Diamond Chemical Vapor Deposition: Nucleation and Early Growth Stages (Materials Science and Process Technology Series)

DIAMOND CHEMICAL VAPOR DEPOSITION Nucleation and Early Growth Stages

Huimin Liu Department of Chemical Engineering Colorado State University Fort Collins, Colorado and

David S. Dandy Department of Chemical Engineering Colorado State University Fort Collins, Colorado

I

nP

NOYES PUBLICATIONS Park Ridge, New Jersey, U.S.A.

Copyright 0 1995 by Noyes Publications No part of this book may be reproduced or utilized in any form or by any means, electronic or mechanical, including photocopying, recording or by any information storage and retrieval system, without permission in writing from the Publisher. Library of Congress Catalog Card Number: 95-30332 ISBN: 0-8155-1380-l Printed in the United States Published in the United States of America by Noyes Publications Mill Road, Park Ridge, New Jersey 07656 10987654321

Library

of Congress

Cataloging-in-Publication

Data

Liu, Huimin, 1961Diamond chemical vapor deposition : nucleation and early growth stages ! by Huimin Liu and David S. Dandy cm. P. Includes bibliographical references and index. ISBN 0-81551380-I 1. Diamonds, Artificial. 2. Chemical vapor deposition. I. Dandy, David S. II. Title TP873.5DSL58 1995 ^_ ^^^__ 666<,88__dc2(-J y>-3u*jz CIP

Rointan F. Bunshah, University of California, Los Angeles (Series Editor) Gary E. McGuire, Microelectronics Center of North Carolina (Series Editor) Stephen M. Rossnagel, IBM Thomas J. Watson Research Center (Consulting Editor)

Electronic

Materials and Process Technology

HANDBOOK OF DEPOSITION TECHNOLOGIES FOR FILMS AND COATINGS,

Second

Edition: edited by Rointan F. Bunshah CHEMICAL

VAPOR

SEMICONDUCTOR by Gary E. McGuire

DEPOSITION MATERIALS

FOR MICROELECTRONICS: AND PROCESS

HYBRID MICROCIRCUITTECHNOLOGY Enlow

IONIZED-CLUSTER

BEAM DEPOSITION

AND TECHNIQUES:

AND EPITAXY: AND

CONTROL

HANDBOOK:

edited

by James J. Licari and Leonard R.

PROCESSES

DIFFUSION PHENOMENA IN THIN FILMS edited by Devendra Gupta and Paul S. Ho HANDBOOK OF CONTAMINATION Donald L. Tolliver

TECHNOLOGY

HANDBOOK:

HANDBOOK OF THIN FILM DEPOSITION Klaus K. Schuegraf

by Arthur Sherman

by Toshinori

edited by

Takagi

MICROELECTRONIC

MATERIALS:

IN MICROELECTRONICS:

edited

by

HANDBOOK OF ION BEAM PROCESSING TECHNOLOGY: Stephen M. Rossnagel, and Harold R. Kaufman

edited by Jerome J. Cuomo,

CHARACTERIZATION McGuire

Volume 1: edited by Gary E.

OF SEMICONDUCTOR

MATERIALS,

edited by StephenM.

HANDBOOKOF PLASMA PROCESSINGTECHNOLOGY: Jerome J. Cuomo, and William D. Westwood HANDBOOK OF SEMICONDUCTOR SILICON O’Mara, Robert B. Herring, and Lee P. Hunt HANDBOOK OF POLYMER Licari and Laura A. Hughes

COATINGS

DEPOSITION

FOR ELECTRONICS,

HANDBOOK Hayakawa

OF SPUTTER

HANDBOOK N. Helbert

OF VLSI MICROLITHOGRAPHY:

CHEMISTRY

OF SUPERCONDUCTOR

CHEMICAL VAPOR DEPOSITION E. J. Schmitz

TECHNOLOGY:

TECHNOLOGY:

OF TUNGSTEN

V

by William

2nd Edition:

by Kiyotaka Wasaand

edited by William B. Glendinning

MATERIALS:

ELECTROCHEMISTRY OF SEMICONDUCTORS McHardy and Frank Ludwig

edited

Rossnagel,

C.

by James

Shigeru

and John

edited by Terre11 A. Vanderah

AND TUNGSTEN

SILICIDES:

AND ELECTRONICS:

by John

edited by John

Series

vi

HANDBOOK DIAMOND

OF CHEMICAL

VAPOR

DEPOSITION:

FILMS AND COATINGS:

ELECTRODEPOSITION:

by Hugh 0. Pierson

edited by Robert F. Davis

by Jack W. Dini

HANDBOOK OF SEMICONDUCTOR Werner Kern

WAFER

CONTACTS

edited by Leonard J. Brillson

TO SEMICONDUCTORS:

CLEANING

TECHNOLOGY:

HANDBOOK OF MULTILEVEL METALLIZATION FOR INTEGRATED by Syd R. Wilson, Clarence J. Tracy, and John L. Freeman, Jr. HANDBOOK Pierson

OF CARBON,

MOLECULAR HANDBOOK E. McGuire

GRAPHITE,

BEAM EPITAXY: OF COMPOUND

DIAMONDS

DIAMOND

PLASMA

CHEMICAL

SEMICONDUCTORS:

SOURCES:

VAPOR

FRICTION

SPECIAL

PROCESSING

FOR INDUSTRIAL

MELTING

OF MATERIALS:

ELECTRONICS

AND CERAMIC

Volume 1: edited byJon G. P.

by Peter J. Blau

edited by Lawrence E. Murr

TECHNOLOGIES:

REFRACTORIES

FILMS AND COATINGS:

PREFORMS,

edited by K. S. Mazdiyasni

APPLICATIONS:

CORROSION OF GLASS, CERAMICS David E. Clark and Bruce K. Zoitos

CERAMIC

L.

and Technology

ANDTECHNOLOGY,

AND PROCESSING

HANDBOOK OF INDUSTRIAL and Gordon L. Barna

edited by Raymond

by Huimin Liu and David S. Dandy

COMPOSITES:

AND WEAR TRANSITIONS

SHOCK WAVES

by Hugh 0.

edited by Oleg A. Popov

DEPOSITION:

CERAMIC

CERAMIC

edited

edited by Paul H. Holloway and Gary

SOL-GEL TECHNOLOGY FOR THIN FILMS, FIBERS, AND SPECIALTY SHAPES: edited by Lisa C. Klein

ADVANCED Binner

by

edited by Robin F. C. Farrow

Ceramic and Other Materials-Processing

FIBER REINFORCED

CIRCUITS:

AND FULLERENES:

HANDBOOK OF VACUUM ARC SCIENCE AND TECHNOLOGY: Boxman, Philip J. Martin, and David M. Sanders HIGH DENSITY

edited

edited by G. K. Bhat

SUPERCONDUCTORS:

TECHNOLOGY:

edited by John B. Wachtman

edited by

by Stephen C. Carnigiia

and Richard A. Haber

Related Titles CODECOMPLIANCE SEMICONDUCTOR David G. Baidwin

FORADVANCED INDUSTRIAL

TECHNOLOGY

HYGIENE

FACILITIES:

HANDBOOK:

bywilliam

by Michael

R.Acorn

E. Williams

and

Preface

Chemical Vapor Deposition (CVD) process, one ofthe most important technological developments in the past decade, has made production of high-quality diamond coatings on preshaped parts and synthesis of freestanding shapes of diamond a reality. Epitaxial diamond has been grown on diamond and cubic-BN. Polycrystalline diamond films have been deposited on various non-diamond substrates, including insulators, semiconductors, and metals, ranging from single crystals to amorphous materials. However, further technological developments in CVD of diamond films, particularly in such challenging areas as single-crystal growth for electronic applications and low-temperature deposition for coating on optic and plastic materials, requires a detailed understanding and effective control of the fundamental phenomena associated with diamond nucleation and growth. These phenomena, especially the nucleation and early growth stages, critically determine film properties, morphology, homogeneity, defect formation, adhesion, and the type of substrates that can be successfully coated. In an effort to enhance diamond nucleation and control film morphology, extensive studies on the nucleation and earlv growth stages have been performed. A number of surface pretreatment methods have been developed and the dependence of the nucleation process on deposition parameters and substrate surface conditions has been investigated. In particular, highly oriented and textured growth methods provide a novel approach to obtaining diamond films of a near-single-crystal morphology over large areas and may allow diamond to realize its potential as an electronic material inthe near future. Therefore, it is important to summarize these research results in a concise; structured manner in order to improve our understanding and control over the nucleation and early growth stages

vii

viii

Preface

This book presents an updated, systematic review of the latest developments in diamond CVD processes, with emphasis on the nucleation and early growth stages of diamond CVD. The objective of this book is to familiarize the reader with the scientific and engineering aspects of diamond CVD, and to provide experienced researchers, scientists, and engineers in the academic and industry communities with the latest developments in this growing field. The scope ofthe present book encompasses the developments and applications of diamond CVD, starting with a brief description of atomic and crystal structures of diamond and a review of the various processing techniques used in diamond CVD. It is followed by an extensive discussion of fundamental phenomena, principles and processes involved in diamond CVD, with emphasis on the nucleation and early growth stages of diamond during CVD. Diamond nucleation mechanisms, epitaxy and oriented growth are discussed on the basis of experimental observations. The nucleation enhancement methods developed to date are summarized. The effects of surface conditions and deposition parameters on surface nucleation are described. Finally, theoretical and modeling studies of surface nucleation are reviewed. There have been recent books on CVD of diamond films and coatings, but this book is probably the first one specifically addressing the nucleation and early growth stages in diamond CVD. We have attempted to correlate many diverse nucleation effects and mechanisms and to encompass fundamental phenomena, principles and processes associated with the nucleation and early growth stages into this book. Huimin Liu David S. Dandy

November, 1995 Fort Collins, Colorado

NOTICE 10 trrc oest or our krmwiedge the information in this pubiication is accurate; however the Publisher does not assume any responsibility or liability for the accuracy or completeness of, or consequences arising from, such information. This book is intended for informational purposes only. Mention of trade names or commercial products does not constitute endorsement or recommendation for use by the Publishcr. Final determination of the suitability of any information or product for use contemplated by any user, and the manner of that use, is the sole responsibility of the user. We recommend that I”.,A”P ,,,L.,'LU"'~ ;ntS=trrl;nn tn YPl,, ",, nn ‘2L.J sn,r ,""",,,,,,"I,"ULn",I ~P,.AmmP”A9t;AII "I nf ,II‘LL"IIcLIY mnt,v;str nr U","'." L" I"', Y, procedures mentioned in this publication should satisfy himself as to such suitability, and that he can meet all applicable safety and health standards.

Contents

1 General Introduction . .. . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 - -i

2 Atomic and Crystal Structures

ofDiamond

.......... 8

3 Diamond CVD Techniques ................................... 1.O HOT-FILAMENT CVD ......................................................... 2.0 PLASMA-ASSISTED CVD .................................................. 2.1 Microwave Plasma-Assisted CVD ................................... 2.2 Direct-Current Plasma-Assisted CVD ............................... 2.3 Radio-Frequency Plasma-Assisted CVD ........................... 2.4 Direct-Current Thermal Plasma CVD .............................. 2.5 Radio-Frequency Thermal Plasma CVD ........................... 3.0 FLAME CVD ...................................................................... 4.0 GENERAL CHARACTERISTICS OF DIAMOND CVD PROCESSES ...................................................................... 4.1 Crystallite Morphology ..................................................... 4.2 Gas-Phase Activation ........................................................ 4.3 Gas Species and Gas Compositions ................................... 4.4 Gas Flow Rate and Pressure ............................................. 4.5 Substrate Materials and Pretreatment Methods .................. 4.6 Substrate Temperature .....................................................

ix

14 18 .23 .26

27 28 .28 .29

30 31 31 33 35 36 39 .42

x

Contents

4.7 Substrate Position and Size .............................................. 4.8 Effects of Electric and Magnetic Fields ............................ 4.9 Impurities and Defects ..................................................... 5.0 SUMMARY .....................................................................

.43 .44 .44

.45

4 Diamond Nucleation Mechanisms . . . .. . . . . . . . . . . . . . . .. . . . 46 1.O HOMOGENEOUS NUCLEATION-GAS-PHASE NUCLEATION . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . .. . . . . . .. . . . . . . . . . . . . . . . . . . . 47 2.0 HETEROGENEOUS NUCLEATION-SURFACE NUCLEATION . . .. . . . . . . . . . .. . . . . . . . . . ._.... . . . . . . .. .. . .. . . . . . . . . . . . . . . . . . . . . . .. . . 50 2.1 Nucleation Processes and General Features . . . . . . .. . . . . . . . . . . . . . . . 50 2.2 Nucleation on an Intermediate Layer of Diamond-like Amorphous Carbon . . .. . . . . . . . .. . . . . . . . . . .. . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . 59 2.3 Nucleation on an Intermediate Layer of Metal Carbides . . . . .64 2.4 Nucleation on an Intermediate Layer of Graphite . . . . . . . . . . . . . .. 72 3.0 SUMMARY . . . . . . . .. . . . . . . . .. . . . . . . . .. . . . . . . . . . . . . .. . . . .. . . . . . . . . . . . . .. . . . . . . . . . . 77

5 Diamond Epitaxy, Oriented Growth, and Morphology Evolution .......................................... 1.0 EPITAXY ...................................................................... 2.0 ORIENTED AND TEXTURED GROWTH ........................... 3 .O MORPHOLOGY EVOLUTION ............................................. 4.0 SUMMARY ......................................................................

6 Effects of Surface Conditions on Diamond Nucleation .............................................................. 1.0 SUBSTRATE MATERIALS .................................................. 2.0 SURFACE PRETREATMENT METHODS AND NUCLEATION ENHANCEMENT MECHANISMS .......... 2.1 Scratching ...................................................................... 2.2 Seeding .................................................................... 2.3 Biasing .................................................................... 2.4 Covering and Coating ..................................................... 2.5 Ion Implantation ............................................................. 2.6 Pulsed Laser Irradiation .................................................. 2.7 Carburization ................................................................. 2.8 Catalytic Effects ............................................................. 3.0 SUMMARY ....................................................................

79 79 81 87 89

92 92 .94 96 103 106 114 12 1 123 124 124 126

Effects of Deposition Conditions on Diamond Nucleation ............................................................ 1.O SUBSTRATE TEMPERATURE .......................................... 2.0 GAS-PHASE ACTIVATION ............................................... 3.0 GAS PRESSURE AND FLOW RATE ................................. 4.0 GAS COMPOSITION.. ........................................................ 5.0 OXYGEN ADDITION ......................................................... 6.0 SUMMARY ....................................................................

Theoretical and Modeling Studies on Diamond Nucleation ............................................................ 1.O IDENTIFICATION OF NUCLEATION AND GROWTH MODE ............................................................. 2.0 THEORETICAL STUDIES ON NUCLEATION THERMODYNAMICS ...................................................... 3 .O THEORETICAL MODELING OF NUCLEATION KINETICS .................................................................... 4.0 CLUES OF STRUCTURE, CHEMISTRY, AND SIZE OF DIAMOND NUCLEI .......................................... 5.0 SUMMARY ....................................................................

131 13 1 134 134 135 137 141

143 143 145 150 156 158

References . . . . . . .. . . . . . . . .. . . . . . . .. . . . . . . .. . . . . . . .. . . . . . . . . . . . . . . . . . .. . . . . 160 Index . . . . .. .. . . . . . . . . . . . .. . . . . . . .. . . . . . .. . . . . . .. . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . 183

1 General

Introduction

The extreme hardness, high thermal conductivity, excellent infrared transparency, and remarkable semiconductor properties (Table 1) combine to make diamond one of the most technologically and scientifically valuable materials found in nature.l’l-I31 However, natural diamond is rare and only obtainable as gem stones in small sizes and at great expense. The scarcity and high cost have motivated researchers to attempt to duplicate nature and synthesize diamond since it was discovered in 1797 that diamond is an allotrope of carbon. At room temperature and atmospheric pressure, graphite is the stable crystalline form of carbon, with an enthalpy only 2 k.I mol-’ lower than diamond. Diamond is thermodynamically stable relative to graphite only at high pressures, as evident from the carbon phase diagram, Fig. 1.131 Although thermodynamically feasible at relatively low pressures and temperatures (Fig. l), graphite to diamond conversion faces a considerable kinetic barrier, and the rate of conversion apparently decreases with increasing pressure.121 Therefore, the early attempts to convert graphite into diamond by simply increasing pressure were unsuccessful for over one hundred years 111until high-pressure high temperature (HPHT) processes came into being. The development of the HPHT processes is a result of the extensive research during the 1940’s that built upon detailed knowledge of the carbon phase diagram developed by 0. I. Leipunskii et al.1’1 The HPHT synthesis of diamond essentially duplicates the natural process by converting graphite into diamond under conditions at which diamond is the I

2

Diamond Chemical Vapor Deposition

thermodynamically favored phase. Direct conversion of graphite to diamond in static HPHT processes (Fig. 1) requires high pressure (>120 kbar[‘]-[3]) and high temperature (~:300O~C[‘]-[~]) to overcome the kinetic barrier and obtain any observable conversion rate, and hence not economically viable. The difficulties in the direct conversion spurred the development of various processes to lower the temperature and pressure.

Table 1. Properties of CVD Diamond and Single-Crystal Diamondr2] CVD

Diamond Density (g cmm3)

2.83.51t4)

Thermal capacity at 27’C (J mot-’ K-l)

6.12

Single-crystal Diamond 3.515 6.195

Standard entropy at 27% (J mol“ K-l)

2.428

Standard enthalpy of formation at 27’C (J mol-‘)

1.884

Effective Debye temperature at 0-827°C

(K)

1860 * 10

Thermal conductivity at 25’Ca (W m-l KM’) Thermal expansion coefficient at 25-200°Cc

(x 10s6 ‘C-I)

Band gap (eV)

2100b _2.OI2lI51

2200 0 .8-l .2[21[51

5.45

5.45

10’2-10’6

10’6

Dielectric constant at 45 MHZ to 20 GHz

5.6

5.7

Dielectric strength (V cm-‘)

106

106

Electrical resistivity (Q cm)

Loss tangent at 45 MHZ to 20 GHz Saturated electron velocity (x 10’ cm s’) Carrier mobility (cm2 V’ s-l) electron (n) positive hole (p) Young’s modulusa (GPa) Compression Poisson’s Coefficient

~0.0001 2.7

2.7

1350-1500 480

2200 1600

820-900d at 0-800”Ct61

910-1250

strength (GPa)

8.68-16.53

ratio of friction in air

Vickers hardnessa (GPa) [varies with crystal orientation] Index of refraction at 10 pm

0.10-0.16 0.035-0.3t71

0.05-O. 15

50-100

57-104

2.34-2.42

2.40

a Higher than any other known materials b Anisotropic characteristic of thermal conductivity of thick CVD diamond films may be found in Ref. 9. ’ Lower than Invar d Young'smodulus=89_5(1-1.04x1~(7'-20)~, (GPa), where T in “Ct*).

General Introduction

3

Pressure, kbar

loo0

zoo0 3ooo Temperature. ‘c

4ooo

so00

Figure 1. Carbon phase diagram with temperature and pressure ranges corresponding to various diamond synthesis processes, as indicated by the shaded areas. (Reproduced wifh permission

from ReJ 3, 0 American

Chemical

Society,

1989)

The first breakthrough came in 1953 when H. Liander at Allemanna Svenska Elektriska A. B. (ASEA) in Sweden developed a HPHT processl”] using a liquid metal solvent-catalyst at pressures and temperatures where diamond is thermodynamically stable. Independently, the General Electric team, F. P. Bundy and co-workers, synthesized diamond using the HPHT technique in 1954, 1111followed by H. B. Dyer and co-workers at De Beers Adamant Research Laboratory in South Africa .l121Through surmounting the kinetic barrier by the solvent-catalyst reaction with a transition metal-Fe, Ni, Co, Cr, Pt, Pd, Fe-Ni, Co-Fe,121 or Mn, Al as well as B141--the solventcatalytic HPHT process permits graphite to diamond conversion to occur at conditions much nearer the graphite-diamond equilibrium line but at lower temperatures (Fig. 1). Typically, pressures range from 50 to 100 kbar and temperatures from 1300 to 2300”C.111131141 This technique, commercialized by General Electric in the U.S., along with the dynamic HPHT technique, i.e., shock-wave synthesis, U3]industrialized by Du Pont in the U.S., provides a reproducibility and tailorability unavailable in natural diamond in terms of chemistry, morphology, size, shape, toughness, and other properties important for abrasive and heat-sink applications.141 The development of the techniques also increases knowledge ofthe carbon phase diagram, which has

4

Diamond Chemical Vapor Deposition

advanced carbon research in general. Large crystals of approximately 8 mm diameter have been synthesized by Sumitomo Electric in Japan, and recently, even larger single crystals, up to 17 mm, by De Beers in South Africa.[21131 Research in the HPHT synthesis of diamond is still underway in an effort to lower production costs and produce even larger crystals. The second breakthrough came with the discovery by W. G. Eversole of the Union Carbide Corporation in the U.S.l141 that diamond could be deposited on a substrate from a hydrocarbon gas or a CO/CO, mixture by chemical vapor deposition (CVD) at low pressures and temperatures where diamond is metastable with respect to graphite (Fig. 1). Eversole’s effort, starting in 1949 and proceeding in parallel with the early studies ofthe HPHT processes, led to first successful synthesis of diamond by CVD, predating the first successful HPHT synthesis. 1151[161In 1953, H. Schmellenmeier at Potsdam Teachers College in East Germany reported the formation of very hard carbon films from acetylene in an electrical discharge.l171 Although Schmellenmeier was not attempting to synthesize diamond films in his experiments, the x-ray data of the coatings revealed crystalline diamond present in the films. Contemporarily, B. V. Derjaguin and co-workers at the Institute of Physical Chemistry of the Academy of Science of the former U.S.S.R.l181 and J. C. Angus and co-workers at Case Western Reserve University in the U.S. llgl initiated efforts to grow diamond at low pressures. These two groups worked independently of one another, and both were unaware of Eversole’s work. The experiments conducted by these two groups were characterized by the co-deposition of diamond and graphite on diamond seed crystals. The deposition processes required frequent interruptions to remove accumulated graphite by hydrogen etching at temperatures and pressures greater than 1000°C and 50 atm, or by oxidizing in air at atmospheric pressure. 118]The typical growth rates of diamond were less than 0.1 pm h-‘. In 1966 it was found by J. J. Lander and J. Morrison of Bell Telephone Laboratories in the U.S. that hydrogen may permit metastable growth of diamond by impeding the conversion of diamond to graphite at temperatures between 900 and 1300°C.1201 They pointed out that the growth of single-crystal diamond layers on a single-crystal diamond substrate should be possible as long as carbon atoms are added at a rate low enough to prevent stable graphite nuclei from forming. In 1978, E. C. Vickery of Diamond Squared Industries in the U.S. developed a process for growth of diamond layers on a diamond substrate,131 in which the deposition/removal cycles were combined into a single step by using a

General Introduction

5

mixture of 95 vol.% hydrogen and ~5 vol.% hydrocarbon in the presence of catalysts such as Pt or Pd. Although the average growth rates at the time were too low to be of commercial significance, the sustained efforts of Derjaguin, Angus and their co-workers ultimately led to the discovery of the crucial role of atomic hydrogen to preferentially etch graphite deposits and to permit high nucleation and growth rates on non-diamond substrates. This discovery was a historic milestone in the development of diamond CVD techniques.[2’l-1241 It was the recognition of the crucial role of atomic hydrogen that led Derjaguin and co-workers125l to the first successful growth of diamond crystals on nondiamond substrates at a commercially practical deposition rate (> 1 pm h-‘) in the mid 1970’s. This was followed by the development ofvarious methods, to increase the concentration such as electric discharge and hot-filament, 1261 of atomic hydrogen during CVD. These successes triggered considerable interest and an extensive research effort on diamond CVD in Japan. In the early 1980’s, the Japanese research group at National Institute for Research in Inorganic Materials (NIRIM) synthesized individual faceted diamond crystals at growth rates of 10 urn h-’ by microwave plasma and hot-filament assisted CVD.1271-1291 These results, with convincing characterization evidence by electron microscopy, x-ray diffraction, and Raman spectroscopy, confirmed the earlier experiments and refocused worldwide attention on the synthesis of diamond by CVD. In the past decade, a wide variety of energetically assisted CVD processes have evolved for diamond synthesis on various substrates at practical growth rates. 1271-1411 In particular, the strong Japanese research effort led to significant strides in the practical synthesis and applications of polycrystalline diamond films and coatings. Deposition areas as large as 400 cm2 have been achieved.131 Linear growth rates have been increased to the order of hundreds of micrometers per hour,l37l and recently to 930 pm h-l in DC plasma arc-jet CVD.1421 Diamond-coated boring and drilling tools have been developed by Mitsubishi Metals in Japan. High-fidelity loudspeakers for high-frequency sound with a diaphragm coated with diamond thin film have been manufactured by Japanese Victor Corp. and Sumitomo Electric Co. in Japan.131 The commercial production of free-standing shapes of diamond, >lOO cm* in area and 1 mm in thickness, has been realized by Norton and General Electric in the U.S., and others.121 The potential for economic scale-up of diamond CVD techniques qualifies it as a viable processing alternative to the HPHT methods for

6

Diamond Chemical Vapor Deposition

production of diamond abrasives or heat sinks at a cost that is still high but will be reduced as the technology improves. Moreover, CVD processes offer an opportunity to exploit many desirable physical properties of diamond (Table 1). The ability to coat a large area on a variety of substrate materials with diamond films vastly expands the potential application areas of diamond (Table 2) over those possible with natural or HPHT-synthetic diamond. This capability, along with the need to explain the improbable growth of diamond under apparently metastable conditions, has stimulated active research into all aspects of diamond CVD in all the major industrialized countries over the world. As potential applications of CVD diamond are continuously discovered, it may be anticipated that the ultimate economic impact of this emerging technology on the defense, space, and commercial areas will outstrip that of high-temperature superconductors with more immediate applications141 (Table 2). By making an updated and systematic review of diamond CVD processes, the objective of this book is to familiarize the reader with the scientific and engineering aspects of diamond CVD, and to provide experienced researchers, scientists, and engineers in academic and industry community with the latest developments in this growing field. The scope of the present book encompasses the development and applications of diamond CVD, starting with a brief description of atomic and crystal structures of diamond and a review of the various processing techniques used in diamond CVD. It is followed by an extensive discussion of fundamental phenomena, principles and processes involved in diamond CVD, with emphasis on the nucleation and early growth stages of diamond during CVD. Diamond nucleation mechanisms, epitaxy and oriented growth are discussed on the basis of experimental observations. The nucleation enhancement methods developed to date are summarized. The effects of surface conditions and deposition parameters on surface nucleation are described. Finally, theoretical and modeling studies on surface nucleation are reviewed.

General Introduction

7

Table 2. Actual and Potential Applications of CVD Diamond[2][3][431-[46]

Application areas Grinding/cutting tools

Application examples

Physical properties of diamond utilized in the applications

Inserts

greathardness

Twistdrills

great wear resistance high strength and rigidity good lubricating properties general chemical inertness

Whetstones Industrial knives Circuit-board drills Oil drilling tools Slitter blades surgical scalpels .savrr

Wearparts

Acoustical coatings

D@io&corrosion

protection

Optical coatings

Photonic devices Thermal management

Semiconductor devices

Bearings

pat hardness

Jet-nozzle coatings Slurry valves Extrusion dies Abrasive pump seals Computer dish coatings Engine parts Mechanical implants Ball bearings Drawing dies Textile machinery Speaker diaphragms

great wear resistance high strength and rigidity good lubricating properties general chemical inertness

Crucibles Ion barriers (sodium) Fiber coatings Reaction vessels Laser protection Fiber optics X-my windows Anti reflection W to IR windows Radomes Radiation detectors Switches Heat-sink diodes Heat-sink PC boards Thermal printers Target heat-sinks High-power transistors High-power microwave Photovohaic elements Resistors Capacitors Field-effect transistors W sensors

high sound propagation speed high stiffness low weight general chemical inertness high strength and rigidity good temperature resistance transparency from W through visible into IR good radiation resistance

large bandgap high thermal conductivity high electrical reaistivity

high dielectric strength high thermal conductivity good temperature resistance good radiation resistance high power capacity good high- frequency performance low saturation resistance

Atomic and Crystal Structures of Diamond

A comprehensive understanding of diamond nucleation and growth processes during CVD necessitates knowledge of atomic and crystal structures of diamond, as well as of the competing crystallites or amorphous phases that may be produced by CVD. Three hybrid carbon orbitals available for bonding--sp, sp2, and sp3-complete the series of electronic building blocks of all carbon allotropes and compounds. Vapor-grown fullerenes,[47] diamond polytypes,J4*J carbynes J4’j and amorphous hard carbon, a-C (diamond-like carbon, DLC, and diamond-like hydrocarbon, a-C:H)t15J[50]have been found in diamond growth experiments, yet the significant competition during CVD occurs between sp2 and sp3 types of carbon, i.e., graphite and diamond. In the graphite lattice structure, each carbon atom combines with its three neighbors using hybrid sp2 atomic orbitals, forming a series of continuous hexagonal structures, all located in parallel planes, as illustrated in Fig. 1. The strong (J bonds are covalent, forming equal angles of 120’ to each other, with a short bond length of 0.141 nm and a high strength of 524 kJ mo1-1.[2j The fourth orbital, i.e., the delocalized electron, is directed perpendicularly to the planes and paired with another delocalized electron of the adjacent plane by a much weaker van der Waals bond (the subsidiary n:

8

Atomic and Crystal Structures

of Diamond

9

bond) of only 7 kJ mol-i. The weakness ofthe n:bonds plus the large spacing of 0.335 nm between two planes, gives rise to the anisotropic characteristic of graphite. In the diamond lattice structure, each carbon atom is tetrahedrally coordinated, forming strong bonds to its four neighbors using hybrid sp3 atomic orbitals, with equal angles of 109’28’ to each other. Each tetrahedron combines with four other tetrahedra forming strongly-bonded, uniform, three-dimensional, entirely covalent crystalline structure. The covalent bonding between carbon atoms is characterized by a small bond length of 0.154nmandahighbondenergyof711 kJmol-‘.I*]

A

A A

C 0 0

0

A

A

A

GRAPHITE

LONSOALEITE

DIAMONO

Figure 1. Schematic diagram showing crystal structures of hexagonal graphite, cubic diamond, and hexagonal lonsdaleite. The shaded hexagonal rings of carbon: planar for graphite, chair form for cubic diamond and boat form for hexagonal lonsdaleite. The letters depict the stacking sequences of carbon: planar layers for graphite, puckered planes for cubic diamond and lonsdaleite.

Diamond has two basic crystal structures, one with a cubic symmetry (more common and stable) and the other with a hexagonal symmetry (rare but well established, found in nature as the mineral lonsdaleite). The closepacked layers, { 1 1 1} for cubic and { lOO} for hexagonal, are identical. The cubic structure can be visualized as stacking of puckered planes of sixmembered saturated carbon rings in an ABC ABC ABC sequence along (111) direction, referred to as 3C diamond (Fig. 1). All ofthe rings exhibit the chair

10

Diamond Chemical Vapor Deposition

conformation, as indicated by the shaded areas, and all C-C bonds are staggered. In the hexagonal structure, the stacking occurs in an AI3 AB AI3 sequence, known as 2H diamond (Fig. 1). The rings in the stacking direction show the boat conformation, and the C-C bonds normal to the chair planes are eclipsed. The formation of a lonsdaleite layer on a diamond surface is equivalent to the formation of a stacking fault during diamond growth. The cubic structure is the dominant crystal structure in both natural and synthetic diamond since the staggered conformation is more stable than the eclipsed due to the slightly lower energy (0.1-0.2 eV per carbon atom).I*I Diamond polytypes and carbyne phases form only during the homogeneous nucleation and growth of diamond powder.14’I Diamondhas several crystal shapes (habits), including the { lOO}cube, the { 1 lo} dodecahedron, the { 111} octahedron (Fig. 2), and other more complicated shapes (Figs. 3-5). In CVD diamond, the { 11 l} octahedral faces are observed at low temperatures and low hydrocarbon concentrations; the { 100} cubic faces predominate at high temperatures and high hydrocarbon concentrations,1151[261 and cube-octahedral crystals combining both these faces are commonly found. The dependence of crystal shapes on deposition conditions has been correlated to the ratio of growth rates in ( 100) and (111) directions,1261 defined as a growth parameter 01, a = (v~~,,/v~~~)~.I~~I

Cubic (100)

Dodecahedron

{llO)

Octahedron

{ill}

Figure 2. Simple crystal shapes of diamond.

In Fig. 3a, cubic and octahedral faces are evident, and in Fig. 3b the twinned crystals with pseudo-fivefold symmetry can be clearly seen. This twinned fivefold symmetry is prevalent in CVD diamond thin films and apparently never develops on homoepitaxially grown crystals.152al152bIBalllike diamond crystals are grown at high supersaturations1151 (Fig. 3~).

Atomic and Crystal Structures

of Diamond

Figure 3. Diamond crystals grown from 1.5 vol. % CH,-H, at 10 torr using a combined microwave and hot-filament CVD method. (a) Cube and octahedral faces, (71)twinned crystals with pseudo-fivefold symmetry, (c) ball-like diamond grown at high supersaturations; the scale in (a)-@) is 12 pm and in (c) is 15 pm. (Reproduced with permission,[“] 0 American

Associationfor the Advancement

of Science,

1988.)

In Fig. 4, it can be seen that, with the a value increasing from 1 to 3, the crystal shape changes from cube to cube-octahedron and then to octahedron (upper); and the crystals grown with a >1.5 exhibit rough { 1 1 1} facets whereas the crystals grown with 01 Cl.5 show very smooth { 11 l} facets (lower). L511

12

Diamond Chemical Vapor Deposition

tb)

Figure 4. (Upper) Idiomorphic crystal shapes of diamond for different values of the growth parameter CL(CL= (v,c&~~,&&; the arrows indicate the direction of fastest growth. (Lower) SEM micrographs of isolated diamond crystals grown under different growth conditions; the cube-octahedral shapes with values of 2.4,2.05, 1.65, and 1.4 for (a) to (d), respectively.[“l (Reproduced with permission.)

Faceted diamond dendrites may form in flame synthesis (Fig. 5a). Twinning, or stacking fault, occurs frequently on the { 1 11 } planes.t53]-t55] A computer simulation t55]shows that a single stacking fault on parallel { 11 l} planes leads to very slowly growing triangular platelets; two or more stacking faults on parallel { 1 1 1} planes lead to hexagonal, platelet shaped crystals (Fig. 5b); truncated triangular shaped platelets are formed when a small odd number of stacking faults occur on parallel { 1 1 1 } planes; two stacking faults on non-parallel { 1 1 1} planes result in fivefold twinned decahedra1 crystals; and three stacking faults on non-parallel { 11 l} planes result in twinned icosahedral crystals.

Atomic and Crystal Structures

of Diamond

I3

Figure 5. CVD diamond crystals. (a) Faceted dendrites of flame CVD grown diamond,[52a] @) hexagonal platelet crystal with fully developed three-dimensional facets grown from 1 vol.% 0, - 1 vol.% CH, - H, at 30-40 torr and 850°C substrate temperature using microwave plasma assisted CVD, the hexagonal platelet is -2.5 pm in maximum linear dimension.[53] (Reproduced with permission.)

Diamond

CVD

Techniques

A large variety of carbon-containing gas species have been employed to synthesize diamond by CVD. These include methane, aliphatic and aromatic hydrocarbons, alcohols, ketones, amines, ethers, and carbon monoxide t21t31t561 with methane being the most frequently used reagent. In additibn to these carbon carriers, the gas phase usually must contain powerful non-diamond carbon etchants and surface site preparation species such as hydrogen, oxygen, or fluorine atoms. In a methane/hydrogen mixture, for example, hydrogen concentrations must generally exceed 97-99 vol.% in order to grow high quality diamond films. It has been observed that a necessary condition for diamond growth to occur is the presence of a gas-phase non-equilibrium in the region adjacent to the deposition substrate. The gas-phase non-equilibrium is generated through gas-phase activation. The gas-phase activation is achieved typically using one of the three basic methods: 1. External heating (as in hot-filament CVD) 2. Plasma activation (as in plasma assisted CVD) 3. A combination of thermal and chemical activation (as in flame CVD) To obtain the gas-phase activation state required for the stable growth of well-crystallized diamond, a variety of energetically assisted CVD techniques have been developed and employed. These may be classified into three major categories:

14

Diamond CKD Techniques 1. Hot-filament

15

CVD (HFCVD)[271[28~[301

2. Plasma assisted CVD (PACVD)[2g1[311[38] 3. Flame (combustion)

CVD[571[581

Laser-enhanced CVD[401[5gl-[621 h as also been applied to synthesize diamond. In addition, hybrid techniques utilizing various combinations of these processes have been developed and employed, differing primarily in the means of producing the gas-phase activation state. The concept of applying catalytic processes to diamond CVD t63]has not attracted close attention, due primarily to the difficulty in controlling chemistry and the fact that adding catalytic mechanisms to an already complex and poorly understood process involving atomic hydrogen and hydrocarbon species requires considerably more time and effort, in spite ofthe potential importance ofthe approach from an energy and economic viewpoint. The advantages and disadvantages of the diamond CVD techniques, along with typical technical data, are summarized in Table 1,[31[641[65] with a survey of typical growth rates versus gas-phase temperatures given in Fig. 1. Typical reactor operating conditions with C/Hmixtures are listed in Table2, and a schematic diagram of the diamond CVD techniques is depicted in Fig. 2. Table 1. Typical Technical Data and Characteristics

Method Hot-filament DC discharge (low P) DC discharge (medium P) DC plasma jet RF (low P) RF Ithermal, 1 atm) Microwave lo.9 - 2.45 GHz) Microwave (ECR 2.45 GHz) Flame. (combustion)

Rate (clrn h-l) 0.3 - 40

Area (cm2) 100-400

Quality’ (Raman) +++


70

+

20 - 250

<2

+++

10 - 930

2-100

+++


1-

10

-I+

3 - 78

+++

1 (low P) 30 (high P)

40

+++

0.1

<40 1 - 100

30-500

30-200

* - = poor quality. +++ = excellent quality

Substrate material Si, MO. silica, Al2O3. etc. Si, MO. silica, Al2O3, etc. Si, MO. Al203

of Diamond CVD

Advantage

Disadvantage

Simple, large area Simple, area

Contamination, stability Quality. rate

Rate, quality

Mo,Si.W.Ta.Cu,Ni, Ti. stainless steel Si, MO, silica, BN, Ni MO

Highest rate. quality Scale-up

-I+

Si. MO, silica, WC, etc. Si

Quality, area, stability Area, low P

+++

Si,Mo.A1203,TiN

Rate

Simple, rate

Area Stability, homogencitv Quality, rate, contamination Stability, homogeneity Rate Quality,rate.cost. contamination Stability, unifomuty

16

Diamond Chemical Vapor Deposition

sure microwave hot filament

RF gbw discharges 0.01 1000

I

I

I

I

I

2ooo

xboa

4ooo

!ilxQ

6000

gas-phase

temperature

CVD processes

in diamond

7000 [K]

Figure 1. Typical growth rates versus gas-phase temperatures in various diamond CVD processes, illustrating the importance of gas-phase temperature for high rate diamond synthesis.[64] (Reproduced with permission.)

Table 2. Typical Reactor Operating Conditions in Diamond CVD Techniques Deposition

HFCVD

condition [15,65,701 Gas 0.1 - 2 vol.45 cH4 in H2 composition Gas pressure 10 - 100 toll Gas flow rate 20 - 500 seem Subsuate 7OO- 1OOO’C tempera* Filament 2000 - 2400 ‘C temperature Power

MW PACVD 12.15,64,65,71-731 0.2 - 2 vol.% CHq in H2 10 - 100 torr 20 - 1000 xcm 800- 11oO’C -

600- 15OOW

DC Plasma Jet [64.74-771 1 - 3 vol.% CH4 in H2 and Ar 60 - 760 torr, or 1atm 3ooO - 70000 seem 800- 11OO’C

Flame CVD [64,70,781 0.44 C2H2, 0.19 Hz. 0.37 02 - I atm 1000 - IWO0 seem 800- 11OO’C

-

-

I-50kW

-

Diamond CVD Techniques

17

The neutral species gas-phase chemistry in diamond CVD reactors generally consists of a well-understood set of pyrolysis and oxidation reactions, long studied in the combustion community. Several detailed combustion mechanisms exist for non-sooting conditions typically encountered in diamond CVD 1791-1811 as summarized by Coltrin and Dandy,ls21 and Meeks et a1.l83lfor analysis of diamond CVD. Detailed chemical kinetic models of the gas-phase and surface chemistry in diamond CVD have been presented by Tsuda et al., W1[851Fre&la&et a1.,Wl+‘31 &-r-s eta1.,[941-[‘0’1 and others.l82ll1o2ll1o3l Due to uncertainties in identification of important growth species, these models would have to be considered somewhat speculative. While the gas-phase chemistry is viewed to contain little uncertainty, the surface reaction mechanisms in diamond CVD are not as well understood and have garnered a great deal of attention. Research on the surface reaction mechanisms is ongoing and it will be some time before an operative mechanism is universally agreed upon. The various surface mechanisms developed to date typically incorporate some or all of the following general features. The mechanisms allow for the growth of both diamond and graphite phases, with the dominant phase depending on reactor operating conditions; interconversion between graphitic carbon and diamond under conditions of high H atom concentration is included. A single growth precursor for diamond deposition has not been identified unambiguously. Indeed, consistency between a given model and experimental growth rates is not proof of the identity of the growth species. In addition, the most abundant gas-phase species will depend upon reactor operating conditions. For example, modeling studies predict an abundance of gas-phase C atoms under plasma arc-jet conditions,l82ll’02l but few C atoms in hot-filament simulations. [lo41 Therefore, the mechanisms often consider the growth of diamond (and graphite) from more than one precursor (CH,, C2H2, CH, and C atoms). Generally, the first step in the surface mechanisms is the abstraction of a surface H atom by a gas-phase H atom. After the abstraction a reactive radical site is left on the surface. This surface radical site can be terminated efficiently via the addition of another gas-phase H atom to the site, and this reaction is more facile than the initial H abstraction reaction. These two reactions-H abstraction and H termination-are fast compared to the other rate processes occurring during the deposition. The ratio of rates ofthesetwo reactions governs the fraction of open (radical) sites on the diamond surface.

18

Diamond Chemical Vapor Deposition

Hydrocarbon species, particularly the CH, (x = O-3) radicals, compete with atomic H to cap the open radical sites, although the hydrocarbon adsorption reactions are much slower than the H termination reaction. Once a gas-phase hydrocarbon has been adsorbed onto the surface or has inserted into a surface bond, subsequent reactions take place to form additional bonds with adjacent surface carbon species, eventually concluding when a carbon atom has formed all four (if diamond) or three (ifgraphitic material) of its lattice bonds. In the following, various CVD techniques for the growth of diamond crystals and films on deposition substrates are discussed in detail.

1.0

HOT-FILAMENT

CVD

HFCVD, developed by Matsumoto and co-workers at NIRIM,t2’lt2*l is probably the simplest and most reproducible way to grow diamond at low pressures, and appears to be the easiest to understand experimentally and conceptually. It is also the first method to achieve nucleation and continuous growth of diamond on various substrates.13’l In a HFCVD process (Fig. 2a), a gas mixture containing 0.1-2 vol.% CH, in Hz enters a reactor that is evacuated to approximately lo-100 torr, and flows past a wire or mesh made of a metal such as W, Ta, MO, or Re, heated to about 2000-24OOOC (Tables 1 and 2). Under these conditions, 210% of H, is dissociated into atomic H, and CH, undergoes pyrolysis reactions leading to the formation ofradicals such as CH, and CH,, and stable species such as C,H,, C,H,, and C,H,. Diamond is deposited on a substrate made of Si, MO, or silica, etc., which is mounted at a distance of 0.5 to 2 cm from the glowing filament and kept at 700 to 1000°C either by the radiation from the filament or by a separate substrate heater. Deposition rates typically vary from 0.3 to 20 urn h-l. Coated areas as large as 20 x 20 cm2 have been achieved. By adding oxygen-containing growth species, such as carbon monoxide, water, or molecular oxygen, deposition rates can be increased to as high as 40 urn hs1.131Growth rates up to 10 urn h-’ in an extended pressure range up to 1 atm have also been reported for diamond film growth with up to 10 vol.% acetone in hydrogen.l105] However, oxygen-containing gas atmospheres can lead to severe corrosion of the hot filament and none of the high growth rate experiments in HFCVD reactors have proven viable. It is found that a deposition rate of 1 urn h-’ is typical for high-quality films.131

Gas Manifold

(a)

(b>

Figure 2. Schematic diagram of various CVD techniques for diamond synthesis. (a) HFCVD; (71)MW PACED, (c) ECR h4W PACVD; (d) DC PACVD; (e) RF PACED, fl DC thermal plasma CVD, (gj RF thermal plasma CVD; (h) flame (combustion) CvD.[‘*] (Reproduced with permission.)

Water

Substrate

Water- mJ

Gas mlet

Figure 2 (Cont’d.)

Water 4t-h

t”

4

RF generator 113.56 MHz1 “.rnr.“.

Figure 2. (Cont’d)

a,.-

Carder + Reactant Gas

Plasmafw Sheath gas (S) -1

(PI -

Cd inr Holdbr

111 1

3 in Wortlc

0

’

RF

Settling

Chamber

Dmpinc Scrrenl

LL

Substrate holder

Figure 2. (Cont’d)

Water

Diamond CVD Techniques

23

The HFCVD is a good choice for many applications in which an ultra clean film is not mandatory. However, the HFCVD method has some severe limitations associated with long-term stability of the filament and the contamination of the growing film caused by metal evaporation from the filament. The stability of the filament is affected primarily by the corrosion of the filament and the formation of carbides on the filament during the incubation period and during deposition. The filament may then sag or distort, leading to unsteady deposition conditions. Uniform temperature distribution and gas flow are the factors of major importance for obtaining good quality diamond coatings when scaling up the HFCVD.t301 A variation of the HFCVD method employs a bias voltage that is applied between the filament and the substrate.t’06j Electrons emitted from the filament accelerate toward the substrate, while heavier ions in the gas phase are repelled by the positively charged substrate surface. Electron impact enhances not only the reactivity of graphite (if formed) with atomic hydrogen through excitation of 71:electrons, but also the fragmentation of hydrocarbons at or near the substrate surface.t’51 The positive electric potential (>I20 V) of the substrate relative to the filament improves the growth rate, enhances the nucleation density, and permits the deposition of diamond at lower substrate temperatures.p51 However, biasing may be detrimental to the film quality. [lo71 If the bias voltage is sufficiently high, a glow discharge between the substrate and the filament may occur. Such a system is an embryonic form of the direct-current plasma methods discussed below.

2.0

PLASMA-ASSISTED

CVD

A plasma is an efficient way to dissociate gas molecules to produce non-equilibrium concentrations of gas-phase species, such as the high concentrations of atomic hydrogen needed for diamond growth. Plasmas can be generated by a number of energy sources (microwave, radio-frequency, or direct-current electric fields), and can be either cold (non-isothermal, or nonequilibrium plasmas) or hot (isothermal, or equilibrium plasmas). The major characteristics of these plasmas are summarized in Table 3.

24

Diamond Chemical Vapor Deposition

Table 3. Characteristics

Plasma type Frequency Power Electron concentration Pressure Electron temperature

of Plasmast21 Glow discharge

Arc-jet

non-isothermal

isothermal

(non-equilibrium)

(equilibrium)

50 kHz - 3.45 MHz 2.45 GHz (microwave)

-1 MHz

l-1OOkW IO9 - 1012 cmS3 2- 1OOtorr

1 kW-20MW 1014 cmm3 lOOtorr- 1 atm

104K

104K

500 K

104K

Aglow discharge plasma is generated in a low pressure gas by a highfrequency electric field, for example, microwave. Within the electric field, gas is ionized into electrons and ions. The electrons are quickly accelerated to high energy levels corresponding to 5000 K or higher. The heavier ions cannot respond to the rapid change of the field direction, and thus their temperature and that of the plasma remain low. Hence, it is referred to as nonisothermal. The high energy electrons collide with the gas molecules, leading to activation of the gas phase, i.e., dissociation of gas species and generation of reactive precursors for diamondgrowth. Atomic hydrogen concentrations do not generally exceed 10% in glow discharge plasma devices. As a result, the hydrocarbon pyrolysis chemistry produces CH, as the dominant single carbon species, and the growth rates are observed to be proportional to CH, concentration at the substrate surface. Aplasma arc-jet is generated in a gas between two electrodes by either direct or alternating current at high power. In this high-intensity, lowfrequency arc, both electrons and ions respond to the relatively slowly changing field direction, acquire energy, and increase in temperature at comparable rates, reaching 5000 K or higher. Hence, it is referred to as isothermal. At such high temperatures, a large fraction (over 99%)t1081 of hydrogen molecules inthegas phase dissociate, creating a “superequilibrium” of atomic hydrogen, and the pyrolysis reactions strip all ofthe hydrogen from the methane feed, forming atomic carbon-a unique feature of this type of

Diamond CFD Techniques

25

plasma. In addition, a plasma arc is generated at a higher pressure than in glow discharge plasmas, so that the mean free path is reduced. As a result, molecules and ions collide more frequently, and heat more readily. Electrons in plasmas respond to electric fields more quickly than heavier ions due to their smaller mass. This higher kinetic speed makes electrons much more energetic than ions. Consequently, electrons are the dominant charge carriers in plasmas. Due to their higher mobility, electrons recombine at a much faster rate with reactor chamber walls than ions, generating a net positive charge in plasmas. A steady state potential called plasma potential is eventually reached. This potential is several volts more positive than the chamber potential. If a floating (ungrounded) surface (for example, a substrate) is placed in contact with a plasma, it will charge negatively due to the faster rate of bombardment of electrons. An equilibrium surface or floating potential (self-bias) will arise which is negative with respect to the plasma potential. Since the bulk plasma is ofuniform potential, this voltage drop is confined to a thin region above the substrate surface, called plasma sheath. Electrons are further accelerated across the sheath, making this region of very low electron density. With fewer electrons, a reduced number of radiative recombinations occur, giving rise to the “darker” sheath region above the substrate than the rest of the plasma. The influence of the width and electric field intensity of the plasma sheath on diamond thin film growth in CVD has been well documented in related literature.l’0gl-11121 PACVD (Figs. 2b-g) may be further subdivided into the following four major types in terms ofthe methods with which plasmas are generated and the characteristics of plasmas, as listed in Table 4.

Table 4. Four Major Types of PACVD Non-isothermal (a) Microwave

Plasma -

MW PACVD [29,31,321

l

magnetic field at high pressure

. ECR (co.1 tot-r )

(b) Direct Current Plasma -

l

capacitively

DC PACVD [35,361

l

hollow cathode discharge

(c) Radio-Frequency

Plasma -

RF PACVD (33,341

coupled

l

inductively coupled

l

capacitively

l

DC thermal plasma torch

l

RF thermal plasma torch

coupled

Isothermal (d) Thermal Plasma CVD [37,38]

26

Diamond Chemical Vapor Deposifion

In these PACVD methods, gas compositions are similar to those used in the HFCVD,I’51 although gas flow rates in the HFCVD are up to a factor of 1O3lower than in some ofthe PACVD techniques. Substrate temperatures are determined by power densities, placement of substrates, and water cooling rates of substrates, usually maintained at about 800-l 100°C. Gasphase temperatures depend on the method used, i.e., whether the plasma is isothermal or non-isothermal. 2.1

Microwave

Plasma-Assisted

CVD

Research on MW PACVD began from the pioneering work of Kamo et a1.t291The MW PACVD (Tables 1 and 2) is, aside from the HFCVD, the most frequently used method for diamond growth and also the most extensively studied process. The intensive use of the MW PACVD is prompted by the following factors: 1. The stability and reproducibility of microwave non-isothermal plasmas, allowing continuous deposition for tens or hundreds of hours 2. The high energy efficiency (high plasma density and low sheath potentials) 3. The increased availability of l-2 kW microwave (waveguided) power supplies and applicators 4. The potential to scale up the process to larger substrates In commonly used systems, as schematically depicted in Fig. 2b, carbon-containing gases diluted in hydrogen (typically 0.2-2 vol.% CH, in H2) are introduced into a silica tube connected to a vacuum pump and surrounded by a microwave applicator. The tube is typically pumped down to lo-100 torr. At such pressures, an electrical discharge in the gas mixture may be readily sustained by means of microwave radiation at a frequency of 2.45 GHz, the same frequency as in a home microwave oven. The carboncontaining gases and hydrogen molecules are then dissociated within such a discharge. A substrate made of Si, MO, silica or WC, etc., is placed in the middle of the plasma and heated to 800-l 100°C by a radiant or resistance heater so that accurate and stable temperature control may be achieved.I*l Diamond then nucleates and grows on the substrate surface. In spite of some ofthe advantages ofthe MW PACVD over other CVD methods, notably its stability, the deposition rates currently achievable in the

Diamond CKD Techniques

27

MW PACVD remain a drawback. Deposition rates up to 3 0 pm h-’ have been reported, yet for the growth of high-quality films with graphite-free, faceted crystals, the rate is only about 1 pm h-l, and uniform coatings are currently limited to diameters of less than 5 cm.131 Since Hirose and Terasawalio51 demonstrated that higher growth rates and better film quality could be achieved at higher pressures and carbon concentrations using oxygencontaining compounds to supplement the CH, and HZ feedstock, compounds such as O,, H,O, CO, CO,, alcohols, ethers, etc., have been increasingly used as constituents of feed-gas mixtures. Care must be taken, however, because the stability of plasmas may be easily disturbed by the addition ofthe oxygencontaining compounds.12] A microwave plasma can also be generated by electron-cyclotronresonance (ECR). Cyclotron resonance is achieved when the frequency of an alternating electric field is made to match the natural frequency of the electrons orbiting the lines of force of a magnetic field. This occurs in a magnetic field of 875 gauss at the standard microwave frequency of 2.45 GHz. An ECR plasma CVD reactor is shown schematically in Fig. 2c. The gas pressure in the reactor is usually very low (10 mtorr to 0.1 torr). The ECR plasma CVD can minimize the potential substrate damage caused by high-intensity ion bombardment, usually found in a standard high frequency plasma. It can also minimize the risk of damaging heat-sensitive substrates due to its relatively low operation temperatures. The disadvantages of this method (Table 1) are the low growth rates, costly equipment and difficult process control due to the additional complications arising from the magnetic field. 2.2

Direct-Current

Plasma-Assisted

CVD

A typical DC PACVD system is schematically depicted in Fig. 2d, which is based upon the system developed by Suzuki et a1.l35lin 1987. In this setup, a glow discharge occurs in the reactor between the substrate and the other electrode when the bias voltage is sufficiently high. The activation of the gas phase is achieved through the collision of high energy electrons with the neutral gas molecules, leading to the dissociation of the gas species and the generation of reactive precursors for diamond growth. In all DC PACVD methods, substrates are mounted on the anode of the deposition system. Mounting substrates on the cathode results in deposition of graphite or carbides rather than diamond.t31115] A negative substrate bias, however, can significantly enhance nucleation of diamond in the MW PACVD.1711

28

Diamond Chemical Vapor Deposition

In DC glow discharge PACVD systems, larger substrate plates may be used, with coated areas reaching 70 cm2 (Table 1). The gas pressures in these systems are usually on the order of 40 torr. While the DC PACVD method is simple in design and use, it suffers from a very low growth rate (~0.1 urn h-l), making it impractical for many industrial applications. This drawback is mitigated by applying a high DC voltage of more than 1 kV together with a high discharge current.131 A plasma can be sustained at a pressure of -150 torr. Under these conditions, deposition rates may be increased to the range of 20-250 urn h-l, but deposition areas have to be reduced to less than 2 cm2.1641 2.3

Radio-Frequency

Plasma-Assisted

CVD

A typical RF PACVD system is schematically depicted in Fig. 2e, as developed by Matsumoto in 198.5.1331 A RF plasma is a glow discharge plasma, generated in a gas at relatively low pressures (-1 torr) by highfrequency electric field at a frequency of 13.56 MHz. The high energy electrons collide with the gas molecules, leading to the activation of the gas phase. The RF PACVD is more likely to produce diamond-like carbon (DLC) rather than pure diamond. 121Low deposition rates (CO.1 urn h-l) are also a drawback of this method, in spite of the potential of scale-up to large deposition areas (Table 1). 2.4

Direct-Current

Thermal

Plasma CVD

A common DC glow discharge PACVD system may be varied by operating the plasma at higher pressures and powers at which a DC arc discharge between the electrodes can be produced. In 1988, Kurihara et a1.l37lfirst reported the use of the DC plasma arc-jet CVD method, in which methane was mixed into a H,/Ar plasma jet at 25 kPa (-188 torr). In this approach, as schematically shown in Fig. 2f, the substrate is mounted on a separate water-cooled holder, rather than on one of the electrodes. The electrodes usually consist of a water-cooled copper anode and a tungsten cathode, although different cathode materials have been examined. Several gas-jet nozzles may be operated simultaneously and many design variations are possible, such as separate input nozzles for hydrogen and methane, feeding of these gases in a coaxial feed electrode,121 and remote injection of methane.l74l The sudden expansion of the gases into the reactor chamber as

Diamond CVD Techniques

29

they are heated in the plasma arc of H,/Ar introduced at a speed up to several hundred shn gives rise to the formation of a high-speed arc-jet. At the high gas temperatures (>5000 K) in the isothermal plasma, a large fraction of the hydrogen molecules dissociate and a sufficient supply of the active carbon species necessary for diamond growth is created. The atomic hydrogen and the activated carbon species in the high-speed jet are transported to the deposition surface within a few hundred microseconds. Hydrogen recombination reactions in the gas phase are hence reduced. The substrate may be heated to an unacceptable level by the high temperature gases, and active cooling is usually necessary. Temperature control and substrate cooling remain a problem in thermal plasma CVD systems. The distinct advantage of the thermal plasma CVD method over others is the high deposition rates (Table l), which are attributable to the higher concentrations of activated carbon species and hydrogen atoms, and higher mass transfer rates. l1131 The highest known deposition rate, 930 pm h-‘, has been achieved using the DC plasma jet CVD by Ohtake and Yoshikawa,*421 and thick deposits are routinely produced.t1141 The plasma jet can be cooled rapidly prior to impact on the substrate surface by mixing with a cold inert gas fed into an annular fixture. Gaseous boron or phosphorous compounds can be introduced into the gas feed for the deposition of doped semiconductor diamond.t1151

2.5

Radio-Frequency

Thermal

Plasma CVD

Reaction conditions similar to those in the DC thermal plasma CVD can be obtained in an inductively coupled, radio-frequency induced thermal plasma at atmospheric pressure, as shown schematically in Fig. 2g. This technique was first reported by Matsumoto et al. in 1987.t3*l Very high growth rates, 300 to 500 urn h-i, have been achieved by Matsumoto and also by Koshino et al. 131Matsumoto found that RF thermal plasmas are difficult to control and tend to be unstable, whereas the DC thermal plasma CVD method is easier to use but contamination from the plasma nozzle may deteriorate the quality of the as-deposited films. In spite of the poor energy efficiency and difficult process control in both the DC thermal plasma CVD and the RF thermal plasma CVD, the high growth rates achieved with these methods represent a milestone in diamond CVD technology. Within less than a decade, deposition rates have been

30

Diamond Chemical Vapor Deposition

increased more than three orders of magnitude, exceeding the 100 pm h-’ mark considered necessary for CVD to be commercially competitive with HPHT synthesis methods. The DC and RF thermal plasma CVD methods are being actively investigated to attain uniform, large-area, high-speed diamond deposition and improvement in process understanding and contro1.11161-11211 Recently, deposition areas up to 10 cm2 have been achieved in the DC thermal and uniform diamond films up to 78 cm2 plasma CVD by Eguchi et al., 11201 have been grown in the RF thermal plasma CVD at a rate of 30 pm he’ by Kohzaki et a1.,16’lwhich is the largest deposition area of diamond films ever grown with the RF thermal plasma CVD method. Norton in the U. S . can now routinely grow 4.5~inch diameter wafers using DC plasma jet deposition. Attempts have also been made to enlarge the diamond deposition area by changing the arrangement of plasma torches, for example, developing a new type of plasma arc-jet CVD apparatus using one cathode and three anodes.I1lgl

3.0

FLAME

CVD

Combustion synthesis of diamond films from atmospheric pressure oxyacetylene flame was first reported by Hirose in 1988.15’l This technique, as shown schematically in Fig. 2h, has been demonstrated to be a potentially very important means of growing diamond (Table 1). In Hirose’s experiments, linear growth rates of 100-200 pm h-l were achieved. The results were then confirmed by Hanssen et al. 1581In Hirose’s experiments and most of those that followed,[58][1221-[126] th e oxyacetylene torch was typically run with equal volume flow rates of acetylene and oxygen. The mole fractions of the gas mixture are typically 0.44 C2H2, 0.19 H2, and 0.37 021781(Table 2), with [O,]/[C,H,] ranging from 0.83-1.35.168111271 In a flame CVD process, a small premixed flame issues from a nozzle, surrounded by a diffusion flame where excess fuel and CO continue to be oxidized. While hydrogen, oxygen and acetylene are burned, diamond forms on the deposition substrate positioned in the reducing part ofthe flame at a substrate temperature ofabout 800-l 100°C and a gas temperature of about 2000°C in the immediate vicinity of the substrate surface. Extensive studies on the mechanisms of diamond formation from acetylene flames have been conducted by Matsui et a1.11271~li2gl

Diamond CCI) Techniques

31

The atmospheric flame synthesis is simple in design and use. In addition, it shows great promise for high growth rates. Currently, deposition area is small, and stability and uniformity in film structure and composition are not readily achieved. However, by building upon flat flame burner design experience, scale-up to arbitrary areas is feasible. A number of scale-up approaches have been suggested, such as the use of multiple torches or painting the deposition surface by moving a single torch, but these will probably have limitations in achieving large-area film uniformity. As an alternative, premixed flat flames have recently been investigatedl83l and employed in low pressure diamond synthesis,[‘30] in which premixed fuel and oxidizer issue through a flat burner face towards a parallel deposition surface. Another alternative approach using a flat plate perforated with many small holes has also been studied.1681t6gl The configuration of the premixed flat flames is inherently scalable to large deposition areas aside from some unavoidable edge effects. The high gas temperatures (>2OOO”C)and high heat fluxes in the flame CVD make it mandatory to cool substrates. As a result, large temperature gradients are formed which are difficult to control. The deposition efficiency is low due to generally low nucleation rates. Further, gas consumption, energy requirements and hence cost are high.

4.0

GENERAL CHARACTERISTICS PROCESSES

OF DIAMOND

CVD

In spite ofthe differences between the various CVD methods developed so far, it is interesting to note that there exist many common points for these techniques, as summarized below. 4.1

Crystallite Morphology

Single-crystal diamond can be grown epitaxially on a single-crystal diamond or cBN substrate, yet diamond thin films deposited on other nondiamond substrates in various CVD processes all exhibit a generally similar, polycrystalline morphology, consisting of randomly oriented crystals and containing a varying amount of non-diamond carbon and defects. Crystallite size may range from several tens of nanometers to several tens of micrometers 91311411701 depending primarily on gas composition, flow rate and pressure,

32

Diamond Chemical Vapor Deposition

substrate temperature and surface pretreatment, as well as film thickness. At substrate temperatures of SOO-lOOO’C, crystallite size in diamond films of about 5 pm in thickness is typically l-5 pm.t41 Large grains (l-50 pm) are also common in CVD polycrystalline diamond films, approaching 100 urn in thick films (-200 pm in thickness).1701 The top surface of films is usually rough and highly scattering due to the large grain size and random orientation ofthe crystallites that make up the films, limiting optical, electrical and tribological applications. Two approaches have been suggested to overcome this drawback: (a) single-crystal films (essentially zero nucleation density) and (6) submicron grain-sized films (extremely high nucleation density). The latter is a better approach for applications in which deposition occurs on non-diamond substrates and for which mechanical properties of the deposited films are important. Various polishing methods have also been developed and applied to smooth the diamond film surface on the growth side.11311-11351 The polycrystalline morphology may change from well-defined facets to poorly-defined rounded shapes, or highly irregular forms. Within the optimum range of deposition conditions, diamond films are highly faceted.1541 Independent of the deposition methods used, octahedral crystals with { 1 1 1} faceting are the dominant growth form at low temperatures and low carbon supersaturations, i.e., low gas-phase carbon concentrations;I’51 with increasing temperatures and/or carbon concentrations, the morphology evolves to cube-octahedra composed of both { 11 l} and { lOO} facets, then to cubic { lOO} facets, and finally to a spherical shape at the highest supersaturation.11361 Twinning or stacking fault occurs frequently on { 11 l} planes. Spherical clusters of diamond microcrystals and flat hexagonal platelets, as well as multiply twinned decahedrons and icosahedrons are also observed.In’l In MW PACVD experiments, 11361t1371 at a substrate temperature of about 800°C and a gas pressure of 30 torr, { 11 l} faces dominate diamond crystals at CH,concentrations of CO.4 vol.%; at CH,concentrations between 0.4 and 1.2 vol.%, { 100} faces are prevalent; and at even higher CH, concentrations, the deposits become structureless. Another study 113*1 showed that at temperatures of 900°C and lower, { 11 l} faces are dominant in the crystallite morphology, and at 1000°C or higher, { lOO} faces are dominant. These results suggest that low substrate temperatures and low CH, concentrations favor { 11 l} faces, whereas high substrate temperatures and high CH, concentrations favor { 100} faces. Consistent observations have been

Diamond CVD Techniques

33

It should be noted, however, reported by other researchers. [151[261[661[13g1[1401 that the influence of temperature on faceting during diamond CVD is complex, and the change of predominant faces from { lOO} to { 11 l} with increasing temperature has also been reported.t541

4.2

Gas-Phase

Activation

Gas-phase activation above the deposition surface is essential for achieving appreciable diamond growth rates. The various CVD methods differ primarily in the way they produce gas-phase activation. The most abundant carbon-containing gaseous species present in most activated systems are methyl radicals and acetylene molecules which are also considered to be predominant growth precursors for diamond, almost independent of the deposition methods used. However, in systems that dissociate a significant fraction of H2, such as DC plasma arc-jet CVD, carbon atoms, aside from acetylene, are also abundant in the gas phase. One of the critical factors affecting growth rates is the gas-phase temperature which can be reached in a CVD method. A comparison of the various CVD methods to each other demonstrates[64l that typical linear growth rates correlate positively with estimated gas-phase temperatures (Fig. l), approaching 1 mm h-’ in atmospheric pressure plasma arc-jet CVD with temperatures around 6000 to 7000 K. The partial pressures of various gas species in typical CVD processes have been calculated as a function of the gas-phase temperatures and summarized in Fig. 3 for two gas mixtures. High rate CVD methods obviously operate at high temperatures where neutral H and C atoms dominate the gas phase. In the complex C-H-O systems, CO remains present at high temperatures, but the CO partial pressure hardly varies with temperature from 1000 to 6000 K and hence would not explain any rate change. Angus et a1.t1131indicated that linear growth rates in different diamond CVD processes increase approximately linearly with the area1 power densities employed. It is also suggested 11131that the higher concentrations of hydrogen atoms and other activated carbon species, and the higher mass transfer rates are responsible for the much larger growth rates observed in atmospheric pressure DC or RF thermal plasma or flame CVD processes than those in low pressure MW, DC, or RF plasma or HFCVD processes.

34

Diamond Chemical Vapor Deposition

W: -e-H

-4-H

a-c2 -eCH + A

gas-phase

temperatures

4 C2”2 C2H

[K]

(a)

LOQWKI:

%-H +H

2 4-C + CH4 a C2H2 -L C2H -= “2 -A- Co -0 CH -*cn2 -0% -*c3 -* 0 -0 H20 0

loo0

2wo gas-phase

3ooo

4oal

temperatures

5ow

woo

[K]

(b) Figure from a at 37.5 which

3. Calculated temperature dependence of partial pressures of gas species formed gas mixture of (a) 0.5 vol.% CH, in H,, and (b) 53 vol.% C,H, and 47 vol.% O2 torr. The shaded areas represent the approximate gas-phase temperature regimes can be reached in different diamond CVD methods.[64] (Reproduced wiih

permission.)

Diamond CVD Techniques

4.3

35

Gas Species and Gas Compositions

Diamond of similar quality and morphology has been grown using a variety of species, including aliphatic and aromatic hydrocarbons, ketones, amines ethers PlPl[471 &&&,U 1W211[1411~1421 carbon ,ono~de,[6011431-[147J carbon dioxide,11481t14gland halogen (CC14/H~‘501 and CF4/H2t1471t151J). Methane has been the most frequently used gas. Large hydrogen gas dilution, typically 0.1-2 vol.% CH, in H,, is necessary with a trade-off between high rates (at about 5-6 vol.% CH4t4111521) and large crystal size and good quality (at about 0.2-0.5 vol.% CH4141).Increasing CH, concentration results in an increase in growth rate I1531and a decrease in crystal size due to increased nucleation density and number of secondary nucleation on the exiting facets of diamond crystals. ~1 High CH, concentrations may lead to the formation of amorphous carbon or diamond-like carbon. In flame CVD, low 02/C2H2 ratio may lead to the formation of DLC on Mot’241 and suppress the formation of SiO, on Si. Diamond films of good quality have been grown from CO at a concentration of 50 vol.% or even higher.t41 However, it is found that the efficiency of diamond growth from CH, (possibly via CH, radicals) is about two orders of magnitude higher than that from CO, and the growth of diamond from CO occurs via the conversion of CO to hydrocarbons.t1431 The high thermodynamic stability and kinetic inertness of CO make it a relatively inefficient carbon source for diamond deposition.l143l Nevertheless, additions of oxygen and/or oxygen-containing species to HZ-hydrocarbon gas mixtures can lead to higher growth rates and better film qualities, and allow lower growth temperatures and higher carbon concentrations.l65l The presence of halogens in reaction systems has been found to lead to an increased activation of the deposition surface, especially at low temperatures.t471 The rate coefficient of the abstraction reaction of surface H atoms by gaseous F atoms is more than two orders of magnitude greater than that of the abstraction reaction by H at 900°C or lower.t471 The deposition of diamond films at substrate temperatures as low as 250°C and relatively mild gas activation has been achieved in which fluorine was added to the gas mixture. 11541In addition, the sp2 carbon components may be preferentially etched by F atoms. 11541The use of chlorine-permuted methane as a carbou source also favors high-purity diamond deposition at low temperatures and high methane concentrations.t1551

36

Diamond Chemical Vapor Deposition

The gas compositions used in a number of diamond CVD experiments from more than 30 years are summarized in Fig. 4, which also provides a common scheme for all major diamond CVD methods used to date. It can be seen that the gas compositions suitable for diamond deposition are restricted to a well-defined area within the diagram, independent of the deposition methods or carbon species used.

0 n

diamond no growlh

@

nondlmnondcarbon

0

porlllon of undllulal compound otbnlaflon llno

,b*’ flmlf of dkmond domaln /

WI of connacled wqwlnunfal da@

Figure 4. C-H-O phase diagram showing points/regions of gas compositions suitable for diamond growth as well as non-diamond carbon growth region and no-growth region.[56] (Reproduced with permission.)

4.4

Gas Flow Rate and Pressure

While gas flow rates vary between the different systems, ranging from 20 to 300,000 sccm,141[1511651 film quality and to a lesser extent, deposition rates are relatively insensitive to gas flow rates.141

Diamond CVD Techniques

37

Gas pressures may range from 10 mtorr in ECR MW PACVD[‘451 to atmospheric pressure in flame CVD,i70] depending on deposition techniques.[156]-[158] F or example, the gas pressure is typically 10 to 100 torr in HFCVD, DC PACVD, and h4W PACVD, and the plasma may begin to become unstable at gas pressures above 100 torr. A DC glow discharge plasma may be sustained at a medium pressure of about 150 torr, whereas DC or RF thermal plasma CVD and flame CVD may operate at 1 atm. A relatively low gas pressure favors a high nucleation density and a small crystal size, [1561whilea relatively highgas pressure fosters ahigh linear growth rate (Fig. 5). [22] A low gas pressure also lowers the rate of atomic hydrogen recombination in the gas phase.

Figure 5. Temperature and pressure dependence of growth rates of epitaxial diamond films on (111) planes of natural diamond crystals (gas pressure = 12, 66, and 162 torr, respectively); the maximum growth rate in the experiments, 3 pm h-‘, was attained at a gas pressure of 760 torr.lz21 In another experiment, a higher growth rate was also achieved at atmospheric pressure (see Fig. 8). (Reprintedjom ReJ 22, 0 1988, with kind permission jPom Elsevier

Science

Ltd, The Boulevard,

Langford

Lane,

Kidlington,

OXSlGB,

UK.)

A recent study on DC thermal plasma CVD of diarnond[r’*] demonstrated the existence of an optimum pressure at 270 torr, at which the

38

Diamond Chemical Vapor Deposition

maximum growth rate occurs. This optimum pressure was related to the balance between generation and recombination of atomic hydrogen and active carbon-containing species above the substrate surface. Brunsteiner et a1.t15glinvestigated the influence of gas pressure on the growth of diamond films on SiAlON substrates from CH,iHz in HFCVD. The total gas pressures varied from 5 to 500 torr and the filament temperatures were 2200-25OOT. In the lower pressure range (5-10 torr), 0.5 vol.% CH, was the optimum concentration for producing well-faceted diamond, while concentrations above 1.0 vol.% CH, led to ballas-type diamond. Between 20 and 300 torr, faceted diamond films could also grow at high CH, concentrations (1 .O vol.%), while the highest deposition rate (1.44 urn h-l) was obtained at 20 torr, with a filamenttemperature of25OO”C. From 50 to 500 torr the deposition rate decreased with increasing gas pressure, even falling below 0.05 urn h-i at the highest pressure. This drastic decrease in the overall deposition rate with pressure is related to the drastic decrease in diamond nucleation rate. The observed effects of gas pressure on diamond nucleation and growth were also attributed to the changes in the formation and recombination rates and the resultant concentrations of atomic hydrogen with pressure. In another HFCVD experiment, 11531 the influence of gas pressure on diamond film coverage and crystal size was studied. As shown in Fig. 6, the crystal size and surface coverage attain a maximum at 1.3 kPa (-10 torr). Crystal quality and phase purity are both optimized around a pressure value of4 kPa (30 ton-). At 665 Pa (-5 ton) an amorphous film covers the substrate and ball-like particles form.

0.y 0

-+-~sur.~ 2

4

6

1 8

10

12’

Pressure @Pa) Figure 6. Crystal size and film coverage versus gas pressure; growth conditions: 0.5 vol.% CH, in H,, 22OOT filament temperature, 75O’C substrate temperature, 5 mm tilamentsubstrate distance, 45 seem total flow rate and 3 hour run duration.[‘53] (Reproduced with permission.)

Diamond Cvz) Techniques

4.5

39

Substrate Materials and Pretreatment Methods

Diamond has been nucleated and grown on diamond, plus a large variety of non-diamond substrates, including metals, semiconductors, insulators, graphite, and even fused silica glass, such as MO, W, WC, Ta, Cr, Co, Pt, Au, Al, Cu, Ni, Fe, stainless steel, NiAl, NisAl, FeSi,, Ti, Ti-2Al-1 SMn, Ti-6Al-4V, TiN, TIC, Si, SIC, S&N,, silica, SiAlON, MgO, Al,O,, cBN, and Y_ZrO, ~~~~~~1~~~~1~~~1~~~1[1171[160l~~t~~l h ong these, Si wafer is probably the most commonly used substrate in diamond CVD. Diamond nucleation on a substrate surface critically depends on the chemical nature Ill71 and surface condition of the deposition substrate.[52a] Figure 7 shows the dependence of the nucleation and growth of diamond in CVD on the substrates used. There is usually an initial incubation period in diamond CVD, i.e., a delay in nucleation before individual crystallites can be observed, followed by a delay in growth before these crystallites grow and coalesce together to form a continuous film.

2.26 z 1.06 f.

6

lo

&noJtlon time

16

20

[hrrl

Figure 7. Nucleation and growth of diamond on different substrates CVD.[S2a] (Reproduced with permission.)

in low pressure

Diamond Chemical Vapor Deposition

40

Substrate

surface pretreatment (scratchingt73)[‘70]t’72]-t179] or substrate surface with diamond powder or nondiamond powderst181] as well as graphite flakest’61t182)[183]etc., biasing substrate,[‘781[‘841-[‘911 or covering/coating substrate surface with carbon clusters[1921t’93]etc.) is the most effective method for increasing surface nucleation density and decreasing the incubation period. The effect of substrate surface pretreatment on diamond nucleation depends not only on the pretreatment methods but also on the type of substrates.[118] The various surface pretreatment methods developed to date and the nucleation enhancement mechanisms are extensively discussed in Ch. 6. The physical properties of the currently used or potential substrate materials are summarized in Tables 5 and 6. seeding[‘661[‘801

Table 5. Physical Properties of Currently Used or Potential Substrate Materials for Diamond Epitaxy Substrate lllalerlill diamond

Melting

3797

(hexagonal) a-axis c-axis

con\tantb

Dearity’ (k. L m.’

(A)

--

3.567 121 ----

_...__

2.52 I42

[2] 121

3515 (21 3520 [2] ----__

[ 1941 -__ .___

---2.46 121 [2] 6.71

2260 [2] -------

3057

(cubic) (hexagonal) a-axis c-axis

graphite

Lalllce

pomp” (‘c) [ lY4]

cBN (cubic)

2727[197]

3615[lY7]

Co ( a ) (hcp) (p) >390 ‘C (fee)

-1494 [ 19x1

cu (ICC) I~e(y)912-1400~C(fcc) Fe ( a) <912-C (bee)

XYOO [19X]

1084(11)X1 1536llYXJ ----

361 (IYXI 3.56 [I981 2.86 [IYKJ

XYhO[IYX[ 74OullYn~ 7x70 [IYX]

Ni (fee)

14551191(~

3.S2 [19X1

XYOO

Si (dumond-cubic)

1412 [19X]

5.42

234O[lYX]

,WiiC

(cubic)
2697 (hcp)

(B)>YOO’C(bcc)

[IYXI

32lO[lY7] 4SOO[lYX] 4110[191(]

[ 19x1

4Y20

316O[lY7~

TIN

BrO (hexagonal)

4.24 [ 19x1 2Y27 lY7 ] 2 hY,4.37[197] 2452 [ lY7]

5430 [ 1971 3020 [I971

MgO

2X251197]

35x1 [I971

(cubic) at room temperature,

4.3131197]

or at the temperatures

I(I-(’K~’ )

at which

Surhcc

cncrgy’

(JId)

0.x [Z]

.53(111)[1] h.S(llO)[l] 1).2(IW)[I[

__-

“egatlvc 25 I21 [2[

2.80 (lO~O)[l951 0 17 (OWl)[lY6[

---.

0.5’) 12.5

.___

2-lW7[lY7] x.1(1-7 (1971 2 1()-x [ I’)X]

[ 197) [ IYXJ 2.709/2.003[200]

170,1YX] >I46[lYXI 12.1 [IYX]

2.364/l

I IYXI 1lY7[ I IYX]

I .46

4.63

---7.lo.Y

[,9X]

2.10-h

[ I’)71

X.Y 9.0 [IYX] 6 52

IO-X [52h]

2 OX ( 1001 l1’PIl 2.Y3Y/l.Y23[200]

13 J IIYX[ 7.6

7 10.1Sll~7,

[lY7]

TIC (fee) (cubic)

Thcl Illal cxpnwion cucllicx!nP(

----

[ IYX[

4.35 [IYX[

4.32

5-1,

I~lrJ-X[19x[

2.95, 4 6X [IYX] 3.29 [19X]

(1971

--1667 [19X]

(cm*

3490[197[

2.Sl,4.~17[lYX[ 3.54 [IYX]

Ti (a)

C dilfusivityd

,

77~[200~

(I I I) (I I

2.S70/1.723[200~

I lY7[

Y -!s [ lY7] 7 43 [lY7[

----

12.X3 [I971

the phases exist.

at 20 ‘C. calculated 81 800 “C for most metals; phases ex,sr,nX above 800 ‘C. O-100

‘C Car melals.

surface energy

diamond

at 25 ‘C/surface

surface reconstruction,

31 II00

and graphw

‘C for carhidcs

25.500

energy a, melung

physisorplionlchelnlsorptlon.

[52b]:

nt Ihc lower

limit

of tcmpcr;~rc

range

for ~bc

‘C for ccwm~cs. lompcralure.

The surlacc cncrg,cr

or other surlnce rcxtiws.

du no, mclude

the efl’ccts 01

41

Diamond CVD Techniques

Table 6. Physical Properties of Currently Used or Potential Substrate Materials and Their Carbides, Nitrides, or Oxides Subslrue material

McldnX point” (‘C)

Al (ICC) Al4C3 (rhombic)

660 [19X] 2200 [201

Al203 Au

204911971 1064(19X]

(hexagonal)

(kc)

Lattice wnswuh (A) 4.04 [19X]

I

[

BqC (rhomb.,hedral) Cr(a)
2447 1971 1860[198]

CqC2

1895 [I971

(orthorhomblc)

Fc?C~anhorhombic) Hf-( a)<1 3lO’C (hc;)

133.24.YJ [IOX] 4.785, 12.991 [I971 4.07 119X] 5.60, I2 12 [19X] 2.8Y [19X] 2.82.5.52.1

1.46[198]

5.09. 1650-:.12011 4.52.3.19. 5.206.75 [l&l 11981

DensrlyC (kg “i3) 2700[19X~ 2950 12lll

I . .

C d,llus~w~y~ Tbcl.m:~l cxp.m\,r,n sur,acc wgyl coelfxIcnl’( 106 K-1) (cm2 s-l) (J ,,I-~) 23 5 I IYX] I .085/0.939[200~

19300 119X]

_L..., I

---

2510[197] 71lx)[11)8] 670011971

,_..

4.10-X ,I’)71

;

----

3.50 [I981

7400 i20, 13100[198] .-

HfC (fee)

3887 12011 1244 [ IYX]

4.46 1198) X.YlJ [I’)81

12520[1’ 74lJOIlYXI

Mo (kc)

2615 [I981 26Y0 I’)71

3.14 [I981

lo2oo[lYx]

10-11 [IYX,

YIXO[203~

X 10-121521,1

X6OO(IYXI

? 11~121 ,w,

Mn(u)<742'C (kc) Mo2C (hcp)

I

Nb (kc)

24671198)

NbC (bee) Pl (fee) SilNa i Dllhcxnwmal)

3497 11971 1770[198] 2442 II971

~

D-.._.,

._,.._,

3 111, 4.74 3.294

I IWI

[IYK]

4 424-4.457 [19X] 392 [19X] 7603 2YO9 IIYXI

l0_13[52a] ....

,l,

23

.”

,__.., ----

_, .__,___, ____,.._,

3.333/1.701[200~

I IYX]

5 I [I981 7x-Y3

2 YXV2 02212001

,, 9.0 (IYXJ 2 II 1,971

2.691/2OS5[200~

--

1902,198,

3.024 [IYX,

hl(KJ,llj..,

3327 [I971 33x7 IIYXI

4.17 [IYXI 3.16llYXI

5770 [2Ol] 193”OlIYX,

10-13 [,Y7l

7.2 1201 I 4SllYXl

WC (hcp) Ul.0 IL”..,

2627 1201) 2776 [ 1941 2850, I 971

2.90. 2.83 [ 1981 2.99. 4.71 119X1 S.07 11651

15X00 [I971 17150,201~ 5560 ,lY7,

---2 IO-13[52h] ----

442 II971 3.5X l?Ol] JO,lY7,

---1852 ,19X]

3.23, 5.14 [19X] 3.61 IYX]

649011981 5XOOll9Xl

&IO-‘111981 3 10-H ,19X,

59,19X,

II

p_

l,aq,,

zlc (fee) a

326O[lY7]

[

467[19X]

6660

[1971

IO-14 [52;,/ 7 lll-IJ147~1

65,19X, ,170 11w1

vc (fU) W(U)(bcC)

Y-Zr02 (cubic) Zr (a)<X40’C (hcp) (/J)>84O’C (bcC)

1612001

721lYXI

V (bee)

“,.

I .440/l .03Y~2Olll 2.X77/2.1

1x1

__.“,._.,

. ,._I,

Y1317,~,,!

,“,,

33O~lYXI ‘l‘l’(l19XI

._-\___,

? 056/l

I

hO[l9H] .... 11.1,

----

78W 12( -, 21450119K] -ilX71l~Y71 I6600 I IYXl Id‘wII1Y71

6.0 1201

2YXO I IYXl ?CdO,1~,71

‘T2*(lxx) T:,l- ,li?,

I 626/1.345~200~

X.OO’[ 19;1

222711981

(~)>I3IO’C(bcc)

14 I IYX] 45 [I’)71 6SIIYKI

,.,

,I,,,,

----

,._..,

3.1l1X/2.270(2001 _ .._.. _ .._ _,_ . .._. 3 46X/2 4X712”Ol

2.7YO,l.5S4(200, 6 I [I’)71

42

Diamond Chemical Vapor Deposition

4.6

Substrate Temperature

Figures 5 and 8 show the influence of substrate temperature on growth rates of diamond in CVD. Clearly, a small change in the substrate temperature can alter the growth rates markedly, especially at higher gas pressures. This is attributable to the Arrhenius behavior of the deposition reactions. Moreover, in both low-temperature and high-temperature regimes, the growth rates are lower.

I ’

l

I

I

I

t

900

I

I

I

I

1100 T,OC

Figure 8. Temperature dependence of growth rates of polycrystalline diamond films (gas pressure = 760 torr).[**] (Reprinted jFom ReJ: 22, 0 1988, with kind permission j?om Elsevier Science Ltd. The Boulevard, Langford Lane, Kidlington, OXSlGB, UK.)

The optimum substrate temperature, at which a maximum diamond growth rate and a highest level of crystal perfection can be achieved for a given system, is generally in the range of SOO-lOOOT, with a typical low bound approaching 6OO”C, basically independent of deposition techniques. 131141147111181 Such high substrate temperatures obviously limit substrates to rather refractory materials. Ifthe substrate temperature is too low (<8OO”C), a significant amount of amorphous carbon may be co-deposited, whereas if the substrate temperature is too high (> 1 lOOT>, non-diamond components, At substrate temperaincluding microcrystalline graphite, may form. 156112041 tures higher than 13OO”C, only graphitic carbon is deposited.t261t271[2gl[561 In addition, with increasing substrate temperature, the domain of the gas

Diamond CFD Techniques

43

compositions suitable for diamond deposition in Fig. 4 reduces and the domain width narrows down to zero at substrate temperatures >1300°C.[56] The low temperature synthesis of diamond films has been investigated in the substrate temperature range of 350-800°C in RF thermal plasma CVD, and diamond films of reasonable quality have been obtained at 550600°C [I211which are considerably lower than those generally considered as optimum regime (i.e., around 9OO’C) for diamond growth. Diamond was grown even at a substrate temperature of 350°C[121] or 365°C[162] in spite of very low growth rates and containing some non-diamond carbon. It has been reportedL4] that diamond films with crystallite size of about 200 nm form at the substrate temperatures around 65O”C, while lo-20 nm sized crystallites occur at the substrate temperatures between 450 and 600°C. Another study[2051showed that diamond grains of 160, 15, and 10 nm were grown at the substrate temperature of 800,600 and 415”C, respectively, in MW PACVD. Clearly, for substrate temperatures lower than SOO’C, the crystallite size increases with increasing substrate temperature. These results demonstrated that the crystallite size, surface roughness and hence optical transparency as well as other properties of diamond films can be tailored by varying the substrate temperature. 4.7

Substrate

Position and Size

Substrates are usually immersed in the plasmas in MW, DC and RF PACVD processes or separated from the plasmas in DC and RF thermal plasma CVD processes. A substrate is generally placed 0.5-2 cm from the hot filament in the HFCVD, or from the flame burner nozzle in the combustion CVD, and up to 5 cm from the thermal plasma nozzle in the DC thermal plasma CVD. [74] Report on distances greater than several centimeters has not been found in published literature, which may be attributable to the optimum gas-phase conditions resulting from the complex gas-phase reactions and mass transport within the short distance above the substrate. Substrate size which can be coated depends on the CVD technique used, and may vary from less than several cm2 in the DC discharge CVD to 400 cm2 in the HFCVD (Table 1). Depositing diamond on a large area is essential for many applications (cutting tools, optical windows, semiconductor wafers) and also critical for making diamond CVD techniques commercially feasible. For a large coating area, the film homogeneity, uniform

44

Diamond Chemical Vapor Deposition

substrate heating/cooling and temperature addition to other constraints.

4.8

Effects of Electric

control may be a problem in

and Magnetic Fields

The application of electric and magnetic fields in diamond CVD processes offers an opportunity to merge between chemistry-controlled and bombardment-controlled growth of diamond. Electrical biasing of a substrate can enhance nucleation rates and densities, and improve growth rates 1151although the latter effect is not significant and even detrimental to the &n quality. 11071In the DC PACVD1311151and HFCVD,t2061 positively biasing a substrate gives rise to the formation of diamond, whereas negative biasing results in deposition of graphite or carbides.t31ti51 A positive substrate bias may limit the DC PACVD, since DC plasmas may no longer work due to space charge buildup after high-resistivity diamond films form. A negative substrate bias, on the other hand, can significantly enhance nucleation of diamond in MW PACVD.165jt711 At relatively high gas pressures (>l torr), an ECR plasma may not occur due to gas-phase collision. At relatively low gas pressures (CO.1 torr), an ECR plasma occurs, and diamond has been grown at substrate temperatures as low as 500°C with a uniform discharge area of -200 cm2.12071

4.9

Impurities

and Defects

The current CVD diamond films are impure, especially by semiconductor standards.1208j Main impurities are hydrogen and silicon,*4ll208l as well as nitrogen. 120yl12101 Other impurities include graphite in the presence of Oxygen impurity has also oxygen, and Sic in the absence of oxygen. 12111[212] been reported to occur when oxygen or oxygen-containing compounds are added into the reaction gas mixture.l213l Hydrogen is a necessary ingredient in diamond CVD and highly reactive. Silicon is a commonly used material for deposition substrates, reactor tubes and bell jars. Nitrogen is the dominant impurity in commercial high grade hydrogen and any leakage in a deposition system may lead to an increase in the nitrogen contamination. The hydrogen impurity is found to be inversely related to the hydrogen concentration in the reaction gas mixture. Silicon impurities are detected in octahedral holes of the diamond lattice,141 and may give rise to an SO-fold

Diamond CVD Techniques

45

increase in nucleation rate and a 3-fold increase in growth rate.1221 Metal may form due to the interactions of carbon carbides (for example, SIC 12121) with substrate materials. Evaporated metals may remain in the diamond lattice as impurities. Defects in CVD diamond films12141may include the following types: 1. Point lattice defects: vacancies,12151 interstitials,12161 and substitutional elements 2. Line lattice defects: dislocations.126l 3. Planar lattice defects: microtwins,15111531 stacking fhults,1531 and grain boundaries 4. Volume lattice defects: voids,l70l and inclusions (for example, graphite inclusionst2041). These defects are generally detrimental to electronic, thermal and mechanical properties of diamond films, although in some cases they may be neutral or beneficial. The frequently occurring defects appear to be stacking faults (twinned clusters with many re-entrant surfaces/comers), which are deemed to play a major role in enhancing diamond growth rates.I161

5.0

SUMMARY

It is evident that none of the CVD techniques developed so far may be considered to be the best in terms of economic viability, and it still remains to be verified whether the maximum growth rates have been achieved for each of these techniques. Moreover, there may exist more than one set of optimum deposition conditions and care must be taken to define specifically what is being optimized. The trade-offs among stability, deposition rate, coating area, film quality (as defined by scientific or technological criteria), and production cost of each of the techniques have to be determined on the basis of the specific requirements of an application (such as purity, size, flexibility and simplicity of operation, etc.) when choosing a CVD method and deposition conditions for diamond synthesis. Further research is needed to integrate the best features of each technique into a single fabrication process, if it is possible.

Diamond Nucleation Mechanisms

It has become increasingly evident that further technological development in CVD of diamond films, particularly in such challenging areas as single-crystal growth for electronic applications and low-temperature deposition for coating on optic and plastic materials, requires a detailed understanding and control of the fundamental phenomena associated with diamond nucleation and growth. These phenomena, especially the nucleation and early growth stages, critically determine film properties, morphology, homogeneity, defect formation, adhesion, and the type of substrates that can be successfully coated. For example, an increase in surface nucleation density may reduce morphological instabilities 170]and surface roughness of diamond fihn~.l~~~l A high surface nucleation density may also improve homogeneity of films and reduce formation of voids at the substrate/coating interface, leading to a better film/substrate adhesion. Most earlier studies on the low pressure CVD of crystalline diamond have focused on examining various deposition techniques and characterizing the deposited films. These studies have led to a reasonable understanding of growth mechanisms and processing parameters. Recently, studies on the nucleation and early growth stages have attracted an increasing attention. The extensive research efforts have significantly contributed to understanding of diamond nucleation mechanisms in 0. In this chapter, diamond nucleation mechanisms are reviewed on the basis of available literature.

46

Diamond Nucleation Mechanisms

1.0

47

HOMOGENEOUS NUCLEATION-GAS-PHASE NUCLEATION

The emphasis of most studies on nucleation and growth of diamond has been placed on the heterogeneous formation of diamond particles and the crystallization and deposition of diamond films on substrate surfaces. Only a limited number of experiments have been conducted to achieve the homogeneous nucleation of diamond in the gas phase at atmospheric and subatmospheric pressures. However, there is evidence that, at least in some cases, diamond can be nucleated homogeneously in the gas phase 12181-[2211

Derjaguin and Fedoseevt25] presented theoretical arguments, based on classical nucleation theory, that homogeneous nucleation is possible. Matsumoto and Matsui[222] suggested that hydrocarbon cage molecules such as adamantane, tetracyclododecane and hexacyclopentadecane are possible embryos for the homogeneous nucleation of diamond (Fig. 1, upper). The adamantane molecule, C,,H,,, represents the smallest combination of carbon atoms which possess the diamond unit structure, i.e., three six-membered rings in a chair coordination. The tetracyclododecane and hexacyclopentadecane molecules represent twinned diamond embryos that have been proposed as precursors to the fivefold twinned diamond microcrystals prevalent in CVD diamond thin films. From simple atomic structure comparison, one can easily generate the diamond lattice from these cage compounds by simple hydrogen abstraction followed by carbon addition. However, thermodynamic considerations show that these molecules are quite unstable in the harsh environment associated with diamond CVD. Thermodynamic equilibrium calculations of Godleski et a1.t223j and Stein[224] revealed that such low molecular weight hydrocarbons are not stable at high temperatures (>6OO”C). Stein [224]also found that the favorable enthalpies of formation of such diamond precursors as adamantane, diadamantane, and cyclohexane are overwhelmed by unfavorable entropy of reactions above 500 K. Below this temperature, the calculations showed that these diamond-like molecules are unstable with respect to dissociation into methane and other small molecular weight gas species. Above this temperature, the corresponding graphite precursors (polycyclic aromatics) are found both theoretically and experimentally to be more stable.

48

Diamond Chemical Vapor Deposition

adarnantane

cis

boat-boat

tetracycl-

bicyclodecane

hexacyclcwentadecane

tram

boat-boat

bicyclodecare

Figure 1. Proposed nuclei for homogeneous nucleation of diamond in gas phase: cage compounds (upper),[***]and ring compounds (lower). [2251The formation of nuclei is shown with dashed lines and circles. (Reproduced with permission j-om the Annual Review of Materials Science.f’831 0 1991 by Annual Reviews, Inc.)

Angus

et ~1.[2251W61

argued that fully hydrogenated ring compounds structurally related to the boat-boat conformer of bicyclodecane are more plausible nuclei due to their greater abundance in the typical growth environment and having kinetically preferred sites for atom addition (Fig. 1, lower). Such graphite precursors, as described in hydrocarbon flame and combustion literature,[2271-[2301are stable in the high temperature reducing atmosphere of diamond CVD. Indirect evidence for the presence of such molecular precursors has been demonstrated.1881[231] On the contrary, Frenklach and Wang,rs7] in a detailed chemical kinetics modeling of a HFCVD reactor, found that under the operating conditions typical for diamond deposition, atomic hydrogen suppresses the formation of such aromatics.

Diamond Nucleation Mechanisms

49

Mitura[2181 reported the formation of diamond particles from the decomposition of methane diluted by hydrogen, nitrogen, or argon in RF electric field. Frenklach et a1.[221lpresented evidence for the homogeneous nucleation of diamond particles in MW PACVD using a variety of hydrocarbons diluted in hydrogen, argon or oxygen. Although in most cases only nondiamond materials like graphite and carbyne were obtained, the homogeneous nucleation of diamond was clearly observed in the dichloromethane and trichloroethylene-oxygen mixtures using optical and electron microscopy, electron diffraction, and Raman spectroscopy. The particles formed exhibit crystalline shapes and are a mixture of polytypes of diamond, but mostly hexagonal. The particle diameter is on the order of 50 nm, and the largest particles are -200 nm in diameter. In another experiment,r21g1 a mixture of cubic and hexagonal diamond polytypes, lo-500 run in diameter, were produced by MW assisted combustion of acetylene in oxygen. It was found that diamond could form only over a limited range of gas compositions and pressures: C/O of 0.56 1.4 in premixed flame, 0.2-0.9 in diffusion flame, and 50-500 torr. Larger particles were observed at lower reactor pressures (cl50 torr) and higher C/O atomic ratios (0.83-l). The effect ofthe pressure on the particle size was attributed to its influence on nucleation density, growth rate, and reaction time. Buerki and Leutwyler 12321also reported the homogeneous nucleation of spherical diamond powder by CO,-laser-driven gas-phase reactions. Cubic and hexagonal diamond powders of average diameters 6-120 nm (maximum 300 nm) were obtained by laser-induced decomposition of C,H, or mixtures of C,H,, H, and SiH, at low pressures and temperatures. The presence of impurities can increase nucleation densities. Higher nucleation rates can be obtained with halogen/hydrogen mixtures than with methane/hydrogen mixtures. W71 The presence of a large amount of oxygen in a plasma etches the sp2 carbon phase which nucleates more easily than diamond,[2211 and the quality and yield of diamond particles may be improved when oxygen is added. 12201In a low-pressure plasma reactor, the addition of silane, WI,, to acetylene mixed in hydrogen or argon results in powder of weakly bonded, amorphous, hydrogenated carbon-silicon,l220l whereas the addition of diborane, B2&, even at very low concentrations, is very effective and pronounced in inducing diamond nucleation in the gas phase.l220l Substantial production of diamond particles (5-450 run in diameter) have been achieved at linear growth rates on the order of 102- 1O4pm h-’ due to the addition of diborane. Under the same conditions, no diamond forms without

50

Diamond Chemical Vapor Deposition

the presence of diborane. 12201Based on these experimental observations, Frenklach et a1.12201 developed the principle idea of induction nucleation, i.e., adding to a hydrocarbon mixture some compounds of other elements that can quickly decompose and nucleate small clusters. These clusters then serve as seeds for the subsequent deposition of carbon on their surfaces. In spite of these studies and results, the relative importance of the gasphase nucleation compared to the surface nucleation is unclear as yet. In fact, the number of diamond particles collected from the gas phase is very small compared to the typical surface nucleation densities, thus the homogeneous nucleation mechanism cannot account for the large variety of nucleation densities observed on different substrate materials and from different surface pretreatments. It is speculated and also supported by a recent experimentI233l that the nuclei formed in the gas phase may reach the growing surface and increase the surface nucleation density. However, how the diamond particles formed in the gas phase could serve as seeds on the substrate surface for the subsequent growth of a diamond film remains unknown.

2.0

HETEROGENEOUS NUCLEATION

2.1

Nucleation

Processes

NUCLEATION-SURFACE

and General Features

Diamond can grow homoepitaxially on a diamond surface without m p~~~,P371[*381 and nucleation problems in HFCVD, MW4lH*W ECR MW PACVD11461 from a variety of carbon sources. Diamond also nucleates easily on cBN 123gl-12411 due to the identical crystal structure and the nearly identical lattice constant (mismatch 1.4”/0) and thermal expansion. However, in most CVD processes, diamond nucleation on non-diamond surfaces without pretreatment is usually very slow. A conventional growth process in CVD of polycrystalline diamond films typically shows several distinguishable stages,l242l as shown in Fig. 2: (i) incubation period, (ii) 3-D nucleation of individual crystallites on substrate surface (Fig. 2a), (iii) termination of surface nucleation, and 3-D growth of individual crystallites (Fig. 2b-c), (iv) faceting and coalescence of individual crystallites, and formation of continuous film (Fig. 2d), and (v) growth of continuous tilm (Fig. 2e-I-). Before nucleation starts, an incubation

Diamond Nucleation Mechanisms

51

52

Diamond Chemical Vapor Deposition

period may take from a few minutes up to hours,[1741[1811[243] depending on substrate materials, surface pretreatment and deposition parameters. The isolated nuclei/microcrystals formed during nucleation exhibit sphere-like geometry. With increasing time, the nucleation density increases to a certain value, then the surface nucleation terminates. The isolated crystals grow homogeneously in size and faceting develops due to abundant C surface diffusion from the relatively large diamond-free area surrounding them. 152al When the isolated crystals coalesce together, a continuous film forms. Growth spirals (Fig. 3a), steps (Fig. 3b),15281and reentry grooves (Fig. ~c)[~~I as well as various grain boundaries and twins, etc. are common growth helpers fsza]for faceted crystals and are usually found in the early stages of diamond growth. Secondary nucleation sitest2441 may appear at these imperfections, leading to films of poor quality. The time needed to form a continuous film depends on not only growth parameters but also substrate materials and surface conditions, while the growth rate of a continuous film, itself, is independent of substrate materials and/or substrate surface conditions.t52al11741 Under certain growth conditions, the competition growth of crystals may govern the subsequent growth process of a continuous film.[2451 The grains increase in size and perfection [174lwith the direction of fastest growth nearly perpendicular to the substrate, leading to a columnar structure[174] or apber-like crystallographic texturel244Il245l (Fig. 4a). Some crystals grow faster and swallow their neighbors, a phenomenon described as evolutionaryselection by Van der Drift,1246las shown in Fig. 4. It should be noted, however, that diamond thin films synthesized by a variety of CVD processes are of polycrystalline structure with poor surface morphology, mostly consisting of randomly oriented, highly disordered crystals, containing varying amount of non-diamond carbon and defects. The morphology, texture, orientation, crystal size distribution and incorporation of non-crystalline phases in polycrystalline diamond films may be very different when grown under different deposition conditions. The relationship between these microstructural characteristics and growth conditions will be discussed in the next chapter.

Diamond Nucleation Mechanisms

53

Figure 3. (a) Growth spirals, (b) steps[52a] and (c) reentry grooves[53] formed in the early stages of diamond growth. (Reproduced with permission.)

The observation that surface nucleation may terminate while a fraction of the substrate surface remains unnucleated may be rationalized as fol10ws.[~~~] In the initial stages of growth, nuclei form and crystals grow independently of each other. The number of the nuclei is still so small that the formation of new nuclei is unhindered. When the number of the crystallites increases to such a value that the diffusion zones ofthe individual crystallites start to overlap and the crystallites compete for the same growth species, the generation of new nuclei becomes increasingly improbable. Smaller crystallites grow faster than larger crystallites, leading to the formation of relatively uniform-sized crystals. These crystals continue to grow until they coalesce together.

54

Diamond Chemical Vapor Deposition

Figure 4. Columnar (fiber) structured layers developed during continued growth of CVD diamond in preferred growth direction. (a) Columnar morphology with increasing size and perfection of grains during continued growth. (Reproduced with pennissionfiom RejI 54, 0 1990, American Association for the Advancement of Science.) (21)Starting from randomoriented equidistant cubic crystals in a two-dimensional space and assuming infinite surface diffusion along the substrate, the intercrystal boundaries (dashed lines) and the crystal front (solid lines) at different times (i.e., at t = x, t = 5x, and t = 25x, where x is the shortest time required for the coalescence of adjacent nuclei) are constructed on the basis of the evolutionary selection mechanism, a principle governing the growth orientation (texture, generally speaking) in vapor-deposited layers. The crystals with the direction of fastest growth nearly perpendicular to the substrate are in the favor position and will survive.t246] (Reproduced with permission.)

In an atomistic scale, a surface nucleation process may include the following events: 1. Atoms impinge upon the deposition substrate from the gas phase and become adsorbed onto the surface. 2. The adatoms may either re-evaporate (desorb) or diffuse over the surface. The adatoms may also diffuse into the substrate or bond to other surface atoms. 3. With increasing time, the surface concentration of the adatoms increases and clusters form. 4. Through statistical fluctuation in the local adatom concentration, these clusters grow or decay.

Diamond Nucleation Mechanisms

55

5. There exists a critical size above which the probability of growth will be greater than decay, so that the clusters, whose size exceeds the critical size during the concentration fluctuation, become stable. 6. The stable clusters provide suitable sites for growth either from the continued migration of single adatoms or from the direct impingement of atoms from the gas phase. This general scheme of surface nucleation processes, as described above, may be adequate only for nucleation on a perfect substrate. It is well known that in many important practical situations, nucleation occurs at defect sites on the substrate surface. In addition, the interactions ofgas-phase species with the substrate surface in diamond CVD may lead to surface carbon atoms of different chemical bonding states and structures, for example, sp’, sp2, or sp3 bonded carbon, amorphous carbon, diamond-like carbon or carbon in carbides. These factors further increase the difficulty in understanding surface nucleation processes of diamond in CVD. Surface nucleation processes can be described with two quantities: surface nucleation density, Nd (cmm2), and surface nucleation rate, N,. (cmm2h-l). The nucleation density is the number of nuclei grown on unit substrate surface, and the nucleation rate is the number of nuclei formed per unit substrate surface in unit time. The nucleation density depends on the number of activated nucleation sites available on the substrate surface. Nucleation will stop when crystals have nucleated on all available nucleation sites or when the diffusion zones of nuclei overlap each other (as discussed above), whichever occurs first. Both the nucleation density and rate are determined by substrate surface conditions and deposition parameters. Nucleation density critically influences the thickness, crystallite size, homogeneity, morphology, adhesion, and surface roughness of deposited diamond films. A high nucleation density is required for deposition methods possessing intrinsically slow growth rates in order to obtain continuous films within a reasonable time. A high nucleation density allows a complete film of smaller grains to form within a relatively short time, and leads to a smoother film surface than those grown at lower nucleation densities.t’**l A low nucleation density may give rise to large crystallites and a discontinuous film within a limited time, as demonstrated in Fig. 5.

56

Diamond Chemical Vapor Deposition

!

0th films :atments: :m-*; (6) :h lower and the

Diamond Nucleation Mechanisms

57

Spitzyn et a1.t261considered the surface nucleation process to be the primary process responsible for diamond nucleation. In their experiments, diamond was nucleated on non-diamond substrates (poly- or single-crystalline Au, Cu, Si, MO, W). The substrates were polished with abrading grits (diamond, zirconia or others), or etched, or seeded with diamond crystals prior to deposition. Crystals of a few tenth to a few ten micrometers were grown with octahedral and cube-octahedral habits under the same conditions as used for growth on diamond substrates. Nucleation rates of 105- 1OScmm2h-’ were obtained. A continuous film of 2 urn in thickness was grown, corresponding to a minimum nucleation density of 2.5 x lO’cm_*. A 100 nm thick, polycrystalline SIC inter-facial layer was identified by Auger spectroscopy on Si substrates. From their studies, some general features associated with surface nucleation processes of diamond may be summarized as follows: 1. Diamond nucleation rates on non-diamond substrates vary from lo3 to 10’ cmm2h-l, depending on synthesis conditions, substrate materials and surface pretreatment methods (polishing, etching, seeding, or annealing). 2. Nucleation of diamond crystals is observed mostly on defects such as scratches, grain boundaries, dislocation outcrops, etc., suggesting that nucleation occurs on the substrate surface, and not in the gas phase. 3. Diamond nucleation rates are several times lower on single-crystal substratesthanonpolycrystallinesubstrates of the same material after identical surface pretreatment. 4. Diamond nucleation rates on carbide-forming substrates (Si, MO, IV) are one to two orders of magnitude higher than on substrates that do not form carbides (Cu, Au). 5. Diamond nucleation rates on foreign substrates decrease as the substrate coverage with diamond and the crystal size increase. Recent advances in experimental measurement methods make it possible to directly observe the nucleation and early growth stages, and in some cases in-situ or in vacua measurements. P61[1*01[1**1[~4*1 A number of experimental measurement methods have been employed for the identification ofphases and degree ofcrystallinity in diamond films, for the characterization

58

Diamond Chemical Vapor Deposition

of microstructure and morphology of diamond films, for the observation of diamond film surface and diamond-substrate interface (in some cases in the nanometer scale, for example, AFM and STM), and for obtaining information about the chemical nature of the first few atomic layers on a substrate surface (MISM). These methods include, for example: -

optical microscopy

* transmission

(OM)p5’l

electron microscopy

. scanning electron microscopy

(SEM)115711165112501-Iz541

* high

resolution transmission (HRTEM)1*~5~lt*55bl

* X-ray diffraction

electron

microscopy

(XRD)11651t2531

- X-ray photoelectron -

(TEM)11sglI*4gl

spectroscopy

(XPS)t1721t173*

low energy electron diffraction (LEED)12561-12581

* reflection high energy diffraction (RHEED)125gl1260al1260bl * Rutherford backscattering -

spectrometry

(RBS)1260al1260bl

laser interferometry12481[2611

* electron energy loss spectroscopy

(EELS)l’8g1l254l

* micro-IR absorption measurementl250l * Auger electron spectroscopy -

(AES)t2541[25gl

atomic force microscopy (AFM)t2501t260alt2621

* scanning tunneling microscopy * micro&man

(STM)1731

spectroscopy115711’65112521

-

secondary ion mass spectrometry

-

spectroscopic

(SIMS)t1561

ellipsometry (SE)l72ll263l

A number of experimental

observations

on diamond nucleation pro-

~~~~~~I’~l~~~1~~~~lI~~~l~~~~l~~~~l~~~~l~~~~l~*~*l~*~~lI*~01[2~~l~I2~~l~*~~l reveal that

,

b

most cases, diamond does not nucleate directly on a non-diamond substrate surface, but instead, on an intermediate layer which develops at the interface between diamond and the non-diamond substrate during the incubation period before diamond nucleation begins. The intermediate layer, formed due to the chemical interactions of activated gas species with the substrate

Diamond Nucleation Mechanisms

59

surface, may consist of diamond-like amorphous carbon (DLC, a-C, or aC:H), metal carbides, or graphite, depending on substrate materials, pretreatment methods and deposition parameters. It is generally proposed that the intermediate layer provides nucleation sites for diamond crystallite growth, hence enhances diamond nucleation densities on non-diamond substrates, and offers an opportunity for controlling the morphology, orientation, and texture of diamond films during nucleation and growth. In the following sections, the representative mechanisms governing diamondnucleation on non-diamond substrates are discussed on the basis of the experimental and theoretical studies available in published literature. 2.2

Nucleation Amorphous

on an Intermediate

Layer

of Diamond-like

Carbon

The HRTEM study of diamond nucleation and growth on copper TEM grids in HFCVD by Singh1255alprovides direct evidence for the formation of a diamond-like amorphous carbon layer. The intermediate layer is 8-14 nm thick, in which small diamond microcrystallites approximately 2-5 nm across were embedded (Fig. 6a), and large diamond crystallites were observed to grow from these microcrystallites (Figs. 6b-c). that the diamond microcrystallites are formed as It was suggested 1255a] a result of direct transformation of the a-C into diamond, with the intermediate layer providing nucleation sites on which the large diamond crystallites grow. Based on these experimental observations, a detailed mechanism of diamond nucleation and growth on copper in HFCVD was proposed, as depicted in Fig. 7. In step I, carbon clusters are formed on the substrate surface and a change in the bonding structure from sp’ to sp* ties place. In step II, sp* bonded carbon atoms are converted into relatively stable network of sp3 bonded carbon. The continuous molecular rain of activated hydrocarbon and atomic hydrogen on the substrate surface provides sufficient energy for the sp’+sp2+sp3 conversion. At the same time, etching of unstable phases (sp’ and sp*) which is ten times faster than etching of stable phase (sp3) promotes and stabilizes the sp3 phase. Atomic hydrogen plays an important role in the sp2+sp3 transition. It is estimated that approximately 1O4hydrogen atoms are needed to convert a carbon atom to sp3 hybridization.t2651 In step III, transition of the bonding state of the carbon network occurs from a disordered domain with sp3 bonded carbon to diamond with sp3 bonded

60

Diamond Chemical Vapor Deposition

Figure 6. HRTEM micrographs showing (a) the non-uniform amorphous-crystalline interface, the amorphous layer of about S-10 nm with non-uniform surface, and the embedded diamond microcrystallites of 2-3 nm in the amorphous layer; (b) the secondary diamond crystallite (A) embedded in the amorphous layer on the surface of the primary diamond crystal (B), and the amorphous layer of 6 nm around the diamond crystal; (c) the amorphous layer of -14 nm in thickness around the faceted diamond crystal grown at the edge of a Cu TEM grid, the amorphous layer between the diamond and grid, and the lattice planes parallel to the faceted boundaries of the diamond crystal, AB or CD, corresponding to { 111) planes of diamond.t255a] (Reproduced with permission.)

Diamond Nucleation Mechanisms

II

DLC

61

._...._.._._

Amorphous carbon (Diamond-like carbon) III DLC,. .__..._ m

. .._._

VII

Figure 7. Schematic diagram showing the proposed nucleation nuclei form on a DLC interlayer. (I) Formation of carbon clusters and change in bonding structure from sp’ to sp*. @I) Conversion (III) Crystallization of amorphous phase. (1~+7,) Growth and crystal. (‘VIZ) Secondary nucleation and growth of diamond.[*“‘] permission.)

mechanism: diamond on substrate surface of sp2+sp3 bonding. faceting of diamond (Reproduced with

62

Diamond Chemical Vapor Deposition

carbon. Crystallization in the amorphous layer also includes chemical reactions, such as hydrogen abstraction, dehydrogenation of absorbed complexes, recombination of hydrogen atoms, etc. During the crystallization, carbon atoms rearrange towards { 11 l} planes to achieve the minimum surface energy since the { 11 l} planes possess the lowest surface energy in a diamond crystal. 11112661 The crystallized regions then act as nuclei for the subsequent growth of diamond. In steps IV to VI, diamond growth takes place. A disordered domain (amorphous layer) always surrounds diamond microcrystals. Carbon atoms addedto the surface (step IV) diffuse inwards via a solidstate diffusion process. [2671The initial diamond shape is hemispherical (step IV and Fig. 6b), as confirmed through examination of a planar view of diamond growth on iron silicide by the same author.[255bl Once a diamond microcrystal reaches a critical size (step V), it will acquire a faceted crystallographic shape characterized by defects such as stacking faults and twins (step VI). In step VII, secondary nucleation takes place as a result of the concentration fluctuation on the surface of the diamond crystal. This fluctuation leads to an uneven surface of the disordered domain, the thickness of which varies from 8 to 14 mn, depending on deposition conditions. Once the thickness of the disordered domain exceeds the critical thickness (>15 nm), there will not be enough localized thermal energy or time available for carbon atoms to diffuse into the diamond crystal, leading to the secondary nucleation on the surface. The amorphous layer will recrystallize following the steps I-VII. An amorphous layer always forms initially between the substrate surface and the primary diamond crystallite as well as between the primary and secondary diamond crystallites. The amorphous layer between the primary and secondary diamond crystallites will eventually convert to a diamond crystal via atomic diffusion during the course of time, if localized thermal energy is sufficient. The formation of the DLC interlayer has also been observed by other investigators in the experiments examining diamond nucleation on Si substratesl1751l18gll24gll25oll268land MO substrates.l70ll124l Csencsits et a1.l18gl employed TEM and EELS to investigate the initial stage of diamond nucleation in biased MW PACVD and found that diamond nucleation occurred on an a-C intermediate layer formed during biasing. Tamaki et al.125ol used SEM, AFM, micro-IR absorption and Raman spectroscopy to study the nucleation and growth of diamond particles

Diamond Nucleation Mechanisms

63

from CH,-H, on an abraded Si substrate in HFCVD. Their experiments provide detailed information on the diamond nucleation process. It was observed that a hydrogenated amorphous carbon (a-C:I-I) film initially formed on the substrate after 30 min (Fig. Sa), and the surface ofthe a-C:H layer was covered with numerous protrusions after 60 min (Fig. 8b). These protrusions developed into layered triangular pyramids after 120 min (Fig. 8~). The aggregation ofthe triangular pyramids was suggested to be a precursor for the diamond particles. Dubray et a1.I268linvestigated the influence of a hydrogenated disordered carbon (a-C:I-I)layer on diamond nucleation on Si and SiAlON substrates. The results of AES, SEM, and Raman spectroscopy characterization show that the presence of the a-C:H layer enabled diamond nucleation to occur readily without any polishing pretreatment, and more continuous diamond films were grown when the substrates were polished with diamond powder prior to the deposition of the a-C:H layer.

Figure 8. AFM images of deposition surfaces developed atIer (a) 30 min, (b) 60 min, and (c) 120 min.[2501 (Reproduced with permission.)

64

Diamond Chemical Vapor Deposition

Overall, these experimental results confirm that diamond crystallites are not located directly on the substrate surface but on an intermediate amorphous layer. [175][24g] Nucleation of diamond occurs readily on the disordered carbon surfaces, and the formation of this type of intermediate layers is indeed one step in the surface nucleation mechanism of diamond.[268] Further in-depth studies are needed to ascertain the size, molecular and structural arrangements of nucleation sites and stable nuclei as well as the dependence of these factors on substrate surface conditions and deposition parameters. 2.3

Nucleation

on an Intermediate

Layer of Metal Carbides

Diamond can nucleate on foreign surfaces, notably elements that form refractory carbides (Si, MO, Ta, W), without pretreatment. Thermodynamics calculations reveal that, in the absence of a plasma, SIC (as opposed to diamond or graphite) is the only stable phase expected on a heated Si substrate under conditions of 0.3 vol.% CH, in H2, 30 torr pressure, and 1000°C substrate surface temperature.[26g] Badzian and Badzian[270] suggested that diamond nucleation on Si is preceded by the formation of a P-Sic buffer layer, and diamond nucleation occurs on the surface of the carbide. This is supported by many growth experiments of diamond particles or films on Si substrates in HFCVD and m ~~~~,~~~1~~~1~~~~1~~~~1[1~~1~~~~1~*~~1~*~~1~I*~~1 which show hat &. Si surface is indeed transformed to SIC under conditions leading to diamond growth, and diamond nucleation occurs on the SIC intermediate layer. A recent AFM study[262] provides further evidence for the formation of Sic in the precursor layer. The AFM images show that the Si substrate surface was initially covered by a small-grained polycrystalline-like film with some Sic, and the nucleation of diamond took place on the top of this layer. Substrate pretreatments (such as biasing and carburization)[71][187] and appropriate selection of deposition parameters, [157]which can promote the formation of the intermediate layer of carbides, appear to enhance the diamond nucleation density and reduce the incubation period. Joffreau et a1.[52b]and Lindlbauer[276] conducted systematic studies of diamond growth on carbide-forming refractory metals and observed that diamond nucleation occurred only after the formation of a thin carbide layer. Lux and Haubner[‘*‘] subsequently postulated a model to elucidate the

Diamond Nucleation Mechanisms

65

mechanism governing the diamond nucleation process on a carbide-forming substrate. As shown in Fig. 9, carbon dissolves into the substrate material initially, resulting in the formation of a stable carbide. Diamond nucleation occurs on the carbide layer when the carbon concentration on the surface reaches its saturation value. Thus, C diffusion rate influences both the carbide growth and diamond nucleation. Lux and Haubner[52al compared the time evolution of diamond nucleation densities on Ti, Hf, Nb, Ta, MO, and W, and found that the difference in the nucleation densities and rates is related to the diffisivity of carbon in the respective substrates. Diamond nucleated relatively rapidly on Hf Ta and W due to their low carbon difisivity. The nucleation densities on these substrates were low and the crystals increased in size uniformly after the termination of the nucleation process. On Nb, which has a moderate carbon diffisivity, the nucleation density was high initially, but the number of diamond crystals decreased with time, presumably due to dissolution of carbon from diamond into the substrate. Diamond nucleation on MO was the most rapid and the nucleation density was the highest among Ti, Hf, Nb, Ta, and W, possibly because ofthe rapid formation rate of Mo,C which built a C-diffusion barrier necessary to generate the surface carbon saturation. Diamond nucleation on Ti, which has the highest carbon difisivity, experienced the longest incubation period, presumably due to the long time needed to achieve a carbide layer thick enough to impede diffusion of surface carbon into the substrate. The nucleation density was low initially and increased with time, leading to a broad size distribution of diamond crystals on Ti. Overall, the incubation period for nucleation is the shortest on the metal that can most rapidly achieve a supersaturation of carbon on the surface. The formation of Sic on Si substrates in the initial stages of nucleation was also observed by Waite and Shah[2121in MW PACVD using XPS. A diamond peak appeared after the incubation period with a simultaneous decrease in SIC peak intensity. However, SIC remained as an impurity even after several hours of deposition. Belton et a1.tg6]identified a Sic layer during diamond nucleation on a Si substrate in HFCVD via in situ XPS measurements. The carbonaceous contamination and a SiO, layer present on the substrate prior to growth were removed in the first 1.5min ofgrowth and a SIC layer formed. Diamond then nucleated and grew on the SIC layer.

66

Diamond Chemical Vapor Deposition

jizerburization] Surface

reactions

Nucleation

Growth

Figure 9. Schematic diagram showing the proposed nucleation mechanism: diamond nuclei form on a carbide interlayer on a carbide-forming refractory metal substrate.tsza] Initially, carburization consumes all available C to form a carbide surface layer. A minimum C surface concentration required for diamond nucleation cannot be reached on the substrate surface. With increasing carbide layer thickness, the C transport rate slows and the C surface concentration increases. When the C surface concentration reaches a critical level for diamond nucleation, or a surface C cluster attains a critical size, a diamond nucleus forms. @eproduced with permission.)

Smolin et a1.l248lreported the formation of a MO carbide layer in the initial stages of diamond film deposition in DC arc discharge CVD. Bou et a1.t156]employed a powerful micro SIMS technique coupled with a computerdriven data acquisition system to monitor the transverse cross-sections ofthe MO-diamond interface during nucleation on MO substrates in M&VPACVD. It was demonstrated that diamond nucleation took place on a 2-3 pm thick Mo,C layer. In diamond growth experiments on MO and Si substrates using MW PACVD by Meilunas et al., 12”1Mo,C and SIC layers of approximately 1.5 pm and 10 nm in thickness were observed with SEM within 1 min and after 5 min, respectively. The growth rate of Sic was much less than that ofMo,C. Diamond nanocrystallites were observed after 1 min, and no further carbide layer growth was detected once the surface was covered with diamond. The effects of surface carbon on diamond nucleation depend not only on the diffusion rate of carbon in a substrate, but also on the structure of chemical bonding of carbon to the substrate surface and the uniformity and

Diamond Nucleation Mechanisms

67

thickness of the carbon layer. Michau et a1.t2781discussed the effects of surface carbon bonding structure on the nucleation density and crystal quality of diamond. Two types of carbon were considered: (a) chemically bonded carbon on a carbide interlayer and (6) “free” carbon on a substrate surface. It was speculated that a diamond nucleus may be induced as a result of the interactions of carbonated gaseous species with a substrate surface only if carbon supersaturation is reached locally on the surface. In their studies, a graphite disk substrate and W, MO, and Cu substrates precarburized or deposited with graphite were used to develop carbon supersaturation on the surfaces. X-ray diffraction analyses confirmed the formation of tungsten carbides (W2C and WC), and molybdenum carbides (Mo,C and MoC,_J at the interface between diamond films and W and MO substrates, respectively. Their experiments show that carbon participating in the surface composition of a substrate (as a carbide, for example) plays an important role in diamond nucleation since the nucleation density was enhanced for the precarburized substrate compared to the metallic substrate, in spite ofthe presence ofa small amount of sp2 C in the diamond deposit in the case of precarburization. The local structure ofthe carbide interlayer may assist epitaxy. On the other hand, free carbon deposited onto the substrate surface (as graphite, for example) induced an even higher diamond nucleation density, and the formation of sp2 C was also detected. The nucleation density on the graphite disk substrate was the highest among the substrates used (graphite, W, MO, and Cu), but ball-like crystals were obtained on the graphite substrate, compared with the faceted crystals on the other substrates. These results suggest that free carbon on a substrate surface may lead to a higher diamond nucleation density relative to chemically bonded carbon on a carbide interlayer, and the carbon excess on a substrate surface favors the ball-like (cuul$ower) morphology. Regarding the role of carbide intermediate layers in diamond nucleation processes, some contrary experimental data have also been reported. For example, the experimental results of Williams et a1.l273ldemonstrate that a SIC interlayer grew to its equilibrium thickness within -2 to 5 min ofplasma exposure, but in Bachmann et al.‘s workluhl an incubation period of -10 hours was necessary before diamond crystallization could be detected in the absence of pretreatment. Yarbrough and Messiert541 argued that, if Sic formation is both a necessary and sufficient condition for diamond nucleation, some rationalization for the large difference in time between the carbide formation and the initiation of diamond growth is needed. In addition, Stoner et a1.1r7*l stated that attempts to grow diamond on untreated bulk Sic

68

Diamond Chemical Vapor Deposition

substrates did not yield high nucleation densities. Hartnett et al.12791 reported three times higher diamond nucleation densities on (220) textured, CVD grown P-Sic films than on untreated Si, but no nucleation enhancement on ion-sputtered Sic and (111) textured P-Sic films. Indeed, although the formation of carbide interlayers is an important factor for diamond nucleation on carbide-forming substrates, it seems not to be a sufficient rationalization of all the available data. In other words, the formation of carbide interlayers is not the only mechanism for diamond nucleation. Intermediate layers of a-C:H or a-C have been found to form on Si and MO substrates, as reviewed above. Graphitic intermediate layers, as discussed in Sec. 2.4, have also been shown to form on some substrates prior to diamond nucleation. A possible explanation would be that, what intermediate layers form depends not only on substrate materials and pretreatment methods but also on deposition conditions. For example, Belton and Schmieg12561 have observed distinctly different intermediate layers on different substrates in HFCVD of diamond. In their experiments, graphite carbon deposited on Pt substrates, while a thick graphite layer developed on Ni substrates prior to diamond nucleation. Diamond eventually nucleated on defect sites in these graphite deposits. Moreover, different gas compositions or different substrate temperatures can produce different intermediate layers on the same substrate. For example, in MW PACVD of diamond on a Si substrate scratched with diamond paste,l271ll273l an interfacial, single-crystal P-Sic layer of 5 nm grew for 0.3 vol.% CH, in H,, whereas an amorphous layer, instead of the SIC layer, was observed for 2 vol.% CH, in H2. It is speculated that at low CH, concentrations, non-diamond phases are etched more effectively and Si atoms can diffuse to the surface to form SIC. Hence, it is safe to say that the formation of the interlayers is a necessary step in the spontaneous nucleation processes of diamond on nondiamond substrates, but this alone is not sufficient for nucleation to occur.l178l Surface carbon saturation and defects or high-energy sites,1280l1281l present in a-C:H or a-C intermediate layers,1124112821 in carbide interlayers (due to island growth of carbides and carbon accumulation on carbides),l’78l or in graphite interlayers,l2561constitute a sufficient condition for diamond nucleation on these interlayers. Since a DLC layer has a high concentration of active surface sites for diamond nucleation due to high surface defect density and high hydrogen concentration, it is then easy to explain why nucleation enhancement is more significant on a DLC layer than on abraded substrates or carbide layers.l124ll282l

Diamond Nucleation Mechanisms

69

In addition to these observations and speculations, the in vucuo surface analyses by Stoner et a1.t178]provide further detailed information about the role of Sic interlayers in enhancing diamond nucleation. They employed several analytical techniques, including XPS, AES, EELS, HRTEM, SEM, and Raman spectroscopy, to examine diamond nucleation processes on negatively biased, mirror-finished Si substrates in MW PACVD. The HR cross-section TEM confirmed that a complete, I-10 nm thick, amorphous Sic layer developed before diamond could be detected, and a small amount of a-C on the surface of the a-Sic layer played a major role in the nucleation sequence, as shown in Fig. 10. In Fig. lOa, the a-Sic interfacial layer between the Si substrate and diamond can be readily observed. Several diamond nuclei can be seen emerging from the interfacial layer, and none of the nuclei are in direct contact with the Si substrate surface, suggesting that diamond nucleation did not occur on the Si substrate directly, but rather on the Sic inter-facial layer at the locations close to the Si/SiC interface, as depicted in Fig. 1Ob. This phenomenon was attributed to the preferential etching of Si from the Sic layer. During this non-uniform etching, Si was preferentially depleted and excess carbon could then saturate the surface locally. This local C saturation would allow critical carbon clusters, and eventually diamond nuclei, to form at the locations where the Sic layer was relatively thin. Based on these results and the basic three-dimensional nucleation steps on a flat surface,t2831 a detailed nucleation mechanism was proposed,t’78] as described below: 1. Atoms impinge upon the substrate surface from the gas phase and become adsorbed onto the surface. 2. The adatoms may either re-evaporate (desorb) or diffuse over the surface; the adatoms may also diffuse into the substrate or bond to other surface atoms. At standard CVD temperatures, it is more energetically favorable to form Sic than for the adatoms to accumulate on the surface and eventually form clusters; thus Sic forms almost immediately after the Si substrate is exposed to the CVD environment. 3. The formation of Sic is kinetically limited by Si diffusion to the surface through the growing carbide layer (note that Lux and Haubner[‘*‘] suggested that it is limited by the C diffusion through the carbide layer); thus, even if there

70

Diamond Chemical Vapor Deposition

(4 Diamond nuclei A

Amorphous

. carbon

Initial nucleation point ’

Amorphous silicon carbide Silicon substrate

(b) Figure 10. (a) Low-magnification HRTEM micrograph of cross-section of a sample biased for 1 h and further grown for 5 h showing several diamond nucleation sites and an interfacial layer of a-Sic between Si and diamond. Twin lamellae, prominent defects in CVD diamond, can be observed where the nuclei begin to coalesce, as indicated by the arrows. (8) Schematic diagram showing nucleation details and how nucleation may have occurred closer to the substrate than it was observed in cross-section TEM.Ir’*l (Reproduced with permission.)

Diamond Nucleation Mechanisms

71

exists initially a high C concentration on the surface, it may not be kinetically favorable for critical nuclei of diamond to form until the carbide layer has reached a critical thickness such that the outward diffusion of Si is significantly slowed down (or the inward transport rate of C significantly slows down).t52~ Due to the decreased C consumption and/or preferential Si etching, the surface C concentration increases to a critical level such that the adatoms can form clusters; the carbon on the surface has been identified as neither diamond nor graphite, but instead amorphous carbon. 4. Through statistical fluctuation in the local adatom concentration, the clusters grow or decay. 5. There exists a critical size above which the probability of growth will be greater than decay, so that the clusters, whose size exceeds the critical size during the concentration fluctuation, become stable. 6. The stable clusters provide suitable sites for growth either from the continued migration of single adatoms or from the direct impingement of atoms from the gas phase. These steps are consistent with the two criteria for “spontaneous” (non-epitaxial) diamond nucleation on a surface, i.e., (a) carbon saturation of the surface, and (b) presence of high-energy sites (unsatisfied va1ences).[280j[281]He rice, it is reasonable to speculate that the excess amorphous carbon on the substrate surface contributes significantly to the nucleation of diamond,t178] and any means that can accelerate surface carbon saturation may help to expedite diamond nucleation. Moreover, judging from the much higher nucleation densities achieved on carbide-forming substrates compared with those on non-carbide-forming substrates, either the amount or type of the excess carbon, which has the correct crystal structure and bonding configuration to promote the formation of diamond structure, must be related to the carbide-forming nature of the substrates. The role of a carbide interlayer in promoting diamond nucleation[178] is that it provides a temporary but critical host on which carbon can accumulate until clusters of the appropriate size and structures required for diamond nucleation can develop. However, the size, molecular and structural arrangements of nucleation sites (clusters) are still unclear. Further studies are needed to address these issues.

72

Diamond Chemical Vapor Deposition

2.4

Nucleation on an Intermediate Layer of Graphite

The prevailing view on the role of graphite particles is that graphite formation competes with diamond formation and is to be avoided.[26][87] However, enhancement of diamond nucleation on graphite powder,[284] graphite fiber,[1s11[284]graphite disk,[278] and graphite fih11[~~~1 has been reported. Microbalance studies of diamond nucleation on Pt[“j] show an initial incubation period during which an oriented graphite deposit formed. The deposit subsequently disappeared, and the final deposit contained only Several other ~~p~~~~~~~‘~~l~~~l~~~~l~~~~l~~~~l polycrystalliue diamond. have also provided direct evidence for the formation of graphite on Ni, Pt, Si, Cu substrates prior to diamond nucleation. Belton and Schmieg ‘s experimental studies[2561provide detailed information on the formation of graphite particles and layers on Ni and Pt substrates prior to diamond nucleation. They used XPS, EELS and LEED to examine the time evolution of surface carbon species during diamond nucleation and growth on the substrates in HFCVD. The results reveal that on Ni substrates (Fig. 1 l), very early in the growth, a thick graphite layer was deposited onto both the scratched (with smooth surface) and unscratched surfaces (in an ordered array parallel to Ni(lOO)). Diamond nucleation occurred readily on the scratched substrate, while the graphitic layer on the unscratched substrate underwent a roughening or disordering process which occurred at a very slow rate. After a very disordered surface of the graphite layer was generated, diamond nucleation was observed. The disordered surface has graphite edge sites similar to those introduced by scratching. These sites are favorable for diamond nucleation. On Pt substrates (Fig. 12), early growth on both scratched and unscratched surfaces underwent the same event, i.e., the formation of graphite and an adsorbed hydrocarbon deposit. On the unscratched foil, the surface graphitic carbon was eventually removed (etched) without nucleation or growth of diamond to follow; on the scratched surface, diamond nucleation and growth were observed before the graphitic component was completely etched. These results suggest that the graphitic deposits provide nucleation sites for diamond, and defects such as the edges ofgraphite planes are necessary for diamond nucleation. Scratching pretreatment increases the number ofnucleation sites by introducing defects into the graphite layers and stabilizes the graphitic carbon deposits such that they remain on the surface long enough for diamond to nucleate, thereby enhancing the probability of the formation of a suitable nucleation site.

Diamond Nucleation Mechanisms

73

Scratched Foil ‘Belore prow31 ‘Ordered C-atomArray’ “Small C-AtomConcentration”

‘Graphite Deposition’ Ordered Graphite Overlayer (LEED Pattern)

Smooth Graphite Overlayer (SW

‘Graphite C&order’ -waphite

Diamond ‘Nucleation’

c-

Graphite

Figure 11. Schematic representation of nucleation and early growth stages of diamond on an unscratched Ni(lOO) single crystal surface and a scratched Ni foil. Graphite is formed initially on both surfaces, however, the highly ordered graphite layer on Ni(100) does not nucleate diamond until the surface becomes disordered. Defect sites, introduced by scratching or disordering of the graphite, are necessary for diamond to nucIeate.t256] (Reproduced

with permission.)

74

Diamond Chemical Vapor Deposition

Unrcratched

Screlched

Graohile

Qraphite

Before Growth ‘Mostly Graphite’

5!!&%Y+

&aphite

Early Growth ‘Graphite Etching’ WC Reptacement’ Graphite

Diamond /

‘Stabilized CH Surface’ Later Growth ‘Nucleation’

Gkphite

Diamond

‘Film Growth

Figure 12. Schematic representation of nucleation and early growth stages of diamond on scratched and unscratched l?t foils. Diamond nucleation occurs only on the scratched foil which has a stable graphitic carbon deposit, and hydrocarbon species adsorbed on Pt substrate surfaces appear to be unimportant for diamond nucleation processes on Pt since they are also present on unscratched Pt surfaces which do not nucleate diamond.tzs6t (Reproduced with permission.)

Diamond Nucleation Mechanisms

75

In MW PACVD of diamond on Si WA and (=~[~871[~99lWl s&c&-&+ it was observed that, as the substrates became supersaturated with carbon, graphite began to form in the early stages of diamond nucleation. The graphite was then preferentially etched during the growth of diamond. The graphite layer formed greatly enhanced diamond nucleation.[1gg1*285] On the basis of these experimental findings and variable metric static energy minimization calculations using the Tersoff semi-empirical manybody potential, Lambrecht et a1.[2641proposed a detailed nucleation mechanism, as shown schematically in Fig. 13. It was suggested that graphite initially condenses on the substrate surface and the { ITOO} prism planes are subsequently hydrogenated. Diamond nuclei grow preferentially on the prism planes of graphite, with kinetically preferential nucleation at the emerging graphite stacking faults, and with an almost perfect interface between graphite and the diamond nuclei. There exists a preferential epitaxial lattice registry relationship between graphite and diamond, i.e., (11 l)diamond11(OOOl),.+;,, and (1 lO)diamond11(1 120)~~~~~~~~~‘6~~1gg~~284~~285~ This relationship means that the puckered six-member rings in the diamond (111) plane retain the same orientation as the flat six-member rings in the graphite (0001) basal plane. Etching of graphite occurs simultaneously during diamond nucleation. Atomic hydrogen, known to be important in diamond growth, also plays an important role in nucleation: by terminating the dangling surface bonds it stabilizes sp3 nuclei with respect to sp2 nuclei. It also serves as a reactive solvent which permits the conversion of graphite nuclei into diamond nuclei, hence circumventing the large activation barrier between graphite and diamond. Lambrecht et a1.t2641stated that this detailed nucleation mechanism is the dominant channel for the spontaneous nucleation of new, independent diamond crystals in the absence of preexisting diamond seeds. They indicated that the elimination of graphite precursors is necessary for the suppression of secondary nucleation, which limits the growth of large diamond single crystals. On the other hand, for faster growing highly oriented, epitaxial, polycrystalline diamond films, enhancing nucleation with graphite precursors is desirable during the early stages of heteroepitaxial deposition, when rapid coalescence of oriented diamond nuclei is required. The epitaxial relationship of diamond with graphite may provide important clues for new routes to heteroepitaxy if methods can be found to orient the initial graphitic precursors.

76

a

Diamond Chemical Vapor Deposition

[llll

[OOOl]

t

[ii001 C B A

C B A

A

B C B A C

Figure 13. Schematic diagram showing the proposed nucleation mechanism: diamond nuclei form on a graphite interlayer. Initial condensation of graphite and subsequent hydrogenation of the { ITOO} prism planes along the edges of graphite particles are followed by kinetically preferential nucleation of diamond at the emerging graphite stacking faults, and with an almost perfect interface between the graphite layer and the diamond nuclei. (Upper) Cubic diamond on perfect hexagonal graphite. (Lower) A twinned diamond nucleus adjoining a graphite stacking fault. Twin boundaries in diamond are indicated by the dashed lines, H atoms by the small open circles, and C atoms by dark solid circles. The larger open circles indicate the initial nucleus formed at the interface by tying together the graphite layer with tetrahedrally bonded C atoms. (Macmillan Magazines Limited, 0 1993. Reproduced with permission J;om Nature.[2641 )

Diamond Nucleation Mechanisms

3.0

77

SUMMARY

Homogeneous nucleation of diamond in the gas phase and its contribution to different deposition processes are poorly understood. A limited number of experiments have been conducted to examine the homogeneous nucleation at atmospheric and subatmospheric pressures. The number of diamond particles collected from the gas phase is very small compared to typical surface nucleation densities. Therefore, the homogeneous nucleation mechanism cannot account for the large variety of nucleation densities observed on different substrate materials and from different surface pretreatments. Whether and how the diamond particles formed in the gas phase could serve as seeds on the substrate surface for subsequent growth of a diamond film remain unknown. Two criteria must be satisfied for spontaneous (non-epitaxial) surface nucleation: (a) carbon saturation of the substrate surface, and (b) presence of high-energy sites (unsatisfied valences). Diamond nucleation on nondiamond substrates occurs mostly on an intermediate layer of a-C, DLC, metal carbides, or graphite formed at the diamond/substrate interface due to the chemical interactions of activated gas species with the substrate surface during the incubation period. Such intermediate layers provide nucleation sites for diamond crystallite growth, and hence enhance diamond nucleation density on non-diamond substrates and provide an opportunity for controlling the morphology, orientation, and texture of diamond films. The formation of the interlayers is a necessary step in the spontaneous nucleation processes of diamond on non-diamond substrates, but it alone is not suffkient for diamond nucleation to occur. Surface carbon saturation and defects or high-energy sites constitute a sufficient condition for diamond nucleation on the interlayers. The formation of the interlayers and the formation rate depend not only on substrate materials and pretreatment methods but also on deposition conditions. Distinctly different intermediate layers may form on different substrates at different rates, and different gas compositions or different substrate temperatures may produce different intermediate layers on the same substrate. Low C/H ratio and/or high substrate temperature may favor the formation of carbides, while high C/H ratio and/or low substrate temperature may lead to the formation of a-C or DLC, or even the direct nucleation and growth of diamond on the substrate surface. The thickness of the interlayers is also dependent on these factors and may range from several angstroms up

78

Diamond Chemical Vapor Deposition

to a few micrometers, as summarized in Table 1. Further in-depth studies are needed to ascertain the size, molecular and structural arrangements of nucleation sites and stable nuclei as well as their dependence on substrate surface conditions and deposition parameters.

Table 1. Type and Thickness of Interlayers Formed on Different Substrates in Different Deposition Processes

4nm,lR9,

MWPACVD

-

865

10 nm ,277,

MWPACVD

-

950

0.3.ICHq 0.202

200

5 nm ,271.273,

MW PACVD

-

BOO

0.3

Y

I-lOnm117K]

MW PACVD

-

1

2

IIWN)

MWPACVD

-

1.6

XM

with 10 pi dsamond Pwdcr ruapcnded ,n ethyl .Ilcohol:DC h,armg:(140”, 37 5 H2 plasma lor 15 m,n: DC h,armE (-200”)

St

Corbndes

SIC

12121 Mr$

2-3pm,156, MWPACVD -

polished with 0 25 pm d,amond p;ls!c. ullrason~cally cleaned in methanol nnd DI water 30 puhshcd w,d, 0 25 ,,m diamond parle.ul~~omcally cleaned I” acea,ne.cthanul and d,ald,cd wi,,cr 15 ullr.lrwlc;dly LIC;IIICd111 TCE.acclone.methan,rl. 2-propanol,rmscd in DI water.d!ppcd Ior I min in HCI DI (I:lO).rinsed m D1 waler. blown dry w,,h N2: DC hosing (-25OV) 40 nolushed wth 0 25 ,,m duunond pnrte 30 ? 30 see,277, ;,hove IO “ftlaN,“lcidlY c,e;med

900

0.6

?

910 950 1075

2 0 3.ICH4.0 202 0.4

102 200 5112

I 0.4 0.4 0.4

XX, 502 502 502

50 10

2”

1.5 ,m, ,277, I6 wr, 12751

MWPACVD HFCVD (Ts)

12781 2 ,un ,27S, 4 pm ,275, (I pm 12751

HFCVD (W,

2ODo

BSO

HFCV” CT.,) HKvD (‘l_i,, HFCVD (Ta)

2NYI 2,““) 2wO

,075 1075 1075

6 A ,256]

HFCVD (W)

2327

827

0.5 IO]/[C]~O.S

IwJaccn,

8 nrn 12561

HFCVD (W)

2327

827

0.5 [o]/[clio.s

IN, LCC,”

10A[1871

MWPACVD

-

650/6W’

2/l

?

,212,

MW PACV”

-

9,HI

06

?

2000

30

I”

W&WC Nh~(‘.NhC Ta2C,TaC (;r,,Ph,,P P&kS

Layers

I0 10

SI

SI

s,

SI Mo MI, MO-

nCeil”C

prccarhurinmun bee ,275, ahwe 127.5,;,huve xc ,275, .,hovc

ret

acrrahcd wth 6 pm dnmond pess.ul,rasun,c:,lly clcancd wth mchlorwthanc. accmne. nnd mcthunol 20 xc ,256, uhwc

IS/25 polished wlh S1C.30.6. md d,;tmond P;,\le. ” 3 and 0 “5 pm AI>O,: The XII II \:I,nc ;,* 178,

I ),,n

W W Nh ‘I:, 1’1

NI Cu

,

T,, 1; nlamcn, and ruh\l,a,c ,emper.,wc: * htasing condiuanslgrowh cnndiuonr

P ,o,:,, gas prc\r”rc.

40

TCE. ,r,chlon,e,hy,cnc:

\cc ,212, :,hwc “I, dc-,on,,cd

St

5 Diamond Epitaxy, Oriented Growth, and Morphology Evolution

1.0

EPITAXY

Homoepitaxial growth of diamond has been demonstrated by many invcstigators. [26][146] Some semiconductor devices have been developed on homocpitaxial diamond films. [286][287] Diamond films epitaxially deposited on both diamond and cBN single-crystal substrates provide evidence that under typical low pressure processing conditions, single-crystal diamond deposition is possible. However, deposition of monocrystalline diamond films on substrates other than diamond and cBN is desirable due to the difficulty in obtaining single crystals of either of these materials large enough to be used as substrates for many applications of practical interest. Methods for obtaining large-area single crystals or at least highly oriented diamond films must be developed for large scale electronic applications. Therefore, nucleation and growth of heteroepitaxial diamond films are of prime importance if full advantages of the unique properties of diamond arc to be exploited. Hetcrocpitaxial growth of diamond has been and will continue to be a major research objective. It is crucial to develop a technique to accommodate this objective and to make it economically viable.

79

80

Diamond Chemical Vapor Deposition

The primary difficulty inherent in this issue is the small number of materials with suitable crystal structures and lattice constants. Some transition metals and ceramics, such as Ni, Cu, Fe, and cBN (Table 5, Ch. 3), are the few isostructural materials with sufficiently similar lattice constants (mismatch ~5%). In addition, the extremely high surface energies of diamond (ranging from 5.3 to 9.2 J mm2for the principle low index planes)[‘l and the existence of interfacial misfit and strain energies between diamond films and non-diamond substrates constitute the primary obstacles in forming oriented two-dimensional diamond nuclei.t1661 Earlier attempts to grow heteroepitaxial diamond on the transition metals were not successful. The reasons may be related to the high solubility/ mobility of C in/on the metals (for example, Fe, Co, Ni)[541 or the formation of an intermediate layer. For example, in Belton and Schmieg’s experiments of diamond epitaxy on Ni( 100) single crystals,l256l a thick, poorly oriented graphite and glassy carbon interlayer formed between the Ni substrate and the subsequently grown diamond layer. This interlayer prevented the diamond layer from growing in lattice registry with the Ni surface. The progress in experiments [166112881[28gl has demonstrated a promising potential for the epitaxial growth of diamond on Ni. Sato et a1.[28gl reported the feasibility of the epitaxial growth of diamond on Ni and Co in MVVPACVD with CO.9 vol.% CH, in H2. Badzian and Badzian12881achieved the heteroepitaxial growth of diamond on Ni by transforming Ni into Ni hydride and suppressing graphite nucleation. Yang et al.ll@jl deposited wellcrystallized and highly oriented diamond on seeded Ni substrates without graphite formation in a multistep HFCVD process. Epitaxial diamond has also been grown on Cu by means of a laser heating method.l2g0l Heteroepitaxy of diamond on cBN powder has been achieved by Yarbrough 123gland others.12401t2411Heteroepitaxial diamond films, 0.7 to 5 urn thick, have also been grown on HPHT-synthesized cBN crystals in MW PACVD.12**1 Jiang et a1.l184lused AFM to study the early stages of diamond heteroepitaxy on negatively biased single crystalline Si( 100) substrates in MW PACVD. They observed that non-faceted nuclei of -5 nm in diameter formed initially. These nuclei exhibited crystalline structures and increased in size with deposition time (Fig. 1). The initial nuclei had a preferential (100) orientation and could grow epitaxially once a critical nucleus radius was reached through proper control of the nucleation and growth process.

Diamond

Epitaxy,

Oriented

Growth, and Morphology

Evolution

81

Vi

tct

Figure 1. Nucleation and growth processes of an epitaxial diamond Iilm on a negatively biased single crystalline Si(lO0) substrate monitored with AFIVI[‘~‘] (a) Si surface cleaned with acetone in an ultrasonic bath and etched with a microwave hydrogen plasma, (b) diamond nuclei after 5 min deposition, (c) diamond nuclei aller 20 min deposition, and (d) epitaxial diamond film grown for 22 h after 20 min nucleation. (Reproduced wifh pf?-~iSSiO?,.,

2.0

QRIENTED

AND TEXTURED

GROWTH

It is known that the relative growth rates on { 100) and { 1 I I} planes v,,,&~~, determine the crystal habit of diamond,[26] and the appearance or disappearance of crystallographic planes in diamond films depends on the growth velocities of the corresponding planes. The facets that appear on a crystal are those for which the normal growth velocity is the slowest. On the basis of the simple Wulff criterion ~‘1 for ctystal habit, the most stable growth planes in diamond are the octahedral { I1 1) planes, followed by the

82

Diamond Chemical Vapor Deposition

cubic { lOO} planes and the { 1 lo} planes.[54] Spitzyn et a1.rz6]indicated that at a low substrate temperature (vu,O/vlll 2 $), the crystal habit of diamond is octahedral, at a high substrate temperature (vroO/vl11I d/3), it is cubic, and inbetween(vI,,/vrII - $/2), it is cube-octahedral. With increasing substrate temperature, the relative growth rates decrease, and hence the morphology of polycrystalline diamond films evolves from octahedral, to cube-octahedral, to cubic. Increasing the hydrocarbon concentration in the gas phase has the similar effect on the crystal morphology. Void-type defects produce stresses in homoepitaxial diamond films. Decreasing substrate temperature from 1100 to SOO”C,the stresses increase by a factor of -3. [26] The stresses may result in deterioration of the morphology and a gradual transformation from a single crystalline film into a polycrystalline film. When the stresses exceed a certain threshold, lattice discontinuity and microtwin lamellae will form in a film. Twinning probability is known to be proportional to supersaturation. Thus, the high twindensity observed in CVD diamond films demonstrates that syntheses take place at a relatively high supersaturation. Regarding the formation of texture in polycrystalline diamond films and the growth of oriented, textured diamond films, extensive studies and Aposc&leme&_ progresses have beenmade. ~~~1~~~1~~~~1~~~~1~~~~1~~~~1~~~1~~~~~1 nism for the growth of oriented crystals from randomly oriented nuclei has been proposed by Van der Drift t246](see Fig. 4b, Ch. 4). The model, named evolutionary selection, takes into account the competition between crystals in a polycrystalline film during growth. As growth proceeds, an increasing fraction of the crystals are overgrown by the adjacent crystals, and the number of the remaining crystals decreases progressively. The resultant texture is a function of the growth competition between different crystal orientations. The crucial factor, which determines the probability of survival of an individual crystal, is its vertical growth rate. Only those crystals with a rapid growth direction perpendicular to the substrate surface will survive, whereas crystals with all other orientations will be gradually buried. Based on this mechanism, Wild et a1.[245jdeveloped a two-dimensional computer program to simulate the crystal growth evolution and the surface morphology of polycrystalhne films grown from square nuclei with random orientations and random positions on a substrate. The calculated[245] and expe&3n~l[2451[2951 results show that, for the diamond films (150-400 urn in thickness) investigated under the experimental conditions, (110) is the direction of fastest growth with the film surface consisting of { 11 l} planes.

Diamond Epitaxy, Oriented Growth, and Morphology Evolution

83

The degree of the (110) fiber texture improves (i.e., a preferential orientation of { llO} planes develops) with increasing film thickness. However, the film texture and surface morphology are critically dependent on the growth conditions. Under conditions where { 100} or both {100} and { 111) planes develop, the preferred orientations different from the (110) fiber texture may form. Recent success171111861 in highIy oriented, textured diamond growth on Si represents a novel approach to obtaining diamond films of a near-singlecrystal morphology over large areas. This advance may allow diamond to realize its potential as an electronic material in the near future. In initial growth experiments, no texture formed due in part to the formation of a-Sic interlayers, which led to randomly oriented diamond particles. Subsequently, highly oriented, textured growth of diamond films on single crystal Si( 100) substrates was achieved in h4W PACVD by combining a two-step nucleation process using CH,/H, and a growth process using CH&I.&O. Textured diamond films were first nucleated via an in situ carburization and a biasenhanced nucleation process (Table 1). It is speculated that the carburization converted the Si surface to an in situ epitaxial P-Sic layer, and at the same time, reduced the nucleation incubation period, and thus diminished the eroding effects of the subsequent biasing process. In the absence of this carburization step, an amorphous Sic interfacial layer formed prior to diamond nucleation, which destroyed the crystallographic registry relationship. The bias-enhanced nucleation took place on the l3-SiC, maintaining crystallographic registry across the interface and enabling textured, oriented growth of diamond. The nucleation stage, which produced partially (50%) oriented diamond nuclei, was immediately followed by a (100) textured growth process, similar to Wild et al.‘s approach,l245l12g5l12g6l resulting in a diamond film of -15 urn in thickness, in which essentially all of the grains were both (100) textured and epitaxially oriented relative to the Si substrate (Fig. 2). The (100) textured diamond films were grown at high growth rates under conditions of lower substrate temperatures (650-7OOOC)and higher CH, concentrations (lo-15 seem with 180 seem H, and 5-20 seem CO, i.e., 5-7 vol.% CH,). The films exhibited columnar grain structure and low surface roughness. Low-angle (O-6”) grain boundaries also developed between (100) oriented grains. This misorientation was shown to result from the large concentration of misfit dislocations at the diamond/SiC interface due to the large lattice mismatch between diamond and P-Sic (-20%). Since the degree of texture improves with increasing film thicknessl245land the

84

Diamond Chemical Vapor Deposition

misoriented grains could be eventually buried as the film thickness increases, it should be beneficial to continue growing the films out to greater thicknesses in order to reduce the overall n&orientation. Table 1. Deposition Parameters Used in In Situ Carburization, Biasenhanced Nucleation and Growth Processes[186] Parameters Bias voltage (V)

Carburization -

Bias nucleation -170 to - 200

Bias current (mA)

-

300-400

MW power (W) Pressure (Torr) Gas flow rate (seem) H2 CH4 co

Growth -

1000

1000

1300

20

20

3s

300 6

300 6

180

-

-

Substrate temperature (‘C)

800

800

Duration

3h

15 min

1s

5

650 lo-40h

(4

Figure 2. (a) Cross-sectional SEM of a (100) textured diamond film of high-quality by bias enhanced nucleation and textured growth of 40 h under the conditions that promote (100) texture development. (b) Cross-sectional SEM of a highly (100) oriented, textured diamond film of high-quality by carburization, bias enhanced epitaxial nucleation and textured growth of 40 h under the conditions that promote (100) texture development, showing coalescence of diamond grains and low-angle grain boundaries between them. The film exhibits low surface roughness and columnar grain structure; nearly 100% of the grains are both (100) textured and in an epitaxial alignment relative to the Si substrate.t’8~ (Reproduced with permission.)

Diamond Epitoxy, Oriented Growth, and Morphology

Evolution

85

Most recently, Wild et a1.t5r]quantitatively investigated the dependence of film texture and morphology on growth conditions. They considered three types of oriented diamond films: (a) strongly fiber-textured films, (b) epitaxially textured films grown on { lOO} Si, and (c) homoepitaxial films grown on { lOO}diamond. The { 100) faceted surfaces were selected because CVD diamond grown on {100) is known to contain much fewer structural defects than on {111}. Diamond films were grown from a CH4-Hz gas mixture under various growth conditions in MW PACVD. Wild et al. introduced a new growth parameter CL,cc= (vIce.&r1&b, and indicated that the microstructure (texture and orientation) and morphology of diamond films can be controlled by controlling this parameter, which in turn can be changed by altering the growth conditions, as shown in Fig. 3for the case of thejber-texturedfjlms. In this figure, three different growth regimes can be distinguished: a I 1.5, 1.5 < a I 3, and a > 3. At low CH, concentrations and increasing substrate temperatures (01< 1S), the films exhibit pronounced ( 110) texture; at intermediate CH, concentrations and substrate temperatures (1.5IaI3),atransitionofthefiberaxisfrom(110)to(100)occurs;afurther increase in CH, concentrations or decrease in substrate temperatures (a > 3) leads to a sudden deterioration of the film morphology. Under these conditions, the films are line grained and the film surface does not show any crystalline facets. Hence, the a = 3 line demarcates the sudden transition from sharply (100) textured films to disordered, fine-grained materials. In the case of the homoepitaxial or heteroepitaxially texturedjlms of { lOO} orientation, the microstructure and morphology of the films are strongly affected by twin formation, which may lead to complete deterioration of the epitaxial orientation. Large values of a (a = 2.5) result in suppression of twin formation, and the subsequent film growth improves epitaxial alignment of crystals. In contrast, small values of a (a = 1.5) lead to deterioration of the film orientation and morphology due to twinning. In the heteroepitaxial growth, the primary, epitaxially oriented crystals were superseded by the crystals with the orientation of first-order twins, and the epitaxial orientation appeared unstable with respect to twinning. In the homoepitaxial growth, the film surface exhibited macroscopic steps and a number of misoriented, secondary crystals.

86

Diamond Chemical Vapor Deposition

METHANE CON~ENTRATIOi

(X)

Figure 3. Dependence of film texture and morphology on CH, concentration and substrate temperature.[“] The parameter ~~cc is the tilt angle of (100) directions with respect to the substrate normal. (Reproduced with pemission.)

Clearly, the substrate temperature and gas composition are the most important factors governing film texture and surface morphology. By employing different combinations of these two process parameters (Fig. 3) it is possible to control film texture, surface morphology and stability with respect to twin formation, and in particular, to grow (100) textured diamond films with { 100) faceted surfaces. Therefore, criteria such as film quality or deposition rate may be used to determine the optimum growth conditions for a desired film texture. Moreover, the variation and control of film texture are also achievable by adding oxygen-containing species or nitrogen to the gas mixture. The addition ofthese species is found to catalyze and stabilize (100) texture.[511[20g]For CH, concentrations in the range of l-2 vol.%, the addition of small amounts (40-200 ppm) of nitrogen leads to a transition of the film surface morphology from nanocrystalline to large, coplanar { 100} facets with concomitant change of the crystal structure from a (110) to a (100) fiber texture and a significant improvement of the crystalline quality. The effect of nitrogen on film texture and morphology is attributed to the growth-sector-dependent incorporation of nitrogen,[210]which leads to significantly larger nitrogen concentrations in { 11l} growth sectors relative to

Diamond Epitaxy, Oriented Growth, and Morphology

Evolution

87

{lOO} sectors. In addition, nitrogen affects the growth rates of {lOO} and { 11 1 } planes in different ways, and the surface poisoning or the generation ofgrowth steps may be a possible mechanism explaining the different effects of nitrogen on the respective growth rates. Although the operative mechanism is not yet understood, these studies clearly show the effects of oxygencontaining species or nitrogen on the resultant film texture and morphology. The results are, therefore, of particular importance to the applications requiring the incorporation of impurities such as for electronic devices.

3.0

MORPHOLOGY

EVOLUTION

Regarding the crystal size distribution and evolution during growth, quantitative studies on nucleation and growth of diamond films on Si(100) in MW PACVD11741reveal that, with increasing growth time, both the crystal size and the size spectrum increase. During the early stages of growth, crystals are small and relatively uniform-sized. As films grow, the average crystal size increases, and at the same time, the size distribution broadens. To address the incorporation of non-crystalline phases in polycrystalline diamond films and the morphological instabilities at high growth rates, Ravi1701conducted an experimental investigation of the combustion synthesis of diamond and proposed a model for the development of morphological instabilities in diamond films, as schematically depicted in Fig. 4. The morphological instabilities are found to be operative in the low pressure, high rate CVD of thick diamond films or slabs. The morphological instabilities are more pronounced for high growth rate techniques (such as plasma arc-jet and combustion synthesis) than for those of lower growth rates (such as HFCVD and MW PACVD). The instabilities do not appear to be pronounced for films of thickness below -20 pm. The films thinner than 20 pm are usually dense and uniform, an observation that is consistent with the finding of Bauer et a1.[1741The lack of surface diffusion and re-evaporation at diamond film surface during growth accelerates the stochastic interface perturbations, leading to the faster growth of some crystals relative to others. The morphological instabilities are then magnified during sustained growth as a result of competitive shadowing and nutrient starvation/overfeeding.

88

Diamond Chemical Vapor Deposition

t

nnnnl

f” ‘Idrn

Figure 4. Development of morphological instabilities in thick diamond films synthesized at high rates. (i, Early stages of growth are characterized by uniform, void-free, dense films. Morphological instabilities are not evident in thin films. (ii, The onset of morphological instabilities during sustained growth is accompanied by smaller crystals suffering from nutrient starvation due to the competitive shadowing. (iii, Taller crystals are closer to the high temperature nutrient source and hence grow at higher rates than shorter crystals due to nutrient overfeeding and higher temperatures. (iv) Presence of oxidizing species in growth ambient, along with higher surface temperatures of the taller crystals, leads to lateral growth, further increasing the competitive shadowing, while the lower surface temperatures of the shorter crystals (as a result of the competitive shadowing), along with the nutrient starvation, promote the formation of non-diamond phases (such as DLC) around them.t701 (Reproduced with pemission.)

In addition to the statistical nature of the interface instabilities active in diamond CVD, the orientation effect and anisotropic growth of crystals (i.e., evolutionary selection) play an important role inthe observed instability phenomenon. Surface chemical reactions that occur preferentially between the growing diamond surface and oxidizing species in the combustion synthesis ambient also influence the development of the microstructure and morphology of crystals in diamond films. For example, in combustion CVD, the growth of { lOO} planes is promoted by the presence of oxidizing species in the flame. The morphological instabilities not only cause extreme variations in the sizes of individual crystallites in polycrystalline films, but also lead to the

Diamond Epitaxy, Oriented Growth, and Morphology Evolution

89

incorporation of voids and non-crystalline phases in the films. As growth proceeds, small crystallites are found to be coated with DLC, a phenomenon attributable to the nutrient starvation and the low surface temperatures of small crystallites caused by the competitive shadowing. In order to increase the uniformity and density of diamond films and reduce the phase impurity in the films synthesized at high rates, appropriate processes need to be developed if the advantages offered by the high growth rates are not to be completely offset. IG~vi[~~l suggested two approaches to overcome the morphological instabilities. The first approach is to increase the nucleation density of diamond on substrates, so that the onset of the instabilities can be delayed. The second approach involves the periodic interruption of a growth process and the renucleation of diamond through depositing very thin DLC films on the initially grown diamond layer, followed by recommencing growth. This latter approach can minimize or eliminate the morphological instabilities, and makes it possible to grow thick, dense, void- and flaw-free diamond films with very narrow grain size distributions. When diamond films are deposited on non-diamond substrates, stresses may be generated in the films due to lattice mismatch and/or differences in thermal expansion coefficients between diamond and the substrate materials.i7] In addition, lateral variations in the grain size, density, or impurities incorporated during growth may also lead to stresses, which may be either tensile or compressive. The stresses are known to generally build up with increasing film thickness, and will influence diamond-substrate adhesion and properties of diamond fihns.t7]

4.0

SUMMARY

The primary difficulty in diamond epitaxy is the small number of materials (Ni, Cu, Fe, Co, Si, and cBN) with suitable crystal structure and lattice constants. The extremely high surface energies of diamond and the existence of inter-facialmisfit and strain energies between diamond films and non-diamond substrates constitute the primary obstacles in forming oriented 2-D diamond nuclei and single-crystal growth. The high solubility/mobility of C in/on the metals (Fe, Co, Ni), or the formation of an intermediate layer (carbides or graphite) may inhibit the possible development of an orientational, epitaxial relationship between diamond films and the substrates.

90

Diamond Chemical Vapor Deposition

Diamond homoepitaxy and heteroepitaxy have been achieved on diamond and cBN, respectively. Epitaxially textured and fiber textured, highly oriented diamond films have been grown on Si( 100). The feasibility of diamond heteroepitaxy on Ni, Ti, Co and Cu has also been confirmed by several experiments. Diamond can grow on Fe, but not yet by epitaxy. According to the principle of evolutionary selection, the development of fiber textures is a result of the growth competition between randomly oriented diamond crystals, and the resultant direction of fiber axis coincides with the direction of fastest growth. With increasing film thickness, both crystal size and size spectrum increase, and the degree of texture improves with concomitant reduction of overall n&orientation. The microstructure (texture, orientation) and morphology of diamond films can be controlled by varying the growth parameter o1(a = (v~~~/v~ 1J $), which depends primarily on gas composition and substrate temperature. For fiber-textured films, at low CH, concentrations and increasing substrate temperatures (cc < 1.5), the films exhibit pronounced (110) texture; at medium CH, concentrations and substrate temperatures (1.5 I cc I 3), a transition of the fiber axis from ( 110) to ( 100) occurs; a further increase in CH, concentrations or decrease in substrate temperatures (a > 3) leads to fine-grained, non-faceted films. For homoepitaxial or heteroepitaxially textured films of { 100} orientation, the microstructure and morphology of the films are strongly affected by twin formation, which may lead to complete deterioration of the epitaxial orientation. Large values of cc (CX= 2.5) result in suppression of twin formation and competitive growth which favors epitaxially oriented crystals, while small values of ar.(a x 1S) lead to deterioration in the film orientation and morphology due to twinning. The addition of oxygen-containing gas species or nitrogen to the gas mixture catalyzes and stabilizes (100) texture. Films thinner than 20 urn are usually dense and uniform. Morphological instabilities occur in high rate CVD of thick diamond films or slabs (>20 urn, typically) due to the lack of surface diffusion and re-evaporation at the diamond film surface. The morphological instabilities are magnified during sustained growth as a result of competitive shadowing, nutrient starvation/ overfeeding, orientation effect and anisotropic growth of crystals as well as surface chemical reactions between the growing diamond surface and oxidizing species. The morphological instabilities cause not only extreme variations in crystal size, but also incorporation of voids and non-crystalline phases in diamond films. Two approaches have been suggested to overcome

Diamond Epitaxy, Oriented Growth, and Morphology

Evolution

91

the morphological instabilities: (a) increasing surface nucleation density and/ or (6) periodic interruption of a growth process each 20 pm for renulceation on the diamond surface through depositing very thin DLC films on it, followed by recommencing growth. A detailed understanding of the mechanisms governing the development of crystal orientation, film texture and morphology is still desirable for the growth of high quality, highly oriented diamond films, and ultimately large-area single crystal diamond films, if possible.

Effects of Surface Conditions on Diamond Nucleation

In an effort to enhance diamond nucleation and to control film morphology, extensive work on the nucleation and early growth stages has been performed. As a result, technology problems associated with the nucleation ofpolycrystalline diamond films have been adequately addressed. A number of nucleation enhancement methods have been developed that enable the control of nucleation density over several orders of magnitude. Nucleationdensityhas beenincreasedfrom<105cm-*onuntreatedsubstrates up to 10” cm-* on scratched or biased substrates. The effects of surface conditions on nucleation processes have been investigated to provide the guideline for the selection of optimum surface pretreatment methods. In this chapter, substrate materials, surface pretreahnent methods and their influences on diamond nucleation are discussed.

1.0

SUBSTRATE

MATERIALS

The chemical properties and surface conditions of substrate materials critically influence surface nucleation processes of diamond in 92

Effects of Surface Conditions on Diamond Nucleation

93

e main reason for the existence of the incubation period frequently observed in diamond nucleation is the chemical interactions of the gas phase with the substrate surface.[52a] Lux and Haubnerts2~ classified substrate materials into three major groups in terms of carbon-substrate interactions, as listed in Table 1. cq).[52a][179][181][270][277][297]-[299]

fi

Table 1. Classification of Substrate Materials Carbon-substrate interactions Little or no solubility or reaction C-diffusion only, C dissolves in MeC mixed crystals

Substrate materials Diamond, graphite, carbons, Cu. A& Au. Sn, Pb. etc. Pt. Pd. Rh, etc.

Carbide formation metallic covalent ionic

Ti,Zr,Hf,V,Nb,Ta,Cr.Mo,W,Fe.Co.Ni(metastable) B. Si, etc. AI,Y, rare earth metals, etc.

Although at present a complete picture is not yet available regarding which substrate materials favor diamond nucleation most, the following trends may be summarized from available experimental results and theoretical speculations:

Although at present a complete picture is not yet available regarding which substrate materials favor diamond nucleation most, the following trends may be summarized from available experimental results and theoretical speculations: 1. Diamond surbaces or particles appear to provide the best nucleation p~~~~~~~~~~~~l~~~~l~~~~l~~~~l~~~~l 2. Nucleation on cBN occurs readily.[23g]-[2411[300] 3. Nucleation rates on stable carbide-forming substrates

(Si, MO,W) are much higher than on substrates that do not form carbides (Cu, Au).[~~] 4. Among carbide-forming substrates (Si, MO,Al, Ni, and

Ti), MO substrates give the highest nucleation density under the same deposition conditions, and the nucleation densities on all other substrates are approximately one order of magnitude lower.t161] The nucleation of diamond on these substrates is controlled by neither the thermal conductivity of the substrates nor the interface misfit between diamond and the substrates, but rather the ease ofthe formation of carbide as a buffer layer.t161]

94

Diamond Chemical Vapor Deposition 5. Refractory metal carbides (TaC, WC, MO&) and some covalent carbides (Sic, B,C) have a positive effect on diamond nucleation, while effects of ionic carbides (Al&, liquid salts, etc.) are still less lcr~owr-r.[~~~] 6. Nucleation occurs readily on substrates forming amorphous DLC (mostly MOw41wi and SiW81) without my polishing pretreatment.[2681 7. Graphite deposits form mostly on Ni,[256]Ft,[161[256] Cu,[1871[1991[2851 Fe, and SiL212] prior to diamond nucleation when the substrate surface becomes supersaturated with carbon,[2121and may enhance diamond nucleation W’lP851 8.

adhesion of diamond films on carbide-forming substrates (Si) is better than on non-carbide-forming substrates (Cu, Au), particularly those with little or no solubility of carbon.[17gl

The

In addition to the beneficial effect of promoting surface nucleation and adhesion, certain substrates with high carbon solubility or difisivity delay diamond nucleation by acting as carbon smk.*52~[52b]At high deposition temperatures, large amounts of carbon are transported into the substrates to form either carbides or solid solutions, leading to a substantial decrease in surface carbon concentration and hence a considerable delay in the onset of nucleation. Thin films reach carbon saturation more rapidly than bulk materials. Hence, carburization of substrates and control of diffusion layer thickness may be effective means for utilizing the beneficial effect and avoiding the detrimental effect ofthe substrate materials which form carbides or possess high C solubility. To this end, some surface pretreatment methods, such as precarburization and thin metal film coating of substrates, have been developed, as will be discussed below. 2.0

SURFACE PRETREATMENT METHODS AND NUCLEATION ENHANCEMENT MECHANISMS

Diamond nucleation on non-diamond surfaces without pretreatment is usually too slow to obtain continuous films within a reasonable tune. A variety of surface pretreatment methods have been developed to enhance diamond nucleation rate and density on various substrates, including:

Effects of Surface Conditions on Diamond Nucleation

95

* Scratching substrate surfaces with ab~ives[73~[170~[1n~-[17g] *

Seeding

substrates

with diamond

grit and other

powders[l66111801[3011 -

Electrical biasing substrates1178111841-11g11

* Covering substrate surfaces with graphite fibers,1280112811 clusters,11511and fihns[252112781 -

Coating substrate surfaces with thin films ofmetals,1302113031 C,, [1921[1931 a-C, [1751[2g7] Y-Zr02,t1651 cBN, Sic and & , [254][274][297][303] and hydrocarbon oi1t3041

*

Ion implantation of substrate surfaces11991

*

Pulsed laser irradiation of substrate sur-faces[2g7]

*

Carburization of substrate surfaces1711[2781[302]

-

Chemical etching of substrate surfaces13051

These nucleation enhancement methods make it possible to control diamond nucleation density over several orders of magnitude. The nucleation density has been increased from
96

Diamond Chemical Vapor Deposition

2.1

Scratching

The most often used method for enhancing diamond nucleation is scratching, abrading or blasting non-diamond surfaces with diamond particles or paste, although the method is acceptable only for limited applications (not for optical, for instance). Other abrasives include borides, carbides, nitrides, silicides, oxides (BN, Sic, A&O,, MOB,LaB,, TaB,), and graphite, etc [170][174][175][177][315] D’ uunond nucleation densities after scratching preas corn_ treatments range typically from lo6 to 1Orocm-2, ~~~1~~~~1~~~~1~~~~1~~~~1 pared to
Effects of Surface Conditions on Diamond Nucleation

97

form the precursor molecules. Finally, the removal of surface oxides is also suggested to be a possible operating mechanism of the nucleation enhancement by scratching.[3151[317]

Figure 1. SEh4 of diamond crystals nucleated on a scratched Si substrate. Nucleation occurs more easily on scratched lines and other surface irregularities.[167] (Reproduced with permission.)

98

Diamond Chemical Vapor Deposition

Iracturbd dlemond particle trOm pOllrhinO @I meed

TI

V

Zr Hf

Nb Ta

Cr MO W

Figure 2. Schematic diagram of mechanisms for diamond nucleation enhancement on scratched substrates.[52] (Reproduced with permission.)

DeAg

et al+[2981[2991employed

topography patterning techniques to investigate the influence of substrate topography and pretreatment on diamond nucleation. Their results reveal that, although residual diamond abrasive powder may enhance nucleation, the nucleation event is promoted by topographical features alone, and the presence of some residual abrasive is not necessary to initiate nucleation. The experimental results of Polini[31gl partially support these findings and confirm that the seeding effect alone cannot explain the increased nucleation density on substrates scratched with diamond paste. It is further indicated thatthe presence of edges is a necessary but not a sufficient condition for diamond nucleation. Some specific atomic arrangements at the edges are required for diamond nuclei to form, although these arrangements are unknown at present. On the other hand, the ineffectiveness of Al,O, powder in enhancing diamond nucleation density on Cu and Si substrates[2g7]or on MO substrates[156Isuggests that the strain centers produced by polishing or abrading do not play a significant role.[2g71 Similar results have also been reported by Bauer et a1.,I174lwho found that pretreating Si substrates with graphite, Sic, A&O,, and polymethyl methacrylate had no effect on diamond nucleation density. Therefore, the role of the scratching pretreatment in enhancing diamond nucleation may be attributed to different mechanisms, and more than one mechanism may operate, depending on pretreatment processes, abrasive and substrate materials, deposition methods and conditions. Regarding the effect of various abrasives (Table 2) on diamond nucleation, it is found that polishing with diamond has the most pronounced effect, and the effect decreases in the order diamond > cBN > SiC.l244l For

Effects of Surface Conditions on Diamond Nucleation

99

example, in HFCVD, [244]nucleation was observed within 5 ruin on diamondpolished substrates; however, no evidence of nucleation could be detected even after 30 min on substrates polished with other abrasives (cBN, Sic, TaC, WC, Al,O,, S&N,, etc.). Maeda et a1.t177] used the ultrasonic method with various ceramic particles (borides, carbides, nitrides and silicides) suspended in acetone to abrade a p-type Si(100) wafer, and subsequently investigated morphology and nucleation densities during the early stages of diamond growth in MW PACVD. The abrasion treatment is believed to implant fine abrasive particles on the substrate surface and these implanted particles then act as nucleation sites for diamond. The effect ofthe abrasives on nucleation enhancement increases in the order silicides < nitrides < carbides < borides, with the resultant nucleation densities ranging from lo4 to 4 x lo7 cmw2.Among the abrasives, MOB, LaB,, and TaB, particles show the best effect on increasing diamond nucleation density, nearly equivalent to the effect of cBN particles. Because of the quite different sizes and shapes ofthese particles (Table 2), the equivalent effect appears to demonstrate that the efficacy in enhancing diamond nucleation would be dependent upon the chemical properties of the implanted ceramic compounds, rather than the surface characteristics produced by the abrasion processes. Therefore, Maeda et al. concluded that the nucleation enhancement by abrading with MOB, LaB,, TaB,, or cBN is due to the presence of boron atoms and the stability of these seed ceramics in the microwave plasma. The active participation of boron compounds in surface nucleation processes is similar to that in the gas-phase nucleation processes.[220] Regarding the effect of abrasive particle size, Ohtake et a1.[321] reported that, in the arc discharge plasma CVD, the nucleation density on a diamond-lapped MO surface attained a maximum when 1 pm diamond paste was used, approximately 25 times larger than that on an untreated surface. The nucleation densities on the surfaces lapped with 15 and 6 pm diamond paste were only slightly larger than that on the untreated surface. Zhang et a1.t’70]observed that, in the oxyacetylene combustion synthesis on Ti-2Al1SMn alloy substrates, nucleation density tended to increase with decreasing size of Sic grit used for scratching. However, there exist also some contrary arguments and observations, i.e., scratching with coarse particles increases nucleation density more effectively due to the increased roughness.[52a] Polishing with 1 pm diamond grit did enhance diamond nucleation on Nb, Ta, and W substrates, but not on MOsubstrates.t528]On MO substrates, polishing/scratching was effective only with 7 pm diamond grit. This behavior is believed to be caused by the dissolution of the very small diamond

100

Diamond Chemical Vapor Deposition

particles into the MOsubstrates prior to diamond nucleation, since the diffusion rate of carbon in MO is higher than in Nb, Ta, and W, and the high substrate temperatures favor the dissolution.[5&IA similartendency was also observed on nucleation densities of 3.4 x lo’, Si substrates. In HFCVD experiments, t3161 2 x lo’, 5.4 x 106,and 7.7 x lo5 m2 were obtained on Si substrates lapped with 6,1 and 0.25 pm diamond grit, as well as on untreated surfaces, respectively. Table 2. Size and Physical Properties of Diamond and Various Ceramic Compounds Used for Scratching Pretreatment Abrasive material Diamond Oxides A1203 ZtQ2 SiO2 Borides [ 1771 TiB2 CTB aB2 NbB2 MOB m6 TaBZ WB Carbides [ 1771 B4C SiC Tic VaC7 Cr3C2 Ztc NbC Mo2C TaC WC Nitrides [ 1771 BN AIN Si3N4 TiN VN Cr2N ZrN NbN TaN SiKcides [ 1771 TiSi2 CrSi2 ZrSi2 NbSi2 MoSip TaSi2 WSi2

Patti& size (run) 0.25 - 40 (ultrasonic)[317.320] 0.25 - 15 1316,317,321]

Density (kg nr3) 3515 [2]

HardocSS (kg mnr2) 5700 - 10400 [2]

CIyStal stmchlre Cubic/ hexagonal [2]

0.3 - 1 [52,164,174] 0.1 - 0.3 [305] [305]

3970 [I971 5560 [ 1971 2320 [198]

2OcG [197] 1019 [I971 790 [ 1971

Hexagonal [ 1971 Cubic [ 1971 Hexaeonal~202~

2- 10 5 - 20 5- 15 0.5 - 1 l-5 10 - 40 0.5 - 1 l-5

4530 6110 6090 7000 8670 4720 12620 15730

3370 1250 2252 2600 2350 2770 2500 3700

Hexagonal Orthorhombic Hexagonal Hexagonal Tetragonal Cubic Hexagonal Temagonal

10 - 20 5 - 20 10 - 30 20 - 40 2- 10 2- 10 10 - 30 0.5 - 3 10-20 20-30

2510 3220 4920 5480 6740 6660 7820 9180 14400 15770

2750 2550 3170 2480 1800 2950 2170 1499 1720 1716

Rhombohedml Hexagonal Cubic Cubic Ortborhombic

8- 12 I- 10 0.2 - 1 3- 15 l-5 0.5 - 3 3 - 15 5 - 30 l-5

3480 3260 3180 5440 6100 6510 7350 8310 14360

4530 1200 2100 2050 1310 1571 1670 1461 2416

Cubic Hexagonal Hexagonal Cubic Cubic Hexagonal Cubic Cubic Hexagonal

10-30 10 - 50 10-30 IO-40 10-40 IO-60 5- 15

4040 4980 4860 5660 6240 9100 9860

892 1131 1063 1050 1200 1407 1074

Orthorhombic Hexagonal Onhorhombic Hexagonal Tetragonal Hexagonal Teuagonal

Cubic

Cubic Hexagonal Cubic Hexagonal

Effects of Surface Conditions on Diamond Nucleation

101

Ascarelli and Fontanat31~ correlated the nucleation densities of diamond on Si in HFCVD to substrate pretreatment processes and diamond abrasive particle (grit) sizes, as summarized in Figs. 3a and b. The results in these figures clearly show dissimilar grit-size dependence of diamond nucleation densities on substrate surface pretreatments. From Figs. 3a and b, it is evident that diamond nucleation densities after polishing and ultrasonic pretreatments range from 10’ to lop and from 10’ to 10” cm2, respectively. Diamond nucleation densities decrease with increasing mean particle size of diamond abrasive paste used in the polishing pretreatment. Conversely, diamond nucleation densities increase with increasing mean size of diamond particles used in the ultrasonic pretreatment. Further, it was observed that, after the polishing pretreatment, diamond nucleated at a higher density along scratches and at a lower density away from the scratches, while the ultrasonic pretreatment with diamond particle suspensions generated a more homogeneous morphology over the entire deposition surface. 10’0

10' 0.1

Figure 3. Dissimilar grit-size dependence of diamond nucleation densities on substrate surface pretreatments. (a) Nucleation density versus inverse abrasive paste mean-size used in polishing pretreatment.t3r7) (a) Nucleation density versus single-particle mean-size used in ultrasounding pretreatment; solid squares: Ref. 317, blank squares: Ref. 322, solid circles: Ref. 323. (Reproduced with permission.)

102

Diamond Chemical Vapor Deposition

To explain the dissimilar grit-size dependence of diamond nucleation density, Ascarelli and Fontana suggested that, for brittle materials, the scratching pretreatment results in breaking of a certain number of surface bonds, a fraction of which remains organized in bonding configurations that would act as suitable “sites” for diamond nucleation. If the load and time of the scratching pretreatment are kept constant, this fraction of the broken bonds is in proportion to the energy Edissdissipated during the scratching pretreatment in a volume given by a depth dg (i.e., the mean particle size of the diamond abrasive paste) and a surface area S,. Accordingly, diamond nucleation density is directly proportional to the density of the “sites” and inversely proportional to the condensation energy Econdper unit volume: Ediss

const .

1

Nd CC--‘XS&g Econd

dg

Hence, diamond nucleation density is inversely proportional to the mean particle size of the diamond abrasive paste used in the scratching pretreatment. In the ultrasonic pretreatment, the fraction of the broken bonds is in direct proportion to the collision energy of the particles suspended in the solution. A particle generates a deformation depth L on a contact surface SP on a substrate. The collision energy may be correlated to the mass of a particle mP and the speed of sound v,. The total number of the collisions is assumed to be roughly independent of the particle concentration due to the high sound frequency ( lo4 Hz) and the relatively short pretreatment duration (-30 min). If the particle concentration in the solution is kept constant, diamond nucleation density can then be correlated to the mean size of diamond particles suspended in the solution such as ‘hrn&

1

d3 &ocdg

spL

Econd

d;

Nd ~--

i.e., diamond nucleation density is proportional to the mean size of diamond particles in the ultrasonic pretreatment. The similar trend has also been reported in Refs. 322 and 323.

Effects of Surface Conditions on Diamond Nucleation

103

The size and shape of scratches affect diamond nucleation by altering the migration of carbon atoms on the substrate surface, the formation of critical nuclei, and the re-evaporation of the nuclei. Yugo et a1.l32ol correlated the optimum size of scratches produced by the ultrasonic pretreatment to the nucleation density and the critical nucleus size of diamond on Si( 100) substrates in MW PACVD. It was found that scratches less than 5 mn in depth are ineffective for diamond nucleation, whereas the smooth, Vshaped scratches of 5-15 nm (resulted from the ultrasonic pretreatment with diamond powder of 20-40 urn) act as effective nucleation centers at which carbon atoms can aggregate rapidly to form critical nuclei against the reevaporation. The small scratches may pin the arriving carbon atoms and inhibit their migration, making difficult the aggregation of a sufficient number of carbon atoms and the formation of critical nuclei within the average residence time of carbon atoms. It was estimated that the optimum size of scratches necessary for generating diamond nuclei is 30 nm, and the diameter of the critical nucleus is 3 nm. Correspondingly, the possible maximum nucleation density of diamond would be - 10’1cm‘*. However, the optimum size of scratches depends on growth conditions, especially the saturation degree of carbon atoms in the gas phase. This size will decrease as the deposition rate of carbon atoms increases or as the substrate temperature decreases. In addition, etching the substrates in both a hydrogen plasma and an acid solution decreased the size of scratches, leading to a decrease in diamond nucleation density. Clearly, the optimum size of abrasive particles depends on pretreatment processes, deposition methods, growth conditions, and substrate materials. Moreover, the efficacy of scratching also depends on substrate materials, descending in the order Si > MO > WC.l54l 2.2

Seeding

Substrate surface seeding with submicron diamond powder has been employed to enhance surlhce nucleation density and rate of diamond.13241-13261 The diamond seeds littered on the surface serve as nuclei for immediate growth. Other non-diamond powders used for seeding include Si, cBN, Al203 , TaC , and Sic , [1811[3241 and graphite flakes. [161[1**1[1*31 Several

104 Diamond Chemical Vapor Deposition seeding techniques have been attempted, including dipping, spinning, and spraying.t324] Uniform seeding densities of lo8 to log cmm2submicron diamond particles on Si wafer were obtained through electrophoretic seeding pretreatment.t32fl This technique has been demonstrated to be simpler than alternative methods and far less damaging to substrates.[308] In Smolin et al.‘~t~~~]experiments using DC arc discharge, diamond nucleation density was increased to 2 x log cmm2by seeding MO substrates with ultrafine diamond grit (-20 run), and smooth thin films were produced from CH,-H, mixtures. Seeding also offers the possibility for epitaxial or oriented growth of diamond films. tr‘WW81 WJI ~ighl y oriented diamond films have been grownt32g] in MW PACVD and HFCVD by seeding Si substrates with diamond crystals such that most of the crystals had a {11 l} orientation. Large-area mosaic diamond films approaching single-crystal quality have also been obtained by a seeding technique. t328]In this process, pyramidal pits or sawtooth-profile gratings were made by etching (lOO)-oriented Si wafers. Diamond seeds of 75 to 100 urn diameter were deposited onto the substrates in a slurry. The diamond films were then grown in HFCVD, PACVD and flame CVD, respectively. As discussed previously, a graphite interlayer generally forms immediately when a Ni substrate is exposed to a methane-hydrogen CVD environment.[256] This interlayer excludes the possible development of an orientational relationship between a diamond film and a Ni substrate, even though diamond eventually nucleates and grows on the graphite interlayer. However, through seeding and using a multistep HFCVD process, Yang et al.t1a6]have deposited well-crystallized and highly oriented diamond on Ni substrates without initial graphite formation. The oriented nucleation and growth of diamond and the suppression of graphite formation on Ni substrates are speculated to be achieved through the dissolution of seeded diamond fragments into Ni lattice and the subsequent alignment of the partially dissolved diamond with the orientation of the Ni substrates. Diamond nucleation density has been found to be linearly proportional to diamond seeding site density; it is approximately a tenth of the seeding site density,t301]as shown in Fig. 4.

Effects of Surface Conditions on Diamond Nucleation

105

Residual Diamond Dust Density, RDD [ #/cm21

Figure 4. Dependence of diamond nucleation site density (NSD) on residual diamond particle density (RDD) under various deposition conditions. The dotted line shows NSD equal to RDD and the solid line shows NSD being 10% of RDD. Solid circles: ultrasonic polishing, followed by MW PACVD; crosses: hand polishing, followed by hIW PACVD; b&-&es: ultrasonic polishing, followed by HFCVD (filament temperature = 2973 K); squares: ultrasonic polishing, followed by HFCVD (filament temperature = 2773 K); open circles: ultrasonic polishing, followed by HFCVD (filament temperature = 2573 K); dotted circles: fluidized-diamond polishing, followed by Mw PACVD,t30’] (Reproduced with permission.)

The grain boundary and grain size of substrate materials also influence diamond nucleation density. t1701t3301 It was observed that the effect of the grain boundary on diamond nucleation is more important than that of the crystal lattice of substrate materials. ~~‘1This effect is related to the intrinsic characteristics of the grain boundary. Nucleation on the grain boundary results in a minimization of the interface energy. The presence of more vacancies, dislocations and dangling bonds at the grain boundary helps in the chemisorption of nucleation species. Thus, diamond nucleation density increases with decreasing grain size or increasing number ofgrain boundaries of substrate materials. In MW PACVD, diamond nucleation on a WC substrate tended to occur selectively at the edges of WC grains.l33ol For coarse WC grains (-1 pm), diamond nucleation density was -9 x lo6 cmSz. It increased to -5 x lo7 cm2

106

Diamond Chemical Vapor Deposition

when using a finer-grained substrate (-0.5 pm). A considerable enhancement of diamond nucleation density (up to 5 x 10’ cmv2)was achieved through introducing a number of fine microflaws onto the substrate surface. In the microflawing treatment, fine diamond powder (co.25 pm) was suspended in an ultrasonic cleaner bath. The size of diamond crystals decreased with increasing microflawing time. 2.3

Biasing

Scratching and seeding cause surface damage and contamination, and cannot be easily applied to substrates of complex geometry and shape. Therefore, these pretreatment methods are incompatible with many applications that require extremely smooth, clean surfaces, such as diamond films for electronic devices, optical window materials and smooth wear-resistant coatings. Alternative pretreatment methods that can yield high diamond nucleation densities without damaging substrate surfaces are therefore of particular importance. Recently, biasing pretreatment of substrates has been increasingly employed to enhance surface nucleation of diamond [8][106][178][186]-[190][331]-[333] u sing a positive or negative bias to obtain large nucleation densities on unscratched substrates provides an opportunity to control nucleation densities by varying the applied voltage and current, while at the same time reducing surface damage. Suzuki et a1.t35]reported the same nucleation density (10’ cmW2 after 30 min) in DC PACVD on a mirror-polished Si wafer as on diamond-scratched Si substrates. Kobayashi et a1.t260b] investigated the infhrence of a positive bias of 0 to 140 V applied to substrates. For voltages of 60 to 140 V, the substrates ultrasonically-scratched with diamond powder were completely covered by diamond. However, this was not the case when biasing was not used. Diamond films of good crystalline quality were obtained with a concomitant increase in nucleation and growth rates when the gas pressures ranging from 30 to 50 torr and the voltages from 100 to 140 V were applied. These were found to be the optimum growth conditions for the system used. Yugo et a1.[333] employed a negative bias and high CH, concentrations during pretreatment to generate diamond nuclei on a Si mirror surface in PACVD. The several-minute pretreatment resulted in an enormous nucleation enhancement. Diamond nucleation densities as high as 10” cmm2were achieved. For the onset of diamond nucleation, a minimum voltage of -70 V and a minimum concentration of 5 vol.% CH, in H2 were necessary.

Effects of Surface Conditions on Diamond Nucleation

107

Recently, Katoh et a1.t190]presented thorough experimental results for diamond nucleation on mirror-polished Si( 111) wafers in MW PACVD via both positive and negative biasing. The positive or negative biasing pretreatment was conducted for 30 min at substrate bias voltages ranging from -100 to +lOO V, MW power of 300 W, a substrate temperature of 900°C, gas pressures from 0.2 to 15 torr, CH,/H, from 2 to 40 vol.% and a gas flow rate of 100 seem. The pretreatment was followed by the conventional h4W PACVD growth for 3 h at a substrate temperature of 950°C, 0.5 vol.% CH,/H,, agas pressure of 35 torr and a gas flow rate of 200 seem. By varying the negative or positive substrate bias voltages, diamond nucleation density may be changed over six orders of magnitude, as shown in Fig. 5.

1. log 10' a 0 % e

+-

10s

0.2TorrcH42+

-t_

2TonCnb2X

+

l!firOKCH4 2%

-+

0.2TonCH4 10%

%Id

-f-

PTorCH4lMC

*

1sT0nCH410%

-100 -50 T tw+loo SubstrateBias Voltage I V

m

3 log 10'

0C

i! B 2

105 ld-

-100 -50

0

-o_

0.2TonCH44o%

-t_

2TomCH44O9b

-o_

1!monCH440%

40 tloo

Substrate Bias Voltage I V

Figure 5. Diamond particle density as a function of substrate bias voltage. During bias pretreatment, the gas pressure was 0.2,2, and 15 torr, and CH, was (a) 2 vol.%, (b) 10 vol.%; and (c) 40 VOI.~~.[‘~~](Reproduced with permission.)

108

Diamond Chemical Vapor Deposition

It is evident from this figure that: 1. both positive and negative biasing are, effective for enhancing diamond nucleation density; 2. lower pressures, higher CH4 concentrations, and/or larger absolute values of substrate bias voltages lead to higher nucleation densities; 3. at the same absolute value of substrate bias voltage and for CH,concentrations ranging from 10 to 40 vol.%, the nucleation densities on the negatively biased substrates are one to two orders of magnitude higher than those on the positively biased substrates. For example, nucleation densities of lo’, lo*, and log cmm2were obtained for bias voltages of 0, +20 and -20 V, respectively, under the same deposition condition. The attraction of cations in negative biasing led to rough Si surfaces, whereas the positively biased substrates maintained smooth surfaces. It was then proposed that, for Si substrates, the positive biasing is a more suitable pretreatment condition than the negative biasing.ligOl Mechanical properties of diamond films are greatly affected by both the grain size and the non-diamond carbon content incorporated in the films, which in turn can be affected by biasing current. In the I-IFCVD experiments conducted by Baba and Aikawa, 1’1relatively large compressive stresses, -100 MPa, were observed in the diamond films grown at DC bias currents less than 500 mA. Stresses turned to be tensile when the bias current was increased to 700 mA. With increasing bias current (from 0 to 700 mA), both the grain size (from 0.6 to 0.3 pm) and the non-diamond carbon incorporation in the diamond films decreased, while Young’s modulus increased from -240 to 860 GPa with a concomitant increase in fracture strength. Young’s modulus reached a maximum of 860 GPa, which is close to the value of single-crystal diamond (Table 1, Ch. 1). In an in-depth study of diamond nucleation on Si in MW PACVD,l”*l substrates were pretreated by negative biasing in a 2 vol.%CH,-H2 plasma. The biasing enhanced diamond nucleation density on unscratched Si wafers, up to 10” cmm2,as compared to 10’ cmm2 on scratched Si wafers and 103- 1O5 cm-2127g* on untreated Si wafers (Fig. 6).

Effects of Surface Conditions on Diamond Nucleation

0.5

1.0 Bias Time (h)

109

1.5

(4

10” Legend*

___--

I? 6 0 . !z 2 a

+Silicon +Hafnium 10’

+Titanium *Tantalum

a” 4

+Niobium -Tungsten

10’

+-Copper

:: G d lo5 0

20

40

60

80

100

Bias Time (minutes)

Figure 0. Diamond nucleation density as a function of bias time: (a) on a Si substrate,[‘78] (b) on different metal substrates.[275] (Reproduced with permission.)

110

Diamond Chemical Vapor Deposition

The results shown in Fig. 6 suggest that diamond nucleation density may be varied over seven orders of magnitude by controlling the duration of the biasing pretreatment. In addition, the SEM observations show that, once the bias was turned off, nucleation no longer continued and diamond grew on the existing nuclei. Ifthe bias remained on throughout the deposition process, the resulting film was of much poorer quality than if the bias was turned off, suggesting that the conditions which are favorable for diamond nucleation are not necessarily ideal for diamond growth. The efficacy of biasing in enhancing diamond nucleation not only depends on the applied voltage, current, and duration, as well as the deposition conditions (gas composition and pressure), as discussed above, but also appears to depend on substrate materials to some extent. In an invacua surface study of diamond nucleation on Cu and Si in MW PACVD,t’*‘l a negative substrate bias was employed. The biasing pretreatment step proved to have a tremendous influence on the nucleation density on Si( loo),but the nucleation density on polycrystalline Cu was only slightly increased. The Si( 100) substrates were characterized by the formation of a carbide with some form of non-diamond carbon present on the surface throughout the biasing, whereas the Cu substrates were covered by a 10 A thick, highly graphitic layer that was quite stable in thickness after 15 min of the biasing. Another study of diamond nucleation on Cu, Ni, and Si substrates11581demonstrates a different tendency. Diamond was nucleated at low pressures and low temperatures in ECR MW PACVD. Microwave power of 1.5 kW was used for 30 min with 15 to 100 vol.% CH&I,, a substrate bias voltage of 0 to -60 V, a substrate temperature of 500°C, and a gas pressure of 0.1 torr. The experiments reveal that under the identical pretreatment conditions the resultant deposits on Cu, Ni, and Si substrates were different in both nucleation density and morphology. The nucleation density on Cu was higher than on Si. A completely contiguous film formed on Cu, and a non-contiguous film on Si, while no diamond nuclei formed on Ni. The morphology of diamond on Si was irregular, whereas most of diamond crystallites on Cu exhibited ‘a morphology with the { 11 l} planes parallel to the substrate surface. These differences were attributed to the different physical processes occurring on the surfaces during the pretreatment. Si was etched and roughened by the large amount of H ion species present in the plasma, while Cu was not etched to that extent. On Cu, non-diamond phases

Effects of Surface Conditions on Diamond Nucleation

111

(graphite or a-C) tended to deposit from the CH,-H, plasma at a negative substrate bias. As discussed previously, graphite can be well fitted to the diamond lattice depending on the orientation of planes, and a-C also easily takes the form of the sp3 bond. Ni did not create diamond nuclei because it effectively interacted with carbon through adsorption and dissolution. Stoner et al.t”*] proposed a model to explain diamond nucleation mechanism on Si during biasing pretreatment, as displayed in Fig. 7: 1. Before biasing begins, there may exist both adsorbed oxygen and amorphous carbon on the Si surface. 2. The adsorbed carbon is either etched away or converted to Sic, and the physisorbed oxygen is converted into a thin SiOz layer. 3. As biasing continues, the SiO, layer is completely removed while the carbide islands continue growing. Preferential etching of Si from the Sic and continued high flux of carbon to the surface create an excess concentration of carbon on the surface which is calculated to be -5 A thick at 5 min ,a;“dto increase to 10 A by 1 h. 4. When the local carbide islands reach a critical thickness (9 nm under the experimental conditions) such that continued carbide growth is unlikely, the excess carbon on the surface becomes free to form small clusters. Surface mobility of the carbon may be enhanced by the bombardment during the biasing. Some of these clusters become stable and form diamond nuclei. The SIC layer is calculated to reach a maximum of -9 nm by 1 h and then to decrease to -5 nm by 2 h. 5. As most of the carbide islands reach the critical thickness, more free carbon becomes available to form new diamond nuclei. 6. As biasing continues, there are ongoing adsorption of carbon and etching of the surface, with SIC etched preferentially relative to the more stable diamond nuclei. The preferential etching creates a rougher Sic surface.

112 Diamond Chemical Vapor Deposition Since Si is preferentially depleted from the carbide, carbon concentrations in those local regions are increased so that carbon clusters may actually form on thinner regions of the carbide, close to the Si substrate. The etching, cluster formation, and diamond nucleation continue until the surface is eventually covered with diamond nuclei.

(a) Before

Bias

Cd) 30min

(b)

5 min

Cc) 15 min

(4 1 hr

(f) 2 hrs

Figure 7. Model of diamond nucleation on Si substrate via biasing.[“*l permission.)

(Reproduced wifh

The mechanisms of the nucleation enhancement by biasing have been addressed in several studies,t1781t190] as depicted in Fig. 8. In negative biasing of a Si substrate (Fig. Sa), the role of the biasing is suggested to: 1. increase the flux of carbon-containing cations (C?, CH+, CH,+, CH,+, CL&,+)to the surface, expediting the local carbon saturation on the surface and leading to a thin layer of amorphous carbon on the Sic layer to form small clusters favorable for diamond nucleation; 2. transfer higher energy to the surface due to ion bombardment, resulting in an increased surface mobility of adsorbed species;

Effects of Surface Conditions on Diamond Nucleation

113

3. reduce and suppress oxide formation on the surface, and remove native oxides (SiOJ which impede diamond nucleation process;[30gl-[311] and 4. enhance reactions abovethe substrate surface as a result of the increased ion-neutral collisions and the higher energy within the sheath region, leading to much higher concentrations of dissociated hydrocarbons and atomic hydrogen near the substrate surface.

Figure 8. Schematic diagram ment on biased substrates. (a) toward the substrate surface. substrate surface and bombard (Reproduced with permission.)

showing the mechanisms of diamond nucleation enhanceNegative biasing: carbon-containing cations are accelerated (b) Positive biasing: electrons are accelerated toward the carbon-containing molecules adsorbed on the surface.t’gO]

114

Diamond Chemical Vapor Deposition

The nucleation enhancement on positively biased substrates is deemed to be due to the presence of a high electron density (Fig. Sb). The electron velocity (in terms of kinetic energy) is approximately one hundred times greater compared to the carbon atom velocity, and the electron impingement rate on the substrate is approximately ten times greater. Although the electron mass is approximately four orders of magnitude smaller than that of carbon, the effect of electrons on surface processes cannot be neglected due to their high velocities and impingement rates. Through the high-speed impingement, electrons play an important role in the decomposition of molecules physisorbed on the substrate surface, such that hydrogen is sequentially removed from hydrocarbon species, CH,+CH,+CH,+CH+C, as shown in Fig. 8b. The decomposed hydrocarbons contribute to diamond nucleation more effectively than those formed in the gas phase since they are already on the substrate surface. Hence, the electron bombardment of the adsorbates on the surface may enhance diamond nucleation. Finally, the process parameters used in biasing pretreatments are summarized in Table 3 for an overview. 2.4

Covering and Coating

Another alternative surface pretreatment method, which can enhance diamond nucleation and avert surface damage, is covering or coating substrate surfaces with overlayers. The overlayers may involve clusters ~1 or films,[2521~2781 thin films of metgraphite fibers, [2801[2811 BN, Sic, als W'*lPO31 C70,W’*1tW a-c, [17b’71 DLC,f3341 Y_ZrO,,[lW &

[254][274][297][303]

or hydrocarbon oil. t304] Covering or coating substraies with carbon clusters[151~[1g2~[1g3] has been reported to yield an effect on diamond nucleation equivalent to that of diamond seeding or substrate biasing, while the use of thin metal overlayers or other materials , t303]or fluorinated plasmas, [151]etc., does not produce as significant a degree of nucleation enhancement as that obtained by diamond seeding or substrate biasing.

Effects of Surface Conditions on Diamond Nucleation Table 3. Process Parameters Used in Biasing Pretreatments Parameters Bias voltage (V)

Bias current (mA) m

power (w)

Pressure (torr)

Gas flow rate (seem)

CHq (vol.%)

Substrate temperature CC)

Duration (min)

Substrate

Values 0 to -60 [ 158] -170 to - 200 [ 1861 -100 to 100 [190] -200 to -300 [192] 100 to 140 [260b] -70 to -200 [333] 0 - 700 [S] 300 - 400 [ 1861 1500 [ 1581 1000 [186] 300 [ 1901 400 [ 1921 350 [333] 150 [S] 0.1 [158] 20 [ 1861 0.2 - 15 [190] 15 [ 1921 30 - 50 [260b] 15 [333] 306 [ 1861 100, [ 1901 100 11921 1 [81 15 - 100 [ 1581 2 [186] 2 - 40 [190] 10 [ 1921 1 [260b] 5 - 40 [333] 850 [S] 500 [US] 800 [186] 900 11901 <600 [ 1921 800 [260b] 900 [333] 30 [ 1581 15 [186] 30 [ 1901 15 [192] 120 [26Ob] 2 - 15 [333] Cu, Ni, Si, in ECR plasma [ 1581 Si, in MW PACVD [ 1861 C70 on Si, in MW PACVD [ 190, 1921 Si, in HFCVD [260b] Si, in MW PACVD 13331

115

116 Diamond Chemical Vapor Deposition Pehrsson et a1.t2801[2811 proposed a model for diamond nucleation mechanism, in light of Rudder et al.‘st1511experimental observations of very rapid, dense diamond nucleation (>lOg cmm2)on unscratched Si overlaid with graphite clusters. It was suggested that nucleation occurs on the edges of etch pits and carbon-rich particles. Both the etch pits and the particles satisfy the two criteria for spontaneous nucleation, i.e., carbon saturation and the presence of high-energy sites on edges and steps. To examine the effects of graphite film coating and film thickness, Feng et a1.I252lconducted diamond nucleation experiments on Si substrates in MW PACVD. They found that the application of a hydrocarbon oil or the evaporation of a thin carbon film on a variety of polished substrates increased both the nucleation density and uniformity of diamond films subsequently grown. Nucleation densities on the order of lo6 cmm2were obtained on the substrates coated with a carbon film thinner than 1 pm. The SEM and Raman spectroscopy results suggested that the nucleation enhancement and good-quality cube-octahedral diamond crystals may be attributed to the physical and chemical effects associated with changes of both the Si surface and gas chemistry. The change on the surface involves the development of a porous, ultra-thin residual carbon film which provides sites for diamond nucleation; the change in the gas-phase chemistry involves the variation of the local carbon concentration above the surface. Reactions between the solid carbon in the overlayer and the activated gas-phase species may increase the activated carbon concentration in the plasma, particularly in the case of thick carbon films, thereby shifting the gas-phase condition from the diamond growth domain to the non-diamond carbon growth domain, according to the CVD phase diagram developed by Bachmann et a1.t561Hence, thick carbon films will exert a detrimental influence on diamond nucleation. A critical thickness of carbon films was found to be less than 1.7 pm.[2521 Coating substrates with a thin film of metals, such as Fe, Cu, Ti, Nb, MO, or Ni, has been used to enhance diamond nucleation.[3021[3031[3151 Diamond nucleation on FeSi, substrates in HFCVD was enhanced by more than an order of magnitude relative to that on a bare Si substrate, as reported by Godbole and Narayan. I335lThe FeSi, was formed by laser deposition of Fe on Si substrates, followed by alloying during thermal annealing at 700°C. To enhance diamond nucleation by covering/coating substrate surfaces with carbon overlayers, several conditions are necessary:[rg21

Effects of Surface Conditions on Diamond Nucleation

117

1. A structured, pure carbon source on the surface is necessary to act as a diamond seed, which must be more thermodynamically stable than any hydrocarbon molecules under the typical diamond CVD conditions. 2. The structured carbon source needs to be both air-stable to facilitate application, and durable enough to withstand the environment in diamond CVD. 3. The structured carbon source must have steps or ledges to serve as diamond nucleation sites. 4. A means of initiating diamond nucleation is necessary during the nucleation stages. On the basis ofthese considerations, Meihmas et a1.t1g21t1g31 developed a unique method to use thin solid films of C& and C7,,clusters as nucleating agent layers for the growth of diamond films on different substrate surfaces, including metals (MO,W), insulators (quartz), and semiconductors (Si). It was found that, compared to other forms of carbon, such as graphite, amorphous carbon, soot, etc., the nucleation density on a C,, thin film in a microwave plasma discharge after a critical activation step is equivalent to those obtained on diamond-polished surfaces, v&h a nucleation enhancement of nearly ten orders of magnitude higher than on an untreated Si surface. The nucleation on a C&,tihn is less favorable. The activation step necessary for promoting diamond nucleation on C,, consists of negatively biasing (-200 to -300 V) the C,, film in a low power (400 W), hydrogen/methane (lo-20 vol.“/o>microwave plasma at a pressure of 15 ton; flow rate of 100 seem, and substrate temperature below 600°C for 15 min before the diamond deposition is initiated. The experiments show that the nucleation density of diamond on C,, is not critically dependent on the C,, film thickness. A C,, layer of -100 nm is sufficient for the nucleation and growth of fine-grain polycrystalhne diamond films and may substitute the diamond polishing pretreatment. The favorable nucleation on a CL,0film results from its chemical stability (against hydrogen plasma etching) and geometry: (a) the relatively planar surface of C,, allows diamond nucleation without major lattice distortion; (b) a grouping of four fused hexagonal rings serve to “lock in” a diamond lattice on the surface of the C,,, cluster; (c) the hollow shell structure of C,Oprovides both an inner and outer surface for chemical reactions to take place; and (d) the cage structure of C,Oprovides a non-rigid surface which relaxes to relieve any stress generated by bond angle strain between the nucleated diamond

118 Diamond Chemical Vapor Deposition sheet and the CT0 cluster base. Accordingly, this method may be applied to the growth of diamond on a wide range of substrates, and has the potential as a one-step lithographic template for growing diamond on selected regions of substrates. Hence, the carbon clusters could be of general use in the optical, electronic, and protective coating applications of diamond thin films. Very high nucleation densities, averaging around 3 x 1O’Ocm-*, have been obtained on Si substrates scratched with 0.5 pm diamond paste, cleaned in acetone and deposited with an a-C film in MW PACVD.[175j The nucleation densities are several orders of magnitude higher than those achieved on Si scratched alone (2x lo6 cm-*) under the same deposition conditions. Coating substrates with BN or SIC is considered to be anew, practical method for generating diamond nuclei to promote diamond film growth on non-diamond substrates.[274j Using HFCVD, Hirata et a1.[254]succeeded in preparing polycrystalline diamond films on non-scratched crystalline Si by depositing a very thin microcrystalline Sic layer. In Shing et al.‘s experiments,[274] the initial nucleation of diamond was achieved in ECR plasmas at a gas pressure of 0.013 torr on a-BN and cSiC coated Si substrates at 600°C. Diamond films with well-faceted crystallites were grown in MW PACVD using high methane concentration and oxygen addition (2-15 vol.% CH4 and 2-10 vol.% 0, in H2) at 10 torr pressure and at low substrate temperatures (400-75OOC). Morrish and Pehrsson[304] compared the effects of scratching with submicron diamond grit to those of scratching plus coating with a low vapor pressure, high thermal stability hydrocarbon oil, or a lo-20 nm thick layer of evaporated carbon on diamond nucleation. All these pretreatments enhanced the nucleation density and uniformity, led to good quality, and reduced the incubation period, relative to untreated surfaces. On diamondscratched plus oil-coated MO substrates, 0.5 urn thick, continuous diamond films were obtained within 4 h, while on non-oil-coated substrates, only large, isolated diamond crystais were grown. The same effect was also observed on Si, Cu, and Ag substrates. Nucleation density and crystal size can therefore be manipulated either by the use of a coating, or a coating in conjunction with scratching, thereby permitting tailoring of film structure, morphology and properties. Since coating with oil or evaporated C is easier, more controllable and less damaging than scratching, the method is potentially important to the exploitation of CVD diamond. In the combustion synthesis of diamond crystals and films on MO substrates,[124j[282] by reducing the 02/C2H2 ratio in the gas mixture to -0.75

Effects of Surface Conditions on Diamond Nucleation

119

for less than 30 s, thin (cl00 mn) DLC layers initially formed on the substrates. Followed by increasing the O,/C,H, ratio to -0.9 (the condition that produces diamond), diamond nucleation density was increased by an order of magnitude and growth rate by -6O%, relative to those on abraded MO and Mo,C substrates where no DLC films were present. The morphology was typified by dendritic growth on the substrates with the DLC layers versus well-faceted cube-octahedrons on the substrates without the DLC layers. Nucleation enhancement through the formation of the DLC layers was postulated to be a result of a high concentration of active surface sites for diamond nucleation, formed primarily due to the high surface defect density (in the form of dangling bonds) and the high hydrogen concentration in the DLC layers. The thin DLC layers not only substantially enhanced diamond nucleation density, but also facilitated microstructure control duringthegrowth ofthick diamond films, as discussed in Sec. 3, Ch. 5, above. In the HFCVD experiments conducted by Kanetkar et a1.,[1651 enhanced nucleation of polycrystalline diamond was achieved on Si( 100) coated with an epitaxial layer of yttrium-stabilized zirconia (Y-ZrO,). The Y-ZrO, layer, 150 nm in thickness, was grown by pulsed excimer laser ablation prior to diamond deposition, Since the diamond lattice constant is 3 567 1 A, while that of cubic Y-ZrO, is 5.13 A, the semiface diagonal of Y-ZrO, matches diamond within 1.85%. Diamond nucleation on the Y-ZrO, layer was the fastest at a gas pressure of 120 torr among the pressure values of40,80, and 120 torr, with CH,:H, of 1: 100, a substrate temperature of 850°C and W filament temperature of 2000°C. The use of a barrier layer in the form of oxides has three implications in terms of the growth of high-quality diamond films on different substrates at high growth rates. 1165lFirst, in a reducing environment such as the one present in a typical diamond CVD reactor, an oxide may release oxygen atoms to the ambient via the formation of O-H bonds (i.e., the Y-ZrO, layer reduces to Zr due to H,). The presence of such bonds is known to aid diamond growth. 1561[3361 Second, the continuous seeding of the freshly created metallic surface with reactive carbon radicals from the gas phase may lead to the formation of surface carbides (i.e., the reduced Zr evolves into a carbide), which may provide carbon-carbon bonding to the newly arriving carbon atoms and also may act as a Cdiffusion barrier. Third, the presence of the reactivity between the growing front and the gas phase may lead to a dynamic creation of a surface morphology in the form of steps and kinks, and thereby a

120

Diamond Chemical Vapor Deposition

continuous creation of nucleation sites. These three factors are considered to be responsible for the nucleation enhancement on the Y-ZrO, coated Si. Coating substrates with other C-diftision barrier layers has also been attempted. A 25 nm thick film of TiN coated on a Fe substrate was found to be sufficient to prevent soot from formation on the Fe substrate and to inhibit C diffusion into the Fe substrate.t3371 The thickness of the overlayers used in covering/coating pretreatments is summarized in Table 4. Table 4. Thickness of Overlayers in Covering/Coating Pretreatments Typ;opoverlayer Cmphile Fe

Thickness Depowon of averlayer method Cl pm MWPACVD LOOnm HFCVD(W)

Tf (%) 2100

T, (‘C) 8.50 800

CH4 in H2 gun flow ( “0, c, mtc(wcm) 1 -loI 1.5 ‘!

P (a,*, 100 10

Nd (c,,r2) IO6 4.84 x I05 7.07

FL?

2-8nm

MWPACVD

-

835

?

MWPACVD

-

835

IOOnm

MWPACVD

-

>650

? 150 “m 25.50nm

MW PACVD HFCVD(W) MW PACVD

aBN

?

ECR PACVD

csic

IO0 nm

ECR PACVD

? ?

HFCVD(W) HFCVD(Re)

Cu.Ti.Ni.Mo.Nb C70 a-C Y-Zro2 TIN

2oW -

940 850 9W

I .95 CH4 0.49 co* I .95 cH4 0.49 co* I

205

40

205

40

0.7

503.5

I

IO0

I

?

?

x I07

lSXlO6 enhanceman

? =sccding effect 3x 10’0 40 40.80.120 enhnncemcn, 30 diffusion

Subrtma maerial~ St St

Rcfcrur ,252, 13021

so* SIC

(3031

Sic

13031

Mo.W.Sa S, St Fe

11921 ,175, ,164, 13371 I2741

bxrw

a-SiC:H Hydmcabon ail EvaporatedC WC r-C. cBN DLC ion sputtered O-SIC a-C:H

IOOnm

HFCVD(?)

600-750 2-15 CH4 2.,002 6&X750 2.15,Ch 2-1002 I 2000 850 2100. 850 2200 I 210% 850 2200 1.48 2003 800

?

HFCVD(?)

2wO

IO-20 nm

HFCVD (Re)

-

I

800

1.48

FlameCVD

-

SWII00

f

MW PACVD

-

850

3CH4 0.5 02

4Wom

MWPACVD

-

850

3CH4 0.5 02

;rBN

ZOOnm

MWPACVD

-

x50

Amarphic diamond@

8OOnm

MWPACVD

-

850

2 mm

MW PACVD

-

850

?

HFCVD (W)

2000

830

3 CH4 0502 3CH4 0.5 02 0.5

?

HFCVD (W) HFCVD (W)

2lKIo 2MK1

X30 830

0.5 0.5

HFCVD (W)

2000

830

thin coating HFCVD (W) of tlaker

2000

830

C60 3.4.Y.10 psrylenc levacarboxvlic acid dianhydd&~ Owphiw powder Highly oriented pyrolydc graphite

paw&r ? powder

cnhancemem

S,

?

IO

enhancelnr”,

Si

IO

IO-IO0 45

I”1

‘? S, enhanccmcnl Mo.Si.Cu.Ag

12741 ,254, 1XWI

101

45

cnhancemcnt Mo.Si.Cu.Ag

[ScLi,

?

20

enhancement

I2971

?

20

0.95 6-10~10~ 02M32H2


3CH4 0.5 o*

(220) lcxlured.CVD bulk O-Sic P.C:H

10

?

760

Cu.%, slainless we, enhsnccmen, C”.S,, cnhanccmcn, E’Bi”;:s’=e’

,297, [,24.282

40

2.4~10’

500

40

3.5x103

SI, ZnS

(2791

500

40

1.2x105

Si. ZnS

12791

5W

Si. ZnS

12791

500

40

1.0~106

St. zns

[2791

500

40

7x107

-

I2791

IIns

20

4.7x103*

Si

,IX21

IM.5 loo.5

20 20

1.4x104* 8.3X104’

Si Si

,182,

0.5

100.5

20

1.2x105’

SI

llX21

0.5

loo.5

20

2.2x105’

SI

II821

* Relalrve nucleildon densay where sub-micron dtamond crysrnl, were cxcludcd.

II821

Effects of Surface Conditions on Diamond Nucleation

2.5

121

Ion Implantation

Ion bombardment is a well established process for improving and controlling the nucleation of metallic films on a variety of substrates.[3381 Since the surface energies of diamond are very high,[ll ion implantation method was motivated to modify the surface energies of substrates in order to enhance diamond nucleation. However, implantation of different ions on different substrates has been reported to have distinctly different effects on diamond nucleation, either enhancing or impeding the nucleation, as exampled below. Carbon ion implantation on a single crystal copper surface at a temperature of 82O”C, an ion dose of 10” ions cm-’ and a beam energy of 65 to 120 keV[lggl was found to result in an enhancement of diamond nucleation. The nucleation enhancement was postulated to be due to the formation of a graphite film on the copper surface, with subsequent diamond nucleation occurring preferentially on the edges of the graphite lattice. A focused ion-beam pretreatment on Si( 100) was reported to enhance diamond nucleation density. [33gl A precise array of diamond grains was grown in HFCVD using a 25 keV Ga+ focused ion beam. The ion beam created a crater array on the Si substrate and diamond nucleated only at the crater sites. Diamond nucleation enhancement on a 0.5 keV Ar+ sputtered Si(100) surface was also observed. [3401As+ and Si+ implantation on Si substrates without diamond scratching pretreatment[3411led to selective deposition of diamond only for low dose (100 keV, 1014ions cmW2), but not forhighdose(lOOkeV, 10’6ionscm-2). Nucleationdensitiesashighas 105-lo6 cm-* were achieved on As+ ion implanted Si, as compared to lo’-lo* cme2 after abrading Si substrate with diamond grit under the same deposition conditions. It is then interesting to know why both the abrasive scratches and the ion sputtered craters act as good nucleation sites. It also needs to be made clear whether the local surface facets in the ion sputtered craters or in the scratches are sufficient to cause diamond nucleation, or the lattice damages created by sputtering or scratching are responsible for the nucleation of diamond. In an attempt to address these issues, Lin et a1.[3411 conducted a series of experiments. Their results demonstrated that the local facets in the craters or scratches are not active sites for diamond nucleation, but instead, the lattice damages created by ion implantation or abrading are responsible for diamond nucleation. The lattice damages caused by ion implantation may

122

Diamond Chemical Vapor Deposition

involve strain, amorphous disorder and twins. Strain is most likely the physical reason for diamond nucleation enhancement on both ion implanted and mechanically abraded substrates, since strain energy can be released and thus enhance diamond formation reactions in the initial stages of nucleation. Hence, the strain field effects have been proposed as one of the operating mechanisms for diamond nucleation enhancement after different pretreatment processes, such as diamond abrasion, ion implantation and ion sputtering. Yosbikawa et a1.t3051investigated the effects of various surface pretreatments on diamond nucleation. The surface pretreatments of Si substrates included grinding with diamond wheels, polishing with alumina or diamond, mechanochemical polishing with zirconia, blasting with glass beads and etching with mixed acid. The results also show that the nucleation densities on crystallographically damaged surfaces were higher than those on non-damaged surfaces. Carbon ion implantation has been used as a pretreatment process to control the nucleation of diamond particles on surgical alloy Ti-6Al-4V substrates.t171] Carbon ions at 30 keV were implanted at room temperature into masked regions on the substrates, up to doses of 1016-7 x 1017ions cmW2. The SEM, microfocus Raman scattering and RBS analyses of diamond nucleation in MW PACVD indicated that the carbon ion implantation was very effective in controlling diamond nucleation on Ti-6Al-4V, and diamond nucleation density depended on ion dose. The carbon ion implantation gave rise to: (a) a decrease in diamond nucleation density, up to 8 times smaller than that on the substrate polished with 0.25 pm diamond paste (2-5 x lo5 cmm2vs. 2.2 x lo6 cmW2);(b) near perfect diamond particles; and (c) large internal stresses leading to partial flaking and poor adhesion when a continuous f&n formed on the entire surface. The formation of carbides at the substrate surface appeared not to be important for diamond nucleation on the Ti-6Al-4V alloy. Some ofthe implanted carbon was either incorporated into the growing diamond crystals or present at the diamond-substrate interface. Contrary to the strain field effects discussed above, the ion damages to the substrate surface or to the diamond seeds embedded into the surface during mechanical polishing were deemed to be responsible.for the observed decrease in diamond nucleation. In Kobayashi et al.‘s experiments,t260a] 100 keV Ar ions were implanted into scratched Si substrates and diamond films were deposited on the implanted Si. Diamond nucleation density decreased as the ion dose

Effects of Surface Conditions on Diamond Nucleation

123

increased. For the ion dose greater than 3 x 1015ions cmw2,virtually no diamond particles grew on the implanted area. In order to ascertain the reasons why the Ar+ implantation on Si substrates caused a decrease in diamond nucleation density, the surface morphology and crystallinity were examined using AFM in air, RHEED and RBS, respectively. It was found that the surface morphology changed with the ion dose, and the decrease in diamond nucleation density by the Ar+ implantation might be related to the change in the surface morphology and crystallinity. The decrease in diamond nucleation density after Ar+ implantation on Si has also been observed in other studies and utilized for the selective nucleation of diamond pa~icles_[3101[3111

2.6

Pulsed Laser Irradiation

The enhancement of nucleation and adhesion of diamond films has been achieved on Cu, stainless steel and Si using a pulsed laser irradiation pretreatment, followed by HFCVD. 12g7]A thin buffer layer of a-C, WC or cBN was first deposited on the substrates by pulsed laser evaporation, followed by pulsed laser irradiation. For the Cu substrates, diamond nucleation densities onthe a-C overlayer and WC (or cBN) buffer layer were 4-5 times and more than an order of magnitude, respectively, higher than that on an untreated Cu substrate. It is postulated that the irradiation converted some of the a-C on the surface into diamond or led to a reaction product that facilitates the nucleation of diamond. Laser can evaporate carbon preferentially, leaving diamond particles unaffected. In addition, the irradiation might melt the substrate surface and embed diamond particles into it, leading to an improvement in the adhesion of the diamond films with the substrates. The deposition of diamond on stainless steel substrates is quite diffkult but technologically very important. The difficulty is due partly to the very high diffusion rate of C into steel, leaving insufficient C or DLC on the substrate surface for diamond nucleation. The laser irradiation process12g71 may lead to the formation of carbide phases or diamond microcrystallites in the C overlayer on a stainless steel substrate, providing nuclei for diamond growth during subsequent HFCVD. The WC or cBN buffer layers on stainless steel substrates act as a diffusion barrier which makes the surface rapidly achieve carbon saturation.

124

Diamond Chemical Vapor Deposition

2.7

Carburization

Nucleation enhancement on precarburized W and MO substrates has been observed.12781These metals form carbides such as W,C and WC, as well as Mo,C and MoC,, during the carburization. However, MO and Mo,C have higher thermal expansion coefficients, 5.1 x 1Oa and 7.8-9.3 x lOa, respectively, as compared to that of diamond, 2 x 1Oa. Diamond films were found to crack off and separate from Mo,C/Mo substrates due to the high residual stresses between the films and the substrateslfl Since Sic has a thermal expansion coefficient halfthat ofMo,C, it was suggested171thatathin layer of Si (several to tens of nanometers) should be evaporated onto MO substrates. The carbon diffusion and reactions subsequently lead to the formation of molybdenum carbide, molybdenum silicide and silicon carbide layers on MO substrates, which build a C-diffusion barrier to reduce Mo,C thickness and provide strong atomic bonding between diamond films and the substrates. An epitaxial SIC conversion layer may form during the carburization of single-crystal Si substrates and the Sic layer facilitates the deposition of ordered diamond fihn~.l~~lThe in situ carburization provides an economical alternative to obtaining epitaxial diamond films on single-crystal Sic. The enhancement of diamond nudleation on both Si and Fe/Si substrates has been achieved by the carbonization pretreatment of the substrates.l302I 2.8

Catalytic Effects

It has been known for decades that transition metals such as Fe, Co, Ni, Cr, Pt and Pd can be used as solvent-catalysts for diamond synthesis under HPHT conditions.[2113421 The most common and effective catalysts are FeNi and Co-Fe.I*l Although the detailed mechanisms of the solvent-catalytic effects have not been completely understood, it has been proposed that the strong reactivity ofthese metals with carbon, the formation of metal carbides, the supersaturation of carbon in/on the metals and/or the deformation of graphite sheets by metal atoms to form diamond structure are important factors in the catalytic HPHT diamond growth processes. It is therefore interesting to consider whether the same effects can be utilized iu low pressure CVD processes. Attempts have been made to explore such catalytic effects in low pressure CVD processes.[302113031[3151

Effects of Surface Conditions on Diamond Nucleation

125

studied the effect of catalytic material (a lOOKobayashi et al. 1302113151 mu thick Fe film) on diamond nucleation on Si and SiO, substrates in HFCVD. It was found that, while diamond particles scarcely grew on a Si wafer, the catalytic material (thin Fe film) promoted the growth of diamond and enhanced the nucleation density by a factor of -50 and 7000 on the Fecoated Si and SiO,, respectively, compared to that observed on an untreated Si substrate. This catalytic effect on diamond nucleation was attributed to the diffusion of carbon atoms into the catalyst film and the increased sites for diamond nucleation created by the catalyst. Due to the high C diffusivity in Fe, the Fe film promoted the formation of carbide layers on the substrates. Hence, the catalytic effect of Fe in CVD diamond synthesis is essentially the same as in HPI-IT processes. l315] Diamond nucleation enhancement by a thin Fe film evaporated or laser-deposited on a Si wafer was also reported in RF PACVD13431 and HFCVD.13351It was confirmed thatC and Fe atoms difX.rse deeply into the Fe film and the Si substrate, respectively, and diamond particles were observed at the interface between the Fe film and the Si substrate.l343l Yehoda et a1.13031 presented a method to “catalyze” the nucleation and growth of diamond films in MW PACVD without seeding substrates. A thin film of Fe, Cu, Ti, Nb, MO, or Ni was abraded or deposited onto Sic-coated substrate surfaces. The metal films resulted in varying degrees of diamond nucleation enhancement. A qualitative ordering of the “best” to the “worst” nucleating metals was established to be Fe, Cu, Ti, Ni, MO to Nb at the substrate center (hot area), and Fe, Nb, Cu, MO, Ti, to Ni away from the center (cold area), with Fe exhibiting the most pronounced effect on diamond nucleation enhancement. For Fe films of thickness ranging from 0.5 to 8 nm evaporated on the Sic, it was found that the 0.5 nm Fe film did not show nucleation enhancement, the thicker films (2,4, or 8 run) did, though. The thicker the Fe films, the higher diamond nucleation density was achieved. Nevertheless, the degree of nucleation enhancement was not as significant as that obtained by using the simple method of polishing with diamond grit. Although the operative mechanism is not completely understood, the presence of the Fe film indeed played an active role in the diamond nucleation enhancement since diamond nucleation on Sic appeared to proceed no faster than on Si. The Sic layer merely played the role of eliminating the competition between diamond nucleation and Sic formation usually experienced in diamond CVD on Si. Fe,C was not detected in x-ray diffraction of the diamond films. However, the role of the Fe carbide or Fe silicide

126 Diamond Chemical Vapor Deposition formation in diamond nucleation enhancement should be an important area to address. The catalytic effect of Pd was examined by Brower et a1.[344]The wiping pretreatment of Si( 100) with Pd/A1203 and Pd/graphite led to a shorter incubation period relative to those on untreated, diamond-polished, A&O,-polished and graphite-wiped substrates. Nickel is one of the few materials possessing a close lattice match with diamond. However, its high carbon solubility and strong catalytic effect on hydrocarbon decomposition may be the reason that graphite layers develop on Ni substrates under the typical diamond CVD conditions. This effect impedes the direct nucleation of diamond on Ni substrates and excludes the eventual development of an orientational relationship between the diamond films and the Ni substrates, although diamond may eventually nucleate and grow on the graphite interlayers. Cobalt suppresses diamond nucleation by promoting soot formation.[156]

3.0

SUMMARY

Surface nucleation rates and densities of diamond on non-diamond substrates vary from lo3 to 10” cm-* h-l, and from lo3 to 10” cm-*, respectively, depending on substrate materials, surface pretreatment methods, and synthesis conditions. The possible maximum nucleation density of diamond would be -10” cm-*. Diamond surfaces or particles provide the best nucleation potential. Nucleation on cBN readily occurs. Nucleation rates on stable carbideforming substrates (Si, MO, W) are one to two orders of magnitude higher than on non-carbide-forming substrates (Cu, Au). Among carbide-forming substrates (Si, MO, Al, Ni, Ti), the nucleation density on MO is about one order of magnitude higher than on all other substrates under the same deposition conditions. Nucleation rates are several times higher on polycrystalhne substrates than on single-crystal substrates of the same material after identical surface pretreatment. Nucleation density increases with decreasing grain size or increasing number of grain boundaries of substrate materials. Refractory metal carbides (TaC, WC, Mo,C) and some covalent carbides (SIC, B,C) have a positive effect onnucleation, while effects ofionic carbides (Al,C,, liquid salts, etc.) on nucleation are less known. Nucleation occurs readily on substrates forming amorphous DLC without any polishing pre-

Effects of Surface Conditions on Diamond Nucleation

127

treatment. Graphite interlayers or particles form mostly on Ni, Pt, Cu, and Si, and favor diamond nucleation. Surface pretreatment methods, including scratching, seeding, electrical biasing, covering, coating, ion implantation, pulsed laser irradiation, and carburization, can enhance diamond nucleation on non-diamond surfaces, with ultrasonic-scratching and biasing having the best efficacy on nucleation enhancement, followed by scratching, seeding, covering, and ion implantation, etc., as summarized in Table 5. Nucleation on pretreated surfaces are observed to occur primarily on defects, such as scratches, grain boundaries, dislocations, electron bombardment damages, or the edges of etch pits and carbon-rich particles.

Table 5. Typical Surface Nucleation Densities of Diamond After Various Pretreatments Pretreatement Method No pretreatement Scratching Ultrasonic scratching Seeding Biasing Covering/Coating with Fe film graphite film graphite fiber a-C film (first scratched) C70 clusters + biasing hydrocarbon oil (first scratched) evaporated carbon layer (first scratched) Y-Zr02, a-BN, cSiC layer C+ ion implantation on Cu As+ ion implantation on Si Pulsed laser irradiation + coating a-C, WC, cBN layer Carburization

Nd (cm-*) 103 - 105 lob- 10’0 107- 101’ 106 - 10’0 108- 10’1

Reference

,

[ 167,174,279] [72,174,178.316,317] [54,167,317,318] [180,301] ]2s,178]

4.84~10~ 106 >109 3x1010 = seeding effect enhancement enhancement enhancement enhancement 105 - 106 enhancement

[3151 12521 [1511 V751 ~1921 [3041 [3041 [ 165,254,274,303] H991 [341]

enhancement

[278.302]

I2971

128

Diamond Chemical Vapor Deposition

Nucleation enhancement by scratching is attributed to (@ seeding effect, (6) minimization of interfacial energy on sharp convex surfaces, (c,J breaking of a certain number of surface bonds and the presence of a number of dangling bonds on sharp edges, (d) strain field effects, (e) rapid carbon saturation on sharp edges, fl removal of surface oxides, and (j$ formation ofnon-volatile graphitic clusters. Abrasives used for scratching pretreatment include diamond, oxides, silicides, nitrides, carbides and borides (Table 2). The effect of the abrasives on nucleation enhancement increases in the order: silicides < SiO, < nitrides < ZrO, < carbides < borides < Al,O, < BN < diamond. Nucleation density decreases with increasing particle size of diamond abrasive paste in the polishing pretreatment, but increases with increasing particle size in the ultrasonic scratching pretreatment. Generally, the optimum size of abrasive particles depends on pretreatment processes, deposition methods, growth conditions, and the nature of substrate materials. The efficacy of scratching on different substrates descends in the order: Si > MO > WC substrate. Dipping, spinning, spraying, and electrophoretic seeding have been employed to seed diamond, Si, cBN, A&O,, TaC, or Sic on various substrates. The size of seeds may vary from tens of nanometer to hundred of micrometer, with submicron diamond powder being used most frequently. Nucleation density is linearly proportional to the diamond seeding site density, being approximately a tenth ofthe seeding site density. The residual diamond seed particles on the substrate surface are the predominant nucleation sites (or the seed particles themselves are nuclei) and diamond growth occurs by means of homoepitaxy on these seed particles. Seeding also provides the possibility for epitaxial or highly oriented growth of diamond films on non-diamond substrates. Large-area mosaic diamond films approaching single-crystal quality have been grown by a seeding technique. Biasing substrates can enhance diamond nucleation without significantly damaging the substrate surfaces. Heteroepitaxial, highly oriented diamond films may be generated by means of bias-enhanced nucleation (BEN), followed by conventional growth under deposition conditions that favor oriented growth. Biasing can help to reduce and suppress oxide formation on the substrate surface, remove native oxides, and overcome the energy barrier for the formation of stable diamond nuclei by more effectively activating the substrate surface and increasing the flux and mobility of nucleation species. In DC PACVD and HFCVD, positive substrate biasing is effective for increasing diamond nucleation, while in MW PACVD both

Effects of Surface Conditions on Diamond Nucleation

129

positive and negative biasing can enhance diamond nucleation. By varying the duration of biasing pretreatment, the applied voltage and current, and/or the gas composition and pressure, diamond nucleation density can be controlled over several orders of magnitude. Lower gas pressures, higher CH, concentrations, longer biasing time, and/or larger absolute values of substrate bias voltage lead to higher nucleation densities. At the same absolute values of substrate bias voltage and for CH, from 10 to 40 vol.%, the nucleation densities on negatively biased substrates are one or two orders of magnitude higher than those on positively biased substrates. However, the attraction of cations in negative biasing leads to roughening of Si surfaces, whereas positively biased Si substrates maintain smooth surfaces. Therefore, positive biasing is a more suitable pretreatment condition for Si substrates. With increasing bias current, the grain size and non-diamond carbon incorporation in diamond films decrease with a concomitant increase in Young’s modulus and fracture strength, while large compressive stresses in films decrease and turn to tensile stresses. Nucleation enhancement has also been achieved by covering substrate surfaces with graphite fibers, clusters, or films, and coating substrate surfaces with thin films of metals (Fe, Cu, Ti, Nb, MO, Ni), CTO,a-C, DLC, Y-ZrO,, a-BN and SIC, or hydrocarbon oil, etc. The nucleation enhancement is attributable to the physical and chemical effects associated with changes to the substrate surface (the overlayers promote carbon saturation on the substrate surface, and provide high-energy sites or nucleation centers), and changes to the gas chemistry above the surface. The efficacy of overlaid materials on nucleation enhancement decreases inthe order: C,,, > a-C > DLC > graphite fiber > graphite film > Fe > Cu > Ti >Ni > MO> Nb. The thickness of the overlayers ranges typically from a few nanometers to about one micrometer (Table 4). The ion implantation method has been used to modi@ the surface energy and surface structure of substrates in order to enhance diamond nucleation. The effect of ion implantation on diamond nucleation depends on the type of ions and substrates, and ion dose and energy. Implantation of C+ (1018ions cmq2,65-120 keV) on Cu andAs+(1014 ions cme2, 100 keV) on Si enhances diamond nucleation, while Ar+ implantation (3 x 10” ions cmm2, 100 keV) on Si decreases diamond nucleation density. The lattice damages (strain, amorphous disorder and twinning) created by ion implantation are responsible for diamond nucleation enhancement. The strain is probably the

130

Diamond Chemical Vapor Deposition

primary physical reason for diamond nucleation enhancement on ion implanted substrates. Pulsed laser irradiation of a thin buffer layer of a-C, WC or cBN deposited on Cu, stainless steel, and Si substrates leads to enhancement of nucleation and adhesion of diamond films on the substrates. The irradiation may convert a portion of the a-C on the surface into diamond or result in the formation of a reaction product that facilitates diamond nucleation. Carburization of substrates (MO, W, Si, Fe/Si) leads to diamond nucleation enhancement due to the formation of carbides and the saturation of carbon at the substrate surface. Among Fe, Cu, Ti, Ni, MO and Nb, Fe shows the best catalytic effect on diamond nucleation. Pd has also a positive catalytic effect on diamond nucleation, while Co suppresses diamond nucleation by promoting soot formation. The strong reactivity of these metals with carbon, the formation of metal carbides, the supersaturation of carbon in/on the metals and/or the deformation of graphite sheets by metal atoms to form diamond structure have been proposed as possible mechanisms governing the catalytic effects. Scratching and seeding are simple and effective for diamondnucleation enhancement, but cause surface damages and contamination. These pretreatment methods cannot be easily applied to,substrates of complex geometry and shape, and are incompatible with many applications requiring extremely smooth, clean surfaces, such as diamond films for electronic devices, optical window materials and smooth wear-resistant coatings. As an alternative, biasing or covering/coating substrates can yield high nucleation densities comparable to, or even over those achieved by seeding or scratching, without significantly damaging the substrate surface, and therefore is of particular importance.

Effects of Deposition Conditions on Diamond Nucleation

Deposition conditions, such as substrate temperature, gas pressure, and gas composition, critically influence diamond nucleation density and rate. It has been noted that ideal growth conditions may not be optima1 for nucleation. For example, the in-vacua surface analyses and microstructure characterization ofthe diamond nucleation processes on negatively biased Si substrates[‘781 reveal that the biasing could enhance diamond nucleation significantly, whereas a much poorer-quality diamond film was grown if the biasing was continued during the growth. Similarly, the optimum values of gas pressure and substrate temperature for growth are not identical to those for nucleation, as will be discussed below,

1.0

SUBSTRATE

TEMPERATURE

Haq et al.[3451 presented the time evolution of diamond nucleation on Si(100) in MW PACVD, as shown in Fig. 1. The effect of substrate temperature on the nucleation processes is evident in the figure. For each substrate temperature, the nucleation density increases drastically after the incubation period, and then rapidly attains its saturation level. For substrate temperatures less than 8 16”C, higher substrate temperatures lead to shorter incubation periods, higher nucleation rates and higher nucleation densities. 131

132 Diamond Chemical Vapor Deposition

\

t 2.0 Q

‘E 1.6 r? B $ 1.2 2 %08-

s ‘c‘ go.42 2 0

_.-.. .x . . . . . . . . . . . x . . . . . . . . . . . . . x 0 i. j

:I ;o/ - . :

,_CL-;--q-_,+~~

b

p

-

30

P 60

I 90

* 120

1 750

z 180

y/fI I 210 240 270 time (min I -

300

Figure 1. Time evolution of nucleation density and influence of substrate temperature on nucleation. Gas pressure: 30 torr; CH&2: 0.4 vol.%; Mw power: 700 W, and substrate temperature: (a) 81692, (b) 716°C and (c) 516°C. [3451 (Reproduced with permission.)

JSim et a1.[346]studied the effects of deposition conditions on diamond nucleation on Si substrates ultrasonically pretreated with SIC powder. Their HFCVD experiments reveal that the nucleation rate increased with increasing substrate temperatures up to -9OO”C, reached a maximum at a substrate temperature around 950°C, and then decreased with further increase in substrate temperatures. A similar tendency was also observed by Hayashi et a1.[72] during diamond nucleation on Si in MW PACVD. The Si( 100) substrates were pretreated by rubbing with submicron diamond powder and wiping off the residual powder until visually clean. The temperature dependence of the nucleation density was measured with an ellipsometric monitor for two values of applied microwave power, as summarized in Fig. 2. For substrate temperatures less than 860°C, the nucleation density increased rapidly with increasing substrate temperature, and reached a maximum at 860°C; for substrate temperatures higher than 860°C, the nucleation density gradually decreased, but remained above lOlo cms2 in the range of the higher temperatures considered. This overall dependence of the nucleation density on substrate temperature was speculated to be caused by the change in the adsorption state and surface diffusion length of growth precursors. The precursors are adsorbed on the substrate surface mainly by physical

Effects of Deposition Conditions on Diamond Nucleation

133

adsorption below 900°C and predominantly by chemical adsorption above this temperature. This change in the adsorption state results in an abrupt increase in diffusion length of the precursors around 900°C. Consequently, the capture rate of the precursors (sticking probability) on the substrate surface, and hence the nucleation rate and density, drastically increased when the substrate temperature approached 86OOC. These results reveal that an optimum substrate temperature exists around 860°C, at which a maximum nucleation density can be achieved.

I

/---

c---*

/

f

i

/

I 109 t

1 850

900

SubstrateTeapenture

950

CC>

Figure 2 Temperature dependence of nucleation density measured with an ellipsometric monitor. Closed circles and solid line show values for 1000 W microwave power; open circles and broken line show values for 1400 W microwave power. Other deposition conditions: 5 vol.% CO/Ha, flow rate of 100 seem, and pressure of 50 tom Using CO as reactive gas led to diamond films containing hardly any non-diamond phases.[‘*] (Reproduced with pemission.)

134

Diamond Chemical Vapor Deposition

2.0

GAS-PHASE ACTIVATION

In Brunsteiner et al’s experimentsI15gl it was found that increasing filament temperature led to a high nucleation density and a higher deposition rate over the pressure range of 5-500 torr in HFCVD. Park and Lee’s experiments [347j show that the effect of filament temperature in HFCVD on diamond nucleation on Si is qualitatively similar to the effect of substrate temperature, i.e., with increasing filament temperature, the nucleation density initially increased, reached a maximum at 2 100°C and decreased thereafter. The drop-off for T > 2 100°C was explained by the observation that the etching of nucleation sites was enhanced with increasing filament temperature. Similar results were also reported in Ref. 346. In DC PACVD 1348]a nucleation density of 6 x 10’ cmm2was achieved on untreated substrates by increasing the discharge current to 1 A and the cathode temperature to 1400°C. It also has been suggested that diamond nucleation may be enhanced by using high power densities, such as in plasma jet, or DC plasma discharge CVD,134glin which H2 and CH, dissociation is promoted.

3.0

GAS PRESSURE

AND FLOW RATE

Chang et a1.I350linvestigated the effect of gas pressure on diamond nucleation density in microwave ECR discharge. It was found that the nucleation density decreased slightly, from 2.2 x IO6to 0.5 x lo6 cms2, as the gas pressure increased from 2 to 20 torr. Rim et a1.[157l studied the effect of gas pressure on the nucleation behavior of diamond on a Si( 100) substrate in I-IFCVD. The pressure was varied from 2 to 50 torr, while a filament temperature of 2200°C, a substrate temperature of 85O”C, a total flow rate of 20 seem and a CH, concentration of 0.8 vol.% were used. The characterization of diamond deposits using micro-Raman spectroscopy, SEM and OM revealed that the maximum nucleation density of diamond particles on the unscratched Si substrate occurred at a pressure of 5 torr. The pressure dependence of the nucleation density was explained by the competition effect between P-Sic formation, which increases the diamond nucleation density, and atomic-hydrogen etching, which decreases the number of nucleation sites. On the basis of this finding, a new fabrication approach for high-quality diamond films without

Effects of Deposition

Conditions on Diamond Nucleation

135

any substrate pretreatment was proposed, i.e., diamond nucleation at a lower pressure (5 ton-) and growth at a higher pressure (30 ton-). Increasing gas pressure may eliminate Si contamination from reactor walls to a substrate in MW PACVD, but it reduces diamond nucleation density.[1561An in situ treatment with an optimum combination of deposition conditions was proposed to reduce the nucleation delay and enhance the nucleation density on MO substrates, i.e., a higher CH, concentration at a lower pressure (12 seem CH,/4 seem 0,/184 seem HZ at 5 torr) to promote diamond nucleation by carburizing the substrate surface and enhancing the surface C concentration within a minimum time, followed by a lower CH, concentration at a higher pressure (12 seem CH,/4 seem 0,/392 seem HZ/ 196 seem Ar at 10 torr) to preserve diamond quality. Under these conditions, dense, homogeneous and well-crystallized diamond films were grown that would not have been achieved without using the initial nucleation stage at the higher CH, concentration and the lower pressure. Singh et al.‘st361experiments with hollow-cathode PACVD show that diamond nucleation density increased as the gas flow rate increased. A high gas flow rate may increase the mass transfer of gas species to the substrate surface, and hence diamond nucleation densities are enhanced.

4.0

GAS COMPOSITION

In HFCVD of diamond on Si1153113461t3511 and Cu13511substrates, diamond nucleation density was found to increase as the CH, concentration increased. Similar phenomena were also observed in MW PACVD W61[1371[1561[1W F'igure 3 clearly shows the effect of the gas composition on the nucleation density. As is evident in the figure, the nucleation density increases with increasing CH, concentration. The gas composition influences not only the nucleation density, but also the nucleation behavior and the resultant crystal morphology. Hyer et a1.t2471studied nucleation and growth of diamond in HFCVD under the following deposition conditions: 25 torr gas pressure, 250 seem gas flow rate, 1100 K substrate temperature and 2000 K filament temperature. At a lower CH, concentration (0.25 vol.% CH, in HZ), diamond nucleation terminated with a significant area of the substrate still unnucleated, and growth followed, leading to good-quality, well-faceted, isolated diamond

136

Diamond Chemical Vapor Deposition

crystals of cube-octahedral morphology. At higher CH, concentrations (1.2-1.4 vol.% CH, in Hz), diamond nucleation did not show obvious termination. Non-diamond components were incorporated into the films with “cauliflower” and ball-like morphology forming during the nucleation and growth.

Figure 3. SEM micrographs of diamond grown on a Si(100) wafer at various CH,:H,, (a) 0.1 vol.%, (6) 0.5 vol.%, (c) 1.2 vol.%, and (d) 2 vol.%, in HFCVD at a W-filament temperature of 2200°C, a substrate temperature of 750°C, a total gas flow rate of 45 seem, a filament-substrate distance of 5 mm, a growth time of 3 h, and a total gas pressure of 30 tor~.[‘~~] (Reproduced with permission.)

The change in nucleation behavior is also closely related to the etching of the substrate surface. In HFCVD experiments of diamond films and particles on Si substrates, t347]the nucleation density decreased markedly with increasing hydrogen-treatment tune prior to deposition. The enhanced etching ofthe substrate surface with increasing hydrogen-treatment time was thought to be responsible for the decrease in diamond nucleation density.

Effects of Deposition Conditions on Diamond Nucleation 5.0

137

OXYGEN ADDITION

A number of diamond CVD experiments demonstrated that oxygen species in the gas phase can improve both the growth rate and quality of diamond films.[41~~3361[352]-t35*~ Shah and Waitef211]used XPS to study the effects of oxygen on the nucleation and growth of diamond films on a Si substrate during MW PACVD. The presence of oxygen was found to accelerate the saturation of carbon on the substrate surface, reduce the incubation period, and promote a much faster diamond nucleation and growth than with oxygen-free plasmas. The incubation period was 40-50 min in the absence of oxygen and decreased to 1 min in the presence of oxygen. These effects were attributed to the preferential formation of SiO, rather than SIC during the very early stages of growth due to the presence of oxygen. The formation of SiO, was postulated to impede Si diffusion to the substrate surface, and thus the available carbon rapidly saturates the surface, thereby accelerating diamond nucleation. Moreover, in the absence of oxygen, the substrate temperature must be higher than 850°C for growth to occur, while the presence of oxygen allows lower substrate temperatures due to the lower binding energy of OH to C (85 kcal mol-‘) relative to that of CH to C (107 kcal mol-‘). This lower binding energy makes it easier to abstract OH by H to form H,O, thus lowering the substrate temperature required for growth. Bou et al.l156]also reported that a small amount of oxygen favors diamond nucleation and can enhance diamond deposition while preserving a good film quality at higher methane concentrations. They argued that, in MW PACVD, oxygen may be considered to be a diamond nucleation activator since it enhances the concentration of H atoms in the gas phase by suppressing heterogeneous recombination on the reactor walls, and allows high methane concentrations. In addition, oxygen suppresses the surface pollution by Si (from the reactor walls due to plasma etching). The surface pollution may inhibit or delay diamond nucleation, introduce defects into the diamond lattice, and result in bulk contamination of diamond. However, oxygen addition has also been reported to reduce diamond nucleation.[2561[35glThe HFCVD experiments conducted by Belton and Schmieg12561 reveal that, although the addition of oxygen in the gas mixture did not suppress growth of existing diamond and generally gave better

138 Diamond Chemical Vapor Deposition Raman spectra and more sharply faceted crystals, it suppressed diamond nucleation by eliminating the nucleation sites on Ni and Pt substrates. As shown schematically in Figs. 4 and 5, at the very high gas-phase oxygen concentration (e.g., [O]/[C] = I), the surface carbon was very rapidly etched from the substrate, leading to an essentially clean substrate surface. When the oxygen concentration was lowered to [O]/[C] = 0.75, the deposition of graphitic carbon was observed. This graphitic layer did not lead to diamond nucleation due to the absence of nucleation sites at that oxygen concentration. Only at low oxygen concentrations ([O]/[C] I 0.5), could nucleation sites be formed. Once nucleation occurred, growth was not prohibited by high oxygen concentrations. Diamond nucleation took place on the defect sites in the graphite deposits, which were susceptible to attack by the gas-phase oxygen species. The reason, why oxygen addition leads to larger, more perfect crystals, is that oxygen suppresses secondary nucleation on existing diamond by etching nucleation sites.[256] Ohtake and Yoshikawa[35g]examined the effects of oxygen addition on nucleation, growth and crystallinity of diamond films in plasma arc jet CVD. The etch rate of graphite by an Ar-0, plasma was 130 times larger than by an Ar-H, plasma. The nucleation density ofdiamond decreased with oxygen addition. The growth rate of diamond decreased with the increase in the amount of added oxygen. The surface unevenness and porosity also decreased with the increase in the amount of added oxygen. The optimum oxygen concentration was determined to be -33 vol.%. At this oxygen concentration, the surface unevenness of a 1.2 mm thick diamond film was -100 pm and the film showed good crystallinity due to the isotropic growth rate of { 1 1 1} and { 100} planes. Regarding the optimum oxygen content, it was proposed[360j that a high oxygen content should be used for etching non-diamond components in a growing diamond film, and a low oxygen content for deposition to achieve a higher diamond growth rate without sacrificing diamond quality. Oxygen contents of 2-10 vol.% have been employed in low pressure, low temperature MW PACVD of diamond films at high methane concentrations (2-15 vol.% CH, in H2).[274]

Effects of Deposition

Conditions on Diamond Nucleation

139

Before GroMh

-

--.

Oeposition of thick graphite deposit for a’f IO,1

l>,, o/c>,,

--

-_

_-

Nucleation Sites

o/c

VI Nucleation occurs only when OK: < 0.5

- (D

“Formation of nucleation sites’:

--

-_

_-

n

‘Nucleation’

i lb,O/C1,

0

‘Film Growih’

Assume growth proceeds for all 1021 as onpt

Figure 4. Schematic diagram showing the effect of oxygen on diamond nucleation and growth on scratched Ni foil substrates: oxygen suppresses the formation of diamond nucleation sites on the pre-deposited graphite layer. [*w (Reproduced with permission.)

140

Diamond Chemkal

o/c -

w -

1

Surface C rapidly etched by 0 species

"Clean"

--_

o/c

Vapor Deposition

= 3l4

Most slabte sites occupied

-

-v

“Deposition in scratches”

Nucleation Sites

3 (D “Formation of nucleation sites”

Nucleation occurs

1 o/c

= ‘I2

---

-

Diamond

02 does not impede growth

OIC - 1

“Growih”

Figure 5. Schematic diagram showing the effect of oxygen on diamond nucleation and growth on scratched Pt substrates: high oxygen concentrations clean the Pt surface of all graphite deposits so that diamond does not nucleate; if diamond has nucleated at low oxygen concentrations, growth will continue at the high oxygen concentrations at which diamond does not nucleate.t2s6] (Reproduced with permission.)

Effects of Deposition Conditions on Diamond Nucleation 6.0

141

SUMMARY

The optimum deposition conditions (ranges, values or tendency) for diamond nucleation reported in available literature are summarized in Table 1.

Table 1. Optimum ranges, values or tendency of deposition conditions for diamond nucleation

Parameters Substrate temperature MW PACVD HPCVD

Gas-phase activation Filament temperature in HPCVD Discharge current in DC PACVD Power density Gas pressure MW PACVD

Optimum ranges, values or tendency 860°C 1721 830 - 860°C [305] 925°C 900 - 1OOOT (3461 85O“C 1361.3621 2100 T 13471 Nd increases with increasing discharge current [348] Nd increases with increasing power density [349] 5 torr W51

q Gas flow rate hollow-cathode PACVD Gas composition MW PACVD Oxygen addition MW PACVD

HPCVD Plasma arc jet CVD Flame CVD

Ndincreaseswith

increasing gasflowrate

[36] Nd increases with increasing CHq vol.% in H2 f136.137.156.1741

Accelerates nucleation (21 l] Favors nucleation [ 1561 2 - 10 vol.% 02 + 2 - 15 vol.%CH4 in H2 12741 Suppresses nucleation [256] Decreases nucleation density, 33 vol.% 02 [359] 0.75 - OB/C2H2 [282]

An optimum substrate temperature exists around 860°C, at which a maximum nucleation density can be achieved. The effect of filament temperature in HFCVD on diamond nucleation is similar to that of substrate temperature, with 2 100°C being a possible optimum value. Nucleation may be enhanced by using high power densities, such as in plasma jet, or DC PACVD. Low gas pressures (-5 torr), high CH, concentrations, and/or high gas flow rates lead to high nucleation densities. The addition of oxygen in gas mixture can accelerate the saturation of carbon on the substrate surface,

142 Diamond Chemical Vapor Deposition reduce the incubation period, and promote a much faster diamond nucleation and growth than with oxygen-free plasmas. The presence of oxygen inthe gas phase allows low substrate temperatures, preserves a good film quality at high CH, concentrations, and suppresses eventual surface pollution by Si. However, the addition of oxygen has also been reported to suppress diamond nucleation by etching nucleation sites (graphite) on Ni and Pt substrates. A plausible optimum oxygen concentration is about 33 vol.% in plasma arc jet CVD, 2-10 vol.% in low pressure low temperature MW PACVD, and 0.75 0.9 O,/C,H, in flame CVD.

8 Theoretical and Modeling Studies on Diamond Nucleation

Compared to the significant development in nucleation enhancement methods, fundamental scientific issues related to diamond nucleation processes remain less well addressed. In this chapter, the theoretical and modeling studies on surface nucleation of diamond are reviewed on the basis of available literature.

1.0

IDENTIFICATION MODE

OF NUCLEATION AND GROWTH

In thermodynamic theory of heterogeneous nucleation of solid from vapor,[3631 a nucleus on a substrate is considered to be cap-shaped with a contact angle to the substrate. Young’s equation dictates the relationship among the surface and interfacial energies as: OS= cri + oo,cos e

143

144

Diamond Chemical Vapor Deposition

where the subscripts s, i, o refer to substrate, interface, and overgrowth, respectively. Depending on the material parameters of the substrate and deposit, and the supersaturation of the vapor, [3641three different nucleation and growth modes can be distinguished using the value of the contact angle 0: 1. 8 > 0: Volmer-Weber three-dimensional (3-D) island growth.l365l In this case, o, = CT;+o, cos 8. The cohesion energy between the adsorbing and condensing adatoms (i.e., the forming nuclei) is greater than the adhesion energy between the substrate atoms and the adatoms. Island growth occurs from the beginning of deposition, i.e., nucleation. This nucleation and growth mode is typical for the materials with higher surface energies relative to the substrate. 2.

0 = 0: Frank-van der Merwe two-dimensional (2-D) layer growth.l366l In this case, a,> Oi+ 6,. The adhesion energy exceeds the cohesion energy and the film tends to grow layer-by-layer. This growth mode occurs typically for the overlayers with lower surface energies relative to the substrate.

3.

8 = 0: Stranski-Krastanov layer plus island growth.l367l In this case, KS,2 (Ti+ o, and the strain energy in the One or more overgrowth is large compared to o,. complete monolayers are initially formed during growth. The total strain in the layer(s) increases as growth proceeds due to lattice mismatch. The existence of the strain in the initially deposited monolayers facilitates the breakdown of monocrystalline to polycrystalline structure and induces 3-D island growth similar to the Vohner-Weber mode on the top of the layers, leading to a decrease in the total energy of the substrate/film system. This mode has been observed in most heteroepitaxial growth processes in which the associated materials satisfy the surface energy criterion (0 = 0). 13681Only if the substrate is chosen in such a way as to minimize or eliminate the onset of strain in the deposit, will it be possible to grow out a singlecrystal film.11g61

Theoretical and Modeling Studies on Diamond Nucleation

14.5

the Vohner-Weber 3-D island nucleation As indicated by Davis, [3681 and growth mode is the only mode predicted for the case of diamond nucleation and growth directly on almost all non-diamond substrates, since chemically unterminated diamond surfaces possess the highest surface energies of any known materials (Tables 5 and 6, Ch. 3). The high surface energies of diamond largely limit the number of substrates that could potentially stabilize 2-D layer growth. Analyses ofthe surface and interfacial energies of various diamond-substrate systemst36glreveal that Be0 and BN may be suitable substrate candidates for diamond heteroepitaxy, and the substrate bonding type is more important than the lattice constant match. For example, Cu has a good lattice match with diamond, however, its utility as a heteroepitaxial substrate may be limited because of its relatively low surface energy and low adhesion energyto diamond.t3701Appropriate surface modification is therefore necessary to produce epitaxial diamond on CU.[~~] Further, the high surface energies of diamond, along with the small sticking probability of diamondnucleation precursors on non-diamond substrates, are deemed as the reason for small nucleation densities of diamond.t721t1841 The Volmer-Weber 3-D island nucleation and growth mode has been observed in a number of experiments of diamond nucleation. ~~~~1~~~~1~~~~1~~~~1~~~~1~~~~11~~~1~~~~1~~~~1~~~~1~~~~1 Asageneralnucle_ ation model, the Volmer-Weber nucleation theoj13651may be applied to the calculation of the activation energy for nucleation when the nucleation density is known from experimental measurements. Bauer et a1.[174l estimated the activation energy for diamond nucleation on Si to be 52 kcal mol-’ on the basis of their experimental results (Nd = 1.29 x lo4 cmm2). However, such calculations could not provide detailed information about diamond nucleation processes or mechanisms.

2.0

THEORETICAL STUDIES ON NUCLEATION THERMODYNAMICS

Diamond is a metastable phase under CVD conditions. The processes competing with diamond nucleation and growth are the nucleation and growth of graphite and/or amorphous carbon, as well as spontaneous graphitization of the diamond surface. The graphite nuclei may also contribute to creation of planar defects, which along with other defects, constitute the major obstacles to growth of single crystal diamond fihn~.l~~l The free-energy

146

Diamond Chemical Vapor Deposition

differences between these carbon phases are relatively small, so that kinetic factors, size effects, surface reconstruction, etc., during diamond nucleation may be more important than thermodynamic factors, and hence metastable diamond can be formed under kinetically controlled conditions. Searching such preferential conditions for diamond growth has taken decades of research.t21] Thermodynamic calculations [372]demonstrate that the equilibrium pressure of carbon vapor over diamond is -2 times higher than over graphite in the temperature range of 1000-2000 K. In the absence of other factors, graphite nucleation is more probable than diamond. To explain the improbable nucleation and growth of diamond under apparently metastable conditions, Derjaguin and Fedoseev[372]considered chemical kinetics based on macroscopic concepts of classical nucleation theory, adsorption-desorption kinetics and equilibrium, as well as surface mobility of the adsorbed carbon species, and revealed several key points: 1. The presence of a diamond surface, for example, an epitaxial or diamond-seeded surface, increases the probability of diamond nucleation. 2. There should be a small range of conditions under which the nucleation rate of diamond is greater than that of graphite. 3. Hydrogen dilution decreases the growth rate of graphite more than diamond. 4. Atomic hydrogen etches graphite faster than diamond. Thus, there exists a narrow range of conditions, such as pressure (supersaturation), temperature, and composition as well as substrate surface state (structure, roughness, etc.), under which the nucleation and growth of diamond are significant and graphite is etched. In this competition, graphite nucleation and growth will essentially cease or will be covered by diamond.[22] To determine this potentially small range of conditions, Derjaguin and Fedoseev[372]derived the ratio of the nucleation rates of diamond to graphite on the (111) diamond surface as a function of supersaturation, given by 0.9(++Yj2(I x(x -1)

-0.6~)

1

Theoretical and Modeling Studies on Diamond Nucleation

147

with

under the assumption of a two-dimensional nucleus. In these expressions, I is the nucleation rate, x is the supersaturation, +o is the bond energy of the closest neighbors in diamond crystal, K is the Boltzmann’s constant, T is the temperature, p is the partial pressure of carbon-containing vapor, pea is the equilibrium pressure of the vapor over graphite, and the subscripts D and G denote diamond and graphite, respectively. The calculated results show that a maximum of Idr, exists at x = 2.7. witbin a very narrow range around this value of X, the nucleation rate of diamond is high relative to that of graphite, but outside this small range, the nucleation rate of diamond is close to zero. Therefore, graphite nucleation will generally predominate, and only within a small range of conditions will diamond nucleation occur. Data on diamond deposition rates as a function of temperature118~t22~ show amaximum at about 1000°C. Surface reconstruction and relaxation on the { 1I 1} diamond surfacest221may occur in the range of 900 to 1000°C.t373~ Other phenomena and properties which critically depend on temperature in the same range (i.e., 900-1000°C) includet221adsorption/desorption and migration of atomic hydrogen, etch pit orientation, coefficient of static friction with metals, and oxidation rate. For example, two peaks in hydrogen desorption rates are present at -900°C and 1000°C.1221This temperature dependence of desorption clearly suggests the existence of a critical temperature in nucleation and growth processes of CVD diamond. This has been confirmed by recent experimental and theoretical results.t721t3461t3611t362] It has been recognized that large hydrogen gas dilution, typically 9899 vol.% H2, is the key to successful diamond growth under the metastable conditions.t221 Atomic hydrogen plays an important role in stabilizing diamond structure on the substrate surface relative to graphite:t221t’96113681~3741 1. As an adsorbent on the surface, atomic H acts to maintain the bonding of the surface carbon atoms in sp3 form, and the interchange of adsorbed H atoms with C atoms allows continuation of growth of the diamond structure; 2. Atomic H helps to reduce the amount of graphitic or amorphous C deposited on the substrate surface by rapidly reacting with these phases but allowing sp3 bonded diamond component;

148 Diamond Chemical Vapor Deposition 3. The dangling bonds at the surface of diamond are energetically unstable, and the surface will reconstruct to a graphite-like surface containing a mixture of single and double bonds to reduce the surface energies. A monolayer of reactive atomic H bonded to the dangling bonds will prevent the surface reconstruction and sp2 bond formation, and hence stabilizes the diamond structure relative to graphite. Additionally, by bonding to the surface and capping dangling bonds at the surface, atomic H reduces the surface energies of diamond. The role of atomic H in the nucleation processes of diamond is similar to that in the growth processes of diamond. It has been inferred that the influence of atomic H on diamond nucleation and growth is more important than the influence of temperature.t221 The role of various substrates in stabilizing diamond relative to graphite was evaluated by Machlin 11”1for conditions under which graphite is thermodynamically stable relative to diamond in the bulk. The theoretical calculations of entropy, bond, and bond angle strain reveal that many metallic substrates that bond to carbon can stabilize the diamond phase even in the absence of atomic hydrogen. This stability af the diamond structure relative to graphite is attributable to the large differences in the bond energies or strain energies between diamond-substrate and graphite-substrate systems so that the small difference in the free-energy between diamond and graphite in the bulk (about 0.454 kcal mol-‘) is overwhelmed. The substrate surfaces, that are most effective in achieving the thermodynamic stabilization of metastable phases, are those that can produce an epitaxial fit in atom arrangement between the crystal planes of the metastable phases and the substrate surfaces. Therefore, substrate materials should be limited to those that can minimize the difference in atom configuration between the substrates and diamond crystals. Pseudomorphic stabilization of diamond is possible on the following substrates:11g6l 1. Unreconstructed, clean diamond surfaces in ultra high vacuum (UHV) or exposed to atomic hydrogen. 2. Reconstructed, clean diamond surfaces in UHV or exposed to atomic hydrogen above the temperature for surface reconstruction.

Theoretical and Modeling Studies on Diamond Nucleation

3.

149

Clean surfaces of many metals in UHV or exposed to atomic hydrogen.

4. Clean liquid metals exposed to atomic hydrogen when the average of the metal-metal and carbon-carbon bond energies is less than the metal-carbon bond energies, and the metals do not bond strongly to hydrogen. The influence of substrate materials on the relative nucleation rates of diamond to graphite was quantitatively determined by Kernt1g5]on the basis of classical nucleation theory. The crux of the theory is the formation of a nucleus of critical size. The Gibbs free-energy of the formation of a critical nucleus of spherical geometry, AG, is expressed as

where CTis the surface free-energy at the nucleus-vapor interface, and A,u is the difference in volume free-energies between the vapor and solid phase. Assuming that the temperature is high enough for the thermodynamic driving forces of the nucleation of diamond and graphite, from a supersaturated vapor to be nearly equal, i.e., Ap,_, = A,u~_~,but not so high that diamond, when formed, reverts to graphite, the ratio of the Gibbs free-energies of the nucleation of diamond to graphite reduces to

where V is the volume of an atom. The relative nucleation rates of diamond to graphite are then governed by the surface energies. From the bond strength, lattice parameters and densities of diamond and graphite, the ratio is calculated to be AGdAGG = 16. However, on a substrate, this ratio must be modified by a factor

150

Diamond Chemical Vapor Deposition

where oi is the surface energy of the deposit crystal face i, which comes into contact with the substrate, s, and pj,#is the adhesion energy given by the Dupre relation Pi,8 =

6

+

us- ui,.s

where bi,.sis the interface energy between the deposit crystal face i and the substrate. On a clean, unreconstructed diamond substrate, for example, ifthe crystal plane of the diamond deposit is parallel and identical to that of the substrate, then oi = cr,and Q = 0, so that [ 1 - &l(2ui) = 0],, while that of graphite is non-zero. Thus, diamond is expected to nucleate faster than graphite. It should be noted, however, that most conditions of deposition from the vapor phase have been shown to be such that classical nucleation theory is not well-suited to describe the nucleation kinetics of diamond, since the critical nucleus size is on the order of a few atoms. 13”1The small size of the critical nucleus makes it quite inappropriate to use the classical thermodynamic variables to describe the nucleation processes. Under such conditions, the Gibbs free-energy of the formation of a critical nucleus cannot be expressed with the above formulation. The surface energy contribution may cause a reverse effect on the phase stabilityl**l and dG may be less than zero.[376l Hence, a nanometer-sized diamond nucleus may be more stable at subatmospheric or atmospheric pressures than agraphite nucleus containing the same number of atoms.12211377] A quantitative calculationl377]shows that surface energies are an important aspect in the stabilization of nanocrystalhne diamond, and for surface bonds terminated with hydrogen atoms, diamond crystals smaller than -3 nm in diameter are energetically favored over polycyclic aromatics (the precursors to graphite). The case of dG < 0 has been referred to as a nonclassical nucleation process. In such a case, the surface energy contribution to nucleation must be evaluated on the basis of a microscopic framework of a nucleus,l376]and atomistic theory[375]should be employed for studying the nucleation process.

3.0

THEORETICAL

MODELING OF NUCLEATION KINETICS

Based on the work of Avrami1378]and Frenke1,[37glTomellini et a1.12i71 developed a kinetic nucleation model to describe the time evolution of the

Theoretical and Modeling Studies on Diamond Nucleation

151

surface nucleation densities and rates of diamond in CVD. In this kinetic scheme (Fig. l), it is assumed that a surface nucleation process starts from nucleation sites on the substrate surface that are generated by a scratching pretreatment. A nucleation site is first transformed into a germ, and a nucleus then forms from the germ. Agerm is defined as an intermediate species during the formation of a nucleus. A germ may be desorbed from a nucleation site, whereas a nucleus can only grow. Nucleation sites and germs may be captured by the growing diamond phase. Nucleation sites may also destruct due to thermal and/or chemical attack during the nucleation process.

K

v

c

Nucleus

GWKI A ~ “d

7 ’ Capture

_

“f

1

Active site

-

__---

+

Center

1

Figure 1. Schematic diagram of the kinetic nucleation model developed in Ref. 217. (Reproduced with permission.)

152

Diamond Chemical Vapor Deposition

By assuming first-order kinetic processes, Tomellini et al. formulated the rates of each of the four steps in the kinetic scheme as a function of the surface densities of the transforming species, with nf, Q, K and Kc as rate constants for the corresponding steps. The equations for diamond nucleation density and rate were then derived by solving these coupled kinetic equations. For small values of surface coverage (S(t)
P =NoD l+ -exp(-at)--% w*>c? a-B [

a_p

W-B*)

1

and

J% (2)

k(t),

= N, $+exp(+*)

-

exP(-a*)1

with Eq. (3)

N, = KnfN+(0)l(alS)

Es. (4)

a=

$(nf+nd+K+Kc)+,/(n~-n~-K+Kc)2+4npd]

and

Es. (5)

j3 = $(nf+nd+K+Kc)-,/(n~-n~-K+Kc)2+4np~]

where N(t), andk(t), are the surface nucleation density and rate, N, is the surface nucleation density when t ---, w, t is the time, N+ (0) is the surface density of the nucleation centers at t = 0, a and pare the parameters related to the rate constants. When the surface coverage is so large that the capture of the nucleation sites and germs by the growing diamond phase may not be neglected during the entire nucleation process, the nucleation rate must be modified by a factor [l - S(l)]:

Theoretical and Modeling Studies on Diamond Nucleation

Eq-(6)

153

$(t)s = H(t),[1 -s t ]

To predict the time evolution of the nucleation process using these equations, thevaluesofNm, aandpand S(t) must be known for a general range of time and deposition conditions. The parameters N,, a and /3 are constant with respect to time and depend primarily on the deposition conditions, such as gas composition CH,/Ha and substrate temperature T,, and substrate pretreatment. As a first approach, a linear growth law of the form,

Eq.(7)

R

= a(z-x)

was assumed, whereR is the grain radius,a is a constant,z is the deposition time, x is the time at which the grain starts nucleation, and it is the power index, 0 < ns 1. By analyzing the SEM data of the diamond particles grown in HFCVD on Si(ll1) and Si(100) scratched with 0.25,~m diamond paste, the diamond particle size distribution and surface density were obtained as a function of deposition time. Thevalues of theparametersN,, aand pwere then determined by fitting Eq. 1 to the experimental data for the nucleation of well-separated particles on the surfaces (i.e.,,S(t)
4?_

(a>

Cd-H+ H(g)

09

C, + CH,(,) -

(4

cd + H(s) -

kb

42

‘d

+ H2(,)

C,CH,

Cd-H

154

Diamond Chemical Vapor Deposition

which excluded details of the incorporation reactions, and any reactions leading to thedeposition of non-diamondcarbon. This simplification, which was made on the basis of the sensitivity analyses in Ref. 87 and the observations inRef. 381 significantly reduced the number of the reactions to be considered in the kinetic scheme. The modified expression for the grow rate couples the nucleation density with the crystal growth, and relates the growth rate to gas-phase composition, kinetic rate constants and surface reactions as well as gas dynamic conditions at the gas-substrate interface:

where c is a constant characteristic of the crystal shape (c = 3.42 for a cube or a sphere), [CH4]QH2] is the gas composition, [HlsWfand [H] are the atomic hydrogen concentration at the substrate surface and in the gas phase, 1 is the characteristic length of the system (boundary-layer thickness above the substrate surface), cH is the mean velocity of atomic hydrogen at gas temperature Tg, D is the diffusivity of H in H,, NU is the Nusselt number,f and g are the constants related to the recombination coefficient of atomic hydrogen on Si, yDis the recombination coefficient of atomic hydrogen on a diamond surface, k,, kb and kc are the kinetic rate constants for the reactions (a)-(c), K, is the equilibrium constant for the reaction:

(4

CH,+H*CH3+H2

and S(t) is the surface coverage fraction formulated as a function of the surface nucleation density and crystal growth rate. The recombination coefficient depends on the Debye temperature of a solid material. Since the Debye temperatures of diamond and silicon are 1860 + 10 and 650 K, respectively, it was assumed thatyr,= 3y,;. The modified formulation of the

Theoretical and Modeling Studies on Diamond Nucleation

155

growth rate, Eq. 8, introduces two additional parameters, i.e., the concentrationofHatoms inthegasphaseand therecombinationcoefficientofHatoms on Si substrates, which need to be determined by fitting the model to experimental data. The HFCVD experiments of diamond on Si(100) were run under the following conditions: substrate temperatures of 500-95O”C, gas compositions of OS-2 vol.% CH, in H,, and total gas pressures of 15-76 torr. The fitting of the model to the experimental data showed that lo-* 5 ysi s lo-‘, and [Hk&[H] = 0.16 and 0.84 for the gas pressures of 76 and 15 torr, respectively. The series of diamond growth experiments conducted by Molinari et al.t361]demonstrate that thesurfacenucleationdensity increased with increasor with decreasing particle size of the ing PWP21 Pm 0~{CCH,1/[H21~2~3>, diamond paste used for substrate scratching. At a substrate temperature of -8SO”C, diamond nucleation density experienced a maximum, and both a and j?, obtained from fitting the data, exhibited minima. The fitting of the model to the experimental data further demonstrates that, at this substrate temperature, the recombination coefficient also attains a maximum value, which partly explains the dependence of the nucleation density on the substrate temperature. The experiments also reveal that the observed number of surface nuclei was a fraction of the total nucleation sites available on the substrate surface, and this fraction varied with substrate temperature, gas composition and surface pretreatment. To elucidate thedependenceof the nucleation density on the deposition conditions, Molinari et a1.f3’j2]further modified the kinetic model, assuming the chemisorption of acetylene on the nucleation centers being the first step in a nucleation process, and the GH, and qH3CH3 species converted from the chemisorbed C$H2 being the active sites. A germ then forms from such an active site, followed by the irreversible formation of a nucleus from the germ. Two Si-Si dual sites suitable for the initial chemisorptionof two GH, molecules, separated by a distance smaller than a critical value, are defined as a nucleation center. The distance criterion apparently dictates that only a fraction of the nucleation centers which satisfy the corresponding energy criterion (H > Hcut_&) can eventually transform into nuclei, i.e., only nucleation centers with the desorption enthalpy (I$,) larger than the cut-off value (Ifcut&) can remain on the surface long enough for a germ of average size to be created. Since the value of HcUt-orris dependent on substrate temperature, gas composition and surface pretreatment, the surface nucleation density is then also governed by the deposition and surface conditions.

156

Diamond Chemical Vapor Deposition

The theoretical modeling of the kineticaspects of diamond nucleation processes is indeed scarce in published literature. Although attempts have been made to model the nucleation kinetics,l217ll361ll362ll380las reviewed above, the approaches require an accurate estimation of the kinetic rate constants, necessitating fitting the kinetic model to experimental data, thereby making the model system- (or experiment-) dependent. In addition, the kinetics of surface diffusion of adatoms and the formation of intermediate carbonaceous phases were not considered in these studies. As indicated in Ref. 217, a kinetic model is expected to contribute to a better understanding of the role of Sic formation in the nucleation of diamond on a Si substrate. However, the kinetic scheme employed in these studies was, in fact, unable to distinguish between a Si and a SiCsurface. To capture the possibility that an intermediate carbonaceous phase (such as DLC, carbide or graphite) may form prior to diamond nucleation, the kinetic model should be modified to include a time dependence of the density of nucleation sites determined by the kinetics of the formation of the intermediate phase. Further studies are therefore needed to construct a clear picture of the kinetics of diamond nucleation processes in CVD.

4.0

CLUES OF STRUCTURE, DIAMOND NUCLEI

CHEMISTRY,

AND SIZE OF

As discussed in Ch. 6, diamond nucleation enhancement may be achieved by, for example, predeposition of a thin DLC layer in combustion flame or DC glow-discharge plasmas,l124ll282lor, by carburization/biasing to developacomplet&iClayeron thesubstratesurfaceinMWPACVD.1178111871 In spite of these advances in experiments, the structure, chemistry, size, molecular or structural arrangement of nucleation sites and diamond nuclei are still less clear. If the structure, chemistry, and size as well as the detailed formation mechanism of nucleation precursors can be identified, the nucleation of diamond in CVD may then be more easily controlled and better tailored toward specific applications. To this effect, Meilunas and Chang lr921developed a model to calculate the minimum energy geometry using a force-field method (SYBYL). Shen et a1.l382lmodeled diamond-based precursors employing a finite cluster approximation approach. This theoretical modeling investigated carbon cluster fragments of the diamond lattice which are saturated by hydrogen

Theoretical and Modeling Studies on Diamond Nucleation

157

atoms, i.e., so-called adamantane-based or polymantane-derived carbon clusters. Ab initio quantum mechanical computations were performed to study the convergence behavior of the properties of the clusters toward the properties of bulk diamond. Stable polymantane-based clusters, up to &Hj6, were considered in the calculations. However, little indication was given regarding the stability of such molecules in the diamond CVD environment. Computer simulation by Sunkara et a1.l53ll55lusing Monte Carlo method predicted the morphology evolution (twinning and faceting) of diamond crystals in the early stages of growth. In this simulation, only the addition and removal of single carbon species (CH,) on an unreconstructed diamond surface were considered, although the direct addition of two and/or three carbon atoms (qHX and/or C$HJ could also contribute to growth. The effect of surface diffusion of adatoms (or hydrocarbon fragments such as methyl groups) was not included in the modeling. For a given nucleation site, the relative probabilities of the removal of singly bonded, doubly bonded, and triply bonded surface atoms were assumed to be 1, 10e9and 0, respectively, and the probability of the addition relative to the removal of a singly bonded carbon atom was l/20. The simulation by Sunkara et a1.l53ll55lstarted from nuclei of an assumed structure, chemistry, and size, such as a twinned nucleus containing 18 carbon atoms with two stacking errors (boat-chair-chair-boat tetracyclo octadecane), or a nucleus containing 14 carbon atoms with a single stacking error (chair-boat-chair tricycle tetradecane), etc. These structures were chosen to examine the influence of specific defects on the final morphology. As indicated by the authors, although these molecules are not necessarily the true precursors, the true nuclei may likely contain some of these structures within highermolecularweightcompounds formed by the partial hydrogenation of graphitic or polyaromatic hydrocarbons. The simulation shows that a reentrant corner will form at the intersection of a pair of parallel { 111) stacking errors with the surface of a diamond-cubic crystal. The reentrant corner provides a self-regenerating site for the nucleation of new layers and leads to rapid growth of the crystal within the plane of the twin band. This occurs because only two atoms are required to form a surface nucleus at the reentrant corner, and once the second atom is added the probability of the removal of the doubly bonded atoms is greatly reduced. On the contrary, a single stacking error does not yield a self-regenerating site for atom addition to readily occur and therefore results in a much lower growth rate. The

158

Diamond Chemical Vapor Deposition

simulation of the growth from the twinned nucleus led to a flat, hexagonally shaped platelet containing a microtwin. Such hexagonal platelets have been observed in the MW PACW experiments conducted by the same authors. Hirabayashi and Matsumoto 11441also reported the formation of flattened diamond crystals in MW PACW using oxygen-containing gas, and attributed the phenomenon to the reentrant corner effect and the resultant formation of the twin boundaries. These results suggested that diamond nuclei may contain multiple twinned structures.I53ll383l

5.0

SUMMARY

Frank-van der Merwe 2-D layer-by-layer growth or in some cases Stranski-Krastanov layer plus island growth is the growth mode of diamond homoepitaxy and heteroepitaxy. Volmer-Weber 3-D island growth is the mode of nucleation and growth of polycrystalline diamond films directly on non-diamond substrates due to the highest surface energies of diamond among any known materials. In theoretical studies on diamond nucleation, thermodynamic theory, homogeneous and heterogeneous chemical kinetics, classical nucleation theory, adsorption-desorption kinetics and equilibrium have been considered to predict preferential conditions for diamond nucleation and growth. A narrow range of conditions, such as pressure (supersaturation), temperature, and composition as well as substrate surface state (structure, roughness, etc.), have been derived, under which nucleation and growth of diamond are significant and graphite is etched. The theoretical studies uncovered the crucial role of hydrogen atoms in stabilizing the diamond structure on substrate surfaces relative to graphite. The role of substrate surfaces in stabilizing the diamond structure has also been recognized. Classical nucleation theory may be not well suited to describe the nucleation kinetics of diamond in CW, since the critical nucleus size under the typical CW conditions may be on the order of a few atoms. The surface energy contribution may cause a reverse effect on the phase stability and the Gibbs free-energy of the formation of a critical nucleus may be negative, a case referred to as nonclassical nucleation. In such a case, atomistic theory should be employed as the starting point of theoretical analyses.

Theoretical and Modeling Studies on Diamond Nucleation

159

Theoretical modeling of diamond nucleation kinetics is scarce in published literature. Some existing approaches require an accurate estimation of kinetic rate constants, which has been made on the basis of empirical and experimental data, and hence is system- (or experiment-) dependent. Therefore, further studies are still needed to improve the fundamental understanding of diamond nucleation kinetics. The current information on size, structure and chemistry of diamond nuclei is primarily speculative, with a small number of conclusive results. It has been proposed that diamond nuclei may be multiple twinned particles, likely containing some of the structures related to the boat-boat conformer of bicyclodecane (10 carbon atoms) or boat-chair-chair-boat tetracyclo octadecane (18 carbon atoms) within higher molecular weight compounds formed by the partial hydrogenation of graphitic or polyaromatic hydrocarbons. The diameter of a critical nucleus of diamond is presumably around 3 nm. It is evident from published literature that the technological problems associated with the nucleation of polycrystalline diamond films have been adequately addressed, as demonstrated by the development of the numerous nucleation enhancement methods, the selective growth of diamond films of desired shape and thickness, and the textured, oriented growth. Particularly, the advances in thegrowth of highly oriented diamond films representa novel approach to obtaining diamond films of a near-single-crystal morphology over large areas and may allow diamond to realize its potential as an electronic material. However, the scientific issues integral to diamond nucleation processes remain less well understood. Further work is needed in the areas such as single crystal growth and heteroepitaxy. A comprehensive theoretical model is required to achieve a thorough understanding of diamond nucleation processes and to obtain more predictable results.

References

1.

J. E. Field, The Properties of Diamond, Academic Press, London, (1979)

2.

H. 0. Pierson, Handbook of Carbon, Graphite, Diamond and Fullerenes, Noyes Publications, Park Ridge, New Jersey (1993)

3.

P. K. Bachmann, and R. Messier, Emerging C&EN, 67(20):24-39 (May 15, 1989)

4.

Status and Applications of Diamond and Diamond-like Materials: An Emerging Technology, National Materials Advisory Board, NMAB-445, National Academy Press, Washington, DC (1990)

5.

S. B. Qadri, C. Kim, E. F. Skelton, T. Hahn, and J. E. Butler, Thermal expansion of chemical vapor deposition grown diamond films, Thin Solid Films, 236(12):103-105 (1993)

6.

Y. Seino, and S. Nagai, Temperature dependence of the Young’s modulus of diamond thin film prepared by microwave plasma chemical vapour deposition, J. Mater. Sci. Lett., 12(5):324-325 (1993)

7.

M. S. Wong, R. Meilunas, T. P. Ong, and R. P. H. Chang, Tribological properties of diamond films grown by plasma-enhanced chemical vapor deposition, Appl. Phys. Lett., 54(20):2006-2008 (1989)

8.

K. Baba, and Y. Aikawa, Mechanical properties of diamond films by a filament assisted chemical vapor deposition, NEC Research and Development, 34 (2):176182 (1993)

9.

J. E. Graebner, S. Jin, G. W. Kammlott, B. Bacon, L. Seibles, and W. Banholzer, Anisotropic thermal conductivity in chemical vapor deposition diamond, J. Appl. Phys., 71(11):5353-5356 (1992)

10.

H. Liander, ASEA JZ, 28:97-98 (1955)

160

technology

of diamond thin films,

References

161

11.

F. P. Bundy, H. T. Hall, H. M. Strong, and R. H. Wentorf, Man-made diamonds, Nature, 176:51-55 (1955)

12.

H. B. Dyer, F. A. Raal, L. du Preez, and J. H. N. Loubser, Phil. Mug., 11:763-775 (1965)

13.

G. Cowan, B. Dunnington, and A. Holtzman, Process for Synthesizing Diamond, U.S. Patent 3,401,019 (Sept. 10, 1968)

14.

W. G. Eversole, Synthesis of Diamond, U.S. Patents 3,030,187, (April 17, 1962)

15.

J. C. Angus, and C. C. Hayman, Low-pressure, metastable growth of diamond and ‘diamond-like’ phases, Science, 241 (4868):913-921 (1988)

16.

J. C. Angus, A. Argoitia, R. Gat, Z. Li, M. Sunkara, L. Wang, and Y. Wang, Chemical vapour deposition of diamond, Phil. Trans. Royal Sot., Ser. A (Whys. Sci. Eng.), 342 (1664):195-208 (1993)

and 3,030,188

17.

H. Schmellenmeier,

18.

B. V. Derjaguin, D. V. Fedoseev, B. V. Spitzyn, D. V. Lukyanovich, B. V. Ryabov, and A. V. Lavrentev, Filamentary diamond crystals, J. Cyst. Growth, 2:380-384 (1968)

Exp. Tech. Phys., 149 (1953)

19.

J. C. Angus, H. A. Will, and W. S. Stanko, Growth of diamond seed crystals by vapor deposition, .I Appl. Phys., 39:2915-2922 (1968)

20.

J. J. Lander and J. Morrison, Surf: Sci, 4(3):241 (1966)

21.

R. C. DeVries, Synthesis of diamond Mater. Sci., 17:161-176 (1987)

22.

A. R. Badzian, and R. C. DeVries, Crystallization of diamond from the gas phase, part 1, Mater. Res. Sot. Bull., 23:385-400 (1988); A. R. Badzian, T. Badzian, R. Roy, R. Messier, and K. E. Spear, Crystallization of diamond crystals and films by microwave assisted CVD: Part II, Muter. Res. Sot. Bull., 23:531-548 (1988)

23.

R. E. Clausing, L. Heatherly, and E. D. Specht, in Diamond andDiamond-like Films and Coatings, (R. E. Clausing, L. L. Horton, J. C. Angus, and P. Koidl, eds.) Plenum Press, New York, pp. 619-626 (1991)

24.

J. C. Angus, History and current status of diamond growth at metastable conditions, in Proc. First Int. Symp. on Diamond and Diamondlike Films, Pennington, New Jersey, Electrochemical Society, pp. l-13 (1989)

25.

B. V. Derjaguin and D. V. Fedoseev, Vapor growth of diamond on diamond and other surfaces, Scientific American, 233(5):102-109 (1975)

26.

B. V. Spitzyn, L. L. Bouilov, and B. V. Derjaguin, Vapor growth of diamond on diamond and other surfaces, .I C?ystul Growth, 52:219-226 (1981); B. V. Spitzyn, and B. V. Derjaguin, in: Problems of Physics and Technology of Wide-Gap Semiconductors, Akad. Nauk SSSR, Leningrad, pp. 22-34 (1979)

27.

S. Matsumoto, Y. Sato, M. Kamo, and N. Setaka, Vapor deposition of diamond particles from methane, Jpn. .I. Appl. Phys., 21, L183 (1982)

28.

S. Matsumoto, Y. Sato, M. Tsutsumi, and N. Setaka,J. Muter. Sci., 17:3106 (1982)

29.

M. Kamo, Y. Sato, S. Matsumoto, and N. Setaka, Diamond synthesis from gas phase in microwave plasma, .I. Cyst. Growth, 62(3):642 (1983)

under me&table

conditions,

Ann. Rev.

162 Diamond Chemical Vapor Deposition 30.

R. Haubner, and B. Lux, Diamond growth by hot-filament chemical vapor deposition, state of the art, Diamond R&at. Mater., 2(9):1277-1294 (1993)

31.

S. Saito, S. Matsuda, and S. Nogita, .I Mater. Sci. Left., 5565 (1986)

32.

H. Kawarada, K. S. Mar, and A. Hiraki, Large area chemical vapor deposition of diamond particles and films using magneto-microwave plasma, Jpn. J. AppZ. Phys., 26:L1032 (1987)

33.

S. Matsumoto, Chemical vapour deposition of diamond in RF glow discharge, J. Mater. Sci. Len., 4(5):600-602 (1985)

34.

R. Mania, L. Stobierski, and R.Pampuch, Diamond synthesis in cool plasma, Cv.st. Res. Tech., 16:785 (1981)

35.

K. Suzuki, A. Sawabe, H. Yasuda, and T. Inuzuka, Growth of diamond thin films by DC plasma chemical vapor deposition, Appl. Phys. Lett., 50 (12):728 (1987)

36.

B. Singh, 0. R. Mesker, A. W. Levine, and Y. Arie, Diamond synthesis by hollow cathode plasma assisted chemical vapor deposition, in Proc. SPZE, 877:70 (1988)

37.

K. Kurihara, K. Sasaki, M. Kawaradi, and N. Koshino, High rate synthesis of diamond by DC plasma jet chemical vapor deposition, Appl. Phys. Lett., 52:437 (1988)

38.

S. Matsumoto, M. Hino, and T. Kobayashi, Synthesis of diamond films in a RF induction thermal plasma, Appl. Phys. Lett., 51:737 (1987)

39.

W. A. Yarbrough, and R. Roy, in Diamond and Diamond-like Materials Synthesis, Spring MRS Meeting Extended Abstracts, Symposium D, (lvf. W. Geis, G. H. Johnson, and A. R. Badzian, eds.), MRS, Pittsburgh, PA, p.33 (1988)

40.

K. Kitahama, K. Hirata, H. Nakamatsu, S. Kawai, N. Fujimori, T. Imai, H. Yoshino, and A. Doi, Synthesis of diamond by laser-induced chemical vapor deposition, Appl. Phys. Len., 49:634 (1986)

41.

F. G. Cehi, and J. E. Butler, Diamond chemical vapor deposition, Ann. Rev. Phys. Chem., 42643-684 (1991)

42.

N. Ohtake, and M. Yoshikawa, Diamond film preparation by arc discharge plasma jet chemical vapor deposition in the methane atmosphere, J. Electrochem. Sot., 137(2):717-722 (1990); N. Ohtake, and M. Yoshikawa, Nucleation effects and characteristics of diamond film grown by arc discharge plasma jet chemical vapor deposition, Thin Solid Films, 212(1-2):112-121 (1992)

43.

R. F. Davis, Diamond Films and Coatings: Development, Properties, and Applications, Noyes Publications, Park Ridge, New Jersey (1993)

44.

E. F. Borchelt, and G. Lu, Applications of diamond made by chemical vapor deposition for semiconductor laser submounts, in Proc. SPZE, 185164-77 (1993)

45.

D. C. Harris, Diamond: the ultimate durable infrared window material, Naval Res. Rev., 3:3-16 (1992)

46.

M. N. Yoder, Diamond: its impact on electronics, Naval Res. Rev., 3:17-21 (1992)

47.

K. E. Spear, and M. Frenklach, Mechanisms for CVD diamond growth, in Synthetic Diamond: Emerging CVD Science and Technologv, (K. E. Spear, and J. P. Dismukes, eds.), John Wiley & Sons, Inc. (1994)

References 163

48.

K. E. Spear, A. W. Phelps, and W.B. White, J. Muter. Rex, 5(11):2271-2285 (1990)

49.

M. Frenklach, and E. D. Feigelson, Astrophys. J., 341:372 (1989)

50.

S. Aisenberg, and R. Chabot, Ion-Beam deposition of thin films of diamondlike carbon, J. Appl. Phys., 42:2953-2958 (1971)

51.

C. Wild, R. Kohl, N. Herres, W. Mtiller-Sebert, and P. Koidl, Oriented CVD diamond films: twin formation, structure and morphology, Diamond Refut. Muter., 3:373-381 (1994)

52a.

B. Lux, and R. Haubner, Nucleation and growth of low-pressure diamond, in Diamond and Diamond-like Films and Coatings, @. E. Clausing, L. L. Horton, J. C. Angus, and P. Koidl, eds.), Plenum Press, New York, pp. 579-609 (1991)

52b.

P. 0. Joffreau, R. Haubner, and B. Lux, Int. J. Rej HardMater.,

7(4):186 (1988)

53.

J. C. Angus, M. Sunkara, S. R. Sahaida, and J. T. Glass, Twinning and faceting in early stages of diamond growth by chamical vapor deposition, J. Mater. Res., 7(11):3001-3009 (1992)

54.

W. A. Yarbrough, and R. Messier, Current Issues and problems in the chemical vapor deposition of diamond, Science, 247(4943):688-696 (1990)

55.

M. Sunkara, “Monte-Carlo simulation of diamond nucleation and growth,” Ph.D. Thesis, Case Western Reserve University, Cleveland, Ohio (1993)

56.

P. K. Bachmann, D. Leers, and H. Lydtin, Diamond Relut. Muter., l:l-12

57.

Y. Hirose, in First Intern. Conj on the New Diamond Sci. TechnoL, New Diamond Forum, Japan, p. 38 (1988)

58.

L. M. Hanssen, W. A. Carrington, J. E. Butler, and K. A. Snail,Muter. (1988)

59.

P. A. Molian, B. Janvrin, and A. M. Molian, Laser chemical vapor deposition of fluorinated diamond thin films for solid lubrication, Wear, 165(2):133-140 (1993)

60.

J. H. D. Rebello, D. L. Straub, and V. V. Subramaniam, Diamond growth from a CO/ CH, mixture by laser excitation of CO: laser excited chemical vapor deposition, J. Appl. Phys., 72(3):1133-1136 (1992)

61.

J. Krishnaswamy, A. Rengan, A. R. Srivatsa, G. Matera, and J. Narayan, Laser and plasma enhanced deposition of diamond and diamondlike films by physical and chemical vapor deposition techniques, in Proc. SPIE, 1190:109-l 17 (1989)

62.

Y. Goto, and H. Nagai, Synthesis of diamond films by laser-induced chemical vapor deposition, in Laser- and Particle-Beam Chemical Processes on S&aces Sjvnposium, (A. W. Johnson, G. L. Loper, and T. W. Sigmon, eds.), Pittsburgh, PA, MRS, pp. 213-218 (1989)

63.

E. Vickery, Process for epitaxial growth of diamonds, U.S. Patent 3,714 334, (1973)

64.

P. K. Bachmann, and H. Lydtin, in Diamond and Diamond-like Films and Coatings, (IX.E. Clausing, L. L. Horton, J. C. Angus, and P. Koidl, eds.), Plenum Press, New York, p. 829 (1991)

65.

C. P. Klages, Chemical vapor deposition of diamond, Appf. Phys. A, 56:513-526 (1993)

(1991)

Lett., 7:289

164 Diamond Chemical Vapor Deposition 66.

Z. P. Lu, J. Heberlein, and E. Pfender, Process study of thermal plasma chemical vapor deposition of diamond. I. Substrate material, temperature, and methane concentration, Plasma Chem. Plasma Proc., 12( 1):35-53 (1992)

67.

M. Kohzaki, K. Uchida, K. Higuchi, and S. Noda, Large-area high-speed diamond deposition by RF induction thermal plasma chemical vapor deposition method, Jpn. .I AppI. Phys., 32(3B):I438-440 (1993)

68.

N. G. Glumac, and D. G. Goodwin, Thin Solid Films, 212(1-2):122-126

69.

N. G. Glumac, and D. G. Goodwin, Mater. Lett., 18(3):119-122

70.

K. V. Ravi, Morphological instabilities in the low pressure synthesis of diamond, J. Mater. Rex, 7(2):384-393 (1992)

71.

S. D. Wolter, B. R. Stoner, J. T. Glass, P. J. Ellis, D. S. Buhaenko, C. E. Jenkins, and P. Southworth, Textured growth of diamond on silicon via in-situ carburization and bias-enhanced nucleation, Appf. Phys. Lett., 62(11):1215-1217 (1993)

72.

Y. Hayashi, W. Drawl, and R. Messier, Temperature dependence of nucleation density of chemical vapor deposition diamond, Jpn. J. Appl. Phys., 31(2B):L193196 (1992)

73.

K. F. Turner, B. R. Stoner, L. Bergman, J. T. Glass, and R. J. Nemanich, Observation of surface modification and nucleation during deposition of diamond on silicon by scanning tunneling microscopy, J. Appl. Phys., 69(9):6400-6405 (1991)

74.

S. W. Reeve, W. A. Weimer, and D. S. Dandy, Diamond growth using remote methane injection in a direct current arc jet chemical vapor deposition reactor, Appl. Phys. Lett., 63( 18):2487-2489 (1993)

75.

F. M. Cerio and W. A. Weimer, Rev. Sci. Instrum., 63(3):2065-2068

76.

R. L. Woodin, L. K. Bigelow, and G. L. Cann, in Applications of Diamond Films and Related Materials, (Y. Tzeng, M. Yoshikawa, M. Murakawa, and A. Feldman, eds.), Elsevier Science Publishers, Amsterdam, pp. 439-444 (1991)

77.

G. A. Raichie, G. P. Smith, and J. B. Jeffries, in New Diamond Sci. Technol., @. Messier, J. T. Glass, J. E. Butler, and R. Roy, eds.), MRS, Pittsburgh, PA., pp. 25 I256 (1991)

78.

M. Murayama, S. Kojima, and K. Uchida, J. Appf. Phys., 69(1 I):7924 (1991)

79.

J. A. Miller, and C. T. Bowman, Prog. Energy Combust. Sci., 15:287 (1989)

80.

J. A. Miller, and C. F. Melius. Combust. Flame, 91:21 (1991)

81.

G. P. Smith, and J. B. Jeffries, in Diamond and Diamond-like Films, (A. J. Purdes, J. C. Angus, R. F. Davis, B. M. Meyerson, K. E. Spear, and M. Yoder, eds.), Electrochemical Sot., Pennington, NJ., 91-8:194 (1991)

82.

M. E. Coltrin, and D. S. Dandy, J. AppI. Phys., 74(9):5803-5820

83.

E. Meeks, R. J. Kee, D. S. Dandy, and M. E. Coltrin, Combustion and Flame, 92(12):144-160 (1993)

84.

M. Tsuda, M. Nakajima, and S. Oikawa, J. Amer. Chem. Sot., 108:5780 (1986)

85.

M. Tsuda, M. Nakajima, and S. Oikawa, Jpn. J. Appl. Phys., 26:L527 (1987)

86.

M. Frenklach, J. Appf. Phys., 65:5142 (1989)

(1992)

(1993)

(1992)

(1993)

References

165

87.

M. Frenklach, and H. Wang, Phys. Rev. B, 43:1520 (1991)

88.

M. Frenklach, and K. Spear, J. Mater. Res., 3:133 (1988)

89.

D. Huang, M. Frenklach, and M. Maroncelli, J. P&s.

90.

D. Huang, and M. Frenklach, J. Phys. Chem., 95:3692 (1991)

91.

D. Huang, and M. Frenklach, J. Phys. Chem., 96:1868 (1992)

92.

M. Frenklach, J. Chem. Phys., 97:5794 (1992)

93.

M. Frenklach, Phys. Rev. B, 45:9455 (1992)

94.

S. J. Harris, J. A&.

95.

S. J. Harris, D. N. Belton, A .M. Weiner, and S. J. Schmieg,J. (1989)

96.

D. N. Belton, S. J. Harris, S. J. Schmieg, A. M. Weiner, and T. A. Perry, In situ characterization of diamond nucleation and growth, Appl. Phys. Left., 54(5):416417 (1989)

97.

S. J. Harris, and A. M. Weiner, J. Appl. Phys., 67:6520 (1990)

98.

S. J. Harris, Appf. Phys. Letf., 56:2298 (1990)

99.

S. J. Harris, and D. N. Belton, Thin Solid Films, 212:193 (1992)

100.

Chem., 92:6379 (1988)

Phys., 65:3044 (1989) Appl. Phys., 66:5353

D. N. Belton, and S. J. Harris, J. Chem. Phys., 96:2371 (1992)

101.

S. J. Harris, and D. G. Goodwin, J. Chem. Phys., 97:23 (1993)

102.

B. W. Yu, and S. L. Girshick, J. AppZ. Phys., 75:3914 (1994)

103.

B. J. Garrison, E. J. Dawnkaski, 255:835 (1992)

D. Srivastava,

and D. W. Brenner,

104.

D. S. Dandy, and M. E. Coltrin, J. Appl. Phys., 76:3102 (1994)

105.

Y. Hirose, and Y. Terasawa, Jpn. J. Appl. Phys., 25:L519 (1986)

Science,

106.

A. Sawabe, and T. Inuzuka, Thin Solid Films, 137:189 (1986)

107.

W. Banholzer, and R. Kehl, Surf: Coat. Technol., 47:51 (1991)

108.

K. Suzuki, A. Sawabe, and T. Inuzuka, Growth of diamond thin films by DC plasma chemical vapor deposition and characteristics of the plasma, Jpn. J. Appl. Phys., 29(1):153-157 (1990)

109.

K. Suzuki, S. Okudaira, N. Sakudo, and I. Kanomata, Jpn. J. Appl. Phys., 16:ll (1979)

110.

B. Chapman, Glow Discharge Processes, John Wiley, New York (1980)

111.

J. Wei, H. Kawarada, J. Suzuki, J. S. Ma, and A. Hiraki, Effects of plasma potential on diamond deposition at low pressure using magneto-microwave plasma chemical vapor deposition, Jpn. J. Appl. Phys., 30(6):1279-1280 (1991)

112.

G. J. Vandentop, P. A. P. Nascente, M. Kawasaki, D. F. Ogletree, G. A. Somojai, and M. Salmeron, J. Vat. Sci. Technol. A, 9:2273 (1991)

113.

J. C. Angus, F. A. Buck, M. Sunkara, T. F. Groth, C. C. Hayman, and R.Gat, URS Bull., 14(10):38 (1989)

166 Diamond Chemical Vapor Deposition 114.

S. Matsumoto, Y. Manabe, and Y. Hibino, Diamond deposition using an X-Y stage in a D plasma jet chemical vapour deposition, J. Muter. Sci., 27(21):5905-5910 (1992)

115.

N. Fujimori, New Diamond,

116.

S. Matsumoto, and K. Okada, Modified thermal-plasma chemical vapor deposition of diamond films and the observation of surface morphology, Oyo Buturi, 61(7):726730 (1992)

117.

X. Chen, and J. Narayan, Effect of the chemical nature of transition-metal substrates on chemical-vapor deposition of diamond, J. Appl. Phys., 74 (6):4 168-4 173 (1993)

118.

Z. P. Lu, J. Heberlein, and E. Pfender, Process study of thermal plasma chemical vapor deposition of diamond. II. Pressure dependence and effect of substrate pretreatment, Plasma Chem. Plasma Proc., 12( 1):55-69 (1992)

119.

A. Hirata, and M. Yoshikawa, Enlargement of the diamond deposition area by onecathode three-anode arc discharge plasma jet chemical vapour deposition, Diumend R&t. Muter., 2(11):1402-1408 (1993)

120.

K. Eguchi, S. Yata, and T. Yoshida, Uniform and large-area deposition of diamond by cyclic thermal plasma chemical vapor deposition, Appl. Phys. Lett., 64(1):5860 (1994)

121.

I. Watanabe, T. Matsushita, and K. Sasahara, Low temperature synthesis of diamond films in thermoassisted RF plasma chemical vapor deposition, Jpn. J. AppZ. Phys., 31(5A):1428-431 (1992)

122.

P. G. Kosky, and D. S. McAtee, Muter. Lett., 8:369 (1989)

123.

M. A. Cappelli, and P. H. Paul, J. Appl. Phys., ,67:2596 (1990)

124.

K. V. Ravi, and C. A. Koch, Nucleation enhancement of diamond synthesized combustion flame techniques, Appl. Phys. Lett., 57(4):348-50 (1990)

125.

Y. Tseng, C. Cutshaw, R. Phillips, and T. Srivinyunon, Appl. Phys. Lett., 56:134 (1990)

126.

D. B. Oakes, and J. E. Butler,J.

127.

Y. Matsui,A.

128.

Y. Matsui, H. Yabe, and Y. Hirose, Jpn. J. Appf. Phys., 29:1552 (1990)

129.

Y. Matsui, H. Yabe, T. Sigimoto, and Y. Hirose, Diamond Relat. Muter., 1:19 (1991)

130.

J. A. Cooper, and W. A. Yarbrough, Diamond Optics, SPIE, 1325:41 (1990)

Yuuki,M.

2(2):10-15

(1988)

by

Appl. Phys., 69:2602 (1991)

Sahara, and Y. Hirose, Jpn. J. Appl. Phys.,28:1718

(1989)

13 1. S. Jin, The mechanisms and implications of CVD diamond etching by chemical reactions with metals, Presentation at Gordon Research Conference on Cm Diamond Synthesis, Plymouth, (June 19-24, 1994) 132.

A. B. Harker, Reactive polishing of polycrystalline diamond and measured spectroscopic properties, in Applications of Diamond Films and Related Materials, (Y. Tzeng, M. Yoshikawa, M. Murakawa, and A. Feldman, eds.), Elsevier Science Publishers, Amsterdam, pp. 223-225 (1991)

References

167

133.

A. Hirata, H. Tokura, and M. Yoshikawa, Smoothing of diamond films by ion beam irradiation, in Applications ofDiamond Films and RelatedMaterials, (Y. Tzeng, M. Yoshikawa, M. Murakawa, and A. Feldman, eds.), Elsevier Science Publishers, Amsterdam, pp. 227-232 (1991)

134.

H. Tokura, and M. Yoshikawa, Polishing of CVD diamond film, in Applications of Diamond Films and Related Materials, (Y. Tzeng, M. Yoshikawa, M. Murakawa, and A. Feldman, eds.), Elsevier Science Publishers, Amsterdam, pp. 241-248 (1991)

135.

V. N. Tokarev, and V. I. Konov, Light-induced polishing of diamond films, in Applications of Diamond Films and Related Materials, (Y. Tzeng, M. Yoshikawa, M. Murakawa, and A. Feldman, eds.), Elsevier Science Publishers, Amsterdam, pp. 249-255 (1991)

136.

K. Kobashi, K. Nishimura, (1988)

137.

K. Kobashi, K. Nishimura, K. Miyata, K. Kumagai, and A. Nakaui, (1 lO>oriented diamond films synthesized by microwave chemical-vapor deposition, J. Mater. Res., 5(11):2469-2482 (1990)

138.

A. R. Badzian, in Advances in X-Ray Analysis, (C. S. Barrett, J. V. Gilfrich, R. Jenkins, J. C. Russ, J. W. Richardson, and P. K. Predecki, eds.), Plenum Press, New York, 31:113-128 (1988)

139.

B. V. Spitzyn, and L. L. Bouilov, in Diamond and Diamond-like Materials Synthesis, Spring A4RS Meeting Extended Abstracts, Symposium D, @I. W. Geis, G. H. Johnson, and A. R. Badzian, eds.), MRS, Pittsburgh, PA, pp. 3-14 (1988)

140.

B. V. Spitzyn, L. L. Bouilov, and B. V. Dejaguin, 17:79 (1988)

141.

R. A. Rudder, J. B. Posthill, G. C. Hudson, D. P. Malta, R. E. Thomas, R. J. Markunas, T. P. Humphreys, and R. J. Nemanich, Chemical vapor deposition of diamond films using water:alcohol:organic-acid solutions, in Wide Band Gap Semiconductors Symposium, (T. D. Moustakas, J. I. Pankove, and Y. Hamakawa, eds.), Pittsburgh, PA, MRS, pp. 23-30 (1992)

142.

R. A. Rudder, G. C. Hudson, J. B. Posthill,R. E. Thomas,R. C. Hendry, D. P. Malta, R. J. Markunas, T. P. Humphreys, and R. J. Nemanich, Chemical vapor deposition of diamond films from water vapor RF-plasma discharges, Appl. Phys. Lett., 60(3):329-33 1 (1992)

143.

C. E. Johnson, and W. A. Weimer, Identification of hydrocarbon precursors to diamond in chemical vapor deposition using carbon monoxide reagent, J. Mater. Res., 8(9):2245-2249 (1993)

144.

K. Hirabayashi, and S. Matsumoto, Flattened diamond crystals synthesized by microwave plasma chemical vapor deposition in a CO-H, system, J. Appl. Phys., 75(2):1151-1154 (1994)

145.

C. R. Eddy, Jr., D. L. Youchison, and B. D. Sartwell, Comparison of (CH,-0,)-H, and CO-H, plasmas for low temperature diamond film deposition by electron cyclotron resonance plasma-assisted chemical vapor deposition, Diamond Relat. Mater., 3(1-2):105-111 (1994)

Y. Kawate, and T. Horiuchi, Whys. Rev. B, 38:4067

prog. Cqutal Growth Charact.,

168

Diamond Chemical Vapor Deposition

146.

M. Komori, T. Maki, T. Kim, G. Hou, Y. Sakaguchi, K. Sakuta, and T. Kobayashi, Homoepitaxial growth of diamond thin films by electron cyclotron-resonance microwave plasma chemical-vapor-deposition apparatus with CO/H, gaseous source, Appl. Phys. Lett., 62(6):582-584 (1993)

147.

R. J. Graham, J. B. Posthill, R. A. Rudder, and R. J. Markunas, Cathodoluminescence from diamond films grown by plasma-enhanced chemical vapor deposition in dilute CO/H,, CF,/H,, and CH,/H, mixtures, Appl. Phys. Lett., 59( 19):2463-2465 (1991)

148.

C. Chen, and T. Hong, The role of hydrogen in diamond synthesis from carbon dioxide-hydrocarbon gases by microwave plasma chemical vapor deposition, SU$ Coat. TechnoL, 54-55(1-3):368-373 (1992)

149.

S. Katsumata, Diamond synthesis by the microwave plasma chemical vapor deposition method using the pretreated carbon dioxide and hydrogen mixed-gas system, Jpn. J. Appl. Phys., 31(3):868-871 (1992)

150.

F. C. Hong, G. T. Liang, J. J. Wu, D. Chang, and J. C. Hsieh, Diamond deposition from halogenated methane reactants in a hot-filament chemical vapor deposition reactor, Appl. Phys. Lett., 63(23):3149-3151 (1993)

15 1. R. A. Rudder, G. C. Hudson, J. B. Posthill, R. E. Thomas, and R. J. Marcunas, Appl. Phys. Lett., 59:791 (1991) 152.

T. Tsuno, T. Tomikawa, S. Shikata, T. Imai, and N. Fujimori, Diamond (001) single-domain 2x1 surface grown by chemical vapor deposition, Appl. Phys. Lett., 64(5):572-574 (1994)

153.

J. Mercier, A. M. Bonnot, E. Caignol, and E. Gheeraert, Investigations on growth mechanisms of diamond thin films by hot filament assisted CVD, in Diamond and Diamond-like Films and Coatings, (R. E. Clausing, L. L. Horton, J. C. Angus, and P. Koidl, eds.), Plenum Press, New York, pp. 533-539 (1991)

154.

D. E. Patterson, B. J. Bai, C. J. Chu, R. H. Hauge, and J. L. Margrave, in New Diamond Sci. Technob, (R. Messier, J. T. Glass, J. E. Butler, and R. Roy, eds.), MRS, Pittsburgh, PA., pp. 433438 (1991)

155.

T. Nagano and N. Shibata, Diamond synthesis by microwave-plasma chemical vapor deposition using CH,Cl and CH,CI, as carbon source, Jpn. J. Appf. Phys., 32(11A):5067-5071 (1993)

156.

P. Bou, L. Vandenbulcke, R. Herbin, and F. Hillion, Diamond film nucleation and interface characterization, J. Muter. Res., 7(8):215 l-2159 (1992)

157.

D. Kim, H. Lee, and J. Lee, Effect of reaction pressure on the nucleation behaviour of diamond synthesized by hot-filament chemical vapour deposition, J. Mater. Sci., 28(24):6704-6708 (1993)

158.

S. Ojika, S. Yamashita, K. Kataoka, T. Ishikura, A. Yamaguchi, and H. Kawarada, Low-pressure diamond nucleation and growth on Cu substrate, Jpn. J. Appl. Phys., 32(2A):L200-203 (1993)

159.

R. Brunsteiner, R. Haubner, and B. Lux, Hot-filament chemical vapour deposition of diamond on SiAlON at pressures up to 500 Ton, Diamond Relat. Mater., 2(9):1263-1270 (1993)

References 160.

169

H. Matsubara, and T. Sakuma, Diamond deposition on cemented carbide by chemical vapour deposition using a tantalum filament, J. Mater. Sci., 25 (10):44724476 (1990)

16 1. F. Akatsuka, and F. Shimura, Effect of substrate materials on CVD diamond growth, in Applications ofDiamond Films and RelatedMaterials, (Y. Tzeng, M. Yoshikawa, M. Murakawa, and A. Feldman, eds.), Elsevier Science Publishers, Amsterdam, pp. 379-381 (1991) 162.

Y. Liou, A. Inspektor, R. Weimer, and R. Messier, Low-temperature diamond deposition by microwave plasma-enhanced chemical vapor deposition, Appl. Phys. Lett., 55(7):631-633 (1989)

163.

R. Haubner, A. Lindlbauer, and B. Lux, Diamond deposition on chromium, cobalt and nickel substrates by microwave plasma chemical vapour deposition, Diamond Relat. Mater., 2(12):1505-1515 (1993)

164.

H. Chen, M. L. Nielsen, C. J. Gold, R. 0. Dillon, J. DiGregorio, and T. Furtak, An investigation of diamond growth on ferrous metals, in Applications of Diamond Films and Related Materials, (Y. Tzeng, M. Yoshikawa, M. Murakawa, and A. Feldman, eds.), Elsevier Science Publishers, Amsterdam, pp. 137-142 (1991)

165.

S. M. Kanetkar, A. A. Kulkami, A. Vaidya, R. D. Vispute, S. B. Ogale, S. T. Kshirsagar, and S. C. Purandare, Diamond nucleation on epitaxially grown Y-ZrQ, layers on Si(lOO), Appf. Phys. Lett., 63(6):740-742 (1993)

166.

P. C. Yang, W. Zhu, and J. T. Glass, Nucleation of oriented diamond films on nickel substrates, J. Mater. Rex, 8(8):1773-1776 (1993)

167.

G. Popovici, and M. A. Prelas, Nucleation and selective deposition of diamond thin films, Physica Status Solidi A, 132(2):233-252 (1992)

168.

S. Park, and J. Lee, Synthesis of diamond films on titanium substrates by hotfilament chemical vapor deposition, J. Appl. Phys., 69(4):2618-2622 (199 1)

169.

H. C. Shih, W. T. Hsu, C. T. Hwang, C. P. Sung, L. K. Lin, and C. K. Lee, Application of nickel plating for the synthesis of chemical vapour deposition of diamond on steels, Thin Solid Films, 236(1-2):ll l-l 14 (1993)

170.

G. F. Zhang, X. Zheng, and Z. T. Liu, Nucleation and growth behaviour of diamond films deposited on chemically etched titanium alloy substrates, J. Cvsf. Growth, 133(1-2):117-122 (1993)

t

17 1. D. K. Sood, W. R. Drawl, and R. Messier, The effect of carbon ion implantation on the nucleation of diamond on Ti-6Al-4V alloy, Surf: Coat. Technol., 51(1-3):307312 (1992) 172.

M. T. McClure, J. A. von Windheim, J. T. Glass, and J. T. Prater, Effect of native SiO, layer on the nucleation of diamond using a combustion flame, Diamond Relat. Mater., 3(3):239-244 (1994)

173.

Z. Ajji, M. Buck, and C. Wall, Investigation of nucleation centres in diamond chemical vapour deposition with spatially resolved x-ray photoelectron spectroscopy, Diamond Refat. Mater., 2(12):1525-1528 (1993)

174.

R. A. Bauer, N. M. Sbrockey, and W. E. Brower, Jr., Quantitative nucleation and growth studies of PACVD diamond film formation on (100) silicon, J. Mater. Res., 8(11):2858-2869 (1993)

170 Diamond Chemical Vapor Deposition 175.

P. N. Barnes, and R. L. C. Wu, Nucleation amorphous films, Appl. Phys. Lett., 62(1):37-39

enhancement (1993)

of diamond

with

176.

E. Molinari, R. Polini, and M. Tomellini, Diamond crystallite formation on Si(100) from the gas phase: Seeding or heterogeneous nucleation?, Appl. Phys. Lett., 61(11):1287-1289 (1992)

177.

H. Maeda, S. Masuda, K. Kusakabe, and S. Morooka, Nucleation and growth of diamond in a microwave plasma on substrate pretreated with non-oxide ceramic particles, J. C@. Growth, 121(3):507-515 (1992)

178.

B. R. Stoner, G. H. M. Ma, S. D. Wolter, and J. T. Glass, Characterization of biasenhanced nucleation of diamond on silicon by in vucuo surface analysis and transmission electron microscopy, Phys. Rev. B, 45( 19): 11067-l 1084 (1992)

179.

C. M. Niu, G. Tsagaropoulos, J. Baglio, K. Dwight, and A. Wold, Nucleation and growth of diamond on Si, Cu, and Au substrates, J. SolidState Chemistry, 91(1):4756 (1991)

180.

A. A. Smolin, V. G. Ralchenko, S. M. Pimenov, T. V. Kononenko, and E. N. Loubnin, Optical monitoring of nucleation and growth of diamond films, Appl. Phys. Lett., 62(26):3449-3451 (1993)

18 1. P. K. Bachmann, W. Drawe, D. Knight, R. Weimer, and R. Messier, Diamond and Diamond-like Materials Synthesis, in Diamond and Diamond-like Materials Synthesis, Spring MRS Meeting Extended Abstracts, Symposium D, (M. W. Geis, G. H. Johnson, and A. R. Badzian, eds.), MRS, Pittsburgh, PA, pp. 99-102 (1988) 182.

J. C. Angus, Z. Li, M. Sunkara, R. Gat, A. B. Anderson, S. P. Mehandru, and M. W. Geis, Nucleation and growth processes in chemical vapor depostion of diamonds, Electrochem. Sot. Symp. Ser., Pennington, New Jersey, Electrochemical Society (1991)

183.

J. C. Angus, Y. Wang, and M. Sunkara, Metastable diamondlike phases, Ann. Rev. Mater. Sci., 21:221-248

184.

X. Jiang, K. Schiffmann, A. Westphal, and C. P. Klages, Atomic-force-microscopic study of heteroepitaxial diamond nucleation on (100) silicon, Appf. Phys. Lett., 63(9):1203-1205 (1993)

185.

L. Chang, Z. Lin, J. Chen, and F. Chen, Transmission electron microscopy study of diamond nucleation on 6H-Sic single crystal with possibility of epitaxy, Appl. Phys. Lett., 62(26):3444-3446 (1993)

186.

B. R. Stoner, S. R. Sahaida, J. P. Bade, P. Southworth, and P. J. Ellis, Highly oriented, textured diamond films on silicon via bias-enhanced nucleation and textured growth, J. Mater. Res., 8(6):1334-1340 (1993)

187.

S. D. Wolter, B. R. Stoner, G. H. M. Ma, and J. T. Glass, In-vacua surface analytical study of diamond nucleation on copper vs. silicon, in Novel Forms of Carbon Sjmposium, (C. L. Renschler, J. J. Pouch, and D. M. Cox, eds.), Pittsburgh, PA, MRS, pp. 347-352 (1992)

188.

B. R. Stoner, B. E. Williams, S. D. Wolter, K. Nishimura, and J. T. Glass, In-situ growth rate measurement and nucleation enhancement for microwave plasma CVD of diamond, .I. Mater. Res., 7(2):257-260 (1992)

growth (1991)

of diamond

and

References

171

189.

R. Csencsits, J. Rankin, R. E. Boekenhauer, M. K. Kundmann, and B. W. Sheldon, Diamond nucleation during biased chemical vapor deposition, in Evolution of Surface and Thin Film Microstructure Symposium, (H. A. Atwater, E. Chason, M. H. Grabow, and M. G. Lagally, eds.), Pittsburgh, PA, MRS, pp. 695-700 (1993)

190.

M. Katoh, M. Aoki, and H. Kawarada, Plasma-enhanced Jpn. J. Appl. Phys., 33(2A):L194-L196 (1994)

191.

R. Kohl, C. Wild, N. Herres, P. Koidl, B. R. Stoner, and J. T. Glass, Oriented nucleation and growth of diamond films on g-Sic and Si, Appl. Phys. Lett., 63(13):1792-1794 (1993)

192.

R. J. Meilunas, and R. P. H. Chang, Diamond nucleation on surfaces using carbon clusters, J. Mater. Res., 9( 1):61-79 (1994)

193.

R. J. Meilunas, R. P. H. Chang, S. Liu, and M. M. Kappes, Nucleation of diamond films on surfaces using carbon clusters, Appl. Phys. Lett., 59(26):3461-3463 (1991)

194.

R. Hultgren, P. D. Desai, D. T. Hawkins, M. Gleiser, and K. K. Kelley, Selected Values of the Thermodynamic Properties of Binary Alloys, American Society for Metals, Metals Park, OH (1973)

195.

R. Kern, in Current Topics in Materials Science, (E. Kaldis, ed.), North-Holland, Amsterdam, 1283-l 17 (1985)

196.

E. S. Machlin, Stabilization 3(5):958-968 (1988)

197.

J. Shackelford, and W. Alexander, The CRC Materials Science and Engineering Handbook, CRC Press, Boca Raton (1992)

198.

E. A. Brandes, Smithells Metals Reference Book, Butterworth ed. (1983)

199.

T. P. Ong, F. Xiong, R. P. H. Chang, and C. W. White, Nucleation and growth of diamond on carbon-implanted single crystal copper surfaces, J. Mater. Res., 7(9):2429-2439 (1992)

200.

H. Jones, The surface energy of solid metals, Metal Sci. J., 5:15-18.L.Z. (1971); Mezey, and J. Giber, The surface free energies of solid chemical elements: calculation from internal free enthalpies of atomization, Jpn. J. Appl. Phys., 21(11):1569-1571 (1982)

of diamond

diamond nucleation on Si,

relative to graphite,

J. Mater. Res.,

& Co, London, 6th

20 1. P. T. B. Shaffer, Plenum Press Handbooks of High-Temperature Materials, No. lMaterials Index, Plenum Press, New York (1964); G. V. Samsonov, Plenum Press Handbooks ofHigh-Temperature Materials, No. 2-Properties Index, Plenum Press, New York (1964) 202.

G. V. Samsonov, The Oxide Handbook, IFUPlenum Data Corporation, New York (1973)

203.

J. R. Davis, P. Allen, S. R. Lampman, T. B. Zorc, S. D. Henry, J. L. Daguila, A. W. Ronke, J. Jakel, K. L. O’Keefe, and R. L. Stedfeld, Metals Handbook, vol. 2, Properties and Selection: Nonferrous Alloys and Special Purpose Materials, 10th ed., ASM International (1990)

204.

W. Zhu, C. A. Randall, A. R. Badzian, and R. Messier, J. Vat. Sci. Technol., A7:2315 (1989)

172

Diamond Chemical Vapor Deposition

205.

T. P. Ong, and R. P. H. Chang, Low-temperature deposition of diamond films for optical coatings, Appl. Phys. Letf., 55(20):2063-2065 (1989)

206.

Y. H. Lee, P. D. Richard, K. J. Bachmann, and J. T. Glass, Bias-controlled chemical vapor deposition of diamond thin films, Appl. Phys. Lett., 56(7):620-622 (1990)

207.

A. Hiraki, H. Kawarada, J. Wei, and J. I. Suzuki, Preparation and characterization of wide area, high quality diamond film using magnetoactive plasma chemical vapor deposition, Sud Coat. Technol., 43-44(1-3):10-21 (1990)

208.

R. J. Graham, T. D. Moustakas, and M. M. Disko, Cathodoluminescence imaging of defects and impurities in diamond films grown by chemical vapor deposition, J. Appl. Phys., 69(5):3212-3218 (1991)

209.

R. Lecher, C. Wild, N. Herres, D. Behr, and P. Koidl, Nitrogen stabilized texture in chemical vapor deposited diamond films, Appl. Phys. Lett., 65(1):34-36 (1994)

210.

Y. Yokota, H. Kawarada, and A. Hiraki, MRS Symp. Proc., 162:231 (1990)

211.

S. I. Shah, and M. M. Waite, Effect of oxygen on the nucleation and growth of diamond thin films, Appl. Phys. Lett., 61(26):3113-3115 (1992)

212.

M. M. Waite, and S. I. Shah, X-ray photoelectron spectroscopy of initial stages of nucleation and growth of diamond thin films during plasma assisted chemical vapor deposition, Appl. Phys. Lett., 60( 19):2344-2346 (1992)

213.

J. Ruan, W. J. Choyke, and K. Kobashi, Oxygen-related centers in chemical vapor deposition of diamond, Appl. Phys. Lett., 62(12):1379-1381 (1993)

214.

S. A. Stuart, and S. Prawer, TEM and CBED observation of defects in isolated diamond crystals synthesized by microwave plasma chemical vapor deposition on tungsten wire tips, J. Computer-Assisted Microscopy, 4(2):201-204 (1992)

215.

T. D. Moustakas, Nucleation and growth of diamond thin films, in 20th Int. Co& on the Phys. of Semiconductors, Thessaloniki, Greece, 6-10, Aug., 1990, (E. M. Anastassakis, and J. D. Joannopoulos, eds.), Singapore, World Scientific, 1:320327 (1990)

216.

A. T. Collins, M. Kamo, and Y. Sato, Optical centres related to nitrogen, vacancies and interstitials in polycrystalline diamond films grown by plasma-assisted chemical vapour deposition, J. Phys. D: Appl. Phys., 22(9):1402-1405 (1989)

217.

M. Tomellini, R. Polini, and V. Sessa, A model kinetics for nucleation at a solid surface with application to diamond deposition from the gas phase, J. Appl. Phys., 70(12):7573-7578 (1991)

218.

S. Mitura, J. Cv.rl. Growth, SO:417 (1987)

219.

W. Howard, D. Huang, J. Yuan, M. Frenklach, K. E. Spear, R. Koba, and A. W. Phelps, Synthesis of diamond powder in acetylene oxygen plasma, J. Appl. Phys., 68(3):1247-1251 (1990)

220.

M. Frenklach, W. Howard, D. Huang, J. Yuan, K. E. Spear, and R. Koba, Induced nucleation of diamond powder, Appl. Phys. Left., 59(5):546-548 (1991)

221.

M. Frenklach, R. Kematick, D. Huang, W. Howard, K. E. Spear, A. W. Phelps, and R. Koba, Homogeneous nucleation of diamond powder in the gas phase, .J. Appl. Phys., 66(1):395-399 (1989)

References

and Y. Matsui, J. Mater. Sci., 18:1785-1793

173

222.

S. Matsumoto,

223.

S. A. Godleski, P. V. Schleyer, E. Osawa, and W. T. Wipke, Prog. Phys. Org. Chem., 13:63 (1981)

(1983)

224.

S. E. Stein, Nature, 346(6284):517

225.

J. C. Angus, R. W. Hoffman, and P. H. Schmidt, in Sci. Technol. ofNew Diamond, (S. Saito, 0. Fukunaga, and M. Yoshikawa, eds.), KTK Terra Scientific, Tokyo, pp. 9-16 (1990)

226.

J. C. Angus, J. A. Mann, Y. Wang, M. Sunkara, P. A. Washlock, M. Kuczmarski, and Z. Li, Progress Report, DARPA University Research Initiation Grant, Contract #NOOO14-895-3167 (Jan. 31, 1991)

227.

S. E. Stein, J. Phys. Chem., 825-566 (1978)

(1990)

228.

S. E. Stein, and A. Fahr, J. Phys. Chem., 89:3714 (1985)

229.

S. E. Stein, Accounts Chem. Res., 24(11):350 (1991)

230.

S. E. Stein, and R. L. Brown, .I American Chem. Sot., 113(3):787 (1991)

23 1. G. A. Raiche, and J. B. Jeffries, Carbon, 28:796 (1990) 232.

P. R. Buerki, and S. Leutwyler, Homogeneous nucleation of diamond powder by COs-laser-driven gas-phase reactions, J. Appl. Phys., 69(6):3739-3744 (1991)

233.

Z. Ajji, M. Buck, H. Schwenk, and C. Wall, Diamond nucleation by seeding from the gas phase, Presentation in 41st National Symposium and 3rd International Conference on Nanometer-Scale Science & Technology, p. 137, American Vacuum Society, Denver, Colorado (Oct. 24-28, 1994)

234.

E. Kondoh, K. Tanaka, and T. Ohta, Homoepitaxial advanced hot-filament chemical vapor deposition 74(3):2030-2035 (1993)

235.

M. P. D’Evelyn, C. J. Chu, R. H. Hange, and J. L. Margrave, Mechanism of diamond growth by chemical vapor deposition: carbon-13 studies, J. AppI. Phys., 71(3):1528-1530 (1992)

236.

C. J. Chu, M. P. D’Evelyn, R. H. Hauge, and J. L. Margrave, Mechanism of diamond growth by chemical vapor deposition on diamond (loo), (11 l), and (110) surfaces: carbon-13 studies, J. AppI. Phys., 70(3):1695-1705 (1991)

237.

Z. M. Zhang, H. M. Cheng, S. H. Li, Q. Y. Cai, D. L. Ling, S. J. Wang, Z. W. Hu, S. S. Jiang, C. Z. Ge, and N. B. Ming, Morphologic study of (100) single crystal diamond films epitaxially grown by chemical vapor deposition, J. Cryst. Growth, 132(1-2):200-204 (1993)

238.

S. Katsumata, Cathodoluminescence from epitaxial diamond layer grown by plasma-assisted chemical vapor deposition on high-pressure synthetic diamond, Jpn. J. AppI. Phys., 31(11):3594-3597 (1992)

239.

W. A. Yarbrough, J. Vacuum Sci. Technof. A, 9:1145 (1991)

240.

W. P. Chai, Y. S. Gu, M. Li, Z. H. Mai, Q. Z. Li, L. Yuan, and S. J. Pang, Orientation influence of cubic boron nitride crystal facets on the epitaxial growth of diamond film by microwave plasma chemical vapor deposition, J. Cyst. Growth, 135(34):639-642 (1994)

growth of diamond by an method, J. Appl. Phys.,

174

Diamond Chemical Vapor Deposition

24 1.

S. Koizumi, T. Murakami, T. Inuzuka, and K. Suzuki, Epitaxial growth of diamond thin films on cubic boron nitride (111) surfaces by DC plasma chemical vapor deposition, Appl. Phys. Lett., 57(6):563-565 (1990)

242.

M. Yoshikawa, Diamond Films and Technol., 1:l (1991)

243.

S. J. Harris, A. M. Weiner, and T. A. Perry, J. Appl. Phys., 70:1385 (1991)

244.

W. A. Yarbrough, A chemical perspective on the nucleation and growth of diamond from hydrocarbons, in Applications of Diamond Films and Related Materials, (Y. Tzeng, M. Yoshikawa, M. Murakawa, and A. Feldman, eds.), Elsevier Science Publishers, Amsterdam, pp. 25-33 (1991)

245.

C. Wild, N. Herres, and P. Koidl, Texture formation in polycrystalline films, J. AppJ. Phys., 68(3):973-978 (1990)

246.

A. van der Drift, Evolutionary selection, a principle governing growth orientation in vapor-deposited layers, Philips Res. Repts, 221267-288 (1967)

247.

R. C. Hyer, M. Green, K. K. Mishra, and S. C. Sharma, Nucleation and growth of diamond in hot filament assisted chemical vapour deposition, J. Mater. Sci. Lett., 10(9):515-518 (1991)

248.

A. A. Smolin, S. M. Pimenov, V. G. Ralchenko, T. V. Kononenko, V. I. Konov, and E. N. Loubnin, In-situ laser interferometry characterization of nucleation and growth of diamond films, Diamond Films and Technol., 3( 1): l-l 5 (1993)

249.

C. Trevor, D. Cherns, and P. Southworth, TEM studies of the nucleation of CVD diamond on silicon, in Proc. Inst. Phys. Electron Microscopy and Analysis Group ConJ, (F. J. Humphreys, ed.), Bristol, UK, IOP, pp. 275-278 (1991)

250.

K. Tamaki, Y. Watanabe, Y. Nakamura, and S. Hirayama, Nucleation of diamond particles by hot filament chemical vapour deposition, Thin Solid Films, 236(12):115-119 (1993)

251.

R. Ramesham, and C. Ellis, Selective nucleation of diamond crystals on the apex of silicon pyramids, in Wide Band Gap Semiconductors Symposium, (T. D. Moustakas, J. I. Pankove, and Y. Hamakawa, eds.), Pittsburgh, PA, MRS, pp. 6977 (1992)

252.

Z. Feng, K. Komvopoulos, I. G. Brown, and D. B. Bogy, Effect of graphitic carbon films on diamond nucleation by microwave-plasma-enhanced chemical-vapor deposition, J. Appl. Phys., 74(4):2841-2849 (1993)

253.

X. Peng, and H. Li, Nucleation and growth observations of diamond grown on cast tungsten carbide (W,C+WC) particles by low-pressure CVD, Mater. Left., 17(5):223-226 (1993)

254.

G. A. I&rata, L. Cota-Araiza, M. Avalos-Boja, M. H. Farias, 0. Contreras, and Y. Matsumoto, Nucleation and growth of diamond films on u c-SiCx-Si by hot-filament CVD, J. Phys., 5(33A):A305-306 (1993)

diamond

255a.

J. Singh, Nucleation and growth mechanism of diamond during hot-filament chemical vapour deposition, J. Mater. Sci., 29( 10):2761-2766 (1994)

255b.

J. Singh, and M. Vellaikal, Surj Coat. Technof., 64:131 (1994)

256.

D. N. Belton, and S. J. Schmieg, Nucleation of chemically vapor deposited diamond on platinum and nickel substrates, Thin Solid Films, 212(1-2):68-80 (1992)

References

175

257.

G. Casalone, M. G. Cattania, M. Simonetta, and M. Tescari, Surf: Sci., 62:321 (1977)

258.

G. Ertl, and J. Kuppers, Low Energy Electrons and Surface Chemistty, Chernie, Weinheim (1974)

259.

B. Y. Lin, C. P. Beetz, D. W. Brown, and B. A. Lincoln, The influence of diamond surface perfection on the preferential nucleation of sp* carbon during methane pyrolysis, in Evolution of Surface and Thin Film Microstructure Symposium, (H. A. Atwater, E, Chason, M. H. Grabow, and M. G. Lagally, eds.), Pittsburgh, PA, MRS, pp. 705-710 (1993)

260a.

K. Kobayashi, M. Kumagai, S. Karasawa, T. Watanabe, and F. Togashi, Effect of ion implantation and surface structure of silicon on diamond film nucleation, J. Cryst. Growth, 128(14):408-412 (1993)

260b.

K. Kobayashi, S. Karasawa, T. Watanabe, and F. Togashi, J. Cyst. Growth, 99(14):1211 (1990)

261.

C. H. Wu, W. H. Weber, T. J. Potter, and M. A. Tamor, Laser reflective interferometry for in-situ monitoring of diamond film growth by chemical vapor deposition, J. Appl. Phys., 73(6):2977-2982 (1993)

262.

M. A. George, A. Burger, W. E. Collins, J. L. Davidson, A. V. Barnes, and N. H. Tolk, Investigation of nucleation and growth processes of diamond films by atomic force microscopy, J. Appl. Phys., 76(7):4099-4 106 (1994)

263.

R. W. Collins, Y. Cong, H. V. Nguyen, I. An, K. Vedam, T. Badzian, and R. Messier, Real time spectroscopic ellipsometry characterization of the nucleation of diamond by filament-assisted chemical vapor deposition, J. Appl. Phys., 71(10):5287-5289 (1992)

264.

W. R. L. Lambrecht, C. H. Lee, B. Segall, J. C. Angus, Z. Li, and M. Sunkara, Diamond nucleation by hydrogenation of the edges of graphitic precursors, Nature, 364 (6438):607-610 (1993)

265.

T. R. Anthony, in The Physics and Chemistry of Carbides, Nitrides andBorides, Freer, cd.), Kluwer Academic Press, Netherlands, p. 133 (1989)

266.

J. Kim, M. Kim, S. Park, and J. Lee, The morphology changes in diamond synthesized hot-filament chemical vapor deposition, J. Appl. Phys., 67(7):33543357 (1990)

267.

B. Liu, G. S. Sandhu, N. R. Pa&h, M. L. Swanson, and W. K. Chu, Nucl. Instr. Methods Phys. Res., B45:420 (1990)

268.

J. J. Dubray, C. G. Pantano, M. Meloncelli, and E. Bertran, Nucleation of diamond on silicon, SiAlON, and graphite substrates coated with an a-C:H layer, J. Vacuum Sci. & Technol. A, 9(6):3012-3018 (1991)

269.

J. A. Baglio, B. C. Farnsworth, S. Hankin, C. Sung, J. Hefter, and M. Tabasky, Studies on the formation of diamond nucleation sites on <100> silicon substrates, in Diamond and Diamond-like Films and Coatings, (R. E. Clausing, L. L. Horton, J. C. Angus, and P. Koidl, eds.), Plenum Press, New York, pp. 635642 (1991)

270.

A. R. Badzian, and T. Badzian, Surf Coat. Technol., 36:283-293

271.

B. E. Williams, and J. T. Glass, J. Mater. Res., 4:373 (1989)

(1988)

(R.

I76

Diamond Chemical Vapor Deposition

272.

S. Yugo, T. Kimura, and T. Muto, Vacuum, 41:1364 (1990)

273.

B. E. Williams, J. T. Glass, R. F. Davis, and K. Kobashi, J. Crystal Growth, 99:1168 (1990)

274.

Y. H. Shing, F. S. Pool, and D. H. Rich, Low-pressure microwave plasma nucleation and deposition of diamond films, Thin Solid Films, 212(1-2):150-155 (1992)

275.

S. D. Wolter, J. T. Glass, and B. R. Stoner, J. Appl. Phys., 77(10):5119 (1995)

276.

A. Lindlbauer, Ph.D. Thesis, TU-Vienna, Austria (1991)

277.

R. Meihmas, M. S. Wong, K. C. Sheng, R. P. H. Chang, and R. P. VanDuyne, Appl. Phys. Lett., 54(22):2204-2206 (1989)

278.

D. Michau, B.Tanguy, G. Demazeau, M. Couzi, and R. Cavagnat, Influence on diamond nucleation of the carbon concentration near the substrate surface, Diamond Relat. Mater., 2(1):19-23 (1993)

279.

T. Hartnett, R. Miller, D. Montanari, C. Willingham, and R. Tustison, J. Vat. Sci. Technol. A, 8:2129 (1990)

280.

P. E. Pehrsson, J. Glesener, and A. A. Morrish, The CVD diamond nucleation mechanism on Si overlaid with spz carbon, in Wide Band Gap Semiconductors Symposium, (T. D. Moustakas, J. I. Pankove, and Y. Hamakawa, eds.), Pittsburgh, PA, MRS, pp. 57-62 (1992)

281.

P. E. Pehrsson, J. Glesener, and A. Morrish, Chemical vapor deposition diamond nucleation induced by sp2 carbon on unscratched silicon, Thin Solid Films, 212(12):81-90 (1992)

282.

K. V. Ravi, C. A. Koch, H. S. Hu, and A. Joshi, The nucleation and morphology of diamond crystals and films synthesized by the combustion flame technique, J. Mater. Res., 5( 11):2356-2366 (1990)

283.

J. L. Robins, Appl. Surf: Sci., 33-34:379 (1988)

284.

Z. Li, L. Wang, T. Suzuki, A. Argoitia, P. Pirouz, and J. C. Angus, J. Appl. Phys, 73:711-715 (1993)

285.

T. P. Ong, F. Xiong, R. P. H. Chang, and C. W. White, Mechanism for diamond nucleation and growth on single crystal copper surfaces implanted with carbon, Appl. Phys. Lett., 60( 17):2083-2085 (1992)

286.

H. Shiomi, K. Tanabe, Y. Nishibayashi, and N. Fujimori, Epitaxial growth of high quality diamond film by the microwave plasma-assisted chemical-vapor-deposition method, Jpn. J. AppI. Phys., 29(1):34-40 (1990)

287.

G. S. Gildenblat, S. A. Grot, C. W. Hatfield, C. R. Wronski, A. R. Badzian, T. Badzian, and R. Messier, Mater. Res. Bull., 25:129 (1990)

288.

A. R. Badzian, and T. Badzian, in Proc. MRS FallMeeting, (T. M. Besmann, B. M. Gallois, and J. Warren, eds.), 250B:339-349, Boston, MA (Dec. 4-6, 1991)

289.

Y. Sato, I. Yashima, H. Fujita, T. Ando, and M. Kamo, in New Diamond Sci. Technol., (R. Messier, J. T. Glass, J. E. Butler, and R. Roy, eds.), MRS, Pittsburgh, PA., pp. 371-376 (1991)

290.

J. Narayan, V. P. Godbole, and C. W. White, Science, 252:416-418

291.

M. Moore, Ind. Dia. Rev., 2:67 (1985)

(1991)

References

177

292.

K. Ikoma, and M. Yamanaka, Preparation of textured diamond films on Si subsbrates by hot-fdament chemical-vapor deposition, J. Appl. Phys., 74(5):35 193522 (1993)

293.

B. R. Stoner, and J. T. Glass, Textured diamond growth on (100) g-Sic via microwave plasma chemical vapor deposition, Appl. Phys. Lett., 60(6):698-700 (1992)

294.

F. G. Celii, D. White, Jr., and A. J. Purdes, Deposition of smooth, oriented diamond films using microwave plasma chemical vapor deposition, Thin Solid Films, 2 12( l2):140-149 (1992)

295.

C. Wild, N. Herres, J. Wagner, P. Koidl, and T. R. Anthony, Electrochem. Sot. Proc., 89-12:283 (1989)

296.

C. Wild, W. Mtkller-Sebert, T. Eckermann, and P. Koidl, Electrochem. Sot. Proc., 91-8:224 (1991)

297.

J. Narayan, V. P. Godbole, G. Matera, and R. K. Singh, Enhancement of nucleation and adhesion of diamond films on copper, stainless steel, and silicon substrates, J. Appl. Phys., 71(2):966-971 (1992)

298.

P. A. Dennig, H. Shiomi, D. A. Stevenson, and N. M. Johnson, Influence of substrate treatments on diamond thin film nucleation, Thin Solid Films, 212(1-2):63-67 (1992)

299.

P. A. Dennig, and D. A. Stevenson, Influence of substrate topography on the nucleation of diamond thin films, Appl. Phys. Lett, 59(13):1562-1564 (1991)

300.

R. Haubner, J. ReJ: Met. Hard Mat., 9(2):70-77

(1990)

30 1. M. Ihara, H. Komiyama, and T. Okubo, Correlation between nulceation site density and residual diamond dust density in diamond film deposition, Appl. Phys. Lett., 65(9):1192 (1994) 302.

K. Kobayashi, T. Nakano, N. Mutsukura, and Y. Machi, Characterization of diamond nucleation on Fe/Si substrate by hot-filament chemical vapor deposition, Vacuum, 44 (l):l-5 (1993)

303.

J. E. Yehoda, R. I. Fuentes, J. C. Tsang, S. J. Whitehair, C. R. Guarnieri, and J. J. Cuomo, Enhanced nucleation and growth of diamond on SIC by plasma enhanced chemical vapor deposition using thin metal films, Appl. Phys. Lett., 60 (23):28652867 (1992)

304.

A. A. Morrish, and P. E. Pehrsson, Effects of surface pretreatments on nucleation and growth of diamond films on a variety of substrates, Appl. Phys. Lett., 59 (4):417-419 (1991)

305.

M. Yoshikawa, Y. Kaneko, C. F. Yang, H. Tokura, and M. Kamo, J. Jpn. Sot. Precision Eng., 54 (9): 1703 (1988); M. Yoshikawa, Y. Kaneko, C. F. Yang, and H. Tokura, Gas phase synthesis of diamond on silicon with microwave plasma CVD, lI. Effects of substrate surface state on nucleation, J. Jpn. Sot. Precision Eng., 55 (1):155-160 (1989)

306.

J. S. Ma, H. Yagyu, A. Hiraki, H. Kawarada, and T. Yonehara, Large area diamond selective nucleation based epitaxy, Thin Solid Films, 206(1-2):192-197 (199 1)

307.

J. S. Ma, H. Kawarada, and A. Hiraki, Selective nucleation based epitaxy of CVD diamond, Solid State Phys., 26 (8):496-500 (1991)

178

Diamond Chemical Vapor Deposition

308.

J. L. Valdes, J. W. Mitchel, J. A. Mucha, L. Seibles, and H. Huggins, Selected-area nucleation and patterning of diamond thin films by electrophoretic seeding, J. Electrochem. Sot., 138 (2):635-636 (1991)

309.

J. S. Ma, H. Kawarada, T. Yonehara, and A. Hiraki, Nucleation control and selective growth of diamond crystals on Si substrate, Technol. Reports of the Osaka University, 40( 1986-2003):63-72 (1990)

310.

J. S. Ma, H. Kawarada, T. Yonehara, J. Suzuki, Y. Yokota, and A. Hiraki, Nucleation control and selective growth of diamond particles formed with plasma CVD, J. Cyst. Growth, 99(14, pt. 2):1206-1210 (1990)

311.

J. S. Ma, H. Kawarada, T. Yonehara, J. I. Suzuki, J. Wei, Y. Yokota, and A. Hiraki, Selective nucleation and growth of diamond particles by plasma-assisted chemical vapor deposition, Appl. Phys. Lett., 55(11):1071-1073 (1989)

312.

S. J. Lin, S. L. Lee, and J. Hwang, Selective deposition of diamond films on ionimplanted Si(100) by microwave plasma chemical vapor deposition, J. Electrochemical Sot., 139( 11):3255-3258 (1992)

313.

S. Yu, Z. Jin, X. Lu, and G. Zou, Selective deposition of diamond films on silicon wafer with SiO, masks by tungsten filament chemical vapor deposition, Mater. Lett., 10(7-8):375-378 (1991)

314.

C. P. Chang, D. L. Flamm, D. E. Ibbotson, and J. A. Mucha, J. Appf. Phys., 63(5):1744 (1988)

3 15.

K. Kobayashi, N. Mutsukura, and Y. Machi, Catalyst effect for diamond nucleation in the low pressure process, Mater. and Manufacturing Processes, 7(3):395403 (1992)

316.

Y. Shu-Cheng, C. Hon-Sho, and G. Fon-Shu, J. Cyst.

317.

P. Ascarelli, and S. Fontana, Dissimilar grit-size dependence of the diamond nucleation density on substrate surface pretreatments, Appl. Surf: Sci., 64(4):307311 (1993)

Growth, 99:1196 (1990)

3 18.

S. Iijima, Y. Aikawa, and K. Baba, Growth of diamond particles in chemical vapor deposition, J. Mater. Res., 6(7):1491-1497 (1991)

3 19.

R. Polini, Diamond 2519 (1992)

320.

S. Yugo, A. Izumi, T. Kanai, T. Muto, and T. Kimura, in New Diamond Sci. TechnoL, (R. Messier, J. T. Glass, J. E. Butler, and R. Roy, eds.), MRS, Pittsburgh, PA., pp. 385-389 (1991); S. Yugo, T. Kanai, and T. Kimura, Effects of hydrogen plasma on the diamond nucleation by chemical vapour deposition, Diamond Relat. Mater., 1(9):929-932 ((1992)

nucleation on cleaved Si(l1 l), J. Appf. Phys., 72(6):2517-

32 1. N. Ohtake, M. Yoshikawa, K. Suzuki, and S. Takeuch, in Applications of Diamond Films and Related Materials, (Y. Tzeng, M. Yoshikawa, M. Murakawa, and A. Feldman, eds.), Elsevier Science Publishers, Amsterdam, pp. 4311138 (1991) 322.

K. Higuchi, and S. Noda, Diamond Relat. Mater., 1:220 (1992)

323.

W. Mtiller-Sebert, C. Wild, P. Koidl, N. Herres, J. Wagner, and T. Eckermann, in Sic Natural and Synthetic Diamond and Related Materials, (A. A. Gippius, R. Helbig, and J. P. F. Sellschop, eds.), North-Holland, Amsterdam, p. 173 (1992)

References and D.E. Kuhman,J.

179

324.

F. Jansen, M. A. Macho&in, (1990)

Vacuum Sci. Technol. A, 8:3785

325.

J. J. Cuomo, J. P. Doyle, D. L. Pappas, K. L. Saenger, C. R. Guamieri, and S. J. Whitehair, in Applications of Diamond Films and RelatedMaterials, (Y. Tzeng, M. Yoshikawa, M. Murakawa, and A. Feldman, eds.), Elsevier Science Publishers, Amsterdam, pp. 169-178 (1991)

326.

R. Beckmann, W. Kulish, H. J. Frenk, and R. Kassing, in Applications ofDiamond Films and Related Materials, (Y. Tzeng, M. Yoshikawa, M. Murakawa, and A. Feldman, eds.), Elsevier Science Publishers, Amsterdam, pp. 543-548 (1991)

327.

J. L. Valdes, J. W. Mitchel, J. A. Mucha, L. Seibles, and H. Huggins, in Applications ofDiamondFilms andRehtedMaterials, (Y. Tzeng, M. Yoshikawa, M. Murakawa, and A. Feldman, eds.), Elsevier Science Publishers, Amsterdam, pp. 4231128 (1991)

328.

M. W. Geis, A. Argoitia, J. C. Angus, G. H. M. Ma, J. T. Class, J. Butler, C. J. Robinson, and R. Pryor, Appl. Phys. Lett., 58~2485 (1991)

329.

M. W. Geis, Appt. Phys. Lett., 55550 (1989)

330.

H. Itoh, T. Osaki, H. Iwahara, and H. Sakamoto, Nucleation control of diamond synthesized by microwave plasma CVD on cemented carbide substrate, J. Muter. Sci., 1991,26(14):3763-3768 (1991)

331.

G. H. Ma, Y. H. Lee, and J. T. Glass, Electron microscopic characterization of diamond films grown on Si by bias-controlled chemical vapor deposition, J. Muter. Rex, 5(11):2367-2377 (1990)

332.

X. Jiang, K. Schiffmann,

333.

S. Yugo, T. Kanai, T. Kimura, and T. Muto, Generation of diamond nuclei by electric field in plasma chemical vapor deposition, Appt. Phys. Lett., 58( 10): 10361038 (1991)

334.

D. E. Meyer, R. 0. Dillon, and J. A. Woollam, Radio-frequency plasma chemical vapor deposition growth of diamond,J. Vacuum Sci. h Technol. A, 7(3, pt. 2):23252327 (1989)

335.

V. P. Godbole and J. Narayan, Nucleation and growth of diamond on FeSi,/Si substrates by hot filament chemical vapor deposition, J. Appl. Phys., 71(10):49444948 (1992)

336.

Y. Liou, A. Inspektor, R. Weimer, D. Knight, and R. Messier, The effect of oxygen in diamond deposition by microwave plasma enhanced chemical vapor deposition, J. Mater. Res., 5(11):2305-2312 (1990)

337.

P. S. Weiser, S. Prawer, A. Hoffman, R. R. Manoty, P. J. K. Paterson, and S. A. Stuart, Carbon diffusion in uncoated and titanium nitride coated iron substrates during microwave plasma assisted chemical vapor deposition of diamond, J. Appl. Phys., 72(10):4643-4647 (1992)

and C. P. Klages, Phys. Rev. B, 50(12):8402 (1994)

338.

M. Marinov, Thin Solid Films, 46:267 (1987)

339.

A. R. Kirkpatrick, B. W. Ward, and N. P. Economou, Focused ion-beam crater arrays for induced nucleation of diamond film, J. Vacuum Sci. & Technol. B, 7(6):1947-1949 (1989)

180

Diamond Chemical Vapor Deposition

340.

K. Hirabayashi, Y. Taniguchi, 0. Takamatsu, T. Ikeda, K. Ikoma, and N.I. Kurihara, Appl. Phys. Lets, 53(19):1815 (1988)

341.

S. J. Lin, S. L. Lee, J. Hwang, C. S. Chang, and H. Y. Wen, Effbets of local facet and lattice damage on nucleation of diamond grown by microwave plasma chemical vapor deposition, Appl. P&s. Z&t., 60(13):1559-1561 (1992)

342.

F. P. Bundy, H. M. Strong, and R. H. Wentorf, Jr., Methods and mechanisms of synthetic diamond growth, in Chemisby and Physics of Carbon, (P. L. Walker, Jr., and P. A. Thrower, eds.), 10, Marcel Dekker, Inc., New York (1973)

343.

Y. Shimado, N. Mutsukura, and Y. Machi, Synthesis of diamond using Fe catalysts by RF plasma chemical vapor deposition method, Jpn. .I Appl. Phys., 31(6B):1958-1963 (1992)

344.

W. E. Brower, R. A. Bauer, and N. M. Sbrockey, Catalytic effects on diamond film formation by plasma-assisted chemical vapor deposition, Diamond Refat. Mater., 1(8):859-&l (1992)

345.

S. Haq, J. A. Savage, and D. L. Tunnicliffe, in Applications ofDiamond Films and Related Materials, (Y. Tzeng, M. Yoshikawa, M. Murakawa, and A. Feldman, eds.), Elsevier Science Publishers, Amsterdam, pp. 405410 (1991)

346.

J. W. Kim, Y. J. Baik, andK. Y. Eun, in Applications ofDiamondFilmsandRelated Materials, (Y. Tzeng, M. Yoshikawa, M. Murakawa, and A, Feldman, eds.), Elsevier Science Publishers, Amsterdam, pp. 399-403 (1991)

347.

S. Park, and J. Lee, Nucleation behaviour of diamond particles on silicon substrates in a hot-filament chemical vapour deposition, J. Mater. Sci., 28(7):1799-1804 (1993)

348.

Y. Hibino, and Y. Hayashi, Increase in nucleation density of diamond film growth by D plasma chemical vapor deposition, Surf: Coat. Technob, 54-55(1-3):365-367 (1992)

349.

L. Vandenbulcke, (1991)

350.

J. J. Chang, D. Mantei, R. Vuppuladhadium, 59:1170 (1991)

P. Bou, and G. Moreau, J. Electrochem. Sot.,

138(10):2985

and H. E. Jackson, Appl. Phys. Lett.,

35 1. J. Singh, M. Vellaikal, and R. Dat, Gas flow effects in synthesis of diamond by hottilament chemical vapor deposition, Thin Solid Films, 238( 1): 133-140 (1994) 352.

Y. Muranaka, H. Yamashita, and H. Miyadera, Thin Solid Films, 195:257 (1991)

353.

Y. Muranaka, H. Yamashita, and H. Miyadera, J. Appf. Phys., 69:8145 (1991)

354.

Y. Muranaka, H. Yamashita, and H. Miyadera, J. Cry~l. Growth, 112:808 (1991)

355.

M. Nunotani, M. Komori, M. Yamasawa, Y. Fujiwara, K. Sakuta, T. Kobayashi, S. Nakashima, S. Minomo, M. Taniguchi, and M. Sugiyo, Jpn. J. Appf. Phys., 3O:L1199 (1991)

356.

F. M. Cerio, W. A. Weimer, and C. E. Johnson, J. Mater. Res., 7:1195 (1992)

357.

W. A. Weimer, F. M. Cerio, and C. E. Johnson, J. Mater. Res., 6(10):2134-2144 (1991)

358.

K. Takeuchi, and T. Yoshida, J. Appl. Phys., 71:2636 (1992)

Refeences

181

359.

N. Ohtake, and M. Yoshikawa, Effects of oxygen addition on growth of diamond film by arc discharge plasma jet chemical vapor deposition, Jpn. J. Appl. Phys., 32(5A):2067-2073 (1993)

360.

J. Wei, J. M. Chang and Y. Tzeng, Deposition of diamond films with controlled nucleation and growth using hot filament CVD, Thin Solid Films, 212(1-2):91-95 (1992)

361.

E. Molinari, R. Polini, V. Sessa, M. L. Terranova, and M. Tomellini, Diamond nucleation from the gas phase: a kinetic approach, J. Muter. Res., 1993, 8(4):785797 (1993)

362.

E. Molinari, R. Polini, and M. Tomellini, Patterns of diamond nucleation from the gas phase, J. Muter. Rex, 8(4):798-810 (1993)

363.

J. W. Christian, The Theory of Transformations in Metals and Alloys, Part Z, Equilibrium and General Kinetic Theory, 2nd ed., Pergamon Press, Oxford, Chap. 10, pp. 418475 (1975)

364.

K. Reichelt, Vacuum, 38:1083 (1988)

365.

M. Vohner, and A. Weber, Z. Phys. Chem., 119:277 (1926)

366.

F. C. Frank, and J. H. van der Merwe, Proc. Royal Sot. London Ser. A, 198:205 (1949)

367.

I. N. Stranski, and L. Krastanov, Sitzungsber. Akad. Wiss. Wien, Math-Naturewiss. Kl ZZb, 146:797 (1938)

368.

R. F. Davis, J. Vuc. Sci. Technol. A, 11:829 (1993)

369.

J. C. Angus, W. R. L. Lambrecht, P. Pirouz, and B. Segall, Ann&Report, #NOOO14-89-1631, (Sept. 30, 1990)

370.

P. E. Pehrsson, F. G. Celii, and J. E. Butler, Chemical mechanisms of diamond CVD, in Diamond Films and Coatings, Development, Properties, andApplications, (R. F. Davis, ed.), Noyes Publications, Park Ridge, NJ (1993)

Contract

37 1. M. P. Everson, and M. A. Tamor, Investigation of growth rates and morphology for diamond growth by chemical vapor deposition, J. Mater. Res., 7(6):1438-1444 (1992) 372.

B. V. Derjaguin, and D. V. Fedoseev, Surf Coat. TechnoZ., 38:131-250 (1989); Growth of diamond and graphite from the gas phase, Jzd. Nauka, Ch. 4, Moscow (in Russian) (1977)

373.

B. B. Pate, The diamond surface atomic and electronic structure, Surf Sci., 165:83 (1986)

374.

W. Piekarczyk, and S. Prawer, Role of atomic hydrogen in preventing surface reconstruction and sp* bond formation during chemical vapour deposition of diamond, Diamond Relat. Mater., 2(1):41-47 (1993)

375.

J. W. Matthews, Epitaxial Growth, Part A and B, Academic Press, New York (1975)

376.

M. Tomellini, Evidence for nonclassical nucleation at solid surfaces in diamond deposition from the gas phase, J. Muter. Res., 1993, 8(7):1596-1604 (1993)

377

P. Badziag, W. S. Verwoerd, W. P. Ellis, and N.R. Greiner, Nature, 343:244 (1990)

182

Diamond Chemical Vapor Deposition

378.

M. Avrami,J.

379.

J. Frenkel, J. Chem. Phys., 7538 (1939)

Chem. Phys., 7:1103 (1939), and 8:212 (1940)

380.

E. Molinari, R. Polini, M. L. Terranova, P. Ascarelli, and S. Fontana, Uncoupling crystal growth and nucleation in the deposition of diamond from the gas phase, J. Muter. Res., 7(7):1778-1787 (1992)

381.

S. J. Harris and L. R. Martin, J. Muter. Res., 5:2313 (1990)

382.

M. Shen, H. F. Schaefer, C. Liang, J. Li, N. L. Allinger, and P.R. Schleyer, J. Am. Chem. Sot., 114:497 (1992)

383.

N. Jiang, B. W. Sun, Z. Zhang, and Z. Lin, J. Mater. Res., 9(10):2695 (1994)

Index

A A-C 8 A-C:H 8 Abrading 96 Abrasion treatment 99 Abrasive. residual 98 Abrasive particle size 99, 101 Abrasive scratches 121 Abrasives 95, 98, 128 Abstraction hydrogen 62 of H atoms 17 rate coefficient 35 Acetylene 155 Activation energy 14.5 Activation of the gas phase 24 Activation step 117 Adamantane 47 Adamantane-based clusters 157 Adatoms 54 Adhesion 46, 94 Adhesionenergy 144, 145, 150 Adsorption state 133 AES 58

AFht 58 Allotrope ofcarbon 1 Amorphouscarbon 42, 59, 63 Amorphous tilm 38 Amorphous hard carbon 8 Amorphous layer 62 Amorphous-crystalline interface 60 Analytical techniques 69 Angus, J. C. 4, 5, 33, 48 Anisotropic growth 88 Anode materials 28 Applications abrasive 3 boring and drilling tools 5 electronic 46, 79 emerging 6 heat-sink 3 sound diaphragm 5 AP implantation 123 Arrheoius behavior 42 As+ implantation 121 Atomic carbon 24 Atomic diffusion 62 183

I84

Diamond Chemical Vapor Deposition

Atomic force microscopy 58 Atomic hydrogen 5, 29, 59, 147, 148, 154 Atomic hydrogen concentration 24 Atomic structures 8 Auger electron spectroscopy 58

B Bachmann, F? K. 67, 116 Badzian, A. R. 64, 80 Ball-like diamond crystals 10 Ballas-type diamond 38 Barrier layer 119 Be0 145 Bias substrate 44 Bias voltage 23, 108 Bias-enhanced nucleation 83 Biasing 64, 95, 106, 128, 156 Si substrate 112 Biasing pre-treatment 107 duration 110 Bicyclodecane 48, 159 Blasting 122 BN 145 Boat conformation 10 Boltzmaun’s constant 147 Bond angle strain 148 Bond energy 9, 147 Bond length 9 Bonded carbon 59 Bonding covalent 9 orbitals for 8 Bonding state transition 59 Boron gaseous 29 Boron atoms 99 Buffer layer 123 carbide 93 Bundy, F. R 3

C 3C diamond 9 C diffusion rate 65 C concentration 71 C-H-O phase diagram 36 C-H-O systems 33 C/O atomic ratios 49 C35H36 157 C70 117 Cage compounds 47 Capture rate 133 Carbide as a buffer layer 93 Carbide interlayer 68, 71 Carbide islands 111 Carbide layer 64, 66, 68, 71 Carbide-forming substrate 57, 71, 93, 126 Carbides 94 Carbon atomic 24 Carbon atoms 55, 59 average residence time 103 rearrange 62 Carbon bonding 67 Carbon carriers 14 Carbon clusters 40, 69, 112, 114 Carbon concentrations 32 Carbon ion implantation 122 Carbon phase diagram 1, 3 Carbon radicals 119 Carbon saturation 68, 71, 77, 112, 128 Carbon sink 94 Carbon solubility 94 Carbon species 24 Carbon supersaturation 67 Carbon-containing gas 14,26 Carbon-rich particles 116 Carbon-substrate interactions 93 Carburization 64, 83, 94, 95, 123, 124, 156

Index

Carbynes 810 Catalysts 124 Catalytic processes 15 Cathode materials 28 Cations carbon-containing 112 Cauliflower morphology 67 CBN 50, 90, 93 CBN powder 80 single-crystal substrates 79 substrate 3 1 Ceramic particles 99 Ceramics 80 Cl$ 18, 32, 35, 106 Cl& concentration 83, 85, 86, 135 CH,/H, 83 Chair conformation 9 Chair coordination 47 Chair planes 10 Characterization of diamond films 57 Chemical kinetics model 48 Chemical vapor deposition (CVD) 4 Close-packed layers 9 Cluster approximation 156 Cluster formation 112 Clusters 32, 45, 50, 59, 114, 117, 128 grow or decay 71 stable 55 Co-deposition of diamond and graphite 4 Co-Fe 124 Coated area 28, 43 Coaxial feed electrode 28 Cobalt 126 Cohesion energy 144 Columnar gram growth 83 Combustion CVD 15 Combustion mechanisms 17 Combustion synthesis 30, 87, 88, 99, 118

18.5

Competition growth of crystals 52 Compressive stresses 108 Contact angle 144 Contamination from plasma nozzle 29 of the growing film 23 Continuous film 57 Conversion graphite to diamond 1 Conversion rate 1, 2 Copper anode 28 Copper surface 12 1 Corrosion of the hot filament 18 Covalent carbides 94 Covalent crystalline structure 9 Craters 121 Critical nuclei 103 Critical size 71 Crystal growth evolution computer simulation 82 Crystal habit of diamond 81, 82 Crystal morphology 82 Crystal quality 38 Crystal shapes 10, 12 Crystal size 38 critical 62 distribution 87 Crystal structures 8 Crystallite morphology 3 1 Crystallite size 31, 43 Crystallites 53, 59 small 89 Crystallization in the amorphous layer 62 Crystallographic planes 8 1 Crystallographic registry 83 crystals ball-like 67 cube-octahedral 116 faceted 52 graphite-free 27

186

Diamond Chemical Vapor Deposition

large 4 octahedral 32 single 4 CU nucleation density on 110 Cu substrates 75, 123 Cubic diamond 9 Cubic faces 10 Cubic structure 10 Cubo-octahedral crystals lo,82 CVD advantages and disadvantages CVD diamond 10 CVD methods high rate 33 CVD techniques 18 Cyclohexane 47 Cyclotron resonance 27

15

D Dangling bonds 75, 148 DC glow discharge PACVD 28 DCPACVD 27, 28, 37, 44 Debye temperature 154 Decahedrons 32 Defects 44, 62 structural 85 void-type 82 Deformation depth 102 Dehydrogenation 62 Delocalized electron 8 Dendrites 12 Dendritic growth 119 Deposition large-area 30 low-temperature 46 Deposition area 5, 28, 31 Deposition conditions 10, 52, 68, 131, 135, 141, 153, 155 optimum 135 Deposition efficiency 3 1

Deposition rate 18, 38, 147 function of temperature 147 Deposition rates 18, 26, 28, 29, 36 Dejaguin, B. V. 4, 5, 47, 146 Desorption enthalpy 155 Diadamantane 47 2H diamond 10 Diamond lattice constant 119 properties 1 unit structure 47 { lOO}Diamond 85 Diamond CVD 5, 6 Diamond films characterization of 57 Diamond grains 43 Diamond lattice structure 9 Diamond nucleation 13 1 Diamond polytypes 8 Diamond powder 132 Diamond seed crystals 4 Diamond wheels 122 Diamond-like carbon 8, 28 Diamond-like hydrocarbon 8 Diborane 49 Diffusion solid-state 62 DiEusion barrier 120, 124 Direct conversion 2 Direct-current plasma-assisted CVD 27 Disordered surface 72 Dissociation of gas species 24 DLC 8, 35, 94, 119 DLC interlayer 62 DLC layer 68, 156 Domain disordered 59, 62 Doped semiconductor diamond 29 Dupre relation 150 Dyer, H. B. 3

Index

E ECR MW PACVD 37 ECR plasma 44 ECR plasma CVD 27 EELS 58 Electric discharge method 5 Electric fields 44 Electrical biasing 44 Electron density 114 Electron energy loss spectroscopy 58 Electron-cyclotron-resonance (ECR) 27 Electronic material 83 Electrons in plasmas 25 Electrophoretic seeding 104, 128 Energies misfit and strain 80 Energy sources 23 Enthalpies of formation 47 Enthalpy graphite 1 Entropy 148 Epitaxial growth onNi 80 Epitaxially textured films 85 Epitaxy 79 Etch pits 116 Etching 95 atomic-hydrogen 134 of graphite 75 nonuniform 69 preferential 111 of unstable phases 59 Eversole, W. G. 4 Evolutionary selection 52, 82, 90

F F atoms 35 Faceted crystallographic shape Faceted crystals 52 Faceted diamond films 38

62

I87

Faceting 157 Facets 32 Fe films on the Sic 125 Fe substrate TiN coated 120 Fe-coated Si 125 Fe-Ni 124 Feed-gas mixtures 27 Fiber textures 90 Fiber-textured films 85 Filament 18, 23 Filament temperature 38, 134 Film morphology 85 Film texture 85 Fivefold symmetry 10 Fivefold twinned decahedra1 crystals 12 Flame CVD 14, 37 Flame synthesis 12 Flow rate 141 Force-field method 156 Fracture strength 108 Frank-van der Merwe 158 Frank-van der Merwe layer growth 144 Free-standing shapes 5 Frenklach, M. 17, 48, 49, 50, 153 Fullerenes 8

G Gas composition 36, 86 Gas dynamic conditions 154 Gas flow rate 36, 135 Gas pressure 14 1 effect on nucleation 134 Gas pressures 37 Gas species 35 Gas-jet nozzles 28 Gas-phase activation 33 Gas-phase non-equilibrium 14 Gas-phase nucleation 50

188

Diamond Chemical Vapor Deposition

Gas-phase temperature 15, 33 Gem stones 1 Germ 151, 155 Gibbs free-energy 149, 150, 158 Glass beads 122 Glass, J. T. 163, 164, 169, 170-172, 175-177, 179 Glow discharge 23 Glow discharge PACVD 28 Glow discharge plasma 24, 28, 37 Gram boundaries 52 low-angle 83 Grain boundary 105 Grain size 105 Graphite 59, 129 anisotropic characteristic 9 etch rate 138 stable 1 Graphite carbon onPt 68 Graphite clusters 116 Graphite deposits 94 Graphite disk substrate 67 Graphite fibers 95 Graphite flakes 40 Graphite impurities 44 Graphite interlayer 104 Graphite lattice structure 8 Graphite layer 75 Graphite nucleation 146 Graphite particles 72 Graphite precursors 48, 75 Graphite to diamond conversion 1 Graphite-diamond equilibrium 3 Graphitic carbon 42, 138 Graphitic deposits 72 Graphitic intermediate layers 68 Gratings sawtooth-profile 104 Grinding 122 Grit 95, 101, 118 Grit-size 102 Grooves reentry 52

Growth bombardment-controlled 44 chemistry-controlled 44 early stages 46 homoepitaxial 79 interruption 89 single-crystal 46 Growth conditions 85, 86 Growth modes 144 Growth parameter 85, 90 Growth rate 23, 28, 49, 52, 82 ofMo 66 of SC 66 Growth rates 4, 5, 10, 15, 29, 30, 33, 42, 87, 89 Growth species oxygen-containing 18

H H atom gas-phase 17 H atoms 35 H ion species 110 H termination reaction 18 HZ-hydrocarbon gas mixtures 35 H,/Ar plasma 28 Habits 10 Halogen/hydrogen mixtures 49 Halogens 35 Haubner, R. 64, 65, 69, 93 Heteroepitaxial diamond on transition metals 80 Heteroepitaxial diamond films 79 Heteroepitaxial growth 144 Heteroepitaxy 90 Hexacyclopentadecane 47 Hexagonal graphite 9 HPCVD 15, 18, 23, 26, 37, 38, 44, 56, 59, 63, 65, 68, 72, 99, 121, 134 reactor model 48 High energy sites 96

Index 189 High resolution transmission electron microscopy 58 High-energy sites 68, 71, 77 High-pressure high temperature @PI-IV 1 Hirose, Y. 27, 30 Homoepitaxially grown crystals 10 Homoepitaxy 90 Homogeneity 46 Homogeneous nucleation 47 Hot-filament CVD 14, 18 Hot-filament method 5 HPHT 1, 124 solvent-catalyst 3 HPHT-synthesized cBN 80 HRTEM 58 Hybrid carbon orbitals 8 Hydrocarbon cage molecules 47 Hydrocarbon concentration 82 Hydrocarbon deposit 72 Hydrocarbon fragments 157 Hydrocarbon gas 4 Hydrocarbon oil 116, 118 Hydrocarbon species 18 Hydrocarbons 113, 157, 159 aliphatic and aromatic 14, 35 Hydrogen abstraction 47 Hydrogen atoms needed to convert 59 recombination of 62 Hydrogen concentration 14, 44 Hydrogen dilution 146 Hydrogen gas 147 Hydrogen impurities 44 Hydrogen recombination reactions 29 Hydrogen-treatment 136 Hydrogenated amorphous carbon 63

I Icosahedral crystals Icosahedrons 32

12

Idiomorphic crystal shapes 12 Incorporation reactions 154 Incubation 39, 40,50,58,65,67, 72,77,83,93,131, 137, 142 Induction nucleation 50 Instability phenomenon 88 Interfacial layer 69 Intermediate amorphous layer 64 Intermediate layer 58, 68, 77 Interruption ofgrowth 89, 91 Ion bombardment 12 1 Ion implantation 95, 129 Ionic carbides 94 Irradiation 95 pulsed laser 123 Island growth 68, 144, 158 Isostructural materials 80 Isothermal plasma 24, 29

K Kinetic barrier 2 Kinetic equations 152 Kinetic processes 152 Koidl, P 163, 171, 172, 174, 177, 178

L Lamellae 82 Lander, J. J. 4 Large-area single crystals 79 Laser interferometry 58 Laser irradiation 123 Laser-driven reactions 49 Laser-enhanced CVD 15 Lattice bonds 18 Lattice constant 50, 80 Lattice damage 12 1 Lattice discontinuity 82 Lattice mismatch 89, 144 Lattice registry 75 Layer growth 144, 158

190

Diamond Chemical Vapor Deposition

LEED 58 Liander, H. 3 Linear growth law 153 Lonsdaleite 9, 10 Low energy electron diffraction 58 Low temperature synthesis 43 Lux, B. 64, 65, 69, 93

M Magnetic fields 44 Many-body potential 75 Matsumoto, S. 18, 28, 29, 47, 158 Messier, R. 67 Metal carbides 45, 59 Metal evaporation from the filament 23 Metal films 125, 129 Metastable growth 4 Metastable phase 145, 148 Methane 14, 35, 137 chlorine-permuted 35 decomposition of 49 Methane-hydrogen environment 104 Methane/hydrogen mixtures 49 Methyl groups 157 Micro-JR absorption measurement 58 Micro-Raman spectroscopy 58 Microcrystalline graphite 42 Microcrystallites 59 Microflaws 106 Microstructure 90 Microtwin lamellae 82 Microwave 24 Microwave frequency 27 Microwave plasma assisted CVD 26 Microwave plasma method 5 Microwave power 132 Microwave radiation 26 Mirror-polished Si 106, 107 Misfit dislocations 83

MO DLCon 35 nucleation on 65 MO substrates 66, 93, 104, 118, 124, 135 MO& 124 Molybdenum carbides 67 Monocrystalline diamond 79 Monte Carlo method 157 Morphological instabilities 88, 90 model 87 Morphology 87 Morrison, J. 4 Mosaic large-area 104 Multiply twinned decahedrons 32 MWPACVD 26, 37, 44, 66, 68, 69, 75, 83, 118, 131 biased 62 oxygen 137

N Native oxides 95 Natural diamond 1 Ni substrate 72, 80, 104 Nickel 126 NJRlM5 Nitrogen addition of 86 Nitrogen impurities 44 Non-crystalline phases 52, 87, 89 Non-diamond substrates 57, 118 Non-isothermal 24 Nozzles 28 Nucleation 8, 39, 46, 68, 72, 92, 153 effect of gas composition 135 effect of gas pressure 134 gas-phase 50 homogeneous 49, 77 on MO substrates 62 modeling 143 on non-diamond 50

Index non-epitaxial 7 1 on Si(100) 131 on Si substrates 62 temperature dependence 133 thermodynamic factors 146 Nucleation center 155 Nucleation density 23, 32, 37, 40, 46, 49, 52, 55, 57, 65, 89, 95, 96, 99, 102, 103, 106, 108, 132, 155 on graphite 67 Nucleation enhancement 114, 116, 121, 128, 156 by scratching 96 on Y-ZrO, coated Si 120 Nucleation enhancement by biasing 112 Nucleation enhancement methods 92 Nucleation kinetics 159 model 156 Nucleation mechanism 61, 64, 75 Nucleation model 116, 150 Nucleation modes 144 Nucleation process 54, 55, 57, 63 two-step 83 Nucleation rate 31, 38, 44, 45, 55, 57, 132, 147, 149 Nucleation sites 72 secondary 52 Nucleation theory 47, 146, 149, 150, 158 Nuclei non-faceted 80 Nusselt number 154 Nutrient starvation 89 0 Octahedral faces 10 OM 58 Optical microscopy 58

191

Optical transparency 43 Orbitals 8 Orbitals for bonding 8 Orientation effect 88 Overlayers 114 Oxidation reactions 17 Oxide formation 113 Oxyacetylene combustion synthesis 99 Oxyacetylene torch 30 Oxygen 137 in the gas phase 142 Oxygen addition 137, 138 Oxygen impurities 44 Oxygen in a plasma 49 Oxygen-containing compounds 27 Oxygen-containing gas atmosphere 18 Oxygen-containing species 86

P K bond

8 n: electrons 23 PACVD 15, 25, 26 Partial pressures 33 of gas species 34 Particle diameter 49 Particles 77 ball-like 38 Pd catalytic effect 126 Phase purity 38 Phosphorous compounds 29 Plasma arc-jet 24 Plasma arc-jet CVD 28 Plasma-assisted CVD 14, 23 Plasma characteristics 23 Plasma potential 25 Plasma sheath 25 Platelets 158 triangular 12 Poisoning 87

192

Diamond Chemical Vapor Deposition

Polishing 98, 122 Polyaromatic hydrocarbons 157 Polycrystalline Cu 110 Polycrystalline diamond films 50 Polycrystalline films 42 morphology 82 Polycrystalline morphology 3 1, 32 Polycrystalline structure 52 Polycrystalline substrates 57 Polycyclic aromatics 47 Polymantane-derived clusters 157 Polytypes 10, 49 Precursor layer 64 Precursors 24, 33, 47 graphite 47, 48 Predeposition 156 Pressure dependence 37 Pretreatment 39, 57, 63, 64, 72, 94, 135, 151 biasing 106 seeding 104 wiping 126 Pretreatment methods 127 smooth surfaces 106 Pretreatments 96, 118 of Si 122 Prism planes of graphite 75 Process parameters 86 Properties of CVD diamond 2 Properties of diamond 3 Properties of single-crystal diamond 2 Protrusions 63 Pt substrates 72 Puckered planes 9 Pulsed laser irradiation 123 Pyramidal pits 104 Pyramids 63 Pyrolysis 17, 18 hydrocarbon 24

R Radio-frequency plasma-assisted CVD 28 Radio-frequency thermal plasma CVD 29 Ravi, K. ‘I! 87, 89 RBS 58 Reentrant comer 157 Re-evaporation 103 Reactions entropy 47 Reactive precursors 24 Reactivity of graphite 23 Reactor operating conditions 15 Reactor pressure 49 Recombination coefftcient 154 Reducing environment 119 Reflection high energy diffraction 58 Refractory carbides 64 Refractory materials 42 Refractory metal carbides 94 Refractory metals carbide-forming 64 Renulceation 9 1 RPPACVD 28 RF thermal plasma CVD 37, 43 RHEED 58 Ring compounds 48 Rings six-member 75 Roughness 46, 83 Rutherford backscattering spectrometry 58

S CT bonds

8 Saturation of carbon 69 Scale-up 3 1 Scanning electron microscopy 58 Scanning tunneling microscopy 58

Index

Schmellemneier, H. 4 Scratched substrate 72 Scratches size of 103 Scratching 40, 95, 96, 102, 118, 128, 151 SE 58 Secondary ion mass spectrometry 58 Secondary nucleation 62, 75 Seed ceramics 99 Seeding 40, 95, 103, 114, 128 Seeding effect 96, 98, 128 Seeding techniques 104 Self-bias 25 Self-regenerating site 157 SEM 58 Semiconductor devices 79 Sheath region 25 Shock-wave synthesis, 3 Si 154 nucleation mechanism 111 SiO, on 35 a-Sic interlayers 83 P-Sic buffer layer 64 P-Sic films 68 P-Sic formation 134 P-Sic layer 68, 83 Si diffusion 137 Si mirror surface 106 Si substrates 66, 69, 75, 104, 108, 121, 122, 125 abraded 63 Si wafer 39, 108 Si(100) substrate 134 single crystal 83 Si+ ion implantation 121 SiAlON substrates 38, 63 Sic 64 on Si substrates 65 SIC formation 156 Sic impurities 44 Sic interfacial layer 57 Sic interlayer 67

193

Sic layer 124, 125 amorphous 69 Sic powder 132 Sic-coated substrate 125 Silane 49 Silicon impurities 44 SIMS 58 Single-crystal diamond 3 1, 79 Single-crystal substrates 57 SiOa 113 SiO, layer 111 SiO, on Si 35 SiO, 137 Slabs 87 Smooth surfaces pre-treatment methods 106 Solvent-catalyst reaction 3 Solvent-catalysts 124 Soot 120, 126 Sp2 atomic orbitals 8 Sp3 atomic orbitals 9 Space charge buildup 44 Spectroscopic ellipsometry 58 Spirals 52 Spitzyn, B. V 57, 82 Spontaneousnucleation 75, 77, 116 Stability of the filament 23 Stabilization of diamond 148 Stacking errors 157 Stacking fault 10, 12, 32, 45, 62 Staggered conformation 10 Stainless steel substrates 123 Steps 52 Sticking probability 133, 145 STM 58 Strain 122 Strain centers 98 Strain energy 144 Strain field effects 128 Stranski-Krastanov 158 Stress 89 in homoepitaxial films 82 Structures of diamond 8

194 Diamond Chemical Vapor Deposition Substrate bias 27 is mounted 28 Substrate bias 44, 114 Substrate cooling 29 Substrate materials 6, 26, 39, 92, 148 physical properties 40 Substrate position 43 Substrate size 43 Substrate temperature 26, 42, 43, 83, 86, 132, 141 influence on growth rates optimum 133 Substrates complex 106 mounting 27 non-diamond 32, 39 Superconductors 6 Supersaturated 94 Supersaturation 32,82 of carbon 65 Surface conditions effects on nucleation 92 Surface energy 62, 149 high 80 Surface reaction mechanisms 17 Surface treatment methods 95 Surgical ahoy 122 Suzuki, K. 27, 106 SYBYL 156 Synthesis low temperature 43 Synthesis of diamond 1 byCW 4

T TEM 58 Temperature 147 gas-phase 33 influence of 33 optimum 133, 141 substrate 26, 32, 42

Temperature control 29 Temperature dependence 37 of nucleation 133 Tersoff many-body potential 75 Tetracyclododecane 47 Tetmhedra 9 Thermal expansion coefficients 89 Thermal plasma CW 29, 37 Thermal plasmas RF 29 Thermodynamic equilibrium calculations 47 Thermodynamic stabilization 148 Thermodynamics of nucleation 146 Thick deposits 29 Ti nucleation on 65 Ti-6Al-4V substrates 122 TiN on Fe substrate 120 Tools diamond-coated 5 Topographical features promote nucleation 98 Transforming species 152 Transition metals 3, 80, 124 Transmission electron microscopy 58 Tungsten carbides 67 Tungsten cathode 28 Twin formation 85 Twinned clusters 45 Twinned crystals 10 Twinned embryos 47 Twinned nucleus 157 Twinning 12, 32, 82, 157 Twins 52, 62

U Ultrasonic pre-treatment 101, 102 Ultrasonically damaged surface 96

Index V Valences unsatisfied 7 1 Van der Drift 52, 82 Van der Waals bond 8 Vickery, E. C. 4 Voids 89 Volmer-Weber 145, 158 Volmer-Weber island growth

144

W W substrates 124 WC substrate 105 Wulff criterion 8 1

X X-ray X-ray XPS XRD

diffraction 58 photoelectron spectroscopy 58 58

Y Y-ZrOz 119 Yarbrough, W. A. 67, 80 Yoshikawa, M. 29, 122, 138 Young’s equation 143 Young’s modulus 108

Z Zirconia 122 yttrium-stabilized

119

58

195