degrees of localization. The observed dynamical optical response, induced by the absorption of an intense femtosecond pulse, is well understood in terms of the time-dependent dielectric function of the metal. 2. Static and dynamical studies of Au nanostructures We have studied periodic arrays of circular holes made by a Focused Ion Beam technique on a 300 nm thick gold film evaporated either on a sapphire or a glass substrate. Although the sub-wavelength size of the holes forbids light propagation in the spectral region of interest (X>500 nm), extraordinary peaks of transmission have been observed in these nanostructures [1]. The spectral position of theses peaks has been shown to depend on the geometrical design and their line-width on the thickness of the metallic films [4]. We found out that these parameters do not necessarily affect all the spectral features present in the transmission. Indeed, two kinds of resonances should be distinguished. In Figure 1., we show the linear transmission of an array with a periodicity of 360 nm and a hole diameter of

664

180 nm. The spectrum exhibits two main resonances Rl and R2 centered respectively at 527 nm and 670 nm. They have a different physical origin since, in contrast to the resonance R2, Rl is not red-shifted when the period of the array increases. Moreover, when the refractive index of the holes is modified, both peaks shift to the red as the index is increased. This behavior is seen in figure 1 where the two spectra have been obtained with and without immersion of the nanostructure in dioxane which has a refractive index n=1.46. This shift has been checked to be reversible as the nanostructure is taken out of the solvent and dried. The above spectral behavior of resonance Rl is similar to the one of gold nanoparticles embedded in a dielectric matrix. Such nanoparticles display a surface plasmon resonance near the interband transition (527 nm). The spectral position of this plasmon mode is well-known to be red-shifted for a larger refractive index of the matrix. Since, in addition, the resonance Rl is not affected by the design of the nanostructure, we attribute it to the individual response of the holes in the nanostructure. It may therefore be viewed as the optical response of an inverse nanoparticle (a dielectric cylindrical hole in a metallic matrix). Fig. 1. Linear transmission of a gold nano-array with a period P=360 nm and hole diameter d=180 nm on a glass substrate with (n=1.46) and without (n=l) immersion in dioxane

500

550

600

650

700

Wavelength (nm) The dynamical behavior of the nanostructure has been studied with an amplified femtosecond laser system centered at 800 nm and cadenced at 5 kHz. The differential spectra are measured by a pump-probe technique in a confocal configuration with a spatial resolution of ~1 jurn^. In Figure 2, we have represented the spectrally resolved dynamics of a nanostructure for an excitation density of the pump of ~1 mJ/cm^. The corresponding linear transmission is shown in the inset. The differential spectrum exhibits several interesting features. First, its shape is characteristic of a broadening of the resonance Rl. Secondly, its relative amplitude change is very large (-15 % for a pump-probe delay of 500 fs) in the spectral region 500-600 nm, that is near the interband transition of gold. In contrast, in the spectral region of R2, the shape of the differential spectrum is characteristic of a red shift. These results are a strong indication that the dynamics is dominated by the gold dielectric function B(>^,T). In the vicinity of interband transitions, it is mostly the imaginary part of 8(X.,T) which is modified via the electron-electron scattering. Correspondingly, it leads to a broadening of the resonance Rl. It is noticeable that this spectro-temporal dynamics is very similar to the one associated to surface plasmons in metallic clusters [5]. On the long wavelength side of the interband transitions it is the real part of E(X,T) which is predominantly affected. It manifests

665

as a red shift of the resonance R2. We have also studied the time-resolved differential transmission of the above nanostructure for different probe wavelengths by pre-selecting a narrow bandwidth in the continuum spectrum. Fig. 2 Spectrally resolved differential transmission of the array of holes with a period of 300 nm designed on a gold film deposited on a sapphire substrate, for a pump-probe delay T=0.5 ps

"T 500

550

600

\VaVelength (nrn) 650'

Wavelength (nm) To model the dynamics of the nanostructure, we have considered the resolution of Maxwell equations taking into account the appropriate boundary conditions as well as £(k,x) either described with a random phase approximation model or obtained from the measurements of the differential reflection and transmission of a thin gold film. This model reproduces the qualitative behavior of the dynamics of resonance R2 without including explicitly collective electronic excitations. Concerning the dynamics of resonance Rl, it cannot be reproduced by the above model since it results from the behavior of individual holes. A proper model would require to take into account the difficult problem of light scattering in a stratified medium with a complex dielectric function. In conclusion, the study of the static and dynamical optical response of periodic arrays of holes in metallic films shows that the spectrally enhanced transmission involves two mechanisms. One implicates the entire array via its diffractive character. The other one results from the individual response of the holes without influence of their spatial arrangement. In both cases, these resonances can be significantly modified, spectrally and temporally, with intense femtosecond pulses. 3. References [1] T. W. Ebbesen, H. J. Lezec , H. F. Ghaemi, T. Thio, P. A. Wolff, Nature 391, 667-669 (1998). [2] Ph. Lalanne, J. P. Hugonin, S; Astilean, M. Palamaru, K. D. Moller, J. Opt. A: Pure Appl. Opt. 2,48-51 (2000). [3] H. F. Ghaemi, T. Thio, D. E. Grupp, T. W. Ebbesen, H. J. Lezec, Phys. Rev. B. 58, 6779-6782 (1998). [4] A. Degiron, H. J. Lezec, W. L. Barnes, T. W. Ebbesen, Appl. Phys. Lett. 81, 4327-4329 (2002). [5] J.-Y. Bigot, V. Halte, J.-C. Merle, A. Daunois, Chem. Phys. 251, 181-203 (2000).

666

Surface plasmon assisted 26 fs, 0.4 keV electron pulse generation S. E. Irvine and A. Y. Elezzabi' ' Ultrafast Photonics and Nano-Optics Laboratory, Department of Electrical and Computer Engineering, University of Alberta, Edmonton. Alberta, Canada. T6G 2V4 E-mail: [email protected], [email protected] Abstract. We report on the generation of 0.4 keV. 26-fs electron pulses using laser pulses from a low-intensity, 80 MHz repetition rate. Ti:Sapphire oscillator. This work is complemented with particle simulations using thefinite-differencetime-domain method. A high spatial gradient can be created through the excitation of surface plasmon (SP) waves in a thin metal film, which results in a local electric field enhancement, and therefore, a larger ponderomotive potential for acceleration. Furthermore, by using the same metal film to eject electrons via multiphoton excitation, the generation and acceleration can be accomplished using the same optical pulse. This provides a simple, all-optical process for the generation and acceleration mechanisms. The synchronous generation/acceleration of the electron pulses is ideally suited for time-resolved optical pumping and electron beam probing studies [1,2]. Previous experiments [3,4] have demonstrated the potential for SP-assisted electron acceleration; however, the case where the SP's are excited on a time-scale below the plasmon lifetime (-30-50 fs) has not been studied. The focus of the current work is the impulsive excitation of SP waves for the generation and acceleration of electron pulses, where it is shown that energetic electrons, up to 0.4 keV, are produced using a low intensity, 80 MHz Ti:Sapphire oscillator. Furthermore, the experimental work is complimented with an ultrafast photoemission and acceleration simulation (UPAS). The UPAS model accounts for the electromagnetic field profile using a finite-difference time-domain (FDTD) method, as well as the intensity-dependent electron emission process of the metallic surface. SP waves are excited using 25 fs, 1.5 nJ pulses delivered by a 80 MHz, 800 nm Ti:Sapphire oscillator. Since coupling between electromagnetic radiation and SP waves cannot occur directly, a silver coated (50 nm) prism has been used in the Kretschmann geometry. Pulses fr*om the Ti:Sapphire oscillator are focused to a spot size of 60 jiim, and the SP coupling condition is optimized by maximizing the electron emission. The photocurrent is monitored using an electron multiplier (0.28 sr. collection angle), which along with the prism, has been placed in a vacuum chamber that has been evacuated to 10'^ Torr.

667

By tracking the dependence of the photocurrent on incident power, the electron emission mechanism can be estabHshed. The results of this experiment are shown in Figure la, which clearly illustrates the third-order multiphoton dependence of the photocurrent on incident power, indicating that a three-photon relation {^electron'^^hn-Wf) is domiuaut. cr 10.0 1.0 H

c^ 0.1 intensity (GW/cm")

delay (fs)

Fig. 1. (a) Measured variation of photocurrent on excitation intensity, demonstrating a third-order process, (b) Measured 3PP correlation trace for -26 fs pulses. Since the electron pulse duration cannot be measured directly, the three-photon photoemission (3PP) process is employed in a third-order correlation experiment. The lifetime for plasmons in a thin silver film is ~48 fs [5], and since the excitation pulse (25 fs) is lower than this value, the SP's will remain coherent for the duration of the electron generation and acceleration. Electron generation will only occur while the optical field illuminates the silver film, as indicated by the multiphoton emission (Fig. 1), and the pulse duration can be deduced from a 3PP correlation. Figure lb contains the results of an interferometric two-pulse 3PP correlation, demonstrating 26 fs laser pulses. Given that the electron emission is a 3PP process, the duration of the electron pulses are <26 fs. As the plasmons are excited on a time-scale below their lifetime, electron emission will be in phase with the SP waves, and the resultant ponderomotive potential for acceleration will be higher. Therefore, it is expected that more energetic electrons, as compared to the case of incoherent SP launching using longer optical pulses [3], will be generated. This is illustrated in Figure 2a, which shows a typical electron pulse kinetic energy spectrum. The center of the Gaussian-like spectrum is 315 eV, with energies ranging from 250 to 400 eV. The measured full-width at halfmaximum is 83 eV, only 26% of the mean value. This is in contrast to other experiments [3] that have obtained similar energies using intensities (//ascv=40 TW/cm", 150 fs pulses) that are four orders of magnitude larger than the intensities used here (4;vcv = 1.8 GW/cm"). This is a clear indication that the excitation pulse duration is a fundamental parameter in the determination of the ponderomotive potential, and impulsive SP excitation is necessary for efficient electron acceleration. The UPAS model is employed to study electron pulse emission and acceleration in the presence of an SP field, which accounts for the spatial/temporal distribution of the electromagnetic field and the 3PP mechanism. In the UPAS code, an input

668

optical pulse with a duration of 25 fs and a central wavelength of 800 mn is used to excite SP waves in a 50 nm metal film, as illustrated in Figure 2b. Clearly, the enhanced field decays away from the prism surface, which is ideal for the ponderomotive acceleration of electrons. Figure 2c illustrates several electron trajectories, as calculated by the UPAS code, which demonstrate the 'quivering' of the electrons as they gain energy from the ponderomotive potential of the highgradient electromagnetic field. Figure 2a displays the experimental and simulated kinetic energy spectra, which are in good agreement. The enhanced electric field used in the simulation is 1.38x10^ V/cm, whereas the experimental Eiasc- =1.2x10^ V/cm. Based on a comparison between the two, the effective SP enhancement factor is -lO"". The large noise features of the modeled kinetic energy spectrum are expected since only a solitary pulse is simulated, which is in contrast to the smooth experimental data that has been obtained from averaging over 80 million shots per second.

270 320 370 kinetic energy (eV)

Fig. 2. (a) Simulated (solid line) and experimental (circles) kinetic energy distributions of the SP generated electron pulses, (b) Simulated electromagnetic field distribution resulting from the excitation of SP waves in a thin metal film, (c) Several electron trajectories, as calculated by the UPAS model, illustrate the complex motion of the electrons in an intense electromagnetic field.

References 1 B. J. Siwick, J. R. Dwyer, R. E. Jordan, and R. J. D. Miller, Science 302. 1382 (2003). 2 J. C. Williamson, J. Cao, H. Ihee. H. Frey, and A: H. Zewail. Nature 386, 159 (1997). 3 J. Zawadzka, D. Jaroszynski, J. J. Carey, and K. Wynne, Appl. Phys. Lett. 79, 2130 (2001). 4 J. Kupersztych, P. Monchicourt, and M. Raynaud, Phys. Rev. Lett. 86, 5180 (2001). 5 M. van Exter and A. Lagendijk, Phys. Rev. Lett. 60, 49 (1988).

669

Space-Time Control in Ultrafast Nano-Optics T. Brixner^ J. Schneider\ W. Pfeiffer^ and F. J. Garcia de Abajo^ ^ Physikalisches Institut, Universitat WUrzburg, Am Hubland, 97074 Wiirzburg, Germany E-mail: [email protected] ^ Centre Mixto CSIC-UPV/EHU, Apartado 1072, 20080 San Sebastian, Spain Abstract. The interaction of optimally polarization-shaped laser pulses with nanostructures allows simultaneous spatial and temporal control of electric fields below the diffraction limit. Simulations demonstrate controlled localization of the nonlinear response in space and time, opening new possibilities for spatially resolved femtosecond spectroscopy. Electric near-field distributions are at the center of many experimental techniques in fundamental and applied sciences such as in scanning tunneling microscopy (STM), scanning near-field optical microscopy (SNOM), near-field two-photon fluorescence microscopy [1], spatially localized photochemistry, or novel nanophotonic devices for communication and computing purposes [2]. Ultrahigh spatial resolution below the diffraction limit can be provided by making use of the optical field enhancement in the vicinity of a sharp tip. It had already been shown that the linear chirp of femtosecond laser pulses has an influence on the energy localization in nanosystems [3]. Here we use not only linear chirp but we rather optimize many-parameter spectral phase and polarization profiles [4] to localize the nonlinear response at specifically desired points in space and time. The proposed experimental schematic is shown in Fig. 1. Femtosecond laser pulses are modified in a polarization pulse shaper [4], and near-field enhancement is provided under plane-wave illumination by a nanoscale model system. Fig. 1. Schematic setup. A femtosecond polarization pulse shaper manipulates the spectral phases of two polarization components at 128 frequencies (between 670 nm and 890 nm), leading to ultrafast timepolarization varying polarization states. pulse shaper These pulses are irradiated onto a nanoscale model system, in the example here a silver tip-sphere combination (« = 10 nm, Z? = 25 plane-wave illumination nm, J = 5 nm, L = 1500 nm. a=5°, 0= 45°). We calculate the full vectorial electricfieldat any point P as a function of the input pulse shape by solving Maxwell's equations with help of the boundary-element method. Control over local electric field properties is achieved by optimizing the pulseshaper settings with an evolutionary algorithm and various different control objectives 670

The calculation of the local field is based on the time dependence of the vectorial external field, the detailed geometry, and electromagnetic propagation and retardation effects along with specific material properties [5]. As a result, we obtain the vectorial time-dependent electric field at points P in the vicinity of the nanosystem [6]. Different objectives of space-time nanoscale control are then defined by a suitable "fitness function", and an evolutionary algorithm is used to optimize the polarization pulse-shaper settings. In our first example, we localized a nonlinear second-order response at different points within an xy-plane located between tip and sphere. For this purpose, we maximized the fitness function

in which the dependence of the signal on the time-integrated intensity squared of the field describes the probability of an anisotropic two-photon excitation (e.g. of molecules absorbed on a surface with transition dipole moments in the xy-plane). The goal was to get high excitation probability at r, and low excitation at all other points Tn within that plane, where the summation with Gaussian weighting factors g(r) and 25 nm FWHM introduces penalties for signals away from the target area. Typical results of optimizations with different target points are shown in Figs. 2b-c, plotting the nonlinear responses throughout the xy-plane. The points of high nonlinear excitation (dark shading) can indeed be located in different places as desired. Optimization results for a number of spatial points r, are shown as a function of position in Fig. 2a. The excitation can be varied in a two-dimensional fashion, with a degree of localization that is the better the darker the shading. The area of best localization reflects the lateral extension of the optical near-field modes that are excited with the polarization-shaped laser pulse. The use of polarization shaping is crucial in this scheme because linearly 1-polarized or 2polarized (phase-shaped) laser pulses alone do not allow controlling the localization of the nonlinear response in the xy-plane. This scheme is an attractive tool for near-field microscopy without mechanical movements. Thus it is possible to avoid artifacts that can arise from topographical surface features during the mechanical scanning procedure. Another application would be nanoscale all-optical spatial multiplexing.

-20

0

20

x-Axis [nm]

-50

0

50

Fig. 2. (a) Contour plot of the maximum fitness obtained by optimizing Eq. (1). The optimized distribution of the timeintegrated nonlinear response is shown for two sample points (white crosses) in (b) and (c). High nonlinear response corresponds to dark shading

x-Axis [nm]

671

In the next step, we localized the light intensity at certain spatial points r^ and specific times U, For this purpose, we minimized the fitness function

(

^sr.

V

dt (2) a \ a J in which the Ua (cc = x,y,z) are coefficients that determine which field components are optimized and pi(t) reflects the desired time dependence of the normalized nonlinear response at point r^. The goal in this particular example was to achieve a high intensity at point ri = (-30, -30, 27.5) nm and a time of ti = -30 fs but also at point r2 = (-30, +30, 27.5) nm at the later time t2 = +30 fs, with low intensity at all other times. The resulting optimized spatial-temporal evolution of the nonlinear response (Fig. 3) indeed peaks at -30 fs for point ri, whereas at r2 it is practically zero. Similarly, at the desired time +30 fs the nonlinear response is concentrated at r2 while it vanishes at ri. Such control opens the door to novel pump-probe schemes with simultaneous ultrahigh temporal and spatial resolution. For example, one could perform experiments on macromolecules or other nanosystems in which a single laser pulse excites the system in one place at time ^i, and the same laser pulse probes the system at a different location at a later time ^2. We have verified that the pump-probe delay time (60 fs in Fig. 3) can be varied in a general fashion within the temporal window of the pulse shaper. Accordingly, this scheme offers a new way to study energy transport by direct spatial and temporal probing. / • •

Fig. 3. Time evolution of the nonlinear response / / + / / on a line defined by jc = -30 nm and z = 27.5 nm. The signal strength is given in arbitrary units

-30

0 30 Time [fs]

References 1 2 3 4 5 6

672

E. J. Sanchez, L. Novotny, X. S. Xie, Phys. Rev. Lett. 80, 5180,1998. T. Kawazoe, K. Kobayashi, S. Sangu, M. Ohtsu, Appl. Phys. Lett. 82, 2957, 2003. M. L Stockman, S. V. Faleev, D. J. Bergman, Phys. Rev. Lett. 88, 067402, 2002. T. Brixner, G. Gerber, Opt. Lett. 26, 557, 2001. F. J. Garcia de Abajo, A. Howie, Phys. Rev. B 65, 115418, 2002. T. Brixner, W. Pfeiffer, F. J. Garcia de Abajo, Opt. Lett., in press, 2004.

Coherent Control of Ultrafast Linear and Nonlinear Optical Phenomena in Nanostructures Mark I. Stockman^ David J. Bergman^, and Takayoshi Kobayashi^ ^ Department of Physics and Astronomy, Georgia State University, Atlanta, Georgia 30303, USA E-mail: [email protected] " School of Physics and Astronomy, Raymond and Beverly Sackler Faculty of Exact Sciences, Tel Aviv University, Tel Aviv, 69978 Israel E-mail: [email protected] ^ Department of Physics, University of Tokyo, Kongo 7-3-1 Bunkyo-Ku, Tokyo 113-033, Japan E-mail: [email protected] Abstract. We theoretically investigate the unique possibility to control distribution of ultrafast local optical fields, both linear and nonlinear, in nanosystems with nanometer spatial resolution and in time on the femtosecond scale. Various applications are discussed.

1.

Introduction

The concentration of optical excitation energy on the nanoscale is one of the most important problems in nanoscience. Metal nanosystems, such as fractal clusters, rough surfaces and films, and nanostructured systems, such a tip of the apertureless near-field scanning tunneling microscope (NSOM), are known to produce highly enhanced local optical fields, sufficient to observe single molecules ^'^. Obtaining such high enhancements is a formidable problem: due to the very strong electromagnetic interaction between nanoparticles separated by distances on the nanometer scale, the excitation tends to spread over the system and delocalize. The ultrafast concentration of optical energy on the nanoscale is even more difficult, for which there are two reasons. First, it is impossible to directly focus the optical radiation on the nanoscale to achieve the nanolocalized optical excitation because the light wavelength A is so much larger. Second, due to the strong interaction between the constituent nanoparticles, the excitation transfer times in nanostructured systems are very short, typically on femtosecond scale. We theoretically investigate a unique way to control the localization of optical excitation with nanometer spatial resolution and femtosecond temporal resolution using the phase modulation of the excitation radiation as a functional degree of freedom '^' ^, For linear responses, this allows one to change the temporal distribution of the ultrafast local fields at a given spatial point to concentrate the excitation in time while the integral (time-averaged) local optical energy in each

673

point is not controllable and does not change. In contrast, for nonlinear phenomena, also the time-integrated local response is coherently controllable ^.

2. Theory and Numerical Results The theory is based on the^spectral expansion in the Green's function method presented in Refs. '^' ^. This method represents Green's function in the form that exactly preserves its analytical properties such as symmetry, dispersion (KramersKroning) relations, and causality. As an example of the nanosystem, we consider a silver V-shape embedded into a transparent host medium with dielectric constant £^ = 2. The geometry of this Vshape in its plane is shown in Fig. 1. The thickness of this V-shape in the normal ( y ) direction is 2 grid steps (2 to 6 nm). We have studied effects of four excitation pulses shown in Fig. 2 (a)(d). Pulses (a)-(c) have the same envelope; pulses (b)-(d) have the same spectral composition. Pulses (b) and (c) are phase modulated [negative chirp: (b) and negative chirp: (c)]. Pulses (a) and (d) are transform limited. The comparison of the corresponding responses allows one to distinguish the effects of the pulse shape, spectral composition, and phase modulation.

30 20 10

0

V 10

20

Fig. 1. Geometry of V-shape. The coordinates are shown in the grid units. One such a unit corresponds to 1-3 nm. 3,0

For the comparison with experimental data, the measurement of spatio-temporal responses on the nanometer -femtosecond scale is the most formidable problem. Much simpler a problem would be the measurement of time-integrated (averaged) responses. However, it can easily be shown that the integrated linear responses are insensitive to the phase modulation of the excitation pulses for their given spectrum, i.e., not coherent-controllable. In contrast, nonlinear responses are controllable by the phase modulation, even after averaging (integration) over time. In Fig. 2 (e)-(h), we show the time-averaged spatial distribution of the intensity of the two-photon excitation, an incoherent nonlinear response. It is shown relative to such a response induced by the excitation pulse, scaled by 10"^ The two-photon excitation is responsible for the anti-Stokes fluorescence in metals used for study of nanostructures ^. Comparison Fig. 2 (a) with (f)-(h), we see that the spectrum of the pulse is a very important factor: the wide-band pulses (b)-(c) induce much stronger two-photon responses. Comparison of Fig. 2 (f), (g) with (h) shows that

674

the chirped pulses (c) and (d) induce the two-photon local intensity that is much better localized at the tip of nanostructure. Finally, comparing Fig. 2 (f) and (g), we conclude that in our case the negative-chirp pulse (b) induces the maximum and best-localized two-photon response Fig. 2 (f), which is strongly localized at the tip of the nanostructure. Thus, all the factors of the excitation pulse (spectrum, phase modulation, and length) are important. Interestingly enough, the shortest and highest-amplitude pulse of Fig. 2(d) is not the optimum one for the two-photon excitation. The enhancement coefficient of the two-photon excitation for the silver V-shape is very high, by a factor of ~ 10^. Ei-(t)

Ei°) (t)

(3)

E^' (t)

(d)

1

00

(e)

t(fs) 200

0

DO

(f)

h^^m

(g)

t(fs) 200

0 ^

100

t(fs) 200

(h)

L i ^^

Fig. 2. Excitation pulses [narrow-band, unchirped (a), negative chirp (b), positive chirp (c), and wide-band, unchirped (d)]. The excitation electric field is shown in arbitrary units. The enhancement factors for the second-order responses are shown in the units of 10^ for corresponding columns (e)-(h)]

Acknowledgements. This work was supported by grants from the Chemical Sciences, Biosciences and Geosciences Division of the Office of Basic Energy Sciences, Office of Science, U.S. Department of Energy, and a grant from the USIsrael Binational Science Foundation. References 1 K. Kneipp, Y. Wang, H. Kneipp, et al., Phys. Rev. Lett. 78, 1667 (1997). 2 S.M.Nie and S.R.Emery, Science 275, 1102(1997). 3 A. Hartschuh, E. J. Sanchez, X. S. Xie, et al., Phys. Rev. Lett. 90, 095503 (2003). 4 M. I. Stockman, S. V. Faleev, and D. J. Bergman, Phys. Rev. Lett. 88, 67402 (2002). 5 M. I. Stockman, D. J. Bergman, and T. Kobayashi, Phys. Rev. B 69, 054202 (2004). 6 A. BouheHer, M. R. Beversluis, and L. Novotny, Appl. Phys. Lett. 83, 5041 (2003).

675

SPASER as Ultrafast Nanoscale Phenomenon and Device Mark I. Stockman^ and David J. Bergman^ ^ Department of Physics and Astronomy, Georgia State University, Atlanta, Georgia 30303, USA E-mail: [email protected] " School of Physics and Astronomy, Raymond and Beverly Sackler Faculty of Exact Sciences, Tel Aviv University, Tel Aviv, 69978 Israel E-mail: [email protected] Abstract. We theoretically introduce surface plasmon amplification by stimulated emission of radiation (SPASER) as an ultrafast phenomenon and device. Spaser is predicted to generate femtosecond, nanoscale-localized pulses of nearly atomic-strength local fields. Various applications are discussed.

1.

Introduction

There has recently been great interest in the nanoscale optical phenomena, including plasmonic nanooptics. The external laser fields exciting a metal nanostructured system cause generation of high-intensity local optical fields. These fields are greatly enhanced due to the high quality factors of surface plasmon resonances in nanostructures made of noble metals. The creation of ultrafast, strongly enhanced field on the nanoscale is a formidable problem. It s impossible to directly focus the optical radiation on the nanoscale, which is due to the wavelength of light being on the microscale, many orders of magnitude too large. Coupling laser radiation to the nanoscale through, e.g., tapered optical fibers ' or by focusing on metal tips ^ leads to an enormous loss: only a miniscule part of the excitation energy is transferred to the nanoscale. The known methods and applications of nanooptics are based on the excitation of local fields by an external laser source. Here we propose a principally different approach to use the recently introduced phenomenon of surface-plasmon amplification by stimulated emission of radiation (SPASER) ^ for the generation of ultrastrong (comparable to the atomic-scale strength), ultrafast (~ 1 - 50 fs) pulses directly on the nanometer scale. The idea of spaser is based on the fact that surface plasmons (SP's) have the same properties necessary for the stimulated emission as photons: (i) SP's are bosons; (ii) they are neutral excitations. (///) SPP's are highly harmonic excitations. Of principal importance for the nanooptics, in contrast to photons, SP's are known to localize on the nanoscale due to the negligible amplitude of their magnetic component: SP's are purely electric oscillations '^. A possibility that spaser will be

676

a femtosecond generator is based on the facts that the coupling to the SP's is strong and the spectral width of the amplification region is large enough (-- 1 eV).

2.

Theory of Spaser as an Ultrafast Phenomenon and Device Quantum Dots

Active Medium''

Ftesonait I\lanoparticle Quantum Dots Substate

Fig. 1. Schematic of a possible design of spaser (arrows indicate assembly steps). Spaser incorporates an active medium formed by two-level emitters, excited in the same way as a laser active medium: optically, or electrically, or chemically, etc. One promising type of such emitters are quantum dots (QDs). These emitters transfer their excitation energy by radiationless transitions to a resonant nanosystem that plays the role of a laser cavity. These transitions are stimulated by the SPs already in the nanosystem, causing buildup of a macroscopic number of SPs in a single mode. A possible design of a spaser is illustrated in Fig. 1. We have carried^ out numerical computations for a silver V-shape as resonant nanoparticle [Fig. 2(a)] in the visible/near-ir spectral region. The spaser predicted gain [Fig. 2(c)] is large: the increase time of SP population is r^/a^ ^ 6 fs for the maximum a^ [cf. Fig. 2(b)]. The amplification contour is wide enough to support the ultrafast SP generation. Thus, spaser is predicted to be an ultrafast device. The eigenmodes (SP's) with highest gains shown in Fig. 3 comprise both luminous [(a) and (d)] and dark [(b) and (c)],delocalized and strongly localized. Importantly, the electric fields EJN„+\/2 that are generated by a spaser can be as large as atomic fields. A spaser generating a dark SP mode provides a unique possibility for a background-free nano-spectroscopy, where the spaser by itself does not radiate in the far zone. However, molecules in spaser vicinity are strongly excited by the SP local fields, and their radiation is further enhanced by the metal nanoparticles (similar to SERS) and can be detected in the far field.

677

OCn

(b)

15 10 5 0

10 20 30

h(jOn(eiV)

1 2

3

Fig. 2. (a) Geometry of V-shape as resonant nanoparticle (one grid unit corresponds to 1-2 nm); (b) Lifetime of SPs as function of their frequency; (c) Gain a,^ of a spaser as a function of frequency for three monolayers of quantum dots as active medium; (d) Gain of spaser vs. oscillator strength /.ofSPs. l.lSeV,

Fig. 3. Eigenmode amplitudes for (SPs) with highest gains for the frequencies co^, gains a^, and oscillator strength f^ shown. The local electric field amplitude is EJN,, +1/2Acknowledgements. This work was supported by grants from the Chemical Sciences, Biosciences and Geosciences Division of the Office of Basic Energy Sciences, Office of Science, U.S. Department of Energy, and a grant from the USIsrael Binational Science Foundation. References 1 2 3 4

678

A. A. Mikhailovsky, M. A. Petruska, M. I. Stockman, et a!., Opt. Lett. 28, 1686 (2003). A. Hartschuh, H. N. Pedrosa, L. Novotny, et al., Science 301, 1354 (2003). D. J. Bergman and M. I. Stockman, Phys. Rev. Lett. 90, 027402 (2003). M. I. Stockman, S. V. Faleev, and D. J. Bergman, Phys. Rev. Lett. 87, 167401 (2001).

Ultrafast quenching of ring closure in molecular switches^ self-assembled on gold nanoparticles Ralph Hania, Audrius Pugzlys, Tibor Kudemac, Harry Jonkman and Koos Duppen Materials Science Center, University of Groningen, Nijenborgh 4, 9747AG Groningen, The Netherlands E-mail: Hania(g),chem.rug.nl Abstract. We report the ultrafast quenching of the ring-closure reaction in BTE-based photochromic switches self-assembled on gold nanoparticles. The photoinduced population dynamics of the switches reveals that the electronic states of the switch molecules are strongly mixed with the states of the gold particles.

1. Introduction In recent years, there has been increasing interest in the synthesis, investigation and application of organic photochromic materials as a possible basis for optoelectronic and photo-optical devices [1, 2 and references therein]. Bisthienylethene (BTE) based compounds in particular attracted much attention because of their lov^ fatigue, high thermal stability, remarkable switching sensitivity and rapid response [3]. Implementation of such photochromic switches into electronic elements is expected to open new perspectives in the creation of novel opto-electronic devices with unique functionalities. Therefore, determination of the pathways of excitation relaxation in photochromic switches assembled on metal contacts, such as occur in devices, is an important issue from both fundamental and practical point of view. Ring opening and ring closure reactions in free BTE-based photochromic switches were extensively studied in the past decade. It was demonstrated that the ring closure reaction in the switch molecule proceeds via ultrafast equilibration of the initially excited state with an electronic state of different symmetry, from which a transition to the ground state of the closed form occurs in a few picoseconds [2,3]. The reaction quantum yield (QY) was shown to be close to 0.5. Very recently, it was demonstrated that in a BTE-based molecular switch attached to gold contacts, the ring closure reaction is fully quenched and switching occurs only in one direction, namely from the closed to the open form [4]. It was suggested that the absence of the ring closure is caused by the quenching of the precursor state of the open form by gold. Here, we discuss the dynamics of the ring closure of a thiophene substituted bisthienylcyclopentene photochromic switch (T-DTCP) in solution and compare this to the dynamics of the same switch attached to gold nano-particles. The studies are performed by means of femtosecond, polarization selective, frequency resolved pump-probe experiments.

679

2. Experimental Methods The isolated T-DTCP molecules, the same molecules self-assembled on gold nano-particles and reference dodecanethiol-coated gold nanoparticles were all dissolved in toluene. In order to avoid photodecomposition, the solution was flowed through a 100 ^m cell by a peristaltic pump. Pump-probe experiments were performed in the standard geometry [2]. All samples were excited with 100 nJ, 350 nm pulses, while the photoinduced population dynamics was probed in the 490-620 nm spectral region. Gaussian fits of two-color two-photon absorption in a 100 |Lim glass plate indicated a time resolution of the experiments of 63 fs at wavelengths larger than 520 nm and up to 90fsat490nm.

3. Results and discussion Characterization of the T-DTCP photochromic switch. In Fig. (la) the chemical structure of the T-DTCP switch molecule in the open and the closed form is shown. Switch molecules in the open-ring form absorb only in the nearUV spectral region (solid line in Fig. lb). Ground-state interconversion between the open-ring and closed-ring isomers does not occur, but upon UV-irradiation the open-ring isomer converts to the closed-ring one. As a consequence of the near-coplanar geometry of the thiophene moieties in the closed-ring form, TT-conjugation spreads throughout the molecule resulting in the appearance of a new red shifted absorption band (dashed line in Fig. lb). Upon visible irradiation the ring opens, which breaks this conjugation. Our estimate of the QY of the ring closure reaction in toluene solution is 0.4+0.1. This is mainly determined by the co-existence in a close to 1:1 ratio of switchable and non-switchable conformers (for details see ref. 2). (a)

(b)

. 8 r\

T\

0.25 0.20

y Q

VIS

uv

o ,8 I

'/<

0.15 \ 1

o 0.10 0.05

X

S S ""%

open-ring isomer closed-ring Isomer -

J //\ \ 1/' *1 1 ' M

^

\\

1 /

M

r

H

// '*• \

, / ,' ' » ' \ \ -^ \\ /

\ \ \ ^

V

0.00

500

"•?

.'

600

X, nm Fig. 1. Panel (a): chemical structure of the thiophene substituted BTE molecule in the open-ring (top) and closed-ring (bottom) forms. Panel (b): steady state absorption spectra of the molecules depicted in panel (a).

Dynamics of the ring closure reaction in the isolated T-DTCP switch. The dynamics of T-DTCP, measured at different probe wavelengths, reveal that the initially excited state, similar to the case of the benzene substituted bisthienyl-

680

cyclopentene molecular switch (B-DTCP) [2], decays via mixing 0.02•^-*-«*.*.*^,^^,^490 nmwith an intermediate state. From •/ "^ this latter state a transition occurs ^ 530 nm to the ground state of the closed 580 nm •• • 0.01form. The time constants of the J< dynamics are obtained by fitting pump-probe transients measured 0.000.0 0.5 1.0 1.5 at probe wavelengths of 490 nm Probe delay (ps) and 580 nm (see Fig. 2) at which, respectively, the initially excited Fig.2. Pump probe transients of T-DTCP in and intermediate states are probed. solution at different probe wavelengths. This reveals an evident precursorsuccessor relation with a transition time of 95 fs. This resembles the reported scheme of the ring closure reaction of B-DTCP [2]. The ring closure itself is observed by monitoring the formation of the ground state of the closed form. The time constant, thus obtained, is 3.7 ± 0.3 ps, which is typical for ring closure of perhydro-BTE derivatives [3]. 1

•

—

1

1

—

•

—

'

—r

•

1

^

~;

Dynamics in T-DTCP self-assembled on gold nanoparticles. In the case of T-DTCP molecules attached to gold nano-particles with sub 5-nm diameter, the dynamics change dramatically. The pump-probe transients measured at different probe wavelengths are, after normalization, identical within noise level. The signals are formed instantaneously, which is followed by bi-exponential decay with time constants of 300±30 fs and 3.1±0.2 ps. The observed instantaneous formation of the signal reveals that the energy transfer from the intermediate state to the gold manifold, suggested in [4], is oversimplified. The observed decay times are close to the time constants that were observed for the reference gold particles (0.3±0.1 ps and 3.310.3 fs). These times are characteristic for the surface-plasmon dynamics of gold, due to electronelectron and electron-phonon scattering, respectively [5]. This indicates that strong mixing of electronic states takes place after the covalent self-attachment of the switch molecule to the gold surface. An independent treatment of the switch molecule and the gold surface is then impossible.

References 1 S. Shim, T. Joo, S. C. Bae, K. S. Kim and E. Kim, J. Phys. Chem. A 107, 81068110,2003 2 P. R. Hania, R. Telesca, L. N. Lucas, A. Pugzlys, J. van Esch, B. L. Feringa, J. G. Snijders and K. Duppen, JPCA 106, 8498-8507, 2002 3 M. Irie, Chem. Rev. 100, 1685-1716, 2000 4 D. Dulic, S. J. van der Molen, T. Kudemac, H. T. Jonkman, J. J. D. de Jong, T. N. Bowden, J. van Esch, B. L. Feringa and B. J. van Wees, Phys. Rev. Lett. 91, 207402-1-4, 2003 5 S. Link and M. A. El-Sayed, JPCB 103, 8410-4826, 1999

681

Part X

Terahertz Wave and Applications

Temporal spectroscopic behavior of terahertz pulses transmitted through metal hole arrays Fumiaki Miyamaru^ Masanori Hangyo^ ^Institute of Laser Engineering, Osaka University, 2-6 Yamadaoka, Suita, Osaka 565-0871, Japan E-mail: [email protected]

Abstract. The temporal spectroscopic behavior of the terahertz single pulse, which transmitted through the metal hole array, was investigated using the short-time Fourier transform (Gabor transform). In the Gabor spectrum, it is observed that the THz wave at the resonant frequency of the surface plasmon-polariton (SPP) is delayed for several picoseconds with compared to that at other frequencies which are higher than the diffraction frequency. Since such delay time depends on the thickness of the metal slab, it is considered that the delay at the SPP resonant frequency is attributed to the deceleration of the group velocity of the THz wave inside the metal hole.

1.

Introduction

Metallic photonic crystals have possibilities for fabricating attractive optical devices by constructing periodic structures. For the two-dimensional metallic hole array, the extraordinary optical transmission property was demonstrated in the optical region by Ebbesen et al. [1] The peak transmittance of the band-pass property is very high and several times of magnitude compared to the porosity of the metal hole array. Up to now, it is well established that the transmission peak of the metal hole array is attributed to the resonant coupling of the incident light to surface plasmon-polaritons (SPPs) excited on the metal-dielectric boundary not only in the optical region [2,3] but also in the terahertz (THz) region [4]. Not only the transmittance for the metal hole array but also the temporal behavior of the electromagnetic wave are considered to be very important properties for applying the metal hole array to ultrafast optical devices. However, in the optical region, it is difficult to observe the temporal spectroscopic behavior of light because of its high frequency. In the terahertz (THz) region, on the other hand, the temporal spectroscopic behavior can be observed because the temporal wave form of the THz wave can be measured using the THz time-domain spectroscopy (THz-TDS). The temporal wave form transmitted through the metal hole array has been already reported experimentally in the THz region [5]. However, physical insight of the temporal behavior of the transmitted electromagnetic wave has not been investigated in detail so far. In this paper, we investigate the transmission property of the THz pulse in the spectral-time space using the short time Fourier transform (Gabor transform).

685

2.

Experimental Methods

The metal hole array used in this study was an aluminum slab which was perforated in the triangular structure with the circular holes [Fig. 1(a)]. The transmission characteristics of this type of metal hole array is mainly determined by the geometrical parameters, such as a hole diameter d, a spacing between holes s^ and a slab thickness t. Since all metals can be treated as a perfect conductor in the THz region because of the very large absolute value of the real part of the permittivity, the species of the metal is not important for the transmission characteristics of the metal hole array in the THz region. We measured the wave form of the THz wave transmitted through the metal hole array by using the THz time domain spectroscopic (THz-TDS) system [4, 7]. The schematic diagram of the THz-TDS system is shown in Fig. 1(b). The fundamental principle of the THz-TDS is as follows. The THz wave was radiated from a photoconductive antenna which was illuminated by 100 fs laser pulses. The THz radiation was collimated and focused on a detector with a pair of offaxis paraboloidal mirrors. The detector was illuminated by gating laser pulses divided from the pump laser pulses. By changing the delay between the pump and gating laser pulses, the signal which was proportional to the electric field of the THz radiation was measured in time domain. The transmission and phase shift spectra was obtained by comparing the Fourier components of the measured signals with and without the sample inserted on the path of the THz wave. In order to investigate the temporal spectroscopic behavior of the THz wave, the Gabor transform was used with the Gaussian time window. By changing the delay time of the Gaussian time window with respect to the signal of the THz wave, the transmission spectrum at each delay time was obtained.

Optical Delay

A rk

(b) Mode Locked Ti: sapphire Laser

iGateBeam Pump Beam' DetectorW

^---^Emi^ Emitter Sample

5 mm

Off-Axis Paraboloidal Reflector

Fig, 1 (a) Photograph of the metal hole array, (b) Schematic diagram of the optical setup of the THz-TDS system.

686

3.

Results and Discussion

We measured the temporal wave form of the THz wave transmitted through the metal hole array, the geometrical parameters of which were d=0.6S mm, ^=1.13mm and ^O.SOmm. Figure 2(a) shows the measured THz wave forms transmitted through the metal hole array (solid line) and the incident THz wave (dashed line). The incident THz wave is almost single cycle pulse. The transmitted THz wave has sharp peak around / = 0 ps and has over ten cycle oscillations after the sharp pulse, corresponding to the band pass characteristic of the metal hole array. The transmission spectrum of the metal hole array is shown in Fig. 2(b). The transmission peak is clearly observed at 0.27 THz, which is attributed to the SPP excitation on the metal surface. In the spectral region higher than the diffraction frequency of 0.31 THz, the transmittance is relatively flat. To investigate the temporal transmission property for each frequency, we calculated the Gabor transform spectrum [Fig. 2(c)]. In this calculation, the M l width of half maximum (FWHM) of the Gaussian window function was 34 ps. The THz wave at the frequency higher than the diffraction frequency (> 0.31 THz) has very short rise and decay times and has the maximum of the Gabor intensity at around / = 0 ps. On the other hand, the Gabor intensity of the THz wave at the transmission

0 0.5 N

I

Transmittance 0.5 1.0 (b) 0.5

0.4 0.3 0.2

-0.0 ^ <— AT=10 ps

0.1 (a) 10 J3

3

I< 20

40 Time (ps)

Fig. 2 (a) Measured wave form of the THz wave transmitted through the metal hole array, and (b) transmission spectrum, (c) Gabor spectrum of the THz wave calculated from the wave form with the Gaussian window function.

687

peak frequency (0.27 THz) rises slowly and becomes maximum at ^ 1 0 ps. It is considered that the THz wave higher than the diffraction frequency transmits through the air hole with small interaction with metal holes. At 0.27 THz, on the other hand, the THz wave interacts strongly with the metal holes, leading the time delay for transmitting through the metal hole. Two mechanisms are considered the reason of the delay of the THz wave at the transmission peak frequency. One mechanism is the conversion time from the SPP mode to the emitted THz wave into the free space. The other mechanism is the deceleration of the group velocity of the THz wave inside the metal hole. In order to investigate which mechanism is dominant for the delay time observed in Fig. 1(c), we measured the dependence of the delay time on the thickness of the metal slab. For the former mechanism, the delay time is considered to be independent on the slab thickness because the conversion time of the SPP is determined only by the surface structure. For the latter mechanism, on the other hand, the delay time will increases with the length of the metal hole, and consequently depends on the slab thickness. Figs. 3(a)-(d) show the Gabor spectra of the THz wave transmitted through metal hole arrays with the slab thicknesses of ^=0.20, 0.50, 1.00 and 2.00 mm, respectively. The hole diameter and the lattice constant are (#=0.70 nam and 5=1.12 mm for all metal hole arrays. It is clearly recognized that the delay time Ax between the high frequency region (> 0.31 THz) and the transmission peak frequency (0.27 THz) increases from AT=3.5 to 18 ps with increasing the slab thickness in the range from 0.20 to 2.00 mm. This slab thickness dependence implies that the delay time of the THz wave at the transmission peak frequency is

^ = 0.5 mm

t = 0.2 mm

0

0

20

20

Time (ps)

Time (ps)

t= 1.0 mm

K^)

0.4-

0.20

20

Time (ps)

il L,

-20

r = 2.0mm 10.8

1 1

1

—>{— ^ 1— 0

-AT=18.0ps 20

40

Time (ps)

Fig. 3 Measured Gabor spectra of the THz wave transmitted through the metal hole arrays for four slab thicknesses of (a) 0.2, (b) 0.5, (c) 1.0 and (d) 2.0 mm.

688

due to the deceleration of the group velocity inside the metal hole. From the result of Figs. 3, the group velocity of the THz wave can be estimated at 0.32c, where c is the velocity of light in vacuum. The group velocity of the electromagnetic wave inside the metal hole can be estimated from the formula as follows [7].

= cfVJff

(1)

YIQXQ, fc=\.M\clnd

is the cutoff frequency of the TEn mode which is the lowest order of the waveguide mode of the metal hole, / is the frequency of the electromagnetic wave. The group velocity at 0.27 THz is estimated to be 0.37c from Eq. (1), being in good agreement with that observed in our experiment. From this result, it is concluded that the delay at the transmission peak frequency observed in Fig. 2(c) is mainly attributed to the deceleration of the group velocity of the THz wave in the metal hole.

4.

Conclusions

In summary, we investigated the temporal spectroscopic behavior of the THz wave transmitted through the metal hole array using the Gabor transform. In the Gabor spectrum, the time delay of the THz wave at the transmission peak frequency compared to that at higher frequency range is observed. This time delay is not attributed to the conversion time of the SPP but to the deceleration of the group velocity of the THz wave inside the metal hole. Acknowledgements. This work was partially supported by the Gant-in-Aid for Sientific Research from the Ministry of Education, Culture, Sports, Science, and Technology, Japan.

References 1 2 3 4 5 6 7

T. Ebbesen, H. Lezec, H. Ghaemi, T. Thio, and P. Wolff, Nature 391, 667 (1998). E. Popov, M. Neviere, S. Enoch, and R. Reinisch, Phys. Rev. B 62, 16100 (2000). A. Krishnan, T. Thio T. Kim, H. Lezec, T. Ebbesen, P. Wolff, J. Pendry, L. MartinMoreno, andF. Garcia-Vidal, Opt. Comm. 200, 1 (2001). F. Miyamaru and M. Hangyo, Appl. Phys. Lett. 84, 2742 (2004). C. Winnewisser, F. Lewen, M. Schall, M. Walther, and H. Helm, IEEE Trans. Microwave Theory Tech. 48, 744 (2000). S. Nashima, O. Morikawa, K. Takata, and M. Hangyo, J. Appl. Phys. 90, 837 (2001). N. Marcuvitz, in Waveguide Handbook (McGraw-Hill, New York, 1951).

689

Surface-plasmon-polariton enhanced tunneling of THz radiation through arrays of sub-wavelength apertures J. Gomez Rivas, C. Janke, P. Haring Bolivar and H. Kurz Institut fiir Halbleitertechnik, RWTH Aachen, D-52056 Aachen, Germany E-Mail: [email protected] Abstract. Terahertz transmission through periodic arrays of sub-wavelength apertures made from highly doped silicon is analyzed. Sharp resonances with enhanced transmission (up to 70% of the amplitude incident on the apertures is transmitted) are observed. Recently, intriguing measurements reported enhanced optical transmission through thin-film metal gratings with sub-wavelength holes [1]. This anomalous transmission is a consequence of a grating assisted generation of surface plasmon polaritons (SPPs) [2]. Such an enhanced transmission can find various applications in fields such as near-field microscopy, photolithography and high-density optical data storage. At THz fi*equencies these gratings could be used as frequency selective filters for high-throughput biomolecular sensing or as efficient subwavelength imaging apertures. To date most experiments on enhanced transmission through gratings of subwavelength holes have been done at optical frequencies. Studies of the THz transmission through subwavelength gratings have only recently been reported on gold grating structures [3]. However, these results could be explained using solely classical diffi*action theory. Here, we report on time-resolved analysis of the THz propagation through arrays of sub-wavelength holes in highly doped semiconductors, extending earlier work [4,5]. For optimum generation and propagation of SPPs an interface between a dielectric and an ideal metal is needed. An ideal metal is a material with a permittivity 8(co)= £r((o) + i ei((o), where 8r(co) is negative and -8r(co)» 8i(co). At THz frequencies the real part of the permittivity of metals is large and negative, but -er(co) « ei(co). For instance, gold has a -£r(co) / 8i((o) ~ 0.16 at 00/271 = ITHz; whereas at optical frequencies -8r(co) / 8i((o) ~ 13 at X, == 1pm. A small ratio between real and imaginary parts of the permittivity is equivalent to a poor generation of SPPs and leads to a low enhancement of the SPPs assisted transmission through gratings [6]. In the following we demonstrate that doped semiconductors are a better material choice to realize SPP tunneling structures. The samples presented here, consist of highly doped silicon (S-doped at N=10^^cm'^) with 100 ]xm thickness. Cuts with a depth of 50]im, i.e., one half the wafer thickness, and a separation of ao are made on both sides of the wafer. The cuts at one side are perpendicular to the ones on the other side, providing a square grating of 70 x 70 |im apertures with a lattice constant ao in the x and y directions. Fig. 1 shows a scanning electron microscope image of such a grating. We measure the zero-order transmission, i.e., the transmission of a wave incident normal to the sample surface. The size of the apertures is constant for all samples; the lattice constant ao is varied from 200 pm to 400 ]xm. As will later be experimentally demonstrated these structures deliver resonant transmission from 0.62 to 1.15

690

THz. These frequencies correspond to vacuum wavelengths of 485 and 260 ]im respectively, i.e., wavelengths 6.5 and 3.5 times larger than the aperture dimensions, demonstrating sub-wavelength THz transmission. The choice of doped semiconductors for the excitation of SPPs at THz frequencies is justified by their permittivity, corresponding to 8r= -18.1 and 8i= 91.8 for n-doped Si with N^IO ^ cm"^. The relatively large value of 8i reflects its low electron mobility ]i. However, it is important to note that in spite of this, the ratio -£r /ei of doped Si (-er /8i = 0.2) is higher than the one of metals in this frequency range.

2

3 4 5 Time delay (ps)

Fig 1. Field anplitude of the zero-order transmission through a square apertures array with dimension 70pm x 70|im and a lattice period of ao = 400]im (straight line, magnified by factor 50) and reference transmission through air (dashed line). Inset: SEM image of a THz subwavelength hole array with a lattice constant of ao == SOOpm. The transmission through the silicon hole gratings is measured using standard femtosecond laser based time domain THz spectroscopy. A typical THz pulse transmitted through air is plotted with a dashed line in Fig. 1. The transmission through a grating with ao= 400 jiim is plotted as a continuous line, indicating a large pulse dispersion associated with spectral resonances. In the following, analysis is performed in frequency domain by Fourier transformation. Measurements of transmission amplitude spectra through three Si gratings, normalized by the aperture filling fraction are plotted in Fig. 2. The solid line corresponds to a grating with a lattice constant ao= 200 ]im, the dashed line to ao= 300 ]im and the dotted to ao= 400 jim. The black arrows indicate the first long wave resonance. In all the measurements the polarization of the THz beam was linear an along the xdirection. Due to the procedure employed to define the grating in the Si wafers, the samples are asymmetric along their x and y directions. However, we checked the reproducibility of the measurements for different orientations of the sample relative the polarization of the THz radiation and no differences in the resonance wavelength nor in the transmitted amplitude are observed within the experimental accuracy. As evident from Fig. 2, the normalized transmission decreases as the lattice constant decreases or as the hole density increases. Interestingly, this

691

behavior is opposite to the one observed at optical frequencies in metal films [6]. The resonance wavelength shifts to larger values as the lattice constant ao is increased. As can be appreciated in the right part of Fig. 2 the shift of the resonance is linear with ao. This linear dependence is explained by the excitation of SPPs on squares lattices [1,4-6] 0.8 !


••"'

r

'

••

1

-«'•

T — ' " •

1

c

o t) b 2 0.6 h CO

O

r c

QJ C

o »^ CO CO

CD

£ C/5 C

0.4

O

sz

o

0.2

. 1 . 1 . 1 . 0.0 200 300 400 500 Wavelength (|im)

1

^"^m^e

c o n m t , a^^^

Fig. 2: Left: Transmission amplitude spectra normalized by the aperture areal filling fraction. The solid line corresponds to a grating with lattice constant ao = 200]Lim, the dashed line to ao = 300]Lim and the dotted line to ao = 400)Lim. The black arrows indicate the first (1,0) resonance. Right: Dependence of the resonance wavelength on the grating lattice constant. The squares correspond to the (1,0) resonance. The line is a linear fit to the data, where the slope indicates the effective refractive index of the gratings In conclusion, we measured the enhanced transmission of THz radiation through gratings of sub-wavelength apertures in highly doped silicon. This anomalous transmission is attributed to the resonant tunneling of surface plasmons polaritons. The resonance wavelength dependence with the grating lattice constant is consistent with previous measurements at optical frequencies and with a simple analytic model. This observation paves the way for using sub wavelength apertures for the development of enhanced THz biosensing devices and subwavelength THz imaging systems. We gratefiilly acknowledge support by the European Union under the project Interaction and by the Deutsche Forschungsgemeinschaft. 1. T.W. Ebbesen, H.J. Lezec, H.F. Ghaemi, T. Thio, and P.A. Wolff, Nature 391, 667 (1998). 2. L. Martin-Moreno, F.J. Garcia-Vidal, H.J. Lezec, K.M. Pellerin, T. Thio, J.B. Pendry, and T.W. Ebbesen, Phys. Rev. Lett. 86, 1114 (2001). 3. Filin, M. Stowe, and R. Kersting, Opt. Lett. 26, 2008 (2001). 4. J. Gomez Rivas, C. Schotsch, P. Haring Bolivar, and H. Kurz, Phys. Rev B 68, 201306(R) (2003). 5. C. Janke, J. Gomez Rivas, L. Beckmann, C. Schotsch, P. Haring Bolivar, and H. Kurz, Phys. Rev. B. 69, 205314 (2004). 6. T. Thio, H.F. Ghaemi, H.J. Lezec, P.A. Wolff, and T.W. Ebbesen, J. Opt. Soc. Am B 16, 1743-1748 (1999).

692

Terahertz Access to the Nanoworld R. Kersting^ H.-T. Chen^ N. Karpowicz^ and G.C. Cho^ ^ Department of Physics, University of Munich, Amalienstr. 54, 80799 Munich, Germany E-mail: [email protected] ^Department of Physics, Rensselaer Polytechnic Institute, Troy, NY 12180, USA ^ IMRA America, 1044 Woodridge Avenue, Ann Arbor, Michigan 48105, USA Abstract. We report on time-resolved terahertz near-field microscopy with a spatial resolution of 150 nm. Terahertz time-domain experiments identify a novel imaging mechanism, which allows for nanoscale THz studies of organic and inorganic materials.

1. Introduction Some of the most attractive application areas for terahertz (THz) imaging techniques are in the biosciences and in semiconductor technology. Hov^ever, many of these opportunities require a microscopic spatial resolution because, for instance, biological cells themselves can be of the order of 10 |im. Similar limitations are set by today's semiconductor technology, which currently enters the nanoworld with device dimensions of less than 100 nm. Up to now, THz imaging of such small objects has been out of reach, because Rayleigh's criterion restricts spatial resolutions in far-field imaging to the wavelength scale, i.e. to about 300 [xm for a frequency of 1 THz. Recently, we developed an apertureless THz-microscope. Spatial resolutions of 150 nm allow for THz-spectroscopy of nanoscale objects [1].

Incident THz Pulse

Tungsten J\Q .

Specular Reflection^

Fig. 1, Left side: Schematic of the head of the THz-SNOM. Right side: Terahertz image of a metallic grating having lines of 10 )Lim width. The subwavelength resolution demonstrates that the near-field under the tip is imaged.

693

2. Experimental Methods In our apertureless THz-SNOM, spatial resolution is achieved by scanning the object with a tungsten tip (Fig. 1). The images map the dielectric properties of the surface, because the specular reflection of the THz pulse depends on the induced dynamic dipole of the tip-surface system. Depending on the desired resolution, we use tungsten probes that have a tip-radius of either 100 nm or 1 |im. In our experiments, the THz pulses are generated by femtosecond laser excitation of InAs and have a center frequency of about 2.0 THz. After transmission through the microscope head, the THz pulses are time-resolved by electro-optic sampling [2].

3. Results and Discussion In order to deduce the spatial resolution of the THz-SNOM, we fabricated organic and inorganic structures of microscopic size. They consist either of metallic grating lines on the top of silicon, or of 350 nm high grating lines, which were made of the organic compound cresol resin. Figure 1 shows a typical THz image of a metallic grating, which was recorded with an acquisition time of about 10 minutes. High-resolution measurements revealed a spatial resolution of about 150 nm. On metal structures, the image contrast is typically of about 0.5 %. In case of organic structures, the changes of the surface permittivities are smaller and thus the image contrast. We deduced a minimum detectable volume of about 1.4x10'^^ m^ from the product of resolution and thickness of the organic film. Time-resolved experiments give direct insight into the physical mechanism that allows for THz near-field imaging, because electro-optic sampling provides amplitude and phase information of the THz signal that forms the image. From data such as shown in Fig. 2, we deduce the following properties of the imaging process: i) The image contrast of the specular signal is of the order of 0.5%. ii) Higher field strengths of the specularly reflected pulses are detected when the tip approaches the surface, iii) The signal consists mostly of lower frequency components, iv) The optical phase of the differential signal is zero with respect to the incident THz pulse. All four findings contradict the Mie scattering model [3] that is commonly used for the interpretation of image data. In order to interpret THz-image data, we developed a model, which is based on a nearly resonant AC coupling of the tip-surface system [4]. The background is that the system has a resonance frequency of u -M-sITc - Thus, the resonance depends on the surface permittivity. This results in a decrease of the system's radiation dissipation when the permittivity under the tip increases. For the model calculations we deduced an inductance of 3 nH for a tip with 1 |im radius. The capacitance between tip and surface was measured to change between 100 aF and 600 aF depending on the tip-surface distance. The results of our model calculations agree with the experimental data in terms of i) the enormous signal intensities, ii) the dependence on the dielectric constant of the surface, iii) the spectral dependence as shown in Fig. 2b), and iv) the optical phase, which is nearly zero according to our model.

694

- 1 0 1 Time delay (ps)

2

1 2 3 Frequency (THz)

4

Fig. 2. a) Terahertz transients, which were recorded on a gold surface. Shown are the incident THz pulse (solid line) and the differential signal (symbols), which is the difference between the field transients obtained at tip-surface distances of 0.1 |Lim and 1.6 \im, respectively, b) Corresponding amplitude spectra, which show that the image signal mostly consists of low-frequency components.

4. Conclusions In summary, we developed an apertureless SNOM for terahertz imaging of organic and inorganic structures. Spatial resolutions as small as 150 nm were demonstrated. We identified a new mechanism that allows for near-field imaging in the far-infrared. The AC coupling of the tip-surface system is distinguished from Mie scattering by its unique properties such as spectral dependence and optical phase. Applications may arise in particular in the field of semiconductor device inspection, where our technique may allow to map ultrafast charge carrier dynamics in submicron devices.

References 1 2 3 4

H.-T. Chen, R. Kersting, and G.C. Cho, Appl. Phys. Lett. 83, 3009 (2003). Q. Wu and X. -C. Zhang, Appl Phys. Lett. 68, 2924 (1996). M. Ohtsu (Ed.), Near-field nano/atom optics and technology. Springer, 1998. H.-T. Chen, S. Kraatz, R. Kersting, and G.C. Cho, submitted to Phys. Rev. Lett. (2004).

695

Terahertz surface plasmon polariton coupling on metallic grating structures John F. O'Hara, Richard D. Averitt, aiid Antoinette J. Taylor Materials Science and Teclinology Division, Los Alamos National Laboratory, MS K764, Los Alamos, NM 87545 Abstract We report efficient coupling and decoupling of free-space terahertz radiation to surface plasmon polaritons on various metallic gratings.

1.

Introduction

Recent interest in optical sinface plasmon polaritons (SPPs) has been based on the role they play in transmission through sub-wavelength aperture arra) s, tliin film guided-wave spectroscopy, and non-linear interactions via surface field enliancement [1]. SPPs have already been studied in detail at other frequencies including microwave [2], infrared [3], and more recently THz [4-6]. Due to their coherent nature, THz-TDS (time-domain spectroscopy) systems are ideally suited for phase-sensitive, broadband SPP measurements, wliich were not previously possible. And due to tlie wavelength of THz radiation, SPP research can be scaled to a more convenient size. This simplifies fabrication of guiding and coupling structures and helps elucidate basic SPP behavior on complex surfaces. Coupling and decoupling of SPPs can occur \ia several metliods including attenuated total reflection (ATR), edge excitation by either prisms or apertures, and grating coupling [7]. In tliis work, we use THz-TDS to provide insight about the behavior and efficiency of grating couplers. (a) ^ '

Grating on rotation stage

(b)

(c)

Fig. 1. Diagram of (a) reflection-mode THz system, (b) brass rod grating, large arrow shows grating rotation direction, laige circle describes hole in Lexan holder, small circle describes THz spot size, (c) aluminum plate grating, circle describes THz spot size.

2.

Experiment

Tlie experimental setup, shown in Fig. la, is a reciprocal THz-TDS system modified for reflection-mode measurements. Tlie p-polarized THz pulses travel

696

from the emitter to a paraboloidai collimating mirror, pass tlirough a Teflon lens. aiid are focused to a planar-wavefront spot witii a lie amplitude diameter of -5 mm on the grating. Tlie bistatic angle between tlie emitter and detector assemblies is approximately 37 degrees. To examine polarization effects, tlie gratings were rotated about the axis normal to the grating surface, as shown in Fig. lb.

3,

Results

In Fig. 2a we show the reflected signals from an array of 0.5 mm diameter brass rods. The rods were moimted in contact with each other over a hole in a Lexan plate, as shown in Fig. lb. The scans were recorded while the axes of the rods were parallel (thin curve) and perpendicular (tliick curve) with the plane of incidence. Only the peipendicular grating configuration permits coupling between tlie p-polarized incident radiation and the SPPs by using the grating's periodicit}^ to match the momenta of the free-space radiation and the SPP [8].

-5 0 5 10 15 20 25 30 350 Time Delay (ps)

5 10 15 20 25 30 35 Hme Delay (ps)

0.5 1.0 Frequency (THz)

0.5 1.0 Frequency (THz)

Fig. 2. Signals and spectra: (ax) reflection from 0.5 mm rod grating, (b,d) reflection from 0.5 mm grooved plate grating. For (a) and (b), the tliiii and tliick curves correspond to parallel and perpendicular grating configurations, respectively. Signals sliifted horizontally for clarity. Window boundaries for spectra in (c) and (d) are indicated by the solid vertical lines in (a) and (b). In (c) and (d), tlie Hiick dashed curves are reference spectra from flat mirror targets, \^hile the thin dashed curves correspond to parallel grating configuration. The solid curves are sj^ctra of the windowed signals tbr the perpendicular grating configuration. Labels Wl and W2 designate the spech'a of the Window 1 and Window 2 regions labeled in (a) and (b). The sharp peaks of the W2 spectra occur at: (c) 450 GHz, (d) 455 GHz. All spectra were normalized \^itli the same value.

Figure 2c shows the energ}^ spectra obtained from tliese data and a reference waveform collected with a flat plate target. The spectrum for tlie parallel grating configuration shows a -20% loss at 450 GHz, which is predominately due to scattering. In tlie perpendicular configiu-ation there exists a sharp peak feature at 450 GHz in tlie spectrum of the ringing (Window 2) portion of the data. We believe this peak corresponds to the coupling and propagation of a SPP. The peak indicates that about 57% of the total 450 GHz energy incident upon the gmting is returned to tlie detector, but delayed in time. Nonnalizing tliis to tlie energ}^ retimi from the parallel configuration, we estimate that about 70% of the energ}^ not scattered by or transmitted tlirough the grating is coupled into a SPP mode. The limited duration of the ringing in Fig. 2a indicates that the SPPs are coupled to the

697

grating, travel a short distance along it (<8nmi), and are quickly decoupled. This ver>^ rapid coupling and decoupling of tlie SPPs would obviously limit tlie utilit} of tliis grating metliod m terms of launcliing usable SPPs on guiding stnictures. Though not shown, veiy similar data were found in transmission mode. We found 1-2% energ>^ transmission tlirough the rod grating at tlie SPP frequency. Otlier reflection data were collected for the target shown in Figure Ic: an aluminmn plate with v-shaped grooves. The grooves were about 0.25 nmi deep and were separated by 0.5 nmi. As shown in Fig. 2b, tliis structiu*e produced ver} similar data to the 0.5 mm rods. The main difference is the shape of the initial peak m the time-domain data. This is due to tlie fact that tlie cur\ ed surfaces of tlie rods cause multiple scattering events tliat are detectable from each rod, whereas tlie grooved plate largely pemiits only a single detectable scattering event from each ridge. The coupling efficiency of the SPP on this grating is about 63%, when normalized to the parallel configuration data. Tlie overall diminished signal strengtli is likely due to the imperfect macliining of the grooves and the fact that grooves cause additional scattering losses. Clearly, transmission is not possible in tliis gratmg. Simple calculations of dispersion and gmting diffi-action reveal that coupling to SPPs in tliese reflection arrangements should occur via the m = -1 diffraction order at 455 GHz, assuming tlie complex permittivity of the metal is that of gold and tliat the bistatic angle of the system is 37°. Physical imperfections in our gratings and their current mounting structiu-es forbid us from verify ing a perfect match to tlieoty. Nevertheless, witliin our measurement tolerances, our data exhibit good agreement with initial theoretical predictions. Additional measurements witli smaller tolerances are planned to remove this ambiguit>.

4.

Conclusions

Reflection and transmission-mode THz-TDS was used to excite siuface plasmon polaritons on various metallic gratings. Coupling and decoupling between freespace radiation and SPPs was found to be -63-70% efficient and SPP propagation distances were short (<8 imn) due to rapid decoupling. A small portion of energ}^ (1-2% at the SPP frequency) was found to transmit through the rod gratings. Our resuhs are in good agreement with tlieoretical grating-coupled SPP predictions.

References 1 2 3 4 5

W. Barnes, A. Dereiix and T. Ebbesen, Nature 424, 824, 2003. H. Barlow and A. Ciillen, Proc. I.E.E. 100, 329, 1953. J. Schoenwald, E. Biirstein and J. Elson, Solid State Coniinun. 12, 185, 1973. D. Begley, R. Alexander, C. Ward, R. Miller aiid R. Bell, Surf Sci. 81,245, 1979. J. Saxier, J. Gomez Rivas, C. Jaiike, H. Pellemans, P. Hariiig Bolivar and EI. Kurz. Phys. Rev B. 69, 155427-1, 2004. 6 D. Qu, D. Grischkowsky and W. Zhang, Opt. Lett. 29, 896,2004. 7 V. Agronovich and D. Mills, Surface Polaritons, North-Holland, New York, 1982. 8 H. Raether, Surface Plasmons on Smooth and Rough Surfaces and Gratings, Springer-Verlag, New York, 1988.

698

Control of THz transmission througli twodimensional metallic photonic crystals Ci-Ling Pan^ Cho-Fan Hsieh^ and Ru-Pin Pan^ Masaki Tanaka^, Fumiaki Miyamaru^, Masahiko Tani^, and Masanori Hangyo'^ ^ Institute of Electro-Optical Engineering and Departments of Photonics and Electrophysics, National Chiao Tung University, 1001, Ta-Hsueh Rd, Hsinchu, Taiwan 300, RO.C. E-mail: [email protected]^v ^ Institute of Laser Engineering, Osaka University, 2-1 Yamadaoka, Suita, Osaka 565-0871, Japan Abstract. We demonstrate frequency tuning of resonantly enhanced THz radiation transmitted through a two-dimensional metallic photonic crystal on a liquid cr}^stalline "substrate".

!• Introduction Metal films with two-dimensional periodic arrays or metallic photonic crystals {2DMPC), can exhibit extraordinary optical transmission characteristics [1]. Enhanced THz emissions through 2D-MPC [2] were recently reported. This development could lead to the realization of ultra-sensitive THz sensing and imaging systems. Control of THz emission through 2D-MPC is also expected to add new flmctionalities to future THz photonic crystal devices. For the 2D-MPC, frequency of the well-defined peak in the transmission spectra is determined by the geometry, both of the 2D-MPC and the incident light, as well as the refi-active index of the adjacent dielectric media,

where c is the speed of light in vacuum, k_^ is the in-plane wave vector of the incident light, £^and ^ are the reciprocal lattice vectors of the periodic structure. s^ and £d are the dielectric constants of the metal and the adjacent dielectric media, respectively. For THz waves, i^^nl^^-^d, Eq. (1) becomes VR~ Vspp/na where Vspp is the surface plasmon-polariton (SPP) resonance frequency in the absence of the dielectric and {n^^ ~ s^ and n^ is the index of refraction of the dielectric. It is then possible to tune the SPP resonance frequency monotonically by varying n^. We have recently shown that the 4'-/^pentyl-4-cyanobiphenyl or 5CB, a nematic liquid crystal (NLC), exhibits large birefringence and negligible absorption in the THz frequency range [3]. A magnetic-field tunable liquid crystal THz phase shifter was realized [4]. In this work, we demonstrate for the first time, control of the

699

transmission of the THz wave by using a 2D-MPC on the liquid crystalline "substrate".

2. Experimental Methods The 2D-MPC sample was a 0.5 mm-thick aluminum plate perforated with circular holes in a triangular lattice of lattice constant, s = 0.99 mm and the diameter of the circular hole, J = 0.56 mm. The 2D-MPC was sandwiched by a pair of Mylar sheets (75 j^m in thickness) and machined on the one side so that a boxlike structure of dimension 15.Ox 15.0x0.2 mm^ in the center of the 2D-MPC is formed for holding the NLC (5CB, Aldrich). A pair of horizontally arranged peimanent magnets is used to align the liquid crystal molecules. By rotating the magnet assembly at an angle 9 with respect to the propagation direction of the incident THz beam, the effective refractive index, n^j , of the liquid crystal infiltrated in and on top of the 2D-MPC is changed from n^ = 1.75 to no == 1.62, COS ^ 0

sin ^ ^

^.#

(2)

We use conventional THz time-domain spectroscopy for characterizing the transmission of the 2D-MPC. Photoconductive antennas fabricated out of LTGaAs are used for generation and detection of THz generation with a femtosecond Ti:Sapphire laser for optical excitation^

3. Results and Discussion The spectral transmittances of the 2D-MPC itself and 2D-MPC with liquid crystals of several orientations are shown in Fig. 1. The frequency of the main transmission peak is red-shifted from that of the bare 2D-MPC (- 0.3 THz) to 0.192 THz and 0.184 THz for 5CB aligned perpendicular and parallel to the polarization of the incident THz wave, respectively. Using the indices of refraction of 5CB, we estimate that Vc = 1.841c/7rd - 0.2 (0.183) THz, while Vdiff =2c/sv3 - 0.216 (0.2) THz for the perpendicular (parallel) geometries respectively. Varying the effective refractive indices of 5CB, neff, we can tune the peak transmission frequency of the 2D-MPC. It is shown in the inset that VR increases linearly with the inverse of neff, in good agreement with the prediction of Eq. [1]. As we change from the perpendicular to the parallel geometry, the peak amplitude transmission (see Fig. 1) increases from -- 0.6 to -- 0.7. Considering the porosity (fractional area of the THz beam occupied by the holes) of the tunable 2D-MPC, -- 0.3, we observe an enhancement factor of over two compared to 2.76 for the bare 2D-MPC. The comparatively smaller enhancement factor than the visible and near IR can be understood by (1) a smaller value of the ratio of the real and imaginary parts of permittivity of metal at THz frequencies; and (2) the diameters of the holes are not substantially subwavelength. Higher transmission for the parallel geometry could be

700

understood as the matching of SPP energies on either side of the 2D-MPC, as ne 1.75 for 5CB is closer to that of Mylar (n --1.7) than no == 1.62.

Fig. 1. Spectra transmitted through the MPC with the effective indices of the LC molecules tuned by rotating the magnets. The peak transmission frequencies are plotted versus 1/ ngg-in the inset. The O'^ magnetic field v^ould block the THz wave, a equivalent vertical magnetic field was used instead..

4. Conclusions In summary, we demonstrate control of the transmission of the THz wave through a 2D-MPC. The device structure consists of a 2D-MPC infiltrated with liquid crystals (5CB) and also on the liquid crystalline "substrate". The peak transmission fi-equency can be tuned from 0.192 THz to 0.184 THz for 5CB aligned perpendicular (ordinary index, no-1.62) and parallel (extraordinary index, ne=1.75) to the polarization of the incident THz wave. Because of its extreme sensitivity to the optical constants of the material infiltrated into and surrounding the 2D-MPC, the present device can also be considered as the prototype of a SPP sensor for the THz frequency range. Acknowledgements This work was performed while Ci-Ling Pan was on sabbatical and Cho-Fan Hsieh was a research student at Osaka University.

References 1 T. W. Ebbesen, H. J. Lezec, H. F. Ghaemi, T Thio, and P. A. Wolff, Nature (London) 351,667, 1998. 2 F. Miyamaru, M. Hang>^o, Appl. Phys. Lett. 84,2742,2004. 3 T.-R. Tsai, C.-Y. Chen, C.-L. Pan, R.-P. Pan, and X.-C. Zhang, Appl. Opt. 42, 2372, 2003. 4 C. Y. Chen, T. R. Tsai, C. L. Pan, and R P. Pan, Appl. Phys. Lett. 83,4497,2003.

701

Teflon photonic crystal fiber as polarizationpreserving waveguide in THz region M. Goto, A, Quema, H. Takahashi, S. Ono, and N. Sarukura Institute for Molecular Science, 38 Nishigonaka, Myodaiji, Okazaki 444-8585, Japan E-mail: [email protected] Abstract. The construction of long and non-polarization changing photonic fiber waveguide was demonstrated using highly flexible plastic materials. Due to its relatively low-loss coefficient, the possibility of preparing longer photonic fiber waveguide can be easily attained.

1.

Introduction

The progress of THz spectroscopy in recent years has been enormous and several THz-radiation emitters have been reported as a result of the development of ultrafast optical pulses [1,2]. However, the current THz spectroscopic technique utilizes mainly free space propagation of this far-infrared electromagnetic (EM) wave and it is to some extent difficult to control and guide. To overcome this difficulty, it is thus necessary to develop and utilize waveguides that are transparent in the THz region [3]. Photonic crystal fiber (PCF) is very attractive from the scientific and technological point of view due mainly to its single-mode propagating characteristic. Silica is commonly used in fabricating PCF. However, this material is not transparent in the THz region. As such, alternative material like plastic, which is transparent in the THz region and is flexible, is a good candidate for the construction of PCF. At present, most plastic THz waveguide uses highdensity polyethylene. In this study, we propose an alternative PCF material for THz waveguide. Utilizing polytetrafluorethylene (commonly known as teflon), which is a readily and commercially available material, highly flexible and highly polarization-preserving PCFs are assembled.

2.

Experimental Methods

THz radiation from an undoped bulk InAs with (100) surface immersed in a magnetic field provided by a 2.5-T pennanent magnet and excited by a modelocked Ti: Sapphire laser with an average power of 1-W was used. The excitation laser delivered 100-fs optical pulses with center wavelength of 800 nm at a repetition rate of 82-MHz. The emitted THz radiation was collimated and focused using paraboloidal mirrors. A silicon hyper-hemispherical lens was used to focus the THz beam into the waveguides and a liquid-He cooled Si-bolometer was used for detecting the THz radiation. The PCFs were constructed using teflon rods as core and teflon tubes for the periodic cladding. The periodic cladding mounted in

702

the samples had two layers, wherein the first and second layers had 8 and 16 tefion tubes, respectively. The diameter of the core used was 1 mm while the clad had an outer diameter of 0.75 mm and an inner diameter of 0.25 mm. For comparison purposes, a metal tube (inner diameter of 0.25 mm, outer diameter of 5 mm) and a solid teflon rod whose diameter is the same as that of the PCF was also used.

3.

Results and Discussion

Figure 1(a) shows the absorption coefficient plot of a 5 cm solid teflon rod. It can be seen that teflon has low absorption for the THz radiation with an absorption coefficient of approximately 0.3 cm"^ at 1 THz. Teflon is chosen for this study due to the fact that reports have shown that this material is highly transparent in the THz frequency region [4]. A photograph of the constructed PCFs is shown in Fig. 1(b). The upper right-hand comer inset shows the cross-section of the PCF. The cladding layers and the core are fused together without any furnace heating. Holding the 1 ^^ cladding layer and the core together is ordinary plastic tube, which is placed between the 1^^ and 2"^ cladding layers. Another thin plastic tube is placed on the outer portion to envelop the whole PCF thereby fusing together the core, 1 ^^ and 2" layers into one fiber. To test the confinement of the incident THz wave inside and to confirm that the wave propagation is mainly in the core of the constructed PCF, knife-edge measurements at the exit point are conducted. The intensity profile of the beam in one of the PCFs is shown in the lower left-hand comer of Fig. 1(b). From the plot shown, it can be seen that the data is well described via Gaussian fitting. This confirms that almost all the THz radiation propagates in the core. f~" 0.4

1

1

.s u

0-3

A^^l^^^

1 |..a c

}"•'

/^

•

^ ^ ^ ^ ^ ^ ^ ^ " ^ ^ ^

\K

r

^

i

•

1

:

Frequency (THz)

Fig. 1. (a) Absorption coefficient versus frequency plot of the 5 cm solid teflon rod with a diameter similar to that of the PCF. (b) Photograph showing the PCFs constructed. Upper right-hand comer inset shows the cross-section of the PCF while the lower left-hand inset shows the intensity profile of the emitted radiation. Figure 2(a) shows the length dependence of the transmitted intensity through the waveguides. It can be inferred from Fig. 2(a) that the intensity is fitted by an exponential function of length. Here, it can be deduced that at the PCF-hyperhemispherical lens interface 10% of the total radiation propagates into the PCF while 30% transmits through the metal tube. From the best-fit lines, the loss coefficient for the PCF and the metal are determined to be about 0.12 cm"^ and 0.10 cm"^ respectively. Based on the relatively low loss coefficient of the PCFs,

703

t Incident THz power

ISa

10.1 L r • \ O

1 1 Vf7|jJ?v'' 1

metal tube PCF

0

5

-

^

0.6 1-

:

1

'^^ 10 Length (cm)

I I

15

I

•

\ -3-PCF

^ %

0.2 10

20 30 40 SO 60 70 80 Polarization Angle (degress)

90

Fig. 2. (a) Length dependence of the transmitted intensity through the waveguides. The dotted line indicates the incident THz radiation while the solid and dashed lines denote the best-fit lines for the metal and PCFs, respectively. Inset shows the spectra of the PCFs with various lengths, (b) Polarization angle dependence of various waveguides. the possibility of constructing long PCFs for potential THz v^aveguide can be easily attained. The inset shows the spectra of the PCFs with various lengths. The reference is the based on the emitted THz pulses from a single hyperhemispherical lens placed at the beam waist of the radiation. High frequency cutoff is observed for each PCF and such cut-off shifts toward low frequency with increasing PCF length. By placing a wire grid polarizer in front of the bolometer and rotating it at increments of 5 degrees, the polarization angle dependence of the various waveguides are determined. As seen in Fig. 2(b), the metal tube is found to be non-polarization preserving. This is possibly due to multiple reflections inside the metal tube, which tends to rotate the wave polarization. On the other hand, all of the constructed PCF show good polarization conservation. Considering the characteristics of the PCFs, i.e. intensity profile well fitted via Gaussian function with high polarization preserving quality, it is likely that the PCFs have singlemode propagation. To confirm this, thorough mode propagation analysis is required.

4.

Summary

In summary, we have demonstrated the construction of reasonably long, easilyprepared and non-polarization changing photonic fiber waveguide using readily/commercially available and highly flexible material. Based on the relatively low loss coefficient obtained from the PCFs, the possibility of constructing long PCFs for potential THz waveguide can be easily achieved. Modifications are now being done to improve the efficiency of the PCF,

References 1 2 3 4

704

D. H. Auston, K. P. Cheung, and P. R. Smith, Appl. Phys. Lett. 45, 284, 1984. K. Kawase, M. Sato, T. Taniuchi, and H. Ito, Appl. Phys. Lett. 68, 2483, 1996. H. Han, H. Park, M. Cho, and J. Kim, Appl. Phys. Lett. 80, 2634, 2002. J. Birch, J. Droney and J. Lesurf, Infrared Phys. 21, 225, 1981.

Generation of coherent tunable THz waves by using birefringent crystal and grating pair Ryuzi Yano\ Hideki Gotoh^ and Toshiaki Hattori^ ^ NTT Basic Research Laboratories, NTT Corporation, 3-1, Morinosato-Wakamiya, Atsugishi, Kanagawa, 243-0198, Japan E-mail: [email protected] ^ Institute of Applied Physics, University of Tsukuba, Tsukuba, 305-8573, Japan E-mail: [email protected] Abstract. Laser pulses with temporally sinusoidal intensity modulation excited a photoconductive antenna to produce coherent tunable THz waves. The laser pulses with sinusoidal intensity modulation were generated by modulating a femtosecond laser pulses with a birefringent crystal and a grating pair. The carrier-envelope phase (CEP) of the THz waves created by this method was unaffected by the instability of the optics.

1. Introduction Coherent tunable THz electromagnetic sources are useful for the studies of spectroscopy, pulse propagation phenomena, and coherent phenomena, including the coherent control of materials [1]. Coherent tunable THz electromagnetic waves can be generated by exciting THz wave emitters with intensity-modulated laser pulses. The intensity-modulated laser pulses are created by modulating a femtosecond laser pulse with a Michelson interferometer and a grating pair [2]. The THz waves thus generated have a stable frequency. However, if the optics is unstable or the optical table vibrates, the carrier-envelope phase (CEP) of the THz waves becomes unstable. Here, we propose a method of generating coherent tunable THz waves that uses a birefringent crystal instead of a Michelson interferometer. This method is basically the same as the method mentioned above. However, it is unaffected by the instability caused by the vibration of the optical tables, because a pulse pair is produced within the crystal. Therefore, the CEP of the THz waves is stable.

2. Coherent tunable THz wave generation To obtain a sinusoidally intensity-modulated laser pulse, a linearly polarized laser pulse is first guided to a grating pair to produce a chirped pulse. The chirped pulse is then guided to a birefringent crystal that has two principal axes with wavelengthindependent refractive indices HQ and rie (rie > rio). The polarization angle of the linearly polarized laser pulse is set to 45 degrees with respect to both principal axes of the refractive indices of the crystal.

705

When the chirped laser pulse transmits through the birefringent crystal, it splits into two pulses with the time separation t^ given by t^ = {nf.-no)Llc, where L is the length of the crystal and c is the speed of light. As the pulse pair is produced in the crystal, their time separation t^ is fixed. The chirped pulse pair produces a laser pulse with a sinusoidal intensity modulation. The stability of the CEP of the THz wave depends on the stability of the time separation t^. Therefore, the THz waves generated by this method have a fixed CEP. The crystal used in the experiment was YVO4, which has a tetragonal structure. It is birefringent and transparent in the near-infrared wavelength region. The refractive indices are roughly wavelength independent and n^- rio- 0.21. The output of a mode-locked Ti:Al203 laser (a pulse width of -150 fs, a repetition rate of 100 MHz, laser wavelength set to 810 nm) was divided by a beam splitter into pump and gate pulses. A laser pulse with sinusoidal intensity modulation was generated by modulating the pump pulse with the YVO4 crystal and a grating pair (1800 lines/mm goldcoated holographic grating). This laser pulse was focused by an objective lens onto a PC-antenna (emitter) fabricated on low-temperature grown (LT-) GaAs to produce a frequency-tunable THz wave. The gate pulses were guided and focused onto another PC-antenna (receiver) by another objective lens. An optical chopper modulated the pump beam with a frequency of 1.3 kHz. The current in the receiver was amplified and fed to a lockin amplifier. To obtain the temporal profiles of the THz waves, the output of the lock-in amplifier was measured as a function of the delay time between the pump and probe pulses.

3. Experimental results and discussions Figure 1(a) shows the temporal profile of the laser pulse modulated by a l.O-mm thick YVO4 crystal and a grating pair. The distance between the gratings was set at -49 mm. The laser pulse had a periodic intensity modulation, which lasted for -80 ps. There was virtually no change of the refractive indices or the crystal thickness, the time separation between the pulses was fixed. Therefore, the periodic modulation was stable.

1 3

illUi^ '"\

CO

^0.51CO

c

CD

^ ~

0 0

^

JU

20

40 60 80 Time (ps)

100

20

40 60 Time (ps)

Fig.l. (a) Intensity-modulated laser pulse and (b) THz wave.

706

80

100

Figure 1(b) shows the THz wave emitted from the PC-antenna excited by the laser pulse shown in Fig. 1(a). Since the THz wave was proportional to the temporal change of the carrier density created in the PC-antenna, the THz wave showed amplitudes with both plus and minus signs. The Fourier-transform of Fig. 1(b) showed that the THz wave has only the ~0.24 THz frequency component; the spectral component corresponding to the overall pulse width of ~30 ps observed in Fig. 1(a) is eliminated. Since the interference spectrum created by the crystal is stable, the CEP of the THz wave is stable regardless the instability of the optical table. Figure 2 shows the grating distance dependence of the center frequency (v) and spectral width (Av) of the THz wave. Since the crystal length is fixed and the distance b between the gratings is changeable, both v and Av are inversely proportional to b. The solid curves are fitting ones using this relation. The agreement between experiment and theory is satisfactory.

^

0

10 20 30 40 50 60 70 Grating distance (mm) Fig. 2. Grating distance dependence of THz wave centerfrequency(circle) and spectral width (square).

4. Conclusions By using a birefringent crystal and a grating pair, we generated coherent tunable THz waves. As the pulse pair is created in the crystal, the CEP of the THz waves is stable regardless of the instability of the optical table. The insertion of a Babinet-Soleil compensator will control the CEP of the THz waves. We consider that our method of generating THz waves with a stable CEP will be useful for nonlinear processes and coherent control of materials in the THz region.

References 1 R. Ascazubi et al, Appl. Phys. Lett. 81, 4344, 2003. 2 A. Welling and D. H. Auston, J. Opt. Soc. Am. B13, 2783, 1996.

707

Ultra-wide bandwidth THz emission from a semiconductor irradiated with intense, radiallypolarized, Bessel-Gauss pulses K. J. Chau and A. Y. Elezzabi Ultrafast Photonics and Nano-Optics Laboratory. ECERF Building. Department of Electrical and Computer Engineering. University of Alberta. Edmonton Canada T6G 2V4 E-mail: [email protected], [email protected] Abstract. THz generation from a semiconductor irradiated with a focused, high intensit) radially-polarized Bessel-Gauss pulse is modeled. When compared to a Gaussian pulse, the Bessel-Gauss pulse exerts stronger ponderomotive force on photocarriers. resulting in an order of magnitude increase in the THz emission power.

1.

Introduction

The interaction of intense laser light and plasma has attracted much attention as a source of free space THz radiation. Hamster et al. [IJ have demonstrated high power, broadband THz emission resulting from the high intensity (~10'^W/cm') photo ionization and ponderomotive acceleration of plasmas originating from gas and metallic targets. Alternatively, semiconductor electron-hole (e-h) plasmas are attractive THz sources due to the low excitation energies and low effective electron mass. Ponderomotive acceleration of e-h plasmas in a semiconductor can be achieved by producing either extreme spatial field gradients or by high fluence excitation. The damage threshold of the semiconductor restricts the latter requirement. However, the polarization state and intensity distribution of the laser beam can be manipulated to produce the spatial gradients necessaiy to achieve strong ponderomotive interaction. We model THz generation based on the ponderomotive acceleration of carriers by an intense, 50 fs radially-polarized, Bessel-Gauss, laser pulse. The advantage of these pulses over linearly-polarized, Gaussian pulses is that at the focus, extreme field gradients [2,3] can accelerate electrons ponderomotively with higher efficiency.

2.

Model

As shown in Fig. 1 (a), the radially-polarized beam is focused onto an intrinsic semiconductor slab with its surface at z= 0. The fields in the semiconductor half-space are obtained by angular summation over a Bessel-Gauss weighted amplitude function [4]. The spatial distribution of the focused (NA=0.9) radial and longitudinal intensities inside a GaAs semiconductor (62 = 13.4, a = 0.8 i.mi) surface are shown in Fig. 1 (b) for aA = 800 nm laser beam.

708

(b)

0

0,4

0.8

0

0.45 0.9

Fig. 1. (a) Schematic of the THz emitter, (b) The intensity components of the focused beam inside GaAs are depicted. When a femtosecond pulse is focused onto a semiconductor slab, the laser field generates a dense e-h plasma near the surface and accelerates carriers across the high spatial gradients. The potential responsible for this effect is 2 UAr.zj)

=-

2m * t

-V(£(r,-,/) focus • E

{r.zj)focus)

(1)

where £(r, z, t) focus ^^ ^^^ focused field, co is the fi*equency of the laser pulse, and e and w* are the carrier charge and effective mass, respectively. The current density, driven by carrier density and field gradients, is expressed as J{r,-j)

= -e jLi^n(/\zJ)

UAr.z^t) V ^(/\--./) + -

(2)

where the holes are taken to be stationary, i/A.nzj) is the electrostatic potential, and //,> and n{r,z,t) are the electron mobility and density, respectively. The emitted THz field is evaluated by the temporal derivative of the integrated current density.

3.

Results and Discussion

Fig. 2 (a) depicts the spatio-temporal evolution of the radial and longitudinal current density components,yX^.^.O and jz{r,z,t), near the surface when a GaAs slab is excited with a 50 fs pulse. In the radial direction, at / = -5 fs, the photo-excited electrons are bunched and ejected away from the high intensity radial lobe of the pulse. When r = 0 fs, ponderomotive electron confinement near r = 0 results in a high current density. After the peak of the pulse, screening forces dominate ?iX\d j,{r,z,t) is reduced. In the z-direction, the leading edge of the pulse at / = -5 fs induces a localized electron current density near z = 0. A large current density is attained at the peak of the pulse, and at t = 5 fs, screening effects diminish yX/^z,/). The THz emission using an optical fluence of 99 mJ/cm~ is depicted in Fig. 2 (b). The peak-to-peak THz field exceeds 1 V/cm, and the FWHM is less than 22 fs, corresponding to a bandwidth greater than 25 THz.

709

.A.fe'>0

(a)

J^^Kt)

(a.u.

X 10 90"

\

180"f fS<¥YYJO" 0.04

o.o:

(b)

>

(X)

-50

0

50

100

150

r (j.im) Time (fs) Fig. 2. (a) Spatio-temporal evolution ofj,{r,z,t) and jA^zj) at times - 5 . 0, and 5 fs relative to the peak of the optical pulse. The (b) radially and (c) longitudinally-polarized THz pulse components detected at D = 1 cm and ^ = 14° for a tluence value of 99 mJ/cm' is depicted. As shown in Fig. 3, for fluence values from 5 mJ/cm^ up to 100 mJ/cm^, the power of the THz emission driven by the radially-polarized Bessel-Gauss pulse is an order of magnitude greater than with a Gaussian pulse.

Fluence (mJ/cm) Fig. 3. THz emission power versus tluence using Bessel-Gauss and Gaussian pulses.

4.

Conclusions

Ponderomotive acceleration of e-h plasma by intense, radially-polarized, BesselGauss pulses results in the emission of large bandwidth, high power THz radiation. When compared to a Gaussian pulse of equivalent fluence, excitation with a Bessel Gauss pulse results in an order of magnitude increase in THz emission power.

References 1 2 3 4

710

H. Hamsteretal., Phys. Rev. Lett. 71.2725 (1993). F. Gori, G. Guattari, and C. Padovani, Opt. Commun. 64,491 (1987). R. Dom, S. Quabis, and G. Leuchs. Phys. Rev. Lett. 91, 233901 (2003). B. Richards and E. Wolf Proc. Roy. Soc. A 253, 358 (1959).

2aj

Mechanism crossover of terahertz radiation from InAs surface induced by a magnetic Held at high density excitation Makoto Nakajima\ Yuji Oda^ Shingo Saito^, and Tohru Suemoto^ ^ The Institute for Solid State Physics, The University of Tokyo, Kashiwa, 277-8581, Japan E-mail [email protected] ^ Kansai Advanced Research Center, Kobe, 651-2494, Japan Abstract. The excitation-density dependences of the terahertz radiation power and waveform from InAs surface under magnetic field of 2 T were investigated. At high density excitation, the drastic changes were observed and it is explained by the crossover of the radiation mechanism of the magnetic field induced component of the THz radiation.

1.

Introduction

Terahertz (THz) radiation from semiconductor surfaces induced by femtosecond laser pulses [1-5] attracted much attention. To develop applications using the THz wave such as an imaging and a spectroscopy, high-power THz radiation sources are required. Recently, an enhancement of the THz radiation power from InAs surface by applying a magnetic field has been reported by Sarukura et al [1], and several groups reported the radiation mechanism from InAs surface under a magnetic field [1-3,5]. The enhancement by the magnetic field is induced by Lorentz force, which modifies the direction of the photocurrent [3]. However, there is no complete understanding of the THz radiation mechanism especially under a magnetic field at high-density excitation.

2.

Experimental IVIethods

THz radiation power and temporal waveforms from undoped InAs with a (100) surface were measured under a magnetic field using a Liq.-He-cooled Si bolometer and a dipole-type low-temperature-grown GaAs photoconductive antenna, respectively. The magnetic field of ±2 T was applied parallel to the sample surface as shown in the inset of Fig. 1. Optical excitation was mainly provided by an amplified Ti:sapphire laser with a center wavelength of 800 nm, a repetition rate of 200 kHz and a pulse duration of 80 fs. We also used an amplified Ti:sapphire laser pulses with a repetition rate of 1 kHz and a unamplified Ti: sapphire laser with a repetition rate of 80 MHz to investigate the wide range excitation density dependence. The pump beam was incident to the sample surface at Brewster angle and the spot size of the pump beam was 3 mm for the power measurements and 2 mm for the waveform measurements.

711

10^ k

.0^

+ 2T 2T Pump beam

\

iTHz wave

^ 10"'

..m

>k

.X10-°

m i_i

X

or

D •

+2T -2T

XI J_ 10-^

10-'

10"

10^

Excitation Density (\x3/cm)

Fig. 1. The excitation density dependence of the THz radiation power from the InAs surface.

3.

Results and Discussion

Figure 1 shows the excitation density dependence of the THz-radiation power under magnetic fields of 0, ±2 T. The enhancement of the radiation power is observed under a magnetic field of ±2 T as reported previously [3]. The anomalous dependence was observed at ~ 1 |iJ/cm^, at which the radiation powers for B = 0 and ±2 T are crossing. In order to clarify the origin of this anomaly, we have done the THz radiation waveforms measurement. Figure 2(a) shows the excitation density dependence of the THz waveforms without magnetic field. The increase of the amplitude is observed with increasing the excitation density and no change of a phase is observed. In contrast, the observed THz waveforms under the magnetic field show not only the amplitude changes but also the phase changes on the excitation density. We extracted the magnetic field induced (B-induced) component as shown in Fig. 2 (b) by subtracting the waveform without the magnetic field (Fig. 2(a)) from the THz waveforms under the magnetic field. The excitation density dependence of the B-induced component for B = +2 T shows the polarity reversal 100 F

120 h

2.8|iJ/cm'

B-induced component

I 2.0^J/cm' 60 H l.O^J/cm

40 k 0.3^J/cm' 20 F 0.1 ^ / c m 2 3 Delay Time ^ s )

2 3 Etelay Time (ps)

Fig. 2. (a) THz waveforms without magnetic field and (b) the waveforms for the B-induced component at B = 2T.

712

at ~1.0 ^J/cm^. This excitation density corresponds to the cross-point of the power dependence at +2 T and 0 T in Fig. 1. The B-induced component for -2 T shows the identical features except for the polarity opposite to that for +2 T. We discuss the origin of the polarity reversal in Fig. 2(b) below. It was reported that the radiation mechanism of the THz radiations from InAs surface is the transient diffusion current by the photogenerated carriers [2]. The polarity of the THz waveform corresponds to the direction of the photocurrent in the plane normal to the radiation direction [4]. We confirmed by comparing the THz waveforms from the photoconductive antenna that the polarity for InAs without the magnetic field agrees with the direction of the diffusion current by the photogenerated carriers while the polarity is disagreement with the direction of the drift current. The polarity of the waveforms under the magnetic field at highdensity excitation corresponds to the direction of Lorenz force for the diffusing electrons. However, the polarity under a magnetic field at low-density excitation is opposite to that predicted from the model of the diffusion current. We take an effect of the electrons in a surface accumulation layer into consideration to explain these results. The THz radiation from the electrons in a surface accumulation layer under a magnetic field has been reported and the contribution appears only under a magnetic field [5]. The polarity of the B-induced component originating from the electrons in the accumulation layer, which is triggered by the screening of the intrinsic surface electric field, is expected to be opposite to that of the diffusion current by the photogenerated carriers, and it agrees with the observations. The excitation density at 1.0 jiJ/cm^ corresponds to the surface density 4 x 10^^ cm'^, which is close to the electron sheet density of 10^^ cm"^ in the accumulation layer. We conclude that the dominant THz radiation mechanism of the B-induced component is the electrons in the accumulation layer at low-density excitation and the diffusion current by the photogenerated electrons at high-density excitation. In summary, we investigated the excitation density dependence of the THz radiation waveforms and power from InAs surfaces imder a magnetic field of 2 T. The polarity reversal in the B-induced components was observed at high excitation density. We demonstrated that the radiation mechanisms of the B-induced component crossover from the electrons in the accumulation layer at low-density excitation to the photogenerated carriers at high-density excitation. Acknowledgements. This work has been supported by the Grant-in-Aid for Scientific Research from the Ministry of Education, Culture, Sports, Science, and Technology, Japan.

References 1 2 3 4

N. Sarukura, H. Ohtake, S. Izumida, and Z. Liu, J. Appl. Phys. 84, 654 (1998). M. B. Johnston et al, Phys. Rev. B 65, 165301 (2002). H. Takahashi et al, Appl. Phys. Lett. 83, 1068 (2003). M. Nakajima, M. Hangyo, M. Ohta, and H. Miyazaki, Phys. Rev. B 67, 195308 (2003). 5 J. N. Heyman, P. Neocleous, D. Hebert, P. A. Crowell, T. Mailer, and K. Unterrainer, Phys. Rev. B 64, 085202 (2001).

713

Transient Grating Generation and Waveform Shaping of Free-Space Propagating, Picosecond, Narrow-Band THz Radiation Andrei G. Stepanov \ Janos Hebling \ and Jiirgen Kuhl ^ ' Max-Planck-Institut fur Festkorperforschung, 70569 Stuttgart, Germany E-mail: [email protected], [email protected] ^ Research Group of Nonlinear and Quantum Optics, Hungarian Academy of Sciences, and Department of Experimental Physics, University of Pecs, H-7624, Hungary E-mail: [email protected] Abstract. We demonstrate the generation of free-space, picosecond, narrow-band, tunable THz radiation by a transient polarization grating induced by two femtosecond pulses propagating in a LiNbO, crystal. Furthermore, we show that this technique provides an attractive opportunity for a high contrast shaping of the generated THz waveform. The experimental realization of the THz shaping is quite simple. During the last few years optical rectification methods utilizing shaped ultrafast laser pulses [1] or periodically poled lithium niobate (PPLN) [2,3] for the generation and shaping of narrow band picosecond THz pulses have been developed. Here, we report on the generation of narrow-band picosecond pulses of THz radiation via optical rectification with transient polarization gratings induced by femtosecond laser pulses propagating in a bulk LiNb03 crystal. Although transient grating or two pump pulse techniques have been widely exploited for coherent phonon-polariton generation in many different solids (see for example [4]), to the best of our knowledge, this work is the first demonstration of this technique as a source of THz radiation propagating in free space. Furthermore, we show that this technique allows to tune the frequency and manipulate the temporal shape of the generated THz waveform. Experimental realization of this pulse shape control is quite easy in comparison to other techniques [1,3]A schematic illustration of the transient grating or two-pump-pulse THz generation is shown in Fig. 1. The small black circles represent interference maxima formed by the two laser pulses propagating in the sample at a small angle as shown by the two black arrows. The gray tilted lines represent maxima of the generated THz field propagating in the directions marked by the gray arrows. Roughly, we can assume that each laser interference maximum represents a Cherenkov type source of almost single cycle THz radiation. The superposition of the radiation emitted by all maxima determines the resulting THz waveform. It is easy to see, that a manipulation of the intensity distribution of the laser spot causes a corresponding change of the time profile of the generated THz pulse. This spacetime conversion provides an opportunity for THz pulse shaping. THz radiation in our experiments has been generated in a nearly stoichiometric LiNbO^ crystal with 2 mol % Mg doping and shaped in such a way that the angle between the front and one of the side surfaces is equal to 64 degree in order to achieve normal incidence on the surface for the incoming exciting laser beams and

714

THz

LiNbOj

^

/

Fig. 1. Schematic illustration of transient grating THz generation. for the outgoing generated THz radiation (Fig.l). Near-infrared pulses with a pulse duration of 150 fs, a repetition rate of 200 kHz and an average power of 600 mW have been generated by a commercial mode-locked Ti: sapphire system (Mira/ Rega-9000 Coherent). After the amplifier, the femtosecond radiation was collimated into a narrow (~0.5 mm diameter) beam and then split into two parts by a 50% beam splitter. Spatial and temporal superposition of the two beams with a small variable angle (0.3-1.9 degree) between them results in an interference pattern with variable period. The diameter of the laser spot on the sample was 660 |im. The energy of the generated THz pulses was measured by a calibrated liquidHe cooled Si-bolometer, and the temporal profile of the electric field was detected by a standard electro-optic sampling set-up employing a 0.6-mm-thick ZnTe crystal as the sensor. The measured temporal profiles of the THz field and their normalized power spectra obtained by Fourier transformation are plotted in the left and middle panel of Fig. 2. The spectra of the THz pulses reveal relatively narrow (~ 0.1 THz) peaks at 0.6, 1.9, and 2.7 THz.

pa 0

2 -2

-9

-AA/yyyvH -6

-3 0 3 Time delay (ps)

9 0

1K.

J

1 2 Frequency (THz)

1 2 Frequency (THz)

Fig. 2. Left panel: THz waveforms measured for 3 different angles (0.350*^ (c), 1.30° (b), and 1.90° (a)) between the two excitation beams. Middle panel: normalized power spectra calculated by fast Fourier transformation of the time domain data. Right panel: Measured frequency dependence of the generated THz pulse energy.

715

The right panel of Fig. 2 exhibits the energy of the generated THz pulses as a function of frequency. The highest energy achieved by our setup was 1.8 pJ at 1.6 THz, and the frequency dependence shows a broad peak around this value. Three examples of the THz waveform shaping are presented in Fig. 3. In the

•

3

-

2

-

1

0

1

2

3

3 - 2 - 1 0

1

Time delay (ps) Fig. 3. Illustration of the THz waveform shaping . first example (Fig. 3 (a)), the central part of the laser pump spot has been blocked by a 80 |im wire inserted vertically in front of the crystal. Such a blocking of the central part of the transient grating results in a corresponding minimum in the temporal profile of the generated THz waveform. We also performed measurements for blocking approximately half of the laser spot by a vertical shield. When the shield was placed on that side where the THz radiation was measured (Fig. 3 (b)) the generated THz waveform reveals a sharp rising and a slow trailing edge. When the shield was placed on the opposite side the generated THz waveform has a slowly rising edge and a sharp trailing edge (not shown in Fig. 3). A narrow slit in front of the crystal resulted in a THz wave form shown in Fig. 3 (c). More complicated and rapidly changing THz waveforms can be achieved by variation of the laser intensity distribution with a liquid crystal modulator. Such a device could be used in THz signal processing systems [5]. Acknowledgements. A.G.S. and J.H. acknowledge funding of their stays in Stuttgart by the Max-Planck-Institute for Solid State Research. This work was supported by the Hungarian Scientific Research Fund under Grant No. T 038372.

References 1 J. Ahn, A. V. Efimov, R. D. Averitt, and A. J. Taylor, in Optics Express, Vol. 11, 2486, 2003. 2 Y.-S. Lee, T. Meade, M. DeCamp, T. B. Norris, and A. Galvanauskas, in Applied Physics Letters, Vol. 77, 1244, 2000. 3 Y.-S. Lee, N. Amer, and W. C. Hurlbut, in Applied Physics Letters, Vol. 82, 170 2003. 4 A. G. Stepanov, J. Hebling, and J. Kuhl, in Physical Review B, Vol. 63, 104304, 2001. 5 B. E. Cole, J. B. Willams, B. T. King, M. S. Sherwin, C. R Stanley, in Nature, 63, Vol. 410, 60, 2001.

716

Typesetting THz Waveforms Joshua C. Vaughan, T. Feurer, Thomas Homung, and Keith A. Nelson Massachusetts Institute of Technology, Cambridge, MA 02139, USA E-mail: [email protected] Abstract. We demonstrate generation of temporally shaped THz waveforms m LiNbOs by spatially shaping femtosecond excitation pulses with a spatial light modulator. The generated THz waveforms are approximately proportional to the first spatial derivative of the excitation beam profile.

Femtosecond pulse shaping has found numerous applications in fundamental and applied science. Although pulse shaping techniques are well established from the ultra-violet throughout the mid-infrared spectral region [1], less attention has been devoted to pulse shaping methods in the THz spectral range. In conventional femtosecond pulse shaping, the phase and/or amplitude of the spectral components of a broadband laser pulse are modulated at the spectral Fourier plane of a zerodispersion compressor. Due to many practical and technical reasons this approach is not easily adapted to the THz regime. Instead, the THz pulse shaping methods reported to date have utiUzed either temporally shaped femtosecorid excitation waveforms [2] or specially fabricated materials or devices [3]. 1.0

^^ / c«>

k I1

i'\

2i as c no

a £ -0.5

/*

l''\ l p• •1 '

5/^u w^\s^

Z8

Z9

aO

3.1

32 Z8

Z9

3.0

31

position [mm]

position [mm]

d)

iffl

U

iWJ M\

y jy

32 28

29

32

ai

\M^M

4

delay [ps]

3.0

position [mm]

delay [ps]

6

8

delay [ps]

Fig. 1. Two complementary excitation profiles a) and b) whose sum is spatial'^ uniform c), and the corresponding measured THz waveforms d) and e) whose sum is unifoMy zero f).

Here, we propose and demonstrate a different approach to shaping THz waveforms, based upon shaping the spatial profile of femtosecond laser pulses which are used to excite phonon-polaritons [4] in LiNbOs through impulsive stimulated Raman scattering. Since phonon-polaritons excited in an appropriate

717

crystal propagate in a plane almost perpendicular to the direction of tfee excitation beam, the spatial profile of the femtosecond laser pulse exerts a strong influence on the temporal profile of the generated phonon-polariton. As the phonon-polariton wave intercepts the crystal-air interface, the electromagnetic part, i e. the THz waveform, is able to radiate out into air and may be used for further experiments or applications. It can be shown analytically in one dimension that in ttie impulsive excitation limit, the observed THz waveform E(x,t) resembles the spatial derivative of the excitation beam profile [5]. Therefore, the spatial excitation profile that is needed to produce a user-defined THz waveform is to a first approximation calculated by integration. Corrections may then be made to conipensate for dispersion and absorption effects. All that is required, then, is to simplj> 'typewrite' that profile onto the crystal by shaping the spatial profile of the excitation pulse. An illustrative THz pulse shaping example is shown in figure 1. The laser pulses used to excite phonon-polaritons had a central wavelength of 790 nm, a duration of 40 fs, and 10-100 pJ of energy. A computer-controlled spatial light modulator (SLM, Hamamatsu model SLMM X7550-800) was used to generate user-defined one-dimensional excitation profiles by amplitude' modulation of the incident pump beam. The SLM was operated in amplitude-modulation mode, in which the nematic liquid crystal layer of the SLM was aligned such that it variably rotated the polarization of the pump beam at different spatial locations. This spatially varying polarization was converted to an amplitude-mo4ulated beam with a polarizer, and then imaged onto a 0.5 mm thick x-cut LiNbOs crystal. In all cases, the polarization of the excitation beam at the sample was parallel to the optic axis, which is oriented vertically in figure 2. The excitation profiles '"-aried along the x-direction but were constant along the y-direction, typically over a few millimeters. A charge-coupled device camera was used to record the spatial excitation profile of the excitation pulse used to generate all THz jvaveforms. Phonon-polaritons were detected with a variably delayed probe pulse via electrooptic sampling in the same LiNbOa crystal used for generation.

LiNbOa

pump

probe

Fig. 2. Experimental setup for the generation and detection of shaped THz waveforms.

To demonstrate that the present technique is suitable for encoding signals in THz waveforms, we have generated spatially varying excitation profiles that are able to imprint a waveform resembling one byte worth of information, the bits are equally spaced, and each bit is realized through a single Gaussian shape»d feature of the excitation profile. The spatial Ught modulator allows single bits to be switched on or off in a simple manner. A compilation of nine consecutive experiments is

718

shown in figure 3. An intensity plot of a vertical section of the excitation profile together with the corresponding byte is displayed in (a), and the measured phononpolariton responses are shown in (b). Each excitation line generates a $ingle-cycle response that corresponds to a single Tjit'. The center frequency of a J^ingle 'bit' is 1.1 THz and the bandwidth around 1 THz. Intensity [arb. units]

Amplitude [arb. units]

ii|pipiiooo|

IHidii'i i|i(iioioi| Ipiciioi ^^^(l^fjitopbod |iOD1100|

x[mm]

Fig. 3. Generation of THz waveforms representing a byte, (a) Intensity plo^ of a vertical section of the excitation profile as a function of the spatial position together with the corresponding number in binary coding, (b) Measured THz waveform versu*' probe delay time. The narrow dark line at time zero in (b) is caused by an instantane(>».is electronic contribution to the Kerr signal. In conclusion, we have demonstrated that spatial shaping of femtosecond excitation pulses leads to well defined THz phonon-polariton waveforms in LiNbOs. The method requires no specialized materials or techniques pther than a commonplace electrooptic crystal and the ability to create spatially shaped beam profiles. By coupling the generated THz radiation into free space, the corresponding THz electric field may be used for further expMments or applications. Acknowledgements. This work was supported in part by Army Research Office Grant No. DAAD10-01-1-0674 and NSF Grant No. CHE-0212375. T.H. acknowledges the Deutsche Forschungsgemeinschaft for financial support.

References

5.

A.M. Weiner, Rev. Sci, Instr. 71, 1929 (2000). J. Ahn, A.V. Efimov, R.D. Averitt, and A.J. Taylor, Opt. Express l i 86 (2003). N.M. Froberg, B.B. Hu, X.-C. Zhang, and D.H. Auston, IEEE J. Quant. Elec. 28, 91 (1992). M. Bom, K. Huang, Dynamical theory of crystal lattices, (Oxford Classic Texts, Oxford 1988). D.A. Kleinman, D.H. Auston, IEEE J. Quant. Elec. 20, 964-970 {\9U).

719

Magnetically induced evolution of terahertz radiation spectrum emitted from InAs up to 27T Hiroshi Takahashi^ Alex Quema^, Masahiro Goto^, Shingo Ono^'^, Nobuhiko Sarukura^'^, Gen Nishijima , and Kazuo Watanabe^ ^ Department of Photo Science, The Graduate University for Advanced Studies , Shonan Village, Hayama 240-0193, Japan E-mail: [email protected] ^ Institute for Molecular Science (IMS), Myodaiji, Okazaki 444-8585, Japan ^ High Field Laboratory for Superconducting Materials, Institute for Materials Research, Tohoku University, Katahira, Aoba-ku, Sendai 980-8577, Japan Abstract. THz-radiation from femtosecond-laser-irradiated InAs (100) surface is investigated. It is found that THz-radiation spectrum exhibits two inter-related phenomena in a strong magnetic field under the Voigt configuration.

1. Introduction Since Zhang et al. reported the quadratic magnetic-field dependence of THzradiation power from GaAs [1] many studies have shown that application of a magnetic field causes an order of magnitude enhancement of THz-radiation [2,3]. Among various semiconductors, InAs is widely accepted as the practical THzradiation source, and the quadratic magnetic field and excitation-fluence dependence of THz-radiation power have been reported [4,5]. Additionally, THzradiation spectrum is found to exhibit periodic structure at magnetic fields above 12 T, which is explained by the change of refractive index induced by the strong magnetic field [6-8]. In this report, the physical origin of this phenomenon is discussed by considering the THz-radiation propagating in a direction perpendicular to the magnetic field.

2. Experimental Methods The experimental setup is described in Ref. [7]. A mode-locked Ti.Sapphire laser was used as an excitation source, which delivered 100-fs optical pulses at a center wavelength of 800 nm with 82-MHz repetition rate. The laser power was adjusted to 300 mW, and its spot size on the sample surface was approximately 2 mm. The sample used was an undoped bulk InAs with a (100) surface. A hybrid magnet, consisting of an outer superconducting magnet and an iimer water-cooled resistive magnet, provided the static magnetic field. The hybrid magnet generated a maximum magnetic field of 28 T in a 52-mm room temperature bore [9]. A liquidHe cooled Ge bolometer was used for detecting the power of the total radiation.

720

The Fourier spectrum of THz-radiation was measured by a Michelsoninterferometer.

3. Results and Discussion Figure 1 (a) illustrates the THz-radiation power measured at various magnetic fields. THz-radiation power increases, and exhibits a quadratic dependence on the magnetic field until it saturates at around 3 T. The main mechanism of magneticfield induced enhancement of THz-radiation power in this case is explained by the Lorentz force. Then it decreases and reaches a minimum value. To investigate this in more detail, THz-radiation spectra were measured by Michelson interferometer. Figure 1(b) presents the THz-radiation spectra at selected magnetic fields. There are two phenomena observed under the existence of strong magnetic fields. The low-frequency component begins to appear, and a periodic structure is clearly observed in the spectrum, (a) . , (b) DOWN 24 T

? c

€3

5 (ft

-jwlMAArAAiWUu.A*.www.s-.*^ ^ .,J1»AMAAAAAA>*,, V«-,-.--,

.n.

.„^,

,wJ*^«'^^AM#^A^<-AA-/^/>.,*ww^

.W•"'W*#M^v«^Aw^w.--^^

1 Magneticfield(T)

20 T 16T 14T 12 T 3T

Frequency
Fig. 1. (a) THz-radiation power measured at various magnetic fields, (b) THz-radiation spectrum measured at selected magnetic fields. To reveal the physical origin of this phenomenon, the propagation of THzradiation into InAs is considered under the existence of strong magnetic field. For the radiation propagating in a direction perpendicular to the magnetic field, Voigt effect emerges, and the refractive index for radiation with polarizations parallel (ri//) and perpendicular («^) to the static magnetic field are given as:

n„=el'\l-col/a)'y^''

(1)

n^= nAl-[col/(o)' -col)][co^/((o' -0)1-0)^)]}'^' . (2) where Sg is the static dielectric constant of InAs, oi is the cyclotronfi*equency,and Of, is the plasmafi*equency.Figure 2 presents the calculated values of refractive index in THz region. At frequencies below oj^, the refractive index becomes negative, which implies that THz-radiation cannot propagate into the InAs. At the frequency region between co^ and ojjj, the refractive index of InAs is smaller than the static dielectric constant, and exhibits a positive dispersion. At fi"equencies above 6^, the dispersion becomes large and the refractive index increases until the frequency reaches OJ^. The lower frequency limit of the transparent region is given by w^, and shifts toward lower frequency with increasing magnetic field. To discuss the experimental results, the calculated transparent regions are plotted in Fig. 2 (b). The calculated co^ at each magnetic field is also shown in Fig. 1 (b) as

721

triangle marks. Good agreement with experimental results is observed, and the physical origin of the refractive index change can be explained by the Voigt effect, which is related to the emergence of the magneto-plasma effect, 25r (a) n,i 20 -

Z2r

— « i

IS 10

/ i.


5

y

/ i LI

1

1-

1

1

Frequency (THz)

1

Frequency (THz)

Fig. 2. (a) The refractive index of InAs for THz-radiation calculated by eq. (1) and (2). (b) A diagram of radiation frequency given by (O2 and CO^. The marked areas present the radiationfrequency,which is not allowed to propagate into InAs.

4, Summary In summary, the magnetic-field dependence of THz-radiation power from InAs surface is investigated up to 27 T. The behavior of the THz-radiation spectrum exhibits two inter-related phenomena, namely, peak shift toward lower fi-equency and appearance of periodic structure. Both phenomena are found to be interrelated, and induced by the emergence of the magneto-plasma effect.

References 1 X. C. Zhang, Y. Liu, T. D. Hewitt, T. Sangsiri, L. E. Kingsley, M. Weiner, Appl. Phys. Lett. 62, 2003, 1993. 2 C. Weiss, R. Wallenstein, and R. Beigang, Appl. Phys. Lett. 77,4160, 2000. 3 H. Takahashi, Y. Suzuki, M. Sakai, S. Ono, N. Sarukura, T. Sugiura, T. Hirosumi, and M. Yoshida, Appl. Phys. Lett. 82, 2005, 2003. 4 R. McLaughlin, A. Corchia, M. B. Johnston, Q. Chen, C. M. Ciesla, D. D. Amone, G. A. C. Jones, E. H. Linfield, A. G. Davies, and M. Pepper, Appl. Phys. Lett. 76, 2038, 2000. 5 H. Takahashi, A. Quema, R. Yoshioka, S. Ono, and N. Sarukura, Appl. Phys. Lett. 83,1068,2003. 6 H. Ohtake, H.Murakami, T. Yano, S. Ono, N. Sarukura, H. Takahashi, Y. Suzuki, G. Nishijima and K. Watanabe, Appl. Phys. Lett. 82, 1164, 2003. 7 H. Takahashi, Y. Suzuki, A. Quema, M. Sakai, T. Yano, S. Ono, N. Sarukura, M. Hosomizu, T. Tsukamoto, G. Nishijima, and K. Watanabe, Jpn. J. Appl. Phys. 42, L532, 2003. 8 H. Takahashi, A. Quema, M. Goto, S. Ono, N. Sarukura, G. Nishijima, and K. Watanabe, Jpn. J. Appl. Phys. 43, L1017, 2004. 9 Y. Muto, Y. Nakagawa, K. Noto, S. Miura, A. Hoshi, K. Watanabe, G. Kido, H. Ichikawa, T. Fujioka, Y. Sato, O. Osaki and H. Takano, Sci. Rep. RITU, A 33, 221,1986.

722

A Liquid Crystal Phase Shifter with a Tuning Range of over 360 degrees around 1 THz Chao-Yuan Chen\ Cho-Fan Hsieh^ Ru-Pin Pan^ and Ci-Ling Pan^ ^ Institute of Electro-Optical Engineering, National Chiao Tung University, Hsinchu, Taiwan 300, R.O.C. E-mail: [email protected],tw ^ Department of Electrophysics, National Chiao Tung University, Hsinchu, Taiwan 300, ROC. E-mail: [email protected] Abstract. Tunable phase shift over 2n radians (360 degrees) at THz frequencies is achieved with a sandv^iched liquid crystal (LC) cell operated over a broad range using magnetically controlled birefriengence.

1. Introduction In recent years, terahertz (THz) photonics have undergone remarkable growth with intense interests for applications ranging from investigations of ultrafast dynamics in materials to medical, environmental sensing and imaging [1], For these and future applications in THz communication and surveillance, quasi-optic components such as phase shifters, modulators, attenuators and polarizers are indispensable. For tunable phase shifting application, several groups have demonstrated devices based on optically or electrically controlled carrier concentration in quantum well structures [2-4]. These quantum-well-based THz phase shifters, however, have not yet demonstrated In phase shift and need be operated at temperatures far below room temperature. The large birefringence of liquid crystals (LCs) is well known and has been employed successfiiUy for phase shifting of microwave and millimeter wave signals previously [5,6]. In THz region, we have recently determined the optical constants of a nematic LC, 5CB, which exhibits relatively large birefiingence and small extinction coefficient at frequencies around 1 TITz, at room temperature [7]. An electrically controlled room temperature THz phase shifter with 5CB has also been demonstrated by the authors [8]. A maximum phase shift of 4.07"^ was shown at 1.07 THz. The high voltage (~177V) was required. By using magnetically controlled birefringence in a 5CB cell with nominal thickness of 1.5 mm, we were able to achieve a maximum phase shift of 141*^ at 1.025 THz [9]. The alignment of the LC layer becomes increasingly difficult with thicker cells. In this work, we show a phase shift of 2n phase shift around 1 THz is possible with a sandwichedstructured LC cell

723

2. Operating Principles and Device Configuration

Propagation diiirttion

Fig. 1. The schematic diagram of theLC THz phase shifter.

The tunable THz Phase shifter consists of a homeotropically aligned LC cell and a rotary magnet as shown in Fig. 1. The magnetic inclination angle, 9, is defined as the angle between the magnetic field and the propagation direction. We can assume that the LC molecules are reoriented parallel to the magnetic field when the magnetic field is large enough as in this work [10] . The phase shift, 5(0), due to magnetically controlled birefi*ingence is given by 6{0)^27tL-^~<

cos\0)

%mHO)

(1)

where no and ne are the ordinary and extra-ordinary refi'active indices of the LC, L is the thickness of LC layer,/is the frequency of the THz waves and c is the speed oflight in vacuum. Three fiased silica plates are employed for the sandwiched cell. The two compartments of the cell are filled with a common LC, E7 (MerckO. Teflon spacers are used for controlling the thickness of LC layers to L5 mm each. The alignment of the LC molecules is homeotropic [10]. We employ an Nd-Fe-B sintered magnet on a rotation stage, which provides a tunable magnetic field for tuning the phase shift of the THz wave. We use THz-TDS method for characterizing the device. The experimental setup has been described previously [9]. We have also measured the extraordinary and ordinary refractive indices of E7, n^ and no in the THz frequency range.

3. Results and Discussions In the 0.2-1.2 THz range, ne varies fi-om 1.69 to 1.80, while no varies fi*om 1.51 to 1.63 for E7 at room temperature (25''C). The birefringence of E7 is thus 0.12 to 0.21 for the samefi*equencyrange. The corresponding imaginary indices of E7 are relatively small (< 0.04). The deduced phase shifts of the phase shifters are plotted as a fiinction of the magnetic inclination angle in Fig. 2 for 0.49 and 1.025 THz waves. The maximum phase shift achieved was 368"* at 1.025 THz and 9 == 54^. The theoretical predictions are plotted as the sohd curves in Fig. 2. They show good agreements with the measured results.

724

e (degrees)

Figure 2 The phase shift of the THz waves versus the magnetic inclination angle. ITie solid curves are theoretical predictions. The open and solid circles are experimentally measured phase shift at 0.49 and 1.025 THz.

4. Conclusion In summary, we have demonstrated a room temperature THz phase shifter by using a sandwiched LC cell. A maximum phase shift of over 360^ around 1 THz has been achieved successfiilly. The phase shift can be tuned magnetically by controlling the effective refractive index of LC layer. Measured results are in good agreements with theoretical predictions. A c k n o w l e d g e m e n t s . This work was supported in part by the National Science Council of R.O.C. under Grants No. NSC 92-2215*E-009-030 and Program for liie I\irsuit of Academic Excellence of the Ministry of Education, R.O.C.

References 1.

B. Ferguson and X.-C. Zhang, Nature Materials 1, 26, 2002. I. H. Libon, S. Baumg^rtner, M. Hempel, N. E. Hecker, J. Feldmann, M. Koch, and P. Dawson, Appl Phys. Lett. 76, 2821, 2000. 3 R. Kersting, G. Strasser, and K. Unterrainer, Electron. Lett. 36, 1156, 2000. 4 T. Kleine Ostmann, M. Koch^ and P. Dawson, Microwave Opt. Tech. Lett. 35, 343, 2002, 5 K. C. Lim, J. D. Margerum, and A. M. Lackner, Appl. Phys. Lett. 62, 1065, 1993. 6. D. Dolfi, M. Labeyrie, P. Joffre, and J. P. Huignard, Electronics Lett. 29, 926, 1993. 7. T.-R. Tsai, C.-Y. Chen, C.-L. Pan, R.-P. Pan, and X.-C, Zhang, Appl. Opt. 42, 2372, 2003. T.-R Tsai, C.-Y. Chen, R.-P. Pan, C,-L. Pan, and X.-C. Zhang, IEEE Microw^ave Wireless Comp. Lett. 14, 77, 2004. 9. C.-Y. Chen, T.-R Tsai, C.^L. Pan, R.-P. Pan, Appl. Phys. Lett. 83, 4497, 2003. 10. P, G, de Gennes and J. Prost, % e Physics of Liquid Crystals, 2nd ed. (Oxford, New York, 1983).

2.

725

Cooper pair breaking dynamics in MgB2 using optical-pump terahertz-probe spectroscopy J. Demsar^^ R. D. Averitt^*, A. J. Taylo^^ V. V. Kabanov^ ^ Los Alamos National Laboratory, MS K764, Los Alamos NM 87545, USA E-mail: [email protected] Ja ^ "Jozef Stefan" Institute, Jamova 39, Ljubljana, Slovenia E-mail: [email protected] Abstract. Cooper pair-breaking dynamics (PBD) have been resolved for the first time in MgBz- We present an analysis of the PBD using the Rothwarf-Taylor model, enabling the determination of the bare quasiparticle recombination and phonon pair-breaking rates.

Nonequilibrium effects have been studied in superconductors since the 1960's. One of the primary goals of such investigations has been to determine the bare quasiparticle recombination rate. However, directly measming this rate is difficult since, as first noted by Rothwarf and Taylor [1], other phenomena contribute to the measurements. In the case of superconductors, the bare quasiparticle recombination rate is masked due to pair-breaking by above gap phonons. To date, it has not been possible to extract the bare quasiparticle recombination rate. In the case of conventional superconductors, one problem is that the superconducting transition temperatures are quite lov^, making detailed fluence dependent measurements all but impossible. However, with the discovery of MgBa (Tc = 39K) it is now possible to perform detailed fluence dependent ultrafast measurements on a conventional (i.e. phonon mediated pairing) superconductor. Our recent experiments of the picosecond photoinduced far-infrared conductivity dynamics on MgB2 have, for the first time, revealed the dynamics of Cooper pair breaking following photoexcitation [2]. Here we present a detailed analysis of the pair breaking dynamics which suggests that in MgB2 photoexcitation is initially followed by energy relaxation to high frequency phonons instead of, as commonly assumed, e-e thermalization. Furthermore, the bare quasiparticle recombination rate and the probability for pair-breaking by phonons have been determined for thefirsttime.

726

1H

d

1

(0,

0.1 n

^—f

0.1

t[ps]

t[ps]

Fig. 1. a) Photoinduced terahertz (THz) conductivity dynamics on MgB2 at 7K at different excitation fluences F in jiJ/cm^. Inset: the expanded timescale, showing that the superconducting recovery dynamics proceeds on tlie timescale of several hundred picoseconds, b) Solutions of Equation (2) for different values of parameter K compared to the trace taken at F = 0.6 |iJ/cm^. The experiments were performed using optical-pump THz-probe spectroscopy as described in [3]. Figure 1 a) presents the early stage dynamics of photoinduced changes of the conductivity in MgB2 thin films [4]. Unlike in lead, where the PBD proceeds on the sub-picosecond timescale [5], in MgB2 the process is much longer. Moreover, the PBD in MgB2 are temperature and photoexcitation fluence dependant. In order to understand the PBD we analyzed the data in terms of the Rothwarf-Taylor model, where the dynamics of the quasiparticles (QP) and high frequency (o) > 2A) phonon densities, n and N, are described by a set of t\^^o coupled differential equations [1]. Since the SC recovery dynamics proceed on a much longer timescale than the PBD (see inset to Figure 1 a)), the term describing the loss of co > 2 A phonons by processes other than pair excitation can be neglected, and the equations are given by:

dnldt = /3N-Rn^

dN/dt =

-[Rn'-/3N]

(1)

R is the bare quasiparticle recombination rate and p is the probability for pairbreaking by phonons. With the initial condition that after photoexcitation (and the initial sub-picosecond e-e and e-ph dynamics) the QP and high frequency phonon densities are no and No, the subsequent time evolution of n is given by [6]

n{t)-.

_i_J_ 1 4

2r

1

Tl-KQxp(-tfi/T)

(2)

where K and T are dimensionless parameters determined by the initial conditions:

727

Equation (2) has three distinct regimes, depicted in Fig. lb). K = 0 corresponds to the stationary solution, when no and No, following photoexcitation, are already in quasi-equilibrium and n(t) is a stepftmction.For 0 < K < 1 the number of QPs is higher than the quasi-equilibrium value, while -1 < K < 0 represents the opposite situation when the initial T of co > 2A phonons is higher than the quasiequilibrium one. As Fig. 1 b) shows, the latter situation seems to be realized in the case of MgB2. The most probable reason why in MgB2 the photoexcited electrons initially relax mainly by electron-phonon scattering is a strong coupling of electrons with high frequency optical phonons (in particular the coupling of a band electrons to the E2g phonon mode at 60-80 meV [7,8]). At low T, when the density of thennally excited QPs and co > 2A phonons is exponentially small, the change in conductivity is proportional to the photoinduced quasiparticle density. Therefore we can fit the conductivity dynamics using Equation (2). Best fits to the data taken at various intensities are plotted by solid lines in Fig. la). This allows us to determine the microscopic parameters R and p byfittingthe x/p and K vs. the absorbed energy density Q to Equations (3). A best fit to the two data sets yields p~^ = 15 +/- 2 ps, R = 100 +/30 ps"^ unit cell [2]. In conclusion, we presented the first observation of time-resolved pair-breaking dynamics in a superconductor. For MgB2, we have determined the microscopic parameters R, the bare quasiparticle recombination rate and p, the probabilit>^ for pair-breaking by phonons.

References 1 A. Rothwarf, B.N. Taylor, Phys. Rev. Lett. 19,27 (1967). 2 J. Demsar, R. D. Averitt, A. J. Taylor, V. V. Kabanov, W. N. Kang, H. J. Kim, E. M. Choi, S. I. Lee, Phys. Rev. Lett. 91, 267002 (2003). 3 R. D. Averitt, A. J. Taylor, J. Phys.: Condens. Matter 13, R1357 (2002). 4 W.N. Kang, H. J. Kim, E. M. Choi, C. U. Jung, S. I. Lee, Science 292, 1521 (2001). 5 J.F. Federici, B. I. Greene, P. N. Saeta, D. R. Dykaar, F. Sharifi, R. C. Dynes, Phys. Rev. B46, 11153 (1992). 6 A detailed Rothwarf-Taylor analysis will be presented elsewhere.. 7 LI. Mazin O. K. Anderson, O. Jepsen, O. V. Dolgov, J. Kortus, A. A. Golubov, A. B. Kuzmenko, D. van der Marel, Phys. Rev. Lett. 89, 107002 (2002), 8 A.A. Golubov, J. Kortus, O. V. Dolgov, O. Jepsen, Y. Kong, O. K. Anderson, B. J. Gibson, K. Ahn, R. K. Kremer, J. Phys.: Condens. Matter 14, 1353 (2002). 9 P.B. Allen, Phys. Rev. Lett. 59,1460 (1987).

728

Femtosecond formation of phonon-plasmon coupled modes studied by ultrabroadband THz spectroscopy R. Huber\ C. Kubler^ S. Tubel\ A. Brodschelm\ F. Kohler^ M.-C. Amann^ and A. Leitenstorfer^ ^ Physik-DepartmentEll, Technische Universitat Munchen, D-85747 Garching, Germany, Email: [email protected] ^ Fachbereich Physik, Universitat Konstanz, D-78457 Konstanz, Germany ^ Walter-Schottky-Institut (E26), Technische Universitat Munchen, D-85747 Garching, Germany Abstract. The ultrafast transition of an optical phonon resonance to a coupled phononplasmon system is studied. After 10-fs photoexcitation of i-InP, the buildup of coherent beats of the emerging hybrid modes is directly monitored by means of ultrabroadband THz spectroscopy. Mutual repulsion and redistribution of the oscillator strength of the interacting phonons and plasmons is seen to emerge within approximately one oscillation cycle of the upper branch of the mixed modes. Longitudinal optical (LO) lattice vibrations in a polar semiconductor couple to charge-density waves of a carrier plasma via long-range electrostatic forces [1,2]. The resulting LO-phonon plasmon coupled (LOPC) resonances play a central role for transport properties and relaxation dynamics. Ultrashort laser pulses allow for the generation of a dense electron-hole plasma in an intrinsic semiconductor within a time that is shorter than a typical oscillation cycle of the lattice and the photogenerated plasma. Quantum kinetic studies have predicted that the LOPC resonances themselves are not established instantaneously in this extreme nonequilibrium situation [3]. The femtosecond buildup of carrier-carrier correlations has recently been observed by means of ultrabroadband THz spectroscopy [4], however the role of the polar lattice has not been resolved. Here, we report on the first direct observation of the ultrafast transition of a pure LO-phonon dielectric oscillator to LOPC resonances in photoexcited intrinsic InP. The 230-nm-thin sample of i-InP is kept at room temperature. In the first step of the experiment, a dense carrier plasma is generated via interband absorption of a 10-fs laser pulse with a central photon energy of 1.55 eV. After a time delay /QJ a single-cycle electromagnetic pulse (see Fig. 1(a)) of a duration of 27 fs and a central fi-equency of 28 THz [5] probes the polarization response of this nonequilibrium system in real time. Resonances in the sample modify the wave form and lead to trailing oscillations after the THz transient transmitted through the semiconductor. Both amplitude and phase of the probe electric field JE'THZ are resolved as a function of a second delay time T by means of ultrabroadband electro-optic detection [6,7]. The pump induced change Afi'iHz of the THz electric

729

-50

0

50

100 150 200 250

T(fs) -20

-10

0 10 AE^(V/cm)

Fig. 1. Ultrabroadband twodimensional THz data of the photoinduced phonon-plasmon system in i-InP: (a) real-time evolution of the electric probe field Ejuz as a function of time T. (b) pump induced changes A£'XHZ of the transmitted electric field are recorded as a function of the sampling time T and the delay /Q after photoinjection of 2 x 10^^ electron-hole pairs per cm^. The dotted line indicates the temporal position of the maxium of the pump pulse. A sequence of up to 5 retarded half oscillation cycles (labels ' 1 ' through '5') builds up within/D< 150 fs.

20

field is measured as a function of the pump-probe delay /D and the THz sampling delay T (Fig. 1(b)): The polarization response of the carrier plasma follows instantaneously upon the single-cycle probe pulse at early delay times t^ <20 fs. With increasing values of ^D a trailing oscillatory response after the THz probe burst emerges. Up to five delayed half-cycles are observed for tx)> 150 fs (label T to '5'). Most interestingly, the oscillation cycles are not equidistant indicating beating between two newly emerging resonances. From the two-dimensional real-time data we extract the dynamics of the complex dielectric function of the excited semiconductor at frequencies between 7 THz and 60 THz [4]. Longitudinal electromagnetic resonances may be identified from the peaks of the imaginary part of the inverse dielectric function -Im(l/s^=o)) Fig- 2(a). In the unexcited sample or equivalently at negative delay times t^ <-70 fs, only a single resonance due to the LO phonon oscillator is observed. This pole vanishes rapidly after photoinjection of the plasma. A spectrally broad feature at 30 fs 150 fs, the resonance signature of fully developed dynamic screening is observed at frequencies around the more plasmon-like L+ branch: Re(l/8q=o) assumes negative values on the low-energy side due to resonant

730

energy (meV) 40 60 80 100

10 15 20 25 frequency (THz)

energy (meV) 40 60 80 100

10 15 20 25 frequency (THz)

Fig. 2: (a) Negative imaginary and (b) real part of the long-wavelength limit of the inverse dielectric function of InP versus frequency for different pump-probe delays /DLO and TO phonon frequencies of the bare ion lattice are indicated by broken lines.

overscreening by the carrier plasma. In particular, the macroscopic electric field of the phonon oscillator is overcompensated, giving rise to an effective sign reversal of the restoring force of the lattice vibration. Therefore, the L. mode is softened below the TO phonon frequency. At early delay times, ^D ^ 1 3 0 fs, screening of the lattice and the plasma mode builds up. Many-particle correlations which are a prerequisite for welldefmed collective LOPC resonances form approximately within the inverse eigenfrequency of the upper LOPC

mode - the shortest characteristic time scale of the system. Interestingly, this dynamics is not accompanied by a continuous shift of the phonon-like mode from COLO to below COTO- In fact, no resonance is observed in the Reststrahlen region (CDTO < co < COLO) at any delay time. Our findings support predictions of a quantum kinetic theory [3] and are well reproduced by simulations based on the non-equilibrium Greens functions formalism [8].

References A. Mooradian and G. B. Wright, Phys. Rev. Lett. 16, 999 (1966). G. Abstreiter et al., in Light scattering in Solids VI, edited by M. Cardona and G. Giintherodt (Springer, Berlin, 1984). Q. T. Vu and H. Haug, Phys. Rev. B 62 (2000) 7179. R. Ruber et al., Nature 414 (2001) 286. R. Ruber et al., Ultrafast Phenomena XIII, Springer Series in Chemical Physics 71 (2002) 365. R. Ruber et al., Appl. Phys. Lett. 76 (2000) 3191. A. Brodschelm et al., Ultrafast Phenomena XII, Springer Series in Chemical Physics 66 (2000) 215. Q. Wu and X.-C. Zhang, Appl. Phys. Lett. 71 (1997) 1285. A. Leitenstorfer et al., Appl. Phys. Lett. 74 (1999) 1516. R. Ruber, C. Kubler, S. Tubel, A. Leitenstorfer, Q. T. Vu, R. Raug, F. Kohler, and M.-C. Amann, to be published.

731

Solid-state phase transition onset detection in estrogen-like chemical via terahertz transmission spectroscopy Alex Quema^'^, Masahiro Goto^ Masahiro Sakai\ Gerardo Janairo^, Riadh El Ouenzerfi\ Hiroshi Takahashi\ Shingo Ono\ and Nobuhiko Sarukura^ ' Institute for Molecular Science, 38 Nishigonaka, Myodaiji, Okazaki, 444-8585, Japan E-mail: [email protected] ^ De La Salle University, Physics Department, 2401 Taft Avenue, Manila, Philippines 3T De La Salle University, Chemistry Department, 2401 Taft Avenue, Manila, Philippines Abstract. Solid-state phase transition onset in an endocrine-disrupting estrogen-like chemical (1,4-naphthol) is detected using terahertz transmission spectroscopy. Differential scanning microscopy and temperature-dependent X-ray diffraction analysis confirmed the occurrence of such phenomenon.

1.

Introduction

Since terahertz (THz) radiation can penetrate through a w^ide range of substances, this frequency range is now considered an important region in vibronic spectroscopy of hydrogen-bonded biomolecules [1]. Recently, it was reported that 1,4-dihydroxynaphthalene (1,4-naphthol) manifested an absorption peak at around 0.4 THz in room temperature [2]. Such peak is ascribed to some interaction between the THz radiation and the hydrogen bonds that link into chains the adjacent molecules in the isomer. Thus far, a response of organic materials to THz radiation has yielded results of conformational change and was depicted by the continuous shifting of the absorption peak with varying temperature [3]. In this report, the detection of solidstate phase transition onset in the THz frequency region of 1,4-naphthol (a chemical that exhibits estrogen-like property) is presented. As the temperature is increased, a discontinuous evolution of the absorption peak frequency is observed suggesting the onset of a certain solid-state phase transition. The occurrence of the phase transition is confirmed by differential scanning microscopy (DSC) and temperature-dependent X-ray diffraction (XRD) measurements. The results here show the enhanced sensitivity of THz spectroscopy in detecting phase transition.

2.

Experimental Methods

The sample was prepared using 0.2 grams of 1,4-naphthol (CioH6(OH)2) chemical powder. The powder was pressed into a 10-mm diameter pellet and was placed on a copper sample holder, which was attached to the cold finger of a cryostat. The cryostat had a 10-mm diameter window equipped with a special quartz glass for

732

THz transmission spectroscopy. The THz transmission spectroscopy setup was similar to that used in Ref. (2) except in this case a 2.5T magnet was used. Briefly, the surface of an undoped bulk (100) InAs was illuminated with 800 nm, 82-MHz repetition rate, 100 fs pulse from a mode-locked Ti:Sapphire laser. The excitation source with a 1-W average power was incident at 45° on the InAs surface and had a spot size of 2 mm. Paraboloidal mirrors collimated and focused the radiation. The cryostat together with the sample was placed at the beam waist (2-mm diameter). A liquid-He cooled Si-bolometer was used as THz radiation detector while a polarizing Michelson interferometer provided its Fourier spectrum.

3. Results and Discussion Figure 1 (a) shows the plot of the absorption spectra of 1,4-naphthol at various temperatures. The absorption peak at 4 K is relatively sharp and well defined compared to the distorted and broad peak at 300 K. However, the relative intensity of the peak at 300 K is higher than that at 4 K. At temperatures 200 K and below, a steady shift of the absorption peak toward lower frequency coupled with a decrease in absorption intensity are observed as the temperature is increased. It has been reported that a steady shift of the absorption peak toward lower frequency with increasing temperature is typical for molecules in hydrogen-bonded networks [3], which implies that such peak is due to intermolecular hydrogen bonding. This absorption feature of the current sample suggests that this low-energy vibrational mode is strongly coupled to the lattice since it also reflects the response of the lattice itself. In general, a continuous shift of absorption peak as a function of temperature would indicate conformational change. However, a discontinuous evolution of absorption peak due to temperature change would entail a different explanation, which in this case the observation solid-state phase transition. (a)

V - 20

IS

• ^i ; wk

A yjt \\

-i» tafif / > ^ ' ^

^^ Hif / ^m

'K' W A

1

I i

vA/j

yw ] ^ y ^ . \ ^ 0.6 0.9 1.2 Frequency (THz) 1

I,

/v ^W

rO, v' ^^,«^•- fM jjjA. 5 k 'ft, fr^-^ t

""oSoSSsa 1.0 g

4K 50K •~»>«-100K 150K 200K —--210K -«~220K -t-230K -X—240K Q 270K 300K

100 150 200 Temperature (K)

Fig. 1. (a) Temperature-dependent absorption of 1,4-naphthol. (b) Temperature dependence of transmitted THz radiation and absorption peak fi-equency. A plot of the absorption peak frequency and transmitted THz radiation with respect to temperature is shown in Fig. 1(b). Initially the peak frequency shifts monotonically toward lower frequency as the temperature is increased. Then at 210 K, an abrupt shift toward much lower frequency (below 0.6 THz) is observed. A rather similar behavior is observed with the temperature-dependent transmitted

733

THz radiation power. In this case, the initial steady increase of THz radiation power suddenly manifested a sudden upsurge THz radiation at 210 K. These results cannot be explained by simply considering conformational change but rather the occurrence of solid-state phase change may have taken place.

e( ^0.12 O)

1, 1 0.08 E o X 0.04

llo.»

1

1

1

220

1

1

•

1

240

Temperature (K)

Fig. 2. DSC plot showing the endothermic peak observed when the sample is heated. To verify this notion, differential scanning calorimetric (DSC) measurement was conducted. It can bee seen in Fig. 2 that as the temperature is increased, a peak is observed indicating an endothermic reaction. This result reveals that solid-state phase transition indeed occurred and that it is a first-order type. Temperaturedependent x-ray diffraction (XRD) analysis was also conducted. Gradual peak shifting toward lower 20 is initially observed as the temperature is increased from 50K to 230K and a sudden shift is observed at 240K. Both DSC and XRD measurements show that phase transition indeed occurred at 240K. The difference in the observed transition temperature between the THz spectroscopic measurement and the DSC and XRD measurements indicates that the former detected the onset while the latter detected the actual phase transition.

4.

Summary

Solid-state phase transition in 1,4-naphthol (an estrogen-like chemical) is found to have indeed occurred, as confirmed by DSC and XRD measurements, and that detection of such phenomenon is possible via THz spectroscopy. The difference in the observation temperature of the phase transition points out that it is the onset that is observed by THz spectroscopy and not the actual phase transition itself. This clearly indicates that transmission spectroscopy in the THz region is a sensitive technique in detecting the occurrence of phase transition.

References 1 M. Walter, B. Fischer, and P. Uhd Jepsen, Chem. Phys. 288, 261, 2003. 2 A. Quema, H. Takahashi, M. Sakai, M. Goto, S. One, N. Sarukura, R. Shioda and N. Yamada, Jpn. J. Appl. Phys. 45, L932, 2003. 3 Y. Shen, P. Upadhya, E. Linfield and A. Davies, Appl. Phys. Lett. 82, 2350, 2003.

734

Maximum Entropy Method for Misplacement Phase Error Correction in Terahertz TimeDomain Reflection Spectroscopy Y.Ino\ R.Shimano^*, M.Kuwata-GonokamiVErik M.Vartiainen^, Yuri.P.Svirko^ and K.-E.Peiponen^ ^Department of Applied Physics, the University of Tokyo, and SORST, Japan Sceience and Technology Corporation(JST), 7»3'-L Hongo, Bunkyo-ku Tokyo, 113-8656, Japan E-mail: [email protected] ^ Department of Electrical Engineering, Lappeenranta University of Technology, P.O.Box 20, Fin-53851 Lappeenranta, Finland ^ Department of Physics, University of Joensuu, P.O.Box 111, Fin-80101 Joensuu, Finland Abstract. We develop a numerical method for the misplacement phase error correction in terahertz time-domain reflection spectroscopy. The method is based on the maximum entropy algorithm, dramatically which simplifies the experimental procedure.

Owing to recent advances in ultrashort laser pulse technology, measuring and utilizing the time domain information of electromagnetic wave has been attracting wide areas of new studies. In such experiments, precis determination of the phase of the measured electromagnetic wave precisely is of critical importance. Specifically in reflection measurements, the phase of the wave depends on the position of the reflected surface in the experimental setup. Therefore if we can't put the sample and the reference exactly in the same position, the measured phase will contain a frequency dependent error 2coALcosr]i/c, where rji is the incidence angle, and AL is misplacement.

Fig. 1. The formation of misplacement between the sample and reference in THz-TDS in reflection configuration

735

Here we propose a new approach to retrieve the phase error by numerical analysis, which enables us to correct the misplacement error without additional measurements. The developed algorithm is based on maximum entropy (ME) principle[l]. According to this, the power spectrum S(a)) of the dscrete time domain signal x,=x(nAt) is expressed as S(co)^/\l +lli,^j^exp(-ikcoAt)\, where a^. are given as functions of C,„=< x„ x„+,„ >. .By using these parameters, the complex spectral amplitude X(co) of x(t) is obtained as X(a))=b^^^exp(i(l)(a)))/(1+I,,,^j^exp(ika)At)[2]. Therefore 6(a))=3rg X(a)) is decomposed to two parts, d(a))=(t)(a))+ip((o), where Jlj((jo)=-(l-\-I.,^^j^Qxp(-ika)At) is determined from C,„, and 4(^0 is known as the error phase[2]. We apply this approach to the measured reflection coefficients rexp(i6(a))) in the spectral range (D^KOXCD^, which includes the phase error described above. Firstly, we determine C,„ from r^, which implies that the procedure determining ipio)) is independent on the phase error aw. This allows us to write (i>(co) as (p(a))=(pj„^Ja))+aa), where (l>true(^) is the error phase without misplacement error. Typically (ptmei^) is a smooth function of frequency i.e. ^j^,^(a))=(j)Q+^+0((D^), therefore we obtain d(^(co)=a+p+0((D). Since (t)true(^) originates solely from the ME procedure, it depends on the number of spectral points and shape. These can be changed by squeezing procedure [2] i.e. increasing the number of spectral points by factor 2K-\-l, replace the "red" and "blue" part of the increased spectral points with values at o)^ and co^, respectively. In the frequency region where the contribution of po) is negligible in comparison with (pg, dtp^'^/do) and d^^'^/dw should be nearly same. In such frequency regions dp/cko is mainly governed by the phase error due to misplacement, therefore one can obtain a=2ALcosr]i/c

2.0 ^c) o 1.5

^1.0

'

1

1

o

o K=0 • K=1

~ _ = 0

~

-0.5 -1.0

1

1

-^^^**

-

4

1

1

^

0.5

1.0

1.5

2.0

1 0

4

o K=Q • K=1

a.

I 0.0 •s

g 0.0 UJ

I 1.0 ^d)

%# » • •

• — •

P-0.5 8

0.5

1

1

1.0

1.5

1 20

4 2.5

Fig.2 The measured a) amplitude and b) phase of complex reflection coefficients. The corrected phase is also shown in b). c) The obtained error phase 0^"^and (^^'\ d) Their deri vativ es d(p^'^/d(D and J0^" ^/do).

736

We apply this analysis to the THz-TDS experiment of n-type InAs[3]. From Fig.2(d), we can obtain a and correct the obtained phase. After the correction, the phase shows good accordance with Drude fitting. We find that our analysis can correct the misplacement as small as 2fj,m. In summary, we develop a numerical method for the misplacement phase error correction in terahertz time-domain reflection spectroscopy. The method which is based on the maximum entropy algorithm dramatically simplifies the experimental procedure dramatically.

References [l]J.P.Burg, "A New Analysis Technique for Time Series Data," in 37*'' Ann.Intern.Meeting.Soc.Explor.Geophys., Oklaholma City [21E.M.Vartiainen, K.-E.Peiponen, and T.Asakura, "Phase Retrieval in Optical Spectroscopy: Resolving Optical Constants from Power Spectra", Appl.Spectrosc.5 0, 1283(1996) [3]R.Shimano, Y.Ino, Yu.P.Svirko, and M.Kuwata-Gonokami, "Terahertz frequency Hall measurement by magneto-optical Kerr spectroscopy in InAs," AppLPhys.Lett.8 1, 199 (2002) *The present address is "Department of Physics, the University of Tokyo, Hongo, Bunkyo-ku, Tokyo, 113-0033, Japan"

737

Pulsed terahertz spectroscopy and imaging applied to inspection of explosives and inflammable liquids Kohji Yamamoto^, Mariko Yamaguchi^ Fumiaki Miyamaru^ Masahiko Tani\ Masanori Hangyo\ Takeshi Ikeda , Akira Matsushita , Kenji Koide^, Michiaki Tatsuno^, and Yukio Manami^ ^ Institute of Laser Engineering, Osaka University, Yamada-oka 2-6, Suita, Osaka 5650871,Japan E-mail: kohji@ ile.osaka-u.ac.jp ^ Forensic Science Laboratory, Osaka Prefectural Police Headquarters, Honmachi, chuo-ku, Osaka 541-0053, Japan Abstract. We carried out applied researches on the pulsed terahertz spectroscopy and the terahertz imaging to detect the C-4 explosive hidden in a mail and inflammable liquids stored in plastic bottles. We confirm that THz techniques are quite efficient at contactless and nondestructive inspection of hazardous materials used in terrorist activities.

1.

Introduction

Terror threat has been substantially increasing worldwide. If we can inspect hazardous materials hidden inside packages, we can take effective precautions against terrorist activities. The THz techniques, which have been successfully applied to plastic-package imaging of an IC chip [1], are one of possible measures. In this study we examined application feasibility of the THz time-domain spectroscopy (THz-TDS) and the THz imaging to inspecting the C-4 explosive [2] and inflammable liquids.

2.

Experimental

We carried out the THz-TDS with a standard system (Fig. 1(a)), which uses photoconductive emitter and receiver antennas. A mode-locked Ti:sapphire laser with 50 fs in pulse width pumps the antermas for generation and detection of pulsed THz radiation. An example of detected signals in the time domain and its power spectrum are shown in Fig. 1. The refractive index and the absorption coefficient were obtained by the THz-TDS analysis. A C-4 pellet sealed inside an envelope was measured by the THz transmission and the THz imaging. We conducted the THz-TDS of alkanes (carbon number=5~12) and inflammable liquids, including gasoline, diesel oil, benzine, and some other oils. These liquids were enclosed in polyethylene terephthalate (PET) bottles (6 cm in thickness) or in home-made stainless steel cells (3 cm in thickness) with silicon windows for the THz measurements.

738

THz emitter 20

40 60 80 .^ 100 Wavenumber / cm

120

Fig. 1. A schematic diagram of the THz-TDS apparatus (a). Detected signals of pulsed THz radiation in the time domain (b) and its power spectrum (c).

3.

Results and Discussion

THz spectra of C-4 are shown in Fig 2. Six bands of C-4 are observed in a THz region from 5 to 90 cm"^ Among the six bands, C-4 has the most distinctive band at 26.9 cm"^ around which the largest dispersion of refractive index (Fig. 2(a)) and the strong absorption (Fig. 2(b)) are observed. These spectral features enable us to identify C-4 by the THz measurement. THz absorption spectra of the C-4 sealed in the envelope are shown in Fig. 2(d). By subtracting the featureless absorption of the envelope paper, the spectrum almost identical with that of C-4 can be obtained (Fig. 2(e)). Our results reveal that the inspection combining the THz spectroscopy and the THz imaging (not shown) can be a powerful tool for the screening of C-4.

10 20 30 40 50 60 70^80 90 Wavenumber / cm

10 20 30 40 50 60 70 ^80 90 Wavenumber / cm

10 20 30 40 50 60 70 30 Wavienmnber I cm

10 20 30 40 50 60 70^80 Wavenumber / cm

Fig.2. THz spectra of C-4: (a) refractive index, (b) absorption coefficient, (c) a photograph of C-4 hidden in an envelope, (d) THz absorption of C-4 with the envelope (solid line) and that of the envelope (dashed line), and (e) THz absorption spectrum of C-4 obtained by

739

subtraction of envelope absorption (solid line) and C-4 absorption of (a) multiplied by 0.05 (dashed line). THz absorption spectra of the inflammable liquids and the plastic bottle are shown in Fig. 3(a). The low THz absorption of plastic bottles allows us to examine liquids even if they are stored in the bottles. THz radiation, which is transmitted through all the inflammable liquids stored in the plastic bottle, gives sufficient intensity above the noise level in detection. No signal of transmitted THz radiation, however, can be observed for water or alcohol in plastic bottles. The difference in transmitted intensity of THz radiation for these liquids can be used for screening hazardous liquids. The refractive indexes of alkanes, shown in Fig. 3(b), increase with the carbon number of alkanes. Alkanes are major components of inflammable liquids originating in petroleum. The refractive index as well as the absorption in the THz region can be used for classifying the inflammable liquids. (b)

1.43^ 1.42 k-^ ^ 1.41 p : ^fl 1.40 r ~ nr—

1.39[7 1.38f 1.37r~ 1-36L 1.35F 1.34'^ 10 20 30 40.J 50 Wavenumber / cm

10 20 30 40^ 50 Wavenumber / cm Fig. 3. THz spectra of inflammable liquids, (a) THz absorption of gasoline (open circle), diesel oil (open square), benzine (open triangle), and plastic bottle (closed triangle). Absorption of the plastic bottle is subtracted in the spectra of the inflammable liquids, (b) the refractive index of w-alkanes, carbon numbers of which are 5 to 12frombottom to top.

4. Conclusion We succeeded in detection of the C-4 hidden inside an envelope and inflammable liquids packed in plastic bottles by the THz spectroscopy and the THz imaging. We demonstrated that the combination of these THz techniques provides a powerful tool for inspection and screening of hazardous materials. The THz inspection is promising for security application to chemical and biological hazardous materials.

References 1 B. B. Hu and M. C. Nuss, Opt. Lett. 20, 1716, 1995 2 K. Yamamoto et al, J. J. Appl. Phys. 43, L414 , 2004

740

Terahertz time-domain spectroscopy of surface plasmon polaritons on semiconductor surfaces J. Gomez Rivas, J. Saxler, M. Kuttge, P. Haring Bolivar and H. Kurz Institut flir Halbleitertechnik, RWTH Aachen, Sommerfeldstr. 24, D-52056 Aachen, Germany E-Mail: [email protected] Abstract; We present time domain measurements of terahertz surface plasmons SPPs on semiconductors. The low permittivity of semiconductors at THzfrequenciesgives rise to a strong SPPs confinement to the surface, while maintaining long propagation lengths. Surface plasmon polaritons (SPPs) are electromagnetic waves coupled to the oscillation of free electrons on the surface of conductive materials like metals [1]. These excitations propagate along the surface, have a strong field enhancement at the surface, and decay evanescently away from it. SPPs have therefore proven to be extremely usefiil for the detection and investigation of thin films deposited on metals, e.g. for chemical sensing or biomolecular analysis [2,3]. To date most of the work on plasmonics has been done at optical and near-infrared wavelengths using metals as the conductive material. However, when extending the operation of SPPs to longer wavelengths the very large permittivity of metals at low frequencies leads to a weak confinement of the SPPs to the surface. In this article we present terahertz time domain measurements of SPPs propagating on semiconductor surfaces, specifically on p-doped Si with a carrier concentration of N-lO^'^cm'^ and on undoped InSb. Semiconductors behave at THz frequencies as good metals in the optical range and they can therefore act as the basic building block for low frequency plasmonic applications. In order to appreciate the basic material dependencies of THz SPPs, it is usefiil to determine the value of characteristic length scales [1]. The decay length of SPPs into free space is given by Ld=l/Re{(27rv/c)[-l/(l+8ni)]^^^}, where c is the speed of light in vacuum, v the frequency, and Sm the complex permittivity of the conductive material. The SPPs propagate on the surface with an attenuation length given by La=l/Ini{(27cv/c)[8m/(l+ Sm)]^'^^]}. A large permittivity of the conductive material leads to a large decay length Ld, and to a long attenuation length La. In table 1 we have listed the values of 8m, L^ and L^ for gold, p-doped Si (A^=10^^cm' ^ ), and undoped InSb at 1 THz. To investigate the propagation of SPPs on semiconductor surfaces we use a modified THz-time-domain spectrometer, where we detect the time dependent electromagnetic field with sub-picosecond time resolution. A detailed description of the setup can be found in Ref [4].

741

1

^^ Ld (mm)

1 U (mm)

Gold -8.6-10^+i-6.2-10^ 50 6-10'

p-doped Si -31 + i-3.5-10^ 1.22 34

InSb

1

-91 + i-30 0.47

28

1

Table 1. Decay length intofreespace La and attenuation length La of SPPs at afrequencyof ITHz propagating on gold, doped Si and InSb surfaces with permittivitty 8m. Time domain transients of SPPs propagating 20 mm on the p-doped Si wafer are represented in Fig. 1(a) for different values of the out-coupling height of the knife over the sample /z2. The SPPs amplitude first increases as h2 augments, reaching a maximum when /?2 ~ 300|Lim. Further increase of h2 leads to a decrease of the amplitude. This dependence of the SPPs amplitude with h2 can be appreciated better if we calculate the Fourier integral (FI), by Fourier transforming the transients to obtain the spectra and integrating these spectra in the frequency range containing most of the THz amplitude, i.e., 0.2-2 THz. In Fig. 1(b) we plot the FI as a function of /22, where the black squares correspond to the measurements on the doped Si wafer and the grey circles to similar measurements on the InSb wafer. The equal increase of the FIs for small h2 in both samples reflects the increase of the coupling efficiency. For larger values ofh2 the FI is given by a superposition of the evanescent decay of SPPs away from the semiconductor surface and the coupling efficiency of the apertures, which difficult an accurate determination of Ld. However, we can see that the decrease of the FIs is on the order of 1mm as expected from the values of Ldin Table 1, and this decrease is more pronounced in the InSb sample.

time (ps)

500 1000 outcoupling height h, (nm)

Fig 1. (a) THz time domain transients of SPPs on a p-doped Si surface for different outcoupling heights hi. The transients are offset vertically for clarity, (b) Fourier integral of SPPs transients versus hi. The black squares correspond to SPPs on a p-doped Si wafer and the grey circles to SPPs on an InSb wafer. To obtain the attenuation length of SPPs along the semiconductor surface we measured the THz transients for different distances ?y between the in- and outcoupling apertures. These measurements for the p-doped Si wafer are plotted in Fig. 2(a), where, as can be appreciated, there is a monotonous decrease of the amplitude as ?y increases. The FI as a fimction of the distance between apertures or the longitudinal propagation distance is plotted in Fig. 2(b) with black squares

742

for the measurements on the Si wafer and with grey circles for those on the InSb wafer. Due to the small size of the InSb wafer (2") it was only possible to perform the measurements for small ?y. However, we can see that, as expected, the decrease of FI with ?y is similar in both semiconductors. As can be appreciated in the Si measurements, for propagation distances smaller than 20 mm the FI decreases more abruptly than for longer ?y. As the two apertures get closer a significant fraction of the scattered THz radiation at the first aperture gets through the second aperture and is detected. This undesired signal increases the FI for small ?y. For ?y = 20 mm the mentioned contribution of the scattered THz radiation is negligibly and the decrease of the FI can be attributed solely to the attenuation of SPPs. The solid line in Fig. 2(b) is an exponential fit to the long ?y measurements fi-om which an attenuation length La of 51mm with an asymmetric error of -18 mm and + 1 1 0 mm is found. This attenuation length has to be compared with the theoretical value of Za weighted over the setup response in the fi-equency range 0.2-2 THz, i.e., the range in which the FI is calculated, which in our case is La =114 mm. Within our experimental accuracy we find thus a good agreement between experiments and theoiy and confirm the long propagation lengths of SPPs on semiconductor surfaces. 0.4

(a) _^_^,,,,.>.-^^.,^^ -_——'O"^ ————«.._

.2, 0.2 T3

i

(b)

Ajfa4.D mm

.„. 0.3

w>—*.

0.1

.

^""^

6 time (ps)

12.5 17,5... '22.5

-

2Z5_ 32.5 37.5 42.5

0

10 20 30 40 50 longitudinal propagation distance 4 / ( m m )

Fig 2. (a) THz time domain transients of SPPs propagating a distance ?y on the surface of a pdoped Si wafer. A vertical offset has been introduced for clarity, (b) Fourier integral of SPPs transients versus the propagation distance ?y. We gratefully acknowledge financial support fi"om the European Union through the TMR project Interaction and by the Deutsche Forschungsgemeinschaft. 1.

2. 3. 4.

H. Raether, "Surface plasmons on smooth and rough surfaces and on gratings", Springer Tracts in Modem Physics Vol. I l l (Springer-Verlag, Berlin 1988). B. Liedberg, C. Nylander, and I. Limdstrom, Sensors and actuators 4 (2), 299(1983). S. Lofas, M. Malmqvist, I. Ronnberg, E. Stenberg, B. Liedberg, and I. Lundstrom, Sens. Actuators B 5, 79 (1991). J. Saxler, J. Gomez Rivas, C. Janke, H.P.M. Pellemans, P. Haring Bolivar, and H. Kurz, Phys. Rev B 69,155427 (2004).

743

Evaluation of Complex Optical Constants of Semiconductor Wafers Using Terahertz Ellipsometry Takeshi Nagashima and Masanori Hangyo Laser Terahertz Division, Institute of Laser Energetics, Osaka University 2-6 Yamadaoka, Suita, Osaka 565-0871, Japan E-mail: [email protected] Abstract. We have developed terahertz ellipsometry by combining ellipsometry with time domain spectroscopy in the terahertz frequency region. Complex optical constants of Si wafers with various carrier concentrations are measured by the terahertz ellipsometry.

1. Introduction Transmission and reflection measurements using the time domain spectroscopy (TDS) in the terahertz (THz) region have been used to evaluate complex optical constants of materials. Usually, in order to obtain transmittance or reflectivity, one needs not only measurements for samples but also those for reference. For in situ measurement, some sort of TDS system without sample replacement is required. So far, we have proposed and developed "THz ellipsometry" in which the THzTDS is combined with ellipsometry that needs no replacement of the samples in principle [1]. Purpose of this study is to show that the complex optical constants of semiconductor with various carrier concentrations (or resistivities) can be deduced by using the THz ellipsometry.

2. Experiment Figure 1 shows a schematic diagram of the THz ellipsometry system [1]. THz pulses were generated by a photo-conductive antenna excited by a mode locked Tirsapphire laser with -80 fs time width pulses. S or p polarized THz pulses were extracted by a wire grid type polarizer in front of a sample. The sample was mounted on a sample holder with an aperture of 80 mm in diameter. The polarized THz pulses with the incident angle of 45° were reflected by the sample. The reflected THz pulses were focused on a detection antenna triggered by delayed laser pulses. Changing the delay time of the trigger pulse, the waveforms of the reflected THz pulses were measured. By using the complex Fourier transformation, the waveforms of s and p polarized THz pulses were transformed to complex spectra rs=|rs|exp(i6s) and rp=|rp|exp(i6p), respectively. From rs and rp, ellipsometric angles W=tan"^(|rp|/|rs|) and A=5p-5s are obtained.

744

Fig. 1. Schematic diagram of the THz ellipsometry system. The complex optical constants of the sample can be deduced from W and A by using analyses appropriate for the samples.

3. Results and Discussion Figure 2 shows the frequency dependence of the ellipsometric angles of Si wafers with resistivities of 0.136 Qcm and 126 Qcm, respectively. As seen in Fig. 2, tan^ and A change monotonically with frequency, which exhibits that there is no multiple reflection of the THz pulses in the sample due to strong attenuation of the THz pulses in the wafer with the low resistivity (0.136 Qcm). For this wafer, the frequency dependence of the complex refractive index n-iK can be deduced by using well known analytical expressions used in the conventional ellipsometry for bulk sample [2].

0.0

0.2

0.4 0.6 0.8 Frequency (THz)

1.0

0.0

0.2

0.4 0.6 0.8 Frequency (THz)

Fig. 2. Frequency dependence of ellipsometric angles W and A for a Si wafer with resistivity of (a) 0.136 Qcm and (b) 126 Qcm. Figure 3 shows the frequency dependence of the complex refractive index deduced from the data shown in Fig. 2. The precision of the data points in the spectra is not so good, however we have confirmed that the precision is improved when the measurement of W and A are carried out at angles near the principle angle of the samples. The result for the low-resistivity sample shown in Fig. 3(a) exhibits the Drude type dispersion caused by the response of free carriers and

745

show a good agreement with previous report on the same wafer using a normal reflection measurement in the THz frequency region [3].

r- (a) I '.

r

. 0.2

Si 0.136 QcrrJ eo=45cieg. 1

. . K 1 0.4 0.6 0.8 1.0 Frequency (THz)

0.2

0.4 0.6 0.8 Frequency (THz)

1.0

Fig. 3. Frequency dependence of ellipsometric angles W and A for a Si wafer with resistivity of (a) 0.136 Qcm and (b) 126 Qcm. On the other hand, periodic structures are observed in the spectra of W and A for the high resistivity wafer (126 Qcm) as seen in Fig. 3(b). These periodic structures originate from the multiple reflections of the THz pulses in the sample caused by the higher transmittance of the sample. Numerical calculations taking into account such multiple reflections of the THz pulses in the wafer are carried out to deduce the complex refractive index from the spectra of the ellipsometric angles shown in Fig. 3(b). The results are shown in Fig. 3(b). There are small periodic variations in the real and imaginary parts of the complex refractive index spectra. The small periodic variations originate from the fact that the interference effect of the THz pulses in the wafer is not eliminated completely by the numerical calculations. However, we can see that the real and imaginary parts of the complex refractive index have almost no frequency dispersion except the artificial periodic variations. The values of n and K are consistent with ones reported previously.

4. Summary It has been shown that the complex optical constants of the semiconductor wafers with various resistivities are deduced by using the THz ellipsometry. For the low resistivity wafer in which no multiple reflections of the THz pulses occur, the complex refractive index is deduced by the simple analytic expression. For the high resistivity sample, the complex refractive index spectra are deduced by using the numerical calculations taking into account the multiple reflections of the THz pulses in the sample.

References 1 T. Nagashima and M. Hangyo, in AppL Phys. Lett. Vol. 79, 3917 (2001). 2 M. Born and E. Wolf, Principles of Optics, 7th ed. (Cambridge University Press, Cambridge, 1999), Chap. 14. 3 S. Nashima, O. Morikawa, K. Takata, and M. Hangyo, in AppL Phys. Lett. Vol. 79, 392 (2001).

746

Terahertz two-dimensional spectroscopic imaging with a high speed CMOS camera Hideaki Kitahara, Taijiro Yonera , Fumiaki Miyamaru, Masahiko Tani, and Masanori Hangyo Institute of Laser Engineering, Osaka university The 21st Century Plaza, 2-6 Yamadaoka, Suita, Osaka 565-0871, Japan E^nail: [email protected] Abstract. We have developed a THz imaging system based on two-dimensional EO sampling using a high-speed CMOS camera. With the dynamic subtraction and laserintensity normalization synchronized with the laser repetition rate, a high SNR THz spectroscopic imaging is possible in a short time.

1.

Introduction

Imaging with pulsed THz radiation (THz imaging) is attracting considerable attention recently because of its potential applications in many fields. By using the characteristic absorption or dispersion spectra in the THz frequency region, we can obtain THz spectroscopic images, with which the chemical structures of the imaging objects can be specified. In the standard THz imaging system based on a pair of photoconductive emitter-detector, however, the spectroscopic THz imaging requires considerable measurement time because of the raster scanning of the imaging target at the focal plane of the THz beam in addition to the scanning of the time-delay to obtain the signal waveforms. The method of two-dimensional electro-optic (2-D EO) sampling [1] with an IR digital camera enables us to take THz images without the raster scan and significantly shortens the measurement time. However, the imaging quality is poor compared to that of the raster-scan system because of the reduced intensity of the expanded THz beam to cover the imaging area on the samples. In addition, the large intensity fluctuation of the low-repetition-rate amplified laser used in 2D-E0 measurements further reduces the signal-to^oise ratio (SNR) of the images. By averaging images we can increase the SNR but the averaging efficiency is low for the low repetition laser (Note that to increase the SNR by N times, averaging of N^ times events are required). In this paper, we report a 2D-B0 THz spectroscopic imaging system using a high speed CMOS camera operated in a differential mode with dynamic normalization by the laser intensity. A considerable improvement of the image SNR was achieved with this method, making it possible to take spectroscopic THz images in a short time with a reasonable SNR.

747

2. Experimental The setup was a standard 2D-EO sampling except that we employed a high-speed CMOS camera (128 x 128 pixels) with a maximum frame rate of 1 kHz [2], in stead of a low frame^'ate CCD camera. The CMOS camera, synchronized to the \-kHz amplified femtosecond laser (X~ 800 nm, 6t~ 100 fs), took every shot of the image sampled by each probe laser pulse. The EO signals were detected in the cross-polarized configuration with a small^hase-bias [3]. For the differential detection of EO signals, an optical chopper transmitted every second pump laser pulses, which excited a ZnTe emitter (3 mm) and generated THz radiation pulses at 500 Hz. The THz radiation beam and the probe laser beam was combined coUinearly using a Si plate and directed to the detector ZnTe (4 mm). The CMOS camera detected the sequence of the EO-modulated optical images at 1 kHz, which alternately contained the THz EO signals. By subtracting the (2/+l)th background image (Imgbkg[2/+1]) from the 2/th image with THz signal on the background (ImgTHz[2i]), we obtained the ith differential image, Imgsig[/], which contained the THz signal alone. To reduce the noise due to the intensity fluctuation in the amplified laser ("5%), we normalized the images, ImgTHz[2/] and Imgbkg[2/+1], respectively by the corresponding prove laser intensity, I[2z] and I[2/+l], measured by selected pixels on the CMOS camera. The ith normalized differential image is thus expressed as

Img,^J2/]_Img,,J2/ + l]^

(1) I[2i] I[2i +1] The influence of the fluctuation in the pump-4aser intensity is not included in this equation since it is difficult to evaluate correctly the saturation effect for the laser intensity and the nonlinearity of the EO signals to the THz radiation field amplitude in the fmite^hase-bias condition. Even though the compensation for the pump4aser4ntensity fluctuation is not perfect, we can significantly increase the SNR with the dynamical normalization by the probe laser intensity. Img,ig[/] =

10

Time(ps)

Fig. 1 THz waveform measured with a 2-ms integration time for each delay step.

748

1.0

1.5

2.0

2.5

Frequency (THz) Fig. 2 THz power spectra for the reference (solid line) and for a L5-iTim L-cystine pellet (broken line).

3. Results and Discussion Figure 1 shows a signal waveform measured at a pixel near the THz beam center with a 2^iis integration time (single differentiation) at each delay step. The SNR (the signal peak to rms noise floor) is approximately 100. Figure 2 shows Fourier transformed spectrum of THz radiation without sample (reference spectrum) and that with a sample taken with a 1-sec integration time at each step. The sample was a 1.5-mm thick pellet of L-cystine (an oxidation product of two L-cysteine amino acids). The spectrum of L-cystine shows a distinct absorption band at 0.73 THz. By using this spectral maker we can make a spectroscopic imaging for this substance. We measured a polyethylene pellet containing a small piece of Lcystine. Figure 3 shows the spectroscopic images ( 4 x 4 mm^) of the sample pellet at 0.73 THz, where a strong absorption band of L-cystine exists. In the THz image the L-cystine is clearly distinguished from the surroundings of polyethylene. Fig. 3. THz spectroscopic images at 0.73 THz of a sample. The 0.73-THz image shows a piece of L-cystine (the oval bright area) contained in the polyethylene pellet. The bright area near the bottom of the image is resulted from an interference effect by an aluminum plate used to limit the THz beam. One pixel corresponds 40 |um on this image. 80

60

40

20

Pixel numbers

4. Conclusions We have demonstrated a fast THz imaging system based on 2D-E0 sampling using a high-speed CMOS camera operated in the differential mode with dynamical normalization by the laser intensity. A high SNR spectroscopic imaging is possible with our system, which is useful for practical applications. Acknowledgements. This work was supported by the Ministry of Education, Culture, Sports, Science and Technology, Japan, through the Terahertz Optics Project for Medical Application lead by Dr. J. Nishizawa.

References L Q.Wu, T.D.Hewitt, and X.-C.Zhang, Appl. Phys. Lett. 69, 1026, 1996. 2. F. Miyamaru, T. Yonera, M. Tani and M. Hangyo, Jpn. J. Appl. Phys., 43, L489, 2004. 3. Z. Jiang, F. G. Sun, Q. Chen and X.-C. Zhang: Appl. Phys. Lett. 74,1191, 1999.

749

Single-shot terahertz imaging Rakchanok Rungsawang, Aya Mochiduki, Shin-ichi Ookuma and Toshiaki Hattori Institute of Applied Physics, University of Tsukuba, 1-1-1 Tennodai, Tsukuba, Ibaraki, 305-8573 Japan E-mail: [email protected]

Abstract. Single-shot detection of two-dimensional terahertz (THz) imaging was conducted using a half-cycle THz pulse generated from a large-aperture photoconductive antenna. The appropriate timing of the probe pulse for imaging an object was discovered. The single-shot imaging enables observation of ultrafast phenomena at a frame rate of 1000 frames/s.

1.

Introduction

Imaging with electromagnetic pulses in THz region (10 GHz-10 THz) is attractive for real-world applications because it can provide a noninvasive monitoring method. Furthermore, it can be applied for time-resolved spectroscopy that has the potential to read out both amplitude and phase information simultaneously. Since the THz pulses have pulse widths less than 1 ps, they can be used in many types of time-resolved measurements. For the observation of ultrafast single-time events, such as explosion and melting, singleshot THz imaging should be introduced. Single-shot THz imaging can also be applied to real-time imaging of various objects. Two-dimensional simultaneous detection of THz electric field generated from an unbiased photoconductive antenna using electro-optic (EO) detection was first reported by Wu et al. [1]. Their readout time of the experiment was 0.133 s. Spatio-temporal detection with a higher frame rate was performed by Jiang and Zhang [2]. They achieved a capture rate up to 69 frames/s with an improved signal-to-noise ratio using the dynamic subtraction technique. The THz movie of a moving biological object was reported at a frame rate of 10 frames/s [3]. The essential components required for the single-shot THz imaging are a powerful THz source, a sensitive detection method, and a high-speed camera. We used a large-aperture biased photoconductive antenna to produce a high electric field. This source gives a simple output waveform, /.e,, almost half-cycle shape at focus. The pulse width was about 500 fs. The waveform can change a great deal along its propagation, and deviate from the half-cycle pulse at the observation plane [4]. We found that there is a certain time window that is suitable for imaging at a fixed time delay. We adopted the optical heterodyne detection method in th'" EO sampling for sensitive field image detection [5,6]. We used a high-speed CCD camera, which enabled to capture one THz image for every THz

750

pulse. Our system is run at a 1 kHz repetition rate. The raw data were passed through digital imaging processes to reduce noises.

2.

Experiment

The imaging setup was similar to that described in a previous paper [5]. By scanning the delay time of the probe pulse, we found that images are clearly seen only at a specific time window. Interestingly, that is not the peak time of the THz pulse. In Fig. 1, the THz field waveforms observed at the focus and on the image plane are shown. The focal length of the lens w a s / = 98.5 mm, and the image plane was at 1.5/from the focus. From the plot, it is seen that the main pulse shape at the focus (dotted line) is almost half-cycle having a tail with a small negative value. The negative tail is attributed to diffraction of low-frequency components before being focused. At the object location, on the other hand, the pulse (solid line) is broadened. The long rise time shows diffraction of highfrequency components after being focused. The period when images were clearly observed corresponds to the time region where the THz field has steep transient, namely at the time delay 0-lps in the figure, where the high frequency components dominate. By fixing the time delay at time delay 0.5ps, we succeeded in observing images in real time.

Time Delav (n.s>

Fig. 1. On-axis temporal waveforms on the focal plane (dotted line) and on the image plane (solid line) measured using the conventional balanced detection method.

3.

Results and Discussion

The THz pulses passed through an object sample that was placed at distance 3 / The image of the object was projected to a 18 x 20 mm^ ZnTe crystal. The sample object was a metal rod of 2.6 mm in diameter. The rod was hung by a string and, when at rest, placed vertically at the center of the THz beam. Realtime images of the rod, while swinging, were obtained. The electric field images

751

were calculated from the CCD data and divided by the reference image obtained when the sample was removed. By this procedure, the artificial pattern attributed to the characteristics of the focused THz beam itself was removed [7]. The resulted images contained noises which originated from the CCD camera and the instability of the laser. By processing the images using a Gaussian filter, good image quality was achieved. Two images of the rod selected from a series are illustrated in Fig. 2. The spatial resolution of the images is limited by the low central frequency of the THz pulses.

1.20 L58 Fig. 2. Single-shot images of a metal rod which moved from left to right. Image (b) was obtained 20 ms after image (a). These images were divided by the reference image obtained without the sample object and processed using a Gaussian filter. The dimension of images is 8.8 x 7.4 mm The bar indicates the gray scale showing in percentile range 10%-90%.

4.

Conclusions

Using THz pulses from a large-aperture biased photoconductive antenna and a high-speed CCD camera, single-shot detection of THz images was achieved. Using this technique, we obtained a high-speed movie of a moving object in real time at a rate of 1000 frames/s. Potential applications of this technique include high-speed movies and time-resolved spectroscopic studies of single-time events in the THz frequency region. References 1 Q. Wu, T.D. Hewitt, and X.-C. Zhang, Appl. Phys. Lett. 69, 1026, 1996. 2 Z. Jiang and X.-C. Zhang, Opt. Express 5, 243, 1999. 3 M. Usami, T. Iwamoto, R. Fukasawa, M. Tani, M. Watanabe, and K Sakai, Phys. Med. Biol.. 47, 3749,2002. 4 T. Hattori, R. Rakchanok, K. Ohta, and K. Tukamoto, Jpn. J. Appl. Phys. 41, 5198,2002. 5 T. Hattori, K. Ohta, R. Rungsawang, and K. Tukamoto, J. Phys. D: Appl. Phys. 37, 770, 2004. 6 Z. Jiang, F.G. Sun, and Q. Chen, Appl. Phys. Lett. 74, 1191, 1999. 7 R. Rungsawang, K. Ohta, K. Tukamoto, and T. Hattori, J. Phys. D: Appl. Phys. 36,229,2003.

752

Ultrabroadband detection of multi-THz field transients with GaSe electro-optic sensors Carl Kubler\ Rupert Huber^, Stefan TlibeP, and Alfred Leitenstorfer^ ^ Fachbereich Physik, Universitat Konstanz, D-78457 Konstanz E-mail: [email protected] ^ Physik-Department Ell, TU Munchen, James-Franck-Str., D-85748 Garching, Germany Abstract. Electro-optic detection with unprecedented bandwidth is implemented with GaSe sensors utilizing phase matching. We directly record transform-limited 28-fs pulses containing spectral components beyond 120 THz. Continuous tunability of the center frequency is demonstrated between 14 THz and 31 THz. Two distinct methods are commonly used today to measure the electric field of broadband terahertz (THz) waves: photoconductive and freespace electro-optic sampling [1]. Photoconductive sampling only recently entered the frequency regime above 20 THz [2]. However, electro-optic sensing [3, 4] is still superior for detection of even higher frequencies throughout the midinfrared range [5]. Unfortunately, the response bandwidth of isotropic electro-optic materials like ZnTe and GaP is limited by the mismatch between the THz phase velocity and the group velocity of the optical gating pulse [3, 6]. For certain applications, the severe structuring of the amplitude and phase spectrum requires a correction of the recorded transient [4]. The solution to this problem is to employ a birefringent detector that allows for phase matching of the involved pulses. In this contribution, we demonstrate such a phase matched electro-optic detection scheme utilizing GaSe crystals. In our experiment, we employ 10-fs pulses derived from a Ti: sapphire laser that operates at a 64 MHz repetition rate with an average output power of 1 W. The pulses are split into a pump for the generation and a probe for the detection of the THz transients. The generation is based on optical rectification in a z-cut GaSe and implemented as described in Ref. 7. The THz radiation is focused on a second zcut GaSe crystal. This crystal may be tilted by an angle 0det, the phase matching angle, about a horizontal axis perpendicular to the direction of the time-delayed probe beam. The horizontally polarized probe beam interacts with the THz beam in a nonlinear optical process which may be interpreted as the inversion of the emission [7, 8]: Higher frequency near infrared light is generated and superimposed on the orthogonally polarized probe beam. The overall polarization state of the probe beam is modified depending on the THz field strength. The difference AI between two balanced photodiodes, caused by the change in polarization, is proportional to the applied THz electric field present in the electro-optic crystal during propagation of the probe pulse. For a maximum detector response, it is crucial to minimize the phase mismatch Ak between the involved pulses. This condition may be satisfied for different frequency ranges by varying the angle 9det exploiting the natural birefringence of GaSe. Using lock-in amplification the detection sensitivity

753

15

20

wavelength (|jm) 7.5 5.0 3.8 3.3

2.5

40 60 80 100 120 frequency (THz)

Figure 1. (a) THz field transient generated in a 20-|im-thick GaSe crystal as detected with a GaSe sensor of 30 \xm thickness. (b) Corresponding amplitude and phase of the transient in (a). The dashed curves are spectra of the same pulses recorded with a 12-|im-thick ZnTe crystal.

0

10

20 30 40 frequency (THz)

Figure 2. Normalized amplitude spectra detected with a 50-|im-thick GaSe crystal at various phase matching angles Bdet- The THz pulse is generated in a 30-|am-thick GaSe emitter at a phase matching angle of 6 = 52°. The dotted line represents the calculated emission spectrum.

is only limited by the shot noise (5 x 10'^ Hz"^^^) of the photon flux of the gating beam. Thus, we achieve a high dynamic range larger than 10"^. Figure la depicts the time trace of a linearly polarized THz pulse generated with a 20-jim-thick GaSe emitter (0 = 57°) as detected with a GaSe detector (0det= 60°) of 30 fim thickness. The orientation of the crystals was optimized for type-II phase matching resulting in generation of vertically polarized THz radiation. The pulse duration is determined to be 28 fs (FWHM) via a Gaussian fit to the intensity envelope. Figure lb displays the amplitude and phase spectra of the time trace in Figure 1 a. The amplitude spectrum peaks at 33.8 THz and extends from 7 THz to beyond 120 THz (k = 2.5 |im). Without correcting for the detector response, we find an unprecedented 3 dB bandwidth of 41 THz. Remarkably, at 80 THz the amplitude is reduced with respect to the maximum by only one order of magnitude. From the intensity spectrum, a FWHM of 14.5 THz is extracted. The THz pulse is transform limited with a bandwidth product of 0.41. For comparison, the same transient was detected using a 12-|im-thick ZnTe electro-optic crystal (dashed curve in Figure lb). The advantages of the phase matched detection scheme are obvious: The detector response shows no roll off due to group velocity mismatch as it is found in ZnTe. In fact, the sensitivity of the GaSe detector is maximal around 34 THz where the ZnTe response shows a local minimum. Consequently, the GaSe spectrum lacks the phase jump that is connected with the group velocity mismatch in ZnTe. Instead, we find an almost flat spectral phase from 10 THz up to 105 THz in accordance with the minimal bandwidth product calculated above.

754

The THz wave field amplitude measured with a GaSe crystal exceeds the one detected by a ZnTe crystal with comparable detection bandwidth due to the larger second order electro-optic coefficient and the increased effective interaction length [9]. In principle, the bandwidth of ZnTe detectors could be increased by reducing the crystal thickness. However, this procedure results in a decrease of interaction length and an even smaller detector response. Moreover, fabrication and handling of crystals thinner than 10 \xm is extremely difficult. Tunability of the detector response is demonstrated in Figure 2. We use ultrabroadband THz pulses emitted from a 30-iim-thick GaSe crystal (6 = 52°) with a calculated emission spectrum [7] displayed as a dotted line in Figure 2. From this ultrabroadband spectrum, we selectively cut out spectral windows by changing the angle 0det of a 50-iim-thick detector. Because of the larger thickness and therefore stricter phase matching condition, this sensor has a reduced phase matching bandwidth as compared to the 30-|Lim-thick sensor. At Gdet = 60° perfect phase matching is achieved around 31 THz. With decreasing angles, the center of the detection window shifts down in frequency to approximately 14 THz at Qdet = 35°. Similar to the emission, the detection is limited at the low frequency end by the strong dispersion in GaSe close to the Reststrahlen band. In conclusion, we have presented coherent phase matched detection of ultrabroadband infrared pulses in GaSe. If combined with phase matched optical rectification, a powerful multi-THz spectrometer is obtained that represents a versatile tool for time and field resolved infrared spectroscopy: On the one hand, narrow frequency bands may be selected and investigated with enhanced sensitivity by tuning of the phase matching angle. Alternatively, ultrabroadband measurements which cover the complete frequency range from the far to the near infrared all at once may be performed with an undistorted spectral phase.

References 1 2 3 4 5 6 7 8 9

S.-G. Park, M. Melloch, and A. Weiner, Appl. Phys. Lett. 73, 3184, 1998. S. Kono, M. Tani, and K. Sakai, Appl. Phys. Lett. 79, 898, 2001. Q. Wu and X.-C. Zhang, Appl. Phys. Lett. 71, 1285, 1997; 70, 1784, 1997. A. Leitenstorfer, S. Hunsche, J. Shah, M. Nuss, and W. Knox, Appl. Phys. Lett. 74, 1516, 1999. A. Brodschelm, F. Tauser, R. Ruber, J. Sohn, and A. Leitenstorfer, in Ultrafast Phenomena XII, Springer Series in Chemical Physics, Edited by T. Elsaesser, S. Muhkhamel, M.Mumane, and N.Scherer (Springer, Berlin, 2000), Vol. 66. G. Gallot, J. Zhang, R.W. McGowan, T.-I. Jeon, and D. Grischkowsky, Appl. Phys. Lett. 74, 3450, 1999. R. Ruber, A. Brodschelm, F. Tauser, and A. Leitenstorfer, Appl. Phys. Lett. 76, 3191,2000. K. Liu, J. Xu, and X.-C. Zhang, accepted for publication in Appl. Phys. Lett. 85, 2004. R. Kaindl, F. Eickemeyer, M. Woemer, and T. Elsaesser, Appl. Phys. Lett. 75, 861, 1999.

755

Terahertz field detection beyond 30 THz by proton-bombarded InP photoconductive antennas Tze-An Liu ^ Masahiko Tani^, Makoto Nakajima^, Masanori Hangyo^, Kiyomi Sakai^, Shin-ichi Nakashima"*, and Ci-Ling Pan ^ ^ Institute of Electro-Optical Engineering, National Chiao Tung University, 1001, Ta-Hsueh Rd, Hsinchu, Taiwan 300, R.O.C. E-mail: [email protected] ^ Research Center for Superconductor Photonics, Osaka University, 2-1 Yamadaoka, Suita, Osaka 565-0871, Japan. E-mail: [email protected] ^ Kansai Advanced Research Center, Communications Research Lab, 588-2 Iwaoka, Kobe 651-2492, Japan. 4 National Institute of Advanced Industrial Science and Technology, Power Electronics Research Center 1-1-1, Umezono, Tsukuba, Ibaraki, 305-8568, Japan Abstract, We show that photoconductive antennas fabricated on proton-bombarded InP (InP:H^) substrates with a proton dosage of 1x10^^ ions/cm^ are promising for THz detection.

h

Introduction

Recently, we demonstrated ultrabroadband (>30THz) THz radiation detection using arsenic-ion-implanted GaAs (GaAs: As^) photoconductive antenna.[l] However, the shallow arsenic ion-implanted layer (--lOO nm), resulted in lower signal to noise ratio (SNR) for THz detection than LT-GaAs. In this work, we investigate the performance of InP:H^ PC antennas as ultrabroadband THz detector. The bandwidth and SNR of the InP:H^ PC antenna were compared to those of an LT-GaAs one and discussed in relation to the photoconductive gain and resistivity.

2. Experimental Methods Four InP:H^ PC antennas were prepared by bombarding {100)-oriented Sl-InP substrates with 180 keV proton with dosages of 1x10^^ 3x10^^ 1x10^^ and 3x10^^ ions/cm^, respectively. The carrier lifetimes of all InP:H^ and LT-GaAs samples used in this experiment were found to be shorter than 2 ps. A microstripline dipole antenna with a 30-jj,m dipole length and a 5-^m PC gap was fabricated on each PC substrate. The experimental setup was as reported previously. [1]

756

3. Results and Discussions Figure 1 shows the current-voltage (I-Y) characteristics for InP:H'^ and LT-GaAs PC gaps at nearly zero bias {< 0.2 V) without and with laser excitation. The resistance of the InP:H^ PC gaps, which decreased as the ion dosage increased, which can be explained by the presence of the shallow defects. [2] The THz waveforms detected by each InP:H^ and the LT-GaAs PC antenna are shown in Fig. 2(a). The signal amplitudes in InP:H^ PC antennas are all larger than our previous GaAs.As^ device.[1] Figure 2(b) illustrates the Fourier transformed spectra of the PC-detected THz waveforms. The high frequency ends of the spectral distribution for InP: H^ and LT-GaAs PC antenna are higher than 30THz.

0.05

0.10

0.15

0.05

Voltage (V)

0.10

0.1 S

Voltage (V)

(a)

(b)

Fig. L Current-voltage (I-V) measurement (a) without and (b) with optical illumination for InP:H^ doses of 10^^ (square dot), 3x10^^ (circle dot), 10^^ (triangular dot), 3x10^^ (diamond dot) ions/cm^ and LT-GaAs (open circle) PC antennas with 5 \xm gap in the weak bias case.

nP ;H ', 1 0 " Sons/cm '

.0

0.2

0.4

0.6

8.8

1.0

T i m e d e l a y (p s )

(a)

1.2

1.4

10 20 30 F r e q u e n c y (THz)

40

(b)

Fig. 2. Time-resolved ultrabroadband THz radiation (a) waveforms and (b) Fourier transformed amplitude spectra detected by a PC antenna fabricated on InP:H^ at various dosages and LT-GaAs. Some spectra are enlarged for easier viewing.

757

The main contributions to the noise in a PC detector are the Johnson noise (or thermal noise) and the laser shot noise.[3] The Johnson noise is inversely proportional to the square-root of the resistance, 1/4R • We have plotted the noise against l/V^ in Fig. 3. •

Data point linear fitting curve

IS'

•'i»-

t3 «

0.01

0.02

0.03

mP:H , 3x10

0.04

Relative conductivity ( 1 / a f * - J j -

Fig, 3. Noise level of different InP:H^ samples and relative to that of a LT-GaAs PC antemia are plotted as a function of conductance. Circle dot is the data point; dashed line is the fitting curve.

4. Conclusions In summary, we have investigated the performance of InP:H^ PC antennas as ultrabroadband THz wave detector. With THz radiation generated from a thin ZnTe emitter excited by 15-fs optical pulses, the detectable frequency distribution was confirmed to be about 30 THz. The peak THz signal of the InP.H"^ (10^^ ions/cm^) PC antenna is slightly higher than that of the LT-Ga.A.s one, while the SNR of the former is about half as high as the latter. This can be improved by increasing the resistivity of InP:H^ through optimizing the ion dosage level and/or the annealing condition. InP:H^ could thus be a promising material as the photoconductive substrate for ultrabroadband PC antennas. A c k n o w l e d g e m e n t s . Hiis work was supported in part by the National Science Council of R.O.C. under Grants No. NSC 92-2215-E-009-030 and Program for the Pursuit of Academic Excellence of the Ministry of Education, R.O.C.

References 1 T. A. Liu, M. Tani, M. Nakajima, M. Hangyo and C. L. Pan, Appl. Phys. left 83, 1322, (2003). 2 H. Boudinov, J.P. De Soiiza, and C. Jagadish, Ntwl Instr. AndMeth. B. 175,235, (2001). 3 M. Tani, K. Sakai and H. Mimura, Jpn. 1 Appl Phys. 36, LI 175, (1997).

758

THz Wave Near-Field Emission Microscope Tao Yuan\ Hongkyu Park^'^, Jingzhou Xn\ Haewook Han^'^, and X.-C. Zhang^ ^ Center for THz Research, Rensselaer Polytechnic Institute, Troy, NY ^ Center for Terahertz Photonics, POSTECH, Pohang, Kyugbuk, Korea Abstract: THz wave near-field emission microscopes provide THz emission from nanoscale semiconductor through femtosecond laser excitation. Metal/semiconductor surface measurements show that the resolutions are less than 1 and 100 nm in vertical and lateral directions, respectively.

1. Introduction A THz time-domain spectroscopy (THz TDS) allows an exploration of the rich spectroscopic information on molecular vibrations, rotations, and other lowenergy transitions in biological and organic compounds, and semiconductor structures. When a near-field technology is applied to THz imaging, the interaction between THz wave and sample surfaces is limited effectively to a small area, making it possible to achieve microscopic imaging with sub-wavelength resolution. There have been several approaches for the sub-wavelength resolution such as sub-wavelength aperture [1], dynamic aperture [2-4], and recently apertureless THz near-field microscopes [5-7]. Here, we report an apertureless THz near-field emission microscope which the THz signal is emitted from the sample surface by optical pumping. Our measurements on metal/semiconductor surfaces show that with this technique it is possible to obtain nanometer spatial resolutions.

2. Experimental set-up Figure 1 is the schematic diagram of a THz near-field emission microscope, consisting of a metal tip mounted on a piezo actuator, and a THz detection system. The femtosecond laser is incident on a semiconductor wafer at the Brewster angle with a power ranging from 10 to 150 mW. The generated THz wave was collected by an off-axis parabolic mirror, and detected by EO methods. With the presence of a metal tip, the laser induced dipole moments are coupled with the tip, and part of the THz wave is scattered by the tip. It is the scattered signal that provides the local spectroscopic information. The typical bias for the tip is from -3 to 3 V for DC, and from 0 to 3 V for AC. These bias voltages will also induce additional dipole moments at the sample surface and thus change band bending at the semiconductor surface near the tip, providing additional means to characterize the surface. To measure the THz signal

759

scattered from the tip, the AC bias voltage to the tip was used as a reference input for the lock-in amplifier. Various tips, Tungsten or Pt-Ir, with tip diameters ranging from less than 100 nm to 1 jxm were used in this experiment. THz pulse AC bias

DC bias

Fig. 1. A schematic of the THz frequency microscope setup.

3. Results and discussion —o— Normalized Current —•— Normalized THz . Step Size ~0.6nm

/"X:::-A/

0.0 o
436

440

444

Relative Tip Z Position (nm)

Fig. 2. Approach curves for peak THz signal and tip current

Fig. 3. THz signal measured while scanning a tip across the edge of a gold film on an InAs substrate.

Shown in Figure 2 is the THz peak signal and tip current, measured simultaneously, while the tip was approaching a p-type (1x10^^ cm"^) InAs surface. The tip was driven by a piezo actuator with a step size of ~ 0.6 nm. It should be noted that the onsets of the tip current and tip signal appeared almost exactly at the same tip location. This indicates that the tip current and THz signal are closely correlated to each other. The 10% to 90% transition in the tip signal occurred within 1 nm, demonstrating nanometer resolution in the vertical direction. Figure 3 shows the THz signal scanned across the edge of a gold film on a ptype InAs substrate. The thickness of the gold film was 20 nm, and the tip diameter was ~ 100 nm. The scanning range was ~ 5 //m with a step size of 100 nm.

760

It was found that the temporal shape and polarity of the THz signal changed abruptly within 100 nm when the tip was scanned from the metal to InAs surfaces. It is expected that the lateral resolution will be improved if the tip diameter is reduced. The proposed THz emission microscope has advantages over the conventional THz microscope. The coupled dipole interaction between the tip and sample surface can be tailored by tuning the wavelength of the pump laser and thus selectively exciting dipole moments below the sample surface. Further, if the exited dipole moments are localized in nanostructures such as quantum dots and wires, the spatial resolution will be significantly enhanced.

4. Conclusion We have observed the THz emission from the STM tips on semiconductor surfaces. The measured THz signal shows a sharp transition within nanometer scales during the tip approach and lateral scanning, which indicates the potential of the THz near-field emission microscopes for THz imaging and spectroscopy tools with nanometer resolutions.

Acknowledgements: This work was supported by the U.S. National Science Foundation and the Korean Nano Research and Development Program.

References [1] O. Mitrofanov, M. Lee, J.W.P. Hsu, L.N. Pfeiffer, K.W. West, and J.D. Wynn, "Terahertz pulse propagation through small apertures", Appl. Phys. Lett. 79, 907-909 (2001). [2] D. V. Palanker, G. M. H. Knippels, T. I. Smith and H. A. SchweUman, "IR microscopy with a transient photo-induced near-field probe (tipless near-field microscopy)", Opt. Commun., 148, 215-220 (1998). [3] Q. Chen, Z. Jiang, G.X. Xu, X.-C. Zhang, "Near-field terahertz imaging with a dynamic aperture". Opt. Lett. 25,1122 (2000). [4] J. Z. Xu and X.-C. Zhang, "Optical rectification in an area with a diameter comparable to or smaller than the center wavelength of terahertz radiation". Opt. Lett. 27, 1067 (2002). [5] N. van der Valk and P. Planken, "Electro-optic detection of sbwavelength terahertz spot sizes in the near field of a metal tip", Appl. Phys. Lett., 81,1558-1560 (2002). [6] Kanglin Wang, A. Barkan, D.M. Mittleman, "Propagation effects in apertureless nearfield optical antennas", Appl. Phys. Lett., 84, 305-307 (2004). [7] Hou-Tong Chen, R. Kersting and Cyu Cheon Cho, "Terahertz imaging with nanometer resolution", Appl. Phys. LeU., 83, 3009-3011 (2003).

761

Part XI

Optoelectronics and Other Applications

Optimization of a 40 GHz regeneratively and harmonically mode-locked fiber laser under PLL operation and its longitudinal mode characteristics Masato Yoshida, Taro Yaguchi, Shinji Harada, and Masataka Nakazawa Research Institute of Electrical Communication, Tohoku University, 2-1-1 Katahira, Aoba-ku, Sendai, 980-8577 Japan E-mail: [email protected] Abstract. We report the optimum operating conditions for a 40 GHz regeneratively and harmonically mode-locked erbium-doped fiber laser. We point out that there is temperature-dependent longitudinal mode-hopping with time that is not controlled with a phase-locked-loop operation.

1. Introduction Ultrashort optical pulse sources with repetition rates of a few tens of GHz are very important for realizing ultra-high-speed optical communication in terms of both transmission and ultrafast signal processing. In this paper, we report in detail the optimum operating conditions and longitudinal frequency hopping of a 40 GHz regeneratively and harmonically mode-locked fiber laser (MLFL)[l]-[2]. We measure the pulse characteristics of three lasers with different cavity lengths, and we point out that there is an optimum range for the cavity length with which to obtain long-term stable operation without supermode noise [3]. Furthermore, we show that even under PLL operation, there is longitudinal mode-hopping with time, and we reveal that this mode-hopping depends mainly on small temperature fluctuations in the cavity.

2. 40 GHz regeneratively and harmonically mode-locked fiber laser The configuration of a 40 GHz MLFL is shown in Fig. 1. The laser consisted of a polarization-maintaining erbium-doped fiber, a polarization-maintaining dispersion-shifted fiber, a 30 % output coupler, a polarization-dependent isolator, a LiNbOs intensity modulator, an optical bandpass filter, and a WDM coupler. All the fibers in the cavity were polarization-maintaining to prevent any polarization fluctuation. A long term stability in mode-locking process was achieved by use of regenerative mode-locking technique[4]. To stabilize the repetition rate, a PLL operation was also used by negatively feeding the error signal from the synthesizer back to the PZT in the laser cavity [5].

765

OQ •a CO

DC

Q. Q. 13 CO 0 •a

o E o

High Voltage Controller

Fig. 1 Experimental setup for 40 GHz regeneratively and harmonically modelocked fiber laser under PLL operation.

Q. 13 CO

100

150

200

250

300

Time [s] Fig. 2 Time dependence of supermode suppression ratio of fiber lasers with fibers A and B.

It has already been clarified that a stable pulse train can be obtained only when the GVD of the cavity is anomalous [6]. Here we prepared three DSFs A, B, and C with anomalous GVDs. Fibers A, B, and C were 200, 50, and 30 m long, and had GVDs of +2.2, +5.4, and +8.0 ps/km/nm at 1.55 ^im, respectively. The average GVDs of the cavities for fibers A, B, and C were +1, +0.75, and +0.5 ps/km/nm, respectively.

3. Experimental results and discussion for 40 GHz MLFL Figure 2 shows the time dependence of the supermode suppression ratio of fiber lasers with fibers A (200 m) and B (50 m). When fiber A was used in the cavity, the long-term stability was poor because of the long cavity length. In contrast, when fiber B was used, the long-term stability of the laser remained excellent when the pump power was higher than 100 mW. In the long cavity configuration with fiber A, approximately 46,000 supermodes were in competition and modehopping occurred easily. This mode-hopping causes a detuning of the phase between the pulse and the modulation signal in the intensity modulator and this leads to supermode noise. Therefore, it is important to shorten the cavity length to reduce the number of beat notes to a proper number (^ 10) in the microwave filter. On the other hand, when fiber C (30 m) was used, the peak power inside the cavity was 160 mW for a pump power of 140 mW, where the nonlinear phase shift was as small as 0.01 JC rad. In this case the laser operated in the linear regime, and it was not possible to suppress the supermode noise. We measured the change in laser frequency of a 40 GHz MLFL with fiber B by detecting the beat signal between one of the longitudinal modes filtered out from the 40 GHz MLFL and a stable pulse beam from a 10 GHz MLFL. Figure 3 shows changes in the beat signal and repetition rate of the laser as a function of temperature change AT in the cavity. The repetition rate was varied by the cavity length change caused by AT, and the shift in the beat frequency (lower traces) was

766

8

without PLL !

;

^ -- -^ \

1 —

r-^

N X 6 O^

with PLL AT=0.02°C :

-.z

4

^

c

0)

2 SAT=0.05''C

^

:

~x

^ 0

{.,_ AT=0.02°C' ^'1


-4 O

:

: I . I .

100 150 200 250 300 350 400

Time [sec] Fig. 3 Changes in pulse repetition frequency and heterodyne beat frequency with temperature change.

50

, AT=0.05^pH

. 1... -8 100 150 200 250 300 350 400

O) O

Time [sec] Fig. 4 Changes in pulse repetition frequency and heterodyne beat frequency with and without PLL operation.

roughly proportional to the repetition frequency shift (upper traces). It should be noted that there was no mode-hopping when AT was less than 1 x 10"^ degrees, where this small AT was estimated from the change in the repetition rate. Figure 4 shows the change in the beat frequency and repetition rate of the laser with and without PLL operation. Although the repetition rate can be completely stabilized with PLL operation, the beat frequency shift still occurred. This result indicates that mode-hopping, which may be caused by a small temperature change in the present laser, plays an important role in the change in the absolute optical frequency.

4. Conclusions We reported the optimum operating conditions for a 40 GHz regeneratively and harmonically mode-locked fiber laser. It is important to reduce the number of beat signals within a microwave filter, while continuing to suppress the supermode noise. We also pointed out that there was longitudinal mode-hopping, and that it could be suppressed by precisely controlling the temperature in the cavity to within 1 X 10"^ degrees.

References 1 2 3 4 5 6

E. Yoshida, N. Shimizu, and M. Nakazawa, IEEE Photon. Technol. Lett., vol. 11, p. 1587,1999. M. Nakazawa, E. Yoshida, IEEE Photon. Technol. Lett., vol. 12, p. 1613, 2000. M. Nakazawa, K. Tamura, and E. Yoshida, Electron. Lett., vol. 32, p. 461, 1996. M. Nakazawa, E. Yoshida, and Y. Kimura, Electron. Lett., vol. 30, p. 1603,1994. M. Nakazawa, E. Yoshida, and K. Tamura, Electron. Lett., vol. 33, p. 1318, 1997. E. Yoshida, K. Tamura, and M. Nakazawa, lEICE Trans. Electron., vol. E81-C, p. 189, 1998.

767

Femtosecond Synchronization of RF-Signals with Optical Pulse Trains J. Kim\ M. H. Perrott^ and F. X. Kaertner^ ^ Department of Electrical Engineering and Computer Science, Research Laboratory of Electronics, Massachusetts Institute of Technology, Cambridge, MA 02139-4307, USA E-mail: [email protected] ^ Department of Electrical Engineering and Computer Science, Microsystems Technology Laboratories, Massachusetts Listitute of Technology, Cambridge, MA 02139-4307, USA Abstract. A synchronization scheme for extraction of low jitter RF-signals from optical pulse trains, which is robust against photodetector nonlinearities, is described. Sub-100 fs timing jitter between the extracted RF-signal and the pulse train is demonstrated. As has been shown recently [1], the extraction of a microwave signal from an optical pulse train emitted by a mode-locked laser by use of direct photodetection is limited in precision by excess noise. The origin of this excess noise has been identified to be amplitude-to-phase conversion in the photodetection process, beam-pointing variations, and pulse distortions due to photodetector nonlinearities [1]. For precise timing synchronization of an RF-signal to an optical pulse train, the typical methods are limited by the detector nonlinearities. In this paper, a synchronization scheme that avoids these limitations is proposed. As the first experimental demonstration, sub-100 fs timing jitter between the extracted RFsignal and the optical pulse train is demonstrated. Repetition Rate: f^

Ai

AAA\

rr

I I r^

M ! Amplitude Modulators

§ -[M]*©-H>"' vco

•D ^

>

• RF:f=mfR Recovered from optical pufse train

180" Shift

Fig. 1. Schematic setup for RF-signal extractionfroman optical pulse train. The general idea for suppression of excess noise due to the photodetection process is shown in Fig. 1. While still in the optical domain, the timing information is transferred into an intensity imbalance between two beams when the pulse train is sent through a pair of amplitude modulators. The modulators are driven by the output signal from a vohage-controlled oscillator (VCO) with 180° phase difference. The intensity difference is detected with a balanced detector and this signal controls the input to the VCO via a loop filter. The 180° out-of-phase amplitude modulators can be realized by a Sagnac-loop interferometer with a

768

phase modulator in the loop. Fig. 2 shows the synchronization scheme with the measurement setup. A 100 MHz repetition rate Ti:sapphire mode-locked laser is used as the pulse source. After a bandpass filter at 800 nm to limit the pulsewidth to about 100 fs, the input pulse train is sent into the Sagnac-loop. A resonant phase modulator at 2 GHz is positioned in the Sagnac-loop in such a way that the optical delay between counter propagating pulses at the phase modulator is set to half of the RF-signal period. This assures that the two pulses experience opposite phase modulation. The output beams are detected by a balanced detector which drives the VCO after proper filtering. For a stable and drift-free biasing of the interferometer, a quarter-wave plate is inserted in one of the beams using a thinfilm coating covering only half of the substrate.

Loop Filter [Balanced • [Detector / O V C O (2 GHzi Phase Noise Test System Rep rate • 100 MHz

putptit 2 Output 1

^

Ti:sapphire ML-laser

Oscilloscopa ' ' 800 nm BP filter

L ^ r[ >S T

Phase Modulator @ 1 •

= j^Mixer

i Vector Signal Analyzer

b ^

ri2 Phase nil Shift Device

^

LP filter Baseband Amp

Direct p—n MeasuremenNr

2 GHz BP filter

RFAmp

Fig. 2. Extraction of a 2GHz signal from lOOMHz Ti:sa laser. The VCO output is characterized (i) by a commercial phase noise test system and (ii) by mixing in quadrature with the 2 GHz component of the directly detected signal. The phase noise of the RF-output signal from the VCO is characterized in two ways: (i) by the frequency discriminator technique using a commercial phase noise measurement setup (PN9000, Aeroflex) and (ii) by mixing the output signal of the VCO in quadrature with the 2 GHz component of the directly detected pulse train to measure the relative phase noise between the pulse train and the extracted RFsignal. The measured single-sideband phase noise spectra from 1 Hz to 10 MHz are shown in Fig. 3. Curve 1 shows the phase noise spectrum of the free-running VCO measured with the Aeroflex phase noise measurement system. Curve 2 shows the phase noise measured by the same method when the system is locked. The locking is clearly visible in the spectrum covering the range of 100 kHz to 10 MHz. At lower frequencies, the phase noise of the Ti:sapphire pulse train

769

dominates. To verify this assumption, we measured the relative phase noise between the pulse train and the RF-signal by using the second characterization method. The result is shown in curve 3 of Fig. 3. Due to the noise floor of the vector signal analyzer (curve 4 in Fig. 3) and excess noise in the photodetector, the high frequency noise floor is increased in comparison to method (i). But this measurement clearly shows that the noise increase at low frequency in curve 2 is the phase noise of the free-running Ti:sapphire laser. The origin of the enhanced phase fluctuations below 1 kHz may be either due to mechanical vibrations in the Sagnac-loop or excess phase noise in the direct photodetection process. The relative timing jitter between the RF-signal and the pulse train integrated from 100 Hz to 10 MHz can be estimated by the area underneath curve 5, which lines up with the high frequency noise of the Aero flex measurement (curve 2 in Fig. 3) and results in approximately 60 fs timing jitter.

10

100

Ik

10k

100k

Frequency (Hz)

Fig. 3. Measured single-sideband (SSB) phase noise spectra. In summary, we have demonstrated a novel synchronization scheme for extracting low jitter RT-signals from optical pulse trains. Sub-100 fs timing jitter measured from 100 Hz to 10 MHz between the extracted RF-signal and the optical pulse train is demonstrated. With improved system design and implementation, it is expected that this method is able to reduce the relative jitter to the sub-fs range, which has been so far only achieved by purely optical means [2].

References 1 E.N. Ivanov, S.A. Diddams, and L. Hollberg, IEEE J. Sel. Top. Quant. Elec. 9, 1059(2003). 2 T.R. Schibli, J. Kim, O. Kuzucu, J.T. Gopinath, S.N. Tandon, G.S. Petrich, L.A. Kolodziejski, J.G. Fujimoto, E.P. Ippen, F.X. Kaertner, Opt. Lett. 28, 947 (2003).

770

Fast photo4nduced phase switching in organic conductor crystal; (EDO-TTF)2PF6 Matthieu Chollef; Laurent Guerin^^ Naoki Uchida^ Souichi Fukaya^ Tadahiko Ishikawa^ Shin-ya Koshihara^*^, Kazunari Matsuda®, Akira Ota^, Hideki Yamoehi*^ and Gunzi Saito ^Department of Materials Science, Graduate School of Science and Engineering, Tokyo Institute of Technoldgy 2-12-1 Oh-okayama, Meguro-ku, Tokyo 152-8551, Japan; e-mail: [email protected]; 'T)epartement of Materials Science, University of Rennes 1, Rennes, France; ^ERATO, JST *^Division of Chemistry, Graduate School of Science, Kyoto University, Sakyo-ku, Kyoto, 606-8502, Japan; ^KAST, JST, PRESTO Abstract: Organic conductor (EDO-TTF) aPFg crystal shows metal (M)-insulator (I) transition at 280 K. Here, we report the occurrence of highly efficient photo-conversion from insulator to metal phase within a few pico-seconds.

1. MetaHnsulator (M-I) traasition in (EDO-TTF)2PF6 crystals Organic donor EDO-TTF:ethylenedioxytetrathiaftilvalene based molecular conductor (EDO-TTF) aPFg crystal (see Fig. 1) shoves a phase transition from metal (M) to insulator (I) as decreasing temperature. In addition, this transition is accompanied with Peierls, charge ordering, and anion ordering transition at critical temperature Tc = 280 K [1]. This M-I transition in (EDO-TTF) aPFe can be sensitively probed by spectroscopic method [2]. ^

_ High Temp. Phase (300 K)

9H^

Low Temp. Phase (150 K) t

f

*

f it

^ b a

Figure 1* Schematic view graph of the structural changes accompanied with M-I transition in (EDO-TTF) 2PF6 crystal.

771

This crystal is an important candidate for the study of photo-induced phase transition, because the M-I transition accompanied with lattice, electronic structural and magnetic changes is expected to be triggered by weak photoexcitation reflecting cooperative interaction in the crystal. In addition, the large photo-induced reflectance change in near IR and visible region can be expected around room temperature for this material, because it is often the case that photoinduced cooperative response is largely enhanced around Tc (280 K).

2. Reflectance change in (EDO-TTF) 2PF6 crystals induced by femto-second laser irradiation As discussed in the previous section, the reflectivity change in (EDO-TTF) 2PF6 crystal (AR/R) in photon energy region between 1.2 and 2 eV becomes good probe for phase transition. Therefore, in this study, AR/R induced by the irradiation of femto-second (fs) pulsed laser (pulse width: 120 fs, photon energy: 1.55 eV, repetition rate: 1 kHz) was utilized for the confirmation of M-I phase transformation and its dynamics. The crystal surface with the size of 0.1x0.3 mm^ was excited by laser light, and change in reflection spectrum was probed utilizing pulsed white light (200-250 fs in width) generated by nonlinear optical method. The excitation photon energy was nearly resonant with the charge transfer excitation among constituent EDO-TTF molecules (excitation from D^D^ to D^'^D^) [2]. The Fig. 2 shows the calculated AR/R spectra (dashed line) expected for I to M phase transition based on the reflection spectra [2]. 0.6

• - 1 6 psec • + 5 psec

0.4 V-

" calculated AR/R

::a^ 0.2 h o 00 0.0 -0.2

<

-0.4 -0.6 -0.8 [1.2

1.3

1.4

1.5

1.6

1.7

1.8

1.9

Photon Energy (eV) Figure 2. Photo-reflectance spectra observed 16 ps before (black line), 5 ps after (gray line) photo-excitation by 120 fs laser pulse at 180 K and the calculated AR/R spectra (dashed line). The spectral shape of AR/R observed at 5 ps afl:er photo-excitation (gray line in Fig. 2) can be well explained by the about 50% transformation from I-phase to Mphase on the surface of the crystal. This result strongly suggests the occurrence of the photo-induced I-to-M phase transition in this crystal within 5 ps.

772

To obtain the information of the transition dynamics, time-dependence of AR/R at the photon energies of 1.38 and 1.72 eV have been observed at 180 K (Fig. 3). The time profiles of AR/R values show large change just after photo-excitatioii and kept the same value until a few hundreds pico-second if the crystal was illuminated by rather strong excitation light (6.4x10^^ photons /cm\ 12-

Pump: 1.55 eV 6.4x10''^ photons/cm^

-100

0

100

200

300

400

Time Delay (psec) Figure 3. Time-dependence of AR/R at the probe photon energy of 1.38 and 1.72 eV measured at 180 K. The obtained result clearly shows that I-to-M phase transition in (EDO-TTF) 2PF6 crystal occurs within 5 ps, and it is also well consistent with the observed spectral changes (see Fig. 2). Such an ultra fast phase conversion process cannot be driven by simple heating effect by laser irradiation. In addition, based on the estimated excitation photon density from the optical conductivity measurement [2], 50% I-to-M conversion has been achieved by only one excitation photon for every 1000 EDO-TTF molecules. Observed ultra fast and highly efficient photo-switching from I-phase to M-phase at rather high temperature (80-270 K) seems to be reflecting ah important role of spin-latticecharge coupled cooperativity in the photo-excited state. In addition, the noble and large photo-response of (EDO-TTF) 2PF6 is demonstrating that organic correlated system with multi-instability is an important candidate even for application in optical signal processing.

Acknowledgements The authors are grateful to Dr. O. Drozdova and Professor K. Yakushi (IMS Okazaki) for useftil discussions.

References [ 1] A. Ota, HT Yamochi, and G. Saito, J. Mater. Chem. 12,2600 (2002). [2] O Drozdova, K. Yakushi, A. Ota, H. Yamochi, and G. Saito, Synth. Met. 133434,277 (2003), 773

Molecular phase-to-amplitude converter using femtosecond wave packet engineering Isao Matsuda, Kazuhiko Misawa, Naoyuki T. Hashimoto, and Roy Lang Department of Applied Physics, Tokyo University of A & T, and CREST, JST 2-24-16 Naka-cho, Koganei, Tokyo 184-8588, Japan E-mail: [email protected] Abstract We show a molecular phase-to-amplitude converter, which convertsthe optical phase infomiation of femtosecond pulses into spontaneous emission amplitude in a cyanine dye molecule through the coherent excitation of the quantum wave packet.

!•

Introduction

Ultrafast dynamic control of optical phases in the femtosecond regime is expected to make a large impact on a wide range of teclmology such as optical communication, infomiation storage, and material processing. There is rapid progress in geration and shaping of a femtosecond pulses, keeping all spectral comjx)nents in the whole bandwidth of pulses in constant relative phase. Coherent control of quantum wave packets has also been studied in various systems such as atoms, molecules, and semiconductors using chir|3-controlled and phase-locked double pulses [1-3]. We developed a phase-programmable femtosecond optical source to show nonlinear optical phenomena sensitive to the optical phase [4]. In the present paper, we show a molecular phase-to-amplitude converter, wliich suggests a large possibility^ of writing and reading phase information in condensed matters using phase-programmed femtosecond pulses. The phase-toamplitude converter can convert magnitude and the cliirp direction and rate of the input pulses directly into luminescence intensity from the material.

2,

Experimental Methods

The intensity of the spontaneous emission from a cyanine dye molecule (IR-140) is measured to evaluate the excited-state population after photoexcitation with femtosecond chirped pulses. The chirped pulses are prepaied by using the chirp variable device with a chiiped miiTor pair [5]. The center wavelength, pulse energy, and duration were 790 nm, 2^, and 40fs, respectively. The positive chirp of a fixed rate is undercompensated with the prism compressor in the amplifier. The chir|) variable device gives negative chirp in proportion to a number of reflection times on the cliirped mirror pair. The spectral profile does not change irrespective of the chirp condition. The ethanol solution of IR-140 at a concentration of 0.4mM is circulated in a 0.5-mm thick quartz cell. The cyanine dye molecules are excited by the cliirped

774

pulses, and fluorescence spectra from the sample are measured with a CCD spectrometer to evaluate the remaining excited-state population.

3,

Results and Discussion

Figure 1 shows the difference fluorescence spectra of the positively chirped (PC) and negatively chirped (NC) excitations with respect to the Fourier transform limited (TL) excitation. The fluorescence spectrum of the sample solution by TL excitation is shown in the inset of Fig. 1 as well as the excitation pulse spectrum. Compared with TL excitation the luminescence intensity is increased and decreased in case of PC and NC excitations, respectively, even with the equal GDD. The energy, duration, and spectrum of the pulses are all the same, betw^een -600 and +660 fs", but only the chirp direction is opposite. The most essential result is that the luminescence intensity is dependent on the direction and rate of the pulse chirp.

75D

800 850 900 Wavelength (nm)

Fig. 1. Difference fluorescence spectra of the PC and NC excitations with respect to the TL excitation. Fluorescence and excitation pulse spectrum are shown in the inset.

The luminescence intensity integrated over the whole spectral profile is also plotted in Fig.2 as a function of the excitation pulse energy. In tliis excitationintensit}' range, the luminescence intensity is always lai'ger in case of PC excitations, and smaller for NC excitations. The intensity difference is up to about 25 percent of the total intensity at TL excitation. To explain the dependence on the excitation intensity, we perform a quantum mechanical calculation based on a three-level model [6]. The level structure consists of two ground levels at 0, £'2, and one excited level at E3. Tlie bottom level is assumed to be initially occupied by 100 percent. According to this model, the population left in the third level after pulse transit was calculated. The center wavelength and the TL width of die incident pulses are supposed to be 790nm and 40fs, respectively. The center wavelength of the excitation pulse conesponds to the transition energy E3-E2I2. The dipole moment of the transitions betv^^een the first and the tliird level, and the second and the third is assumed to be equal to each other. .4ny relaxation process is neglected in the present calculation. The chiiped pulses are assumed to have a Gaussian envelop function with linear chiip. Figure 3 shows the calculated population in the third level as a function of the excitation intensity. At low excitation the stimulated emission is not effcient and

775

CDF is negligible. As excitation is increased, CDF becomes remarkable. Theoretically, an oscillatory dependence like Rabi oscillation is expected in case of NC excitation. Strictly speaking, this is different from the Rabi oscillation, because a molecule is not an ideal two-level system. Anyhow, such strong stimulated emission pumping is not experimentally remarkable as shown in Fig. 2. This is due to not only low excitation, but also dephasing in dye molecules.

1

2

Pulse excitation (jiJ) Fig.2. Intensity dependence on the excitation pulse energy with the PC (crosses), TL(squares), and NC(circIes) excitations

4.

Pulse Intensity (pulse areaFig,3. Calculated intensity dependence of the excited-state population excited by PC (thick line), TL (thin line), NC (thick dashed line) pulses

Conclusions

In conclusion, we demonstrate the phase-to-amplitude converter in a cyanine dye molecule using phaseprogrammable femtosecond pulses. The luminescence intensity proportional to the excited-state population in IR-140 molecules was observed to be remarkably dependent on the pulse chirp. The pulse chirp information is converted to the amplitude information of luminescence, induced by wave-packet shaping. The observedshaping of quantum' wave packet opens a new possibility to process the intra-pulse phase information..

References

3. 4 5.

776

Cohernet Control in Atoms, Molecules, and Semiconductors, Ed. by W. Potz and W. A. Schroeder (Kluwer, 1999). G. Cerullo, CJ. Bardeen, Q. Wang, and C V. Shank, Chem. Phys. Lett. 262, 362 (1996). K. Misawa and T. Kobayashi, J. Chem. Phys. 113, 7546 (2000). L Matsuda, K. Misawa, and R. Lang, Technical Digest of CLEO02, 113 (2002). I. Matsuda, K, Misawa, and R Lang, Opt. Gommun., in press (2004). H. T. Hashimoto, K. Misawa, and R. Lang, Appl. Phys. Lett. 82, 2749 (2003).

Femtosecond pulse recoding and regeneration by a two-photon gated periodic diffractive optics Hajime Nishioka, Hitoshi Tomita, and Ken-ichi Ueda Institute for Laser Science, University of Electro-communications, 1-5-1 Chofligaoka, Chofu, Tokyo 182-8585 Japan E-mail: [email protected] Abstract. A new scheme for pulse recording, time-reversed playback, and self-phase regeneration in a nonlinear diffractive device was proposed. The pulse regeneration has been demonstrated in a two-photon absorbing glass-chip. Numerical calculation shows that sufficient visibility in two-photon interference is possible for 100 times pulse compression.

1.

Introduction

In this paper, v^e have demonstrated automatic phase correction based on nonlinear interferometoric devices. This device records phase structure of a phase-modulated input (signal) wave and regenerates a time-reversed replica of the signal vy^ave. The time-reversed operation i.e. the frequency-domain phase conjugation (FDPC), regenerates initial pulse shape after passing through again the same GDD medium. The FDPC operation generally requires frequency-resolving mechanism for example atomic resonance in the photon-echo system, or gratings in the 4-f spectral holography. In this work, we have firstly demonstrated the FDPC operation with no dispersive optics using time to space conversion by two-photon interference, as schematically shown in Fig.l. Gate pulse 1

nonimeartmefiiiuni

^ ml.m......

< ,i..f

, Frequency chirped pulse

^^^U,yi:iJi^

Two-photon interference —»JL^

Recorded chirpedgrating

Readout pulse

Time reversed replica

Fig.l Schematics of two-photon-gate pulse recording and time-reversed read out.

777

A frequency chirped signal pulse interacts with a counter propagating transform-limited reference pulse in a nonlinear medium. The pulse duration of the reference pulse is chosen to match spectral shape to the signal wave. Typically two-photon absorbing (TPA) medium records the two-photon cross correlation function as its index change due to the TPA. The recorded phase information can be read out by injecting another TL pulse (read pulse). One of the advantages of the two-photon interference is that the two pulses can be interacted each other with a long time delay exceeding their coherent length. We have numerically examined the interaction length and the contrast ratio in the two-photon cross correlation between the TL-reference and chirped pulses. A visibility of 0.5 has been evaluated when the ratio of pulse duration is 50, as shown in Fig.2. ^^ JQ

2 1.5

Urn

o T3

^

0

5

-0.5

^

-1 -1.5

E <

(a)

0.5

-15

-10

10

15

Fig.2 A frequency chirped pulse having pulse duration of 50xTL (a) , and the two-photon interference fringes between the chirped and TL pulses (b).

0 200 400 Delay time (fs)

600 -200

0

200 400 Delay time (fs)

600

Fig.3 Autocorrelation trace of the GTI-encoded signal pulse and the recompressed single pulse appeared in the first order diffraction. The schematics, Fig.l, is showing reflective type Bragg gratings but also this scheme is effective for the Bragg diffraction. The diffraction type configuration has an advantage because diffraction efficiency of the phase-modulated volume gratings is higher than its reflectance in normal incidence. A Ti:Al203 laser system producing a 120 fs pulses at 800 nm was used for the two-photon recording and the readout experiments.

778

2.

Pulse recording and regeneration experiments

An TL pulse from the laser system is splitted into two beams. One of them is reflected by a 25 mm air-gap Gires-Toumios interferometer (GT-I) to form a phase-distorted multi-spiked pulse that is used as the test signal. We have used semiconductor-doped yellow-colored glass (Schott OG-530) as the two-photon recording nonlinear medium The temporal structure of the first order readout has been examined by a phase-sensitive autocorrelator as shown in Fig.3. These autocorrelation traces were taken under the repetition rate of 10 Hz with no time averaging. Each fringe in the traces corresponds to a record by single shot. The multiple spikes in the signal compressed to a single pulse by the diffraction. We have measured spatial frequency of recorded grating by angular dispersion with the TL reference beam as shown in Fig.4. We have observed four spatial 80S r'^u%';^'i:,',,'C'i ' ^'\^^y modes (grating period: A = 8.59, 8.52, 8.45, and 8.38 |Lim). The mode 1^:/ 8.45 urn 3 I'ih^^l 800 - j separation corresponds to Av = 3.22 ll,:.'. THz in optical frequencies and the pulse separation by the GTI (1/Av = sz 795 310 fs). The result shows that phase structure of the signal wave has been recorded as the spatial frequencies of ^vthe grating. / : • •

785

Fig. 4 Spatial frequencies of the recorded grating measured with the reference beam.

3.

% 8.52 (im ^//,T-'^^^ 8.38 jim iZ ).;:^.j ,;•-/%

92

93 94 Diffraction Angle (mrad)

Summary

In conclusion, we have proposed new scheme for optical pulse recording and regeneration in the two-photon recorded diffraction optics. This device requires no external dispersive optics and has possibility for ultrafast pulse recoding, GDD compensation by the time-reversed playback and self-regeneration in a single glass tip. The essence of the diffraction device is a wavelength multiplexed optical memory with sequential I/O having femtosecond time response. The readout after multi-time accumulative recordings gives coherent sximmation of the signals so multi-step optical programming is possible. Similarly, stacked or multi-layered devices will be used for multi-step all-optical signal processing. This work was carried out under the 21st Century Center of Excellence (COE) program on "Innovation in Coherent Optical Science" supported by the Ministry of Education, Culture, Sports, Science and Technology.

779

Linewidth and RIN measurements of longitudinal modes in ultrahigh-speed modelocked laser diodes Kentaro Haneda \ Hiroyuki Yokoyama^, Yo Ogawa^, and Masataka Nakazawa^ ^ Research Institute of Electrical Communication, Tohoku University, 2-1-1 Katahira, Aoba-ku, Sendai-shi, Miyagi-ken 980-8577, Japan E-mail: [email protected] ^ New Industry Creation Hatchery Center, Tohoku University, Aoba, Aramakiaza, Aoba-ku, Sendai-shi, Miyagi-ken 980-8577, Japan ^ Oki Electric Industry Co., Ltd., 550-1 Higashiasakawa-machi, Hachioji-shi, Tokyo, 1938550, Japan Abstract. The longitudinal linewidth and corresponding RIN of 10 and 40-GHz modelocked laser diodes are measured for the first time. The cavity Q-value is a dominant parameter of the linewidth. The mode-dependence of the RIN is observed.

1. Introduction To realize stable short pulse sources, further improvements must be made to such pulse characteristics as chirp and phase noise. Pulse characteristics in the time domain are already a major research subject. However, it is also important to investigate the longitudinal mode linewidth and relative intensity noise (RIN) to evaluate the phase coherence of each mode. In this paper, we describe for the first time the measurement of the linewidth and RIN of 10 and 40 GHz MLLDs with different configurations. We discuss in detail the relationship between the linewidth and RIN of the two MLLDs in terms of the difference in the cavity Q value and the change in the RIN.

2. Longitudinal mode linewidth and relative intensity noise measurements Table 1 shows principal specifications for MLLDs we used. In the measurement of the 10 GHz MLLD, we measured the linewidth and RIN of the center longitudinal mode and the ±8 th and ±16 th modes from the center. In the measurement of 40 GHz MLLD, the linewidth and RIN were measured in the longitudinal modes of ± 1 St to ± 3 rd modes. Table 1. Principal specifications for MLLDs. 10 GHz fll 40 GHz [2]

780

Cavity External Monolithic

Cavity Length 15 mm 3.8 mm

Mode-locker Saturable Absorber (SA) EA modulator

At [psl 1.6 3.1

2000 '

\<-A

pgBEHE^

N

5

^

4

4 0

§ 3

3a

i5.

77 1500 X §

•

\

^ A

i1- ^\ 7i t1i y• ^

"

'

!

"

"

^ i

•

~

1000

2

1J

-5

0

5

10

15

20'

Ljongitucf nal-mode nunnber

4

^

9

500

0 '-20 -15 -10

^

3 QI.

0) C

5

>"!

0 W O

LJ,,,],..]..^ - 4 - 3 - 2 - 1 0 1 2 3 4

Longitudinal-mode number

(a) (b) Figure 1. Change in the linewidth and RIN for each longitudinal mode of (a) 10 GHz and (b) 40 GHz MLLD. We used a self-delay heterodyne detection method to measure the longitudinal mode linewidth of the 10 GHz MLLD [3]. The linewidth of the center mode was 5.6 MHz. Figure 1 (a) shows the linewidth of the individual modes. In Fig. 1 (a), we also plotted the measured RIN, which we estimated by frequency-integrating the noise power spectrum. The linewidth was almost the same for each mode as shown in Fig. 1 (a). This result showed that the phase noises of each mode are almost the same in the mode-locked condition. In contrast, the RIN value was larger for modes that were distant from the center. Since the self-heterodyne method revealed that the linewidth of the 40 GHz MLLD was very wide, we newly adopted the heterodyne method. The linewidth of the center mode of the 40 GHz MLLD was 1.6 GHz. Figure 1 (b) shows the linewidth and the RIN of each longitudinal mode. The linewidth of the individual mode was almost constant as found with the 10 GHz MLLD. Also, as with the 10 GHz MLLD, the RIN value became larger for longitudinal modes as their distance from the central mode increased. From the results in Figs. 1 (a) and (b), there was an approximately three hundredfold difference between the linewidths of the 10 and 40 GHz MLLDs. In the next section, we discuss this difference in detail.

3.

Discussion

The linewidth of a "CW" LD is given by the modified Schawlow-Townes formula [4,5]:

Av =

•ju(l + a )

PQ'

(1)

where hv is the lasing energy, P is the power in the cavity, /u is the population inversion parameter, a is the linewidth enhancement factor [5], g is the Q value of the laser cavity: ^ Q

2jmlv

c

1 ,

.

. ^

a^l -In^R^R^

(2)

781

c is the speed of light, RiR2^^^{-2ail) is the loss of the light traveling inside the cavity, and nl is the optical length of the cavity. The parameters for each laser are listed in Table 2. We evaluate the MLLD linewidth by assuming that the light power P in equation (1) is replaced with the light power of one longitudinal mode P;. Here Nm is the effective mode number within a 3 dB bandwidth and each longitudinal mode power at the output was roughly estimated by Pout IN^ - From Table 2, Pi of the 10 GHz MLLD is given by

P P _

out

N^ 1-R

^ = 0.15mW

(3)

e

m

Similarly, for the 40 GHz MLLD, Pi = 0.19 mW. From equation (2), the Q values of the 10 and 40 GHz MLLDs are calculated to be 90900 and 6900, respectively. With these parameters, and by assuming // = 3, a = 5 in both lasers, from equation (1), the linewidths of the 10 and 40 GHz lasers are estimated to be 2 MHz and 0.3 GHz, respectively. Theoretical values are not far from the experimental results. The cavity Q value seems to be a dominant parameter in terms of the MLLD linewidth as in the case of a CW LD. The mode-dependence of the RIN that was observed in both MLLDs (Fig. 1 (a) and (b)) can be explained as a consequence of the mode-partition noise [6]. Table 2. Parameters of each laser.

10 GHz 40 GHz

4.

Pout [mW] 0.3 0.7

e' 0.5 1

Nm 20 4

nl [mm] 15 3.8

RiR2Gxp(-2ail) 6dB 20 dB

Pi [mW] 0.15 0.19

Q 90900 6900

Conclusion

We have reported the linewidth and RIN characteristics of 10 and 40 GHz MLLDs for the first time. We found that the linewidth is narrower for an MLLD with a larger cavity Q value. In addition, linewidth measurements for each mode showed that the phase noise of each mode is almost the same in the mode-locked condition. The RIN value was greater for modes that were distant from the center longitudinal mode.

References 1 2 3 4 5 6

782

H. Yokoyama, lEICE Trans. Elec, Vol.E85-C, p.27, 2002. Y. Katoh, S. Arahira, and Y. Ogawa, OFC 2001, WC 5, 2001. T. Okoshi, K. Kikuchi, and A. Nakayama, Electron. Lett., Vol.16, p.630, 1980. A. L. Schawlow and C. H. Townes, Phys. Rev. Vol.112, p.l940, 1958. C. H. Henry, IEEE J. Quantum Electron., Vol. QE-18, p.259, 1982. K. Ogawa, IEEE J. Quantum Electron. Vol. QE-18, p.849,1982.

Charge Generation in Inorganic/Organic Photovoltaic Blends Sebastian Westenhoff, Sophia C. Hayes, Neil C. Greenham and Carlos Silva Cavendish Laboratory, University of Cambridge, Madingley Road. Cambridge CBS OHE, United Kingdom E-mail: [email protected] Abstract. We apply ultrafast spectroscopy to investigate charge generation kinetics in blends of semiconducting polymers and cadmium selenide nanocrystals with photovoltaic applications. The rate is limited by diffusion of the primary excitons in the polymer network.

1.

Introduction

Conjugated polymers provide a convenient class of solution-processable semiconductors to achieve low-cost, large area devices, including photovoltaic cells [1]. In order to overcome the difficulty of dissociating the charges in photovoltaic devices, an important approach is to blend an electron accepting material with the semiconducting polymer [2, 3]. The charge generation efficiency is enhanced and electrons are extracted more efficiently in these devices. Colloidally grovvai semiconducting nanocrystals can be designed with a high electron affinity such that they are excellent electron acceptors. Hence, blends of semiconducting polymers and nanocrystals have proven to be efficient photovoltaic materials [4, 5]. The optimization of the power conversion efficiency has motivated intensive research into polymer/nanocrystal architectures. Optimizing nanoparticle shapes and altering the morphology have led to a much improved power conversion efficiency at AM 1.5 global conditions of 3% (1 sun) [5-7]. However, the detailed physics that determine the efficiency of these devices is still not completely understood. The focus of this study is to gain a deeper understanding of the elementary process of charge generation at the polymer/ nanocrystal interface.

2.

Experimental Methods

Our samples consist of a soluble derivative of poly(;?-phenylenevinylene), OCiCioPPV (see inset in figure 1 for chemical structure), blended with spherical cadmium selenide nanocrystals of 4-nm diameter synthesized via the method reported by Katari [8]. Details of the sample preparation are described elsewhere [9]. We use two time-resolved optical spectroscopy methods, namely femtosecond transient absorption (FTA) and time-correlated single photon counting (TCSPC) as described elsewhere [10, 11], to investigate the initial formation and

783

recombination of charge carriers. The detection system for FTA employed realtime analysis of the diode readings to reject "bad" laser shots and thus improve the signal to noise ratio. Both methods employed 3.04 eV pulses, with durations of 100 fs and 70 ps respectively, to excite the samples. In FTA measurements, a white light probe pulse spanning from 2.75 eV to 1.23 eV was used.

3.

Results and Discussion

Figure 1(a) shows photoluminescence decays measured by TCSPC of a series of samples with increasing concentration of cadmium selenide nanocrystals. The photoluminescence is almost entirely due to OCiCio-PPV because the luminescence of the cadmium selenide nanocrystals is very weak. The graph reveals that with increasing content of nanocrystals the exciton lifetime becomes shorter. The neat polymer displays essentially mono-exponential kinetics, while the decay of the 50% nanocrystal sample is clearly non-exponential. These findings can be rationalized with a competing charge transfer pathway opening up when nanocrystals are added to the sample acting on a nanosecond timescale [4].

5

10 15 Time /ns

1 10 100 Delay time /ps

1000

1 10 100 Delay time /ps

1000

Fig 1: (a) Photoluminescence decay recorded at 1.98 eV for spin-cast films of OCiCio-PPV with varying cadmium selenide nanocrystal content: 0% weight (squares), 10% weight (open dots) and 50% weight (open triangles). The inset shows the chemical structure of OCiCio-PPV. (b) Normalized ground state bleach dynamics of 0% weight (squares) and 50%) weight content (open triangles) recorded by FTA. Detection is at 2.48 eV. The inset summarizes the charge generation related elementary processes, (c) Difference between the two kinetics shown in (b). In order to probe charge generation dynamics on faster timescales we have carried out FTA measurements. Comparison of the ground state bleach dynamics of the 50% nanocrystal blend and pristine polymer in figure 1(b) reveals that the decay (recovery of ground state population) for the pure polymer is significantly faster and that at long times the 50% nanocrystal blend decays to a residual relative value of 0. Thus, in the blend some excitons do not recover to the ground state, but undergo an additional decay channel. This is thought to be charge generation although triplet exciton formation would also be a possibility on this timescale. We can estimate the timescale on which charge generation occurs from

784

the difference of the two transients, shown in figure 1(c). Assuming that the rate of the ground state recovery does not change by adding nanocrystals to the polymer, we conclude that the charge generation occurs on a timescale of ~15 ps, while the decay at later times is likely to be due to recombination of charges at the inorganic/organic interface. It is difficult to extract an exact rate constant for the charge generation, because the timescales for generation and recombination may overlap and the exciton lifetime may be different in the two samples. Additionally, excitons migrate extremely fast in the films causing a distribution of rate constants for the ground state bleach even in the pristine polymer sample. The remaining ground state bleach for the blended sample of 0.2 corresponds directly to about 20% of excitons that form charges within 500 ps. The data presented are consistent with a diffusion-limited model in which the excitons migrate through the polymer network until they reach a site where they can dissociate. Compared to similar studies on blends of PPV derivatives and Ceo, where charge generation usually occurs faster than 1 ps, our system exhibits a much slower rate [12].

4.

Conclusions

We have shown that charge generation occurs on nanosecond and picosecond timescales. This leads to the conclusion that the charge generation rate is limited by exciton migration through the polymer network. In order to gain more insight into accurate values for rate constants of the charge related processes we will carry out FTA measurements at 0.5 eV probe energy. It has been established that the photoinduced absorption in that region is due to the low energy polaron absorption by steady state photoinduced absorption measurements [9]. At this spectral position we should be able to separate charge-related dynamics from radiative and non-radiative decay, triplet formation and energy migration effects.

References 1. C. J. Brabec et al., Adv. Funct. Mater., 11, 15 (2001). 2. J. J. M. Halls et al., Nature, 376, 498 (1995). 3. G. Yu et al, Science, 270, 1789 (1995). 4. N. C. Greenham et al., Phys. Rev. B, 54, 17628 (1996). 5. W. U. Huynh et al., Science, 295, 2425 (2002). 6. B. Q. Sun et al., Nano Lett., 3, 961 (2003). 7. B. Q. Sun et al., submitted to Journal of Applied Physics, (2004). 8. J. E. B. Katari et al., J. Phys. Chem., 98, 4109 (1994). 9. D. S. Ginger and N. C. Greenham, Phys. Rev. B, 59, 10622 (1999). 10. C. Daniel et al., Phys. Rev. B, 68, art. no. (2003). 11. A. C. Morteani et al.. Adv. Mater., 15, 1708 (2003). 12. N. S. Sariciftci et al., Synth. Met., 59, 333 (1993).

785

Enhanced polariton decay in LiNb03 due to stimulated emission of acoustic phonons J. Hebling'' A. G. Stepanov', G. Almasi', and J. Kuhl' ^ MPI fur Festkorperforschung, D-70569 Stuttgart, Germany " Department of Experimental Physics, University of Pecs, H-7624 Pecs, Hungary E-mail: [email protected] Abstract. Optical rectification reveals saturation of the THz pulse energy with increasing pump energy. Electro-optic sampling of the THz pulse shape exhibits strong evidence that for LiNbOg this is mainly caused by stimulated emission of acoustic phonons. Recent experiments of THz pulse generation by optical rectification of ultrashort light pulses have shovm [1,2] that the THz conversion efficiency is decreasing with increasing pump pulse energy. For GaSe increased absorption of free carriers generated by two-photon absorption can be responsible [1] for that. For LiNbOa (LN) and pump pulses with 800 nm [2], however, two-photon absorption is impossible.

I^ 3

•e ^ „ ^ 0

. /\

;;

a T = 32K

/ \ 'A /'

« o '« c O)

Q.-1

o o

^

UJ . 9

0

1 Time (ps)

2 Frequency (THz)

Fig. 1. (a) Electro-optic sampling signals of the THz pulses generated in LiNb03 at 32 K using pump intensities of I^ (dashed curve), 1.5 x I^ (dotted curve), and 2.2 x I^ (solid curve). The curves are normalized for the first positive peak, (b) The corresponding normalized spectra obtained by Fourier transformation of the time domain data.

First a simple heating of the LN crystal by the pump pulses was ruled out as the source of the efficiency drop by measuring the THz generation efficiency using chopped pump pulses with different duty cycles. This measurement proved that the efficiency depends on the pump pulse energy, but not on the average pump power. In order to identify the origin of the saturation of the THz pulse energy with increasing pump pulse energy, we measured the generated THz pulse shape at different temperatures and pump pulse energies by electro-optic sampling. The THz pulses were generated in a nearly stoichiometric LN crystal doped with Mg at a level of 0.6 mol%. The experimental conditions were very similar to those used

786

in ref. 3. The diameter of the pump spot on the crystal was about 0.6 mm, and the THz pulses were generated in a volume of about 0.5 mm^. At 2.3 |iJ excitation energy, the energy of the THz pulses leaving the crystal was 100 - 400 pJ depending on temperature. Figure la depicts electro-optic sampling signals of the THz pulses generated at 32 K for three pump pulse energies. The data show an enhanced decay of the signal for the highest pump intensity. Principally such behavior could result from increased absorption of free carriers generated by threephoton absorption of the pump pulses. However, this explanation can be ruled out since free carrier absorption is stronger for smaller frequencies [4], but our measurements (see. Figure lb) indicate a larger increase of the absorption with increasing pump intensity at higher frequencies.

CO

_0

2

0

1

Time (ps)

1

2

3

Frequency (THz)

Fig. 2. (a-c) Electro-optic sampling signals of the THz pulses generated in LiNbOjat 14, 95 and 300 K using pump intensities of 1.5 x I^ (dotted curves), and 2.2 x I^ (solid curves). The curves are normalized for the first positive peak, (d-f) The corresponding normalized spectra obtained by Fourier transformation of the time domain data.

We suggest a very different explanation: The THz pulses inside the LN crystal form phonon-polaritons which decay into pairs of acoustic phonons by anharmonic interaction. The decay rate is given by [5]

T{v) = const, • («i + /72 +1) • A (y)

(1)

787

where v, Vj, V2 are the frequencies of the polariton and the two acoustic phonons, D^iv) is the two-(acoustic)phonon density of states (it is proportional to v^), 0(v,VpV2) is the Fourier transform of the cubic anharmonic force constant (it is proportional to Vi-V2) and n^ and n^ are the occupation numbers of the acoustic phonon modes. According to Eq. 1 besides the spontaneous decay of the polariton, decay processes stimulated by acoustic phonons play an important role if the occupation numbers are non-negligible as compared to 1. In thermal equilibrium this process depends only on the temperature. However, in our experiment the number of the generated phonon-polaritons can be higher than the number of the acoustic phonon states (a few times 10^^). This high number of coherent polaritons obtained at high pump intensities can create a number of acoustic phonons comparable to the number of the thermal acoustic phonons at low or medium temperature and thus results in a faster polariton decay. According to Eq. 1, the enhancement of the polariton decay, and consequently the efficiency drop is strongly increasing with the polariton frequency. This is in fiill agreement with our observations (see Fig. lb). Our assumption is further supported by the fact that at room temperature the polariton decay is less sensitive (see Fig. 2c and 2f) to the pump intensity than at 32 or 95 K. At elevated temperatures the thermal acoustic phonon population is so high, that the absorption of the THz pulse does not lead to a significant increase. At 14 K, the observed polariton decay is also less sensitive to the excitation intensity than at medium temperatures. This finding also agrees with our explanation, since at low temperature the occupation numbers n^ and n^ are much smaller than 1, and thus the induced decay remains relatively low. In conclusion, to the best of our knowledge this is the first report on enhanced decay of coherent phonon-polaritons caused by stimulated emission of acoustic phonons and also the first nonlinear optical process investigated by THz pulses obtained by optical rectification. Acknowledgements. Support from the Hungarian Scientific Research Fund Grant No. T038372 is kindly acknowledged.

References 1 K. Reimann, R. P. Smith, A. M. Weiner, T. Elsaesser, and M. Woemer, Optics Lett. 28,471,2003. 2 A. G. Stepanov, J. Hebling, and J. Kuhl, Appl. Phys. Lett. 83, 3000, 2003. 3 J. Hebling, A. G. Stepanov, G. Almasi, B. Bartal, and J. Kuhl, Appl. Phys. B 78, 593, 2004. 4 P.Y. Yu, and M. Cardona, Fundamentals of Semiconductors, Springer-Verlag, Berlin, 2001. Chapter 6.5. 5 U.T. Schwarz, M. Maier, Phys. Rev. B 53, 5074, 1996.

788

Ultrafast Electrooptic Deflector Using Quasi-Velocity-Matching Kyoji Shibuya, Shintaro Hisatake, Haruya Kitano, and Tetsuro Kobayashi Division of Advanced Electronics and Optical Science, Graduate School of Engineering Science, Osaka University, 1-3 Machikaneyama-cho, Toyonaka, Osaka 560-8531 Japan. E-mail: shibuya@ laser.ee.es .osaka-u.ac.jp Abstract. We propose an ultrafast electrooptic deflector using quasi-velocity-matching and demonstrate the beam deflection with the repetition of 16.25 GHz and the maximum resolvable spot number of 12.5.

1.

Introduction

Optical deflectors are one of the basic components in an optoelectronics system. Up to now various kinds of optical deflectors have been developed and applied in various applications. Electrooptic deflectors (EODs), which have the advantages of high operating speed and excellent controllability, have potentialities for many applications such as the ultrashort pulse generation with high-repetition-rate [1, 2], a timedomain demultiplexer for optical information systems and an optical streak camera. However, high-performance EODs have not been realized so far because of difficulties for electrooptic modulators to operate sufficiently in the microwave region. The phase modulation is cancelled every an interaction length by velocity mismatch between light and microwave. A method for compensating the velocity mismatch using periodic domain inversion, which was named quasi-velocity-matching (QVM), has been proposed [3]. Using the method, electrooptic modulators with large modulation index are obtained easily. In this paper, we propose and demonstrate an EOD with high-speed and large amplitude of deflection using the QVM technique. And we discuss the application of the EOD.

2.

Theoretical Considerations

Fig. 1 shows a schematic of the QVM EOD. When a light with the group velocity VQ propagates along t/-axis with a coUinear traveling microwave with the phase velocity Vm in the device, the electric field of the microwave E{y, t) which the light entering the crystal at time t encounters at position y can be written as E{y, t) = Em cos (27r/^t - ^ i / )

(1)

where E^ is the amplitude of electric field of microwave, fm is the microwave frequency and L is a half period of domain inversion for the condition of the QVM given by 1/ = l/[2/^((l/'i;y^) — {l/vo))\. Considering that the sign of change of refractive index induced by the traveling microwave is inverted in domain inversion 789

E(y. t)

A(t)(x, t)

0 /

L

electrooptic crystal

2L

3L I

4L

2mL (2m+1)L

2(m+1)L

doTiain inversion region

Fig. 1. Schematic of the QVM electrooptic deflector regions, the phase shift A0(a?, t) which is given to a light passing through the length of 2JL in the device is given by

A(f){x, t)

r'^L 1 A6 / '^'^hssEiy, t)g{x, y)dy = —j^^ A 1 (non-domain-inversion regions) —1 (domain inversion regions)

27r

sin 27r/^t

(2)

{

where A is the wavelength of light in a vacuum, Ue is the extraordinary refractive index of the crystal, 733 is the electrooptic coefficient of the crystal, d is the width of the device in x direction and A(j)rn{= ^Lnl^yssEm/X) is the maximum phase modulation index obtained at x = ±d. Since (2) expresses a spatial linear phase shift whose slope varies sinusoidally with time, this device operates as a deflector whose deflected angle varies sinusoidally with time. The far-field electric field of the deflected Gaussian beam, Efar (^51), is approximated as Efari^^t)

OC exp

X —

XfmA4>ri 27rrf

sin27rfmt

(3)

where wis 3. half Gaussian beam width, / is the focal length of a Fourier transform lens and m is the number of the section of 2L. Equation (3) is vaUd for w < d. The resolvable spot number N which represents the performance of the EOD is defined by the ratio of the full amplitude of deflection to the beam spot size in the far-field plane. We can derive that N is proportional to mA(j)m from (3).

3. Experimental Results and Discussion We fabricated a QVM EOD with LiTaOs crystal and demonstrated the ultrafast beam deflection using a 514.5 nm wavelength Ar laser as a light source and a magnetron operating at 16.25 GHz as a modulating microwave source. The deflected beam was observed with a streak camera whose minimum time resolution is less than 300 fs. Fig. 2 shows a streak camera image of the deflected beam at the maximum modulating power. The modulation index of the deflector can be estimated by 790

time

Fig. 2. Streak camera image of the deflected beam at the maximum modulation index 25 rad spectrum analysis of output light. In this case, the maximum modulation index is estimated to be 25 rad, and resolvable spot number is estimated to be 12.5. This value is still low comparison with expected performance of the QVM EOD. We think that larger modulation index can be obtained by improving the device. As an application of the ultrafast EOD, we propose an optical streak camera system. A beam in the far-field plane passing through two EODs arranged at right angles to each other constructs a Lissajou's figure on a screen. We can obtain a circular track by giving the appropriate phase difference between the electric fields of modulating microwave applied to two EODs. Since the beam spot in the circular track moves at fixed velocity, an arc on the track corresponding to fixed time. Therefore this system works as an optical streak camera. The time resolution of this streak camera system is estimated by the number of resolvable spots in a circumference of the beam track. Using (3), the time resolution A r is given by Ar:

7rfmmA(l)mW

Using an experimental result mA
(4)

16.25 GHz, the time

4. Conclusions We proposed an ultrafast EOD using quasi-velocity-matching and demonstrated the beam deflection with the repetition of 16.25 GHz and the maximum resolvable spot number of 12.5. As an application of the EOD, we also proposed an optical streak camera system. The time resolution of the streak camera system is estimated to be about 1 ps if the EOD used in this experiment is used for the system.

References 1 T. Kobayashi, H. Ideno and T. Sueta, J. Quantum Electron. QE-16, 132-136, 1980. 2 B. Y. Lee, T. Kobayashi, A. Morimoto and T. Sueta, J. Quantum Electron. 28, 17391744, 1992. 3 D. -S. Kim, M. Arisawa, A. Morimoto and T. Kobayashi, IEEE J. Select. Topics Quantum Electron. 2,493-499, 1996.

791

Ultrafast Control of a Surface Plasmon Resonance via the Insulator to Metal Transition in V02Nanoparticles Matteo Rini\ Andrea Cavalleri\ Rene Lopez^, Lynn A. Boatner^, Richard F. Haglund, Jr.^, Tony E. Haynes^, Leonard C. Feldman^'^ Robert W. Schoenlein^ ^Materials Sciences Division, Lawrence Berkeley National Laboratory, Berkeley CA 94720, USA E-mail: [email protected] ^ Department of Physics and Astronomy, Vanderbilt University, Nashville TN 37235, USA ^ Condensed Matter Sciences Division, Oak Ridge National Laboratory, Oak Ridge TN 37831, USA

Abstract. We report on the study of the ultrafast insulator-to-metal transition in nanoparticles of strongly correlated VO2. The particles are grown by ion-implantation and self-assembly in a Silica matrix and can be switched between the insulating and metallic phase within less than 100 fs. The prompt formation of the metallic state results in the appearance of a surface-plasmon resonance that is absent in the bulk and can be further tailored by controlling the particle shape. The physics of strongly correlated compounds has important ramifications in both fundamental science and technology. The many phase transitions exhibited by these systems are in fact associated with large changes in their electrical, optical and magnetic properties. Such multi-stable behavior is also extremely sensitive to external stimuli, due to the many interacting degrees of freedom and competing states of the system. Recently, a number of studies have reported large optical nonlinearities in such strongly correlated solids \ How^ever, compatibility with existing technologies, operating temperatures and wavelengths are often critical limitations for practical applications. Recently, the design of VO2 nanoparticles in silica and in optical fibers has been demonstrated, opening attractive new avenues for optical-device applications^. Fabrication of nanometer-scale structures allows for the design of specially tailored materials that adapt the property of extended solids to the specific application^. Important features of the VO2 phase transition are: (1) the possibility of photoinducing it on the ultrafast timescale"^'^, (2) the large optical changes in the near IR and (3) the near-room temperature transition point (340 K). The availability of these crystallites makes it possible to exploit the change in optical properties associated with its insulator-to-metal transition in a fiber-optic environment. Here, we study the near and mid IR photo-response of VO2 nanoparticles in silica, showing that a dramatic increase of absorption can be photo-induced on ultrafast time scales. In brief, the transition from the insulating to the metallic state is responsible for the large, ultrafast changes in the optical properties, with a

792

combined contribution from the rearrangement in the density of electronic states (bandgap collapse) and from the formation of a dielectrically confined surface plasmon in the metallic phase. The semiconducting gap, around 0.7 eV, disappears upon the phase transition to the metal state. Furthermore, since the features of the surface-plasmon resonance depend on the size and shape of the VO2 nanoparticles, the optical switching characteristics of nanorods of different geometry exhibit substantial differences. Proper nanoscale materials design can thus lead to optimal response at telecom wavelengths.

100

^ o

* \

90

40-nm spheres

^ •

CO

I 80

b/a = 2

V^

rods

_

.

4//g^

H 70^

1

-0.5

•

1

•

0.0

1

0.5

•

1

•

1.0

1

1.5

•

1

2.0

Pulse Delay [ps] Figure 1. Ultrafast response at 1.55 jim for nanospheres and nanorods. The excitation wavelength is 800 nm. The particles are obtained by ion implantation and high temperature annealing. Size and aspect ratio can be controlled by varying the annealing time. The total amount of material is nearly identical in the two cases, distributed over a layer of 100-nm thickness. Fig. 1 shows the ultrafast response in the near IR induced by 100-fs excitation at 800 nm with 1 mJ/cm^. The initial drop in the optical transmission of single monolayers of spheres and rods is associated with the formation of the metallic phase and is pulse-width limited, i.e. faster than 120 fs. This is consistent with the fundamental timescale of 75 fs found in the bulk^. Importantly, we find that the response can be strongly affected by the shape of the nanoparticles, with the optical switching efficiency of the nanorods twice that of the nanospheres, despite the fact that in both cases the two samples contained the same amount of VO2. To further substantiate our assignment of the physical origin of this optical switching, we note that the dynamics exhibit a well-defined fluence threshold and saturation, indicative of a photoinduced phase-transition. The transient AT/T spectrum measured 200 fs after 800-nm excitation is reported in Fig. 2 for the case of spherical particles. A clear resonance in the AT/T signal between 1 and 2 microns is observed, in good agreement with static spectroscopic studies of the thermally induced phase transition^. This is evidence that the VO2 nanoparticles have reached the high temperature metallic phase immediately after excitation.

793

thermally induced photoinduced @ 200 fs

0.0 h

1.0

1.5

2.0

2.5

Wavelength [|Lim] Figure 2. Relative change of transmission across the phase-transition for the VO2 nanospheres. Solid lines: thermally induced. Squares: measured at a delay of 200 fs after photoexcitation. In summary, we have studied the photo-induced response of VO2 nanoparticles in a silica matrix. We find that the photo-induced insulator-to-metal transition is responsible for very large changes of the optical properties in the infrared, in particular at the technologically relevant wavelength of 1.3-1.5 jLim. Furthermore, we find that the magnitude of the response can be tailored by means of controlling size and shape of the particles during the self-assembly step of the growth procedure. Shape and size act on the optical properties through the surface plasmon resonance. Higher modulation depths can presumably be obtained by increasing the density and implantation depth of the VO2 nanoparticles, whereby a true transparent-to-opaque transition could be obtained. It is also important to point out that the 300 jiJ/cm^ threshold for the photo-induced phase transition is equivalent to 150 pJ pulse for a typical 50-|Lim^ mode size in a single-mode fiber, making this and similar schemes attractive for real-world applications. This work was supported by the U.S. Department of Energy under Contract No. DE-AC03-76SF00098.

References 1 H. Kishida et al. Nature 405, 929 (2000); T. Ogasawara et al. Phys. Rev. Lett 5, 2204 (2000). 2 Lopez et al. Opt. Lett. 27, 1327 (2002). 3 U. Huynh, J. J. Dittmer, A. P. Alivisatos Science 295, 2425 (2002). 4 Becker et al. Appl Phys. Lett. 65, 1507 (1994). 5 Cavalleri et al. Phys. Rev. Lett. 87, 237401 (2001). 6 A. Cavalleri et al. Cond. Mat. 0403214.

794

External Generation of Flat Power-envelope THz Modulation Sidebands from a CW Laser based on an Electrooptic Phase Modulator Shintaro Hisatake, Yuko Nakase, Kyoji Shibuya, Masato Tobinaga, and Tetsuro Kobayashi Division of Advanced Electronics and Optical Science, Graduate School of Engineering Science, Osaka University, 1-3 Machikaneyama-cho, Toyonaka, Osaka 560-8531 Japan. E-mail: [email protected] Abstract Flat power-envelope terahertz-wide modulation sidebands have been generated by only electrooptic phase modulation of continuous wave laser light. Generation of 46 sidebands spaced by 16.25 GHz within -3 dB bandwidth has been demonstrated.

1.

Introduction

An optical frequency comb of phase modulation sidebands forming ultrashort pulses is expected to play a major role in various fields as a new light source, which consists of a set of continuous wave (CW) laser light sources that are separated by a constant frequency spacing with a fixed phase relationship. An external phase modulation method to generate the sidebands has several attractive advantages: low power CW laser of any wavelength with narrow linewidth is applicable for light sources, wide (>10 GHz) and tunable frequency spacing can be reahzed easily, stable, and so on. However, there is one big drawback; the amplitudes of some of generated spectral components, which are determined by the Bessel function of the first kind, can be extremely small, or even be zero for the specific modulation index. This drawback limits the flexibility to use the modulation sidebands as a set of CW lasers. In this paper, we propose and demonstrate a new technique to directly generate the flat power-envelope terahertz-wide modulation sidebands from a CW laser based on a bulk type external electrooptic phase modulator (EOM).

2.

Principle of Operation

The basic structure of the proposed sideband generator is shown in Fig. 1(a). An incident CW laser beam with an amplitude distribution function of A{x) is modulated with the modulation index of AOmi^) depending on the position x in the beam cross section. The power of the nth sideband in the output beam, which is taken out through the Fourier transform lens and the slit with narrow enough width, can be written as. In oc

oo

/

A{x)Jn{Aem{x))dx

(1)

-oo

where Jn is the nth Bessel function of the first kind. In the proposed scheme, the 795

output

2L=5.69{toi inverted

^ phase modulator "^T lens (focal length: f) FT lens -0.25 0 0.25 >t[nfim] (a) Basic structure of the proposed flat power-envelope (b) Designed shape of domain inversed region sidebands generator. FT lens: Fourier transform lens.

Fig. 1. Schematic of proposed flat power-envelope sidebands generator.

amplitudes of each sideband are manipulated by designing the distribution function of the modulation index, AOmiof^)- Based on the least squares fitting, we have determined AOm{x)to realize aflatterpower-envelope: A6m{x) = A0yn{(O.556/D^)a?^ d= {I.bb6/D)x + 1}, where "+" for x < 0 and "-" for a? > 0, respectively. In the optimization, we assumed the Gaussian beam of A{x) oc exp[—{x/D)^], Here we call A^rn as a maximum modulation index. Note, optical sidebands with power-envelope of other shape can also be synthesized from a CW laser through this method. At the low modulation frequency, AOm{x) can be controlled by the electrode shape as shown in Fig. 1(a). To generate widespread sideband spectrum, deep modulation is required. Considering a traveling-wave EGM, when it has a suitable half domain-inverted period L of L = l/[2fm{l/vm -^ l/'^o)]> where Vm is the phase velocity of the microwave, -^o is the group velocity of the light, and fm is the modulation frequency, the quasi-velocity matching (QVM) occurs and accordingly a large modulation index is achieved [1]. In the QVM phase modulator, the spatial distribution of the modulation index can be achieved by distributing the length of the domain inversion within the optical beam cross section by designing the shape of each domain-inverted region in the periodic domain inversion. The relation between the spatially distributed length, W{x)., of the domain inversion in the elemental section (see Fig.l(b)) and the modulation index, AOm{^), is simply calculated as [2]: AOm{x) = A4>msH^W{x)/{2L)), Fig. 1(b) shows the designed shape of the domain inverted region (2L) realizing the optimized phase modulation index distribution. We fabricated the designed QVM phase modulator with z-cut LiTaOs crystal (length: 30 mm, thickness: 0.5 mm) cascading five elemental sections (5 x 2L = 28.45 mm).

3. Results and Discussion In the experiment, a 514.5 nm Ar laser was used as a light source, and a 16.25 GHz pulsed magnetron (pulse width: 1/is, repetition rate: 1 kHz) was used as a modulating microwave source. ' Fig. 2 (a) show a streak trace (Hamamatsu: C3735-01S) and intensity profile of the output pulses. The focus spot around the slit shown in Fig.l (a) moves periodically according to tlie distribution function of the modulation index, hence the optical pulses are extracted periodically through the narrow slit. The repetition rate 796

—WmM

— Calculated A(t>. = 33.3

T i m e [ps]

(a)

-o-CalculatedA^-33.3

Offset frequency [GHz]

(b)

Fig. 2. Experimental results, (a) Streak trace and intensity profile of the output beam, (b) Measured and calculated sidebands. is 32.5 GHz and achieved pulse width is 1.7 ps. In the figure, dotted line indicates the theoretical intensity profile calculated with A^rn = 33.3 rad. Fig. 2 (b) show the measured (solid line) and the calculated (open circle) output spectra. The observed spectrum agree well with calculation. Forty-six sideband lines forming optical comb of over 0.8 THz have been generated within -3 dB bandwidth. The spectrum span over 1 THz has been achieved within ±3 dB bandwidth. The amplitudes and the phases of the produced sidebands can be controlled individually by spatial filter in a conventional 4-f Fourier transform optical layout to shape arbitrary complex waveform of the ultrashort pulses.

4.

Conclusions

We have proposed a new technique of realizing the power-envelope manipulation simultaneously with the generation of widespread sidebands and experimentally demonstrated the generation of flat power-envelope terahertz-wide sidebands from a CW laser based on an external phase modulation method. Forty-six sideband lines forming optical comb of over 0.8 THz have been generated within -3 dB bandwidth (1 THz within ± 3 dB). The output pulse width was 1.7 ps and the repetition rate was 32.5 GHz. The generated flat power-envelope sidebands are applicable to not only seed light for the ultrashort optical pulse synthesis but also WDM optical communication system, spectroscopy, frequency measurements and so on. Acknowledgements. This research was partially supported by the Ministry of Education, Science, Sports and Culture, Grant-in-Aid for Young Scientists (B), 16760040, 2004.

References 1 A. Motimoto, M. Tamaru, Y Matsuda, M. Arisawa, and T. Kobayashi, in Pacific Rim Conf. on Lasers and Electro-Optics, TiA, 1995. 2 T. Khayim, M. Yamauchi, D. -S. Kim, and T Kobayashi, in IEEE J. Quantum Electron., Vol 35, 1412, 1999.

797

Part XII

Microfabrication by Femtosecond Laser Pulses

3D photonic devices fabricated in glass by a femtosecond oscillator A. M. Kowalevicz^ V. Shanna\ E. P. Ippen\ J. G. Fujimoto^ and K. Minoshima^ ^ Department of Electrical Engineering and Computer Science and Research Laboratory- of Electronics, Massachusetts histitute of Technology, Cambridge, MA 02139, USA ^National histitute of Advanced Industrial Science and Technolog\^, AIST Tsukuba Central 3, M 4 , Umezono, Tsukuba, Ibaraki, 305-8563, JAPAN Abstract: Photonic structures are fabricated in glass using high energy pulses from an extended cavity femtosecond oscillator. Several novel tliree-dimensional devices are demonstrated and characterized. The ability to fabricate in transparent materials enables an entirely new class of 3D photonic devices that are not possible in planar geometries.

1. Introduction Recent work on photonic device fabrication in transparent materials using femtosecond pulses has enabled the creation of a wide variety of devices. Because the material modification is initiated by a nonlinear interaction, structures can be created within the bulk of the substrate. Two-dimensional devices, such as X and directional couplers [1, 2], and interferometers [3] have been demonstrated. A three-dimensional coupler was fabricated using an amplified femtosecond laser [4], but oscillator machining has been limited to straight 3D waveguides spaced over several tens of microns in depth [5]. In these proceedings, we report the demonstration of several novel 3D photonic devices.

2. Experimental description and results Our devices were fabricated using pulses generated from a high pulse energy femtosecond oscillator capable of generating 150-nJ pulses [6]. The increased energy gives greater flexibility and versatility to the machining process. We use an effective 0.9 NA, 70 X, microscope objective to focus -30 nJ pulses (--15 nj transmitted) into the glass (Coming 0215) substrate. Our experimental setup utilizes high resolution, three-axis computer controlled stages to translate the glass sample through a predetermined path. By scanning the substrate at a constant speed of 10 mm/s in directions transverse and longitudinal to the incident radiation, we were able to fabricate structures at various levels within the bulk glass. The glass substrates were then cut and polished to permit access to the device for characterization and analysis. Figure 1 (a) shows a schematic of a l-to-4 waveguide coupler. The device is composed of a total of 3 X-couplers, which allow for a single input and a total of

801

fiOfon

*

•i-

2Spni

(a)

(tj}

Figure 1, (a) Schematic of a l-to-4 3D coupler with outputs on 3 separate levels of the substrate, (b) CCD image of the output modes from the waveguides; the output ports are separated by 50 jam along the x-direction and 25 {xm along the z-axis. four outputs at three separate depths within the glass. Figure 1 (b) shows a CCD image of the output facet of the device in reverse gray scale after light at 800 nm was coupled into the input port. The waveguides at the output are separated by 50 \xm in the horizontal direction and 25 jim vertically. The slightly varying intensities are hkely due to minor imperfections in the movement of the stages during the writing of the -2 \xm diameter waveguides used in these devices. It is possible to combine two different couplers in 3D to take advantage of the operating characteristics of each device. Figure 2 (a) shows the schematic of a device with a horizontally oriented broad bandwidth X-coupler and a vertically oriented directional coupler. For spectral characterization, broadband light from a mode-locked Ti: Sapphire laser was coupled into the device. Figure 2 (b) shows a comparison of the reference spectrum and the output of the broad bandwidth port. The excellent agreement of the spectra demonstrates near wavelength independent coupHng over the fiill 150 nm of the light source. The output of the top and bottom guides are also shown in Figure 2 (b), normaUzed to the output of the broad bandwidth port. The wavelength dependence of the directional coupler divides the spectral components between the two output ports. Shorter wavelengths are coupled to the lower port, while longer wavelengths are transmitted through the upper port. Broadband Output

"Jppe^'^ort ' l ^ r {a)

Porl^ VAvriMt^TtM^vCnm)

Figure 2. (a) Schematic of the device with a broad bandwidth coupler in the horizontal plane, and a directional coupler in the vertical plane, (b) Spectral comparison of the output ports, demonstrating both wavelengtli division capability and broadband coupling.

802

•^m 50nm

/

\ (b)

Figure 3. (a) Schematic of a symmetric 3-waveguide directional coupler. Waveguides are separated by 50 |Lim at ends and 5 \xm within the interaction region, L. (b) hiverse gray scale CCD image of the waveguide outputs. The increased dimensional freedom enables the fabrication of a 3-waveguide directional coupler (Figure 3 (a)). The initial and final waveguide separation is 50 lam, while the separation in the interaction region is 5 \xm. Figure 3 (b) shows the characterization of a coupler with L = 2.00 mm. The CCD image shows that the output of the input guide maintains 43% of the total power, v^th the remaining 28% and 29% being transferred to the upper right and left outputs, respectively.

3, Conclusion Using a high pulse energy femtosecond oscillator, it is possible to fabricate photonic devices in glass. We report the demonstration several novel 3D devices. The ability to fabricate waveguides in three dimensions will enable higher device densities, as well as an entirely new class of 3D photonic devices that were not possible in planar geometries. Acknowledgements. This research was supported by AFOSR F49620-01-10084, the AFOSR Medical Free Electron Laser Program F49620-01-1-0186, and NSF ECS-0119452, A.M. Kowalevicz is also with the Division of Engineering and Applied Sciences of Har\'ard University, Cambridge, MA 02138, USA.

References: 1 D. Homoelle, S. Wielandy, A. L. Gaeta, N. F. Borrelli, and C. Smith, Optics Letters 24,1311,1999. 2 A.M. Streltsov and N. F. Borrelli, Optics Letters 26,42,2001. 3 C. Florea and K. A. Winick, Joumal of Lightwave Technology 21,246,2003. 4 S. Nolte, M. Will, J. Burghoff, and A. Tuennermann, Applied Physics A 77, 109, 2003. 5 K. Minoshima, A. M. Kowalevicz, I. Hartl, E. P. Ippen, and J. G. Fujimoto, Optics Letters 26, 1516,2001. 6 A. M. Kowalevicz, A. T. Zare, F. X. Kaertner, J. G. Fujimoto, S. Dewald, U. Morgner, V. Scheuer, and G, Angelow, Optics Letters 28,1597, 2003.

803

Writing of photonic devices and waveguide lasers by a diode-pumped femtosecond oscillator Roberto Osellame', Nicola Chiodo', Giuseppe Delia Valle', Stefano Taccheo', Roberta Ramponi', Giulio Cerullo', Alexander Killi^, Uwe Morgner^, Max Lederer\ Daniel Kopf^ ' National Laboratory for Ultrafast and Ultraintense Optical Science - INFM, Istituto di Fotonica e Nanotecnologie - CNR, Dipartimento di Fisica - Politecnico di Milano, Piazza L. da Vinci 32, 20133 Milano, Italy e-mail: [email protected] ^ Max-Planck-Institut fur Kernphysik, Saupfercheckweg 1, D-69117 Heidelberg, Germany ' HighQLaser Production GmbH, Kaiser-Franz-Josef-Str. 61, A-6845 Hohenems, Austria Abstract. We demonstrate active waveguide writing by a compact diode-pumped cavity dumped femtosecond Yb:glass oscillator. The waveguides are perfectly mode-matched to standard single-mode telecom fibers at 1.55 |,im and show internal gain and laser action.

1.

Introduction

Direct writing of waveguides and integrated optical devices by femtosecond laser pulses has attracted much attention in recent years due to its strong potential for telecom applications. Most experiments on this subject have used amplified, kHz repetition rate, Ti:Sapphire lasers [1, 2]. These systems provide high pulse energy but are complex, expensive and poorly suitable to an industrial environment. A few groups are working with more compact sources [3,4], but still the complexity of these laser systems can be reduced. In this paper we show the possibility of writing high quality waveguide devices with a diode-pumped cavity-dumped Yb:glass femtosecond laser oscillator. The waveguides are single-mode in the telecommunication band and show laser action, for the first time to our knowledge, in a cavity with fiber Bragg gratings.

2.

Experimental Methods

The femtosecond laser oscillator used in these experiments is based on an Ytterbium doped LG760 phosphate glass pumped by a single diode [5]. A semiconductor saturable absorber mirror (SESAM) is used to mode-lock the laser. Specially designed dispersive mirrors generate a sufficient amount of negative dispersion for soliton mode-locking. The laser includes a cavity dumper able to dump more than 30 percent of the intracavity energy and still have stable operation. The contrast ratio is better than 1:500. An average power of 45 mW is obtained in the dumped beam, yielding a pulse energy of 270 nJ. The spectral width of the pulses is about 4 nm at 1040 nm resulting in a Fourier limited pulse width shorter than 300 fs, which corresponds to the measured autocorrelation. An excellent pulse to pulse stability of <0.8% was measured.

804

The waveguides were written in a transversal geometry. The laser beam was deflected to the vertical direction and then focused by a microscope objective into the glass sample which was translated horizontally by a motorized stage (Polytec PI, Inc., M-511.DD). Two objectives were alternatively used: a 50X long working distance one (N.A. 0.6) and a lOOX oil immersion one (N.A. 1.4). The substrate used was a QX phosphate glass (Kigre, Inc.) doped with erbium (2%wt) and ytterbium (4%wt). Waveguides were written at a depth of 170 |Lim from the sample surface.

3.

Results and Discussion

Several waveguides were fabricated with the two microscope objectives at different writing velocities, ranging from 20 |Lim/s to 20000 |Lim/s, and using the maximum available pulse energy, 270 nJ. Waveguides written with the milder focusing (0.6 N.A.) show a transversal shape which is strongly asymmetric due to the writing geometry, as discussed in several previous works [1,2]. If the laser pulse is focused with the higher N.A. objective, a different behavior is found. A nearly circular cross-section is observed and the waveguide lateral dimension (about 5|Lim) is larger than that obtained with equal writing parameters but milder focusing. Waveguides written at a speed of 100 |Lim/s support a single mode at 1.55 jiim. I.U

^ C Z3

0.8

JZi

0,6

gS^^m^^^v

_ "

CO ^

0.4

•

^^^H W' ^Im

\\ V i

/ // /

-

"co c c -1

faite^^kiail

'"''^^- / ...1

-4

-2

0

1

2

4

6

8

10

Transversal axis [j,inn]

Fig. 1. Intensity profile in a transversal direction of the waveguide mode (continuous line) and of the fiber mode (dashed line). Near field image of the waveguide mode (inset). A near field of the single mode at a wavelength of 1.60 jum (chosen to avoid absorption from the Er ions), acquired with a Vidicon camera, is reported in Figure 1 (inset), together with the intensity profile in one transversal direction (continuous line). The measured intensity profile of a standard telecom fiber at the same wavelength is also reported (dashed line). The mode matching between the fiber and the waveguide is very good, giving a theoretical coupling efficiency as high as 96%. Numerical simulafions have been performed to estimate the index change inside the waveguides. Assuming a Gaussian refractive index profile with a 1/e full-width of 3.2 jum we obtained the best fit to the measured mode profile with a maximum index change of 9x10'^ Active optical characterization of the 12 mm waveguides was performed in a fiber butt-coupling configuration with index matching fluid, using bi-directional

805

diode-pumping at 980 nm. An internal gain peak of 2.7 dB at 1533 nm with 250 mW of pump power is observed. The measured insertion losses (including coupling and propagation losses) are of 1.5 dB, corresponding to a net gain of 1.2 dB at 1533 nm. By terminating the waveguide with two fiber Bragg gratings (see Fig. 2) laser action was observed, with output power of 200 ]xW. Much better results are anticipated using longer waveguides. His is the first time, to our knowledge, that laser action is demonstrated in a waveguide fabricated by femtosecond laser pulses. Diode pump

Active waveguide

K??t^R-:-U-^nrTTrT--><

^

WDM

Diode pump

:zL

/I

0.01 1E-3

> fiber Bragg gratings

' 0.1

1

1E-4 Optical spectrum ^^^^ analyzer

1E-5

1

L

]

lE-e 1530 1531 1532 1533 1534 1535 1536 Wavelength (nm)

Fig. 2. Setup for laser action and optical spectrum analyzer image of the laser line.

4,

Conclusions

The feasibility of waveguide writing with a very compact, diode-pumped, femtosecond laser oscillator has been demonstrated. The obtained waveguides are single mode in the telecom range, show nearly perfect mode-matching with standard single mode fibers and low propagation losses. As a consequence of such high-quality parameters, laser action has been observed for the first time in a waveguide directly written with a femtosecond laser. Acknowledgements. This research was funded by the European Union within the contract GlST-CT-2002-50266 (DACO)

References 1 K.M. Davies, K. Miura, N. Sugimoto, and K. Hirao, in Optics Letters, Vol.21, 1729, 1996. 2 R. Osellame, S. Taccheo, M. Marangoni, R. Ramponi, P. Laporta, D. Polli, S. De Silvestri, and G. Cerullo, in Journal of the Optical Society of America B, Vol.20, 1559,2003. 3 C.B. Schaffer, A. Brodeur, JT. Garcia, and E. Mazur, in Optics Letters, Vol.26, 93,2001. 4 K. Minoshima, A.M. Kowalevicz, I. Hartl, E.P. Ippen, and J.G. Fujimoto, in Optics Letters, Vol.26, 1516, 2001. 5 A. Killi, U. Morgner, M. J. Lederer, D. Kopf, accepted for publication in Optics Letters, 2004.

806

Toward the Fabrication of Hybrid Polymer/Metal Three-Dimensional Microstructures T. Baldacchini\ C. N. LaFratta^ R. A. Farrer\ A. C. Pons\ J. Pons\ M. J. Naughton^ B. E. A. Saleh^ M. C. Teich^ and J. T. Fourkas^ ^ Eugene F. Merkert Chemistry Center, Boston College, Chestnut Hill, Massachusetts E-mail: [email protected] ^ Department of Physics, Boston College, Chestnut Hill, Massachusetts ^ Department of Electrical & Computer Engineering, Boston University, Boston, Massachusetts Abstract. A method for coating three-dimensional microstructures fabricated by multiphoton absorption polymerization (MAP) with metallic patterns in a selective manner is introduced in this paper. We first use a small fraction of the output of a mode-locked Ti:sapphire laser to induce MAP in a homemade acrylic-based resin featuring a commercially-available photoinitiator. Then, through a novel multiphoton-induced process, we deposit metallic silver onto the surfaces of microstructures fabricated by MAP. In this paper we present preliminary results on the fabrication of hybrid polymer/metal microstructures.

1. Introduction Different groups have used multiphoton absorption polymerization (MAP) for the fabrication of three-dimensional microstructures having complex topologies [1,2]. In MAP, excitation of the photoinitiator occurs only in the small volume of a tightly focused laser beam, the position of which can be controlled in three dimensions. For this reason, MAP presents unique advantages with respect to more traditional microfabrication techniques. The properties of the materials used in MAP, which are generally acrylic or epoxy resins, pose the greatest limitations to the applications of this technique. Specifically, they lack electrical conductivity and are incompatible with most standard techniques for patterning electronic circuits. As a consequence, although MAP has been used for the fabrication of objects with complicated geometries with sub-micrometer resolution, few functional devices have been fabricated to date [3,4]. We introduce a direct laser writing method that allows for the deposition of metallic silver onto glass and polymer surfaces. In combination with MAP this technique permits the selective coating of three-dimensional microstructures with metals, and thus addresses some of the current limitations of this fabrication method.

807

2. Experimental Methods The experimental apparatus used in MAP is also used in metal laser deposition and it has been described elsewhere [5]. A mode-locked Ti:sapphire laser at a repetition rate of 76 MHz is used as the light source. The composition of the resin for polymerization consists of a commercially available photoinitiator (Lucirin TPO-L) and two monomers, ethoxylated(6) trimethylolpropane triacrylate and tris(2-hydroxyethyl)isocyanuratetriacrylate. Laser deposition of metallic silver is achieved utilizing a solution of polyvinylpyrrolidone (PVP) and silver nitrate in ethanol. While mixing, the solution changes color from clear to yellow because of the formation of silver nanoparticles. Spin-coating or slow evaporation of the solvent then affords a thin film of PVP on top of glass or polymeric substrate.

3. Results Figure 1 shows SEM images of microstructures fabricated by MAP. In Figure 1(a) a complex microstructure consisting of three bridges 12, 24 and 42 \xm tall is represented. In this case a power of 7.2 mW was used in conjunction with a stage velocity of 20 \im/s. The micro-cup in Figure 1(b) was fabricated with a laser average power at the sample of 5.8 mW and a stage velocity of 10 ^im/s. In both cases an objective with numerical aperture of 1.3 was used. Shrinkage upon polymerization is limited by the use of a monomer that presents stretchable moieties such as ethoxylated groups. When the laser is focused onto a PVP film in which silver nanoparticles and silver nitrate are embedded, material is deposited. Energy dispersive spectroscopy analysis has shown this material to be silver. Deposition occurs only when the laser is mode-locked, demonstrating that a multiphoton process is responsible for the reduction of silver nitrate. Two-dimensional silver patterns fabricated with this method were not conductive. Gold enhancement by electroless deposition of a solution of HAUCI43H2O and NH2OHHCI can be used to grow a conductive film of gold on top the deposited silver patterns. Figure 2(a) illustrates a two dimensional Ag pattern deposited on a glass surface. An objective with NA = 0.5 was used with 35 mW of power at the sample and the stage was moved at 35 ^im/s. Figure 2(b) shows a silver microstructure that has been coated with gold.

808

Fig.l. MAP-fabricated microstructures. (a) Interconnected micro-bridges (b) micro-cup.

Fig. 2. Laser metal deposition (a) Optical micrograph in transmission mode of a silver pattern with line connecting squares being 200 \im long (b) gold coated silver microstructure; the length of the c is 70 \xm.

4. Conclusion We have fabricated three-dimensional polymer-based microstructures by MAP. We have introduced a novel method to deposit selectively metallic silver in twodimensional patterns. The combination of the two techniques will facilitate the fabrication of hybrid polymer-metal microstructures.

References 1

S. Kawata, H.-B. Sun, T. Tanaka and K. Taksidsi, Nature 412, 697-698 (2001).

2

B. H. Cumpston, S. P. Ananthavel, S. Barlow, S. M. Kuebuler, D. M. Maugon, J. Qin, H. Rokel, M. Rumi, X.Wu, S. R. Marder and J. W. Perry, Nature 398, 51-54 (1999).

3^

S. Maruo, K. Ikuta and H. Korogi, Appl- Phys.Lett. 82, 133-135 (2003).

4

K. Kaneko, H.-B. Sun, X.-M. Duan, S. Kawata, .4;7/7/. Phys, Lett. 83,1-4 (2003).

5

T. Baldacchini, R. A. Farrer, J. Moser, J. T. Fourkas and M. J. Naughton, Synt. Met. 135-136,11-12 (2003). 809

Production of 30, dichroitic microstructures in nanocomposite glasses by femtosecond laser pulses Gerhard Seifert, Alexander Podlipensky, Amin Abdolvand, Jens Lange and Heinrich Graener Martin-Luther-University Halle, Physics Institute, Hoher Weg 8, D-06099 Halle, Germany E-mail: [email protected] Abstract. Deforming metal nanoparticles in glass by femtosecond laser pulses, 3dimensional dichroitic microstructures have been produced. This technique has a large potential for production of optoelectronic elements and long-time data storage.

Surface plasmon resonances (SPR) dominate the linear and nonlinear optical properties of metallic nanoparticles m dielectrics. These SPR can be varied within a wide spectral range throughout the visible and near infrared by choice of the metal and the dielectric matrix, or manipulation of size, shape and spatial arrangement of the metal clusters [1,2]. Thus, these compound materials are very promising candidates for a great number of applications in the field of photonics [3]. One of the main issues in this context is to structure the optical properties of such materials on a micro- or even nanometer scale. While current research on the latter is often aiming at fixture, sub-wavelength optics, the micrometer scale is appropriate for a lot of standard and advanced optical elements. Recently, we have Irradiation wavelength SSOnm , 500nni, ^OOnin

^^1^1 !^^%PJ^HH|^^^^H|'

2 mm

T

Fig. 1: Photograph of fs laser irradiated region in Ag nanoparticle containing glass sample, using linear polarization of monitoring light with a) polarization parallel to laser polarization, and b) polarization perpendicular to laser polarization, as indicated by the large arrows. The square areas of 2x2 mm^ have been produced irradiating with 400, 500, and 550 nm laser wavelength (see legend).

810

shown that irradiation of femtosecond laser pulses on glass containing silver nanoparticles (typical size 10 to 50 nm), using a wavelength near to the SPR, can lead to a permanent deformation of the nanoparticles to ellipsoidal shapes of uniform orientation; this results in dichroitic extinction spectra in the irradiated region [4]. A number of studies on the physical mechanism leading to these permanent changes (including electron microscopy [5], photoluminescence [6] and fs time-resolved experiments [7]) showed that ejection of electrons and Ag ions driven by the strong laser field and the ultrafast heating and cooling of the particle surroundings play the key role for the anisotropy of the deformation process. We have also demonstrated the potential of this technique to produce micro-structured dichroitic optical elements, e.g. polarization-sensitive diffraction gratings [8]. Here we show new results on soda-lime glass containing silver nanoparticles in a surface layer of » 6 jam thickness, with a significant depth gradient of particle size and volume fill factor. These particles with a mean diameter of 30-40 nm, which have been prepared by Ag-Na ion exchange and following annealing in hydrogen atmosphere, show depth-dependent SPR wavelengths due to their different sizes and, in particular, interaction-induced SPR red-shift. Thus, it becomes possible to address (deform) silver particles in different distances from the sample surface by irradiating the samples with femtosecond laser pulses of different wavelengths. As an example, we have irradiated our samples with 150 fs pulses derived from a mode-locked Ti:Sapphire laser with regenerative amplifier and frequency doublmg or OPA frequency conversion, at wavelengths of 400, 500 and 550 nm. Dichroitic regions have been produced in square areas of 2x2 mm by writing lines in a multi-shot regune at typically 10"^ shots/mm^ and a peak power density of 0.5 TW/cm^. The three different wavelengths have been applied successively onto areas slightly shifted with respect to each other, producing a red (irradiation at 400nm), a green (500nm) and a blue square (550nm) in different depths of the particle-containing layer. The result is given in Fig.la and b as photographs (reproduced as gray scale here), using linearly polarized light with the axis of polarization parallel (Fig. la) or perpendicular (Fig. lb) with respect to the polarization of the laser. Although the depth dependence here can only be seen by the imperfect focusing in the deeper layer, the nearly perfect dichroism of the structures produced is apparent comparing Fig. la and b. To prove the visual impression that the colored regions are located in different depths, the sample has been etched in 12% HF acid carefully from the surface; during this procedure, the three different colors could in fact be removed successively. The extinction spectra of the individual colored squares, as given in Fig. 2, have been recorded by an experiment applying the same irradiation parameters to an identical sample. So, it is an important conclusion of the experiments that the nanoparticles located in different depths and showing different SPR absorption, can be addressed and deformed separately by irradiation of fs laser pulses at different wavelengths. The resulting, strongly dichroitic extinction in the irradiated regions can be used to create various optical elements by this procedure; in particular, since the lower limit for the size of structures made with this method is the minimal focal spot size achievable with focusing optics, in principle arbitrary 3D-microstructures can be produced. Furthermore, the deformation of the nanoparticles produced has (as far as it is known at present) a very high long-time stability, so the technique has the potential to provide a new medium for long-term

811

data storage. Since it is an optical method, also advanced (e.g. holographic) storage technologies can be used in these materials. Finally, it should be mentioned that also the nonlinear susceptibility (spherical nanoparticles have a large y^^\ ellipsoidal ones even a considerable x^^^ [9]) of glass-based nanocomposite materials can be manipulated by femtosecond laser irradiation. So, in general dielectrics containing metal nanoparticles and the method described here provide a very broad and promising basis for future optoelectronic components and long-time data storage.

400

500

600

700

800

W a v e l e n g t h , nm Fig.2: Extinction spectra of the different square areas after irradiation with 400, 500, and 550 nm, recorded in linear polarization (parallel to polarization of fs laser pulses).

References 1 U. Kreibig, M. Vollmer, Optical Properties of Metal Clusters, Springer, Berlin, 1995. 2 V.M. Shalaev, Optical Properties of Nanostructured Random Media, Springer, Berlin, 2002. 3 R. Jin, C. Cao, E. Hao, G.S. Metraux, G.C. Schatz, C.A. Mirkin, Nature 425, 487, 2003. 4 M. Kaempfe, T. Rainer, K.-J. Berg, G. Seifert, H. Graener, Appl. Phys. Lett. 47, 1202, 1999. 5 M. Kaempfe, H. Hofmeister, G. Seifert, H. Graener, J. Phys. Chem. B 104, 11847, 2000. 6 A.V. Podlipensky, V. Grebenev, G. Seifert, H. Graener, J. Luminesc. 109, 135, 2004. 7 G. Seifert, M. Kaempfe, K.-J. Berg, H. Graener, Appl. Phys. B 71, 795, 2000. 8 G. Seifert, M. Kaempfe, K.-J. Berg, H. Graener, Appl. Phys. B. 73, 355, 2001. 9 A. Podlipensky, J. Lange, G. Seifert, H. Graener, I. Cravetchi, Opt. Lett. 28, 716, 2003.

812

Micrometer and sub-micrometer structures fabrication and analysis with femtosecond laser micro-nanomachining system Egidijus Vanagas, Jouji Kawai, Yury Zaparozhchanka, Dmitri Tuzhilin, Hirofiimi Musasa, Pavel Rutkovski, Igor Kudryashov, and Shoji Suruga Tokyo Instruments Inc., 6-18-14 Nishikasai, Edogawa-ku, Tokyo 134-0088, Japan E-mail: [email protected] Abstract. A concept of modular laser processing system, which can be flexibly optimized for processing by femtosecond laser pulses, is presented. The system capabilities and fabricated structures are demonstrated..

1.

Introduction

Femtosecond laser microfabrication is one of the streamlines of recent nanotechnology development. Physics of a femtosecond pulse and matter interaction have a number of the advantages that allows flexibly manipulate with designed system configuration. The control of a pattern shape and fabrication conditions are easy to implement with a relative low costs than with the other methods. Competitive abilities for small series of products, R&D in industry and academics institutions are making femtosecond microfabrication attractive for growing nanotechnology field. Mira900+RegA9000 (Coherent)

Hurricane (Spectra-Physics)

Step-motor Stage (Sigma)

Fig. 1. The block scheme of femtosecond laser micro-nanomachining system. Exchangeable system parts are shown as dashed squares.

813

2.

System Configuration

The femtosecond laser micro-nanomachining system (Tokyo Instruments, Inc.) is presented here. The core of the system is: main controller, software and optical unit, with build inside confocal laser microscope, pulse energy and polarization attuning, beam delivery part. Exchangeable system parts are: fs-laser, stage unit, optical focusing unit. Block scheme of the system is shown in figure 1. The software of the system is capable to accept pattern files created by standard software (*.dxf, *.txt, *.bmp). Main controller can master every laser pulse up to 300 KHz, and run all system units and functions.

3.

Examples of Applications

The system capabilities are demonstrated with presenting the results of the experiments carried during system development. The matrix of hillock-shape defects is fabricated on the surface of borosilicate glass. The density of more than 1 Gbit/cm^ and the sizes of defects from 100 to 150 nm [1] are obtained. Fabrication possibilities of photonics crystals and high-capacity 3D optical memory devices are demonstrated with the matrix of 6 Tbit/cm"^ density created [2]. The capabilities of the system for waveguides writing using direct writing method are demonstrated and discussed in the paper [2]. Small heat affected zone during material treatment with femtosecond pulses makes the method suitable for dicing, drilling, cutting, surface patterning [2,3]. The experiments of drilling [2], cutting and patterning [3] (figure 2) are performed. High precision, smoothness (less than 1 jim) and pattern manipulation flexibility are demonstrating.

Fig. 2. SEM images of rectangular hole cut out in a cover glass (thickness 170 \xm) (a); pyramid shape patterning of cover glass (b). Smoothness of fabrication is better than 1 jam. SEM courtesy of Prof. H. Misawa (Hokkaido univ.). (a)

(b)

SQum

Njw*"^

Fig. 3. CCD images of the matrix (10,000 points), fabrication time 10 seconds (a) and the portrait (26,000 points), processing time 39 seconds (b). Both patterns are fabricated on the surface of borosilicate glass using femtosecond laser micro-nanomachining system with galvanic mirror scanner option.

814

The system configuration with a galvanic mirrors scanner and Hurricane {Spectra Physics) femtosecond laser system is optimized for maximum efficiency (close to "1") at 1 KHz repetition rate. Almost every pulse is applied for processing. In figure 3 are shown patterns fabricated by the system with galvanic mirror scanner option. For particular pattern the optimization of the parameters of the system possible in order to get the best average performance rate. The confocal laser microscope built in femtosecond laser micronanomachining system allows 3D scan of fabricated pattern immediately after processing. The resolution up to <5r=200 nm (lateral), &=540 nm (axial) possible to achieve. In figure 4 are shown scans of the selected areas with increasing scanning resolution.

161

182

183

184

Fig. 4. The confocal laser microscope 2D scans of fabricated pattern on the glass with increasing scanning resolution. Scan (c) is performed with specified lateral resolution dr=2m nm.

4.

Conclusions

Femtosecond laser micro-nanomachining system structure and capabilities are presented. High-density sub micrometer resolution patterning; micro- cutting, drilling and structuring; confocal laser microscope scanning possibilities are demonstrated. Potential fields of the applications are: micro-, nano-technology appHcations (MEMS, MOEMS opto-electronics chips, biochemical chips, waveguided structures, high-capacity optical memory devices, photonics crystals, etc.), materials sciences, life sciences (medicine, biology, biochemistry, and biophysics). This experimental setup can make speedy, easy, and less expensive the multidisciplinary investigations. High-tech R&D forces supported with flexible faciUties take opto-electronics industry at the vantage for a new technologies implementation.

References 1 E. Vanagas, L Kudryashov, D. Tuzhilin, S. Juodkazis, S. Matsuo, H. Misawa, in Appl. Phys. Lett., Vol. 82, 2901, 2003. 2 E. Vanagas, D. Tuzhilin, M. Zinkou, A. Sedunov, N. Vasiliev, I. Kudryashov, V. Kononov, S. Suruga, in Laser Precision Microfabrication, Edited by I. Miyamoto, A. Ostendorf, K. Sugioka, H. Helvajian, Proc. SPIE, Vol. 5063, 374, 2003. 3 E. Vanagas, J. Kawai, D. Tuzhilin, I. Kudryashov, A. Mizuyama, K. G. Nakamura, K.-L Kondo, S. Koshihara, M. Takesada, K. Matsuda, S. Juodkazis, V. Jarutis, S. Matsuo, H. Misawa, in J. Microlith. Microfab. Microsyst., Vol. 3, 358, 2004.

815

Femtosecond Laser Effects on Osseous Tissues B. Girard', D. Y u ^ M.R. Armstrong % B.C. Wilson \ C.M.L. Clokie^ and R.J.D. Miller^ ' Department of Medical Biophysics, Ontario Cancer Institute, Toronto, ON. Canada E-mail: [email protected] ^ Department of Chemistry and Physics, 245-80 St-Georges street, University of Toronto, ON, Canada ^ Department of Maxillofacial Surgery, Mount Sinai Hospital, Toronto,ON. Canada

Abstract. We have investigated the effects of femtosecond (fs) laser irradiation on bone samples in in vitro and on ex vivo living bone samples. Ablation threshold, material removed per pulse and plasma shielding were examined using in vitro samples. Ablation threshold was found to be 0.9J/cm2 at 775nm and O.SJ/cm^ at 367nm using 200fs pulses. Material removal was found to vary non-linearly with pulse energy. Using in vivo samples we have demonstrated intact enzymatic activity on the surface of cells immediately adjacent to cells removed by fs laser irradiation suggesting no thermal damage.

1.

Introduction

Many factors can influence tissue rate removal: tissue type, pulse duration, pulse energy, wavelength. Bone is a very heterogeneous tissue and has two distinct portions: the dense outer part called the cortex, and a softer inner part known as cancellous. Our study focuses on the cortical portion because we hypothesize that efficient removal of the latter should also mean rapid removal of the cancellous portion. Structurally, cortical bone has a density of up to 1.6g«cm"^, and is composed of 70% inorganic material and 30% proteins. The inorganic mineral portion is mostly composed of hydroxyapatite (OHAP) Caio(P04)6(OH)2, while the proteins form the matrix, a collagen scaffold interwoven among the OHAP crystals. Because of its many different components bone has a very broad single photon absorption spectrum [1]. Because of this broad absorption spectrum no particular laser wavelength is suitable to ablate bone without thermal damage in the single photon absorption regime. Since femtosecond lasers have very high peak powers and can remove material through multiphoton absorption, electron avalanche and microplasma formation, we are investigating their use for bone removal. To determine the effects of femtosecond lasers on osseous tissues, we have separated our experiments in two main categories: 1) Effects of pulse energies and wavelength on hard tissues; 2) Irradiation of in vitro living samples

2. Experimental Methods Two different lasers systems were used in the experiments: 1 kHz, 200fs up to ImJ 'pulse-^ centered around 775 nm and IkHz, 150ns up to 6mJ •pulse" ^ at

816

535nm. Bone samples were retrieved from fresh pig mandibles, kept in saline at 4°C and used within days of harvest. Live tissue samples were harvested from 6day old mice and used immediately.

3.

Results and Discussion

Ablation Threshold. Cortical bone was irradiated at 775nm with pulses of decreasing energy until no further material removal was observed. The calculated ablation threshold for bone at 775nm is 0.9J«cm-2 and this is consistent with [2]. By doubling the laser frequency, we have found the bone ablation threshold at 387nm to be 0.3J«cm-2 Material Removed per Pulse. Cortical bone was exposed to increasing pulse energy. The amount of material per pulse was measured using an optical profilometer (Fig. 1.).

Pulse energy (uJ)

Fig. 1. Amount of material removed per pulse in relation to input pulse energy. The ablated depth increase nonlinearly with pulse energy but is not quadratic. Irradiation of Living Bone. Living cortical bone was obtained from 6-day old mice calvaria and used immediately after excision. To compare the importance of thermal damage caused by femtosecond versus nanosecond laser, the bone was stained for Alkaline phosphatase, an enzyme that is denatured at 56°C (Fig. 2.). The absence of color indicates enzyme denaturation.

817

Fig. 2. Microphotographs of excised mice calvaria after laser irradiation A) A.=775nm 100 HJ/pulse, 1 kHz, T = 200fs, Alkaline phosphatase (AP) staining of calvaria. AP (blue colour) is active on the cell surface in the area immediately adjacent to ablation. B) A.=535nm T = 150ns ImJ/pulse, 1 kHz, this picture clearly shows charring in the wound periphery, AP has been denatured up to 200mm from the ablated area, indicating a temperature rise above 56°C. The brown circular area is charred tissue. C) High power confocal image of wounded area in A showing cell covered with active alkaline phosphatase (white arrow) indicating that the enzymes on the cells have not been denatured by laser heat deposition and that the cell shape has not been altered.

4. Conclusions We have investigated the effects of femtosecond lasers on bony tissues. In living specimen, we have demonstrated that no thermal damage occurs on cells immediately adjacent to the wounded area as seen by evidence of active cell surface enzymes. These results w^ould suggest that bone healing may not be impaired in femtosecond laser surgery as is the case with nanosecond lasers [3]. These results will need to be validated in vivo. To our knowledge, intact enzymatic activity in cells immediately adjacent to cells removed by laser irradiation, has not been demonstrated before. The nonlinear dependence of energy vs depth per pulse was expected, although we didn't obtain a quadratic nonlinearity. This might be due to the extreme structural and chemical heterogeneity of bone. This non quadratic dependence remains to be explored further with bone, as well as with different biological tissues. Acknowledgements. Photonics Research of Ontario for providing funding for this research. #454192.

References 1 p. Spencer, J.M. Payne, et al. J Periodontol 70, 68, 1999. 2 W.B. Armstrong, J.A. Neev, et al. Laser Surg Med 30, 216-220 2002. 3 M. Buchelt, H. P. Kutschera . Lasers Surg Med 15, 373-381. (1994)

818

Femtosecond Laser Material Processing - How Short is Short? Yehiam Priori Kaiyin Zhang\ V. Batenkov^ Yuri Paskover^ I.Sh. Averbukh^ F. Korte^ and C. Fallnich^ ^ Department of Chemical Physics, Weizmann Institute of Science, Rehovot, Israel 76100 E-mail: [email protected] ^ Laser Zentrum Hannover e.V., Hollerithallee 8, D-30419 Hannover, Germany E-mail: [email protected] Abstract. Ultrashort pulses are routinely used for material processing. Since the ablation cannot be faster than the electron phonon equilibration times, temporal pulse shaping, and proper selection of pulse duration may offer advantages. We find that the shortest pulse is not always the best in terms of ablation efficiency and quality, and develop tools for using adaptive pulse shaping for the optimization process. During the last decade, femtosecond lasers v^ere demonstrated to be useful tools for material processing [1-8]. The fact that the deposited energy cannot diffuse aw^ay during the pulse gives rise to more efficient ablation and a reduced Heat Affected Zone [5,9]. The process is a nonequilibrium one: the energy is absorbed by the electrons raising the electron temperature to very high values, and only later is this energy redistributed, mostly via the electron-phonon coupling, causing the lattice to heat as well. Thus, typically, two (electron and lattice) temperature thermal models are used to describe these ultrashort interactions[10]. Femtosecond lasers offer an additional advantage resulting from the well defined ablation threshold typical for many materials. For a tightly focused laser beam, outside a certain diameter the energy density is below the ablation threshold, and therefore, ablation occurs only in the central part of the focused region, which may "nominally" be smaller than the diffraction limit [5]. This sub-micron structuring is not possible with longer pulses, where the heat diffusion during the pulse extends for several microns. For massive material processing on commercial scales, however, one must use a large fraction of the available laser power for faster, more efficient and cost effective processing. Ultrashort femtosecond laser systems are commercial available, delivering around 1 Watt at pulse duration of less than 100 fsec, but these are hardly used in industrial applications. Other types of lasers such as fiber amplified lasers may deliver higher power, but with much longer pulses, (i.e. ~1 psec). Thus, the crucial question "how short is a short pulse" still remains unanswered, and in this work we tried to address this question, and offer a new insight. The experiments were carried out with an amplified regenerative laser system delivering up to 1 mJ of pulse energy (at 1 KHz) around 800 nm. The laser light was mildly focused onto a flat surface of silicon, and single pulse holes were drilled into the surface. The holes were measured by optical and Atomic Force Microscopy, and for each hole the volume removed by the pulse (volume below the surface plane) was extracted from the AFM scan. Figure 1 depicts the results. At low fluence (near threshold) the short pulse is much more efficient in removing material, whereas at high fluence, short AND long pulses are efficient, with a minimum around pulse length of a few picoseconds. 819

14 10

IS

20

?&

iMm wasih Cp8i9o)

Fig. 1. Ablation rate vs. pulse duration near the ablation threshold (left) and high above the threshold (right) In addition to the different ablation rates, the morphology of the drilled hole is also different, as can be seen in the AFM images in figure 2.

0

20

40

60p

60 urn

Fig. 2. Typical AFM pictures for short (60 fsec, left) and long (35 psec, right) pulses high above threshold. For the long pulse melting is observed. In a separate set of experiments on various thin metal films, the effect of the variation in electron-phonon coupling (EPC) strength was measured. The temporal profile of the lattice temperature was extracted and correlated to the morphology of the observed ablation. Since the effect of heat flow away from the focal region is much more critical for smaller dimensions, these experiments were performed with a tight focus and holes of diameter around 1 micron. The experiments show that metals with weak EPC can be molten by a single laser pulse whereas metals with strong EPC are ablated without melting. Figure 3 demonstrates this contrast by the sub-micrometer melting/ablation of Gold and Chromium by a 100 fs laser pulse with 0.15 J/cm2 fluence. The experimental results were compared with numerically simulated two-temperature analysis, which provided a satisfactory understanding of the temporal and spatial heat transfer processes involved[13]. Table 1 gives the values of the delay between the irradiation pulse and the time the lattice reaches its maximal temperature. In summary, both sets of experiments point towards the same conclusion. The two-temperature characteristic behavior of laser ablation by ultrashort pulses leads to reaching consequences. The energy is deposited into the electronic system on a very short time scale (on the time scale of the exciting pulse), but the material response of melting, material flow and removal, sputtering and all other mechanical phenomena that are part of the ablation process occur only after sufficiently long delays. The typical electrpn-lattice equilibration times are of order of a few picoseconds, and thus, while short pulses offer the advantages

820

Material Ag Au Al Cu W Cr Fig.3. Comparison of single pulse ablation of thin Gold and Chromium lavers.

Delay 70 ps 56 ps 54 ps 23 ps 13 ps 9,8 ps

Table 1. Delay time between the nulse and maximum lattice

discussed above, it is not apriori obvious what pulse duration qualifies as short. Different metals exhibit very different delay times, which are the result of different electron-phonon coupling strengths, and the observed ablated hole morphologies are correspondingly different. Thus, for the ablation of noble metals, even pulses much longer than 100 fsec may still give the same results as the much shorter pulses, with the potential practical advantages in terms of power, handling and system cost. It is also shown that longer pulses may be as efficient for material removal, depending on competition with other processes occurring on the same time scale as the electron-phonon equilibration time. The present results appear to indicate that specific tailoring of pulse duration to the desired goals of a given ablation experiment might be useful[14], and the optimal shape and duration may be searched by means of femtosecond pulse shapers and self-learning algorithms if proper diagnostic tools can be developed. For that purpose, we studied the emission from the ablation plasma, and found that the normalized intensity of the Silicon I atomic line at 288 nm [11] is proportional to the ablated volume over a large range of pulse durations and intensities, and may therefore be used as our online spectroscopic signature for the ablation volume. Details of these experiments will be provided elsewhere [12]. Support from the BMBF is gratefully acknowledged.

References 1 2 3 4 5 6 7 8 9 10 11 12 13 14

C. Momma, B.N. Chichkov, S. Nolte et al., Opt.Commun.129,134 (1996) B.C. Stuart, M.D. Feit, S. Herman et al., J.Opt.Soc.Am.B 13, 459 (1996). P. Simon and J. Thiemann, Appl.Phys.A 63, 505 (1996). W. Kautek, J. Kreger, M. Lenzner et al., Appl.Phys.Lett. 69, 3146 (1996). S. Nolte, C. Momma, H. Jacobs, et al., J.Opt.Soc.Am.B, 14, 2716 (1997). Laser Ablation, Proceedings of the 5th International Conference, Eds. J. S. Horwitz, H.-U. Krebs, K. Murakami; M. Stuke, Appl. Phys. A, 69, [Suppl], 1999. P. P. Pronko, S.K. Dutta, J. Squier et al., Opt.Commun.114,106 (1995). F. Korte, J. Serbin, J. Koch et al., App. Phys. A, 77, 229 (2003). F. Korte, S. Nolte, B.N. Chichkov et al., App. Phys. A, 69, S7 (1999). A.P. Kanavin, I.V. Smetanin, V.A. Isakov et al., Phys, Rev. B57,14698 (1998). A. Durandet, C.A. Davis and R.W. Boswell, Appl.Phys.Lett. 70,1814 (1997). Y. Prior, K. Zhang, V. Batenkov et al., SPIE, proceedings of the HPLA 2004 conference. In press; and to be published.. S.I. Anisimov und B.S. Lukyanchuk, Physics-Uspekhi 45, 293 (2002) R. Stoian, M.Boyle, A.Thoss et al., Appl.Phys.A 77, 265 (2003). 821

Diode-pumped Cr^^:LiCAF Laser for Ultrahigh Resolution Optical Coherence Tomography Philipp Wagenblast\ Tony H. Ko^ Vikas Sharma\ Uwe Morgner^, James G. Fujimoto\ and Franz X. Kaertner^ ^ Department of Electrical Engineering and Computer Science and Research Laboratory of Electronics, Massachusetts Institute of Technology, 77 Massachusetts Ave., Cambridge, MA 02139, USA E-mail:[email protected] ^ Max-Planck-Institut fiir Kernphysik, Saupfercheckweg 1, 69117 Heidelberg, Germany Abstract. Ultrahigh resolution optical coherence tomography (OCT) is performed with a broadband, Kerr-lens mode-locked, diode-pumped Cr^^:LiCAF laser. The laser source has a bandwidth of 89 nm with an output power of 37 mW. OCT imaging is demonstrated with 3.4 yUm axial resolution in tissue. Optical coherence tomography (OCT) is an emerging technique for highresolution, two-dimensional in vivo biomedical imaging [1]. In order to increase the axial resolution of OCT imaging, broadband mode-locked lasers have been used. Broadband light from a mode-locked Tiisapphire laser has been utilized to demonstrate outstanding axial resolution of 2-3 /urn in ophthalmic imaging [2]. Lasers that are directly diode-pumped are promising candidates for widespread use of OCT imaging because of the reduced cost and complexity of the pump laser as compared to the frequency-doubled pump sources required in Tiisapphire systems. A Cr"^'^:LiCAF laser has been demonstrated to generate broad spectra with bandwidth exceeding 150 nm and output powers on the order of tens of mW [3]. However, the spectrum of this laser exhibits relatively strong modulation due to non-uniformity in dispersion compensation. For OCT applications, a smooth spectral shape is indispensable to avoid pedestals in the interferometric point spread function. Since the spectral shape of the broadband Cr'^"^:LiCAF laser is predominantly influenced by dispersion characteristics, we accomplish spectral shaping by using a combination of double-chirped mirrors to achieve an overall smooth dispersion within one round-trip in the resonator, and a significant reduction of spectral modulation as compared to the laser of Ref. 3. The laser setup is shown in Fig. 1. The different mirrors have dispersion oscillations that are out of phase, as can be seen in Fig. 2(a). With a combination of 3:1 reflections on the two different mirror sets, cancellation of dispersion oscillations can be achieved between 750 and 900 nm. A combination of 5:2 reflections in the laser setup results in sufficiently flat dispersion characteristics so that broadband emission is obtained with a smooth mode-locked spectrum. The mode-locked spectrum in Fig. 2(b) is centered at 815 nm and has a modulation-free shape with a full-width half-maximum bandwidth of 89 nm. The output power of the mode-locked laser is 37 mW. In order to demonstrate imaging of the human retina, an ultrahigh resolution ophthalmic OCT system was used. The interferometric point spread function of

822

Fig. 1. Schematic of the laser setup. the ophthalmic imaging system, shown in Fig. 2(c) and (d), is determined by a calibrated measurement of the reflection from a mirror in the sample arm after attenuation by 60 dB. The resolution is 4.5 jum in air, which corresponds to 3.4 jum in tissue. A sensitivity of 95 dB was obtained at an electronic detection bandwidth of 170 kHz, an axial scanning velocity of 410 mm/s, and a Doppler frequency of 1 MHz. — A

(a)

1(b)750

800

B

3A + B 1 • ( c )

850

I

-5

900

1

. \l,M—4.5

\im

3 I 0 Q.

1

0

5

V - i — 8 9 nm

750

800 850 wavelength [nm]

900

-100

-50

0 50 position [|jnn]

Fig. 2. a) Dispersion characteristics of the double-chirped mirrors, b) mode-locked spectrum of the laser, c) Interferometric point spread function of the imaging system in air after 60 dB attenuation, d) same as c) but demodulated and logarithmic. The resolution is 4.5 ^m in air, which corresponds to 3.4//m in biological tissue. OCT imaging was performed in the macular region of the retina. In Fig. 3 (a), an ultrahigh resolution OCT image of the retina obtained with the Cr^'^iLiCAF source is shown. For comparison, the retina of the same healthy subject is imaged at the same location with a commercial OCT imaging system (Fig. 3 (b)). By comparing the ultrahigh resolution OCT images obtained with the Cr"^'^:LiCAF source to the standard resolution OCT images obtained from the commercial system with a SLD light source, it can be seen that the Cr^'^-.LiCAF source produces OCT images with much better resolution because of its higher source bandwidth. Retinal structures and retinal layer boundaries are much better delineated. With a resolution of 3.4 jxm and a sensitivity of 95 dB, the ultrahigh

823

resolution imaging performance of the Cr^'^:LiCAF light source approaches the performance of Ti:sapphire lasers.

Fig. 1. a) Ultrahigh resolution OCT image taken with the Cr^^:LiCAF laser with -3.4 //m axial resolution, b) Image from commercial OCT system with -10 jum axial resolution. In summary, we have demonstrated ultrahigh resolution OCT imaging with a diode-pumped, broadband Cr^"*":LiCAF laser. Retinal images with an axial image resolution of 3.4 jum in tissue were obtained with a sensitivity of 95 dB. The resolution of this light source is three times better than that of standard commercial OCT systems and is comparable to resolutions achieved with mode-locked Tiisapphire lasers. It has the advantage of direct diode pumping, whereas Tiisapphire lasers are generally pumped by diode-pumped, frequency-doubled solid-state lasers. The advantage of the diode-pumped Cr "^iLiCAF laser is that it considerably reduces the complexity, cost, and power consumption. Acknowledgements. This research is supported by the AFOSR Program F4962001-1-0186 and F49620-01-01-0084, NSF grants ECS-0119452 and BES-0119494, and NIH grants R01-CA75289-06 and R01-EY11289-18.

References 1 D. Huang, E. A. Swanson, C. P. Lin, J. S. Schuman, W, G. Stinson, W. Chang, M. R, Hee, T. Flotte, K.Gregory, C. A Puliafito, J. G. Fujimoto, Science 254, 1178, 1991. 2 W. Drexler, U. Morgner, R. K. Ghanta, F. X. Kaertner, J. S. Schuman, J. G. Fujimoto, Nature Med. 7, 502, 2001. 3 P. Wagenblast, R. Ell, U. Morgner, F. Grawert, F. X, Kaertner, Opt. LeU. 28, 1713,2003.

824

Time-resolved electron imaging of femtosecond laser ablation Yasuaki Okano, Yoichiro Hironaka, Ken-ichi Kondo, and Kazutaka G. Nakamura Materials and Structures Laboratory, Tokyo Institute of Technology, 4259 Nagatsuta, Midori, Yokohama 226-8503, Japan E-mail: [email protected] Abstract. Pulsed electrons generated by intense femtosecond laser irradiation onto a metal target were applied for electron imaging. An instantaneous charge-separated field was observed on a femtosecond-laser-irradiated surface of a copper film. From deflection of the probe electrons, the electric field was estimated to be 1.5 MV/m at a pump-laser intensity of 10^^ W/cm^.

1

Intruduction

T h e interaction of an intense femtosecond laser with m a t t e r has been widely studied in connection with energetic particle generation. In an early process of femtosecond laser ablation, a charge-separated field is formed in front of t h e target surface as energetic electrons are ejected into t h e vacuum faster t h a n ions. This field is considered to be t h e most probable mechanism of fast ion acceleration [1]. In this work we applied laser-driven electron beams for picosecond time-resolved imaging, in order t o investigate the charge-separated electric field in t h e femtosecond-laser ablation process. As electrons are strongly affected by electrical and magnetic fields, the electron flux can be used to visualize the space-charge fleld induced by t h e ablation plasma, in contrast to conventional optical probes.

2

Experimental Methods

T h e experiment was performed using a Ti:sapphire laser system (B. M. industries, alOus, 4 T W ) . T h e amplifled laser beam had a pulse duration of approximately 60 fs and a central wavelength of 790 nm. In a vacuum chamber, t h e laser beam was divided into two beams by a beam splitter. T h e reflected beam was used for generation of a probe pulsed-electron-beam by focusing onto a molybdenum disk with an off-axis parabola at a laser intensity of 3 X 10^^ W / c m ^ . T h e other transmitted beam was passed through an optical delay line, and t h e n t h e beam (energy of approximately 1 m J/pulse) was focused onto a copper fllm to p-polarization at an intensity of 10^^ W / c m ^ . Distances from t h e electron source to t h e sample target and t h e detector were 53 m m and 192 m m respectively, with a resulting magniflcation of approximately 3.6 for imaging.

825

11

o

1.5

>

1.0

1

1

1

r

r *^^/^

0.5 r

. AE=280keV(FWHM) 1

>v

!1

1

1|

(b)

•]

! 2

0.0

J—

200

1

1

1

400

600

800

Electron energy (keV)

Velocity (10 m/s)

Fig. 1. Typical features of a probe electron beam for the electron imaging, (a) Signal intensity as a function of probe-electron energy, (b) Velocity distribution of the electron probe beam derived from the energy distribution. The dotted line represents extrapolated values.

3

Characteristic of p r o b e electron pulse

Electron image was recorded on an imaging plate (IP) (Fuji P h o t o Film, FDL-UR-V) with a light shielding [2]. Figure 1(a) shows an energy spect r u m of t h e probe electrons obtained by a magnetic-deflector-type electron spectrometer with the detector [3]. T h e distribution of electrons has a broad spectrum and spread around 170 keV, with a full-width at half-maximum ( F W H M ) energy bandwidth of 280 keV. This spectrum depends on t h e I P efficiency, and the light shielding also limits t r a n s m i t t a n c e of electrons for low-energy electrons less t h a n 90 keV. This spread of t h e electron energy affects a temporal resolution of the time-resolved measurement due to t h e resulting velocity distribution. T h e electron beam also becomes an energychirped pulse as a result of t h e range of flight times. T h e pulsewidth at t h e target position is estimated to be 64 ps F W H M by considering the velocity distribution (see Fig, 2(c)).

4

Electron images of femtosecond-laser plasma

Figure 2(a) shows a typical time-resolved image of t h e femtosecond laser plasma. Typical features of "dark" and "bright" parts were clearly obtained. T h e "dark" part is considered to be a shadow of the ions or the plasma, depending on their density. T h e "bright" part is due to deflection of t h e probe electrons in t h e charge-separated field, and constructed by focused of electrons. Figure 2(b) shows an image plot of signal intensities along a line normal to t h e target, as a function of t h e relative delay time. T h e deflected signal of t h e "bright" part showed a decrease with increasing t h e relative

826

Position at target (mm)

(a)

EQ (keV) 800

|Pump pulse Copper filmy

Scale at detector:

-| Q mm

Intensity: 1.2 i i i ^ i M i 0.4

0 2 4 6 8 Position at detector (mm) (^Q9 c(f^nts/ps)

Fig. 2. Time-resolved electron image of a femtosecond laser plasma for relative delay time. Time zero was defined as the time when an ablation image appeared. (a) Typical electron image obtained by one laser pulse. The contribution of xrays, which were ejected coincidentally with electrons from the souce, was less than 2% of the intensity, (b) Image plot of signal intensities along the distance from the target normal as a function of relative delay time. The dashed line represents the calculated position of the deflected electron signal at an electric field of 1.5 MV/m. (c) Temporal distribution of the probe electron intensity (solid line) and corresponding electron energy (dotted line), adjusted to the relative delay time .

delay time. Figure 2(c) shows t h e temporal distribution of t h e electron probe adjusted to t h e relative delay time. T h e duration of t h e electron signal corresponds to t h e pulse-width of t h e probe electrons as t h e duration of t h e electric field is much shorter t h a n t h e duration of probe electrons. T h e decay of t h e "bright" signal can be explained as follows: At an early delay time, low energy electrons are deflected by t h e electric fleld. W i t h increasing delay time, t h e effective electron energy of t h e probe pulse increases because of its chirped property. At t h e same time t h e deflection angle, corresponding to t h e detected position of t h e "bright" signal, decreases because of an increase in t h e probe electron energy. This feature was verified by calculating t h e expected "bright" positions of t h e electron signal; this is shown in Fig. 2(b) as a dashed line. This result agrees well with t h e experimantal electron signals, and t h e electric field is estimated to be 1.5 M V / m .

References 1. P. Mora, Phys. Rev. Lett. 90, 185002 (2003). 2. Y. Okano, Y. Hironaka, K. G. Nakamura, and K. Kondo, Appl. Phys. Lett. 83, 1536 (2003). 3. Y. Okano, Y. Hironaka, K. G. Nakamura, K. Kondo, Y. Oishi, T. Nayuki, and K. Nemoto, J. Appl. Phys. 95, 2278 (2004).

827

Part XIII

Frequency Stabilization and Ultrawide Frequency Comb

A new ultrastable cesium optical atomic clock with a 9.1926-GHz regeneratively mode-locked fiber laser Masatsugu Yakabe^ Ko Nito\ Masato Yoshida^ Masataka Nakazawa\Yasuki Koga^, Ken Hagimoto^, and Takeshi Ikegami^ ^ Research Institute of Electrical Communication, Tohoku University, 2-1-1 Katahira, Aoba-ku, Sendai, 980-8577 Japan E-mail: [email protected] ^ National Institute of Advanced Industrial Science and Technology, 1-1-4 Umezono Tsukuba, Ibaraki, 305-8563 Japan Abstract. A new Cs atomic clock is demonstrated by using a mode-locked fiber laser. Frequency stabilities of 4.8x10"^^ for T=1 S and 6.3x10'^^ for x=50 s are obtained when using a Cs beam cavity as long as 1-m.

1. Introduction By using a regeneratively mode-locked fiber lasers (MLFLs) [1], a new type of Cs optical atomic clock can be realized that has neither multipliers nor a quartz oscillator [2]. The Cs optical atomic clock can achieve a better performance because the short-term stability of the MLFLs is superior to that of quartz oscillator and the noise generated at multipliers is not added on the microwave signal. In addition, the obtained frequency standard signal can be delivered throughout the world via optical network. In this paper, we show the measurement of the frequency stabilities of the Cs optical atomic clock by using a 1-m long Cs beam cavity.

2. Principle of Cs optical atomic clock The configuration of our new Cs optical atomic clock is shown in Fig. 1. The laser consisted of a 15-m polarization-maintaining erbium-doped fiber, a wavelengthdivision multiplexing coupler for pumping the erbium fiber with lA8-jum laser diodes, a 200-m polarization-maintaining dispersion-shifted fiber, a 10 % output coupler, a polarization-dependent isolator, a lithium niobate (LN) modulator, and an optical filter with a 5 nm bandwidth. To obtain the sinusoidal harmonic beat signal between the longitudinal laser modes, we coupled part of the output beam to a clock extraction circuit. We then amplified the beat signal and fed it back to the modulator to allow regenerative mode-locking. Part of the fiber cavity was wound onto a PZT to allow us to change the cavity length, therefore this laser operates as an opto-microwave oscillator whose frequency is tunable by controlling the voltage applied to the PZT.

831

CsBEAM TUBE

•^^A

NEGATIVE FEEDBACK CIRCUIT

PHASE I DISCRIMINATOR!

|l.48nmLD|

7^—L-2

OSCILLATOR 1 FOR PHASE I MODULATOR j

£ , I FILTER I MICROWAVE PHASE MODULATOR

I

PM-EDF

INTENSITY I MODULATOR!

PHASE CONTROLLE

ISOLATOR

5t

CLOCK EXTRACTIOI CIRCUIT DIVIDING CIRCUIT

1 ^

1 sec OUTPUT

Fig. 1. A new Cs atomic clock with a 9.1926-GHz regeneratively mode-locked fiber laser. The 9.1926-GHz opto-microwave oscillator signal is phase-modulated at 137 Hz to carry out the phase-sensitive detection. After that, the signal is injected into a Cs beam tube in which we use the Ramsey fringe to detect the frequency deviation [3,4]. Phase-sensitive detection is achieved by the use of a lock-in amplifier, in which a frequency deviation from the Cs resonance can be converted into a voltage-error signal. The error signal is negatively fed back to the PZT of the fiber laser cavity after passing through proportional and integral circuits.

3. Experimental results Figure 2 shows the Ramsey fringe patterns of the Cs resonance and its first derivative signal, which are obtained for the first time by using an MLFL. Until now an electrical microwave synthesizer has been used to observe these curves. This can be accomplished by applying a voltage to the PZT, as shown in Fig. 2(a) below. Figure 2(b) shows the first derivative signal of the Ramsey fringe obtained with phase sensitive detection, where frequency modulation is applied to the microwave signal.

E

CO

Time [s] 0.5 s/div

10 s/div

(a) Fig. 2. Characteristics of a Cs beam tube observed with a MLFL. (a) Ramsey fringe pattern of the Cs resonance, (b) First-derivative signal of the Ramsey fringe shown in Fig. 2 (a).

832

10"^' —•—Cavity length 1 m "-•-•Cavity length 0.25 m

L

©

1

T '''°'•••

-

>

"**»^

"•

•

10"'' 1

10

100

Integration time [s]

(b) Fig. 3. Output characteristics of Cs optical atomic clock, (a) Frequency fluctuation, (b) Allan variance. Figure 3 shows the frequency stability characteristics of the Cs optical atomic clock. Figure 3(a) shows the way in which the repetition rate fluctuates with time when the feedback control is on. The upper and lower results were obtained with commercially available and optically pumped Cs beam tubes, respectively [5]. The commercially Cs beam tube has a 0.25 m cavity and its central fringe width is 394 Hz. The optically pumped Cs beam tube has a long cavity (1 m) and the obtained central fringe width is 110 Hz. It is clearly seen that the fluctuation in the laser repetition frequency is less than 1 Hz in both cases over a long period. Figure 3(b) shows the Allan variance estimated from the frequency deviation [6]. The Allan variance was 2.6x10"^^ for T= 1 s with the commercial Cs beam tube. However, it was greatly improved to 4.8x10'^^ for T= 1 s with the optically pumped Cs beam tube. The stability was improved five times when we adopted the Cs tube with a long cavity. For T = 5 0 S, the oscillator stability reached as high as 6.3x10"^^.

4. Conclusion We proposed a new Cs atomic clock with an opto-microwave oscillator, namely a 9.1926-GHz MLFL. The Allan variance stability thus obtained was as high as 4.8x10"^^ for T= 1 s and 6.3x10^^^ for T= 50 s.

References 1 2 3 4 5

M. Nakazawa, E. Yoshida, and Y. Kimura, Electron. Lett., 30,1603 (1994). M. Nakazawa, K. Suzuki, Opt. Lett., 26, 635 (2001). H. Lyons, Ann. N.Y. Acad. Sci., 55, 831 (1952). N. F. Ramsey, Phys. Rev., 78, 695 (1950). S. Ohshima, Y. Nakadan, T. Ikegami, and Y. Koga, IEEE Trans. Instrum. Meas., 40,1003 (1991). 6 D. W. Allan, Proc. IEEE., 54, 221 (1966).

833

Frequency transfer of optical standards through a fiber network using 1550-nm mode-locked sources Kevin W. Holman\ David J. Jones^, R. Jason Jones^ and Jun Ye^ ^ JILA, NIST and University of Colorado, 440 UCB, Boulder, CO 80309, USA E-mail: [email protected] ^ Dept. of Physics and Astronomy, University of British Columbia, 6224 Agricultural Road, Vancouver, B.C. V6T 1Z1, Canada Abstract. A 1.5-nm mode-locked laser source phase locked to an optical atomic clock is used to transfer precise optical / radio frequency signals over a fiber network with ultrahigh stability. Active stabilization of the transfer medium is demonstrated as well.

With recent advances in the development of optical atomic clocks with superior short-term frequency stability [1], the transfer of signals linked to these clock / frequency standards over appreciable distances with minimal loss of stability has become an important research subject. A transfer process which is capable of maintaining the superior stability of these optical frequency standards has many applications, including the comparison of optical frequency standards developed in remote laboratories. The comparisons of frequency standards derived from different atomic species would enable tests of fundamental physical laws. A highfidelity transfer medium would also allow synchronization of remote lasers, which would be useful for coordinating components of an experiment conducted over large distances, such as in a linear accelerator. Optical fiber networks are ideal transfer media due to their flexibility of use and wide availability. However, the stability of an optical clock based on a particular optical frequency standard must first be transferred to a 1.5-jLim laser source for fiber transmission. Using a 1.5-|im mode-locked source allows simultaneous transfer of optical and radio frequency (RF) signals, both phase locked to the optical frequency standard. Optical transfer is realized by detecting the absolute positions of the frequency-domain comb lines of the transferred pulses, whereas RF transfer is achieved by detecting the repetition frequency of the transferred laser pulses. For stabilizing a 1.5-iLim source to an optical clock, we have achieved synchronization of repetition rates and coherent phase-locking of optical carriers between a 1.5-|am mode-locked laser diode and a Ti:sapphire (Ti:s) laser [2, 3]. The Ti:s laser is phase locked to the optical frequency standard in the implementation of an optical clock.

834

Of course RF signals can be transferred via direct amplitude modulation on an optical carrier. Here we utilize the phase stabilized 1.5-|Lim mode-locked laser source for simultaneous distribution of optical and radio frequency standards over a 7-km-roundtrip installed fiber network. The transfer instability for the repetition rate of a 1.5-|im mode-locked fiber laser is determined by comparing the pulse rate detected after transmission through a roundtrip of the fiber network with that of the pulses before transmission. To minimize instability introduced during detection, it is critical to minimize the light power incident on the photodetector while maintaining a sufficient signal-to-noise ratio (SNR). Figure 1. shows the Allan deviation, a measure of the fractional frequency instability, for the transfer of the repetition rate of the mode-locked laser. The measured instability is below 3 x 10" ^"^ at one second, comparable to optical carrier-transfer of a narrow-linewidth cw laser [4]. RF transfer with mode-locked pulses provides an order of magnitude lower instability than that achieved through modulation on an optical carrier [5]. # - Modulated CW over fiber network e - CW optical transfer over fiber network A Mode-locked fiber laser over fiber network

4 m of fiber (noise floor of measurement system) 10

-t 5 6 7 89

10

Averaging Time (s)

I tII 7 89

100

Fig. 1. Allan deviation showing instability of RF transfer of mode-locked fiber laser's repetition frequency, along with that for modulated optical carrier and cw optical carrier transfers. Measurement with a 4-m fiber represents measurement system's noise floor. The use of mode-locked pulses for RF transfer over optical fibers requires the minimization of dispersive effects on the pulses. The resultant temporal stretching reduces the SNR of the RF harmonic recovered from the photodetection. One method for reducing the effects of dispersion is to reduce the spectral bandwidth of the transmitted pulses, which is used to obtain the results in Figure 1 for pulse transfer. Another approach is to compress the pulses after transmission. This approach can be investigated by substituting the installed dispersive fiber with dispersion shifted fiber (DSF), which exhibits negligible dispersion at the wavelength of operation. Alternatively, dispersion compensation fiber (DCF) can be used to recompress the pulses after passing through the installed dispersive fiber. Both methods have been successfully used to recover a high SNR for the transferred RF signal with a relatively low level of incident light power. The phase error between the RF signal derived from pulses after a roundtrip and that obtained from pulses prior to transmission can be used to actively stabilize fluctuations in the transmission network. Reducing the level of incident light

835

power on the photodetector insures the detected fluctuations represent noise introduced by the transmission medium and not in the photodetection process. Figure 2. shows the instability for RF transfer over 4 km of DSF can be reduced to the measurement noise floor with the use of active stabilization. A retro-reflecting mirror actuated by a long-range, low-bandwidth (100 Hz) shaker is used to accomplish this stabilization.

c o • >

c (0

100

Averaging Time (s) Fig. 2. Allan deviation showing transfer instability over DSF of mode-locked laser's repetition frequency, both with and without active stabilization. For reference, instability for modulated optical carrier transfer and system noise floor also shown. A natural extension of this work in the time domain is to minimize the rms timing jitter introduced during the transfer process. This will be an important step toward exploration of tight synchronization of remotely located pulsed lasers and radio frequency sources. Though for the purpose of frequency transfer the existing stabilization loop has sufficient bandwidth, increasing this bandwidth will need to be explored for reducing the total integrated timing jitter.

References 1 J. Ye, L.-S. Ma, and J. L. Hall, Phys. Rev. Lett. 87,270801 (2001). 2 D. J. Jones, K. W. Holman, M. Notcutt, J. Ye, J. Chandalia, L. A. Jiang, E. P. Ippen, and H. Yokoyama, Opt. Lett. 28, 813 (2003). 3 K. W. Holman, D. J. Jones, J. Ye, and E. P. Ippen, Opt. Lett. 28,2405 (2003). 4 J. Ye, J.-L. Peng, R. J. Jones, K. W. Holman, J. L. Hall, D. J. Jones, S. A. Diddams, J. Kitching, S. Bize, J. C. Bergquist, L. W. Hollberg, L. Robertsson, and L.-S. Ma, J. Opt. Soc. Am. B 20, 1459 (2003). 5 K. W. Holman, D. J. Jones, D. D. Hudson, and J. Ye, Opt. Lett. 29, 1554 (2004).

836

Femtosecond laser optical frequency synthesizers with uncertainty at the 10"^^ level Long-Sheng Ma^^, Zhiyi Bi^, Albrecht Bartels"^, Lennart Robertsson\ Massimo Zucco', Robert Windeler^ Guido Wilpers^ Chris Oates^ Leo Hollberg^ Scott Diddams"^ ' Bureau International des Poids et Mesures, Pavilion de Breteuil, 92312 Sevres, FRANCE ^ Physics Department, East China Normal University, Shanghai 200062, CHINA ^ OPS Laboratories, 700 Mountain Avenue, Murray Hill, New Jersey 07974, USA "^ Time and Frequency Division, National Institute of Standards and Technology, 325 Broadway, M.S. 847, Boulder CO 80305

Abstract. We verify the reproducibility of femtosecond laser optical frequency synthesizers that employ microstructure fibers with those that directly generate a broadband output. No limitation of either system is found at fractional frequency levels approaching 1 x 10'^^.

A femtosecond laser optical frequency synthesizer [1-3] generates a broadband comb of optical frequencies that can be phase-coherently referenced to an optical or microwave frequency standard (Fig 1). Such synthesizers have found widespread use in optical frequency metrology [4] and emerging optical atomic clocks [5,6]. They are anticipated to play an increasingly important role in laboratorybased tests of symmetries in physics, searches for possible time-variations of fundamental constants [7], and the coherent control of ultrafast pulses [8]. In this context, it is important to investigate the potential limitations of different types of femtosecond laser optical frequency synthesizers. When referenced to an optical frequency standard, we demonstrate that the relative frequency uncertainty in the output comb elements from such a synthesizer is near 1x10"^^. The reproducibility of this performance is verified by comparison of four synthesizers of different construction from three laboratories. The optical synthesizers and control techniques we employ have been described in detail previously [5,9-12]. Notably, we compare synthesizers that employ nonlinear microstructured optical fibers (Fig. 1(b)) [5,9,12] with synthesizers that directly emit a broadband spectrum (Fig. 1(c)) [10,11]. Each synthesizer employs a self-referencing scheme [2] to measure and phase-lock its offset frequency j ^ . A cavity-stabilized diode laser at^ref ^456 THz (657 nm) is heterodyned with mode no of the synthesizer, and the control of the resuhing beat^f, is used to fix the mode spacing (i.e. repetition rate) to be /.=(/./-A-/o)/no

(1)

The k* output mode of the synthesizer relative to mode no is then given by

837

Jk ~ Jref

(a)

Jh- '

\fr,-h-h)

Phase Locks Femtosecond Laser Synthesizer

532 nm

V

^ith

^ = 0,1,2

(2)

k

P VA

f.

Ti; Sapphire Gain

Fig. 1. (a) Schematic of the femtosecond laser synthesizers, (b) 1 GHz Ti:sapphire femtosecond laser that is spectrally broadened to more than an octave in nonlinear microstructure fiber [5,9,12], (c) 1 GHz Ti:sapphire femtosecond laser that directly emits a broad spectrum [10].

The basic scheme of our measurements is a comparison of pairs of femtosecond laser synthesizers that are both phase-locked to/ef- We verify Eq. (1) using two methods: photodetection of/ from each laser ifollowed by electronic mixing, or optical nonlinear cross-correlation. In each case the difference mf^ from the two lasers is measured and compared to the value predicted by Eq. (1). In a similar fashion, we verify Eq. (2) using optical heterodyne techniques [13]. Four femtosecond laser synthesizers (named BIPM-C2, ECNU-Cl, NIST-BBl, NISTBB2) were involved in these comparisons over a period of ~9 months. NIST-BBl and -BB2 are broadband lasers (Fig. 1(c)), while BIPM-C2 and ECNU-Cl employ microstructure fiber (Fig. 1(b)). A summary of these frequency comparisons is given in Table 1. In all cases, the difference between the measurements and the values predicted by Eqs. (1) and (2) is consistent with zero, and the relative uncertainty is reported with a confidence level of 95%.

838

The lowest uncertainty (~1 x 10"^^) is achieved with exceptionally low shortterm instabilities (2.3 x 10"^^ at 1 s of averaging). This requires that the path length fluctuations from the two synthesizers to the heterodyne photodetector be common mode at an approximate level of < 10 nm (~ 20 attoseconds) in 1 s of averaging. Detection off^ results in larger uncertainties, principally because the short term instability is higher, which subsequently leads to longer required averaging times. Nonetheless, our data do not point to the existence of any fundamental limitations to the achievable uncertainty at the 10"^^ level. This establishes the femtosecond laser synthesizer as a reliable tool for optical and microwave frequency comparisons and precision measurements in experimental physics. Table 1. Summary of comparison of femtosecond laser frequency synthesizers Systems Compared BIPM-C2, ECNU-Cl, NIST-BB1,NIST-BB2 NIST-BB1,NIST-BB2 NIST-BB1, ECNU-C1

NIST-BBl, ECNU-Cl

Equation Tested 2

Compared Frequency(ies) 333-456THz

2 456-473THz 1 (photodetection and electronic 995 MHz mixing) 1 (optical nonlinear 995 MHz cross-correlation)

Avg. Time 51 943 s

Relative Uncertainty 1.4 x lO'^

21060s

1.3 x IQI^

12 293 s

9.8 x lO'^

8 125 s

2.7 x lO'^

The work at NIST was funded in part by NASA. The project at ECNU was ftmded in part by NSF of China (10274020 and 60490280) and SMEC in Shanghai

References 1. Th. Udem, J. Reichert, R. Holzwarth, T. W. Hansch, Phys. Rev. Lett. 82, 3568 (1999). 2. D. J. Jones, et al., Science 228, 635 (2000). 3. R. Holzwarth, et at., Phys. Rev. Lett. 85, 2264 (2000). 4. T. Udem, R. Holzwarth, T. W. Hansch, Nature, 416, 233 (2002). 5. S. A. Diddams, et al., Science 293, 825 (2001). 6. J. Ye, L.-S. Ma, J. Hall, Phys. Rev. Lett. 87, 270801 (2001). 7. S. Bize, et al., Phys. Rev. Lett. 90, 150802 (2003). 8. R.K. Shelton, et al., Science 293, 1286 (2001). 9. A. Bartels, T. Dekorsy, H. Kurz, Opt. Lett. 24, 996 (1999). 10. A. Bartels, H. Kurz, Opt. Lett. 27, 1839 (2002). 11. T. M. Ramond, S. A. Diddams, L. Hollberg, A. Bartels, Opt. Lett. 27, 1842 (2002). 12. J. K. Ranka, R. S. Windeler, A. J. Stentz, Opt. Lett. 25, 25 (2000). 13. L.-S. Ma, et al.. Science 303, 1843 (2004).

839

Femtosecond Laser Frequency Combs with linewidths at the 1-Hz Level* A. Bartels\ S.A. Diddams^ C.W. Oates^ J. C. Bergquist^ L. Hollberg^ ^ Time and Frequency Division, National Institute of Standards and Technology, 325 Broadway, M.S. 847, Boulder CO 80305, USA Abstract. Femtosecond laser frequency combs (FLFCs) can be phase-locked with subHertz linewidth and a coherence time of 5 s relative to a low-noise reference laser. Thus we are able to detect high contrast spectral interferograms at up to 10 s integration time between two FLFCs sharing a common optical reference. We also establish an upper limit of 4 Hz on the absolute linewidth of the FLFC components. Visible continuous-wave (CW) lasers v^ith sub-hertz linew^idth are essential parts of optical atomic frequency standards v^ith projected fractional frequency uncertainties of wlO"^^ [1-3]. When locked to the narrov^ atomic clock transition such lasers serve as the local oscillator of the frequency standard and provide a stable optical frequency output. A means to transfer the lov^ noise properties of the optical standard to other optical frequencies or the microw^ave domain is important for applications, such as optical frequency metrology, spectroscopy, or the generation of highly stable microwave signals. This task can be conveniently solved with a FLFC that operates as an extremely broadband optical frequency synthesizer. Such systems are capable of synthesizing optical and microwave frequencies with instabilites of a few parts in 10^^ or below in 1 s of averaging [4, 5]. While these previous experiments have shown how an FLFC might compromise the stability of a reference laser, in this paper we directly address the question of what the linewidth of an FLFC can be relative to the reference laser by comparing two systems that are locked to a common reference. Furthermore, we establish a limit on the absolute linewidth of the FLFC components. We employ two femtosecond lasers (referred to with indices i=l,2) that emit a broadband continuum at 1 GHz repetition rate [6]. They are phase-locked to a common cavity-stabiHzed laser diode at 657 nm (fLD=456 THz) as has been described in references [5] and [7]. In this case, the repetition rates are given by fR,i=(fLD-fb,i-fo,iyni and the frequency comb components are fki,i=fLD+ki/niX(fLD-fbifo,i), with ki=0, ±l,±2,.... Subsequently, we set the comb spacings to be equal by choosing ni=n2 and fo,i+fb,i=fo,2+fb,2. With this scheme, the residual noise on the laser diode largely cancels out in the following intercomparison experiments which are rigorous tests of the excess phase-noise introduced to the frequency combs by the femtosecond lasers. The average linewidth of a group of comb components is measured around 900 nm by detecting a heterodyne beat signal between the FLFCs after a filter that transmits the IR parts of the combs. This beat appears at frequency Afo=fo,i-fo,2 when the pulses from both lasers are temporally overlapped on the photodetector. * Contribution of an agency of the U.S. government; not subject to copyright 840

Fig. la shows the RF spectrum of this optical beat taken at 1 kHz resolution bandwidth (RBW) with a RBW-limited peak at the carrier frequency and a broad noise pedestal. The inset of Fig. la was taken with 1 Hz RBW and again shows an RBW-limited carrier and a number of lines that result from mechanical resonances. Spectra with higher resolution were obtained with a FFT-spectrum analyzer (see Fig. lb). With the RBW of 3 mHz, we see a linewidth of «23 mHz and the carrier still containing «59% of the total RF-power.

-400

-200

0

200

400

offset from Af^ (kHz)

frequency (Hz) -1

0

1

offset from Af^ (Hz)

Fig. 1. (a) RF-spectra of the Afo-beat signal measured with an RF spectrum analyzer, (b) measured with an FFT-analyzer. (c) Phase-noise spectrum of the Afo-beat signal. The carrier frequency is 333 THz. A rigorous determination of the coherence time of the FLFCs relative to the reference laser comes from a phase-noise measurement of the Afo-beat. The phase noise spectrum S<{)(f) is shown in Fig. Ic. It shows a 1/f^-behavior up to 10 Hz before a multitude of acoustic resonances between 10 Hz and a few ten kilohertz appear. The peaks around 50 kHz and 100 kHz contain contributions of the feedback loop resonances. The integrated phase-noise (|)int. (f) (see Fig. Ic) reaches 1 rad at f=0.2 mHz from which we infer a coherence time of 5 s. The present limitation to the measured coherence time is not of fundamental nature but due to mechanical vibrations and air currents in our setup. The high mutual coherence between the FLFCs is confirmed by a measurement of high contrast spectral interferograms (Sis) between them at wavelengths around 850 nm [8]. Fig. 2a shows a SI recorded on a CCD-array after a grating spectrometer with an integration time of Tint=1 s. The contrast of the SI is 65% for the data of Fig. 2a. Fig. 2b shows a SI with Tint.=10 s. Although the contrast has decreased to C=45% and despite the higher noise due to background Hght on the CCD-array, interference fringes are still clearly visible. A detailed analysis has shown that the observed contrasts can be quantitavely explained when one accounts for phase noise of the FLFCs and the spectrometer resolution [9]. While the previous measurements were relative to a reference laser, we have also investigated the absolute linewidth of the components of a single FLFC by locking it to a highly-stabilized CW-laser at 563 nm and beating one mode of it

841

against a second CW-laser at 657 nm. This beat signal is shown in Fig. 2c. The observed linewidth of 4 Hz was entirely limited by the slow drift of the reference cavity used with the 657 nm laser.

820

840

860

880

wavelength (nm)

I/AAAJ -50

0

frequency offset from ^^-L^, : 820

840

860

50

(Hz)

880

wavelength (nm)

Fig. 2. (a) and (b) Spectral interferograms between two FLFCs at integration times of Is and 10 s (gray lines: Individual spectra; dashed line: calculated SI, see Ref 9 for details), (c) beat signal between one FLFC locked to a CW-laser at 563 nm and a CW-laser at 657 mn. We have demonstrated that the components of an FLFC can have a linewidth of 23 mHz and a coherence time of 5 s relative to a low-noise reference laser. To confirm this high degree of coherence, we have detected high contrast spectral interferograms between the FLFCs at integration times of up to 10 s. The longterm coherence has been entirely limited by environmental influences on the measurement. We have established an upper limit of 4 Hz on the hnewidth of the components of a single FLFC that is stabilized to a high-quality CW laser. Acknowledgements. We are grateful to R. Fox for helping us with the microwave electronics. This work has been funded in part by NASA.

References 1 2 3 4 5 6 7 8 9

842

Ch. Salomon et al, J. Opt. Soc. Am. B 5, 1576 (1988). B.C. Young et al., Phys. Rev. Lett. 82, 3799 (1999). R.J. Rafac et al., Phys. Rev. Lett. 85, 2463 (2000). S.A. Diddams et al.. Opt. Lett. 27, 58 (2002). A. Bartels et al.. Opt. Lett. 28, 663 (2003). A. Bartels and H. Kurz, Opt. Lett. 27, 1839 (2002). T.M. Ramond et al.. Opt. Lett. 27, 1842 (2002). L. Lepetit et al., J. Opt. Soc. Am. B 12,2467 (1995) and references therein. A. Bartels et al.. Opt. Lett. 29,1081 (2004).

Frequency metrology with a turnkey all-fiber system T. R. Schibli^ K. Minoshima\ F.-L. Hong\ H. Inaba\ A. Onae\ H. Matsumoto^ I. HartP, and M. E. Fermann^ ^ National Institute of Advanced Industrial Science and Technology, AIST, 1-1-1 Umezono, Tsukuba, Ibaraki 305-8563, Japan E-mail: [email protected] ^ MRA America, Inc., 1044 Woodridge Ave., Ann Arbor, MI 48105, USA E-mail: [email protected] Abstract. The repetition-rate and carrier envelope offset frequency of a turnkey, all-fiberbased continuum generator are phase-locked to a highly stable atomic clock, a H-maser. With this source, optical frequency measurements at 1064nm and at 1542nm are demonstrated. The performance of this turnkey system was found to be comparable to a traditional Ti:sapphire-based comb.

1.

Introduction

Current systems for frequency metrology are mainly based on Ti: sapphire lasers, which are prohibitively bulky and greatly restrict applications beyond basic research. The need for a more robust system for metrology recently inspired several groups[l-3] to investigate fiber lasers as a potential alternative. However, the increased noise of such fiber systems provided little hope that fiber lasers could be used for any but the least challenging frequency measurements. In this paper, we demonstrate for the first time optical frequency measurements with a fully turnkey fiber based frequency-comb and verify that it provides an accuracy comparable to a traditional Ti:sapphire-based comb source.

Oscillator

n

c Amplifier

HNLF

SMF28 IF.

Phase locking electronics ^) H-maser

Laser to be measured

1500

2000

2500

Wavelength [nml

Fig. 1. a) Experimental setup of the turnkey frequency measurement system; b) Spectral output of the comb generator.

843

2.

Experimental Setup

The experimental setup is shown in Fig.la). A pulse compression mode-locked Fabry-Perot type Er fiber oscillator in conjunction with an Er fiber amplifier, generates 70fs pulses at an average power of 120mW and a pulse repetition rate frep of 86MHz [4]. The amplified pulses are injected into a spliced 12cm long highly nonlinear fiber (HNLF) to generate an octave-spanning comb (Fig. lb). The carrier envelope offset frequency {fcEo^ Fig.2a) is detected in a fiber coupled, common path f-2f interferometer. Finally, a 2x2 fiber coupler was used to split the continuum into two parts. One part was used to detect the repetition rate and the carrier envelope offset frequency and the other one for frequency metrology experiments and the performance evaluation of the comb. Hereunto, the laser to be measured was directly connected to the remaining input port of the said coupler.

10 20 30 40 50 Frequency [MHz;

4

8

12

16

Frequency [MHz]

Fig. 2. Beat notes: sCjfcEo- FWHM ~200kHz; h)fbeat with Acetylene stabilized diode laser: FWHM -100.. 150kHz, limited by the line width of the stabilized diode laser; c)fbeat with Iodine stabilized Nd:YAG laser: FWHM ~30kHz.

3.

Results and Discussion

To evaluate the performance of this turnkey system, two absolute-opticalfrequency measurements at two different wavelengths (1064nm and 1542nm) were conducted and compared to the results obtained by a Ti: sapphire system. For the measurement in the telecom region we used an Acetylene stabilized laser diode locked on P(16) of ^^CiHi [5]. The beat note between the stabilized laser diode and the comb revealed an S/N ratio of 40dB in a lOkHz bandwidth (Fig.2b). The measured absolute frequency was always within the reproducibility of the stabilized laser and matched well to the previously measured value [5]. The stability of the system is shown in Fig. 3 with a comparison to the Ti:sapphire system. To evaluate the performance of the comb close to its edge we measured the absolute frequency of an Iodine stabilized Nd: YAG laser locked on R(56)320:alO of ^^\ [6].The FWHM of the beat note was 30kHz in a 30kHz bandwidth (Fig.2c). The absolute frequency measured with the fiber and the Ti:sapphire source were within 20Hz (determined from the average of 100 points taken with a 10s gate time) corresponding to a fractional uncertainty of 3.6-10"^'^. The stability of this system is shown in Fig.3.

844

4.

Conclusions

In this paper we demonstrated the usefulness of a novel turnkey all fiber continuum source for high precision frequency metrology. This novel source not only greatly facilitates the measurement of absolute optical frequencies but it also opens a door to long-term measurements such as required for optical atomic or molecular clocks. This system demonstrates a big step towards a frequency measurement system with the ease-of-use of an ordinary wave meter. IO-

CS 1 0 "

10-

10-

3 4 5 6 7

2

3

3 4 5 6 7

10

100

4

5 6 7

1000

Averaging time [s]

Fig. 1. Root Allan variance of the measured beat frequencies between the comb and the stabilized lasers: Acetylene measured with the turnkey fiber system (solid circles) and measured with a Ti:sapphire laser (solid squares). Iodine measured with the turnkey fiber system (open circles) and with a Tiisapphire laser (open squares). Dotted line: stability of the Hydrogen maser (frequency reference); Dash-dotted line: root Allan variance between two Iodine stabilized lasers of the same type.

References 1 B. R. Washburn, et al., Phase-locked, erbium-fiber-laser-based frequency comb in the near infrared,' Opt. Lett, 29, 250 (2004) 2 F. Tauser et al., 'Amplified femtosecond pulses from an Er:fiber system: Nonlinear pulse shortening and self-referencing detection of the carrier-envelope phase evolution,'Opt. Express 11, 594 (2003) 3 F.-L. Hong et al., 'Broad-spectrum frequency comb generation and carrierenvelope offset frequency measurement by second harmonic generation of a mode-locked fiber laser,' Opt. Lett., 28, 1 (2003) 4 M.E. Fermann et al., ^Femtosecond fiber laser,' Electron. Lett., 26, 1737 - 1738 (1990) . 5 Feng-Lei Hong et al., 'Absolute Frequency Measurement of an AcetyleneStabilized Laser at 1542 nm,' Opt. Lett., 28, 2324 (2003) 6 F.-L. Hong et al., 'Frequency reproducibility of an iodine-stabilized Nd:YAG laser at 532 nm,'Opt. Commun. 235, 377 (2004)

845

Optical frequency measurement precision of femtosecond laser optical comb system and the stability of its HF reference frequency Hiroyuki Ito, Ying Li, Miho Fujieda, Michito Imae and Mizuhiko Hosokawa National Institute of Information and Communications Technology E-mail: [email protected] Abstract. Repetition rate of femtosecond frequency comb is controlled by HF frequency standards. We investigate the relation between the stability of HF frequency standards and the optical frequency measurement capability of femtosecond frequency comb.

1

Introduction

Femtosecond laser optical comb is one of great applications of ultrashort pulse laser that brought about a breakthrough on the absolute measurement of the optical frequency [1]. There are two key techniques; one is the control of the repetition rate of the femtosecond laser with the accuracy of 10~^^ or 10~^^ and the other is the widening of the output frequency range up to over one octave for the calibration of the frequency offset. These technologies have made it possible to measure the optical frequency linking with the microwave frequency standards without loss of the accuracy. Femtosecond laser optical comb system open up a way to the absolute comparison of optical transition frequency [2] with the Cs hyper fine transition frequency that is now used for the definition of the second. In National Institute of Information and Communications Technology (NICT), we have commenced the research on the optical region frequency standards and have installed a femtosecond laser frequency comb system (FC8003, Menlo Systems GmbH [3]) for the frequency measurement in this region. The block diagram of this system is shown in Fig. 1. 1 —•

1

lIUJj/v

I

I

t '

£

820MHz ' '

ring T i : saf

Figure 1: Block diagram of FC8003 The performance of the femtosecond laser frequency comb system is really remarkable. Its measurement accuracy depends on many factors, such as the quality of 846

supplied high frequency (HF) standard, phase noise on the phase locking of the repetition rate to the synthesized frequency and phase noise of the photodetector. Here we concentrate on the stability of the HF standard and investigate its influence on the optical frequency measurement precision.

2 Measurement of diode laser system with high frequency stability We proceed with a research and development of an optical frequency standard based on a ^^Ca+ in an RF trap[4]. In order to observe this high Q-factor clock transition, we are developing an ultranarrow linewidth diode laser system. The required reduction of the laser linewidth was achieved by locking the laser to an ultrahigh-finesse ultralow expansion glass (ULE) reference cavity[5]. In order to evaluate the performance of the laser system, we employ a femtosecond (fs) laser frequency comb system (FC8003). Its repetition frequency and offset frequency are stabilized to a lOMHz reference frequency, provided by an H-maser standard (Fig. 1). We measured a beat-note frequency of the diode laser beam and the fs laser frequency comb. The short term stability of the beat-note is 9.6 x 10~^^ for 1 second averaging time. It seems that the result is limited by the stability of the fs laser frequency comb. Therefore, we investigate the stability-degradation factors of the fs laser frequency comb, especially on the stability of HF standards.

3

Reference frequency

NICT is one of the time and frequency standards institutes that participate the construction of International Atomic Time (TAI) whose accuracy is about 2 x 10~^^ and stability is 6 x 10~^^ at 30 days. There we generate and keep a highly accurate timescale that is linked to TAI. As mentioned in the previous section, usually we use an H-maser as the source of the reference frequency of FC8003. There are, however, many other equipment that can supply the reference frequencies, such as a high stability frequency synthesizer for primary frequency standards. These equipments have own stability characteristics. They can be measured easily and precisely by using some commercial high-precision frequency-stability measurement systems [6]. Here we used Timing Solutions 5110A time-interval analyzer for the evaluations. The results are shown in Fig. 3. We have used three types of reference frequency sources, hydrogen maser (KVARZ CHl-75) and synthesizer (NIST HR2-9 and SDI CS-1). All synthesizers use 5MHz frequency reference of hydrogen maser.

4

Experimental results and discussion

We have measured frequency stability of ECDL using fs optical frequency comb with various HF sources as a reference frequency source. In the measurement, we have observed a trend that the precision of optical frequency measurement depends on the stability of supplied HF frequency standard (Fig. 3). We have also measured the influence of DDS (Fig. 1) by using two synthesizer (SRS345 and HP8663A). We cannot, however, observe any clear difference in measured frequency stability. 847

r V»

—-— —•— ...A... -—^—

System H-maser - H-maser |-|_nnaser - NIST Syntesizer H-maser - SDI Syntesizer

1 1 |

—<= HM —n—SDI

—.^—. NIST
X

/'/'

10"

ja 1

10^ x(s)

Figure 2: The optical stabilization scheme of the laser system

Figure 3: The optical stabilization scheme of the laser system

We are planning to measure the frequency stability of multiplied frequency (800MHz) for more precise evaluation of influence of HF source on frequency stability of the comb system. We are also planning to build another independent frequency stabilized 729nm ECDL to measure the line width and the short term frequency stability by observing these beat note to check the frequency stability of ECDL is better than the measured frequency stability by the comb.

References [1] David J. Jones et al., "Carrier-Envelope Phase Control of Femtosecond ModeLocked Lasers and Direct Optical Frequency Synthesis" Science 288, 635639(2000) [2] H. Dehmelt, "Mono-Ion Oscillator as Potential Ultimate Laser Frequency Standard" IEEE Trans. Instrum. Meas., IM- Vol.31, pp.83-87 1982 [3] R. Holzwarth et al., "Optical Frequency Synthesizer for Precision Spectroscopy," PRL 85, 2264-2267(2000) [4] K. Matsubara et al., "Study for a 43Ca+ optical frequency standard," submitted to the Conference on Precision Electromagnetic Measurements 2004 [5] Ying Li et al., "Frequency stabilization of a diode laser for the S-D transition of Ca+ ion," 7th International Symposium on Contemporary Photonics Technology 2004 Technical Digest, pp.95-96 [6] L.Sojdr, J. Cermak, and G. Brida, "Comparison of high-precision frequencystability measurement systems" in International Frequency Control Symposium and PDA Exhibition Jointly with the 17th European Frequency and Time Forum of Proceedings, (Institute of Electrical and Electronics Engineers, New York, 2003),pp.317-325

848

Evaluation of oscillation frequency stability of a diode laser using a fs laser optical comb H. Kobayashi^ T. Nimonji^ and A. Sawamura^ T. Sato^ M. Ohkawa^ and T. Mamyama^ T. Yoshino^ H. Kunimori^ M. Hosokawa^ H. Ito^ Y. Li\ and S. Nagano^ S. Kawamura"^ ^C/0 Sato Lab., Graduate School of Science and Technology Niigata University, Ikarashi 2.no-cho, Niigata 950-2181, JAPAN [email protected]. ac jp faculty of Engineering Niigata University, Ikaraslii 2-no-cho, Niigata 950-2181, JAPAN [email protected]. ac.jp National institute of Information and Communications Technology, 4-2-1 Nukui-Kitamachi, Koganei, Tokyo 184-8795 Japan "^National Astronomical Observatory, 2-21-1 Osawa, Mitaka, Tokyo, 181-8588, Japan Abstract.We have stabilized the oscillation frequency of a diode laser using the Faraday effect of Rb absorption lines. The stabilized laser frequency was measured by means of a femtosecond mode-locked pulse laser optical comb generator.

1. Introduction There is a plan to measure the time varying gravity field for the global observation of the environmental changes, using the laser interferometric technique for satellite-to-satellite tracking instead of using the radio technique as used in the current inter-satellite system, GRACE. The laser interferometric technique w^ill improve the sensitivity of the system by as much as 10 nm/s. We are trying to stabilize their oscillation frequencies of the lasers better than 10"^^ in the square root of the Allan variance of the frequency stability from 1 s to 100 s in the averaging time and are investigating the availability of a diode laser as a light source in a satellite-to-satellite laser interferometer. In recent years, a femtosecond mode-locked pulse laser optical comb generator has been developed as a new reference frequency source for absolute frequency measurement. We have measured the frequency stability of our system using the optical comb and obtained the results of about 10"^^ from I s to 100 s in the averaging time.

2, Femtosecond laser optical comb generator A titanium sapphire laser is used as a femtosecond mode-locked pulse laser of the optical comb generator system. The mode-locked laser emits optical pulse sequences of a fixed time interval in a time region. They are a large number of modes in the frequency range separated by a fixed frequency interval determined by the repetition rate of pulses. The frequency of the n-th order mode in a

849

mode-locked laser can be expressed as: /(n)=/?/;+/, (1) /^is referred to as the "carrier-envelop offset frequency follows: and expressed as

,/;=(A^/2;r)y; A^ is called the "carrier-envelop offset phase". (2) Each modefrequencyof a mode-locked laser is defined by /^ and / , , as shown in Equation L Therefore, the frequency in each mode is determined by these two frequencies stabilized, /^is detected by observing a portion of optical comb frequencies through the beat note, and the frequency /^ is stabilized through its harmonics. As shown in Fig. 1, the spectrum width of the optical comb is extended to more than one octave by introducing its output to a photonic crystal fiber and then f^ can be determined from the beat signal of Eq.3 of the secondary harmonics in the n-th mode and Eq.4 of the basic wave in 2n-th order mode.

2/(«)=2«/; + 2/„

(3)

/(2«)=2«/;+/„ (4) The frequency fi of a laser can be defined as Eq. 5 because /^ and f^ are fixed. fj^^mf, + f,±f, (5) Where /^ is the beat frequency of the comb generator mode and the measured laser frequency. If both the optical comb and the measured laser frequency are known beforehand within /^ / 2 , each oscillation frequency will be defined. The stability and accuracy of the optical comb are limited by the stability and accuracy of the microwave standardfrequencywhich controls f^ and /^ [2].

:(2n) 2\\m f(")

\/.

LiLh

^

fr

I , « , M

fo

I

I Frequency

fo;OfTset frequency fr: Repeat frequency

Fig. 1 Optical comb spectrum

3. Experimental setup

We measured the stability of the laser diode using the Faraday effect-based PEAK method by introducing the optical comb generator. The PEAK method is a frequency stabilization method utilizing an envelope detection principle. This method can produce a large control signal, hereby improving frequency stability. Figure 2 shows the optical system used in our experiment. The laser beam divided by BSl is introduced to an optical comb generator in the National institute of Information and Communications Technology (NiCT) in order to measure its frequency stability. This optical comb generator is controlled by the microwavefrequencystandard systems and provides stability of 4x10"^^ at an averaging time of 1 s and at the order of 10'^^ at 1000 s averaging time.

850

RlK^ Lens ISO BSl LPl

f%H^

Optical Comb

LD; Laser dio
ISO: Optical Isolator B S : Beam splitter ^ : Magnetic field

Fig.2 Optical setup

10

10 10 Averaging Time x (s)

Fig.3 Frequency stabilities

4* Experimeatal result The obtained stability is shown in Fig. 3. The open circle indicates the results we obtained at Niigata University, using the beat frequency between two similarly stabilized lasers. The filled circle shows the results obtained at NiCT by means of the optical comb generator. The stability measured by means of the optical comb generator is better than the stability measured using the beat note in our laboratory (Niigata), by a factor of approximately 2. This may be due to the fact that only one laser frequency variation is measured through the optical comb generator, in 10*^*^ or better stability, in the latter system.

5. Conclusion The absolute stability of a laser diode can be measured about 10"^° from 1 s to 100 s in the averaging time by means of an optical comb generator. Because few laboratories possess optical comb generators, we want to measure stability using both the optical comb generator and the beat note between two similarly stabilized lasers at the same time in order to verify the accuracy of our frequency stability evaluation system using the beat note detection method.

References [1] K. Sugiyama:"Optical frequency measurement with mode-locked lasers". Opt, Vol.31, N a l 2 (2002) 870-876. [2] S. Ito, T. Nimonji, T. Saga, A. Sawamura, T. Sato, M. Ohkawa, T. Maruyama: "Oscillation frequency stabilization of the semiconductor laser using the Faraday effect of Rb atom". Technical Report of lEICE, LQE2002-47,2002-06.

851

Part XIV

Coherent Control and Other Topics

Coherent cooling of molecular vibrational motion with laser-induced dipole forces Hiromichi Niikura, P. B. Corkum, and D. M. Villeneuve National Research Council of Canada 100 Sussex Dr., Ottawa, Ontario Kl A 0R6 Canada [email protected]

Abstract. Applying an intense laser pulse whose duration is shorter than one vibrational period at the appropriate time, we show that a vibrational wave packet can be accelerated, decelerated, or dissociated. 1. Introduction When an intense laser field ('-lO^^W/cm^) is applied to a molecule, electrons in the molecule are moved and a laser-induced dipole force is generated. The dipole force tends to pull the molecule apart against the binding force of the molecule. If the laser pulse is applied to the molecule in a sufficiently short time compared to the vibrational period, then the vibrational motion is influenced, depending on the phase of the vibrational motion. For example, if the molecule is stretching w^hen the laser pulse is applied, then the vibrational motion is accelerated. Conversely, if the molecule reaches the outer turning point when the laser field is still present, then the molecule dissociates. If the molecule is shrinking when the laser pulse is applied, the vibrational motion is decelerated or even stopped. We demonstrate these processes both experimentally and theoretically using a simple molecule, H2 or D2. We suppose that two short laser pulses are used. One creates the wave packet and the other controls it. The first pump pulse removes an electron from H2 (D2) by tunnel ionization, which produces the vibrational wave packet on the D2^2g surface [1]. After a delay time, the control laser pulse lowers the Zg surface by laser-induced coupling between the Sg and lu surfaces. It produces the dipole forces that act to pull the molecule on Eg apart. At a laser intensity of ^lO^'^W/cm^, the dipole force is '^1 eV/A, comparable to the binding forces of H2^(D2^). Experimentally, we use two approaches. One is to use two -^8 fs laser pulses for

855

pump and control. The other uses the signal and idler from an optical parametric amplifier (OPA) to modulate the envelope of a 40 fs laser pulse. Each modulation crest is used for pump or control pulse. 2. Gated dissociation [2] First, we show that the dissociation yield of H2"^ or D2^ can be controlled by modulated laser pulse. We combine two laser pulses with different wavelengths, signal and idler output from an OPA pumped by 800 nm, 40 fs, 800|^J laser system. By tuning the angle of the crystal in OPA, we change both signal and idler wavelengths simultaneously in the range 1550 nm - 1200 nm for the signal, and 1700 nm - 2500 nm for the idler. By overlapping the two pulses with same polarization direction, we modulate the envelope whose half period (defined as a "gate" time) is from 5 fs to 30 fs. The D2 is preferentially ionized at one of the envelope peaks, and forms the vibrational wave packet on the D2^i;g surface. If the laser is present when the wave packet reaches the outer turning point, then the wave packet dissociates through the field-modified potential, producing the D^ fragment. Thus, we expect that changing the modulation period will control the dissociation yield of D2^. We measured the fragmentation yield of D2^ as a function of the gate time. We found that at the gate time of ^11 fs, the dissociation yield is mostly suppressed. For longer or shorter gate time, the yield increases. We compared H2 and D2 to verify that the dissociation depends on the vibrational frequency. The minimum shifts to '^7fs for H2. In both cases, the gate times that give the lowest dissociation yield are approximately half the vibrational period consistent with the qualitative picture that the wave packet is trapped if the laser field is absent when it reaches the outer turning point. The experimental results agree with a wave packet calculation. 3. Cooling vibration motion [3] Using the same model, we show how to improve the controllability by using two short laser pulses. We can influence the internal energy of the molecule. The first pulse launches a wave packet on D2^Ig at t =0, and the control pulse is applied at some time later. With numerical calculations, we find that the wave packet picks up '-0.3 eV of kinetic energy from the laser field when the control pulse is applied at 1/4 of a vibrational period {T^). The laser intensity is 2xlO^'^W/cm^ and the pulse duration is '^ 12 fs. On the other hands, ^0.3 eV is extracted from the molecule when the control pulse is applied at '^3/4 of Ty.

856

Fig. 1 plots the calculated time-evolution of the population of each vibrational eigenstate at a delay time of ~3/4 T^. At t = 0, the population is distributed over ten eigenstates, but after the laser pulse is over, most of population coherently moves to V = 0. This corresponds to coherent cooling of the vibrational motion.

oUvA 0.6 0.5

V=0

0.4 0.3 j^ 0.2 o

V^2 MiiilliillMlpiMN^^

20

40 60 Time / fs

• -'-••'•'I"

80

_

0.1

1 '"--10.0

100

Fig. 1. Calculated time-evolution of the population of each vibrational eigenstate. The upper panel shows the laser field. After the laser pulse is over, most population moves to v = 0.

4. Conclusion Controlling atoms in molecules by laser fields is one of the dreams of coherent control. We show that combining wave packet motion with carefully timed laser-induced dipole forces, the vibrational wave packet exchanges its energy with the laser field. The laser induced dipole force is one of the mechanisms responsible for control of larger molecules in learning loop experiments. References [1] H. Niikura et aL, Nature 417, 917 (2002); Nature 421, 826 (2003). [2] H. Niikura et at., Phys. Rev. Lett. 90, 203601 (2003). [3] H. Niikura et al., Phys. Rev. Lett. 92, 133002 (2004).

857

Molecular orientation of CH3F induced by phase-controlled lights Hideki Ohmura, Fumiyuki Itoh and M. Tachiya National Institute of Advanced Industrial Science and Technology (AIST), Tsukuba Central 5, 1-1-1 Higashi, Tsukuba, Ibaraki 305-8565, Japan E-mail: [email protected] Abstract. We investigate molecular orientation of CH3F induced by phase-controlled twocolor pulses consisting of a fundamental pulse (co) and its second harmonic pulse (2co) with an intensity of 1.0x10^^ W/cm^ and a pulse-duration of 130 fs.

1.

Introduction

Recent developments in laser technology have allowed us to investigate the ultrafast phenomena in femtosecond or attosecond timescale. When a pulseduration is comparable to an oscillation period of an optical field, such as fewcycle laser pulses, a direction of the electric field at a maximum point is dependent on a phase so that ultrafast processes are strongly affected by the phase of the optical fields. Such absolute-phase phenomena have been observed in photoionization by using few-cycle laser pulses [1], and have a potential to open the door for a new type of the manipulation for atoms and molecules [2]. Another method to induce phase-related phenomena without few-cycle laser pulses is a use of phase-controlled two-color laser pulses consisting of a fundamental pulse (co) and its second harmonic pulse (2co). The total electric field of the linearly polarized optical fields of the two frequencies, the fundamental (co) and its second harmonic (2co), is given by E(t)=E]Cos(cot)-\-E2Cos(2cot-^ (p), where Ex and E2 are the amplitudes of the electric fields and 0 is the relative phase difference between the fundamental and the second harmonic. The amplitude of the electric field in the positive direction is twice that in the negative direction when 0 = 0 and Ex = Ei. The field asymmetry induces phase-related phenomena. Theoretical works have suggested the application of the phase-controlled co+2co fields to the orientafion of molecules [3,4]. The phase-controlled co+2co optical field discriminates between parallel and antiparallel configurations of polar molecules. We have reported molecular orientation of IBr, CH3I and C3H5I by phase-controlled co+2co laser pulses experimentally [5,6]. It is expected that molecular orientation improves with decreasing the moment of inertia. In this paper, we have investigated molecular orientation of CH3F induced by phasecontrolled co+2co laser pulses with an intensity of 1.0x10^^ W/cm^ and a pulseduration of 130 fs. We improved molecular orientation by reducing the moment of inertia from CH3I to CH3F.

858

2.

Experimental Methods

The experimental apparatus, which has been described previously [6], consisted of a Tiisapphire laser system, a Mach-Zehnder interferometer, and an ion time-offlight (TOF) mass spectrometer. Briefly, laser pulses of 400 nm and 800 nm (the fundamental and the second harmonic), 130 fs duration, and 20 Hz repetition rate were focused on the molecular supersonic beam in the reaction chamber using a spherical mirror. The polarizations of the co and 2co pulses were both parallel to the detection axis in the TOF mass spectrometer. The relative phase difference between the co and 2co pulses was scanned by rotating the quartz plate (3 mm thickness) inserted m the path of the 2co beam in the Mach-Zehnder interferometer. A gas sample consisting of CH3F and CH3I diluted with helium was introduced via a supersonic beam source. Photofragments were measured by ion time-of-flight (TOF) mass spectrometer in a condition for high velocity resolution.

3.

Results and Discussion

Figure 1 shows the TOF mass spectra of singly charged ions produced by dissociative ionization of CH3F when the molecules were irradiated with both the fundamental and the second-harmonic beams with zero time delay. We observed singly charged CHs"^ ions and F"^ ions. The main dissociation channels were CHsF^ -^ CHs^ + F and CHsF^ -^ CH3 + F^. Each photofragments were observed as a pair of peaks, one resulting from ions flying directly toward the detector, the other from those ions which first flew in the backward direction before being reversed by the extraction fields. The spacing of the forward and backward peaks reflects the kinetic energy release. As shown in Fig. 1, the directional asymmetries of the photofragment angular distribution were clearly observed in the TOF spectrum. The F^ ions were preferentially emitted away from the detector, and CH3^ ions were emitted preferentially toward the detector at 0 = 0. Conversely, F^ ions were emitted preferentially in the forward direction, and CH3'^ ions were emitted preferentially in the backward direction at 0 = 7C. Figure 2(a) shows the ratio of forward and backward yields, (/> If/lb as a function of 0 . A CH3" F' c periodicity of 2n was clearly Kinetic energy [eV] Kinetic energy [eV] observed in the If/lb ratio for ^ 5 0 ^5 0 03 p! each photofragments in CH3F. The phase dependencies between the F^ ions and CUs^ ions were almost out of phase with each CO Z other. This result indicates that molecular orientation is achieved ^•^ ^-^ ^-^ by phase-controlled (o+2(o fields. FLIGHT TIME [jas] Figure 2(b) shows the ratio of forward and backward yields If/lb Fig. 1. The TOF spectra of singly charged ions produced by dissociative ionization of CH3F.

r- r^rr T ^ J • i r/:i ^ r CH3I reported previously [ 6 ] .

859

Comparing CH3F with CH3I, the If/lb ratio at 0=0 for halogen ions increased from 1.3 (f) to 2.4 (F^). According to the classical equation of motion and the interaction between the optical fields and the dipole moment, the key parameter for molecular orientation is the permanent 27C 47C 67c 87t 107C dipole Id and the moment of RELATIVE PHASE inertia /. We can neglect a DEFFERENCEd) difference of the permanent dipole because there is no big Fig. 2. The ratio of forward and backward yields, R = 1,/!^, as a function of the relative phase difference (j>; (a) CH3F, (b) CH3I. difference between CH3F (^1=1.86 ) and CH3I (^1=1.64 ). We can evaluate the effect of the moment of inertia ( I ) form the rotational constant ( B ), defined as 5 = Pi^/ll. BCHSI and BCHSF are 0.25 cm"* and 0.85 cm'*,

respectively. Qualitatively, the increase of the y \ ratio is in good agreement with the increase of the rotational constant. Therefore, we conclude that the molecular orientation improves as the mass of the counter-ions decrease.

4.

Conclusions

We investigate molecular orientation of CH3F induced by phase-controlled twocolor pulses consisting of a fundamental pulse (co) and its second harmonic pulse (2co) with an intensity of 1.0x10^^ W/cm^ and a pulse-duration of 130 fs. The orientation of CH3F is monitored by the forward^ackward yield ratio of photofragments in dissociative ionization. The molecular orientation is observed when a relative phase difference between co and 2co pulses (9) is 0 and 71. The forward/backward yield ratio at (p= 0 for CH3F is twice larger than that for CH3I. This result indicates that molecules with small moments of inertia are easy to orient by phase-controlled two-color laser pulses.

References G. G. Paulus, F. Grasbon, H. Walther, P. Villoresi, M. Nisoli, S. Stagira, E. Proori and S. De Silvestri, Nature 414, 182-184(2001). H. Niikura, F. Legare, R. Hasbani, M. Y. Ivanov, D. M. Villeneuve, and P. B. Corkum, Nature 421, 826-829(2003). E. Charron, A. Giusti-Suzor, and F. H. Mies, Phys. Rev. Lett. 75, 2815-2818 (1995). C. M. Dion, A. D. Bandrauk, O. Atabek, A. Keller, H. Umeda, and Y. Fujimura, Chem. Phys. Lett. 302, 215-223 (1999). H. Ohmura, T. Nakanaga and M. Tachiya, Phys. Rev. Lett. 92, 113002(2004). H. Ohmura and T. Nakanaga, J. Chem. Phys. 120, 5176-5180(2004).

860

Observation and manipulation of quantum interferences in ladder climbing B. Chatel, J. Degert, S. Stock and B. Girard Lab. Collisions Agregats Reactivite, CNRS UMR 5589, IRSAMC, UPS, 118 route de Narbonne, 31062 Toulouse, France E-mail: [email protected] Abstract: Two-photon excitation of a quantum ladder system by a chirped pulse leads to strong interferences in the excited state population. The interest of such scheme in the project of the artificial star is discussed.

1. Introduction Many results have been obtained in the high [1] or intermediate-field regime involving adiabatic transfer in multilevel ladder climbing [2,3] in atomic [4] or molecular systems [5]. The advent of ultrafast lasers combined with pulse shaping techniques have led to new schemes [6,7] where the two-photon transitions representing the lowest-order non linear interaction, are used as a benchmark system. The broad bandwidth of ultrashort lasers allows one to take advantage of interference effects between quantum paths in multiphoton transitions. Considering the simplest case of linearly chirped pulses where the laser frequency is regularly swept during the pulse, Noordam et al [8] observed quantum interferences between sequential and direct two-photon transitions in rubidium. These interferences are due to the phase difference accumulated by the various quantum paths, namely the direct non-resonant two-photon transition and the sequential ladder climbing. They are observed as a function of the laser chirp. In this contribution, we extend the scheme to the case of several intermediate states which provide as many sequential excitation paths and new interference possibilities. In particular, interferences between these sequential paths produce totally different patterns as compared to the sequential-direct interferences [9]. Moreover we study the conditions to observe a maximal contrast. This scheme is applied to the 3s-3p-5s ladder system in sodium vapor. Highly contrasted interferences are observed, as a direct consequence of this new interference scheme. The principles of new pulse shaping scheme are presented and under investigation experimentally. For example, cubic phase is added to enhance the population probability. Finally the interest of such a scheme in the project of artificial star is discussed.

861

2. Experimental Set-up This scheme is illustrated in sodium atoms [9]. The (3s-5s) two-photon transition (at 603 nm) is excited with a chirped pulse. The laser bandwidth allows us to excite the sequential (at 589 and 615 nm) and the direct two-photon transitions. The (4p-3s) fluorescence is recorded as a function of the chirp. This two-photon transition involves two intermediate states 3p (P1/2 and P3/2) and one excited state 5s,S1/2. The one-photon detunings are close. The laser system is based on a conventional Ti: Sapphire amplifier (800 |LIJ -130 fs - 1 kHz -795 nm). A fraction of the energy feeds a home made Non-coUinear Optical Parametric Amplifier (NcOPA), without compression, which delivers pulses of 10 |ij, 28 nm bandwidth centered around 603 nm. To vary the chirp on the interval (- 40 000 fs^ 40 000 fs^ ) with 100 fs^ steps, we combine glass rods (4 cm SF58, 6 cm SFIO) with a tunable double pass gratings pair (600 gr/mm). The 4P-3S fluorescence signal at 330 nm is recorded as a function of chirp rate.

3. Results

^

-35

-30

-25

-2D

-15

-10

-5

0

5

10

)"(1Cffs')

Fig. 1: Quantum interferences of a two-photon excitation by a chirped pulse. Figure 1 presents the experimental result compared with a theoretical simulation based on the numerical calculation of the excited state probability summing over both 3p intermediate states. In the calculation, the complete phase of the laser pulse is included instead of a pure quadratic phase to take into account the higher order dispersion terms. However, for the sake of simplicity, the results are plotted as a function of the chirp. The agreement between theory and experiment is excellent without any adjustment. The high contrast and the beat of slow and fast oscillations are clearly observable for negative chirp. For positive chirp, the signal is rapidly decreasing and is only due to the direct two-photon transition, since the

862

sequential paths are forbidden. The sharp step corresponds to the passage of the pulse around the zero chirp. The maximum enhancement of the population transfer compared to the population obtained with a transform limited pulse is about a factor of 3. For negative chirps, the combination of slow and fast oscillations leads to a maximum enhancing factor of 20 between the minimum at a chirp of -7500 fs^ and the maximum at -500 fs^ The oscillation periods are easy to deduce as well from the analytic expression as from the dressed states picture [9]. The maximum enhancement is obtained for a small negative chirp corresponding to the constructive phase of sequential and direct path. By adding more complex phase like cubic phase, it is possible to lead to constructive interferences also between both sequential paths.

4. Application to the polychromatic artificial star project Observations from terrestrial observatories are strongly limited by wavefront perturbations due to atmospheric turbulences. One way to reduce this effect is to use adaptive optics combined with a laser guided artificial star based on mesospheric sodium. A polychromatic star is required to correct the tilt induced by the atmosphere's dispersion. This can be achieved using a two-photon transition followed by cascading fluorescence [10]. The large spectral width of ultrashort pulses allows to cover the whole Doppler profile so that all velocity classes can be simultaneously excited. Adiabatic transfer is in principle the best way to achieve nearly 100% transfer efficiency per atom. However, this cannot be used for mesospheric sodium for the following reasons : (i) Obtaining the necessary power density to reach the adiabatic criterion at a distance of 100 km is a real challenge; (ii) When total fluorescence is to be maximized, it is more efficient to increase the excited volume than the power density. Therefore the optimal conditions are in the lowfield or intermediate field regime.

References 1 2 3 4 5 6 7 8 9 10

R. R. Jones, Phys. Rev. Lett. 74, 1091-94 (1995). J. Greg, G. Kazak and J. H. Eberly, Phys. Rev. A 32, 2776 (1985). S. Chelkowski, et al, Phys. Rev. Lett. 65, 2355-8 (1990). B. Broers, et aL, Phys. Rev. Lett. 69, 2062-65 (1992). J. S. Melinger, et al, J. Chem. Phys. 95, 2210-13 (1991). D. Meshulach and Y. Silberberg, Nature 396, 239-42 (1998). T. Homung et al, Appl. Phys. B 71, 277-84 (2000). P. Balling et al, Phys. Rev. A 50, 4276-85 (1994). B. Chatel et al, Phys. Rev. A 68, 041402R (2003). R. Foy et al, Astron. Astrophys. Suppl. Series 111, 569-78 (1995).

863

Adaptive polarization control of molecular dynamics T. Brixner', G. Krampert', T. Pfeifer\ R. Selle', G. Gerber', M. Wollenhaupt', O. Graefe', C. Horn', D. Liese', T. Baumert' ^ Physikalisches Institut, Universitaet Wuerzburg, D-97074 Wuerzburg, Germany E-mail: [email protected] ^ Experimentalphysik III, Universitaet Kassel, D-34132 Kassel, Germany Abstract. We demonstrate that the use of time-dependent hght polarization opens a new level of control over quantum systems. With K^ molecules we show that polarizationshaped laser pulses increase the multiphoton-ionization yield compared to linearlypolarized laser pulses. Coherent control is a powerful method which allows to "steer" quantummechanical processes toward a desired outcome by applying optimal light fields [1]. The main experimental tool for achieving this goal has been spectral phase shaping of femtosecond laser pulses [2]. Numerous implementations were reported in recent years, but in all these experiments only the scalar properties of ultrashort laser pulses were optimized. The light-matter interaction is governed by the scalar product ju-E{t) with E{t) a vectorial quantity. If the momentary polarization state of the applied electromagnetic field is varied, ^ . ~E{t) can be optimized throughout the complete temporal evolution of a quantum system. In the work reported here, we carry out molecular quantum control making explicit use of polarization variation [3] on an ultrashort time scale. In order to illustrate the novel features of such experiments, we maximize photoionization in a small prototype system, the potassium dimer K^. Within the bandwidth of our laser system, the dominant transition pathway that contributes to the K^^ yield [4] populates the 2^11^ state as an intermediate before the final ionization step (Fig.la). The 2^11^ state can be reached from the X^Sg"^ ground state by a twophoton process with intermediate wavepacket propagation in the A^IL^ state. This pathway is strongly polarization dependent because according to selection rules the two involved electronic transitions require electromagnetic fields with polarizations parallel and perpendicular to the molecular axis, respectively. In order to illustrate these issues further, we performed a "conventional" pump-probe experiment in partially aligned K^ molecules where the alignment is due to suitable molecular beam conditions. First, we used equally intense pump and probe laser pulses both polarized parallel to the mass-spectrometer axis. The amount of K2"' as a function of pump-probe delay (Fig. lb, solid line) is symmetric and shows a minimum with respect to time zero. On the other hand, if the probepulse polarization is perpendicular to that of the pump pulse, the signal is asymmetric (Fig. lb, solid/dashed line), and for negative time delays the K/ production is significantly enhanced. This result proves the polarization dependence of the K/ ionization pathways.

864

45 cm'' (740 fs)

2 3 4 5 6 7 8 9 1011 distance [A]

delay [fs]

wavenumber[cm']

Fig. 1. a) Calculated potential energy curves of the potassium dimer system, b) Section of pump-probe transients with mutually parallel (solid line) and crossed (solid/dashed line) linear polarizations around delay time zero, c) FFT-analysis of pump-probe transients with mutually parallel (lower graph) and crossed (upper graph) polarizations. Fourier analysis of the pump-probe transients reveals that for mutually parallel pump and probe polarizations (Fig.lc, bottom graph) only the dynamics in the 2^ng state with a vibrational period of 740fs [4] is visible. For mutually perpendicular pump and probe polarizations (Fig.lc, upper graph) an additional Fourier peak is found at 65cm'^ (corresponding to 510fs) which can be attributed to vibrational dynamics in the A^I^J state. This result proves that a time-dependent polarization of the controlling laser field gives access to the observation of additional dynamics on different electronic states. Note that all other experimental parameters in the two cases of Fig. 1 were identical (pulse intensities etc.), so that the differences in the observed transients are entirely due to light polarization properties. This type of polarization sensitivity can then be exploited even in much more generality in connection with femtosecond laser pulse shaping. For this purpose, the experimental setup is complemented by a polarization pulse shaper [5] and a computer with the optimization algorithm. We performed two types of adaptive control experiments to maximize the K2'' yield: spectral polarization-and-phase laser pulse shaping as well as phase-only shaping. In both cases the same number of free parameters is optimized and the two strategies are run in a parallel implementation. This ensures identical experimental conditions in terms of the molecular beam parameters and laser performance, allowing us to compare the results directly. The evolution of the K/ signal as a function of generation number within the evolutionary algorithm is shown in Fig.2. The increase for phase-only pulse shaping (solid circles) is due to the adaptation of the laser pulse structure to the vibrational dynamics of the potassium dimer, providing high laser intensities when the wavepacket is in a suitable Franck-Condon region. This general type of mechanism is what had been exploited and discussed in the theoretical and experimental literature on quantum control to date. However, when the additional mechanism of light-polarization control is used (open circles), one can go beyond the limitations set by linearly polarized fields, and achieve significantly higher product yields. This demonstrates not just a quantitative improvement but rather a qualitative extension of quantum control mechanisms, because it goes beyond one-dimensional addressing of transition dipoles and rather makes use of their directional properties by shaping the polarization state of the controlling laser pulse.

865

o pctiartzeikin epiimilzaftion

* ptiaeQ-Qhly Dptlmizatlon

goneralfon

Fig. 2. a) Evolution curves show the K^"^ ion yield relative to that obtained with an unshaped laser pulse, b) Quasi-three-dimensional representation of the optimal polarization-shaped laser pulse. Time evolves from -1.5ps (left) to -i-1.5ps (right), and electric field amplitudes are indicated by the sizes of the corresponding ellipses. The momentary frequencies are indicated by grey scale and the shadows represent the amplitude envelopes of the two orthogonal components. Fig.2b) shows a representation of the best laser pulse shape reached in the final generation of the polarization optimization. The momentary frequency and the polarization state of the optimized pulse changes substantially in a complex fashion as a function of time. Some reasons for this complexity are briefly discussed now. First, the detection step in this experiment (i.e., the ionization) needs to be considered in more detail. Observation of 2^ng state dynamics in our pump-probe measurement (Fig.l) proves that ionization from the l^U^ is dependent on the internuclear distance and occurs predominantly at the outer turning point [4]. Another reason for the complicated pulse structure is the broad spectrum of the ultrashort laser pulse. The vibrational dynamics of the potassium dimer are known to depend strongly on the center frequency of the excitation laser pulse [4]. In our case this means that the optimized polarization needs to be provided for a wide distribution of frequencies and timings. While all these factors complicate the analysis and interpretation efforts of the optimal pulse shape, the important point is that despite of the complexity an optimized electric field timevarying polarization indeed has been exploited by the evolutionary learning algorithm as a novel control agent. In conclusion we have demonstrated that time-dependent shaping of femtosecond light polarization can give access to a further level of control of quantum systems. Comparative optimizations of K^"^ yield show that polarization laser pulse shaping is superior to phase-only shaping, because the vectorial electric field can adapt to the time evolution of the polarization-dependent transition dipole moments. We have hence exploited the vectorial properties of light-matter interaction to achieve quantum control in a molecular model system.

References 1 2 3 4 5 866

T. Brixner and G. Gerber, ChemPhysChem 4, 418, 2003. A. M. Weiner, Rev. Sci. lustrum. 71, 1929, 2000. T. Brixner and G. Gerber, Opt. Lett. 26, 557, 2001. C. Nicole et al., J. Chem. Phys. I l l , 7857, 1999. T. Brixner, G. Krampert, P. Niklaus, and G. Gerber, Appl. Phys. B 74, S133, 2002.

Two-photon absorption imaging with shaped femtosecond laser pulses W. S. Warren^'^ A. Miller\ W. Wagner^ T. Ye^ M. Fischer^ and G. Yurtsever^ ^ Department of Chemistry, Princeton University, Princeton, NJ 08544, USA ^Department of Radiology, University of Pennsylvania, Philadelphia, PA 19104, USA ^Department of Biomedical Engineering, Rutgers University, Piscataway, NJ 08854, USA E-mail: [email protected] Abstract. Femtosecond laser pulse shaping permits background-free detection of twophoton absorption, which tends to refill spectral holes. This opens up new spectroscopic windows for monitoring tissue characteristics. Optical methods for determining structure and characteristics of tissue in vivo have seen significant advances over the last few^ years. Absorption, reflectance, Raman, fluorescence, fluorescence lifetime and backscattering measurements have been shown to be useful in a wide variety of clinical applications where the target is near a surface. For example, fluorescence readily locates metabolic co-factors (NAD(P)H, FAD), structural proteins (collagen or elastin crosslinks), and aromatic amino acids. However, the exciting UV light typically penetrates only a few tens of microns in the near-UV. Near-IR imaging (ca. 800 nm) can detect oxygenation state with very modest laser energies ( « 1 mW) [1] because of the low tissue absorption coefficient and differences between hemoglobin and deoxyhemoglobin in this spectral region. Major applications to date include tumor detection (e.g. in breast) and functional brain imaging. Unfortunately, images typically have 1 -2 cm spatial resolution. The fundamental problem is light scattering at 800 nm. Ultrafast laser pulses provide some advantages in imaging applications. One approach is to use two-photon fluorescence microscopy, starting in the water window near 800 nm [2]; since the signal is proportional to the square of the intensity, scattered light induces minimal fluorescence. Despite significant performance advantages over more traditional methods such as confocal microscopy (most notably in penetration depth, which is 1 mm), multiphoton fluorescence imaging still has some significant limitations. Only species that fluoresce are visible (thus, for example, while NADH can be seen, the oxidized version NAD+ cannot); the fluorescence still has to get out of the sample; and unwanted higher-order multiphoton processes commonly cause cell damage. Detection of absorption instead of fluorescence permits direct observation of important endogenous molecular markers which are invisible in multiphoton fluorescence microscopy. However, conventional methods for measuring twophoton absorption (TPA), such as the z-scan approach, require very large pulse energies. Here we show that using appropriately designed shaped femtosecond pulses permits detection of TPA with modest laser powers. It can also permit signal detection from deeper tissue, using the long-wavelength water windows (around 1.05 and 1.3 microns) which have significantly reduced optical scattering, but little endogenous two-photon fluorescence. TPA spectroscopy at these

867

wavelengths will give greater penetration depth and less potential for photodamage than conventional fluorescence-based methods. The essential trick in efficient measurement of two-photon absorption (which, in a typical application, might be orders of magnitude smaller than light scattering or one-photon effects) is to tailor the pulse or pulse sequence such that a nonlinear response appears at a different frequency than linear ones. For example, Tian and Warren noted that an amplitude-modulated modelocked laser pulse train (see Fig. 1) would produce frequency sidebands Siinfo ±f, but that only nonlinear processes (not one-photon absorption or scattering) would produce frequency sidebands at nfo±2f. Thus a lock-in amplifier could efficiently detect two-photon absorption [3]. Data for oxygenated hemoglobin and a suspension of eumelanin obtained at a wavelength of 800 nm are displayed in Fig. 2 and confirm the quadratic dependence of TPA on pulse intensity. This loss modulation method decreased power requirements by a factor of 100 relative to the conventional zscan measurements, but is still limited by shot noise and instability in the laser pulse train. 1 Tim e

...... •\

IJI 'Frequency

Natural Eumelanin

4

Oxygenated Hemoglobin

Ikl t

1 ^^

JiLl Lll u

LU

Fig. 1. (top) TPA of a modulated pulse train creates additional frequency components. Fig. 2. (right) TPA measurement of melanin and oxygenated hemoglobin.

1

Average Power (mW)

2

3

4 5 6 78

Average Power (mW)

A variant which avoids these problems is to shape the laser pulses themselves in such a way that new light components are created by nonlinear effects [4]. Consider, for example, a laser pulse with a hole in its spectrum (Fig. 3); in the time domain, such a pulse must consist of a intense and short part, balanced against a long and weak part (with the areas of these two parts the same, if the pulse is to have zero amplitude at the center frequency). TPA tends to decrease the amplitude of the intense part more than the weak one, thus refilling the spectral hole. We have demonstrated this spectral hole refilling at the three water window wavelength ranges (0.8, 1.1 and 1.3 jum) in a variety of test samples, including excised tissue, with a wide range of waveforms produced by acousto-optic pulse shaping [5]. The acousto-optic approach lets us toggle between different waveforms at very high speed, and thus provides a clean method of detecting pulse shape effects. As a general rule, the extent of the refilling depends on the frequency dependence of the two-photon absorption, the precise shape of the laser pulse, and on self phase modulation (which serves as a competing process). In the simplest case (starting with an in-phase pulse such as shown in Fig. 3), we observe

868

Frequency J k domain ^ B ^ ^

- NoTPA;SPM - Detuning=0.00; no SPM - • Detuning=0.00; SPM •— Detuning=0.05; no SPM • - Detuning=0.05; SPM Detuning=0.01; no SPM Detuning=0.01;SPM

kk ^ ^ ^ TPA

1 •

Time domain

'

jd

^B

^k

Fig. 3 (top). Hole refilling effect of TPA. •^

ann

BUU -

•

1 700 - • Water 0.00

•e 600 - 1 a 5 mM R6G/water ffi

1

400 •

(0

'•= 200 S 100-

• ^ ^^

0

0.10

Fig. 4 (top). Hole refilling for an anti-symmetric laser pulse as a function of detuning from twophoton resonance.

B

£ 300 -

3:

0.05

Frequency Offset [10'^ Hz]

» 500 -

1

A il 2

3

Average power (mW)

4

5

Fig. 5 (left). Difference between hole refilling for symmetric versus anti-symmetric laser pulses.

the refilling to be quadratic in the laser power at low powers. For an antisymmetric pulse in the frequency domain, the spectral hole refilling is expected to be smaller than for a symmetric pulse (which we also observe experimentally), and Fig. 4 shows that the expected refilling depends strongly on offset from two-photon resonance. In more complex samples, one often finds power-law dependence significantly different from 2 because of pulse reshaping. Figure 5 shows one such example, in this case a dye solution (R6G) where the excitation is toggled back and forth between symmetric and antisymmetric pulses. The difference between the amount of hole refilling in these two cases rises much faster than quadratically with laser power in this case. The interaction between normal dispersion and selfphase modulation can lead to the minimum pulse length before or after the focus; in the latter case, increasing the peak power increases the self-phase modulation, hence shortening the pulse at the focus and leading to a higher apparent power law dependence. Since the hole refilling effect is sensitive to the details of the TPA profile we can investigate a range of different species in tissue, including melanin and hemoglobin, by tuning to different wavelength ranges (and combinations corresponding to sum-frequency absorption instead of two-photon absorption).

References 1 B. Chance, E. Anday, S. Nioka, S. Zhou, L. Hong, K. Worden, C. Li, T. Murray, Y. Ovetsky, D. Pidikiti and R. Thomas, Optics Express 2, 411-423 (1998) 2 W. Denk, J. H. Strickler, and W. W. Webb, Science 248, 73 (1990) 3 P. Tian, W. S. Warren, Optics Letters 27, 1634-1636 (2002) 4 W. S. Warren, W. Wagner and T. Ye, Mag. Res. Imag. 21, 1225-1233 (2003) 5 C. Hillegas, J. X. Tull, D. Goswami, D. Strickland and W. S. Warren, Optics Letters 19,737-739 (1994)

869

Selective two-photon functional imaging through scattering media based on binary phase shaping Igor Pastirk, Johanna M. Dela Cruz, M. Comstock, Vadim V. Lozovoy, Marcos Dantus Michigan State University, East Lansing MI 48824 Tel. No 1 (517) 355 9715. Email: [email protected] Abstract: We demonstrate experimentally selective two-photon functional imaging through a scattering medium (a thin slice of chicken) based on concepts of coherent control. The selectivity, achieved using binary phase shaping, is maintained by the ballistic photons.

Two-photon imaging has provided researchers with unique possibilities for fluorescence imaging. It has the advantages of high resolution, lower background scattering, better penetration in thick samples, and reduced photo-induced damage [1-3], which arise from the basic physical principle that the absorption depends on the square of the excitation intensity. In previous work we have shown that phase modulated femtosecond pulses can be used for selective two-photon excitation [46]. Here we explore this control as the laser transmits through scattering media. The experiments are carried out using a (10 fs) Ti-sapphire oscillator. The spectral phase was calibrated by Mil phase-scan, first to obtain transform-limited pulses (TL) and then to introduce the desired phase [7]. Ultrashort pulses overlap the two-photon absorption spectrum of the pH sensitive chromophore in acidic and alkaline buffer. The spectral phase functions used for this experiment were obtained by an algorithm that optimized the contrast between excitation of the chromophore in acidic versus alkaline buffer. The search space was reduced by restricting the values that each pixel can take to 0 and it [8]. The first sample consisted of five capillaries (1 mm i.d.), four contained aqueous solutions of HPTS, two in acidic and two in alkaline pH buffers; the fifth tube is filled with water. In front of the capillary tubes a -- 1 mm slice of tissue is placed and the whole sample is translated in front of the laser beam, which is focused by a 1 OX objective. The fluorescence as a function of position measured with TL pulses is plotted in Figure 1 (black lines). A binary phase mask was optimized for excitation of HPTS in pH 6, while another phase mask was optimized for excitation of HPTS in pH 10 solution. The difference of the two-photon fluorescence signals obtained using these two masks is shown in Figure 2 (red lines). Notice that the contrast, and therefore coherent control, remains high even with the presence of ~ 1 mm of chicken tissue. These results were obtained with 1 nJ, 10 fs pulses centered at 830 nm. We expect that thicker slabs of tissue could be used, when more intense (factor of a million) pulses from an amplified laser source are used.

870

1

^^

1.2Difference 1.0- 1 0.80.6- With tissue 0.40.20.0-0.2-0.4-0.6-0.8-1.0-1.2-1.4-1.6-

1 8

1

^

h

f. CT"^ - • s ,

i

0

2

4

ii\r^.f^<' 11V

6

8

10

1 12

14

16

J

18

Position [mm]

Fig. 1. Selective two-photon excitation without (left) and with (right) tissue. The signal obtained for TL pulses is shown in the negative direction for clarity. The difference signal (optimized for pH 6 - Optimized for pH 10) is shown in the positive direction. The first and fourth capillaries have HPTS in pH 10 buffers. The second and fifth have HPTS in pH 6 buffer. The third capillary (~ 9 mm) has water and gives no signal.

For functional imaging experiments we the sample configuration shown in Figure 2. Three capillaries with HPTS in acidic buffer were submerged in a 2 mm cell containing HPTS in an alkaline buffer. A mask with small letters (MSU, 1.5 mm in height) was used to help us localize the capillaries. Once the figures were obtained using the two different phase masks, a functional image was computed by taking the ratio of the two images pH6/pH10.

HPTSinpMOsoln

Fig. 2. Functional two-photon imaging without (left) and with (right) biological tissue. The setup for this experiment is shown schematically in the left panel. The letter S coincides with a capillary filled with HPTS in acidic buffer. The brighter regions correspond to the capillaries with pH6 solution.

The images in Figure 2 show that it is possible to use coherent control methods to discriminate between the two solutions. TL pulses give the same intensity for both solutions (not shown) and give no pH sensitive contrast. The pH functional

871

contrast achieved by phase shaping is clearly preserved even when the laser transmits through scattering media such as biological tissue. We have performed a number of experiments to determine the limitations to our observations. We have established that the ballistic photons, those that are not scattered, are responsible for tv^o-photon excitation. The ballistic photons are lost exponentially with depth, thus limiting our method to 3-4 mm. We have imaged fluorescent beads that are 15 microns in diameter after the laser transmits through one millimeter of tissue, demonstrating that functional imaging is capable of --1 micron resolution. Finally, we have directly measured the spectral phase of pulses after they transmit through biological tissue using MIIPS and found that the ballistic photons suffer no phase deformation. These results will be published elsewhere [9]. Acknowledgements We gratefully acknowledged funding from the Chemical Sciences, Geosciences and Biosciences Division, Office of Basic Energy Sciences, Office of Science, U.S. Department of Energy, for their support of this research.

References 1 2 3 4 5 6 7 8 9

872

W. Denk, J. H. Strickler and W. W. Webb, Science 248,73,1990. K. Konig, J. Microsc. -Oxf., 200,83, 2000. P. T. C. So, C. Y. Dong, B. R. Masters and K. M. Berland, Annu. Rev. Biomed. Eng. 2, 399, 2000. K. A. Walowicz, I. Pastirk, V. V. Lozovoy and M. Dantus, J. Phys. Chem. A, 106, 9369, 2002. V. V. Lozovoy, I. Pastirk, K. A. Walowicz and M. Dantus, J. Chem. Phys., 118, 3187,2003. J. M. Dela Cruz, I. Pastirk, V. V. Lozovoy, K. A. Walowicz and M. Dantus, J. Phys. Chem. A 108,53, 2003. V. V. Lozovoy, I. Pastirk, M. Dantus, Optics Letters, 29, 775, 2004. M. Comstock, V. V. Lozovoy, I. Pastirk, M. Dantus, Optics Express, 12, 1061,2004. J. Dela Cruz, I. Pastirk, M. Comstock, V. V. Lozovoy, M. Dantus, Proc. Nat. Acad. Sci. USA, submitted 2004.

Photon number squeezing of ultrabroadband pulses generated by microstructure fibers H. Furumochi\ A.Tada\ K.Hirosawa\ F. Kannari\ M. Takeoka^ and M.Nakazawa^ ^ Department of Electronics and Electrical Engineering, Keio University, 3-14-1, Hiyoshi, Kohoku-ku, Yokohama, 223-8522, Japan E-mail: [email protected] ^ National Institute of Information and Communications Technology (ex NICT), 4-2-1 Nukui-kitamachi, Koganei-city, Tokyo 184-8795 Japan ^ Research Institute of Electrical Communication, Tohoku University, 2-1-1, Katahira, Aoba-ku, Sendai 980-8577, Japan Abstract. Quantum correlation in broadband fiber(MF) is experimentally and theoretically highest squeezing of -4.6 dB using a 100-fs among the spectral components is analyzed with

spectra generated with a microstructure studied. We experimentally obtained the femtosecond pulse. Quantum correlation QNLSE including Raman effect.

1. Introduction Quadrature entanglement of optical solitons created from the Kerr nonlinearity in an optical fiber is an alternative scheme to generate entanglement in photonics domain in order to realize a practical quantum information network[l]. More than 50% photon-number squeezing of optical soliton pulses is obtainable with a nonlinear fiber and a spectral filtering[2]. In the spectral filtering method, photon number squeezing is obtained when only the negatively correlated parts of the spectrum are preserved. Figure 1 shows plots of squeezing level measured by changing the shorter wavelength edge of the BPF. The highest squeezing of —4dB was obtained for a 766-917 nm spectrum band.

Wsfv-elengthlnm ] Fig. 1. Plots of squeezing level measured for band-pass filters(BPF) with various higher frequency cut-off. The broadband spectrum output from the MF is also shown. The center wavelength of the input pulse is 810nm.

873

Recently, MFs, of which zero dispersion wavelength locates around 700-800 nm, have been employed to generate squeezing with an 800 nm femtosecond pulse. However, the relatively higher transmission loss of these fibers was more pronounced and higher optical nonlinearity cannot be well utilized yet [3].

2. Numerical Methods We regarded the quantum fluctuation as perturbation and adopted a backpropagation method that can set up the variance from commutation relation. In the sequence of the backpropagation method, first, we calculated pulse propagation classically in the following equation and set up the classical solution.

Then, we calculated the backpropagation of M^ using equation (2), which describes the propagation of the perturbation field. ^^" - j ^ ^ ^ ' ^ " ^ m^- fMA'^" + jf^yich^it-T)\A\^dru''

- j(^- fR)y^^"^

+ A^h^{T -t)A*uUr

(2)

- A^h„{T -t)Au''*dT

Using u'^, which was worked out from the equation (2), we obtained the variance. For treating Raman scattering, we calculated the Raman response function: 13

A'

h^ = y ^-exp(-r,0exp(-r,'rV4)sinK,0

(3)

Raman scattering couples to a heat bath, and will degrade squeezing. We calculated the intensity of the quantum correlation of intrapulse using the following equations.

In the equation (4), when we obtain the negative correlation components and eliminate the positive correlation components, we can get high squeezing levels[4]. We calculated the QNLSE to examine the influence of Raman effect on the quantum correlation in a soliton pulse. We set a standard glass fiber (I32=-10 ps^km, y=0.01 soliton period of 40cm. We let the soliton pulse propagate along 3soliton periods in the fiber. The quantum correlation map among spectral components in the output spectrum is slightly modified with the Raman term. The photon-number squeezing level obtained by the same BPF also slightly degraded with the Raman effect. As the input pulse energy increases, the soliton splitting due to Raman effect and FWM processes generate broadband spectrum structures..

3. Results and Discussion We repeated similar analysis for more intense input laser pulses with parameters of the MF. We set the MF (P2= -12.76 ps^ /km, y=0.0452W"^ m"\ zero dispersion wavelength 800nm) and the laser pulse with the center wavelength of 850nm, the

874

transmission power of llkW, and the pulse width of 50fs. We let the ultrashort pulse propagate 2cm in the MF.

600 700 800 90010001100120G Q

Wavelength (nm) 600700 600900100011001200

Wavelength (nm)

Fig. 2. Calculated spectrum output from a fiber and the quantum correlation map for this spectrum Figure 2 is the spectrum output from the fiber and the map of the quantum correlation. The spectrum is highly broadened and exhibits complicated structures. It is clear that the Stokes components negatively correlate with the anti-Stokes components. However, there is only positive correlation inside of the broad Stokes spectrum. It doesn't agree with our experimental result where the highest squeezing was obtained by preserving Stokes or anti-Stokes near the center wavelength, although the propagation length was much longer in our experiment than this calculation.

4. Conclusions In the experiment, we observed negative quantum correlation in the Stokes region of the fs-laser pulses propagated through a PCF. SPM, SRS, and FWM play a crucial role to construct the multi-mode quantum correlation. And we obtained the photon-number squeezing, which was generated by proper spectral filtering. Numerical analysis was carried out to understand the quantum correlation structure in more detail. Negative quantum correlation is obtained in limited regions of Stokes and anti-Stokes wavelengths. Quantum correlation depends on the input pulse (wavelength, power, pulse width) and PCF(GVD, length).

References 1 G, Leuchs and et. al. "Quantum Information with Continuous Variables" Ed. S.L. Braunstein and A.K. Pati, Kluwer Academic Publishers, p.379, 2001. 2 S.R. Friberg and et. al. Phys. Rev. Lett. Vol. 77, 3775,1996. 3 S. Lorenz and et. al., Appl. Phys. B Vol. 73, 855, 2001. 4 S. Spatter and et. al. Phys. Rev. Lett. Vol. 27, 786,1998.

875

Real-time, ultrahigh-resolution optical coherence tomography at 1*5 fim using a femtosecond fiber laser continuum N. Nishizawa, Y. Chen, P. Hsiung, V. Sharma, T. H. Ko, E. P. Ippen and J. G. Fujimoto Department of Electrical Engineering and Computer Science and Research Laboratory' of Electronics, Massachusetts Institute of Technology, Cambridge, MA 02139, U.S.A. E-mail: [email protected] Abstract. Ultrahigh-resolution, real-time OCT is demonstrated with --5.5 }^m resolution in tissue at 1.4-1.7 i^m wavelengths using a high-power continuum generated by a mode-locked, stretched-pulse, Er-dopedfiberlaser and highly nonlinear fiber.

1. Introduction Optical coherence tomography (OCT) is an emerging technology for micron-scale cross-sectional imaging of biological tissue and materials [1]. One of the hmitations in OCT has been the lack of compact, high-performance, low-coherence light sources with sufficient bandwidth and power to enable ultrahigh-resolution, real-time imaging. Recently, compact, broad bandwidth femtosecond fiber laser sources have been demonstrated for uhrahigh-resolution OCT in 1300 nm wavelength region [2]. However, low output powers and high excess noise made real-time imaging difficult. Because optical scattering in tissues is reduced at longer wavelengths, the 1.4 to L7 jLim range is interesting for OCT imaging. In this paper, we demonstrate an all-fiber, ultrahigh-resolution, high-speed OCT system using a stretched pulse, passively mode-locked, high-power fiber laser that is spectrally broadened in a nonlinear fiber. Real-time, high resolution OCT in the 1.4 to 1.7 \xm wavelength region is demonstrated for the first time. This system is compact, easy to use, and promising for OCT imaging applications.

2. Experiments Figure 1 shows the experimental setup. We developed a stretched-pulse passively mode-locked, high-power, Er-doped fiber laser for OCT imaging [3]. The fiber laser is pumped by two 350 mW laser diodes at 976 nm wavelength using a polarization combiner. The fiber laser generates stable, linearly chirped, 1.7 ps pulses with 100 mW output at a repetition rate of 51 MHz. The output pulses are compressed using a single-mode fiber, then coupled into a highly nonlinear fiber to generate a low-noise, Gaussian-like, supercontinuum.

876

SMF28

Nonlinear fiber w

i—~~0

Sample

.||i^s|| Detector vf

corrtroller

High speed scanning ddiay line

Computer

Fig. 1: Experimental schematic of the all-fiber, ultrahigh-resolution OCT system using a broadband femtosecondfiberlight source.

Zi (0

":

ilnll \ji ^^

^

s

c .1

E

1

•

f

«

-

0) Q.

/

J

1

"'

'*

Output of ^ V fiber laser"

' .-^

,\; r

I*''

•

V

1

0,5

^

•; .

- 7.4 um

iiiil

:

0

P< -0,5 :

• -

|||H

"^ -1

CO -vW .1 • ... ... ,^'). > / , , , . , .*0. ,t ,'

1300

1400

1500

. 1600

Wavelength (nm)

.. i 1 .. r* 1700

-1.5 -30

-20

-10

0

10

20

30

Delay (urn)

Fig. 2: (a) Optical spectnun of thefiberlaser output and continuum, (b) Interferometric point spreadfiinctionwith 7.4 j,im axial resolution in air, corresponding to 5,5 jam in tissue. Figure 2(a) shows the resulting spectrum using a 95-meter length of normal dispersion, highly nonlinear (ND-HNL) fiber. Using operation in the normal dispersion regime, it is possible to generate a broadband, Gaussian-hke continuum without an increase in noise [4]. The bandwidth is broadened from 70 nm to 180 nm full-width half-maximum (FWHM) in the ND-HNL fiber. The OCT imaging system consists of a 50/50 fiber coupler and a circulator used in a dual-balanced detector configuration to optimize signal power coupled back to the detectors. The reference arm is scanned using a reflective scanner at a speed of 7.6 m/s and a repetition rate of 1900 Hz. Polarization controllers are used in the sample and reference arms to match the polarizations. In vivo imaging is performed with a scanning probe. The interference signal is electronically bandpass filtered, log demodulated, low-pass filtered, and digitized. The performance of the femtosecond fiber laser and OCT system is measured using an attenuated isolated reflection fi'om a mirror. To achieve optimum axial resolution, the dispersion in the interferometer sample and reference arms is

877

matched using identical materials in the two arms. Figures 2(b) shows the resulting interference signal. Since the spectral shape is close to Gaussian, almost no side lobes are visible on the linear scale. The measured axial resolution is 7.4 jam FWHM in air, corresponding to 5.5 jim in tissue. The detection sensitivity is measured to be 98.7 dB with a power of--10 mW on the sample.

s|2«)!!ijWil

iiilililiiiii?lfi^|iiij

Fig.3: Ultrahigh resolution OCT of (a) human skin and (b) human tooth and gingiva in vivo imaged at 4framesper second.

The feasibility of using this light source for high-speed in vivo OCT was demonstrated by imaging human skin, as shown in Figure 3(a). The image consists of 500 transverse scans with 1000 pixels per scan, covering a 2.5 x 1.8 mm area. The epidermis, sweat ducts, and junction of the epidermis and dermis are clearly visible. An image of the tooth and gingival junction is shown in Figure 3(b). These ultrahigh-resolution images are comparable to those obtained with bulk solid-state femtosecond hght sources. Fiber lasers have the advantage of being significantly more compact and robust compared to bulk solid-state lasers. Acknowledgements. We thank Furukawa Electric Co., Ltd. for providing the normal dispersive highly nonlinear fiber. This research was sponsored in part by NIH R01-CA75289-07 and ROl-EY 11289-18, NSF ECS-01-19452 and BES0119494, and AFOSRF49620-01-1-0186 and F49620-01-01-0084. N. Nishizawa is visiting from the Department of Quantum Engineering, Nagoya University, Japan, and is supported by the Telecommunications Advancement Foundation.

References: [1] D. Huang, E.A. Swaiison, C.P. Lin, J.S. Schuman, W. G. Stinson, W. Chang, M.R. Hee, T. Flotte, K. Gregory, C.A. Puliafito, and J.G, Fujinioto, Science 254, 1178-1 ISl, (1991). [2] K. Bizlieva, P. Pova^ay, B. Hemiann, H. Sattmaiin, W. Drexler, M. Mei, R. Plolzwarth, T. Hoelzenbein, V. Wacheck and H. Pehamberger, Opt.Lett.28, 707-709 (2003). [3] K. Tamura, C. R. Doerr, L. E. Nelson, H. A. Haus, and E. P. Ippen, Opt. Lett. 19, 46, (1994). [4] S. Bourquin, A. D. Aguiixe, L Hartl, P. Hsiung, T. H. Ko, J. G. Fujimoto, T. A. Birks, W. J. Wadsworth, U. Biinting and D. Kopf, Opt. Express 11, 3290 (2003).

878

Ultrafast Exciton Transport in Organic Nanotubes Audrius Pugzlys, Ralph Hania, Catalin Didraga, Victor Malyshev, Jasper Knoester and Koos Duppen Materials Science Center, University of Groningen, Nijenborgh 4, 9747AG Groningen, The Netherlands E-mail: [email protected] Abstract. The dynamics of exciton transport between the inner and outer walls of doublelayer cylindrical aggregates is measured. Downhill transport is fast (275 fs) and excitation intensity independent. Uphill transport is much slower (3.5 ps), but this rate increases when the excitation density is raised. This suggests that exciton-exciton annihilation is involved.

1. Introduction Organic supramolecular systems are prime candidates to serve as molecular wires for electronic energy transport. A possible application for such systems is light harvesting in artificial photosynthetic units. Here, the energy of a light source is absorbed by the supramolecular structure and transported as excitons to a special site where the electronic excitation is converted to another form of energy. Among the many systems that form supramolecular structures, the recently discovered molecular aggregates based on the 5,5',6,6'tetrachlorobenzimidacarbocyanine (TBC) chromophore are particularly interesting, since their morphology, and thereby their energy transport capabilities, can be easily controlled. The molecular structure and absorption spectrum of the aggregates are shown in Fig. 1. By varying the substituents R and R', either linear (quasi-onedimensional), planar, spherical or cylindrical aggregates are formed [1]. We report here time and frequency resolved optical studies of excitations in double wall cylindrical aggregates. These systems have a rather rigid structure that resembles rod-like elements in natural light harvesting chlorosomes of green bacteria. Experiments on the dynamics of excitons in chlorosomes were recently published [2]. We present the first study of the exciton dynamics of artificial TBC-based aggregates, which differ from the natural systems both in the nature of their building blocks and in details of their morphology.

Fig. 1. Absorption (sohd line) and fluorescence (short dashed line) spectrum of cyhndrical TBC aggregates, referred to as CgOa. The TBC molecule is shown in the inset. For CgOa-aggregates the substituents are R=C8Hi7 and R'=(CH2)3COOH. 440

480

520

560

600

X,/nm

879

2. Experimental Methods The mesoscopic structure of the aggregates is known from cryogenic electron transmission microscopy [1]. In these experiments, the CgOa-aggregates appear as Hnear rope-hke assembhes of tubular strands with a total thickness in the order of a few tens of nanometers and hundreds of nanometers length. A single strand represents a double wall cylinder with a total diameter of 10 ± 1 nm and a doublewall thickness of 4 ± 0.5 nm. Femtosecond, polarization selective, frequency resolved pump-probe experiments were performed on aggregates in room temperature water. Details of the experimental setup will be published elsewhere.

3.

Spectroscopy of the Aggregates

Four J-type absorption bands can be discerned in the absorption spectrum of the aggregates (1 to 4 in Fig. 1). Linear dichroism (LD) studies show that the transition moments of bands 1, 3 and 4 are preferentially oriented along the cylinder long axis while transition 2 has perpendicular orientation. Apart from band 1, which may originate from aggregates with a different morphology, the absorption and LD spectra can be simulated very well with a theoretical model based on a rolled bricklayer structure [3]. In this report we will concentrate on the lowest state of each manifold, since these are the most relevant when transport properties are considered. In Fig. 2 pump-probe spectra are shown, measured at zero delay. In case of excitation at 600 nm, the signal consists of a negative feature due to bleaching/ stimulated-emission, and a positive feature due to excited state absorption. This dispersive-type shape is typical for excitation of a single excitonic state [4], in this case involving transition 4. Since no optical density changes around 580 nm are induced, the excitons related to transitions 3 and 4 have different ground states. Pumping at 580 nm again gives a dispersive shape, now related to transition 3. The smaller features around 600 nm are caused by very fast energy transfer from the higher energy exciton to the lower energy exciton (see next section). Excitation at 560 nm gives small photobleaching around all transitions, indicating that the absorption is due to the same excitonic manifolds as 3 and/or 4.

3

A

\ ^ . . . . \

^,1

\ / \ /

\

r

540 550 560 570 580 590 600 610 620

X/nm

880

Fig. 2. Zero delay pump-probe spectra in the case of excitation at 600 nm (solid line), 580 nm (dashed line) and 560 nm (short-dashed line); the dasheddotted line is the steady state absorption spectrum.

j(a) a 0.00

9 — J 1 560

A

0.10-

A ^i ioofs\ yj i50fs|-^*..;/ 200 fs U - ' - / 500 fs I y 570

580

590

(b)

0.050.00-

600

X/nm

-0.10-

i

•'"•V^"*

-0.05-

1

11

- - 1ps • • • • 5 ps 20 ps

'// //

-0.15-

610

570 580 590 600 610 X/nm

570

580 X/nm

Fig. 3. Panel (a): dynamics of the pump-probe spectrum after excitation at 580 nm. Panel (b): dynamics of the pump-probe spectrum after excitation at 600 nm.

4. Exciton dynamics In Fig. 3 the dynamics of the excitons is shown. After excitation at 580 nm a very rapid decay occurs around 580 nm (higher energy exciton), while around 600 nm delayed formation occurs (lower energy exciton). The excitation energy transfer can be described with a single time constant (275±10 fs) that is independent of the excitation intensity. The weak interaction between the two excitonic manifolds suggests that the transport occurs incoherently as a Forster-type hopping process. The excitation of the lowest energy exciton at 600 nm is not only followed by decay, but also by build-up in the spectral region 595-610 nm. The enlargement of the spectrum around 580 nm demonstrates that the exciton of transition 3 is involved. At low excitation density (1 absorbed photon per 10"^ molecules), the transfer from low to high energy takes place with a time constant of 3.5 ± 1 ps. The difference of about a factor of 10 in the energy transfer back and forward, indicates that the Forster model may be an oversimplification of the true dynamics. Possibly a barrier is present that influences the transfer process. The transfer rate from lower to higher energy exciton was measured to be intensity dependent with distinct non-exponential behavior. A possible mechanism involves exciton annihilation, in which two colliding one-excitons populate a two-exciton state, followed by fast relaxation back to the one-exciton manifolds [4].

References 1 H. von Berlepsch, C. Bottcher, A. Quart, C. Burger, S. Dahne and S. Kirstein, J.Phys.Chem. B 104, 5255-5262, 2000 2 V. I. Prokhorenko, D. B. Steensgaard and A. R. Holzwarth, Biophysical Journal 79,2105-2120,2000 3 C. Didraga, A. Pugzlys, P. R. Hania, H. von Berlepsch, K. Duppen and J. Knoester, acceptedfor publication in J. Phys. Chem. 4 M. van Burgel, D. A. Wiersma and K. Duppen, J. Chem. Phys. 102, 20-33, 1995

881

Ultrafast molecule to semiconductor electron transfer via different anchor groups in ultrahigh vacuum R. Ernstorfer, L. Gundlach, S. Felber, R. Eichberger, C. Zimmermann, W. Storck, and F. Willig Hahn-Meitner-Institute, Dept. Dynamics of Interfacial Reactions, Glienicker StraBe 100, 14109 Berlin, Germany E-mail: [email protected] Abstract. Electron transfer in the wide band limit occurs from the donor perylene to Ti02 via the carboxyl group in 13 fs and via the phosphonate group in 28 fs under ultra-high vacuum (UHV) conditions.

1. Introduction Electron transfer from a molecular donor to a solid functioning as an electrode is a necessary ingredient of most scenarios for molecular electronic devices [1]. The molecule is attached to the surface of the solid via an anchor group. The latter forms up to three chemical bonds to specific surface atoms such that the system is stable up to temperatures much higher than room temperature [2]. Hitherto, the influence of specific anchor groups on molecule-to-semiconductor electron transfer has not been investigated. Previous experiments have addressed additional spacer groups that were inserted between the anchor group and the chromophore part of the molecule. These additional spacer groups slow down and control the electron transfer time compared to the influence of the anchor group [3]. We report here on transient absorption experiments that probed directly the influence of two specific anchor groups on the interfacial electron transfer dynamics. The two different anchor groups, i.e. carboxylic acid and phosphonic acid, were attached via a covalent bond to the same atom of the chromophore perylene. The latter was protected against dimerization on the surface of Ti02 by covalently attaching two bulky tertiary butyl groups.

2. Experimental Methods The transient absorption measurements were carried out under UHV conditions, i.e. with a base pressure in the range of 10'^^ mbar. UHV conditions offer two obvious advantages for the investigation. Firstly, the complicated energetic influence and response behavior of a solvent environment was avoided. Only the influence of the anchor group was of interest here. Secondly, tools of surface science like UPS and XPS could be applied and the energetics of the interface were determined by direct experiments carried out at the same interface that was

882

investigated with transient absorption measurements. The interface was prepared by a wet chemistry preparation procedure. Electron transfer was measured here as time-resolved rise of the absorption signal for the ionized perylene, i.e. the cation, applying the well-known pump-probe technique. Our group has shown before that the absorption spectra for the ground state, the excited singlet state, and the cationic state of the chromophore perylene are spectrally well separated from each other in the visible part of the optical spectrum. Pump (central wavelength 435 nm) and probe (central wavelength typically at 570 nm) pulse were generated with two NOPAs and the pulse duration (FWHM) was typically in the range of 15 to 17 fs. A novel feature of the experimental set-up with the two NOPAs was the low pulse energy in the range of several nJ combined with a high repetition rate, i.e. between 100 kHz and 150 kHz.

3. Results and Discussion From UPS data the energy of the molecular ground state was determined with respect to the Fermi energy of the solid. Inserting the excitation energy to the first excited singlet state of the chromophore perylene gave the location of the donor orbital and thus also the energy difference between the donor orbital and the lower edge of the conduction band. UPS showed that the latter energy difference was about 0.7 eV for each of the two different anchor groups. Thus, the molecular donor orbital of the excited singlet state of perylene was located sufficiently high above the lower edge of the conduction band of Ti02 such that the so called "wide band limit" was realized [4]. In this case the complete energy range of the FranckCondon weighted density of states, estimated for example from the envelope of the ionization spectrum of perylene, was available in parallel to the electron transfer reaction. In this limit electron transfer is controlled only by the strength of the electronic interaction and not by a Franck-Condon factor. This situation is particularly well suited for investigating the electronic coupling through different anchor groups for the same chromophore attached to the same semiconductor surface. Fig. 1 shows the transient absorption of the ionized chromophore. Electron injection via the carboxyl anchor group was very fast, the fit gave a time constant of 13 fs, about two times faster than the 28 fs measured with the phosphonic acid anchor. The rise of the cation absorption signal did not show any dependence on the probing wavelength when chosen within the absorption spectrum of the perylene cation. The faster rise for the carboxylic anchor group was accompanied also by a faster initial decay. The latter was ascribed to the disappearance of the cation state via recombination of an electron in Ti02 with the cation to form the neutral ground state of perylene. Thus, the link between faster rise and faster decay for the carboxylic group compared to the phosphonic group suggested that the electronic coupling was stronger for electron injection in one direction via the carboxylic anchor group and also for electron transfer in the opposite direction. Thus, the apparent electronic tunneling barriers are greater for electron transfer in both directions in the case of the phosphonic anchor group compared to the carboxylic anchor group. This result is non-trivial, an experimental example for a different behavior is represented by coupling through acrylic acid.

883

& o

<

DTB-Pe-COOH: TET=13^S DTB-Pe-PO(OH): x =28 fs

300

delay [fs] Fig. 1. Rise of the absorption of the ionized chromophore, i.e. electron transfer product state, for the two different anchor groups indicated in the inset along with the time constants for the rise. Hitherto, theoretical calculations of through bond electron transfer [3] have been compared only to measurements that were carried out in a solvent environment and with molecular donor-acceptor pairs where additional assumptions had to be made about the role of the Franck-Condon factors and the role of the solvent. The present type of experimental data on molecule to semiconductor electron transfer obtained in UHV in the "wide band limit" should provide a far more reliable basis for a comparison with theoretical calculations that predict the strength of the electronic coupling through a molecular bridge or bridge/anchor group. Acknowledgements. We gratefully acknowledge support by the Deutsche Forschungsgemeinschaft (DFG).

References 1

Molecular Electronics, Edited by J. Jortner and M.A. Ratner, Blackwell Science, Oxford 1997. 2 R.E. Tanner and Y. Liang, E.I. Altman, Surface Science 506, 251 (2002). 3 R. Pati and S. P. Kama, Chem. Phys. Lett. 351, 302 (2002). 4 S. Ramakrishna, F. Willig, V. May, A. Knorr, J. Phys. Chem. B 107, 607 (2003); S. Ramakrishna, F. Willig, V. May, Phys. Rev. B 62, R16330 (2000)

884

Controllability in dissociative ionization of organic molecules with pulse-shaped intense laser fields H. Yazawa\ T.Okamoto\ T. Yamanaka\ F. Kannari\ R. Itakura^, and K. Yamanouch^ ^ Department of Electronics and Electrical Engineering, Keio University, 3-14-1, Hiyoshi, Kohoku-ku, Yokohama, 223-8522, Japan E-mail: [email protected] ^ Department of Chemistry, School of Sciene, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033 Japan Abstract. Fragmentations at dissociative ionization of various molecules in intense laser fields are investigated. Some yield ratios exhibit monotonous changes against the laser pulse width. Fine pulse shape does not affect those fragmentations.

1. Introduction Adaptive control of ultrashort laser pulses in both amplitude and phase through a self-learning feedback laser system has been widely applied to study interaction of molecules with intense laser fields, where its interaction Hamiltonian cannot be fully described. However, in general, the optimized pulse shape is too complicated to interpret the physics by decoding the optimized pulse shape. More physical insights are obtainable rather through experiments by systematically changing one of laser pulse parameters such as interval of a periodic pulse train, linear frequency chirp rate, or pulse width.

2. Dependence of fragmentation on linear frequency chirping Recently, we performed experiments to study the dissociative ionization of ethanol in an intense laser field with chirped laser pulses [1]. The femtosecond pulses (central wavelength of SOOnm, optical intensity ~4 PW/cm^, Fourier transform limited pulse width of 35fs) were irradiated to ethanol molecules. The dissociated ions were measured with a Wiley-McLaren type mass spectrometer. As the linear chirp rates increases while keeping the pulse energy constant, the dissociative ionization becomes preferred to the parent ion formation, and the fragment ions such as C2H5"^ and OH"*" becomes preferred to formation of CH20H^ and CHs"^. The further increase in the chirp rate promotes the enhanced ionization into the doubly charged ethanol ions resulting in Coulomb explosion either at C-C or C-0 bond. Figure 1 shows the dependence of the yield ratio of [C2H5"^]/[CH20H'*"] upon the linear frequency chirp rate. The yield ratio increases as the chirp rate increases up to 2.7x10'^ ps^. As the linear chirp increases, the

885

ratio [(CH20H^)*]/[CH20H^] also increases by a factor of 10. Here, the fragment ions produced from the Coulomb explosion from doubly charged ethanol ions are denoted with an asterisk. 0.9 0.8 0.7 0.6 0.5 0.4

...

^ , V

0.2 - 0 08

-• ••• .

• t

0.3

-0.06

-0.04

-0.02

0

0.02

0.04

0.06

0.0

Linear chirp ra te (ps'ps)

Fig. 1. The chirp dependence of the yield ratio of [C2H5'"]/[CH20H'']

In these fragmentation processes, the direction of the chirp does not influence the relative yields of the fragment ions, suggesting that the dynamics in the intense laser field is primarily governed by the temporal width of the laser field.

3. Adaptive control of pulse shape We have applied self-learning system to adaptively control excitation laser pulse shapes so that the laser pulse promotes specific dissociation channels. We used a Simulated Annealing algorithm to control a liquid crystal-spatial light modulator (LC-SLM) arranged in a 4-f optical setup, which was placed before a regenerative amplifier. By referring to a ratio of specific fragment signals, such as[C2H5"']/[CH20H'-] or [ ( C H 2 0 H O * ] / [ C H 2 0 H ^ ] , the pulse shape was iteratively optimized so that the targeted fragment peak became highest or lowest. The selflearning feedback system successfully converged typically after -1000 iterations. However, the yield ratios are always almost the same as that obtained by increasing the linear frequency chirp rate. Moreover, the optimized pulse shapes via the adaptive control vary at each the optimization run. The only common feature among the optimized pulse shapes is their long pulse duration. Figure 2 shows four different pulse shapes. They produce almost the same yield ratio of [C2H5^]/[CH20H^]. Figures 2(a) and (b) were the pulse shapes generated via adaptive control. Figure 2(c) is a pulse shape output from the regenerative amplifier, where the seed pulse was shaped by the pulse shaper with a linear frequency chirp rate of 2.7x10"^ ps^. The fine modulation in the amplitude was caused due to the finite size of liquid-crystal segments in LC-SLM. Such higher linear chirp rates cannot be attained in a smooth phase function with the segmented SLM. Figure 2(d) is a pulse stretched by adjusting the pulse compressor to impose a linear frequency chirp rate of 2.7x10" ps . It is surprising that these four pulses produce almost the same fragmentation for ethanol. The nuclear wave packet of ethanol cannot response to the fast changes in the optical field. The nuclear wave packet can be guided to a certain dissociation channel as long as the intense optical field is maintained, which sustains the modified reaction potential surface.

886

Fig. 2. Four different pulse shapes that generate the same yield ratio of [C2H5'^]/[CH20H"^]. The upper two pulses were generated via adaptive control. The lower left is a chirped pulse generated by the pulse shaper placed before the amplifier. The lower right is a chirped pulse at the same chirp rate generated by adjusting the pulse compressor. We repeated the similar experiments for 1-propanoic. When comparing with mass spectra obtained with a transform limited pulse, linearly chirped pulses, there are always significant difference between that obtained with a transform limited pulse and that with chirped long pulses. However, there is less difference between those obtained by the chirped pulses with opposite chirp signs. In fact, when we systematically varied the chirping rate, all of the yield ratios of fragments indicate the V-shape functions similar to that shown in Fig.l. Therefore, the mechanism of dissociative ion generation of 1-propanoic is also dependent on duration of electric field.

4. Conclusions Since the unclear wave packet of moleculer does not respond to fast change in amplitude or phase of optical pulse, dissociative ionizations of ethanol and 1propanole do not exhibit strong dependence on the excitation pulse shape. Their ionization fragmentation are controllable only by the duration of optical field maintaining the dressed state reaction surface to guide the wave packet toward energy level crossing for dissociation.

References 1 R. Itakura, K. Yamanouchi, T. Tanabe, T. Okamoto, and F. Kannari, "Dissociative ionization of ethanol in chirped intense laser fields" , J. Chem. Phys. 119, 4179 (2003). 2 T. Tanabe, M. Yamanaka, T. Okamoto, and F. Kannari, "Compensation for a transfer function of a regenerative amplifier to generate accurately shaped ultrashort pulses in both the amplitude and phase" , IEEE J. Selected Topics in Quantum Electron.

887

Laser Coulomb explosion imaging for probing molecular structure and dynamics Frangois Legare^'^ Kevin F. Lee^'^ I.V. Litvinyuk^'^ P.W. Dooley^'^ A.D. Band^auk^ D.M. Villeneuve\ P.B. Corkum^ ^National Research Council of Canada, Ottawa,Ontario, Canada KIA 0R6 ^Departement de Chimie, Universite de Sherbrooke, Sherbrooke, Quebec, Canada JIK 2R1 ^Department of Physics & Astronomy, McMaster University, Hamilton, Ontario, Canada KIS 5P3 "^Department of Physics, Kansas State University, Manhattan, Kansas, 66506, USA Abstract: Using the molecular clock, we time-resolved the double ionization of D2. When 8 fs pulse -10^^ W/cm^ are used, there is a 4 fs delay between the first and the second ionization. We measure molecular structure of D2O and SO2 with -O.S-Angstrom resolution using laser Coulomb explosion imaging.

1. Introduction Coulomb explosion imaging requires ionizing molecules to a high charge state in a very short time [1]. Using conventional femtosecond pulses, it was only possible to measure structure and dynamics of system with heavy atomic constituents such as iodine using Coulomb explosion. Using nuclear motion (a "molecular clock" [2]) to time-resolve the ionization of D2 molecules, we show that --8 fs optical pulses can remove both electrons within -- 4 fs. Such an ionization rate is fast enough to inertially confine virtually any molecule or molecular ion during its ionization phase [3]. Thus, sub-Angstrom spatial resolution combined with sub-5 fs temporal resolution is available using Coulomb explosion for measuring the dynamic molecular structure of small molecules. Showing this is the main purpose of this paper.

2. Experimental methods To determine the structure of polyatomic molecules using Coulomb explosion imaging, the correlated velocity vectors of all fragments must be measured. Using coincidence imaging [4] and 8 fs laser pulses, we measured those vectors. For each event in the correlated momentum distribution, we find a threedimensional structure that reproduces the measured fragment velocities.

3. Results and discussion We begin with 3-D structure, but no dynamics. Using coincidence imaging and 8 fs laser pulses, we measured those vectors for the explosion of D2 (D2^^ -^ D^ +

888

DO, D2O (D20^^ -> D^ + O^^ + DO and S02(S02^^ ^ O^^ + S^' + O^O- In Fig. 1(A), we present the correlated D"^ kinetic energy spectra. On this figure, using the concept of molecular clock [2], we show the expected kinetic energy spectra if nuclear motion occurs for 2.66 fs and 5.22 fs on D2'^ ground electronic state. Our experimental spectra peaks in between this two curves showing that the double ionization of D2 is complete in within 4 fs. Enhanced ionization is completely suppressed [5,6]. The intensity of the pulse is 2x10^^ W/cm^ and the regime of double ionization is sequential. In the low intensity regime, < 5x10^"^ W/cm^, non-sequential ionization induced by electron recollision [7] becomes the predominant channel for double ionization like for attosecond pulse generation [8]. The first recollision is dominant (see Fig. 1(B)).

1.0

I 0.8

5.22fs^:i/ .

I 0.6 5 0.4 .2) 0.2 CO

> M*-T

0.0 • ^ - T " » " T

1.0^

1 0.8-1 2 0.6J

5 0.4 &0.2' CO

0.0 2

Ai

.

(B)

J:

4 6 8 10 12 D* kinetic energy / eV

14

Fig. 1: (A) Experimental kinetic energy spectra for the explosion of D2 (D2^'' -> D"*- + D"*-) with 8 fs pulse -2x10^^ W/cml Square: experimental. Dotted: expected kinetic spectra if there is 2.66 fs of delay between the first and the second ionization. Shord dot: 5.22 fs of delay. (B) Pulse intensity -2x10^"^ W/cm^. Dotted: expected kinetic energy spectra for first recollision (1.8 fs). In Fig. 2, we present the measured structure for D2O. We exploded D2O by using 8 fs pulse with intensity of 5x10^^ W/cm^. The highest charge state observed is D20^'*^ but the count rate observed is too low to present structure with reasonnable statistics. The results presented in Fig. 2 are obtain by using D20'*^ -> D^ + O^"^ +

889

D^. These experiments show that we have the possibility of measuring 3dimensionnal structure with reasonnable resolution for pump-probe measurement. On the bottom curve of Fig. 2, we present radial and bond angle distribution from our experiments and expected from harmonic oscillator approximation. The spatial resolution of our measurement is about 0.3 Angstrom which will be sufficient for imaging transient structure of small molecules during molecular processes such as dissociation and important molecular rearrangement like isomerization and proton transfer.

(a)

^

\ ' I—'-"T" • 2 - 1 0 1 2 X (Angstrom)

(c)

(b)

:A 1.0

1,5

R^,,, (Angstrom)

2.0

oO

90

120

150

180

Angle {degree?

Fig. 2: (a) Structure of D2O using the 020^'' charge states (D20^^ -> D"^ + O^"*^ + D"*^). The reconstruction is achieved using the Coulombic potential. The center of mass is at x=0, y=0, and the y axis is the bisector of the angle, (b) Radial distribution, (c) Angular distribution. In (b) and (c), the dotted curve represents what we should expect for the v=0 stationary state structure of D2O. References: 1. 2. 3. 4. 5. 6. 7. 8.

890

Z. Vager, R. Naaman, and E. P. Kanter, Science 244, 426 (1989). H. Niikura, F. Legare, R. Hasbani, A.D. Bandrauk, M. Yu Ivanov, D.M. Villeneuve, and P.B. Corkum, Nature 417, 917 (2002). F. Legare, I. V. Litvinyuk, P.W. Dooley, F. Quere, A.D. Bandrauk, D.M. Villeneuve, and P.B, Corkum, Phys. Rev. LeU. 91, 093002 (2003). A. Hishikawa, A. Iwamae, K. Hoshina, M. Kono, and K. Yamanouchi, Chem. Phys. Lett. 282, 283 (1998). T. Seideman, M. Yu Ivanov, and P.B. Corkum, Phys. Rev. Lett. 75, 2819 (1995). T. Zuo et A. D. Bandrauk, Phys. Rev. A 52, R2511 (1995). P.B. Corkum, Phys. Rev. Lett. 71,1994 (1993). A. Baltuska, Th. Udem, M. Uiberacker, M. Hentschel, E. Goulielmakis, Ch. Gohle, R. Holzwarth, V.S. Yakovlev, A. Scrinzi, T.W. Hansch, and F. Krausz, Nature 421, 611 (2003).

Ultrafast electron transfer via a bridge-extended donor orbital R. Ernstorfer, L. Gundlach, S. Felber, W. Storck, R. Eichberger, C. Zimmermann, and F. Willig Hahn-Meitner-Institute, Dept. Dynamics of Interfacial Reactions, Glienicker StraBe 100, 14109 Berlin, Germany E-mail: [email protected] Abstract. Electron transfer from the excited aromatic donor perylene to Ti02 occurred with 10 fs time constant via the conjugated -CH=CH- bridge unit compared to 57 fs in the presence of the saturated -CH2-CH2- bridge unit.

1. Introduction Transport of electrons through molecular bridges (wires) of different chemical nature is an important topic for the postulated field of molecular electronics [1], Hitherto, much of the work in this area has focused on saturated aliphatic chains or on ring structures with saturated C-C bonds, however, some theoretical papers and some of the experiments have dealt also with conjugated molecular bridges [2]. So far, all the theoretical work has addressed molecular bridges in ultra-high vacuum (UHV), whereas all the measurements have been carried out in a solvent environment. We report here on transient absorption experiments probing electron transfer via different molecular bridges in UHV. Electron transfer occurred from the excited singlet state of the aromatic chromophore perylene to nano-sized particles of anatase Ti02. The chromophore was linked covalently to a carboxylic group, which binds to surface Ti atoms and thus functions as the anchor group. Optionally, bridge groups were inserted in-between the perylene backbone and the anchor group. Three different experimental systems were compared: firstly, no bridge, secondly a conjugated -CH=CH- bridge unit, and thirdly a saturated -CH2CH2- unit. Bulky side groups were attached to the perylene chromophore (2,5-bitertiary-butyl-perylene-9-yl, DTB-Pe) to prevent the formation of dimers by neighboring perylene chromphores on the surface. A particularly attractive feature of the perylene/Ti02 system is the validity of the so called "wide band limit", which is realized if the molecular donor level is energetically far above the conduction band minimum compared to the width of the electron transfer spectrum [3]. In this regime the electron transfer time is controlled by the strength of the electronic coupling and not by the magnitude of individual Franck-Condon factors [3]. UPS measurements revealed that for all three investigated systems the donor level is at least 700 meV above the conduction band minimum. The electron transfer time was determined from the time-resolved rise of the absorption signal of the ionized perylene at around 570 nm, i.e. the molecular product state of electron transfer. Our group has shown before that the absorption

891

'

'

'

'

I

1

1

T

» - |

1

•

•

1

1

•

1

1

•

1

•

•

1

•

1 0•W^^ ?5

/A^^^^^^

'"^^"^

0.80.6-

-§ •i

Jr K

040.2-

or i4

O G

DTB-Pe-COOH, .^= 13 fs DTB-Pe-CH=CH-COOH, "^=10fs

A

DTB-Pe-CH-CH-COOH, . .= 57 fs

inj

2

0.0-

-50

2

' uy

pump: 440 nm, probe: 570 nm -

50 100 delay [fs]

150

200

Fig. 1. Rise of the absorption of the product state of heterogeneous electron transfer, i.e. the cation of perylene, for DTB-perylene attached to Ti02 via different bridge-anchor units. The fit to the rise of the pump-probe signals showed an injection time of 10 fs for CH=CH-COOH, of 57 fs for -CH-CH-COOH, and of 13 fs for -COOH. spectra for the ground state, the excited singlet state, and the cationic state of the chromophore perylene are spectrally well separated from each other. The laser system is described in another contribution to this issue. Briefly, pump (central wavelength 435 nm) and probe (central wavelength typically at 570 nm) pulses were generated with two NOPAs at a repetition rate of 150 kHz and the crosscorrelation (FWHM) was typically in the range of 25 to 30 fs.

2. Results and Discussion Fig. 1 shows the time-dependent absorption of the perylene cation for the three investigated systems. There are characteristic differences in the rise (electron injection) for the two different bridge units. Electron injection via the carboxyl anchor group alone gave a time constant of 13 fs, whereas 10 fs were determined when the -CH=CH- bridge unit was inserted. In contrast, electron transfer in the presence of the -CH2-CH2- bridge unit was found to be much slower with a time constant of 57 fs. The spatial separation of the perylene ring structure from the Ti02 surface was most likely comparable for the conjugated and the saturated bridge units. The latter might even allow for configurations with a shorter distance. The extension of the donor orbital onto the conjugated bridge can be seen in Fig. 2 (center) as derived from a semi-empirical calculation. The perylene SQ-SI transition has strong HOMO-LUMO character [4] so that the LUMO wavefunction can be considered a good approximation to the excited state wavefunction. Fig. 2 refers to the free molecule, whereas in the actual system the molecule was attached to the surface of Ti02. The proximity of empty isoenergetic electronic states in the

892

'ft'

,. • • • « ' • •

'

«^

,

,

''«'••'

0t

it ?§,

'm m

Fig. 2. LUMOs of DTB-Perylene with three different bridge-anchor groups: -COOH (left), -CH=CH-COOH (center), and -CH2-CH2-COOH (right). The energy of the LUMO for the neutral molecule is sufficiently close to that of the conjugated bridge and has pure K* character. The orbital representing the donor wavefunction extends onto the sp^-hybridized conjugated bridge unit (center) but not onto the saturated bridge unit (right). Wavefunctions calculated with ZINDO/S on a ZINDO/1 optimized geometry. Ti02 conduction band can lead to a further extension of the excited state donor orbital right into the electronic states of the solid. Thus, a direct optical charge transfer contribution appears possible that is not included in the picture derived for the donor orbital of the isolated neutral molecule shown in Fig. 2. Nevertheless, the very different extensions of the chromophore's donor orbital, i.e. onto the conjugated bridge but not onto the saturated bridge, are clearly borne out by the different chemical structure of the bridge units. For electron injection the latter functioned as insulator whereas the conjugated bridge as conducting molecular wire. Acknowledgements. We gratefully acknowledge support by the Deutsche Forschungsgemeinschaft (DFG).

References 1

Molecular Electronics, Edited by J. Jortner and M.A. Ratner, Blackwell Science, Oxford 1997. 2 R. Pati and S. P. Kama, Chem. Phys. Lett. 351, 302 (2002); W. B. Davis, M. A. Ratner, M. R. Wasielewski, Chem. Phys. 281, 333 (2002). 3 S. Ramakrishna, F. Willig, V. May, A Knorr, J. Phys. Chem. B 107, 607 (2003); S. Ramakrishna, F. Willig, V. May, Phys. Rev. B 62, R16330 (2000). 4 T. M. Halasinski, J. L. Weisman, R. Ruiterkamp, T. J. Lee, F. Salama, M. HeadGordon, J. Phys Chem. A 107, 3660 (2003).

893

Multiple Ionization of Atoms by 25 and 7 fs Laser Pulses A. Rudenko, B. Feuerstein, K. Zrost, V.L.B. de Jesus, CD. Schroter, R. Moshammer and J. Ullrich Max-Planck-Institut fiir Kemphysik, Saupfercheckweg 1, 69117 Heidelberg, Germany e-mail: [email protected] Abstract. Double, triple and fourfold ionization of rare gas atoms has been studied using a "reaction microscope". We show that ionization dynamics drastically depends on atomic species and present the first data obtained with few-cycle pulses. Since the first experimental observation of doubly charged ions produced by intense laser field [1], many-electron dynamics in laser-induced ionization has been the subject of numerous experimental and theoretical studies. It was soon found out that the yields of second and higher charge states produced by intense linearly polarized laser pulses exceed significantly those expected for independent successive removal of two or more electrons. The origin of this enhancement (which was called "non-sequential" ionization) remained controversial until 2000 when first measurements of recoil ion momentum distributions [2,3] along with the results of earlier experiments using circularly polarized light [4] provided convincing evidences in favour of the so-called "recollision" model. Since then, for double ionization ion momentum distributions, coincident photoelectron energy spectra and correlated electron momenta were measured for different rare gases in a broad range of experimental parameters. However, the details of the double ionization dynamics are still far from being completely understood. In particular, one of the most intriguing quesdons is whether and how atomic structure can influence ionization dynamics [5]. For the case when more than two electrons are ionized, there is much less experimental results available. Intensity dependence of ion yield was measured for different rare gases (see, e.g. [6]). Ion momentum distribution obtained in [3] for Ne^"^ at 1.3 PW/cm^ remains, to the best of our knowledge, the only differential data published up to now. Here we report on the systematic experimental study of multiple (up to fourfold) ionization of Ar and Ne by intense laser field and present the first results obtained with the laser pulses as short as 7 fs. The experiments were performed using a new "reaction microscope" [7] designed to meet the specific requirement of the experiments with high-intensity lasers. We used linearly polarized radiation of a Kerr-lens mode locked Tiisapphire laser at 795 nm wavelength with 25 fs pulse width (FWHM). To generate few-cycle pulses they are spectrally broadened in a gas-filled hollow fiber and then compressed to 6-7 fs (FWHM) by chirped mirrors and a prism compressor.

894

Fig. 1 shows momentum distributions of Ne^'^"^ ions along the laser polarization direction. All spectra exhibit a clear double peak structure which is a signature of the recollision process In the momentum distributions of Ne"^"^ and Ne"^"^ ions this structure is even more pronounced than for double ionization, with almost no ions produced with zero momentum. Momentum distribution of Ne^"^ ions created by 7 fs laser pulse (solid line in Fig.la) closely follows the results obtained with 25 fs pulses, indicating that momentum distribution is mainly defined by the recollision within one or two cycles after the removal of the first electron. The widths of the spectra are in good agreement with kinematical constraints following from the classical consideration [8]. In general, confirming the results of [3] we can state that ion momentum distributions of Ne""^ ions agree well with the model assuming direct impact ionization by rescattered electron via (e,ne) process.

P„, a.u.

Fig. 1. Longitudinal momentum distributions of Ne"'*' ions. Arrows indicate ion momenta of ± 2n JiJ^ However, considering the momentum distributions of Ar^'"*"*" ions (Fig. 2), we can conclude that for Ar other mechanisms play an important role. In longitudinal momentum distributions of Ar^"'^"^ ions obtained with 25 fs pulses (squares in Fig.2a,b) minimum at zero momentum is pronounced much less than in the case of Ne. For Ar"^"^ ions created by 25 fs laser pulse we do not observe double peak structure at any intensity! Such 'filling of the valley' can be attributed (at least for the case of double ionization) to the contribution of electron-impact excitation during recollision followed by field ionization of the exited ion in one of the subsequent cycles of the oscillating laser field.

Fig. 2. Longitudinal momentum distributions of Ar""^ ions. Arrows indicate ion momenta of ± 2h Jij^

895

Here the ionization step occurs close to the maximum of the field, leading to relatively small ion momenta. Analysis of the ratios of collision-induced ionization/excitation cross sections shows that for Ne direct electron impact ionization clearly dominates, whereas for Ar (as well as for He) the mechanism involving excitation plays a decisive role [5]. The contribution of the latter mechanism should be suppressed for a shorter laser pulses. This is in a good agreement with the data we have obtained with 7 fs pulses. Comparing the results obtained for Ar with two different pulse lengths, we see that the fraction of ions with zero momentum is considerably reduced for a 7 fs pulse. In this case the minimum at zero momenta is more pronounced for double and triple ionization, and can be also observed for Ar"^^ ions. The same depletion of the region around zero momentum can be also caused by the suppression of the sequential ionization from the ground state. However, the contribution of purely sequential production of Ar^\ Ar'^'^and Ar"*"^ at corresponding intensities is still a small fraction of the total yield [9]. Nevertheless, for the production of triply and fourfold charged ions there is a possibility to reach the final state via a combination of sequential and nonsequential processes, and this dynamics might be changed for a few-cycle pulses. Contribution of such combined processes can also explain the relative narrowing of Ar^"*" and Ar^"^ momentum distributions (compare the position of the arrows at Fig.2 and 3) [8]. In conclusion, we have studied multiple ionization of Ar and Ne by 25 and 7 fs laser pulses. Whereas for all charge states of Ne we have observed ion momentum distributions consistent with the kinematics of direct (e,ne) process induced by the electron recolliding with its parent ion, for Ar both the shape and the width of the spectra indicate more complex process with different mechanisms being of importance. For double ionization recollisioninduced excitation can explain the differences observed. This explanation is strongly supported by the results obtained with 7 fs laser pulses. For higher charge states more elaborated description, which considers the possibility of multiple excitations and different combinations of sequential and nonsequential processes, is required.

References [1] V.V. Suran and LP. Zapesochny, Sov. Phys.- Techn. Phys. Lett. 1,420 (1975). [2] Th. Weber et a/, Phys. Rev. Lett. 84,443 (2000). [3] R. Moshammer et al, Phys. Rev. Lett. 84,447 (2000). [4] D.N. Fittinghoff ^r al, Phys. Rev. A 49, 2174 (1994). [5] V.L.B. de Jesus et al, J. Phys. B 37, LI 61 (2004). [6] S. Larochelle, A. Talebpour and S.L. Chin, J. Phys. B 31, 1201 (1998). [7] V.L.B. de Jesus et al, J. Electron Spectrosc. Relat. Phenom., to be published. [8] B. Feuerstein, R. Moshammer, and J. Ullrich, J. Phys. B 33, L823 (2000). [9] A. Becker and F.H.M. Faisal, J.Phys.B 32, L335 (1999).

896

Time Resolved, Phase-Matched Harmonic Generation from Exploding Noble Gas Clusters Bonggu Shim, Greg Hays, Mykhailo Fomyts'kyi, Alex Arefiev, Boris Breizman, Todd Ditmire, and Michael C. Downer Department of Physics, University of Texas at Austin, Austin, Texas 78712, USA E-mail: [email protected] Abstract. Third-harmonic generation from noble-gas clusters by ultrafast probe pulses is more sharply-enhanced than linear absorption following heating by an ultrafast pump pulse, in good agreement with simulations of cluster expansion and collective electron dynamics with phase matching consideration. Recent experiments [1,2] have shown that the linear absorption and refractive index of van der Waals bonded clusters are resonantly enhanced at time delays Atcrit ~ 1 ps following excitation by a short pump pulse, as illustrated by our timeresolved linear absorption data for Ar clusters in Fig. la. This resonance occurs as the natural plasma frequency cjp of the expanding clusters drops through the probe laser frequency co, and thus occurs earlier for smaller clusters and/or higher pump intensity. The resonance is quite broad, spreading over ~ 1 ps, prompting Milchberg et al [3] to propose that a moving surface critical density layer, rather than a bulk collective resonance [4], dominates the linear optics of the expanding clusters. We have conducted experiments aimed to measure the third order susceptibility of the plasma inside of a cluster as a function of density using a pump(400nm,200fs)-probe(800nm,80fs) scheme at a small angle (--1^ with respect to the pump propagation direction and generated a 3 ^ probe signal ( ^ = 266 nm) (Fig. lb). By varying the time delay between the pump and probe beams, we have found evidence of a resonance condition in the third harmonic production that is different from linear absorption cases (Fig. lc).For given pump intensity, the THG resonance occurs earlier than the linear absorption maximum (Fig. la), as expected for cj^ resonance with the higher frequency 3 ^ . In addition, and quite interestingly, it is much narrower than the linear absorption resonance, suggesting the dominance of collective electron dynamics with phase matching consideration. a)

1

b)

Pump lnt«nsity:8 1 8 X 1 0 " W / c m ' Probe lnt»nsity:7 3X 1 O " W/cn-. '

p;;j].p;;'r{j>-;.^'-Ar 400 psi

O.iO O O.iSQ.0JO. O 0*1

1

"^jr^mx

too

•x:r I'O'GD

Delay(fs)

190)0

c)

pump 1 .0 -

Cluster W) 0.8 -

A SFG-

d \

THG

tr U

0.4 0.2 -

1

]

^FWM

•400

-200

6 ,

200.

400

«00

Delay(fs)

897

Fig. 1. Pump-probe experiments and results from argon clusters, a) Time-resolved linear absorption of probe pulse (800nm, 80 fs) following excitation of Ar clusters by pump pulse (200 fs, 400nm). Higher backing pressure corresponds to larger clusters (800 psi: R - 21 nm; 600 psi: R ~ 17nm; 400 psi: R ~ 13 nm). b) Non-coUinear pump-probe scheme used for results in c), showing signals at ^ probe/3 resulting from Third Harmonic Generation (THG) by the probe, Sum Frequency Generation (SFG) and Four Wave Mixing (FWM) by the pump and the probe at zero delay, which are separated spatially by momentum conservation, c) Delayed third harmonic resonant enhancement from Argon clusters (800 psi., R~ 21nm), and SFG peak at zero delay.

The simple model calculation based on references [4,5,6] shows that the coherence length (Lcoh) for THG, normally very small ( <|im), peaks sharply as clusters expand and the SOOnm and 266nm light become phase matched. During this short peak (~ several fs), Lcoh becomes comparable to the collinear interaction length (~lmm), greatly enhancing conversion efficiency of the THG. In order to verify the phase matching theory, WQ made the overlap length (Loverlap) small (~30^m) by using a large angle (--20^ pump (800nm)-probe (SOOnm) geometry. Compared with the nearly collinear case (~1^) (Fig.2.a), the THG resonance enhancement for the large angle case is greatly suppressed because the interaction length is quite small (Fig. 2b). The delayed resonance is also quite sensitive to the position of the pump-probe focus within the cluster jet, which is also consistent with phase-matching sensitivity. Therefore, the fact that only phase matched THG signal is enhanced even though the THG resonance itself has long time duration explains the narrow peak of THG resonance. ^)

b) 13nm Cluster

21nm cluster

Pump Intensity; 9X i 0 •' W/<: m ^ Prob» IntftnsHy: 2X10 •'- W/cnt-'

to 9-

• •

...*,.. At

•, *

.... HeFYv-M

Ik

dw-Mtf

7 '

iiiff

z

a-

Small ^ntx^nc^xnent

large enhanceff^ertt -«0o

zm

o

'^<:ii> 4«x)

»oc(

-^00

-aQ'J

O

2?>0

400

yoo

800

Fig. 2. Comparison between a small angle(~l^) and a large angle (~20^) pump-probe experiments, a) The second peak (THG) is quite pronounced for a nearly collinear case {Loverlap ~lmm). There is no second peak for He gas (dotted line) because of no cluster resonance, b) The enhancement for the phase matched THG resonance is suppressed. There is no SFG and 4WM peak near zero delay because of 800-800 pump-probe interaction. The pump intensity was 9xl0'^ W/cm^ and the probe was 2xlO^^W/cml

The pump intensity scan with fixed probe intensity (Fig. 3) shows that at lower pump intensities, the SFG and 4WM processes peak at zero delay, whereas the THG resonance peaks at a later delay and is well separated from the first peak. At

898

higher pump intensities, the two peaks merge because the clusters expand more rapidly.

— Putnp Inl&nsity 8.3X10'* W^cm— Pump Inlensity 2.0X10" W/cm^

lies liOO-

4

tMD1103: 1

mjl^THG

lOO-

IIOO.

*»•

1

•i ^I

1 lOD .

SB 1 too• ^ MOO-

/

>

ICD-

C^

too. lOOlOD-

*•*!

•i

• » » • 100 •

i \ THG

.:.:»^-^ ^ - " S . , . .

Kfi' a•400

-»0

D

100

Delay(1s)

«I0

SM

Fig. 3. Pump intensity scan with Argon clustered jet (R~21nm).When the pump was 8.2x10'^ W/cm^ two peaks were well separated. At the higher intensity (2x10^^ W/cm^), two peaks merge because of faster expansion of clusters. The probe intensity was 7x10^^ W/cml

Theoretically, we developed a PIC code for describing an individual fully ionized cluster [8].Third hannonic generation results from nonlinear response of the cold electron core [7] in the potential of the positively charged ion background. A more comprehensive model describing the macroscopic nonlinear optical properties of the cluster gas medium is being developed. This model will address the phase matching issue in the non-collinear geometry and the dependence of the third harmonic signal on the relative polarization and angle of the pump and the probe. Acknowledgements. This work is supported by Department of Energy Grant No. DEFG03-96-ER-40954 and NSF Physics Frontier Center Grant No. PHY0114336.

References 1 2 3 4 5 6 7 8

J. Zweiback, T. Ditmire, and M. D. Perry, Phys. Rev. A, 59, R3166 (1999).. K. Y. Kim, I. Alexeev, E. Parra, and H. M. Milchberg PRL, 90(2), 023401 (2003). H. M. Milchberg, S. J. McNaught, and E. Parra, Phys. Rev. E, 64, 056402 (2001). T. Ditmire, T. Donnelly, A.M. Rubenchik, R.W. Falcone, and M.D. Perry, Phys. Rev. A, 53, 3379 (1996). T. Tajima, Y. Kishimoto, and M. C. Downer, Phys. Plasmas, 6, 3759 (1999). J. W. G. Tisch Phys. Rev. A, 62, 041802(R) (2000). B. N. Breizman and A. V. Arefiev, Plasma Physics Reports, 29(7), 593(2003). M. Fomytskyi, B. Breizman, A. Arefiev, and C. Chiu, "Harmonic generation in clusters", to be published in Physics of Plasmas.

899

Index of Contributors

A Abdolvand, A. 810 Abe,M. 613 Adachi, M. 49,158 Adachi, S. 64,563 Aiba,M. 328 Akagawa, T. 115 Akahane, K. 257 Akturk, S. 97, 109, 112 Albrecht,M. 316 Alexandrou, A. 575, 628, 634 Almasi,G. 786 Amann,M.-C. 729 Andrade, L.H.F. 316 Andreev, A.A. 225 Anfinrud,P. 581 Aoshima, S.-i. 106,219 Apkarian, V.A. 377 Apolonski, A. 25 Arabski, J. 316 Aranson, I. 325 Archundia, L. 118 Arefiev, A. 897 Arisholm, G. 61 Armstrong, M. R. 816 Ashihara, S. 76 Atas, E. 456 Attwood, D.T. 189 Averbukh,I.S. 819 Averitt,R.D. 269,313,325, 696,726

B Bachilo, S.M. 331 Backus, S. 40, 175, 192 Baldacchini, T. 807 Baltuska, A. 8, 73, 94 Bandrauk, A.D. 888 Bargheer, M. 304 Baronavski, A.P. 437 Bartels,A. 837,840

Baum,P. 79, 130 Baumert, T. 864 Beaurepaire, E 316 Beers, J.D. 236,298 Beigang, R. 661 Belabas, N. Beljonne, D. 281 Bell,A.F. 610 Benabbas, A. 664 Bergman, D.J. 673,676 Bergquist, J.C. 840 Bhat,R.D.R. 287 Bi,Z. 837 Biegert,J. 22,61,204 Bigot, J.-Y. 316,664 Blank, D. A. 413 Blome, C. 170 Blums, J. 170 Boatner, L.A. 792 Bolivar, P.H. 690,741 Bonora, S. 207 Borca, C.N. 566 Botan,V. 443 Bower, J.E. 231 Bowes, B.T. 334 Branciard, C. 231 Braun, M. 462 Bredenbeck, J. 542 Breizman,B. 897 Brixner,T. 554,670,864 Brodschelm, A. 729 Bruck, M. 204 Bruner, B.D. 407 Buck,M. 260 Bucksbaum, P.H. 100 Buckup, T. 368 Bulaevskii, L.N. 325 Bulanov,S.V. 85,216 Buma,T. 231 Buma,W.J. 548 Buntinx, G. 465 Butkus,R. 8,73

C Cavalleri,A. 170,343,346, 792 Cerullo,G 363,804 Chatel,B. 70,91,861 Chau,K.J. 708 Chen,C.-W. 103 Chen,C.-Y. 723 Chen,H.-T. 693 Chen, J. 640 Chen,Y. 876 Chi,X. 269 Chiba,H 526 Chiodo,N 804 Cho,B.M. 622 Cho,G.C. 693 Cho, S. 517 Chollet,M. 771 Chong,H.H.W. 343,346 Chosrowjan, H. 604 Christov, LP. 40, 175, 192 Chrysostom, E.T.-H. 496 Chung, H.S. 551 Cina,J.A. 514,523 Clarke, S.A. 149 Clokie, C.M.L. 816 Cogdell,R.J. 363,368 Colonna, A. 616 Comstock,M. 870 Constantinescu, A.M. 242 Corkum,P.B. 139,155,164, 855,888 Cormier, E. 124 Corner, L. 124 Couairon, A. 22 Cowan, M.L. 407 Cringus,D. 386,431 Cundiff,S.T. 287,566 Czycholl,G. 260

D Daido,H. 85,216

901

Daly, Brian C. 231 Daniel, C. 281 Danielius, R. 8, 73 Dantus, M. 870 DeCamp, M.F. 422 De Flores, L.P. 422 de Jesus, V.L.B. 894 De Silvestri, S. 210 DeWaele,V. 465 Dechant, A. 658 Degert,J. 861 Degiron, A. 664 Dekorsl^, T. 346 DelaCruz,J.M. 870 Delfyett,P.J. 118 Delia Valle, G. 804 Demsar, J. 269,726 Dexheimer, S.L. 331 Diddams, S.A. 837,840 Didraga,C. 879 DiMauro, L.F. 124 Ditmire, T. 897 Dombi, P. 25 Domcke,W. 613 Dooley,P.W. 888 Dorrer, C. 127 Downer, M.C. 334,897 Dreyer,J. 407,448,485 Duppen,K. 679,879 Dwyer, J.R. 144

Fernandez-G., A. 25 Feuerstein, B. 894 Feurer,T. 152,236,239,298, 569,717 Fischer, M. 867 Fleming, G.R 331,425,554, 586 Fomyts'kyi, M. 897 Forstner, J. 19 Fortier, T.M. 287 Fourkas, J.T. 807 Fourmaux, S. 343 Franzen, S. 634 Eraser, J.M. 628 Friend, R.H. 278 Fromme, P. 586 Fuji,T. 8,25,225,383 Fujieda, M. 846 Fujimoto, J.G. 801,822,876 Fujimoto, M. 106,219 Fujimura,Y. 508,511,613 Fujiwara, K. 526 Fukaya, S. 771 Fukuda,Y. 322,482 Furbach, A. 25 Furue,S. 322,482 Furuichi, T. 257 Furukawa, N. 133 Furumochi, H. 873 Fushitani, M. 304

Eaves, J.D. 410,535 Ebbesen,T.W. 664 Edler,J. 401 Efimov, A. 52 Eichberger, R. 882,891 El Quenzerfi, R. 732 Elezzabi, A.Y. 658, 667, 708 Elsaesser, T. 292, 389, 407 Ema,K. 295 Emstorfer,R. 882,891

Gaarde, M. 204 Gabolde,P. 109 Ganim, Z. 551 Garcia de Abajo, F.J. 670 Ge,N.-H. 545 Geissler, P.L. 535 Gerber,G. 178,505,864 Gershgoren, E. 239 GeBner, O. 496 Gibson, E. A. 40,175,192 Giessen,H. 19 Gilch,R 462 Girard,B. 91,816,861 Glover, T.E. 343 Goldschleger, I. 377 Gong, Q. 354 Gorbunov, V.A. 502 Goto,M. 702,720,732 Goto,T. 31,34 Gotoh,H. 705 Gould, I.R. 622 Goulielmakis, E. 8,94 Graefe, O. 864

Fainberg, B.D. 502 Fallnich, C. 819 Falvo,C. 401 Fang,X. 67 Farrer, R.A. 807 Farrow, D. 380 Fecko,C.J. 410,535 Felber, S. 882,891 Feldman,L.C. 792 Fermann, M.E. 843

902

Graener,H. 392,810 Greenham, N.C. 783 Groma, GI. 616 Grondelle, van, R. 592, 601, 610 Gu,X. 97 Guerin, L. 771 Guhr,M. 245,304 Guidoni, L. 664 Gundlach,L. 882,891 Gutowski, J. 260

H Hagihara,Y. 526 Hagimoto, K. 831 Hagiri,M. 395 Haglund, RF. Jr. 792 Haguida, K.-i. 43 Halte,V. 664 Hama,Y. 337 Hamada,N. 604,607 Hamaguchi, H. 468,560 Hamazaki, J. 295 Hamm,R 401,443,542, 631,637 Han,H. 759 Hanamura, E. 357 Hane,H. 371,471 Haneda,K. 780 Hangyo,M. 685,699,738, 744, 747, 756 Hania,R. 679,879 Harada,S. 765 Harris, S.E. 3 Hartl, I. 843 Hase,M. 242,248 Hasegawa, A. 272 Hashimoto, H. 363, 368, 589 Hashimoto, N.T. 774 Hashimoto, Y. 319 Hattori,T. 337,705,750 Hauri,C.R 22,61,204 Hayden,C.C. 496 Hayes, S.C. 783 Haynes,T.E. 792 Hays,G. 897 He,Xiang 610 Hebeisen, C.T. 144 Hebling,J. 714,786 Heilweil,E. 167 Heimann, P.A. 343 Heinrich, A. 22,204 Heinz, B. 368 Helbing,F.W. 22,61,204 Helbing,J. 542,631

Herek,J.L. 368 Heroux,J.B. 319 Herz,L.M. 278,281 Hey,R. 292 Heyne,K. 389 Hirao,K. 349 Hirasawa, M. 371,471 Hiratsuka,H. 395 Hironaka,Y. 222,825 Hirooka, T. 43 Hirosawa, K. 873 Hisatake,S. 789,795 Hochstrasser, R.M. 539 Hoeben, F.J.M. 281 Hoki,K. 508 Hollberg,L. 837,840 Holman,K.W. 834 Holme, Niels C.R. 231 Homma, T. 88 Honda, M. 526 Honer zu Siederdissen, T. 19 Hong,F.-L. 843 Hori,T. 31 Horioka,K. 225 Horiuchi,H. 395 Horn,C. 864 Hornung,T. 569,595,717 Horn-von Hoegen, M. 170 Hosokawa, M. 846,849 Hou,B. 334 Hsieh,C.-F. 699,723 Hsiung,P. 876 Huber,R. 729,753 Humble, T.S. 514 Huse,N. 407

Ishii,K. 459 Ishii,N. 8,73,563 Ishii,Y. 295 Ishikawa, T. 771 Ishioka,K. 248 Ishizawa, A. 37,201 Isobe,M. 328 Itakura, R 885 Itatani, J. 164 Ito,H. 846,849 Ito,T. 219 Itoh,F. 858 Iwai, S. 340 Iwata, K. 468 Iye,Y. 319

I Ichida,H. 607 Ichinose, N. 395 Ikeda,M. 76 Ikeda,T. 738 Ikegami,T. 831 Ikezawa, M. 301 Ikuta,M. 374,520 Imae, M. 846 Imahori, H. 474 Imai, H. 313 Imasaka, T. 55 Imura,K. 434,655 Inaba,H. 843 Ino,Y. 735 Inoue, K. 357 Ippen,E.P. 801,876 Irvine, S.E. 667 Ishibashi,Y. 474

K Kabanov, V.V. 726 Kaertner, F.X. 768,822 Kakehata, M. 88 Kaku,M. 195 Kammler,M. 170 Kanai,T. 161,184,198,310 Kanematsu, Y. 607 Kannari, F. 28, 640, 873, 885 Kano,H. 560 Kaplan, D. 70 Kapteyn,H.C. 40,175,189, 192,239 Karaiskaj, D. 545 Karavitis, M. 377 Karpowicz, N. 693 Kato, S. 158 Kato,T. 511 Katsumoto, S. 319

Jahnke, F. 260 Janairo, G. 732 Janke, C. 690 Jason Jones, R. 16, 834 Jiang, H. 354 Jiang, Y. 545 Joannopoulos, J.D. 298 Joffre,M. 575,616,628,634 Johnsson, P. 204 Joly,N. 52 Jonas, D.M. 380,572 Jones, D.J. 287,834 Jonkheijm,P. 281 Jonkman, H.T. 679 Joo,T. 419 Jordan, R.E. 144 Jung,Y.S. 645

Kawada,Y. 158 Kawai,J. 813 Kawamura, S. 849 Kawano, H. 640 Kazansky, P.G. 349 Keller, U. 22,61,204 Kersting, R 693 Khalil,M. 551 Kida,Y. 55 Kieffer, J.-C. 201,343,346 Kienberger, R. 8 Kikuchi,A. 295 Killi,A. 804 Kim,D.S. 453,517,650 Kim,H.K. 645 Kim, H.T. 213 Kim, I.J. 213 Kim, J. 557 Kim,J.-W. 768 Kim,S. 625 Kimmel,M. 112 Kimura,T. 374,520,563 Kimura,Y. 398 Kishida,H. 340 Kishimoto, T. 272 Kishimura, H. 222 Kishino,K. 295 Kita,T. 263 Kitahara,H. 747 Kitajima, M. 242,248 Kitano,H. 789 Kito,T. 13 Kleiman,V.D. 133,456 Klug, D.R. 622 Knight, J.C. 52 Knoester,J. 879 Knorr,A. 19,292 Ko,T.H. 822,876 Kobayashi,H. 849 Kobayashi,T. 64,67,82, 121,371,374,383,471, 520,563,619,673,789, 795 Kobayashi, Y. 88 Koch, S.W. 19 Koga,Y, 831 K6hler,F. 729 Kohmoto,T. 322,482 Koide,K. 738 Kojima, E. 319 Kojima,0. 451,257 Koller,F. 462 Kompa, J. 167 Kompa,K.-L. 595 Komukai, M. 589 Kondo,K.-I. 222,825

903

Kono,H. 508,511 Kopf,D. 804 Kornelis,W. 22,61,204 Korte,F. 819 Koshihara, S.-y. 771 Kosik, E.M. 124, 127 Kosugue, A. 184 Kosumi,D. 589 Kowalevicz, A.M. 801 Koyama, K. 158 Koyama,Y. 619 Kozawa, T. 479 Kozich,V. 485 Krampert,G. 505,864 Krausz,F. 8,25,73,94 Kubarych, K.J. 575 Kubler,C. 729,753 Kubo,A. 645 Kubo,M. 474 Kubo,Y. 313 Kubuchi,M. 266,275 Kudemac, T. 679 Kudryashov, I. 813 Kuhl,J. 19,714,786 Kuhn,0. 389 Kumada,T. 377 Kumar, K. 539 Kumbhakar,P. 82 Kunimori, H. 849 Kunimoto,M. 322,482 Kunugita, H. 295 Kuroda,H. 468 Kuroda,K. 76 Kurz,H. 690,741 Kuttge,M. 741 Kuwata-Gonokami, M. 266, 275,319,735

L'Huillier, A. 204 Laenen, R. 428 LaFratta, C.N. 807 Lambry, J.-C. 616 Lang,R. 115,774 Lange,J. 810 Langhoff,H. 334 Lanzani, G. 363 Larsen,D.S. 592,610 Larsen, O.F.A. 548 Laubereau, A. 428 Lederer,M. 804 Lee,A.M.D. 496 Lee,C.-K. 103 Lee,K.F. 888 Lee,Y.S. 213

904

Legare,F. 888 Leigh, D.A. 548 Leitensdorfer, A. 729, 753 Levsque, J. 164 Lezec, H.J. 664 Lezius, M. 94 Li,B. 416 Libertun, A.R. 189 Lienau, C. 650 Liese, D. 864 Lim,M. 625 Lin, S.H. 502 Linde, von der, D. 170 Lindner, F. 94 Lindner, J. 431 Litvinyuk, I.V. 888 Liu,M. 499 Liu,T.-A. 756 Liu,Y. 189,354 Lochbrunner, S. 79, 130, 491 Loparo, J.J. 410,535 Lopez, R. 792 Lopez-Martens, R. 204 Lozovoy, V.V. 870 Lu,R. 248 Luo,C.W. 292

M Ma,L.-S. 16,837 Ma,Y.-Z. 331,425 Maeda, A. 340 Magnes,B.-Z. 448 Mair,C.E. 133,456 Makereel, F. 281 Maley, M.P. 325 Malyshev,V. 879 Manami, Y. 738 Mariette, H. 263 Marsal, L. 263 Martin, J.-L. 616,628,634 Maruyama, T. 849 Mashiko,H. 181 Masuda,K. 195 Masuda, S. 158 Masumoto, Y. 301 Mataga,N. 604 Matsubara,E. 357 Matsuda,H. 374,520,563 Matsuda, I. 774 Matsuda,K. 771 Matsui,S. 295 Matsumoto, H. 843 Matsumoto,Y. 307 Matsushita, A. 738 Matsuzaki, H. 340

Mauritsson, J. 204 Mayer, H. 25 McCracken, J.M. 422 Meijer,E.W. 281 Melinger, J.S. 456 Mercer, LP. 622 Meulen, van der, P. 499 Midorikawa, K. 28,46,58, 181,640 Milder, Maaike T.W. 431 Miller, A. 867 Miller, R.J.D. 144,407,601, 816 Minami, F. 272 Minemoto, S. 161, 198,310 Minoshima, K. 801,843 Misawa,K. 115,774 Misochko, O.V. 248 Mitsumori, Y. 272 Miura,E 158 Miyaji, G. 195 Miyamaru, F. 685, 699, 738, 747 Miyasaka,H. 474,598 Miyawaki, A. 640 Miyazaki, K. 195 Mizoguchi,K. 251,257 Mizuno, H. 640 Mochiduki,A. 750 Mohammed, O.F. 448 Monmayrant, A. 70, 91 Morgner,U. 804,822 Mori,Y. 474,598 Morishita, H. 222 Morita, R. 13 49 Morteani, A.C. 278 Moshammer, R. 894 Motzkus,M. 167,368,595 Mourou, G. 334 Muller,R. 650 Murakami, Y. 474 Mumane,M.M. 40,175,189, 192,239 Musasa, H. 813 Mysyrowicz, A. 22, 266

N Nabekawa, Y. 640 Nagahara, S. 263 Nagahara, T. 434, 655 Nagai, H. 34 Nagano, S. 849 Naganuma, R. 263 Nagasawa, Y. 598 Nagashima, T. 744

Nakagawa, Y. 598 Nakajima, M. 328 Nakajima,M. 711,756 Nakamura, K.G. 222,825 Nakamura, R. 607 Nakamura, T. 158 Nakano,H. 37,201 Nakase,Y. 795 Nakashima, S. 474 Nakashima, S.-I. 756 Nakayama, M. 251,257 Nakayama, T. 395 Nakazawa,M. 43,765,780, 831,873 Nakazono,K. 322 Nam,C.H. 213 Naughton, M.J. 807 Naumov, S. 112 Nayuki,T. 225 Nees,J. 334 Nelson, K.A. 152,236, 239, 252, 298, 529, 569,717 Nemoto,K. 222,225 Nibbering, E.T.J. 389, 407, 448 Nicoul,M. 170 Nielsen, N.C. 19 Nienhaus, G.U. 631 Nienhaus, K. 631 Niikura,H. 155,855 Niklaus,P. 505 Nimonji, T. 849 Nishijima, G. 720 Nishijima,K. 88 Nishimura, K. 619 Nishioka,H. 777 Nishiyama, Y. 511 Nishizawa,N. 31,34,876 Nisoli,M. 207,210 Nito,K. 831 Nomura, Y. 198 Norris,T.B. 231 Nurhuda,M. 46,58

O 0'Hara,J.F. 696 Oane,A. 843 Gates, C.W. 837,840 Oda,Y. 711 Ogasawara, M. 598 Ogawa,Y. 780 Ogilvie,J.P. 575,634 Ohkawa,M. 849 Ohmori, K. 526

Ohmura,H. 858 Ohsuka,S. 219 Ohtani,N. 257 Ohtsuki,Y. 508,511,613 Oishi,Y. 28,222,225 Okada,T. 598 Okamoto,H. 340,434,655 Okamoto, T. 885 Okamura,H. 313 Okano,Y. 222,825 Okazaki, S. 219 Okihira,S. 219 Okimoto, Y. 340 Omenetto, F.G. 52 Onda,K. 416,645 Ono,M. 340 Ono,S. 43,702,720,732 Ookuma, S.-I. 750 Osellame, R. 804 Osgood, R.M. 298 Ota, A. 771 Otani, M. 474 Owrutsky, J.C. 437 Ozaki,T. 201

Pines, E 425,448 Pirozhkov, A.S. 85,216 Piskarskas, A. 8, 73 Ploog,K.H. 292 Podlipensky, A. 810 Polack,T. 634 Poletto,L. 207,210 Polli,D. 363 Pon,A.C. 807 Pons, J. 807 Poulin,P.R. 529 Pouthier, V. 401 Pradarutti, B. 661 Prasankumar, R.P. 313 Prior, Y. 819 Prokhorenko, V.I. 601 Pshenichnikov, M.S. 386, 404,431 Pugzlys, A 679,879

Padmore, H.A. 343 Pan,C.-L. 103,699,723, 756 Pan,R.-P. 699,723 Papagiannakis, E. 592 Paradowska-Moszkowska, K. 392 Park,D.-J. 650 Park,H. 759 Park,J.-S. 419 Park,S. 557 Pascolini, M. 207,210 Paskover,Y. 819 Pastirk, I. 870 Patzlaff,T. 392 Pau,S. 231 Paul, A. 175,189,192 Paulus,G. 94 Paxton,B.J. 236,254 Pearson, B.J. 100 Peiponen, K.-E. Peng,Z. 456 Perrott, M.H. 768 Petek,H. 242,416,645 Petkovic,M. 389 Pfeifer,T. 178,864 Pfeiffer,W. 670 Pfister,R. 401 Pines, D. 425,448

R Radunsky, A.S. 127 Ragozin, E.N. 85 Ramirez, A.P. 269 Ramponi, R. 804 Rau, C. 661 Raymondson, D. 189 Reimann, K. 292 Ren,H.-W. 301 Resan, B. 118 Riedle,E. 79, 130,465, 491 Rini,M. 448,792 Rivas,J.G. 690,741 Rizo,P.J. 121 Roberts, S.T. 410,535 Robertsson, L. 837 Rodriguez, G. 149 Rohrdanz, M.A. 523 Rondonuwu, F. S. 619 Roos, P.A. 287 Ropers, C. 650 Roth,R.M. 298 Rubtsov,I.V. 539 Riickmann, I. 260 Rudenko, A. 894 Rungsawang, R. 750 Russell, P.S.J. 52 Rutkovski, P. 813

Q Qian,W. 380 Qiu,J. 349 Quema, A. 702,720,732

905

Saeki,A. 479 Saeta, P. 664 Saito, G. 771 Saito,N. 158 Saito, S. 711 Sakai,H. 161,198,310 Sakai,K. 756 Sakai,M. 732 Sakazaki,Y. 371,471 Saleh,B.E.A. 807 Salen,P. 499 Sando,G.M. 437 Sansone, G. 207,210 Santoro, F. 505 Sarukura,N. 702,720,732 Sasaki, M. 272 Sato,Y. 526 Sato;T. 849 Savolainen, J 368 Sawamura, A. 849 Saxler,J. 741 Schaarschmidt, M. 19 Schafer,K.J. 204 Schanz,R. 443 Schatzel,M. 94 Schenning, A.P.H.J. 281 Scherer,N.F. 557 Schibli,T.R. 843 Schlup,P. 61,204 Schmidhammer, U. 465 Schmidt, B. 462 Schneider,!. 670, Schoenlein, R.W. 343, 346, 792 Schotte,F. 581 Schrader, T. 462 Schreier,W. 462 Schriever,C. 491 Schroter, CD. 894 Schumacher, S. 260 Schwentner, N. 245,304 Seifert,G. 392,810 Sekikawa, T. 184 Sekiya, T. 225 Selle,R. 864 Shaffer, J.R 496 Sharma,V. 801,822,876 Shibuya,K. 789,795 Shih,T. 292 Shim,B. 897 Shimakawa, Y. 313 Shimano,R. 266,275,319, 735 Shimotsuma, Y. 349 Shimura,T. 76

906

Shu,S.-R 103 Shverdin, M. 3 Shymanovich, U. 170 Sieg,A. 462 Siemens, M. 239 Silva,C. 278,281,783 Sipe,J.E. 287 Siwick,B.J. 144 Skenderovic, H. 595 Skryabin, D.V. 52 Smalley,R.E. 331 Smilgevicius, V. 8, 73 Smith, A.W. 551 Smith, D.L. 269 Smith, E.R. 380 Sobotta,C. 462 Sokolowski-Tinten, K. 170 Sorokin, E. 112 Sorokina, I.T. 112 Spanner, M. 164 Spielmann, C. 178 Sreearunothai, P. 278 Stagira, S. 207,210 Statz,E.R. 298 Steinmeyer, G. 650 Stenger, J. 586 Stepanov, A.G 714,786 Stibenz,G. 650 Stiopkin,I. 554 Stock, K. 491 Stock, S. 861 Stockman, M.I. 673,676 Stokkum, van, I.H.M. 592, 610 Stolow,A. 496 Storck,W. 882,891 Strenger, J. 331 Suda,A. 28,46,58, 181 Suemoto,T. 328,711 Sul,S. 545 Sun,Q. 354 Sun, Zhijun 645 Suruga, S. 813 Suto,R 301 Suzuki,!. 161 Svirko,Y.R 735

Taccheo, S. 804 Tachiya,M. 858 Tada,A. 873 Tagawa, S. 479 Tahara,T. 459,488 Takada,H. 88 Takagi,N. 307

Takahashi,H. 88,219,702, 720, 732 Takahashi, J.-I. 357 Takamiya, H. 88 Takaya,T. 468 Takayanagi, J. 34 Takeda,J. 133 Takeoka,M. 873 Takeuchi,S. 459,488 Takizawa, Y. 225 Tanabe,T. 640 Tanaka, M. 699 Taneichi, T. 383 Tani, M. 699, 738, 747, 756 Taniguchi, S. 604 Tanimoto, M. 158 Tarasevitch, A. 170 Tatsuno,M. 738 Taylor, A.J. 52,149,231, 269,313,325,696,726 Teich,M.C. 807 Tennant,D.M. 231 Terazima, M. 398 Thaller, A. 428 Thorsm0lle, V.K. 269, 325 Tobey,R. 175,239 Tobinaga, M. 795 Tokmakoff, A. 410, 422, 535, 551 Tokunaga, E. 563 Tokunaga,F. 604,607 Tokura,Y. 340 Tomita,H. 77 Tondello,G. 210 Tonge,P.J. 610 Torchinsky, D.H. 152 Torizuka,K. 88 Torosyan, G. 661 Tosa,V. 213 Toumois, P. 70 Trebino, R. 97, 109, 112 Tsuchiya,Y. 106,219 Tubel,S. 729,753 Tuzhilin, D. 813

U Uchida,N. 771 Ueda,K.-I. 526,777 Ueda,Y. 328 Uiberacker, M. 8 Ullrich,!. 894 Underwood, D.F. 413 Unno, M. 604

V Valloresi,P. 207,210 Vanagas, E. 813 Varju,K. 204 Vartiainen, E.M. 735 Vaswani,H.M. 586 Vaughan,J.C. 569,717 Vengris,M. 592,610 Ventalon,C. 628 Villeneuve, D.M. 155,164, 855,888 Vogt,G. 505 Vohringer, P. 431 Volkov,V. 637 Vomir,M. 316 Vos,M.H. 628,634 Vozzi,C. 207,210

White, J.L. 100 Wiersma,D.A. 386,404, 431 Wilcox, M. 334 Willig,F. 882,891 Wilpers,G. 837 Wilson, B.C. 816 Windeler,R. 837 Winterfeldt, C. 178 Wischmeier, L. 260 Witte,T. 167 Woemer, M. 292 Wohlleben,W. 368 WoUenhaupt, M. 864 Woutersen, S. 548 Wu,Z. 354 Wulff,M. 581

W Wada,0. 263 Wadsworth, W.J. 52 Wagenblast, P. 822 Wagner, N. 40, 192 Wagner, W. 867 Waldmuller, I. 292 Walker, D.R. 3 Walker, R.C. 622 Walmsley, LA. 124, 127 Walter, D. 178 Walther,H. 94 Wang,X. 25 Ward,D.W. 298 Warren, W.S. 867 Watanabe,K. 307,720 Watanabe,N. 337 Watanabe, S. 184 Webb,K.J. 298 Weisman, R.B. 331 Wemcke,W. 485 Westenhoff, S. 281,783

X Xu,J. 759

Y Yaguchi,T. 765 Yakabe,M. 831 Yakovlev, V.S. 8 Yamaguchi, M. 236,254, 738 Yamaki,M. 508 Yamamoto, K. 738 Yamamoto, N. 257 Yamamoto, Y. 398 Yamanaka, T. 885 Yamane,K. 13,49 Yamanouch, K. 885 Yamashita,M. 13,49 Yamazaki, T. 225 Yamochi,H. 771 Yanagi,K. 589 Yang,H. 354

Yang, J. 479 Yang,M. 586 Yano,R. 705 Yasuda, M. 474 Yavuz,D. 3 Yazawa, H. 885 Ye, J. 16,834 Ye,T. 867 Yeremenko, S. 386,404 Yin,G.-Y. 3 Ying,Li 846,849 Yogi,T. 337 Yokoyama,H. 780 Yonera,T. 747 Yoon,M.-C. 517 Yoon,Y.-C. 650 Yoshida,M. 34,765,831 Yoshida,Y. 479 Yoshino, T. 849 Yoshioka,K. 266,275 Yoshizawa, M. 589 Yu,D. 816 Yuan,T. 759 Yuasa,Y. 374,383,520 Yurtsever, G. 867

Z Zaitsu, S. 55 Zaparozhchanka, Y. 813 Zeek,E. 97 Zeidler,D. 164 Zhang, K. 819 Zhang, T. 566 Zhang, X.-C. 189,759 Zhao,C. 395 Zhong, Q. 437 Zimmermann, C. 882,891 Zinth,W. 462 Zrost,K. 894 Zucco,M. 837

907