Materials Characterisation V Computational Methods and Experiments
WIT Press publishes leading books in Science and Technology. Visit our website for new and current list of titles. www.witpress.com
WITeLibrary Home of the Transactions of the Wessex Institute. Papers presented at Materials Characterisation V are archived in the WIT eLibrary in volume 72 of WIT Transactions on Engineering Sciences (ISSN 1743-3533). The WIT eLibrary provides the international scientific community with immediate and permanent access to individual papers presented at WIT conferences. http://library.witpress.com
FIFTH INTERNATIONAL CONFERENCE ON COMPUTATIONAL METHODS AND EXPERIMENTS IN MATERIALS CHARACTERISATION MATERIALS CHARACTERISATION 2011
CONFERENCE CHAIRMEN A.A. Mammoli University of New Mexico, USA C.A. Brebbia Wessex Institute of Technology, UK A. Klemm Glasgow Caledonian University, UK
INTERNATIONAL SCIENTIFIC ADVISORY COMMITTEE G. Badalians Gholikandi A. Bayton A. Galybin H. Huh G. Moriconi P. Prochazka I. Sanchez P. Viot
Organised by Wessex Institute of Technology, UK University of New Mexico, USA Sponsored by WIT Transactions on Engineering Sciences
WIT Transactions Transactions Editor Carlos Brebbia Wessex Institute of Technology Ashurst Lodge, Ashurst Southampton SO40 7AA, UK Email:
[email protected]
Editorial Board B Abersek University of Maribor, Slovenia Y N Abousleiman University of Oklahoma,
G Belingardi Politecnico di Torino, Italy R Belmans Katholieke Universiteit Leuven,
P L Aguilar University of Extremadura, Spain K S Al Jabri Sultan Qaboos University, Oman E Alarcon Universidad Politecnica de Madrid,
C D Bertram The University of New South
USA
Spain
A Aldama IMTA, Mexico C Alessandri Universita di Ferrara, Italy D Almorza Gomar University of Cadiz, Spain
B Alzahabi Kettering University, USA J A C Ambrosio IDMEC, Portugal A M Amer Cairo University, Egypt S A Anagnostopoulos University of Patras, Greece
M Andretta Montecatini, Italy E Angelino A.R.P.A. Lombardia, Italy H Antes Technische Universitat Braunschweig, Germany
M A Atherton South Bank University, UK A G Atkins University of Reading, UK D Aubry Ecole Centrale de Paris, France H Azegami Toyohashi University of Technology, Japan
A F M Azevedo University of Porto, Portugal J Baish Bucknell University, USA J M Baldasano Universitat Politecnica de Catalunya, Spain J G Bartzis Institute of Nuclear Technology, Greece A Bejan Duke University, USA M P Bekakos Democritus University of Thrace, Greece
Belgium
Wales, Australia
D E Beskos University of Patras, Greece S K Bhattacharyya Indian Institute of Technology, India
E Blums Latvian Academy of Sciences, Latvia J Boarder Cartref Consulting Systems, UK B Bobee Institut National de la Recherche Scientifique, Canada
H Boileau ESIGEC, France J J Bommer Imperial College London, UK M Bonnet Ecole Polytechnique, France C A Borrego University of Aveiro, Portugal A R Bretones University of Granada, Spain J A Bryant University of Exeter, UK F-G Buchholz Universitat Gesanthochschule Paderborn, Germany
M B Bush The University of Western Australia, Australia
F Butera Politecnico di Milano, Italy J Byrne University of Portsmouth, UK W Cantwell Liverpool University, UK D J Cartwright Bucknell University, USA P G Carydis National Technical University of Athens, Greece
J J Casares Long Universidad de Santiago de Compostela, Spain
M A Celia Princeton University, USA A Chakrabarti Indian Institute of Science, India
A H-D Cheng University of Mississippi, USA
J Chilton University of Lincoln, UK C-L Chiu University of Pittsburgh, USA H Choi Kangnung National University, Korea A Cieslak Technical University of Lodz, Poland
S Clement Transport System Centre, Australia M W Collins Brunel University, UK J J Connor Massachusetts Institute of Technology, USA
M C Constantinou State University of New York at Buffalo, USA
D E Cormack University of Toronto, Canada M Costantino Royal Bank of Scotland, UK D F Cutler Royal Botanic Gardens, UK W Czyczula Krakow University of Technology, Poland
M da Conceicao Cunha University of Coimbra, Portugal
L Dávid Károly Róbert College, Hungary A Davies University of Hertfordshire, UK M Davis Temple University, USA A B de Almeida Instituto Superior Tecnico, Portugal
E R de Arantes e Oliveira Instituto Superior Tecnico, Portugal L De Biase University of Milan, Italy R de Borst Delft University of Technology, Netherlands G De Mey University of Ghent, Belgium A De Montis Universita di Cagliari, Italy A De Naeyer Universiteit Ghent, Belgium W P De Wilde Vrije Universiteit Brussel, Belgium L Debnath University of Texas-Pan American, USA N J Dedios Mimbela Universidad de Cordoba, Spain G Degrande Katholieke Universiteit Leuven, Belgium S del Giudice University of Udine, Italy G Deplano Universita di Cagliari, Italy I Doltsinis University of Stuttgart, Germany M Domaszewski Universite de Technologie de Belfort-Montbeliard, France J Dominguez University of Seville, Spain K Dorow Pacific Northwest National Laboratory, USA W Dover University College London, UK C Dowlen South Bank University, UK
J P du Plessis University of Stellenbosch, South Africa
R Duffell University of Hertfordshire, UK A Ebel University of Cologne, Germany E E Edoutos Democritus University of Thrace, Greece
G K Egan Monash University, Australia K M Elawadly Alexandria University, Egypt K-H Elmer Universitat Hannover, Germany D Elms University of Canterbury, New Zealand M E M El-Sayed Kettering University, USA D M Elsom Oxford Brookes University, UK F Erdogan Lehigh University, USA F P Escrig University of Seville, Spain D J Evans Nottingham Trent University, UK J W Everett Rowan University, USA M Faghri University of Rhode Island, USA R A Falconer Cardiff University, UK M N Fardis University of Patras, Greece P Fedelinski Silesian Technical University, Poland
H J S Fernando Arizona State University, USA
S Finger Carnegie Mellon University, USA J I Frankel University of Tennessee, USA D M Fraser University of Cape Town, South Africa
M J Fritzler University of Calgary, Canada U Gabbert Otto-von-Guericke Universitat Magdeburg, Germany
G Gambolati Universita di Padova, Italy C J Gantes National Technical University of Athens, Greece
L Gaul Universitat Stuttgart, Germany A Genco University of Palermo, Italy N Georgantzis Universitat Jaume I, Spain P Giudici Universita di Pavia, Italy F Gomez Universidad Politecnica de Valencia, Spain
R Gomez Martin University of Granada, Spain
D Goulias University of Maryland, USA K G Goulias Pennsylvania State University, USA
F Grandori Politecnico di Milano, Italy W E Grant Texas A & M University, USA
S Grilli University of Rhode Island, USA
R H J Grimshaw Loughborough University, D Gross Technische Hochschule Darmstadt,
M Karlsson Linkoping University, Sweden T Katayama Doshisha University, Japan K L Katsifarakis Aristotle University of
R Grundmann Technische Universitat
J T Katsikadelis National Technical
A Gualtierotti IDHEAP, Switzerland R C Gupta National University of Singapore,
E Kausel Massachusetts Institute of
UK
Germany
Dresden, Germany
Singapore J M Hale University of Newcastle, UK K Hameyer Katholieke Universiteit Leuven, Belgium C Hanke Danish Technical University, Denmark K Hayami University of Toyko, Japan Y Hayashi Nagoya University, Japan L Haydock Newage International Limited, UK A H Hendrickx Free University of Brussels, Belgium C Herman John Hopkins University, USA S Heslop University of Bristol, UK I Hideaki Nagoya University, Japan D A Hills University of Oxford, UK W F Huebner Southwest Research Institute, USA J A C Humphrey Bucknell University, USA M Y Hussaini Florida State University, USA W Hutchinson Edith Cowan University, Australia T H Hyde University of Nottingham, UK M Iguchi Science University of Tokyo, Japan D B Ingham University of Leeds, UK L Int Panis VITO Expertisecentrum IMS, Belgium N Ishikawa National Defence Academy, Japan J Jaafar UiTm, Malaysia W Jager Technical University of Dresden, Germany Y Jaluria Rutgers University, USA C M Jefferson University of the West of England, UK P R Johnston Griffith University, Australia D R H Jones University of Cambridge, UK N Jones University of Liverpool, UK D Kaliampakos National Technical University of Athens, Greece N Kamiya Nagoya University, Japan D L Karabalis University of Patras, Greece
Thessaloniki, Greece
University of Athens, Greece
Technology, USA
H Kawashima The University of Tokyo, Japan
B A Kazimee Washington State University, USA
S Kim University of Wisconsin-Madison, USA D Kirkland Nicholas Grimshaw & Partners Ltd, UK
E Kita Nagoya University, Japan A S Kobayashi University of Washington, USA
T Kobayashi University of Tokyo, Japan D Koga Saga University, Japan S Kotake University of Tokyo, Japan A N Kounadis National Technical University of Athens, Greece
W B Kratzig Ruhr Universitat Bochum, Germany
T Krauthammer Penn State University, USA C-H Lai University of Greenwich, UK M Langseth Norwegian University of Science and Technology, Norway
B S Larsen Technical University of Denmark, Denmark
F Lattarulo Politecnico di Bari, Italy A Lebedev Moscow State University, Russia L J Leon University of Montreal, Canada D Lewis Mississippi State University, USA S lghobashi University of California Irvine, USA
K-C Lin University of New Brunswick, Canada
A A Liolios Democritus University of Thrace, Greece
S Lomov Katholieke Universiteit Leuven, Belgium
J W S Longhurst University of the West of England, UK
G Loo The University of Auckland, New Zealand
J Lourenco Universidade do Minho, Portugal J E Luco University of California at San Diego, USA
H Lui State Seismological Bureau Harbin, China
C J Lumsden University of Toronto, Canada L Lundqvist Division of Transport and
Location Analysis, Sweden T Lyons Murdoch University, Australia Y-W Mai University of Sydney, Australia M Majowiecki University of Bologna, Italy D Malerba Università degli Studi di Bari, Italy G Manara University of Pisa, Italy B N Mandal Indian Statistical Institute, India Ü Mander University of Tartu, Estonia H A Mang Technische Universitat Wien, Austria G D Manolis Aristotle University of Thessaloniki, Greece W J Mansur COPPE/UFRJ, Brazil N Marchettini University of Siena, Italy J D M Marsh Griffith University, Australia J F Martin-Duque Universidad Complutense, Spain T Matsui Nagoya University, Japan G Mattrisch DaimlerChrysler AG, Germany F M Mazzolani University of Naples “Federico II”, Italy K McManis University of New Orleans, USA A C Mendes Universidade de Beira Interior, Portugal R A Meric Research Institute for Basic Sciences, Turkey J Mikielewicz Polish Academy of Sciences, Poland N Milic-Frayling Microsoft Research Ltd, UK R A W Mines University of Liverpool, UK C A Mitchell University of Sydney, Australia K Miura Kajima Corporation, Japan A Miyamoto Yamaguchi University, Japan T Miyoshi Kobe University, Japan G Molinari University of Genoa, Italy T B Moodie University of Alberta, Canada D B Murray Trinity College Dublin, Ireland G Nakhaeizadeh DaimlerChrysler AG, Germany M B Neace Mercer University, USA D Necsulescu University of Ottawa, Canada F Neumann University of Vienna, Austria S-I Nishida Saga University, Japan
H Nisitani Kyushu Sangyo University, Japan B Notaros University of Massachusetts, USA P O’Donoghue University College Dublin, Ireland
R O O’Neill Oak Ridge National Laboratory, USA
M Ohkusu Kyushu University, Japan G Oliveto Universitá di Catania, Italy R Olsen Camp Dresser & McKee Inc., USA E Oñate Universitat Politecnica de Catalunya, Spain
K Onishi Ibaraki University, Japan P H Oosthuizen Queens University, Canada E L Ortiz Imperial College London, UK E Outa Waseda University, Japan A S Papageorgiou Rensselaer Polytechnic Institute, USA
J Park Seoul National University, Korea G Passerini Universita delle Marche, Italy B C Patten University of Georgia, USA G Pelosi University of Florence, Italy G G Penelis Aristotle University of Thessaloniki, Greece
W Perrie Bedford Institute of Oceanography, Canada
R Pietrabissa Politecnico di Milano, Italy H Pina Instituto Superior Tecnico, Portugal M F Platzer Naval Postgraduate School, USA D Poljak University of Split, Croatia V Popov Wessex Institute of Technology, UK H Power University of Nottingham, UK D Prandle Proudman Oceanographic Laboratory, UK
M Predeleanu University Paris VI, France M R I Purvis University of Portsmouth, UK I S Putra Institute of Technology Bandung, Indonesia
Y A Pykh Russian Academy of Sciences, Russia
F Rachidi EMC Group, Switzerland M Rahman Dalhousie University, Canada K R Rajagopal Texas A & M University, USA T Rang Tallinn Technical University, Estonia J Rao Case Western Reserve University, USA A M Reinhorn State University of New York at Buffalo, USA
A D Rey McGill University, Canada
D N Riahi University of Illinois at Urbana-
Champaign, USA B Ribas Spanish National Centre for Environmental Health, Spain K Richter Graz University of Technology, Austria S Rinaldi Politecnico di Milano, Italy F Robuste Universitat Politecnica de Catalunya, Spain J Roddick Flinders University, Australia A C Rodrigues Universidade Nova de Lisboa, Portugal F Rodrigues Poly Institute of Porto, Portugal C W Roeder University of Washington, USA J M Roesset Texas A & M University, USA W Roetzel Universitaet der Bundeswehr Hamburg, Germany V Roje University of Split, Croatia R Rosset Laboratoire d’Aerologie, France J L Rubio Centro de Investigaciones sobre Desertificacion, Spain T J Rudolphi Iowa State University, USA S Russenchuck Magnet Group, Switzerland H Ryssel Fraunhofer Institut Integrierte Schaltungen, Germany S G Saad American University in Cairo, Egypt M Saiidi University of Nevada-Reno, USA R San Jose Technical University of Madrid, Spain F J Sanchez-Sesma Instituto Mexicano del Petroleo, Mexico B Sarler Nova Gorica Polytechnic, Slovenia S A Savidis Technische Universitat Berlin, Germany A Savini Universita de Pavia, Italy G Schmid Ruhr-Universitat Bochum, Germany R Schmidt RWTH Aachen, Germany B Scholtes Universitaet of Kassel, Germany W Schreiber University of Alabama, USA A P S Selvadurai McGill University, Canada J J Sendra University of Seville, Spain J J Sharp Memorial University of Newfoundland, Canada Q Shen Massachusetts Institute of Technology, USA X Shixiong Fudan University, China G C Sih Lehigh University, USA L C Simoes University of Coimbra, Portugal
A C Singhal Arizona State University, USA P Skerget University of Maribor, Slovenia J Sladek Slovak Academy of Sciences, Slovakia
V Sladek Slovak Academy of Sciences, Slovakia
A C M Sousa University of New Brunswick, Canada
H Sozer Illinois Institute of Technology, USA D B Spalding CHAM, UK P D Spanos Rice University, USA T Speck Albert-Ludwigs-Universitaet Freiburg, Germany
C C Spyrakos National Technical University of Athens, Greece
I V Stangeeva St Petersburg University, Russia
J Stasiek Technical University of Gdansk, Poland
G E Swaters University of Alberta, Canada S Syngellakis University of Southampton, UK J Szmyd University of Mining and Metallurgy, Poland
S T Tadano Hokkaido University, Japan H Takemiya Okayama University, Japan I Takewaki Kyoto University, Japan C-L Tan Carleton University, Canada E Taniguchi Kyoto University, Japan S Tanimura Aichi University of Technology, Japan
J L Tassoulas University of Texas at Austin, USA
M A P Taylor University of South Australia, Australia
A Terranova Politecnico di Milano, Italy A G Tijhuis Technische Universiteit Eindhoven, Netherlands
T Tirabassi Institute FISBAT-CNR, Italy S Tkachenko Otto-von-Guericke-University, Germany
N Tosaka Nihon University, Japan T Tran-Cong University of Southern Queensland, Australia
R Tremblay Ecole Polytechnique, Canada I Tsukrov University of New Hampshire, USA R Turra CINECA Interuniversity Computing Centre, Italy
S G Tushinski Moscow State University, Russia
J-L Uso Universitat Jaume I, Spain E Van den Bulck Katholieke Universiteit
Z-Y Yan Peking University, China S Yanniotis Agricultural University of Athens,
D Van den Poel Ghent University, Belgium R van der Heijden Radboud University,
A Yeh University of Hong Kong, China J Yoon Old Dominion University, USA K Yoshizato Hiroshima University, Japan T X Yu Hong Kong University of Science &
Leuven, Belgium
Netherlands
R van Duin Delft University of Technology, Netherlands
Greece
Technology, Hong Kong
P Vas University of Aberdeen, UK R Verhoeven Ghent University, Belgium A Viguri Universitat Jaume I, Spain Y Villacampa Esteve Universidad de
M Zador Technical University of Budapest,
F F V Vincent University of Bath, UK S Walker Imperial College, UK G Walters University of Exeter, UK B Weiss University of Vienna, Austria H Westphal University of Magdeburg,
R Zarnic University of Ljubljana, Slovenia G Zharkova Institute of Theoretical and
Alicante, Spain
Germany
J R Whiteman Brunel University, UK
Hungary
K Zakrzewski Politechnika Lodzka, Poland M Zamir University of Western Ontario, Canada
Applied Mechanics, Russia
N Zhong Maebashi Institute of Technology, Japan
H G Zimmermann Siemens AG, Germany
Materials Characterisation V Computational Methods and Experiments
Editors A.A. Mammoli University of New Mexico, USA C.A. Brebbia Wessex Institute of Technology, UK A. Klemm Glasgow Caledonian University, UK
Editors: A.A. Mammoli University of New Mexico, USA C.A. Brebbia Wessex Institute of Technology, UK A. Klemm Glasgow Caledonian University, UK Published by WIT Press Ashurst Lodge, Ashurst, Southampton, SO40 7AA, UK Tel: 44 (0) 238 029 3223; Fax: 44 (0) 238 029 2853 E-Mail:
[email protected] http://www.witpress.com For USA, Canada and Mexico Computational Mechanics Inc 25 Bridge Street, Billerica, MA 01821, USA Tel: 978 667 5841; Fax: 978 667 7582 E-Mail:
[email protected] http://www.witpress.com British Library Cataloguing-in-Publication Data A Catalogue record for this book is available from the British Library
ISBN: 978-1-84564-538-0 ISSN: 1746-4471 (print) ISSN: 1743-3533 (on-line) The texts of the papers in this volume were set individually by the authors or under their supervision. Only minor corrections to the text may have been carried out by the publisher. No responsibility is assumed by the Publisher, the Editors and Authors for any injury and/ or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or operation of any methods, products, instructions or ideas contained in the material herein. © WIT Press 2011 Printed in Great Britain by Martins the Printers. All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise, without the prior written permission of the Publisher.
Preface
The increasing demands for high-quality products from both industry and consumers are the driving force for the rapid developments in materials science and engineering. In principle materials science involves relating the desired properties and relative performance of a material to its microstructural features through characterization. The major determinants of the structure of a material and hence its properties are its constituent chemical elements and the way a material has been processed into its final form. Over the years, a variety of experimental techniques have been developed for characterizing the physical and chemical properties of materials. Unfortunately due to a number of simplifying assumptions and limitations on the use of individual methods, it is not often possible to describe in a qualitative, reliable way the microstructural features of many materials. Triangulation of different experimental methods as well as computer simulations may become essential to achieve a thorough, comprehensive analysis. Simulations can contribute to the understanding of the phenomena and to provide a good basis for the development of durable materials and components which can withstand ambient and extreme environmental conditions. The way forward in material characterisation is to develop new experimental techniques or apply existing methodologies adopted from other related disciplines. A very wide range of materials, starting with metals through polymers, semiconductors to composites, necessitates a whole spectrum of experimental techniques and numerical models, which are specific for material types. Some of these well established methodologies could potentially find applications in new fields. In this context a multidisciplinary approach in material characterisation and the exchange of original ideas is indispensible. The aim of the International Conference on Computational Methods and Experiments in Materials Characterisation held in Kos, Greece, was therefore to facilitate such interdisciplinary interactions within the research community. The
resulting conference book has been arranged in several chapters addressing various experimental and numerical methods. The wide range of topics covers mechanical characterisation and testing, corrosion problems and thermal analysis as well as recycled materials, nano-composites and energy materials. The editors would like to express their gratitude to all authors without whose involvement this book could not have been produced. We wish to aknowledge the valuable input of the members of the Scientific Advisory Committee in attracting and selecting many high quality contributions. We trust that this book will present some innovative ideas and will facilitate further developments in materials science. The Editors, Kos, Greece 2011
Contents Section 1: Micro and nano characterisation of cementitious materials (Special session organised by A. J. Klemm) Application of positron annihilation lifetime spectroscopy to nano-characterisation of polymer-modified mortars P. Guagliardo, A. J. Klemm, S. N. Samarin & J. F. Williams ............................. 3 Multi-technique investigation of calcium hydroxide crystals at the concrete surface E. Gueit, E. Darque-Ceretti, P. Tintillier & M. Horgnies ................................. 15 Characterization of the influence of the casting mould on the surface properties of concrete and on the adhesion of a protective coating M. Horgnies, P. Willieme, O. Gabet, S. Lombard & M. Dykman...................... 27 Section 2: Nano-materials HRTEM techniques applied to nanocrystal modeling: towards an “atom-by-atom” description D. G. Stroppa, L. A. Montoro, E. R. Leite & A. J. Ramirez ............................... 41 Ca(OH)2 nanoparticle characterization: microscopic investigation of their application on natural stones V. Daniele & G. Taglieri ................................................................................... 55 Nanocarbon composite materials with optical response on radioactive waste M. Vantsyan, G. Popova, E. Karpuzova, M. Bobrov, O. Plaksin & E. Dabek ........................................................................................................ 67
Section 3: Corrosion problems Evaluation of the fretting corrosion mechanisms on the head-cone interface of hip prostheses I. Caminha, C. R. M. Roesler, H. Keide, C. Barbosa, I. Abud & J. L. Nascimento ............................................................................................ 77 Improving corrosion performance by surface patterning M. Bigdeli Karimi, V. Stoilov & D. O. Northwood ............................................ 85 Material characterisation to understand various modes of corrosion failures in large military vehicles of historical importance A. Saeed, Z. Khan, N. Garland & R. Smith........................................................ 95 Section 4: Computational models and experiments A multi-factor interaction model (MFIM) for damage initiation and progression C. C. Chamis.................................................................................................... 109 Analytical solution of a two-dimensional elastostatic problem of functionally graded materials via the Airy stress function H. Sakurai........................................................................................................ 119 Moment curvature analysis of concrete flexural members confined with CFRP grids A. Michael & P. Christou ................................................................................ 131 Application of effective media theory in the characterization of the hygrothermal performance of masonry Z. Pavlík, E. Vejmelková, L. Fiala, M. Pavlíková & R. Černý......................... 143 3D FIB reconstruction and characterisation of a SOFC electrode S. Chupin, N. Vivet, D. Rochais & E. Bruneton............................................... 155 Modelling of load transfer between porous matrix and short fibres in ceramic matrix composites J. G. P. Silva, D. Hotza, R. Janssen & H. A. Al-Qureshi................................. 165 Modeling aspects concerning the axial behavior of RC columns H. O. Koksal, T. Turgay, C. Karakoç & S. Ayçenk.......................................... 175
Section 5: Innovative experiments Surface characterization of eucalyptus and ash wood veneers by XPS, TOF-SIMS, optic profilometry and contact angle measurements G. Vázquez, R. Ríos, M. S. Freire, G. Antorrena & J. González-Álvarez........ 187 Interface resistances in heat and moisture transport: semi-scale experimental analysis Z. Pavlík, J. Mihulka, J. Žumár, M. Pavlíková & R. Černý ............................. 199 Section 6: Mechanical characterisation and testing Tension/compression test of auto-body steel sheets with the variation of the pre-strain and the strain rate G. H. Bae & H. Huh ........................................................................................ 213 Definition of averaged elastic-plastic characteristics of sandwich panel structures I. I. Zakirov, V. N. Paimushin & I. M. Zakirov ................................................ 227 Hot deformation and mechanical properties of P/M Al special M. Tercelj, P. Cvahte, I. Perus & G. Kugler ................................................... 239 Coarsening kinetics of the bimodal γ′ distribution in DS GTD111TM superalloy V. S. K. G. Kelekanjeri, S. K. Sondhi, T. Vishwanath, F. Mastromatteo & B. Dasan ...................................................................................................... 251 Effect of the elastomer stiffness and coupling agents on rheological properties of magnetorheological elastomers A. Boczkowska & S. F. Awietjan...................................................................... 263 Optimization of magnetoelastic properties of pure nickel by means of heat treatments A. L. Morales, A. J. Nieto, J. M. Chicharro, P. Pintado, G. P. Rodríguez & G. Herranz................................................................................................... 275 Nanomechanical structure-property relations of dynamically loaded reactive powder concrete P. G. Allison, R. D. Moser, M. Q. Chandler, T. S. Rushing, B. A. Williams & T. K. Cummins ............................................................................................. 287 Dynamic strength of concrete under multiaxial compressive loading Y. P. Song & H. L. Wang ................................................................................. 299
Modelling and simulation of the rutting resistance of bituminous mixes: experimental and stochastic approaches A. E. Ouni, A. Dony & J. Colin........................................................................ 307 Laboratory tests on the cleanliness of soil materials used as subgrades in pavement structures A. Athanasopoulou & G. Kollaros................................................................... 315 Use of additives to improve the engineering properties of swelling soils in Thrace, Northern Greece A. Athanasopoulou & G. Kollaros................................................................... 327 Characteristics of a bolted joint with a shape memory alloy stud N. Ould-Brahim, A.-H. Bouzid & V. Brailovski............................................... 339 Section 7: Thermal analysis Experimental validation of a thermal model of adhesively bonded scarf repairs for CFRP composite materials incorporating cure kinetics C. C. N. Bestley, S. G. R. Brown & S. M. Alston ............................................. 351 Computational and experimental characterization of building envelopes based on autoclaved aerated concrete V. Kočí, J. Výborný & R. Černý....................................................................... 363 Section 8: Recycled materials Quantitative description of the morphology of polyurethane nanocomposites for medical applications J. Ryszkowska & B. Waśniewski ...................................................................... 377 Description methods of the properties of composites from oxybiodegradable foil waste and wood J. Ryszkowska & K. Sałasińska........................................................................ 387 The effect of slag composition on recycling of “OFHC” through the “ESCM” process S. Ketabchi, F. K. Ahadi, K. Hanaee & S. H. Alhoseini .................................. 397 Author Index .................................................................................................. 407
Section 1 Micro and nano characterisation of cementitious materials (Special session organised by A. J. Klemm)
This page intentionally left blank
Materials Characterisation V
3
Application of positron annihilation lifetime spectroscopy to nano-characterisation of polymer-modified mortars P. Guagliardo1, A. J. Klemm2, S. N. Samarin1 & J. F. Williams1 1
ARC Centre of Excellence for Antimatter-Matter Studies, School of Physics, University of Western Australia, Australia 2 School of Built and Natural Environment, Glasgow Caledonian University, UK
Abstract Positron annihilation lifetime spectroscopy (PALS) has been applied to study the microstructural features of immature cement mortars. Two types of cement mortars containing superabsorbent polymers (SAPs) were studied, in addition to Ordinary Portland Cement (OPC). The ortho-positronium lifetimes for all samples were in the range of 1.70-1.73 ns, values that are close to that of free water (1.7 ns) and hence suggest the presence of water-filled pores. Periodic lifetime measurements showed that the intensity of this component decreased slightly over a period of four weeks, indicating water loss associated with the curing process, evaporation or a combination of the two. Keywords: cement mortar, superabsorbent polymer, positron annihilation lifetime spectroscopy, positronium, porosity, hydration, curing.
1 Introduction Cement is a material of immense practical importance, but in spite of its almost ubiquitous use, its microstructural characterisation still proves to be problematic. The complexity of the problem is enhanced when the composition of cement is modified through the use of auxiliary agents. Ordinary Portland cement is a combination of gypsum (CS*H2) and clinker. Gypsum acts to prevent rapid setting of the cement paste while clinker is the hydraulic binder. The major constituents of clinker are tricalcium silicate (C3S), dicalcium silicate (C2S), WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line) doi:10.2495/MC110011
4 Materials Characterisation V tricalcium aluminate and tetracalcium alumina-ferrite (C4AF); C = CaO, S = SiO2, A = Al2O3, F = Fe2O3, S* = SO3 and H = H2O (abbreviations commonly used in industrial nomenclature) [1, 2]. The presence of water among the active compounds of cement leads to a rearrangement of their structure and the initiation of the hardening process. The initial hardening is caused by the hydration of C3S, which forms a gel-like silicate and calcium hydrate phase referred to as C-S-H gel. These particles ultimately crystallise and bind together the particles of sand or stone into a hard mass. Other hydrates of the complexes described above are also formed, although the reaction rates may differ considerably. The final product is a hardened cement paste which is a mixture of unreacted cement particles, hydration products and pores. Two types of pores are usually distinguished capillary pores, which comprise the water- or air-filled spaces between the hydrates, with sizes in the range of 10-1000 nm, and smaller pores referred to as gel pores, which are contained in the amorphous hydrate phase. Gel and capillary pores form a continuous network of pores throughout the material [3, 4]. The mechanical properties of cementitious materials are heavily influenced by porosity, with the volume and size distribution of pores controlling both the strength and durability of the material [5]. A detailed characterisation of the pore structure and the factors that affect it are thus crucial to advancing the design of these materials; however, classical porosimetry methods such as mercury intrusion porosimetry (MIP) and gas adsorption cannot always be relied upon to extract information on pores smaller than a few nanometres [6]. In this regard, positron annihilation lifetime spectroscopy (PALS) has considerable advantages over these classical methods. PALS is sensitive to both open and closed porosity, for example, whereas isolated pores are invisible to the aforementioned techniques. In addition, PALS is sensitive to pores in the size range of 0.3-30 nm (that is, in the size range of gel pores), and both pore sizes and relative concentrations can be measured (refer to the following section for a detailed explanation) [7]. In this work, PALS was used to study the features of the porosity of three cement mortars. Two types of cement mortars containing superabsorbent polymers (SAPs) were studied, in addition to a sample composed only of OPC. SAPs are cross-linked networks of hydrophilic polymers with a high capacity for water uptake - they can absorb and retain up to 500 times their weight in water. This makes them ideal for use in water-absorbing applications such as absorbent medical dressings and controlled release media [8]. During the mixing process the polymers absorb the pore solution immediately after their addition to the mortar, reaching saturation within minutes [9, 10]. They then swell to form spherical cavities filled with water. At later stages of the hydration process the water is released to the concrete matrix and the cavities remain as empty pores. For mortars with a low water-to-cement ratio, it is possible to replace part of the irregular capillary pores with larger spherical pores formed by saturated polymers. The dispersion and size of these pores can be estimated by the material attributes of polymers.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
5
The samples were studied with PALS over a period of 4 weeks with measurements starting when the samples were aged 21 days, in order to elucidate the changes in nano-porosity caused by the addition of SAPs as well as the changes arising from the curing/hydration process. In addition, samples aged at 3 and 6 weeks were studied with MIP to determine changes in their pore distributions and bulk densities. MIP can provide information on larger pores (from 100 nm to 100 µm) to which PALS is not sensitive.
2 Positron annihilation background Positron annihilation techniques have been applied to study nano-porosity in a wide range of materials, such as zeolite, silicates and polymers [7]. In positron annihilation lifetime spectroscopy (PALS), positrons from a 22Na radioisotope are injected into the sample under study and a positron lifetime spectrum is recorded [11]. As a 1.274 MeV gamma-quantum is emitted during the decay of 22 Na almost simultaneously with a positron, the lifetime is defined as the time between detection of the 1.274 MeV photon and the subsequent 511 keV photons created from the positron’s annihilation with an electron. The measured spectrum is a histogram of the time periods between these two events. Ideally, it is a sum of decaying exponentials of the form
∑
/
(1)
positron state with lifetime . An where is the intensity or weighting of the experimentally obtained spectrum differs from this form however, in that it is convolved with the time response function of the apparatus, usually approximated by a sum of Gaussians. Porosimetry is made possible by exploiting the phenomenon of positronium (Ps) formation, Ps being the hydrogen-like bound state of a positron and an electron [12]. In brief, positrons penetrate the material under investigation and rapidly lose energy, predominately through ionising collisions. Ps formation can then take place as a reaction between the positron and a secondary electron produced during the positrons’ thermalisation (these secondary electrons are often termed spur electrons [13]). Positronium is formed in two spin states with dramatically different annihilation characteristics. The singlet state (with antiparallel orientation of spins) is termed para-positronium (p-Ps); it has a vacuum lifetime of 125 ps and decays via the emission of two gamma quanta. The triplet state (with parallel spins of electron and positron) is referred to as orthopositronium (o-Ps) and has a much longer vacuum lifetime of 142 ns, decaying via three gamma emission. Due to its short lifetime, p-Ps is not significantly perturbed by the material; however, o-Ps, because of its intrinsically longer lifetime, interacts strongly with the material and its pore structure. In the presence of matter, o-Ps can decay into two gammas via a process known as pick-off annihilation - the positron in o-Ps annihilates with an electron of opposite spin in the material via a two-gamma process. As a result of pick-off WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
6 Materials Characterisation V annihilation, the o-Ps lifetime in matter is significantly shorter than its vacuum value (and can be as low as ~ 1 ns) [7]. The basis for studying porosity is that o-Ps tends to localise in regions of low density, such as pores. This is because in the bulk of the material the positron and the electron in o-Ps experience repulsive Coulomb and exchange forces respectively. As a result, most of these materials (porous insulators) have a negative work function for positronium [7]. Porosimetry is possible because the annihilation rate (the inverse of the lifetime) in the pore is a function of the pore size – smaller pores have a shorter lifetime compared to large pores, where the overlap of the Ps wave function with the pore walls is reduced. A semi-empirical quantum mechanical relation has been developed by Tao [14] and Eldrup et al. [15] to relate the o-Ps lifetimes to pore radii: 1 τ
= 2 1-
R R+ΔR
+
1 2π
sin
2πR
(2)
R+ΔR
where is the lifetime, R+R is the pore radius and R is empirically determined to be 0.16-0.17 nm. This model is applicable to sub-nanometre pores of a spherical geometry. Gidley et al. [16, 17] extended this model to include larger pore sizes and varying pore geometries (cylinders, cubes, channels and sheets).
3 Experimental 3.1 Materials and mixes For the purpose of this research Portland cement (BS EN 197-1 CEM II/B-V 32,5) was mixed at 1:1 ratio with fine sand (the vast majority of particles were distributed below 0.6 mm). Throughout the investigation the total water-tocement ratio of 0.45 was maintained. The mix compositions are presented in Table 1. The pastes were shaped into cylinders with diameters of about 25 mm and thicknesses in the range of 5-10 mm. Table 1:
Composition of cement mortars.
Sample designation
OPC
SAP-A
SAP-B
Mix code
R
A
B
(Water/Cement)total
0.45
0.45
0.45
(Water/Cement)effective
0.45
0.425
0.438
Sand/Cement
1
1
1
SAP content [%] (by cement weight)
0
0.25
0.25
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
7
The SAPs used in this study were cross-linked polymers provided by BASF. SAP-A is a copolymer of acrylamide and acrylic acid and SAP-B is a polymer based on acrylic acid. The products had absorption capacities of 200-250 ml/g in demineralised water, though the absorption in mortar depended on the product and was approximately 10 mL/g for SAP-A and 5 mL/g for SAP-B. Both materials were prepared by grinding and screening to sizes of 63-125 µm, but there was also a minor (less than 10%) content of finer particles. The SAP particle shapes were irregular and their sizes after initial absorption were in the range of 135-270 µm for SAP-A and 105-210 µm for SAP-B. 3.2 Positron lifetime and mercury intrusion measurements For positron lifetime measurements, approximately 30 μCi of 22NaCl was deposited on 7μm Kapton foil and covered with an identical foil. The edges of this foil-sandwich were then sealed, and this source foil was placed between two identical pieces of sample. The positron lifetimes were measured with a fast-fast coincidence system. The gamma-ray detectors consist of a truncated cone (31.8 mm diameter tapering to 12.7 mm with a height of 12.7 mm) BC418 scintillator coupled to a Burle 8850 photomultiplier tube. The time resolution of the system is approximately 220 ps, as determined from analysis of a spectrum of highpurity annealed nickel. The spectra comprise of at least 2 million counts and have been analysed using PAScual version 1.3.0 [18]. Mercury intrusion porosimetry (MIP) has been carried out with the use of a Porosimeter Autopore IV 9500 by Micromeritics, with a pressure range up to 60000 psi for all samples at ages 3 and 6 weeks.
4 Results and discussion In Table 2 the fitted lifetimes and intensities for all samples are given at four different ages. In each case, the lifetime spectra could be fitted with three discrete components. In porous media a number of annihilation processes are possible because multiple positron and positronium states exist. It is likely that the two shorter lifetime components (1 and 2) in Table 2 are the averages of several different annihilation modes. These could include positron annihilation (of both free and trapped positrons), p-Ps self-annihilation and o-Ps pick-off annihilation in the bulk of the material, and annihilations in the source foil. Nanosecond lifetime components (3) on the other hand are typically associated with o-Ps pick-off annihilation in nano-pores. For all of the cement samples the average 3 over the course of the measurements was in the range of 1.70-1.73 ns. The presence of a nanosecond component is often an indicator of porosity or free volume, and these values would correspond to pores having diameters in the range of 0.51-0.52 nm (using the TE model, equation 2). However, these lifetimes are very close to that observed in pure water, approximately 1.7 ns [19]. Given that cement pastes are known to contain a significant fraction of water, and that the samples were tested at a relatively young age (testing began at 21 days after mixing), it seems likely that these WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
8 Materials Characterisation V nanosecond components are associated with annihilation in water contained in capillary pores of the cement and hence cannot be directly related to a pore size. Consolati and Quasso [20, 21], Consolati et al. [22] and Salgueiro et al. [23] observed similar lifetimes in Portland cement pastes and attributed this component to water-filled pores.
OPC
SAP-B
SAP-A
Table 2: Age (weeks) 3 4 5 6 3 4 5 6 3 4 5 6
Lifetimes and intensities; standard deviations are given in brackets. 1 (ps)
2 (ps)
218 (9) 227 (9) 242 (7) 233 (8) 219 (5) 229 (5) 221 (6) 237 (6) 224 (6) 239 (5) 238 (5) 212 (9)
410 (10) 410 (10) 430 (10) 420 (10) 409 (7) 421 (9) 409 (8) 420 (10) 402 (8) 430 (10 420 (10) 390 (10)
3 (ns)
I1 (%)
I2 (%)
I3 (%)
2
1.67 (0.02) 1.67 (0.03) 1.73 (0.03) 1.72 (0.03) 1.73 (0.02) 1.76 (0.03) 1.71 (0.03) 1.63 (0.03) 1.71 (0.02) 1.67 (0.02) 1.67 (0.03) 1.73 (0.03)
43 (4) 46 (5) 55 (4) 50 (4) 45 (2) 49 (3) 46 (3) 53 (4) 47 (3) 57 (3) 53 (3) 43 (3)
50 (3) 48 (5) 40 (4) 44 (4) 49 (2) 46 (3) 50 (3) 43 (4) 48 (3) 39 (2) 42 (3) 53 (4)
5.9 (0.1) 5.7 (0.2) 5.2 (0.2) 5.2 (0.2) 5.0 (0.1) 4.0 (0.1) 4.3 (0.1) 4.2 (0.2) 4.6 (0.1) 4.3 (0.1) 4.2 (0.1) 4.3 (0.1)
0.93 1.04 0.99 0.99 1.15 1.08 1.04 1.08 1.01 1.01 1.09 1.07
In Figure 1, o-Ps lifetimes (3) and intensities (I3) are plotted as a function of sample age. Lifetime spectra were recorded every 7 days for a 4 week period. Over the course of the measurements, I3 for the cement sample containing SAPA is somewhat higher than for the cement containing SAP-B and the pure cement sample. If these lifetimes are indicative of water-filled pores then this suggests a higher concentration of pores in the sample containing SAP-A. As the results for SAP-B are closer to those of pure cement, this suggests that the presence of the polymer does not significantly alter the pore concentration. For the SAP-A sample, a gradual decrease in I3 is observed over weeks 3-5, whereas for the SAP-B and pure cement samples there is a decrease after the initial measurement (weeks 3-4), and then I3 is relatively constant. While these changes are small, they may be indicative of various stages of the hydration process, as the decrease in I3 is consistent with the reduction in total porosity that is known to occur as hydration progresses [3]. The gradual decline seen in the SAP-A sample suggests that this polymer releases its water more slowly compared to SAP-B. At the onset of the measurement (age 21 days) the hydration process is likely to be still underway, with both di- and tri-calcium silicates reacting with water to produce a network of calcium silicate hydrates. Water could then be consumed in the hydration process for at least the next two weeks. However, during this time it is difficult to say if the loss of water seen by the decrease in I3 is also associated with evaporation. As the samples lose water one might expect the lifetimes to increase as there would be a transition from filled to empty pores.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
9
The fact that this does not occur suggests that pores collapse with water loss (the volume of capillary pores is known to decrease with hydration [3]). These results are consistent with the work of Consolati and Quasso [20, 21] who observed a systematic decrease in I3 with sample age, but they are in contrast to the work of Myllylä and Karras [24] where the opposite trend was seen. It should be noted however that a direct comparison with these studies is not possible due to differing experimental conditions (namely the composition and age of samples).
o-Ps intensity, I3 (%)
6.5 6.0 5.5 5.0 4.5 4.0 3.5 3
4
5
6
Sample age (weeks)
o-Ps lifetime, 3 (ns)
2.2 2.0 1.8 1.6 1.4 1.2 3
4
5
6
Sample age (weeks) Figure 1:
o-Ps intensities (top) and lifetimes (bottom) vs. sample age for SAP-A (circles), SAP-B (triangles) and Ordinary Portland Cement (squares).
In addition to PALS, MIP measurements were carried out on samples aged 3 and 6 weeks; the resulting pore distributions are shown in Figure 2. It is apparent that the samples contain a wide distribution of pores, with the majority of pores WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
10 Materials Characterisation V
0.25 0.2 0.15 0.1 0.05 0 1000000 100000
10000
1000
100
10
Log differential intrusion (mL/g)
in the size range of 10-100 nm. The MIP results are in qualitative agreement with the PALS data in the sense that the sample containing SAP-B displays similar behaviour to pure cement - at 6 weeks their pore distributions almost coincide and the total porosity for both samples has decreased with aging. This decrease is also consistent with the reduction in I3 seen in the PALS data.
1
0.25 0.2 0.15 0.1 0.05 0 1000000 100000
10000
1000
100
10
1
Log differential intrusion (mL/g)
Pore size diameter (nm)
Pore size diameter (nm) Figure 2:
Pore size distributions determined by MIP at age 3 weeks (top) and 6 weeks (bottom); OPC – diamonds, SAP-A – squares, SAP-B – triangles.
Conversely, the SAP-A sample shows a significant increase in the pore concentration within the 10-100 nm range at 6 weeks. This is in contrast to the decrease in I3 seen in PALS, although PALS may not be sensitive to the porosity in this size range. In spite of the differing trend, both PALS and MIP show that WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
11
the presence of SAP-A significantly alters the hydration process of the cement mortar. The bulk densities of the samples were also determined by MIP measurements (Table 3). It was found that the density of all samples increased with aging time. Again, the changes in density for the pure cement and SAP-B samples are similar. The change in density for SAP-A is much smaller, which is consistent with the slower rate of change of I3. Table 3: Sample OPC SAP-A SAP-B
Bulk densities determined by MIP at 3 and 6 weeks. Bulk density at 3 weeks (g/mL) 1.83 1.82 1.84
Bulk density at 6 weeks (g/mL) 1.89 1.86 1.91
In order to gain further insight into the nature of the porosity, the samples were heated at 116°C for 4 hours under vacuum conditions (10-5 Torr) and lifetime measurements were repeated. Lifetime spectra were recorded directly after heating and the samples were sealed in a plastic bag containing a desiccant (to prevent water absorption) during the measurement. The resulting lifetimes and intensities are given in Table 4 and the corresponding pore sizes are given in Table 5. At least three types of water can be distinguished in a hydrating cement paste. Chemically-bound water is directly incorporated into the structure of the hydration products, physically-bound water is absorbed on the surfaces of cement particles and reaction products, and there is also “free” water contained in the capillary and gel pores [25]. The heat treatment will have removed a significant fraction of the latter two sources of water; however, much higher temperatures (above 950°C) are required to remove chemically-bound water. Table 4:
Lifetimes and intensities obtained after heat treatment.
1 (ps) 2 (ps) 3 (ns) 4 (ns) I1 (%) I2 (%) I3 (%) I4 (%) 2
OPC 263 (2) 536 (7) 4.3 (0.1) 75 (2) 24 (1) 1.3 (0.1) 1.13
Sample SAP-A 214 (9) 380 (9) 1.32 (0.07) 12.6 (0.7) 36 (8) 60 (7) 3.3 (0.4) 1.0 (0.1) 1.05
SAP-B 219 (4) 413 (5) 1.62 (0.05) 14.9 (0.6) 44 (2) 52 (2) 2.5 (0.1) 0.70 (0.01) 1.01
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
12 Materials Characterisation V After heat treatment, 3 for OPC increases to 4.3 ns. This corresponds to a gel pore diameter of 0.86 nm using the TE model (equation 2). However, the intensity of this component decreases by a factor of about 3.3; this could suggest that a significant fraction of the pores collapse upon water removal. There is also a significant redistribution of intensity from I2 to I1 after heat treatment. These are the shorter components associated with multiple positron states in the material. The redistribution of intensity from the longer component (2) to the shorter one (1) indicates that the heat treatment has reduced the number of positron traps in the bulk material. This could be associated with an increase in the bulk density after heating. Table 5:
Pore diameters, d3 and d4, calculated from o-Ps lifetimes, and total o-Ps intensities (Io-Ps).
d3 (nm) d4 (nm) Io-Ps before (%) Io-Ps after (%)
OPC 0.86 4.3 1.3
Sample SAP-A 0.42 1.45 5.2 4.3
SAP-B 0.50 1.56 4.2 3.2
For the cement samples containing SAPs an additional low intensity, long lifetime component (4) appears in the spectra after heat treatment (and due to the presence of this long component it was necessary to fix the background to the left of the main coincidence peak to the average value in the analysis). 4 may be associated with pores formed by the polymers which have replaced some of the ordinary pore structure, and its emergence shows that the presence of SAP significantly alters the structure of the cement paste. For the sample containing SAP-A, 3 = 1.32 ns and 4 = 12.6 ns, with respective intensities of 3.3% and 1.0%. These lifetimes would correspond to pores having diameters of 0.42 and 1.45 nm respectively. There is also a redistribution of the weighting of the shorter components (1 and 2); however this is not as pronounced as for ordinary Portland cement. For the sample containing SAP-B, 3 = 1.62 ns and 4 = 14.9 ns, with intensities of 2.5% and 0.7% respectively. These lifetimes correspond to pores having diameters of 0.42 and 1.45 nm respectively. The value of 3 is comparable to that obtained before heat treatment (which was similar to the o-Ps lifetime in water). This could suggest that the heat treatment did not result in the complete removal of free water. For both SAP-A and SAP-B the total o-Ps intensity (I3 + I4) decreases after heat treatment, signalling a reduction in porosity; however, the reduction is larger for the ordinary Portland cement sample.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
13
5 Conclusion This study has shown that PALS can be used to monitor the hydration process in cement samples by using the o-Ps intensity as an indicator of the concentration of water-filled pores. In addition, PALS has shown that the addition of a SAP to the cement mixture alters the structure and hydration process of cement. The addition of SAP-A caused the hydration process to progress more gradually compared to OPC; however, the addition of SAP-B gave comparable results to that of OPC. Both PALS and MIP showed that the presence of SAP-A has a significant effect on the hydration process. In addition, MIP showed that the concentration of pores in the 10-100 nm range increases with aging for the SAP-A sample. For the samples with SAP added, heat treatment under evacuation (resulting in the removal of free water) gave rise to an additional long lifetime component and a higher overall o-Ps intensity compared to OPC. This shows that the SAP had modified the pore distribution, replacing some of the regular pore distribution with larger pores formed by saturated polymers.
References [1] [2] [3] [4] [5] [6] [7] [8] [9]
[10]
[11] [12] [13] [14]
Bogue, R. H. The Chemistry of Portland Cement. (Reinhold, 1955). Lea, F. M. The Chemistry of Cement and Concrete. (Arnold, 1970). Neville, A. M. Properties of Concrete. 4th edn, (Longman, 1995). Taylor, H. F. W. Cement Chemistry. (Academic, 1990). Ghosh, S. N. Advances in cement technology (Pergamon, Oxford, 1983). Gregg, S. J. & Sing, K. S. W. Adsorption, Surface Area and Porosity. (Academic, 1982). Jean, Y. C., Mallon, P. E. & Schrader, D. M. Principles and applications of positron and positronium chemistry (World Scientific, 2003). Buchholz, F. L. & Graham, A. T. Modern Superabsorbent Polymer Technology (John Wiley & Sons, 1997). Jensen, O. M. & Hansen, P. F. Water-entrained cement-based materials: I. Principles and theoretical background. Cement and Concrete Research 31, 647-654, (2001). Jensen, O. M. & Hansen, P. F. Water-entrained cement-based materials: II. Experimental observations. Cement and Concrete Research 32, 973-978, (2002). Krause-Rehberg, R. & Leipner, H. S. Positron Annihilation in Semiconductors. Vol. 127 (Springer, 1999). Charlton, M. & Humberston, J. W. Positron Physics. (Cambridge University Press, 2001). Mogensen, O. E. Spur reaction model of positronium formation. The Journal of Chemical Physics 60 (1974). Tao, S. J. Positronium Annihilation in Molecular Substances. The Journal of Chemical Physics 56 (1972).
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
14 Materials Characterisation V [15] Eldrup, M., Lightbody, D. & Sherwood, J. N. The temperature dependence of positron lifetimes in solid pivalic acid. Chemical Physics 63, 51-58, (1981). [16] Gidley, D. W. et al. Positronium annihilation in mesoporous thin films. Physical Review B 60, R5157 (1999). [17] Gidley, D. W. et al. Determination of pore-size distribution in lowdielectric thin films. Applied Physics Letters 76 (2000). [18] Pascual-Izarra, C. et al. Advanced fitting algorithms for analysing positron annihilation lifetime spectra. Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment 603, 456-466, (2009). [19] Patro, A. P. & Sen, P. Anomalous parapositronium lifetime in water at 20 degrees C. Journal of Physics C: Solid State Physics 5, 3273 (1972). [20] Consolati, G. & Quasso, F. Evolution of porosity in a Portland cement paste studied through positron annihilation lifetime spectroscopy. Radiation Physics and Chemistry 68, 519-521 (2003). [21] Consolati, G. & Quasso, F. A positron annihilation study on the hydration of cement pastes. Materials Chemistry and Physics 101, 264-268 (2007). [22] Consolati, G., Dotelli, G. & Quasso, F. Positron lifetime spectroscopy as a probe of nanoporosity of cement-based materials. Radiation Physics and Chemistry 58, 727-731 (2000). [23] Salgueiro, W., Somoza, A., Cabrera, O. & Consolati, G. Porosity study on free mineral addition cement paste. Cement and Concrete Research 34, 9197, (2004). [24] Myllylä, R. & Karras, M. Positron Annihilation Probing for the Hydratation Rate of Cement Paste. Applied Physics 7, 303-306 (1975). [25] Sen Wang, P. U. et al. 1H nuclear magnetic resonance characterization of Portland cement: molecular diffusion of water studied by spin relaxation and relaxation time-weighted imaging. Journal of Material Science 33, 3065-3071 (1998).
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
15
Multi-technique investigation of calcium hydroxide crystals at the concrete surface E. Gueit1, E. Darque-Ceretti1, P. Tintillier2 & M. Horgnies2 1
MINES ParisTech, Center for Material Forming, Sophia Antipolis, France 2 Lafarge Centre de Recherche, St Quentin Fallavier, France
Abstract The durability and aesthetic qualities of high-performance concrete, which makes it particularly suitable for architectural applications, are constantly compromised by environmental aggressions. In this study, an innovative solution was developed to protect the concrete from these aggressions, which consists of growing a mineral coating on the concrete surface. The coating is composed of layered calcium hydroxide crystals, whose nucleation and growth are triggered byvarious non-ionic surfactants (the details of the process will not be presented). This paper describes the procedure used to investigate the structure of the formed crystals. Scanning Electron Microscopy and optical microtopography were used to determine the morphology of the crystals. Image analysis allowed the quantification of their amount, size and shape. The contribution and limits of each technique are discussed. Keywords: concrete, scanning electron microscopy, image analysis, optical microtopography, surface.
1 Introduction The mechanical and aesthetic durability of concrete is often compromised by the constant environmental aggressions to which the structures are exposed (organic or inorganic particles, algae, micro-organisms, staining from various sources). It is possible to protect concrete from these attacks and increase its durability by applying organic coatings on the hardened surface, but this comes with operational and environmental costs.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line) doi:10.2495/MC110021
16 Materials Characterisation V An innovative solution was proposed [1], which consists on covering and protecting high-performance concrete (HPC) with a mineral coating made of calcium hydroxide crystals (CH). The crystals growth happens during the concrete setting and is triggered by the presence of non-ionic surfactants at the concrete/formwork interface. One of the difficulties of this study was to determine the adequate characterization methods to properly investigate the influence of various surfactants on the amount, morphology and size of the crystals. Several techniques exist to observe, measure and study concrete hydrates, but not all of them are directly suitable for surface investigation. An original procedure had to be developed, combining several techniques. The structure and morphology of the crystals were assessed through Scanning Electron Microscopy (SEM) both on the concrete surface and on polished section. The SEM observations were completed by microtopography on the concrete surface. The quantification of the amount and size of the crystals was made by image analysis on binocular images of the concrete surface. The purpose of this paper is to illustrate what each of these techniques can bring to concrete surface studies.
2 Morphology of the crystals The most common method to observe concrete hydrates morphology is Scanning Electron Microscopy [2]. Most of the time, the observations are made on polished section in secondary electron mode. When higher magnifications or no polishing are required, fresh fractures can also be observed. Surprisingly, it is very rare to find published picture of concrete surface directly observed by SEM. In this study, SEM observations were conducted on both surfaces and crosssections of the concrete. Small cubes (1 cm x 1 cm x 1 cm) were cut from each concrete sample. For surface observation, the cubes were directly carbon-coated on the adequate face and observed in secondary electron mode. For crosssections observations, the cubes were impregnated, polished and carbon-coated, and then observed in back-scattered electron mode. A SEM FEG Quanta 400 from FEI Company was used at an accelerating voltage of 15kV and current intensity of 1mA.
Figure 1:
SEM observations of a standard high-performance concrete surface in secondary electron mode (A) and in back-scattered electron mode on a polished section (B).
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
17
Figure 1 shows observations of both the surface (A) and the cross-section (B) of a standard high-performance concrete. No specific features are visible on the surface, except for a couple of scratch due to the formwork defects. As for the polished section, it reveals a classical concrete microstructure, with dark grey aggregates and light grey un-hydrated cement grains surrounded by cement paste. Figure 2 shows SEM observation of two cross-sections from two different samples. These observations reveal the presence at the concrete surface of a thin layered structure presenting various orientations and organization, sometimes well-aligned parallel to the surface (A), sometimes arranged in a more chaotic way in the first micrometers of the surface (B). At this level of observation, it is not possible to clearly indentify the nature of this unexpected phase. The limited magnification in back-scattered mode, the thin structure of the hydrates and the surface damages due to the polishing all complicate the interpretation. This is why it is necessary to make complementary observations of the surface itself.
Figure 2:
SEM observation in back-scattered electron mode of two polished sections from two different concrete samples from the study.
Figure 3 shows the same samples as figure 2 observed from the surface in secondary electron mode. These micrographs allow a better understanding of the structure of the crystals. All of them share the same layered and flaky structure with different orientations. In all cases, the crystals are composed of thin leaves that grow around the nucleation point. For some crystals, these leaves are strictly parallel to the surface and grow as ‘flower petals’ around the central point, reaching sometimes a perfect hexagonal shape. For other crystals, the leaves are strictly perpendicular to the surface, forming a very regular spherulitic structure. This very organized structure was not visible on the polished sections, where the leaves appeared randomly implanted in the surface. This is due to the polishing, which damaged the first micrometers of the surface and disturbed the structure. The characteristic hexagonal shape of the crystals, as well as their layered organization, allow identifying them as calcium hydroxide – no other concrete phase would present this morphology. This is confirmed by EDS analysis, where mostly calcium is detected is these areas (spectrum not showed). WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
18 Materials Characterisation V
Figure 3:
SEM observation in secondary electron mode of the surface of two different samples from the study.
3 Amount and size of the crystals 3.1 Microtopography SEM observations were very useful to highlight the presence of an unusual phase and to identify it as CH crystals, but they are not sufficient to get a complete insight on the structure of the crystals. Even though the secondary electrons give information on the surface topography, it is difficult to evaluate from the micrographs how thick the crystals are or how deep they are embedded in the surface. As for the polished section, the crystals layers are too strongly delaminated by the polishing for the observation to be conclusive. This is why the observations were completed with profilometry of the surface. Mechanical and optical profilometry are sometimes used on concrete to assess its roughness and evaluate its behavior regarding adhesion problems [3–5]. In our case, the measurements were made using optical profilometry only, because a mechanical probe is likely to damage the very fragile CH crystals. The measurements were made on a confocal full-field 3D surface profilometer with a spot of 2 micrometers and a working distance of 4.5 mm. In these conditions, the vertical resolution is 0.01 micrometers and the lateral resolution 0.1 micrometer. Areas of 4 x 4 mm² were scanned with a step of 10micrometer. The data were computed using the software MountainsMap. In the examples given below, there were only two steps of data treatments: the maps were straightened to compensate for the horizontality defects, and profiles were extracted. Figure 4 shows an example of profilometry on a sample where the crystals are oriented preferentially parallel to the surface. Two patterns appear on the 2D mapping: regular and slightly curved lines, which are due the formwork texture, and irregular spots, which are the CH crystals. Below the mapping is a roughness profile extracted along the dotted line. The narrow peaks correspond to the lines, the larger one at 1.7 mm corresponds to the CH crystal in the middle. The profile allows measuring the size of the crystal (400 micrometers) and the height between the concrete surface and the top of crystal (5 micrometers). WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
19
Figure 4:
Example of a mapping of the concrete surface. The profile was extracted from the mapping along the black dotted line.
Figure 5:
Example of a mapping of the concrete surface. The profile was extracted from the mapping along the black dotted line.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
20 Materials Characterisation V The same measurements can be made of figure 5. The mapping was acquired on the same sample, but in an area were the CH crystals have been pulled out of the surface during the demoulding phase, leaving hollows the size of the crystals. The depth of the hollows gives the height between the surface and the bottom of the crystals. The measurements from figures 4 and 5 give an estimation for the total thickness of the crystals of 20 micrometers. Of course, the operation has to be repeated a large number of times to obtain statistical results. This is only one example taken from one of the samples. It should be noted that profilometry gives no conclusive information for the samples where the crystals are oriented perpendicular to the surface. Figure 6 shows a mapping of such a sample (A): the spherulitic structures appear slightly lighter than the surface, but are difficult to distinguish. A zoom on the mapping (B) reveals that the crystals are indeed apparent, but that the resolution is far from being sufficient to separate the different flakes composing the structure.
Figure 6:
Example of a mapping from a sample with perpendicular crystals. The 3D mapping on the right is a zoom from the 2D mapping on the left.
3.2 Image analysis The only way to know the amount of CH crystals as well as their dimension and geometrical parameters is to count and measure them one by one. Image analysis software are capable of automatically achieve this procedure on a large number of images, provided that they are correctly settled by the operator. Image analysis has been successfully used in previous studies [2, 6] to assess the amount of calcium hydroxide in concrete as well as the geometrical parameters of the particles (size, shape, etc.). In those cases, image analysis was conducted on SEM micrographs from polished sections. In our case, there is no need to work on sections as the crystals are on the concrete surface. Furthermore, the crystals are so large that SEM micrographs are not suitable for a proper counting. Even at low magnification, there are too few particles in the observation field. On the contrary, images taken under a binocular have the right scale to count and measure the particles. WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
21
The CH crystals are shinier than the cement paste, which makes them visible with the naked eye, but this is not sufficient for image analysis software to distinguish them. To overcome this problem, the samples were coloured with a black felt-tip pen prior to analysis. The black ink penetrates the cement paste, strongly colouring it, but is not absorbed by the crystals which then become clearly distinguishable. For each concrete samples, five small areas of the surface (1 cm2) were coloured and photographed under a binocular. The images were treated and analysed using the open-source software ImageJ [7]. The most difficult step in image analysis is to separate the studied particles from the background. In our case, the following procedure was used. Figure 7 shows an example of an image prior to treatment. The brightness and contrast of the images were optimized using the automatic function on the software. The optimization was not based on the whole image, but on the histogram analysis of a small area of one of the particle on the image. This step created a highly contrasted image where particles appeared red and the background black (figure 8).
Figure 7:
Figure 8:
Example of an image before treatment by the software imageJ.
Example of an image after optimization of the contrast and brightness.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
22 Materials Characterisation V The noise was removed using the ‘remove outliers’ function of the software, which replaces each pixel by the median value of its neighbours in a given area (figure 9). In this case, a radius of 20 pixels was considered, and the removing was set to occur if the difference between the considered pixel and the median of its neighbour was higher than 10.
Figure 9:
Example of an image after noise removal.
The images were then converted to binary and inverted. At this point, the background appeared white and the particles black (figure 10).
Figure 10:
Example of an image after conversion to binary.
The holes in the particles were closed using the ‘Fill holes’ function of the software (figure 11). Finally, the ‘watershed’ function was used to automatically draw the outlines of the particles (figure 12). This step might be the main cause of errors in the measurements, as watershed segmentation works best for convex objects that do not overlap too much. In some cases, the software was not able to correctly separate clusters of particles, which were then counted as one big particle.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
Figure 11:
Figure 12:
23
Example of an image after filling the holes.
Example of an image after the software automatically delimited the particles.
Once the image was treated, the particles were counted one by one. The smallest ones (with a surface smaller than 0.1 mm²) were not considered as they are more likely residual noise than actual crystals. The particles overlapping with the boarder of the image were not considered either as their size would not be correctly measured. This means than the measured amount of crystals (from 1 to 20 particles per mm² depending on the samples) is slightly smaller than reality. The following parameters were measured for each particle: - The coordinates of the centre of the particles. These coordinates were used to calculate the distance between each crystal and its closest neighbour and to verify that the nucleation appeared randomly and homogeneously on the surface (figure 13). - The Feret’s diameter, which is the longest distance between two opposing points of the particle. - The fraction area of the surface covered by the crystals. As these crystals are artificially grown to act as a mineral coating, this fraction area must be as high as possible. The results showed that even if the crystals seemed very large, they hardly cover 50% of the surface, except in one promising case where almost 100% of the surface was coated. WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
24 Materials Characterisation V -
The circularity of the particle, which is defined by 4.area / perimeter². A perfect circle has a circularity of 1, whereas a very elongated particle has a circularity close to 0 (figure 14). This geometrical parameter is very convenient to distinguish the crystals that grow parallel to the surface with regular shapes (high circularity) from the crystals that grow perpendicular to the surface, which have a low circularity.
Figure 13:
Example of spatial distribution of the particles on the concrete surface, showing that the nucleation occurs rather homogeneously on the surface.
Figure 14:
Two examples of the circularity distribution. The curve marked with squares corresponds to a sample where the crystals have a low circularity because they grow perpendicular to the surface. The curve marked with dots corresponds to a sample where the crystals have a high circularity because they grow parallel to the surface and adopt very regular shapes.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
25
The results of the image analysis allowed a better understanding of the influence of various surfactants on the nucleation (amount of particles) and growth (size and orientation of the particles) of the CH crystals. Two types of sample could not be properly analyzed though, one because it exhibited excessive roughness that induced too much background noise on the images, and one because the crystals overlapped too much to be clearly separated. In those cases, a more sophisticated procedure has to be developed. In particular, software capable of separating overlapping particles exists, but they were not tested yet in this study.
4 Conclusion Three techniques – scanning electron microscopy, microtopography and image analysis – were successfully used to assess the morphology and geometrical parameters of calcium hydroxide crystals at the concrete surface. These techniques are well-known and developed, but not necessarily widely used in concrete research. Yet, they have proved very efficient in this case to make a preliminary study of a new phenomenon – the massive growth of CH crystals in presence of surfactants, bringing complementary information and results. Of course, they absolutely do not make further investigation any less necessary. For example, a proper crystallographic study would be essential to fully understand the growing mechanisms and the action of the surfactants. In the field of materials characterization, it is important to be creative and to combine and adapt existing techniques. This is particularly true in the field of concrete research, and even more when it comes to concrete surface, a topic which is slowly emerging and where a lot of fascinating research still waits to be done.
References [1] Gueit, E., Darque-Ceretti, E., Tintillier, P. & Horgnies, M., Surfactantinduced growth of calcium hydroxide at the concrete/formwork interface as a mineral coating for concrete, Manuscript submitted for publication. [2] Skalny, J., Gebauer, I. & Odler, I., (eds). Calcium Hydroxide in Concrete, The American Ceramic Society: Westerville, 2001. [3] Garbacz, A., Courard, L., & Kostana, K., Characterization of concrete surface roughness and its relation to adhesion in repair systems, Materials Characterization, 56, pp. 281-289, 2006. [4] PER09 Perez, F., Bissonette, B. & Courard, L., Combination of mechanical and optical profilometry techniques for concrete surface roughness characterisation. Magazine of Concrete Research, 61(6), pp. 389-400, 2009. [5] Ramirez, A.M., Demeestere, K., De Belie, N., Mäntylä, T., & Levänen, E, Titanium dioxide coated cementitious materials for air purifying purposes: Preparation, characterization and toluene removal potential, Building and Environment, 45, pp. 832-838, 2010.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
26 Materials Characterisation V [6] Gallucci, E. & Scrivener, K., Crystallisation of calcium hydroxide in early model and ordinary cementitious systems, Cement andConcrete Reseach,37, pp. 492-501, 2007. [7] NIH, http:\\rsbweb.nih.gov/ij
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
27
Characterization of the influence of the casting mould on the surface properties of concrete and on the adhesion of a protective coating M. Horgnies, P. Willieme, O. Gabet, S. Lombard & M. Dykman Lafarge Centre de Recherche, St Quentin-Fallavier, France
Abstract Protective coatings are deposited on concrete to improve aesthetics and to prevent ageing. However, their adhesion on concrete depends on several interlinked parameters. In this study, the surfaces of concrete are characterized according to the process of casting and post-treatment used (sandblasting) by using Scanning Electron Microscopy (SEM), Fourier Transformed-Infrared (FTIR) spectroscopy and profilometry. The surface properties are correlated to the adhesion force of a polyurea (PU) coating. The development of a specific peel test (a strengthened and porous membrane is introduced into the layer of liquid coating before its crosslinking) ensures a reproducible debonding of the coating/concrete system and allows measuring the fracture energy. Moreover, the interface after debonding is analyzed by FT-IR to highlight the presence of concrete/coating residues and to determine the locus of failure. Results underline that the nature of casting mould influences the concrete surface and modifies the adhesion of PU coating. The mould made of polyoxymethylene (POM) induces a micro-tearing of the extreme surface of concrete during demoulding. By increasing the roughness and the open porosity of the concrete surface, this tearing enhances the adhesion of the coating. On the contrary, the smooth concrete surface, induced by the use of a polyvinylchloride (PVC) mould, decreases the anchorage of the coating. Finally, the sandblasting of the surface, by increasing the roughness and the interface area, is an interesting treatment to promote the adhesion of PU coating, whatever the mould used for the casting. Keywords: concrete, coating, roughness, FT-IR, SEM, peel test, adhesion.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line) doi:10.2495/MC110031
28 Materials Characterisation V
1 Introduction The staining of concrete could occur due to its specific microstructure that retains the liquid and dust particles. The deposition of a coating is then important to close the surface porosity and protect concrete against acid rains, settlement of algae and lichens…etc [1, 2]. PU coatings are commonly used due to their high resistance against chemical and mechanical aggressions [3, 4]. However, the surface properties of concrete depend on several interlinked parameters as chemical composition, intrinsic porosity and roughness. The aim of this study concerns the influence of the casting process and post-treatment on the adhesion between PU coating and concrete surface. The influences of the casting conditions on the hardened concrete surface were characterized by SEM, FT-IR and profilometry. Secondly, this study was undertaken to determine if the surface properties of concrete could influence the adhesion of PU coating. Fracture energies were measured by a specific 90°-peel test. This method was retained because it is appropriate to characterize the adhesion of thin films [5–8]. Some publications have already showed the use of strengthened membrane or mesh sheet that were incorporated into the bulk of soft material to characterize [9]. Concerning our system, the introduction of a polymer membrane into the bulk of the coating was necessary to strengthen the system and measure a reproducible adhesion of the PU coating. After the peel tests, the FT-IR analyses of the debonded faces were undertaken to detect the residues of concrete or coating and determine the loci of failure. FT-IR spectroscopy allows detecting organic compounds of coatings [10, 11] and several components of concrete [12].
2 Material and methods 2.1 Material 2.1.1 Substrates made of hardened concrete A high-performance concrete was prepared by mixing 31% of white Portland cement (CEM I 52.5 PMES from Lafarge), 9% of limestone filler (DURCAL 1), 7% of silica fumes (MST), 43.5% of sand (BE01) and 1.5% of admixture. A water to cement ratio (W/C) of 0.26 was used. The samples were prepared by pouring the fresh concrete mixture into horizontal and rectangular formwork (15x12x1 cm) made of PVC or POM. The concrete samples were removed from their formworks after 18 hours and were stored during 28 days under ambient conditions (25°C; 50% relative humidity) to complete their hydration. Some concrete samples were then sandblasted after demoulding (by using a powder of corindon) to increase their roughness. The sandblasted samples were cleaned by air flow to remove the dust before the deposition of coating. 2.1.2 Coating and conditions of deposition The PU coating was composed of 50% of isocyanates diluted into 45% of solvent (butyl acetate). A catalyst (dibutylétain laurate, DBTL) was added into the mix to initiate the reaction with water (present in concrete or in atmosphere). WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
29
The isocyanates units reacted with water to produce a polyurea-based film [13]. The coatings were sprayed (120 g/m²) with air pressure of 3 atmospheres on the concrete surface. The drying period was about 3 days under ambient conditions (25°C, 50% relative humidity) before peeling. 2.2 Methods of characterization 2.2.1 Scanning electron microscopy (SEM) Samples were characterized by using a high-resolution field-effect gun digital scanning electron microscope (SEM FEG Quanta 400 from FEI Company; using an accelerating voltage of 15 keV and a current intensity of 1 nA). Images of the cross-sections were obtained after being polished. 2.2.2 Profilometry The roughness of concrete samples was measured with a Surftest SJ-201 M mechanical profilometer (Mitutoyo) in order to calculate the arithmetic mean of the profile deviations from the mean line (Ra). The Ra value was obtained by compiling the arithmetic mean of 5 profiles of 12.5 mm. 2.2.3 Fourier transform-infrared spectroscopy (FT-IR) The FT-IR spectrometer Nicolet iS10 (Thermo Fisher Scientific Inc.) was equipped with a deuterated triglycine sulfate (DTGS) detector and controlled by OMNIC software. The Attenuated Total Reflexion (ATR) mode was mainly used in this study. FT-IR (in ATR mode) characterized the sample over a thickness of a few µm. The sampling area analyzed was approximately 1 mm². The crystal used was made of diamond and 16 scans were routinely recorded over the range 4,000-650 cm-1 with a spectral resolution of 4 cm-1. The background was collected at ambient atmosphere before analyzing each sample. Spectra were corrected with a linear baseline. No specific preparations of the samples of concrete and coating were performed before FT-IR analyses: they were studied just after demoulding or after debonding. 2.2.4 Specific peel tests of concrete/coating system Peel test allows measuring the debonding force (F). According to the peel angle (θ) and the width of the adhesive coating (w), the fracture energy (G) could be calculated according to [14]: F G w
1 cos
In the specific case of 90° peel angle, G is equal to the peeling force (F) divided by the width (w) of the adhesive coating. The coating was strengthened by inserting a membrane. The membrane was firstly cut in polyamide 6.6 tissues (Nytex from Dutscher SAS) and deposited on each concrete samples. The coating was directly sprayed on this membrane in order to avoid any air bubble at the interface. Finally, the edges of the membrane were cut to avoid any shear during the debonding. The membrane was flexible but non-stretched under the
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
30 Materials Characterisation V solicitations that occurred during the peel test. The dimensions of each membrane were 200 mm long, 20 mm wide, 120 µm thick, with an open porosity of 50% and a mesh opening of 150 µm. We observed by SEM (Figure 1) a crosssection of the concrete/coating interface in order to describe the system. No mechanical step was used to initiate the crack at the interface. All the 90°-peel tests were performed with a 1000 N sensor, equipped with a specific mobile table (as detailed by Figure 2). All the peel tests were performed under ambient conditions (25°C, 50% relative humidity) and by using a constant speed of peel (0.2 mm/s). Strengthened membrane
PU coating (130 µm thickness)
Concrete surface
Figure 1:
Figure 2:
SEM image of a cross-section of the concrete/coating system.
Schema and image of the experimental test of peeling.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
31
3 Results 3.1 Analyses of the surfaces of reference materials 3.1.1 Concrete surfaces according to the nature of the mould Table 1 details the IR bands of concrete (demoulded with PVC). Portlandite (at 3640 cm-1) and C-S-H*/silica (large band at 1078 cm-1) were well detected. Moreover, three strong bands were also assigned to the presence of CaCO3. Their detection confirmed previous studies [12]. Detection of CaCO3 could be due to the carbonation process but also due to the use of limestone filler in concrete mix. As presented by Table 1, concrete demoulded with POM (spectrum not shown) underlines distinct IR bands at 935 and 900 cm-1 that were assigned to alite [15] (a component of the cement before hydration). This FT-IR spectrum showed also peaks at 1092 and 800 cm-1, assigned to Si-O bonds from silica fumes [12]. These silica fumes and alite are usually present into the bulk of concrete sample. Moreover, no IR band of the portlandite (Ca(OH)2) was detected. Table 1:
Assignment of FT-IR bands recorded on reference materials. FT-IR bands (cm-1)
Concrete (PVC mould)
O-H; Ca(OH)2
3640
X
C=O; CaCO3
1410; 872; 710
X
X
Si-O; silicates, CSH
1080-970
X
X
X
X
Assignment
Concrete (POM mould)
PU coating
Si-O; C3S
935; 900
Si-O; silica
797; 777
X
CH2/CH3; methyl units
2850; 2950; 1450
X
N-H; urea
3335
X
NCO; isocyanates
2270
X
C=O; urea
1690
X
The concrete samples were also studied by SEM. Concrete demoulded with POM (Figures 3a and 3b) appears to be rougher than concrete demoulded with PVC (Figures 3c and 3d). The flat topography observed after using PVC mould contrasted with the heterogeneous surface obtained after using a POM mould. These observations confirmed the measurements obtained by profilometry (Table 2): the Ra values of concrete could vary according to the demoulding process. Indeed, the results obtained by SEM suggest that a tearing of the extreme surface of concrete could happen during the removing of the POM mould. This hypothesis could explain why silica fumes were easily observed by SEM (Figure 3b) and why alite was detected by FT-IR spectroscopy (Table 1).
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
32 Materials Characterisation V (a)
(b)
Silica fumes
(d)
(c)
Figure 3:
SEM images recorded on concrete samples: (a, b) if POM mould; (c, d) if PVC mould.
Table 2:
Roughness of concrete according to the demoulding and posttreatment.
Ra roughness (µm)
Concrete (POM mould)
Reference
0.7 µm +/- 0.3
Concrete (PVC mould) 0.5 µm +/- 0.3
After sandblasting
12.2 µm +/- 0.4
12.1 µm +/- 0.5
3.1.2 Concrete surfaces after sandblasting Table 2 summarizes the roughness of sandblasted concrete, compared to the one before demoulding. The initial roughness of 0.5-0.7 µm range (depending on the nature of mould used) increased to a higher Ra values of 12 µm after sandblasting (whatever the nature of the mould used). SEM images of the sandblasted concrete, presented in Figure 4, could be compared to the ones of the reference concrete samples (Figure 3). The sandblasted surface was rougher than after demoulding. Silica fumes were also detected after sandblasting. Figure 5 compares FT-IR spectra recorded on concrete (PVC mould) samples before and after sandblasting. The intensity of the bands assigned to the CaCO3 compounds (at 1410, 870 and 710 cm-1) decreased after sandblasting while the intensity of peaks of alite (at 930-900 cm-1) and silica (at 1100 cm-1) increased. WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
33
Silica fumes
Figure 4:
SEM images of concrete (PVC mould) samples after sandblasting.
CaCO3
(b)
(a) Silica
Figure 5:
Alite
FT-IR spectra of concrete samples (PVC mould): (a) before sandblasting; (b) after sandblasting.
3.1.3 PU coating with membrane FT-IR spectroscopy is known to determine the degree of crosslinking of the isocyanates because a band at 2270 cm-1 characterizes presence of isocyanates units [16, 17]. In this study, no results of adhesion were linked to the degree of crosslinking of the coating. However, Table 1 underlines other IR bands assigned to the presence of urea units (at 3335 and 1690 cm-1). These bands were used to detect the residues of PU coating after peeling. The membrane made of WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
34 Materials Characterisation V polyamide 6.6. was also analyzed by FT-IR in order to assign the reference peaks (amide units at 3300, 3070, 1630 and 1530 cm-1). However, these IR bands of the polyamide were never detected at the interface after debonding. 3.2 Results of peel tests and FT-IR analyses of the interface after debonding 3.2.1 Influence of the mould used to cast the concrete Figure 6 compares the peel curves obtained during the debonding of PU coating from concrete samples. The adhesion of the coating varied according to the nature of the casting mould. 1400
1200
(a) Concrete (POM mould)
G (N/m)
1000
800
600
(b) Concrete (PVC mould)
400
200
0 0
10
20
30
40
50
60
70
80
Length of debonding (mm)
Figure 6:
Peel curves of coating recorded on concrete samples: (a, in blue) if POM mould; (b, in red) if PVC mould.
As described by Table 3, the fracture energy measured during the peel of the PU coating from concrete (POM mould) is higher (1000 N/m +/- 150) than the one measured during the peel from concrete (PVC mould): 400 N/m +/-50. Figure 7a presents the FT-IR spectrum recorded on concrete (POM mould) side after peeling: almost all the IR bands were assigned to the residues of coating that covered the entire surface. These results confirm that the high fracture energy was linked to a debonding into the bulk of the coating. It could also be deduced that the adhesion between concrete and coating was better than 1000 N/m. On the contrary, Figure 7b shows FT-IR spectrum of concrete (PVC mould) side after peeling: compounds of concrete (such as CaCO3, silica and portlandite) were mainly detected. Concerning the internal side of coating after peeling from concrete (PVC mould) sample, all the IR bands (spectrum not shown) were assigned to the own components of the coating. No FT-IR bands could be assigned to concrete residues.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
Table 3:
Sample
35
Results obtained by peel tests and by FTIR analyses of the interface. Post-treatment of concrete before coating
Fracture energy (N/m)
Failure localization deduced from FTIR analyses
/
1000 +/- 150
Inside the PU coating
Sandblasting
1200 +/- 100
Inside the PU coating
/
400 +/- 50
At concrete/PU interface
Sandblasting
1200 +/- 100
Inside the PU coating
Concrete/POM
Concrete/PVC
(a) Concrete side (POM mould) after debonding
(b) Concrete side (PVC mould) after debonding
Figure 7:
FTIR spectra of concrete sides after peeling: (a) if POM mould; (b) if PVC mould.
These data allow concluding that the debonding occurred at the interface between concrete and coating. The fracture energies and FT-IR analyses of the locus of failure could be correlated to the SEM observations of concrete surface WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
36 Materials Characterisation V before the deposition of the coating. Figure 3 had highlighted that the surface of concrete demoulded with POM was more heterogeneous than concrete demoulded with PVC, which was smoother. We can deduce that the liquid coating (before crosslinking) strongly diffused and anchored into the heterogeneous concrete surface, demoulded with POM. On the contrary, the smooth and close concrete demoulded with PVC did not allow a high mechanical anchorage of the coating. 3.2.2 Influence of the sandblasting of concrete surface The peels tests undertaken on the sandblasted concrete (PVC mould) highlighted high fracture energies (about 1200 N/m) while the peel undertaken on reference concrete (PVC mould) samples induced a low level of adhesion of the PU coating (350-400 N/m), as described in Figure 8a. Concerning concrete demoulded with POM, the fracture energies (presented in Figure 8b) did not 1600
(a)
1400
Sandblasted concrete (PVC mould)
1200
G (N/m)
1000 800 600
Not-sandblasted concrete (PVC mould)
400 200 0 0
10
20
30
40
50
60
70
80
Lenght of debonding (mm)
1600
(b) 1400
Sandblasted concrete (POM mould)
1200
G (N/m)
1000
Not-sandblasted concrete (POM mould)
800 600 400 200 0 0
10
20
30
40
50
60
70
80
Length of debonding (mm)
Figure 8:
Peel curves of the coating recorded on concrete with or without sandblasting: (a) if PVC mould; (b) if POM mould.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
37
evolve: they were as high after sandblasting as after demoulding. Thanks to these results, we can conclude that a sandblasting of concrete demoulded with PVC could be interesting to enhance the adhesion of PU coating. As described by SEM images (Figure 4) and values of roughness (Table 2), the open porosity and the roughness of concrete demoulded with PVC were higher after the sandblasting (Figure 4c) than after demoulding (Figure 3c). On the contrary, the sandblasting was less helpful with concrete demoulded with POM because the open porosity was already present after demoulding. FT-IR analyses of concrete sides were performed after peeling but only residues of coating were detected on these concrete sides (whatever the nature of mould used). The debonding occurred inside the layer of the PU coating if the substrate had been sandblasted before the deposition of the coating.
4 Conclusion A specific methodology of 90°-peel test was used to measure the fracture energy between concrete and PU coating. By introducing a porous and thin membrane to strengthen the layer of coating, measurements were reproducible and allowed comparing the fracture energies with the FT-IR analyses of the loci of failure recorded after debonding. These results were correlated to the surface properties of the reference concrete surface after demoulding (topography, composition, roughness) that were characterized by several methods. The influence of the nature of mould was significant. Some moulds, such as POM, could induce a micro-tearing of the extreme surface of concrete during the demoulding. This phenomenon increased the roughness (and the open porosity) of concrete and improved the anchorage of the liquid coating (before its crosslinking). On the contrary, the smooth and flat concrete surface induced after using a PVC mould did not allow a good adhesion of PU coating. The influence of the mechanical anchorage was confirmed by the high fracture energies measured on the sandblasted samples of concrete. In future, analyses of the surface porosity and sorptivity will be performed to confirm these results.
References [1] Dubosc, A., Escadeillas, G. & Blanc, P.J., Characterization of biolical stains on external concrete walls and influence of concrete ad underlying material. Cement and Concrete Research, 31, pp. 1613-1617, 2001. [2] Manoudis, P.N., Karapanagiotis, I., Tsakalof, A., Zuburtikutis, I., Kolinkeova, B. & Panayiotou, C., Surface properties of superhydrophobic coatings for stone protection. Journal of Nano Research, 8, pp. 23-33, 2009. [3] Chattopadhyay, D.K. & Raju, K.V.S.N., Structural engineering of polyurethane coatings for high performance applications. Progress in Polymer Science, 32, pp. 352-418, 2007.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
38 Materials Characterisation V [4] Carmona-Quiroga, P.M., Martínez-Ramírez, S., Sobrados, I. & BlancoVarela, M.T., Interaction between two anti-graffiti treatments and cement mortar (paste). Cement and Concrete Research, 40, pp. 723-740, 2010. [5] Barquins, M. & Ciccotti, M., On the kinetics of peeling of an adhesive tape under a constant imposed load. International Journal Adhesion and Adhesives, 17, pp. 65-68, 1997. [6] Horgnies, M., Darque-Ceretti, E. & Combarieu, R., Adhesion of pressure sensitive adhesives to automotive coatings: Influence of topcoat composition. Journal of Adhesion Science and Technology, 18, pp. 10471061, 2004. [7] Johnson, K.L., Kendall, K. & Roberts, A.D., Surface Energy and the contact of elastic solids. Proceedings of Royal Society of London, 324, pp. 301-313, 1971. [8] Kinloch, A.J., Adhesion and adhesives: science and technology. Chapman and Hall, London, New York, pp. 66-73, 1987. [9] Giannis, S., Adams, R.D., Clark, L.J. & Taylor, M.A., The use of a modified peel specimen to assess the peel resistance of aircraft fuel tank sealants. International Journal of Adhesion and Adhesives, 28, pp. 158-175, 2008. [10] Almeida, E., Balmayore, M. & Santos, T., Some relevant aspects of the use of FT-IR associated techniques in the study of surfaces and coatings. Progress in Organic Coatings, 44, pp. 233-242, 2002. [11] Poliskie, M. & Clevenger, J.O., Fourier Transform Infrared (FT-IR) spectroscopy for coating characterization and failure analysis. Organic Finishing, pp. 44-47, 2008. [12] Chollet, M., Horgnies, M., Analyses of the surfaces of concrete by Raman and FT-IR spectroscopies: comparative study of hardened samples after demoulding and after organic post-treatment. Surface and Interface Analysis, In press. [13] Agrawal, R.K. & Drzal, L.T., Adhesion mechanisms of polyurethanes to glass surfaces. Part I. Structure property relationships in polyurethanes and their effects on adhesion to glass. Journal of Adhesion, 54, pp. 79-102, 1995. [14] Kendall, K., Thin-film peeling – the elastic term. Journal of Physics D: Applied Physics, 8, pp. 1449-1452, 1975. [15] Frost, R.L., Cejka, J. & Weier, M.L., Molecular structure of the uranyl silicates – a Raman spectroscopic study. Journal of Raman spectroscopy, 37, pp. 538-551, 2005. [16] Daniel-da-Silva, A.L., Bordado, J.C.M. & Martin-Martinez, J.M., Moisture curing kinetics of isocyanate ended urethane quasi-prepolymers monitored by IR spectroscopy and DSC. Journal of Applied Polymer Science, 107, pp. 700-709, 2007. [17] Agrawal, K. & Drzal, L., Adhesion mechanisms of polyurethanes to glass surfaces. Journal of Adhesion Sciences Technology, 9, pp. 1381-1400, 1995.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Section 2 Nano-materials
This page intentionally left blank
Materials Characterisation V
41
HRTEM techniques applied to nanocrystal modeling: towards an “atom-by-atom” description D. G. Stroppa1,2, L. A. Montoro1, E. R. Leite3 & A. J. Ramirez1,2* 1
Brazilian Synchrotron Light Laboratory, Brazil Mechanical Engineering School, University of Campinas, Brazil 3 Department of Chemistry, Federal University of São Carlos, Brazil 2
Abstract The development of technologies based on nanostructures presents a wide range of challenges for materials scientists and engineers, including the attainment of well-controlled synthesis procedures, the improvement of characterization techniques down to the atomic scale resolution, and the conception and validation of reliable models that can describe materials properties as functions of their morphology and fabrication process. A relevant topic in this scenario is the correlation among the spatial distribution of chemical elements, the surface energy configuration, the growth mechanism, and the resultant nanocrystal 3D morphology. This work presents an overview on the use of advanced HRTEM techniques for the quantitative analysis of nanocrystals and how these results can be used to implement nanocrystals models, which can analytically describe the material features on an atomic level. The presented findings show the combined use of experimental data and theoretical tools, such as image simulation and ab initio surface energy calculations, for the advanced quantitative characterization of nanocrystalline systems. The combination of experimental and theoretical efforts on HRTEM characterization represents a powerful tool for the nanocrystal 3D morphology elucidation with atomic resolution and the chemical/structural properties assessment in a quantitative way. Thereby, it is presented as the stepping stone towards the development of novel approaches to describe nanostructured systems. Keywords: HRTEM, nanocrystals modeling, quantitative analysis. WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line) doi:10.2495/MC110041
42 Materials Characterisation V
1 Introduction Nanotechnology has been considered one of the most promising branches on the scientific development since the 60’s [1] and its contributions to various fields are notorious nowadays. The “nano” suffix and its effects became an intrinsic subject of materials science and technology on the last few years as the unique properties from nanometer scale are becoming increasingly relevant in those fields. The traditional materials engineering perspective [2], depicted in Figure 1, can also be applied for the analysis of nanostructured systems. This approach considers that the design of materials aiming an optimum performance requires the complete understanding of the correlation among the processing routes parameters, the materials structure and the resultant properties. However, regarding nanostructured systems, this correlation may be stated as the interrelationship among the synthesis parameters, the atomic scale morphology and the energy configuration [3]. As the surface area to volume ratio is enhanced for such systems, unique properties related to the surface energy distribution become relevant. These properties are the source for the unique performance of nanomaterials [4].
Figure 1:
Engineering approach for describing and correlating materials features.
Materials characterization plays a crucial role on the development of reliable models for the description and the design of nanostructured materials with specific features for countless technological areas. Apart from the specific questions that different characterization techniques may answer, there are three general issues that can be highlighted as the most relevant questions for the nanostructured materials modeling and design. The first one is related to the morphology characterization at the atomic scale. As mentioned before, it is vital to assess the nanomaterials structure with high resolution in order to evaluate its correlation with the system properties. The ultimate goal in this sense is the “atom-by-atom” description of nanocrystals, WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
43
which would be specially required for the determination of atomic species segregation on alloyed/doped materials [5] and the high resolution strain state analysis of nanostructures [6]. The second issue is related to the nanocrystals surface structure and its dependency on the synthesis environment and chemical species. This aspect is intrinsically correlated to the nanocrystals morphology and, consequently, to the material surface energy distribution and properties [7]. The third relevant point is the interrelation between the nanocrystals and the system growth behavior [8]. These aspects represent crucial issues for the understanding and optimization of synthesis methodology as they are closely related to the nanostructured system stability and long term reliability [9]. As describing the properties and behavior of each nanocrystal in an actual nanostructured system represents an overwhelming many-bodies problem, simplified models are needed for the engineering of nanomaterials in a reliable manner. Nanocrystal modeling [10] is a wide approach where nanostructured systems components are described as building blocks with specific morphology, energy distribution, and interaction mechanisms. In these terms, the development of accurate models may be a fundamental tool for nanocrystals and mesocrystals shapes prediction, design and control of growth processes, and the resulting properties tuning of an unlimited number of systems. The development of reliable modeling methodologies represents a hard task for materials scientists and engineers due to involved challenges, especially when quantitative high resolution characterization analyses are required. Even though advanced characterization techniques based on electron microscopy, scanning probe microscopy and synchrotron radiation are feasible and complimentary in this scenario, limitations [11] associated to each one of them prevent their isolated use a for reliable analysis. The most relevant limitations of high resolution techniques are summarized on Table 1. Table 1:
Most relevant limitations from high resolution characterization tools.
ISSUE Sample Preparation Sample Stability Energy Resolution Signal to Noise Ratio Sampling Averaging Data Analysis
EFFECT modification on the sample original features characteristics changes during analysis spectroscopic signature may have overlaps poor detection system sensitivity lack of statistical representativeness lack of features from individual particles deconvolution, modeling and fitting
Although continuous improvements on the characterization instrumentation and methodologies enhance the individual techniques performance, the combined use different available experimental techniques and theoretical approaches are widely used to overcome the inherent limitations of the quantitative analysis of nanosized systems. WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
44 Materials Characterisation V Transmission Electron Microscopy (TEM) based techniques stand out among other characterization techniques for nanostructured systems due to the possibility of a comprehensive and versatile approach including high resolution imaging, diffraction and spectroscopy. This multipurpose characteristic allied with simulation procedures provides complementary information that can be used to improve the analysis precision. Recent developments on the TEM instrumentation and analysis methods allowed outstanding advances on nanoscience and nanotechnology [12]. The spherical aberration (Cs) correction [13] can be considered among the most influential improvements on both high resolution TEM (HRTEM) and high resolution scanning TEM (HRSTEM) due to a number of factors. For HRTEM, the Cs-correction of the objective lens resulted in the drastic improvement on achievable spatial resolution, the possibility of larger gaps in the pole piece for in situ experiments, and new imaging modes [14] due to fine tuning of the aberration coefficients. For HRSTEM, the Cs-correction of the condenser lens leads to an improvement on the achievable spatial resolution due to the probe size reduction and the effective beam current increase. In addition to the instrumental TEM improvements, the development of more accurate models to describe the image formation and to support the TEM image simulation [15] provides a more reliable interpretation of experimental data. This work presents an overview on the HRTEM techniques state of art with several examples which indicate the development of such characterization tools towards the quantitative high resolution analysis of nanocrystalline materials.
2 HRTEM and HRTEM image simulation HRTEM image formation is based on the incident electron beam scattering by a thin sample. The technique is especially relevant for very thin crystalline samples, where the interaction of the several diffracted beams forms an interference patterns. As the diffracted components correspond to particular oriented periodic spacings on the sample, their interference may appear as a 2D periodic image which cannot be directly interpreted. The direct interpretation unfeasibility is mainly related to the HRTEM interference pattern dependency with the sample thickness and with the microscope configuration, especially the defocus value. However, the use of HRTEM image simulation can provide information about the sample crystalline arrangement. HRTEM image simulation is a crucial step in HRTEM characterization, which supports the atomic structure analysis. Multislice method [16] is the most widely used image simulation procedure and consists in the electrons wave function calculation after the electron beam interaction with the sample projected potential. Imaging distortions due to the lens aberrations, microscope instabilities and detection system defects [17] may be included in order to better reproduce the experimental imaging condition. The HRTEM imaging and image simulation combination has provided numerous examples of successful materials characterization. Remarkable examples can be noticed on the determination and refinement of crystalline WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
45
structures [18] and the quantitative evaluation of the chemical composition profile along the thin films interface [19]. Figure 2 illustrates the combination of HRTEM imaging and image simulation on the evaluation of an anomalous anisotropic growth mechanism for SnO2 nanocrystals [20]. The used approach provided an unambiguous characterization of the growth mechanism and its preferential direction, which would not be feasible from direct imaging only.
Figure 2:
(a) SnO2 nanocrystal HRTEM image with an indexed FFT inset and a HRTEM multislice simulation (red square). (b) oriented attachment along the [110] is identified as the main growth mechanism. [From ref. [20], copyright © 2011 by RSC Publishing, reprinted with permission of authors.]
In addition, the accurate determination of nanocrystals 3D morphology by HRTEM simulation of nanocrystals [21] can be used to indirectly extract quantitative dopant segregation information [22] from systems where conventional analytical techniques are impracticable due to several experimental restrictions. Figure 3 depicts the 3D morphological modeling for Sb:SnO2 nanocrystals with different dopant contents. The nanocrystals models were compared to the Wulff [23] constructions based on ab initio surface energy calculations, providing the dopant atoms segregation for individual particles and its dependency with the doping level [24]. The presented examples depict the synergism between the HRTEM imaging and theoretical methods such as HRTEM image simulation and ab initio surface energy calculations. Therefore, this combination presents a remarkable potential as a tool for unveiling important nanocrystalline systems features, such as 3D morphology, dopant segregation, surface energy distribution, and growth mechanisms.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
46 Materials Characterisation V
Figure 3:
(a), (e) Trimetric and (b), (f) [111] zone-axis oriented views for the Sb-doped nanocrystals geometric model. (c), (g) Original HRTEM images and (d), (h) superimpositions with the simulated HRTEM images. [Adapted from ref. [22], copyright © 2009 by ACS Publications, reprinted with permission of authors.]
3 Focal series reconstruction and geometric phase analysis Although the HRTEM image simulation can support accurate nanocrystals 3D morphology analyses, it is not always feasible to use such procedure to precisely evaluate the position of individual atoms or atomic columns in crystalline structures. To extract information which is directly related to the atomic positions, such as strain/stress state, a refinement on the HRTEM analysis is needed. Focal Series Reconstruction (FSR) [25] is a proposed calculation which is applied for the electron wavefunction restoration on the HRTEM imaging process and allows obtaining the sample projected potential without the interference of the microscope optical aberrations. The FSR implementation is based on the acquisition of a set of images under different objective lens defocus conditions, which is subsequently submitted to a restoration procedure that estimates the contrast transfer function (CTF) and the aberration coefficients implicated on the imaging process. A number of successful uses of FSR aiming a straightforward image interpretation [26], the correction of residual aberration [27], and the improvement of the HRTEM technique spatial resolution [28] have been already described in the literature. An example which illustrates the image improvement when FSR method is applied to structural analysis is depicted on Figure 4. These images show the comparison between an original HRTEM image and a phase image of the reconstructed wavefunction [29]. Geometric Phase Analysis (GPA) [30] is a methodology for measuring and mapping structural displacement fields on HRTEM images using a reference lattice. Its application to HRTEM image analysis allows the local lattice distortion evaluation, which can be directly related to the local strain state. This WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
47
method has been effectively employed to study strain fields in semiconductors and metals [31]. The GPA implementation is based on the reciprocal space evaluation of a HRTEM image with respect to a reference undistorted lattice image, either obtained from a different region in the analyzed sample or from an image simulation procedure. As a spatial frequency spread around the characteristic lattice parameters frequencies is existent on strained samples, the determination of lattice distortions can be performed by the comparison with a reference image. The distorted components can be extracted on the reciprocal space and further translated to real space components through an inverse Fourier transform. Such analysis can be performed for two non-collinear spatial frequencies, resulting in a projected 2D distortion description of the sample which can be related to its 2D strain/stress state.
Figure 4:
(a) Example of an original HRTEM image and (b) the output from a FSR restoration from a CoSi2 sample. The insets exhibit a higher magnification detail with the crystal structure superimposition. [Adapted from ref. [29], reprinted with permission of authors.]
The GPA application is reported on the literature as an efficient tool to calculate strain maps on microelectronic components [32] and for the strain state analysis on individual nanocrystals [33]. A typical output from a GPA analysis is the distortion map from the analyzed HRTEM image, as exemplified on Figure 5, which includes the perpendicular, parallel and rotational strain components. An example of the combined user of FSR and GPA is presented on Figure 6 [34]. Cross-section HRTEM images of epitaxially grown Si1-xGex:Si(001) alloyed islands were used to characterize in a self-consistent way the strain configuration and the local chemical composition with high spatial resolution. By the combination of projected 2D chemical composition mapping from two different cross-sectioned zone-axis views, [100] and [110], it was possible to infer the chemical composition in a three-dimensional fashion [35].
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
48 Materials Characterisation V
Figure 5:
(a) Original Cs-corrected HRTEM image and (b) distortion mapping from a single dislocation on a CeO2 sample.
Figure 6:
(a) FSR reconstruction from a Si-Ge island including a higher magnification detail on the inset. (b) A simplified 3D chemical reconstruction was obtained after the strain chemical mapping of several Si-Ge islands at different zone-axis. An example from a single island: (c) chemical composition map, (d) parallel and (e) perpendicular strain projections. [Adapted from refs. [34] and [35], copyright © 2009 by ACS Publications, reprinted with permission of authors.]
The successful application of this technique to quantify the elastic behavior of the Si-Ge:Si(001) system shows that this methodology arises as a remarkable tool for accurate chemical and elastic state evaluation, which can be applied to several strained alloyed nanostructures, such as epitaxial islands, nanowires, nanocrystals, and thin films.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
49
4 A novel approach for HRTEM imaging The improvement of HRTEM characterization depends on the comprehensive understanding of the underlying image formation theory. An important imaging condition representation is given by the Contrast Transfer Function (CTF), which represents how the spatial frequencies are transmitted through microscope imaging system. Since the 40’s, microscopists have been trying to improve the CTF characteristics by tuning the instruments parameters during the experiments or by the posterior treatment of the images. The main goal in this sense is to extend information transfer for high spatial frequencies. The most remarkable improvements were achieved by using induced defocus values on the objective lens in order to compensate its inherent positive spherical aberration coefficient [36]. However, a ground-breaking evolution happened on HRTEM with the Cscorrection possibilities. At a first glance, the use of Cs-correction adds an additional degree of freedom to the microscope tuning, allowing the CTF direct improvement for high spatial frequencies and improving the microscope information limit. A comparison between the HRTEM images from the same sample obtained by conventional and Cs-corrected microscopes, which is presented on Figure 7, clearly shows the image enhancement provided by the aberration corrector hardware.
Figure 7:
Raw HRTEM images from the same CeO2 nanocrystal from a (a) standard TEM microscope (1.25 Å information limit) and from a (b) Cs-corrected microscope (0.8 Å information limit).
A detailed investigation [37] of the electron scattering on the dynamical regime pointed out that non-linear contributions would enable the achievement of enhanced contrast transfer for low-Z elements by the use of negative Cs values. The Negative Cs Imaging (NCSI) [14] aims the contrast maximization at high frequencies by the microscope parameters optimization, including defocus and Cs adjustments. The NCSI application allowed some unique measurements WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
50 Materials Characterisation V including the high resolution assessment of light elements atomic columns information, such as occupancy next to crystalline defects [38] and sub-angstrom displacement in magnetic domain walls [39]. Examples of NCSI HRTEM characterization for CeO2 sintered and nanocrystalline samples are depicted on Figure 8. Although the contrast for Oxygen columns is observed in both images, the direct analysis of a bulk-like sample HRTEM image (Figure 8a) is unfeasible due to a combination of factors. As the accurate thickness estimation is not achievable by HRTEM image simulation for this specific combination of crystalline system and imaging conditions, the intensity distribution cannot be directly associated to the Oxygen columns scattering. A solution for this issue is the sample projected potential assessment by the FSR application on NCSI HRTEM images (Figure 8b).
Figure 8:
(a) NCSI HRTEM image from a CeO2 bulk-like sample on the [100] ZA and (b) FSR from NCSI HRTEM images from a CeO2 nanocrystal on the same ZA orientation including a unit cell superimposition.
5 Concluding remarks State of art HRTEM techniques indicate that their use in conjunction with theoretical procedures for HRTEM image simulation and data analysis represents a vital tool for the quantitative evaluation of nanostructured systems. The depicted examples illustrate how this technique can simplify, or even present completely novel approaches, a number of materials characterization challenges where a quantitative high resolution assessment is required. The current development of HRTEM instrumentation and related methodologies indicates that its importance on materials characterization will grow on this decade, especially due to the novel possibilities achieved by aberration correction implementation and by on-growing combined use with HRSTEM imaging and spectroscopy techniques.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
51
Acknowledgements The authors acknowledge the financial support of the Brazilian research funding agencies FAPESP and FINEP; the German research funding agency DAAD. The authors would also like to thank the Ernst Ruska Centre staff at the Forschungszentrum Jülich and the QFA staff at the Universitat Jaume I for the fruitful discussions and scientific support.
References [1] [2] [3]
[4]
[5]
[6]
[7]
[8]
[9] [10] [11] [12] [13] [14]
Feynman, R. P.; There’s plenty of room at the bottom; Eng. Sci., 23(5), 22-36, 1960. Callister, W. D.; Fundamentals of Materials Science and Engineering; John Wiley & Sons: New York, pp. 2-5, 2001. Burda, C.; Chen, X.; Narayanan, R.; El-Sayed, M. A.; Chemistry and properties of nanocrystals of different shapes; Chem. Rev., 105(4), 1025-1102, 2005. Kolmakov, A.; Moskovits, M.; Chemical Sensing and Catalysis by OneDimensional Metal-Oxide Nanostructures; Annu. Rev. Mater. Res.; 34, 151-180, 2004. Shibata, N.; Pennycook, S. J.; Gosnell, T. R.; Painter, G. S.; Shelton, W. A.; Becher, P. F.; Observation of rare-earth segregation in silicon nitride ceramics at subnanometre dimensions; Nature; 428, 730-733, 2004. Brune, H.; Giovannini, M.; Bromann, K.; Kern, K.; Self-organized growth of nanostructure arrays on strain-relief patterns; Nature, 394, 451-453, 1998. Yang, H. G.; Sun, C. H.; Qiao, S. Z.; Zou, J.; Liu, G.; Smith, S. C.; Cheng, H. M.; Lu, G. Q.; Anatase TiO2 single crystals with a large percentage of reactive facets, Nature, 453, 638-641, 2008. Dalmaschio, C. J.; Ribeiro, C.; Leite, E. R.; Impact of the colloidal state on the oriented attachment growth mechanism; Nanoscale, 2, 2336-2345, 2010. Yu, W. W.; Wang, Y. A.; Peng, X.; Formation and Stability of CdTe Nanocrystals; Chem. Mater.; 15, 4300-4308, 2003. Barnard, A. S.; Modelling of Nanoparticles: Approaches to Morphology and Evolution; Rep. Prog. Phys., 73, 1-52, 2010. Brundle, C. R.; Evans, C. A.; Wilson, S.; Encyclopedia of Materials Characterization, Butterworth-Heinemann, Reed Publishing USA, 1992. Hawes, P.; Advances in Imaging and Electron Physics: Aberrationcorrected Electron Microscopy; Elsevier, 153, 2008. Haider, M.; Uhlemann, S.; Schwan, E.; Rose, H.; Kabius, B.; Urban, K.; Electron Microscopy Image Enhanced; Nature, 392, 768-769, 1998. Jia, C. L.; Lentzen, M.; Urban, K.; High-resolution transmission electron microscopy using negative spherical aberration; Microscopy and Microanalysis, 10(2), 174-184, 2004.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
52 Materials Characterisation V [15]
[16] [17] [18]
[19]
[20]
[21]
[22]
[23] [24]
[25]
[26]
[27]
[28]
Stadelmann, P. A.; EMS - A Software Package for Electron Diffraction Analysis and HREM Image Simulation in Materials Science; Ultramicroscopy, 21(2), 131-145, 1987. Cowley, J. M.; Moodie, A. F.; The Scattering of Electrons by Atoms and Crystal; Acta Crystallographica, 10(10), 609-619, 1957. Thust, A.; High-Resolution Transmission Electron Microscopy on an Absolute Contrast Scale; Physical Review Letters, 102(22), 1-4, 2009. Kilaas, R.; Radmilovic, V.; Structure determination and structure refinement of Al2CuMg precipitates by quantitative high-resolution electron microscopy; Ultramicroscopy, 88(1), 63-72, 2001. Tillmann, K.; Luysberg, M.; Specht, P.; Weber, E. R.; Direct compositional analysis of AlGaAs/GaAs heterostructures; Ultramicroscopy, 93, 123-137, 2002. Stroppa, D. G.; Montoro, L. A.; Beltran, A.; Conti, T. G.; Silva, R. O.; Andres, J.; Leite, E. R.; Ramirez, A. J.; Anomalous oriented attachment growth behavior on SnO2 nanocrystals; Chem. Commun.; 47, 3117-3119, 2011. Stroppa, D. G.; Montoro, L. A.; Righetto, R. D.; Ramirez, A. J.; MEGACELL: A nanocrystals models construction software for HRTEM multislice simulation; Ultramicroscopy; 2011, doi:10.1016/ j.ultramic.2011.03.013. Stroppa, D. G.; Montoro, L. A.; Beltran, A.; Conti, T. G.; Silva, R. O.; Andres, J.; Longo, E.; Leite, E. R.; Ramirez, A. J.; Unveiling the Chemical and Morphological Features of Sb:SnO2 Nanocrystals by the Combined Use of HRTEM and ab initio Surface Energy Calculations; J. Am. Chem. Soc.; 131, 14544-14548, 2009. Herring, C.; Some Theorems on the Free Energies of Crystal Surfaces; Phys. Rev.; 82(1), 87–93, 1951. Stroppa, D. G.; Montoro, L. A.; Beltran, A.; Conti, T. G.; Silva, R. O.; Andres, J.; Leite, E. R.; Ramirez, A. J.; Dopants segregation analysis on Sb:SnO2 nanocrystals; submitted for publication, 2011. Thust, A.; Coene, W. M. J.; de Beeck, M. O.; Van Dyck, D.; Maximumlikelihood method for focus-variation image reconstruction in high resolution transmission electron microscopy; Ultramicroscopy, 64(1), 211-230, 1996. Barthel J.; Weirich T. E.; Cox, G.; Hibst, H.; Thust, A.; Structure of Cs0.5[Nb2.5W2.5O14] analysed by focal-series reconstruction and crystallographic image processing; Acta Materialia, 58(10), 3764-3772, 2010. Kirkland A. I.; Meyer, R. R.; "Indirect" high-resolution transmission electron microscopy: Aberration measurement and wavefunction reconstruction; Microscopy and Microanalysis; 10(4), 401-413, 2004. Biggemann, D.; da Silva, M. H. P.; Rossi, A. M.; Ramirez, A. J.; High-Resolution Transmission Electron Microscopy Study of Nanostructured Hydroxyapatite; Microscopy and Microanalysis, 14, 433438, 2008. WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
[29] [30]
[31]
[32]
[33]
[34]
[35]
[36] [37]
[38] [39]
53
Montoro, L. A. et al; unpublished data. Kret, S.; Ruterana, P.; Rosenauer, A.; Gerthsen, D.; Extracting quantitative information from high resolution electron microscopy; Phys. Stat. Sol., 227(1), 247-295, 2001. Hytch, M. J.; Putaux, J. M.; Penisson, J. M. ; Measurement of the displacement field of dislocations to 0.03 angstrom by electron microscopy; Nature, 423, 270-273, 2003. Hue, F.; Hytch, M.; Bender, H.; Houdellier, F.; Claverie, A.; Direct mapping of strain in a strained silicon transistor by high-resolution electron microscopy; Phys. Rev. Lett., 100(15), 1-4, 2008. Johnson, C. L.; Snoeck, E.; Ezcurdia, M.; Rodrıguez-Gonzalez, B.; Pastoriza-Santos, I.; Liz Marzan, L. M.; Hytch, M. J.; Effects of elastic anisotropy on strain distributions in decahedral gold nanoparticles; Nature Materials, 7, 120-124, 2008. Montoro, L. A.; Leite, M. S.; Biggemann, D.; Peternella, F. G.; Batenburg, K. J.; Medeiros-Ribeiro, G.; Ramirez, A. J.; Revealing Quantitative 3D Chemical Arrangement on Ge-Si Nanostructures; J. Phys. Chem. C., 113(21), 9018-9022, 2009. Montoro, L. A.; Medeiros-Ribeiro, G.; Ramirez, A. J.; Novel Approach for High-Resolution Elastic Behavior Assessment of Alloyed Strained Nanostructures; J. Phys. Chem. C., 114(29), 12409-12415, 2010. Scherzer, O.; The Theoretical Resolution Limit of the Electron Microscope; J. Appl. Phys., 20(1), 20-29, 1949. Jia, C. L.; Houben, L.; Thust, A.; Barthel, J.; On the benefit of the negative-spherical-aberration imaging technique for quantitative HRTEM; Ultramicroscopy, 110(5), 500-505, 2010. Jia, C. L.; Lentzen, M.; Urban, K.; Atomic-resolution imaging of oxygen in perovskite ceramics; Science, 299, 870-873, 2003. Jia, C. L.; Mi, S. B.; Urban, K.; Vrejoiu, I.; Alexe, M.; Hesse, D.; Atomic-scale study of electric dipoles near charged and uncharged domain walls in ferroelectric films; Nature Materials, 7, 57-61, 2008.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
This page intentionally left blank
Materials Characterisation V
55
Ca(OH)2 nanoparticle characterization: microscopic investigation of their application on natural stones V. Daniele & G. Taglieri Department of Chemistry, Chemical Engineering and Materials, University of L’Aquila, Italy
Abstract Owing to conversion of lime into calcium carbonate, lime is usually adopted for conservative surface treatments. However, some critical aspects concerning the treatments reduced penetration depth, the binder concentration and the incomplete lime carbonation process still represent undesired limits and hindrances. In order to improve lime treatments, Ca(OH)2 particles with nanometric dimensions (nanolime) have recently been introduced in Cultural Heritage conservation (frescoes, stuccoes, ..). The aim of the present work is to characterize Ca(OH)2 nanoparticles synthesized by a chemical precipitation process starting from two base supersaturated aqueous solutions of calcium chloride and sodium hydroxide. After several washes, necessary to remove the sodium chloride, the aqueous medium is partially substituted by 2-propanol to improve the suspension stability; an alcoholic nanolime suspension is obtained, characterised by a given concentration and a residual water content. In order to identify the structure of the formed phases and the particles reactivity, the obtained nanolime is characterised by X-ray diffraction (XRD) and profile analysis; scanning and transmission electron microscopy (SEM-TEM) are performed too. The results show hexagonally plated and regularly shaped particles with side dimensions equal to or less than 300nm; moreover particles have pure crystalline features and a high reactivity in terms of the carbonation process. Finally, the Ca(OH)2 nanoparticles are applied on some natural lithotypes; SEM analyses are performed to evaluate penetration depth and grain adhesion of the nanolime treatments itself. From SEM micrographs a partial filling of the lithotypes pores, located at a distance from the surface of more than 200µm, is observed. Keywords: calcium hydroxide, consolidation, lime, nanoparticles, protection. WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line) doi:10.2495/MC110051
56 Materials Characterisation V
1 Introduction Thanks to conversion of lime into calcium carbonate, lime water and milk are usually adopted for conservative surfaces treatments. In particular, lime water consolidation is generally obtained spraying the lime solution on the cleaned surface. To reach a good penetration, the treatment is repeated several times until the surface is able to absorb lime water [1]; some Authors indicate that could be necessary to repeat the application for 30-40 times [2]. Lime milk is used on the same basis as lime water [3]; nevertheless it involves greater amounts of lime with the same water volume, reducing the water percentage brought to the stone. However, some critical aspects concerning the treatments reduced penetration depth, the binder concentration and the incomplete lime carbonation process still represent undesired limits and hindrances. In order to improve lime treatments, Ca(OH)2 particles with submicrometric dimensions (nanolime) are recently introduced in Cultural Heritage conservation. Lime nanoparticles present the following advantages in stone, mortar and plaster consolidation: the possibility to penetrate deep into damaged zones (no limitations due to the particle size), high reactivity and fast reactions (such as carbonation) in the treated zones, high purity and defined composition [4]. Nanolimes are successfully employed on mural paintings, stuccoes and frescoes [5–9]; refurbishments of architectonical surfaces are considered too [10–12]. Lime nanoparticles are typically produced by a chemical precipitation process in supersaturated aqueous solutions of the reactants (calcium chloride and sodium hydroxide). To improve nanolime particles dispersion, the use of alcoholic solutions in place of aqueous ones is adopted; in fact, when 2-propanol alcohol is used as a solvent, dispersions of calcium hydroxide particles show a slower rate of agglomeration (and therefore, slower sedimentation rates) in comparison to aqueous media. This reduces the tendency for a white film to form on surfaces to be consolidated [13]. In this paper a nanolime suspension, characterised by a residual water content in the precipitated phase, is synthesised. X-ray diffraction (XRD) measurements and profile analysis are performed to characterise the nanolime samples and to evaluate the carbonation efficiency (yield). To correlate the produced nanolime to its properties, morphological characterisation is performed by scanning and transmission electron microscopy (SEM and TEM). Finally, the nanolime suspension is applied by brush on cleaned stone surfaces of natural lithotypes; in particular, the suspension is used in a diluted form (concentration of about 1mg/ml) in order to avoid the risk of leaving a white film on treated materials surfaces [11]. Several lithotypes, largely employed in architectonic fields, are considered: “Travertine”, “Pietra di Lecce”, “Pietra Serena” and “Basalto”. A local calcareous lithotype, “Poggio Picenze” is investigated too. SEM analyses allow one to investigate how the nanolime penetrates inside the stone porosity, in terms of the reached depth and the grains adhesion of nanolime treatments itself.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
57
2 Experimental section Calcium chloride dihydrate (CaCl2·2H2O), sodium hydroxide (NaOH) and 2propanol pro analysi products, supplied by Merck, are used without further purification. Water is purified by a Millipore Organex system (R ≥ 18 M cm). Two aqueous solutions of 400ml, containing 0,3mol/l of CaCl2 and 0,6mol/l of NaOH respectively, are prepared. The NaOH solution (used as precipitator) is added dropwise into the CaCl2 solution (speed ≈ 4ml/min, temperature of 90°C). After about 24 hours two distinct phases are observed: a limpid supernatant solution and a white precipitated phase (NLW sample). In order to remove the NaCl produced, several deionised water washings are performed. Subsequently, in order to improve the suspension stability, the water content is partially substituted by 2-propanol; the obtained suspension is characterised by a water/2propanol ratio (W/A) of 0,75 and a final concentration of about 10mg/ml (NLA sample). SEM analyses (Philips XL30CP) are performed, depositing 0,2ml of the suspension on the specimen; TEM investigations (Philips CM100) are carried out dispersing 0,2 ml of the suspension in 50 ml of 2-propanol and depositing the sample on the suitable grid. As concerns XRD measurements, the sample is prepared maintaining the nanolime preparation for 20’ in ultrasonic bath (US) and then depositing 0,2ml of the suspension on a silica sample holder; measures are performed on dry sample, in laboratory conditions (T=20°C, relative humidity RH=40%). Each experimental diffraction spectrum is elaborated by a Profile Fit Software (Philips PROFIT v.1.0) and each crystalline phase is attributed by JCPDS patterns; the ratio between the CaCO3 peaks area and the spectrum total area is assumed as the carbonation process efficiency (yield). 2.1 Results and discussions TEM micrographs, obtained on NLA sample, are reported in Fig.1; in particular, hexagonally and regularly shaped particles of side dimension ranging from 100 to about 300nm are observed. Moreover, in Fig.1b) it is possible to note that all the particles appear overlapped in an ordered way and are so thin to be transparent to the electron beam.
a)
b)
Figure 1:
TEM micrographs on NLA sample: a) scale bar 500nm; b) scale bar 200nm.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
58 Materials Characterisation V In Fig. 2, SEM images on NLA sample are shown. The particles, characterised by a prismatic features are recognizable to calcium carbonate of side dimension less than 500nm (Fig.2a). In Fig.2b) a perfectly hexagonal and regular shaped particle (marked as A) is observed, characterized by a side dimension less than 250nm.
A
a)
b) SEM micrographs on NLA sample (scale bar 1µm).
Figure 2:
XRD pattern on NLA sample shows the presence of Ca(OH)2 and CaCO3 phases (84-1276 and 85-1108 JCPDS patterns, respectively) with a corresponding yield value of about 80% (Fig.3a). On the contrary, XRD results obtained on NLW sample show as the corresponding yield decreases, reaching a value of about 50% (Fig.3b). The NLA higher yield values can be attributed to the presence of 2-propanol, that tends to “disagglomerate” the nanolime particles; so, a greater specific surface of the Ca(OH)2 particles exposed to air, can leads to a better carbonation process. *
* CaCO3 ° Ca(OH)2
* CaCO3 ° Ca(OH)2
° *
°
°
° *
a)
°
*
*
*
°
°
°
*
*
*
*
b) Figure 3:
XRD patterns: a) NLA sample; b) NLW sample.
3 Treatment on natural lithotypes: characterisation and results The considered samples, largely employed in the Italian historical architecture, are natural calcareous stones - Travertine, Pietra di Lecce and Poggio Picenze (the last one is a local lithotypes), a national sandstone - Pietra Serena - and an igneous lava stone of volcanic origin - Basalto. WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
59
The nanolime treatment is carried out by brush, in laboratory conditions (T=20°C; R.H. 50%), applying the diluted alcoholic suspension (characterised by a concentration of 1mg/ml) on the dry and clean stones surface. In particular, 1ml of the suspension is applied on a side (5*5cm2) of the stone surface; when the solvent is evaporated, this procedure is repeated for 100 times (giving about 100mg Ca(OH)2 for each stone). For SEM investigations, each stone sample is broken along a plane perpendicular to the treated side and the section surface is observed. On the following figures, SEM micrographs, referred to untreated and treated samples section, are shown; in particular, in order to evaluate the treatment penetration inside the stone, pores situated at different surface distances are considered. 3.1 Travertine In Fig. 4 SEM micrographs on untreated Travertine stone sample is reported; in particular, a typical region, containing macropores and areas with a compact matrix is shown. For what concerns the treated stone, it is possible to note as the nanolime treatment covers the internal stone pore without filling it completely (Fig.5). In particular, the treatment, well recognisable in SEM micrograph referred to a pore at 150μm from the surface (Fig.5a-b), is found also in pores located at about 1mm from the surface itself (Fig.5c-d). This result can be attributed to the Travertine porosimetric structure, constituted by macropores, by which the nanolime treatment can be carried inside the stone.
A
a)
b)
Figure 4:
SEM micrographs on untreated Travertine stone: a) sample section (scale bar 500μm); b) zoom view on A pore (scale bar 100μm).
3.2 Lecce stone Pietra di Lecce is an organogenic marl calcareous stone, compact, generally porous and characterised by a fine-grained structure [14]. In Fig. 6, SEM micrographs on untreated sample are reported; it is possible to note the granular stone structure, constituted by various size clasts and by an intergranular porosity (Fig.6b).
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
60 Materials Characterisation V B
B
C a)
b) C
C
c)
d)
Figure 5:
SEM micrographs on treated Travertine stone: a) sample section (scale bar 200μm); b) zoom view on B pore localised at 150μm from the surface (scale bar 10μm); c)-d) different zoom views on C pore at about 1mm from the surface (scale bar 10μm and 5μm, respectively).
A
a)
b)
Figure 6:
SEM micrographs on untreated Lecce stone: a) sample section (scale bar 500μm); b) zoom view on A pore (scale bar 10μm).
Considering the treated sample, the treatment covers homogeneously the internal stone pores without filling them completely (Fig.7b-c). In particular, the treatment is also well recognisable in the pore located at 800μm from the surface, underlining the penetration depth reached from the treatment itself (Fig.7c).
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
61
B
B
C B a)
b) C
c) Figure 7:
SEM micrographs on treated Lecce stone: a) sample section (scale bar 200μm); b) zoom view on B pore localised at 250μm from the surface (scale bar 10μm); c) zoom view on C pore at about 800µm from the surface (scale bar 10μm).
3.3 Poggio Picenze stone Poggio Picenze lithotype, a local calcareous stone, is constituted by heterogeneous grained material; the stone surface and its internal structure are characterized by small-medium size pores and vacuoles (Fig.8) [14].
A
a)
b)
Figure 8:
SEM micrographs on untreated Poggio Picenze stone: a) sample section (scale bar 500μm); b) zoom view on A pore (scale bar 20μm).
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
62 Materials Characterisation V SEM micrographs referred to treated stone, show as the nanolime treatment tends to fill the internal cavity of the material, adhering to its grain, without occlude the pores. In particular, the treatment, recognisable in a pore located at a distance of 300μm from the surface (Fig.9), is not found in more internal cavities. B
B a)
b) B
c) Figure 9:
SEM micrographs on treated Poggio Picenze stone: a) sample section (scale bar 200μm); b)-c) different zoom views on B pore at about 300µm from the surface (scale bar 10 and 5μm, respectively).
In particular, in Fig.9b-c) internal stone cavities, that can be related to intergranular space, don’t appear covered by the precipitated particles; this fact can indicate that the nanolime treatment doesn’t modify the porosimetric system of the lithotype. 3.4 Pietra Serena stone This lithotype is different from those discussed above; in fact, it is a sandstone characterised by clay fractions. Moreover, this stone is constituted by different grains size, ranging from medium-fine to coarse ones (Fig.10) [14]. SEM micrographs referred to the treated sample are shown in Fig.11; the internal voids appear filled, on the walls, by precipitated particles (shaped like a small “flakes”) that tend to adhere to the clasts edges, without occluding the pores system (Fig.11b-c).
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
63
A
a)
b)
Figure 10:
SEM micrographs on untreated Pietra Serena stone: a) sample section (scale bar 500μm); b) zoom view on A pore (scale bar 20μm). B C
B a)
b) C
c) Figure 11:
SEM micrographs on treated Pietra Serena stone: a) sample section (scale bar 200μm); b) zoom view on B pore localised at 750μm from the surface (scale bar 10μm); c) zoom view on C pore at about 450µm from the surface (scale bar 10μm).
3.5 Basalto stone Basalt lithotype is an igneous rock lava of volcanic origin, typically composed by gray-black silicates [14]. It is an hard and resistant stone characterised by many pores and voids, of varying sizes, usually medium-large ones (Fig.12). WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
64 Materials Characterisation V
A
a)
b)
Figure 12:
SEM micrographs on untreated Basalto stone: a) sample section (scale bar 500μm); b) zoom view on A pore (scale bar 20μm). B
B
C a)
b) B
C
c)
d)
Figure 13:
SEM micrographs on treated Basalto stone: a) sample section (scale bar 200μm); b)-c) different zoom views on B pore localised at 500μm from the surface (scale bar 10 and 5μm, respectively); d) zoom view on C pore at about 700µm from the surface (scale bar 5μm).
SEM micrographs on the treated stone, show as the nanolime treatment tends to fill the internal cavity of the material, adhering to its grain, without occluding the porosimetric system. In particular, the treatment, is not found in pores more internal than 700μm.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
65
4 Conclusions Nowadays lime nanoparticles are synthesised and employed for protection and superficial consolidation of several artworks such as stones, plasters, frescoes, wall paintings and paper documents. The obtained Ca(OH)2 nanoparticles are hexagonally plated and regularly shaped, with side dimensions generally less than 300nm, as shown by TEM observations. From XRD analyses, the carbonation process efficiency of the aqueous nanolime is about 50%, while the use of 2-propanol improves it reaching a yield value of about 80%. This result underlines the 2-propanol role in disagglomerating the particles, leaving a higher specific surface exposed to air. SEM analyses are performed to evaluate the interaction between the nanolime treatments and the porosimetric system of some natural lithotypes. The images show the partial filling of the pores in all the examined lithotypes, without occluding them. On the contrary, considering all the lithotypes, the treatment penetration is observed at a limit distance from the surface, ranging from 300µm (Poggio Picenze stone) to about 1mm (Travertine stone).
References [1] Vinardi, M.G., Cully, M.H., Brunetto, A., La reversibilità nel restauro. Riflessioni, esperienze, percorsi di ricerca, In: XI National Congress Scienza e Beni Culturali: “La Pulitura delle Superfici dell’Architettura”, Bressanone, XIX, pp. 399-405, 2003. [2] Sten, P., Lime Water Consolidation, ICCROM, 53-61, 1981. [3] Turriziani, R., I leganti ed il calcestruzzo, Edizioni Sistema, Roma, pp. 70-82, 1972. [4] Pianski, J., Brümmer, K., Ziegenbalg, G., Nano-particles for stone conservation–state of the art, characteristics and recent developments, In: Stonecore – “Recent progress in the consolidation of calcareous materials”, Litomyšl, Czech Republic, Aprile 2010. [5] Baglioni, P., Cesari, C., Dei, L., Giorgi, R., Grassi, R., Lorenzetti, M., Mauro, M., Pinzani, D., Ruffo, P., Schonhaut, G., Stucco forte veneziano con “Tecniche della Tradizione”: proprietà chimico-fisiche, degrado in ambiente lagunare e conservazione mediante nanofasi cristalline di Ca(OH)2, In: National Congress Scienza e Beni Culturali: “Lo Stucco. Cultura, Tecnologia, Conoscenza”, Bressanone, XVII, pp. 289-298, 2001. [6] Ambrosi, M., Baglioni, P., David, P.R., Dei, L., Giorgi, R., Lalli, C., Lanterna, G., Mairani, A., Matteini, M., Rizzi, M., Schonhaut, G., Inorganic consolidants and protectives for architectonic surfaces: experimental tests on Santa Prisca Church Apse in Rome, In: 2nd International Congress on “Science and Technology for the safeguard of Cultural Heritage in the Mediterranean Basin”, Vol. 2, pp. 873-877, 1992. [7] Dei, L., Salvadori, B., Arlango, E., Pietropoli, F., Scardellato, C., Gli affreschi del XIII e XIV secolo nella Cripta di San Zeno a Verona: la sperimentazione della nanocalce dispersa in alcol iso-propilico durante WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
66 Materials Characterisation V
[8]
[9]
[10]
[11]
[12]
[13]
[14]
l’intervento conservativo, In: International Congress Scienza e Beni Culturali: “Sulle pitture murali. Riflessioni, conoscenze, interventi”, Bressanone, XXI, pp. 293-302, 2005. Dei, L., Radicati, B., Salvadori, B., Sperimentazione di un consolidante a base di idrossido di calcio nanofasico sugli affreschi della Cappella del Podestà al Museo del Bargello di Firenze: aspetti chimico-fisici e prove di colore, In: International Congress Scienza e Beni Culturali: “Sulle pitture murali. Riflessioni, conoscenze, interventi”, Bressanone, XXI, pp. 99-108, 2005. Dei, L., Favaro, M., Giorgi, R., Portieri, R., Gli stucchi della Cappella della Madonna di Lourdes nella Chiesa di S. Giovanni Evangelista in Venezia: caratterizzazione del degrado e pre-consolidamento mediante dispersioni alcoliche di calce, In: National Congress Scienza e Beni Culturali: “Lo Stucco. Cultura, Tecnologia, Conoscenza”, Bressanone, XVII, pp. 557-566, 2001. Slížková, Z., Drdácký, M., Moreau, C., Frankeová, D., Nosál, L., Consolidation of porous limestone with suspensions of calcium hydroxide nano-particles in alcohols, In: Stonecore – “Recent progress in the consolidation of calcareous materials”, Litomyšl, Czech Republic, 2010. Daniele, V., Taglieri, G., Nanolime suspensions applied on natural lithotypes: the influence of concentration and residual water content on carbonatation process and on treatment effectiveness, Journal of cultural heritage, 11, pp. 102-106, 2010. Ambrosi, M., Giorgi, R., Salvadori, B., Dei, L., Nanotecnologie per la conservazione di affreschi e materiali lapidei a base carbonatica, In: 1st International Workshop on Science, Technology and Cultural Heritage, Venezia, pp. 29-33, 2004. Hansen, E., Doehne, E., Fidler, J., Larson, J., Martin, B., Matteini, M., et al., A review of selected inorganic consolidants and protective treatments for porous calcareous materials, Rev Conserv., 4, pp. 13–25, 2003. http://www.italithos.uniroma3.it/index.php.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
67
Nanocarbon composite materials with optical response on radioactive waste M. Vantsyan1, G. Popova1, E. Karpuzova1, M. Bobrov1, O. Plaksin2 & E. Dabek3 1
D. Mendeleyev University of Chemical Technology of Russia, Russia A.I. Leypunsky Institute for Physics and Power Engineering, Russia 3 Science & Technology Branch Environment Canada, Ottawa, Canada 2
Abstract Nanocarbon materials have numerous unique features – high porosity, large specific surface area, chemical inertness, radiation stability, etc. We applied nanocarbon/nanodiamond and silicon carbidecomposites as matrix for optical chemochips construction. Composite elements consist of porous nanocarbon substrate with specific chromophores introduced to nanosize pores and siliconorganic coating. Similar multicomponent composition has a response to radiation, in particular, -irradiation. By using diarylethenes as sensitive chromophores, their electronic and/or luminescent spectra data may be applied for doze power detection. Quantum chemistry methods, computer simulation were considered for optimal design of nanocarbon/organic chromophore hybrid. Experimental data and modeling have shown that diarylethenes are able to change color under -irradiation in molecular crystal phase only. Weak interactions (inter-, intramolecular and binding with hydrogen-containing walls in pores) play key role under irradiation. Induced self-organization in limited volume is considered. SWAXS and AFM data are discussed. Nanodiamond composition possesses luminescent response to -, - and -irradiation. Siliconorganic polymer and SiC composition are neutral to irradiation, they are stable in extreme conditions. Application of composite elements with optical sensing as multifunctional chemochip fragment for atmospheric media monitoring is discussed. Keywords: nanocarbon, nanodiamond, silicon, composite, chemochip, chromophore, radiation, computer simulation, optical response, nuclear waste.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line) doi:10.2495/MC110061
68 Materials Characterisation V
1 Introduction One of the most important directions of modern nanoscience and nanotechnology is the so-called sensor technology combining the latest achievements in highly sensitive intelligent materials, first of all, multifunctional chemochips with immobilized sensing fragments with nanoscale function [1–4] and different microdevice construction with high processing speed and reliability. Of especial interest are the materials with high sensitivity and adaptability [5]. In this context, great attention is being given to the selfassembling systems capable of hierarchical ordering from the nano- to mesoscopic level [6], which may lead to elaboration of intelligent, in particular, optically sensing, materials responsive to weak external factors such as temperature, irradiation, pH, pollutants in the environment, etc. Typically, these devices (chemochips) consist of solid inorganic support and sensing fragments immobilized on it. Nanocarbon materials are promising inorganic supports as they have a large variety of unique and specific properties – high porosity, large specific surface area, chemical inertness, radiation stability and others [7]. To date, highly porous (up to 40% vol.) composites can be prepared on the basis of nano-sized diamond. Diarylethenes seem to be the most suitable objects to be used as sensing units. These are photochromes changing from colourless to coloured under irradiation; normally, the colour disappears upon UV or visible-light irradiation [8]. Diarylethenes have numerous advantages such as thermal stability, linear dose dependence of the coloration, compared to other photochromic compounds. On the other hand, diarylethenes exhibit their sensing properties in highly ordered state (e.g. in single crystal). In this context, of great importance is prediction of diarylethene molecules self-organization pattern. Computer modeling and quantum chemistry calculations help greatly to successfully solve applied science problems of predicting optimal synthetic pathways, structure and properties of materials resulting in decrease in time and costs of initial laboratory investigations. For integrated solution of the problem it was necessary to predict diarylethenes self-organisation in a limited volume i.e. to estimate the possibility for chromophores introduction into porous composite and to evaluate the assembly parameters. In this study, we applied experimental techniques together with a set of up-todate computing chemistry methods, Cambridge Structural Database and specialized software. We obtained models of diarylethene molecules and their assemblies; non-covalent interactions providing formation of 2D and 3D supramolecular clusters and crystal structures were revealed.
2 Results and discussion In this study, for multifunctional detector device creation we used efficient adsorption properties of nanoporous carbon composites nanocarbon/nanodiamond and silicon carbide [9, 10] subsequently covering the composite matrix with polymeric photochromic composition. WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
69
As nanodiamond composite matrix is black it is reasonable to introduce only luminescent molecules whereas silicon carbide composites are suitable for colour-changing chromophores. Therefore, we applied diarylethenes that change both luminescence and colour under different types of irradiation. Moreover, diarylethenes (DAE) have a unique property to change their colour under radioactive irradiation and to reversibly restore it under UV-irradiation. We have chosen a diarylethene – 1,2-bis(2,5-dimethyl-3-thienyl)perfluorocyclopentene changing from colourless to coloured under -irradiation; the colour disappears upon UV or visible-light irradiation [11] (Fig.1). F
F
F F F
F F
S
H 3C
S
CH3 H3C
Open form colourless
Figure 1:
F F
F F
UV or visible-light irradiation
CH3 H3C
-irradiation
F
CH3 S H3C
S
CH3
Closed form coloured
Coloration under -irradiation and bleaching under UV or visiblelight irradiation.
Nanodiamond/nanocarbon composites samples were black tablets with diameter of 8 mm and thickness of 3 mm. They contained 28% vol. nanodiamond, 15% vol. graphite-like carbon with porosity of the material 57% vol. and pore size 8 – 10 nm (as found by capillary condensation method). Silicon carbide nanocomposite samples were grayish tablets with a diameter of 8 mm and thickness of 3 mm. Composites surface was studied by atomic force microscopy method (AFM); the average pore size was found to be up to 10 nm [10]. However, pore shape and depth was not estimated with high accuracy. Theoretically, different pores in composites can have conic and cylindrical shape. For calculations we used an approximation that all pores in composites are cylindrical with pore diameter not more than 10 nm. Average grain size of the composites was estimated by wide-angle x-ray scattering method (WAXS). For nanodiamond/nanocarbon composite it was 4 nm, for silicon carbide – 12.5 nm. Chromophores were introduced by impregnation method: composites were treated by 2% mass solution of diarylethene in chloroform with subsequent drying under vacuum. Luminescence spectra of modified nanoporous matrices exhibit a new emission band corresponding to starting diarylethene. The intensity of this new
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
70 Materials Characterisation V band is approximately 15–20% from intensity of the luminescence peak of starting diarylethene. A preliminary study has shown rather high radiation stability of nanocarbon composites , silicon-organic coating and chromophores. Diarylethene molecules are conjugated aromatic systems capable not only of hydrogen bonding but also of stacking interactions [12]. Pore size and shape also influence self-organization of chromophores when filling the limited volume. A molecular geometry modeling for diarylethene molecule and for its molecular aggregates was performed by semi-empirical quantum chemistry methods АМ1 by programme complex GAMESS [13]. The initial approximation of the molecule (derived from x-ray diffraction data) was taken from Cambridge Structural Database [14]. After optimizing the geometry parameters of diarylethene molecule, we have estimated maximal distances between atoms centres along the axes of three main inertias. These distances were ~9,9 Å (along axis 1), ~11,5 Å (along axis 2), ~6,6 Å (along axis 3) The optimization of geometry parameters of two diarylethene molecules assembly was performed by non-empirical quantum chemistry method GAMESS DFT 6-31G(d,p) [13]. These methods are also suitable for calculation of bond critical points, in particular, bond critical points of non-covalent interactions [15]. The size of the optimized model of the molecular aggregate including two chromophore molecules (Fig. 2) have also been estimated by maximal distances between atom centres along three axes. Two diarylethene molecules are arranged in two parallel planes and are bonded by weak bonds. One can assume two probable aggregate structures – one is formed by four СF…H bonds and the other – by two СS…H bonds. Both aggregates dimensions are close to one another – these are 19* 11* 9 Å. The proposed four chromophore molecules aggregate structure (Fig. 3) has maximal distance between atom centres along one of the axes, equal to 29 Å. The distance along two other axes is 18 Å and 9 Å. As a result of the calculation performed, it was found that two diarylethene molecules are also arranged in two parallel planes. In order to reveal weak noncovalent interactions in two diarylethene molecules assembly we have applied R. Bader quantum topological analysis of atoms in molecules [15]. By programme complex AIM-2000 bond critical points were calculated and a molecular graph was drawn for two diarylethene molecules assembly [16]. We have calculated bond critical points between carbon atoms, distances and angles of weak noncovalent interactions, local characteristics in critical points (Table 1). This calculation has elucidated weak bonds between fluorine and hydrogen atoms, these bonds are pairwise symmetrical. Moreover, weak bonds were found inside the molecule, by bonds fluorine – hydrogen an intramolecular cycle is formed.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
71
(a)
(b) Figure 2:
Assemblies of two diarylethene molecules; (a) – formed by four СF…H bonds, (b) – by two СS…H bonds. Non-covalent bonds are shown by dotted line.
Table 1:
Distances, angles, and local parameters in critical points (3,-1) corresponding to weak interactions.
CP number 1 2 3 4 5 6 7 8 9
R А 2.54 2.53 2.53 2.54 2.60 2.60 2.78 2.78 4.20
Angle, degrees 139.5 146.8 146.8 139.4 108.5 108.5 138.3 138.3 -
Contact CH38…F45 CH38…F47 CH77… F8 CH77… F6 CH61…F47 CH22… F8 CH22…H70 CH38…C23 C21…C60
CP number 55 48 52 35 36 24 43 47 33
b
2b
gb
vb
Econt
a.u.
a.u.
a.u.
a.u..
kcal/mole
0.00636 0.00645 0.00645 0.00635 0.00677 0.00677 0.00195 0.00195 0.00117
0.02822 0.02825 0.02825 0.02822 0.03319 0.03320 0.00619 0.00619 0.004150
0.005769 0.005794 0.005794 0.005768 0.006470 0.006471 0.001114 0.001114 0.0007669
0.004483 0.004525 0.004525 0.004482 0.004642 0.004642 0.0006811 0.0006811 0.0004963
1.41 1.42 1.42 1.41 1.46 1.46 0.21 0.21 0.16
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
72 Materials Characterisation V
(a)
(b) Figure 3:
Assemblies of four diarylethene molecules: (a) – formed by two СS…H bonds and eight СF…H, (b) – by four СS…H bonds and four СF…H bonds.
Energies of the above mentioned interactions were estimated by local density of the potential energy (νb) in a critical point. For this, an empirical formula was used: E (kcal/mole) = 313.754•vb (atomic units) [17]. So, the overall energy of interatomic non-covalent interactions was Econt = -9.15 kcal/mole.
3 Conclusions Experimental techniques (e. g. AFM, x-ray scattering and others) applied together with a set of up-to-date computing chemistry methods aided greatly for subsequent creation of a chemochip sensitive to -irradiation based on WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
73
nanocarbon composites. Computer modeling has revealed the self-organization pattern for diarylethene molecules in limited volume. Diarylethene molecule dimensions (as well as those for its aggregate of two and four molecules calculated by semi-empirical quantum chemistry methods) are estimated. By density functional theory geometry parameters in assembly of two diarylethene molecules were optimized. The quantum-topological electron density analysis has revealed weak non-covalent interactions fluorine-hydrogen, carbon-hydrogen, carbon-carbon. In separate diarylethene molecules a cycle formed by non-covalent interactions fluorine-hydrogen is found. The overall energy of interatomic non-covalent interactions was Eконт = -9.15 kcal/mole. Quantum chemistry and quantum topological analysis methods are promising tools for revealing interatomic interactions when studying formation of molecular assemblies at the nanoscale in pore filling modeling.
Acknowledgements The research is being supported by Ministry of Education and Science or Russian Federation (Project 2.2.2.2.325) and ISTC (Project #3891).
References [1] Al-Azzawi A. (ed). Photonics: Principles and practices, CRC Press: Boca Raton, 2007. [2] Davies A.G. and Thompson J.M.T. (eds). Advances in nanoengineering: Electronics, materials and assembly, Imperial College press: London, 2007. [3] Fryxell G.E. and Cao G. Z. (eds). Environment application of nanomaterials – Synthesis, sorbents and sensors, Imperial College Press: London, 2007. [4] Lu G.Q. and Zhao X.Z. (eds), Nanoporous materials: Science and Engineering, Imperial College Press: London, 2004. [5] Ong K. G., Yang X., Mukherjee N., Wang H., Surender S., Grimes C.A. A wireless sensor network for long-term monitoring of aquatic environments: Design and implementation, Sensor Letters, 2 (1), pp. 48-57, 2004. [6] Grimes C.A, Dickey E.C., Pishko M. V., Encyclopedia of Sensors, ASP Press: N.-Y., 2004. [7] Seki Y., Impact of low activation materials on fusion reactor design, J. Nucl. Mater, 258-263, pp. 1791-1797, 1998. [8] Irie S., Irie M., Radiation-induced coloration of photochromic dithienylethene derivatives in polymer matrices, Bull. Chem Soc. Jpn., 73, pp. 2385-2388, 2000. [9] Lisichkin G.V. (ed.) Chemistry of Grafted Surface Compounds, Fizmatlit: Moscow, 2003. [10] Gordeev S.K., Nanocarbon Materials, Nanotechnics (Rus.), 2005, pp. 3-11. [11] Morimoto M., Kobatake S. Irie. M., Photochromism of diarylethenes in nanolayers of a single crystal, Photochem. Photobiol. Sci., pp. 1088-1094, 2003. WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
74 Materials Characterisation V [12] Yamada T., Kobatake S., Muto K., Irie M., X-ray Crystallographic Study on Single-Crystalline Photochromism of Bis(2,5-dimethyl-3-thienyl) perfluorocyclopentene, J. Am. Chem. Soc., 122, pp. 1589–1592, 2000. [13] http://classic.chem.msu.su/gran/gamess/index.html. [14] CSDB, http:// www.ccdc.cam.ac.uk [15] Bader R., Atoms in Molecules: A Quantum Theory, Oxford University Press, 1994. [16] http://www.aim2000.de [17] Espinosa E., Alkorta I., Rozas I., Elguero J., Molins, E., Topological Analysis of the Electron Density Distribution in Perturbed Systems. I. Effect of Charge on the Bond Properties of Hydrogen Fluoride, Chem. Phys. Letts., 336, pp. 457-464, 2001,.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Section 3 Corrosion problems
This page intentionally left blank
Materials Characterisation V
77
Evaluation of the fretting corrosion mechanisms on the head-cone interface of hip prostheses I. Caminha1, C. R. M. Roesler2, H. Keide1, C. Barbosa1, I. Abud1 & J. L. Nascimento1 1
Laboratory of Characterization of Mechanical and Microstructural Properties, National Institute of Technology (INT), Rio de Janeiro (RJ), Brazil 2 Laboratory of Biomechanical Engineering, University Hospital, Federal University of Santa Catarina (UFSC), Brazil
Abstract Fretting-corrosion is one of the main concerns in the application of hip prosthesis. This type of information is very important in the stage of orthopedic implants design, with the purpose of minimizing the amount of tissue exposed to corrosion products which are released during the permanence of the prosthesis in the patient. The residual corrosion products of stainless steels are associated to the occurrence of several adverse reactions in the human body. The knowledge about these corrosion products is extremely important in the phase of the project of hip prostheses, aiming at the minimization of the amount of exposure of the organic tissues to corrosion products released during the permanence of the prosthesis inside the patient. In the present work the mechanical stability and fretting corrosion resistance of modular hip prosthesis, which was fabricated with ASTM F 138 austenitic stainless steel, were evaluated according to the criteria of ASTM F 1875 standard, method I, which prescribes long term test, with the purpose of determining the amount of damage through the quantification of the corrosion products and debris which resulted from the fretting corrosion conditions. The mechanical tests were performed in a servohydraulic mechanical testing machine and the modular interfaces were exposed to an electrolytic 0.9% NaCl in distilled water solution and subjected to a minimum load of 230 N and a maximum load WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line) doi:10.2495/MC110071
78 Materials Characterisation V of 2.3 kN, frequency of 5 Hz for 10 million cycles, according to ASTM F 1875 standard, thus simulating real conditions of use. Load ratio (R = 10) was determined in ASTM F1440 standard. A significant amount of fretting-corrosion products were observed: 0.22 g resulted from five specimens after 107 million cycles, and the diffraction X-Rays tests showed the presence of the crystalline phases Fe2O3, Fe3O4 and Cr2O6. It can be concluded that the mass loss in the head-cone connection allowed the entrance of the physiological saline solution in the inner region of the head, increasing the predicted micro movement in the head–cone interface, resulting in an accelerated process of fretting-corrosion, and consequently the liberation of debris and corrosion products that could lead to adverse biological reactions. Keywords: fretting corrosion, hip prosthesis, modular components, mechanical properties.
1 Introduction The increasing life expectancy in the last decades and trauma risk due to the more frequent exposure of individuals to different types of accidents and current insecure life conditions has led to an increased search for orthopedic implants with high performance, capable of resisting to even more severe loads for longer times. For this reason, metallic materials, which were developed for applications in orthopedic implants, must present some specific properties, such as biocompatibility, strength and resistance to degradation (by wear or corrosion). In spite of the increasing use of titanium and its alloys for this application in the last decades, in Brazil ASTM F138 austenitic stainless steel [1, 2] is still very used in the fabrication of modular components for total hip arthroplasty due to its lower cost, together with good mechanical properties and satisfactory corrosion resistance, since the main customer is Health Ministry, which supplies the demands of public hospitals in Brazil. The use of modular components in the total hip arthroplasty presents several advantages, such as a great variety of stem geometry, which allows a better choice of modular component for the surgeon, according to the patient disease, mainly in the case of revision surgeries, thus minimizing the risk of inadequate procedures. However, the modular interfaces are subjected to micro movements, which can result in fretting and corrosion, leading to the release of debris, which can cause adverse reactions and accelerated wear in the articulation interface. “Fretting” can be defined as a wear phenomenon which occurs when two solids in contact are subjected to tangential oscillatory movement with small displacement amplitude. It can be described more accurately as a movement whose amplitude is smaller than the contact extent. The presence of a corrosive environment contributes to accelerate the wear process and this condition is defined as “fretting-corrosion”. The occurrence of corrosion results in degradation processes which reduce the structural integrity of the implant and release products whose reaction can be harmful in contact with organic tissue.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
79
When the prostheses are fabricated with ASTM F138 stainless steel, the generation of a passive barrier in the surface (passive layer) is the main obstacle to corrosion process. This film, which is composed by metallic oxides, formed spontaneously on the metal surface, and whose thickness can be increased with surface treatments, avoids the migration of metallic ions from the metal to the solution and the migration of anions from the solution to the metal through the metal-solution interface (physiological environment). For being effective barriers these films must be compact and cover completely the metallic surface, they must have an atomic structure which limits the migration of metallic ions and/or electrons through the metallic oxide/solution interface, and they must be capable of remain on the surface of these alloys even when subjected to mechanical and abrasive stresses in the clinical use of prostheses. The most frequently mentioned mechanisms of failure in these components are aseptic loosening, periprosthetic osteolysis and metallosis and are directly related to the debris released by the prosthesis components, mainly the non cemented prostheses [3]. According to the literature [4], the residual products of corrosion in stainless steel based implants are associated to tissue necrosis, inflammation, allergenic reactions and even cancer. The aim of the present work is an evaluation of fretting corrosion in modular prostheses fabricated with ASTM F 138 stainless steel, in order to minimize the adverse reactions during the permanence of the prosthesis in the patient.
2 Material and methodology In this work five modular hip prostheses, composed by heads and stems, five stems and five heads, fabricated with ASTM F 138 from the same batch, were analyzed. A SZX16 model Olympus stereomicroscope was used for comparing the surfaces of the modular components before and after fretting corrosion test. In Fig.1 an as received representative sample of the batch can be observed.
Head Figure 1:
Cone
Stem
Head and stem of the hip prosthesis fabricated with ASTM F 138 stainless steel.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
80 Materials Characterisation V The fretting corrosion tests on the cone-head interface were performed according to method I of ASTM F 1875 standard, which prescribes a 0.9% NaCl physiological saline solution, thus simulating in vivo conditions, since that when implanted in patients these components are in contact with body fluids. With the purpose of analyzing both fatigue resistance of hip stems and mechanical stability of modular components the experimental procedures described in ASTM F 1440 [5] and ISO 7206-4 [6] were also adopted. These tests were performed in two universal machines for mechanical testing (Instron model 8872) with a 25 kN load cell with the following characteristics, described in ASTM F 1875 standard [7]: -
Maximum load: 3.3 kN;
-
Frequency: 5 Hz;
-
Test finish after 10 million cycles;
-
Error of applied load lower than 1% of the maximum load;
-
Control function of cyclic load (sinusoidal);
-
Monitoring of vertical displacement of the prosthesis head and registering of cycles counting;
-
Computerized operation;
-
Equipment for recirculation and heating of solution test with temperature control within 37ºC±1ºC. Before starting the test each modular component composed by head and stem was prepared according to the following procedure: - Assembling the modular component and checking the fittings; - Placing the modular component in a device which allows alignment of angles referred in the standard, in concern to the load direction; - Drying the mounting; - Introducing the environment chamber together with the metallic cup with the mounted prosthesis; - Adding the saline solution (0.9% NaCl in distilled water) to the environment chamber; - Starting the pump for recirculation and heating the test solution; - Fitting the complete assembly in the test machine (Fig. 2). The machine was programmed for the application of the load in the center of the prosthesis head, with minimum 0.3 kN and maximum 3.3 kN loads, with sinusoidal 5 Hz frequency according to the conditions specified in ASTM F1875 standard. Maximum load requirements are specified in ISO 7206-4-2010 standard. A limiting condition was imposed for automatically stop ping the machine: a vertical displacement of 5 mm. If this value is not reached the test is finished after completing 10 million cycles of loading and unloading. The parameters of sample placement follow the requirements of ASTM F1440 standard. WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
Figure 2:
81
Complete assembly before starting test in INSTRON 8872 machine.
X-ray diffraction analysis was carried out in the following conditions: anode material: Cu Kα, angle between 10 and 100º, step size 0.05º, scan step time: 1 s, receiving slit size: 0.03 mm, current: 40mA, voltage: 40 kV.
3 Results and discussion The images of the modular surfaces (head and cone), which were obtained by optical microscopy, did not present surface defects such as cracks other types of flaws, introduced by the fabrication process, which could influence the results of this test. In Figs. 3 and 4 the aspect of the surface of the modular components (head and cone) of a hip prosthesis, before and after testing, are presented. The other prostheses have similar aspects. A significant corrosive attack inside the head and in the cone can be observed after 1 million cycles with increasing rates along the test, releasing corrosion products which resulted in mass loss in the head-cone connection. The mass loss, which was observed in the test solution, allowed the introduction of saline solution inside the head, increasing the predicted micro movements, leading to an accelerated wear (friction), with tearing of debris, according to Fig. 6. No type cracks and mechanical failure could be observed in the five hip prostheses, which were analyzed, thus meaning that the mechanical stability of the modular component was preserved along the test.
a
Figure 3:
b
Detailed view inside the heads: (a) before and (b) after the test. Magnification: 10X.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
82 Materials Characterisation V
b
a
Head-cone interface
Figure 4:
Detailed view of the cone region: (a) before (b) after. Magnification: 10X.
Figure 5:
Detailed view of the head-cone interface connection shown in Fig. 4-b. Magnification: 60X.
Figure 6:
Detailed view of tearing which was observed on the cone. Magnification: 60X.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
83
In Figs. 7 and 8 the corrosion products and particulate residues (debris), resulting from the fretting corrosion test can be observed in the bottom of the environment chamber (indicated by an arrow in fig. 8). These ones are consequence of the damages caused by fretting and corrosion fretting in the head-cone interface of the primary hip prosthesis. These images represent the five hip prostheses subjected to this analysis.
Figure 7: Test assembly after 106 cycles.
Figure 8: Detail.
Until 1 million cycles neither corrosion products nor particulate debris could be observed, but after this period the degradation of the modular component was increasing and localized on the head-cone, as can be seen in detail in Fig. 5. After 10 million cycles the corrosion products and particulate debris were quantified, resulting in an average value of 0.22 g. In the literature there is only one citation of fretting-corrosion products quantification, but in their experiment the cycle number was varied, thus hindering any type of comparison with the results presented in this work. For better comprehension of fretting corrosion mechanisms in hip modular components the variation of load and number of cycles is necessary. The characterization of the corrosion products and particulate residues by X-ray analysis is also very important, in order to identify the phases and type of metallic particles which were released, with the aim of minimizing the incidence of aseptic loosening, periprosthetic osteolysis and metallosis in the surgeries of total hip arthroplasty. In the present work a complete characterization of the fretting corrosion products was not concluded, but it can be observed that they presented a predominantly ellipsoidal shape with size distribution approximately uniform, whose larger diameter was around 500 μm. WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
84 Materials Characterisation V Anyway, the X-ray diffraction analysis performed in the corrosion products showed the presence of the following crystalline phases: Fe2O3, Fe3O4 and incipient Cr3O8, compatible with the ASTM F138 stainless steel.
4 Conclusion The evaluation of fretting corrosion on the head-cone connection interfaces of five modular hip prosthesis fabricated with stainless steel led to the following results: A significant grade of localized corrosion in the inner region of the head and in the head-cone interface was observed, originated from the fretting corrosion and fretting during 10 million cycles, resulting in particulate debris and corrosion products. In order to achieve a better comprehension of the fretting corrosion mechanisms, more tests are necessary, including changes in the applied load and number of cycles. At first glance, fretting corrosion does not affect the mechanical stability of the modular components, at least until 107 cycles.
References [1] Cavalcanti, E.H. de S., Souza, S.M.C. de, Ferreira, C. de A., Campos, M.M., Abud, I. de C. and Palmeira, L., (2002), “Avaliação da Resistência à Corrosão de Prótese Total de Quadril de Aço Inoxidável Austenítico Removida de Paciente”, 22° CONBRASCORR – Congresso Brasileiro de Corrosão, Salvador – Bahia. [2] Giordani, E.J., Ferreira, I. and Balancin, O., (2007), “Propriedades mecânicas e de corrosão de dois aços inoxidáveis austeníticos utilizados na fabricação de implantes ortopédicos”, REM: R. Esc. Minas, Ouro Preto, 60(1): 55-62. [3] Crestani, M.V., Boschim, L.C. and Schwartsmann, C.R., (2004), “Metalose Simulando Tumor Abdominal” , Revista Brasileira de Ortopedia, Nov/Dez. 2004. [4] Merritt, K.; Brown, S.A.; (1985), “Biological Effects of Corrosion Products from Metals”, Corrosion and Degradation of Implant Materials: Second Symposium, ASTM STP 859, A.C. Fraker and C.D. Griffin, Eds., American Society for Testing and Materials, Philadelphia, 1985, pp. 195-207. [5] ASTM F 1440 – 92 (Reapproved 2008), Standard Practice for Cyclic Fatigue Testing of Metallic Stemmed Hip Arthroplasty Femoral Components Without Torsion. [6] ISO 7206-4-2010 - Implants for surgery - Partial and total hip joint prostheses -- Part 4: Determination of endurance properties and performance of stemmed femoral components. [7] ASTM F 1875 – 98 (Reapproved 2004), Standard Practice for Fretting Corrosion Testing of Modular Implant Interfaces: Hip Femoral Head-Bore and Cone Taper Interface. WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
85
Improving corrosion performance by surface patterning M. Bigdeli Karimi, V. Stoilov & D. O. Northwood Department of Mechanical, Automotive and Materials Engineering, University of Windsor, Canada
Abstract Based on the idea that hydrophobic (low or non-wettable) surfaces can decrease the contact area between a corrosive solution and a surface, thereby potentially rendering the material more corrosion resistant, the effect of surface patterning on the corrosion behaviour of nickel was investigated. The surface patterning consisted of an array of holes of various diameters (D) and inter-hole spacings (L) that were produced by a laser ablation process. The corrosion behaviour of the patterned surfaces was studied using a potentiodynamic polarization method in a 0.5M H2SO4 electrolyte and compared with that of a polished reference sample. Following the potentiodynamic polarization corrosion test cycle, the corroded surfaces were examined using scanning electron microscopy (SEM) for morphological features and white light interferometry (WLI) to determine the surface roughness. The changes in surface morphology were related to the corrosion behaviour. A relationship was found between D, L, and the corrosion current density (Icorr), whereby the higher the (D/L )2 ratio, the higher the Icorr value. The corrosion potential (Ecorr) of all surface patterned samples was lower (less noble) than that of the reference sample in all tests. Keywords: corrosion resistance, surface patterning, hydrophobicity, laser ablation.
1 Introduction Wettability of solid surfaces is an important property and depends both on the surface chemistry and on the surface topology. Recently, hydrophobic and superhydrophobic surfaces (water contact angles larger than 150°) have received a lot of attention, due to their important applications ranging from self-cleaning WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line) doi:10.2495/MC110081
86 Materials Characterisation V materials and in-flight icephobic treatment [1] to microfluidic devices. The archetype superhydrophobic surface is that of the lotus leaf, on which a water droplet apparently forms a sphere, reducing the area of contact. This behavior, known as the lotus-leaf-effect, is found to be a result of the hierarchically patterned structure. Effectively the fluid cannot penetrate the air gaps between the patterned pillars on the surface, and therefore it forms a heterogeneously wetted contact surface of alternating liquid/solid contacts and air pockets. As a result the overall solid liquid contact area is significantly reduced [2–5]. In the present work we employ surface patterning to create heterogeneous wetting on metallic surfaces. We show that heterogamous wetting in metallic surfaces leads to reduction in corrosion rates and increase in corrosion resistance.
2 Experimental procedures 2.1 Specimen preparation Pure nickel (99.7 Wt.%) was selected as a model metal. Samples of 1.5 cm × 1.5 cm size were polished to a standard finish with a roughness that did not exceed 50 nm. Laser ablation with a single pulse copper bromide (CuBr) metal vapour laser was used to create special surface textures. During laser ablation, nitrogen (N2) was blown to protect the surfaces from oxidation and clean any debris. The pulse duration was selected at 30 ns. Different hole/pattern sizes were achieved with different laser output power within the range of 20-80 W. Figure 1 shows a schematic of the applied pattern. The diameter of the holes and the inter-hole spacing were varied to obtain different textures. Table 1 presents the selected hole size and the inter-hole spacing. The distances between the holes are labeled L1, L2 and L3, where as the D1, D2, D3 etc are the hole diameters.
Figure 1:
Schematic presentation of the proposed surface texture; D relates to the hole diameter and L defines the inter-hole spacing. Table 1:
Hole diameter (D) µm 5 10 20
The hole sizes and inter-hole spacing. Inter-hole spacing L1, µm 5 10 20
Inter-hole spacing L2, µm 10 20 30
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Inter-hole spacing L3, µm 20 30 40
Materials Characterisation V
87
For easy identification of the samples the following sample labeling system has been adopted: DxLy, where x is the diameter of the hole in m and the y is the inter-hole spacing in m. 2.2 Corrosion tests The corrosion resistance of the samples was determined using a potentiodynamic polarization corrosion test, wherein the samples were immersed into a 0.5M H2SO4 solution at room temperature (24 ◦C). A conventional three-electrode system was used in which a standard calomel electrode (SCE) served as the reference electrode and a platinum electrode as a cathode electrode. In the tests, the applied potentials were in the range of -0.7 to 1.5 V (with respect to SCE) with a scan rate of 1.0 mV/s. Corrosion rates were calculated in terms of the corrosion current density, Icorr, by using linear polarization resistance curves (LPR). The relationship between Icorr and the polarization resistance, Rp, is obtained from the Stearn–Geary equation [5]:
Rp where a and
c
a c E iapp 2.3( I corr )( a c )
(1)
are the Tafel slopes of the anodic and cathodic reactions,
respectively, Icorr is the corrosion current density and the
E is polarization iapp
resistance. After the corrosion test the sample surfaces were examined using JEOL 5800 scanning electron microscopy (SEM), and energy-dispersive X-ray spectrometry (EDS).
3 Results and discussion 3.1 Corrosion tests The electrochemical characteristics of the patterned Ni samples were investigated by potentiodynamic techniques described in Sec 2.2. Polarization curves for all patterned samples and the reference sample were analyzed and the corresponding corrosion current density, Icorr, and potential, Ecorr ,were obtained. Figure 2, shows the variation of the Icorr for different hole diameters (1/D) in samples with the same ratio of patterned to non-patterned area (D/L)2. Clearly samples with D=10 µm(1/D=0.1 µm-1) have the worst corrosion properties compared to any other samples with either pattern density((D/L)2=0.25 and (D/L)2=1.0). On the other hand, samples with D=20 µm(1/D=0.05 µm-1) show the lowest Icorr , and therefore better corrosion resistance at either pattern density. In addition, the corrosion resistance observed in samples with D=20 µm exceeds the performance of the reference sample (1/D=0.0 µm-1) which is a significant improvement of the corrosion properties of the metal surface. The observed WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
88 Materials Characterisation V trends in Figure 2 suggest that a decrease of the hole size (D) µm towards nanoscale size could decrease the corrosion current density, Icorr, even further. Figure 3 presents the Ecorr values for samples with pattern density of (D/L)2=1 and (D/L)2=0.25. The corrosion potential (Ecorr) of all surface patterned samples was lower (less noble) than that of the reference sample (1/D=0.0 µm-1) in all tests.
a)
b)
Figure 2:
The Icorr values versus inverse hole diameters (1/D) for all samples with (D/L)2=1 and (D/L)2=0.25 after the corrosion test.
a)
b)
Figure 3:
The Ecorr values versus inverse hole diameters (1/D) for all samples with (D/L)2=1 and (D/L)2=0.25 after the corrosion test.
3.2 SEM images after corrosion tests The patterned samples were examined by SEM before and after above corrosion tests (see Figs. 4–6). The original REF sample showed some severe, localized corrosion. It is evident that those areas were formed by coalescence of small pits. Sample D10L30 showed a severe corroded appearance (Figs. 5a and 5b). According to Icorr of this sample there is a good agreement between corrosion current density and the surface condition.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
a)
89
b)
Figure 4:
a)
SEM images of the reference sample(REF) a) before and b) after corrosion testing.
b)
Figure 5:
a)
SEM images of the D10L30 a) before and b) after corrosion testing.
b)
Figure 6:
SEM images of the D20L40 a) before and b) after corrosion testing.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
90 Materials Characterisation V Sample D20L40 exhibited the lowest amount of patterning degradation. Comparing the original sample and after the corrosion test there was no obvious change of the surface pattern [Figures 6(a) and 6(b)]. The measured Icorr for this sample was the lowest of all samples including reference and patterned. No surface damaged was detected also in samples D20L20 and D20L40. Clearly the observed surface damage in all samples was in a good agreement with the predicted trends by the measured Icorr values. Sample REF
a) Sample REF
b) Figure 7:
EDS analysis of the reference sample REF a) before and b) after corrosion testing.
3.3 EDS analysis of the patterned samples In order to investigate the significant improvement of the corrosion resistance in some samples energy-dispersive x-ray spectrometry (EDS) was used. The main goal was to evaluate the change of the O concentration on the surface of the samples and on the inside of the patterned holes. Fig. 7(a) refers to sample REF before the corrosion test. In the spectrum three elements, nickel (Ni), oxygen (O) and carbon (C) are observed. The intensity of the most intense peaks (Ni: Lα=0.851 and O: Kα=0.523 keV) was used to calculate the ratio of Ni/O (see Fig 7(a)). Comparing sample REF before the test [Fig. 7(a)] and after the test [Fig. 7(b)] shows that there is virtually no change in the O concentration on the surface. The WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
91
Ni/O ration remained at 49/5, indicating that passive oxide layer was not formed. Similar EDS spectra were obtained for all corrosion damaged samples. For instance the EDS spectra of sample D5L20 on the surface [Fig. 8 (a)] and in the hole [Fig. 8 (b)] after corrosion testing do not show any significant difference and the Ni/O ratio is the same as for the polished REF sample. However, the EDS spectra for the pattered samples with the improved corrosion properties (D20L20, D20L30, and D20L40) exhibit different behaviour. The Ni/O ratios for sample D20L40 before and after the corrosion test are shown in [Figs. 9 (a),(b) and Figs. 10(a),(b)]. The concentration of the O on the surface before and after the corrosion test slightly changed, which is consistent with the other corrosion damaged samples. However, the bottom of the patterned holes shows significant increase in O concentration (Figs. 8(b) and 10(b)). This clearly indicates that Ni oxides were formed but were not dissolved by the electrolyte. A possible explanation could be that the fluid actually did not reach the bottom of the patterned hole. In other words, in the samples with better corrosion resistance, the contact between the electrolyte and the metal surface is heterogeneous wetting – alternating solid/liquid zones and air pockets.
Sample D5L20-surface
a)
Sample D5L20-hole
b) Figure 8:
EDS analysis of the sample D5L20 after the corrosion test a) surface and b) hole.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
92 Materials Characterisation V
Sample D20L40-surface
a)
Sample D20L40-hole
b) Figure 9:
EDS analysis of the virgin sample D20L40 a) surface and b) hole. Sample D20L40-surface
a)
Sample D20L40-hole
b) Figure 10:
EDS analysis of the sample D20L40 after the corrosion test a) surface and b) hole.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
93
4 Conclusions Surface patterns with different pattern density were successfully created on Ni by laser ablation. The patterns had a distinct effect on the corrosion properties of the metallic surfaces. A specific group of patterns led to significant decrease of the corrosion current density and the corresponding corrosion rate. For this group of patterns it has been shown that a possible reason for the decrease in corrosion rate is the detected heterogeneous wetting on the patterned surface.
References [1] K. Varanasi, T. Deng, J. David Smith, M. Hsu, and N. Bhate, “Frost formation and ice adhesion on superhydrophobic surfaces” Appl. Phys. Lett. Vol. 97, 2010, No 234102. [2] M. Shafiei and A. T. Alpas, “Nanocrystalline nickel films with lotus leaf texture for superhydrophobic and low friction surfaces”, Applied Surface Science Vol. 256, 2009, pp. 710–719. [3] W. Barthlott and C. Neinhuis, “Purity of the sacred lotus, or escape from contamination in biological surfaces, Planta, Vol. 202, 1997, pp. 1-8. [4] S.J. Lee, C.H. Huang, J.J. Lai and Y.P. Chen, “Corrosion-resistant component for PEM fuel cells”, Journal of Power Sources, Vol. 131, 2004, pp. 162-168. [5] T. Liu, Y. Yin and L. Dong, “New application of the underwater superhydrophobic surface in the corrosion protection”, Advanced Materials Research, Vols. 79-82, 2009, pp. 1115-1118. [6] M. G. Fontana and N. D. Greene, “Corrosion engineering”, 2nd edition, McGraw Hill, 1978.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
This page intentionally left blank
Materials Characterisation V
95
Material characterisation to understand various modes of corrosion failures in large military vehicles of historical importance A. Saeed1, Z. Khan1, N. Garland1 & R. Smith2 1
Sustainable Design Research Centre, School of Design, Engineering and Computing, Bournemouth University, UK 2 The Tank Museum, Bovington, UK
Abstract Large military vehicles within museum collections are stored in two distinct environments, controlled and uncontrolled, with an intermittent transitional mode where vehicles travel between the two. Variable environmental conditions combined with operational factors pose significant risks to the reliability, durability and longevity of these vehicles. Although there are methods for retarding or decelerating aspects of failure, to maintain the integrity and originality of these vehicles as artefacts a sustainable methodology for conserving these vehicles should be developed. Corrosion is one of the significant contributors to the structural damage and material aging of historical military vehicles; therefore an experimental study was conducted to understand the prevailing mechanisms of failures due to corrosion with various types occurring in these vehicles identified. This paper presents various modes of corrosion in historic vehicles while X-ray Fluorescence and ultrasonic scanning corrosion mapping techniques characterise corrosion inhibiting materials and subsequent material loss. Understanding material profiles and their link to environmental exposure during use and non use of these vehicles will lead to a sustainable methodology for their conservation. Keywords: sustainable, military, vehicle, corrosion, materials, conservation.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line) doi:10.2495/MC110091
96 Materials Characterisation V
1 Introduction The structural and functional deterioration of the large metal structures due to their environmental exposure and the operating conditions are responsible for their structural aging [1]. Aging mechanisms in large metal structures such as bridges, aircrafts, vehicles etc is recognised as a growing problem [2]. Aging mechanisms such as corrosion, surface and sub surface cracks, wear due to interacting surfaces and undesired stresses are leading to mechanical failures, higher maintenance costs and compromises the structural integrity. Corrosion has been identified as one of the major contributor to the aging mechanisms; it affects all metal structures by deteriorating the material’s properties [3]. This research focuses upon the corrosion damage within the large historic vehicles in the museum environment. Corrosion is indiscriminate in affecting metal structures and museum artefacts are no exception. The military vehicles from World War I and II were designed and manufactured according to the automotive technologies in place at those times [4] with only a limited lifeexpectancy, however their participation in the war and the subsequent historic significance means it is important to devise sustainable methodologies for longterm preservation. Continuous corrosion problems within large historic military vehicles in the museum environment are leading to the concerns of their longevity, in particular where failures in protective measures are also identified [5]. For the purpose of sustainable methods it is important to understand the various corrosion modes, material characterisation and the material loss profile due to corrosion. For this research military vehicles were inspected for visual signs of corrosion and novel methods such as X-ray fluorescence for elemental identification and ultrasonic scanning for material loss were conducted. Visual inspection identified different modes of corrosion depending on their exposure to extreme environments during the war, operating conditions at the time of their service life, types of materials used and their current working or non-working environments. X-Ray fluorescence method found Fe, the main constituent; other elements such as W, Mo and Mn were also identified. Ultrasonic methods found severe material loss due to corrosion in the selected samples.
2 Experimental methodologies The following methods are used in the research. Only the most common and reoccurring types of corrosion which were found on a larger scale in the vehicles are presented here. 2.1 Visual inspection Visual inspection was completed on more than 20 large military vehicles from the World War I, II and post war era. The investigation found one or more forms of corrosion in the same vehicles. Some widespread and most occurring modes of corrosion are described from four military tanks Table 1. WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
Table 1:
97
Military tanks presented in this research.
No
Vehicle
Environment
1 2 3 4 5
Mark II Sherman* King Tiger Centurion Scorpion
Controlled Uncontrolled Controlled Uncontrolled Uncontrolled
Manufacture
Country of Origin
Era
1917 United Kingdom WW 1 1940-1941 United States WW 2 1943 Germany WW 2 1945 United Kingdom Post war 1973 United Kingdom Post War
XRF and Ultrasonic scanning is conducted on Sherman.
2.2 X-ray fluorescence X-ray fluorescence was conducted on Sherman M4A1 for elemental identification. The test sample was bombarded by an intense x-ray beam resulting in the emission of fluorescent x-rays from the test object. The emitted fluorescent x-rays are then detected by XRF analyser [6] which identifies the elements by measuring the energies of the emitted fluorescent x-rays and counting the number of rays in each energy spectrum [6, 7]. 2.3 Ultrasonic scanning Ultrasonic scanning uses pulse waves from 0.10 to 15.0 MHz frequencies range to measure material loss and other dimensional anomalies in the test objects [8]. This research work presents ultrasonic scanning using pulse echo method where pulses are transmitted and received back on the same side of test object to detect and characterise defects.
3 Results 3.1 Uniform corrosion With uniform corrosion a direct chemical attack affects the surface of the metal and then spreads out evenly across the entire surface, results are not usually fatal and it is not classified as one of the severe forms [9]. Figure 1 of the King Tiger’s flywheel shows uniform corrosion spreading across the surface. Uniform corrosion was also found in Tog II and FT 17 on considerably larger surface areas. Uniform corrosion, if left to continue, could lead to general thinning of the surface and may transform in to another form of corrosion [10]. 3.2 Fretting corrosion Fretting corrosion occurs at the interface of the interacting surfaces which are subject to relative motions. The King Tiger planetary gears (Figure 2) are a classic example of this form. Interacting mechanical wear lead to the removal of material particles, followed by the oxidation of the material debris and the newly exposed surface [11]. The oxidised debris would act as an abrasive media leading to further wear and fatigue failures could occur [9, 12]. WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
98 Materials Characterisation V
Figure 1:
Figure 2:
King tiger flywheel.
King tiger planetary gears.
3.3 Stress corrosion cracking Stress corrosion cracking is considered one of the severe types of corrosion, it could cause a partial or complete failure of a structure especially in the presence of dynamic or static loading and initiation of a stress corrosion crack [13]. The Mark-II Tank glacis plate in Figure 3 shows stress corrosion cracking where a crack is initiated and penetrating through the surface with no significant material loss on the surface. Stress corrosion is difficult to detect, hard to predict and unexpected failure can occur [13, 14]. 3.4 Inter-granular/exfoliation corrosion Inter-granular corrosion describes the attack on a metal’s or alloy’s grains or boundaries [15]. Corrosion products between the grain boundaries exert pressure on the grains and the result is exfoliation corrosion. The Centurion armoured skirt in Figure 4 is an example of exfoliation corrosion leading to lifting or leafing effect. Leafing or lifting effect is a direct result of the expanding corrosive products among the grains of the metals which forces the grains apart [9]. WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
Figure 3:
Figure 4:
99
Mark II glacis plate.
Centurion armoured skirt.
3.5 Galvanic corrosion Galvanic corrosion is the electrochemical reaction between two dissimilar metals in the presence of electron conductive path and the preferential corrosion of one metal more than the other [16]. This mode of corrosion was identified in the Scorpion, see Figure 5. The presence of moisture among the bolt, washer and the armoured skirt has caused this type of corrosion. 3.6 Pitting corrosion Centurion in Figure 6 shows signs of pitting corrosion. In this case the corrosion is confined to a small point with resulting cavities in the metal. The localised phenomenon of corrosion creates holes in the metal and then penetrates inwards, deteriorating metal from the inside. The corrosion products covers the pit and creating a small hole with no significant material loss on the surface leading to a catastrophic malfunction, it is one of the severe forms of corrosion which is very difficult to detect and hard to predict the result [14, 17, 18].
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
100 Materials Characterisation V
Figure 5:
Galvanic corrosion in scorpion.
Figure 6:
Pitting in centurion.
3.7 X-ray Fluorescence of the Sherman M4A1 X-Ray fluoresce (XRF) was conducted on three points A, B, C on corroded, E, F, and G on sand blasted surfaces and point D on the cross section of the sample, see Figure 7. Table 2 shows results of the points A, B and C. On averages Fe 92.07% is the highest, second highest is the Si 5.39%, Mn 0.91% was the third highest and the lowest was Cu with a trace of 0.04%. Figure 8 shows the graphical representation of the XRF conducted at point D where Fe 98.30% is the highest, second highest is Si 0.65%, Mn 0.63% the third highest and traces of Ti, V, Co and Mo were identified. XRF conducted at points E, F and G on the sandblasted surface shown on averages Fe 98.12% the highest, Mn 0.66% second highest, Si 0.59% the third highest while W 0.02% the lowest and no traces of Co or V were identified. Figure 9 shows the graphical representation of the XRF analysis at point G.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
Figure 7: Table 2:
101
XRF points on Sherman.
XRF of the corroded surface of M4A1.
Vehicle Type: Sherman Sample- 1 Time duration: 30 seconds
Surface: Corroded MQ for points: A= 5.70 B= 4.50 C= 4.60 06 American 06 American
Material Standard
06 American
Constituents %
A- Reading 254
B-Reading 255
C-Reading 256
4.89 0.46 0.19 0.05 0.11
5.92 0.54 0.19 0.05 0.08
5.38 0.85 0.2 0.05 0.09
1 2 3 4 5
Si P Ti V Cr
6
Mn
0.86
1.11
0.78
7 8 9 10
Fe Co Ni Cu
92.68 0.21 0.06 0.05
91.48 0.22 0.08 0.03
92.05 0.12 0.1 0.04
11 12
Mo W
0.05 0.33
0.06 0.2
0.06 0.12
3.8 Pulse echo ultrasonic scanning Pulse Echo scanning was conducted on the Sherman M4A1 with a probe of 4.00 MHz twin crystal and 10.00 mm diameter using 0.25 MHz Hi pass filter and 10.00 MHz low pass filter frequencies with a probe delay of 7.630 µS. Compression waves with a velocity of 5903.00 M/sec are used to scan a total area of 174.00 mm length with 6.040 pulses/mm and 68.00 mm width with 3.580 pulses/mm mapping material loss every 2.00 mm on the sample. WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
102 Materials Characterisation V
Figure 8:
XRF analysis of point D, No Fe (98.30%) included.
Figure 9:
XRF analysis of point G, No Fe (98.09%) included.
Figure 10 shows the end views and the c-scan of the sample with implementation of colour coding Red for 0.00 mm and Blue for 10.00 mm. Two cursors Red and Blue show the X-axis, Y-axis location and the corresponding depths of the particular points on the sample. A user defined area is taken for finding the maximum and minimum thicknesses starting at 2.00 mm on the X-axis, 2.00 mm on the Y-axis in Figure 11 and finishing at 172.00 mm on the x-axis and 66.00 mm at the y-axis in Figure 12 respectively to avoid any distortion in signals at the edges. The scan detected maximum remaining thickness of 9.80 mm at horizontal position of 150.00 mm and vertical position of 26.00 mm with a drastic loss in material to a minimum of 4.60 mm at horizontal position of 28.00 mm and vertical position of 8.00 mm. The maximum thickness of 9.80 mm is recorded at two different points. The second lowest depth is 4.70 mm and is detected at seven different WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
Figure 10:
Figure 11:
103
Pulse echo scan of M4A1.
User defined area starting at 2.00 mm on x-axis and 2.00 at y-axis.
positions; third lowest of 4.80 mm at six various points and the remaining thickness of 4.90 mm are recorded at four different points in the samples. The minimum remaining thickness of 4.60 mm is recorded at three different points.
4 Discussion The modes of corrosion identified above can be retarded by using the protective measures such as coatings and paint for uniform corrosion, application of lubrication and exclusion of air in the fretting corrosion, alleviating the stress in stress corrosion cracking, using same metal types in galvanic or applying insulation between two different metals. WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
104 Materials Characterisation V
Figure 12:
User defined area finishing at 174 mm on x-axis and 66 at y-axis.
The alloying elements such as Mn, Cr, V W and Mo identified though XRF are used as hardening agents to improve hardness, ductility and tensile strength [19]. Also Mo, Mn and P are used for the rust and corrosion prevention. Mo is used in high strength alloys and super alloys for strength and wear resistance [20]. Sherman M4A1 was designed by Lima Locomotive Works in the United States. At the time of manufacture the armoured thickness range was kept between maximum of 62.00 mm and a minimum of 12.00 mm [21, 22]. However it is not practical to follow any possible modification in the armour thickness and because of this reason the highest thickness recorded in the scan is considered the maximum thickness. During the service life, exposure to extreme environments has led to the deterioration and material loss. The ultrasonic scan of has shown a remarkable material loss of 5.20 mm between the maximum remaining thickness of 9.80 mm and the minimum remaining thickness of 4.60 mm. Currently in the museum collection, the Sherman travels between two controlled and uncontrolled environments. Varying environments and the likelihood of failing protective measures, temperature fluctuations, humidity ratio variations, rain water and the extent of chemical reaction etc will influence the rate of corrosion and in the controlled environment corrosion residues from may result in indoor atmospheric corrosion [23, 24].
5 Conclusion The research work uses a conventional method of analysing corrosion through surveying vehicles in the museum environment which is followed by novel methods of XRF for the elemental identification and Ultrasonic scanning to measure and map corrosion. Research work will be conducted for corrosion WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
105
modelling and for the life expectancy of large historic vehicles based upon the results from this research work in order to maximise the longevity, durability and reliability.
Acknowledgements The authors would like to thank Mark Nel, Mark Clark in AGR, Jon Previtt West-Dean College and Mike Hayton in The Tank Museum Bovington for their in kind support during this research work.
References [1] M. Colavita, "Service Life and Aging of Military Equipment," Corrosion: Environments and Industries (ASM Handbook), vol. 13C, pp. 220-228, 2006. [2] A. S. Vinod, "Corrosion in the Military," presented at the Corrosion: Environments and Industries (ASM International), 2006. [3] Z. Ahmad, "Basic Concepts in Corrosion," in Principles of Corrosion Engineering and Corrosion Control, Oxford: Butterworth-Heinemann, 2006, pp. 9-56. [4] I. C. Handsy and J. Repp, "Ground Vehicle Corrosion," Corrosion: Environments and Industries (ASM Handbook), vol. 13C, pp. 148-150, 2006. [5] M. T. Gudze and R. E. Melchers, "Operational based corrosion analysis in naval ships," Corrosion Science, vol. 50, pp. 3296-3307, 2008. [6] B. Beckhoff, Handbook of practical X-ray fluorescence analysis. Berlin; London: Springer, 2006. [7] R. v. Grieken and A. A. Markowicz, Handbook of X-ray spectrometry, 2nd ed., rev. and expanded. ed. New York, N.Y.: M. Dekker, 2002. [8] J. C. Drury, Ultrasonic flaw detection for technicians, 3rd ed. [Swansea]: Silverwing Ltd., 2004. [9] R. Baboian, Corrosion tests and standards : application and interpretation. Philadelphia, Pa.: ASTM, 1995. [10] H. H. Uhlig and R. W. Revie, Corrosion and corrosion control : an introduction to corrosion science and engineering, 4th ed. / R. Winston Revie. ed. Hoboken, N.J.: Wiley-Interscience; Chichester: John Wiley [distributor], 2008. [11] R. C. Bill, "The role of oxidation in the fretting wear process," 1981. [12] A. Neyman and O. Olszewski, "Research on fretting wear dependence of hardness ratio and friction coefficient of fretted couple," Wear, vol. 162164, pp. 939-943, 1993. [13] R. P. M. Procter, et al., "Stress-corrosion cracking of C---Mn steels in methanol-ammonia environments--I. Effects of environmental and mechanical variables," Corrosion Science, vol. 33, pp. 1009-1031, 1992. [14] R. C. Newman, "Stress Corrosion Cracking," in Shreir's Corrosion, J. A. R. Tony, Ed., Oxford: Elsevier, 2010, pp. 864-901. WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
106 Materials Characterisation V [15] C. A. e. Hampel, The encyclopedia of electrochemistry. NY: [s.n.], 1964. [16] H. P. Hack, "Galvanic Corrosion," in Shreir's Corrosion, J. A. R. Tony, Ed., Oxford: Elsevier, 2010, pp. 828-856. [17] Z. Szklarska-Smialowska, Pitting corrosion of metals. Houston, Tex.: National Association of Corrosion Engineers, 1986. [18] R. Baboian, "Automotive corrosion tests and standards," 1996. [19] M. F. Ashby and D. R. H. Jones, Engineering materials 2 : an introduction to microstructures, processing and design, 3rd ed. Oxford: ButterworthHeinemann, 2006. [20] F. F. Schmidt, The engineering properties of molybdenum and molybdenum alloys. Columbus, Ohio: Defense Metals Information Center, 1963. [21] P. Chamberlain and C. Ellis, British and American tanks of World War II: the complete illustrated history of British, American, and Commonwealth tanks, 1939-1945. London New York: Cassell & Co. ;Distributed in the USA by Sterling Pub., 2000. [22] R. Jackson, Tanks : and armoured fighting vehicles. Bath: Parragon, 2007. [23] A. R. Mendoza and F. Corvo, "Outdoor and indoor atmospheric corrosion of non-ferrous metals," Corrosion Science, vol. 42, pp. 1123-1147, 2000. [24] I. Odnevall and C. Leygraf, "The formation of Zn4Cl2(OH)4SO4 · 5H2O in an urban and an industrial atmosphere," Corrosion Science, vol. 36, pp. 1551-1559, 1994.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Section 4 Computational models and experiments
This page intentionally left blank
Materials Characterisation V
109
A multi-factor interaction model (MFIM) for damage initiation and progression C. C. Chamis NASA Glenn Research Center Cleveland, USA
Abstract A Multi-Factor-Interaction-Model (MFIM) is briefly described to represent complex point material behavior in a single equation. The model is of product form in order to represent coupled interactions and to be computationally effective. The model describes a continuum or surface in space that represents the complex material behavior in terms of the various factors that affect a specified material behavior. The material specified behavior is inclusive of all material properties, mechanical, thermal, physical and effects thereon, such as temperature, time, cyclic loadings, etc. Sample case results simulated by using MFIM are compared with test data to illustrate its versatility and its relevance to reality. These results show that the MFIM can accurately predict metal matrix composite fatigue data and mechanical properties of a steel alloy. Helpful guidelines for its effective use are also included. Keywords: material properties, high temperature, nonlinearities.
1 Introduction The simulation of complex material behavior resulting from the interaction of several factors (such as temperature, nonlinear material due to high stress, time dependence, fatigue, etc), has been mainly performed by factor-specific representations. For example, entire text books are devoted to plasticity, creep, fatigue and high strain rate to mention only a few. Investigators have derived equations that describe material behavior for each factor-specific effect. Suppose we visualize that the material behavior is a continuum represented by some surface. Then, we can think of some representation which describes that surface which is inclusive of all participating factors that affect material behavior either singly or interactively in various combinations. To that end, research has WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line) doi:10.2495/MC110101
110 Materials Characterisation V been a continuing activity at Glenn Research Center for about twenty-five years. It started with a primitive form of MFIM representation for describing complex composite behavior in polymer matrix composites [1]. It was extended to metal matrix composites [2] and continues to be evolved in Aerospace Plane and the High Speed support of the National Research Programs. The result of all this research is that general guidelines for its usage are briefly described. Simulation results are presented and compared (where available) with experimental data to illustrate its versatility and perhaps demonstrate the claim for its uniqueness.
2 Fundamental considerations and the (MFIM) We start with the premise that if we are to quantify the range of factors affecting material properties, we need a description of material behavior. In this context, it is reasonable to consider that material behavior constitutes an n-dimensional space (Material Behavior Space (MBS)) where each point represents a specific aspect of material behavior. It is further reasonable to assume that MBS can be described by an assumed interpolation function. One convenient interpolation function is a polynomial of product form because mutual interactions among different factors can be represented by the overall product, and includes those cross products in common algebraic polynomials. In this investigation, MBS is assumed to be described by the following multifactor interaction equation (MFIM): N MP Aimi (1) M P0 i 1 where M P is the property affected to be evaluated. MPo corresponds to the initial (reference) material state or condition. Ai represents the ith factor that influences material behavior, and mi is an exponent. Ai is further defined by: B (2) Ai 1 B0 Here B represents a specific cause factor for behavior (for example, temperature), and Bo is the corresponding final value. This concept is schematically represented in Fig. 1, the development of the multi-factorinteraction model (MFIM) to represent complex material behavior by a single equation. The objective of the present paper is to briefly describe MFIM and present results obtained there from to illustrate its uniqueness and its versatility. Values for Bo and mi for specific behavior are selected either from known behavior or more likely from a best judgment in conjunction with consultations with seasoned professionals for that behavior. By representing the MBS with the MFIM of product form (Eq. (1)), we gain another distinct advantage. The behavior factors, B, can also be represented by another level of MFIM or progressive substructuring of equation (1). The progressive substructuring leads to a multitier representation of the MBS that permits intrinsic lower tier behaviors to influence more than one factor at the next higher tier. In other words, the observed specific behavior (Bi may depend on another set of lower WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
111
tier elemental behaviors). Further, the behavior factors in this lower set of specific behaviors may depend on yet another next lower tier of elemental behaviors. That is, there are usually sets and subsets of specific behaviors that hierarchically influence the higher level behaviors. When this is done, N can be limited to 6, (for example), but the number of factors influencing material behavior at the next lower tier will increase exponentially as Nj where j is the number of 6-factor tiers. For example, when j = 3, N = 216, and so forth. This representation is natural for multiparallel processing computers where the tiers are programmed with different granularities. Obviously, then, the motivation for selecting such a form is for computational and programming effectiveness. Another reason for selecting an MFIM of product form is that the effect of each factor can be evaluated separately. The interpretation of Bo is that it represents a scale, whereas mi represents a shape or path. For example, (1 – B/Bo)mi where 1 > B/Bo and + ∞ < mi < – ∞, covers the whole space as is illustrated in Fig. 2.
Figure 1:
Conceptual schematic of material behaviour through a MultiFactor Interaction Model (MFIM).
The inclusiveness of this particular form, combined with its simplicity, makes it very attractive for a computational simulation. An expanded form of MFIM is shown in Eq. (3) below: m
Mp Tgw T Mpo Tgw To E ...1 e E ef
u
1 Sf Ce 1 Cef
n
p
t 1 Sf tf
v
...
Cc 1 Ccf
M NM 1 Sf Nf M
q
r
T NT 1 Sf Nf T
s
1 f
...
The Multifactor Equation
(3)
w
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
112 Materials Characterisation V for inclusiveness and convenience of presentation. The top line includes six factors in parentheses: ( Tgw – T) / (Tgw – T0 ) denotes temperature effects: (1 – σ/Sf ) denotes combined stress effect; (1 – σ t / Sf tf ) denotes time dependent effects: ( 1 – σ m Nm / Sf Nfm ) denotes mechanical cyclic load effects where σm is the cyclic stress due to NM cycling; (1 – σT NT / Sf Nft ) denotes thermal stress due to NT cycling; ( 1 – ω / ωf ) denotes frequency effects due to ω – frequency. The factors in the second line denote erosion, corrosion and chemical or metallurgical effects. The dots between the first and second lines indicate that several other factors can be included. Suffice it to say that the MFIM is generic and inclusive. Two points to be noted are: (1) not all terms have to be included; and (2) substructuring may be appropriate after six factors. For example, the three factors in the second line affect some, if not all the factors in the first line. This definitely is the case for the chemical/metallurgical factor.
Figure 2:
Effect of variation in the exponent.
3 Guidelines for usage of the (MFIM) Some general guidelines for usage of the MFIM are appropriate: (1) Several factors may be programmed for inclusiveness. (2) Factors that do not contribute to that simulation assign zero (0) exponents. (3) Selecting exponents may be intriguing in the absence of data. Start with some guess of expected behavior and let the feedback from the material property behavior to guide the next update. (4) The exponents can also be evaluated if there is some data available especially from combined testing. This can be done by expressing the expanded equation using logs and then evaluate them by using the least squares method WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
113
since the exponents will be the coefficients of linear algebraic equations. (5) To simulate block fatigue evaluate the degradation up to the end of the first block then use the degraded properties for the second block and repeat for additional blocks – degradation after second input to third etc. (6) Unloading generally requires different exponent then loading. Select the unloading exponent to simulate the expected loading-unloading by stress cycle. (7) Intermittent time effects are handled the same way as block fatigue. (8) Combined fatigue (thermo mechanical) is simulated by using the corresponding cycles and stresses in the respective factors. (9) Frequency effects are simulated in combination with cyclic loads. Stress value is not needed here since it is included in the cyclic effects factor. (10) The static stress and the temperature factors when present are used in combination with all active factors for that simulation effect. (11) To simulate fabrication process use the temperature factor, the stress factor due to pressure, a flow model and update the geometry through the finite element model of the component. The user will invent other ways to use MFIM that the author has not even thought of yet. The most important point to remember is that MFIM is to be used incrementally where the current values are updated from global thermo structural, etc. analyses. This may require a two level iteration: (a) local to achieve local equilibrium. During the iteration, the factors are changed until local equilibrium is reached; and (b) global to achieve global equilibrium with respect to boundary loads and supports conditions and internal stress field or energy input in the increment equals the energy added to the structure. Another important point to note is that the exponent for specific factor in MFIM represents a general trend and not the entire precise path from its reference value to its final value. Also, note that the MFIM is used for all the properties that is the same factors but with different exponent. Results from sample cases for MFIM application are described for: metal matrix composites (MMC), and a nickel base alloy. As mentioned previously, these results are presented to illustrate a few of the generic features of the MFIM applicability to convey to the reader what has be done and infer therefrom what can be done. The specific how is described in the references.
4 MFIM application to MMC In this sample case MFIM is part of the METCAN (Metal Matrix Composite Analyzer) [3]. The simulation for the properties to be shown starts from the fabrication process as depicted in Fig. 3. The constituent material properties used are summarized in Table 1. MMC Simulation of the transverse strength is shown in Fig. 4. For this simulation the factors for temperature, stress, and time were activated with respective exponents about 0.5. It can be observed that the simulation predictions track the data almost exactly. That simulation is very complex because it incorporates matrix, interphase and fibers. It starts from the fabrication process and ends with monotonic loading to fracture. Sample simulation results for fatigue of the same composite are compared with data in Fig. 5. WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
114 Materials Characterisation V
Figure 3: Table 1:
METCAN Metal Matrix Composite Behavior – computational simulation sequence. Constituent (fiber/matrix) material properties used in METCAN.
It is observed that the agreement is very good. In addition, the simulation identifies the types of failures that lead to specimen fracture that is achievable by the use of the MFIM. For this case, the mechanical cycle term in the MFIM was also activated with an exponent of about 0.5 for all the constituent properties. Simulation results for stress at times are compared in Fig. 6. As can be seen the agreement is very good. WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
115
Figure 4:
Transverse stress strain curve of SiC/Ti-6-4 T = 73 oF, FVR = 0.34.
Figure 5:
Predicted isothermal fatigue life [0] SCS6/Ti-24A1-11 Nb, 0.35 FVR (70 oF (23 oC), R = 0.1).
Figure 6:
Creep behavior of [0] SCS6/Ti-24Al-11Nb composite at 815oC and 310 MPa; FVR = 0.35; stress-free temperature = 815oC.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
116 Materials Characterisation V
Figure 7:
Calibration of nickel base alloy 713C elastic modulus as a function of temperature.
Figure 8:
Calibration of nickel base alloy IN-100 coefficient of thermal expansion as a function of temperature.
The time factor was activated for the fabrication process and the tensile load to fracture. Additional simulation results are described in [3, 4].
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
117
5 Properties for nickel base alloy Simulation results of temperature effects on nickel base alloy 713C properties [5] are shown respectively in Fig. 7 for modulus, Fig. 8 for thermal expansion coefficient, Fig. 9 for thermal heat conductivity, and in Fig. 10 for heat capacity. Collectively the comparisons are in good agreement. Collectively these results show the effectiveness of MFIM.
Figure 9:
Calibration of nickel base alloy MAR-M 200 thermal conductivity as a function of temperature.
Figure 10: Calibration of nickel base alloy IN-100 specific heat as a function of temperature. WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
118 Materials Characterisation V
6 Concluding remarks The salient concluding remarks from a description of the multi-factor interaction model (MFIM) to predict complex material behavior by using a single equation and select applications are: (1) MFIM is a material point simulator. It is unique and inclusive for both composites and metals. (2) It is used incrementally in combination with analyses that provide local stress information. (3) It usually requires alternative solutions consisting of both local and global convergence. (4) Applications for meta matrix composites under different loading conditions predicted results that were in very good agreement with test data. (5) Familiarity and confidence and extensive usage will be gained by continuing usage application to complex solutions of material behavior and to simulate difficult and even untried problems.
References [1] Chamis, C.C., Lark, R.F. and Sinclair, J.H.: Integrated theory for Predicting the Hydrothermal Mechanical Behavior of Composite Structural Components. ASTM pp. 160-192. [2] Chamis, C.C. and Hopkins, D.A.: Thermo Viscoplastic Nonlinear Constitutive Relationships Structural Analysis of High Temperature Metal Matrix Component. NASA TM 87291, 1985 [3] Chamis, C.C., Murthy, P.L.N. and Hopkins, D.A.: Computational Simulation of High Temperature Metal Matrix Composites Cyclic Behavior. ASTM, STP 1080, 1990, pp 56–69. [4] Tong, M.T., Singhal, S.N., Chamis, C.C. and Murthy, P.L.N.: Simulation of Fatigue Behavior of High Temperature Metal Matrix Composites. ASTM Reprint from Standard Technical Publication, 1253, 1996, pp. 540-551. [5] Boyce, L. and Chamis, C.C.: Probabilistic Constitutive Relationships for Cyclic Material Strength Models. AIAA.ASME/ASCE/AHS 29th Structures, Structural Dynamics and Materials Conference. Part 3, AIAA, 1988, pp. 1299-1306. [6] Progressive Fracture Structural Analysis of National Wind Tunnel Structures by L. Minnetyan, NASA CR 198485, May 1996.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
119
Analytical solution of a two-dimensional elastostatic problem of functionally graded materials via the Airy stress function H. Sakurai Co-operative Education Centre, Sendai National College of Technology, Japan
Abstract Functionally Graded Materials (FGMs) possess properties that vary gradually as a function of spatial coordinates. They are different from conventional composite materials in that they have no distinct interfaces at which their material properties change abruptly. These FGMs are suitable for various applications, such as aerospace, nuclear fusion, biomaterial electronics, etc. In practice, applications of analytical solutions are limited. However, the analytical solutions are very important as standards for evaluating numerical simulation results and they are also important to mathematical understanding. Little research on the analytical solutions of two-dimensional elastostatic problems has been reported. Furthermore, few analytical solutions using Airy stress functions have been published. The purpose of this paper is to propose an analytical method for the two-dimensional elastostatic problems of FGMs using the Airy stress function. In the present investigation, FGMs in which the properties of the materials vary exponentially in one direction are examined. A few numerical examples are presented and the validity of the method is shown by comparisons with the results of past studies. Keywords: analytical solution, functionally graded material, two-dimensional problem, Airy stress function.
1 Introduction The Functionally Graded Materials (FGMs) possess properties that vary gradually as a function of spatial coordinates. They are different from conventional composite materials in that they have no distinct interfaces at which WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line) doi:10.2495/MC110111
120 Materials Characterisation V their material properties change abruptly [1]. These FGMs are suitable for various applications, such as aerospace, nuclear fusion, biomaterial electronics, etc. Chakraborty et al. developed a new beam element solving FGM beam structures based on the first order shear deformation theory [2]. Nguyen et al. proposed a first-order shear deformation plates models for modeling structures made of FGMs [3]. Xinag and Yang studied free and forced vibrations of a laminated FGM Timoshenko beam with variable thickness under heat conduction [4]. Sankar and his co-workers have been reporting analytical methods for the thermomechanical analysis of FGM beams [5–7]. Zhu and Sankar presented an elasticity solution of a simply supported FGM beam having variation of Young’s modulus distributed by a polynomial in the thickness direction [6]. In Ref. [6], the Fourier series method is used to reduce the governing partial differential equations to the ordinary equations that are then solved by the Galerkin method. Miers and Telles proposed the Boundary Element-Free Method belonging to a meshless technique, for two-dimensional elastostatic analysis of FGMs [8]. Zhong and Yu presented explicit solutions of a cantilever FGM beam having arbitrary graded variations of material properties distributed in the thickness direction based on two-dimensional theory of elasticity [9]. Applications of analytical solutions are limited to practical shapes of analysis regions and boundary conditions. However, the analytical solutions are very important as standards for evaluating numerical simulation results, such as finite element method etc., and they are also important to mathematical understanding. Little research on the analytical solutions of two-dimensional elastostatic problems has been reported. Furthermore, to the best of the author’s knowledge, few analytical solutions using Airy stress functions have been published. The objective of the present paper is to describe the analysis of the twodimensional elastostatic problems of FGMs using the Airy stress function. In this study, FGMs in which the properties of the materials change exponentially in one direction are treated. A few numerical examples are presented and the validity of the method is shown by the comparisons with the results of the present method and results of past studies.
2 Two-dimensional problem of FGMs and basic equations In the Cartesian coordinate system O-xz , we consider a simply supported FGM beam subjected to a transverse load as shown in Fig. 1. The length in the x direction is l and the length in the z direction is h . In Fig. 1, the uniform transverse load and the simply supported boundary condition are one of the examples. In the absence of body forces the equilibrium equations are given as
xz zz 0 x z where xx , zz , zx are stress components and zx xz . xx zx 0 z x
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
(1)
Materials Characterisation V
121
p z ( x) p 0
O
x
h l
z Figure 1:
An example of a simply supported FGM beam.
The relationships between strains and displacements are
xx where
u x
xx , zz , zx are
zz
w z
zx
strain components and
u w z x
(2)
u, w are the displacement
components in the x and the z direction respectively. The strain components should also satisfy the following compatibility condition.
2 xx 2 zz 2 zx 0 zx z 2 x 2
(3)
The constitutive equations are given as
xx s11 xx s13 zz where
zz s13 xx s33 zz
zx s44 zx
(4)
s11 , s13 , s33 , s 44 are elastic moduli. The material properties of FGMs
change gradually as a function of spatial coordinates. We assume material properties varying exponentially in the z direction, i.e. z
s ij s F ( z ) s e h
(5)
where sij are their corresponding values in the plane
z z 0 with F(z0 )=1 and
0 ij
0 ij
0
F(z) , is called graded function, which expresses the distribution of material properties, and the parameter α is called the graded index [9]. Now, we introduce Airy stress function Φ ( x,z ) expressed by the following equations.
xx
2 2 z
zz
2 x 2
zx
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
2 zx
(6)
122 Materials Characterisation V Substituting Eqns. (4)-(6) into Eqn. (3), the next governing equation with respect to Airy stress function ( x , z ) is obtained. 4 3 4 4 0 0 0 0 s s s s ( 2 ) 2 13 44 33 11 h z 3 z 4 x 2 z 2 x 4 2 2 3 2 2 0 0 0 s s (2 s130 s 44 ) 0 11 13 h x 2 z h 2 z 2 h 2 x 2
s110
(7)
The above equation is the same form as the governing equation of the plates bending with variable rigidity, and it is possible to adopt the same method for the solution [10].
3 Method of solution In Eqn. (7), we introduce the following Airy stress function Φ ( x,z ) .
( x, z ) Ai ( z ) f i ( x)
(8)
i 1
The function f i ( x) should be assumed to satisfy the mechanical boundary conditions at the x 0 and x=l . For instance, under the simply supported condition and the cantilever condition, it should be chosen to satisfy σxx=0, σzz=0 at the both ends and σxx=0 and σxz=0 at the free end respectively. 3.1 Formulation for simply supported FGM beam subjected to uniform pressure In this problem, we assume the following function
Ai ( z ) f si ( x ) i 1
where
i
A ( z ) sin x
i 1, 3, 5
i
i
(9)
i , f si ( x) sin i x . And the stress components are as follows l d 2 Ai ( z ) xx f si ( x) dz 2 i 1, 3, 5
zz
i 1, 3, 5
zx
i 1, 3, 5
2 i
Ai ( z ) f si ( x )
i
dAi f ci ( x) dz
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
(10)
Materials Characterisation V
123
f ci ( x) cos i x . Substituting Eqn. (9) into Eqn. (7), the following ordinary differential equation with respect to the unknown function Ai (z ) is where
obtained easily. 3 2 2 d 4 Ai 2 d Ai 0 d Ai 0 0 0 s s s s 2 { ( 2 ) } 11 11 13 44 i h dz 3 dz 4 h2 dz 2 2 2 dAi 0 0 (2 s130 s 44 ) i ( s33 i 2 s130 2 ) i 2 Ai 0 h dz h
s110
(11)
The solution of the above equation is given as
Ai ( z ) C i ,1e i ,1 C i , 2 e i , 2 C i ,3 e i , 3 C i , 4 e i , 4 r z
where the
r z
r z
r z
(12)
ri , j ( j 1, 2, 3, 4) are the roots of the following the 4-th order
equation and the
C i , j ( j 1, 2, 3, 4) are arbitrary constants to be determined by
the boundary conditions at z h 2 and z h 2 .
2 2 1 3 0 ) i2 ri , j ri , j 2 0 (2 s130 s 44 h s11 h 1 1 0 2 2 0 0 (2 s130 s 44 ) i2 ri , j 0 ( s 33 i s130 2 ) i2 0 h s11 s11 h
ri , j 2 4
Considering Fourier expansion of the transverse pressure
(13)
p z ( x) p0 , the
following equation is obtained.
pz
4 p0 1 1 f si ( x) sin i x i 1,3,5 i i 1,3,5 i
4 p0
(14)
The boundary conditions at the both surfaces are as follows. zz p0 , zx 0 at z h 2
zz 0
,
zx 0
at z h 2
(15)
Substituting Eqn. (10) and Eqn. (14) into Eqn. (15), the following four equations for determining the arbitrary constants C i , j are derived.
4 p0 h Ai ( ) 2 i 2 i h Ai ( ) 0 2
dAi h ( ) 0 dz 2 dAi h ( )0 dz 2
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
(16)
124 Materials Characterisation V
3.2 Formulation for cantilever FGM beam subjected to uniform pressure We deal with the formulation of a cantilever FGM beam such as shown in Fig. 2.
p z ( x) p0
O
x
h l
z Figure 2:
An example of a cantilever FGM beam.
It is necessary to assume the function f i ( x) satisfying the mechanical boundary conditions at the free end. One of the simplest functions satisfying the above conditions is
A2 ( z ) f 2 ( x) A2 ( z ) x 2 where
(17)
f 2 ( x) x 2 . The stress components are as follows.
xx zz zx
d 2 A2 2 2 d 2 A2 f ( x ) x 2 z 2 dz 2 dz 2
d 2 f 2 ( x) 2 A ( z ) A2 ( z ) 2 2 x 2 dx 2
dA ( z ) df 2 ( x) dA ( z ) 2 2 2 2x zx dz dx dz
(18)
Substituting Eqn. (17) into Eqn. (7), the following ordinary differential equation is derived again. 3 2 2 d 4 A2 0 2 d A2 0 0 0 2 d A2 s x 2s11 x {2(2s13 s44 ) s11 2 x } 2 h dz 4 dz 3 h dz (19) 2 0 dA2 2(2s130 s44 ) 2s130 2 A2 0 h dz h 0 2 11
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
125
We consider the above equation as the following ordinary differential equation with respect to z at the given section x x0 (constant). 3 2 2 d 4 A2 2 d A2 2 d A2 0 0 0 0 s x 2s11 x0 {2(2s13 s44 ) s11 2 x0 } 2 h dz 4 dz3 h dz (20) 2 0 dA2 2(2s130 s44 ) 2s130 2 A2 0 h dz h 2 0 11 0
The solution of the ordinary differential equation is given in the same form as Eqn. (12). The boundary conditions on the top z h 2 and on the bottom
z h 2 at the x x0 (constant) are expressed as follows.
h A2 ( ) 2 p 0 2 h A2 ( ) 0 2
dA2 h ( ) 0 dz 2 dA2 h ( )0 dz 2
(21)
4 Numerical examples 4.1 Analysis of simply supported FGM beam under sinusoidal pressure Let us consider a simply supported FGM beam subjected to a sinusoidal pressure as shown in Fig. 3. For convenience, the results are indicated in the coordinate system of Fig. 3 [6]. The Young’s modulus is assumed to be of the form E E0 e αz h and the constitutive equation is the same as that of Ref. [7].
x
O
h l
z Figure 3:
A simply supported FGM beam subjected to sinusoidal pressure.
Two types of the material properties are considered, and the ratios of Young’s moduli of the top surface and the bottom surface are Eh E0 =10 and
Eh E0 =0.1. In the case of the former, the graded index in Eqn. (5) corresponds to α 2.30 , and in the case of the latter, it corresponds to α=-2.30 . Further, in the case of the former, the load is applied on the softer WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
126 Materials Characterisation V surface and in the case of the latter, on the harder surface. In both cases, Young’s modulus is E0 =1 [Gpa], Poisson’s ratio is ν 0.25 and the thickness is h 10 [mm]. The type of the applying force is sinusoidal pz ( x) p0 sin ξx=p0 sin iπx l , ξ=iπ l , i=1,3,5 . Figure 4 shows the axial stress distribution at the middle span for Eh E0 =10 ,ξh=1 , and Fig. 5 shows the shear stress distribution at the middle span for Eh E0 =0.1 ,ξh=3 . In the figures, “Galerkin method” indicates the results of Ref. [6]. The vertical axis is the normalized stress values. The stress xx is divided by the xx ( x, h) , and the stress xz is divided by the average value at the middle section. From these results, it can be noted that the present results agree well with the results by Ref. [6] for the various conditions of the analysis. Analysis of cantilever FGM beam under uniform pressure The analysis of the cantilever FGM beam subjected to the uniform pressure as shown in Fig. 2 have been carried out [9]. The length l is l = 1 [m] and the thickness h is h = 0.2 [m]. The material properties at z = z0 = 0 are given Table 1, and the magnitude of the transverse pressure p0 is p0 = 1 [Pa]. ξh=1, graded index=+2.30 Present solution
Galerkin method
σxx(x,z)/σxx(x,h)
2.0 0.0 0.0
0.2
0.4
0.6
0.8
1.0
-2.0 -4.0 -6.0 z/h
Figure 4:
Axial stress
xx ( x, z ) / xx ( x, h)
through the thickness of FGM
beam at x l 2 . For the graded index component
xx
3 ,
Fig. 6 shows the distributions of the stress
at the clamped edge. The horizontal axis is the non-dimensional
coordinate z and the vertical axis is the stress values. In this figure, “Ref. [9]” means the results of Ref. [9]. The present results again are in good agreement with those of Ref. [9]. It was also confirmed that the distributions of the stress components zz and zx are also in good agreement. WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
127
ξh=3 , graded index=-2.30 Present solution
Galerkin method
σxz/σxz(average)
2.0 1.5 1.0 0.5 0.0 -0.5 0
0.2
0.4
0.6
0.8
1
z/h
Figure 5:
Shear stress
xz / xz (average)
through the thickness of FGM
beam at x l 2 . Table 1:
Materials properties.
Material constants
Values [1/Pa]
s
5.41 10 12
s
1.51 10 12
0 11 0 13
0 s33
9.52 10 11
0 s44
1.37 10 10 graded index α=+3, -3
σxx(l, z)
Present (α=+3) Present (α=-3)
Ref.[9] (α=+3) Ref.[9] (α=-3)
250.0 200.0 150.0 100.0 50.0 0.0 -50.0 -0.5 -0.4 -0.3 -0.2 -0.1 0.0 0.1 -100.0 -150.0 -200.0 -250.0 z/h
Figure 6:
Stress at the clamped end
0.2
0.3
0.4
0.5
xx (l , z ) ( 3 ).
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
128 Materials Characterisation V 4.2 Analysis of simply supported FGM beam under uniform pressure The simply supported FGM beam as shown in Fig. 1 is analyzed. The sizes, the elastic constants and the applied load are the same as those of the previous section. The graded index α is assumed as α 3 . The distributions of the stress components σ xx at the x l 2 0.5 and σ zx at the x 0 are plotted in Figs. 7– 8. The number of terms adopted in the Fourier series of Eqns. (9) and (14) is ten. We confirmed that taking at least eight terms is sufficient accuracy for the solution in this problem. graded index α=+3.0 20.0 10.0
σxx
0.0 -10.0 -0.5 -0.4 -0.3 -0.2 -0.1
0.0
0.1
0.2
0.3
0.4
0.5
-20.0 -30.0 -40.0 -50.0 -60.0 -70.0 z/h
Figure 7:
Stress σ xx through the thickness at x 0.5. graded index α=+3.0
5.0 4.5 4.0
σzx
3.5 3.0 2.5 2.0 1.5 1.0 0.5 0.0 -0.5
-0.3
-0.1
0.1
0.3
z/h
Figure 8:
Stress σ zx through the thickness at x 0 .
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
0.5
Materials Characterisation V
From these results, the value of the stress value stress zx . Comparing the tensile stress compressive stress
xx
xx
xx
129
is greater than that of the
on the bottom surface with the
on the top surface, the greater compressive stress is
caused on the softer surface. The shear stress distribution
zx
is different from
that for homogeneous beams. The point where the maximum stress appears moves toward the softer surface. It may be possible to control distribution of stresses by adjustment of materials properties.
5 Concluding remarks In this paper, one analytical solution method for two-dimensional elastostatic problems in FGMs in which the properties of the materials vary exponentially in one direction is proposed. The method is based on the idea that the governing equation of this problem is of the same form as the governing equation for the plate bending with variable rigidity. The analysis method uses the Airy stress function. From analysis of a few numerical examples, most of the results correlate well with other solutions and the validity of the method is shown. The distributions of displacement components are not stated, however we can obtain them by integrating the strains and displacements relationships. Future work will be focused on development of an analytical method of solution for analyzing problems of FGMs having an arbitrary variation of their material properties expressed by functions other than exponential.
References [1] Suresh, S., and Moretensen, A., Fundamentals of Functionally Graded Materials, IOM Communications Ltd, London, 1998. [2] Chakraborty A., Gopalakrishnan S. and Reddy JN., A new beam finite element for the analysis of functionally graded materials, International Journal of Mechanical Science, Vol.45, pp.519-539, 2003. [3] Trung-Kien Nguyen, Karam Sab and Guy Bonnet, First-order shear deformation plate models for functionally graded materials, Composite Structures, 83, pp.25-36, 2008. [4] H.J. Xinag and J. Yang, Free and forced vibration of a laminated FGM Timoshenko beam of variable thickness under heat conduction, Composites: Part B Vol.39, pp.292-303, 2008. [5] Sankar BV., An elasticity solution for functionally graded beams, Composite Science Technology, 61, pp.689-696, 2001. [6] Zhu H. and Sankar BV., A combined Fourier series-Galerkin method for the analysis of functionally graded beams, Journal of Applied Mechanics, Vol.71, pp.421-424, 2004.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
130 Materials Characterisation V [7] Sankar BV. and Tzeng JT., Thermal stresses in functionally graded beams, AIAA Journal, Vol.40, pp.1228-1232, 2002. [8] L.S. Miers and J.C.F. Telles, Two-dimensional elastostatic analysis of FGMs via BEFM, Engineering Analysis with Boundary Elements, Vol.32, pp.1006-1011, 2008. [9] Zheng Zhong and Tao Yu, Analytical solution of a cantilever functionally graded beam, Composite Science and Technology, Vol.67, pp.481-488, 2007. [10] Mansfield, E.H., The Bending and Stretching of Plates, Pergamon Press, pp.64-73, 1964.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
131
Moment curvature analysis of concrete flexural members confined with CFRP grids A. Michael & P. Christou Department of Civil Engineering, Frederick University, Cyprus
Abstract The moment-curvature (M-Φ) diagrams define the maximum capacity of structural elements that are primarily subjected to bending moments and therefore their failure mode is flexural. The moment-curvature diagrams are also used to assess the ductility of structural elements and are therefore very important for the determination of the amount of plastic energy a structural element can absorb. One way of improving the reinforced concrete (RC) members is the confinement of concrete with FRP composites. The work presented in this paper includes the development of moment curvature diagrams for RC members using a fiber model. The RC members were designed as compression controlled members meaning that their failure initiates in the concrete prior to yielding of the steel tension reinforcement. These types of members have limited ductility. The introduction of a specific amount of Carbon Fibre Reinforced Polymer (CFRP) composite grid as confining reinforcement improves the ductility of the RC members by as much as 30%. The results from the fiber model are compared to available results from an experimental program conducted to evaluate the experimental improvement of the ductility of compression controlled members. The experimental and analytical results are a good match indicating that the fiber model is accurate and can be used to develop the moment-curvature diagrams of RC members confined with a CFRP composite grid. Keywords: CFRP grid, concrete confinement, moment curvature, ductility.
1 Introduction Compression controlled steel reinforced members, fail through crushing of concrete and therefore lack the ductility that is associated with yielding of the WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line) doi:10.2495/MC110121
132 Materials Characterisation V reinforcement. This is due to the fact that concrete as a material does not behave in a ductile manner. Concrete may exhibit limited ductility (also called pseudo ductility) but this ductility varies depending on the aggregate material and the strength of concrete used. High strength concrete is more brittle than low strength concrete and therefore for high strength concrete the pseudo ductility diminishes. This is similar to concrete members reinforced with Fibre Reinforced Polymer (FRP) composites. FRP reinforcing bars when loaded in tension, exhibit linear stress-strain behaviour up to rupture. There is no yield point and associated plateau to provide a ductile response when used as tensile reinforcement in concrete. Rooney and Taylor [1], Toutanji and Deng [2], and Grace et al. [3] found that the post cracking beam behaviour was linear to failure in concrete beams reinforced with glass FRP rods which shows the lack of a yield plateau and therefore the lack of a ductile response. There have been attempts to improve the ductility of FRP with little or no success. If these members are designed in such a way in order to fail through concrete crushing limited ductility may be exhibited much like in the case of compression controlled members. Therefore, the focus of improving member ductility should be on the concrete for compression controlled members. If the concrete in the compression zone of a flexure dominated member is confined then degradation of the compression zone at capacity is delayed, resulting in a more ductile response. Carbon Fibre Reinforced Polymer (CFRP) composite grids have been used to reinforce concrete decks and beams for both strength and crack control [4–8]. The CFRP composite grid has been used as confinement reinforcement with promising results. In a series of cylinder tests conducted by Michael et al. [9] it was found that the crushing strain of CFRP composite grid confined concrete was more than 2 times higher than the crushing strain of unconfined concrete. The advantage that the CFRP composite grid has over the use of FRP composite wraps that have been used for both confinement of concrete or strengthening of concrete members is the ability of the CFRP composite grid to be embedded in the concrete. The CFRP composite grid has openings that allow concrete to flow through thus can be embedded in the member during construction. This also provides environmental protection to the CFRP composite since the composite is not exposed to the natural elements that can cause environmental degradation to the FRP composite. Moment-curvature diagrams are used to assess the ductility of structural elements and are therefore very important for the determination of the amount of plastic energy a structural element can absorb. The research presented in this paper evaluated the moment-curvature response of compression controlled RC members confined with CFRP composite grid with the use of a fiber model that plots the moment-curvature diagram of such sections.
2 Experimental investigation The unique application of the CFRP composite grid as confining reinforcement was evaluated with the construction of two beams. Beam 1 had no CFRP composite grid tubes and served as the control beam and beam 2 had two CFRP WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
133
composite grid tubes placed in the compression zone. Details of the dimensions, amount and location of steel reinforcement and CFRP composite grid tubes are provided in Figures 1 and 2. The CFRP composite grid was formed into a circular tube using a 152 mm diameter plastic pipe and was held in place by a thin rope wrapped around the tube along its length. 5 in (127 mm)
3 in (76 mm)
2.75 in (70 mm) 18 in (457 mm) 6.65 ft (2.03 m)
2.7 ft (0.81 m)
6.65 ft (2.03 m)
16 ft (4.87 m) 1 in (25 mm)
1 in (25 mm)
#3 (#10) Stirrups
18 in (457 mm) 8 #10 (#32) Bars 2.75 in (70 mm)
2 in (51 mm)
3 in (76 mm) 12 in (305 mm)
Figure 1: 5 in (127 mm)
Details of control beam. 13 ft (3.96 m)
2.75 in (70 mm) 18 in (457 mm) 6.65 ft (2.03 m)
2.7 ft (0.81 m)
6.65 ft (2.03 m)
16 ft (4.87 m) 0.75 in (19 mm)
1 in (25 mm) 1 in (25 mm)
6 in (152 mm)
#3 (#10) Stirrups
18 in (457 mm) 8 #10 (#32) Bars 2.75 in (70 mm)
2 in (51 mm)
3 in (76 mm) 12 in (305 mm)
Figure 2:
Details of grid beam.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
134 Materials Characterisation V The beams were tested in a simply supported four-point bending configuration (see Figure 3) in displacement control mode, that is, a constant displacement rate was applied independently of the amount of load. The span length between the two supports was 4.27 m, whereas the total length of the beams was 4.87 m. Actuator 1 Load Cell
Spreader Beam D15
D14
D13
D12
D11
D10
D9
D8
D7
D6
D5
D4
D3
D2
D1
18 in (457 mm)
1 ft (305 mm) Typ.
North Support
South Support 5.5 ft (1.7 m)
LVDT
3 ft (0.91 m)
Tilt Meter
Figure 3:
5.5 ft (1.7 m)
60 mm Foil Gauge
Test set-up.
3 Moment curvature analysis The moment-curvature (M-Φ) diagrams of the beams were calculated based on experimental data and also based on a theoretical fiber model. The experimental moment-curvature diagrams were calculated based on the displacement profile curves at increasing loads. The displacement profile equation was generated by fitting a 3rd order polynomial line to the displacement profile curves. The fitted lines were in good agreement with the displacement data (average R2 = 0.99). The polynomial displacement profile equation of the beam was used to determine the curvature at specific points along the length of the beam. The second derivative of the polynomial displacement profile equation is the curvature equation. The M-Φ curves of the beams were calculated using a fiber model. The compression zone of the tube beam was separated into two regions with regard to concrete properties. The first region was the concrete cover on top of the CFRP grid tubes and the second region was from the neutral axis to the top of the CFRP grid tubes (see Figure 4). The first region was extended approximately 0.25 in. (6.5 mm) below the top of the CFRP grid tubes to account for the unconfined concrete at the sides and between the grid tubes. The first region was assigned unconfined concrete properties and the second region confined properties. The concrete area from the neutral axis to the bottom of the grid tubes was unconfined but was treated as confined concrete for simplicity and because WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
135
the strains in that area close to the neutral axis were small and therefore the effect on the strength was minor. The unconfined region was divided into one rectangular layer while the confined region was divided into eight rectangular layers. For the control beam, the compression zone was treated as an unconfined concrete region and was divided into eight layers. 1 in. (25 mm)
c - 25 mm (c - 1 in.)
0.75 in. (19 mm)
Grid Tube N. A.
Unconfined concrete Confined concrete
Figure 4:
Tube beam cross-section fiber model.
Some of the assumptions employed in this fiber model include perfect bond between concrete and the reinforcing bars (strain compatibility), plain sections remain plain (linear strain distribution), the area below the neutral axis was considered cracked and was ignored in the force and moment calculation and monotonic loading and deformation of the section. The compressive force in each layer was calculated as the product of the area of the layer and the average stress. The average stress was determined based on the strain in the layer and the stress-strain curves of concrete. The Hognestad parabola was used to calculate the stress-strain curve of unconfined concrete. The modified Hognestad was used to calculate the stress-strain curve of confined concrete [10]. The parameters of the modified Hognestad were determined based on the available data from confined concrete cylinder tests [9]. The tensile force was calculated as the product of the area of steel reinforcement and the steel stress. Steel stress was determined based on the average strain of the two steel layers and by assuming elastic perfectly plastic stress-strain curve. The moment in the cross-section for the applied curvature distribution was determined by summing moments about the neutral axis. The moment curvature diagram was calculated by applying increasingly larger top concrete compressive strains (εc) to the cross-section and varying the location of the neutral axis (c) until equilibrium of the axial forces (zero) in the WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
136 Materials Characterisation V cross-section for the applied strain was achieved. The curvature of the section (Φ = εc / c) was then calculated based on the concrete compressive strain and the location of the neutral axis. The experimental and fiber model M-Φ diagrams were calculated for both beams. The theoretical and fiber model M-Φ diagrams for the control beam (see Figure 5) were approximately the same. The fiber model M-Φ behaviour of the tube beam was comparable to the experimental behaviour (see Figure 6). The improvement in the ductility of the Curvature x 1000 (rads/m) 16 24 31
39
500
678
400
542
300
407
200
271 Experimental Fiber Model
100
136
0
0 0
0.2
0.4 0.6 0.8 Curvature x 1000 (rads/in)
Figure 5:
0
Moment (kip-ft)
Moment (kN-m)
8
8
1
M-Φ curves for control beam.
Curvature x 1000 (rads/m) 16 24 31
39
500
678
400
542
300
407
200
271
Expiremental Fiber Model
100
136
0
0
0
0.2 0.4 0.6 0.8 Curvature x 1000 (rads/in)
Figure 6:
M-Φ curves for tube beam.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
1
Moment (kN-m)
Moment (kip-ft)
0
Materials Characterisation V
137
tube beam compared to the control beam, although relatively small, was evident when the experimental M-Φ curves for both beams were compared (see Figure 7). The area under the load-displacement curve represents the amount of energy a structure can absorb before failure. The amount of energy is also a good indicator of ductility. Ductile structures usually can absorb higher amounts of energy compared to non-ductile (brittle) structures.
8
Curvature x 1000 (rads/m) 16 24 31
39
500
678
400
542
300
407
200
271
Control Grid
100
Moment (kN-m)
Moment (kip-ft)
0
136
0
0
0
0.2 0.4 0.6 0.8 Curvature x 1000 (rads/in) Figure 7:
1
Experimental M-Φ curves.
The areas under the experimental moment-curvature curves, of both beams, were calculated and compared. The area under the load-displacement curve of the tube beam was approximately 37% more than that of the control beam. The energy up to peak load was elastic energy while the post peak energy was inelastic. Therefore, the 37% extra energy was primarily inelastic energy which indicated the ability of the tube beam to undergo inelastic deformation compared to the control beam. This represents an improvement in the ductility of the member and is due to the confining effect of the CFRP composite grid. Another way to measure ductility is through ductility factors. Curvature ductility factors are a good way to evaluate the ductility of RC members. The curvature ductility factor (μΦ) is defined as the ratio of the ultimate curvature (Φu) over the curvature at first yield (Φy). The definition of ultimate and yield curvatures is simple when the behaviour is elasto-plastic. Definition of these parameters is more complicated when the behaviour is not elasto-plastic. Usually the behaviour of reinforced concrete is not perfectly elasto-plastic. Therefore, the need for a consistent definition of the ultimate and yield curvatures was realized. Researchers have proposed definitions of ultimate and yield displacement and curvature as well as instructions on how to calculate them. WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
138 Materials Characterisation V The ultimate curvature has been defined as the curvature at which the lateral load in the descending portion of the load-displacement curve is not less than 80% of the maximum load (Pu) [11–14]. Park and Paulay [15] have argued that the available ultimate deformation (and therefore curvature) is not necessarily the deformation that corresponds to the maximum load capacity. They further argued that “when survival without collapse is the criterion, it is too conservative to define ultimate deformation as the deformation corresponding to the maximum load-carrying capacity. It would seem reasonable to recognize at least some of this deformation capacity after the maximum load has been reached and to define the available ultimate deformation as that deformation when the loadcarrying capacity has reduced by some arbitrary amount after maximum load. For example, a 10 or 20% reduction in maximum load-carrying capacity could be tolerated in many cases, but the exact amount would depend on the particular case.” The yield displacement (and therefore curvature) can be defined as the displacement at the intersection of the horizontal line representing the ideal lateral capacity, Pi, (nominal capacity using the ACI 318 approach and a reduction factor of unity) and the straight line that passes through zero and the point in the load-displacement curve at 75% of the ideal lateral capacity [11–13]. The definitions of yield and ultimate displacements are depicted in Figure 8. P Pu Pi 0.8 Pu 0.75 Pi
0
Figure 8:
d y or d c
y
du
u
d
Ductility factor component definitions.
The curvature ductility (μΦ) was approximately 1.5 and 2 for the control and tube beam respectively. The schematic for determining Φu and Φc for the control beam is shown in Figure 9. The same approach was used in determining the ductility factor for the tube beam. The curvature ductility factor of the tube beam was approximately 33% higher than the curvature ductility factor of the control beam. WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
0
8
Curvature x 1000 (rads/m) 16 24 31
39 678
400
Mu Mi
542
300
0.8 Mu 0.75 Mi
407
200
271
100
136 c
0 0
Figure 9:
0.2
u
0.4 0.6 0.8 Curvature x 1000 (rads/in)
Moment (kN-m)
500
Moment (kip-ft)
139
0 1
Curvature ductility factor for control beam.
The curvature ductility factor using the fiber model moment-curvature diagram of the control beam was approximately 1.6 which was approximately 4% higher than the experimental. On the other hand, the curvature ductility factor of the tube beam was approximately 2.45 which was an overestimate of approximately 21%.
4 Conclusions In this paper the development of moment-curvature diagrams for compression controlled RC sections confined with CFRP composite grids was presented. These diagrams were compared with diagrams developed from experimental data for the same sections. Based on the results from this comparison the following conclusions can be drawn: 1. The results from the two beams indicate that the CFRP grid provided confinement to the concrete. Confinement was achieved using a series of small diameter CFRP grid tubes rather than wrapping the entire cross-section. 2. The curvature ductility factor of the beam utilizing the CFRP grid tubes was improved by 33% compared to the control beam with no grid tubes. The improvement was significant, considering the small amount of CFRP that was used to confine the compression zone. 3. The energy dissipation of the tube beam was 37% higher compared to the controlled beam due to the effect of concrete confinement. This increase in the energy dissipation of the element also represents an improvement in the ductility due to concrete confinement from the CFRP grid tubes. WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
140 Materials Characterisation V 4.
5.
The results from the fiber model are fairly accurate compared to the experimental data and therefore the model is considered valid. However, there is a divergence after the peak load and further refinement of the model is required. The number of test specimens is limited and therefore the more testing is required. Due to the small number of test specimens the conclusions are drawn with great caution.
References [1] Rooney, J., and Taylor, S., 2004, “Flexural Behaviour of Steel and GFRP Reinforced Concrete Beams,” Composites: Part B, Vol. 27, No 3-4, pp. 38-40. [2] Toutanji, H., and Deng Y., 2003, “Deflection and Crack Width Prediction of Concrete Beams Reinforced with Glass FRP Rods,” Construction and Building Materials, Vol. 17, No 1, pp. 69-74. [3] Grace, N. F., Soliman, A. K., Sayed, G. A., and Saleh, K. R., 1998, “Behavior and Ductility of Simple and Continuous FRP Reinforced Beams,” Journal of Composites for Construction, Vol. 2, No 4, pp. 186-194. [4] Rahman, A. H., Kingsley, C. Y., and Kobayashi, K., 2000, “Service and Ultimate Load Behavior of Bridge Deck Reinforced with Carbon FRP Grid,” Journal of Composites for Construction, Vol. 4, No 1, pp. 16-23. [5] Yost, J. R., Goodspeed, C. H., and Schmeckpeper, E. R., 2001, “Flexural Performance of Concrete Beams Reinforced with FRP Grids,” Journal of Composites for Construction, Vol. 5, No 1, pp. 18-25. [6] Harries, K. A., and Gassman, S. L., 2003, “Load Tests of Reinforced Concrete Catch Basing Knockout Panels,” Department of Civil and Environmental Engineering, University of South Carolina, Report No ST03-01, p. 21. [7] Shao, Y., Johnson, C., and Mirmiran, A., 2003, “Control of Plastic Shrinkage Cracking of Concrete with TechFab Carbon FRP Grids,” Department of Civil, Construction, and Environmental Engineering, North Carolina State University, Report to Tech-Fab Inc, p. [8] Hamilton, H. R, Cook, R. A., and Alfonzo, L., 2006, “Crack Control in Toppings for Precast Flat Slab Bridge Deck Construction,” Department of Civil and Coastal Engineering, University of Florida, Engineering and Industrial Station, Final Report Submitted to FDOT, UF Project No 00030907, p. 120. [9] Michael, A. P., Hamilton, H. R. III, and Ansley, M. H. (2005). “Concrete confinement using carbon fiber reinforced polymer grid”, American Concrete Institute, SP230-56, Vol. 230, pp. 991-1010. [10] Park, R., and Paulay, T., 1975 (a), “Chapter 6: Ultimate Deformation and Ductility of Members with Flexure,” Reinforced Concrete Structures, John Wiley & Sons, New York, NY, pp. 195-269.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
141
[11] Priestley, M. J. N., and Park, R., 1987, “Strength and Ductility of Concrete Bridge Columns under Seismic Loading,” ACI Structural Journal, Vol. 84, No 1, pp. 61-76. [12] Zahn, F. A., Park, R., and Priestley, M. J. N., 1990, “Flexural Strength and Ductility of Circular Hollow Reinforced Concrete Columns without Confinement on Inside Face,” ACI Structural Journal, Vol. 87, No 2, pp. 156-166. [13] Sheikh, S. A., and Khoury, S. S., 1993, “Confined Concrete Columns with Stubs,” ACI Structural Journal, Vol. 90, No 4, pp. 414-431. [14] Yeh, Y. K., Mo, Y. L., and Yang, C. Y., 2002, “Seismic Performance of Rectangular Hollow Bridge Columns,” Journal of Structural Engineering, Vol. 128, No 1, pp. 60-68. [15] Park, R., and Paulay, T., 1975 (b), “Chapter 11: Strength and Ductility of Frames,” Reinforced Concrete Structures, John Wiley & Sons, New York, NY, pp. 496-609.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
This page intentionally left blank
Materials Characterisation V
143
Application of effective media theory in the characterization of the hygrothermal performance of masonry Z. Pavlík, E. Vejmelková, L. Fiala, M. Pavlíková & R. Černý Department of Materials Engineering and Chemistry, Faculty of Civil Engineering, Czech Technical University in Prague, Czech Republic
Abstract Hygric and thermal properties of several types of materials which are used in reconstructions of historical stone masonry on Czech territory are experimentally measured at first. The obtained data are then subjected to a homogenization procedure, and the effective heat and moisture transport and storage parameters of the typical fragment of masonry are calculated. For the homogenization, a mixing model originally derived for applications in dielectric studies was employed taking into account the theoretical bounds of the investigated material parameters. The obtained data can find use in damage assessment of historical masonry using methods of computational simulations and analysis. Keywords: hygric properties, thermal properties, historical stone masonry, argillite, sandstone, lime-based plaster, homogenization techniques.
1 Introduction The hygrothermal behaviour of climatically exposed components and structures of historical buildings is related to the hygric, thermal, mechanical and other physical and chemical properties of inbuilt materials. Their knowledge represents necessary information for the understanding of the building performance as well as the first step in avoiding damage, degradation or the undue heat loss from the constructions. The damage assessment of historical masonry and other structures due to the negative effects of moisture and temperature can be done effectively by means of mathematical and computational modelling. In this way, the time development of WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line) doi:10.2495/MC110131
144 Materials Characterisation V moisture and temperature fields can be obtained which is a crucial for a proper assessment of possible future damage [1]. Their prediction is a very important task when preserving historical bridges and buildings or insulating existing buildings and components. The moisture and temperature values can be then assigned to the mechanical properties and to the risk of consequent damage. For instance Charles Bridge in Prague, Czech Republic, which was subjected to reconstruction works lately, is a typical example of extensive damage mostly brought about by temperature and moisture impacts. For the effective application of computational modelling in the buildings’ performance analysis, there is necessary complete knowledge of material properties of applied materials as well as of initial and boundary conditions of simulations. As the initial conditions for the hygric and thermal analysis, the moisture and temperature fields experimentally measured in the investigated structure can be used. Boundary conditions of the computer simulations are given by the climatic loading that can be simply adopted using meteorological data. In the technical practice, simplified calculations of building structures’ behaviour are very often used. Here, in the computational simulations of heat and moisture transport processes, brick or stone masonry is often understood as a single material, the brick or stone itself. This simplification may work reasonably well when the properties of the brick or stone and the mortar are similar. However, this may not always be the case in some historical buildings, where low-quality mortars were often used [2]. On that account, the results of such calculations are diverse from the real state of the buildings, and cannot be effectively used for estimation of some appropriate materials’ damage and problems in hygrothermal behaviour. Therefore, solution of this problem must be found in order to improve the accuracy and reliability of computational modelling of building structures performance. There are two ways how to meet the above given requirements on reliability and accuracy of computational simulation of heat and moisture transport. The first possibility is the use of sophisticated simulation tools that allow discretization of studied structure in such resolution, that all the properties of inbuilt materials can be assigned to the specific places of the structure. Although this procedure is generally correct, it is too complex and sophisticated for wider application in engineering practice, especially in case of conservationists of historical buildings that are usually well oriented in materials’ problems without knowledge of methods of computational modelling. The second, and probably the most straightforward way to solve the problem of differences in the properties of the particular components of masonry is the utilization of the effective media theory. This procedure we introduce in the presented work as an effective tool for simplification of computational modelling of transport processes in the masonry. Application of homogenization principles produces the macroscopic equations which may be used when analyzing the masonry as a whole. In these equations, the effective parameters are used instead of the parameters of the brick or stone.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
145
Quite apparently, the homogenization process cannot be done without the exact knowledge of the properties of all materials constituting the masonry and of the amount of each material in the analyzed wall. Therefore, the experimental measurement of all the material parameters characterising the process of heat and moisture transport and accumulation in inbuilt masonry materials is necessary at first. These data are than implemented in the mixing models based on effective media theory, and the effective thermal and hygric parameters of the whole masonry can be calculated.
2 Studied materials Two types of stones that were in the past often used in Czech region as bearing masonry materials are analysed together with lime-based mortar with pozzolana admixture. Many historical buildings in the Czech Republic were built using similar kinds of sandstone. Siliceous raw-grained sandstone was usually used for historical architectural constructions (walls, portals, window frames) for its strength. Ornamental parts of the architecture (gothic flowers, romantic shells) and sculptures (from the Romanesque period up to now) were made of finegrained calcite-argillaceous sandstone. In this work, sandstone from Mšené-lázně quarry, Czech Republic, is chosen. It is fine-grained psamitic equigranular rock, about 95% of which is made up of suboval quartz clasts. Other mineral grains are present only as minorities (tourmaline, epidote, muscovite and zircone). Quartz grains reach up to 0.1 mm in diameter, but those of muscovite are larger, up to 0.3 mm [3]. The matrix is formed by clay minerals (mainly kaolinite). Also the argillite was very popular material in historical architecture. It was used for sacral as well as for secural buildings, flagstone pavements, roof slabs, and facing. The studied argillite is coming from quarry Džbán, Czech Republic. Its main constituents are illite, calcite, minerals on the basis of SiO2 having granularity 0.3 – 0.15 mm, feldspar, and mica, whereas rigid materials form 40 – 60% of argillite volume. Within the reconstruction of historical buildings, renewal and restoration of interior and exterior plasters50and mmortars is often done. From the point of view of a historian, it is not acceptable to use lime-cement mortars in Romanesque, Gothic, Renaissance, and Baroque buildings. However, the pure lime-based plaster does not exhibit sufficient resistivity against moisture action. On that account, proper hydraulic admixtures must be used that enhance the durability of mortars. These materials must have similar composition as the historical materials and they have to be applicable by the original technological processes. As the chemical analyses of many plasters from historical buildings show, the past centuries external plasters that are preserved until today contain products formed by lime reaction with pozzolanic or hydraulic admixtures. Pozzolanic admixtures appeared to have positive effect on properties of lime binder in the past. On that account, lime-based mortar with metakaolin addition as pozzolana admixture was chosen. The composition of the lime-metakaolin plaster was as
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
146 Materials Characterisation V follows: hydrated lime – 400 g, metakaolin - 80 g, natural quartz sand with continuous granulometry 0 to 2 mm – 1 440 g, and water – 480 g.
3 Experimental methods Basic material properties of all tested materials were determined at first. Bulk density and matrix density were measured using gravimetric method and helium pycnometry, and then total open porosity was calculated. The samples’ dimensions for these measurements were 40 x 40 x 20 mm. For determination of water vapour transmission properties we applied cup method, according to the European standard EN ISO 12571 [5]. The measurements were carried out in steady state under isothermal conditions. We used dry cup arrangement of the experiment. Here, the sealed cup with the studied material sample containing burnt CaCl2 (0% relative humidity) was placed in a controlled climatic chamber at 25±0.5°C and 50% relative humidity and it was weighed periodically. The circular samples had diameter 95 mm and thickness of 20 mm. The steady state values of mass gain were utilized for the determination of the water vapour permeability. On the basis of measured water vapour permeability, the water vapour diffusion coefficient and water vapour resistance factor were calculated using the simple formulas given in [6, 7]. For determination of moisture diffusivity as function of moisture, the moisture profiles were measured at first. The measurements were done on samples having dimensions of 40 x 20 x 300 mm in 1-D experimental arrangement of water transport. The moisture content in specific position in sample was measured by capacitance technique calibrated by gravimetric method. The moisture dependent moisture diffusivity was then calculated using inverse analysis of measured moisture profiles by means of Boltzmann-Matano treatment [8]. In the sorption isotherm measurement, the samples were placed into the desiccators with different salt solutions to simulate different values of relative humidity. The mass of samples was measured in specified periods of time until steady state value of mass was achieved. Then, the volumetric moisture content was calculated and sorption isotherm of each tested material was plotted. The thermal conductivity as the main parameter of heat transport was determined using the commercial device ISOMET 2104 (Applied Precision, Ltd.). ISOMET 2104 is a multifunctional instrument for measuring thermophysical parameters which is based on the application of an impulse technique and is equipped with various types of optional probes. Thermal conductivity was measured in the moisture range from the dry state to full water saturation on the 70 mm cubes.
4 Effective media theory and homogenization techniques Application of effective media theory allows determination of the properties of the whole masonry or building structure instead of parameters of the particular inbuilt materials. In terms of effective media theory, the final composite structure WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
147
(in our case stone masonry wall) can be considered basically as a mixture of walling blocks and mortar. In more precise calculations, each of these materials can be further considered as a mixture of three phases, namely solid, liquid and gaseous phase (in four phase systems, the effect of bound water can be included) that form their matrix and porous space. There are two basic approaches that can be applied for determination of thermal and hygric properties of the stone masonry. The first possibility is to apply homogenization techniques on the moisture dependent material data of the materials involved in the masonry. This simplified procedure was used in this work. The second possibility is based on the complete knowledge of material properties of the particular components forming the porous body of the structure (dry stone, dry mortar, free water, bound water and air). From the properties of particular components and their volumetric fractions the effective properties of the stone masonry can be accessed. On the basis of previous experience with application of homogenization techniques, we have chosen for the calculations performed in the presented work original Lichtenecker’s formula that was proved to be satisfactory for evaluation of moisture dependent thermal and hygric properties of porous building materials [9]. Also the application of four phase models looks promising. However, in case of the masonry studied in this work, determination of amount of bound water represents very complex problem. The Lichtenecker’s equation adjusted for the studied masonry wall assumes that the effective hygric and thermal parameter of the considered material satisfies equation
peff f b pb f m pm , k
k
k
(1)
where peff is calculated effective parameter of masonry, fb volumetric fraction of walling blocks in masonry, fm volumetric fraction of mortar, pb measured material parameter of walling block, pm measured parameter of mortar, and k is free parameter describing basically the path from the anisotropy at k = -1.0 to another anisotropy at k = 1.0. However, the Lichtenecker’s equation may be also applied for isotropic materials. On the basis of previous experience, we have used the value of the parameter k = 0. The effective parameters of multi-phase material cannot exceed the bounds given by the parameters of particular fractions of its constituents. Here, the Wiener bounds according to the Wiener’s original work were used [9]. These bounds can be expressed by the following relations
peff
1 n
i 1
fi pi
,
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
(2)
148 Materials Characterisation V n
peff f i pi ,
(3)
i 1
where Eq. (2) represents the lower limit and (3) the upper limit of the investigated effective material parameter (fi is the volumetric fraction of the particular phase, in our case argillite, sandstone and mortar), pi its material parameter).
5 Studied masonry For application of homogenization theory for the evaluation of the effective hygric and thermal parameters of masonry, the typical fragment of the wall was constructed in a simplified way similar to common brick masonry as shown in Figure 1.
Figure 1:
Scheme of the reference masonry wall (dimensions in mm).
Using this scheme, the volumetric fractions of the particular materials in the wall were calculated. The volumetric fraction of walling materials is equal to 0.824, whereas the volumetric participation of mortar is 0.176. Within the calculations, effective parameters of two different walls were accessed. The studied walls consisted always of lime based metakaolin mortar, and of one walling material, nominally sandstone or argillite.
6 Results and discussion Basic properties of all tested materials are summarized in Table 1. Each result represents the average of five measured values. All the studied materials have WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
149
proved high porosity, what is very positive factor from the point of view of their presumed application in reconstruction of historical buildings. Water vapour transmission properties of masonry materials are presented in Table 2. We can see systematically very low values of water vapour resistance factor what is again very prospective finding for application of all tested materials in historical masonry. Since the historical masonry usually exhibits increased moisture content, there is necessary to apply within the reconstruction processes such materials that will allow moisture evaporation from the renovated structures within the warm periods of the year. Table 1: material sandstone argillite lime-metakaolin mortar Table 2:
Basic material properties of masonry materials. bulk density (kg/m3) 1807 1353 1690
matrix density (kg/m3) 2627 2235 2620
total open porosity (-) 0.31 0.39 0.35
Water vapour transmission properties of masonry materials.
material
sandstone argillite lime-metakaolin mortar
dry cup arrangement, 0 – 50% water vapour water vapour water vapour permeability (s) diffusion resistance factor coefficient (m2/s) (-) 2.4E-11 3.3E-06 7.0 2.9E-11 4.1E-06 5.7 1.8E-11 2.5E-6 10.1
In Table 3, there are presented the effective diffusion parameters of studied stone masonry calculated on homogenization principles. For verification of calculated results, also the limiting bounds of effective parameters are presented. Looking at the obtained data of water vapour transmission properties, one can see that all the effective parameters lie between the Wiener’s bounds which basically verifies their correctness. Effective moisture diffusivity as function of moisture content is given in Figures 2, 3. We can see high dependence of this moisture transport parameter on moisture content. This materials’ behaviour significantly affects the hygrothermal performance and consequently the durability. Systematically, the highest moisture diffusivity exhibits sandstone. Its highly porous structure formed by high radius pores allows fast liquid moisture transport in comparison with argillite that is characteristic also by high total open porosity, but its pore size is much smaller and structure more fine-grained.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
150 Materials Characterisation V Table 3:
Water vapour transmission properties of stone masonry.
Sandstone wall – homogenization, dry cup arrangement – 0-50% RH water vapour permeability (s)
water vapour water vapour diffusion resistance coefficient (m2/s) factor (-)
Wiener’s lower bound
2.29E-11
3.11E-06
7.40
Wiener’s upper bound
2.24E-11
3.05E-06
7.55
Lichtenecker model, k=0
2.27E-11 3.08E-06 7.47 Argillite wall – homogenization, wet cup arrangement – 0-50% RH water vapour permeability (s)
water vapour water vapour diffusion resistance coefficient (m2/s) factor (-)
Wiener’s lower bound
2.74E-11
3.73E-06
6.17
Wiener’s upper bound
2.61E-11
3.55E-06
6.47
Lichtenecker model, k=0
2.69E-11
3.65E-06
6.30
Figure 2:
Effective moisture diffusivity of sandstone masonry.
The calculated effective sorption isotherms of studied masonry are given in Figures 4 and 5. Typically, the highest capacity for water vapour adsorption can be observed for lime-metakaolin mortar. However, in hygroscopic moisture range, the highest values of adsorbed moisture were measured for argillite. The data obtained for sandstone were systematically the lowest. WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
Figure 3:
Effective moisture diffusivity of argillite masonry.
Figure 4:
Effective sorption isotherm of sandstone masonry.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
151
152 Materials Characterisation V
Figure 5:
Effective sorption isotherm of argillite masonry.
Measured and calculated thermal conductivity of all studied materials and two types of masonry is presented in Figures 6 and 7. The values of effective thermal conductivity in dependence on moisture content are systematically between the range of Wiener’s bounds, which basically proves reliability of applied homogenization technique.
Figure 6:
Effective thermal conductivity of sandstone wall.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
Figure 7:
153
Effective thermal conductivity of argillite wall.
7 Conclusions An example of application of homogenization technique for determination of hygric and thermal parameters of the stone masonry was introduced in the paper. The obtained results indicate that the Lichtenecker’s equation may be successfully used for such type of applications. Nevertheless, there is still an open task to verify the reliability of applied homogenization model by laboratory experiments that should confirm or controvert the presented results.
Acknowledgement This research has been supported by the Czech Ministry of Education, Youth and Sports, under project No MSM 6840770031.
References [1] Pavlíková, M., Pavlík, Z. & Černý, R., Hygric and Thermal Properties of Materials Used in Historical Masonry. Proc. of the 8th Symposium on Building Physics in the Nordic Countries, Technical University of Denmark: Lyngby, pp. 903-910, 2008. [2] Vejmelková, E., Pavlík, Z., Fiala, L., Pavlíková, M. & Černý, R., Heat and Moisture Transport Properties of Stone Masonry Materials. Proc. of the Building Physics Symposium, Katolieke Universiteit Leuven, Leuven, pp. 113-116, 2008.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
154 Materials Characterisation V [3] Pavlík, Z., Michálek, P., Pavlíková, M., Kopecká, I., Maxová, I. & Černý, R., Water and Salt Transport and Storage Properties of Mšené Sandstone. Construction and Building Materials, 22(22), pp. 1736-1748, 2008. [4] Pavlík, Z., Identification of parameters describing the coupled moisture and salt transport in porous building materials, CTU Prague, p. 145, 2009. [5] EN ISO 12572, Hygrothermal performance of building materials and products, determination of water vapour transmission properties, the European Committee for Standardization, Brussels, 2001. [6] JIřičková M., Application of TDR Microprobes, Minitensiometry and Minihygrometry to the Determination of Moisture Transport and Moisture Storage Parameters of Building Materials, CTU Prague, p. 102, 2004. [7] Roels S., Carmeliet, J., Hens, H., Adan, O., Brocken, H., Černý R., Pavlík Z., Hall Ch., Kumaran K., Pel L. & Plagge R., Interlaboratory Comparison of Hygric Properties of Porous Building Materials. Journal of Thermal Envelope & Building Science, 27(4), pp. 307-325, 2004. [8] Roels, S., Carmeliet, J., Hens, H., Adan, O., Brocken, H., Černý, R., Pavlík, Z., Hall, C., Kumaran, K., Pel, L., A Comparison of Different Techniques to Quantify Moisture Content Profiles in Porous Building Materials. Journal of Thermal Envelope & Building Science, 27(4), pp. 261-276, 2004. [9] Pavlík, Z., Vejmelková, E., Fiala, L., Černý, R., Effect of Moisture on Thermal Conductivity of Lime-Based Composites. International Journal of Thermophysics, 30(6), pp. 1999-2014, 2009.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
155
3D FIB reconstruction and characterisation of a SOFC electrode S. Chupin, N. Vivet, D. Rochais & E. Bruneton CEA-Le Ripault, France
Abstract SOFC (Solid Oxide Fuel Cells) appear to be a great alternative way to produce electricity from hydrogen with high efficiency and no greenhouse gas emissions. SOFC are efficient at high temperatures (around 800°C) and are meant to be used for stationary applications as heat and electric co-generation devices. To understand how gases, electricity and heat flow through these media and to improve their efficiency, it is critical to know the actual microstructure of these electrodes. The studied electrodes are Ni-YSZ cermets in which characteristic element sizes are around 1m. The 3D microstructure has been reconstructed using FIB (Focused Ion Beam) tomography. This technique has been used on several samples (different Ni-YSZ proportions) and gives representative 3D volumes of around 10×10×10µm with a 10nm resolution. Theses 3D volumes are then analysed to extract some important structural parameters such as volumetric proportions, active surfaces, connectivity of each components and “three phase boundaries” (TPB). Then, the 3D reconstructed volumes have been used to determine homogeneous media equivalent properties such as thermal, ionic and electric conductivities. These homogeneous equivalent properties are estimated using a hot guarded plate simulation that takes into account each component properties and the 3D structure. Keywords: SOFC, FIB, tomography, thermal conductivity.
1 Introduction Solid oxide fuel cell (SOFC) based-technology is one of the most promising energy conversion systems due to its high efficiency and fuel flexibility [1]. The materials involved in their constitution have become increasingly sophisticated, both in composition and microstructure. One of the most common anode WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line) doi:10.2495/MC110141
156 Materials Characterisation V materials for SOFC is a porous Ni-YSZ cermet. The most important requirements for the anode are catalytic activity, stability, ionic conductivity, compatibility and porosity. For most of these conditions some experimental studies can be investigated to estimate the corresponding parameters. Recently, 3D reconstructions of SOFC electrodes have been performed by X-ray computed tomography (XCT) [2, 3] and mainly by focused ion beam - scanning electron microscopy (FIB-SEM) [4–7]. FIB tomography principle consists in ablating a structure physically by FIB slicing (Ga+ ions) followed by a digital reconstruction since SEM images are collected after each ablation step. The obtained images are then aligned and superimposed in order to generate the volume [8]. In the present study, the optimization approach to obtain a high quality 3D reconstruction of a Ni-YSZ anode using FIB-SEM is described first. Then, from these 3D data, various microstructural parameters, and interfacial parameters are quantified. Moreover, by solving the diffusive transport equation on the analysed volume the effective thermal, electrical and ionic conductivities of the sample have been estimated.
2
Experimental
2.1 Preparation of the anodic cermet The deposition of the analysed NiO-8YSZ anode has been performed by screenprinting onto circular 8mol.% yttria doped-zirconia (8YSZ) supports using ink with optimized composition and viscosity. The anode is a three layers coating on the electrolyte substrate. NiO from J.T. Baker (USA) and 8YSZ from Tosoh (Japan) have been used as raw materials to prepare the NiO-8YSZ cermet at different Ni volume ratio. The preparation has been done to expect the 8YSZ/Ni volume ratio to be 60/40.
Figure 1:
Schematic representation of the experimental method used to obtain the 3D reconstruction of the SOFC anode from FIB tomography.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
157
2.2 FIB procedure The image set has been collected with a FEI dual-beam Strata 400-S FIB system. The procedure to obtain the 3D data from the initial sample is schematically presented in Figure 1. 2.2.1 Preparation of the sample and of the region Of interest (ROI) In order to easily distinguish the pores during SEM observation and to avoid mistakes due to the depth of field in the SEM images, the sample has been infiltrated with an epoxy resin under vacuum. To protect the sample from accidental ion milling and erosion, a representative cross sectional area of the sample has been located by electron imaging and coated with a Pt protective layer of 2-3µm (Figure 2) using an in-situ metallorganic ion source. Above all, Pt deposition associated with sample infiltration and surface polishing were necessary to avoid, or at least strongly reduce, a common artefact so-called “waterfalling” or “curtaining” effect. The FIB has been used to mill wide and deep trenches around the ROI with a maximal aperture current of 21nA, constituting a “U-shaped trench” [9]. A quite high dimension free space around the ROI is essential to prevent re-deposition of the sputtered material during the sectional milling process. If the space around the ROI is too small, images with high concentration of artefact features or uneven brightness are obtained.
Figure 2:
SEM image of the sample before the “milling and imaging” procedure. The stage tilt is kept constant at 52° from the electron beam while the sample surface (xz plane) is perpendicular to the ion beam (y direction). The region of interest (ROI) is displayed in the orange rectangle. The sample is sliced in the z direction. The small rectangle on the Pt deposition (reference mark) is used to measure the interslice (z) and correct from drifting effects.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
158 Materials Characterisation V 2.2.2 Milling and imaging process Since the “milling and imaging” procedure is very time consuming, uncorrected drift of electron beam, stage and sample can greatly affect the interslice real value. For this reason a correction procedure has been used [10]. The SEM and FIB operating conditions for the sequential ion milling and electron imaging have been optimized. The FIB milling has been performed with an ion-beam current of 460 pA at 30 kV. For SEM imaging, a magnification of 12 kX and a Through the Lens Detector (TLD) operating in BSE mode with low scan rate have been used. As the difference between the backscatter coefficients of Ni and 8YSZ are maximum for accelerating voltage smaller than 1kV, a 0.5kV accelerating voltage has given an optimal contrast between Ni (bright), 8YSZ (grey) and pores (dark) (Figure 3(a)). In this case, 115 images have been recorded using a manual procedure along the sample thickness (Figure 3(a)). 2.2.3 Data processing Data processing includes the following steps: i) alignment of the consecutives slices, ii) correction of the dimensions taking into account the tilt of the electron beam, iii) thresholding of the grey levels and labelling of phases, iv) resampling of the data to obtain cubic voxels and v) 3D image generation. Most of the processing steps have been performed using the AVIZO 6.2 software [11].
Figure 3:
Images of:(a) one of the original 115 serial-sections collected with a BSE detector, so that Ni appears brighter than 8YSZ and that the impregnated pores appear dark, (b) final 3D image after alignment, delineate, cropping and adjustment of voxel size operations showing Ni (bright), 8YSZ (grey) and pore (dark) phases.
The absolute dimensions in the x direction are obtained from a calibrated SEM magnification. Because of oblique SEM imaging at an angle of 52°, distances in the y direction have been corrected. The 3D image dimensions is finally 8.66×9.79×11.41µm3 (volume= 967.36µm3 corresponding to about 1 billion voxels).
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
159
3 Quantification results and discussion First, volume fractions of each phase and their corresponding connectivity across the analysed volume have been determined. The interfacial planes surface area between two phases and the three phase boundaries lines length have been evaluated. Then, effective thermal, electronic and ionic conductivities have been estimated by solving numerically the diffusive transport equation. The most part of this quantitative analysis has been performed with home made C programs on a personal computer [dual CPU Intel® Xeon® E5440 @ 2.83GHz, 32 GB RAM]. 3.1 Volume fraction The volume fractions calculated from 3D data have given the following results: 41% for pores, 33% for 8YSZ and 26% for Ni. The solid volume fractions have been estimated to 44% Ni and 56% 8YSZ in the studied anode, which are in good agreement with the expected values (40 vol.% Ni / 60 vol.% 8YSZ). This result provides evidence of the good representativity of the analysed volume. 3.2 Volumetric connectivity In order to check the connectivity of each phase, a home made program based on a Hoshen-Kopelman algorithm has been used [12]. The voxels linked together by a face form a cluster [13]. When two voxels are only linked by a vertex or an edge, they are not considered to be connected. Finally, if a cluster is connected to the six boundary faces of the studied volume it is defined as “percolated” (Figure 4).
Figure 4:
Ni clusters representation showing that the main part (87.4%) of the Ni phase is percolated, i.e. in contact with the six boundary faces of the volume (in yellow).
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
160 Materials Characterisation V This analysis reveals that the three phases are mainly constituted of a very large “percolated” cluster which represents 99.8%, 99.1% and 87.4% of pore, 8YSZ and Ni total volume respectively. This high percolation of all the three phases suggests good transport ability for gases in the pores, as well as for charge carrier in the solid phases. 3.3 Interface properties Electrochemical and catalytic activities of an electrode are generally described by only one parameter: the TPBL which is correlated to the number of regions where the electrochemical reactions is able to take place. However not all of them may be active due to non-contiguous regions (isolated cluster). So, the specific surface and interface areas have been calculated only inside the percolated volumes. In the pie diagram of Figure 5, the specific interface areas are indicated. Thus, about 50% of the total nickel surface is exposed to the porous phase and can be used for surface catalytic reactions with the fuel.
Figure 5:
Pie diagram representing the specific surface areas of pores, 8YSZ and Ni and the specific interface areas between the three neighbouring phases (surface and interface are normalized by volume sample). The active TPBL is 7.4µm.µm-3.
The TPB are the regions where the three phases (pores, 8YSZ, Ni) meet each other. The TPB analysis procedure has been apply first to the initial volume and yields to a TPB density of 11.2µm.µm-3. The “activity” of a TPB is subject to numerous definitions in the literature. The major part of the authors defined a TPB as “active”, if it lies three percolated phases (pore, Ni, 8YSZ) [4, 7]. The “active” TPBL has been calculated and equal to 7.4µm.µm-3. This value is relatively high compared to those reported in the literature. This suggests a high electrochemical performance for the studied electrode. WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
161
3.4 Effective conductivities calculations 3.4.1 Effective diffusion coefficient calculation In the absence of external sources and in the case of the steady state condition, the distribution of (for all conduction or diffusion phenomena) in the material is governed by the diffusive transport equation [14]:
. D 0
(1) D is defined as the diffusion coefficient in the material. Each voxel is considered uniform with a given diffusion coefficient depending of its phase (Dvoxel. Table 1). This equation is solved using an implicit finite difference method based on a conjugate gradient algorithm [14]. Different Dirichlet boundary conditions are fixed on two opposite faces ( 1 on one face and
2
on the opposite). Null Neumann boundary conditions are imposed on
the other faces. Once the distribution within the volume is obtained (Figure 6), the average flux (Φ) over a cross section of surface area S can be calculated:
Φ D voxel θ . n dS
(2)
S Table 1:
Thermal, ionic and electronic conductivities at 1123 K of pores, 8YSZ and Ni data [17–21]. (W.m .K ) ionic (10-3 S.cm-1) electronic (S.cm-1) -1
Figure 6:
-1
Pores (H2) 0.48 0 0
8YSZ 2 14 10-6
Ni 50 10-5 104
Reconstructed 8YSZ-Ni SOFC Anode and the electric field resulting of a 1V potential difference between two opposite faces.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
162 Materials Characterisation V For the equivalent homogeneous material, the average conduction flux Φ can be written [14]:
Φ Deff S
θ1 θ 2
(3)
xi max
The combination of eqns (2) and (3) leads to estimate Deff as:
Deff
θ Dvoxel xi S θ1 θ 2 S xi max
dS
(4)
At 1123K and under one atmosphere the mean free path of H2 is around 900nm, and is bigger than our voxel size [15]. Then the concentration cannot be considered uniform in a voxel and the bulk diffusion assumption is not valid anymore. In the case of electron, ion and heat transport the corresponding mean free path are smaller than 100nm. So, only heat, ion and electron diffusive transport have been calculated in this work. The results are given in Table 2 where the different values of Deff are reported under the form of thermal, electronic and ionic effective conductivities. The effective thermal conductivity appears quite isotropic with a slighter lower value in the z direction, close to 4 W.m-1.K-1. This value has the same order than those generally reported for SOFC anodes [16]. Table 2:
Thermal, electronic, ionic effective conductivities in the x, y and z directions from the calculation of Deff at T= 1123 K. (W.m .K ) ionic (10-3 S.cm-1) electronic (S.cm-1) -1
-1
x 4.23 1.43 281
y 4.54 1.19 252
z 3.27 1.03 66
The effective electronic conductivity appears anisotropic: the electronic value in the z direction is smaller than the ones in the x and y direction. The evolution of the Ni concentration along each axes has shown that the Ni proportion along z in more important than along the other axes. It creates a stricture effect that decreases the effective conductivity. Experimental four-point electrical measurements performed at 1123 K on the anode surface lead to a value of the same range: 81 S.cm-1. Ionic conductivity in the studied anode is quite isotropic and close to 10-3. As a value of 1.4 10-2 had been used for the calculation, the microstructural effect for the loss of conductivity is closed to a factor 10.
4 Conclusion This work has focused on the 3D analysis of the microstructure in a Ni-8YSZ SOFC anode reconstructed by FIB tomography. Sample preparation, milling and WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
163
imaging conditions and data processing have been optimized to obtain a high quality representative 3D microstructure having the following dimension of 8.66×9.79× 11.41µm3. Home made programs have led to the calculation of some microstructural parameters such as volume fraction, connectivity, specific surface and interface areas and TPB length including the proportion of active and non active ones. These calculations have shown that for an anode with an initial 8YSZ/NiO volume ration of 60/40 leading after reduction to volume relative amounts 41% of pores, 33% of 8YSZ and 26% of Ni, all the phases were highly percolated. The active TPB length has been estimated as 7.4µm/µm3 which comparing to the literature results corresponds to an effective anode. By solving the diffusive transport equation with finite difference calculations, the effective thermal, electronic and ionic conductivities could be determined.
References [1] S. Singhal, K. Kendall, High-Temperature Solid Oxide Fuel Cells: Fundamentals, Design and Applications, Elsevier Advanced Technology, Oxford, UK, 2003. [2] P.R. Shearing, J. Gelb, N.P. Brandon, X-ray nano computerised tomography of SOFC electrodes using a focused ion beam samplepreparation technique, J. of the Europ. Ceram. Soc., 30 (2010) 1809-1814. [3] K.N. Grew, A.A. Peracchio, W.K.S. Chiu, Characterization and analysis methods for the examination of the heterogeneous solid oxide fuel cell electrode microstructure, Part 2 : Quantitative measurement of the microstructure and contributions to transport losses, J. Power Sources, 195 (24) (2010) 7943-7958. [4] J.R. Wilson, W. Kobsiriphat, R. Mendoza, H.-Y. Chen, J.M. Hiller, D.J. Miller, K. Thornton, P.W. Voorhees, S.B. Alder, S.A. Barnett, Threedimensional reconstruction of a solid-oxide fuel-cell anode, Nature Materials, 5 (2006) 541-544. [5] S.A.Barnett, J.R. Wilson, W. Kobsiriphat, H.-Y. Chen, R. Mendoza, J.M. Hiller, D.J. Miller, K. Thornton, P.W. Voorhees, S.B. Alder, Three dimensional analysis of solid oxide fuel cells using Focused ion beamScanning electron microscopy, Microsc. Microanal., 13 (2007) 596-597. [6] D. Gostovic, J.R. Smith, D.P. Kundinger, K.S. Jones, E.D. Washman, Three-dimensional reconstruction of porous LSCF cathodes, Electrochem. Solid-State Lett., 10 (12) (2007) B214-B217. [7] L. Holzer, B. Munch, B. Iwanschitz, M. Cantoni, T. Hocker, Th. Graule, Quantitative relationships between composition, particle size, triple phase boundary length and surface area in Ni-cermet for Solid Oxide Fuel Cells, J. Power Sources, doi : 10.1016/j.jpowsour.2010.08.006. [8] L. Holzer, B. Muench, M. Wegmann, P.H. Gasser, R.J. Flatt, FIB-nanotomography of particulate systems – Part 1 : Particle shape and topology of interfaces, J. Am. Ceram. Soc., 89 (2006) 2577-2585. [9] J.R Wilson, J.S Cronin, S.A. Barnett, Linking the microstructure, performance, and durability of Ni-Yttria-Stabilized Zirconia solid oxide WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
164 Materials Characterisation V
[10]
[11] [12]
[13]
[14]
[15] [16] [17] [18]
[19]
[20] [21]
fuel cell anodes using three dimensional focused ion beam – Scanning electron microscopy imaging, Scripta Materialia, doi 10.1016/j.scriptamat.2010.09.025. B. Ruger, J. Joos, A. Weber, T. Carraro, E. Ivers-Tiffée, 3D-modeling and performance evaluation of mixed conducting (MIEC) cathodes, ECS Transactions, 25 (2) (2009) 1211-1220.
J. Hoshen, R. Kopelman, Percolation and cluster distribution. I. Cluster multiple labelling technique and critical concentration algorithm, Phys. Rev. B., 1 (14) (1976) 3438-3445. Y. Nakashima, S. Kamiya, Mathematica programs for the analysis of three-dimensional pore connectivity and anisotropic tortuosity of porous rocks using X-ray computed tomography image data, J. Nucl. Sci. Technol., 44 (9) (2007) 1233-1247. D. Rochais, G. Le Meur, V. Basini, G. Domingues, Microscopic thermal characterization of HTR particle layers, Nucl. Eng. And Design, 238 (2008) 3047-3059. T.G. Sherwood, R.L Pigford, C.R. Wilke, Mass transfer, In:Clark, B.J. Maisel, J.W. (Eds.) McGraw-Hill Inc., New York, pp. 39-43, 1975. S. Kakaç, A. Pramuanjaroenkij, X.Y. Zhou, A review of numerical modelling of solid oxide fuel cells, J. Hydrogen Energy, 32 (2007) 761-786. D.R. Lide, Handbook of Chemistry and Physics, Thermal Conductivity of gases, chap. 6, p 206, 89 th. ed., Taylor and Francis, Boca Raton (2008). K.W. Schlichting, N.P. Padture, P.G. Klemens, Thermal conductivity of dense and porous yttria-stabilized zirconia, J. Mat. Sci., 36 (2001) 3003-3010. Y.S. Touloukian, R.W. Powell, C.Y.Ho, P.G. Klemens, Thermophysical Properties of matter, vol. 1, Thermal conductivity, metallic elements and alloys., IFI/Plenum, New-York-Washington, 1970. R. Landauer, The electrical resistance of binary metallic mixtures, J. Appl. Phys., 23 (1952) 779-784. Z. Wu, M. Liu, Modelling of ambipolar transport properties of composite mixed ionic-electronic conductors, Solid State Ionics, 93 (1997) 65-84.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
165
Modelling of load transfer between porous matrix and short fibres in ceramic matrix composites J. G. P. Silva1,2, D. Hotza1, R. Janssen2 & H. A. Al-Qureshi1 1 2
Federal University of Santa Catarina (UFSC), Florianópolis, SC, Brazil Technical University of Hamburg-Harburg(TUHH), Germany
Abstract The aim of this work is to present a model of load transfer between porous matrix and short fibres in ceramic matrix composites. This analysis is based on the earlier shear-lag models used for polymeric composites. However, geometry and strength of fibres in addition to the matrix porosity are included in the present analysis. The theoretical curves for the longitudinal and shear stresses distributions along the fibre-porous matrix interface are presented. They exhibited a maximum strength point at the middle of the short fibres. It became evident that the critical length is governed by the relative properties of the fibres, matrix and porosity, which greatly influenced the load carrying capacity of the fibres in the composites. In addition, the present simplified solution facilitates the understanding of the interface mechanism using porous matrix. Keywords: modelling, ceramic matrix composites, shear-lag models, porous ceramics.
1 Introduction Modern structural ceramic composites possess a number of unique properties that cannot be achieved by other materials. Therefore, they have a potential for saving energy, reducing wear, and increasing the lifetime of components [1]. However, regardless of their remarkable properties, structural ceramics are not as widely used in industry as they should and could be. Among the reasons for the reluctance of industry to introduce structural ceramic as components are [1]: • high price of ceramic parts, • insufficient knowledge in “traditional” metal-oriented engineering, WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line) doi:10.2495/MC110151
166 Materials Characterisation V • low toughness, • need of redesigning components to meet specific requirements. In the case of ceramic composites, the price factor is still the major limitation in terms of commercial applications. This is especially true for components made of composites with complex structures and requiring mechanical performance and reliability. To this aim, fibre-reinforced ceramic composites have been developed to overcome the fragility and unreliability of monolithic ceramics. Their main advantages are high-temperature resistance, low density, better corrosion resistance and adequate damage tolerance [2]. There is a wide range of fibre-reinforced ceramic composites depending on the chemical composition of the matrix and reinforcement, although currently only Cf-C/SiC composites produced by silicon infiltration have reached commercial production. However, fibre-reinforced ceramic composites based on oxides (alumina, mullite) can provide key benefits about long-term stability under oxidizing atmospheres [2]. Despite considerable interest in oxide ceramic composites in the past decade [3–5], there is still no production concept that meets the requirements in view of cost and performance. This study aims to investigate the load transfer phenomenon in ceramic-ceramic short fibre composites, using a simplified mathematical model that aims to predict the actual effectiveness of reinforcement depending on the material properties.
2 Modelling 2.1 Previous considerations and analysis Considering a loaded composite made of a dense fibre with length 2L embedded in a porous matrix made of the same material of the fibre, hereby is assumed that no slippage occurs between fibre and matrix. It should be also considered that the Poisson’s ratio of fibre and matrix is the same, which implies on the inexistence of transversal stress when the loading is applied along the fibre. Considering the displacements in the fibre u and far away from the fibre v: (Fig. 1):
Figure 1:
Simplified scheme of the stress field around the fibre. (a) without loading; (b) loaded [6].
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
167
From Hooke’s law:
E
P E A L
(1)
Taking the differential:
dP EA d dx L dx
(2)
Cox proposes a similar behaviour [7]:
dPf dx
B (u v )
(3)
where Pf is the load acting on the fibre and B is a constant that depends on the fibre distribution and the Young’s Modulus of fibre and matrix. Differentiation of Eq. (3) leads to:
d 2 Pf dx
2
du dv B dx dx
(4)
The derivatives of u and v can be taken as the deformations in the fibre and matrix, respectively:
Pf du dx Af E f
(5)
dv dx
(6)
Substitution of (5) and (6) in (4), gives:
d 2 Pf dx
2
P B f Af E f
(7)
A solution to this differential equation leads to:
Pf E f A f S sinh x T cosh x where:
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
(8)
168 Materials Characterisation V
B E f Af
(9)
and S and T are constants defined by the boundary conditions. To simplify the solution, instead of the stress distribution proposed by Cox, a model proposing that the stress drop in the extremity of the fibre follows a quadratic behaviour. To evaluate the model herein described, it is possible to apply the equations to a model composite, made of a porous alumina matrix and alumina fibres. Table 1 summarizes the properties that are relevant for the calculation. Table 1:
Properties of the model composite.
Property Fibre volume fraction Matrix porosity fraction Fibre length – 2l (mm) Fibre diameter (μm) Ratio Critical length/Length or α
Value 0.45 0.24 50.8 10 0.25
2.2 Proposed model According to Fig. 2, let us consider a composite with fibres whose length is 2L, diameter 2r and Young’s modulus Ef, embedded in a matrix with porosity ρ, made of the same material of the fibre. Hereby we define the critical length Lc, in which from the tip of the fibre the stress distribution isn’t constant by the shearlag between matrix and fibre. It is more feasible to work with α, the ratio between the critical length and fibre length, being Lc =α·L.
Figure 2:
Proposed stress distribution and boundary conditions.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
169
It can be proposed that the stress distribution between the points L-αL and L follows a behaviour such as:
f ( x ) Ax B
(10)
By using the boundary conditions given in Fig. 2, and substituting them in (10):
E f A( L L) B
(11)
(1 ) E f AL B
(12)
Isolating B in (12) and replacing in (11):
(1 ) E f AL E f A( L L)
(13)
E f A L
(14)
And then:
A
Ef L
(15)
By replacing A from (12):
(1 ) E f
Ef B
(16)
Therefore, B is given by:
1 B 1 1 E f
(17)
By replacing the constants in (10), we have the stress distribution behaviour:
f ( x)
Ef 1 x 1 1 E f L
(18)
To determine the shear stresses along the fibre, the force equilibrium in a fibre element with diameter 2r and length dx is made in the x direction, resulting in: (19) r 2 2 rdx 0 Then, the shear stresses are given by:
r d 2 dx
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
(20)
170 Materials Characterisation V By the differential of (18):
f ( x)
Figure 3:
r E f 2 L
(21)
Stress distribution along the fibre for different matrix porosities. α=0.25.
Figure 4:
Force equilibrium in a fibre element whose length is dx.
With the stress distribution along the fibre, it is possible to calculate the average stress carried by the fibre in the composite, given by:
f
1 L f ( x )dx L 0
(22)
For α ≥ 1, i.e. the fibre is shorter than the critical length:
f
1 L Ef 1 x 1 1 E f dx 0 L L
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
(23)
Materials Characterisation V
Figure 5:
171
Shear stress distribution along the fibre for different matrix porosities. α=0.25.
Then,
Ef f 1 2
(24)
And for 0 < α < 1, i.e., the fibre is longer than the critical length:
f
L L
0
E f dx
L
L L
f ( x )dx
(25)
L
Therefore:
f E f 1
2
(26)
With the average stresses well defined, we can define the stresses in the ply longitudinal and transversal directions. When the matrix material is the same as the fibre, it is possible to write the elastic modulus of the matrix as a function of the fibre modulus:
Em E f e bp where b is a shape factor that depends on the pore shape and distribution.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
(27)
172 Materials Characterisation V
Figure 6:
Longitudinal resistance of the composite [6].
The stress on the transversal direction is proportional to the matrix maximum stress, given by:
T m E f e bp
(28)
The stress on the longitudinal direction is given by the average value between matrix and fibre, based on the volumetric fractions of fibre and matrix:
L m 1 f f f 1 f E f e bp f f
(29)
Therefore for 0 < α < 1:
L 1 f E f e bp E f 1
f 2
(30)
And for α > 1:
L 1 f E f e bp E f 1
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
f 2
(31)
Materials Characterisation V
Longitudinal ply ultimate strength, for different α.
Figure 7:
Figure 8:
173
Longitudinal ply ultimate strength for different porosities.
3 Conclusions A load transfer model was proposed to ceramic composites, which relates the matrix porosity and fibre length with the mechanical strength of such composites. It is hoped that the experimental validation of the model can be established or any changes aimed at increasing its accuracy can be made.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
174 Materials Characterisation V
Acknowledgements The authors would like to thank the agencies CAPES (Coordenação de Aperfeiçoamento de Pessoal de Nível Superior), CNPq (Conselho Nacional de Desenvolvimento Científico e Tecnológico) and DFG (Deutsche Forschungsgemeinschaft), for funding this research under grant Bragecrim 015/09.
References [1] Janssen, R., Scheppokat, S. & Claussen, N., Tailor-made ceramic-based components – Advantages by reactive processing and advanced shaping techniques. Journal of the European Ceramic Society, 28, pp. 1369-1379, 2008. [2] Janssen, R. & Hotza, D. Low-cost and reliable production of oxide ceramic matrix composites, Bragecrim Project Proposal, December 2008. [3] Wendorff, J., Janssen, R. & Claussen, N., Saphirfaserverstärkung reaktionsgebundener oxider Keramiken. Verbundwerkstoffe und Werkstoffverbunde, ed. G. Ziegler, pp. 421-424, 1996. [4] Lundberg, R. & Eckerbom, L., Design and processing of Al-oxide composites. Ceramic Transactions, 58, pp. 95-104, 1995. [5] Levi, C.G., Yang, J.Y., Dalgleish, B.J., Zok, F.W. & Evans, A.G., Processing and performance of an all-oxide ceramic composite. Journal of the American Ceramic Society, 81, pp. 2077-2086, 1998. [6] Casaril, A.; Gomes, E.R.; Soares, M.R.; Fredel, M.C. & Al-Qureshi, H.A., Análise micromecânica dos compósitos com fibras curtas e partículas. Matéria, 12(2), pp. 408-419, 2007. [7] Cox, H.L., The elasticity and strength of paper and other fibrous materials. British Journal of Applied Physics, 3, pp. 72-79, 1952
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
175
Modeling aspects concerning the axial behavior of RC columns H. O. Koksal1, T. Turgay 2, C. Karakoç3 & S. Ayçenk4 1
Construction Technology Program, Çanakkale Onsekiz Mart University, Turkey 2 Faculty of Engineering and Architecture, Abant İzzet Baysal University, Bolu, Turkey 3 Civil Engineering Department, Boğaziçi University, Turkey 4 M.S. Student, Civil Engineering Faculty, Yıldız Technical University, Turkey
Abstract This paper is concerned with the axial behavior of the RC columns. Stress-strain relationships of experimentally tested RC columns under concentric loading are compared with the predictions of the Koksal model. Moment-curvature analyses of RC sections are also performed employing the same model in a self-developed moment-curvature program for confined concrete. Results are compared with the output of EXTRACT which uses the Mander concrete model. Keywords: confinement, reinforced concrete columns, concentric loading, stress–strain relations.
1 Introduction Confining pressure on RC columns increases the concrete strength. Transverse reinforcements such as steel stirrups, FRP wraps, and steel jackets create a triaxial compressive stress state producing confinement action around the concrete core. Extensive research on the improvement of concrete confinement has been carried out since the pioneering study of Richart et al. [1]. The constitutive model for confined concrete plays an important role on the accuracy of the moment-curvature curves to perform a reliable pushover analysis of RC frames. There are some frequently cited models (Hognestad [2], Kent and Park [3], Sheikh and Uzumeri [4], Mander et al. [5], Saatcioglu and Razvi [6]) to predict WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line) doi:10.2495/MC110161
176 Materials Characterisation V the peak stress or the stress–strain curve of confined concrete. Köksal has also proposed a failure criterion for concrete under tri-axial compression stress state [7]. In this paper, four square RC columns tested in Yıldız Technical University (Turgay [8], Köksal et al. [9]) are evaluated implementing Koksal model. The flexural behavior of the columns is also simulated and their moment-curvature relations corresponding to different axial load levels are obtained using the same model. Moment-curvature curves are compared with the outputs of EXTRACT.
2 Experimental study The square specimens in the test program have 200x200 mm cross-section dimensions and 1000 mm height. The columns were tested in the structural laboratory of Yıldız Technical University [8]. Figure 1 shows the details of test setup and instrumentation for the two specimens.
Figure 1:
Test setup and details of test specimens C1L4S8 and C1L8S8 [8].
The experimental study was limited to one type of concrete mix design typically. C1 type columns are tested at 30 days. All longitudinal bars are 10mm in diameter and L4 and L8 shows the number of the bars in the cross-section of a column. The tie spacing is 100mm and the S8 represents diameter of stirrups. For measurement of axial strains, four linear variable displacement transducers (LVDTs) are placed over the central 400mm gage length at each side of a column in a similar way used to assess any eccentricity of the applied load as recommended in the study of Shrive et al. [10]. Although the LVDT readings were provided very close to each other, next to the maximum axial load there can be a significant variation between the minimum and maximum values of shortening reaching very high values [9]. Figure 2 shows the axial load-axial WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
177
shortening curve for C1L8S8 for which almost perfectly homogeneous strains were obtained. However, for other columns the differences between LVDT readings are somewhat greater indicating a localized damage zone at one specific region of the specimen.
Figure 2:
Four separate LVDT readings of axial load-shortening curves for C1L8S8 [9].
3 Simulation of flexural behavior and moment curvature relations The limit state design procedure of reinforced concrete elements has undergone major revision by most of the international codes in harmony with the performance-based design engineering approach. In this approach, momentcurvature behavior of flexural members is needed to define the deflection demand and to simulate the behavior of the reinforced concrete members under lateral loads such as earthquake actions. Moment-curvature plots readily illustrate stiffness, strength, and cross-sectional ductility, and allow the calculation of deflections after materials become nonlinear. In a flexural member the shear reinforcement or any other confinement mechanism applies pressure to the concrete in the compression zone and affects energy dissipation capacity directly. Therefore, to predict the moment-curvature behavior of a flexural member, the stress–strain behavior of confined concrete under axial compression is vital. With the development of performance-based design methods, there is an increasing need for simplified but reliable analytical models capable of predicting the flexural behavior of reinforced concrete members. The flexural behavior of reinforced confined concrete sections is introduced by Koksal model [7]. The failure criterion proposed by Koksal is given as:
f 1.1317 0.7645 2 k1, 2 ,3 0 WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
(1)
178 Materials Characterisation V
where k k f l , f c' , and are deviatoric and hydrostatic lengths, respectively. In this equation k, as a function of the lateral confinement pressure, f l , and the cylinder compressive strength of concrete, f c' can be expressed as [7] : 2 fl fl k 4.07 ' 0.89 ' 0.807 fc' fc fc
(2)
The theoretical confining pressure, f l , can be found in usual way as [11] : or (3) f l , x k e sx f yh f l , y k e sy f yh In this relation, sx and sy are the confinement proportion both x and y direction respectively, ke is the confinement effectiveness coefficient, and fyh is the yield strength of transverse reinforcing steel. The Saenz’s equation [12] is adopted for describing the monotonic stress-strain relationship for confined concrete:
1
1E0
E 1 0 2 1 1 cc cc Es
2
(4)
where 1 and 1 are axial compressive stress and strain of concrete, respectively; E0 is the initial tangent modulus of elasticity in MPa and can be calculated as E0 4750 f co' ; Es is the secant modulus at the point of maximum compressive stress f cc' which can be determined using Equation 1. The strain cc corresponding to the maximum compressive stress f cc' can be found employing the recommended relations of Richart et al. [1] :
cc ' c 1 k 2
fl f 'c
(5)
where c' is the peak strain at the strength of plain concrete cylinders. In this equation, k2 is taken as 5k1. Figure 4 aims to compare experimental axial stress-strain curves of the four column specimens with those predicted from Eqs (1)-(4). In this study, a simple program is written in FORTRAN to produce momentcurvature plots for confined column section under bending and axial compression. The cross section is divided into 1mm thickness slices. For a given depth of natural axis, the strain at the extreme slice in compression is found by iterative procedure. For each slices, the stress and strain are calculated using force equilibrium and compatibility requirements respectively. Stress acting on the core concrete (cover concrete is neglected) calculated using Koksal’s model. Koksal’s model has also been utilized through the 3D finite element analysis of RC and FRP-confined concrete columns [11, 13] successively. Figure 3(a) WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
179
demonstrates the typical axial stress-strain curves recommended by Saenz for a concrete member [12]. Finite concrete forces for unconfined and confined core of each slice are found by multiplying stress by corresponding areas. Stress at the reinforcement bars is found by entering a simple bi-linear stress-strain curve in Figure 3(b) with the strain value found from compatibility requirements at each load level. Finite forces on steel can be found by multiplying stress by the area of the reinforcement.
Figure 3:
(a) Saenz curve for uniaxial behavior of concrete; (b) stress-strain relation for steel.
Figure 4:
Experimental and predicted stress-strain curves for the specimens.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
180 Materials Characterisation V Figure 5 show the moment-curvature relations for C1L4S8 and C1L8S8 at various load levels. Increasing the axial load from 200kN to 400kN always results an increase in the moment carrying capacity. But when a load near to the axial load capacity is considered, moment carrying capacity does not increase any more. Also decrease in the curvature is very obvious when the axial load increases as expected.
Figure 5:
Moment-curvature diagrams for the four columns at various axial load levels.
30
Moment (KNm)
C1L4S8
20
10 Model L200 EXTRACT L200 0 0
Figure 6:
20
40 60 Curvature x 10e6 (rad/m)
80
100
Output of EXTRACT program and the proposed model for moment-curvature diagrams for C1L4S8 at the axial load level of 200kN.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
181
Figures 6–8 show the output of EXTRACT program [14] for C1L4S8 subjected to the axial load levels of 200kN, 400kN and 600kN respectively. EXTRACT program employs the Mander model for confined concrete. There are not significant differences if one compares the curvature values for both programs. 40 C1L4S8
Moment (KNm)
30
20
10 Model L400 EXTRACT L400 0 0
20 40 Curvature x 10e6 (rad/m)
60
Figure 7:
Output of EXTRACT program and the proposed model for moment-curvature diagrams for C1L4S8 at the axial load level of 400kN.
Figure 8:
Output of EXTRACT program and the proposed model for moment-curvature diagrams for C1L4S8 at the axial load level of 600kN.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
182 Materials Characterisation V
4 Conclusion This paper has been primarily concerned with the modeling aspects for the axial behavior of RC columns. Experimental axial stress-strain curves of the four column specimens are predicted employing Koksal model. Furthermore, a spreadsheet-based program is constructed to produce moment-curvature plots for the performance-based design engineering approach. The following conclusions can be drawn based on the results of the analyses: 1. As can be seen in Figure 5, Koksal model successively predicts the axial behavior of the four RC column specimens tested under concentric loading. 2. It can be easily observed that Koksal model results into somewhat higher values for the moment carrying capacities than the outputs of EXTRACT program which employs Mander model in Figure 6–8, but the general trend of the curves are the same. The developed program gives 27kNm and 38kNm for moment carrying capacities at the axial load levels of 400kN and 600kN while EXTRACT program results are between 22kNm and 27kNm.
Acknowledgement The support of Boğaziçi University Research Fund (ref: research project 5232) for this paper is gratefully acknowledged.
References [1] Richart, F.E., Bradtzaeg, A. and Brown, R. L. 1928. A study of the Failure of Concrete under Combined Compressive Stresses. Bulletin Np. 185, Engineering experimental station University of Illinois, Urbana, pp. 104. [2] Hognestad, E., 1951. A Study of Combined Bending and Axial Load in Reinforced Concrete Members. Bulletin Series No.399, University of Illinois Eng. Exp. Station, Urbana. [3] Kent, D.C. and Park, R. 1971. Flexural Members with Confined Concrete. Journal of the Structural Division, Proc. of the American Society of Civil Engineers, 97(ST7), pp.1969-1990. [4] Sheikh, S.A. and Uzumeri, S.M. 1982. Analytical Model for Concrete Confinement in Tied Columns. Journal of the Structural Division, Proc. of the American Society of Civil Engineers, 108(ST12), pp. 2703-2722. [5] Mander, J.B., Priestly, M.J.N. and Park, R. 1988. Theoretical Stress-Strain Model for Confined Concrete. Journal of the Structural Engineering, ASCE, 114(8), pp.1804-1826. [6] Saatcioglu, M. and Razvi, S.R. 1992. Strength and Ductility of Confined Concrete. Journal of the Structural Engineering, ASCE, 118(6), pp.1590-1607. WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
183
[7] Köksal, H.O. 2006. A Failure Criterion for RC Members Under Triaxial Compression. Structural Engineering and Mechanics, Techno-Press, 24(2), pp.137-154. [8] Turgay, T. CFRP Uygulanmıs_ betonarme elemanların performansı (The performance of FRP strengthened structural members). PhD thesis, Submitted to Yıldız Technical University; 2007. [9] Köksal, H.O., Karakoç, C., Polat Z., Turgay T. and Akgün Ş. 2007.Evaluation of Experimental Procedures for Confined Concrete Columns. Computational Methods and Experimental Measurements XIII, WIT Transactions on Modeling and Simulation, 46, 233-242. [10] Shrive, P.L., Azarnejad, A., Tadros, G., McWhinnie, C. and Shrive, N.G. 2003. Strength of Concrete Columns with Carbon Fibre Reinforcement Wrap. Canadian Journal of Civil Engineering, 30, pp. 543-554. [11] Doran, B. 2009.Numerical simulation of conventional RC columns under concentric loading. Material and Design, 30(6), 2158-2166. [12] Saenz, L.P. 1964. Discussion of Equation for Stress-Strain Curve of Concrete by Desai and Krishnan. ACI, 61(9), 1229-1235. [13] Köksal, H.O, Doran, B., and Turgay, T. 2009. A practical approach for modeling FRP wrapped concrete columns. Constr. and Build. Mat., 23(3), 1429–1437. [14] EXTRACT-v.3.0.8, Cross section analysis program of structural engineers, TRC/Imbsen Software Systems.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
This page intentionally left blank
Section 5 Innovative experiments
This page intentionally left blank
Materials Characterisation V
187
Surface characterization of eucalyptus and ash wood veneers by XPS, TOF-SIMS, optic profilometry and contact angle measurements G. Vázquez, R. Ríos, M. S. Freire, G. Antorrena & J. González-Álvarez Department of Chemical Engineering, University of Santiago de Compostela, Spain
Abstract Composition and properties of the wood surface are very important in board manufacture as they determine wood-adhesive bonding and the quality of the final product. In this work two spectroscopic techniques, X-ray photoelectron spectrometry (XPS) and time of flight secondary ion mass spectrometry (TOF – SIMS), have been employed to study the surface composition of eucalyptus (Eucalyptus globulus) and ash (Fraxinus excelsior) sliced cut veneers. Both wood species are widely used for decorative veneers in the finishing of wood panels (particle boards, medium density fiber–boards, etc.). Further characterization of the wood surface was carried out by optic profilometry and wettability analysis, using the sessile drop method to measure contact angles. Wood is a very heterogeneous material and its composition can vary significantly depending on the sampling area; then, the influence of the radial position of the veneer in the trunk was also analyzed. From the low resolution XPS spectra the oxygen to carbon (O/C) ratio was calculated and, resolving the C1s signal, the percentages of the different carbon peaks corresponding to different functional groups (C1, C2, C3 and C4 carbons) were also calculated. The lower O/C ratio and the higher C1/C2 ratio for eucalyptus than for ash was attributed to the higher concentration of extractives on the eucalyptus wood surface that was confirmed by the TOF-SIMS spectra, which additionally revealed a patchy distribution of the extractives. The higher hydrophobicity of the eucalyptus wood surfaces was also supported by the results on the wetting properties of the veneers. Eucalyptus veneers exhibited a significantly higher constant wetting rate angle (cwra) and, consequently, a lower wettability than WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line) doi:10.2495/MC110171
188 Materials Characterisation V ash veneers, which can also be related with the higher values of the rugosity parameters for the latter. There were no significant influences of the radial position of the veneer on the surface properties. Keywords: wood veneers, wettability, contact angles, XPS, TOF-SIMS, optic profilometry.
1 Introduction Wood is a renewable resource composed mainly of three polymers, cellulose, hemicelluloses and lignin, and various extractives such as fatty acids, sterols, tannins, etc. Wood surface characteristics condition the interaction with adhesives and binders in wood based products and, therefore, have influence in the quality of the final product. Several methods have been proposed to characterize material surfaces. Contact angle analysis is one the traditional methods for the characterization of surfaces on a macroscopic scale, and provides information on surface wetting properties [1]. Surface chemical composition of complex materials such as wood can be determined using two complementary spectroscopic techniques, X-ray photoelectron spectroscopy (XPS) and time-of-flight secondary ion mass spectrometry (TOF-SIMS), the former generating more quantitative data and the latter more qualitative data [2]. X-ray photoelectron spectroscopy (XPS) provides the elemental composition for the outermost 5–10 nm surface layers and functional groups, chemical bonding types, oxidation state, etc. can be additionally deduced from XPS data. Time-of-flight secondary ion mass spectrometry (TOF-SIMS) provides data for depths of 1–2 nm, including detailed chemical information via identification of intact molecular ions or characteristic molecular fragments that are emitted from the surface [3]. Additionally, it provides ion images, mapping the lateral distribution of secondary ions signals from lignin, carbohydrates, extractives and metals within an analyzed wood surface area [2, 4]. The aim of this paper was to evaluate and compare the surface properties and composition of eucalyptus (Eucalyptus globulus) and ash (Fraxinus excelsior) sliced cut veneers used as decorative veneers in the finishing of wood panels. A wettability analysis, using the sessile drop method to measure contact angles, was applied in combination with two spectroscopic techniques, X-ray photoelectron spectrometry (XPS) and time of flight secondary ion mass spectrometry (TOF –SIMS). Roughness of the wood surface was analysed by optic profilometry. Wood is a very heterogeneous material and its composition can vary significantly depending on the sampling area; then, the influence of the radial position of the veneer in the trunk was also analyzed.
2 Methods 2.1 Contact angle determination Samples of eucalyptus (Eucalyptus globulus) and ash (Fraxinus excelsior) sliced cut veneers with a humidity of 8% (wet basis) were used to prepare 2 cm (length) WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
189
x 2cm (width) x 0.6 mm (thickness) wood pieces. Contact angle measurements were made perpendicular to the grain direction of the wood according to the sessile drop method using a Dataphysics OCA 15 Plus equipment (Filderstadt, Germany) with a video measuring system with a high-resolution CCD camera. The data were analyzed with the Dataphysics software SCA 20. To compare the wettability of both wood species and to analyze the influence of the radial position of the veneer in the trunk, 20% (by weight) aqueous solutions of chestnut shell extracts, potential components of wood adhesives [5], were used. Ten μL drops were added and the variation of contact angle with time was registered till 240 s. Contact angles were calculated according to the Young-Laplace method. The constant wetting rate angle (cwra) was determined as per Nussbaum [6] plotting the wetting rate (dθ/dt, being θ the contact angle) against time and selecting the θ value (cwra) corresponding to a constant wetting rate. Initial (θ0, t=0), final (θf, t=240 s) and cwra contact angles were calculated as the average of thirty measurements on each wood species and radial position. 2.2 X-ray photoelectron spectrometry (XPS) Analysis of the samples was performed using a Thermo Scientific K-Alpha ESCA instrument equipped with aluminium Ka1,2 monochromatized radiation at 1486.6 eV X-ray source. Due to the non conductor nature of samples it was necessary to use an electron flood gun to minimize surface charging. Neutralization of the surface charge was performed by using both a low energy flood gun (electrons in the range 0 to 14 eV) and an electrically grounded stain steel screen placed directly on the sample surface. The XPS measurements were carried out using monochromatic Al-K radiation (hν=1486.6 eV). Photoelectrons were collected from a takeoff angle of 90º relative to the sample surface. Measurements were done in a Constant Analyser Energy mode (CAE) with a 100 eV pass energy for survey spectra and 20eV pass energy for high resolution spectra. Charge referencing was done by setting the lower binding energy C1s photopeak at 285.0 eV C1s hydrocarbon peak. The spectra fitting was based on Chi-squared algorithm used to determine the goodness of a peak fit. Surface elemental composition was determined using the standard Scofield photoemission cross sections. The chemical functional groups identity was obtained from the high-resolution peak analysis of carbon-1s (C1s) and oxygen1s (O1s) envelopes. 2.3 Time of flight secondary ion mass spectrometry (TOF –SIMS) The TOF-SIMS analyses were made on a TOF-SIMS IV (ION-TOF GmbH, Germany).The sample was bombarded with a pulsed Bismuth ion beam. The secondary ions generated were extracted with a 10 kV voltage and their time of flight from the sample to the detector was measured in a reflectron mass spectrometer. Mass spectra of positive and negative secondary ions were acquired from randomly selected areas of 500 x 500 µm2 on each sample. In all WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
190 Materials Characterisation V cases, a 25 keV pulsed Bi3+ beam at 45º incidence was used. Low–energy electron flooding was used for charge compensation. All analyses were done under static SIMS conditions (1012 ions/cm3). 2.4 Optic profilometry Interferometric profilometry measurements were carried out in the Vertical Scanning Interferometry (VSI) mode using an Interferometric Microscope WYKO NT-1100. For each sample, three measurements were carried out at a 5X magnification (1.2x0.9 mm2), to determine statistically representative values of surface roughness parameters, and a measurement at magnifications of 20X (298x227 µm2) and 50X (119x91 µm2) to see the details of the surface topography. Great field measurements covering an area of 3x2 mm2 at a 5X magnification were also performed to obtain a more representative image of the surface.
3 Results and discussion 3.1 Wettability of eucalyptus and ash wood veneers Figure 1 shows the evolution of contact angle with time for an aqueous solution of chestnut shell extracts on eucalyptus and ash veneers. For each species, veneers obtained from different positions in the trunk, interior and exterior, were compared. Throughout the time range tested, contact angles for eucalyptus veneers were significantly higher than those for ash veneers. Additionally, for both wood species there were not significant differences between the contact angles for the interior and exterior veneers.
Figure 1:
Contact angle versus time for an aqueous solution of chestnut shell extracts on interior and exterior eucalyptus and ash sliced veneers.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
191
The values calculated for the initial (t=0; θ0), final (t =240 s; θf) and constant wetting rate (cwra) angles are presented in Table 1. Not only initial contact angles but also final and constant wetting rate angles were remarkably higher for the eucalyptus veneers. Higher contact angles are indicative of a worse wetting behaviour and could be related with a greater migration of hydrophobic substances to the surface of eucalyptus veneers [7]. Differences between the θ0, θf and cwra values of interior and exterior veneers were almost negligible which confirms the previous qualitative observation. Table 1:
Initial (θ0), final (θf) and constant wetting rate (cwra) angles for exterior and interior veneers of eucalyptus and ash wood with an aqueous solution of chestnut shell extracts.
Veneer Eucalyptus Ash
θ0 (º) 99.72 (5.83) 75.22 (1.44)
Exterior θf (º) cwra (º) 45.25 50.8 (1.88) (0.11) 28.01 27.99 (0.37) (0.37)
θ0 (º) 95.34 (4.77) 73.42 (9.91)
Interior θf (º) cwra (º) 46.70 50.12 (6.77) (0.10) 27.75 30.6 (8.43) (0.76)
(Standard deviation)
3.2 X-ray photoelectron spectrometry (XPS) Figure 2a shows the XPS survey spectra of eucalyptus and ash samples of interior veneers. The surface elemental compositions (in atomic %) for the different sample groups are summarized in Table 2. The percentages shown are the mean value of three determinations. Table 2:
Surface elemental composition (in atomic %) for eucalyptus and ash veneers.
Veneer Exterior eucalyptus Interior eucalyptus Exterior ash Interior ash
C 80.7 (2.2) 79.2 (0.1) 77.2 (1.0) 76.9 (1.4)
O 18.0 (2.3) 19.3 (0.7) 20.8 (1.1) 20.5 (1.2)
N 0.7 (0.2) 1.0 (0.4) 0.6 (0.1) 0.9 (0.1)
Si 0.6 (0.1) 0.8 (0.2) 1.2 (0.5) 1.2 (0.5)
Ca 0.3 (0.1) 0.5 (0.1)
(Standard deviation)
As seen in Table 2, carbon and oxygen are the main components in the wood surfaces and small amounts of nitrogen and silicon were also detected in both wood species, whereas calcium was present only in the ash samples. Nitrogen is present in the form of amine/amide functional groups. This assignment is as a result of the N1s peak position, which is at approximately 400 WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
192 Materials Characterisation V eV (with C-C carbon set at 285.0 eV as the binding energy scale reference). The presence of such compounds is mainly as a result of naturally occurring proteins in wood but some contamination by adsorption from the laboratory air cannot be excluded [8]. From the high resolution spectra (Figure 2b) the functional groups which made up the C1s spectra have been identified. The C1s spectra of all samples comprised four peaks with binding energies (BE) of approximately 285.0 (C1), 286.6 (C2), 288.1 (C3) and 289.2 (C4) eV. The C1 carbon component is related to C–C or C–H bonds, the C2 carbon component represents single C–O bonds, the C3 carbon component is bonded to a carbonyl or two non-carbonyl oxygen (C=O or O-C-O) and the C4 class of carbon atoms is bonded to a carbonyl and a non-carbonyl oxygen (O=C-O) [9].
Figure 2:
XPS survey spectra (a) and high resolution C1s spectra (b) for interior eucalyptus and ash veneers.
The relative percentages of the C1s components are shown in Table 3 together with the O/C and C1/C2 ratios. The C1 component arises from lignin and extractives and the C2 component can arise from all wood components but predominantly from cellulose. For both species, most carbon atoms (60–66.5%) present in the wood surface layer were C1-carbons, and between 29.2 and 33.5% of the total carbon signal was from the C2 type. According to a normal wood WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
193
composition and the structure of wood components, carbon atoms bonded with one oxygen atom (C2 component) should be greater. The observed increase in the presence of C1 components, which was more significant for eucalyptus samples, might be due to the migration of lipophilic extractives to the wood surface [9]. Table 3:
Percentages of the C1s spectra components and C1/C2 and O/C ratios for eucalyptus and ash veneers.
Veneer Exterior eucalyptus Interior eucalyptus Exterior ash Interior ash
C1 66.5 (4.7) 65.4 (2.9) 60.1 (1.7) 60.0 (2.2)
C2 29.2 (3.4) 29.3 (3.2) 32.4 (0.4) 33.5 (2.4)
C3 2.3 (1.6) 2.8 (0.5) 5.0 (1.2) 4.1 (1.0)
C4 2.0 (0.4) 2.8 (0.3) 2.6 (0.1) 2.4 (0.7)
C1/C2
O/C
2.3
0.22
2.2
0.24
1.9
0.27
1.8
0.27
(Standard deviation)
This theory is reinforced by the values obtained for the O/C and C1/C2 ratios, which are directly related to the chemical composition of wood constituents (polysaccharides, lignin and extractives). The O/C ratio varied between 0.22 and 0.27 (Table 3). If experimental O/C ratio values are compared to those of pure compounds, 0.83 for cellulose, 0.33 for lignin and 0.1 for extractives [10] the preferential presence of lignin and extractives on the wood surface is demonstrated, mainly in the case of eucalyptus veneers. Moreover, the C1/C2 ratio provides an additional evidence to support the above interpretations. Theoretically calculated values for the C1/C2 ratio are equal to 0 for pure cellulose, around 1 for lignin and 10 or higher for extractives [11]. Therefore, values in the range from 1.8 to 2.3 confirm the presence of extractives on the wood surfaces and especially on eucalyptus wood. Finally, when comparing the mean values of the C1/C2 and O/C ratios and the C1-carbon percentages, more hydrophobic material was found in exterior eucalyptus veneers. The more hydrophobic character of the eucalyptus wood samples is in agreement with the results obtained from contact angle measurements which revealed better wetting properties for ash veneers. 3.3 Time of flight secondary ion mass spectrometry (TOF–SIMS) TOF-SIMS spectra were prepared to represent the masses from 1 to 700 u, although the range from 100 to 200 u is the most interesting because in this spectral region fragments of lignin and carbohydrates are detected. On the contrary, in the region below 100 u, multiple fragments which are common to all organic materials, the so-called non-specific organic fragments appear and, therefore, fragments of this region of the spectrum are rarely used in the study of WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
194 Materials Characterisation V organic compounds. Na, Ca or Al, are also detected in the area of the spectrum below 100 u. However, the intensity of these ions is low when compared with the intensity of the peaks of hydrocarbons. Extractive organic compounds appear in areas of the spectrum with m/z values between 230 and 700 u. Figure 3 shows the comparative positive polarity spectra for ash and eucalyptus veneers at m/z values from 135 to 185 u. Negative polarity spectra were nor included due to the low intensity of the observed signals. For both samples, the characteristic ions of lignin, at m/z =137 and 151 for lignin guaiacyl units (G), and at m/z=167 and 181 for lignin syringyl units (S), were observed.
Figure 3:
Comparison of the positive ion TOF-SIMS spectra from 135 to 185 u for eucalyptus and ash veneers.
Typical peaks of cellulose and hemicelluloses, localized at m/z = 115, 127, 133 and 145 u, also appeared. Peaks at m/z =115 and 133 u are attributed to xylan, the peak at 127 u to mannan and the one at 145 u to cellulose. Additional peaks at m/z =147 and 149 u are attributed to mannan and cellulose. WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
195
TOF-SIMS technique is not quantitative and, therefore, direct comparison of the intensity of a particular ion in different samples does not allow extracting conclusions about its abundance. However, it is possible to compare different samples by referring the intensities of the ions of interest to that of another ion not related to them. In this way, the intensities of the lignin ions at m/z = 137, 151, 167 and 181 u in eucalyptus and ash wood samples were referred to that of the ion at m/z =147 u, attributable to hemicelluloses and present in both kind of samples. The relative intensity of lignin ions in eucalyptus samples was slightly higher than in the ash ones, which indicated that lignin was more abundant on the eucalyptus veneer surfaces. Figure 4 presents the comparison of the positive polarity TOF-SIMS spectra for eucalyptus and ash veneers in the region from 135 to 450 u. Signals at 383, 397 and 425 u are attributed to organic wood extractives. As shown in the spectra, the content of extractives is higher in the eucalyptus sample than in the ash one which is consistent with XPS results.
Figure 4:
Comparison of the positive ion TOF-SIMS spectra from 135 to 450 u for eucalyptus and ash veneers.
Positive ion images for selected ions signals from eucalyptus and ash samples are shown in Figure 5. Black colour indicates no ion signal and white colour maximum signal intensity. Images of all the characteristic ions for lignin (137, 151,167 and 181 u) and extractives, mainly sterols (383, 397 and 425 u), are presented together. Ion images showed that lignin and mainly extractives were more prominent on eucalyptus than on ash veneer surfaces and, moreover, revealed that both components were inhomogeneously distributed on the surface.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
196 Materials Characterisation V
Figure 5:
Positive ion images for eucalyptus (a) and ash (b) samples.
3.4 Optic profilometry Table 4 shows the average values of the roughness parameters Ra (arithmetic mean of the absolute value of the distances from the mean line to the profile) and Rq (mean value of the square roots of the distances from the mean line to the profile) for eucalyptus and ash veneer samples. Due to the heterogeneity of wood samples, surface roughness values depended strongly on the topographic characteristics of each particular area, so that the values showed a high dispersion. Therefore, the most representative roughness values might be those of the larger areas (called stitch), as the measure itself represents the average of a larger area of the samples. The main conclusion drawn from the values of surface roughness is that ash samples are rougher than eucalyptus samples, approximately 1.7 times based on the results of the stitch areas, which can be related with the better wettability behaviour previously encountered.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
Table 4:
197
Roughness average values for eucalyptus and ash veneer surfaces.
Sample
Magnification 5X (1.2x0.9 mm2) 20X (298x227 µm2 )
Eucalyptus 50X (119x91 µm2) STITCH (3x2 mm2) 5X (1.2x0.9 mm2) 20X (298x227 µm2 ) Ash 50X (119x91 µm2) STITCH (3x2 mm2)
Ra (µm) 11.31 (2.44) 3.44 (1.24) 2.30 (0.91) 15.91 (0.45) 12.17 (1.23) 8.78 (2.60) 6.10 (2.56) 27.38 (2.55)
Rq (µm) 16.26 (4.63) 5.11 (0.66) 2.93 (1.00) 23.56 (4.70) 15.92 (1.23) 11.61 (4.14) 8.05 (3.54) 39.36 (6.80)
(Standard deviation)
Acknowledgements This work was funded by Ministerio de Ciencia e Innovación, FEDER Funds and Plan E Fundy (CTQ2009-07539).
References [1] Bryne, L.E. & Walinder, M.E.P., Ageing of modified wood. Part 1: Wetting properties of acetylated, furfurylated, and thermally modified wood. Holzforschung, 64, pp. 295-304, 2010. [2] Englund F., Bryne L.E., Ernstsson M., Lausmaa J. & Walinder M., Spectroscopic studies of surface chemical composition and wettability of modified wood. Wood Mater. Sci. Eng., 1-2, pp. 80–85, 2009. [3] Bryne, L.O., Lausmaa, J., Ernstsson, M., Englund, F. & Walinder, M.E.P., Ageing of modified wood. Part 2: Determination of surface composition of acetylated, furfurylated, and thermally modified wood by XPS and ToFSIMS. Holzforschung, 64, pp. 305–313, 2010. [4] Tokareva, E.N., Fardim, P., Pranovich, A.V., Fagerholm, H.P., Daniel, G. & Holmbom, B., Imaging of wood tissue by ToF-SIMS: Critical evaluation and development of sample preparation techniques. App. Surf. Sci., 253, pp. 7569–7577, 2007. [5] Vázquez, G., González-Alvarez, J., Santos, J., Freire, M.S. & Antorrena, G., Evaluation of potential applications for chestnut (Castanea sativa) shell
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
198 Materials Characterisation V
[6]
[7]
[8] [9]
[10] [11]
and eucalyptus (Eucalyptus globulus) bark extracts. Ind. Crops Prod., 29, pp. 364-370, 2009. Nussbaum, R.M., Natural surface inactivation of Scots pine and Norway spruce evaluated by contact angle measurements. Holz Roh- Werkstoff, 57, pp. 419-424, 1999. Christiansen, A.W., Effect of ovendrying of yellow-poplar veneer on physical properties and bonding. Holz Roh- Werkstof , 52, pp. 139-149, 1994 Popescu, C.N, Tibirna, C.N. & Vasile C., XPS characterization of naturally aged wood, App. Surf. Sci., 256, pp. 1355–1360, 2009. Sinn, G., Reiterer, A. & Stanzl–Tschegg, S.E., Surface analysis of different wood species using X-ray photoelectron spectroscopy (XPS). J. Mater. Sci., 36, pp. 4673-4680, 2001. Barry, A.O., Koran, Z. & Kaliaguine, S., Surface analysis by ESCA of sulfite post-treated CTMP. J. Applied Polym. Sci., 39, pp. 31–42, 1990. Sernek, M., Kamke1, F.A. & Glasser, W.G., Comparative analysis of inactivated wood surface, Holzforschung, 58, pp. 22–31, 2004.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
199
Interface resistances in heat and moisture transport: semi-scale experimental analysis Z. Pavlík, J. Mihulka, J. Žumár, M. Pavlíková & R. Černý Department of Materials Engineering and Chemistry, Faculty of Civil Engineering, Czech Technical University in Prague, Czech Republic
Abstract A sophisticated semi-scale system which allows experimental simulation of heat and moisture transport in building structures is employed for analyzing heat and moisture transfer across interfaces in stone masonry formed by argillite walling blocks and lime-pozzolana mortar. The dimensions of investigated structure are close to a real wall but the measuring accuracy is the same as in a laboratory experiment. Combined relative humidity and temperature mini-sensors are utilized in continuous long-term monitoring. Measured temperature and relative humidity profiles provide information on the properties of the interface between the walling material and mortar which makes possible to identify the interface permeability. This parameter can then be used in computational models of heat and moisture transport in masonry which adds to the accuracy of model predictions. Keywords: heat and moisture transport, interface resistances, semi-scale experiment, stone masonry, argillite, lime-pozzolana mortar.
1 Introduction Computational modelling of coupled moisture and heat transport represents an effective tool for prediction of behaviour of building structures exposed to climatic load. For reliability of computational analysis, two requirements must be met. At first, mathematical-physical model describing the coupled moisture and heat transport must be formulated. From the theoretical point of view, this model should be sufficiently sophisticated and complex, and should take into account all the external and internal effects that affect the moisture and heat WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line) doi:10.2495/MC110181
200 Materials Characterisation V transport. Typical example of such complex models represents for instance hybrid model developed by Grunewald [1, 2] that takes into account both diffusion and convection effects on moisture and heat transport. Although these models describe the physical reality of coupled moisture and transport in sophisticated way, their practical application for hygrothermal analysis of building structures is often limited by the accuracy and availability of input parameters that represent the second limiting requirement of computational modelling. There are two types of input parameters which have to be known in advance for precise computational analysis. The first are initial and boundary conditions. In case of analysis of existing structure, for example by reason of intended reconstruction, initial conditions can be determined using on site analysis of moisture and temperature fields in the studied structure. Boundary conditions are of two types. The first of them are meteorological data for temperatures, relative humidities, rainfall and solar radiation, possibly also concentration of acidforming gases in the atmosphere. This type of data can be obtained from meteorologists in the form of so-called TRY (Test Reference Year) data which present certain average values over a sufficiently long time period. The second type of boundary conditions involves water content in the underground soil close to the studied building. These data can be obtained again by on site analysis [3]. The second type of input parameters are hygric and thermal transport and storage parameters of the materials of the structure which appear in water and heat mass balance equations implemented in the models. In case of study of coupled moisture and heat transport, these parameters include moisture diffusivity and diffusion coefficient of water vapour, sorption isotherms and water retention curves that can be optionally expressed as moisture potential curves, thermal conductivity and specific heat capacity. Within the application of computational modelling, one must take into account that all above given parameters are functions of both temperature and moisture. Therefore, there is a need to determine these parameters in dependence on moisture and temperature changes what is highly time-consuming, and for some types of materials practically unfeasible. On that account, one must assume diversion of computational simulations from reality, and the computer codes must be calibrated and validated using experimentally measured moisture and moisture profiles. Only after this validation is done, the computational modelling can be applied with sufficient accuracy. Specific problem for validation of computational codes solving the coupled moisture and heat transport represent composite materials and structures, where interface resistances crucially affect the reliability of modelling. Most classical models of heat and moisture transport in porous materials do not deal with moisture transport across interfaces between two porous materials in an explicit way. In practical calculations, the simplest solution is commonly used. It is assumed that there is ideal contact between the two materials, which means the equality of temperatures and moistures, taken as limits on the right and the left to the interface. However, this requirement can sometimes lead to dramatic but unclear moisture profiles in the regions close to the interface, particularly if the WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
201
bordering materials have considerably different transfer coefficients. Therefore, the condition that the macroscopic capillary pressure must be continuous across the interface is adopted in more sophisticated models, implying an ideal hydraulic contact. Then, at the interface between two different materials a moisture jump across the interface appears, which can be theoretically determined on the basis of the measured water accumulation functions of both materials. In the case a non-ideal hydraulic contact is formed on the interface due to the different pore size distributions of the adjacent porous materials, a jump of capillary pressure along the interface characterized by the interface permeability can appear. Typical example of non-ideal hydraulic contact of two materials represent masonry, where the water transport properties of walling blocks and mortar are different due to the differences in their pore size distribution. In order to decide which of the particular transport models is realistic for a specific interface, an experimental analysis is required. On that account, an experimental analysis of temperature and moisture transport across the material interfaces in stone masonry is studied in this paper to reveal the interface effects on the moisture and temperature distribution.
2 Semi-scale experiments The semi-scale experiments are presently very popular in verification and calibration of HAM (Heat, Air, and Moisture transport) models since they allow monitoring of hygrothermal changes in the studied structures in more detailed way and the certain specific cases can be studied what is not very common in the case of full test house measurement [4, 5]. The semi-scale measuring system for determination of temperature and moisture fields is designed in such a way that it simulates conditions, which are as close as possible to the real conditions on building site, but it still maintains its laboratory character, so that the expenses are kept considerably lower compared to a real test house. Also the accuracy of applied measuring methods for moisture content and temperature measurement is much higher compared to insitu measurements [6].
3 Experimental Within the performed experiments, the moisture and heat transport in the fragment of argillite wall was studied. Two separate semi-scale experiments were designed and performed in order to evaluate the effect of interface resistances on the transition between the argillite walling blocks and mortar. Within the first experiment, the heat transport was simulated, whereas the relative humidity was maintained on constant level. In the second experimental arrangement, relative humidity transport was monitored at constant temperature conditions. 3.1 Studied masonry and inbuilt materials Two materials typical for historical masonry in Central European territory were used. The researched fragment of the stone masonry consisted of argillite blocks WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
202 Materials Characterisation V and lime mortar with pozzolana admixture based on calcined kaoline mixed with milled mudstone. The composition of applied mortar is given in Table 1. Table 1: lime hydrate (kg) 200
Composition of applied plaster. pozzolana (kg) 50
sand 0 – 4 mm (kg) 750
w/d ratio (-) 0.23
In Table 1, w/d represents the water/dry substance ratio. Lime hydrate is product of lime kiln Čertovy schody, Inc., Czech Republic, whereas the silicious sand is coming from sand-pit Hlavačov, Czech Republic. The applied argillite is coming from quarry Džbán, Czech Republic. It is highly heterogeneous fine-pore material with anisotropic structure. It is characteristic by diversity in composition, whereas its main constituents are illite, kaolin, minerals of SiO2, spar and mica. Basic physical properties of both masonry materials are given in Table 2. Table 2: material mortar argillite
Basic properties of masonry materials.
bulk density (kg/m3) 1688 1353
matrix density (kg/m3) 2560 2235
total open porosity (-) 0.34 0.39
3.2 Experimental arrangement, measuring technology For simulation of moisture and heat transport in the investigated structure, a semi-scale system was used. The device (Figure 1) consists of two climatic chambers for simulation of climatic conditions, connecting tunnel, where the investigated structure is placed, and commercial devices for continuous monitoring of field variables as relative humidity, temperature, liquid moisture content, salinity, heat flux, etc. It is also equipped with optional sprinkling device for rainfall simulation. In the presented experiments, monitoring of relative humidity and temperature changes in the specific places of studied structure was done. For this purpose, sophisticated technique from Ahlborn was used. The accuracy of particular sensors was as follows: ±2% for capacitive relative humidity sensors applicable in the range of humidities 5-98%, for temperature sensors ± 0.4°C in the temperature range from –20°C to 0°C and ± 0.1°C in the range from 0°C to 70 °C. For testing the tightness of climatic chamber system, the anemometers for air flow velocity measurements were used. The whole measuring system was operated by a computer, including the climatic data entry into the particular climatic chambers. The details on the system including the measuring technology and sensors calibration can be found in [6]. The measuring process can be divided into several basic phases: walling the investigated structure, sensors’ installation into the built wall fragment, sample WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
Figure 1:
203
Climatic chamber system for semi-scale experiments.
positioning into the tunnel between the climatic chambers, climatic loading simulation, temperature and relative humidity monitoring, data evaluation. The walling of the studied structure was done in the standard way using dry mortar mix and wet technological process. The sensors for monitoring temperature and relative humidity were placed to the investigated construction to beforehand bored holes. The upper part of the bore opening was closed by silicon sealing. Placing of the sensors was done regarding to the study of interface moisture and heat transport (Figure 2). For a proper setting of climatic conditions, it was necessary to achieve nearsteady-state conditions in the studied wall. Therefore, there was necessary to dry freshly built wall to remove technological water. The drying was done at 50°C and 10% of relative humidity for one month. After that, the first experiment simulating the heat transport at constant moisture conditions was started. On one side of the studied wall, constant temperature of -9.5°C was maintained, whereas on the other side of the masonry fragment, constant temperature of 25°C was simulated. The relative humidity varied during the experiment in the range of ±10%, typically between 40 and 50% in respect to performance of climatic chambers. Duration of the first experiment was 84 days.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
204 Materials Characterisation V
Figure 2:
Sensors’ positioning and wall arrangement (dimensions in mm).
The second experiment was designed to simulate gaseous moisture transport without temperature effects. Also in this case, the climatic chambers’ parameters were limiting factor. On one side of the wall, relative humidity of 45% was simulated, whereas on the opposite wall side, 95% was reached. The temperature varied within the range of 25 to 28°C. Duration of this experiment was 81 days.
4 Results and discussion The temperature profiles measured in the first semi-scale experiment are shown in Figures 3–6. The results are presented in two cross sections (A-A, B-B), according to the scheme in Figure 2. The measured temperature profiles in Figures 3–6 are more or less linear. This indicates that steady state heat transport was reached. We can also observe the expected jumps in temperatures at the contacts of measured wall with ambient environment. They are caused by thermal resistances between the material surface and ambient air. The thermal resistances between the argillite blocks and mortar are manifested as well. The effects of interface resistances are not so distinct as the surface resistances as solid-solid heat transfer can be generally realized in a more ideal way than solid-gas transfer. Nevertheless, their significance is perceptible. WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
Figure 3:
Figure 4:
205
Temperature profiles in cross section A-A (first experiment).
Temperature profiles in cross section A-A (end of the first experiment).
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
206 Materials Characterisation V
Figure 5:
Figure 6:
Temperature profiles in cross section B-B (first experiment).
Temperature profiles in cross section B-B (end of the first experiment).
The relative profiles measured in the second semi-scale experiment are shown in Figures 7 and 8.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
207
Figure 7:
Relative humidity profiles in cross section A-A (second experiment).
Figure 8:
Relative humidity profiles in cross section B-B (second experiment).
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
208 Materials Characterisation V Apparently, in case of simulation of water vapour transport, similar features as in heat transport simulation are observed. The experimental relative humidity profiles reveal significant effects of interface resistances on gaseous moisture transport at the contact of argillite blocks and mortar. The surface resistances remarkably affect the water vapour distribution in the studied masonry as well. The steep parts of relative humidity profiles are highly important for calibration and validation of HAM models of transport phenomena in multilayered systems. For demonstration of temperature conditions within the water vapour transport simulation, Figure 9 is introduced. Clearly, the effect of interface resistances in the stone-mortar-stone transition zone and surface resistances can be observed even in the case of low temperature differences on opposite wall sides.
transition zone
Figure 9:
Temperature distribution in cross section B-B (second experiment).
5 Conclusions Experimental investigation of water vapour and heat transport across the interfaces in stone masonry was done in the conditions of 1-D semi-scale experiment. The climatic loading of the studied structure was chosen in such a way that water vapour transport under isothermal conditions and heat transport at constant relative humidity were simulated. The data obtained in both experiments confirmed the significant effect of both interface and surface resistances on water vapour and heat propagation, thus the necessity to include these effects in computational modelling of coupled moisture and heat transport. On the basis of measured data, determination of heat and water vapour resistances between particular materials’ layers is possible using methods of WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
209
inverse analysis. Also, the surface resistances on the contact of studied building structure with ambient air can be accessed. In this way, the results can contribute to higher accuracy of future computational modelling. Neglecting the behaviour of the transition zone between the particular materials forming stone masonry structures, on the other hand, may lead to possible improper design of renovation or construction solutions and subsequent material and structural damage.
Acknowledgement This research has been supported by the Czech Ministry of Education, Youth and Sports, under project No MSM 6840770031.
References [1] Grunewald, J., Diffusiver und konvektiver Stoff- und Energietransport in kapillarporösen Baustoffen, Ph.D. Thesis, TU Dresden, Dresden, 1997. [2] Černý, R. & Rovnaníková, P., Transport Processes in Concrete, 1st ed., Spon Press: London, 547 pp., 2002. [3] Pavlík, Z., Michálek, P., Pavlíková, M., Kopecká, I., Maxová, I. & Černý, R., Water and Salt Transport and Storage Properties of Mšené Sandstone. Construction and Building Materials, 22(22), pp. 1736-1748, 2008. [4] Krus, M., Rösler, D. & Sedlbauer, K., New model for the hygrothermal calculation of condensate on the external building surface. Research in Building Physics and Building Engineering, Fazio, Ge, Rao, Desmarais (eds.), Taylor & Francis Group, London, pp. 329-333, 2006. [5] Pavlík, Z., Mihulka, J., Žumár, J. & Černý, R., Experimental monitoring of moisture transfer across interfaces in brick masonry. Structural Faults and Repair 2010 [CD-ROM], Engineering Technics Press Edinburgh, Edinburgh, 2010. [6] Pavlík, Z., Pavlík, J., Jiřičková, M. & Černý, R., System for Testing the Hygrothermal Performance of Multi-Layered Building Envelopes. Journal of Thermal Envelope & Building Science, 25(1), pp. 239-249, 2002.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
This page intentionally left blank
Section 6 Mechanical characterisation and testing
This page intentionally left blank
Materials Characterisation V
213
Tension/compression test of auto-body steel sheets with the variation of the pre-strain and the strain rate G. H. Bae & H. Huh School of Mechanical, Aerospace and Systems Engineering, KAIST, Republic of Korea
Abstract This paper investigates the tension/compression hardening behaviour of autobody steel sheets with the variation of the pre-strain and the strain rate. To conduct tension/compression tests with the variation of the pre-strain and the strain rate, an experimental method was established by using a newly developed clamping device to suppress buckling of a specimen. The clamping device provides the supporting force from compression-type coil springs during the test with a conventional dynamic material fatigue testing machine. From experiments, the tension/compression hardening behaviour was observed with the variation of the pre-strain and the strain rate. Effects of the pre-strain and the strain rate on the hardening behaviour were also investigated based on the tension/compression test results. Keywords: tension/compression test, clamping device, pre-strain, strain rate.
1 Introduction Spring-back caused by the elastic recovery of the residual stress inside a formed part has been one of the most significant sources of defects in the sheet metal forming process in recent years. Spring-back predictability of numerical simulation, however, has not been satisfactory because its performance is not good enough to calculate accurate residual stress during the forming process. To improve spring-back predictability in numerical simulation, many researchers have been trying to use accurate information of the hardening behaviour of sheet materials in finite element analysis. The hardening behaviour of sheet materials
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line) doi:10.2495/MC110191
214 Materials Characterisation V has been measured with various mechanical testing methods in the quasi-static region [1–3]. It has been successful to predict spring-back accurately with numerical simulation by applying the precise hardening behaviour obtained from tension/compression tests at the quasi-static state [4, 5]. Numerical simulation still cannot accurately predict the final shape of the formed part after spring-back for a higher strain rate because of lack in information about the tension/compression hardening behaviour at high strain rates. To enhance spring-back predictability in numerical simulation for actual formed parts, the accurate hardening behaviour should be applied in finite element analysis by performing tension/compression tests with the variation of the strain rate as well as of the pre-strain. This paper investigates the tension/compression hardening behaviour of autobody steel sheets with the variation of the pre-strain and the strain rate. To conduct tension/compression tests, a simple clamping device was newly developed to suppress buckling of a specimen during the compression loading. The clamping device provides the side force from compression-type coil springs. The compression-type coil springs were selected for the desired clamping pressure calculated by the plate buckling theory adopted by Cao and Wang [6]. Based on the secant formula and the Euler method adopted by Boger et al. [2], specimen dimensions were also designed to prevent buckling along the width direction in the gauge section and along the longitudinal direction in the unclamped region. The strain in the gauge region of the specimen was measured directly by the digital image processing technique. The hardening behaviour of auto-body steel sheets was investigated precisely based on the tension/compression test results with the variation of the pre-strain and the strain rate.
2 Test preparation To obtain a larger compressive strain range in the tension/compression test, buckling of a specimen should be prevented by using a properly-designed specimen and by imposing a sufficient clamping force. Fig. 1 shows three representative buckling modes in the tension/compression test: (1) buckling in the thickness direction in the gauge region (T-buckling); (2) buckling in the
Figure 1:
Three representative buckling modes in the tension/compression test.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
215
unclamped region (L-buckling); (3) buckling in the width direction in the gauge region (W-buckling). T-buckling mode can be suppressed by imposing a sufficient clamping force on a specimen with a special clamping device. Lbuckling and W-buckling modes can be prevented by designing an appropriate specimen shape. 2.1 Specimen shape Boger et al. [2] used the secant formula and the Euler method to calculate the attainable compressive strain for L-buckling and W-buckling modes, respectively. When the flow curve is expressed by the Swift model, the attainable compressive strain for two buckling modes can be calculated as follows: 1 é ù êæç B ö÷n ú eL = e0 êç ÷÷ -1ú êèçW ø ú ëê ûú (1) eW =
p 2 nW 2 - e0 3G 2
(2)
where B and W are the specimen width in the gripping region and the gauge region, respectively. G is the gauge length of the specimen. ε0 and n are the plastic strain for the yield stress and the hardening exponent in the Swift model, respectively. If the gauge length of the specimen and the coefficients of the Swift model are given, the attainable compressive strains are a function of the gauge width. More detailed formulation procedure can be found in the reference [2]. The specimen design was performed using the above equations formulated for the attainable compressive strain of L-buckling and W-buckling. The selected steel sheets are SPCC and DP590 which are commonly used auto-body steel sheets. Strains at uniform elongation before necking initiates are approximately 0.15 and 0.12 for SPCC and DP590, respectively. These values are assigned as the maximum pre-strain for tension/compression tests without necking of a specimen. Table 1 shows the coefficients of the Swift model and r-values for SPCC and DP590 at a quasi-static state. A specimen shape for the uniaxial tensile test [7, 8] was utilized as reference specimen dimensions for tension/compression tests as shown in fig. 3(a). In order to reduce the required clamping force to suppress T-buckling, the gauge length is first reduced from 30 mm to 20 mm. The small clamping force is favourable for the reliable tension/compression test, which can reduce the frictional and biaxial effect on a specimen. The attainable compressive strain was plotted with respect to the gauge width as shown in fig. 2. From the plotted curves, the optimal gauge width can be selected as 8.8 mm for SPCC and 8.2 mm for DP590, respectively. To ensure a stable testing region for the two steel sheets, the final gauge width can be determined to be 8.8 mm as shown in fig. 3(b).
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
216 Materials Characterisation V Table 1:
Coefficients of the Swift model and r-values of SPCC and DP590. K [MPa]
ε0
n
r0
r45
r90
R
SPCC
555.772
0.01121
0.253
1.381
1.094
1.697
1.317
DP590
1079.261
0.00879
0.220
0.704
0.784
0.948
0.805
Attainable compressive strain
Material
0.4
L-buckling SPCC DP590
0.3
W-buckling SPCC DP590
0.2
SPCC, 0.15 DP590, 0.12
0.1 0.0
0
5
10
15
20
25
Width of the gauge region [mm]
Figure 2:
Attainable compressive strain with respect to the width at the gauge region for SPCC and DP590.
(a)
(b) Figure 3:
Specimen dimensions for the uniaxial tensile test and the tension/compression test: (a) uniaxial tension test [7, 8]; (b) tension/compression test (suggested).
2.2 Clamping force Cao and Wang [6] proposed an equation to calculate the blank holding force to suppress flange wrinkling in the sheet metal forming process. To calculate the clamping force to suppress T-buckling of a specimen, the plate buckling theory was also employed by assuming the gauge region of a specimen to be a rectangular plate model. The critical normal force required to suppress buckling of a specimen can be calculated by WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V n +1 éï ù æ 2u x ö÷ïüï êì ú ï ç e ln 1 ÷ í ý ç ê 0 ú çè G ÷øïþ 3KGWt êïîï ú ï F= ê ú n +1 n 4(n + 1)d ê 2 æ ct ú ö÷ æ 1 æ ö ö 1 t t 1 ç ÷ ç ÷ ê ú ç ê- G èçç 2 + e0 èçç m 2 d + 2 ø÷÷ø÷÷÷ èçç m 2 d + 2 ø÷÷ tan (md )ú ë û
217
(3)
2u x ( l - u x ) 2p ,d= G 2 u p 1+ 2R x where t and ux are the thickness of a specimen and a certain edge displacement. K, n and ε0 are coefficients of the Swift model. For the tension/compression test, the specimen thickness and width change in the gauge region during deformation. These values can be calculated based on anisotropy of material. where c =
1+ R
, m=
W = W0 -
t = t0 -
æ 2u ö W0 ln çç1- x ÷÷÷ 1 + R çè G ø
(4)
t0 R æç 2ux ö÷ ln ç1÷ 1 + R çè G ø÷
(5)
The critical normal force defined in eqn. (3) is now used to calculate the critical clamping force to suppress T-buckling in the tension/compression test. More detailed formulation procedure of the critical normal force can be found in the reference [6]. Fig. 4 shows the critical clamping force with respect to the compressive plastic strain. From the plotted curves, the required clamping force for SPCC and DP590 can be determined to be 2.003 kN and 2.275 kN, respectively.
Clamping force [kN]
10
G20W8.8 SPCC 8 DP590
DP590 0.12
SPCC 0.15
6 4 2 0 0.00
0.05
0.10
0.15
0.20
Compressive strain
Figure 4:
Required clamping force with respect to the compressive strain for SPCC and DP590.
2.3 Design of a new clamping device Based on specimen dimensions and the required clamping force, a new clamping device was developed to suppress T-buckling of specimens. Fig. 5(a) gives a schematic diagram for the clamping device developed. A controllable clamping WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
218 Materials Characterisation V system was suggested by using compression-type coil springs. Four coil springs on each side compress a specimen by fastening the bolts up to the desired deflection distance which is determined by the gap controlling bar. Before the tension/compression test starts, the gap controlling bar is uninstalled from the clamping device in order to allow thickness change of the specimen during the tension/compression test. For lubrication between the specimen and the clamping plate, a Teflon film (3M 5490 model) with a thickness of 0.09 mm is attached on the clamping plate. A compression-type coil spring was selected based on the clamping force calculated in Section 2.2 for SPCC and DP590. The maximum clamping force was selected at 4 kN to ensure performance of the device although the required clamping forces were 2.003 kN for SPCC and 2.275 kN for DP590. To provide the required clamping forces, Φ16L40 coil spring was selected for the clamping device. The maximum load and the spring constant of the selected coil spring are 1.265 kN and 0.158 kN/mm, respectively. The clamping device was manufactured with the specified dimensions based on the specimen shape and the spring size. As shown in fig. 5(b), the springloaded clamping device has a small and simple structure with light weight. The clamping force is controllable by determining the deflection distance of the coil springs. These advantages are favourable for easy and convenient tension/compression tests at high strain rates. Specimen Teflon film
Coil springs Gap controlling bar
Bolt
Nut
Clamping plate
Compression plate
(a) Figure 5:
(b)
Spring-loaded clamping device for the tension/compression test: (a) schematic diagram; (b) manufactured clamping device.
3 Experiments 3.1 Test material The tension/compression test of auto-body steel sheets was conducted with the variation of the pre-strain and the strain rate. Test materials were SPCC and DP590 with a thickness of 1.2 mm. The chemical composition of the sheet metals is presented in table 2. Table 2: Material SPCC DP590
Chemical composition of SPCC and DP590 [wt.%]. C 0.085 0.326
Mn 0.420 1.700
Si 0.040 0.123
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
P 0.0012 0.0162
S 0.0150 0.0009
Materials Characterisation V
219
3.2 Testing machine A clamped specimen is gripped by the dynamic material fatigue testing machine, INSTRON8801, to perform tension/compression tests with the variation of the pre-strain and the strain rate. The dynamic material fatigue testing machine shown in fig. 6 has a maximum stroke of ±75 mm in the vertical direction actuated by a hydraulic system. The load cell installed in the testing machine can measure a load of up to ±100 kN. The gripper is also actuated by a hydraulic system.
Figure 6:
Tension/compression testing system utilizing the dynamic material fatigue testing machine, INSTRON8801, and the spring-loaded clamping device.
True strain in the T/C test
0.15
Actuator displacement: 3 mm Strain rate: 0.001 /sec
0.10
0.05 SPCC DP590 0.00
0
1
2
3
4
5
6
Total actuator movement [mm]
(a) Figure 7:
(b)
Strain measuring method using the digital image processing technique with the high speed camera: (a) recording of the deformation history; (b) measured strain during the tension/compression test.
3.3 Strain measurement Because the specimen is fully covered by the clamping plates, conventional extensometers cannot be employed to measure the strain in the side of the gauge region during the tension/compression test. As an alternative, a digital image processing technique was utilized to measure the strain directly. Before the tension/compression test, a speckled pattern was sprayed on the side wall of the specimen as shown in fig. 7(a). A high speed camera captures a series of frames for deformation during the tension/compression test. The strain can be measured WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
220 Materials Characterisation V from a graph in fig. 7(b) obtained by analyzing recorded frames using the digital image processing technique. 3.4 Tension/compression test Stable tension/compression testing conditions were established by comparing the response of the testing machine with the variation of the actuator displacement and the strain rate. Table 3 presents experimental conditions for the tension /compression test. From the experiments, load-displacement curves and strains in the gauge region were determined for various pre-strains and strain rates. The tension/compression hardening curves can be obtained after correction procedures to subtract the frictional and biaxial effects induced by the clamping force. Table 3:
Material
SPCC
DP590
Experimental table for tension/compression tests of SPCC and DP590. Displacement [mm] 1.0 2.0 3.0 4.0 1.0 2.0 3.0
0.001 O O O O O O O
Strain rate [/sec] 0.01 0.1 O O O O O O O O O O O O O O
1.0 X X O O X X O
3.5 Correction of acquired load All acquired stress−strain results require corrections for frictional and biaxial effects induced by the clamping force, Fc , because the specimen is compressed by the spring-loaded clamping device in the thickness direction to suppress T-buckling. The measured force from the load cell, Fm , is the addition of the force from the specimen deformation, Fd , and the frictional force, Ff .
Fd = Fm - Ff
(6)
The friction behaviour is represented by the Coulomb friction law as follows [1, 2]:
Ff = m Fc
(7)
The frictional conditions can be changed according to the material, the geometrical change of a specimen and the accumulated damage to the Teflon film. The friction coefficient is generally known to be in the range of 0.03~0.09 when the Teflon film is used for tension/compression tests [2–4]. The friction coefficients for SPCC and DP590 are selected as 0.06 and 0.08, respectively, to correct the acquired curve obtained from the tensile test with clamped into a WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
221
corrected curve by comparison with the acquired curve obtained from a tensile test with unclamped. In order to subtract the through-thickness stress, σt , induced by the clamping force, the von Mises yield function is utilized to calculate the effective stress corrected from the biaxial effect.
1é 2 (sm - st ) + sm2 + st2 ùúû ê ë 2
s=
(8)
where σm is the measured stress from the load cell. After both frictional and biaxial corrections, the flow curve of the clamped specimen agrees well with that of the unclamped specimen as shown in fig. 8.
400
SPCC
1000
Strain rate = 0.001 /sec Clamping force = 2.003 kN = 0.06
300 200 Acquired curve (unclamped) Acquired curve (clamped) Corrected curve
100 0 0.00
0.05
0.10
True stress [MPa]
True stress [MPa]
500
800
DP590 Strain rate = 0.001 /sec Clamping force = 2.275 kN = 0.08
600 400 Acquired curve (unclamped) Acquired curve (clamped) Corrected curve
200 0 0.00
0.15
0.05
True strain
(a) Figure 8:
Frictional and biaxial corrections in the uniaxial tension state: (a) SPCC, 0.001/sec; (b) DP590, 0.001/sec. 500
SPCC, 0.001 /sec
250
Simple tension Displacement (pre-strain) 1 mm (0.040) 2 mm (0.080) 3 mm (0.120) 4 mm (0.162)
0 -250 -500 0.00
0.05
0.10
0.15
SPCC, 0.01 /sec
250
Simple tension Displacement (pre-strain) 1 mm (0.037) 2 mm (0.075) 3 mm (0.114) 4 mm (0.156)
0 -250 -500 0.00
0.20
0.05
True strain
Simple tension Displacement (pre-strain) 1 mm (0.038) 2 mm (0.077) 3 mm (0.118) 4 mm (0.161)
0 -250
0.10
0.15
True strain
0.20
SPCC, 1.0 /sec
250
Simple tension Displacement (pre-strain) 3 mm (0.116) 4 mm (0.159)
0 -250 -500 0.00
0.05
0.10
0.15
0.20
True strain
(c) Figure 9:
0.20
True stress [MPa]
True stress [MPa]
SPCC, 0.1 /sec
0.05
0.15
(b) 500
250
-500 0.00
0.10 True strain
(a) 500
0.15
(b)
True stress [MPa]
True stress [MPa]
500
0.10
True strain
(d)
Tension/compression test results of SPCC at various strain rates: (a) 0.001/sec; (b) 0.01/sec; (c) 0.1/sec; (d) 1.0/sec.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
222 Materials Characterisation V 1000
DP590, 0.001 /sec
500
Simple tension Displacement (pre-strain) 1 mm (0.032) 2 mm (0.070) 3 mm (0.112)
0 -500 -1000 0.00
0.05
0.10
True stress [MPa]
True stress [MPa]
1000
DP590, 0.01 /sec
500
Simple tension Displacement (pre-strain) 1 mm (0.034) 2 mm (0.073) 3 mm (0.118)
0 -500 -1000 0.00
0.15
0.05
True strain
(a) DP590, 0.1 /sec
500
Simple tension Displacement (pre-strain) 1 mm (0.029) 2 mm (0.066) 3 mm (0.109)
0 -500 -1000 0.00
0.05
0.10
DP590, 1.0 /sec
500
Simple tension Displacement (pre-strain) 3 mm (0.106)
0 -500 -1000 0.00
0.15
0.05
True strain
Tension/compression test results of DP590 at various strain rates: (a) 0.001/sec; (b) 0.01/sec; (c) 0.1/sec; (d) 1.0/sec. 500
SPCC, 1 mm
250
Strain rate [/sec] 0.1 0.01 0.001
0 -250 -500 0.00
0.05
0.10
0.15
SPCC, 2 mm
250
Strain rate [/sec] 0.1 0.01 0.001
0 -250 -500 0.00
0.20
0.05
True strain
Strain rate [/sec] 1.0 0.1 0.01 0.001
0 -250
0.10
0.15
0.20
True stress [MPa]
True stress [MPa]
250
0.05
0.20
SPCC, 4 mm
250
Strain rate [/sec] 1.0 0.1 0.01 0.001
0 -250 -500 0.00
0.05
True strain
(c) Figure 11:
0.15
(b) 500
SPCC, 3 mm
-500 0.00
0.10 True strain
(a) 500
0.15
(d)
True stress [MPa]
True stress [MPa]
500
0.10
True strain
(c) Figure 10:
0.15
(b) 1000 True stress [MPa]
True stress [MPa]
1000
0.10
True strain
0.10
0.15
0.20
True strain
(d)
Tension/compression test results of SPCC at various actuator displacements (pre-strains): (a) 0.001/sec; (b) 0.01/sec; (c) 0.1/sec; (d) 1.0/sec.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
1000
DP590, 1 mm
500
Strain rate [/sec] 0.1 0.01 0.001
0 -500 -1000 0.00
0.05
0.10
True stress [MPa]
True stress [MPa]
1000
DP590, 2 mm
500
Strain rate [/sec] 0.1 0.01 0.001
0 -500 -1000 0.00
0.15
0.05
True strain
0.10
0.15
True strain
(a)
(b)
1000 True stress [MPa]
223
DP590, 3 mm
500
Strain rate [/sec] 1.0 0.1 0.01 0.001
0 -500 -1000 0.00
0.05
0.10
0.15
True strain
(c) Figure 12:
Tension/compression test results of DP590 at various actuator displacements (pre-strains): (a) 0.001/sec; (b) 0.01/sec; (c) 0.1/sec.
4 Hardening behaviour True stress−true strain curves were obtained after frictional and biaxial corrections, as shown in figs. 9 and 10. Hardening curves expand with the increase of the pre-strain at various strain rates. To investigate the strain rate effect on the hardening behaviour, true stress−true strain curves were redrawn with the variation of the strain rate as shown in figs. 11 and 12. Hardening curves also expand with the increase of the strain rate at various pre-strains. It is well known that the strain and the strain-rate hardening are caused by micro-structure changes inside the materials. To confirm this general expectation, Huh et al. [9] demonstrated by observing TEM experimental results that the mechanism of the strain-rate hardening is the change of the dislocation structures and the increase of the dislocation density. The mechanism of the hardening behaviour during tension/compression loading also can be explained by the micro-structure change of steel sheets caused by the pre-strain and the strain rate effect. From the experiments, it is clear that the tension/compression hardening behaviour is changed by the strain rate as well as by the pre-strain. This means that the prestrain and the strain rate effect on the tension/compression hardening behaviour should be considered simultaneously to improve spring-back predictability in numerical sheet metal forming simulation.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
224 Materials Characterisation V
5 Conclusions This paper investigates tension/compression hardening behaviour by performing tension/compression tests using a newly developed spring-loaded clamping device. Contributions in this paper are summarized as follows: 1) A specimen shape was designed with the secant formula and the Euler method adopted by Boger et al. [2] to prevent L-buckling and W-buckling in the desired tension/compression test range. The clamping force to suppress Tbuckling was calculated with the plate buckling theory. A spring-loaded clamping device was newly developed to suppress T-buckling in the tension/compression test. From experiments, it is noted that the dimension of a specimen and the choice of compression-type coil springs are appropriate to execute reliable tension/compression tests. 2) The strain in the gauge region of a clamped specimen was measured directly by using a digital image processing technique in order to overcome the inherent structural problems in the use of conventional contact-type extensometers. The measuring scheme using the digital image processing technique can provide accurate strain history in the gauge region of a clamped specimen during tension/compression tests. 3) With the test conditions established, hardening curves were reliably obtained for various pre-strains and strain rates. The tension/compression hardening curves were expanded with the increase of the strain rate as well as with the increase of the pre-strain. The mechanism of the pre-strain and strain rate hardening was explained by micro-structure changes, such as dislocation structures and dislocation density. Experiments provide good information of the hardening behaviour with respect to the pre-strain and the strain rate. 4) The newly developed clamping device can be directly applied to tension/compression tests at high strain rates because of its compactness and convenience to control the clamping force with various tensile testing machines. This device guarantees the response of the clamping pressure in real time when the clamping condition is changed by the specimen deformation during tension/compression tests.
References [1] Lee, M. G., Kim, D., Kim, C., Wenner, M. L., Wagoner, R. H. & Chung, K., Spring–back evaluation of automotive sheets based on isotropic–kinematic hardening laws and non–quadratic anisotropic yield functions, part II: characterization of material properties. Int. J. Plast., 21, pp. 883–914, 2005. [2] Boger, R. K., Wagoner, R. H., Barlat, F., Lee, M.–G. & Chung, K., Continuous, large strain, tension–compression testing of sheet material. Int. J. Plast., 21, pp. 2319–2343, 2005. [3] Cao, J., Lee, W., Cheng, H. S., Seniw, M., Wang, H. P. & Chung, K., Experimental and numerical investigation of combined isotropic–kinematic hardening behavior of sheet metals. Int. J. Plast., 25, pp. 942–972, 2009.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
225
[4] Lee, M. G., Kim, D., Kim, C., Wenner, M. L. & Chung, K., Spring–back evaluation of automotive sheets based on isotropic–kinematic hardening laws and non–quadratic anisotropic yield functions. part III: applications. Int. J. Plast., 21, pp. 915–953, 2005. [5] Taherizadeh, A., Ghaei, A., Green, D. E. & Altenhof, W. J., Finite element simulation of springback for a channel draw process with drawbead using different hardening models. Int. J. Mech. Sci., 51, pp. 314–325, 2009. [6] Cao, J. & Wang, X., An analytical model for plate wrinkling under tri–axial loading and its application. Int. J. Mech. Sci., 42, pp. 617–633, 2000. [7] Huh, H., Kim, S. B., Song, J. H. & Lim, J. H., Dynamic tensile characteristics of TRIP–type and DP–type steel sheets for an auto–body. Int. J. Mech. Sci., 50, pp. 918–931, 2008. [8] Huh, H., Lim, J. H. & Park, S. H., High speed tensile test of steel sheets for the stress−strain curve at the intermediate strain rate, Int. J. Automotive Technol., 10(2), pp. 195–204, 2009. [9] Huh, H., Yoon, J. H., Park, C. G., Kang, J. S., Huh, M. Y. & Kang, H. G, Correlation of microscopic structures to the strain rate hardening of SPCC steel. Int. J. Mech. Sci., 52, pp. 745–753, 2010.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
This page intentionally left blank
Materials Characterisation V
227
Definition of averaged elastic-plastic characteristics of sandwich panel structures I. I. Zakirov, V. N. Paimushin & I. M. Zakirov A. N. Tupolev State Technical University, Kazan, Russia
Abstract Analytic formulas for averaged elastic and strength characteristics of folded structure’s determination are identified. They are based on the introduction of a hypothesis about momentless action of its elements (web covers) in a charge operation and on phased deletion from action ex-post buckling. Dimensionless coefficients contained in the structural formulas, which appear on the solution process of formulated problems, are to be determined from experimental evidence on tension, compression and pure shear in two-planes. In a tensioncompression process the folded structures should be considered heterogeneous material with averaged plasto-elastic behaviors, which has relations between averaged voltages and averaged appropriate deformations and has a deformations stepwise change area (conditional plastic flow). Keywords: folded structures, cell of cyclicity, averaged elastic and strength characteristics, modulus of elasticity, deformation curve, structural formulas, theoretical and experimental method.
1
Introduction
In many cases, the use of composite materials is the most efficient in constructions (three-layer or multilayer) with cores, which have good mechanical strength, rigidity, vibration, sound and heat insulation characteristics. These characteristics of cores of some structures have been well studied today. A large number of publications, a detailed analysis of which is contained, in particular, in a review article [1] are devoted to their research. Most often in three-layer and multilayer structures, the cores have a honeycomb ([2, 10] etc.) structure. In the calculations of such structures the real cores are usually replaced by some homogeneous continuum, which averaged WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line) doi:10.2495/MC110201
228 Materials Characterisation V mechanical and thermal characteristics are determined by the principle of equivalence of real and replacing core [3–5]. At such replacement of real honeycomb with orthotropic continuum for the strength calculations averaged elastic module in the tangential directions and the shear modulus in the tangential plane are set equal to zero, and the elastic modulus in the transverse direction and shear module in the transverse plane is determined by theoretical or experimental methods [8, 10]. Formulas for their determination are given, in particular, in references [3, 4] that, as a first approximation of formulas, have been obtained by the energy method [3], or the method of displacements and forces [6]. Along with these, there are other theoretical methods for determining the averaged elastic characteristics of periodic structures based, in particular, on the method of averaging the solution of problems by Bakhvalov and Ponasenko [7]. In recent years, ways of production of the folded structures from flat sheets cores [11, 12] develop very intensively. Such cores compared with honeycomb cores have a number of fundamental differences and advantages. In this paper we consider one of these folded structures. We develop relevant theoretical and experimental method to determine the homogenized elastic and strength characteristics of which. 1.1 Formulation of the problem Consider folded structure with a quadriradiare core [11], formed from a flat sheet blank with thickness t by means of its gradual transformation into a state of
relief, as shown in Figure 1a. We assume that, in general, l1 l2 , 1 2 and all four sloping walls 1,1, 2, 2 of the periodicity cells isolated from the core are a parallelogram, symmetrically placed relative to the plane the middle planes
x01 z . Therefore,
k of each of them reasonable classified as a local oblique
xk 0k yk and xk 0k yk , in which (Fig. 1a) xk xk and the radius vectors of the points k given by a representation coordinate system
r xk e1 yk e2 , r k
k
In them unit vectors
k
k
k k xk e1 yk e2
(1.1)
ei k , ei k satisfies the equations
ei ei 1, ei ei k
k
k
k
1, e1 e2 e1 e2 cos k k
k
k
k
(1.2)
which according to Fig. 1a and 1b with unit vectors i , j , k of orthogonal Cartesian coordinate system related with dependencies
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
229
ie1 k ie1 k cos k , ie2 k ie2 k sin
je2 k cos , je2 k cos , je1 k je1 k 0
ke11 ke11 sin 1 , ke1 2 ke1 2 sin 2
(1.3)
ke2 k ke2 k 0
(а)
(b) Figure 1. Parameterizations (1.1) correspond to the basis vectors (for the elements
1 and 2 compiled following relations must be marked with strokes) r1 r k
k
xk e1 , r2 r k
k
k
yk e2
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
k
(1.4)
230 Materials Characterisation V co- and contravariant components of metric tensor aii ri ri 1, a12 a21 r1 r2 cos k k
k
aiik
k
k
k
k
k
k
(1.5)
k 12
aii cos 1 a , a12k a21k 2 k 2 a k sin k a k sin k
k k
k k
where a k a11 a22 a12 a12 1 cos
2
(1.6)
k sin 2 k .
For the displacement vectors of core walls we accept following representations U k u k e1k v k e2k w k m k z k k e1k k e2k
t 2 z k t 2
(1.7)
eik are the reciprocal basis vectors satisfying the conditions
where
eik e j ij ( ii 1, ij 0 where i j ), aijk eik ejk , m k are the unit k
vectors
normal
to
the
plane
k ,
satisfying
the
equations
ei m 0, m m 1 . k
k
k
k
According to the Kirchhoff-Love model, with an average bending for the functions
k
and
k
will take place following dependences.
k w,xk , k w,yk where
, x
xk . By using these relations for the covariant
components of the tangential deformations are established following relations.
ij ij k z k æijk zk
(1.8)
where
1 2 k k k k u, y v,x w,x w,y
1 2
11 k u,xk w,xk w,xk , 22 k v,yk w,yk w,yk k
212
k k æ11 w,xxk , æ22k w,yyk , æ12 w,xyk
(1.9)
(1.10)
Under the assumption of linear elastic behavior of core material when it is loading the contravariant components of the tension tensors in the walls with deformation tensor components
ij zk
ijk
are related by the elasticity
correlation of the following form WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
ijk
E z Eijml k k ml 2 1 v
231
(1.11)
where E , v are the Young’s modulus of elasticity and Poisson's ratio of wall ijml
material, and E k is the tetravalent tensor of the elastic characteristics, components of which for an isotropic material in plane stress state are defined by correlations
Eijml vaijk amlk k
1 v im jl a k a k ailk ajm k 2
(1.12)
Substitution of formulas (1.6) to the correlations (1.12) leads to the formulas v sin 2 k cos 2 k 1 E1111 E2222 , E1122 E2211 , k k k k 2 sin k sin 4 k
E1112 E1121 E1211 E2111 E1222 E2122 E2212 k k k k k k k E2221 k
cos k , sin 4 k
E1212 E2121 E1221 E2112 k k k k
1 cos 2 k v sin 2 k 2sin 4 k
(1.13)
We assume that under the loading of the core in the sandwich and multilayer structural elements in its walls because of their small thickness realized the flat and membranaceous stress-strain state. In this case, accumulated in them potential energy of deformation is permissible to calculate by the following formulas l d 1k k (1.14) sin k dxk dyk Ï k t 11k 11 k 2 12k 12 k 22k 22 200
where d k
ak dxk dyk sin k dxk dyk is the area of infinitely small element of the middle plane of k -th core wall, and the components of the deformation
ij k are defined by correlations (1.9). These correlations are written
in linear approximation by dropping the nonlinear terms. After the transition from a flat membrane state in the perturbed moment state possible due to a loss of stability of the flat form of equilibrium. To study the perturbed neutral equilibrium state we will use the variational equation of the Ritz method, compiled on the basis of relations (1.8)–(1.11). In the space V of a selected cell of periodicity, referred to orthogonal Cartesian coordinates x, y and z , averaged elastic properties of the core should be considered orthotropic. The elastic modulus of the first kind Ez in the direction z and the modules of the transverse shifts Gxz , G yz in the planes xz and
yz must be determined within the bounds of the model of a transversely WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
232 Materials Characterisation V soft core. If the displacement vector of any point M V provide an expansion U ui vj wk and in cross-sections x const , y const , z const introduce the averaged voltage
xz , yz , z
the accumulated in the volume
strain energy will be determined by the expression
Ï
1 xz xz yz yz z z dV 2 V
where for calculation the transverse shear strain
xz , yz
(1.15)
z
and axial strain
of
the homogenized medium in the linear approximation we have the kinematic relations
xz
u w v w w , yz , z z x z y z
(1.16)
Using of the principle of equivalence of real and replacing conditional core with average elastic and strength characteristics requires to preparation of the equality.
Ï Ï
1
Ï
2
Ï
1
Ï
2
(1.17)
which will be used for further calculations.
2 Critical values of the tensions formed in the walls of the core when it is loading Assume that in walls of the core in terms of loading formed homogeneous in their middle plane tensions
11k , 22k
and
12k .
To produce the approximate structural formulas for using of theoretical and experimental method [8], the edges of the walls xk 0, xk lk , yk 0, yk d will be assumed simply supported. Whether for these or other types of loading in the walls of the core forms only compressive tensions
11k
and
22k ,
and
12k 0 ,
when taken under
considerationт pinning the edges of the walls to construct the solution of the stability function w
k
can be represented as
k w k Amn sin
m xk m yk sin lk d
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
(2.1)
Materials Characterisation V
233
where m 1, 2,..., n 1, 2,... are the wave numbers. Then with the introduction of the dimensionless parameters
lk , d
k
11k it is possible to receive the formula
For definition of critical value
k
Kx
11 k
k 22k 11k .
Dk 2E t k K x Dk , Dk sin 4 k 12 1 v 2 lk
2
(2.2)
where k
Kx
m
2
k2 n 2 4m 2 n 2 k2 cos 2 k 2
m k n 2
2 k
2
, Dk
to construct the solution of the stability function w
represented as
w k W k sin In a case
(2.3)
11k 0 , 12k 0
In general, the subcritical tension state, when
22k 0
Dk sin 4 k
11k 22k 0 ,
critical tension
12k êð
êð k 5.34 4
yk lk
sin
k
and
can be
yk kk xk
(2.4)
d
using the (2.4) for determination the minimum
it possible to receive an approximate formula
lk2 2 Dk 1 cos k d2
2 1 cos k 1 cos 2 k
(2.5)
3 Determination of elastic and strength characteristics Displacement vector in a space V of core we can represent in the form
z (3.1) u0 i v0 j w0 k , 0 z h h where u0 const , v0 const and w0 const – moving the points of the z h, with a net in-plane shear boundary plane
U
x01 z u0 0, v0 0, w0 0 , y 01 z u0 0, v0 0, w0 0
and
tension-
k 1, 2,1, 2
(3.2)
compression in the direction of the axis 01 z . We denote
U 0 k u0 k e1k v0 k e2k w0 k m k
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
234 Materials Characterisation V displacement vectors of faces x1 x1 l1 and x2 x2 0 walls of the filler, and for the displacements U
k
of the points M k k we take the
approximations
x1 1 1 x1 1 2 x2 2 U 0 , U U 0 , U 1 U 0 l1 l1 l2 x 2 1 2 U 0 l2
1 U
U
2
(3.3)
The condition of equivalence of the real and the conventional core requires the satisfaction of the equality k k k k k (3.4) u0 i v0 j w0 k U 0 u0 e1k v0 e 2k w0 m k 1, 2,1, 2
k
Scalar multiplication of both sides by vectors ei
and using (1.2) (1.3), we
arrive to the dependencies
u0 u0 cos 1 w0 sin 1 , v0 u0 sin v0 cos 1
1
u01 u0 cos 1 w0 sin 1 , v01 u0 sin v0 cos
u0 u0 cos 2 w0 sin 2 , v0 u0 sin v0 cos 2
(3.5)
2
u0 u0 cos 2 w0 sin 2 , v0 u0 sin v0 cos 2
2
In component form the vector equalities (3.4) lead to the approximating functions
u 1
x1 1 1 x1 1 1 x1 1 1 x1 1 u0 , v v0 , u u0 , v v0 l1 l1 l1 l1
x 2 2 x 2 2 u 1 2 u0 , v 1 2 v0 l2 l2
(3.6)
x x u 2 1 2 u0 2 , v 2 1 2 v0 2 l2 l2 By the substitution into the linear kinematic relations
11 k u,xk , 22 k v,yk , 212 k u,yk v,xk
(3.7)
compute the components of deformation in the walls of the core in the membrane approximation. WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
235
Compiled on the basis of the relations, we consider the case of the periodicity cell of core by tension and compression along the axis z w0 0, uo v0 0 in the case of loading we can get formulas:
11 k 11 k sin 2 k
w0 k k k k , 22 22 12 12 0 h
E E sin 2 k w0 1111 k E 11 1 v2 k 1 v 2 sin 4 k h E E2211 22k 11 k v sin 2 k cos 2 k 11k k 2 1 v 2 E 1211 k E 11 cos k 11k 12k 1 v2 k
(3.8)
11 k
(3.9)
From which follows that in the condition of tensile fracture of core ( w0 0 ) is possible because of the destruction of the adhesive layer connecting the core with a bearing layer, and while compression ( w0 0 ) due to loss of stability of the walls in her bilateral compression and shear. When we approximating the displacements of the averaged core (3.1), it formed the strains and stresses
w0 u v w , xz 0 , yz 0 , 33 Ez 0 h h h h (3.10) u0 23 v0 13 Gxz , G yz h h where the homogenized elastic modules Ez , Gxz and G yz must be determined.
z
By using the relations (3.10) in accordance with expression (1.15) for calculation Ï we arrive at the formulas in the case u0 v0 0
Ï hd ctg1 ctg 2 sin 33 w0
(3.11)
Ï d ctg1 ctg 2 sin Ez w
(3.12)
2 0
and by using relations (3.8) and (3.9), in accordance with (1.14) for calculation Ï k we arrive to the formulas
Ï
Ï
sin k sin k w0 k 1, 2,1, 2 2
(3.13)
sin 3 k w02 E td k 1, 2,1, 2 1 v 2 2sin 3 k h
(3.14)
k
11k td
k
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
236 Materials Characterisation V By substituting relations (3.12) , (3.14) from equation (1.17) we can obtain a formula to determine Ez
Ez
2 sin 3 k E t 1 v 2 h ctg1 ctg 2 k 1 sin 3 k
(3.15)
and using (3.13) from (1.17) we can obtain the following dependence
33
2 t 11k sin k sin k h sin ctg1 ctg 2 k 1
In the case of the maximum tensile stresses
11k , 22k , 12k
(3.16)
are largest in the
k sin 2 k sin 4 k .The process of destruction in it begins when certain tensile strength is reached. As in real cores, angles k and k do not have significant difference, it is possible to 11 11 assume that when the strength reaches 1 or 2 the simultaneous failure of all
wall, in which the maximum value is
four walls of the core begins. Under these assumptions to determine the limit
33
at which the core is destroyed we can create equality
33
2 t sin k sink h sin ctg1 ctg 2 k 1
(3.17)
33 From it by the experimentally determined value determined the quantity
. When compressing of the core begins its destruction due to the destruction of the buckling of the wall, which formed the voltage
k v sin 2 k cos 2 k
11k
and cos k
at a given dimensions
12k 11k
reaches a critical value
11 k
k
Kx
2
t 2 E sin 2 k t k Kx (3.18) 12 1 v 2 sin 4 k lk 12 1 v 2 sin 4 k h
2E
2
Assuming further that for the real core loss of stability of a wall immediately causes a loss of stability of a second wall, we form the equation
11k 11k ,
from which to determine the lateral deformation at the moment of loss of stability [9] by using relations (3.9) and (3.18) follows the formula
2 t w0 K x 12 h h
2
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
(3.19)
Materials Characterisation V
1
2
where K x K x
237
K x . However
2E t K x 2 12 1 v sin ctg1 ctg 2 h 33
For given parameters E , , t h , k and determined value of critical stress
k of
33 of
3
2
sin 3 k
sin k 1
3
k
(3.20)
the core the experimentally
the derived formula (3.20) is
determined by a dimensionless coefficient K x . Elastic and strength characteristics of the core in a shift in planes XoZ YoZ are defined by analogy with the stated above.
4
Conclusions
Theoretical and the experimental methods presented in this article allow to define the averaged elastic and strength characteristics of sandwich folded fillers on the basis of use of analytical expressions for definition of the critical tensions formed in walls of a filler while loading.
References [1]
[2] [3] [4] [5]
[6] [7] [8]
[9]
Noor A. K., Burton W.S., Bert Ch. W., Computational models for sandwich panels and shells. Applied Mechanics Reviews, 1996, V. 49, ¹ 3, p. 155-199. Bersudsky V.E., Krisin V.N., Forest S., Manufacturing technology of cellular aircraft structures. - Moscow: Mashinostroenie, 1975. 216s. Aleksandrov A.Y., Bryukker L.E., Kurshin L.M., The calculation of sandwich panels / M. Oborongiz, 1960. 272s. Panin V.F., Gladkov Yu.A., Constructions with cores. Directory. Moscow: Mashinostroenie. 1991. 272s. Kryutchenko V.E., Analysis of optimum insulation properties of sandwich plates with honeycomb core. /Mechanics of Composite Materials. - 1993, T.29.- № 6.- S.835-839. Relsey S., Gellatly H. and Clark B., The shear modulus of foil honeycomb cores. Aircraft Engineering. 1958. V.30. № 356. P.294-302. Bakhvalov N.S., Ponasenko G.P., Averaging of the processes in periodic media.- Moscow: Nauka, 1984. 352 c. Sachenkov A.V., Theoretical and experimental method for studying the stability of plates and shells. Investigations on the theory of plates and shells. Kazan: Izdatel'stvo Kazan. State. Univ. - 1970, Vol. 6.7.- S. 391433. Volmir AS Stability of elastic systems. M.: Gos. out of Sci. literature. 1963. 880 sec.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
238 Materials Characterisation V [10]
[11]
[12]
Akishev N.I., Zakirov I.I., Paimushin V.N., Shishov M.A., Theoreticalexperimental method for determining the averaged elastic and strength characteristics of honeycomb of sandwich structures. Mechanics of Composite Materials. - 2011 (in press). Zakirov I.M., Alekseev, K.A., Akishev N.I. Kayumov R.A., NikitinA.V., Zakirov I.I., Manufacturing of sandwich panels with folded core of polymer paper Kazan: Publishing house «Fan» 2009 Zakirov I.M., Alekseev K.A., Determining of the parameters of the four radial helical folded structure / IVUZ “Aviatcionnaya technika” 2005
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
239
Hot deformation and mechanical properties of P/M Al special M. Tercelj, P. Cvahte, I. Perus & G. Kugler Faculty of Natural Science and Engineering, University of Ljubljana, Slovenia
Abstract Almost pure aluminum composite, P/M Al-special, with an average grain diameter of 1μm with oxides on the grain surface was produced by a powder metallurgy route. Billets were extruded on an indirect press at various temperatures and ram-speeds. Specimens made from an extruded profile were additionally annealed at 300ºC at various times. Hot workability of these various initial states was studied by hot compression tests in the strain rate range 0.01– 10 s-1, the temperature range 300–580ºC up to a strain of 0.9. Additionally tensile tests were carried out to determine the mechanical properties on specimens previously annealed at 300ºC. The oxide surface prevents grains coarsening during hot deformation and consequently excellent mechanical properties of the specimens annealed at a temperature of 300ºC were obtained. Keywords: P/M 1080 Al composite, hot extrusion, annealing, hot compression, mechanical properties.
1 Introduction Demands for weight reduction of parts applied in cars and aircrafts has lead to the replacement of conventional materials with lighter and stronger materials. Aluminum alloys have a great potential for this application but their weaknesses are too low strength and stiffness. On the other hand production of parts from metal-matrix composites which is based on the reduction of grain size is very promising due to their unique mechanical properties in comparison with coarsegrained materials [1–6]. Powder (P/M) and ingot metallurgy which include particle reinforcement mixing with solid or molten metal assure production of composites with WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line) doi:10.2495/MC110211
240 Materials Characterisation V improved mechanical, physical and thermal properties. Hard particles like ceramic, carbides, oxides or nitrides are usually finely dispersed in Al alloy matrix. For P/M the size of the powder can be in the range between several tens of nm to hundred μm, while the sizes of reinforcing particles are accordingly lower. Thus with P/M a finer microstructure with a uniform distribution of micro-constituents can be obtained which leads to mechanical properties, i.e. stiffness, strength and wear resistance, etc. which are usually better than those obtained for materials produced by casting procedures [4–16, 23–25]. Mechanical parts made from aluminum-matrix composites (AMCs) are usually produced by conventional hot deformation processes such as rolling, extrusion, forging, etc. Thus understanding of the behaviour of the material during hot working is essential for optimization of the intrinsic workability, control of evolution of microstructure, and for optimization of the production process. Hot workability studies revealed that AMCs are usually more sensitive to strain rate and temperature in comparison to conventional aluminum alloys. Namely, grain boundary sliding that takes place at higher temperatures and/or lower strain rates can lead to the formation of wedge-shaped cracks at grain boundaries. Additionally, the presence of hard particles in the softer matrix can result in plastic flow localization at the particle-matrix interface. Furthermore during hot extrusion of AMCs prior grain boundaries are usually removed due to diffusion, recovery, etc. that results in reduced mechanical properties. Increasing of grains during processing can be partly prevented by fine dispersion of ceramic particles. For such ultra-fine grained materials with a mean grain size of 1μm and below, different deformation mechanisms are operative than for coarser-grained materials. Namely in very small grains the accumulation of dislocations become difficult that leads to decreased ductility of fine grained AMCs, also at room temperature [5–23]. While for conventionally (by casting) produced AMCs the hot deformability has been studied in several publications [14–22, 26–30], hot deformation of P/M AMCs has found considerably less attention [5, 23–25, 31– 34]. Only a few studies on P/M AMC using pure aluminum as a matrix exist. These studies report that pure Al may benefit more from brittle reinforcement particulates than Al-alloys [5, 35, 36]. The aim of this work is to produce a new fine grained P/M Al-special matrix composite on the base of almost pure Al with an oxide surface on grains aiming to prevent excessive grain coarsening during hot extrusion and followed by a hot deformation process as well as assessment of appropriate hot working parameters. Furthermore the mechanical properties were determined after the extrusion process as well as after various annealing times.
2 Experimental procedures 2.1 Material and characterization of microstructure A JSM-5610 energy dispersive spectrometer from iXRF Systems Inc. with a digital processor 500 for chemical analysis of P/M Al-special was used along
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
241
with a transmission electron microscopic (TEM, JEOL-2021) for microstructural characterization. 2.2 Processing of initial P/M Al-special The P/M Al-special was produced according to the routes given in fig. 1. Powder was produced by Vacuum Powder Metallurgy, with a special grain powder process. Adding of various additives and lubricants is aimed to accelerate the diffusion processes in bulk material that represents the next step of preparation of powder mixture. After Rubber Isostatic Pressing (RIP) or Cold Isostatic Pressing (CIP) with application of a typical pressure of 100 MPa, the achieved density was 85%. Then follows the thermal preparation of billet (dehumidifying cycle) and machining of the billet to dimensions of 274 mm. The hot extrusion was carried out on an industrial 35 MN indirect press. After extrusion the achieved density was 95% while following hot compression this value was 100% (see fig. 1).
Figure 1:
Processing routes for P/M 1080Al composite.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
242 Materials Characterisation V 2.3 Hot extrusion Indirect hot extrusion at various billets temperature and extrusion speeds was employed to convert the processed powder to bulk material. 2.3.1 Extrusion of P/M Al-special on 35 MN indirect press Initial powder billets of dimensions 274 x 1500 mm were hot extruded to a flat bar. Die dimension was 91.3 x 50.8 mm and extrusion ratio was 13.3. Maximum press pressure in the main cylinder is 250 x 105 Pa, which corresponds to a force of 35 MN. Billets 2 and 3 are heated in a die furnace (indirect) to 340ºC and 320ºC, respectively, while billet 1 was heated in an induction furnace to 420ºC. Ram speeds were in the range 3.89–8.87 mm/s. Other data applied at extrusion of P/M Al-special on the 35 MN indirect press are collected in Table 1. Table 1:
Technological data - Flat bar 90x50 (±0.9) mm, alloy P/M Al-special, indirect extrusion.
Technological data; die opening: billet dimension: 274 mm x 1500 mm, 91.3 mm x 50.8 mm, container dimension 280 mm, extrusion ratio: 13.3 Billet material
P/M Al-special 1st billet
2nd billet
3rd billet
Billet length (mm)
500
500
500
Container temperature (ºC)
360
360
360
Billet temperature (ºC)
418
340
320
Ram speed (mm/s)
8.87
6.2
3.89
Puller speed - bar speed (m/min)
7.06
4.94
3.10
Press rest (mm)
50
50
50
2.4 Annealing In order to investigate the influence of elevated temperature on recrystallization resistance, annealing tests were carried out at temperatures of 300ºC and employed at different times, i.e. 1, 5, 9 and 20 hours. 2.5 Hot compression and tensile tests Hot workability was studied by hot compression tests. Cylindrical specimens of Rastegew type with dimensions =10 mm x 15 mm were cut from the extruded profile (3rd billet) so in non-annealed states as well as in annealed states, i.e. previously annealed at 300ºC for 1, 5, 9 and 20 hours, respectively. Hot compression of extruded material was carried out on a computer controlled servo-hydraulic machine, Gleeble 1500D, in the temperature range 300ºC to 570ºC, at five different strain rates (0.01, 0.1, 1 and 5 or 10 s-1) up to a strain of 0.9, while for the annealed material from a production point of view only at a WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
243
strain rate range 0.1–5 s-1. The heating rate was 1ºC/s and the soaking time at compression temperature was 1 minute. For reduction of friction between the cylindrical specimen and the tool, and in order to avoid their mutual welding, graphite lubricant and tantalum follies were used. After deformation the specimens were water quenched. The conditions for hot compression testing are given in Table 2. For determination of tensile strength and elongation at room temperature tensile tests were carried out on a Zwick Z400 according to standards (SIST EN 10002-1:2002, 5th edition, June 2002). Tensile specimens was cut out from the centre of the extruded bar, the working length of the tensile specimens was 50 mm with a diameter of =10 mm. Tensile tests were also carried out also on a AA 1050. Table 2: State of material Extruded Annealed
Hot compression test conditions. Hot compression tests Temperature, ºC Strain rate, s-1 300–570 0.01–10 300–570 0.01–10
Strain range 0-0.9 0-0.9
3 Results and discussion 3.1 Composition and microstructures of initial materials The bulk materials consist of 99.44 wt% Al, 0.034 wt% Si, 0.159 wt% Fe, while oxygen (0.363 wt%) is contained on the grain surface. Initial microstructures of loose and compacted powder P/M Al-special are given in figs. 2a-b, respectively. It is visible that the average diameter of powder grains is around 1μm and that the surface layer of grains consists of aluminum oxide (Al2O3) to prevent possible diffusion, recovery processes, i.e. to prevent grain growth during hot deformation.
a
b
Figure 2:
Initial microstructure of P/M Al-special: loose powder (a) and compacted powder with visible oxides on grain surface (b), TEM.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
244 Materials Characterisation V 3.2 Indirect hot extrusion of P/M Al-special The time courses of oil pressure in the cylinder for various billet temperatures as well as for ram speeds are shown on fig. 3. The lowest pressure of 215 x 105 Pa (177 bars) was obtained as expected at the highest billet temperature (418ºC) and highest ram speed (8.87 mm/s). On the other hand at the very beginning of the extrusion of the third billet almost the maximal oil pressure was achieved while the highest pressure at the lowest billet temperature (320ºC) and lowest ram speed (3.89 mm/s). The highest oil pressure 215 x 105 Pa (215 bars) was achieved at lowest billet temperature, i.e. at 320ºC.
Figure 3:
Time course of oil pressure in the cylinder during hot extrusion on the indirect press at various ram speed for P/M Al-special, 35 MN indirect press.
3.3 Hot workability of extruded and annealed P/M Al-special Compression tests were carried out for extruded as well as for annealed states. On fig. 4a the typical microstructure of the extruded profile and on fig. 4b the microstructure of deformed samples which have been previously annealed for 20 hours at a temperature of 300ºC are presented. It is visible that the oxide surface on P/M grains is partly damaged but the average diameter of grains is only slightly increased in comparison to the initial state (ca 1μm, see fig. 2b). The figure clearly reveals that the oxide surface prevents grain growth. Obtained flow curves for extruded P/M Al-special are shown on fig. 5. From deformed specimens and the shape of flow curves it can be seen that carrying out the hot deformation at lower strain rates (0.01–0.1 s-1) is not appropriate since on specimens cracks occurred; this results in rapid fall of flow curves in the strain range 0.4–0.5 as can be seen for the strain rate of 0.1 s-1 on fig. 5, left. The behaviour can be ascribed to the usual problem of low-strain rate/hightemperature plasticity of P/M Al-alloys described in several studies such as in [37–41].
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
Figure 4:
245
Microstructures of extruded P/M Al-special: extruded (a), and forged and annealed (b).
At higher strain rates the flow curves after achieving the maximal value at the very beginning of hot compression, i.e. in the strain range up to 0.03, the values of flow stress begin slightly to fall approaching a steady state (see fig. 5, right). Moreover, at higher strain rates (1–10 s-1) the cracks did not occur that is expressed also by the shape of flow curves.
Figure 5:
Flow curves of P/M Al-special (extruded state) at various temperatures and stain rates.
As mentioned the hot compression tests were carried out also for samples annealed at various times. The shapes of flow curves obtained after various annealing times and for a strain rate of 5 s-1 are similar to those obtained for the extruded state but in general their values decrease up to an annealing time of 5 hours (see fig. 6). It is visible that values of flow curves are considerably lower (about 15%) for annealed states in comparison to non-annealed states. During the extrusion a small percentage of the oxide layer on the grains are torn. This may allow re-arrangement of some grains with orientation changes. During annealing of extruded material the process of part polygonization starts. After one hour at 300°C material became stable since grain growth was blocked by oxides.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
246 Materials Characterisation V
Figure 6:
Comparison of flow curves between extruded and at various annealing times for P/M Al-special; strain rate 5 s-1, annealing temperature 300ºC.
3.4 Obtained mechanical properties The general presentation of the obtained mechanical properties, i.e. tensile strength and elongation, of P/M Al-special (indirect extrusion) is shown in fig. 7. The values of tensile strength of around 209 MPa and 205 MPa for the nonannealed and 20 hour annealed state were obtained, respectively. Furthermore, values for elongation for the non-annealed state were around 24% while for the 20 hour annealed samples these values slightly decreased and were around 23%.
Figure 7:
Tension curves σ vs. for P/M Al-special at various states and times of annealing.
For comparison for AA 1050 the values for strength were around 80 MPa and for elongation were around 38%. Thus around 2.6 times higher values for tensile WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
247
strength and 1.6 times lower values for elongation for P/M Al-special in comparison to AA 1050 were obtained. From the obtained results for P/M Al-special it can be derived that this material can be applied also at elevated temperatures. It is also worth mentioning that in our case grains consist only of almost pure Al and without particles in the matrix. Thus the achieved mechanical properties are also on a relative high level.
4 Conclusions New P/M Al-special composite made from almost pure Al with an average diameter of 1μm and with oxides on the grain surface was developed. The achieved density after cold isostatic pressing was 84%, after hot extrusion 96% and 100% after hot compression. The P/M Al-special material exhibited a good combination of strength and elongation properties also at samples previously exposed to elevated temperature. The produced P/M Al-special has a great potential to substitute the classical aluminum alloys and steels in the field of an elevated temperature working environment. The obtained values for tensile strength for P/M Al-special alloy are up to 2.6 times higher and for elongation around 1.6 times lower in comparison to values obtained for AA 1050 aluminum alloy produced by classical technology. The tensile strength of samples taken from extruded billet with an extrusion ratio of 13.3 reaches a value of about 209 MPa while achieved elongation was around 24%. The material exhibits good resistance against increasing of grain size at elevated temperature. After 20 hours of annealing the tensile strength slightly decreased and reached the value of 204 MPa. On the other hand the value for elongation remains almost at the same level, i.e. 23%. Extruded material exhibited low hot deformability at strain rates of 0.1 s-1 and lower, while at higher strain rates the compressed samples were crack free.
Acknowledgements The authors gratefully acknowledge Martin Balogand and Karol Iždinský from Slovak Academy Of Sciences, Institute of Materials and Machine, Mechanics Račianska 75, 831 02 Bratislava 3, for their contribution towards analyzing the powder.
References [1] Sahin, Y., Acılar, M., Production and properties of SiCp-reinforced aluminium alloy composites, Composites Part A: Applied Science and Manufacturing, 34(8),pp. 709-718, 2003. [2] Shi, X.L., Mishra, R.S, Watson, T.J., Effect of temperature and strain rate on tensile behavior of ultrafine-grained aluminum alloys, Materials Science and Engineering: A, 494(1-2), pp. 247-252, 2008.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
248 Materials Characterisation V [3] Flores-Zamora, M.I., Estrada-Guel, I., González-Hernández, J., MikiYoshida, M., Martínez-Sánchez, R., Aluminum–graphite composite produced by mechanical milling and hot extrusion, Journal of Alloys and Compounds, 434-435(31),pp. 518-521, 2007. [4] Kennedy, A.R., Wyatt, S.M., The effect of processing on the mechanical properties and interfacial strength of aluminium/TiC MMCs, Composites Science and Technology, 60(2), pp. 307-314, 2000. [5] Abouelmagd, G., Hot deformation and wear resistance of P/M aluminium metal matrix composites, JournalofMaterialsProcessingTechnology, 155156, pp.1395-1401, 2004. [6] Masahiro K., Properties of nano-structured pure Al produced by mechanical grinding and spark plasma sintering, Journal of Alloys and Compounds, 434-435, pp. 294-297, 2007. [7] Gubicza, J., Dirras, G., Szommer, P., Bacroix, B., Microstructure and tensile strength of ultrafine grained aluminum processed by hot isostatic pressing, Materials Science and Engineering: A, 458(1-2),pp. 385-390, 2007. [8] Song, M., He, Y.H., Effects of die-pressing pressure and extrusion on the microstructures and mechanical properties of SiC reinforced pure aluminum composites, Materials & Design, 31(2), pp. 985-989, 2010. [9] Zhang, W., Chai, D., Ran, G., Zhou, J., Study on microstructure and tensile properties of in situ fiber reinforced aluminum matrix composites, Materials Science and Engineering: A, 476(1-2),pp. 157-161, 2008. [10] Chou, S.N., Lu, H.H., Lii, D.F., Huang, J., Processing and physical properties of Al2O3/aluminum alloy composites, Ceramics International, 35(1), pp. 7-12, 2009. [11] Zhang, S.L., Zhao, Y.T., Chen, G., Cheng, X.N., Dai, Q.X., Microstructures and mechanical properties of aluminium matrix composites fabricated from Al–x wt.% Zr(CO3)2 (x = 5, 10, 15, 20, 25) systems, Journal of Alloys and Compounds, 429(1-2), pp. 198-203, 2007. [12] Yang, Y., Lan, J., Li, X., Study on bulk aluminum matrix nano-composite fabricated by ultrasonic dispersion of nano-sized SiC particles in molten aluminum alloy, Materials Science and Engineering: A, 380(1-2), pp. 378383, 2004. [13] Wu, J.M., Li, Z.Z., Nanostructured composite obtained by mechanically driven reduction reaction of CuO and Al powder mixture, Journal of Alloys and Compounds, 299(1-2), pp. 9-16, 2000. [14] Moreno, M.F., Carlos J.R. González, O., Compression creep of PM aluminum matrix composites reinforced with SiC short fibres, Materials Science and Engineering: A, 418(1-2), pp. 172-181, 2006. [15] Ceschini, L., Minak, G., Morri, A., Forging of the AA2618/20 vol.% Al2O3p composite: Effects on microstructure and tensile properties, Composites Science and Technology, 69(11-12),pp. 1783-1789, 2009. [16] Li, Z.J., Wang, L.D., Fei, W.D., Effect of interfacial Bi2O3 coating on compressive deformation behavior of aluminum borate whisker-reinforced
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
[17]
[18]
[19]
[20]
[21]
[22]
[23]
[24]
[25]
[26]
[27]
[28]
[29]
249
aluminum composite at elevated temperature, Materials Science and Engineering: A, 447(1-2),pp. 314-318, 2007. Cavaliere, P., Cerri, E., Leo, P., Hot deformation and processing maps of a particulate reinforced 2618/Al2O3/20p metal matrix composite, Composites Science and Technology, 64(9), pp. 1287-1291, 2004. Cavaliere, P., Evangelista, E., Isothermal forging of metal matrix composites: Recrystallization behaviour by means of deformation efficiency, Composites Science and Technology, 66(2), pp. 357-362, 2006. Cavaliere, P., Isothermal forging of AA2618 reinforced with 20% of alumina particles Composites Part A: Applied Science and Manufacturing, 35(6), pp. 619-629, 2004. Vedani, M., D`Errico, F.D., Gariboldi, E., Mechanical and fracture behavior of aluminum-based discontinuously reinforced composites at hot working, Composites Science and Technology,66, pp. 343-349, 2006. Li, Z.J., Fei, W.D., Yue, H.Y., Wang, L.D., Hot deformation Behaviors of Bi2O3-coated Al18B4O33 whisker reinforced aluminum matrix composite with high formability, Composites Science and Technology,67, pp. 963973, 2007. Choi, H.J., Lee, S.W., Park, J.S., Bae, D.H., Tensile behavior of bulk nanocrystalline aluminum synthesized by hot extrusion of ball-milled powders, Scripta Materialia, 59(10), pp. 1123-1126, 2008. Fan, G.J., Wang, G.Y., Choo, H., Liaw, P.K., Park, Y.S., Han, B.Q., Lavernia, E.J., Deformation behavior of an ultrafine-grained Al–Mg alloy at different strain rates, Scripta Materialia, 52(9), pp. 929-933, 2005. Ramanathan, S., Karthikeyan, R., Ganasen, G., Development of processing maps for 2124Al/SiCp composites, Materials Science and Engineering: A, 441(1-2), pp. 321-325, 2006. Bardi, F., Cabibbo, M., Evangelista, E., Spigarelli, S., Vukcevic, M., An analysis of hot deformation of an Al–Cu–Mg alloy produced by powder metallurgy, Materials Science and Engineering: A, 339(1-2), pp. 43-52, 2003. Badini, C., La Vecchia, G.M., Fino, P., Valente, T., Forging of 2124/SiCp composite: preliminary studies of the effects on microstructure and strength, Journal of Materials Processing Technology, 116(2-3), pp. 289297, 2001. Luo, H., Shaw, L., Zhang, L.C., Miracle, D., On tension/compression asymmetry of an extruded nanocrystalline Al–Fe–Cr–Ti alloy, Materials Science and Engineering: A, 409(1-2), pp. 249-256, 2005. Cerri, E., Spigarelli, S., Evangelista, E., Cavaliere, P., Hot deformation and processing maps of a particulate-reinforced 6061+20% Al2O3 composite, Materials Science and Engineering: A, 324(1-2), pp. 157-161, 2002. Zhang, H., Ramesh, K.T. Chin, E.S.C., High strain rate response of aluminum 6092/B4C composites, Materials Science and Engineering: A, 384(1-2), pp. 26-34, 2004.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
250 Materials Characterisation V [30] Cöcen, Ü., Önel, K., Ductility and strength of extruded SiCp/aluminiumalloy composites, Composites Science and Technology, 62(2), pp. 275-282, 2002. [31] Malas, J.C., Venugopal, S., Seshacharyulu, T., Effect of microstructural complexity on the hot deformation behavior of aluminum alloy 2024, Materials Science and Engineering: A, 368(1-2), pp. 41-47, 2004. [32] Staehler, J.M., Predebon, W.W., Pletka, B.J., Subhash, G., Micromechanisms of deformation in high-purity hot-pressed alumina, Materials Science and Engineering: A, 291(1-2), pp. 37-45, 2000. [33] Evangelista, E., Forcellese, A., Gabrielli, F., Mengucci, P., in: T.G. Langdon, H.D. Merchant, J.G. Morris, M.A. Zaidi (Eds.), Hot Deformation of Aluminium Alloys, The Minerals, Metals and Material Society, Warrendale, PA, pp. 121-130, 1991. [34] Cerri, E., Evangelista, E., Forcellese, A., Fiorini, P., Stobrawa, J., Strength and ductility of powder metallury aluminum alloys under hot working conditions, in: H. Oikawa, et al. (Ed.), Strength of Metals and Alloys, The Japan Institute of Metals, Tokyo, pp. 807-810, 1994. [35] Lai S.W., Chung, D.L., Superior high temperature resistance of aluminum nitride particle-reinforced aluminum compared to SiC or Al2O3 particlereinforced aluminum. Journal of Material Science,29, pp. 6181–6691, 1994. [36] Mazen, A.A., Ahmed, A.Y., Mechanical behaviour of Al–Al2O3 MMC manufactured by PM technique, Part I: Scheme I processing parameters. J. Mater. Eng. Perf.,73,pp. 393–401, 1998. [37] Park, K.T., Lavernia, E.J., Mohamed, F.A., High-temperature deformation of 6061 Al, Acta Metall. Mater.,42, pp. 667-678, 1994. [38] Li, Y., Nutt, S.R., Mohamed, F.A., An investigation of creep and substructure formation in 2124 Al, Acta Materialia, 45, pp. 2607-2620, 1997. [39] Li, Y., Langdon, T.G., A comparison of creep properties of an Al-6092 composite and unreinforced matrix alloy, Metall. Mater. Trans., 29A, pp. 2523-2531, 1998. [40] Kloc, L., Spigarelli, S., Cerri, E., Evangelista, E., Langdon, T.G., Creep behaviour of an aluminum 2024 alloy produced by powder metallurgy, ActaMaterialia,45, pp. 529-540, 1997. [41] Kloc, L., Cerri, E., Spigarelli, S., Evangelista, E., Langdon, T.G., Significance of continuous precipitation during creep of a powder metallurgy aluminum alloy, Materials Science and Engineering: A, 216,pp. 161-168, 1996.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
251
Coarsening kinetics of the bimodal distribution in DS GTD111TM superalloy V. S. K. G. Kelekanjeri1, S. K. Sondhi2, T. Vishwanath1, F. Mastromatteo3 & B. Dasan2 1
GE Global Research Center, Materials Characterization Laboratory, Bangalore, India 2 GE Global Research Center, Materials Research Laboratory, Bangalore, India 3 GE Oil & Gas, Nuovo Pignone, Materials & Processing Engineering, Florence, Italy
Abstract Coarsening of precipitates in nickel-base superalloys is one of the mechanisms by which creep damage occurs in these alloys. This is brought about by a concomitant increase in the interparticle spacing, which results in faster dislocation movement and therefore, faster creep strain accumulation. Therefore, it is vital to accurately quantify the coarsening kinetics, which is essential for creep prediction via microstructure-based continuum damage mechanics models. In the present article, we report on coarsening studies of bimodally distributed precipitates in DS GTD111TM (Trademark of the General Electric Company). The baseline microstructure of GTD111TM consisted of a cuboidal secondary population and a much finer, spherical tertiary population with respective mean radii of 266 and 34nm. Long-term aging experiments were conducted on baseline samples at temperatures of T1
252 Materials Characterisation V enhancement of coarsening. Separately, the expected rate of tertiary coarsening was computed by compensating for volume fraction loss of tertiary between successive aging steps consumed during secondary coarsening. The coarsening rate constants from these two approaches were comparable. In summary, the present work is a treatment of bimodal coarsening kinetics in GTD111TM where coarsening of the individual distributions are captured using a single consistent LSW rate equation. Keywords: nickel-base superalloy, GTD111TM, precipitate, - microstructure, bimodal distribution, coarsening, ostwald ripening, kinetics, creep, continuum damage mechanics.
1 Introduction GTD111TM is a directionally solidified nickel-base superalloy which is used in hot gas-path sections of gas turbines at temperatures of 750 to 800ºC [1]. The presence of a bimodal distribution of the (Ni3(Al,Ti)) precipitate phase contributes to superior high temperature strength [2] of the alloy. However, upon prolonged thermal exposure, the distribution undergoes coarsening [3], resulting in an increase in the average interparticle spacing. This leads to easier movement of dislocations in the alloy and consequently, lower creep resistance [4]. In this context, coarsening as a damage mechanism warrants a detailed study so that an understanding of the kinetics of microstructure degradation may subsequently be used to develop continuum damage mechanics models. Coarsening studies in multimodal precipitate systems has been reported in literature in both blade alloys such as GTD111 [5] and IN738 [6] and disc alloys such as Nimonic 115 [7] and Waspaloy [8]. Sharghi Moshatghin and Asgari [9] studied coarsening of bimodal distribution in IN738LC in terms of a compound size parameter that incorporates both primary and secondary dimensions. They found that the kinetics analyzed using this size parameter followed the cubic rate law in the range of 850 to 900ºC. Stevens and Flewitt [6] reported volume diffusion controlled coarsening kinetics for both secondary and tertiary in IN738 in the temperature range of 750 to 850ºC. Coakley et al. [7] implemented the numerical mean-field model based on the LSW theory, originally developed by Chen and Voorhees [10] to study bimodal coarsening in Nimonic 115. Experimental data at short aging durations behaved in accordance with the cubic rate law that corresponded to disappearance of the fine followed by a plateau region where the number of primary precipitates was nearly constant. At a later stage, data again followed the cubic rate law where coarsening was deemed to have reached steady-state. The numerical model predicted the trends in the experimental data reasonably well, better than the existing unimodal LSW coarsening models [7]. The present work involves experimental studies of coarsening in the temperature range of 800 through 1000ºC and subsequent treatment of bimodal coarsening kinetics using a single consistent LSW rate equation.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
253
2 Experimental procedure The baseline material for coarsening studies was GTD111TM which was solutionized at 1121C for 2h followed by an inert gas fan cool and subsequently aged at 816C for 4h. In order to conduct careful and thorough coarsening studies of GTD111TM, baseline specimens, 20 in number, were first examined to obtain their key microstructural statistics, viz. secondary radius rS ' , secondary volume fraction f v _ S ' , tertiary radius rT '
and volume
fraction f v _ T ' . Subsequently, the baseline specimens were subjected to aging at temperatures of T1
3 Results The baseline microstructure consisted of a bimodal distribution of cuboidal secondary precipitates and much finer tertiary precipitates as shown in fig. 1(a). The corresponding bimodal size distribution (PSD) is shown in fig. 1(b). The averaged baseline statistics from examination of 20 specimens are as follows: rS ' =266nm, f v _ S ' =37%, rT ' =34nm and f v _ T ' =4%. The microstructures of specimens that are obtained upon aging will be summarized next. At T1, the microstructure consisted of a bimodal distribution after 250h of aging as seen from the micrograph in fig. 2(a). Upon further aging to 4000h, the microstructures did not show any significant coarsening of the secondary distribution (see fig. 2(b)); only a shape change from cuboids to corner-rounded cuboids was noted after 2000h of aging. The number density of tertiary precipitates decreased with progressive coarsening, also evident from the micrographs. At T2, the tertiary distribution was only sighted at 200h (see fig. 2(a)), beyond which, the tertiary precipitates coarsened and merged with the secondary PSD. Furthermore, shape change of the secondary precipitates from WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
254 Materials Characterisation V
Figure 1:
(a) Micrograph of GTD111TM baseline specimen showing bimodal distribution and (b) histogram of the precipitate distribution.
corner-rounded cuboids to completely rounded shapes and accompanying coarsening was clear beyond 800h of aging. A micrograph representative of the microstructure at 4000h is shown in fig. 2(d) where rounded shapes are evident. The kinetics of coarsening was progressively faster at T3 and T4 due to which, snapshots of tertiary coarsening could not be captured even at relatively small aging durations of 100h. A plot of the total and secondary volume fraction versus aging temperature is shown in fig. 3 for the aged specimens and the corresponding baselines. It is clear from the plot that f v _ tot is higher after aging, which is primarily due to discrepancy in the secondary volume fractions before and after aging. This unexpected rise in the volume fraction upon aging, across the temperature range of 800 to 1000C is hard to reconcile because the thermodynamically permissible volume fraction resulting from the standard aging treatment should be higher
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
Figure 2:
255
Representative microstructures of specimens aged for (a) 250h at T1, (b) 4000h at T1, (c) 200h at T2 and (d) 3200h at T2. 60 fv_S g'‐aged specimens fv_tot‐aged specimens fv_S g'‐baselines
55
fv_tot‐baselines fv_tot‐THERMOCALC
fv (%)
50
45
40
35
30 T (C)
Figure 3:
Plot of the secondary and total volume fractions in the baseline and aged specimens at temperatures of T1
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
256 Materials Characterisation V 8 T1‐tertiary g'
7 6
fvT' (%)
5 4 3 2 1 0 0
1000
2000
3000
4000
5000
t (h)
Figure 4:
Plot showing evolution of the tertiary volume fraction at T1.
600
160 T1‐secondary g' T2‐secondary g'
500
T4‐secondary g'
120 100
300
80
‹r›T' (nm)
T1‐tertiary g'
400
‹r›S' (nm)
140
T3‐secondary g'
60
200
40 100
20
0 0
Figure 5:
1000
2000
3000 t (h)
4000
0 5000
Plot showing evolution of the mean secondary radius with aging through temperatures of T1 through T4 (T1
than at temperatures of T2 to T4. The latter is inferred from the monotonically decreasing trend in the THERMOCALC predicted equilibrium volume fraction WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
257
with increasing temperature, also shown in fig. 3. The evolution of the tertiary volume fraction upon aging at T1 is shown in fig. 4. The gradually decreasing trend is expected for two reasons- (a) consumption of the tertiary by the secondary during coarsening and (b) inherent coarsening of the tertiary , thus merging with the secondary population. Figure 5 shows a plot of the mean secondary radius versus aging time for all the four data sets. The expected increase in rS ' due to coarsening is preceded by a dip early on at T1, T2 and T3. The likely reason for this is attributed to fraction of the coarser tertiary population joining into the secondary distribution. The increase in the mean tertiary radius at T1 is also shown in the same plot. However, the observed increase in rT ' is apparent because of the fact that a fraction of the tertiary population is consumed during coarsening of the secondary distribution (see fig. 4), which otherwise would have participated in inherent coarsening of the tertiary distribution.
4 Analysis and discussion The assumptions made for analysis of the coarsening kinetics using the results presented in the preceding section are stated as follows: 1.
2.
3.
The baseline statistics need further validation because of the unexpected increase in the secondary and total volume fraction upon aging, as mentioned earlier. Therefore, coarsening analysis will be conducted using population statistics of the aged specimens only. The kinetics of secondary coarsening will be analyzed using unimodal secondary statistics at T2, T3 and T4 after disappearance of the tertiary distribution. This ensures that any effects of the tertiary distribution on coarsening of the secondary population are no longer present. Therefore, the commencement of coarsening ( t 0 ) is defined as the earliest aging sampling step at which, the tertiary distribution is non-existent. Longer aging durations are accordingly shifted as per this definition. Data at T1 is not used for this analysis because a bimodal distribution is evidently present up to 4000h of aging. Furthermore, the present data do not show a clear coarsening trend in secondary size and statistics at longer aging times must be considered for obtaining a clearer trend. The expected rate of coarsening of the tertiary distribution at T1 should be faster than the observed rate for reasons specified earlier. Therefore, to arrive at the expected coarsening rate, the loss of tertiary volume fraction between successive aging steps (see fig. 4) is reconciled and accounted for during coarsening.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
258 Materials Characterisation V 1.E+08
T2‐secondary g'‐observed T3‐secondary g'‐observed
1.E+08
T4‐secondary g'‐observed
‹rt›3‐‹ro›3 (nm3)
8.E+07
R² = 0.99
6.E+07 4.E+07 R² = 0.93
2.E+07 R² = 0.99
0.E+00 0
Figure 6:
500
1000
1500 t (h)
2000
2500
3000
Cubic rate law plot of the mean secondary radius at T2, T3 and T4 (T2
As per assumption#2, secondary coarsening kinetics is analyzed using the cubic rate law rSt '
3
rSo '
3
k (T )t , where rSt ' is the secondary radius at Q
k is the radius at t 0 and k (T ) o e RT is the coarsening rate time t, T constant with pre-exponential k o and activation energy Q . A plot of the cubic rate law at T2, T3 and T4 shows reasonably good linear fits at all the three temperatures as shown in fig. 6. As expected and congruent with microscopic observations, the rate of coarsening (slope of the linear fit) is enhanced with increasing aging temperature. Next, a plot of the coarsening rate 1 constant k (T ) versus (see fig. 7) is used to extract the activation energy of RT coarsening as 212.45 (kJ/mol). The expression for the rate constant describing the secondary coarsening kinetics is as follows:
rSo '
212447
k k (T ) o e RT T
(nm3/h)
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
(1)
Materials Characterisation V
259
18.5
k(T) (nm3/h)
18
Q=212,447kJ/mol
17.5
R² = 0.98
17
16.5
16 1/RT (mol/J·K)
Figure 7:
1 Plot of k (T ) vs. using the rate constants at T2, T3 and T4. RT The activation energy of secondary coarsening is obtained as slope of the linear fit.
4.E+06
Observed coarsening rate‐T g' Expected coarsening rate‐ T g'
3.E+06
Predicted coarsening rate‐ T g'
‹rt›3‐‹ro›3 (nm3)
3.E+06
R² = 1
2.E+06 R² = 0.98
2.E+06 1.E+06 5.E+05 R² = 0.80
0.E+00 0
Figure 8:
500
1000
1500
2000 t (h)
2500
3000
3500
4000
Cubic rate law plots of the mean tertiary radius at T1 shown for three cases – measured data, predicted dataset using secondary kinetics and expected dataset by correcting for volume fraction loss of tertiary during coarsening.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
260 Materials Characterisation V Next, this rate constant equation is used to predict coarsening of the tertiary distribution at T1 starting at 250h. However, realizing that eqn. (1) represents coarsening rate of the secondary distribution with a volume fraction of ~0.45, the enhanced coarsening rate due to high volume fraction [3,11,12] must be accounted for prior to applying the rate equation to the tertiary distribution. The pre-exponential in eqn. (1) is divided by a scaling factor k ( f v ) 3.5 to correct for volume fraction effects on coarsening [3]. The resulting equation for the rate constant, listed below, may then be applied to coarsening under low volume fraction scenario such as tertiary coarsening. 212447
0.29 k o RT k (T ) e T 3
Figure 8 shows a plot of rTt '
3
rTo '
(nm3/h)
(2)
vs. t for tertiary coarsening at T1
and the corresponding linear fits for three cases – (a) ‘observed’ using measured data, (b) ‘predicted’ using eqn. (2) and (c) ‘expected’ by correcting measured data. The aging duration of 250h is taken to be the starting point i.e. t 0 and the corresponding tertiary statistics are used to generate the predicted data set. The expected tertiary radii are computed in accordance with assumption#3 by the t
t
t following procedure. Denote f v N_ T ' , N v N_ T ' and rTN'
to be the volume
fraction, particle density (number of precipitates per unit volume) and mean radius of the tertiary distribution at the Nth aging time step corresponding to time t t N . It should be noted that the data point at 250h is denoted using the superscript t o ( t o 0 in this case) and longer aging times are scaled with
t
t
reference to 250h as the starting point. Now, f v N_ T1 ' f v N_ T ' is the tertiary volume fraction which is consumed during secondary coarsening. This lost fraction between successive time steps is accounted for to obtain the expected
t
t
radii. Specifically, the lost volume fraction f v N_ T1 ' f v N_ T '
results in a
volume V t N 1 t N over and above the volume VTN' of a tertiary precipitate t
with mean size rTtN' . The quantities VTtN' and V t N 1 t N are defined as below: t
VTN' and V
t N 1 t N
3 4 rTtN' 3
f
t N 1 tN v _ T ' f v _ T ' t N v N_ T '
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
(3)
(4)
Materials Characterisation V
261
t
Here N v N_ T ' is the number density of tertiary precipitates at the instant t t N . Using this framework, the expected mean tertiary radius t
N rexp ected _ T ' at t t N is computed as:
1
N 3 tN t t VT ' V N 1 N tN 1 (5) rexp ected _ T ' 4 3 The correlation coefficient of the linear fit is improved from 0.8 for the measured data to 0.98 (see fig. 8) for the ‘expected’ data set by making this correction. The agreement between the coarsening rate constants of the ‘expected’ and ‘predicted’ datasets is also better when compared to the rate constant of the measured dataset. The reasons for disparity between the rate constants of the ‘expected’ and ‘predicted’ data sets could be two fold. First, there could be errors associated with truncation of the tertiary PSD due to overlap of the secondary and tertiary distributions, as is the case at 1000 and 2000h at T1. Secondly, any variations in the precipitate statistics that exist among the baselines are ignored in the present analysis. It is quite possible that the truncation errors and initial baseline variations could lead to errors in estimation based on eqn. (5).
5 Conclusions Aging experiments were conducted on GTD111TM in the temperature range of 800 to 1000C with the objective of studying bimodal coarsening kinetics in this alloy. Secondary coarsening kinetics elucidated using data where the PSD’s were clearly unimodal yielded an activation energy of 212.45 (kJ/mol). This kinetics was then applied to tertiary coarsening at the lowest aging temperature by using an appropriate scaling factor to discount volume fraction enhancement of coarsening. Separately, the kinetics of the measured data was corrected approximately by accounting for volume fraction loss of tertiary between successive aging steps accompanying secondary coarsening. The coarsening rate constants obtained from these two approaches showed a reasonable agreement with each other. Thus, the kinetics of bimodal coarsening in GTD111TM could be described using a single consistent LSW rate equation in the chosen range of temperatures.
Acknowledgements The authors greatly appreciate S. Vipin for careful metallographic preparation of GTD111TM aged specimens. Next, we would like to acknowledge the staff at the National Metallurgical Laboratory, Jamshedpur, India for their assistance in WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
262 Materials Characterisation V characterization of the baseline GTD111TM specimens. We also acknowledge the helpful discussions with Drs. M. Karadge and S. Swaminathan on treatment of coarsening data.
References [1] Schilke, P. W. Advanced gas turbine materials and coatings. GE Energy. Schenectady, NY : s.n. [2] Sims, C. T. and Hagel, W. C. The Superalloys. New York : John Wiley & Sons, 1972. [3] Martin, J. W., Doherty, R. D. and Cantor, B. Stability of microstructure in metallic systems. New York : Cambridge University Press, 1997. [4] Dyson, B., Use of CDM in materials modelling and component creep lifeprediction. ASME Journal of Pressure Vessel Technology, 315, pp. 281-296, 2000. [5] Mastromatteo, F., Niccolai F., Giannozzi, M. and Bardi, U., The coarsening kinetic of γ' particles in nickel-based superalloys during aging at high temperatures. Proceedings of the ASME Turbo Expo. 4, pp. 851-856, 2004. [6] Stevens, R. A. and Flewitt, P. E. J., The effects of γ′ precipitate coarsening during isothermal aging and creep of the nickel-base superalloy IN-738. Materials Science and Engineering A 3, pp. 237-247, 1979. [7] Coakley, J., Basoalto, H. and Dye, D., Coarsening of a multimodal nickelbase superalloy. Acta Materialia 58, pp. 4019-4028, 2010. [8] Kelekanjeri, V. S. K. G., Moss, L. K., Gerhardt, R. A. and Ilavsky, J., Quantification of the coarsening kinetics of γ′ precipitates in Waspaloy microstructures with different prior homogenizing treatments. Acta Materialia, 57(16), pp. 4658-4670, 2009. [9] Sharghi-Moshatghin, R and Asgari, S., The effect of thermal exposure on the characteristics in a Ni-base superalloy. Journal of Alloys and Compounds 368, pp. 144-151, 2004. [10] Chen, M. K. and Voorhees, P. W., The dynamics of transient Ostwald ripening. Modelling and Simulation in Materials Science and Engineering, 1(5), pp. 591-612, 1993. [11] Davies, C. K. L., Nash, P. and Stevens, R. N., The effect of volume fraction of precipitate on ostwald ripening. Acta Metallurgica, 28(2), pp. 179-189, 1980. [12] Voorhees, P. W. and Glicksman, M. E., Solution to the multi-particle diffusion problem with applications to ostwald ripening-II. Computer simulations. Acta Metallurgica, 32(11), pp. 2013-2030, 1984.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
263
Effect of the elastomer stiffness and coupling agents on rheological properties of magnetorheological elastomers A. Boczkowska & S. F. Awietjan Faculty of Materials Science and Engineering, Warsaw University of Technology, Poland
Abstract The reported studies are related to a new group of intelligent materials, such as magnetorheological elastomers (MREs), which are composites of ferromagnetic particles embedded in elastomer matrix. Studies on fabrication of MREs were carried out using two different polyurethane elastomers as a matrix, ferromagnetic particles and coupling agent. The matrices were differed in elastomer stiffness and hardness. As a ferromagnetic component carbonyl–iron powder with particle size of 6-9 µm was used. Particles were oriented into chains under the external magnetic field of 240 kA/m. Samples with anisotropic particles arrangement and particles volume fraction equal to 11.5% were examined. Microstructure of MREs was observed using Scanning Electron Microscopy. Structural and magnetic anisotropy of the MREs was derived from the magnetic studies. Rheological properties of the MREs, such as storage and loss modulus and loss factor, were characterized as a function of shear frequency and strength of the magnetic field. Absolute and relative magnetorheological effects were calculated taking the microstructure into account, which was formed in the course of the MRE fabrication, by changing the elastomer type and application of the coupling agent. As a result of the studies it was found that the MREs with stiffer matrix exhibit much lower MR effect than the soft ones. Also the application of coupling agent lead to a decrease of MR effect by the formation of bound elastomer surrounding the particles, which increases the stiffness of the matrix. Keywords: magnetorheological elastomers, polyurethanes, carbonyl iron, rheological properties, coupling agents, microstructure, magnetic properties, magnetorheological effect. WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line) doi:10.2495/MC110231
264 Materials Characterisation V
1 Introduction Magnetorheological elastomers (MREs) are composites of ferromagnetic particles embedded in elastomer matrix. They belong to a new group of intelligent materials, which change their rheological properties under the influence of an applied magnetic field. They are solid analogues of the magnetorheological fluids (MRF) where the matrix is a solid elastic polymer rather than carrier oil. These both consist of micrometer-sized magnetically permeable particles in a non-magnetic matrix material. As in the case of MRF, the particles try to arrange themselves in the direction of the magnetic field. An external magnetic field is applied during the curing process of the elastomer, what is described in many papers, e.g. Zhou [1], Farshad and Benine [2], Jolly et al. [3]. The field induces dipole moments within the particles, which seek minimum energy states. Chains of particles with collinear dipole moments are formed and curing of the polymeric matrix locks the chains in place. As a result special structures, such as chains or columns of particles, remain in the matrix. MREs have controllable magnetic field dependent modulus rather than field dependent yield stress and also stable performance. The advantage of MREs, in comparison to MRF, is that ferromagnetic particles do not undergo the sedimentation. Moreover, the amount of filler is usually lower in MREs. As a result, the weight of sensing and actuating devices based on MREs is reduced and due to their microstructure, the time, following signal introduction, for a strain change response to applying a magnetic field is shortened, as it was described by Carlson and Jolly [4]. MREs exhibit reversible changes of their properties and shape under the magnetic field, what makes them attractive for applications as dampers, sensors or actuators, e.g. An and Shaw [5], Farshad and Le Roux [6]. The automotive bushing based on MREs, adjusted to reduce suspension deflection and to improve passenger comfort, were patented by Watson [7] and Steward et al. [8] for Ford Motor Company. The magnetic particles are usually carbonyl iron and as the matrix soft silicone elastomers, poly(vinyl alcohol), gelatine, natural rubber or polyurethanes can be used. The results of the studies presented in the literature [9–12] show that the properties of MREs depend on the particles shape, size and volume fraction, as well as the magnetic field strength. Only a few papers are focused on the influence of the polymer matrix properties on the magnetorheological effect (MR effect) in MREs. Gong et al. [13] and Lokander and Stenberg [9] reported the advantageously influence on the MR effect of the addition of plasticizers to the silicon rubber due to the decrease of the polymer matrix stiffness. Jiang et al. [14] indicate on the positive effect of the particles surface modification on the increase of the MR effect, while Zhang et al. [15], Fan et al. [16] and Li et al. [17] show that the enhancement of the particle-polymer matrix interaction leads to the decrease of MR effect due to the existence of bound-rubber phenomenon. The improvement of filler-matrix interaction is an effective way to improve the mechanical properties of materials; however in the case of MREs it can affect the MR effect.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
265
In our studies the influence of the matrix and the interaction between matrix and filler on the MREs properties were investigated. We used two polyurethane elastomers as a matrix, which were differed in stiffness and hardness. Particles were oriented into chains under the external magnetic field of 240 kA/m. Samples with the same particles volume fraction equal to 11.5%, were examined. The iron particles-polyurethane matrix interactions were changed by the application of silane coupling agent and the effect of the interaction on the performance of MREs were investigated. Rheological properties of the MREs, such as storage and loss modules and loss factor, were characterized as a function of shear frequency and strength of the magnetic field. Absolute and relative MR effects were calculated taking microstructure into account, which was formed in the course the MREs fabrication, by changing the elastomer type and application of the coupling agent.
2 Materials and methods Magnetorheological elastomers were manufactured using: soft polyurethane (PU), obtained from polyether polyols VORALUX® HF 505 used in a blend with 14922, with the average molecular weight respectively 3596 and 4350 g/mol, and isocyanate compound HB 6013, supplied by Dow Chemical Company, segmented urea-urethane elastomer (EPU), obtained from 4,4’diphenylmethane diisocyanate (MDI), ethylene oligoadipate (OAE) with the average molecular weight about 2000 g/mol and dicyandiamide (DCDA) as a chain extender. PU substrates were mixed in the weight ratio, respectively 30:70:23. Mixing and curing process were conducted at room temperature. EPU with molar ratio of MDI:(OAE+DCDA) equal to 2.5 was synthesized by one-shot method. The curing process was carried out at temperature of 150oC, what makes technology aspects of MRE manufacturing more complicated. The existence in every short hard segment of strong polar urea group and strong polar nitrilimide side-group increases the urea-urethane thermal and mechanical properties, as well as stiffness and hardness in comparison to soft polyether polyurethane (see Table 1). The ferromagnetic component used in the MREs, was carbonyl–iron powder with particles size ranging from 6-9µm, produced by Fluka. The amount of the carbonyl-iron particles was equal to 11.5 vol.%. It was found in our earlier studies, e.g. Boczkowska and Awietjan [18], carried out for the series of samples with the particles content varied from 1.5 to 33 vol.%, that the highest structural and magnetic anisotropy, as well as MR effect was observed for MREs containing 11.5 vol.% of carbonyl-iron. Therefore, in this study such volume fraction of particles was used. As shown in Table 1 polyurethane PU 70/30 is characterized by lower density and mechanical properties than EPU 2.5. Low hardness and stiffness of the polyurethane matrix can lead to the higher relative property changes of the MRE under an external magnetic field. On the other hand, the EPU 2.5 is distinguished WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
266 Materials Characterisation V by lower viscosity of reactive mixture of substrates and good mechanical properties. Low viscosity during the processing of the MRE makes the arrangement of the particles into aligned chains very easy. Table 1:
Selected physical and mechanical properties of elastomers used for MREs fabrication.
Material
Viscosity (mPas)
Density (g/cm3)
PU 70/30 EPU 2.5
8000 1400
1.03 1.26
Glass transition temperature (oC) Soft Hard segments segments -64 41 -23 165
Hardness (oShA)
Young’s modulus (MPa)
< 10 87
0.1 14
The samples were subjected to a magnetic field during curing to produce aligned carbonyl-iron chains within the elastomer. The magnetic field strength used during MREs processing was equal to 240 kA/m. Samples prepared for rheological measurements were in the form of 2 mm thick and 20 mm in diameter round plates. As a coupling agent N-2-(aminoethyl)-3-aminopropyltrimethoxysilane (U-15) was used with the chemical formula H2NCH2CH2CH2Si(OC2H5)3, supplied by Unisil Tarnow Company. Carbonyl-iron particles were mechanically stirred in the 1% solution of coupling agent in toluene before MREs fabrication, until the solvent evaporation. The microstructure of cross-sections of MREs was observed using scanning electron microscopy (SEM) Hitachi S-2600. The smooth surfaces were cut on Rotary Microtome LEICA RM2165 with LN21 cooling device. Magnetic properties were studied by Lake Shore vibrating sample magnetometer (VSM). Tests were carried out parallel and perpendicular to the direction of particles alignment, corresponding to the magnetic field direction during curing. Rheological properties were evaluated at 25oC with the application of Ares Rheometer from TA Instruments with plate-plate setup (plate diameter - 20 mm, gap - 2 mm, magnetic field ranged from 0 to 480kA/m). The chains of particles were aligned parallel to the magnetic field direction during experiment in the case of investigation of the influence of the matrix properties or they were sloped to 45 degrees in the case of the examination of the influence of coupling agent.
3 Results and discussion 3.1 Microstructure observations Microstructure observations by SEM proved the existence of aligned particles chains in the direction of the magnetic field applied during curing of the matrix in PU 70/30 polyurethane soft elastomer (fig. 1a), as well as in EPU 2.5 ureaurethane elastomer (fig. 1b). The direction of the magnetic field during curing is shown by the white arrows in figs. 1a, 1b. WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
267
a)
b)
Figure 1:
SEM images of the cross sections of MREs with 11.5 vol.% of particles, obtained from elastomers: a) PU 70/30 polyurethane, b) EPU 2.5 poly(urea-urethane).
Although the viscosity and the chemical composition of the reactive mixtures of substrates are different, the alignment of particles into chains in both cases was obtained. The average distance among the chains of particles calculated using image analysis, as described in Boczkowska et al. [19], is equal to 40±4 µm. 3.2 Magnetic measurements The existence of structural and magnetic anisotropy was confirmed by VSM studies. Tests were carried out parallel (II) and perpendicular (L) to the direction of particles chains, corresponding to the magnetic field direction during curing.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
268 Materials Characterisation V a) 80
II 60
Magnetization [Am2/kg]
L
40 20 0
-800
-600
-400
-200 0 -20
200
400
600
800
-40 -60 -80 Magnetic field [kA/m]
b) 150
Magnetization [Am2/kg]
II L
100 50 0
-800
-600
-400
-200 0 -50
200
400
600
800
-100 -150 Magnetic field [kA/m]
Figure 2:
Hysteresis loops for MREs with 11.5 vol.% of particles obtained from elastomers: a) PU 70/30 polyurethane, b) EPU 2.5 poly(ureaurethane).
In fig. 2 hysteresis loops obtained for the MREs with different matrix material are shown. From the hysteresis loops the anisotropy coefficient (Ab) was calculated at the selected value of magnetic field strength of 160 kA/m. It is expressed by the ratio of magnetization measured at 160 kA/m, respectively parallel and perpendicular to the particles alignment direction. The Ab values are similar for both kind of MREs matrix. They are equal to 1.50 for PU 70/30 and 1.42 for EPU 2.5, what means that the difference in viscosity and the chemical WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
269
composition of the reactive mixtures of substrates for both elastomers do not significantly influence on the possibility of particles to arrange themselves along the magnetic field direction. 3.3 Rheological properties In this study such rheological properties as storage modulus (G’), loss modulus (G”) and a ratio of G” to G’, so called loss angle, were measured as a function of oscillation frequency and the magnetic field strength. In viscoelastic materials, some of the deformation energy is stored and recovered during each cycle and some is dissipated as heat. The storage modulus represents the capacity of the material to store the energy of deformation, which contributes to the material stiffness. The loss modulus represents the capacity of the material to dissipate the energy of deformation into heat. Loss angle is a measure of material damping properties. Rheological studies were carried out for the MRE samples fabricated using PU 70/30 elastomer without and with U-15 coupling agent. Also the influence of the matrix hardness and stiffness on rheological properties was investigated for PU 70/30 and EPU 2.5 elastomers. The examples of G’, G” and tan δ curves for the MREs based on PU 70/30, obtained without and with coupling agent, as a function of the oscillation frequency, for different magnetic field strengths are shown in fig. 3. Storage and loss modulus grow as a function of oscillation frequency. The values of G’ and G” increase with the increase of the magnetic field strength. The values of G’ and G” for MREs without coupling agent measured without magnetic field exhibit the lowest course, what means that such samples are characterized by lower stiffness than the samples with coupling agent. The improvement of the adhesion between iron particles and polyurethane matrix leads to the immobilizing of the macromolecules around the particles, what increases the stiffness of the polyurethane in the area surrounding the particles. On the other hand MREs with coupling agent exhibit higher values of loss angle what is a measure of damping properties. The higher values of tan δ mean that the better are the damping properties of such material. Loss angle increases with the increase of the magnetic field strength, and the highest values were obtained for MREs with coupling agent. However, significantly higher growth in G’ and G” under the magnetic field was observed for the samples without coupling agent. Absolute (ΔG’) and relative (ΔG’/G’0) magnetorheological effects were calculated from rheological curves shown in fig. 3, for the same selected frequency equal to 10 Hz. The results have been presented in fig. 4 as a function of the magnetic field strength, where G’0 is the “zero field” modulus that means the storage modulus obtained without magnetic field.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
270 Materials Characterisation V 11,5%Fe 0 kA/m 11,5%Fe 160 kA/m 11,5%Fe 480 kA/m 11,5%Fe+1%U-15 80 kA/m 11,5%Fe+1%U-15 320 kA/m
0.90 0.80
GI [MPa]
0.70
11,5%Fe 80 kA/m 11,5%Fe 320 kA/m 11,5%Fe+1%U-15 0 kA/m 11,5%Fe+1%U-15 160 kA/m 11,5%Fe+1%U-15 480 kA/m
0.60 0.50 0.40 0.30 0.20 0.10 0.00 0.1 0.16
f [Hz]
10
100
f [Hz]
10
100
11,5%Fe 0 kA/m 11,5%Fe 80 kA/m 11,5%Fe 160 kA/m 11,5%Fe 320 kA/m 11,5%Fe 480 kA/m 11,5%Fe+1%U-15 0 kA/m 11,5%Fe+1%U-15 80 kA/m 11,5%Fe+1%U-15 160 kA/m
0.14 0.12
GII [MPa]
1
0.10 0.08 0.06 0.04 0.02 0.00 0.1
1
0.35 0.3
tan
0.25 0.2 0.15 0.1
11,5%Fe 0 kA/m 11,5%Fe 160 kA/m 11,5%Fe 480 kA/m 11,5%Fe+1%U-15 80 kA/m 11,5%Fe+1%U-15 320 kA/m
0.05 0 0.1
Figure 3:
1
f [Hz]
11,5%Fe 80 kA/m 11,5%Fe 320 kA/m 11,5%Fe+1%U-15 0 kA/m 11,5%Fe+1%U-15 160 kA/m 11,5%Fe+1%U-15 480 kA/m
10
100
Changes of storage (G’), loss (G”) modulus and loss angle (tan δ) as a function of frequency under magnetic field of 0, 80, 160, 320, 480 kA/m.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
271
0.70 0%U-15 0.60
1%U-15
∆G' [MPa]
0.50 0.40 0.30 0.20 0.10 0.00 0
80
160
240
320
400
480
400
480
560
H [kA/m] 800 0%U-15
700
1%U-15
∆G'/G'0 [%]
600 500 400 300 200 100 0 0
Figure 4:
80
160
240
320
H [kA/m]
560
Absolute (ΔG’) and relative (ΔG’/G’0) MR effects of the MREs based on PU 70/30 obtained without and with the silane coupling agent vs. the magnetic field strength (H).
Both MR effects, absolute and relative, grow with the magnetic field growth. The significant difference is observed between samples fabricated with and without application of silane coupling agent. The difference is much more significant with the magnetic field growth. The MR effect is higher in the MREs obtained without the application of the coupling agent because of the lower stiffness of the polyurethane matrix, what makes the ferromagnetic particles easier to attract under the magnetic field and as a result to introduce the additional strain and stress to the polyurethane matrix. Similar results were obtained for MREs fabricated from elastomers with different stiffness and hardness, as shown in fig. 6. WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
272 Materials Characterisation V
0.30 PU 70/30 0.25
EPU 2,5
∆G' [MPa]
0.20 0.15 0.10 0.05 0.00 0
80
160
240
320
400
480
560
400
480
560
H [kA/m] 600 PU 70/30 500
EPU 2,5
∆G'/G'0 [%]
400 300 200 100 0 0
80
160
240
320
H [kA/m]
Figure 5:
Absolute and relative MR effect of the MREs based on PU 70/30 and EPU 2.5 elastomers vs. the magnetic field strength (H).
MREs with the stiffer matrix (EPU 2.5) exhibit significant lower MR effect because the stiffer matrix makes impossible the particles to displace when they are subjected to the magnetic field.
4 Conclusions The properties of the MREs elastomeric matrix have a great influence on their rheological properties. Elastomer with higher stiffness and hardness leads to the higher storage and loss modulus of the MREs measured without magnetic field, while under the magnetic field the values are significant lower in comparison to
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
273
MREs obtained from soft elastomer. As a result the absolute and relative MR effect is much higher in MREs obtained from soft elastomer. The highest absolute and relative MR effect was obtained for the MREs synthesized from PU 70/30 elastomer with 11.5 vol.% of carbonyl-iron particles. The values of MR effects under the magnetic field of 480 kA/m reached 0.27 MPa and 540%, respectively, when the particles chains were aligned parallel to the magnetic field direction or 0.52 MPa and 750%, when the chains of particles were sloped to 45 degrees. The influence of the particles chains arrangement on the rheological properties is the subject of our different study. The increasing of the adhesion between particles and polyurethane matrix leads to the significant decrease in the MR effect by the existence of bound elastomer around the particles, what makes the matrix stiffer. High stiffness of the MREs matrix do not allow for the particles displacement under the magnetic field and subsequently the deformation of the elastomeric matrix. The rheological properties of MREs depend also on the oscillation frequency and magnetic field strength.
Acknowledgement This work was financed by National Centre for Research and Development (Poland) as a grant no. NR 15 0010 04.
References [1] Zhou G.Y., Shear properties of magnetorheological elastomer, Smart Materials and Structures, 12, pp. 139-146, 2003. [2] Farshad M. and Benine A., Magnetoactive elastomer composites, Polymer Testing, 23, pp. 347-353, 2004. [3] Jolly M. R., Carlson J.D., Munoz B.C. and Bullions T.A., The Magnetoviscoelastic Response of Elastomer Composite Consisting of Ferrous Particles Embedded in a Polymer Matrix, Journal of Intelligent Material Systems and Structures, 7, pp. 613-622, 1996. [4] Carlson J. D., Jolly M. R., MR fluid, foam and elastomer devices, Mechatronics, 10, pp. 555-569, 2000. [5] An Y., Shaw M.T., Actuating properties of soft gels with ordered iron particles: basis for a shear actuator, Smart Materials and Structures, 12, pp.157-163, 2003. [6] Farshad M., Le Roux M., Compression properties of magnetostrictive polymer composite gels, Polymer Testing, 24, pp.163-168, 2005. [7] Watson J. R., Methods and apparatus for varying the stiffness of a suspension bushing, US Patent 5.609.353, 1997. [8] Stewart W. M., Ginder J. M., Elie L. D., Nicholls M. E., Method and apparatus for reducing brake shudder, US Patent 5.816.587, 1998. [9] Lokander M., Stenberg B., Improving the magnetorheological effect in isotropic magnetorheological rubber materials, Polymer Testing, 22, pp. 677-680, 2003. WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
274 Materials Characterisation V [10] Lokander M., Stenberg B., Performance of isotropic magnetorheological rubber materials, Polymer Testing, 22, pp. 245-251, 2003. [11] Zhou, G.Y., Jiang, Z.J., Deformation in magnetorheological elastomer and elastomer-ferromagnet composite driven by a magnetic field, Smart Materials and Structures, 13, pp. 309-316, 2004. [12] Chen L., Gong X.L., Li W.H., Microstructures and viscoelastic properties of anisotropic magnetorheological elastomers, Smart Materials and Structures, 16, pp.2645-2650, 2007. [13] Gong X.L., Chen L., Li J.F., Study of utilizable magnetorheological elastomers, International Journal of Modern Physics B, 21(28,29), pp. 4875-4882, 2007. [14] Jiang W.G., Yao J.J., Gong X.L., Chen L., Enhancement in Magnetorheological Effect of Magnetorheological Elastomers by Surface Modification of Iron Particles, Chinese Journal of Chemical Physics, 21,
pp. 87-92, 2008. [15] Zhang X.Z., Gong X.L., Zhang P.Q, Li W.H., Existence of Bound-Rubber in Magnetorheological Elastomers and Its Influence on Material Properties, Chinese Journal of Chemical Physics, 20(2), pp. 173-179, 2007. [16] Fan Y.C., Gong X.L., Jiang W.Q., Zhang W., Wei B., Effect of maleic anhydride on damping property of magnetorheological elastomers, Smart Materials and Structures, 19(5), pp. 55015-55022, 2010. [17] Li J., Gong X., Zhu H., Jiang W., Influence of particle coating on dynamic mechanical behaviors of magnetorheological elastomers, Polymer Testing, 28, pp. 331-337, 2009. [18] Boczkowska A., Awietjan S.F., Urethane Magnetorheological Elastomers – Manufacturing, Microstructure and Properties, Solid State Phenomena, 154, pp.107-112, 2009. [19] Boczkowska A., Awietjan S., Wejrzanowski T., Kurzydłowski K.J., Image analysis of the microstructure of magnetorheological elastomers, Journal of Materials Science, 44, pp.3135–3140, 2009.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
275
Optimization of magnetoelastic properties of pure nickel by means of heat treatments A. L. Morales, A. J. Nieto, J. M. Chicharro, P. Pintado, G. P. Rodríguez & G. Herranz Department of Applied Mechanics and Project Engineering, ETSII, University of Castilla, La Mancha, Spain
Abstract In this work we include valuable information about the way in which nickel may become a valuable smart material for some applications and we provide a better understanding of the influence of internal stresses on its magnetoelastic effects. The different states of internal stress are achieved via different heat treatments obtained by modifying three main parameters: the heating temperature, the heating time and the cooling method. Then, we carried out the next works: first, a microscopic analysis of the grain size of the samples under different heat treatments; second, all the tested specimens were subjected to an exhaustive internal stress analysis via X-Ray diffraction techniques; and third, the ∆E- and ∆Ψ-effects were estimated in order to link the internal stress state of the sample to its magnetoelastic response. The results can guide us in selecting the most suitable heat treatment in order to make nickel show the smart properties we desire. Keywords: magnetoelasticity, internal stress, heat treatments, nickel, X-Ray diffraction.
1 Introduction Magnetomechanical materials are a kind of smart material in which there is a reciprocal coupling between their mechanical and magnetic properties [1]. In this broad category, described by Jiles in [2], we can find both magnetoelastic and magnetoplastic materials. The former are those in which the variation of the altered properties is reversible, whereas the latter are those in which the recovery of the properties is not necessarily obtained simply on removal of the magnetic or mechanical action. WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line) doi:10.2495/MC110241
276 Materials Characterisation V Some crystalline pure nickel results about ∆E-effect and magnetomechanical damping were previously published in our recent work [3]. In said work, nickel clearly stands out from the rest of classical ferromagnetic materials like iron and cobalt, but yet without showing enough variations to be considered a suitable choice for any application. A bit later, we published our work [4], which also deals with the influence of internal stresses in the magnetoelastic behaviour of nickel and provides a great amount of significant quantitative data related to this topic. In the present work, not only do we want to provide further results, details and discussions related to the microscopic and internal stress characterization of this material, but also we include a qualitative discussion about the way in which this comprehensive characterization can help us to optimize its magnetoelastic response in accordance to different objective functions. Other works such as [5–7] have also dealt with this topic but they considered either different effects or different materials. In this work in particular, two salient magnetoelastic effects will be measured in order to analyze the influence of internal stress on magnetoelasticity in nickel: the dependence of elastic modulus and damping on the magnetic field, i.e., the so-called ∆E- and ∆Ψeffects. The interested reader may find a theoretical background about these effects and the influence of internal stress on them in several works such as [8].
2 Experimental details Pure crystalline rods of nickel 201, all from the same molten material and manufacturing process, were used in this work. They were 110 mm in length and 10 mm in diameter and they showed a purity level of 99.90% and a density of 8912 kg/m3. The magnetoelastic properties were measured via a recently experimental system which we developed in [9] and enhanced later in [3]. It is based on laser Doppler vibrometry and it turns out to be a suitable method for measuring ∆Eand ∆Ψ- effects: not only is it able to measure both effects simultaneously and accurately, but it also allows us to obtain stress- and path-dependencies. The different heat treatments are achieved by modifying three main parameters: the heating temperature, the heating time and the cooling method. The selection of the different heating temperatures for our scheduled heat treatments, probably the more significant parameter from the point of view of internal stresses, was made in accordance with the recommendations found in [10] for nickel specimens. This led us to three different treatments: annealing (705–1205ºC), designed to produce a recrystallized grain structure and softening in work-hardened alloys; stress relieving (425–870ºC), used to reduce stresses in work-hardened non-age-hardenable alloys without recrystallizing the grain structure; and stress equalizing (230–315ºC), used to balance stresses in coldworked materials without appreciably decreasing mechanical strength. Table 1 presents a summary of the heat treatments. In it, each specimen has been identified by three digits (which represent the heating temperature in Celsius), two digits (which indicate the heating time in hours) and one letter (which refers
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
277
to the cooling method, i.e., water, air or furnace). The initial state of the nickel specimens is labelled ‘‘AR’’ (as received). Finally we want to emphasize that, leaving aside field-dependence, other influences such as stress-dependence and path-dependence must be considered in our measurements [3]. In order to avoid their influences, we will report only those results which were obtained under 0.50 MPa and following a magnetization path from the state of zero retentivity (Mr0) to the state of positive saturation magnetization (Ms+). Table 1: Specimen 0 1 2 3 4 5 6 7 8 9 10 11 12 13
Heating temp. (ºC) 300 300 600 600 900 900 900 900 900 900 900 900 900
Scheduled heat treatments. Heating time (h) 2 2 2 2 2 2 2 4 4 4 8 8 8
Cooling method Water Air Water Air Water Air Furnace Water Air Furnace Water Air Furnace
Nomenclature AR 300-02-W 300-02-A 600-02-W 600-02-A 900-02-W 900-02-A 900-02-F 900-04-W 900-04-A 900-04-F 900-08-W 900-08-A 900-08-F
3 Results and discussion 3.1 Microscopic analysis of grain size The scheduled heat treatments were carried out in a Carbolite Eurotherm 2416 furnace. The preparation procedure for microscopic examination of the specimens was made in accordance to [11]. In particular, the specimens were cut with a precision Struers Secotom-10 saw, then mounted in a hard epoxy resin with a Struers LaboPress-3 mounting press and finally ground and polished in a Struers TegraSystem machine (TegraPol-15 and TegraForce-1). The etchant applied to the surfaces was a fresh solution of one part HNO3 and one part acetic acid (glacial), which is the recommended etchant for revealing grain boundaries in nickel 201. For the microscopic examination, the equipment used was a Leica DM IRM inverted research microscope with a Leica DFC 480 high performance digital FireWire camera system which allowed us to obtain micrographies and estimations of the grain size in each specimen. Fig. 1 shows the micrographies for the 8 most significant specimens.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
278 Materials Characterisation V
Figure 1:
Micrographies of specimens (a) AR, (b) 300-02-W, (c) 300-02-A, (d) 600-02-W, (e) 600-02-A, (f) 900-02-W, (g) 900-02-F and (h) 900-08-W.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
279
On the one hand, specimens which were subjected to stress relieving (300-02W and 300-02-A) and stress equalizing (600-02-W and 600-02-A) processes show a grain size similar to that of specimen AR. This fact agrees with the fact that stress relieving and stress equalizing reduce and balance stresses without recrystallizing the grain structure. Besides, in spite of small variations, we can also observe higher grains in specimens which were subjected to stress equalizing due to the higher heating temperature and consequently the higher grain boundary diffusion. On the other hand, annealed specimens (900-02-W, 900-02-F and 900-08-W) clearly shows much larger grains, which will translate into lower internal stresses. Specimen 900-08-F has not been shown because it has got almost the same distribution of grains due to the long heating time, which means high grain boundary diffusion even when cooled in water. Furthermore, the concentration of small grains which appears in specimen 900-02-W corresponds to the centre of the specimen, where the heating time was not high enough to continue increasing the size of inner grains (the diffusion processes goes from the outside to the inside of the rod). Table 2: Specimen AR 300-02-W 300-02-A 600-02-W 600-02-A 900-02-W 900-02-A 900-02-F 900-04-W 900-04-A 900-04-F 900-08-W 900-08-A 900-08-F
Microscopic measurements.
Small grains Perim. Area (µm) (µm2) 350 8200 440 11900 520 14400 600 16900 700 21500 530 16600 620 20700 620 23600 -
Medium grains Perim. Area (µm) (µm2) 1240 41000 1240 53000 1720 144000 1730 157000 1660 147000 1640 137000 1820 167000 1750 143000 1960 194000 1900 195000 1870 170000
Large grains Perim. Area (µm) (µm2) 4000 590000 4200 510000 4100 650000 4200 670000 4400 670000 4700 660000 5300 780000 5300 720000 5600 940000
Table 2 shows the measurements of grain perimeter and grain area for each specimen. The results seemed to indicate the existence of three different distributions: small grains, with a perimeter lower than 1000 mm; medium grains, with a perimeter between 1000 and 3000 mm; and large grains, with a perimeter greater than 3000 mm. Specimens AR, 300-02-W and 300-02-A only contain small grains, although perimeters and areas increase slightly when the specimens have been heated. Specimens 600-02-W and 600-02-A show small grains mainly although medium grains are not uncommon. Finally, all specimens heated to 900ºC show large distributions of medium and large grains due to WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
280 Materials Characterisation V recrystallization, whereas small grains tend to disappear (apart from the previously mentioned significant distribution of small grains in the centre of the section of specimens which were subjected to a time of only 2 hours). In general, the grain size increases with heating time and decreases with cooling velocity. 3.2 Internal stress analysis Internal or residual stress is an extrinsic property which cannot be measured directly but through the previous measurement of an intrinsic property of the material. In this particular case, we have used the strain in the crystal lattice as the intrinsic variable which will allow us to estimate the residual stresses of our specimens. Among all the ways to measure the strain in the crystal lattice, the XRay diffraction technique clearly stands out. X-Ray diffraction techniques can be used to determine both macroscopic (homogeneous) and microscopic (inhomogeneous) residual stresses [12, 13]. In this work we only take into account microstresses, which are those involved in cold-work, plastic strains or heat treatments and lead to the accumulation of a residual dislocation network, producing inhomogeneous strain and an irreversible broadening of the Bragg peaks in X-Ray diffraction. Indeed, the measurement of said broadening of the diffraction peak will be used to estimate the internal stresses of our specimens. Strictly speaking, we have used the Williamson–Hall method [12], which has proven its validity to estimate residual microstresses remaining in materials after heat treatments [13]. The X-Ray diffraction profiles (intensity vs. angle of diffraction) have been recorded using a PHILIPS X’Pert MPD X-Ray diffractometer with CuKa radiation (wavelength λ=1,5405Å) and varying the angle of diffraction (θ) from 40º to 100º with a scan step of 0.04º and a time per step of 5 seconds. Fig. 2 and Fig. 3 show, respectively, the X-Ray full diffraction profile for the specimen AR and the fitting procedure of one peak of the X-Ray diffraction profile for the specimen 900-02-W. No other plots are included since the peaks position in all the specimens remain the same and the variation in width is not high enough to be easily noticed at a glance. The width of a peak can be calculated via two different ways: as the integral breadth β, defined as the width of a rectangle having the same area and height as the observed line profile, or as the full width at half maximum FWHM. WinPLOTR software allowed us to obtain both of them for several reflections as a measure of the peak width, but we only used the integral breadth in our estimations. The strain calculated using WinPLOTR is the so-called maximum strain. Besides, a diffraction pattern has contributions from the instrument optics (even a perfect sample will give lines of finite width), from particle size and from microstrain. The way in which these contributions are added depends on the shape of the peak. In our case, we assume that peaks are Lorentzian in shape, so all these contributions can be directly added: (1)
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
281
Figure 2:
X-Ray diffraction profile for the specimen AR obtained in the software WinPLOTR.
Figure 3:
Detail of the second peak of the X-Ray diffraction profile for the specimen 900-02-W and the fitting procedure in WinPLOTR.
Now, considering that we can make the correction for the instrumental broadening, the corrected integral breath due to microstrain and particle size can be written as follows: .
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
(2)
282 Materials Characterisation V
Figure 4:
Williamson-Hall plots of specimens AR, 300-02-W, 600-02-A and 900-08-F.
where the first addend is the particle size contribution and the second addend is the strain (ε) contribution. Then, left-multiplying by cosθ we obtain: cos
.
4 sin
(3)
This expression leads to the well-known Williamson-Hall plot [12], which is the plot of βcosθ against sinθ. Thus, the strain is directly related to the slope of said plot. Fig. 4 shows some selected Williamson-Hall plots. Table 3 gathers all the internal stresses obtained from the Williamson–Hall plots. In addition, microstrain, demagnetized elastic modulus and lattice parameter have been also included, this latter being known very easily once we know the wavelength of the radiation and the angle of diffraction of each peak. Regarding internal stresses, results provided in Table 3 are in good agreement with what one would expect given the heat treatments carried out. Indeed, we can distinguish three different ranges of residual stresses: High residual stresses (over 500 MPa). These values are shown by the specimens which were subjected to stress equalizing (230–315ºC and air or furnace cooling [10]). Water quenching was also tested obtaining, as expected, the highest value. Medium residual stresses (between 300 and 500 MPa). These values correspond to the specimen AR and to those which have been subjected to stress relieving treatments (425–870ºC and air or furnace cooling [10]). As the name of the treatment indicates, a small stress reduction is achieved. Low residual stresses (below 300 MPa). These values are shown by samples that have been subjected to annealing process (705–1205ºC and air or furnace cooling [10]). In these cases, a high reduction of internal stresses is achieved. Nevertheless, if water quenching is applied, the internal stresses jump into a level of medium stresses. WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
283
Regarding the lattice parameter, no significant variations were found, but the obtained value is in good agreement with the typical values of the lattice parameter in nickel samples (around 3.5 Å). Table 3: Internal stresses and lattice parameter. Specimen AR 300-02-W 300-02-A 600-02-W 600-02-A 900-02-W 900-02-A 900-02-F 900-04-W 900-04-A 900-04-F 900-08-W 900-08-A 900-08-F
Internal stress (MPa) 479.0 579.6 536.6 475.2 498.3 469.0 276.9 223.4 356.6 248.6 106.8 415.8 207.6 86.2
Microstrain (%) 0.23 0.27 0.25 0.22 0.24 0.22 0.14 0.11 0.17 0.13 0.05 0.20 0.10 0.04
Elastic modulus (GPa) 213 213 213 216 210 220 201 199 213 203 203 216 208 203
Lattice parameter (Å) 3.53 3.53 3.53 3.53 3.53 3.53 3.54 3.53 3.54 3.54 3.54 3.52 3.53 3.53
3.3 ∆E- and ∆Ψ-effects The results corresponding to the ∆E- and ∆Ψ-effects of the 14 nickel specimens considered show similarities which can be summarized mainly into two different patterns which are depicted in Fig. 5 and explained next: Pattern I is shown by specimen AR and all those which were subjected to heat treatments of stress relieving and stress equalizing. Indeed, specimens subjected to a heating temperature of 900ºC and water cooled also may be grouped into this patter. On the one hand, the curve of elastic modulus is compound by an initial stage of rapid growth which belongs to the low magnetic field range (less than 150 Oe) and a second stage of slow growth until saturation (around 350 Oe). On the other hand, the curve of damping shows again two stages, an initial rising stage which corresponds to the low applied magnetic field range and the second declining stage until saturation. In both cases, variations between the values at demagnetized and saturated states are lower than those found in pattern II. Pattern II is shown by the specimens which were heated up to 900ºC and slowly cooled down in air or inside the furnace, so they present a very low level of internal stresses. The elastic modulus curves show the same behaviour as in pattern I but variations between the values at demagnetized and saturated states are greater. As far as the damping curves are concerned, they show a very high damping value which quickly start to decrease until saturation without intermediate peaks. WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
284 Materials Characterisation V
Figure 5:
Figure 6:
Different patterns (I: black; II: gray) of the ∆E- and ∆Ψ- effects.
Influence of heating time on (a) ∆E- and (b) ∆Ψ- effects (black: 900-02-A; dark gray: 900-04-A; light gray: 900-08-A).
Since it is not the main goal of this work, we will not include results about the ∆E- and ∆Ψ- effects of each treated specimen. The interested reader may consult such results in our work [4]. Regarding the influence of the parameters involved in the heat treatments, we can claim than the heating temperature is the most significant parameter since only temperatures higher than 700ºC (900º in our tests) may lead to a pattern II behaviour. The cooling methods also stand out as a significant parameter since water cooling may avoid the grain diffusion and increase internal stresses, what would lead to a pattern I behaviour. Finally, the heating time has a very minute influence on the magnetoelastic behaviour when using long cooling methods such as air cooling or furnace cooling. This can be seen in Fig. 6. Another important feature we should analyze is the influence of the heat treatments and the internal stresses on the magnetic field from which the elastic modulus or the magnetomechanical damping remains the same, i.e., on the magnetic field which makes the ∆E- and ∆Ψ- effects saturate. We can conclude that in all the cases the saturating magnetic field is around 350 Oe, so it does not depend on any of the heat treatment parameters which were considered. WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
285
4 Conclusions The previous results can guide us in selecting the most suitable heat treatment in order to make nickel show the smart properties we desire. It is important to notice that the optimization will depend on the objective function we select. Smart materials are usually required to provide a maximum variation of the salient magnitude by supplying a minimum modification of the input. This fact, from the point of view of a magnetoelastic material, means large variations in applied forces, natural frequencies and damping ratios obtained via small changes in the applied magnetic field. Nevertheless, we should not discard that, in some cases, the objective is the opposite: to obtain a constant value of elastic modulus and damping along a range of magnetic field as large as possible. We can extract valuable information about the different possibilities of optimization from Fig. 5. On the one hand, if nickel is required to act as a smart material which provides the highest ∆E- and ∆Ψ- effects, then it should behave following pattern II. In addition, the characteristic damping peak of pattern I disappears, so we get a more predictable behaviour. Thus, in the light of section 3 results, the material should show very low internal stresses, which can be achieved via annealing heat treatments with slow cooling velocities (900-0X-A or 900-0X-F). On the other hand, one may need that nickel show elastic and damping properties as constant as possible under different applied magnetic fields. Although this behaviour cannot be perfectly fitted, it is possible to minimize the ∆E- and ∆Ψ- effects by making the material follow pattern I. Again, we know from discussion in section 3, that this means getting internal stresses as high as possible, which can be achieved via work-hardening, stress equalizing heat treatments and fast cooling velocities (AR, 300-0X-W). Leaving aside optimization rules, this work also provides relevant results, details and discussions related to the microscopic and internal stress characterization of this material. We have carried out an in-depth material characterization in terms of grain distribution and internal stresses when nickel specimens are subjected to several different heat treatments. On the one hand we have included many micrographies which help us to understand not only the way does each heat treatment modify the grain size, but also how the grain diffusion acts depending on the heating temperature, the heating time and the cooling method. On the other hand we have carefully described the way we have obtained the microstrain, internal stress and lattice parameter in all our specimens via X-Ray diffraction and the Williamson-Hall method. All these results agree with the fact that the internal stress decreases when we increase the heating temperature and decrease the cooling velocity, the heating time becoming an almost negligible parameter.
References [1] du Trémolet de Lacheisserie, E., Magnetostriction: Theory and Applications of Magnetoelasticity, CRC Press: Boca Raton, 1993.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
286 Materials Characterisation V [2] Jiles, D.C., Recent advances and future directions in magnetic materials, Acta Materialia, 51, pp. 5907–5939, 2003. [3] Morales, A.L., Nieto, A.J., Chicharro, J.M. & Pintado, P., Field-dependent elastic modulus and damping in pure iron, nickel and cobalt, Journal of Magnetism and Magnetic Materials, 322, pp. 1952–1961, 2010. [4] Morales, A.L., Nieto, A.J., Chicharro, J.M. & Pintado, P., Influence of internal stresses on field-dependent elastic modulus and damping in pure nickel, Journal of Magnetism and Magnetic Materials, 322, pp. 3584–3594, 2010. [5] Motogi, S. & Maugin, G.A., Elastic-moduli of demagnetized polycrystalline ferromagnets, Journal of Physics D-Applied Physics, 26, pp. 1459–1467, 1993. [6] Squire, P.T., Atkinson, D., Gibbs, M.R.J. & Atalay, S., Amorphous wires and their applications, Journal of Magnetism and Magnetic Materials, 132, pp. 10–21, 1994. [7] Kaczkowski, Z., Magnetomechanical properties of rapidly quenched materials, Materials Science and Engineering A—Structural Materials Properties Microstructure and Processing, 226–228, pp. 614–625, 1997. [8] Bozorth, R.M., Ferromagnetism, D. van Nostrand: Toronto, 1951. [9] Morales, A.L., Nieto, A.J., Chicharro, J.M. & Pintado, P., Automatic measurement of field-dependent elastic modulus and damping by laser Doppler vibrometry, Measurement Science and Technology, 19, 2008. [10] International ASM, ASM Handbook Vol. 4—Heat Treating, vol. 4, ASM International, 1991. [11] International ASM, ASM Handbook Vol. 9—Metallography and Microstructures, vol. 9, ASM International, 2004. [12] Williamson, G.K. & Hall, W.H., X-Ray line broadening from filed aluminium and wolfram, Acta Metallurgica, 1, pp. 22–31, 1953. [13] Prevéy, P.S., The measurement of subsurface residual stress and cold work distributions in nickel base alloys, ASM’s Conference on Residual Stress in Design, Process and Materials Selection, ASM International, Cincinnati (Ohio, USA), pp. 27–29, 1987.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
287
Nanomechanical structure-property relations of dynamically loaded reactive powder concrete P. G. Allison1, R. D. Moser2, M. Q. Chandler1, T. S. Rushing1, B. A. Williams1 & T. K. Cummins1 1 2
US Army Engineer Research & Development Center, USA Georgia Institute of Technology, USA
Abstract Low water-to-cement ratio (w/c) reactive powder concretes (RPCs) exhibit much higher compressive strengths compared to conventional concrete through optimized particle packing and specialized curing regimes. The high strain-rate impact behavior of RPCs was investigated at the macroscale. However, little work has been done to study the fundamental material behaviors and failure mechanisms of RPC under high strain impact and penetration loads at lower length scales. These high strain-rate loadings have many possible effects on RPCs at the microscale and nanoscale, including alterations in the composition and bonding present in hydrated phases such as calcium silicate hydrate (C-S-H), in addition to fracture and debonding. In this work, the possible chemical and physical changes of RPCs under high strain-rate impact and penetration loads were investigated using a novel technique wherein nanoindentation measurements were spatially correlated with chemical composition using electron microscopy. Results indicate that high strain-rate impacts degrade both the elastic modulus and indentation hardness of RPCs and, in particular C-S-H, with damage likely occurring due to microfracturing and debonding. Additional studies will be required to better understand degradation phenomena within C-S-H itself. Keywords: nanoindentation, SEM, EDX, RPC, UHPC, C-S-H.
1 Introduction Concretes and Portland cement-based materials are the most-produced manmade materials on earth, with over twenty billion tons produced per year. The WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line) doi:10.2495/MC110251
288 Materials Characterisation V ubiquitous use of cementitious materials within protective infrastructures makes quantifying the response of cementitious materials to high strain-rate penetration and impact loads a critical need as designs become more reliant on computational tools. Low water-to-cement ratio (w/c) reactive powder concretes (RPCs) exhibit superior compressive strength and ductility (when fibers are admixed) compared to conventional concrete and is increasingly being used for high strain-rate loading applications [1]. To fully explore the potential of this material for high strain-rate loading applications, its fundamental failure mechanisms under impact and penetration loads have to be investigated. RPC is a highly heterogeneous material with a microstructure consisting of hydrated cement paste (HCP), unhydrated cement particles, fine aggregates, potentially steel fibers and/or polymer fibers, and pores ranging from nanometer to millimeter in diameter [2]. The addition of pozzolanically reactive silica fume results in an HCP comprised primarily of calcium silicate hydrate (C-S-H). C-S-H is the most important phase of the Portland cement hydration process and functions as the binding component that holds the various other phases of RPC together. C-S-H consists of physically and chemically bound water in nanometer-scale gels, bulk water in gel and capillary pores, adsorbed water on the surfaces of gels, and may behave nano-granularly [3]. Grady [4] suggests that C-S-H may go through chemical changes such as dehydration or vaporization under shock impact loading. Instrumented indentation; namely nanoindentation, techniques were used to quantify the structure-property relations of concrete at lower length scales. Velez et al. [5] performed nanoindentation tests to quantify the elastic modulus and hardness of synthetically manufactured Portland cement clinker phases. Hughes and Trtik [6] used depth-sensing nanoindentation and energy dispersive X-ray (EDX) analysis to correlate the major phase compositions and mechanical properties of hydrated cement paste. Ulm et al. [7] developed a novel statistical nanoindentation technique to characterize cement paste and were able to identify the existence of two distinct types of C-S-H, LD C-S-H and HD C-S-H. DeJong and Ulm [8] and Constantinides and Ulm [9] used a similar approach to study the degrading mechanisms of calcium leaching and high temperature on C-S-H. Sorelli et al. [10] also used similar techniques to characterize the properties and volume fraction of different phases in Ultra High Performance Concrete (UHPC). Their research showed that UHPC with a low w/c (0.2) has a much higher volume fraction of HD C-S-H and unhydrated clinker than LD C-S-H compared to concrete with higher w/c. In this work, the influence of impact loadings on the nanomechanical properties of RPCs is investigated. Specimens were extracted for impacted and non-impacted panels of RPC. A novel technique coupling nanoindentation with spatially correlated scanning electron microscopy (SEM) and chemical analysis using energy dispersive X-ray spectroscopy (EDX) was developed to characterize damage due to high strain-rate impact loadings in the RPC panels.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
289
Based on the results of these studies, a possible mechanism for microstructural damage in RPCs is proposed, and the caveats associated with the techniques utilized are discussed.
2 Experimental procedures 2.1 Materials An RPC mixture with a w/c of 0.2 and an unconfined compressive strength of 180 MPa was cast into panels 305-mm wide by 305-mm long and 25.4-mm thick for ballistic impact testing. The panels were impacted by 11-mm-diameter steel spheres at half the ballistic limit velocity, which was determined according to MIL-STD-662 [11]. The mixing, casting, curing, and ballistic testing is detailed in the manuscript by Rushing et al. [2]. 2.2 Coupled nanoindentation and SEM/EDX analysis Analysis of impacted and non-impacted RPC specimens was performed using a novel technique of nanoindentation coupled with SEM imaging and EDX chemical analysis. In this coupled method, a large number of indents are performed over a standardized indentation grid placed on the RPC sample, after which, each indent is spatially correlated using SEM to obtain an image and EDX to determine the chemical composition. The resulting dataset contains the nanomechanical properties and chemical composition along with an image at each indentation site. This coupled method allows for improved differentiation between the various components present in RPCs and can be used to better correlate alterations in nanomechanical (e.g., due to impact loadings) properties to specific microstructural features. Details on specimen preparation techniques and experimental methods are described below. 2.2.1 Specimen preparation Specimens were extracted from the panel in the impacted zone and a nonimpacted zone as shown in Figure 1. The center of the non-impacted location was 38 mm from both the top edge and side of the panel. This location was selected to minimize the potential for edge effects from casting while also avoiding the damaged zone to the greatest extent. A 25.4-mm-diameter diamondtipped coring bit was used to core the non-impacted specimens, which were then cast into 31.8-mm-diameter cylindrical molds using EpoHeat low-viscosity epoxy supplied by Buehler. The impacted specimen was sectioned using an oilcooled Struers Secotom high precision cut-off saw. The cross section of the specimen was placed into a 31.8-mm-diameter cylindrical mold and mounted in EpoHeat epoxy from Buehler. After the epoxy fully cured, the samples were sectioned in half by the oil-cooled cut-off saw to obtain a cross section from the center of the panel, thus limiting any surface effects such as laitance.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
290 Materials Characterisation V
Nonimpacted specimen location Line on crater represents edge of crosssection analyzed
Impacted specimen location
Figure 1:
Location of impacted and non-impacted specimens removed by coring from the low w/c RPC panel.
A Buehler Ecomet/Automet 250 automatic polishing wheel was used to polish the extracted specimens. The polishing procedure utilized a 240-grit diamond polishing pad at 230 RPM until the specimen was planar, followed by an UltraPadTM with a 9-µm diamond paste at 130 RPM for 5 minutes, and a third polishing step using a TriDentTM pad with a 3-µm diamond paste for 5 minutes. These first three polishing steps used a 50:50 mixture of ethylene glycol and ethanol for a lubricant. Final polishing utilized a ChemoMet® pad with 0.05-µm alumina in ethylene glycol for 5 minutes. All steps used a force of 30 N for polishing. Once polished, the samples were desiccated prior to indentation. 2.2.2 Nanoindentation Polished specimens were examined using an Agilent Technologies G200 nanoindenter to probe microstructural changes across the specimens. Indentations were performed using a pyramid-shaped diamond Berkovich indenter with a tip radius of approximately 20 nm. Prior to each measurement, a 2nd-order area function calibration was performed using a fused silica reference material. Load controlled indentation measurements were performed up to a maximum load of 2 mN at a loading rate of 0.2 mN/s followed by a hold time of 5 s and a 10 s unloading period. Prior to performing the nanoindentation experiments, each specimen was examined in the SEM to create a “map” of images approximately 3 mm by 3mm near the indentation site. This “map” was then used to find a desired location for the indentation grid (i.e., not containing large voids and/or surface defects). For each specimen, a total of 500 indents were placed with a spacing of 10 µm in the X-direction and 20 µm in the Y-direction. Following the indents performed for nanomechanical measurements, 100-mN fiduciary indents were WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
291
placed at a spacing of 245 µm in the X-direction and 20 µm in the Y-direction from the first indent to aid in identifying the start, middle, and end of each line of indents so that the indentations could be precisely located using the SEM. Assuming a Poisson’s ratio of 0.2, mechanical properties such as elastic modulus and hardness were determined for the interaction region of the indentation site [12]. 2.2.3 SEM and EDX measurements Specimens were examined both before and after nanoindentation using an FEI Nova NanoSEM 630 field emission SEM. This device is equipped with lowvacuum capabilities, making it ideal for examining nonconductive cement-based materials without special sample preparation or metallic coating. Imaging was performed at an accelerating voltage of 10 kV using a backscattered electron (BSE) detector to reveal changes in microstructure and the distribution of phases according to their respective densities. When examined after nanoindentation, the “map” of images was used to generally locate the indentation grid, while the fiduciary indents were used to determine the location of each line of indents such that each indent could be correlated with a location on the image. Point chemical analyses were also performed in conjunction with SEM imaging using a Bruker solid-state EDX detector installed in the FEI SEM. Through proper alignment of the indentation grid facilitated by the fiduciary indents, a standardized point chemical analysis grid was developed that resulted in a dataset of point chemical analyses that were spatially correlated with the location of each indent.
3 Results and discussion The highly variable phase composition and distribution present in RPCs presents a variety of different nanomechanical properties that were sampled during nanoindentation measurements. Figure 2 presents the load vs. depth results of five representative indents corresponding to a fine aggregate, an anhydrous cement grain, C-S-H, and an aberrant test. Here, C-S-H was divided into two phases, namely low-density (LD) and high-density (HD), which was determined through molecular simulations and similar nanoindentation studies [9]. Of particular interest is the faulty or “aberrant” test result shown in Figure 2, which defies the traditional stiffening indentation curve typical for homogenous interaction regions but still exhibits an elastic modulus result similar to a homogenous phase (in this case HD C-S-H). These aberrant tests may present themselves as irregular loading and/or unloading curves (as shown in Figure 2). In heterogeneous cement-based materials, aberrant test results may occur due to the presence of voids, polishing defects, cracking during indentation, and as noted more recently, composite or “nanocomposite” multiphase response of material present in the interaction region of the indent [13–15]. The multiphase response of cement-based materials was a topic of recent discussion in the literature and represents an issue that may diminish the utility of nanoindentation as a quantitative microstructural characterization technique. Thus, it is critical to examine each indentation curve and remove aberrant results if reliable WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
292 Materials Characterisation V quantitative information is desired. All indentation curves in the present study were reviewed, and 8% in impacted specimens and 24% in non-impacted specimens were deemed aberrant and removed from the dataset. 2.25
LD C‐S‐H HD C‐S‐H Aggregate Cement Aberrant Test
2.00 1.75
Load (mN)
1.50 1.25 1.00 0.75 0.50 0.25 0.00 0
25
50
75
100
125
150
175
200
225
250
Depth (nm)
Figure 2:
Typical load vs. indentation depth curves for LD and HD C-S-H, aggregate, cement, and an aberrant test result.
Figure 3 (a) illustrates a typical grid of 500 measurement indentation points and fiduciary indents superimposed on a BSE micrograph of an RPC specimen. Using BSE imaging, the anhydrous cement and silica fume can be seen with a bright signature, followed by fine aggregates and silica flour, and finally by the HCP (i.e., solid products of cement hydration with low density) and voids that appear darkest in the image. The HCP appeared to be comprised primarily of C-S-H, with all Ca(OH)2 likely consumed by pozzolanic reactions and subsequently converted into C-S-H. Figures 3 (b) and (c) present contour maps of Ca:Si ratio and elastic modulus results, respectively, corresponding to the indentation grid shown in Figure 3 (a). A clear correlation can be observed between the location of the various components of the RPC and their respective composition and mechanical properties. In addition, the benefits of using this coupled technique are also particularly apparent when trying to differentiate between phases with similar properties. For example, fine aggregate particles and anhydrous cement present in RPCs may exhibit similar mechanical properties, making phase identification/quantification from only nanoindentation results a challenging task. However, when nanoindentation measurements are coupled with chemical composition at the indentation site, the distinction between fine aggregates (with low Ca:Si approaching zero) and cement (with high Ca:Si of 5 to 7) becomes clear. Similar comparisons can be made for the various phases of cement hydration present in the HCP. WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
293
(a) SEM-BSE micrograph of typical RPC microstructure with measurement indentation grid and fiduciary indents superimposed.
(b) Map of Ca:Si corresponding to indentation grid.
(c) Map of elastic modulus (GPa) corresponding to indentation grid. Figure 3:
Typical results from coupled nanoindentation and SEM/EDX studies performed on polished RPC specimens.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
294 Materials Characterisation V Another feature of the nanoindentation measurements clearly shown in the elastic modulus contour map (see Figure 3 (c)) is the gradual transition in nanomechanical properties present at interfaces between two phases. This behavior is likely the result of a composite response of material present within the interaction region below each indent as discussed above. With the aberrant results removed from the dataset, further analysis of “valid” nanoindentation results was performed. Figures 4 and 5 present histograms of the elastic modulus and hardness results of all “valid” indents performed on nonimpacted and impacted specimens in the present study. Results presented in Figures 4 encompass those of anhydrous cement, siliceous fine aggregates, silica flour, the HCP (primarily composed of C-S-H), and voids/porosity. Impacted specimens exhibited a mean elastic modulus of 47.9 GPa compared with 76.7 GPa in non-impacted specimens. In particular, significant reductions in the proportion of indents with elastic moduli between 60 GPa and 110 GPa was observed in impacted specimens, a range common for silica flour and siliceous fine aggregates [10]. 22 20 Non‐Impacted Impacted
18
Frequency (%)
16 14 12 10 8 6 4 2
200
190
180
170
160
150
140
130
120
110
100
90
80
70
60
50
40
30
20
10
0
Elastic Modulus (GPa) Figure 4:
Histogram of elastic modulus results from nanoindentation experiments performed on non-impacted and impacted RPC specimens.
The effect of impact (Figure 5) was much more pronounced in hardness measurements, where a large shift in hardness from a mean of 6.3 GPa to 2.7GPa was observed. In impacted specimens, there was a particularly high increase in the proportion of indents with hardness between 0.5 and 1.5 GPa associated with a reduction in the proportion of indents with hardness above 4 GPa.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
295
16 47.1% 14 Non‐Impacted Impacted
Frequency (%)
12 10 8 6 4 2
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
Indentation Hardness (GPa)
Figure 5:
Histogram of indentation hardness results from nanoindentation experiments performed on non-impacted and impacted RPC specimens.
While these nanoindentation results suggest that impact loadings have a significant influence on both the elastic modulus and hardness of RPCs, it is difficult to determine by what possible mechanism degradation occurs. In order to further evaluate the possible mechanisms causing the observed degradation in nanomechanical properties, indents likely associated with homogenous C-S-H regions were extracted from the full dataset of non-impacted and impacted RPC specimens. Indents identified as C-S-H were selected based on a Ca:Si between 1 and 2 (consistent with C-S-H) with an elastic modulus between 10 and 50 GPa (range typical for LD and HD C-S-H [10, 13]. Out of each dataset, approximately 10% of indents were deemed as homogenous C-S-H. Ca:Si ratios of the C-S-H sub-dataset were 1.44 and 1.42 for non-impacted and impacted specimens, respectively. The histograms shown in Figures 6 and 7 depict the distribution in C-S-H elastic modulus and indentation hardness for non-impacted and impacted specimens. C-S-H present in non-impacted specimens exhibited a mean elastic modulus of 32.5 GPa and mean indentation hardness of 1.56 GPa. C-S-H present in impacted specimens exhibited a mean elastic modulus of 27.4 GPa and mean indentation hardness of 0.66 GPa, with a significant increase in hardness between 0.25 and 0.75 GPa. These reductions in elastic modulus and, in particular, hardness in impacted samples suggests that the C-S-H, in addition to the overall microstructure of RPCs, is degraded under high strain-rate impact loadings. The dehydration/vaporization mechanism proposed by DeJong and Ulm [8] was shown as a possible deterioration mechanism resulting in decreased C-S-H packing factors and in turn reduced elastic modulus and hardness as measured by nanoindentation. However, unless additional secondary chemical bonding was to WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
296 Materials Characterisation V occur between C-S layers (similar to irreversible creep mechanisms [16] during the vaporization event, it is likely that rehydration of the C-S-H would occur slowly. Furthermore, if such bonding were to occur; densifying the C-S-H and limiting rehydration, the elastic modulus and hardness would likely increase rather than decrease. Based on the results presented in Figures 4 and 5, it is clear that a majority of degradation in the full RPC indentation dataset is associated with reductions in elastic modulus and hardness associated with inert particles and unhydrated cement. Therefore, it is likely that a majority of degradation in RPCs following impact loading results from microfracturing and/or debonding. 30 Non‐Impacted
Frequency (%)
25
Impacted
20 15 10 5
50
45
40
35
30
25
20
15
0
Elastic Modulus (GPa)
Figure 6:
C-S-H elastic modulus of impacted and non-impacted specimens.
Frequency (%)
40 35
Non‐Impacted
30
Impacted
25 20 15 10 5
2.25
2.00
1.75
1.50
1.25
1.00
0.75
0.50
0.25
0
Indentation Hardness (GPa)
Figure 7:
C-S-H indentation hardness of impacted and non-impacted specimens.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
297
4 Conclusions A novel statistical nanoindentation technique was developed that spatially correlated the location of indents performed with an image and chemical analyses using EDX. Using this method, the influence of high strain-rate impact loadings on the nanomechanical properties of RPCs was determined. Significant degradation in both the elastic modulus and hardness occurred due to impact loadings of RPCs. Closer examination of C-S-H in particular also showed degradation. However, a link between chemical changes in the C-S-H and deterioration in nanomechanical properties could not be made. Based on analysis of all indents performed on RPC specimens, it is likely that a much of the degradation in elastic modulus and hardness observed stems from microfracturing and debonding, which occurs due to the impact. Future work involves additional experimental studies and quantitative data analysis to further evaluate the possible mechanisms of degradation present in RPCs.
Acknowledgement Permission to publish this article was granted by Director, Geotechnical & Structures Laboratory.
References [1] Millard, S.G., et al., Dynamic enhancement of blast-resistant ultra high performance fibre-reinforced concrete under flexural and shear loading. International Journal of Impact Engineering, 2010. 37: p. 405-413. [2] Rushing, T.S., et al., Independent effects of matrix strength and fiber reinforcement on concrete's ballistic resistance. SAVIAC. [3] Constantinides, G. and F.-J. Ulm, The nanogranular nature of C-S-H. Journal of the Mechanics and Physics of Solids, 2007. 55: p. 65-90. [4] Grady, D., Shock equation of state properties of concrete, in Proc. of Structures under Shock and Impact IV, N. Jones et al., Editor. 1996, Computational Mechanics Publications: Southampton, UK. p. 405-414. [5] Velez, K., et al., Determination by nanoindentation of elastic modulus and hardness of pure constituents of Portland cement clinker. Cement and Concrete Research, 2001. 31: p. 555-561. [6] Hughes, J.J. and P. Trtik, Micro-mechanical properties of cement paste measured by depth-sensing nanoindentation: a preliminary correlation of physical properties with phase type. Materials Characterization, 2004. 53: p. 223-231. [7] Ulm, F.-J., et al., Statistical indentation techniques for hydrated nanocomposites: concrete, bone, and shale. Journal of the American Ceramic Society, 2007. 90(9): p. 2677-2692.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
298 Materials Characterisation V [8] DeJong, M.J. and F.-J. Ulm, The nanogranular behavior of C-S-H at elevated temperatures (up to 700 oC). Cement and Concrete Research, 2007. 37(1): p. 1-12. [9] Constantinides, G. and F.-J. Ulm, The effect of two types of C-S-H on the elasticity of cement-based materials: Results from nanoindentation and micromechanical modeling. Cement and Concrete Research, 2004. 34: p. 67-80. [10] Sorelli, L., et al., The nano-mechanical signature of ultra high performance concrete by statistical nanoindentation techniques. Cement and Concrete Research, 2008. 38(12): p. 1447-1456. [11] MIL-STD-662F, V50 ballistic test for armor. 1997. [12] liver, W.C. and G.M. Pharr, Measurement of hardness and elastic modulus by instrumented indentation: Advances in understanding and refinements to methodology. Journal of Materials Research, 2004. 19(1): p. 3-20. [13] Chen, J.J., et al., A coupled nanoindentation/SEM-EDS study on low water/cement ratio Portland cement paste: evidence for C-S-H/Ca(OH)2 nanocomposites. Journal of the American Ceramic Society, 2010. 93(5): p. 1484-1493. [14] Davydov, D., M. Jirasek, and L. Kopecky, Critical aspects of nanoindentation technique in application to hardened cement paste. Cement and Concrete Research, 2010. [15] Trtik, P., B. Munch, and P. Lura, A critical examination of statistical nanoindentation on model materials and hardened cement pastes based on virtual experiments. Cement & Concrete Composites, 2009. 31: p. 705-714. [16] Mehta, P.K. and P.J.M. Monteiro, Concrete: Microstructure, Properties, and Materials. 2006: McGraw-Hill Companies Ltd.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
299
Dynamic strength of concrete under multiaxial compressive loading Y. P. Song1 & H. L. Wang2 1
State Key Laboratory Of Coastal and Offshore Engineering, Dalian University of Technology, China 2 Civil and Architectural Engineering College, Dalian University, China
Abstract The dynamic characteristics of concrete are important for structures subjected to earthquake actions. Now there is much experimental data about this, and most are uniaxial test data. In large concrete structures, such as concrete dams, concrete reactive power, the stress states usually are in multiaxial states; therefore the main objective of this paper is to study the multiaxial compressive characteristics of concrete subjected to high strain rates under multiaxial compressive loading. Cubic specimens (100mm by side) are subjected to quasistatic and dynamic proportional biaxial and triaxial compression tests. The strain rates used are 10-5/s , 10-4/s, 10-3/s, 10-2/s. The stress ratios are for the biaxial compression 1:0.1, 1:0.25, 1:0.5, 1:0.75,1:1; for the triaxial compression the applied constant confining pressures are 4MPa, 8MPa, 12MPA, 16MPa. The tests are carried out on a concrete triaxial dynamic test system designed by the authors. The strength characteristics of the specimens at different strain rates and stress ratios are given. The experiment results indicate that the dynamic strength of concrete under multiaxial compressive stress states is higher than that under the uniaxial compressive stress state. Based on the test data, the failure criterion is established on the octahedral stress space. Its characteristic is that the effect of the similar angle on the dynamic strength is considered to reflect the changes of the dynamic strength between the tensile and compressive meridians. Keywords: dynamic strength, strain rate, biaxial stress state, triaxial stress state.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line) doi:10.2495/MC110261
300 Materials Characterisation V
1 Introduction Dynamic loading on concrete structures arising from natural hazards such as tornadoes, earthquakes and ocean waves is of great practical significance. Under such dynamic conditions, the loading-rate dependence of material response causes the material behaviour to be significantly different from what is observed under quasi-static conditions. Hence a thorough knowledge of material constitutive relationships and failure criterion, which cover a wide range of strain rates, is very important for the design of structures subjected to all types of loading likely to be encountered during the design lifetime. Bischoff and Perry [1] reviewed and analyzed the response of concrete under dynamic loads and discussed factors that influence the dynamic compressive behaviour of concrete, such as concrete quality, aggregate type, age, curing, and moisture conditions. Malvar and Ross [2] reviewed the existing data describing the effects of strain rate on the compressive strength of concrete and compared the dynamic increase factor (DIF) formulation recommended by the European CEB Model Code [3]. However; extremely rare dynamic experiments in multiaxial stress states are available among the current documents. For material in multiaxial stress states, cases with biaxial compressive or triaxial compressive loading with the other side keeps constant confining pressure are the two typical loading patterns, which is of great significance for studying the behaviour under arbitrary multiaxial stress states. To improve understanding of the mechanical behaviour of concrete, experiments on the dynamic behaviour of concrete under biaxial and triaxial dynamic compression stress states were conducted in this research corresponding to the range of strain rates and stress states encountered in engineering practice.
2 Experimental program 2.1 Preparation of specimens Plain concrete cubes with a size of 100×100 ×100 mm were subjected to biaxial or triaxial dynamic compressive loads. Commercially available Portland cement was used. Crushed natural stones were used as coarse aggregate with maximum particle size of 20 mm. River sand was used as the fine aggregates. The concrete mixture proportions by weight are water: cement: fine aggregate: coarse aggregate =1.00:1.02:4.38:5.35. All specimens required for the aforementioned tests were cast on the same day for each mix and covered with a plastic sheet to prevent moisture loss. They were demoulded after 1 day and cured in the fog room at a relative humidity of 95% and temperature of 27±2°C till the age of 7 days. The 28-day compressive strength of concrete obtained by testing standard cube specimens (150mm×150mm×150mm) is 20 MPa. Then, the specimens are dried in air for 8 weeks before testing.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
301
2.2 Testing of specimens Dynamic tests were conducted on the servohydraulic multiaxial testing system designed and built at Dalian University of Technology, Dalian, China. The experimental apparatus is detailed in Figure 1 with a test setup in one loading direction illustrated in Figure 2.
Figure 1:
Figure 2:
Setup of concrete test.
Illustration of testing system in one loading direction.
The testing system allowed free and independent motion in three directions. Along each direction, a pair of pressure levers loaded a test specimen through two platens located on both sides of the specimen. A spherical hinge was installed between a lever and a platen on the same side of the specimen to ensure WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
302 Materials Characterisation V that the load was exerted exactly along the load axis. The two pressure levers were connected with a load transducer and an oil cylinder, respectively. The nominal capacity of the testing system was 2000 kN in each direction. The servo valve can respond to a command signal at a frequency of up to 10 Hz. Each specimen was instrumented with six linear variable differential transformers (LVDTs), two in each loading direction. Each LVDT had a stroke of 7 mm; it was attached to the two platens that were connected in series with the two opposite faces of a cubic specimen. The measured load and deformation were transmitted to the data acquisition and the processing unit of a computer through a specially allocated amplifier. They were then converted to stress and strain, respectively, using the undeformed area and length of the specimen. The selected loading paths consisted of uniaxial compression, biaxial proportional loading compression and triaxial compression with two constant lateral compressions. The strain rate varied from 10-5/s to 10-2/s. For biaxial proportional compressive loading, the stress ratios of lateral pressure to the axial load are 0:1、0.25:1、0.5:1、0.75:1、1:1. For triaxial compression with constant lateral pressure, the constant pressure is 0, 4, 8, 12, or 16 MPa respectively. To prevent lateral restraint of the loaded specimen, all the loaded surfaces are polished and equipped with three layers of plastic sheet with grease of MoS2 to reduce the surface friction to a minimum.
3 Experimental results and discussion Table 1 and Table 2 provide the dynamic strengths under biaxial and triaxial compression stress states. They represent the average value of each group of at least four specimens. Note that compressive stresses in all directions are positive in this paper. Table 1: Strain rate:/s 10-5 10-4 10-3 10-2
0 9.84 10.63 11.38 12.32
Strengths at different stress ratios(σ2:σ1) 0.25 0.5 0.75 14.86 16.13 16.39 15.48 16.68 16.75 16.17 17.36 17.54 17.15 18.24 18.66 Table 2:
Strain rate:/s 10-5 10-4 10-3
0 9.84 10.63 11.38
Biaxial test results.
4 30.05 32.11 33.70
1 14.00 15.32 16.66 18.01
Triaxial test results. Confinement σl /MPa 8 12 46.27 61.21 48.08 61.42 49.39 61.16
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
16 72.14 75.34 74.08
Materials Characterisation V
303
3.1 Strength characteristics under biaxial compression The ultimate strength of concrete in biaxial compression state is higher than the uniaxial strength at any strain rate owing to the effect of lateral confinement. At a specified strain rate, the strength increment depends on the biaxial stress ratio. The maximum biaxial strength occurs at a stress ratio between 0.5 and 0.75 for any strain rate investigated in the present study. With the increasing strain rate, the ultimate strength at any stress ratio tends to increase. However, the increment at different stress combinations is not identical. Based on the research cited above, it can be concluded that the strength enhancement of concrete in biaxial stress states is attributed to both the strain rate and the lateral confining pressure. A simple expression for the evaluation of dynamic strength of concrete in biaxial stress state is suggested: f bd P3 P4 . . P1 P2 lg s 2 2 fus 1 1
(1)
where s is the quasi-static strain rate, its value being selected as 10-5/s in this . paper; is the current strain rate; fus is the uniaxial compressive strength of concrete at quasi-static loading; fbd is the dynamic strength of concrete in biaxial stress state; α=σ2/σ1 is the stress ratio; P1, P2, P3 and P4 represent the parameters associated with material properties. By fitting to the test, data, P1, P2, P3 and P4 are determined as -0.446, 0.0875, 1.43 and 6.42 respectively. The multiple correlation coefficient being 0.9580, and the mean error being 0.4122 MPa. The suggested relationship in eqn (1) is depicted in Figure 3 and the test results are also shown for comparison. Fairly good agreement is achieved. .
Figure 3:
Biaxial strength envelop under biaxial stress obtained from the proposed criterion (c&c region).
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
304 Materials Characterisation V 3.2 Strength characteristics under triaxial compression In general, the strength enhancement of concrete in triaxial stress states is also attributed to both the strain rate and the lateral confining pressure. Newman [4] recommended a nonlinear relation between the ultimate strength and confining pressure. To take into account the effect of strain rates, Newman’s equation is modified to 2
f A( ) lat B() lat 1 c 0 f f f us us us
(2)
where A( ) a b lg( ) , B ( ) c d lg( ) ; fus is the uniaxial compressive strength of concrete at quasi-static loading rate ; σlat is confining pressure; the material constants a, b, c, and d are determined in this study to be 2.22, -1.54, 23.7, and 1.19, respectively. Figure 4 shows the ultimate strengths at various strain rates as a function of the confining pressure. For comparison, eqn (2) was plotted in Figure 4 as well. It can be seen from Figure 4 that eqn (2) is in excellent agreement with the test data. It should be noted that the values predicted by the formula only make sense when confining pressure is smaller than the uniaxial static strength of concrete.
(a)
(b)
(c) Figure 4:
Triaxial strengths of concrete at various strain rates.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
305
4 United failure criteria in octahedral stress space Based on the strength characteristic of experimental results in Table 1 and Table 2 and theoretical analysis of the failure enveloping plane, the present paper proposes a new failure criterion as the following:
oct f cs
a1 b1 oct c1 oct f cs f cs
2
3
(3)
where oct 1 1 2 3 , oct 1 1 2 2 2 3 2 3 1 2 3
1
2
;a1,b1,c1
are parameters depending on the loading rate, which can be determined by fitting to the test data, as listed in Table 3. Table 3: Strain rate/s-1 10-5 10-4 10-3 10-2
a1 0.2033 0.2865 0.3550 0.4187
Fitting results. b1 -0.9730 -0.7384 -0.5723 -0.4708
c1 -0.4816 -0.2948 -0.1647 -0.0975
R2 0.9547 0.9497 0.9505 0.9407
A total of 236 experimental data points of concrete under dynamic biaxial compressive loads and triaxial compressive loads, were used in the verification of the proposed failure surface for concrete under dynamic loading, as illustrated in Figure 5. Comparisons in Figure 5 indicate that the failure envelope in the octahedral stress space gradually expands with the strain rate.
Figure 5:
Summary of biaxial and triaxial compressive test data and the proposed envelopes.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
306 Materials Characterisation V
5 Conclusions The dynamic behavior of concrete in biaxial and triaxial stress state has been experimentally studied under linearly increasing loads of high strain rates. Based on the extensive test data and analyses, the following conclusions can be drawn: (1) The ultimate strength of plain concrete nonlinearly increases with its confining pressure at all load/strain rates that were considered in this study, but the magnitude of increment depends on the lateral stress ratio. (2) At low confining pressure, the ultimate strength of concrete increased with the strain rate. When the confining pressure was approximately higher than the uniaxial static strength, the ultimate strength tended to vanish with the strain rate. (3) Under biaxial and triaxial stress state, the failure envelope in the octahedral stress space gradually expands with the strain rate. (4) The proposed united strength criterion for concrete under multiaxial stress state reasonably reflects both the effect of strain rate and the effect of lateral confinement. Fairly good agreement with experimental results is achieved.
Acknowledgements This study was supported by the National Natural Science Foundation of China under Grants 90815026 and 51079019 at Dalian University of Technology, Dalian, China, and Grant 50908026 at Dalian University, Dalian, China.
References [1] Bischoff, P. H., & Perry, S. H., Compressive Behavior of Concrete at High Strain Rates. Materials and Structures, 24 (2), pp. 425-450, 1991. [2] Malvar, L. J., & Ross, C. A., Review of Strain Rate Effects for Concrete in Tension. ACI Materials Journal, 95(6), pp. 435-439, 1998. [3] Comité Euro-International du Béton, Model Code 90, CEB-FIP, Redwood Books: Trowbridge, Wiltshire, UK, pp. 48-51, 1990. [4] Newman, J. B., Concrete under Complex Stresses. Development in Concrete Technology-1, F. D. Lydon, ed., Applied Science Pub: London, UK, pp. 151-219, 1979.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
307
Modelling and simulation of the rutting resistance of bituminous mixes: experimental and stochastic approaches A. E. Ouni1,2, A. Dony2 & J. Colin2 1
Arts et Métiers ParisTech (ENSAM), France Ecole Spéciale des Travaux Publics, du Bâtiment et de l'Industrie (ESTP), Institut de recherche en constructibilité (IRC), France
2
Abstract This work deals with the simulation of the rutting resistance of bituminous binders used in road pavement. First, the experimental protocol was assessed to simulate pavement traffic allowing the prediction of the rutting depth evolution versus cycle number of wheel passes under isothermal conditions. Then a probabilistic parametric approach was developed to take into account the different parameter uncertainties related to the changes in experimental conditions. We investigate through the stochastic approach the rutting sensitivity of bituminous mixes under traffic load. A confidence region of a high probability of 99% is defined to allow the prediction of the in situ rutting potential of bituminous specimens. Keywords: rutting, bitumen, uncertainties, stochastic.
1 Introduction The environment protection has been and continues to be the major concern of the road politics in industrialized countries. With the signature of the Kyoto protocol agreement, the European Union is involved to reduce the energy consumption and the resulting greenhouse gases emissions [1]. The warm mix asphalt (WMA) is a promising technology which has been developed to contribute to the protection of the environment and sustainable development program [2–5]. The WMA is a mixture of mineral aggregates and bitumen produced and placed on the road at a lower temperature comparing to the traditional hot mix asphalt (HMA). Reductions in temperature of 20-50°C have WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line) doi:10.2495/MC110271
308 Materials Characterisation V been recorded. As a result, significant benefits were noted through the energy saving and the reduced emissions from burning fuels, undesirable fumes and odors generated at the plant and paving site. Typically, the HMA is produced at 140°C-180°C to make the binder viscous enough to coat the aggregate and to ensure a good workability of the asphalt during laying and compaction. In the case of WMA, there are some techniques developed and used in European countries to produce mix asphalt at low temperature without altering the workability of the mixture [6]. During its service life, the road pavement experiences multiple failure modes such as permanent deformation (rutting), fatigue and thermal cracking [7, 8]. The rutting phenomenon occurring at high service temperature is essentially due to the mechanical traffic load. The accumulated strains resulting from the vehicle passes lead to a plastic deformation in pavement layer. Due to the excessive pavement rutting, water and snow can stagnate in the ruts, for example, and lead to vehicle hydroplaning and accidents. Accordingly, the WMA is required to exhibit a good rutting performance once in pavement. In addition, the laboratory measurements of rutting exhibit very often some dispersion linked to the experimental uncertainties. It was shown by the authors in a previous work [9] that this probabilistic aspect has to be taken into account in order to have an accurate prediction of asphalt rutting performance through bitumen characteristics. In this context, the aim of this paper is to emphasize though a rutting test carried on warm and hot bituminous mixtures the necessity to consider the experimental uncertainties in order to quantify the random response related to the rut. Suitable probability distributions were developed on the light of the statistical information derived from experiments to describe the random stochastic variable attributed to the rut.
2 Experimental program to evaluate rutting performance 2.1 Material description This study is carried out by considering a Semi-coarse asphaltic concrete (BBSG 0/10) formulated as follow: Table 1:
Composition of asphalt mixture.
0/2 Noubleau
2/6 Noubleau
6/10 Noubleau
33%
20%
45%
Filler calcaire Piketty 2%
35/50 Bitumen 5%
The bituminous mixture specimens were prepared in laboratory (ESTP – France). First, hot mix asphalt (HMA) was prepared by mixing aggregates together with bitumen at a mixing temperature of 165°C. Then warm mix asphalt (WMA) was obtained by mixing the components at 110°C. During the preparation of WMA and in order to improve the workability of the asphalt
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
309
binder so that it can be sufficiently viscous to coat the aggregate, a surface active additive was incorporated at 0.3% into the bitumen. 2.2 Laboratory compaction To produce representative specimens for rutting test, slabs were manufactured with the prepared bituminous mixes and afterwards compacted in moulds by a roller compactor device (LCPC – France) according to the standard specifications (NF EN 12679-33). The apparatus (Fig. 1) provides a pneumatically powered means of compacting slabs to reproduce in situ compaction. The dimensions of the obtained slabs are 500 mm x 180 mm x 100 mm.
Figure 1:
Figure 2:
Slabs roller compactor.
French rutting test device.
2.3 Rutting test There are several types of wheel tracking devices that can be used to evaluate the rutting potential of a mixture. The three most known laboratory devices are the Asphalt Pavement Analyser (APA), the Hamburg Wheel Tracking Device (HWTD) and the French Public Works research Laboratory (LCPC) wheel WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
310 Materials Characterisation V tracker commonly known as the French Rutting Tester (FRT). For further information on these devices, the reader could consult [10]. Here, the FRT is used to evaluate the permanent deformation of the bituminous mixtures. This device has been used in France for over 15 years to simulate the traffic load through repetitive passes of a wheel pressurized to 0.6 ± 0.03 MPa. The FRT (Fig. 2) tracks across an asphalt specimen (slab) a loaded wheel of 500 N for many thousands of cycles. The rutting test was performed conforming to the standard specifications (NF EN 12679-22). Accordingly, the temperature was monitored uniformly and maintained to 60°C by temperature sensors inserted within the slab while local deformation is continuously recorded by depth gauge. Here, 2 slabs of HMA and WMA mixtures were tested simultaneously on the rutting device. After the required number of cycles, the rut depth was measured in mm at 15 different positions on the slab and then an average value of the rut is considered. Here, a load cycle corresponds to an outward and return motion of the wheel. In the following, ri denotes the local rut depth measured at the position (i) on the slab. The rut after n cycles is expressed in percent (%) according to eqn (1): Rut %
∑
100
(1)
where r0i represents the initial measurements at the positions (i) located on the slab and h its thickness. In figure 3, we present the evolution of the rut depth in percent versus the number of cycles for both hot and warm asphalt mixes. One can see that the 2 mixtures exhibit nearly the similar performance with a better rutting resistance of the HMA particularly with the increasing of the cycles number. In fact, we recorded after 30 000 cycles a rut depth of 4.23% for the HMA versus 6.53% in the WMA case. We note that in the French specifications [8], the maximum of rut depth shall be less than 10%. This constraint is marked by a dark continuous line in figure 3.
Rut (%)
10
1
HMA WMA
0.1 100
1000
10000
Cycles Figure 3:
Rutting test results.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
100000
Materials Characterisation V
311
3 Parametric probabilistic modelling of experimental uncertainties 3.1 Stochastic approach In this section, we will make use of the experimental uncertainties derived from the aforementioned rut depth values measured on the top surface of the slab. The 15 local measurements ri recorded after n cycles allow the definition of average and standard deviation values for the rut depth. In the following, we will denote by R (in bold letters) the random variable associated to the rut in percent. In order to describe the random uncertainties of this variable, the entropy maximum principle (EMP) is considered for the probabilistic parametric modeling [11, 12]. In general, given a random variable X, The EMP allows the construction of adequate and realistic probability density functions (pdf for short) of X on the light of the given information. Accordingly, the dispersion of X is described by the pdf pX and it is quantified by the entropy S:
S ( X ) p X ( x ) log( p X ( x )) dx
(2)
The given or available information may be the average (or mean) of the variable, the standard deviation, the second or higher order moments, etc. The entropy maximum principle states that the suitable probability density function which describes the random variable distribution is obtained by maximizing the entropy S. This leads to an optimization problem constrained by the known information formulated as follow: Maximize S(X), subject to: p X ( x)dx 1 (3) fi g i ( x) p X ( x)dx i=1 to n, where fi, represent the available information. For instance, if fi = mX (mean of the random variable X), g(x) = x whereas when fi = m2 (second order moment), g(x) = x². This optimization problem can be solved by minimizing a Lagrangian function H which includes (1+n) Lagrange multipliers (0, i, i =1 to n) associated to the (1+n) constraints. The function H is given by:
,
,…,
∑
Π
,
exp
∑
where [a, b] denotes the support of the random variable and Π function defined as:
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
(4) ,
an indicator
312 Materials Characterisation V Π
,
1 if x ∈ a, b 0 if not
x
(5)
Once i are found, the probability distribution of the random variable is known and it is given by: Π
,
exp
∑
(6)
3.2 Results To illustrate this methodology, we focus on the rutting results obtained at 30 000 cycles. The average and standard deviation values of the rut for Hot and warm bituminous mixtures are recapitulated in table 2. Table 2:
Statistical data of the rut in percent. Average_rut
Standard deviation_rut
HMA
4.23
1.62
WMA
6.53
0.74
In our case, given the average and the STD values and assuming that the random variable associated to the rut depth shall not take nor negative neither infinite value, the minimization of Eq. (x) leads to a Gaussian distribution: pRut ( r ) exp( 0 1r 2r 2 )
(7)
Figure 4 presents the probabilistic distributions as well as numerous random trials (1000 in total) of the rut according to the probability density function for both hot and warm asphalt mixes. The red continuous line corresponds to the average value of the rut in percent. Analogically, we can follow the same methodology to determine the rut pdf after 100, 300, 1000...cycles. Then, after a given number of cycles, we carry out some random realizations according to the pdf. This will be useful to define a confidence region including the probabilistic responses of the rut. In figure 5, we show the random variation of the rut value versus the number of cycles for HMA and WMA cases. As can be seen, the mean (or deterministic) model using the average values of the rut lies in the middle of the confidence region. In addition, one can note that even though the confidence regions of the 2 mixtures are very close at the beginning of the test which seems to be positive for warm asphalt rutting performance, the discrepancy increases for higher number of cycles where the intervals become more spaced.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
(a) Figure 4:
313
(b)
Probabilistic density functions of the rut and random realizations, (a) HMA, (b) WMA.
Figure 5:
Confidence regions for rut.
4 Conclusion A series of experimental tests were performed in this paper to compare the rutting performance of HMA and WMA obtained by the incorporation of surface WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
314 Materials Characterisation V active additive. The rutting test FRT shows that both asphalt mixes exhibit comparable rutting performance after 30 000 loaded cycles. This finding added to the environmental advantageous outlines the beneficial use of warm mix asphalt technology in road pavement. The stochastic approach was developed on the light of the available statistical information derived from the experimental uncertainties linked to the rut measurements. Accordingly, confidence regions were defined to estimate the probabilistic rut depth response with respect to the adequate probability density function. Discrepancies in confidence intervals are noticed between the 2 mixtures especially for a high number of wheel passes. In the future, it would be interesting to investigate the modification of rheological and chemical properties of the bituminous binders recovered from the mixtures. This study could improve our knowledge and understanding of the bitumen ageing phenomenon.
References [1] International Technology Scanning Program, Warm-Mix Asphalt: European Practice, 2008. [2] Brosseaud Y., Ecologiques, sécuritaires, confortables, les enrobés de demain se feront autrement : Présentation des enrobes tièdes, 2006. [3] Collectif. Un sujet chaud, les enrobés basses calories. Bitume info. N°12, pp. 12-15, 2006. [4] Brosseaud Y., Warm asphalt-Overview in France. LCPC, France, Présentation to WMA scan team, 2007. [5] Brosseaud Y. & Saint Jacques M, Warm Asphalt Mixes: Overview of This New Technology in France, Second European Road. Transport Research Arena Europe, Ljubljana, Slovénie, 21-24, 2008. [6] Harder, G. et al (2008). Energy and environmental gains of warm and halfwarm asphalt mix: quantitative approach. In: Transportation Research Board, Washington D.C., 2008. [7] “Warm mix asphalt technologies and research”, Government engineering, July-August, 2007. [8] Perraton D., Di Benedetto H., Sauzéat C. et al., Rutting of bituminous mixtures: wheel tracking tests campaign analysis. Materials and structures, Online First, 2010. [9] Eddhahak Ouni A., Dony A. & Colin J., Assessment of a probabilistic parametric rheological model to predict the rutting resistance of bitumen, ASMDO 2010. [10] Cooley Jr., Allen L., Prithvi S. Kandhal, & M. Shane Buchanan. Loaded Wheel testers in the United States: State of the Practice. Transportation Research Board E-Circular number E-C016, 2000. [11] Shannon, C.E., A Mathematical Theory of Communication, Bell System Technical Journal 27, pp. 379–423 & 623–656, 1948. [12] Jaynes, E.T., Information theory and statistical mechanics, Physical Review 106, pp. 620-630, 1957.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
315
Laboratory tests on the cleanliness of soil materials used as subgrades in pavement structures A. Athanasopoulou & G. Kollaros Democritus University of Thrace, Greece
Abstract In order to ascertain the presence of very fine material passing the No. 200 sieve, cleanliness tests are performed in the laboratory. The most important of these tests refer to the determination of the quantity of fines, of the sand equivalent value, of the plasticity index, as well as the methylene blue value of the material tested. Performing the sand equivalent test the percentage of very fine dust in claylike form can be established. Materials with very low sand equivalent are characterized as “impure” and the possibility of the existence of clay size grains in them is very high. In such cases it is necessary to perform a test in order to determine the plasticity of the soil material. The determination of the plasticity index of natural soils specifies their suitability as subgrade layers in pavement structures. Soils having a sand equivalent value lower than 10% will develop excessive swell with a simultaneous decrease in their bearing capacity when a pavement is built on them. The quantity of methylene blue absorbed by clay components of a soil mass is proportional to the specific area of the clay minerals. The methylene blue test supplements the sand equivalent and Atterberg limits tests, since with these the existence of clay-size grains is determined, but not the existence of active clay minerals as well. A number of soil samples have been examined in the laboratory, involving the procedures of cleanliness tests. The quantity of fine grained material was determined and their activity was accessed. The magnitudes were correlated in order to check the suitability of soils as a pavement foundation. Empirical relationships have been established which connect the attributes characterizing the cleanliness of soil materials. These relationships compare very well with most of the findings published worldwide. Keywords: swelling soils, methylene blue test, sand equivalent, cleanliness. WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line) doi:10.2495/MC110281
316 Materials Characterisation V
1 Introduction In most pavement structure techniques, suitability is one of the essential characteristics of soil materials. The term "suitability" characterizes the absence of undesirable elements. These elements are soluble salts, iron hydroxides, organic matter, as well as clays. They can present sickliness because of their fineness, their physicochemical activity or because of their effect in the binding materials. In view of their frequency of appearance in alluvial deposits and rocks, their hydrophilic nature, and of their affinity and plasticity properties, clays are the most harmful. Swelling soils often cause serious damages in pavement structures, while, on the other hand, they cannot be used as embankment material in roads. The swelling potential of soils is mainly affected by their clayey fraction (Kollaros and Athanasopoulou [1]). In order to ascertain the existence of very thin material passing the No. 200 sieve cleanliness tests are performed. The more important tests are consisted in the determination of the quantity of fines, of the plasticity index, the equivalent sand and the methylene blue value. The quantity of material passing the 0,075 mm sieve (#No. 200), expressed as a percentage of the total material, is determined by the sieve analysis test, either through the dry or, more usually, through the wet process, according to AASHTO T88 or ASTM D422 standards. The plasticity index, which is defined as the difference in contained moisture between the liquid and plasticity limits: PI=LL-PL, it is a value that helps to recognize soil characteristics and to properly classify them. The plasticity index determines the appropriateness of soils to be used as subgrades in roadway foundations. Soils having a high PI value, such as clayey, silty and some sandsilt soils, are inadequate for the foundation of pavements. In Table 1 the degree of plasticity is given in terms of PI and of soil characteristics in a dry condition, for the corresponding scale of the PI values. The plasticity index is useful for materials containing high percentages of silt and clay, while in the region not covered by the PI use is made of the sand Table 1: PI >35 16-35 7-15 4-6 0-3
Soil Highly Plastic Plastic Medium plasticity Slightly plastic Non Plastic
Soil characterization as a function of PI. Soil characteristics in dry condition High cohesion value, manual smashing of lumps is impossible Medium to high cohesion, difficulty in smashing lumps manually Low to medium cohesion, lumps are smashed by exerting a low pressure to them Low cohesion, easy manual smashing of lumps Very little cohesion or complete absence of cohesion, lumps are decomposed by a simple touch
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
317
equivalent value. The sand equivalent, SE, test arose in view of the need for a quick quality indication of paving materials. This test was devised by Hveem in 1953, it was put in use in 1956 and it has been adopted for laboratorial use since 1966. More specifically, with the sand equivalent test the proportion of very fine dust of claylike form is determined. For a SE<20 value, the plasticity index should be determined. In the region of higher SE values, the plasticity index is not determined (Non Plastic) and therefore it cannot replace the sand equivalent. On the basis of comparisons of sand equivalent values to the results of other tests it has been found that most of the soils presenting high swell potential when saturated with water, is possible to be recognized by the SE value. Soils on top of which a pavement is to be constructed develop excess swell, with a simultaneous decrease of their bearing capacity in saturated conditions, provided that the SE value is lower than 10%. The different fine grained materials could be classified in three groups, as shown in Table 2, depending on their sand equivalent value. Table 2:
Classification of fine-grained materials as a function of SE.
Sand Equivalent High Values Medium Values Low Values
Fine-grained materials Clean non-cohesive sands Sands blended with some quantities of silt and clay Clays and silt-clayey mixtures
In order to measure the unsoundness of sands, the Laboratoire Centrale des Ponts et Chaussées, in France, examined two different methods. According to the first procedure, the process of the sand equivalent test was modified, to become more compatible for the recognition of sands rich in fine-grained material. Thus, in the AFNOR NF P 18-597 standard, "Propretés des Sables", the following modifications were made: Test sampling using the 2 mm sieve, instead of the 4.75 mm sieve (#No. 4). The percentage of elements finer than 0.08 mm is kept to a constant value equal to 10%. With the second procedure, the test is supplemented with a different measure, namely the capacity of the clayey components to absorb some basic dye such as the methylene blue. The methylene blue test has basically been developed in an effort to determine the clay content of a soil. In 1980 the test was adopted for aggregates and in 1984 the method of methylene blue with agitation was proposed by Tran and Millon-Devigne [2] in order to increase the precision of the test and to make it capable of measuring properties on least clayey samples. The test is based on a successive import of increasing quantities of methylene blue solution in a suspension of the material to be examined, until the claylike particles are saturated with dye. Saturation has been achieved when a spot of soil coloured by the retained dye of soil is shaped, surrounded from a colourless area of humidity ("spot" control) on a filter paper. The control is positive, if in this zone an excess of dye appears in the form of light blue coloration, which surrounds WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
318 Materials Characterisation V radially the central dark blue trace of the drop. The spot test is repeated after two minutes. If the solution continues to demonstrate an excess of dye, the "final point" has been reached. It must be noted that an experienced person can reach faster the final point, omitting some initial doses of methylene solution. The methylene blue test supplements the sand equivalent and Atterberg limits tests, since they determine the existence of particles with claylike dimensions, but not the presence of active clay minerals.
2 Laboratory testing of soil samples Soil samples were collected from twenty six sites located in the Evros prefecture. The criterion used was their proximity to road works and the probability that they could create constructional problems due to large percentages of clay. The properties of these materials ranged in the limits presented in Table 3. Table 3:
Properties of the soils tested.
Soil Property Clay content (material < 2μ) (%) Percentage passing the No. 200 sieve (%) Liquid Limit (%) Plasticity Limit (%) Plasticity Index (%) Linear shrinkage (%) Volume of dye, Vd (cm3) Sand equivalent, SE Skempton Activity [PI/(%material<2μ)]
Mean Value 41.42 74.65 52.73 22.24 29.73 12.93 20.35 14.46 0.71
Range of values 9–54 29.23–90.13 29–76 17–29 17–47 10.0–15.7 6–27.5 5–33 0.58–1.00
The swelling potential of soils is influenced by the percentage of the clay fraction they contain as well as by the composition of this fraction. The grain size analysis of soil samples revealed that most of them had high content of clay (material with grain diameter smaller than 2 microns). When the grain size analysis was performed using sieves with standard openings, the presence of high percentages of very fine material passing the No. 200 sieve has been recorded. Only two samples had particles with a diameter smaller than 0,075 mm in a percentage smaller than 50%. The Atterberg limits tests resulted in high values for the plasticity index. The PI values were above 15 for all soil samples, reaching even a value of 47, suggesting that the soils were very plastic and in a dry condition it was very difficult to smash any formed lump. The determination of PI was not possible for one of the samples. Soils having such PI values are inadequate for the foundation of pavements. The swelling degree could be calculated with the Holtz-Gibbs classification method, which is based on the plasticity index and the colloid content of the samples. Twenty two samples are characterized by a very high or high swelling degree, while only 4 showed medium and low swelling degree. Generally, the WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
319
clays in the region fall in group A-7 according to the AASHTO classification method. This group involves soil materials poor or unsuitable to be used as foundation.
3 Correlation of attributes used for soil characterization With the completion of laboratory testing, a statistical correlation of attributes used for the characterization of soil materials was attempted. Special attention has been focused mainly on those values resulting from the test which specifies the clayey ingredients using the methylene blue and the sand equivalent procedures. The selection of these properties can be easily justified taking into account the high percentages of fine grained material in the samples and the relatively high swelling degrees that are expected in the field, constituting a source of dangers for the integrity and longevity of pavement structures. The sand equivalent values resulted as the mean of three repetitions of the standard process for each sample. In the methylene blue test the volume of dye, Vd was measured. 50
Plasticity Index, PI (%)
40
30
20
10 20
30
40
50
60
Material finer than 2 (%)
Figure 1:
Plasticity index variation with material finer than 2μ.
The distribution of plasticity index values, as estimated by the percentage of material finer than 2μ as well as from the results of the methylene blue test is given in figures 1 and 2, respectively. The linear correlation of these attributes with the plasticity index is particularly good, since the coefficients of determination in these WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
320 Materials Characterisation V cases are R2=0,73 and R2=0,58. Equations (1) and (2) interrelate the results obtained for the plasticity index and the other two properties. PI=0,606 (%material<2μ)+5.02
(1)
PI=1,201Vd+5.79
(2)
In particular, a high degree of correlation has been obtained for the methylene blue value and the percentage of claylike fraction in the samples tested. The linear regression eqn. (3) connecting these attributes gave a coefficient of determination, R2, equal to 0,82. %material< 2μ=2.019Vd+0.34
(3)
In figure 3 the variation of the percentage of material with particle diameter smaller than 2μ against the Volume of dye calculated by the methylene blue method is presented. A similarly good linear correlation appeared between the volume of dye of the samples and the percentage of material passing the sieve No. 200, as shown in figure 4. The coefficient of determination, R2, for the relation connecting the two attributes in eqn. (4) has been found equal to 0,69. %material< No. 200=2.409 Vd+25.63
(4)
50
Plasticity Index, PI (%)
40
30
20
10 10
12
14
16
18
20
22
24
26
28
3
Vd (cm )
Figure 2:
Plasticity index variation with volume of dye, Vd.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
30
Materials Characterisation V
321
Material finer than 2 (%)
60
40
20
0 5
10
15
20
25
30
3
Vd (cm )
Figure 3:
Variation of the percentage of material <2μ as a function of the Vd.
Material passing # No. 200 (%)
100
80
60
40
20 5
10
15
20
25
30
3
Vd (cm )
Figure 4:
Variation of material passing # No. 200 as a function of the Volume of dye, Vd.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
322 Materials Characterisation V
Material passing # No. 200 (%)
100
80
60
40
20 0
5
10
15
20
25
30
35
Sand Equivalent, SE
Figure 5:
Variation of the percent passing # No. 200 as a function of the sand equivalent value.
In eqn. (5) the sand equivalent value, SE, is the independent variable and the percentage of material passing the sieve No. 200 is the dependent variable. This equation yielded a coefficient of determination, R2= 0,66. %material< No. 200=1.994 SE+103.48
(5)
The sand equivalent values have been related to the respective values obtained by the testing of the samples using the methylene blue test. The equation of the curve that fits to the data in figure 6 has a coefficient of determination, R2= 0,43 and is depicted in eqn. (6). Vd=-0,587 SE+28.98
(6)
The negative gradient of the regression lines in figures 5 and 6 means that an increase of the sand equivalent entails a reduction in the respective values of both the fine-grained material and the quantity of dye absorbed by the clay particles. Empirical relations similar with those resulted from the statistical processing of the laboratorial results, have been reported in the international bibliography. Many research efforts have been devoted to the correlation of the Atterberg limits with the attributes tested and with the cation exchange capacity, CEC. High level of correlation (r=0,80) has been obtained between the plasticity index and the CEC (Taylor [3]). Also, a coefficient R2=0,56 was found for the relation of the Skempton activity (PI/% of material<2μ) with the cation exchange WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
323
30
Vd (cm3)
26
22
18
14
10 0
5
10
15
20
25
30
Sand Equivalent, SE
Figure 6:
Variation of Vd as a function of the sand equivalent value.
capacity (Sweere and Galjaard [4]). The plasticity index of 70 Illinois clay samples (Odell et al. [5]) was fairly correlated with the content of material<2μ through the eqn. (7): PI=0,568 (%material<2μ)+1.09
(7)
A very good correlation between the apparent surface area calculated by the methylene blue test and the percentage of particles having a diameter smaller than 1μ has been found by Xeidakis [6]. In this research 8 clayey minerals had been tested and a correlation coefficient r=0,946 was found. A relatively weaker correlation (r=0,65) has been reported for 19 British soils (Farrar and Coleman [7]) which contained some montmorillonite. Nikolaides et al. [8] published the results of SE and MB values obtained at 16 samples from Greek quarries. They found that there is no correlation between MB and SE values. The results of a series of tests performed on aggregates from production sites and 45 mixtures prepared in the laboratory to clarify the effect of potentially harmful fines on the MB and SE values are described by Petkovšek et al. [9]. Particular attention is paid to the prediction of approximate limits for SE and MB values, taking into account the specific geological conditions of the aggregate sources and past experience using the same aggregates.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
324 Materials Characterisation V
4 Conclusions Swelling clay soils are abandoned in the region of Thrace, Northern Greece, presenting many risks for the construction of pavements. In a new classification system, it is possible for soils with homogeneous technical attributes to be distinguished in categories, which they will be based on the activity of the clayey phase -more specifically on the total specific area, that is, on the methylene blue value- in combination with their grain-size gradation. The methylene blue test has been found to be an easy and very repeatable test method. Because of their simplicity, the suitability controls which are supported by the laboratorial tests, can be also conveyed in field testing. The correlation of values of attributes obtained from the tests in methylene blue with the respective values of the plasticity index, the soil gradation (clay fraction, percent of material passing the No. 200 sieve) and with the sand equivalent showed that linearity exists in these relations and that the correlation coefficients were particularly high. For the relations to merit general application, they should be confirmed with the repetition of experimental processes in many more samples from the region of Thrace.
References [1] Kollaros, G. & Athanasopoulou, A., The character and identification of swelling soils in road construction projects, Proc. International Symposium on Engineering Geology and the Environment IAEG, eds. Marinos, Koukis, Tsiambaos & Stournaras Balkema: Athens, pp. 187-192, 1997. [2] Tran, N.L. & Millon-Devigne, P., L'essai au bleu de méthylène en turbidimétrique, Bulletin de l’Association Internationale de Géologie de l’Ingénieur, 29(1), pp. 453-456, 1984. [3] Taylor, R.K., Cation exchange in clays and mudrocks by methylene blue. J. Chem. Tech. Viotechnol., 35A, pp. 195-207, 1985. [4] Sweere, G.T.H. & Galjaard, P.J., The methylene blue test as a rapid means of estimating the cation exchange capacity of soils. International Symposium of Geotechnical Aspects of Mass and Material Transportation, Bangkok, pp. 47-52, 1984. [5] Odell, R.T., Thornburn, T.H. & McKenzy, L., Relationships of Atterberg limits to some other properties of Illinois Soils. Proceedings of the Soil Science Society of America, 24(5), pp. 297-300, 1960. [6] Xeidakis, G.S., Assessment of the engineering and other properties of expansive soils by various methods. Ph.D. Thesis, Dept. of Civil Engineering, University of Leeds, England, 407 p., 1979. [7] Farrar, D.M. & Coleman, J.D., The correlation of surface area with other properties of 19 British clay soils. Journal of Soil Science, 18(1), pp. 118124, 1967.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
325
[8] Nikolaides, A., Manthos, E. & Sarafidou, M., Sand equivalent and methylene blue value of aggregates for highway engineering. Foundations of civil and environmental engineering 10, Publishing House of Poznan University of Technology, Poznan, pp 111-121, 2007. [9] Petkovšek, A., Maček, M., Pavšič, P. & Bohar, F., Fines characterization through the methylene blue and sand equivalent test: comparison with other experimental techniques and application of criteria to the aggregate quality assessment, Bull Eng Geol Environ 69, pp. 561–574, 2010.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
This page intentionally left blank
Materials Characterisation V
327
Use of additives to improve the engineering properties of swelling soils in Thrace, Northern Greece A. Athanasopoulou & G. Kollaros Democritus University of Thrace, Greece
Abstract Highway construction engineers often face the need for more stable, durable and, at the same time, more economic road structures. This is nowadays true because of increased traffic volumes and heavier loads on the roadways. As a consequence, enhanced pavement structures and improved subgrades is a necessity. The international highway “New Egnatia” crosses areas in Thrace, Northern Greece with abundant clayey soils having poor technical properties. The treatment of physical soils with some substances could bring up new materials, which would operate better under the traffic and environmental conditions. This has led to the decision to investigate the possibilities of improving the existing soil materials using chemical additives. Soil samples were collected from the abovementioned area and mixed with lime and fly ash, in various proportions. The modification of the soil properties with special emphasis on their strength has been examined in the laboratory after different curing periods. The experimental results have shown that the unconfined compressive strength increased as a function of both the percentage of additive in the mixture and the time of curing. The improvement depended upon the soil mineralogy and the kind and quantity of exchangeable cations. This holds true when the influence of the kind and quantity of the additive is taken into account. The effects of lime on the swelling clayey soils tested were more beneficial than those of fly ash. The soils after their treatment could be used as a subgrade or even as a subbase layer in roadway pavements. Keywords: stabilization, lime, fly ash, engineering properties, unconfined compressive strength, optimum moisture. WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line) doi:10.2495/MC110291
328 Materials Characterisation V
1 Introduction The continuous growth of the traffic volume and of vehicle's size, has made obvious the need for more stable, durable and, at the same time, more economic road structures. Consequently, a requirement exists for improved properties of the pavement structure as well as of the physical subgrade. The treatment of physical soils and base materials with some substances could bring up new materials, which would operate better under the traffic and environmental conditions. In this procedure, called soil stabilization, such substances as Portland cement, bituminous materials, lime, fly ash or alkali salts could be used. It has been recognized by numerous investigators (Bell [1], Zhang and Cao [2]) that the addition of lime and/or fly ash to soil materials may cause various beneficial changes to the engineering properties of fine-grained soils, such as the reduction of plasticity, the reduction of shrinkage-swelling potential and the improvement of strength characteristics. Roadways have a high potential for large volume use of the fly ash stabilized soils. Arora and Aydilek [3] investigated the use of Class F fly ash amended soilcement or soil-lime as base layers in highways. Unconfined compression, CBR, and resilient modulus tests were conducted. Required base thicknesses were calculated using the strength parameters. The strength of a mixture is highly dependent on the curing period, compactive energy, cement content, and water content at compaction. Lime treatment didn’t provide sufficient strength for designing the mixtures as highway bases. The strength of soil-lime-fly ash mixtures could be estimated by various methods such as unconfined compression, CBR, the Hveem stabilometer and triaxial tests. The most commonly used method is the unconfined compression test, not necessarily being the most appropriate for all purposes. The strength of soil-lime or soil-fly ash mixtures depends on many variables such as the soil type, the lime and fly ash content, the additive type, the time and method of curing (temperature and humidity), the water content, the unit weight and the time interval between mixing and compaction (Çokça [4]). The addition of lime, fly ash, and lime/fly ash to three clayey soils led to a reduction of the plasticity index and contributed to an increase in the optimum moisture content and a decrease in the maximum dry density (Hesham [5]). The optimum lime content ranged from 3 to 5%, while the optimum fly ash content between 16 and 35%. The optimum lime/fly ash content for the three soils was (2.5%L+8%FA), (2%L+12%FA) and (3%L+20%FA). The UCS, Esecant, CBR, and the Velocity of ultrasonic p-waves, Vp, values increased slightly with an increment of the dry density of the untreated compacted soils (due to the compaction process) and strongly due to the addition of the chemical stabilizing agents (lime, fly ash, and lime/fly ash) whereas the formed cementitious compounds (as a result of the chemical reactions between the silica and the alumina and the additives) joined the soil particles. Basic studies on pozzolanes and on the pozzolanic properties of fly ash have been carried out by Adu-Gyamfi [6] and Brooks [7]. Considering the nature of the lime used in the mixtures, many investigators such as Ingles and Metcalf [8], WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
329
have shown that calcarious limes gave higher strength results than those given by dolomitic limes. In Greece, many roads have been built using stabilization techniques for the subbase materials. The behaviour of these roads has been judged as fairly satisfactory. The scope in some of these trials was the use of large quantities of fly ash per surface unit, in order to take advantage of the surplus of this material. However, the cost of transporting the material from places of its production or deposition to the areas of the projects under consideration, which are located to a relatively small distance from the power stations, is a very restrictive factor. The Research Centre of Public Works (KEDE) has carried out significant studies on the stabilization of aggregates and clayey soils, mainly with Portland cement and with lime and fly ash since 1982. The exploitation of Megalopolis fly ash and its applications in highway construction in adjacent areas has been examined by Marsellos et al. [9]. Further studies have been undertaken by Greek Universities, the Technical Chamber of Greece, and other researchers. The main objective of this work is to test the capability of lime and fly ash in improving the engineering properties of clayey soils from the areas of Thrace, in order to use them as stabilized layers in road construction and to apply more economic processes in constructing new pavements or improving existing ones.
2 Materials and methods The soils used in this study have been sampled near the villages Aetolofos and Aetokorifi of the Rhodope prefecture in Thrace. The soils are pleio-pleistocene fluvio-lacustrine deposits resulted by the alteration of andesitic tuffs and tuffites of the Zonaia Mountains surrounding the basin. These clayey soils are of black colour near the surface, but they turn to grey or yellowish in the deeper horizons. In some places they are intercalated by lenses or layers of sand and gravels. Disturbed samples were taken from an excavation about 1 m deep in two different locations, near the New Egnatia highway. The soil "S1" is a fine-grained black clay, while the soil "S2" is a brown clay. Cation concentrations of Mg and Ca 5.76 meq/l and 19.21 meq/l, respectively, have been determined for S1. The respective values for S2 were 8.84 meq/l and 92.32 meq/l. The chemical properties of the soils are summarized in Table 1. Table 1:
Loss on Ignition SiO2 Al2O3 Fe2O3 CaO MgO K2O Na2O
Chemical composition of the soils tested. Black Soil (S1) 11.62 (%) 64.18 (%) 12.73 (%) 5.43 (%) 1.55 (%) 1.50 (%) 1.80 (%) 0.70 (%)
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Brown Soil (S2) 14.95 (%) 57.25 (%) 11.97 (%) 5.43 (%) 7.28 (%) 1.16 (%) 1.55 (%) 0.21 (%)
330 Materials Characterisation V The soils were air-dried and pulverized, in order to pass the Νo. 4 (4.75 mm) sieve. The grain size distribution of the representative soil samples is presented in Table 2. Table 2:
Properties of natural soils.
Properties Specific gravity (kg/cm2) Liquid Limit (%) Plastic Limit (%) Plasticity Index (%) Linear Shrinkage (%) Free Swell Index (%) Maximum Dry Density (kg/m3) Optimum Moisture Content (%) Grain Size Distribution Sand and Gravels (%) Silt (%) Clay (%) Classification AASHO USCS Unconfined Compressive Strength (kg/cm2)
Soils Black Soil (S1) Brown Soil (S2) 2.5 2.7 76 51 29 23 47 28 13.3 10 95 51 1588 1707 21.7 17.8 24.3 22.7 53.0
16.2 35.8 48.0
A-7-6 CH 1.5
A-7-6 CL 2.7
The basic properties of the soils were determined from representative samples and for the soil fraction passing the Νo 40 (425 μ) sieve. The liquid limit of these soils was found using the Casagrande method. The grain size distribution of the soils has been determined by both the dry method (AASHTO T-27) and hydrometer analysis. The physical properties of the soils studied and their unconfined compressive strength (UCS) are presented in Table 2. Both soils are classified as Group A-7-6 according to the AASHO classification system, while, according to the Unified Classification System, are classified as CH and CL respectively. Soil samples passing the No 4 (4.75 mm) sieve were used in order to find the dry density-moisture content relation with the standard Proctor compaction test. The soils were thoroughly mixed with different moisture contents (14% to 30%) and were cured in a moisture room for 24 hours before they were compacted, for uniformity purposes. The maximum dry density and the optimum moisture content are shown in Table 2. The lime used in this study was a typical commercial hydrated calcitic lime, having a high CaO content (65.25%). It was supplied by the AIMOS Lime Company, Drama, Greece which has a 200 ton daily production. The chemical composition of this lime is shown in Table 3. The term fly ash represents the fine-grained ash residue produced from pulverized coal combustion and carried away by the hot gases comes out of the WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
331
chimney. This residue is usually collected with appropriate filter put along the chimney. The fly ash used for the preparation of the laboratory specimens was supplied by the Ptolemaida Power Station (6,000,000 ton/year). The chemical composition of the fly ash is shown in Table 3. Table 3: Properties Loss on Ignition SiO2 Al2O3 Fe2O3 CaO MgO K2O Na2O
Chemical composition of lime and fly ash. Lime 33.25 (%) 0.01 (%) 0.01 (%) 0.11 (%) 65.25 (%) 0.50 (%) 0.01 (%) 0.01 (%)
Fly ash 13.90 (%) 29.95 (%) 10.85 (%) 4.57 (%) 20.00 (%) 1.90 (%) 0.95 (%) 0.32 (%)
3 Treatment of soil samples with lime and fly ash The air-dried soil materials passing the No. 4 (4.75 mm) sieve, were mixed in different proportions by weight with lime (in powder form) and fly ash. Water was added until the optimum moisture content was reached and the mixing process continued till a visually uniform product was achieved. Cylindrical specimens 50 mm in diameter and 100 mm high were then formed in special moulds. The material was placed in the mould in three layers of equal thickness. The quantity of the material for each sample was determined by the optimum moisture-maximum dry density relationship. The compaction to the maximum dry density Proctor was achieved by compressing the required mass in the given volume with an automatic hydraulic press. After their extraction from the mould, the specimens were weighted and sealed in polyethylene bags, in order to keep the moisture content constant during the curing period. For each percentage of additive a set of three specimens was prepared. Care was taken to cure the specimens under stable temperature and moisture conditions. The specimens for the unconfined pressure test were cured for 7, 28 and 90 days before their testing. The additive–soil weight ratios used were: a) for the lime: 4–100, 7–100, 10–100 for both soils. b) for the fly ash: 4–100, 8–100, 12–100. Specimens were also prepared with lime–flyash–soil ratios: 1–3–100, 2–6– 100 and 1–5–100 by weight. For each discrete ratio, the optimum moisture content was determined using the standard Proctor method according to the AASHTO T99 61 specification. The specimens were tested in an unconfined compression machine with strain rate 1.25 mm/min. An X-Ray Diffraction analysis showed that soil S1 had more swelling clay minerals, while soil S2 had more kaolinite and calcite. WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
332 Materials Characterisation V
4 Results and discussion The Atterberg limits, the optimum moisture content and the maximum dry density of the soil-lime, soil-fly ash and soil-lime-fly ash mixtures are presented in Tables 4 and 5 for the black soil S1 and the brown soil S2 respectively. In the same tables the unconfined compressive strength after 7, 28 and 90 days of curing is shown. The addition of lime resulted in a reduction of the liquid limit in comparison with the natural soil. This fact complies to the results obtained by other investigators (Sridharan et al. [10], Akoto [11], Athanasopoulou [12]) who have observed a significant reduction in the LL of fine-grained soils following their treatment with additives. The admixture of lime rapidly initiates flocculation and cation exchange reactions, leading to a reduction of the specific area of the soil. The reduction of the thickness of the diffused double layer causes the reduction of the liquid limit. This reduction was smaller (12% with fly ash and 18% with lime) for the brown clay, in comparison to the black soil (22% with fly ash and 27% with lime) due to higher concentration of calcium and magnesium exchangeable cations and the lower percentage of swelling clay. The admixture of lime and fly ash resulted in a reduction of the maximum dry density (MDD) of the soils. On the other hand, an increase in optimum moisture content (OMC) was observed for the same compaction effort (Tables 4 and 5). The reduction in maximum dry density, following the treatment with lime and/or fly ash, reveals the increased resistance to the compaction effort offered by the flocculated soil-structure. Table 4:
Alteration of properties of black clay treated with lime and fly ash.
Materials Soil g 100 100 100 100 100 100 100 100 100 100
F.A Lime G g 0 0 4 0 8 0 12 0 0 4 0 7 0 10 3 1 6 2 5 1
LL % 76 69 64 59 68 60 55 66 65 59
Atterberg Limits PL % 29 32 35 39 39 41 43 34 48 42
PI % 47 37 29 20 29 19 12 32 17 17
Compaction Characteristics M.D.D. O.M.C. % Kg/m3 1588 21.7 1526 25.6 1487 26.4 1422 31.2 1453 28.8 1449 29.4 1445 29.9 1476 28.2 1453 29.4 1468 28.8
7 Days 1.5 4.3 6.4 7.6 6.0 6.5 7.4 3.3 7.9 3.8
UCS (kg/cm2) 28 90 Days Days 1.5 1.5 4.5 5.6 7.2 7.9 8.8 9.5 9.6 13.6 11.7 22.9 13.5 16.4 4.1 5.0 9.3 11.7 4.3 4.3
The OMC increased as a consequence of the excess of water retained in the voids of the flocculated soil-structure (formation of soil aggregates), which results from the soil-additive interaction. For all the percentages of additive, the mixtures with fly ash showed greater MDD values than those with lime, due to the smaller apparent unit weight of lime. WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
Table 5:
333
Alteration of properties of brown clay treated with lime and fly ash.
Materials Soil F.A. Lime g G g 100 0 0 100 4 0 100 8 0 100 12 0 100 0 4 100 0 7 100 0 10 100 3 1 100 6 2 100 5 1
LL % 51 50 47 45 49 47 42 46 43 40
Atterberg Limits PL % 23 33 37 40 NP NP NP 43 42 39
PI % 28 17 10 5 ---3 1 1
Compaction UCS Characteristics (kg/cm2) M.D.D. O.M.C. 7 28 90 kg/m3 % Days Days Days 1707 17.8 2.7 2.7 2.7 1611 22.1 3.8 4.3 4.5 1577 24.5 6.6 7.9 9.2 1477 28.1 8.8 9.5 12.4 1575 24.6 3.5 5.7 8.5 1559 25.1 3.9 5.8 9.0 1525 26.5 3.9 6.0 8.4 1604 22.6 5.3 6.1 7.1 1558 24.9 8.4 11.7 12.6 1573 23.9 6.3 7.5 8.9
The change of OMC and MDD was gradual when fly ash was used, whereas with the admixture of lime a rapid change existed with small percentages of additive and remained almost constant thereafter. This could be attributed to the reaction rate between the clayey soil and lime, as well as to its quick flocculation due to quick exchange of soil cations with Ca++ from the lime and depression of the double layer. On the other hand, the end change of these properties is greater with fly ash than with lime due to the reaction of the soil with the constituents of fly ash other than CaO, like SiO2, Al2O3 and MgO. Considering the strength change of the soils, the UCS increased both with the percentage of the additive and with the time of curing as it is demonstrated in figures 1 to 4. In the case of lime addition, a dramatic increase occurred in the strength of the soil (more than 10 times) with addition of only 4% lime. This high rate of increase in soil strength was reduced with the increase of lime content and at some 8-10% of lime the UCS remained more or less constant or started to decrease (point of soil satisfaction). This percentage is recognized as the lime modification optimum (LMO) of the soil. The same trends hold true for the fly ash, though the rate of increase was lower and there was no point of soil satisfaction. The strength increased in an almost constant rate when the percentage of fly ash in the mixture was increased. As it is shown in figures 1 to 4 the rate of strength gain and the ultimate strength were different for both the additive and the soil. Lime proved to be much more effective than the fly ash in the case of the soil S1; the opposite has been observed for the soil S2. The difference in the soil behaviour is certainly due to the differences in the mineralogy of the soil and the kind of the exchangeable cations present. The brown soil S2 is almost saturated by Ca++ cations, therefore the addition of lime has little effect on the exchangeable cations of the soil. It is well-known that lime has a more pronounced effect on swelling clay minerals (illite, caolinite) due to the greater depression of the double layer. For the same reasons, fly ash yielded a little better results in soil S1 than in the brown soil S2. WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
334 Materials Characterisation V
Unconfined Compressive Strength (kg/cm2)
24
20 90 D-L
16
28 D-L
12
8 7 D-L
90 D-FA 28 D-FA 7 D-FA
4 Additive: Fly Ash Additive: Lime
0 0
Figure 1:
2
4
6 8 10 Percent Additive (%)
12
14
16
Variation of UCS with the percentage of additive (Black clay, S1).
Unconfined Compessive Strength (kg/cm2)
14 90 D-FA
12 10
28 D-FA 90 D-L
7 D-FA
8 6
28 D-L
4
7 D-L
2 Additive: Fly Ash Additive: Lime
0 0
Figure 2:
2
4
6 8 10 Percent Additive (%)
12
14
16
Variation of UCS with the percentage of additive (Brown clay, S2).
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
335
25 Soil Sample S1 Additive: Fly Ash Additive: Lime Additive: Lime+Fly Ash
7%
Unconfined Compressive Strength (kg/cm2)
20
10%
15 4%
2:6
10 12% 8%
4%
5
1:5 1:3
0%
0 7
28
49
70
91
Curing Period (days)
Figure 3:
Variation of strength values for mixtures after different curing periods (Black clay, S1).
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
336 Materials Characterisation V 15 Soil Sample S2 Additive: Fly Ash Additive: Lime Additive: Lime+Fly Ash
2:6
Unconfined Compressive Strength (kg/cm2)
12%
10 8%
7%
10% 4%
1:5 1:3
5 4%
0%
0 7
28
49
70
91
Curing Period (days)
Figure 4:
Variation of strength values for mixtures after different curing periods (Brown clay, S2).
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
337
Considering the effect of the curing period, the results followed the same general pattern as with the percentage of additive. That is, for soil S1, the effect of time was more significant in the case of lime than in the case of fly ash, whereas in soil S2 the UCS value raised much more with the curing time when fly ash was added than in the case where lime was used. This difference could also be attributed to the abundance of swelling minerals in soil S1. The more the swelling minerals, the more lime is precipitated in the clay surface and the more cementitious materials (CaCO3) are formed. The increase of strength in swelling minerals is due rather to cation exchange (flocculation of the clay) and to the cementitious reaction than to the pozzolanic one (Xeidakis [13], Baykal et al. [14]). Soil mixtures having lime–fly ash ratios of 1–3 and 2–6 have given a little higher strength values than those of the mixture with each additive alone. So, the 2–6 lime–fly ash ratio resulted to a strength two times and 1.5 times greater than that with the addition of fly ash alone and lime alone, respectively (figures 3 and 4). The soil-lime-fly ash mixtures exhibited final strength values intermediate to those found for the soil-lime and soil-fly ash mixtures.
5 Conclusions The admixture of lime and fly ash to two expansive clays have led to a significant decrease of the liquid limit probably due to the depression of the diffuse double layer thickness associated with the clay particles, the aggregation of the clay and the coating by Ca(OH)2. A progressive reduction in maximum dry density and increase in optimum moisture content has been observed with the addition of these materials. The decrease of maximum dry density of clay soils, after their treatment with lime and fly ash, is an indication of the increase of the strength of the soil and the increase of its bearing capacity. The strength of the mixtures tested was much higher than that of the natural soils, in all cases. In general, the strength of soil-fly ash and soil-lime mixtures increased with an increase in the additive content, for all curing periods. For both soils and additives, an increase in curing period resulted to an increase in strength. The increase of the UCS for the soil S1 was greater with the addition of lime (up to 20 times greater than the original), than with the addition of fly ash. The best results obtained when 7% lime was added to the soils and a 90 days curing period followed the compaction of the specimens. This is attributed to less Ca++ and a greater percentage of clay minerals in this soil. The strength increase in soil S2 was greater with fly ash than with lime. This may be due to a higher content in Ca++ and caolinite, as well as to a lower content in swelling minerals. The results showed that the mineralogy of the soil plays a decisive role in the stabilization process and greatly affects the ultimate strength of the mixture. The ultimate strength of the soil after its improvement is adequate for the soil to be used as subgrade or embankment material in main roads, or even as subbase in some secondary roads. WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
338 Materials Characterisation V
References [1] Bell, F.G., An Examination of the Use of Lime and Pulverized Fly Ash to Stabilize Clay Materials. Bulletin of the Association of Engineering Geologists, XXX(4), pp. 469-479, 1993. [2] Zhang, J. & Cao, X., Stabilization of Expansive Soil by Lime and Fly Ash. Journal of Wuhan University of Technology - Mater. Sci. Ed. 17(4), pp. 73-77, 2002. [3] Arora, S. & Aydilek A.H., Class F Fly-Ash-Amended Soils as Highway Base Materials. J. Mat. in Civ. Engrg., 17(6), pp. 640-649, 2005. [4] Çokça, E., Use of Class C fly ashes for the stabilization of an expansive soil. Journal of Geotechnical and Geoenvironmental Engineering, 127(7), pp. 568-573, 2001. [5] Hesham, A.H.I., Treatment and improvement of the geotechnical properties of different soft fine-grained soils using chemical stabilization. PhD. Thesis, Martin Luther Halle-Wittenberg University, Germany, p. 182, 2006. [6] Adu-Gyamfi, G., A Generalized theory for fly ash modified soils. Ph.D. Thesis, Department of Civil Engineering and the Russ College of Engineering and Technology, p. 247, 2006. [7] Brooks, R.M., Soil stabilization with fly ash and rice husk ash. International Journal of Research and Reviews in Applied Sciences, 1(3), pp. 209-217, 2009. [8] Ingles, O.G. & Metcalf, J.B., Soil Stabilization, Butterworths, Melbourne, 1972. [9] Marsellos, N., Christoulas, S. & Kolias, S., Use of Fly Ash In Road Construction. KEDE Bulletin, 3-4, Athens, 1986. [10] Sridharan, A., Rao, S.M. & Murthy, N.S., Liquid Limit of Montmorillonite Soils. ASTM Geotechnical Testing Journal, 9(3), pp. 156-159, 1986. [11] Akoto, B.K.A., Influence of Flyash on the Strength Characteristics of LimeLaterite Soil Mixtures. Australian Road Research 18(4), pp. 224-231, 1988. [12] Athanasopoulou, A., Improvement of the Mechanical Properties of Materials Used in Earthworks and Pavements. Application to Soils Encountered in the Area of Thrace, Ph.D. Dissertation, Department of Civil Engineering, Democritus University of Thrace, p.569, 1995. [13] Xeidakis, G.S., Assessment of the engineering and other properties of expansive soils by various methods. Ph.D. Thesis, Dept. of Civil Engineering, University of Leeds, England, 407 p., 1979. [14] Baykal, G., Arman, A. & Ferrell, R., Accelerated Curing of Fly Ash-Lime Mixtures. Transportation Research Record, 1219, pp. 82-92, 1989.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
339
Characteristics of a bolted joint with a shape memory alloy stud N. Ould-Brahim, A.-H. Bouzid & V. Brailovski École de Technologie Supérieure, Montreal, Canada
Abstract Creep is an important factor that contributes to the load loss and tightness failure of bolted joints. Retightening of the joint can be expensive, time consuming and therefore is an undesirable solution. Currently most efforts are focussed on reducing load losses directly by tightening to yield, improving material creep properties or making joints less rigid. An alternative solution of current interest is the use of bolts in shape memory alloy (SMAs). However, very few experimental studies are available that demonstrate its feasibility. The objective of this study is to exploit the benefit of the shape memory and superelasticity behaviors of a SMA stud to recover the load losses due to creep and thermal exposure of a gasket in a bolted joint assembly. This paper explores several avenues to investigate and model the thermo-mechanical properties of a bolted joint with a Nickel-Titanium SMA stud. A stiffness-based analytical model which incorporates the Likhachev model of SMA is used as a representation of an experimental bolted joint assembly. Using this model the rigidity of the experimental setup is optimized to make the best use of the SMA properties of the stud. This theoretical model is validated by a Finite Element (FE) Model using a custom FE material model which also implements the SMA material model. Finally an experimental test bench with an optimized stiffness derived from analytical simulations is used, with and without gaskets to demonstrate the ability of the SMA stud to recover load losses. Keywords: shape memory alloys, bolted joints, creep, superelasticity, SMA.
1 Introduction Load losses due to creep in any bolted joint can be problematic, even small creep losses of 0.1mm, can cause a total loss of bolt load. Several methods are in use to WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line) doi:10.2495/MC110301
340 Materials Characterisation V attempt to reduce these losses, including tightening to yield, using gasket materials or designs that reduce creep losses, or using less rigid joints. Some research has been done in the use of Shape-Memory-Alloys in a bolted joint in different ways. Peairs et al. [1], conducted research in using a thick SMA washer in a bolted assembly based on a piezo crystal to qualify the load in the joint. When a load loss is detected in the joint a heating element would activate the SMA washer. They demonstrated that load recovery was possible, however due to the nature of piezoelectric crystals the actual load in the joint and the percentage load gained was not determined. Antonios et al. [2] as well as Hesse et al. [3] tested a thick SMA washer with a load cell. Based on the same concept of heating the element when a load loss was detected they succeeded in recovering the load using the SMA washer. Labrecque et al. [4] used a Belleville washer to recover load loss in a bolted joint, by electrically heating the washer when the load was lost. The heating activated the washer, and generated additional load, however after the cooling stage it was found that significant if not all the total load was lost depending on the operating conditions. Ma et al. [5, 6] simulated the use of a bolted joint assembly to absorb seismic energy in structural joints. The simulation was conducted primarily by FEM and demonstrated the ability of SMA bolts to absorb significant stains and return to their initial state, however no experimental work using SMA bolts was conducted. Overall, the concept of using SMA in a bolted assembly has undergone some experimental and theoretical research, however most of the experimental work focuses on directly creating load losses by loosening the bolt, and not by imposing creep losses. Furthermore the bulk of the experimental work was conducted on SMA washers; however for the application of SMA bolts in bolted joint assemblies, little experimental work is available in the literature. This paper investigates the use of a Ni-Ti SMA stud in a bolted joint assembly to recover creep losses. The research will focus on using the combined effects of Shape Memory and Super-Elasticity (see Fig. 1) in order to recapture load losses due to creep. First an analytical model is developed permitting a quick evaluation of the behavior of SMA in a bolted joint. A finite element model is also developed to demonstrate the ability of FEA to tackle more complex joints such as with multiple bolts in SMA. These models incorporate the Likhachev model of SMA [7], incorporated into MATLAB and ANSYS through a user subroutine by Therriault et al. [8]. The models are then validated against an experimental study of the use of a SMA bolt in a bolted gasketed joint.
2 Analytical model In order to make best use of the shape-memory and super-elasticity effects of the nickel-titanium stud in the experimental test apparatus, a simplified analytical model was implemented. The model consists of 4 elements: The SMA stud, a spring element to represent the stiffness of the test rig, a rigid element which is used to simulate
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
341
the thermal expansion of the flange, and a creep element to simulate the properties of the gasket. ∗
∆
1
∆ ,
Shape Memory Effect
1
1
Super-Elasticity
SMA Rod
SMA Rod at hot temperature
Force applied
Force applied Bent rod at low temperature
Releasing force returns rod to undeformed shape
Heating returns rod to original shape
Cold rod retains old shape Shape Memory effect can be used to generate additional force if the rod is constrained during heating
Super-Elasticity effect can be used to recover stains with considerably less load loss than conventional metals
Figure 1:
Schematic representation of shape memory and super-elasticity.
Equal Force and Displacement Constraint Spring Element SMA Element
Thermal Expansion Element
Creep Element
Figure 2:
Analytical model.
The displacement obtained, along with the temperature is forwarded to the SMA element, which yields a new force FSMA. This is converged iteratively for each temperature or displacement change resulting in a 1D simulation of the behaviour of the experimental rig. WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
342 Materials Characterisation V The Likhachev micromechanical model of SMA, implemented into a numerical MATLAB [11] subroutine [8], was used in order to represent the thermo-mechanical behaviour of the SMA Stud. The stiffness element, which in the experiment consists of Belleville washers, is used to simulate the desired flange assembly joint rigidity. The rigid element uses the thermal expansion coefficient of steel and the clamped length of the modelled flange to determine thermal strains of the model. Finally the creep element can use either a logarithmic thermal creep law, experimental curve fits of creep data, or simply consider a creep displacement linearly. Whereas the linear model does not give an accurate time representation of creep over time, it gives accurate stresses and strains before and after creep losses if the creep displacement is known.
Figure 3:
Effect of creep on SMA and steel bolt.
As shown in Fig. 2, there is considerable loss in bolt load during the heating phase, before the austenitic transformation temperature. This is primarily caused by the properties of the shape memory alloy stud. These properties can vary considerably based on the heat treatment given and the nickel titanium ratios [9]. Using the linear creep model, a gasket creep displacement of 1 mm was introduced during heating in a joint having an SMA bolt and compared to a joint with a B7 steel bolt. It is clear that for large creep such as those expressed with PTFE gaskets in industrial flanges, the SMA stud retains much higher loads than the steel bolt.
3 Numerical FE model The axisymmetric arrangement of the simulated flange with an SMA stud in the center lends itself well to a 1 dimensional analysis which can be treated analytically without great difficulties. However in order to model more complex flanges and configurations with multiple bolts, and various non symmetric WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
343
thermo-mechanical loading, requires a more detailed model. In order to accomplish this modeling, a numerical FEM model was developed using ANSYS workbench [10]. Rigid Nut Steel Washer Belleville Washer Centering Plate Flange SMA element Gasket creep induced by displacement constraint
Figure 4:
FE model of the experimental assembly.
The modeling of the shape memory behavior is accomplished using a user subroutine material model that is incorporated to ANSYS [8]. This can adequately represent the tension loads in a bolt. The advantage of using an FEM model is that multiple SMA and non SMA bolts can be used in various configurations to design and optimize a complex structure such as a bolted flange which utilizes SMA to recapture creep losses. Since the contact stresses between the metal elements do not greatly influence the properties of the assembly, they were modeled using point constraints. This results in a model of the experimental assembly which has a constant stiffness. This model is then used to validate the analytical model, with excellent results as the stiffness’s used are the same as shown in Fig. 5.
4 Experimental setup The experimental test rig allows a gasket to be placed between two hollow cylinders with a B7 steel bolt or an SMA bolt in the center. A cartridge heater is placed around the assembly and the exposed surfaces are insulated to maintain an even temperature distribution. The top flange is instrumented with a full strain gauge bridge, calibrated to measure axial load. The SMA bolt is instrumented with thermocouples and strain gauges, to ensure even temperature distribution and measure the strain. Thermocouples are also used to measure air temperature to control the heater and the temperature of the bolt. WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Stress (MPa)
344 Materials Characterisation V 400 350 300 250 200 150 100 50 0
ANSYS Model MATLAB Model
0
50
100
150
200
Temperature (C) Figure 5:
Comparison of the analytical model and FEM variation of load with temperature of the SMA bolt. Top Plate Thermocouples Centering Plates Top Flange
Thermocouple
Strain Gauges for Load Determination
Gasket Postion Bottom Flange Centering Plate Bottom Plate SMA/B7 Rod
Figure 6:
Strain Gauges
Experimental SMA bolt assembly.
5 Results and discussion The experimental results are conducted in two stages. A small scale test using wire samples of the SMA rod (from which the stud is machined) were conducted by straining the samples to various strain levels, then heating them while restraining the displacement, and measuring the force generated. With this characterization available for small scale tests the Analytical and Finite Element models can be supplied with the necessary information they require to model the behavior of this specific sample of Ni-Ti alloy.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
400
5% elongation
3% elongation
345
7% elongation
Stress (MPA)
300 200 100 0 0
20
Figure 7:
40
60
80 100 120 Temperature (C)
140
160
180
200
Small scale characterization of SMA alloy.
30 Ni‐Ti SMA Bolt
B7 Steel Bolt
25
Gasket Stress (MPa)
20 15 10 5 0 0
Figure 8:
20
40
60
80 100 Temperature (C)
120
140
160
Comparison of the behavior of a B7 and an SMA stud in an experimental bolted gasketed joint.
A 3/8” B7 Steel stud is compared to a SMA stud both loaded to a gasket stress of 15MPa (2.1ksi) with a 1/8” expanded PTFE gasket. Both are then heated to 150°C (300°F) and allowed to creep for one hour. The results are compared and visibly demonstrate the advantages of an SMA bolt. The gasket thicknesses were measured post test, and the SMA bolted joint creeped 40% more than the B7 joint, likely due to the increased load. This further shows the advantages of SMA since even with the additional creep, the SMA bolt still generated a considerable amount of load, whereas the steel bolt lost a considerable amount of load. Comparing small scale test and experimental test rig results with Likachev model results yields a good correlation between all 3. The experimental rig results seem to generate higher loads than predicted. This is due to the residual stresses generated by cutting the smaller wire samples from the rod, which has a detrimental effect on their properties.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
346 Materials Characterisation V 600 5% elongation
SMA Bolt
Likachev Model
Stress (MPA)
500 400 300 200 100 0 0 Figure 9:
50
100
Temperature (C)
150
200
Comparison of small scale, experimental and numerical results.
6 Conclusion In conclusion, preliminary results are promising and the use of a SMA rod has considerably reduced the load loss due to creep in the gasket. It was also demonstrated that additional load can be generated to further compensate creep losses at higher temperatures. The application of the Likachev model of SMA in both the analytical and FEM models yields good congruence with experimental results.
References [1] Peairs, D.M., Gyuhae Park; Inman, D.J., 2004, “Practical issues of activating self-repairing bolted joints,” Smart Materials and Structures, v 13, n 6, p 1414-23, [2] Antonios, C, Inman D.J, Smaili, A., 2006, “Experimental and Theoretical Behavior of Self-healing Bolted Joints,” Journal of Intelligent Material Systems and Structures, v 17, n 6, p 499-509, [3] Hesse, T., Ghorashi, M., Inman, D.J., 2004, “Shape memory alloy in tension and compression and its application as clamping-force actuator in a bolted joint, Part 1 – experimentation,” Journal of Intelligent Material Systems and Structures, v 15, n 8, p 577-87 [4] Labrecque, C., Braunovic, M., Terriault, P., Trochu, F., Schetky, M., 1996, “Experimental and theoretical evaluation of the behavior of a shape memory alloy Belleville washer under different operating conditions,” Electrical Contacts, Proceedings of the Annual Holm Conference on Electrical Contacts, p 195-204.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
347
[5] Ma, Hongwei, Wilkinson, T., Cho, C., 2007, “Feasibility study on a selfcentering beam-to-column connection by using the superelastic behavior of SMAs,” Smart Materials and Structures, v 16, n 5, p 1555-1563 [6] Ma, Hongwei, Cho, C., 2007, “Application of superelasticity of SMAs in bolted end-plate connection,” Key Engineering Materials, v 353-358, pt.4, p 3039-42. [7] V.A. Likhatchev, V.G. Malinin, Structure-Analytical Theory of Strength, Nauka, St-Petersburg, 1993 (in Russian). [8] Therriault, P., Viens, F., Brailovski, V., 2006, “Non-isothermal finite element modeling of a shape memory alloy actuator using ANSYS,” Computational Materials Science, v 36, n 4, p 397-410. [9] Brailovski, Vladimir, Prokoshkin, Sergei D., Khmelevskaya, Irina Yu., Inaekyan, Karine E., Demers, Vincent, Dobatkin, Sergei V., Tatyanin, Evgeny V., 2006, “Structure and properties of the Ti-50.0 at%Ni alloy after strain hardening and nanocrystallizing thermomechanical processing,” Materials Transactions, v 47, n 3, p 795-804 [10] ANSYS, 2003, ANSYS Standard Manual, Version 11.0. [11] Matlab, 2007, Version 7.4.0.287 (R2007a).
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
This page intentionally left blank
Section 7 Thermal analysis
This page intentionally left blank
Materials Characterisation V
351
Experimental validation of a thermal model of adhesively bonded scarf repairs for CFRP composite materials incorporating cure kinetics C. C. N. Bestley, S. G. R. Brown & S. M. Alston Materials Research Centre, College of Engineering, Swansea University, UK
Abstract Adhesively bonded scarf repairs are the preferred method of repairing modern composite structures as they provide high strength restoration and aerodynamic flushness. Curing of the adhesive bondline is carried out by locally heating the repair area. To assess repair design and heating practices simulation can be used to model both the transient heat transfer during curing and the level of cure likely to be achieved at different regions in the adhesive joint. In this paper a 3D curing model is described to simulate heat transfer through a composite component. The cure kinetics of a commercial epoxy resin adhesive have been determined using isothermal Differential Scanning Calorimetric (DSC) analysis. Using these kinetics the model is able to determine the influence of the exothermic reaction within the adhesive on the overall temperature variation within the component. An experimental programme has been carried out where composite material and bonded repair patches have been cured with thermocouples providing measured temperature/time data during the cycle. The results from the cure model are then validated by comparison with these experimental results. The cure model is capable of being used to optimise the cure cycle for a bonded repair, ensuring the maximum degree of cure of the adhesive with minimum variation of temperature within the bond line. Keywords: cure kinetics, DSC analysis, numerical model.
1 Introduction The increase in use of composites in the aerospace, automotive and civil engineering industries has led to the increasing demand for development of WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line) doi:10.2495/MC110311
352 Materials Characterisation V suitable repair methods [1]. Bonded repairs have been proven to restore strength without compromising structural integrity [2]. In addition, the weight reduction and aerodynamic properties of bonded repairs make them more favourable to their bolted counterpart [3]. The mechanical properties of the adhesives used for bonded repairs are highly dependent on the curing process [4,5]. Curing involves a complex set of chemical reactions to elongate and crosslink pre-polymer molecules to form a 3D molecular network. This is followed by the materials’ transition from a viscous fluid to a viscoelastic solid [6]. Experimental investigation into the curing of composites produces the most accurate and reliable results, although this is often expensive and time consuming. As a result computational models are often used to simulate the curing process. These models have the advantage of speed, low cost and the ability to simulate ideal conditions. Various studies have been conducted to simulate the curing process for thick thermosetting matrix composites. Early work by Loos and Springer [7] formed the basis for many of the cure models that followed. Their model simulated the curing process of a flat plate in an autoclave using the finite difference method. They were able to simulate residual stress development and void formation. Later work by Bogetti and Gillespie [8] involved a two dimensional cure simulation of a thick anisotropic thermosetting composite using thermal and chemical kinetics. This model also used the finite difference method and was based on the fundamental principles discussed in the Loos and Springer paper, although with an added second dimension. Zhu et al. [9] produced a three-dimensional cure model that simulated heat transfer across a composite component, the cure process, residual stresses developed and deformation. Finally, Cheung et al. [10] produced a threedimensional thermo-chemical cure simulation based on the Galerkin finite element method. The rate and degree of cure throughout the curing process were determined and induced residual stresses and the resulting deformation were evaluated. This paper was the most influential in the formation of the cure model used in this investigation.
2 Model description A flat plate with dimensions of 560mm x 560mm with a uniform thickness of 10mm with a simple 120°C cure cycle was simulated. The repair area consisted of a patch of 260mm diameter with 5mm thickness and a 3° scarf angle. The patch is bonded into the repair area using the same commercial film adhesive as used in the DSC experiments. The cure cycle for this film adhesive is provided by the manufacturer. The equations were solved using a Gauss-Seidel algorithm where the temperature and degree of cure were determined at each time step. The cure kinetics equations in this work were based on the Cheung et al. [10] paper and the heat transfer methodology is based on the work of Patankar [11] entitled ‘Numerical heat transfer and fluid flow’. WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
353
The three-dimensional time dependent energy balance equation is given as: (1) where ρ is the density (kg m-3) , C is the specific heat capacity (J kg-1 K-1), k is the thermal conductivity (W m-1 K-1) and T is the temperature (K). S is a source term. Discretisation of the time dependent energy balance equation forms the following equation: (2) where: ∆ ∆
∆ ∆
∆ ∆
∆ ∆
∆ ∆
∆ ∆
∆ ∆ ∆
(3-9)
The values of aE to aB represent the heat conductance between the grid point P and the corresponding neighbour in the model where Δx, Δy and Δz represent the dimensions of the control volume. The variable ‘b’ is a constant consisting of the internal energy and the rate of heat generation resulting from the term SC. This term is given as: where Q is the heat flow per unit mass (J/kg)
(10)
The harmonic means of the thermal conductivities were used to account for the different materials used. Finally the value of
TP is defined: (11)
This is the temperature at grid point P. The degree of cure, α, is found to be related to the rate of reaction rα by the rate equation [10]:
(12)
This rate of reaction is given by the well established Kamal and Sourour [12] equation: 1 WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
(13)
354 Materials Characterisation V where K is the rate constant of the Arrhenius equation and m and n are constants dependent on the resin system. The Arrhenius equation is used to determine the rate of chemical reactions. It represents the dependence of the rate constant K of chemical reactions on the absolute temperature T and activation energy E and is given as: (14) where A1 represents the pre-exponential factor (sec-1) and E1 is defined as the activation energy (J/mol). The rate constant, K, is the rate of reactions taking place during cure and R is the universal gas constant. The values for these cure kinetics and the value of Hr was found using Differential Scanning Calorimetry (DSC).
3 Cure kinetics evaluation by DSC Differential Scanning Calorimetry can be used to determine the glass transition temperature (Tg) and degree of cure (α) of a commercial film adhesive by measuring the heat flow into and out of a sample as it is heated (at a constant heating rate) at a predefined temperature in a nitrogen purged atmosphere. This method is based on the ASTM E698-05 [13] and involves the analysis of DSC data at various heating rates to derive model-free kinetics. Small samples (weighing approximately 20 mg) were placed in an aluminium pan to be used in a Perkin Elmer Jade DSC. The sample was then rapidly heated to a number of different temperatures. The exothermic reaction produced when the adhesive cured was recorded and a normalised graph was produced of each temperature. The area under the exothermic peak was used to isolate data specific to the cure of the adhesive, and this data recorded for each temperature provided the degree and rate of cure (α, δα/δt respectively) at each time increment.
Figure 1:
Degree of cure of the epoxy film adhesive.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
355
The equation for the rate of cure has been given as: ∝
∝
1 ∝
(15)
where K1 and K2 are rate constants and m and n are reaction orders. Further work using the DSC measurements showed the maximum degree of cure does not reach 100% as towards the latter stages of the reaction it becomes diffusion controlled. The current model was based on a final 100% cure for the adhesive. Some models [14] have used a temperature dependent function αmax as the maximum degree of cure to account for the diffusion controlled process in the final stages of the curing reaction. This modification of the Kamal and Sourour [12] equation is as follows: ∝
∝
∝
(16)
This forms the basis for model 2 cure kinetics (see Table 1). The least square best fit method can be used to minimise this difference between the calculated and measured values by altering the values of K1, K2, m and n. This produces ‘fitted’ values of K1, K2, m and n for all temperatures which can be used to find the cure kinetics of each model.
Figure 2:
Model 1 Arrhenius plot of rate constants lnK1 and lnK2 as a function of cure temperature.
Arrhenius found that by taking the natural logarithm (ln) of both K1 and K2 and plotting them in a graph against the inverse of temperature (1/T) two linear correlations are produced, with gradient and intercept of (-E/R) and A respectively [15]: ln
ln
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
(17)
356 Materials Characterisation V This method was then used to find the cure kinetics for model 2. The cure kinetics for both models were found to be: Table 1: A1 (sec-1)
A2 (sec-1) 5
1
8.2985x10
2
3.435x1029
1.8294x10
12
4.8042x106
Cure kinetics. E1 (J/mol)
E2 (J/mol)
m
n
72203
105960
1.1655
3.0108
257126
64296
0.8892
1.5493
These values were placed in the model to determine the effect of the exothermic reaction during cure on the overall heat transfer through a composite component.
4 Cure model results A cell-centred finite difference model was formulated so that the component geometry, cure cycle, the temperature of the local environment and insulation could be changed with each simulation. This is so that the model can simulate the bonding of repairs on different components in various conditions. Experiments discussed later were based on some of these simulations, so that the model could be validated.
Figure 3:
Model geometry (quarter).
The following table shows an example of the simulations used to determine the effects of component thickness, ambient temperature and ramp rate. Insulation effects were controlled by a change in heat transfer coefficient (HTC). As expected, an increase in component thickness leads to an increase in variation in temperature through the panel, particularly at the edges. An increase in the surrounding temperature led to a more even distribution of heat through the panel. An increase in ramp rate to the desired cure temperature also provided a small improvement of heat distribution but did not significantly reduce the heat WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
Table 2: Model No. 1 2 3 4
Cure Temp (°C) 120 120 120 120
Thickness (mm) 10 20 10 10
357
Model variations. Ambient Temperature (°C) 25 25 50 25
Rate (°C/min)
HTC (Wm-2K-1)
1 1 1 3
15 15 15 15
lost through the edges of the panel. However, the effect of ramp rate on the curing of the adhesive was significant. The increased ramp rate produced a far higher degree of cure of the adhesive during the early stages of the cure cycle.
Figure 4:
Figure 5:
Degree of cure modelling.
Effect of Titanium bolt on α.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
358 Materials Characterisation V The influence of metallic components was investigated in the form of a metallic bolt through the patch and surrounding structure. The bolt (assumed to be manufactured from titanium) has a higher thermal conductivity than the composite structure. However, the results from the simulation showed that the presence of a metallic component increased the temperature of the local area by an insignificant amount and the overall heat transfer through the component was largely unaffected.
5 Experimentation The model is based on the conduction of heat from a source above the component. This is representative of a modern heating method used specifically for repair of large composite components or for in-service applications. This method involves a hot bonder used as the heat source of a number of CFRP composite components of different thickness and geometry. These results were used to validate the model and to suggest areas that required further investigation. Thermocouples were embedded into a 900mmx900mmx10mm panel in the bottom right hand corner of the panel to match the model. Thermocouples 1-16 were placed after plies 8, 16, 24 and 32 in the arrangement shown in figure 6.
Figure 6:
Embedded thermocouple readings and layout.
The centre thermocouples demonstrated little variation in temperature and were able to reach the desired cure temperature with ease. The thermocouples closer to the edges exhibited a significantly lower temperature. The results of these experiments were then compared with the cure model.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
359
6 Results and discussion The thermocouple readings for this trial were comparable to the results produced by the cure model. The panel was inadequately insulated at the edges to clearly demonstrate heat loss through the edges and underside of the panel. This resulted in heat lost through the edges of the panel, particularly in the corner. The model was able to demonstrate this heat loss with a high degree of accuracy. The most significant modification was made to the heat transfer coefficient at the edges and underside of the component. From experimental data and the results from the model a small range of HTCs (between 10-20 Wm-2K-1) were determined based on insulation and environmental temperature. This demonstrates the effectiveness of insulation and the detrimental effect environmental conditions can have on heat transfer through a composite component. The assumptions made in the formation of the cure model should be taken into account when viewing the results.
Figure 7:
Corner view heat distribution.
For example, the thermal conductivity values for each material were considered to be the same in all directions. In practice, the thermal conductivity of carbon fibre in the transverse direction is approximately 25% of that in the longitudinal and through-thickness directions [16]. However, the experimental trials compare favourably with the results from the models. WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
360 Materials Characterisation V
7 Conclusions A finite difference model was produced to simulate heat transfer through a flat CFRP composite panel during the cure cycle of a bonded repair process. The effects of heating rates, insulation and environmental temperature were simulated and the results validated through experimental investigation. Further modification of the model from experimental results has improved both the accuracy of the model and the ability to uniformly heat a composite component. The results from the model have provided insight into the nature of heat transfer under different conditions. The large variance in temperature between the centre and edge thermocouples was demonstrated in both the model and experimentation. The model provided more information as to where the heat loss is most significant, and with further trials in combination with further modelling this heat loss could be significantly reduced. The exothermic reaction during the cure of the adhesive was determined to have an insignificant impact on the overall heat transfer through the component. The incorporation of a Titanium bolt increased the local temperature and therefore degree of cure of the adhesive in the early stages of the cure cycle. However, this became insignificant towards the latter stages of the cure. Ongoing work will involve the optimisation of the cure cycle for bonded repair, residual stresses and deformation, and void formation.
Acknowledgements This work was carried out as part of the CONTOUR project in collaboration with Airbus Operations Limited under Welsh Assembly Government contract reference number HE 09 COL 1030.
References [1] Tomblin, J.S., Salah, L., Welch, J.M., Borgman, M.D., Bonded Repair of Aircraft Composite Sandwich Structures, Final Report, DOT/FAA/AR03/74, Office of Aviation Research, Washington D.C. 20591, February 2004. [2] Baker, G., Bonded Composite Repair of Fatigue-Cracked Primary Aircraft Structure. Composite Structures, 47(1-4), pp. 431-443, 1999. [3] Charalambides, M.N., Hardouin, R., Kinloch, A.J. Matthews, F.L., Adhesively-Bonded Repairs to Fibre-Composite Materials I: Experimental. Composites Part A: Applied Science and Manufacturing, 29(11), pp. 1371-81, 1998. [4] Yi, S., Hilton, H. H., Ahmad M.F., A Finite Element Approach for Cure Simulation of Thermosetting Matrix Composites. Computers & Structures, 64(1-4), pp. 383-388, 1997.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
361
[5] Stewart, I., Chambers, A., Gordon, T., The Cohesive Mechanical Properties of a Toughened Epoxy Adhesive as a Function of Cure Level. International Journal of Adhesion & Adhesives, 27(4), pp. 277-287, 2007. [6] Yu, H., Mhaisalkar, S.G., Wong, E.H., Teh, L.K. & Wong, C.C., Investigation of Cure Kinetics and Its Effect on Adhesion Strength of Nonconductive Adhesives Used in Flip Chip Assembly. IEEE Transactions on Components and Packaging Technologies, 29(1), pp. 71-79, 2006. [7] Loos, A.C., Springer, G.S., Curing of Epoxy Matrix Composites. Journal of Composite Materials, 17(2), pp. 135-169, 1983. [8] Bogetti T.A., Gillespie, J.W., Two-Dimensional Cure Simulation of Thick Thermosetting Composites. Journal of Composite Materials, 25(3), pp. 239-273, 1991. [9] Zhu, Q., Geubelle, P.H., Tucker, C.L., Dimensional Accuracy of Thermoset Composites: Simulation of Process-Induced Residual Stresses. Journal of Composite Materials, 35(24), pp. 2171-2205, 2001. [10] Cheung, A., Yu, Y., Pochiraju, K., Three-Dimensional Finite Element Simulation of Curing of Polymer Composites. Finite Elements in Analysis and Design, 40(8), pp. 895-912, 2004. [11] Patankar, S.V., Numerical Heat Transfer and Fluid Flow, Taylor & Francis: Oxfordshire, 1980. [12] Kamal, M.R., Sourour, S., Kinetics and Thermal Characterization of Thermoset Cure. Polymer Engineering & Science, 13(1), pp. 59-64, 1973. [13] Standard Test Method for Arrhenius Kinetic Constants for Thermally Unstable Materials Using Differential Scanning Calorimetry and the Flynn/Wall/Ozawa Method, ASTM E 698-05, 2005. [14] Lee, C.L., Wei, K.H., Curing Kinetics and Viscosity Change of a Two-Part Epoxy Resin During Mold Filling in Resin-Transfer Molding Process. Journal of Applied Polymer Science, 77(10), pp. 2139-2148, 2000. [15] Lee, J.Y., Choi, H.K., Shim, M.J., Kim, S.W., Kinetic Studies of an Epoxy Cure Reaction by Isothermal DSC Analysis. Thermochimica Acta, 343 (1-2), pp. 111-117, 2000. [16] Mutnuri, B., Liang, R., GangaRao, H., Thermal Conductivity Characterization of FRP Composites: Experimental. ANTEC, May 7-11, Charlotte, NC, 2006.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
This page intentionally left blank
Materials Characterisation V
363
Computational and experimental characterization of building envelopes based on autoclaved aerated concrete V. Kočí, J. Výborný & R. Černý Department of Materials Engineering and Chemistry, Faculty of Civil Engineering, Czech Technical University in Prague, Czech Republic
Abstract Hygrothermal performances of three types of building envelopes based on autoclaved aerated concrete (AAC) provided with different thermal insulating materials (expanded polystyrene, hydrophilic mineral wool, AAC with extended thermal insulation capability) are compared. The simulations are accomplished using the computer code HEMOT based on the finite element method. Results of the simulation are the moisture and temperature fields across the building envelope, which in combination with mechanical parameters present a sufficient data source for service life analysis. Keywords: computational analysis, coupled heat and moisture transport, autoclaved aerated concrete, building envelopes, climatic conditions.
1 Introduction Autoclaved aerated concrete (AAC) is a structural material which is commonly used around Europe, particularly as it combines ease of construction with excellent combination of its mechanical and thermal properties. However, despite the very good thermal properties of AAC, it can be anticipated that with the increasing demand for energy savings it will become a necessity to provide it with thermal insulating system or at least very good thermal insulating plaster to meet stringent conditions given by future thermal standards. The choice of proper insulating material for AAC is though quite difficult with respect to service life of the whole building envelope. As it was demonstrated in [1–3], most common insulation materials cause extreme hygrothermal straining of WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line) doi:10.2495/MC110321
364 Materials Characterisation V external finishes so it is necessary to choose wisely. The proper method, how to compare the results is to use computational tool for simulating coupled heat and moisture transport in multi-layered systems of porous building materials. One of such tools which were already successfully tested in previous calculations is HEMOT developed at the Department of Materials Engineering and Chemistry, Faculty of Civil Engineering, Czech Technical University in Prague. This computer code which will be employed in this paper uses Künzel’s mathematical model for coupled heat and moisture transport. The equations are solved using finite element method.
2 Computational analysis The computer code HEMOT [4] is based on the general finite element package SIFEL [5]. As basic input parameters of the mathematical model, hygric, thermal and basic physical parameters of used materials, construction detail, initial and boundary conditions and time specification of simulation are required. Description of all input parameters in more detail is given later. In the computer simulations we focused on a comparison of hygrothermal behavior of several building envelopes based on AAC provided with different thermal insulating materials. 2.1 Mathematical model Künzel’s mathematical model of heat and moisture transport [6] was used in the simulations which can be formulated as
d v div D grad p grad ps d t
(1)
dH T divgradT Lv div p grad p s dT t
(2)
where v is the partial density of moisture, relative humidity, p permeability of water vapour, ps partial pressure of saturated water vapour, H enthalpy density, Lv heat of evaporation of water, thermal conductivity and T temperature,
D Dw
d v d
(3)
is liquid moisture diffusivity coefficient, Dw capillary transport coefficient. 2.2 Scheme of construction detail Three variations of building envelope based on AAC were chosen for simulation, in order to analyze the consequences of different material combinations. As a WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
365
start-up building envelope we chose AAC without any external finish which allowed us to get real image about hygrothermal performance of AAC layer itself [7]. In the next simulation we provided AAC with hydrophilic mineral wool (marked as Var. 1), expanded polystyrene (Var. 2) and Multipor Ytong produced by Xella CZ (Var. 3). All these envelopes were provided from interior and exterior side with Baumit MVR Uni plaster which is recommended for AAC structures as external finish. On the material interface between mineral wool and AAC an adhesive mortar layer was placed. Description of used materials in more detail is given in next subsection. Scheme of construction detail including the dimensions of each layer is shown in Figure 1. In all investigated variations we focused on the hygrothermal conditions at points within the AAC layer and external plaster just 2 mm under its external surface which can be considered as characteristic position from the point of view of possible frost damage.
Figure 1:
Scheme of AAC-based building envelope.
2.3 Material parameters Aerated autoclaved concrete P4-500 produced by Xella CZ was under consideration in this paper as the load-bearing material. For exterior and interior renders we used Baumit MVR Uni Plaster, which is single-layer plaster for exterior and interior surfaces especially recommended for AAC. As the thermal insulation we assumed Rockwool hydrophilic mineral wool, expanded polystyrene and Multipor. For adhesive layer between AAC and mineral wool we used Mamut M2 mortar. All the material parameters were measured in laboratory of transport processes at the Department of Materials Engineering and Chemistry, Faculty of Civil Engineering, Czech Technical University in Prague [8–11] and are summarized in Table 1 and Figure 2. Data for Mamut M2 mortar were measured by M. Jerman and have not been published yet. We used these symbols: ρ – bulk density [kg/m3], mat – matrix density [kg/m3], porosity%], c – specific heat capacity [J/kgK], μ – water vapour diffusion resistance factor [-], w – moisture content by volume [m3/m3], λ – thermal conductivity [W/mK], – moisture diffusivity [m2/s]. WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
366 Materials Characterisation V Table 1: Parameter
AAC P4-500
Mamu t M2 mortar
ρ [kg m-3] %] c [J kg-1 K1 ] μ [-]
500
λdry [W m-1 K-1] λsat [W m-1 K-1] [m2 s-1]
Expanded polystyrene
Multip or
1430
Hydrophi lic mineral wool 71
50
125
Baumit MVR Uni plaster 1402
80.2 1020 – 1510 3.0 – 9.7
42.6 1020
96.0 810
97.0 1300
12.4
4.3
50
0.114
0.481
0.043
0.040
94.2 2230 – 3500 1.9 – 10.9 0.047
44.4 1020 1780 4.5 – 12.4 0.443
0.454
2.022
0.246
0.560
0.166
1.380
Fig. 2
1.07e9 0.201
8.4e-6
2.10e-11
Fig. 2
1.59e-9
0.000046
0.001
0.0078
0.042
0.01846
whyg [m3 m-3]
Material characteristics.
Moisture diffusivity [m2/s]
1.00E-05
1.00E-06
1.00E-07
1.00E-08
AAC P4-500 Multipor 1.00E-09
1.00E-10 0
0.05
0.1
0.15
0.2
0.25
0.3 3
0.35
0.4
3
Moisture content by volume [m /m ]
Figure 2:
Moisture diffusivity.
2.4 Initial and boundary conditions and time interval of simulation As the initial and boundary conditions climatic data in the exterior in the form of Test Reference Year for Prague which contained average data for 30 years were used. On the interior side constant value of relative humidity 55% and
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
367
temperature 21°C (see Fig. 3) was chosen. The simulation started on 15th July and took 6 years. The final results show data obtained for the last year.
Figure 3:
Boundary conditions.
3 Computational results The results of computational simulations are summarized in a set of figures which describe hygric and thermal performance of studied material of building envelope during a reference year. In all figures we focused on moments when moisture content and temperature reached certain limits simultaneously. In case of moisture content this limit was the value of hygroscopic moisture content (see Tab. 1), in case of temperature the freezing point of water. When these two conditions are fulfilled, contained liquid moisture is getting frozen. This leads to consequent damage of material. The materials capability to resist to freezing of contained water is characterized by its freeze-thaw resistance which can be measured under laboratory conditions [12]. This was accomplished for AAC and its freeze-thaw resistance was set to 25 cycles [13]. Durability of AAC can be then calculated as quotient of freeze-thaw resistance and number of freezing cycles appearing in the material in building envelope during a year. 3.1 AAC wall provided with hydrophilic mineral wool Hygrothermal performance of AAC and external plaster of building envelope provided with hydrophilic mineral wool is captured in Figures 4 and 5. The temperature in AAC block 2 mm under its external surface is lower than in other investigated envelopes, but freezing point of water is not reached and level of moisture content is deeply in underhygroscopic range. That means that AAC block in not threatened by effects of freezing water. Hygrothermal performance of exterior plaster is shown in Figure 5. Although the temperature drops below zero many times during a reference year, this does not happen simultaneously with moisture increase. So, there is not any water which can freeze and damage the structure of plaster. During a reference year, the overhygroscopic moisture content is reached only five times and only in summer months. So we can summarize, that even in the plaster there are not any freezing cycles. WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
368 Materials Characterisation V 0.10 Temperature 273.15 K Temperature Hygroscopic moisture content Moisture content
0.08
3
Temperature [K]
0.07
3
300
0.09
Moisture content [m /m ]
320
280
0.06 0.05
260 0.04 240
0.03 0.02
220 0.01 200 731
0.00 781
831
881
931
981
1031
1081
Time [days]
Figure 4:
Hygrothermal performance of AAC, Var. 1. 0.10 Temperature 273.15 K Temperature Hygroscopic moisture content Moisture content
0.08
3
Temperature [K]
0.07
3
300
0.09
Moisture content [m /m ]
320
280
0.06 0.05
260 0.04 240
0.03 0.02
220 0.01 200 1700
0.00 1750
1800
1850
1900
1950
2000
2050
Time [days]
Figure 5:
Hygrothermal performance of exterior plaster, Var. 1.
3.2 AAC wall provided with expanded polystyrene The results of hygrothermal performance of AAC wall provided with expanded polystyrene are shown in Figures 6 and 7. In AAC, the values of temperatures keep above 0°C for all the year, the level of moisture content stays in underhygroscopic range. It means that any single condition for creation of freezing cycle is not fulfilled. WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
369
0.10 Temperature 273.15 K Temperature Hygroscopic moisture content Moisture content
0.08
3
Temperature [K]
0.07
3
300
0.09
Moisture content [m /m ]
320
280
0.06 0.05
260 0.04 240
0.03 0.02
220 0.01 200 1700
0.00 1750
1800
1850
1900
1950
2000
2050
Time [days]
Figure 6:
Hygrothermal performance of AAC, Var. 2. 0.10
Temperature 273.15 K Temperature Hygroscopic moisture content Moisture content
320
300
0.09 0.08
0.06 0.05
260 0.04 240
3
Temperature [K]
3
Moisture content [m /m ]
0.07 280
0.03 0.02
220 0.01 200 1700
0.00 1750
1800
1850
1900
1950
2000
2050
Time [days]
Figure 7:
Hygrothermal performance of exterior plaster, Var. 2.
A different situation appears in exterior plaster as one can see in Figure 7. There are few moments, when the level of moisture content reaches overhygroscopic range for sufficiently long time. The most interesting moments are these in winter, when the temperature drops below zero (around 1800th day of simulation). In hygrothermal simulation of this type of building envelope, 15 freezing cycles per a year are counted in exterior plaster. 3.3 AAC wall provided with Multipor Figures 8 and 9 capture hygrothermal performance of AAC and exterior plaster of wall provided with Multipor. WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
370 Materials Characterisation V As in the previous variations, the hygrothermal performance of AAC is similar. There are not any freezing cycles because of relatively high temperature and low moisture content (see Figure 8). 0.10 Temperature 273.15 K Temperature Hygroscopic moisture content Moisture content
0.08
3
Temperature [K]
0.07
3
300
0.09
Moisture content [m /m ]
320
280
0.06 0.05
260 0.04 240
0.03 0.02
220 0.01 200 1700
0.00 1750
1800
1850
1900
1950
2000
2050
Time [days]
Figure 8:
Hygrothermal performance of AAC, Var. 3.
In exterior plaster applied on Multipor insulation, 10 freezing cycles are counted (see Fig. 9). Between 1789th and 1814th day of simulation the moisture is in overhygroscopic range and at the same time the temperature drops below zero so the condition for creation of freezing cycles is fulfilled. 0.10 Temperature 273.15 K Temperature Hygroscopic moisture content Moisture content
0.08
3
Temperature [K]
0.07 280
0.06 0.05
260 0.04 240
0.03 0.02
220 0.01 200 1700
0.00 1750
1800
1850
1900
1950
2000
2050
Time [days]
Figure 9:
3
300
0.09
Moisture content [m /m ]
320
Hygrothermal performance of exterior plaster, Var. 3.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
371
4 Discussion The protection of AAC walls against effects of weather conditions and freezing water in particular is certainly a necessity. It was proved already in the previous work [7]. There are two ways how to achieve that. The AAC wall can be protected against increase of moisture content which can be accomplished by suitable waterproof modification of external surface or we can protect the wall against low temperatures using one of many thermal insulating materials. The second possibility seems to be more advantageous with respect to stringent conditions given by thermal standards which have to be met. The advantage of thermal insulation applied on AAC is the fact that together with frost the AAC is protected against moisture penetrating from exterior as well. However, the application of thermal insulation goes along with negative effects. One of the most significant is extreme straining of external finishes, which are applied on. This was already indicated in [1–3] and also results obtained in this paper confirmed it. The main reason can be seen in the low moisture diffusivity of used thermal insulating materials (expanded polystyrene, Multipor). This leads to slowing down the moisture transport from exterior plaster towards the interior and to moisture increase in plaster subsequently. Considering the direct exposition of plaster to weather conditions, temperature in particular, water saturated plaster can be then easily damaged by effects of freezing of contained liquid moisture. It can be spoken in general, the lower moisture diffusivity of thermal insulating material is, the more freezing cycles in exterior plaster will appear. In our case, there were 15 freezing cycles in plaster applied on expanded polystyrene and 10 freezing cycles in plaster applied on Multipor and that corresponds to values of moisture diffusivity. Opposite situation occurred when hydrophilic mineral wool has been used. Thanks to high moisture diffusivity of hydrophilic mineral wool, all the received moisture can be quickly transported away, so the plaster stays relatively dry and does not contain the liquid moisture. On the other hand, moisture content of AAC is higher, which is obvious, when Figures 4, 6 and 8 are compared. Although the moisture content is still very low, the total amount of moisture contained in whole building envelope provided with hydrophilic mineral wool is higher than in other studied cases so that the thermal insulating capability of that building envelope may possibly decrease.
5 Conclusion According to the results obtained in this paper, protection of AAC block in building envelopes against weather conditions is a necessity. It can be accomplished in many ways; one of the best seems to be using thermal insulating materials from exterior side. In this paper, computational simulation of building envelopes provided with different types of thermal insulating materials has been accomplished. The best results were obtained when hydrophilic mineral wool was under consideration. The other variants with expanded polystyrene or WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
372 Materials Characterisation V thermally insulation AAC material Multipor did not lead to suitable results. It was caused by low value of moisture diffusivity which was the cause of increase of moisture content in exterior plaster. The level of moisture content then exceeded the hygroscopic limit and freezing of liquid water was allowed. The building envelope with hydrophilic mineral wool showed different hygrothermal performance. High value of moisture diffusivity of mineral wool allowed water to be transported deeper into the envelope which led to decrease of moisture content in external plaster. Protection against effects of freezing water was sufficiently fulfilled. However, total moisture content in building envelope was then higher. In next research, it would be advisable to focus on design of new insulating materials or modification of material parameters of current materials using inverse analysis in order to find the composition of AAC-based building envelope which will combine good frost resistance and good thermal insulating properties as well.
Acknowledgement This research has been supported by the Czech Science Foundation, under grant No. 103/09/0016.
References [1] Maděra, J., Kočí, V., Vejmelková, E., Černý, R., Rovnaníková, P. et al, Influence of material characteristics of concrete and thermal insulation on the service life of exterior renders. In: Fourteenth International Conference on Computational Methods and Experimental Measurements, Algarve, Portugal. Wessex: WIT PRESS, p. 13-23. ISBN 978-1-84564-187-0. 2009. [2] Maděra, J., Kočí, V., Korecký, T., Černý, R. et al. Computational analysis of hygrothermal performance of building envelope under different climatic conditions In: Thermophysics 2010. Brno: University of Technology, p. 180-187. ISBN 978-80-214-4166-8. 2010. [3] Jerman, M., Kočí, V., Maděra, J., Výborný, J., Černý, R, Water and heat transport parameters of materials involved in AAC-based building envelopes In: 1st Central European Symposium on Building Physics. Lodz: Technical University of Lodz, p. 39-45. ISBN 978-83-7283-367-9. 2010. [4] Černý R., Complex System of Methods for Directed Design and Assessment of Functional Properties of Building Materials: Assessment and Synthesis of Analytical Data and Construction of the System. CTN CTU in Prague, 192 – 201. 2010. [5] Kruis, J., Koudelka, T., Krejčí, T., Efficient computer implementation of coupled hydro-thermo-mechanical analysis. In: Mathematics and Computers in Simulations, vol. 80, no. 8, pp. 1578-1588. ISSN 0378-4754. 2010. [6] Künzel, H.M., Simultaneous Heat and Moisture Transport in Building Components, Ph.D. Thesis, IRB Verlag: Stuttgart, pp. 1–135, 1995. WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
373
[7] Maděra, J., Kočí, J., Kočí, V., Výborný, J., Černý, R., Computational prediction of hygrothermal conditions in innovated AAC-based building envelopes. In: Advanced Computational Methods and Experiments in Heat Transfer XI. Southampton: WIT Press, pp. 291-301. ISBN 978-1-84564462-8. 2010. [8] Jiřičková, M. and Černý, R., Effect of Hydrophilic Admixtures on Moisture and Heat Transport and Storage Parameters of Mineral Wool, Construction and Building Materials, 20, 425-434, 2006. [9] Výborný, J. Stanovení tepelných parametrů a smrštění při vysychání vybraných pórobetonových tvárnic firmy H+H Česká republika s.r.o. a Xella CZ. In: Proceedings of the International Conference 15th Construmat 2009. Praha: České vysoké učení technické v Praze, pp. 449-461. 2009. [10] Fučíková, L., Moisture Properties of AAC Blocks in dependence on the Environment Focused on Problems with Durability. Diploma thesis. Praha: Czech Technical University in Prague, 2009. [11] Jerman, M., Výborný, J., Černý, R., Tepelné a vlhkostní charakteristiky nových pórobetonových výrobků, In: Stavební obzor, roč. 20, č. 1, s. 7-11. ISSN 1210-4027. 2010. [12] ČSN EN 15304. Determination of the freeze – thaw resistance of autoclaved aerated concrete. Praha: Czech Standards Institute, 2007. [13] Výborný, J., Testing of the Drying Shrinkage and Freeze - thaw Resistance of Autoclaved Aerated Concrete. In: XII International scientific conference Technická zařízení staveb a energie budov. Brno: Akademické nakladatelství CERM, pp. 215-218. 2009.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
This page intentionally left blank
Section 8 Recycled materials
This page intentionally left blank
Materials Characterisation V
377
Quantitative description of the morphology of polyurethane nanocomposites for medical applications J. Ryszkowska & B. Waśniewski Warsaw University of Technology, Faculty of Materials Science and Engineering, Poland
Abstract This paper presents the application of stereology methods to the description of morphological properties of polyurethane nanocomposites for medical applications. The study of the cross-section surface structure of the obtained materials was performed by Atomic Force Microscopy. The volume of hard phase agglomerate was used to evaluate the degree of phase separation of the examined nanocomposites. The relationships between the domain agglomerate characteristics and the properties of nanocomposites obtained from them were analysed. The results showed that nanocomposites with non-modified nanosilica dioxide (SiO2) and nanosilica dioxide modified with NH2 groups differs from polyurethane within the following properties: size and volume of the agglomerates of the hard domains, biocompatibility, thermo-mechanical and abrasive wear resistance. Keywords: nanocomposites, polyurethane, structure, image analysis, biomedical application
1 Introduction Polyurethanes (PURs) and their nanocomposites are a versatile plastic material, formulated to provide good biocompatibility, flexural endurance, high strength, high abrasion resistance and processing versatility over a wide range of applications [1]. The most common use in medical devices is in short-term implants [2]. Polyurethane elastomers are linear segmented copolymers WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line) doi:10.2495/MC110331
378 Materials Characterisation V consisting of a relatively flexible component derived from a macrodiol called soft segment, and a relatively hard and stiff component derived from a diisocyanate and a chain extender called hard segment (Fig 1). Thermodynamic incompatibility of these segments leads to microphase separation. The domain structure formed by microscopic phase separation presents similar elastomeric properties to those shown for cross-linked rubber networks. The mechanical strength of this structure can be attributed to hard microdomains physically cross-linked through hydrogen bonding and dispersion forces, acting as fillerlike reinforcement for the soft segment [3]. Polyurethanes are characterized by a complex morphology which is dependent upon the precise nature of the hard and soft segments and their composition, use of nanoparticles, preparation method and its parameters. All these factors influence the morphological factors such as degree of microphase separation, crystallinity, the domain agglomerate characteristics, and define properties such as hardness, stiffness, tensile strength, clarity and biocompatibility [1–7].
Figure 1:
Domains structure of polyurethane.
This paper presents the application of stereology methods to the description of morphological properties of polyurethane nanocomposites for medical application. Stereological parameters [8–10] chosen for analysis were used to evaluate the degree of phase separation of the examined nanocomposites. Relationships between the domain agglomerate characteristics and the properties of the nanocomposites obtained from them were analysed.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
379
2 Experimental 2.1 Materials The following reactants were used in the synthesis of PURs: polyoxythetramethylene glycol (PTMG) of molecular weight 2023 g/mol (Therathane ® 2000) was supplied by Du Pont Nemours Co. ; 4.4’diphenylmethane diisocyanate (MDI) purchased from Aldrich Chemical Co. (Germany); ethylene glycol (GE) and glycerine (G) POCH (Gliwice, Poland). Polyol was dehydrated by mixing under vacuum for 2h at 120C. Ethylene glycol and glycerin was dried under a molecular sieve. Nanosilica powder and nanosilica powder modified with NH2 groups, with primary particle size of 75 nm were used as nanofiller (ICHP Warsaw). To prepare the composites, nanosilica – polyetherodiols 20% wt. concentrate was made. Firstly, polyetherodiol was melted down in the oven under 80˚C degree. Then nanosilica was put into it and mixed with ultrasonic homogenizer VCX 750 by Sonics during 30min in pulse mode 3/3 (3sec mixing, 3sec stop). 2.2 Polyurethane and nanocomposites synthesis Segmented polyether based polyurethanes with substrates molar ratio PTMG: MDI: GE: G equal to 40:80:27:24 (1:2:0.679:0.151), constant isocyanates index of 1.05 and with hard segment contents 20 wt.% was synthesized using a onestep polymerisation method. The polyol was cooled to 70ºC±3ºC, with glycol added and glycerin blended for 5min. Then the mixture was cooled to 60ºC±3ºC and MDI added. The samples were obtained with free casting method. The heat up process was carried out in the oven for 16 hours. To prepare the composites, appropriate quantities of concentrate were added to the polyol. The mixture was dehydrated in temperature of 120ºC±5ºC under 2-5 hPa pressure. Next the process was run in the same way as polyurethane synthesis. Description of achieved materials is presented in Table 1. Table 1: Description of the achieved materials. Sample Amount of nanofiller [wt.%] Modification of nanofiller
0 0 -
1 0.5 -
2 1.0 -
3 2.0 -
1A 0.5 -NH2
2A 1.0 -NH2
3A 2.0 -NH2
2.3 Characterization 2.3.1 Dynamic mechanical analysis Dynamic tests, by dynamic mechanical analysis (DMA) on a Thermal Instruments dynamic mechanical analyzer (Q800 TA), were also carried out in three-point bending mode on specimens with dimensions of 12 x 2 x 60mm3. Tests were performed with the amplitude of deformation during the bending of 25 m. The frequency-dependent storage modulus was also evaluated with a WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
380 Materials Characterisation V 4.5–9 Hz frequency sweep at a constant temperature of 37C with 1.5% strain and 0.01 N static force. Similar parameters were used by Hafeman and coworkers [11]. The storage modulus (E’) value was recorded as a function of frequency. 2.3.2 Abrasion resistance The abrasion resistance of test samples was measured with a Schopper– Schlobach instrument with an APGi circulating roller from Heckert, and the procedure complied with the standard PN-ISO [12]. The test pieces are in the form of a roll (16 0.2 in diameter and 2mm high). Standard rubber from (The Institute for Engineering of Polymer Materials and Dyes, Elastomer and Rubber Technology Division in Piastów) was used as reference material [13]. The abrasion resistance index (V) was calculated with the following relation:
V
Vs , 100% Vt
where Vs is the loss of volume of the standard rubber (mm3) and Vt is the loss of volume of the test sample (mm3). The density figures for the test pieces, which were necessary for calculations, were found by the method described in the standard PN-ISO [14]. 2.3.3 Atomic Force Microscopy (AFM) AFM images were recorded at 37C in air using a Digital Instrument Multimode Nanoscope V (Digital Instruments, Santa Barbara, CA) operating in the tapping regime mode using antimony doped silicon cantilever tips (POL-15, 130 do 250kHz , 48 N/m). Scanner was used with scan rates between 0.5 and 1 Hz. All images are subjected to a first-order plane-fitting procedure to compensate for sample tilt. The microstructure of polyurethane was investigated on micro sections. These were prepared using a rotary microtom RM 2165 (Leica) with a LN 21 cooling device working at -60C. 2.3.4 Image analysis Binary images revealing hard domains agglomerates were produced via digital processing of AFM images. The size and volume fractions of the hard domains agglomerates in polyurethane and nanocomposites were determined by measurements on their sections [15]. Linear covariance method was used for the description of the distribution of particles. The image was transferred to MicroMeter software and quantitative analysis was performed. The diameters were randomly determined on 5 microphotographs. 2.3.5 Fourier transform infrared (FTIR) spectroscopy Infrared spectra of PURs were collected using a FTIR spectrophotometer (Thermo Electron Corporation model Nicolet 6700). Measurements were carried out using attenuated total reflectance (ATR) technique. Each sample was scanned 64 times at a resolution of 4 cm-1 over the frequency range of 4000–400 cm-1. Analysis of FTIR data enabled to determine the carbonyl hydrogenbonding index (R). A straight baseline was drawn in the spectrum between 1780 WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
381
cm-1 and 1640 cm-1 and the carbonyl stretching zone was corrected by subtracting the baseline. To estimate the signal strengths, peak modeling of the infrared active carbonyl bands was carried out using the Gaussian curve-fitting method software OMNIC 7.3. The carbonyl absorption bands were deconvoluted into component bands, the peak area of these bands was measured and carbonyl hydrogen-bonding index R was calculated using Eq. (1) [16, 17]:
R
Abonded A free
(1)
Moreover, the degree of phase separation (DPS) was obtained through Eq. (2):
DPS
R R 1
(2)
3 Results and discussion In polyurethanes application as short-term implants the modulus of elasticity and surface properties must very often be changed. The one way to solve the problem is the use of nanofiller [18–20]. Nanocomposites exhibit advantageous mechanical and physical properties already at small addition of modifying particles, frequently lower than 5 wt%. Most of nanofillers occur without any surface modification. However, sometimes we need to modify nanoparticles to obtain better dispersion of nanofiller in polymer matrix, for instance. One of the modifications is a chemical one relying on attaching functional groups i.e. – COOH, -NH2, -NCO, -OH to nanofiller [20–24]. In this study were used polyetherourethanes with 0,5–2 wt% of nanosilica and nanosilica modified with amino groups. The change in elasticity module was examined in the course of three-point bending as well as abrasion wear of fabricated materials (Table 2). Nanofiller introduction change the storage modulus and abrasion wear of polyurethane matrix. Table 2:
Results of dynamic three point bending and abrasion wear of the PU and PU/SiO2 composites.
Sample E’, storage modulus by frequency 5Hz at 37C, MPa V, abrasion wear, dm3
0
1
24.6
2.8
2
3
32.6 30.8
1A
2A
3.4
40.6 41.6
3A
35.2 26.0 27.7 48.2 23.0 27.3 31.8
The introduction of 0.5 wt% of nanofiller brings about a decrease in storage modulus and abrasion wear; bigger amounts of the filler make the two values grow. The use of modified nanofiller results in the fact that the obtained nanocomposites have a higher storage modulus at lower abrasion wear. In order to explain the mechanism of the influence of nanofiller on polyurethane matrix structure microscopic observations were carried out of the cross-section surface of fabricated materials using AFM. An AFM tapping mode WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
382 Materials Characterisation V is used to depict topographic features and the spatial variation in surface by height and phase imaging. For the examined materials the appearances of topographic and phase images are similar and show less obviously surface morphology related to hard domain and soft domain separation. An exemplary phase image of surface cross-section of examined PU is shown in Fig. 2a. Binary images of all face images of the polyurethanes and nanocomposites were performed and those selected from them were shown in Fig. 2 (b-f). The neat PU cut surface showed microphase separation in phase images, the hard phase of the polyurethane appears brighter than the soft phase. The hard phase during phase separation formed island domain like phase. The results in all nanocomposites are similar, showing a distinct topology of phase separation of hard and soft segments (Fig 2b-f). Images of Pu phase structure and nanocomposites are similar to the images obtained by Zhang et al. [21]. The AFM images could be explained as follows: Glycerine reacted with MDI and then formed island domain like hard phase. This usually accompanied chemical reactions such as cross-link, so the segments shrank as a result [22] and formed caves like those in PU and nanocomposites. The introduction of 0.5 wt% unmodified filler (Fig. 2c) makes the appearance of smaller islands, formed by hard phase, than in PU (Fig. 2b), visible in the pictures of cross-section surfaces of the nanocomposites. The use of 0.5 wt.% SiO2 modified by NH2 groups results in the growth of the size of islands formed by hard phase; moreover, they tend to merge into bigger agglomerates (Fig. 2d). The size of the agglomerates of the hard domains was increased respectively to the nanofiller contents as shown in Fig. 2e and f c. Nanda and his co-workers have reported small amounts of POSS usually tending to combine with hard segments and affect the properties of the hard segments of PU [23]. Changes in the morphology of polyurethanes, caused by the introduction of POSS, were also observed by Zhang and co workers [24]. It can be assumed that the mechanisms of changes caused by the use of nanosilica are similar. Quantitative analysis of binary pictures was used for the description of morphology of PU hard phase and its nanocomposites. Since hard phase created islands, often merging into bigger agglomerates, the agglomerates’ volume formed by hard phase was calculated and the results are presented in Fig. 3 The difference in the level of interaction by hydrogen bond between the hard segments of the materials can regarded as an explanation of the differences in the volume of the agglomerates created by hard phase of the examined materials; hence the determination of hydrogen bond index (R) and phase separation index (DPS). The said difference in the level of interaction was determined basing on the vibrations of carbonyl groups occurring within the scope of wavenumbers 1760-1680cm-1. The comparison of spectra in this scope for materials with equal nanosilica content modified with NH2 groups is shown in Fig. 4.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
a)
Figure 2:
383
b)
c)
d)
e)
f)
AFM phase images at 5 m scan sizes of (a) sample 0 and binary images: (b) sample 0 (c) sample 1 (d) sample 1A (e) sample 3 (f) sample 3A. Z ranges: 30.
The R and DPS values of the polyurethane/nanosilica nanocomposites mixtures are given in Table 3. After the introduction of 0.5 wt.% of nanosilica, the R and DPS is higher than for polyurethane but modification causes a decrease in R and DPS. When a bigger amount of nanosilica is applied then R and DPS increase by adding –NH2 group on the surface of nanosilica. Several factors influence the DPS of polyurethane such as molecular weight, segmental length, crystallizability of soft segments, overall composition and intra- and inter-segments interactions [25–27]. The addition 0.5 wt% nanosilica causes weakening of the hydrogen-bond interactions created between polyurethane hard segments; segmental incompatibility in the polyurethane decreases by adding nanosilicas (Table 3). This explains the formation of the structure visible in the cross-section of the materials (Fig 2c, d). After the introduction of a bigger amount of the filler we observe no weakening of hydrogen-bond interactions created between polyurethane hard segments. A bigger amount of the filler increases incompatibility in the polyurethanes, causing a higher degree of phase separation in the polyurethane (Table 3). WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
384 Materials Characterisation V 80 70 60 Vv, %
50 1.
40
2.
30 20 10 0 0
0,5
1
1,5
2
2,5
SiO2 content, wt %
Figure 3:
Influence of nanosilica content on volume fraction of hard domains agglomerates in PUR: (1) nanosilica, (2) modified nanosilica.
Figure 4:
Influence of modified nanosilica content on FTIR spectrum in the range 1760–1680 cm-1. Table 3: Parameters of polyurethane phase separations.
Sample R DPS
0 0.079 0.073
1 0.041 0.040
2 0.069 0.064
3 0.090 0.082
1A 0.071 0.,066
2A 0.105 0.095
3A 0.109 0.099
Moreover, in surface cross-sections of the examined nanocomposites bigger agglomerates of hard domains can be seen (Fig. 2e and f). Most probably, the increase in the degree of phase separation of this nanocomposites group can be attributed to the growth of the chain mobility in the polyurethane allowing the creation of more ordered phases, with respect to the polyurethane without nanosilica. A smaller amount of nanosilica weakened the interactions between hard segments but caused no growth of the chain mobility; it was only after adding bigger amounts of nanofiller (above 1%) that an increase in nanocomposites chain mobility was observed. Surface modification of nanosilica by NH2 groups results in the increase in phase separation (Table 3), the volume WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
385
of hard domain agglomerates increases (Fig. 2d and f), which is the result of reaction of NH2 groups on nanosilica surface and NCO groups in hard segments. This reaction weakens chain mobility and creates favourable conditions for phase separation in this nanocomposite group.
4 Conclusion Within the framework of the work the influence of the quantity and modification of nanosilica on the properties of polyurethanes was assessed; the polyurethanes are intended to be used as short-term implants exposed to abrasion wear. The addition of 0.5wt% of nanosilica caused a decrease in storage modulus and abrasion wear of nanocomposites. Bigger amounts of nanosilica result in a significant growth of storage modulus. The modification of nanosilica surface with NH2 groups adds to the increase in the modulus and to a favourable fall in the abrasion wear of the examined nanocomposites. The joined qualitative and quantitative analysis of AFM pictures of polyurethane structure and its nanocomposites as well as the analysis of phase separation degree, performed on the basis of FTIR spectra, permitted to explain the causes of changes in the properties brought about by the introduction of nanofillers.
Acknowledgements The authors are grateful to the Warsaw University of Technology for financial support of this study. Thanks are also offered to Dr Maria Zielecka from the Institute of Industrial Chemistry for the carrying out of nanosilica modifications.
References [1] [2] [3] [4] [5]
[6]
[7]
C. Hepburn: Polyurethane Elastomers, Elsev. Sci, Publ., London (1992). M. D. Lelah, S. L. Cooper. Polyurethanes in medicine, CRC Press, Boca Raton, 1986 N.M.K. Lambda, K.A. Woodhouse, S.L. Cooper, Polyurethanes in Biomedical Applications, CRC Press, 1998 Janik H., Postępy w badaniach struktur nadcząsteczkowych segmentowych poliuretanów, Polimery, 6, (2009), 419–500 Janik H., Balas A., Budowa chemiczna i właściwości fizyczne usieciowanych poliuretanów segmentowych, Polimery, 3 (2009), 195– 201 Ryszkowska J., Waśniewski B, Pytek A., Zielecka M; Kompozyty poliuretanowe z nanokrzemionką modyfikowaną grupami izocyjanianowymi, Polimery, 10 (2009), 657–666 Opera S.; Synthesis and properties of new polyurethane elastomers: influence of hard segment structure, Polimery, 2 (2010),111–117
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
386 Materials Characterisation V [8] [9] [10]
[11]
[12]
[13] [14] [15] [16]
[17]
[18] [19] [20] [21] [22]
[23] [24]
[25] [26] [27]
B. Pourdeyhimi, Imaging and Image Analysis Applications for Plastics, Plastics Design Library, New York, 1999. J.C. Russ, R.T. Dehoff, Practical Stereology, Plenum Press, New York 1999. J. Ryszkowska, Quantitative image analysis of polyurethane/carbon nanotube composite microstructures, Materials Characterization, 60 (10), 2009, 1127-1132. Hafeman AE, Li B, Yoshii T, Zienkiewicz K, Davidson JM, Guelcher SA. Injectable biodegradable polyurethane scaffolds with release of platelet-derived growth factor for tissue repair and regeneration. Pharm Res 2008;25:2387–99 Polish Committee for Standardization, Rubber—Determination of Abrasion Resistance Using a Rotating Cylindrical Drum Device, PNISO 4649; Warsaw, Poland, 1999. White, J. R.; De, S. K. Rubber Technologist’s Handbook; Rapra Technology: Shrewsbury, England, 2001. Polish Committee for Standardization, Rubber, Vulcanized— Determination of Density, PN-ISO 27811AC1; Warsaw, Poland, 1996. Ryś J. Stereology of Materials. Kraków, Poland: Fotobit Design; 1995. Tien YI, Wei KH. Hydrogen bonding and mechanical properties in segmented montmorillonite/polyurethane nano composites of different hard segment ratios. Polymer 2001; 42:3213–21. Pretsch T, Jakob I, Müller W. Hydrolytic degradation and functional stability of a segmented shape memory poly(ester urethane). Polym Degrad Stabil , 2009;94:61–73. Gogotsi Y., Nanomaterials Handbook, CRC Taylor &Francis, New York, 2006. Geceler K.E., Rosenberg E., Functional Nanomaterials, American Scientific Publishers, North Lewis Way Stevenson Ranch, 2006. Ajayan P.M., Schadler, Braun P.V. Nanocomposites Science and Technology, Wiley-VCH Verlag, Weinheim, 2003. S. Zhang R. Liu, J Jiang, C.heng Yang, M Chen, X.Ya Liu; Progress in Organic Coatings 70 (2011) 1–8 Bliznyuk, V. N.; Tereshchenko, T. A.; Gumenna, M. A.; Tomza Y. P.; Shevchuk, A. V.; Klimenko, N. S.; Shevchenko, V. V. Polymer 2008, 49, 2298 Nanda, A. K.; Wicks, D. A.; Madbouly, S. A.; Otaigbe, J. U. Macromolecules 2006, 39, 7037. Q. Zhang,H. He, K.ai Xi, X. Huang, X. Yu, X. Jia, Synthesis of NPhenylaminomethyl POSS and Its Utilization in Polyurethanes, Macromolecules, 2011, 44 (3), 550–557 Oertel G. Polyurethane handbook. 2nd ed. New York: Hanser; 1993. 7 [Chapter 2]. Nunes RCR, Fonseca JLC, Pereira MR. Polym Test 2000; 19:93–103. Nunes RCR, Pereira RA, Fonseca JLC, Pereira MR. Polym Test 2001; 20:707–12. WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
387
Description methods of the properties of composites from oxybiodegradable foil waste and wood J. Ryszkowska & K. Sałasińska Warsaw University of Technology, Faculty of Materials Science and Engineering, Poland
Abstract As part of this work oxydegradable polymer was fabricated and analyzed; it was made from the waste following a 30 day exposure in Xenotest, simulating a 2-year exposure in atmospheric conditions and undergoing a triple injection process. Similar examination was applied to composites containing 32 wt.% of wood fabricated from this polymer. In the course of the examinations two methods of degradation process evaluation of the materials were verified. It was stated that the method utilizing the relationship between peak fields originating from scissoring vibrations of the (-CH2-) group with a frequency of ca 1463cm-1 ensures more accurate results. The results of other examinations of the two groups of materials permit us to state that the manufacturing of composites with wood constitutes an interesting form of the utilization of oxybiodegradable polymers. Keywords: oxydegradable polyethylene, foil waste, recycling, wood.
1 Introduction According to the data by Plastics Europe, European association of plastics manufacturers, some 230 million tons of plastics were produced in 2009 in the world (55 million tons in Europe). About 50% of the materials constitute polymers intended for the packaging industry, of which 40% are polyolefines: polyethylene (PE-LD, PE-HD, PE-LLD) and polypropylene (PP). These polymers are made from petroleum-based synthetic polymers that do not degrade in a landfill or in a compost-like environment. However, increased use of synthetic packaging films has led to serious ecological problems. Several WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line) doi:10.2495/MC110341
388 Materials Characterisation V approaches to solve the pollution problems caused by polyolefines were developed in the 1970s [1]. One of the solutions was the introduction of prooxidants to the polymers. The pro-oxidants Co, Mn, Cr, Ni, Mo and Fe on Al2O3 or SiO2 support [2–6], cause that polymers become susceptible to environmental effect. Polyolefines prepared in this way are called oxo-biodegradable. Degradation of all polymers follows a sequence in which they are converted into their single oligomeric or monomeric units and later they are utilized as carbon source by the microbes. Lower molecular weight hydrocarbons are more susceptible to attack by microorganisms than the high molecular weight polymer. The degradation of polyolefines with pro-oxidants addition proceeds in a similar way [7–18]. The presence of pro-oxidants, particularly, provided superior functionality and higher degradation rate in PE films. Though their degradation time varies between 18 and several dozen months, they often land up at the rubbish dump after a month of use. Within the framework of the works a selective collection of this type of waste was proposed along with the production of composites with wood, fabricated from them. Such composites can be used in various fields, i.e. as elements for seasonal gardening, auxiliary materials for agriculture, urban greenery, etc. Later on they can serve as valuable energy raw materials. One of the problems connected with the utilization of products made from such composites is the assessment of their exploitation time. The assessment of the usefulness of climatically hazard materials is carried out by Xenotest type of equipment [15]. After the exposure in such equipment various properties of polymer materials are examined including, first of all, mechanical properties and thermal analysis. However, since the availability of Xenotests is rather limited, other methods are sought after. In the current study an attempt has been made to understand the degradation of composites from waste polyolefines with pro-oxidant and wood. The changes in the various physiochemical properties of the polymer were monitored to elucidate the degradation process. As part of work and in order to assess the oxidation process occurring during the degradation of oxydegradable polymers and their composites, methods were verified utilizing spectra obtained with the use of infrared spectroscopy.
2 Experimental 2.1 Materials Production waste, degradable polyethylene HDPE, containing 1 wt.% of prooxidant d2w, from Ecoplastic Poland, (OXY) was a kind of gift from Ecoplastic Poland. Wood fibers – Lignocel C 120 with particle size 70 – 150 μm, from J. Rettenmaier & Söhne GmbH, Germany.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
389
2.2 Sampling Composite samples were fabricated in two stages. In the first stage, with the use of T-45-25-22T-VS single-screw extruder with screw ratio L/D= 29 and with the head for granulation able to simultaneously extrude six 5mm profiles, a granulate was extruded containing 25 wt.% of wood flour. The extrusion process was conducted at the temperatures of 150/160/170/175/175°C (starting from the charging hopper). After cooling the extruded rods were ground using a mill for plastics. In the second stage wood flour was added to the granulate obtained from the mixture in such quantity so that its total content in the composite should reach 32% mass; next, the granulate of composite mixture (or oxybiodegradable polymer) was placed in the bunker of DrBoy 22A injection moulding machine. The regranulate and wood fibers as well as the obtained granulates were then dried before each processing stage in DAC6 dryer at 80°C for one hour. During the injection of samples the cylinder’s temperature, starting from the bunker, amounted to 155, 160, 170, 178 and 180°C, injection pressure 800 · 105 Pa, and clamp pressure of 800 · 105 Pa. Injection time was ca 2 s, clamp time 12 s, cooling time 8 s; the total cycle time equaled ca 25 s. From all types of materials samples A1 were fabricated in conformity with the norm EN ISO 527-2 using a 2-cavity mould cooled with water at 40°C. In order to verify the influence of injection process on the degradation process of oxydegradable polymers and their composites a 3-time injection process was carried out. Samples obtained in the first injection cycle were ground in an industrial mill, dried and injected. This cycle was performed twice. A description of the achieved materials is presented in Table 1. Table 1:
Description of the achieved materials.
Sample 1 2 3 1.32 2.32 3.32 Composites matrix OXY OXY OXY OXY OXY OXY Amount of wood 32 32 32 [wt.%] Process multiplicity 1 2 3 1 2 3 2.3 Characterization Resistance of the materials to accelerated ageing was determined on the basis of analysis of pictures taken with the use of microscopic scanning, Charpy impact tests as well as FTIR analysis of samples after irradiation. The examination was carried out in conformity with PN-EN ISO 4892-1, PN-EN ISO 4892-2 and PNEN ISO 20105-A02:1996. The samples underwent exposure in Xenotest Alpha High Energy equipped with xenon lamp as radiation source. The examination was carried out for 720h with samples being subjected to irradiation of 388,8MJ/m2 (within the scope of 300-400nm) equivalent to a 2-year exposure in natural conditions [19]. The exposure in Xenotest is shown in Table 2.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
390 Materials Characterisation V Table 2:
Ageing test parameters in Xenotest alpha HE.
Parameter Filter Luminous intensity E, W/m2 Work mode Temperature control Temperature in chamber C Rain Relative humidity, % Phase time, min
Phase 1 Phase 2 Xenochrome 300 1503 1503 without without revolution revolution in chamber in chamber 353 303 no yes rain 503 102 18
SEM picture from the surface of examined samples dusted with gold was achieved using an electronic scanning microscope Hitachi S-2600 with accelerating voltage of 10 kV. Infrared spectra of PURs were collected using a FTIR spectrophotometer (Thermo Electron Corporation model Nicolet 6700). Measurements were carried out using attenuated total reflectance (ATR) technique. Each sample was scanned 64 times at a resolution of 4 cm-1 over the frequency range of 4000–400 cm-1. Analysis of FTIR data enabled to determine the carbonyl index. Charpy impact resistance using Resil 5,5 hammer by Ceast, wg PN-EN ISO 1792:2001 was defined for samples with notch, size 70 × 4 × 10 mm (cut from samples and formed via injection). Mechanical properties at static stretching in conformity with PN-EN ISO 5271:1998 and PN-EN ISO 527-2:1998 were examined using the strength machine MTS Q/Test 10. Five 1A samples were analyzed from each type of composite. The samples were stretched at the speed of 10 mm/min. The measurements were recorded automatically using programme TestXpertII. Determined values: strength in the area of plasticity (), strain at break ( ) and Young elasticity modulus (E) of polymers and composites. Absorption after water soaking was determined basing on the change in the mass of 3 randomly selected samples from a given part of the material. The examination was carried out in conformity with the technology specified in the norm PN-EN 317:1999.
3 Results and discussion In the course of conducted examinations the samples from oxybiodegradable polymer and its composite, fabricated via single injection process, were subjected to exposure in Xenotest. In result of the exposure, simulating a 2-year ageing process in natural conditions a degradation of the examined oxybiodegradable polymers took place (Fig. 1). The degradation was visible in the form of white ovals appearing on a sample not directly subjected to UV radiation (Fig. 1a) as well as in the form of cracking on the surface of a sample WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
391
directly subjected to action of the xenon lamp (Fig. 1b). No such visible changes were observed on the surface of composites (Fig. 2). Degradation of polymers and composites results in the decrease in their impact resistance (Table 3).
Figure 1:
SEM pictures of the surface of oxybiodegradable polymer after 30 days ageing in Xenotest: surface not exposed to a direct radiation of xenon lamp (a), subjected to a direct radiation of xenon lamp (b).
Figure 2:
Rys. 2. SEM pictures of the surface of oxybiodegradable polymer with 32 wt.% of wood, not exposed to a direct radiation of xenon lamp (a), after 30 days ageing in Xenotest (b).
Table 3:
The change in impact resistance of oxybiodegradable polymers and composites containing 32 wt.% of wood, caused by 30 days exposure in Xenotest. 1 – sample of oxybiodegradable polymer, 1.32 – sample of its composite, 1D and 1.32D – samples after exposure in Xenotest. Sample U, impact resistance, kJ/m2
1 1.32 1.D 1.32D 43.9 9.7 33.6 9.6
In oxybiodegradable polymer the impact resistance after exposure in Xenotest decreased by ca 23% while that of the composite only by ca 1%. In order to WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
392 Materials Characterisation V assess changes in the structure of polymer and composite, causing the decrease in impact resistance after exposure in Xenotest a spectroscopic analysis (FTIR – ATR) of them was carried out. The achieved spectra of polymer and composite are shown in Fig. 3.
Figure 3:
FTIR-ATR spectra of oxybiodegradable polymer samples before (1) and after exposure in Xenotest (1D) and composites before (1.32) and after exposure in Xenotest (1.32D).
Basing on FTIR –ATR spectra carbonyl index (COI) was calculated by two methods. In the first one, proposed by Reddy and co-authors [16] and Corti et al. [14], carbonyl index was determined as the ratio of absorbance of band 1716 cm-1 resulting from the vibrations of (C=O) carbonyl group and absorbance of band 1468 cm-1 originating from scissoring vibrations of group (-CH2-).
COI
A1716 A1468
(1)
In the second method, proposed by Douminge and co-workers [18] and Stark and co-workers [15] carbonyl index was determined as the ratio of absorbance of band 1716 cm-1, resulting from the vibrations of (C=O) carbonyl group and absorbance of band 2913cm-1, originating from the asymmetric stretching vibrations of group (-CH2-).
COI
A1716 A2913
(2)
The fields of individual bands were determined with the use of OMNIC 7.3 programme. The analysis was carried out after correction of baseline, the spectrum in the analyzed scope was resolved into component bands using the Gaussian curve-fitting.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
Table 4:
393
Calculation results of carbonyl index.
Sample COI calculated acc. to formula (1) COI calculated acc. to formula (2)
1 0.015 0.005
1.32 0.068 0.019
1.D 0.019 0.006
1.32D 0.041 0.012
In the results of the calculation of carbonyl index using the first method it was stated that the index is three times higher than that calculated by the second method according to formula (2) (Table 4). Therefore, it was agreed that the first method will ensure a more accurate analysis of degradation changes in oxybiodegradable polymers and composites with wood, fabricated from them. To verify this statement an analysis was performed of carbonyl index of the polymer and composite containing 32 wt.% of wood nanofiller after multiple processing and using the first method (Table 5). Carbonyl index is used to monitor the progress of oxidation process. An increase in COI was observed in oxybiodegradable polymers and composites, which indicates that degradation process in these materials occurs after each processing cycle. After consecutive stages of processing the speed of COI changes in composite with wood decreases, which means that the introduction of wood slows down the degradation process. Also examined were the strength properties, impact resistance and water absorption of the materials; the results are shown in Table 5. Table 5:
Examination results of oxybiodegradable polymer and composites with 32 wt.% of wood and after multiple injection process.
Sample COI calculated acc. to formula (1) U, impact resistance, kJ/m2 A, humidity content, % E, elasticity modulus, MPa , stress in the area of plasticity, MPa , strain at break, %
1 0.02
1.32 2 0.03 0.09
2.32 0.06
3 3.32 0.23 0.07
38.5 0.02 820 152
8.0 3.48 1740 19
36.7 0.02 210 105
9.0 3,43 410 26
31,2 11.0 0.08 3,12 205 390 80 27
450
4.4
280
4.9
120
5.7
The examination results of strength properties confirm that after consecutive processing cycles the degradation process of composites proceeds more slowly than the degradation process of matrix polymers. Also, after consecutive processing cycles water absorbency of the composites favourably decreases.
4 Conclusion Within the framework of the work a possibility of manufacturing practical materials from oxybiodegradable polymers waste was assessed as well as their composites with wood flour filler. The possibility of utilizing recycled oxybiodegradable polymers requires the assessment of their degradation degree WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
394 Materials Characterisation V prior to processing. To this end the use of carbonyl index is proposed. Two methods of its determination were verified. It was stated that the method utilizing the ratio of peak field originating from scissoring vibrations of group. (-CH2-) ensures more accurate results. This method was used for the assessment of degradation degree of oxybiodegradable polymers and their composites subjected to the exposure in Xenotest, simulating a 2-year exposure period in atmospheric conditions and following a three-time injection process. Apart from degradation itself, selected properties of the materials were also evaluated. In result of the examinations it was stated that each processing course accelerates the degradation process of oxybiodegradable polymers. The introduction of wood affects the degradation process of the matrix. Further processing causes smaller changes in the properties of composites than in polymers. The results show that oxybiodegradable polymer waste, not often utilized so far, may constitute a valuable raw material for manufacturing practical products.
Acknowledgements The study has been financed by the National Research and Development Centre within the framework of the project N R15 0023 06 / 2009, titled: Polymer Composites with Biomass The examinations in Xenotest were carried out by a team headed by Prof. Ph.D. Eng. K. Czaja at the faculty of Chemistry, Opole University, Poland.
References [1] Arnaud R, Dabin P, Lemaire J, Al-Malaika S, Chohan S, Coker M, et al. Photooxidation and biodegradation of commercial photodegradable polyethylenes. Polym Degrad Stab 1994; 46:211-24. [2] Weiland M, Daro D, David C. Biodegradation of thermally oxidized polyethylene. Polym Degrad Stab 1995;48:275-89 [3] Jakubowicz I. Evaluation of degradability of biodegradable polyethylene (PE). Polym Degrad Stab 2003; 80:39-43. [4] Magagula B. , Nhlapo N., Focke W.W, Mn2Al-LDH- and Co2Al-LDHstearate as photodegradants for LDPE film. Polym Degrad Stab 94 (2009) 947–954. [5] Roy P.K., Surekha P., Raman R., Rajagopal C., Investigating the role of metal oxidation state on the degradation behaviour of LDPE, Polym Degrad Stab 94 (2009) 1033–1039. [6] Wiles DM, Scott G. Polyolefins with controlled environmental degradability. Polym Degrad Stab 2006; 91:1581–92. [7] Chiellini E, Corti A, Swift G. Biodegradation of thermally-oxidized, fragmented low-density polyethylenes. Polym Degrad Stab 2003; 81:341– 51.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
395
[8] Hasan F, Shah AA, Hameed A, Ahmed S. Synergistic effect of photo and chemical treatment on the rate of biodegradation of low density polyethylene by Fusarium sp AF4. J Appl Polym Sci 2007; 105:1466–70. [9] Albertsson A-C, Andersson SO, Karlsson S. The mechanism of biodegradation of polyethylene. Polym Degrad Stab 1987; 18:73–87. [10] Albertsson A-C, Karlsson S. Three stages in degradation of polymers – polyethylene as a model substance. J Appl Polym Sci 1988; 35:1289–302. [11] Albertsson A-C, Barenstedt C, Karlsson S. Susceptibility of enhanced environmentally degradable polyethylene to thermal and photo-oxidation. Polym Degrad Stab 1992;37(2):163–71. [12] Sipinen AJ, Rutherford DR. A study of the oxidative degradation of polyolefins. Proc Am Chem Soc 1992;67:185–7. [13] Hakkarainen M, Albertsson A-C. Environmental degradation of polyethylene. Adv Polym Sci 2004; 169:177–99. [14] Corti A., Muniyasamy S., Vitali M., Syed H. Imam S.H., Chiellini E. Oxidation and biodegradation of polyethylene films containing pro-oxidant additives: Synergistic effects of sunlight exposure, thermal aging and fungal biodegradation, Polym Degrad Stab 95(2010) 1106-1116. [15] Stark N.M., Laurent M. Matuana L.M., Surface chemistry changes of weathered HDPE/wood-flour composites studied by XPS and FTIR spectroscopy, Polymer Degradation and Stability 86 (2004) 1–9. [16] Reddy M.M., Deighton M., Gupta R.K., Bhattacharya S.N., Parthasarathy R., Biodegradation of Oxo-Biodegradable Polyethylene, J Appl. Polym Sci, 111 (2009) 1426–1432. [17] Jakubowicz I., Narahmadi N., Petersen H., Evaluation of the rate of abiotic degradation of biodegradable polyethylene in various environments, Polym Degrad Stab 91 (2006) 1556-1562. [18] Douminge L., Mallarino S., Cohendoz S., Feaugas X., Bernard J., Extrinsic fluorescence as a sensitive method for studying photo-degradation of high density polyethylene part I, Current Applied Physics 10 (2010) 1211–1215. [19] Ryszkowska J., Sałasińska K., Kompozyty z folii oksybiodegradowalnej z recyklingu napelniane drewnem, Polimery 55 (2010) 740-747.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
This page intentionally left blank
Materials Characterisation V
397
The effect of slag composition on recycling of “OFHC” through the “ESCM” process S. Ketabchi, F. K. Ahadi, K. Hanaee & S. H. Alhoseini Academic Center of Education, Culture and Research, Tehran, Iran
Abstract This paper reports the results of an investigation into the role of the slag composition in recycling Oxygen Free High Conductivity (OFHC) copper through a modified Electroslag Remelting Process (ESR) melting technique Electroslag Crucible Melting (ESCM). Materials used for the slag were alumina (Al2O3), cryolite (Na3AlF6), sodium fluoride (NaF), and fluorine (CaF2) at different ratios. The results showed that in addition to purity of the slag and graphite used in the electrode and crucible, the percentage of alumina component in the slag composition was the main factor in the attainment of the purity. The best consequence was achieved when the content of alumina was in the range of 23–27%. Furthermore, in ternary compound systems with constant one component, the influence of the weight ratio of the other components was investigated, from a product purity viewpoint. Finally, in the optimized condition, copper with 99.988% purity was achieved, yielding an electrical conductivity of 100.1% IACS. Keywords: OFHC copper, recycling processes, electro slag remelting, ESCM process
1 Introduction Recent developments in electronic industries and energy fields have pushed the property requirements for oxygen free copper to the extreme. When recycling the OFHC copper, the main problem is to reduce oxygen contamination because it not only impairs the conductivity but also the mechanical properties of the material, possibly leading to high scrap losses [1–4]. Oxygen free copper is used in applications where the parts are going to be annealed in a hydrogen containing
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line) doi:10.2495/MC110351
398 Materials Characterisation V atmosphere. The presence of oxygen in either it's elemental state or as copper oxide, leads to the formation of water vapor, which means the brittleness of the material [5]. Two major techniques are used to reduce the oxygen content in copper. The first one involves casting in an inert atmosphere and fluxing the molten copper with an inert gas to prevent the oxygen involvement. The other approach which is a deoxidizing method consists of adding a reductive material to the melt to form selective oxide (instead of copper oxide). The reactive material must be chosen so that its oxide is stable and could not be reduced by hydrogen during annealing. Unfortunately, most of the reductive materials have highly deleterious effects on electrical conductivity if remain in the solid solution of copper. Because of the nature of these oxidants, it is difficult to accurately control the amount of reactive materials. So it's confronted with diminishing of conductivity. In conventional methods of producing OFHC copper, clean cathodes are melted in contact with carbon in an electrical furnace under a protected atmosphere or in vacuum. The protected atmosphere is a mixture of carbon monoxide and nitrogen gases [3, 6–11]. After charging clean cathodes in an induction melting furnace, providing the necessary vacuum or protected environment, the oxide removal agent is added to the completed melt without any intermission. In this technique, the amount of this agent must be calculated as precisely as possible. It must first react with all the oxygen available in the composition and bring it into the slag and secondly, no residue of it remains as impurity in the melt [8]. Researches on the effect of vacuum on the other elements, also shows that Sb, As, and Bi reduces up to 90% and large fraction of S is removed as sulfur dioxides [8, 12]. Nesslage et al. produced OFHC copper with an electrical conductivity of 101% IACS just by adding approximately 30 ppm manganese [13]. Yurko and Peckens used a method to produce oxygen free copper from copper powders. The powders produced by hydrogen reduction from an aqueous ammoniac solution was formed into briquettes by cold compression and then heated in AC confined area at temperature of 1750°F for 1 h under reducing atmosphere of oxygen [14]. It is evident that, OFHC copper is nearly a precious material; therefore, recycling of the shavings produced in manufacturing processes is an interesting object. Because of a high surface to volume ratio of this kind of scraps, melting them in vacuum furnace is not possible and if so, it is not an economic solution. Consequently, they can not be recycled by ordinary techniques. ESR or electroslag remelting is a process used for remelting and refining of special alloys which are used for critical applications in aircraft, thermal and nuclear power plants, defense hardware, etc. The principal set up of an electroslag remelting plant is shown in Figure 1. An electroslag remelting process (ESR) starts when the lower tip of a consumable electrode is immersed into a pool of molten slag. The pre-melted slag possessing electrical conductivity is located on the water-cooled mold base connected to a power supply. The electric current (commonly AC) passing through the slag keeps it at high temperature, which is about 360ºF (200ºC) higher than the melting point of the WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
399
metallic electrode. The electrode tip is heated by the hot slag and starts to melt forming droplets of liquid metal, which disconnect from the electrode and sink through the slag layer.
Figure 1:
Schematic illustration of Electroslag Remelting (ESR) operation.
The ESCM (Electro Slag Crucible Melting) method was first invented by Borodin and co-workers in 1985 with inspiration from the ESR process [6, 16]. The main target was to design a system for remelting the scraps. The results of analysis showed the fact that applying this method results in decreasing the amount of Sn, Pb and S; although the Fe and Zn content may increase due to the contamination of the charged scraps [2, 16]. The produced ingots are free of impurities, voids, shrinkage and gas porosities. In addition, the problem of hydrogen embrittlement due to annealing (at high temperatures about 400°C) is reduced substantially [6, 16]. In ESCM method, the consumable electrode in ESR (made of the charge material) is replaced by a non-consumable graphite bar and a water-cooled copper mold is superseded by a graphite one. Schematic illustration of ESCM process is shown in Figure (2). This collection is set on a water-cooled base plate. Scrap is fed in small pieces after complete melting of the slag. These shavings are melted through passing the slag and form melt pool behind it. Finally, the molten metal covered by slag is cast into a permanent graphite mould [2]. Use of graphitic crucible and electrode makes it possible to reduce the oxygen content of the melt remarkably. In addition, the high deoxidizing capacity of graphite ceases the addition of deoxidizing agent, e.g. Lithium, copper phosphor, etc. which usually decrease the electrical conductivity of copper. Besides, for the reason of the very low solubility of carbon in copper (about 0.0025% in WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
400 Materials Characterisation V
Figure 2:
Schematic illustration of Electroslag Crucible Melting (ESCM) process.
25°C temperature), it can be used without the problem of carbon dissolution [6]. This method has several advantages such as simplicity of the equipment, the lower cost and possibility of recycling scraps of OFHC copper besides nearly uniform mechanical properties such as toughness in all directions [2, 6]. The results of Prasad et al. experiments have shown that while using cryolite, the content of oxygen decreases considerably. However, the harmful impurities such as Si, Fe, and S attract into the melt to a substantial extent so the melt doesn't have sufficient purity in this respect [17]. Therefore, in this research, it would be attempted to reduce the amount of impurity elements especially Fe and Si via altering slag composition. For this purpose, cryolite was replaced by mixtures of sodium fluoride and alumina, two effective components in preventing oxygen pick-up, with different ratios. An adequate fluidity of the slag as well as its ability of providing the desired temperature was considered in determining the slag composition. 1.1 Experimental procedure An AC ESCM system with a capacity of 150 KVA, comprising a graphite electrode of 50 mm diameter and a graphite crucible with internal diameter of 105 mm was used to do the tests. Copper scrap was in the form of pieces with 5 mm diameters and the lengths of 2-3 cm. These particles were dipped in a 10% HCl solution for 20 min to remove the surface oxides and then dried in the air. The slag combination consisted of alumina, fluorine, cryolite and sodium fluoride in various ratios. The compositions are listed in table 1. It can be seen from the table 1 that, various NaF to Al2O3 and CaF2 to Al2O3 ratios were investigated in constant CaF2 and NaF respectively. The effect of slag recycling was also studied by reusing a certain slag in the next melting. 4 kg of copper scrap and 2 kg of slag were used for each experiment. The slag materials were preheated in a resistance furnace up to 600°C for 2.5 h. The process started with an arc between the graphite crucible and the electrode.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
401
Table 1: Chemical composition of the slag used in various testing conditions. Sample No.
CaF2 (%)
Al2O3 (%)
Na3AlF6 (%)
NaF (%)
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
10.5 30 30 30 40 10.5 30 30 30 23 35 40 33 35 35 37
4.5 40 35 30 15 4.5 12.5 40 35 30 17 20 15 27 15 25 23
85 15 85
30 35 40 30 -
in addition to the recycled slag of test No 6
25 -
30 35 40 60 45 45 40 25 40 40
Once the arc was stabilized, the slag was charged into the crucible. The heat required for the process was generated due to Ohmic resistance of the slag pool by the passage of electric current through it. When the slag was completely melted, the copper pieces were gradually added to the molten slag and melted. The molten slag covered the liquid copper and protected it from atmospheric contamination during the process. The slag composition was selected properly, so, passing the charge through, the oxide particles could be entrapped and the clean liquid metal was collected at the bottom of the crucible. During crucible melting, the melt and slag temperatures were measured using a thermometer. The temperature of the melt was maintained constant at about 1500°C by changing the voltage of the power supply and/or changing the distance between the electrode and the crucible bottom. After the copper was completely melted, the power was kept constant at about 28 KV for 7 min in order to complete the chemical reactions. Chemical analysis of the ingots was determined using an analyzer, and the oxygen content was also specified by LECO (ASTM B 170). The electrical conductivity was measured for some of the specimens by Ely chemical plant (DIN 48200, 48201). Tensile tests were performed by a MTS machine according to ASTM E 8M standard (D = 2.5 mm) and microstructural observations were carried out after preparation of the samples, polishing, and etching by a potassium dichromate solution for 10 sec.
2 Results Chemical analysis of the primary OFHC scrap is presented in Table 2. Table 3 shows the effect of the slag composition on the final analysis. The amount of oxygen content and electrical conductivity for some of the samples WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
402 Materials Characterisation V was recorded. In all the samples, higher contents of impurity elements such as Si, Fe and P were observed as compared with the former analysis due to pick up from slag. Table 2:
Chemical analysis of primary OFHC copper (ppm).
Fe S Te As Se Sb Cd Cr Pb Si Bi Ag Zn Ni Co Mn Sn <1 6.5 <1.2 <1 0.5 <1 3.5 <1 3.5 <1 <0.6 8.6 <3 1 <2 0.34 <1
Table 3: Amounts of different elements of impurity (ppm), the purity of Cu and the electrical conductivity for various testing conditions.
Sample No.
Fe
Si
S
P
Al
O
Other
1 2 3 4 5
2900 500 1500 200 360
2500 2500 2800 1700 730
260 110 30 20 20
750 1300 50 790 <1
700 1700 1000 100 360
15 -
6
230
320
30
<1
450
-
7
40
230
10
<1
<1
-
8
330
410
70
<1
700
-
9
290
420
30
<1
450
-
10
200
300
20
<1
100
-
11
200
570
10
<1
40
-
12 13
140 40
150 80
50 20
<1 <1
90 400
47
14
30
-
30
<1
130
42
15 16 17
11 10 32.6
101 78 <1
29.4 27.3 35.9
<1 <1 <1
10 -
33 12 18
Mn=40 Mg=700 Mn=40 Mg=800 Pb=230 Pb=60 Zn=240 Pb=60 Sn=40 Pb=50 Sn=20 Zn=40 Pb=60 Mn=20 Pb=60 Sn=20 Zn=80 Mn=20 Pb=40 Mn=10 Pb=30 -
Cu purity 99.200 99.150 99.390 99.550 99.680
Conductivity
99.850 99.920
99.90
89.6
99.950 99.940
91.7 95.9
99.980
100
99.974 99.985 99.988
96.1 100 100.1
As it is shown in Figure 3, it can be inferred from the comparison of samples 8 to 11 that the addition of NaF to Al2O3 weight ratio at constant CaF2 percent results in a higher purity of the produced copper (Figure 3). Also, comparing samples 15 to 17 shows that increase of CaF2 to Al2O3 weight ratio at constant NaF percent leads to production of higher purity ingot (Figure 4). It was reported that recycling of slag containing alumina resulted in diminishing pick up of oxygen [17]. Comparison of samples 6 and 7 shows that recycling of slag results in lower pick up of Fe, Si, S and Al into the copper ingots and therefore, the purity of the ingot is increased. The reason is attributed to the impurities that WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
403
transferred from the slag into the first melt and then the recycled slag is of lower contamination. Samples 8, 9 and 10 have the same production parameters as samples 2, 3 and 4 respectively, but the higher purity of graphite used for crucible and electrode; that leads to the higher purity of the final product.
Figure 3:
Influence of increase in the NaF to Al2O3 weight ratio in the slag on purity of the copper.
Figure 4:
Influence of increasing the CaF2 to Al2O3 weight ratio in the slag on purity of the copper.
3 Discussion The aim of this research is the assessment of the capability of ESCM process in recycling of precious OFHC scrap with acceptable product purity. Introduction of oxygen into the melt leads to take place the oxide compounds. Since these kinds of inclusions have fragile natures, the presence of such particles in the microstructure, particularly on the grain boundaries results in lessening the WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
404 Materials Characterisation V workability of the ingot. In addition, from the electrical conductivity point of view, it's necessary to avoid entering of any impurity in the melt. Microstructures of the sample no. 17, one having the highest purity in this work, are shown in figure 5. As can be seen, the grains and grain boundaries are thoroughly free of either oxides particles or other kinds of inclusions. Besides, due to settlement of hot graphite crucible over the water-cooled copper base plate, directional heat flow exists during the solidification process. Furthermore, a thin slag skin is formed around the casting inside the mold and hot slag solidifies above the ingot which protects it from atmospheric contamination. Based on this reason, sound ingots without any shrinkage cavity or the other defects could be obtained in "ESCM" process. It was also observed that, one of the most principal factors in ESCM process affecting on the product purity is the purity of the materials with which the molten copper is in contact (electrodes, crucible and slag components). The major impurities to be considered are iron, silicon, sulfur and oxygen. Entrance of iron to the melt was attributed to the reduction of iron oxide by graphite electrode or crucible by Prasad et al. [17]. In this research, it was also observed that iron content decreases in ingots using recycled slag of the other melts (Table 3, melt No. 7).
Figure 5:
Microstructure of sample No. 17.
Also, increasing silicon and sulfur content in the melt can be attributed to reduction of their oxides in slag composition. These compositions are introduced via fluorine and sodium fluoride [17]. Previous researches have mentioned oxygen content reduction using cryolite as a slag component, but the results of this study determined that it's accompanied by silicon, iron and sulfur increase in the melt. As Table 3 represents, the best slag composition (40% sodium fluoride, 23% alumina, 37% fluorine) results in a product with a purity of 99.988% and electrical conductivity of 100.1% IACS.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
405
4 Conclusions 1. 2.
3. 4. 5. 6. 7.
ESCM is a successful and practical process for recycling of OFHC copper scrap. Considering the results shown in figs 2 and 3, this shows that in these experiments the best results were achieved when the content of alumina in the slag is in the range of about 23-27%. Using graphite with high purity in electrode and crucible has a very important role in obtaining higher quality product. In ternary compound systems with constant NaF percent, when the weight ratio of CaF2 to A12O3 increases, the product purity increases too. At ternary compound systems with constant CaF2 percent, with increasing the weight ratio of NaF to A12O3, the product purity increases too. The best combination of the slag which leads to the best result in this research is 40% sodium fluoride, 23% alumina, 37% fluorine. Application of ESCM process in recycling of OFHC copper scraps, results in a copper with 15-33 ppm oxygen, purity of 99.988%, and electrical conductivity in excess of 100.1% IACS.
References [1] Y. Koshiba, T. Masui & N. Iida., "Mitsubishi Materials' High Performance Oxygen Free Copper And High Performance Alloys", 2nd International Conference on Processing Materials Properties, TMS, 2000. [2] R.G. Baligidad, V.V.S. Prasad, G. Balachandran & V. Ramakrishna Rao., “High Purity Copper Through Electro Slag Crucible Melting”, Transactions of Indian institute of metals,Vol.42, No.3, 1989, pp.339-342. [3] D. Janicijevic., “Modern Plant Produces Oxygen-Free Copper”, Metal Progress, Vol.79, 1961, pp.112-138. [4] H.A. Blank., "Fabricating OFHC Copper ", Metal Industry, 4 June 1964, pp.768-770. [5] J. Crane & E. Shapiro., "High Conductivity High Temperature Copper Alloy", US Patent No.3976477, 1974. [6] A.I. Borodin & A.L. Evlevskii, “Production of Copper Plates By Electroslag Mould Casting”, Advances in Special Electro–Metallurgy, Vol.3, No.1, 1987, pp.8-9. [7] B. Allison., "Copper- The Science and Technology of the Metal, Its Alloys and Compounds", Amer. Chem. Soc., Monograph Series, Reinhold Publishing Corp., New York, 1954, p.244. [8] V.K. Gupta, V.N. Madhav Rao & R.V. Tamhankar., Trans. Indian Inst. Met., Vol.25,1972, pp.33-39. [9] Anon., Metal Progressing 84 (1963)124. [10] OFHC Copper, Production and Properties., TR No.125, International Copper Development Council. [11] H. Eckstein., "Working operations and their results for production of oxygen-free Copper", NEUE HUTTE, Vol.3, No.1, 1958, pp.32-36. WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
406 Materials Characterisation V [12] M. Kameda & A. Yazawa., "Refining of crude copper by Vacuum melting", Tohoku Daigaku Senko Seiren ken kyusho Tho, Vol.19, No.1, 1963, pp.5768. [13] D.J. Nesslage, L.S. Yu & M. F. Shaw., "Oxygen-Free Copper Product and Process", US Patent No.4059437, 1977. [14] W.J. Yurko & D.K. Peckens., "Method of Producing Oxygen-Free High Conductivity Copper", US Patent3No.298070, 1965. [15] A. I. Borodin, B.I. Medovar & B.B. Fedorovskii., "Centrifugal Electroslag Casting of billets from Copper Scrap", E.O. Paton Welding Institute, Kiev, Advances in Special Electro- Metallurgy, Vol.1, 1985, pp.23-25. [16] V.V.S. Prasad, A. Sambasiva Rao, U. Prakash, V. Ramakrishna Rao, P. Krishna Rao & K.M. Gupt., “Alumina Addition to Fluoride Slags for Recycling of Low Oxygen High Conductivity Copper Scrap Through Electroslag Crucible Melting”, ISIJ International, Vol.38, No.12, 1998, pp.1387-1398. [17] V.V.S. Prasad, V. Ramakrishna Rao, U. Prakash, P. Krishna & K.M. Gupt., “Electro Slag Crucible Melting for Recycling of Low Oxygen High Conductivity Copper Scrap”, ISIJ International, Vol.36, No.9, 1996, pp.1113-1118.
WIT Transactions on Engineering Sciences, Vol 72, © 2011 WIT Press www.witpress.com, ISSN 1743-3533 (on-line)
Materials Characterisation V
407
Author Index Abud I. ....................................... 77 Ahadi F. K. .............................. 397 Alhoseini S. H.......................... 397 Allison P. G. ............................ 287 Al-Qureshi H. A....................... 165 Alston S. M. ............................. 351 Antorrena G. ............................ 187 Athanasopoulou A. .......... 315, 327 Awietjan S. F. .......................... 263 Ayçenk S.................................. 175 Bae G. H. ................................. 213 Barbosa C. ................................. 77 Bestley C. C. N. ....................... 351 Bigdeli Karimi M. ...................... 85 Bobrov M. .................................. 67 Boczkowska A. ........................ 263 Bouzid A.-H............................. 339 Brailovski V. ............................ 339 Brown S. G. R.......................... 351 Bruneton E. .............................. 155 Caminha I. ................................. 77 Černý R. ................... 143, 199, 363 Chamis C. C. ............................ 109 Chandler M. Q. ........................ 287 Chicharro J. M. ........................ 275 Christou P. ............................... 131 Chupin S. ................................. 155 Colin J. ..................................... 307 Cummins T. K. ........................ 287 Cvahte P. .................................. 239 Dabek E. .................................... 67 Daniele V. .................................. 55 Darque-Ceretti E. ....................... 15 Dasan B.................................... 251 Dony A. ................................... 307 Dykman M. ................................ 27 Fiala L. ..................................... 143 Freire M. S. .............................. 187
Gabet O...................................... 27 Garland N. ................................. 95 González-Álvarez J.................. 187 Guagliardo P. ............................... 3 Gueit E. ...................................... 15 Hanaee K. ................................ 397 Herranz G. ............................... 275 Horgnies M. ......................... 15, 27 Hotza D.................................... 165 Huh H. ..................................... 213 Janssen R. ................................ 165 Karakoç C. ............................... 175 Karpuzova E. ............................. 67 Keide H...................................... 77 Kelekanjeri V. S. K. G. ............ 251 Ketabchi S. .............................. 397 Khan Z. ...................................... 95 Klemm A. J. ................................. 3 Kočí V. .................................... 363 Koksal H. O. ............................ 175 Kollaros G. ...................... 315, 327 Kugler G. ................................. 239 Leite E. R. .................................. 41 Lombard S. ................................ 27 Mastromatteo F. ....................... 251 Michael A. ............................... 131 Mihulka J. ................................ 199 Montoro L. A. ............................ 41 Morales A. L. ........................... 275 Moser R. D. ............................. 287 Nascimento J. L. ........................ 77 Nieto A. J. ................................ 275 Northwood D. O. ....................... 85 Ould-Brahim N. ....................... 339 Ouni A. E. ................................ 307
408 Materials Characterisation V Paimushin V. N........................ 227 Pavlík Z............................ 143, 199 Pavlíková M. .................... 143, 199 Perus I. ..................................... 239 Pintado P. ................................. 275 Plaksin O.................................... 67 Popova G. .................................. 67
Stoilov V. ................................... 85 Stroppa D. G. ............................. 41
Ramirez A. J. ............................. 41 Ríos R. ..................................... 187 Rochais D. ............................... 155 Rodríguez G. P......................... 275 Roesler C. R. M. ........................ 77 Rushing T. S. ........................... 287 Ryszkowska J. ................. 377, 387
Vantsyan M. .............................. 67 Vázquez G. .............................. 187 Vejmelková E. ......................... 143 Vishwanath T........................... 251 Vivet N. ................................... 155 Výborný J. ............................... 363
Saeed A. ..................................... 95 Sakurai H. ................................ 119 Sałasińska K. ........................... 387 Samarin S. N. ............................... 3 Silva J. G. P. ............................ 165 Smith R. ..................................... 95 Sondhi S. K. ............................. 251 Song Y. P. ................................ 299
Taglieri G................................... 55 Tercelj M. ................................ 239 Tintillier P. ................................. 15 Turgay T. ................................. 175
Wang H. L. .............................. 299 Waśniewski B. ......................... 377 Williams B. A. ......................... 287 Williams J. F. ............................... 3 Willieme P. ................................ 27 Zakirov I. I. .............................. 227 Zakirov I. M............................. 227 Žumár J. ................................... 199
This page intentionally left blank
...for scientists by scientists
Computational Methods and Experimental Measurements XIV Edited by: C.A. BREBBIA, Wessex Institute of Technology, UK and G.M. CARLOMAGNO, University of Naples, Italy
As computer models become more reliable and able to represent more realistic problems, detailed data is leading to the development of appropriate new types of experiments. Experimental measurements need to be conditioned to the requirements of the computational models. Hence it is important that scientists working on experiments communicate with researchers developing computer codes, as well as those carrying out measurements on prototypes. The orderly and progressive concurrent development of all these fields is essential for the progress of engineering sciences. This book resulted from the latest in a series of successful meetings that provide a unique forum for the review of the latest work on the interaction between computational methods and experimental measurements. The topics include: Computational and Experimental Methods; Experimental and Computational Analysis; Direct, Indirect and In-Situ Measurements; Detection and Signal Processing; Data Processing; Fluid Flow; Heat Transfer and Thermal Processes; Material Characterization; Structural and Stress Analysis; Industrial Applications; Forest Fires. WIT Transactions on Modelling and Simulation, Vol 48 ISBN: 978-1-84564-187-0 eISBN: 978-1-84564-364-5 Published 2009 / 672pp / £255.00
WIT Press is a major publisher of engineering research. The company prides itself on producing books by leading researchers and scientists at the cutting edge of their specialities, thus enabling readers to remain at the forefront of scientific developments. Our list presently includes monographs, edited volumes, books on disk, and software in areas such as: Acoustics, Advanced Computing, Architecture and Structures, Biomedicine, Boundary Elements, Earthquake Engineering, Environmental Engineering, Fluid Mechanics, Fracture Mechanics, Heat Transfer, Marine and Offshore Engineering and Transport Engineering.
...for scientists by scientists
Elastic and Elastoplastic Contact Analysis Using Boundary Elements and Mathematical Programming A. FARAJI, University of Tehran, Iran
This book presents a general elastic and elastoplastic analysis method for the treatment of two- and three-dimensional contact problems between two deformable bodies undergoing small displacements with and without friction. The author’s approach uses the Boundary Element Method (BEM) and Mathematical Programming (MP). This is applied to the contacting bodies pressed together by a normal force which is sufficient to expand the size of the contact area, mainly through plastic deformation, acted on subsequently by a tangential force less than that necessary to cause overall sliding. The formulated method described in this book is straightforward and applicable to practical engineering problems. Series: Topics in Engineering, Vol 45 ISBN: 1-85312-733-7 Published 2005 / 144pp / £55.00
Applied Numerical Analysis M. RAHMAN, Dalhousie University, Canada
This book presents a clear and well-organised treatment of the concepts behind the development of mathematics and numerical techniques. The central topic is numerical methods and the calculus of variations to physical problems. Based on the author’s course taught at many universities around the world, the text is primarily intended for undergraduates in electrical, mechanical, chemical and civil engineering, physics, applied mathematics and computer science. Many sections are also directly relevant to graduate students in the mathematical and physical sciences. More than 100 solved problems and approximately 120 exercises are also featured. ISBN: 1-85312-891-0 Published 2005 / 408pp+CD / £165.00
...for scientists by scientists
Development of Petroleum Resources with Emphasis on Offshore Fields O.T. GUDMESTAD, University of Stavanger, Norway, A.B. ZOLOTUKHIN, Gubkin Russian State University of Oil and Gas, Russia and E.T. JARLSBY, Stavanger, Norway
There are large challenges facing the oil and gas industry, e.g. increasing difficulty to find new oil and gas accumulations, a requirement for a clean environment and reductions in CO2 emissions being voiced by public and governments, requirement for long distance transport of natural gas from the producers to the consumers and rapid changes in the price of hydrocarbons on the world market. The decision makers need a wide background for their decisions, and all aspects of the value chain are important in order to identify the optimal field development solution. The objective of this book is to treat in a consistent manner the key aspects related to development of petroleum resources in order to allow an interdisciplinary view of the topics involved in development projects and in order to identify the most optimal development solution. The book is organized to allow the reader to review the different topics individually so the book can serve as a reference source for those interested in topics not familiar to themselves. This book is written as a reference book for training programs, especially courses that are attended by students from many parts of the world, including areas that are in early stages of hydrocarbon development as well as Norway with its established offshore petroleum provinces. Although intended to be useful in a wide range of contexts world wide, it has particular emphasis on offshore developments, and is based to a large extent on insights from Norwegian offshore projects. ISBN: 978-1-84564-478-9 eISBN: 978-1-84564-479-6 Published 2010 / 288pp / £98.00
...for scientists by scientists
Surface Effects and Contact Mechanics IX Computational Methods and Experiments Edited by: J.T.M. de HOSSON, University of Groningen, The Netherlands and C.A. BREBBIA, Wessex Institute of Technology, UK
The importance of contact mechanics and surface effects in modern engineering and their interaction led the organisers to reconvene this 9th International Conference, which covers two related topics: Computational Methods and Experiments in Contact Mechanics and Surface Effects. Nowadays the importance of contact and surface problems in many technological fields is well understood. They are complex and inherently nonlinear due to the moving boundary and the different properties of materials, particularly along the contact surfaces. Structural components fail by wear, corrosion, high cycle fatigue etc, that is to say they are affected by surface conditions. The use of surface treatments can reduce the cost of components and extend the life of the elements for surfaces undergoing contact, a problem that is specifically addressed by the Conference. This book contains papers that discuss the following topics: Surface Treatments; Thick Coatings; Thin Coatings; Surface Problems in Contact Mechanics; Indentation and Hardness; Fatigue; Numerical Analysis; Applications and Case Studies. WIT Transactions on Engineering Sciences, Vol 62 ISBN: 978-1-84564-186-3 eISBN: 978-1-84564-363-8 Published 2009 / 288pp / £109.00
WITPress Ashurst Lodge, Ashurst, Southampton, SO40 7AA, UK. Tel: 44 (0) 238 029 3223 Fax: 44 (0) 238 029 2853 E-Mail: [email protected]