Combustion - Synthesisof Advanced Materials

CHEMISTRY RESEARCH AND APPLICATIONS SERIES
COMBUSTION SYNTHESIS OF ADVANCED MATERIALS
No part of this digital document may be reproduced, stored in a retrieval system or transmitted in any form or by any means. The publisher has taken reasonable care in the preparation of this digital document, but makes no expressed or implied warranty of any kind and assumes no responsibility for any errors or omissions. No liability is assumed for incidental or consequential damages in connection with or arising out of information contained herein. This digital document is sold with the clear understanding that the publisher is not engaged in rendering legal, medical or any other professional services.

Applied Electrochemistry Vijay G. Singh (Editor) 2009. ISBN: 978-1-60876-208-8 Handbook on Mass Spectrometry: Instrumentation, Data and Analysis, and Applications J. K. Lang (Editor) 2009. ISBN: 978-1-60741-580-0 Handbook of Inorganic Chemistry Research Desiree A. Morrison (Editor) 2010. ISBN: 978-1-61668-010-7 Handbook of Inorganic Chemistry Research Desiree A. Morrison (Editor) 2010. ISBN: 978-1-61668-712-0 (Online Book) Solid State Electrochemistry Thomas G. Willard (Editor) 2010. ISBN: 978-1-60876-429-7 Mathematical Chemistry W. I. Hong (Editor) 2010. ISBN: 978-1-60876-894-3 Mathematical Chemistry W. I. Hong (Editor) 2010. ISBN: 978-1-61668-440-2 (Online Book) Physical Organic Chemistry: New Developments Karl T. Burley (Editor) 2010. ISBN: 978-1-61668-435-8 Physical Organic Chemistry: New Developments Karl T. Burley (Editor) 2010. ISBN: 978-1-61668-469-3 (Online Book) Chemical Sensors: Properties, Performance and Applications Ronald V. Harrison (Editor) 2010. ISBN: 978-1-60741-897-9 Macrocyclic Chemistry: New Research Developments Dniel W. Fitzpatrick and Henry J. Ulrich (Editors) 2010. ISBN: 978-1-60876-896-7 Electrolysis: Theory, Types and Applications Shing Kuai and Ji Meng (Editors) 2010. ISBN: 978-1-60876-619-2 Energetic Materials: Chemistry, Hazards and Environmental Aspects Jake R. Howell and Timothy E. Fletcher (Editors) 2010. ISBN: 978-1-60876-267-5
Chemical Crystallography Bryan L. Connelly (Editor) 2010. ISBN: 978-1-60876-281-1 Chemical Crystallography Bryan L. Connelly (Editor) 2010. ISBN: 978-1-61668-513-3 (Online Book) Heterocyclic Compounds: Synthesis, Properties and Applications Kristian Nylund and Peder Johansson (Editors) 2010. ISBN: 978-1-60876-368-9 Influence of the Solvents on Some Radical Reactions Zaikov Gennady E. Roman G. Makitra, Galina G. Midyana and Lyubov.I Bazylyak (Editors) 2010. ISBN: 978-1-60876-635-2 Dzhemilev Reaction in Organic and Organometallic Synthesis Vladimir A.D'yakonov 2010. ISBN: 978-1-60876-683-3 Analytical Chemistry of Cadmium: Sample Pre-Treatment and Determination Methods Antonio Moreda-Pieiro and Jorge Moreda-Pieiro 2010. ISBN: 978-1-60876-808-0
Binary Aqueous and CO2 Containing Mixtures and the Krichevskii Parameter Aziz I. Abdulagatov and Ilmutdin M. Abdulagatov 2010. ISBN: 978-1-60876-990-2 Advances in Adsorption Technology Bidyut Baran Saha.(Editor) 2010. ISBN: 978-1-60876-833-2 Electrochemical Oxidation and Corrosion of Metals Elena P. Grishina and Andrew V. Noskov 2010. ISBN: 978-1-61668-329-0 Electrochemical Oxidation and Corrosion of Metals Elena P. Grishina and Andrew V. Noskov 2010. ISBN: 978-1-61668-329-0 (Online Book) Modification and Preparation of Membrane in Supercritical Carbon Dioxide Guang-Ming Qiu and Rui Tian 2010. ISBN: 978-1-60876-905-6 Thermostable Polycyanurates: Synthesis, Modification, Structure and Properties Alexander Fainleib (Editor) 2010. ISBN: 978-1-60876-907-0
Combustion Synthesis of Advanced Materials B. B. Khina 2010. ISBN: 978-1-60876-977-3 Structure and Properties of Particulate-Filled Polymer Composites:The Fractal Analysis G. V. Kozlov 2010. ISBN: 978-1-60876-999-5 Information Origins of the Chemical Bond Roman F. Nalewajski 2010. ISBN: 978-1-61668-305-4 Cyclic ]-Keto Esters: Synthesis and Reactions M.A. Metwally and E. G. Sadek 2010. ISBN: 978-1-61668-282-8 Wet Electrochemical Detection of Organic Impurities F. Manea 2010. ISBN: 978-1-61668-661-1 Wet Electrochemical Detection of Organic Impurities F. Manea, S. Picken and J. Schoonman 2010. ISBN: 978-1-61668-491-4 (Online Book)
Quantum Frontiers of Atoms and Molecules Mihai V. Putz (Editor) 2010. ISBN: 978-1-61668-158-6 Molecular Symmetry and Fuzzy Symmetry Xuezhuang Zhao 2010. ISBN: 978-1-61668-528-7 Molecular Symmetry and Fuzzy Symmetry Xuezhuang Zhao 2010. ISBN: 978-1-61668-375-7 (Online Book) Molybdenum and Tungsten Cofactor Model Chemistry Carola Schulzke, Prinson P. Samuel 2010. ISBN: 978-1-61668-750-2 Molybdenum and Tungsten Cofactor Model Chemistry Carola Schulzke, Prinson P. Samuel 2010. ISBN: 978-1-61668-750-2 (Online Book) Rock Chemistry Basilio Macas and Fidel Guajardo (Editors) 2010. ISBN: 978-1-60876-563-8
COMBUSTION SYNTHESIS OF ADVANCED MATERIALS
B. B. KHINA
Nova Science Publishers, Inc.

New York
Copyright 2010 by Nova Science Publishers, Inc.

All rights reserved. No part of this book may be reproduced, stored in a retrieval system or transmitted in any form or by any means: electronic, electrostatic, magnetic, tape, mechanical photocopying, recording or otherwise without the written permission of the Publisher. For permission to use material from this book please contact us: Telephone 631-231-7269; Fax 631-231-8175 Web Site: http://www.novapublishers.com NOTICE TO THE READER The Publisher has taken reasonable care in the preparation of this book, but makes no expressed or implied warranty of any kind and assumes no responsibility for any errors or omissions. No liability is assumed for incidental or consequential damages in connection with or arising out of information contained in this book. The Publisher shall not be liable for any special, consequential, or exemplary damages resulting, in whole or in part, from the readers use of, or reliance upon, this material. Independent verification should be sought for any data, advice or recommendations contained in this book. In addition, no responsibility is assumed by the publisher for any injury and/or damage to persons or property arising from any methods, products, instructions, ideas or otherwise contained in this publication. This publication is designed to provide accurate and authoritative information with regard to the subject matter covered herein. It is sold with the clear understanding that the Publisher is not engaged in rendering legal or any other professional services. If legal or any other expert assistance is required, the services of a competent person should be sought. FROM A DECLARATION OF PARTICIPANTS JOINTLY ADOPTED BY A COMMITTEE OF THE AMERICAN BAR ASSOCIATION AND A COMMITTEE OF PUBLISHERS. LIBRARY OF CONGRESS CATALOGING-IN-PUBLICATION DATA Khina, B. B. (Boris B.) Combustion synthesis of advanced materials / author, B.B. Khina. p. cm. Includes bibliographical references and index. ISBN 978-1-61324-254-4 (eBook) 1. Self-propagating high-temperature synthesis. 2. Refractory materials--Heat treatment. 3. Refractory materials--Mathematical models. I. Title. TP363.K46 2010 620.1'43--dc22 2009052731
Published by Nova Science Publishers, Inc. New York
DEDICATION
To the memory of Professor Zinoviy P. Shulman (1924-2007) and Professor Leonid G. Voroshnin (1936-2006) who had taught me the scientific meaning of old Russian proverb, trust, but verify. The important thing in science is not so much to obtain new facts as to discover new ways of thinking about them. Sir William Bragg
CONTENTS
Preface Chapter 1 Chapter 2 Chapter 3 Chapter 4 References Index Advances and Challenges in Modeling Combustion Synthesis Modeling Diffusion-Controlled Formation of TiC in the Conditions of CS Modeling Interaction Kinetics in the CS of Nickel Monoaluminide Analysis of the Effect of Mechanical Activation on Combustion Synthesis xi 1 13 39 75 93 105
PREFACE
Self-propagating high-temperature synthesis (SHS), or combustion synthesis (CS) is a phenomenon of wave-like localization of chemical reactions in condensed media which permits efficiently synthesizing a wide range of refractory compounds (carbides, borides, intermetallics, etc.) and advanced composite materials. CS, where complex heterogeneous reactions proceed in substantially non-isothermal conditions, brings about fine-grained structure and novel properties of the target products and is characterized by fast accomplishment interaction, within ~0.1-1 s, whereas traditional furnace synthesis of the same compounds in close-to-isothermal conditions may take several hours for the same particle size and close final temperature. Uncommon, nonequilibrium phase formation routes inherent of SHS, which have been revealed experimentally, are the main subject of this book. The main goal of this book is to describe basic approaches to modeling nonisothermal interaction kinetics during CS of advanced materials and reveal the existing controversies and apparent contradictions between different theories, on one hand, and between theory and experimental data, on the other hand, and to develop criteria for a transition from traditional solid-state diffusion-controlled phase formation kinetics (a slow, quasi-equilibrium interaction pathway) to non-equilibrium, fast dissolution-precipitation route. Features: analysis of the physicochemical background of modeling approaches to CS; modeling of phase formation kinetics for two typical SHS reactions, Ti+CTiC (CS of an interstitial compound) and Ni+AlNiAl (CS of an
xii
B. B. Khina intermetallic compound), in strongly non-isothermal conditions using the diffusion approach and experimentally known values of the diffusion parameters; novel criteria for the changeover of interaction routes in these systems and phase-formation mechanism maps; analysis of the physicochemical mechanism of the experimentally known strong influence of preliminary mechanical activation of solid reactant particles on SHS in metal-based systems.
It is anticipated that the book will serve the scientists, engineers, graduate and post-graduate students in Solid-State Physics and Chemistry, Heterogeneous Combustion, Materials Science and related areas, who are involved in the research and development of CS-related methods for the synthesis of novel advanced materials.
Chapter 1
ADVANCES AND CHALLENGES IN MODELING COMBUSTION SYNTHESIS

1.1. APPROACHES TO MODELING NON-ISOTHERMAL INTERACTION KINETICS DURING CS
Combustion synthesis (CS), or self-propagating high-temperature synthesis (SHS), also known as solid flame, is a versatile, cost and energy efficient method for producing refractory compounds (carbides, borides, nitrides, intermetallics, complex oxides etc.) and advanced composite materials possessing fine-grain structure and superior properties. Extensive research in this area was initiated by A.G.Merzhanov in Chernogolovka, Moscow district, Russia, in mid 1960es [1,2], who is internationally recognized as a pioneer of SHS. The advantages of CS include short processing time, low energy consumption, high product purity due to volatilization of impurities, and unique structure and properties of the final products. Besides, CS can be combined with pressing, extrusion, casting and other processes to produce near-net-shape articles [3-10]. Despite vast literature available in this area, CS is still a subject of extensive experimental and theoretical investigation. Combustion synthesis can be carried out in the wave propagation mode, or true SHS, and in the thermal explosion (TE) mode. In the former case, a compact reactive powder mixture is ignited at one end to initiate an exothermic reaction which propagates through the specimen as a combustion wave leaving behind a hot final product [3-10]. In the latter case, a pellet is heated up at a prescribed rate (typically 1-100 K/s) until at a certain temperature called the ignition point, Tign, an exothermal reaction becomes self-sustaining and the temperature rises to its final value, TCS, almost uniformly throughout the sample.
B. B. Khina
Typically, the value of Tign is close to the melting point of a lower-melting reactant or to the eutectic temperature. Examples of CS products are listed in Table 1.1, and characteristics of SHS reactions in certain systems are presented in Table 1.2. Table 1.1. Examples of compounds and materials produced by combustion synthesis [3-14]
Type of material Borides Compounds and adiabatic combustion temperature, K (in brackets) TiB2 (3190), TiB (3350), ZrB2 (3310), HfB2 (3320), VB2 (2670), VB (2520), NbB2 (2400), NbB, TaB2 (2700), TaB, CrB2 (2470), CrB, MoB2 (1500), MoB (1800), WB (1700), LaB6 (2800) Carbides TiC (3210), ZrC (3400), HfC (3900), VC (2400), Nb2C (2600), NbC (2800), Ta2C (2600), TaC (2700), SiC (1800), WC, B4C, Cr3C2, Cr7C3, Mo2C, Al4C3 Aluminides Ni3Al, NiAl, Ni2Al3, TiAl, CoAl, Nb3Al, Cu3Al, CuAl, FeAl Silicides Ti5Si3 (2500), TiSi (2000), TiSi2 (1800), Zr5Si3 (2800), ZrSi (2700), ZrSi2 (2100), WSi, Cr5Si3 (1700), CrSi2 (1800), Nb5Si3 (3340), NbSi2 (1900), MoSi2 (1900), V5Si3 (2260), TaSi2 (1800) Intermetallics NbGe, TiCo, NiTi Sulfides and selenides MgS, MnS (3000), MoS2 (2900), WS2, TiSe2, NbSe2, TaSe2, MoSe2, WSe2 Hydrides TiH2, ZrH2, NbH2, CsH2 Nitrides TiN (4900), ZrN (4900), VN (3500), HfN, Nb2N (2670), NbN (3500), Ta2N (3000), TaN (3360), Mg3N2 (2900), Si3N4 (4300), BN (3700), AlN (2900) Carbonitrides TiC-TiN, NbC-NbN, TaC-TaN, ZrC-ZrN Complex oxides Aluminates (YAlO2, MgAl2O4), niobates (NaNbO3, BaNb2O6, LiNbO3), garnets (Y3Al5O12, Y3Fe5O12), ferrites (CoFe2O4, BaFe2O4, Li2Fe2O4), titanates (BaTiO3, PbTiO3), molybdates (BiMoO6, PbMoO4), high-temperature superconductors (YBa2Cu3O7-x, LaBa2Cu3O7-x, Bi-Sr-Ca-Cu-O) Ternary solid solutions TiB2-MoB2, TiB2-CrB2, ZrB2-CrB2, TiC-WC, TiN-ZrN, MoS2based on refractory NbS2, WS2-NbS2 compounds MAX phases Ti2AlC, Ti3AlC2, Ti3SiC2 Cermets TiC-Ni, TiC-Cr, TiC-Co, TiC-Ni-Cr, TiC-Ni-Mo, TiC-Fe-Cr, TiCCr3C2-Ni, TiC-Cr3C2-Ni-Cr, Cr3C2-Ni-Mo, TiB-Ti, WC-Co, TiCTiN-NiAl-Mo2C-Cr Composites and TiC-TiB2, TiB2-Al2O3, TiC-Al2O3 (2300), TiN-Al2O3, B4C-Al2O3, functionally-graded MoSi2-Al2O3 (3300), MoB-Al2O3 (4000), Cr3C2-Al2O3, 6VN5Al2O3 (4800), ZrO2-Al2O3-2Nb, AlN-BN, AlN-SiC, AlN-TiB2, materials Si3N4-TiN-SiC, sialons (SiAlOxNy)
Advances and Challenges in Modeling Combustion Synthesis Table 1.2. Features of combustion synthesis waves for certain typical reactions [3-18]
Type of interaction Reaction Experimental combustion temperature, C 2500 Combustion wave velocity, cm/s 3-4
Solid-solid (formation of a carbide) via a transient liquid phase (melting of a metallic reactant) [17,18] Solid-solid (formation of a complex oxide) via a transient liquid phase with participation of an oxidizing gas Solid-gas with or without melting of a metallic reactant [13] Solid-solid (formation of a carbide) via intermediate gas-transport reactions [12] Liquid-liquid in organic systems with the formation of a solid product [15]
Ti (solidliquid) + C (solid) TiC (solid) 3Cu (solidliquid) + 2BaO2 (solid) + (1/2)Y2O3 (solid) + O2 (gas) YBa2Cu3O7-x (solid) Ti (solidliquid) + (1/2)N2 (gas) TiN (solid) Ta (solid) + C (solid) TaC (solid)
1000
0.2-0.5
1600-2000
0.1-0.2
2600
0.5-2
C4H10N2 (liquid, piperazine) + C3H4O4 (liquid, malonic acid) C7H14N2O4 (solid, salt)
155
0.06-0.15
The unique features of the obtained products, e.g., high purity, small and uniform grain size, etc., are ascribed to extreme conditions inherent in CS, which may bring about unusual reaction routes: (i) high temperature, up to 3500 C, (ii) a high rate of self-heating, up to 106 K/s, (iii) steep temperature gradient in SHS waves, up to 105 K/cm, (iv) rapid cooling after synthesis, up to 100 K/s, and (v) fast accomplishment of conversion, from about 1 s to the maximum of 10 s [3-6]. It should be noted that traditional furnace synthesis of refractory compounds requires a much longer time, ~1-10 h, for the same initial composition, particle size and close final temperature. It has been demonstrated experimentally [16-23] that in many systems phase and structure formation during CS proceeds via uncommon interaction mechanisms from the point of view of the classical Physical Metallurgy [24,25]. Modeling and simulation traditionally play in important part in the development of CS and CS-related technologies (see reviews [3-5,11,26-29] and
B. B. Khina
references cited therein). An adequate mathematical model is supposed to describe both heat transfer in a heterogeneous reactive medium and the interaction kinetics, which is responsible for heat release during CS. In modeling CS, a quasi-homogeneous, or continual model [30,31], which is based on classical combustion theory, is widely used. Heat transfer, which is considered on the volume-averaged basis, and the reaction rate in a sample are described as follows: cp T/t = (T) + Q /t /t = (1)n exp(m) k exp(E/RT), (1.1) (1.2)
where T is temperature, is density, cp is heat capacity, is thermal conductivity, Q is the heat release of exothermal reactions, is the degree of chemical conversion (from 0 in the unreacted state to 1 for complete conversion), R=8.314 J mol1K1 is the universal gas constant, n (the reaction order), k (preexponential factor) and m are formal parameters and E is the activation energy; term Q/t denotes the heat release rate. The thermal structure of a combustion wave according to Zeldovich and Frank-Kamenetskiy [32] is shown schematically in Figure 1.1. Typically, three zones are distinguished: (i) the preheating zone where almost no reaction occurs and the main processes are heat and mass transfer accompanied with evaporation of volatile impurities; in Russian literature it is often termed as the Michelson zone after V.A.Michelson (1860-1927) who described the temperature profile ahead of the moving combustion front [32], (ii) zone of thermal reaction where the conversion degree sharply increases and the heat release rate reaches its maximum and starts decreasing while the temperature almost reaches the adiabatic value, and (iii) the after-burn, or post-reaction zone where the interaction terminates. The latter zone is characterized by a slow increase in both conversion degree and temperature, which finally attain their maximal values =1 and T=Tad, and the heat release rate, Q/t, falls down to zero. The temperature of the reaction front, Tf, corresponds to the onset of fast thermal reaction. In regard to combustion synthesis of materials, it is believed that complex heterogeneous reactions, which may proceed via uncommon (fast) mechanisms and are responsible for major heat release, occur in the thermal reaction zone while the after-burn zone, where the heat release rate is minor, is dominated by the processes bringing about the formation of final structure of the product, such as Ostwald ripening, recrystallization etc.
Advances and Challenges in Modeling Combustion Synthesis
Figure 1.1. Schematic of the thermal structure of a combustion wave.
The approach formulated in Eqs. (1.1) and (1.2) permitted modeling dynamic regimes of SHS, e.g., oscillating [30] and spin combustion [33,34]. It was also used for studying the effect of intrinsic stochasticity of heterogeneous reactions, which can be attributed to a difference in the surface morphology, impurity content and hence reactivity of solid reactant particles, on the dynamic behavior of a solid flame for a one-stage [35] and multi-stage reaction [36] employing the cellular automata method. It should be outlined that this model is not linked to any process-specific phase formation mechanism and hence is referred to as a formal one. When applying this approach to modeling CS in a particular system, the value of the most important model parameter, viz. activation energy E, is supposed to correspond to the apparent activation energy of the CS as a whole. The latter is determined from experimental graphs the combustion wave velocity vs. temperature plotted in the Arrhenius form, and in its physical meaning corresponds to a real rate-limiting stage of phase formation during CS, which may be different in different temperature ranges. For example, below the melting temperature, Tm, of a metallic reactant E always refers to solid-state diffusion in the product while at T>Tm it can refer to processes in the melt (diffusion or crystallization) [37]. This method for choosing the E value was used when studying numerically the conditions of arresting a high-temperature state of substances in the SHS wave by fast cooling for the cases of a one-stage [38] and two-stage exothermal reaction [39].
B. B. Khina
In recent papers [40,41], this formal model [see Eqs. (1.1) and (1.2)] was employed for studying the SHS of a NiTi shape memory alloy. The activation energy used in calculations was E=113 kJ/kg, which is equivalent to 12.05 kJ/mol (because the molar mass of NiTi is 106.6 g/mol). This is an extraordinary low value for a reaction in a condensed system and can correspond only to diffusion in a transient melt formed in the CS wave. However, according to reference data [42], the activation energy for diffusion in some pure liquid metals is the following: Li, E=12 kJ/mol; Sn, E=11.2 kJ/mol; Zn, E=21.3 kJ/mol; Cu, E=40.7 kJ/mol; Fe, E=51.2 kJ/mol. Thus the value of E used for calculations in [40,41] is close to that for diffusion in low-melting metals such as Li or Sn, and is by the factor of 4 lower than for iron whose melting point, Tm, lies between Tm of Ni and Ti (the activation energy for diffusion in liquid metals is known to be proportional to Tm [43]). All the more, this E value is incomparably lower than a typical activation energy for diffusion in intermetallic compounds. Hence in this case the most important parameter of the formal model, E, appears to be physically meaningless. Recently, new features of SHS were observed experimentally [44-47]. First, microscopic high-speed video recording [44,45] and photographing [46] demonstrated micro-heterogeneous nature of SHS which revealed itself in the roughness of the combustion wave front, chaotic oscillations of the local flame propagation rate and new dynamic behaviors such as relay-race, scintillation and quasi-homogeneous patterns. Second, the formation of non-equilibrium structure and composition of SHS products was examined experimentally and interpreted qualitatively in terms of relationships between characteristic times of reaction tr, structuring ts and cooling tc [47]. These features were attributed to two main factors: inhomogeneous heat transfer in the charge mixture and a specific reaction mechanism [46]. These results gave rise to new, heterogeneous models [48-51] involving heat transfer on the particle-to-particle basis [48-50] and percolation phenomena in a system of chaotically distributed reactive and inert particles [51]. However, in these models the traditional formal kinetics for a thermal reaction [Eq. (1.2)] was employed. Thus, an urgent and still unresolved problem in CS is an adequate description of fast interaction kinetics in a unit reaction cell containing particles or layers of dissimilar reactants whose composition corresponds to the average composition of a charge mixture. The most widely used kinetic model, which is connected to a particular phase forming mechanism, is a solid-state diffusion-controlled growth concept first applied to CS in [52] for planar symmetry and in [53] for spherical symmetry of an elementary diffusion couple. As in a charge mixture there are contacts of
dissimilar particles, a layer of an intermediate or final solid product forms upon heating thus separating the initial reactants. The growth rate of the reaction product and associated heat release necessary for sustaining combustion is controlled by solid-state diffusion through this layer. Then, the diffusion-type Stefan problem is formulated instead of Eq. (1.2). However, as demonstrated below in more detail, in most cases modeling was performed not with real diffusion data, which are known for many refractory compounds, but using either dimensionless coefficients varied in a certain range or fitting parameters chosen to match the calculated and measured results of the SHS temperature profile and velocity. It should be emphasized that Diffusion in Materials is a well-developed cross-disciplinary area within Materials Science and Solid State Physics, and the diffusion parameters for many of the phases produced by CS (carbides, nitrides, intermetallics etc.) have been measured experimentally at different temperatures, and these data are supposed to be used in modeling. Besides, in most of the CSsystems fast interaction begins after fusion of a lower-melting-point reactant [35,31] but within this approach melting does not alter the phase layer sequence in an elementary diffusion couple [52,53]. A number of experimental results obtained by the combustion-wave arresting technique in metal-nonmetal (Ti-C [17,18], (Ti+Ni+Mo)-C [19], Mo-Si [20]) and metal-metal (Ni-Al [21,23]) systems gave rise to an qualitative notion of a nontraditional phase formation route. It involves dissolution of a higher-melting-point reactant (metal or non-metal) in the melt of a lower-melting-point reactant and crystallization of a final product from the saturated liquid. Besides, there is much controversy over the presence of an intermediate solid phase in the dissolution-precipitation route. In [21] it is concluded that during SHS in the Ni-Al system, solid Ni dissolves in liquid Al through a solid interlayer separating aluminum from nickel, which agrees with the phase diagram. In this case, the rate-limiting stage is solid-state diffusion across this layer. But in [23] for the same system it is found that above 854 C a solid interlayer between nickel and molten Al is absent; then the overall interaction during CS is controlled by either diffusion in the melt or crystallization kinetics. Such a situation is considered in recent models [54-59], where a solid reactant (nickel [54-56] or carbon [57-59]) dissolves directly in the liquid based on a lower-melting component (Al and Ti, respectively) and product grains (NiAl and TiC, correspondingly) precipitate from the melt; the rate-limiting stage is liquidphase diffusion [54-56] or crystallization kinetics [57-59]. However, within these approaches the fundamental problem of the existence of a thin solid-phase interlayer at the solid/liquid interface is not discussed nor a criterion is obtained for transition between the solid-state diffusion-controlled mechanism and the
B. B. Khina
dissolution-precipitation route with or without a thin interlayer. Hence, the applicability limits of the existing modeling approaches have not been clearly determined so far. The role of high heating rates, which are intrinsic in CS, in most of the models is not accounted for in an explicit form. Thus, adequate description of the interaction kinetics in condensed heterogeneous systems in non-isothermal conditions of CS is an urgent problem in this area of science and technology, and the absence of a comprehensive model hinders the development of new CS-based processes and novel advanced materials. Hereinafter the situation where a reaction between condensed reactants proceeds through a solid layer, i.e. solid reactant (C for the Ti-C system or Ni for the Ni-Al system)/solid final or intermediate product (TiC or one of intermetallics of the Ni-Al system, respectively)/liquid (Ti or Al melt), will be provisionally called solid-solid-liquid mechanism since the interaction occurs at both solid/solid and solid/liquid interface. This term will be used both for the solidstate diffusion-controlled growth pattern where the product layer is growing and for dissolution-precipitation route where the interlayer remains very thin. As the diffusion coefficient in a melt is much higher than in solids, the rate-limiting stage in this mechanism is diffusion across the solid interlayer. The second route, viz. dissolution-precipitation without an interlayer, can be referred to as solid-liquid mechanism since the interaction of condensed reactants (solid C or Ni with molten Ti or Al, respectively) occurs at the solid/liquid interface while the product (TiC or NiAl) crystallizes from the melt. However, up to now the solid-liquid mechanism has not been validated theoretically, nor the applicability limits of the solid-solid-liquid mechanism based on solid-state diffusion kinetics have ever been determined with respect to strongly non-isothermal conditions typical of CS. Thus, the main goal of this work is to develop a system of relatively simple estimates and evaluate the applicability limits of the solid-solid-liquid mechanism approach to modeling CS and determine criteria for a change of interaction routes basing the calculations on experimental data to a maximum possible extent [60,61]. Below, a brief discussion of the diffusion concept of CS is presented. Then, calculations for particulars system, viz. Ti-C and Ni-Al, are performed using available experimental data on both the diffusion coefficients in the growing phase and thermal characteristics of CS. The choice of these binary systems for a modeling study is motivated by the following reasons. First, those are typical SHS systems which have been a subject of extensive experimental investigation (see reviews [3-11,16] and references cited therein). Second, the synthesis products, viz. TiC and NiAl, have a wide industrial application because of their unique physical and mechanical properties.
Hence a large number of parameters needed for numerical calculations can be found in literature. Third, both of these substances are typical representatives of wide classes of chemical compounds that have different properties connected with their intrinsic structural features. Titanium carbide is a typical interstitial compound (like many carbides, nitrides and certain borides) wherein the diffusivities of constituent atoms, Ti and C, differ substantially. Hence the growth of TiC in an elementary diffusion couple Ti/TiC/C during CS is dominated by the diffusion of carbon atoms in the TiC layer and proceeds mainly at the Ti/TiC interface. The experimentally measured parameters such as the chemical diffusion coefficient or the parabolic growth-rate constant for TiC are connected with the partial diffusion coefficient of carbon in this compound. Nickel monoaluminide is a typical substitutional compound with an ordered crystalline structure (like many intermetallics) where the rates of diffusion of Ni and Al atoms are comparable. Thus its growth during CS occurs at both sides of a NiAl layer and can be characterized by a single parameter, namely the interdiffusion coefficient, which is measured experimentally. For the Ti-C system, different situations are considered that can arise during CS within the frame of the above concept and, wherever possible, a quantitative and/or qualitative comparison between the outcome of calculations and experimental results is drawn. Emphasis is made on the structural characteristics of the CS product, titanium carbide, that emerge from this approach. The conditions for a change of the geometry of a unit reaction cell in the SHS wave due to melting of a metallic reactant (titanium) are analyzed and a micromechanistic criterion for the changeover of interaction pathways is derived. For the Ni-Al system, calculations within the frame of the diffusion-controlled growth kinetics are performed taking into account both the growth of the product phase, NiAl, and its dissolution in the parent phases (solid or liquid Ni and molten Al) due to variation of solubility limits with temperature according to the equilibrium phase diagram. Finally, the solid-liquid mechanism concept for CS is justified and phase-formation mechanism maps for these two systems in strongly non-isothermal conditions are plotted.
10
B. B. Khina
1.2. BRIEF REVIEW OF DIFFUSION-BASED KINETIC MODELS OF CS

The interaction kinetics controlled by solid-state diffusion was used for numerical [52,53,62-69] and analytical [70] study of CS for the case of planar diffusion couples (alternating lamellae of reactants) [52,63,64,66,70] and spherical symmetry (growth of a product layer on the surface of a spherical reactant particle) [53,65,67-69]. Inherent in this concept are two basic assumptions: (i) the phase composition of the diffusion zone between parent phases corresponds to the isothermal crosssection of an equilibrium phase diagram, i.e. the nucleation of product phases occurs instantaneously over all contact surfaces and (ii) the interfacial concentrations are equal to equilibrium values. This results in the parabolic law of phase layer growth [71-73]. It should be noted that in many diffusion experiments the phase layer sequence deviates from equilibrium: the absence of certain phases was observed in solid-state thin-film interdiffusion [74,75] and in the interaction of a solid and a liquid metal (e.g., Al) [76,77]. These phenomena were ascribed to a reaction barrier at the interface of contacting phases [78] without considering the nucleation rate of a new phase. The effect of a nucleation barrier was examined theoretically using the thermodynamics of nucleation [79,80] and the kinetic mechanism of phase formation in the diffusion zone [81], and it was shown that in the field of a steep concentration gradient the formation of an intermediate phase is suppressed [79-81]. This effect has never been considered in the diffusion models of CS. As in the theory of diffusion-controlled interaction in solids the nucleation kinetics is not included and it is assumed that critical nuclei of missing phases continuously form and dissolve [72,73], this qualitative concept is sometimes used in interpreting the results of CS [21]. It will be fair to say that deviation of phase-boundary concentrations from equilibrium due a reaction barrier was examined qualitatively for SHS [64] in the case of planar geometry. This effect is noticeable only in the low-temperature part of the SHS wave, and at high temperatures a strong barrier can only slightly decrease the combustion velocity [64]. Also, the influence of such barrier on selfignition in the Ni-Al system at low heating rates, dT/dt<60 K/min, was studied quantitatively using experimental data on both thin-film interdiffusion in the NiAl3 compound [82], which is the first phase to form in Ni-Al diffusion couples, and bulk diffusion data [83]. Similar calculations were performed using an experimental temperature profile of SHS to determine the NiAl3 layer thickness
11
formed below the melting point of aluminum Tm(Al)=660 C [67]. At a thick NiAl3 layer (low heating rates) the reaction barrier is of little significance, but it can slow down the interaction for thin layers (higher heating rates) [83,67]: e.g., at dT/dt > 35 K/min the formation of the primary product can be suppressed [67]. But, as noted in [83,67], these results refer not to the SHS itself but only to a preliminary stage (i.e. the preheating zone of the SHS wave) because fast interaction begins at T>Tm(Al), the combustion temperature reaches 1400 C and the final product is NiAl [67]. It should be outlined that in many works using the diffusion model of CS the calculations were performed with dimensionless (relative) parameters varied in a certain range. A known or estimated value of the activation energy for diffusion in one of the phases was used only as a scaling factor and thus the results obtained revealed only qualitative characteristics of the process [52,53,62,66]. Besides, many of the modeling attempts [52,53] did not account for a change in the spatial configuration of reacting particles due to melting and spreading of a metallic reactant. The effect of melting was reduced to a change of interfacial concentrations and the ratio of diffusion coefficients in contacting phases [62]. In more recent papers [67,68], the parameter values (the activation energy E and preexponent D0) used for calculating the diffusion coefficient in a growing phase were presented. However, those were not the real values measured in independent works on solid-state diffusion but merely fitting parameters calculated from the characteristics of CS. For example, the formation of NiAl above 640 C was modeled using D0=4.8102 cm2/s and E=171 kJ/mol [67]. As noted in [67], this E value was the experimentally determined activation energy for the CS process as a whole. Then the diffusion coefficient in NiAl at T=1273 K is D = D0exp(E/RT) = 4.6109 cm2/s. Lets compare it with experimental data on reaction diffusion in the Ni-Al system. For NiAl, D=(2.53.6)1010 cm2/s at T=1273 K [84]. The parameters for interdiffusion in this phase are E=230 kJ/mol and D0=1.5 cm2/s [85], hence at T=1273 K D=5.41010 cm2/s. Thus, the diffusion coefficient used in modeling SHS exceeds the experimental value by an order of magnitude. SHS wave in the Ti-Al system with the Ti-to-Al molar ratio of 1:3 in the charge mixture was modeled using E=200 kJ/mol and D0=4.39 cm2/s for phase TiAl3 [68]. This E value was obtained from experiments on combustion synthesis using Arrhenius plots, and D0 was chosen to match the calculated and measured results of the propagation speed. Again, these values refer to the SHS wave as a whole but not to interdiffusion in TiAl3. However, experimental data on SHS of TiAl3 for the same starting composition, which were analyzed using the classical
12
B. B. Khina
combustion model [see Eqs. (1.1),(1.2)], gave a substantially higher activation energy: E=483 kJ/mol [86]. If solid-state diffusion in the TiAl3 layer is really the rate-limiting stage of the process, then the values of apparent activation energy ought to agree (within an experimental error) regardless of the particular form of a model. Diffusion coefficients are measured experimentally within a rather wide margin of error using a variety of techniques, and typically various methods yield different values. But since diffusion parameters for many refractory compounds, which can be produced by combustion synthesis, can be found in literature, it appears possible to verify the validity of the diffusion-based kinetic model of SHS employing a somewhat opposite approach: estimating the product layer growth and heat release using the experimental characteristics of SHS and independent diffusion data. The models, parameter values and results of simulations for two classical CS-systems, viz. Ti-C and Ni-Al, will be considered in more detail in the subsequent chapters.
Chapter 2
MODELING DIFFUSION-CONTROLLED FORMATION OF TIC IN THE CONDITIONS OF CS

2.1. INTRODUCTION
CS in the Ti-C system was a subject of extensive theoretical and experimental studies [53,17,18,62,69] because of industrial significance of the product, titanium carbide, which is used for a wide range of applications because of its high melting point, hardness and chemical stability. It is a suitable candidate for theoretical investigation for the following reasons: (i) the Ti-C phase diagram [87] (see Figure 2.1) contains only one binary compound TiC whose melting temperature Tm(TiC)=3423 K exceeds the experimental SHS temperature TCS=3083 K [88] and (ii) numerous diffusion data for titanium carbide are available in literature [89-91]. We consider the case of spherical symmetry which better fits a typical configuration of reacting particles in CS. With respect to the phase diagram, here the solid-solid-liquid mechanism [situation C(solid)/TiC(solid)/ Ti(liquid)] is quasi-equilibrium and the solid-liquid mechanism [situation C(solid)/Ti(liquid)] is truly non-equilibrium. Lets consider solid-state diffusion-controlled formation of the product, titanium carbide, during heating of the Ti+C charge mixture in the SHS wave. Typical particle radii are 5 to 100 m for Ti, about 0.1 m for carbon black and 1 to 30 m for milled graphite [17,18,69,88]. Two scenarios with a different geometry of a unit reaction cell are examined: (1) a solid Ti particle surrounded by carbon particles in a stoichiometric mass ratio at temperatures below the Ti melting point, Tm(Ti)=1940 K [Figure 2.2 (a and d)], and (2) a solid carbon particle surrounded by liquid titanium at T>Tm(Ti) [Figure 2.2 (c and e)].
14
B. B. Khina
Figure 2.1. The equilibrium Ti-C phase diagram [87] and experimental SHS temperature.
Modeling Diffusion-Controlled Formation of TiC
15
Figure 2.2. Schematic of an elementary reaction cell in the SHS wave in the Ti-C system (a and c) and corresponding concentration profiles for solid-state diffusion (d and e) [60]: (a and d) growth of the TiC layer on the surface of a titanium particle at T<Tm(Ti); (b) rupture of the primary product shell and spreading of molten titanium at T=Tm(Ti); (c and e) growth of the TiC layer on the surface of a carbon particle after titanium melting and spreading at Tm(Ti)TTCS.
A condition for the change of the reaction cell geometry due to titanium melting [Figure 2.2 (b)] is analyzed later.
2.2. SCENARIO 1: GROWTH OF S TIC CASE ON THE TITANIUM PARTICLE SURFACE

In scenario 1, at T<Tm(Ti) a thin uniform layer of TiC is formed on the surface of the Ti particle. This is due to fast surface diffusion of C atoms from the Ti/C contact spots in the charge mixture, which agrees with the idea expressed in [88]. Because of low carbon solubility in solid -Ti and high diffusivity of C atoms in the -phase as compared with that in TiC [90,91], we can consider the growth of only one product phase-titanium carbide, i.e. perform an upper-level estimate of the product thickness. Then the TiC layer growth in the SHS wave is described by a diffusion-type Stefan problem written in spherical symmetry neglecting a volume change at the Ti/TiC interface
16
B. B. Khina
c C t
D[T(t)] 2 c C , r r r 2 r
(2.1)
0 ( c0 c1 ) 21
dR 1 dt
= D[T(t)]
c C r
R1(t)
(2.2)
where D is the chemical diffusion coefficient in TiC, which is usually associated with the diffusion coefficient of carbon in the carbide layer, cC is the mass concentration of carbon, R1(t) is the current position of the TiC/Ti interface, R2 is the outer radius of the Ti particle, and c01, c021 and c023 are the interface concentrations [Figure 2.2 (d)] according to the equilibrium phase diagram. The boundary (at r=R2) and initial conditions to Eq. (2.1) are cC(t, R2) = c023, cC(t, r=0) = c01, cC[t, R1(t)] = c021, R1(t=0) = R2. (2.3)
In [62,63] the Stefan-type boundary condition, Eq. (2.2), was posed at both Ti/TiC and TiC/C interfaces. We should outline that in interstitial compounds such as nitrides, carbides and many borides, the partial diffusion coefficient of nonmetal species exceeds that of metal atoms by several orders of magnitude, which is due to the interstitial diffusion mechanism. Hence, the growth of titanium carbide occurs at the Ti/TiC interface and is controlled by the diffusion of C atoms across the TiC layer. But at the C/TiC interface the growth of TiC at the expense of graphite, which requires the supply of Ti atoms, cannot occur. Thus, the first-kind boundary condition, cC(t, R2) = c023 [see Eq. (2.3)] is used for the C/TiC interface, which actually denotes an ideal diffusion contact of carbon particles with the outer surface of the growing TiC layer due to fast surface diffusion of the C atoms from the C/TiC contact spots.
2.3. SCENARIO 2: GROWTH OF A TIC LAYER ON THE SURFACE OF SOLID CARBON PARTICLES
The physical background for scenario 2 [Figure 2.2 (c)] is the following. Spreading of molten titanium towards solid carbon in the SHS wave was observed experimentally [92,93]. Since it is accompanied with chemical interaction, for a sufficiently small C particle size the spreading velocity is not the rate-limiting
17
stage [93,94]. Hence we consider that at TTm(Ti) the carbon particles are completely enveloped with liquid titanium, and a thin TiC layer forms at the Ti/C interface separating the parent phases. The product growth occurs at the Ti(melt)/TiC interface, i.e. outwards, due to diffusion of carbon atoms across the TiC layer [Figure 2.2 (e)]. Since the diffusion coefficient of carbon in the melt is at least an order of magnitude higher than in the carbide (Table 2.1), it is reasonable to presume that the titanium melt is saturated with carbon (otherwise the TiC layer will be dissolving). Then the boundary condition at r=R0 to diffusion equation (2.1) and initial conditions to Eqs. (2.1),(2.2) look as cC(t, R0) = c023, cC(t, r>R1) = c01, R1(t=0) = R0, where R0=const is the initial radius of the carbon particle. (2.4)
2.4. DIFFUSION DATA FOR TIC

The parameters for calculating the diffusion coefficient in TiC in the Arrhenius form D = D0 exp[E/RT(t)] (2.5)
are listed in Table 2.1, wherein the experimental data available in literature [95103] for different temperature intervals T are collected. It is seen that different data sources give substantially different values of both activation energy and preexponential factor, thus it seems necessary to select the parameters values suitable for numerical calculations. Since the extrapolation of D to the whole temperature range of SHS may bring about overestimated values, the diffusion coefficients in TiC calculated at T=Tm(Ti) and TCS must be compared with the diffusion coefficient in molten titanium: it is obvious that the value of D in a solid metal-base refractory compound is at least an order of magnitude lower than in a melt of the corresponding metal. Because of the absence of experimental data, the diffusion coefficient of C in molten Ti is estimated by a simple Stokes-Einstein (or Sutherland-Einstein) formula, which was used for assessing the diffusion parameters of C, N, O and H in liquid metals (Fe, Co, Ni, etc.) [104,105] Di(m) = kBT/(nai), (2.6)
18
B. B. Khina
where numerical factor n=4 for substantially differing atomic radii of the melt components and n=6 for close radii, ai is the atomic radius of i-th diffusing species in the melt, =m is the dynamic viscosity, is the kinematic viscosity and m is the liquid-phase density. For carbon atoms, the covalent radius is aC=0.077 nm [106]. The density of molten Ti is m=4.11 g/cm3 [107]. A typical value of the kinematic viscosity for such liquid metals as Al, Fe, Co, Ni et al. near the melting point is (0.5-1)102 cm2/s [106]. For liquid titanium saturated with carbon, =0.94102 cm2/s at T=Tm(Ti) [107], then DC(m)(Tm) (4.87.2)105 cm2/s. For higher temperatures, the value =1.03102 cm2/s at T=2220 K is known [107]; using it at T=TCS gives DC(m)(TCS) (6.910.4)105 cm2/s. It should be noted that since Eq. (2.6) doesnt account for chemical interaction in the melt, which may be substantial for the Ti-C system, these DC(m) values are upper estimates. Then the values of diffusion coefficients in TiC, which are close to or higher than the upper estimate of DC(m)(TCS), are excluded from consideration (lines 10 to 14 in Table 2.1). Table 2.1. Diffusion data for titanium carbide
Species No. D0, cm2/s E, T, K kJ/mol 235.6 2073-2973 1 5102 2 6.98 398.7 1723-2973 3 4 5 6 7 8 9 10 11 12 13 14 Ti 10 45.44 114 0.1 6.5102 4.2102 0.48 99.48 77.8 220 370 438.9 447.3 460.2 259.4 269.9 307.1 328.42 328.42 338.9 405.8 410.0 1873-2573 1723-2553 2018-2353 1553-1773 1673-1973 1983-2573 1473-2023 1473-2023 1473-1673 2200-2600 2200-2600 1173-1473 2193-2488 D[Tm(Ti)], D(TCS), Note Refs. cm2/s cm2/s 2.3108 5.1106 [90,91,95,101] 1.31010 1.23106 [89,95,96,98] (TiC0.97) 1.51011 3.7107 [95,96] (TiC0.9) 4.11011 1.2106 [96,98] (TiC0.887) 4.61011 1.8106 [96,99] (TiC0.67) 1.0108 4.0106 [90,91] 3.5109 1.7106 [90,102] 2.31010 2.6107 [90,91] (TiC0.9) 6.91010 1.3106 [103] (TiC1.0) 1.4107 2.7104 [103] (TiC0.5) D(TCS) > DC(m)(TCS) 5.8108 1.4104 [89] D(TCS) > DC(m)(TCS) 2.6109 2.9105 [100] (TiCx, D(TCS) x=0.86-0.91) DC(m)(TCS) 3.4109 4.1105 [100] (TiCx, D(TCS) x=0.86-0.91) DC(m)(TCS) 5.8107 1.7103 [90] D(TCS) >> DC(m)(TCS) 6.51016 1.5108 [90,97] (TiCx, DTi << DC x=0.67-0.97)
1.31103 347.3 4.36104 736.4
19
2.5. TEMPERATURE OF THE REACTION CELL IN THE SHS WAVE

Self-heating from ambient temperature, T0, to TCS during the combustion synthesis is due to the adiabatic heat release of chemical reactions which are almost accomplished when maximal temperature is reached, and in the after-burn zone (at TTCS) only coalescence and sintering of the product particles occur with minor heat release [3-5]. Hence calculations of the product layer thickness and relevant heat release should be done in the time interval [0, tCS] corresponding to the attainment of TCS. To perform calculations, we have to know the time dependence of temperature in the reaction cell, T(t). We consider a steady-state combustion regime. For a low-temperature portion of the SHS wave, T0<T<Tm(Ti), where heat release is small, the temperature profile along a sample can be calculated using a known analytical solution [108] T() = T0 + (TmT0) exp(vSHS/), = x vSHSt, (2.7)
where vSHS is the combustion velocity, x is a coordinate along the SHS-sample and is the thermal diffusivity. To determine T(t) for the reaction cell, a coordinate x0 is chosen for which T(x0, t=0) = T0' = T0+0.01Tm, where T0=298 K. Then the heating time, tm, from T0' to Tm is
tm =
(2.8)
v SHS
2
ln
0.01Tm Tm T0
For a stoichiometric Ti-C mixture (20 wt.% C), 0.04 cm2/s [109] and vSHS=6 cm/s [88,18]. For higher temperatures, TmTTCS, we use the spline-approximation of the experimental temperature profile of a steady-state SHS wave in the Ti-C system, which was registered by a micro-thermocouple technique [88] (Figure 2.3). It should be noted that the low-temperature tail [at T<Tm(Ti)] calculated by Eq. (2.7) lies slightly above the corresponding part of the experimental curve, which is not shown in Figure 2.3 merely to avoid encumbering. So, we use an upper estimate of temperature (and hence the grown product layer) in the reaction cell. In Figure
20
B. B. Khina
2.3, x=0 corresponds to the melting point of Ti, consequently the heating time from Tm(Ti) to TCS is tCS = tCStm = x(TCS)/vSHS.
Figure 2.3. Temperature profile of the SHS wave in the Ti-C system [60]: 1, analytical solution for a steady-state SHS wave [Eq. (2.7)] for TTm(Ti); 2, cubic-spline approximation of experimental curve [88] in the range Tm(Ti)TTCS.
2.6. ADIABATIC HEAT RELEASE IN THE REACTION CELL

Having the heating law of the reaction cell, we can calculate the heat release due to diffusion-controlled phase layer growth in non-isothermal conditions and thus the maximal temperature attained, and then compare it with experimental TCS. In adiabatic conditions, a heat balance equation for the formation of stoichiometric TiC1.0 is written as:
Tad
298(TiC1.0)mTiC(t)
= mTiC(t)
298
c p (TiC)dT + m (t) c p (C)dT +

C
Tad
298
mTi(t) c p (Ti )dT + I[Tad Tm (Ti )]H m (Ti ) , 298
Tad
(2.9)
21
where Tad is the adiabatic combustion temperature, cp(i) is heat capacity, mi(t) is a current mass of i-th substance, H0298(TiC1.0) = 3.077 kJ/g is the standard enthalpy of TiC1.0 [110], Hm(Ti) = 0.305 kJ/g is the heat of fusion of Ti [110] and I[TadTm(Ti)] is the Heaviside unit-step function. The masses of all the substances are determined using a solution of the Stefan problem for particular geometry of the reaction cell, and then Tad is calculated from Eq. (2.9).
2.7. MODELING OF TIC LAYER GROWTH ON THE TITANIUM PARTICLE SURFACE

2.7.1. Analytical Solution to Scenario 1
Problem (2.1)-(2.3),(2.5) is non-linear and in a general case can be solved only numerically. However, for a similar linear problem (with D=const) an asymptotic solution for the growth of a spherical phase layer, which is valid for a small layer thickness h=R2R1 << R2, is known [25,111,112]. To apply it, we linearize Eqs. (2.2) and (2.3) using substitution
(t) = D[T()]d .
0
(2.10)
Here t varies from 0 to tCS, where tCS is time at which the temperature of the reaction cell reaches its maximal value TCS. Then, according to [25,111], the asymptotic solution of Eqs. (2.1)-(2.3) with respect to the product layer thickness, h, looks as h() = R2 R1() = 1/2 + 1/R2 + 23/2/(2R22), (2.11)
where 2 = 21 12[24q/+4(5+6q) + 23(1+q)] [32+60q+2(18+20q) + 4(1+q)]1, 1 = 2/(3+2/2), q = (c023c021)/(c021c01), and is a solution of transcendental equation (2.12)
22
B. B. Khina 1/2(/2) exp(/2)2 erf(/2) = q, (2.13)
which arises in a similar Stefan problem for a semi-infinite sample. For scenario 1 [Figure 2.2 (a)], mTiC() = (4/3) [R23R13()]TiC, mC() = mC0 0.2mTiC(), mTi() = (4/3)R13()Ti, where i is the density of i-th substance. For a stoichiometric composition, the initial C-to-Ti mass ratio is mC0/mTi0 = 0.25, where mTi0 = (4/3)R23Ti. Then, ignoring the temperature dependence of heat capacities in Eq. (2.9), the maximal adiabatic heating of the reaction cell, Tad = Tad298, is estimated as
Tad = H 0 (TiC1.0 ) TiC 298
3 3 TiC[c p (TiC) 0.2c p (C)] + Ti [c p (Ti)R 1 () + 0.25c p (C)R 3 ]/[R 3 R1 ()] 2 2
(2.14)
2.7.2. Results of Calculations for Scenario 1

In the temperature range T0'TTm(Ti), equilibrium interfacial concentrations c01= 0.00138 and c021=0.11 corresponding to the Ti-TiC eutectic temperature Teu=1918 K [87] were used. For higher temperatures, TmTTCS, it was assumed that molten titanium remains inside the spherical TiC shell, and the interfacial compositions were taken for an intermediate temperature T=2673 K: c01=0.065, c021=0.14 [87]; at the C/TiC interface c023=0.2 (maximal solubility of C in the carbide). Calculations have shown that varying the c021 and c01 values along the solidus and liquidus lines of the Ti-C phase diagram in the range T=Tm to TCS has a negligible effect on the TiC layer thickness and associated heat release. For the temperature range T0'TTm the calculations were performed with all the diffusion data listed in Table 2.1 [Figure 2.4 (a and b)]. For the whole temperature range, T0'TTCS, only the data giving D(TCS)< DC(m) (lines 1 to 9 in Table 2.1) were used [Figure 2.4 (c and d)]. A maximal TiC layer thickness attained by the time of reaching the titanium melting temperature is small, h(Tm)=0.068 m [Figure 2.4 (a)], and corresponds to diffusion data No.14 in Table 2.1. The relevant adiabatic heating is insignificant, Tad=57 K for R2=10 m [Figure 2.4 (b)], and decreases with increasing the particle radius. For the temperature range T0'TTCS, the maximal TiC layer thickness corresponds to the set of diffusion parameters No. 1 (see in Table 2.1), and this value is small: h(TCS)1.6 m [Figure 2.4 (c)].
23
Figure 2.4. Thickness of the TiC layer formed on the surface of a titanium particle by the time of attainment of Tm(Ti) (a) and TCS (c), and relevant adiabatic heating (b and d) [60]. Numbers at curves correspond to diffusion data sets in Table 2.1.
24
B. B. Khina
The corresponding adiabatic heating is only Tad=1064 K for the Ti particle radius of 10 m and sharply drops with increasing R2 [Figure 2.4 (d)]. Thus, heat release due to product growth is insufficient to sustain the SHS wave (i.e. to reach TCS=3083 K). The obtained result, viz. a small thickness of TiC grown in the temperature range below Tm(Ti), qualitatively agrees with experimental data [17,18]: in rapidly cooled samples almost no interaction was observed in the so-called preheating zone of the SHS wave. However, at the attainment of T=Tm(Ti) the melting of titanium can bring about the rupture of the primary TiC shell and the spreading of the metallic melt. It should be noted that in [69] the diffusion-controlled TiC formation was assessed using 6 different sets of the diffusion data, but only an isothermal situation below the titanium melting point was examined. Besides, the TiC layer growth was considered on the surface of a carbon particle whereas, as mentioned above, the initial TiC film at T<Tm(Ti) will most probably form on the surface of solid Ti particles due to fast surface diffusion of C atoms.
2.8. RUPTURE OF THE PRIMARY TIC SHELL

The density of solid -Ti at T=Tm is s=4.18 g/cm3 while for molten titanium at the same temperature m=4.11 g/cm3 [107]. The conditions for the rupture of the TiC case because of the dilatation of the titanium core during melting can be determined from a continuity equation written for spherical symmetry [113]: grad div Ur = 0. (2.15)
The boundary conditions at r=R1 (expansion due to melting implying that liquid titanium is an incompressible fluid) and r=R2 are written as Ur(r=R1) = R1[(s/m)1/31], rr(r=R2) = p0. (2.16)
Here Ur is the radial displacement, rr is the radial stress and p0=0.1 MPa is the outer pressure. As the plasticity of TiC is low, we consider only elastic deformation. A solution to Eq. (2.15) is: Ur(r) = Ar + B/r2, urr = Ur/r = A 2B/r3, u(r) = Ur/r = A + B/r3, (2.17)
25
where urr and u are the radial and shear strain, correspondingly, and A and B are constants which are determined from boundary conditions (2.16). Hookes law for spherical symmetry looks as rr =
Y Y [(1)urr + 2u], = (u + urr), (1 + )(1 2) (1 + )(1 2)

(2.18)
where rr and are the radial and shear stress, correspondingly, Y is the elastic modulus and is the Poissons ratio [113]. Then the solution for is obtained from Eqs. (2.16)-(2.18):
(r) =
Y s 1 2 m
1/ 3
1+ f R3 R3 f f r 2 2 r ,f= , fr = , 1 p0 3 3 1 + f 1 + f 2 R 2 r 1
(2.19)
1+ . 1 2
Rupture of the primary TiC shell occurs when the maximal shear stress in the spherical layer (at r=R2) exceeds the ultimate tensile stress uts. Then from Eq. (2.19) we obtain a critical thickness, hcr = R2R1, of the TiC layer:
h cr
2 uts + p 0 1 , = = R2 ( uts + p 0 )

1/ 3
1/ 3
3Y s = 1 2 m
1 .
(2.20)
The TiC case can burst at h hcr. This is an upper estimate because we dont take into account partial dissolution of TiC in molten titanium due to the eutectic reaction at 1645 C. To calculate the hcr value, we have to determine the mechanical properties of TiC at the melting temperature of titanium. The temperature dependencies of the elastic modulus, Y, and shear modulus, G, for TiC are known in the following form [96]:
26
B. B. Khina Y(T) = Y0 bYT exp(T0/T), G(T) = G0 bGT exp(T0/T), (2.21)
where T0=320 K, Y0=461 GPa, bY=0.0702 GPa/K, G0=197 GPa and bG=0.0299 GPa/K. Then at Tm(Ti)=1940 K we have Y=346 GPa and G=148 GPa, thus the Poissons ratio is = Y/(2G)1 = 0.17. As for uts values for TiC at elevated temperatures, there are only disembodied data, e.g., uts(T=1073 K) 380 MPa, uts(T=1273 K) 280 MPa [90]. However, available are data on the bending strength, b, of titanium carbide over a wide temperature range because it is a typical test for brittle refractory compounds; b has a maximum of approximately 500 MPa around T=2000 K [96, page 233]. Then, using an estimate uts ~ b/2 = 250 MPa, from Eq. (2.20) we obtain hcr0.6R2. Since the calculated value h[T=Tm(Ti)] is very small, for any initial size of Ti particles used in SHS (R2=5 to 100 m) melting of the titanium core will inevitably bring about the rupture of the primary TiC shell and spreading of the melt. This changes the geometry of a unit reaction cell as shown in Figure 2.2 (a-c).
2.9. GROWTH OF A TIC LAYER ON THE SURFACE OF A SOLID CARBON PARTICLE

2.9.1. Analytical Solution to Scenario 2
For scenario 2 [Figure 2.2 (c and e)], an asymptotic solution to Eqs. (2.2),(2.3)-(2.5) with respect to the TiC layer thickness, h, can be obtained similarly to Eq. (2.11) [25,111,112]: h() = R1() R0 = 1/2 1/R0 23/2/(2R02). (2.22)
Here coefficients , 1 and 2 are defined, as previously, by Eqs. (2.12),(2.13) and is determined according to Eq. (2.10) where integration is performed over the time range 0 t tCS, which corresponds to the temperature range TmTTCS (Figure 2.3). To calculate adiabatic heating, we turn to Eq. (2.9). For the reaction cell shown in Figure 2.2 (c), mTiC() = (4/3) (R13()R03)TiC, mC() = mC0 0.2mTiC(), mC0 = (4/3) R03C and mTi() = 4mC0 0.8mTiC(). Then, ignoring the temperature dependence of heat capacities and neglecting the melting enthalpy of
27
titanium (because Hm(Ti) << |H298(TiC1.0)| [110]), the adiabatic heating of the reaction cell is estimated as
Tad = H 0 (TiC1.0 ) TiC 298 TiC[c p (TiC) 0.2c p (C) 0.8c p (Ti m )] + C[c p (C) +
3 3 4c p (Ti m )]/[R 1 ()/R 0
,
1]
(2.23) were subscript m denotes melt. For calculations, the values of heat capacities (according to [110]) were taken at T=TCS. Eq. (2.23) refers to incomplete conversion of carbon into titanium carbide, i.e. when 0<TiC<1, where the degree of conversion is expressed as TiC = 1 mC()/mC0 = 0.2[R13()/R03 1]TiC /C. (2.24)
For complete conversion TiC=1, the maximal adiabatic heating is Tad(max) = H0298(TiC1.0)/cp(TiC) = 3095 K, and the adiabatic SHS temperature Tad(max) = 298 + Tad(max) = 3393 K. It is somewhat higher than the value Tad=3210 K calculated taking into account the temperature dependence of heat capacities [5,114]. Thus, Eq. (2.23) gives an upper estimate for Tad.
2.9.2. Results of Calculations for Scenario 2

Numerical results are presented in Figure 2.5. The TiC layer thickness, which can form in the SHS wave with the temperature profile shown in Figure 2.3, was calculated using Eq. (2.22) not accounting for the exhaustion of reactants [Figure 2.5 (a)]. The maximal value is h1.5 m for a sufficiently large carbon particle size, R0=12.5 m, at the 1st set of diffusion data in Table 2.1. Adiabatic heating [Figure 2.5 (b)] was calculated taking into account the degree of conversion of carbon into carbide [see Eqs. (2.23),(2.24)]. A plateau with Tad=Tad(max) for small R0 values corresponds to complete conversion (TiC=1). Thus, from Figure 2.5 (b) it is seen that the diffusion-controlled growth mechanism can provide sufficient adiabatic heating to sustain the SHS process, which results from almost complete conversion, only for small-sized carbon particles: R0<3 m. This contradicts numerous experimental works where SHS of TiC was performed with coarse-grained graphite: 7 m [69,115,116], 20 m [69,115], and up to 63 m [117] in diameter.
28
B. B. Khina
Figure 2.5. Formation of the TiC layer on the surface of a carbon particle in the SHS wave after spreading of molten titanium (Tm(Ti)TTCS): (a) product layer thickness and (b) corresponding adiabatic heating vs. carbon particle radius [60]. Numbers at curves correspond to diffusion data sets in Table 2.1.
The results obtained regarding the above concept suggest that fast and complete conversion of reactants into the final product providing the required heat release can be achieved via a different route (without diffusion control of the product formation). For further analysis of the diffusion model it makes sense to estimate a structural parameter of the product, viz. porosity. To do this, it is necessary to evaluate the displacement of the C/TiC interface.
29
2.9.3. Displacement of the C/TiC Interface in the Emptying-Core Mechanism

Above we have calculated the thickness of a spherical product layer formed due to interstitial diffusion of C atoms through titanium carbide, i.e. outward growth of TiC on the surface of carbon particles. Hence the TiC particles formed after complete conversion of the reactants will be hollow. This pattern of diffusion-controlled product formation is sometimes called the emptying-core mechanism [115]. Lets estimate the displacement of the C/TiC interface, i.e. inward growth of the product layer due to diffusion of Ti atoms across TiC. From Figure 2.5 (a) it is seen that at R05 m the effect of curvature is minor: raising R0 from 5 to 12.5 m increases the TiC thickness by less than 10%. Thus the diffusion problem can be considered for a semi-infinite rod. The diffusion equations are written for both C and Ti atoms
c i t
= D i [T ( t )]
ci r
2
, i C,Ti.
(2.25)
The Stefan-type boundary conditions to Eq. (2.25) are formulated at interfaces Ti(melt)/TiC (r=R1) and C/TiC (r=R0) taking into account that here R0=R0(t) and cC + cTi =1
0 ( c0 c1 ) 21
dR 1 dt
= D C[T(t)]
c C r
R1(t)
(1 c0 ) 23
dR 0 dt
= D Ti [T(t)]
c C r
R 0(t)
(2.26)
The initial conditions are R0(t=0) = R1(t=0) = R00. (2.27)
Here DC and DTi are the partial diffusion coefficients of C and Ti atoms in TiC (see Table 2.1), r is the radial coordinate and R00 is the initial position of the C/Ti(melt) interface at which a thin TiC layer originates at t=0. Using substitution
i (t) = D i[T()]d , i C,Ti,

0
(2.28)
30
B. B. Khina
the non-isothermal problem (2.25)-(2.27) is reduced to an isothermal (linear) case which has an analytical solution [118] for the displacement of phase boundaries Ti/TiC (h) and C/TiC (): h(C) = R1(C) R00 = C C1/2, (Ti) = R0(Ti) R00 = Ti Ti1/2. (2.29) The coefficients C and Ti are determined from transcendental equations: 1/2(C/2)exp(C/2)2{erf(C/2) + erf[Ti(Ti/C)1/2/2]} = (c023c021)/(c021c01) 1/2(Ti/2)exp(Ti/2)2{erf(Ti/2) + erf[C(C/Ti)1/2/2]} = (c023c021)/(1c023). (2.30) The calculated displacement of the C/TiCx interface during interaction in the SHS wave (at TmTTCS) is negligibly small: =4.7 nm << h, i.e. only about 10 lattice periods of TiC. This is due to the smallness of the partial diffusivity of Ti atoms, DTi << DC (Table 2.1). That is, the inward growth of the TiC layer is insignificant, and we can reasonably assume R0=const.
2.9.4. Product Porosity in the Emptying-Core Mechanism

Now lets make an upper-level estimate (for TiC1.0) of the apparent density of titanium carbide particles formed through the above described route implying that the starting charge composition also corresponds to the TiC1.0 stoichiometry (20 wt.% C). Then the mass of TiC formed per single carbon particle is mTiC = (4/3)(R13R03)TiC = (4/3)R03C/0.2, and its volume is VTiC = (4/3)R03(1+5C/TiC). Thus the apparent density of hollow TiC particles is eff = [1/TiC+1/(5C)]1 3.3 g/cm3 (where TiC=4.91 g/cm3 for TiC1.0 [89,90,96] and C2 g/cm3 for graphite), which is only 67% of the density of TiC1.0. That is, closed porosity of titanium carbide produced by SHS will be cl = 1eff/TiC 0.33 and the pore size must be about the initial diameter of carbon particles. However, microstructure investigations [17,92,116,117,119] have revealed that as-synthesized TiC samples possess mainly open porosity, op0.50.6. This is ascribed to the presence of green porosity, 0=0.30.65 [94], melt spreading and gas release [92,119]. The fraction of closed porosity is usually small: about 1% for SHS in vacuum [117]. Besides, dense titanium carbide articles (with relative density above 95%) can be readily produced by SHS under high pressure
31
[114,116] or by short-term compaction of a hot pellet immediately after the completion of SHS [3,4 and literature cited therein]. If as-synthesized TiC particles were hollow, which follows from the above described route, the attainment of such relative density would require prolonged pressure sintering at a high temperature. Thus the structural characteristic of the SHS product emerging from the diffusion model disagrees with experimental observations, which therefore supports an idea of a dissolution-precipitation route capable of producing dense TiC particles, which includes dissolution of carbon in molten titanium and subsequent crystallization of the product grains. It should be noted that a conclusion in favor of the diffusion-controlled growth of hollow TiC shells on the surface of carbon particles having a size of 2R0=7 m (initial porosity of a sample was 0=0.2) and 20 m (0=0.4) was made in [115] basing solely on the porosity measurements and microstructures of assynthesized specimens. Lets analyze these experimental data. For all of the samples the initial temperature, T0, was 293 K, and the total porosity measured after SHS was almost the same, t(m)=0.460.5. As shown above, SHS of TiC via this mechanism is possible for small-sized carbon particles, R03.5 m, but closed porosity will be cl=0.33 which greatly exceeds the measured value cl(m)=0.060.08 [115]. Besides, total porosity of an as-synthesized sample for the formation of TiC1.0 is estimated as t = 1 (10)[TiC(0.8/Ti+0.2/C)]1 (2.31)
implying that the specimen volume doesnt change during SHS, which is true for strongly compacted green pellets (as in [115]). Here Ti=4.51 g/cm3 [106] is the density of initial -Ti particles. Then for samples with 7 m diameter carbon particles the formation of dense TiC grains (TiC=4.91 g/cm3) yields the total final porosity t=0.41, which is close to experimental data. But if hollow TiC particles are formed via the diffusion mechanism, then, substituting into Eq. (2.31) eff=3.3 g/cm3 instead of TiC we obtain t'0.13. In this case t' signifies the fraction of pores between the hollow particles. But this value is less than 0.1540.005 (the Scher-Zallen criterion), which is required by the percolation theory [120,121] for the existence of open porosity. Thus, the sample will contain only closed pores (inside the TiC particles and between them) whereas in [115] high open porosity was observed: op(m) = t(m)cl(m) 0.4. For larger carbon particles (2R0=20 m), as demonstrated above, SHS via the diffusion mechanism is impossible (for T0298 K) because of low heat release per
32
B. B. Khina
unit reaction cell [Figure 2.5 (b)]. If dense TiC particles are formed, from Eq. (2.31) for 0=0.4 we have t=0.56 which is close to experimental porosity t(m)=0.460.5. For the formation of hollow particles (cl=0.33), Eq. (2.31) gives t'=0.34, and then the total porosity will be t = t' + cl = 0.67, which substantially exceeds the measured value. Maximal closed porosity in the experiments was cl(m)=0.22 for samples with the particle diameter 17 m (Ti) and 20 m (carbon) [115]. Since SHS was performed under isostatic gas pressure, the origin of closed pores should be ascribed to partial sintering of dense TiC grains (presumably precipitated from melt) around voids formed on the sites of outflown titanium particles. This is supported by the fact that the closed porosity was noticed to increase with raising the gas pressure (from cl(m)=0.12 at 1 bar to 0.22 at 70 bar) while the total porosity remained almost the same, t(m)=0.470.5 [115].
2.10. ANALYSIS OF THE SHRINKING-CORE MECHANISM IN THE TI-C SYSTEM

Lets discuss a dissolution-precipitation route of the TiC formation, which can produce 100% dense particles. According to the idea first proposed in [122] and used for studying SHS in the Ni-Al [67] and Nb-C [123] systems, as soon as a metallic melt spreads and engulfs solid particles, a thin film of an intermediate phase (here TiC) forms around them instantaneously. In this interaction pattern, the phase layer sequence in the reaction cell corresponds to the equilibrium phase diagram (Figure 2.1). Then the product particles (TiC) precipitate from the saturated melt due to diffusion of carbon atoms across this film. The film thickness remains constant: it is believed that its growth rate at the C/TiC interface is equal to the dissolution rate at the melt/TiC interface. Thus, the TiC film shrinks to the center of the carbon particle as the latter dissolves. This pattern is sometimes called the shrinking-core mechanism. It corresponds to the solidsolid-liquid mechanism which, for the Ti-C system, is truly quasi-equilibrium. However, the film thickness has not been previously estimated using realistic diffusion data. The concentration profile of carbon in the reaction cell is similar to that shown in Figure 2.2 (a) but with R0=R0(t); final TiC particles precipitate from the melt in domain [R1(t), R2]. In a general case, the displacement of the melt/TiC and C/TiC interfaces is determined by Eqs. (2.25)-(2.27) with the only difference that Eq. (2.25) should be written in spherical symmetry. But since the TiC film thickness is small and DC>>DTi, outdiffusion of carbon atoms through the film is
33
not the rate-limiting stage. Thus the process is controlled by indiffusion of Ti atoms across the TiC layer, and radial shrinking of the film is described as
(1 c 23 )
dR 0 dt
= D Ti [T(t)]
c C r
R 0(t)
, R0(t=0) = R00, h0 = R1(t)R0(t) = const, (2.32)
where h0 is the layer thickness. For a thin film, a steady-state concentration profile can be used to determine the concentration gradient at r=R0(t) in Eq. (2.32): cC(r) = c023R0/r + (1 R0/r) (R1c021 R0c023)/h0, R0(t) r R1(t). (2.33) Using Ti defined by Eq. (2.28) and introducing z = R0/R00, from Eqs. (2.32),(2.33) we obtain:
c0 c0 dz 23 21 = . 0 0 2 d zR 0 /h 0 + 1 Ti (1 c 23 )(R 0 ) 0 z
(2.34)
By the attainment of the maximal SHS temperature TCS, which corresponds to time t=tCS, the carbon particle completely dissolves, i.e. R0(tCS)=0. Then, integrating Eq. (2.34) from 0 to tCS, we receive a non-linear equation linking the initial radius of the carbon particle R00 with the thickness of the TiC film
h0 R0 0
h 0 ln
R0 0
/ h 0 + 1 = Ti (t SHS )
)]
c0 c0 23 21 1 c0 23
(2.35)
The results of the numerical solution of Eq. (2.35) are shown in Figure 2.6. It is seen that the thickness of the TiC film for the initial radius of carbon particles R00=0.5 m is close to the crystal lattice period: h0=0.5 nm ~ aTiC=0.4327 nm [89], and still decreases with increasing R00. Hence the aforesaid quasiequilibrium solid-solid-liquid mechanism loses its physical meaning: a minimal thickness of a crystalline phase must be about the size of a critical nucleus which is typically of the order of 10 lattice periods.
34
B. B. Khina
Figure 2.6. Calculated thickness of a titanium carbide film vs. carbon particle radius for diffusion-controlled dissolution-precipitation (solid-solid-liquid) mechanism [60].
2.11. PHASE-FORMATION-MECHANISM MAP FOR NONISOTHERMAL INTERACTION IN THE TI-C SYSTEM

The above-presented consistent analysis of the solid-solid-liquid (diffusioncontrolled) mechanism, which was performed using available experimental data on both solid-state diffusion in the product phase and characteristics of the SHS wave, has demonstrated that this widely used concept is actually not applicable to modeling SHS of titanium carbide. This is because the physical meaning of the results obtained within this approach (e.g., the product structure and density) disagrees with experimental data. It is shown that formal calculation of the product-layer thickness and associated heat release for small particles of a nonmetallic reactant (for scenario 2) can bring about numerical data supporting the diffusion model. Thus the comparison of theoretical and experimental results should be performed using structural characteristics of SHS products, e.g., porosity. Therefore, only the solid-liquid mechanism, which for the Ti-C system is truly non-equilibrium, can operate during SHS to produce dense product particles.
35
It involves direct contact of solid carbon with molten titanium without the presence of a continuous TiC interlayer. The product formation occurs via dissolution of carbon in liquid titanium at the C/Ti(melt) interface and precipitation of TiC grains. Because of fast diffusion in high-temperature melts (DC(m)~105104 cm2/s), diffusion in liquid is not the rate-limiting stage and the phase-forming process responsible for major heat release is crystallization of the product particles. Because of the presence of the solid/liquid interface and strong chemical interaction between the C and Ti atoms, crystallization of the TiC particles will occur via heterogeneous nucleation at the C/Ti(melt) interface rather than through homogeneous nucleation in the melt. Besides, the activation energy for the former is generally lower than for the latter. The nucleated TiC grains must detach from the carbon particle surface, otherwise a thin TiC layer separating the parent phases will form and the situation will reduce to the solid-solid-liquid (quasi-equilibrium) pattern considered above. This process continues until complete consumption of solid carbon is achieved. Further, in the after-burn zone of the SHS wave, growth and coalescence of the TiC particles in the metallic melt can occur. In this case, the final size of product particles will depend on the conditions of crystallization and subsequent coalescence/sintering but not on the size of initial reactants. An important factor is the melt lifetime which depends on the Ti-to-C ratio in the charge, structure of a green pellet determining the melt spreading conditions and heat exchange with the environment. This solid-liquid mechanism qualitatively agrees with the results obtained in experiments on arresting SHS wave in the binary Ti-C [18] and multi-component Ti-C-Ni-Mo [19] systems where in rapidly quenched samples the formation of small uniform-sized TiC particles was observed in the molten metal around a graphite particle, which were apparently detaching from the surface of carbon particles [18]. This mechanism may also be valid for other interstitial compounds such as carbides, borides etc. for which the SHS temperature exceeds the melting point of a metallic reactant but is below the melting temperatures of the non-metal and product, and for certain intermetallics. In this situation, the critical thickness, hcr, of the layer of a primary product formed on the surface of a metal particle before the attainment of the melting temperature, which is determined by Eq. (2.20), acquires a precise physical meaning. This is a criterion for the changeover from the solid-solid-liquid (diffusion-controlled) mechanism (a slow route of product formation) to the solid-liquid mechanism (a fast rout). As stated above, in the wave propagation mode this thickness is small (h << hcr) due to high heating rate, and hence the changeover will always take place. But this criterion is important in the thermal explosion (TE) mode when a heating rate to the melting temperature of a lower-
36
B. B. Khina
melting reactant may be small, down to ~110 K/min [67,83], and hence a sufficiently thick case of the primary product can be formed on a metal-particle surface to prevent the liquid core from spreading. Basing on the above, a diagram of interaction routes can be constructed for non-isothermal heterogeneous interaction in the Ti-C system. Consider linear heating with rate vT, which corresponds to CS in the TE mode. Then is the time tm corresponding to the attainment of melting temperature of titanium is determined as tm = [Tm(Ti)T0]/vT. (2.36)
The thickness of a primary TiC case (see scenario 1) corresponding to the changeover of interaction pathways, is written as hTiC[(tm)] = hcr, (2.37)
where is determined by formula (2.10) and hcr is calculated according to Eq. (2.20). Solving Eq. (2.37) together with Eqs. (2.10)-(2.13) and (2.36) permits obtaining the aforesaid criterion for the changeover of interaction mechanisms at different heating rates vT and initial particle sizes of the metallic reactant (titanium) R00. A typical radius of titanium particles used in CS is 1-50 m; in calculations the R00 value was varied within 0.5-150 m. For numerical calculations, we use three data sets, viz. Nos. 1, 11 and 14 from Table 2.1 for the following considerations. As seen in Figure 2.4 (a), sets No. 1 and 11 give a reasonable thickness of the TiC layer grown on the titanium particle surface during heating in the SHS wave while data set No. 14 yields a maximal (probably overestimated) value. The results are presented in Figure 2.7 as a map of phase formation mechanisms in coordinates heating rate and initial radius of a titanium particle where lines 1-3 refer to different sets of the diffusion parameters in the titanium carbide. Parametric domain I, which lies below the line corresponding to the attainment of the critical thickness, hcr, of the primary refractory product (TiC), refers to the diffusion-controlled growth mechanism. This is a slow, quasiequilibrium route typical of diffusion annealing in weakly non-isothermal conditions. Here, a sufficiently thick shell of a primary refractory product (in our system, TiC) is grown of the metal particle surface during heating to Tm so that after melting the metallic reactant remains inside the shell (hTiC>hcr) and further
37
interaction (at T>Tm) proceeds slowly since the rate-limiting stage is still the solid-state diffusion across this spherical layer.
Figure 2.7. Diagram of phase formation mechanisms for synthesis of titanium carbide in non-isothermal conditions: domain I is the solid-state diffusion-controlled TiC growth, or slow route typical of furnace synthesis; domain II is the non-equilibrium crystallization mechanism, or fast route typical of CS. Calculated using different data sets for diffusion in TiC (see Table 2.1): data set No. 1 (line 1), No.11 (line 2) and No. 14 (line 3).
Thus, the models of CS employing this approach [62-66,68,69] are valid in this range of parameters (the heating rate and metal particle size). From Figure 2.7 it is seen that for small-sized metal particles (R000.5 m) the quasi-equilibrium diffusion-controlled growth of the refractory product can proceed at high heating rates typical of the SHS wave, vT~104-105 K/s. This agrees qualitatively with certain results observed during SHS in mixtures of small (nanosized) particles and in mechanically activated SHS [124]. Domain II, in its physical meaning, corresponds to a fast route typical of CS where the non-equilibrium dissolution-precipitation mechanism operates to provide fast completion of the reaction. Here, the refractory product layer formed during heating from T0 to the melting point of the metallic reactant is thin (hTiC<hcr) and cannot keep the melt inside the particle. Thus, melting of the metal core inevitably results in rupture of the primary shell and the melt comes into
38
B. B. Khina
direct contact with the solid non-metal particles, which is a necessary condition for the dissolution-precipitation interaction route to occur. In this parametric domain, the model [57-59] is applicable. The diagram clearly demonstrates the difference, in terms of reaction mechanisms, between CS (domain II), where fast conversion of reactants into the products takes place, and traditional furnace synthesis (domain I) where the interaction proceeds slowly because of a low heating rate and/or large particle size of the metallic reactant.
Chapter 3
MODELING INTERACTION KINETICS IN THE CS OF NICKEL MONOALUMINIDE

3.1. INTRODUCTION
The CS of nickel monoaluminide has gained much attention in literature [21,23,67] because this compound possesses a unique combination of strength properties and resistance to gas corrosion at high temperatures, and is used in a variety of applications as a structural material [125,126] and protective coating [127]. Besides, of substantial interest is CS in multilayer thin-film Ni-Al system where the layer thickness, h, varies from ~1 m to ~10 nm and in stacked foils with h~10-100 m [128]. The former process is used for joining of metallic glasses [129-131], welding of a pure NiAl layer to high-strength superalloys [132], welding/soldering of microscopic objects such as electronic components, and similar applications [133-135] while the latter can be used for near-net-shape manufacturing of NiAl articles [136]. Earlier [137], SHS in laminated Ni/Al foils was used for experimental modeling of the reaction mechanism in Ni-Al powder mixtures. In both cases, CS in this system is a subject of extensive experimental and theoretical investigation. However, an intricate physical mechanism of phase and structure formation during CS of NiAl is not well understood yet. It has been demonstrated experimentally that during SHS in powder mixtures [21,23] and in lamellar Ni-Al systems [136] with h~10-100 m the dissolution-precipitation (DP) mechanism takes place: at heating above the Al melting temperature solid nickel dissolves in liquid aluminum and NiAl grains crystallize from the supersaturated melt. This interaction route may have non-equilibrium nature since the Ni-Al phase diagram [138,139] contains four compound phases: NiAl3 (melts at 1127 K), Ni2Al3
40
B. B. Khina
existing at T<1406 K, NiAl whose congruent melting point is 1911 K and Ni3Al (exists at T<1668 K), and Ni-base solid solution (Figure 3.1).
Figure 3.1. The Al-Ni phase diagram [138,139] and the solubility limits (solid lines) used in modeling. [Reproduced with kind permission from Springer Science+Business Media: International Journal of Self-Propagating High-Temperature Synthesis, Modeling heterogeneous interaction during SHS in the Ni-Al system: a phase-formation-mechanism map, Vol. 16, 2007, pp. 51-61, B.B.Khina and B.Formanek, Figure 1, Copyright Allerton Press, Inc., 2007].
It is important to outline that in literature there is a controversy concerning the presence of solid interlayers of equilibrium phases between liquid Al and solid Ni during the formation of NiAl via this route. According to [21], dissolution of solid nickel in Al-base melt occurs through a continuous forming and dissolving of a reaction diffusion layer on the surface on the Ni particle, which corresponds to the equilibrium Ni-Al phase diagram. This standpoint is based on the classical theory of reaction diffusion in condensed phases [24,25,140] and implies local quasiequilibrium at phase boundaries. But in [23] it is concluded that above the melting point of NiAl3 the system consists of only solid nickel and liquid aluminum solution, i.e. a solid interlayer is absent and hence the formation mechanism of NiAl is substantially non-equilibrium. A criterion for transition between these two possible interaction routes is not known.
Modeling Interaction Kinetics in the CS of Nickel Monoaluminide
41
Besides, there is a difference in interaction patterns between SHS in thin (h<1 m) [129-132] and thick (h~10-100 m) [136] Ni-Al films: in particular, outflow of liquid aluminum through the butt end of stacked thick foils was observed experimentally [136], and the reasons for this behavior were not revealed. In modeling phase formation during CS in a particulate Ni-Al system in the TE mode [141] and during non-isothermal annealing at a prescribed temperature regime [142] using the diffusion-controlled growth kinetics, the choice of parameter values, viz. the activation energy E and preexponential factor D0 for diffusion, was not substantiated. The diffusion model was used for studying SHS in a layered Ni-Al system [143,144,145], but the values of macroscopic parameters, D0 and E, were obtained by fitting the calculated and experimental results of the SHS wave velocity and temperature profile instead of using relevant diffusion data for the solid product phase, NiAl, which are known in literature [146-156]. Hence these parameters refer not to diffusion in a particular phase but characterize the process as a whole, and their intrinsic physical meaning is vague. In modeling CS of NiAl in the TE mode it has been assumed that a solid nickel particle is separated from the Al-base melt by a thin NiAl interlayer of a constant thickness and diffusion across this film is a rate-limiting stage while the product grains precipitate from the melt [67]. This concept was first put forward in [122] and will be hereinafter called the quasi-equilibrium dissolutionprecipitation (QEDP) route. However, the values of E and D0 used in numerical simulation [67] were the fitting parameters but not real diffusion characteristics of NiAl. It should be noted that an important role of Ni melting in the SHS process, which was observed experimentally [137], is not considered in the above models. Basically, structure, composition and properties of the CS product depend to a large extent on the interaction pathway. Current situation in this area necessitates a new insight into the formation mechanism of NiAl in non-isothermal conditions. In particular, it seems important to determine criteria for changeover of reaction routes at high temperatures and a range of parameters where a quasi-equilibrium interlayer of a solid product can exist at the solid/melt interface. In connection with the above, the objective of this chapter of the book is to develop a model for phase formation in the Ni-Al system in substantially nonisothermal conditions. The model is based on the classical solid-state diffusioncontrolled growth kinetics taking into account the independently measured values of diffusion parameters in NiAl, the Ni-Al phase diagram and other properties of phases involved in interaction. The ultimate goal of simulation [157,158] is to determine the ranges of parameters, e.g., the heating rates and initial metal layer thickness, where this quasi-equilibrium approach is valid. Thereby, criteria for transition to uncommon reaction routes, e.g., the dissolution-precipitation
42
B. B. Khina
mechanism with or without a thin solid interlayer, will be found. Basing of the results of numerical calculations, a map of phase-forming mechanisms in nonisothermal conditions in this system will be constructed. As is known experimentally [46], the system geometry substantially influences heat transfer during SHS, which results in non-uniform interaction throughout the specimen. In a particulate system, interparticle contacts yield additional thermal resistance as compared with stacked foils, but in an elementary volume containing both of the reactants the interaction mechanism will be the same. Thus, without sacrificing the generality of approach, it makes sense to limit our consideration to the TE mode of CS where the effect of heat transfer inhomogeneity is eliminated.
3.2. PHYSICAL BACKGROUND OF THE MODEL

3.2.1. Thermal Aspect
First, lets calculate the adiabatic temperature, Tad, of reaction Ni + Al NiAl implying that the reactants are initially at the room temperature, 298 K. As the heat release of the reaction may be insufficient to melt all the product, the heat balance condition is written using the method proposed in [159]:
H 0 298 (pr) +
Tm (pr) 298
c p (pr, s)dT + I[Tad Tm (pr)]f m H m (pr) +
Tad
Tm (pr)
c p (pr, m)dT = 0
, (3.1)
where pr denotes the product (here NiAl), H0298(pr) = 118.4 kJ/mol and Hm(pr) = 62.8 kJ/mol are the standard formation enthalpy and melting heat of NiAl, correspondingly [160], cp(pr,s) and cp(pr,m) are the specific heat of solid (s) and molten (m) product, for NiAl cp(pr,s) = 41.8 + 13.8103T J/(molK) [160], Tm(pr)=1911 K is the NiAl melting point, fm is the mass fraction of melt, 0fm1, I is the asymmetric Heaviside step function: I(x) = 0 at x<0, I(x) = 1 at x0. Upon solving Eq. (3.1) numerically we obtain Tad=1911 K, i.e. the adiabatic combustion temperature coincides with the NiAl melting point, and the fraction of liquid phase at Tad is fm=0.42, that is solid NiAl constitutes 58% of the reaction product. In thin-film combustion in the Ni-Al system, the TCS value may be
43
substantially below Tad because of heat removal into the substrate, thus the formation of only solid final product, NiAl, during CS is possible. As is known [21,23,136,137], fast interaction during SHS in the given system begins after melt formation, i.e. above the aluminum melting temperature, Tm(Al)=933 K, or the eutectic temperature Teu(Al-NiAl3)=913 K (Figure 3.1). According to the Gibbs phase rule, in a binary system the contact of a solid phase layer (here pure nickel or Ni-base solid solution) with a two-phase region (here solid NiAl particles dispersed in Al-base melt at T<Tad) is forbidden; it can take place only in a strongly non-equilibrium situation when the solid product grains (here NiAl) crystallize from a supersaturated liquid phase. Since this research is focused on non-isothermal interaction kinetics in the condensed state, a quasi-equilibrium planar diffusion couple is considered: solid (s) or molten (m) Ni (phase 3)/solid NiAl (phase 2)/Al-base melt (phase 1) (Figure 3.2), which is continuously heated at a prescribed rate vT = dT/dt. This corresponds to a model situation where a unit bilayer thin-film sample without a substrate is heated in the conditions of a high heat exchange coefficient at the outer surface so that the heat released by exothermal reaction Ni+AlNiAl is removed rapidly. Similar experiments using electrical heating of thin wires in a computer-controlled setup were performed in the Ti-N [161] and Mo-Si [162] systems at vT = 101-105 K/s. Higher values, vT 105 K/s, are typical of selfheating during CS.
Figure 3.2. Schematic of the diffusion-controlled growth of the NiAl (phase 2) layer in the diffusion couple Al (phase 1)/Ni (phase 3) during CS. [Reprinted with permission from: B.B.Khina, Journal of Applied Physics, Vol. 101, No.6, 063510 (11 pp.), 2007. Copyright 2007, American Institute of Physics].
44
B. B. Khina
Hence, this situation is close to CS in the TE mode. In this case, the NiAl phase layer growth is due to solid-state interdiffusion across this layer. The role of interface kinetics, which was studied in [81], is ignored, and the Al concentrations at phase boundaries 1/2 and 2/3 correspond to equilibrium values.
3.2.2. Phase Composition of the Reaction Zone

The experimental data concerning the phase formation sequence in Ni-Al thin-film diffusion couples during isothermal annealing or upon heating at low rates, vT ~ 1 K/s, are contradictory. According to [163,164], the first phase to form is NiAl3, and the other equilibrium phases of the Ni-Al system can appear only after the NiAl3 layer exceeds a certain thickness or after complete consumption of one of the starting metals. The final phase composition corresponds to the initial Al-to-Ni thickness ratio of the films. However, in [165,166] the growth of B2NiAl with a metastable concentration of about 63 at.% Al as the first phase was observed. In [167], the formation of metastable -phase (Ni2Al9) during annealing of Ni/Al multilayers was detected, which quickly transformed into stable NiAl3 due to interaction with nickel. At high heating rates, which are typical of CS, the conditions for the first phase nucleation at the Ni/Al interface may be different from those at slow heating. For constructing the model, the following basic assumption is made. It is considered that a thin continuous NiAl layer nucleates at the initial Ni/Al interface at T0 = Teu(Al-NiAl3) = 913 K. During the growth of NiAl compound in the whole temperature range T0 T Tad=Tm(NiAl), the interlayers of other equilibrium phases of the Al-Ni system (Figure 3.1) [138,139], viz. NiAl3 (Tm=1127 K), Ni2Al3 (Tm=1406 K) and Ni3Al (Tm=1668 K), are absent, and metastable equilibria at interfaces Al-base melt/solid NiAl and NiAl/Ni(s) are described by the corresponding equilibrium solubility-limit lines, viz. lines GFE/LK and HIJ/ABC, respectively (Figure 3.1). The latter presumption is only for the sake of simplicity, in order to avoid cumbersome calculation of metastable phase equilibria. Since in intermetallic compounds adjacent to NiAl on the phase diagram, Ni2Al3 and Ni3Al, the diffusion coefficient is lower than in Ni-base solid solution and a fortiori lower than in liquid Al (phase 1), in this situation the phase 2 layer growth will occur faster than in the presence of all the equilibrium phase interlayers. Thus, within this model we will obtain an upper estimate for the NiAl layer grown due to solid-state diffusion across this phase and therefore an upper estimate for the conditions where this phase formation mechanism is possible. This is consistent with the idea of the ex contrario method.
45
It should be outlined that the above assumption concerning the contact of phases NiAl/Ni refers only to temperatures below the melting point of phase Ni3Al: at T > Tm(Ni3Al)=1668 K the phase boundary 2/3 is equilibrium and described by the corresponding solidus and liquidus lines, JD and CD (Figure 3.1). Similarly, regarding the Al(m)/NiAl interface, this assumption is valid only at T < Tm(Ni2Al3)=1406 K: above this point the concentrations at the phase boundary 1/2 correspond to the equilibrium solidus and liquidus lines ED and KD, respectively (Figure 3.1). As the interdiffusion coefficient in phase NiAl, which determines its growth, increases with temperature according to the Arrhenius law, in non-isothermal conditions corresponding to CS the growth of the NiAl layer will occur mainly at high temperatures, hence the effect of the above assumption on the phase 2 layer thickness will be minor. This is clearly demonstrated below by numerical simulation. For modeling, the aforementioned solubility-limit lines as well as the equilibrium liquidus and solidus lines are approximated by splines (solid lines ABCD, HIJD, LKD and GFED in Figure 3.1). In a narrow (60 K) temperature range Tm(Ni3Al)=1668 K < T < Tm(Ni)=1728 K there are 4 phases in equilibrium in this system: Al(m)/NiAl(s)/Ni(m)/Ni(s), while at T < Tm(Ni3Al) and T > Tm(Ni) there are three phases. So, to simplify calculations, it is assumed that Ni-base solid solution melts at 1668 K (a horizontal solid line in Figure 3.1). It is worth noting that after the numerical calculations within this model had been performed [157,158], an original experimental work appeared [168] that provided a more rigorous justification to the aforesaid assumption concerning the presence of only NiAl phase in the reaction zone between nickel and aluminum at high heating rates. SHS in 30 m thick Ni (30 nm)/Al alloy (70 nm) multilayer films, which propagated at the combustion velocity of 2.8 m/s, was studied by in situ time-resolved synchrotron X-ray microdiffraction (with a 60 m diameter incident X-ray beam) using an extremely fast pixel array detector. The thickness ratio of reactants corresponded to the stoichiometry of Ni2Al3. It has been revealed that NiAl appears as the first and only product of interaction between solid Ni and molten Al in the SHS wave where the heating rate was ~106 K [168]. The final product phase, Ni2Al3, was found to form behind the reaction wave front, i.e. during cooling, due to the peritectic reaction at a temperature below its melting point. Also, it has been demonstrated using differential scanning calorimetry (DSC) at the heating rate of 0.7 K/s that the phase formation sequence at low and high heating rates are different. At slow heating, metastable -Ni2Al9 is formed first and later converts into stable NiAl3 according to reaction Ni2Al9 + Ni 3NiAl3 [168], which corresponds to earlier experimental observations [167].
46
B. B. Khina
3.3. MODEL FORMULATION FOR NIAL FORMATION IN NONISOTHERMAL CONDITIONS

3.3.1. Structure of the Model
At a constant temperature, the phase 2 layer thickness, h2, changes with time according to the parabolic law h2(t) (Kpt)1/2 where Kp is the parabolic growthrate constant. In weakly non-isothermal conditions, the growth kinetics of solid NiAl is determined by the diffusion-type Stefan problem [24,25,140] where a number of parameters, e.g., diffusion coefficients, equilibrium concentrations at phase boundaries and phase densities, depend on temperature. According to the Al-Ni diagram (Figure 3.1), the solubility of Ni in phase 1 (Al-base melt) increases with temperature, the same refers to solubility of Al in phase 3 (Ni-base solid solution or melt). As the diffusion coefficient in a melt exceeds that in intermetallics by orders of magnitude, the characteristic time of concentration leveling along the 0x axis in phases 1 and 3 (Figure 3.2) is small. Since solid [at T<Tm(Ni)] or liquid nickel is separated from the Al melt by a solid product layer (NiAl), maintaining of equilibrium compositions of phases 1 and 3 at fast continuous heating is possible only via partial dissolution of phase NiAl. Thus, in strongly non-isothermal conditions the phase 2 layer thickness is determined by the competition of two processes: (i) its diffusion-controlled growth and (ii) dissolution in adjacent phases 1 and 3. The latter is especially important above the Ni melting temperature when the solubility of Al in phase 3 increases substantially (Figure 3.1). In non-isothermal conditions T=T(t), the total differential for the phase 2 layer thickness, h2(T,t), is written as
dh 2 ( t , T) =
h 2 t
dt +
h 2 T
dT ,
(3.2)
or, for the time interval [ti, ti+1] that corresponds to heating in temperature range [Ti, Ti+1], the expression for increment of h2 can be written in the integral form:
t i +1 ti
h 2 = h 2 (t i +1 , Ti +1) h 2 (t i , Ti ) =
h 2 t
dt +
Ti +1 Ti
h 2 T
dT .
(3.3)
47
In formulas (3.2) and (3.3), h2/t is the growth rate of phase 2 due to solidstate diffusion while h2/T is the rate of change of h2 due to variation of equilibrium concentrations c01(T), c021(T), c023(T) and c03(T) according to the AlNi phase diagram (Figure 3.1) [138,139]. Therefore, the model for phase 2 formation should be composed of two parts: (i) a quasi-isothermal (diffusion-controlled growth of NiAl) and (ii) nonisothermal submodel (dissolution of NiAl in neighboring phases due to inclination of the solubility lines). It should be noted that due to competition of two physical processes, the following tentative situations are possible depending on the heating rate and initial metal layer thickness: (1) the dissolution rate of phase 2 becomes comparable with its growth rate, i.e. the phase 2 thickness reaches a certain maximum and starts decreasing. This means that the rest of the product (solid NiAl grains) will be crystallizing from the melt, i.e. the quasi-equilibrium dissolution-precipitation mechanism (with the presence of the NiAl interlayer between phases 1 and 3) starts operating; (2) the NiAl film dissolves completely, h2=0. In the latter situation, two more cases are possible: (a) the NiAl layer disappears at a relatively low temperature, T<Tm(Ni), due to a low D2 value. Then at the non-equilibrium interface Al(m)/Ni(s) a thin continuous layer of phase 2 can nucleate to reestablish the thermodynamic quasi-equilibrium, and its further evolution will again be determined by the competition of growth and dissolution rates; (b) the temperature is high, T>Tm(Ni). In this case two melts, viz. the Al-base (phase 1) and Ni-base (phase 3), merge into one and it will be impossible that a solid NiAl layer forms to separate them once again. This will denote the onset of a nonequilibrium mechanism of NiAl formation through crystallization of the supersaturated melt. The conditions for the changeover of the product-phase formation mechanisms will be determined below by numerical simulation within the frame of the above-stated physical concept. As the concept is based on the fundamental postulate of quasi-equilibrium at phase boundaries, i.e. phases 1 and 3 should always be separated by the phase 2 interlayer, thereby the nonequilibrium interaction route during CS in the Ni-Al system, which was a subject of discussion in literature [21,23,122], will be proved theoretically ex contrario.
3.3.2. Quasi-Isothermal Submodel

The growth of a planar NiAl layer due to solid-state diffusion at a certain temperature T (Figure 3.2) is described by a one-dimensional diffusion-type Stefan problem. As is known experimentally [21,23,136], noticeable interaction
48
B. B. Khina
during CS in the Ni-Al system begins after aluminum melting. So we consider the temperature range from the eutectic point Teu(Al-NiAl3) = 913 K to the adiabatic temperature Tad = Tm(NiAl1.0) = 1911 K. The diffusion coefficients in phases 1 (Al-base melt) and 3 (Ni-base solid solution or melt) substantially exceed that in intermetallic NiAl (the estimates are given below), hence it is reasonable to consider that the concentration of Al in phases 1 and 3 is constant along the 0x axis and corresponds to the equilibrium value at a given temperature T (Figure 3.2); the latter is especially true above the nickel melting point. The diffusion-type Stefan problem for the growth of phase 2 includes the equation for diffusion mass transfer in phase 2
c 2 t
= D 2 (T )
2c2 x 2
(3.4)
and mass balance conditions: (i) at phase boundary 1/2 [Al(m)/ NiAl(s)]
0 [1 (T)c1 (T ) 2 c 0 (T)] 21
dX12 dt
= 2 D 2 (T )
c 2 x
X 21 ( t ) + 0
(3.5)
and (ii) at interface 2/3 [NiAl(s)/Ni(s,m)]

0 [ 2 c 0 (T ) 3 (T )c 3 (T)] 23
dX 23 dt
= 2 D 2 (T )
c 2 x
X 23 ( t ) 0
(3.6)
Here subscripts 1, 2 and 3 denote phases Al(m), NiAl(s) and Ni(s or m), correspondingly, X21(t) and X23(t) are the current coordinates of 1/2 and 2/3 phase boundaries, c2 is the aluminum concentration in phase 2, c01(T) and c03(T) are the compositions of phases 1 and 3 in equilibrium with phase 2 at current temperature, c021(T) and c023(T) are the equilibrium concentrations of Al in phase 2 at the 2/1 and 2/3 interfaces, i is the density of i-th phase, i=1-3, D2(T) is the interdiffusion coefficient in intermetallic NiAl. Hereinafter mass concentration of Al is used. Expressions (3.5) and (3.6) are written taking into account the difference in the density of contacting phases. Initial conditions to Eqs. (3.4)-(3.6) look as
Modeling Interaction Kinetics in the CS of Nickel Monoaluminide h1(t=0) = h01, h3(t=0) = h03, h2(t=0) = 0,
49 (3.7)
where h01 and h03 are the starting thicknesses of phases 1 [Al(m)] and 3 [Ni(s)] at T0 = Teu(Al-NiAl3) = 913 K. As information on the temperature dependence of the density of solid NiAl is not available in literature, in Eqs. (3.4)-(3.6) it is assumed constant: 2=6.02 g/cm3 [169]. For phases 1 (Al-base melt) and 3 (Ni-base solid or liquid solution) the density is expressed as
1 = c1 (T ) Al( m ) (T ) + [1 c1 (T )] Ni ( m ) (T ), 3 = c 3 (T ) Al(s, m ) (T ) + [1 c 3 (T )] Ni (s, m ) (T ) .
0 0 0 0
(3.8) The interdiffusion coefficient in NiAl (phase 2) is defined in the Arrhenius form D2(T) = D0exp[E/(RT)], (3.9)
where E is the activation energy and D0 the preexponent. The choice of parameter values for modeling is substantiated below. It should be noted that within the frame of the Stefan problem a situation may arise where the layer of one of the parent phases, 1 or 3, is spent completely. Then, on the external surface of phase 1 or 3 the second-kind boundary condition, J = D2c2/x = 0, is to be posed instead of Eq. (3.5) or (3.6), correspondingly. This formulation has been widely employed in modeling SHS using the solid-state diffusion-controlled growth kinetics [63,66]. However, in this case it will be impossible to detect the changeover of interaction mechanisms because after complete disappearance of any of the parent phases the diffusion-controlled growth of NiAl will still continue. So, hereinafter this trivial situation is not considered deliberately. The results presented below demonstrate that if the model were supplemented with such conditions, certain phenomena revealed by computer simulation would have remained unnoticed. For a thin planar NiAl (phase 2) film, a steady-state (i.e. linear) profile of Al concentration is typically assumed c2/x = (c023(T)c021(T))/h2, where h2(t)=X23(t)X21(t) is the NiAl layer thickness. Then from Eqs. (3.4)-(3.6) we obtain for given temperature T:
50
B. B. Khina
h2
dh 2 dt
1 1 = 2 D 2 (T )[c 0 (T ) c 0 (T )](k + k ), 21 23 23 21
0 0 0 0
k 23 = 2 c 23 (T ) 3 (T )c 3 (T),
k 21 = 1 (T )c1 (T ) 2 c 21 (T ) .
(3.10)
At T=const Eq. (3.10) has a simple analytical solution for a time interval [ti, ti+1]:
1 1 1 / 2 h 2 ( t i +1 ) = {h 2 ( t ) + 2 2 D 2 (T )t[c 0 (T ) c 0 (T )](k 23 + k 21 )} 2 i 21 23
, (3.11) where t = ti+1 ti is a time increment. Then from Eqs. (3.5), (3.6) and (3.11) the thickness of phases 1 (h1) and 3 (h3) is obtained: h1(ti+1) = h1(ti) h2(ti+1)k23(k21+k23)1, h3(ti+1) = h3(ti) h2(ti+1)k21(k21+k23)1. (3.12)
3.3.3. Non-Isothermal Submodel

To determine the dissolution rate of solid NiAl in adjacent phases, it is necessary to write down mass balance equations for Al at interfaces NiAl/Al(m) (2/1) and NiAl/Ni(s,m) (2/3) and the total mass balance equation in the system bearing in mind that phases 1 and 3 do not contact each other directly but are separated by the NiAl interlayer. As stated above, the interaction kinetics at phase boundaries is not the rate-limiting stage and diffusion rates in metallic melts (Al and Ni base) and Ni-base solid solution are substantially higher than in solid NiAl. Thus, during heating the equilibrium concentrations corresponding to a current temperature are established quickly throughout the whole thickness of phases 1 and 3 (Figure 3.2). This assumption is valid for a small change of temperature, i.e. calculation over the whole temperature range [T0, Tad] should be performed with a sufficiently small step T = Tj+1Tj. Consider the temperature dependence of the phase 2 layer thickness h2/T dh2/dT:
51
dh 2 dT
dX 21 dT
dX 23 dT
(3.13)
where X21 and X23 are the displacements of phase boundaries 2/1 and 2/3, correspondingly. Lets write the total mass balance at interfaces 2/1 and 2/3 during partial dissolution of NiAl in phase 1 (X21) and phase 3 (X23) taking into account that 2=const while the densities of phases 1 and 3 vary with temperature:
dX 21 dT
d(1h 1 ) dT
= 0 , 2
dX 23 dT
d ( 3 h 3 ) dT
=0.
(3.14)
For small T and therefore small X21 and X23 values, the Al mass balance equations at interfaces 2/1 and 2/3 are formulated taking into account the temperature variation of equilibrium concentrations c01, c023, c021 and c03:
d (c 0 21X 21 ) dT
0 d (1c1 h1 )
dT
= 0 , 2
d (c 0 23 X 23 ) dT
0 d( 3 c 3 h3)
dT
(3.15)
=0 .
Then from Eqs. (3.14), (3.15) the following relationships for X21 and X23 are obtained:
dX 21 dT dX 23 dT
0 2 (c 0 21 c1 ) + X 21 2
dc 0 21 dT dc 0 23 dT
+ 1h 1
0 dc1
dT
0 dc 3
=0,
2 (c 0 23
0 ) + X 23 2 c3
+ 3h 3
dT
=0.
(3.16)
Hence for temperature range [Tj, Tj+1], i.e. for a regular step T, from Eqs. (3.13) and (3.14) we have h2(Tj+1) = h2(Tj) + X21(Tj+1) + X23(Tj+1), h1(Tj+1) = (h1(Tj)1(Tj) X21(Tj+1)2)/1(Tj+1),
52
B. B. Khina h3(Tj+1) = (h3(Tj)3(Tj) X23(Tj+1)2)/3(Tj+1). (3.17)
System of equations (3.16) can be rewritten in a more convenient form
d X 2 i dT
(T ) h (T ) dc i0 X 2i dc i0 dc 0 , i=1,3, (3.18) 2i = i 1 i 1 dT (c 0 c 0 ) dT dT c i0 c 0 2i 2 i 2i
which admits an analytical solution. Having initial conditions X21(T0)=0, X23(T0)=0 where T0 = Teu(Al-NiAl3) = 913 K, the solution for temperature range [Tj, Tj+1] is
1 X 2i (T) = i (T )
Tj+1 Tj
Q i ()Fi ()d , Fi (z) = exp Pi ()d , i1,3, (3.19)

Tj
where Fi is the integrating factor and parameters Pi and Qi are defined as

Pi =
(T )h (T ) dc 0 dc 0 dc 0 i , i1,3. i 2i , Q = i j i j i 0 0 0 0 dT dT dT c i c 2i 2 (c i c 2i )
1
(3.20)
On each step over temperature the values of X21 and X23 are found from Eqs. (3.19), (3.20) where the integrals can be evaluated numerically, and then the current thickness of phases 1-3 is determined by formulas (3.17). As mentioned above, at sufficiently fast heating during CS a situation is possible where at relatively low temperatures, T<Tm(Ni), the dissolution rate of NiAl in adjacent phases turns out to be higher than its growth rate, and phase 2 disappears. As we consider a quasi-equilibrium situation when the interaction of liquid phase 1 with solid solution 3 occurs only through phase 2, in such a case we imply the presence of a zero-thickness phase 2 interlayer, i.e. the interface concentrations of phases 1 and 3, c01 and c03, correspond to their equilibrium solubility limits (Figure 3.1). As previously, this assumption is made in order to avoid calculation of metastable phase equilibria, and its effect on the final outcome of simulation is insignificant. In this transient situation, from the mass balance condition at the 1/3 interface [similar to Eqs. (3.14) and (3.15)] the following expressions for the thickness of phase 1 and 3 layers are obtained for heating in temperature range [Tj, Tj+1]:

h 1 (T j )1 (T j )[c1 (T j ) c 3 (T j+1 )] + h 3 (T j ) 3 (T j )[c 3 (T j ) c 3 (T j+1 )] , 1 (T j+1 )[c1 (T j+1 ) c 3 (T j+1 )]
0 0 0 0 h 3 (T j ) 3 (T j )[c1 (T j+1 ) c 3 (T j )] + h 1 (T j )1 (T j )[c1 (T j+1 ) c1 (T j )] . 0 0 3 (T j+1 )[c1 (T j+1 ) c 3 (T j+1 )] 0 0 0 0 0 0
53
h 1 (T j+1 ) =
h 3 (T j+1 ) =
(3.21)
3.3.4. Sequence of Calculations

We study linear heating of a Ni-Al bilayer, which may constitute a unit element of a multilayer Ni-Al film, at rate vT=dT/dt=const in temperature range [T0=913 K, Tad=1911 K]. This corresponds to the TE mode of CS. Then, introducing small steps over temperature T and time t=T/vT, the heating process is divided into a large number of identical stages: (i) isothermal soaking at temperature Tj for time t, and (ii) instantaneous heating to temperature Tj+1 = Tj + T. In the isothermal stage, the thickness of phases 1-3 is calculated by Eqs. (3.11), (3.12). During the heating stage, the displacement of phase boundaries due to variation of the solubility limits is calculated by Eqs. (3.19), (3.20) and (3.17) or Eqs. (3.21). The calculation is performed until complete consumption of one of the parent phases, 1 or 3, is achieved. As the ultimate goal of modeling is to determine the parametric domain where the growth of phase 2 can occur via the solid-state diffusion mechanism, for each initial thickness of the Al layer, h0Al, the heating rate vT is varied to reveal the occurrence (or nonoccurrence) of two physical situations: (1) the onset of shrinkage of phase 2, and (2) complete disappearance of the NiAl film at T>Tm(Ni). Since in SHS the heating rate reaches 105max 106 K/s [3-5] in simulation the vT value was varied within 51005106 K/s. The initial mass ratio of pure Al to Ni corresponds to stoichiometry NiAl1.0. So, in modeling only the initial thickness of the aluminum layer at room temperature (298 K), h0Al, was specified. Then the thickness of the initial Ni layer is calculated by a simple formula h0Ni = h0Al(0Al/0Ni)x0/(1x0), where 0i is the density of pure i-th metal (iAl,Ni) at 298 K, 0Al=2.71 g/cm3 and 0Ni=8.96 g/cm3 [170], and x0=0.314 is the mass content of Al in NiAl1.0. As noted above, interaction during CS in the given system begins at T0=Teu(Al-NiAl3)=913 K. Then the thickness of phase 1 (Al-base melt), h01, and phase 3 (Ni-base solid
54
B. B. Khina
solution), h03, at T0 is determined, similarly to Eq. (3.21), from the condition of mass balance of both Al and Ni:
0 = h0 h1 Al
0 Al
0 x 0 c3 (T0 )
1 (T0 ) x [c 0 (T ) c 0 (T )] 0 1 0 3 0
0 , h 0 = h 0 Al 3 Al
0 c1 (T0 ) x 0
3 (T0 ) x [c 0 (T ) c 0 (T )] 0 1 0 3 0
(3.22) where the densities of phases 1 and 3, 1(T0) and 3(T0), are defined by formulas (3.8). For numerical modeling it is necessary to know (i) the temperature dependence of density of Al and Ni in order to evaluate the densities of phases 1 and 3 by Eq. (3.8), and (ii) diffusion parameters E and D0 for NiAl to calculate the interdiffusion coefficient D2(T) by formula (3.9).
3.4. PARAMETER VALUES FOR MODELING INTERACTION IN THE NI-AL SYSTEM

3.4.1. Densities of Phases
Formula (3.8) includes the density of liquid aluminum Al(m)(T) and that of solid and liquid nickel Ni(s,m)(T). The densities of liquid metals (in g/cm3) are determined as [104] Al(m)(T) = 2.38 Al(T Tm(Al)), Al = (2.4-4.0)104, Ni(m)(T) = 7.90 Ni(T Tm(Ni)), Ni = (8.7-12.5)104. (3.23)
For calculating the density of liquid phases 1 and 3, minimal values Al=2.4104 and Ni=8.7104 were taken. At that, according to formulas (3.8) and (3.23), the density of molten NiAl1.0 (31.4 wt.% Al) at Tm(NiAl1.0)=1911 K is 5.95 g/cm3, i.e. less than the density of solid nickel aluminide 2=6.02 g/cm3 [169] (at higher Al and Ni values the density of liquid NiAl1.0 appears to be higher than that of solid NiAl1.0). To calculate the density of solid phase 3 [Ni-base solid solution at T<Tm(Ni)] by formula (3.8), the temperature dependence of density for pure Ni is estimated
55
using the volumetric expansion coefficient = (1/V)V/T where V=1 is the specific volume, Ni = 3.9105 K1 [170]. Then Ni(s)(T) = Ni(s)[Tm(Ni)] exp{Ni[TTm(Ni)]}, (3.24)
where Ni(s)[Tm(Ni)]=8.10 g/cm3 is the density of solid Ni at its melting point [171]. For Al dissolved in phase 3, the value at a room temperature Al(s)=2.71 g/cm3 [170] was used.
3.4.2. Interdiffusion Parameters

Diffusion in the Ni-Al system and, particularly, in intermetallic compound NiAl has been extensively studied [146-156]. The interdiffusion coefficient in NiAl, D2, was measured experimentally during annealing of diffusion couples at T=650-1300 C [146,147], aluminizing of nickel at T=870-1000 C [148], and reaction diffusion in thin films at low temperatures T=250-440 C [149]. At T=1173-1473 K, the activation energy for interdiffusion in NiAl is E=200 kJ/mol for Al concentration in the NiAl phase xAl40 at.%, and E250 kJ/mol for xAl=4350 at.% [150]. The partial diffusivity of Ni atoms measured experimentally at T=1273-1623 K [151], T=1050-1550 K [152], and T=1273 and 1523 K [153] appears to depend strongly on the composition of NiAl: in stoichiometric NiAl1.0 it is considerably lower than near the upper and lower solubility limits. The concentration dependence of the interdiffusion coefficient in NiAl was measured at T=1223-1773 K [154] but the obtained values appeared to be substantially smaller than those reported in other works [146-148,150,155,156]. Here for numerical modeling we use interdiffusion parameters in NiAl obtained for xAl=38 at.% in the temperature range T=1200-1600 K [155] and for xAl=54 at.% at T=1123-1323 K [156]. The values of preexponent D0 and activation energy E retrieved from the Arrhenius plots log(D2) vs. T1 presented in the cited papers (Figure 10 in [155] and Figure 13 in [156]) are listed in Table 3.1 as data sets No. 1 and No. 2. Since the interdiffusion coefficient for nonstoichiometric NiAl is higher than for NiAl1.0, simulation with these parameters will give an upper estimate for the phase 2 layer growth via the diffusioncontrolled kinetics, which goes in line with the ex contrario method. The calculated D2 values extrapolated to the whole temperature range of CS, T=9131911 K, are shown in Figure 3.3; lines 1 and 2 corresponding to these diffusion data sets intersect at T=1667 K giving D2107 cm2/s.
56
B. B. Khina
Figure 3.3. Temperature dependence of the interdiffusion coefficient in NiAl (D2) according to experimental data [155,156] extrapolated to the temperature range of SHS (solid lines), and according to the parameters D0 and E used in modeling SHS of NiAl by other authors (thin solid, dashed and dash-dot lines). Numbers at lines correspond to the diffusion data set numbers in Table 3.1. Line 7 depicts the interdiffusion coefficient in Ni2Al3 [174] used for modeling SHS in [173].
Table 3.1. Interdiffusion data for NiAl (lines 1-6) and Ni2Al3 (line 7) used in this work (lines 1 and 2) and by other authors (lines 3-7) for modeling phase layer growth during non-isothermal interaction in the Ni-Al system
Data set number 1 2 3 4 5 6 7 D0, cm2/s 1.0 4.75104 1.0103 1.0103 2.18102 34.42 0.1704 E, kJ/mol 222.0 115.9 100 92.05 137 316.8 148.1 xAl, at.% 38 54 42 Data source diffusion experiment, T=1200-1600 K [155] diffusion experiment, T=1123-1323 K [156] modeling CS in the TE mode [141] modeling non-isothermal annealing [142] modeling thin-film CS (fitting parameters) [144,145] diffusion experiment, T=1273-1773 K [154] experimental data for Ni2Al3 [174]
57
At the NiAl melting point, Tm(NiAl)=Tad=1911 K, D2=8.6107 cm2/s for data set No. 1 and 3.2107 cm2/s for data set No. 2 (Table 3.1). Let us compare the obtained values with diffusion coefficients in the melt at the same temperature. For close atomic radii ai (0.143 nm for liquid Al and 0.124 nm for molten nickel [106]), a plausible estimate is given by the Sutherland-Einstein formula (2.6) with n=4. For liquid Ni, m=7.90 g/cm3 at its melting point [104] and the kinematic viscosity is =0.4102 cm2/s at T=1923 K [106]. Then for diffusion of Ni atoms in the Ni-base melt at 1911 K formula (2.6) gives DNi(m)=5.4105 cm2/s, while for diffusion of Al atoms DAl(m)=4.6105 cm2/s. This is two orders of magnitude higher than the interdiffusion coefficient in solid NiAl at the same temperature. Therefore, data sets No. 1 and 2 for diffusion in NiAl (see Table 3.1) give reasonable values of D2 in the whole temperature range of CS. The parameters E and D0 used in modeling SHS in a particulate Ni-Al system in the TE mode [141] and in simulation of non-isothermal annealing [142] are presented in Table 3.1 as data sets No. 3 and 4, respectively. However, the choice of these values in [141,142] was not justified. In modeling SHS in multilayer NiAl thin films [144,145], the values of E and D0 for diffusion in NiAl were obtained by fitting experimental and numerical results of the combustion temperature and velocity (data set No. 5 in Table 3.1). The corresponding diffusion coefficients are presented in Figure 3.3 for the sake of comparison (lines 3 through 5). It is seen that in the temperature range of CS data sets Nos. 3-5 give a substantially overrated diffusion coefficient in NiAl than the upper estimates of the corresponding experimental parameter: D2 exceeds the values given by data set No. 1 (line 1 in Figure 3.3) by 1 to 3 orders of magnitude and those calculated using data set No. 2 (line 2) by one order of magnitude. As noted previously, at T=const the phase layer growth via diffusion mechanism follows the parabolic law h2 [Kp(T)t]1/2, where Kp is connected with the interdiffusion coefficient D2 in the growing phase [see Eqs. (3.10), (3.11)]. In non-isothermal conditions T=T(t), this dependence can be written in the integral form assuming that equilibrium concentrations at phase boundaries are constant:
t h h 2 D 2[T(t)]dt 0
1/ 2
(3.25)
where th~100102 s (depending on the system) is the characteristic time of reactants heating in the SHS wave from T0 to Tad. The heat release, which is
58
B. B. Khina
responsible for self-heating during CS, is proportional to the product phase-layer thickness h2, and hence the adiabatic temperature rise, Tad=TadT0, is estimated as
t 0 H 298 (NiAl)S h T ad D 2[T(t)]dt Vc p (NiAl) 0 1/ 2
(3.26)
where S in the total Ni/Al interface area and V is the specimen volume; here the temperature dependence of heat capacity, cp(NiAl), is neglected. From Eq. (3.26) it is seen that raising the interdiffusion coefficient by the factor of only 3 increases Tad by the factor of 1.73. Therefore, to attain a fair agreement between the calculated and experimental temperature of CS, diffusion coefficient in the growing phase has to be substantially overestimated, otherwise within the diffusion-controlled-growth kinetic model the combustion will be impossible. This is a possible reason of why in many modeling attempts employing this kinetic approach the diffusion parameters E and D0 were either obtained by data fitting [144,145] or chosen without any justification [141,142] instead of using experimental data on solid-state interdiffusion in the product phase. In a recent modeling attempt employing the diffusion-controlled growth concept [172,173], it has been assumed that Ni2Al3 intermetallic phase forms and grows rapidly between the bilayers of Al and Ni following ignition until the temperature reaches 1406 K (i.e., the melting temperature of Ni2Al3) [172, page 2]. However, no experimental evidence concerning the presence of this phase in the SHS wave in multilayer Ni-Al films was presented [172,173]. Let us analyze this work in more detail. As outlined above, it is the NiAl phase that forms and grows at the Ni/Al interface at a high heating rate in the CS wave [168], the maximum temperature observed experimentally in thin-film SHS is about 1650 K, and phase Ni2Al3 appears only behind the reaction front during cooling (see Figure 2(b) in [168]). The diffusion data for numerical calculations in [173] were the following: for NiAl, the values of E and D0 were taken from [154] at the aluminum concentration in this phase xAl=42 at.% (see line 6 in Table 3.1), and for Ni2Al3 the experimental parameters obtained in [174] were used (line 7 in Table 3.1). The calculated interdiffusion coefficients are shown in Figure 3.3: line 6 for NiAl with xAl=42 at.% in the whole temperature range of CS and line 7 for Ni2Al3 in the range [T0, Tm(Ni2Al3)]. It is seen that Ni2Al3 possesses a very high interdiffusion coefficient while that for NiAl with the chosen composition is extremely small. In modeling [173], phase Ni2Al3 grows fast after ignition and at
59
Tm(Ni2Al3)=1406 K converts into NiAl according to the inverse peritectic reaction Ni2Al3 2NiAl + Al(m). After that, growth of the NiAl layer proceeds very slowly because of a low value of D2. Thus, the calculated combustion temperature, TCS, does not exceed 1408 K (see Figures 3, 4 and 7(d-f) in [173]), i.e. the effect of the diffusion-controlled growth of NiAl in the SHS wave on the heat release and hence combustion temperature is insignificant [see formulas 3.25),(3.26)]; almost all heat release is due to Ni2Al3. From the above considerations it follows that the underlying idea in model [173] is to choose arbitrarily a phase having a high interdiffusion coefficient (here Ni2Al3) so that its fast growth in non-isothermal conditions could provide sufficient heat release to sustain the rapidly moving combustion wave, instead of using the real phase composition of the reaction zone (NiAl as the only growing phase [168]) and performing a comparison of the calculated and experimentally observed values of TCS. In the latter case, CS within the frame of this model would be impossible because of a low value of diffusion coefficient in NiAl. The above considerations once again testify to the topicality of the subject of this research, namely determining the applicability limits of the diffusioncontrolled growth approach to modeling heterogeneous interaction in nonisothermal conditions in regard to the combustion synthesis of intermetallic materials.
3.5. NUMERICAL MODELING OF NIAL FORMATION IN NONISOTHERMAL CONDITIONS

3.5.1. Evolution of the Phase Layers
Figure 3.4 displays an example of numerical modeling of phase 2 growth during linear heating of a thin-film Al-Ni diffusion couple at a small initial thickness of the pure aluminum layer, h0Al=0.1 m, using diffusion data set No. 1 for NiAl (Table 3.1). In the range of heating rates from 50 to 7.5105 K/s the conversion of initial metals into the product, NiAl, occurs by the quasiequilibrium solid-state diffusion mechanism. With increasing vT the temperature Tfin at which the process accomplishes, i.e. one of the parent phase layers (1 or 3) is completely consumed, rises, but it is still below the melting temperature of phase 3. At that, the Al layer is consumed earlier than Ni (lines 1-4 in Figure 3.4), but at the heating rate of 7.5105 K/s phase 1 and 3 end simultaneously (lines 5 in Figure 3.4); obviously, this corresponds to a maximum thickness of phase 2.
60
B. B. Khina
Figure 3.4. Change of the layer thickness of phases 1 (Al-base melt), 2 (NiAl) and 3 (Nibase solid or liquid solution) vs. temperature for initial aluminum layer thickness h0Al=0.1 m at linear heating with different rates: vT=50 K/s (line 1), 5102 K/s (2), 5103 K/s (2.1), 7.5104 K/s (4), 7.5105 K/s (5), 5106 K/s (6), and 8.5106 (7). Calculations are performed with diffusion data set No. 1 for phase NiAl (Table 3.1). [Reprinted with permission from: B.B.Khina, Journal of Applied Physics, Vol. 101, No.6, 063510 (11 pp.), 2007. Copyright 2007, American Institute of Physics].
At higher heating rates, phase 1 again ends earlier than phase 3, and the NiAl layer thickness, which is attained by the time of disappearance of the Al layer, decreases. For vT=5106 K/s, temperature Tfin exceeds the Ni melting point (lines 6 in Figure 3.4). In this case, the jags observed on curves h2(T) and h3(T) and the inflection point on line h2(T) are connected with a stepwise increase of the solubility limits due to the melting of Ni (see Figure 3.1). However, at a somewhat higher heating rate vT=8.5106 K/s (lines 7 in Figure 3.4) the phase layers 1 and 3 yet another time end simultaneously, and the thickness of NiAl has a second maximum (lines 7 in Figure 3.4). Only at vT=1107 K/s, which is substantially above a typical heating rate in the SHS wave (105-106 K/s), the phase 2 layer thickness starts decreasing; this is not shown in Figure 3.4 merely to avoid encumbering. The observed behaviors are connected with two reasons [see Eqs. (3.5), (3.6) and (3.16)]: (i) different displacement rates of interfaces 1/2 and 2/3, which are due to the asymmetry of solubility-limit lines in the Ni-Al phase diagram (Figure
61
0 (T ) , 3.1), i.e. different temperature dependencies of boundary concentrations c 1 0 0 c0 21 (T ) , c 23 (T) and c 3 (T ) , and (ii) competition between the diffusion-
controlled growth of NiAl and its dissolution in neighboring phases 1 and 3. It should be noted that the first maximum in the phase 2 layer thickness (lines 5 in Figure 3.4), which is observed at Tfin<Tm(phase 3), does not imply any change in the interaction mechanism. But the second maximum observed at Tfin>Tm(phase 3) (lines 7 in Figure 3.4) denotes the situation when the dissolution rate of NiAl in neighboring melts becomes comparable with its growth rate, i.e. the dissolutionprecipitation mechanism of product formation starts operating, parallel with the diffusion-controlled growth of the continuous NiAl layer. Numerical simulation has shown that for any initial Al layer thickness, h0Al, there always exists a certain critical heating rate, v T , at which this phenomenon is observed. Hence, for a small thickness of the initial metal layers (h0Al0.1 m) the diffusion rate in solid NiAl can provide almost complete conversion of Al and Ni into NiAl via the solid-state diffusion-controlled growth mechanism even at high heating rates that exceed the values typical of an SHS wave. This outcome of numerical modeling demonstrates that the idea expressed qualitatively in [175] about some unusual shear, or martensitic mechanism of atomic mass transfer across a continuous NiAl layer, which is allegedly responsible for fast growth of the reaction product during SHS in a thin-film Ni/Al system, is physically meaningless (in the sense of Occams razor): the reaction can well be accomplished in a very short time via the known solid-state diffusion-controlled route. Martensitic structure of the final product observed experimentally [175] for a certain composition of NiAl, viz. 63.8 at.% Ni which is close to that corresponding to diffusion data set No. 1 (Table 3.1), could most probably be formed during fast cooling of hot NiAl in the after-burn zone of the SHS wave due to heat removal into the substrate. This agrees with earlier results of the same authors [176]: the SHS wave temperature in a thin-film Ni-Al system with h0Al=30-100 nm exceeded the melting point of Al and fast cooling after the completion of high-temperature interaction was observed. At a larger initial thickness of pure Al, increasing the heating rate results in a changeover of interaction mechanisms, which occurs at T>Tm(Ni). Simulation for h0Al=10 m was performed using different diffusion data sets for NiAl (Table 3.1): No. 1 [Figure 3.5(a)] and No. 2 [Figure 3.5(b)]. At relatively small values of vT, below 500 K/s, the system behavior [lines 1 and 2 in Figure 3.5(a)] is similar to that shown in Figure 3.4 for heating rates at which the maximal temperature did not exceed Tm(Ni). The above described situation when the phase 2 layer
( cr ,1)
62
B. B. Khina
thickness attains its first maximum [at T<Tm(Ni)] occurs at vT80 K/s (not shown in Figure 3.5(a) to avoid superimposition of lines).
Figure 3.5. Change of the layer thickness of phases 1 (Al-base melt), 2 (NiAl) and 3 (Nibase solid or liquid solution) vs. temperature for initial aluminum layer thickness h0Al=10 m at linear heating with different heating rates: (a) diffusion data set No. 1 for phase NiAl (Table 3.1): vT=10 K/s (line 1), 100 K/s (2), 1103 K/s (2.1), 2103 K/s (4), and 3104 K/s (5) [Reprinted with permission from: B.B.Khina, Journal of Applied Physics, Vol. 101, No.6, 063510 (11 pp.), 2007. Copyright 2007, American Institute of Physics]; (b) diffusion data set No. 2 (Table 3.1): vT=50 K/s (line 1), 175 K/s (2), 400 K/s (2.1), 1103 K/s (4), and 5103 K/s (5) [Reproduced with kind permission from Springer Science+Business Media: International Journal of Self-Propagating High-Temperature Synthesis, Modeling heterogeneous interaction during SHS in the Ni-Al system: a phaseformation-mechanism map, vol. 16, 2007, pp. 51-61, B.B.Khina and B.Formanek, Figure 3, Copyright Allerton Press, Inc., 2007].
63
However, at a heating rate of about 1103 K/s, which slightly exceeds the vT value at which the second maximum in the phase 2 thickness is observed (above the phase 3 melting temperature), the system behavior changes [lines 3 in Figure 3.5(a)]. The jags on lines 3 are due to melting of phase 3 at 1668 K. At T=1836 K the thickness of phase 3 (Ni-base melt) attains a minimum and starts increasing while the NiAl layer continues growing, and this process lasts till complete consumption of phase 1. This means that the growing phase 2 layer displaces as a whole, and the rest of the reaction product can form only by crystallization of the Ni-base melt. At still faster heating, vT=2103 and 3104 K/s [lines 4 and 5 in Figure 3.5(a)], the phase 2 layer growth proceeds much slower, which is due to a short diffusion time in non-isothermal conditions. By the time when phase 3 melts, the thickness of NiAl ranges from several micrometers to ~0.1 m depending of the heating rate. In this case, during the further temperature rise there exists a thin solid NiAl layer separating two liquid phases, 1 and 3. For a given heating rate, the phase 2 thickness is either almost constant or slightly decreases with rising temperature, while the thickness of Al-base melt (phase 1) decreases and that of phase 3 increases, i.e. boundaries 2/1 and 2/3 move in the same direction with almost same velocity [lines 4 and 5 in Figure 3.5(a)]. The process continues until phase 1 disappears. This phenomenon is conditioned by the fact that at a high temperature the thin NiAl layer is permeable for diffusion of Al and Ni atoms and, as a result, the film displaces as a whole. The NiAl thickness is small: h20.5 m at vT=1104 K/s and ~0.1 m at vT=(2.5-4.0)104 K/s. In this situation, the reaction product can form only by precipitation of NiAl grains from the Ni-base melt (phase 3) during displacement of the thin continuous NiAl film. Thus, it is unambiguously demonstrated that the QEDP mechanism (via diffusion across a thin equilibrium film separating the parent phases) of CS, which was used in modeling on a merely conceptual basis [122,67] can really take place in a certain range of heating rates and initial metal layer thicknesses. In this regime, the direction of displacement of the thin solid-product interlayer (in this particular system, towards point x=0 in Figure 3.2) depends solely on functions
0 (T ) , c 0 (T ) , c 0 (T) and c 0 (T ) [see Eqs. (3.16)], i.e. on the shape of c1 21 23 3
liquidus and solidus lines of the phase diagram (Figure 3.1). Earlier it has been proved [60] that this mechanism cannot operate during SHS in the Ti-C system because of a low value of the diffusion coefficient in titanium carbide: the thickness of the TiC interlayer between liquid Ti and solid C will be below the characteristic size of a critical nucleus. In the Ni-Al system this interaction pattern is possible: at vT=5104 K/s the minimal NiAl film thickness is
64
B. B. Khina
h2min60-30 nm, i.e. (160-80)aNiAl where aNiAl=0.373 nm is the NiAl lattice period [42], which exceeds the typical critical nucleus size hcr~10aNiAl. However, at this high heating rate the NiAl interlayer exists only in a limited temperature range after phase 3 melts. Note that in the given system this route cannot take place below the Ni melting temperature and in this model the elastic stresses that can arise in the phase 2 interlayer and lead to its rupture are not accounted for. A further increase in the heating rate (above 4.25104 K/s) brings about a situation where, at the attainment of a certain temperature Td, the phase 2 layer dissolves completely while phase 1 (Al-base melt) still exists. This transition occurs on a narrow temperature range, Td1 K, which substantially exceeds the step over temperature (T0.2 K) used in modeling, i.e. the calculation accuracy was sufficient. The value of Td almost does not change with increasing vT and is about 1860 K. Apparently, this is connected with the shape of liquidus and solidus lines on the Ni-Al phase diagram (Figure 3.1): at T>1840 K they quickly converge to the NiAl melting point. In this situation, liquid phases 1 and 3 inevitably merge together and hence the only possible product formation mechanism is crystallization of NiAl grains from the melt. As mentioned above, this is a substantially non-equilibrium route. For any initial thickness of the Al , when the thin layer h0Al there must exist a certain critical heating rate, v T NiAl interlayer disappears at T>Tm(Ni), i.e. a transition the non-equilibrium reaction pathway occurs: e.g., v T 4.25104 K/s for h0Al=10 m and v T =350 K/s at h0Al=100 m. Note that the uncommon interaction routes, viz. QEDP via a thin NiAl film and non-equilibrium crystallization of NiAl grains, take place after phase 3 melts,
0 (T ) , c 0 (T ) , c 0 (T) and c 0 (T ) used in i.e. when the functions c 1 21 23 3 simulation correspond to the real phase diagram (Figure 3.1). This confirms the appropriateness of the assumptions made in this model. Simulation using diffusion data set No. 2 (Table 3.1) for the same h0Al value gives similar results [cf. Figures 6(a) and (b)]. The main difference is that the
( cr , 2 ) ( cr , 2 ) ( cr , 2 )
heating rates corresponding to the onset of decrease of the NiAl thickness, v T

( cr , 2 )
( cr ,1)
, and to complete dissolution of phase 2, v T , displace towards lower values. This is due to a smaller D2 values at T>1667 K (Figure 3.3). Besides, the above described case, when at T<Tm(Ni) phases 1 and 3 are consumed simultaneously and thus the h2fin value attains its first maximum, does not occur here; the latter is connected with a higher values of D2 at relatively low temperatures.
65
3.5.2. Estimation of Critical Heating Rates

To correctly determine critical heating rates v T and v T for any thickness of the initial Al layer, the following graphs have been plotted using the results calculated for two diffusion data sets (Table 3.1): the dependence on heating rate of the final thickness of phase 2, h2fin, which is attained by the instant of complete consumption of any of the parent phases (Figure 3.6), and the corresponding temperature value, Tfin, vs. vT (Figure 3.7). The first maximum in lines h2fin(vT) denotes the situation when phase layers 1 and 2 are exhausted simultaneously [Figure 3.6(a)] at T<Tm(phase 3), and in lines Tfin(vT) it is seen as an inflection point [Figure 3.7(a)]. For diffusion data set No. 2 this feature is much less pronounced [Figure 3.6(b)], and the inflection point in curves Tfin(vT) is absent [Figure 3.7(b)]. The second, sharp maximum in all the curves in Figure 3.6(a) corresponds to v T , and in Figure 3.7(a) it is seen as the maximum point. For data set No. 2 this maximum is dome-shaped [Figure 3.6(b)] and in lines Tfin(vT) it corresponds to the transition point to the horizontal branch [Figure 3.7(b)]. These behaviors are determined by different values of D2 (see Figure 3.3). The endpoint of a line corresponds to disappearance of phase 2 interlayer at T>Tm(Ni), which denotes the onset of the non-equilibrium crystallization mechanism for a given h0Al value [Figure 3.7(a) and (b)]; this is the second critical heating rate, v T , for each of the diffusion data sets. It should be noted that all the curves in Figure 3.7(a) and Figure 3.7(b) are similar, and the values of Tfin corresponding to critical heating rates v T and v T are the same for any h0Al at a given set of diffusion parameters. Besides, a minimum in curves Tfin(vT) seen in Figure 3.7(a) does not correspond to a definite physical process and is only a result of competition between growth and dissolution of the phase 2 interlayer, which proceed at different rates in different temperature intervals; for data set No. 2 a minimum is not observed [Figure 3.7(b)].
( cr ,1) ( cr , 2 ) ( cr , 2 ) ( cr ,1) ( cr ,1) ( cr , 2 )
66
B. B. Khina
Figure 3.6. Final thickness of phase 2 (NiAl), h2fin, attained by the time of disappearance of any of the parent phases vs. heating rate for different initial thickness of the Al layer: h0Al=100 m (line 1), 50 m (2), 20 m (2.1), 10 m (4), and 5 m (5); (a) calculated using diffusion data set No. 1 for NiAl (see Table 3.1) [Reprinted with permission from: B.B.Khina, Journal of Applied Physics, Vol. 101, No.6, 063510 (11 pp.), 2007. Copyright 2007, American Institute of Physics]; (b) calculated using diffusion data set No. 2 for NiAl (see Table 3.1) [Reproduced with kind permission from Springer Science+Business Media: International Journal of Self-Propagating High-Temperature Synthesis, Modeling heterogeneous interaction during SHS in the Ni-Al system: a phase-formation-mechanism map, vol. 16, 2007, pp. 51-61, B.B.Khina and B.Formanek, Figure 4, Copyright Allerton Press, Inc., 2007].
67
Figure 3.7. Final temperature, Tfin, at which the growth of phase 2 (NiAl) terminates vs. heating rate for different initial thickness of the Al layer, h0Al: (a) data set No. 1 for interdiffusion in NiAl (Table 3.1): h0Al=100 m (line 1), 50 m (2), 20 m (2.1), 10 m (4), 5 m (5), 1 m (6), and 0.5 m (7) [Reprinted with permission from: B.B.Khina, Journal of Applied Physics, Vol. 101, No.6, 063510 (11 pp.), 2007. Copyright 2007, American Institute of Physics]; (b) data set No. 2 for interdiffusion in NiAl (Table 3.1): h0Al=100 m (line 1), 20 m (2), 5 m (2.1), 1 m (4), and 0.5 m (5) [Reproduced with kind permission from Springer Science+Business Media: International Journal of SelfPropagating High-Temperature Synthesis, Modeling heterogeneous interaction during SHS in the Ni-Al system: a phase-formation-mechanism map, vol. 16, 2007, pp. 51-61, B.B.Khina and B.Formanek, Figure 5, Copyright Allerton Press, Inc., 2007].
68
B. B. Khina
( cr ,1) ( cr , 2 )
Thus, for any h0Al value the critical heating rates v T and v T can be found unambiguously. However, transition to the QEDP mechanism, where the NiAl interlayer with almost constant thickness displaces as a whole between two melts, occurs gradually, but in a rather narrow range of heating rates. Hence there isnt a strict criterion for the onset of this interaction route. So, the third critical heating rate v T
( cr ,3)
corresponding to this transition [obviously, v T
( cr ,1)
<
( cr ,3) ( cr , 2 ) < vT ] was determined, with a small degree of subjectivity, basing of vT
the behavior of lines h2(T) (like those shown in Figures 5 and 6), which were calculated by varying vT for given h0Al; for diffusion data set No. 1 it does not coincide with a minimum in curves for Tfin [Figure 3.7(a)].
3.6. PHASE-FORMATION-MECHANISM MAP FOR NONISOTHERMAL INTERACTION IN THE NI-AL SYSTEM

As a final result of numerical simulation, a diagram of phase and structure formation mechanisms for the synthesis of NiAl in non-isothermal conditions is constructed (Figure 3.8). Lines 1 (1') and 2 (2') correspond to critical heating rates
cr ,1) ( cr , 2 ) and v T , respectively, calculated using diffusion data set No. 1 (lines v( T
1 and 2) and No. 2 (lines 1' and 2'). In parametric domain I, growth of a solid NiAl layer at the expense of liquid Al and solid Ni proceeds via interdiffusion through this layer, which corresponds to a slow reaction route typical of the traditional furnace synthesis, where the heating rate is low, below 10 K/s. Here, widely used models employing solid-state diffusion-controlled kinetics [63-70] can give an adequate description of the product formation, but calculations should be performed with real diffusion parameters. For a reactant particle size of 20-100 m, which is typical of CS in particulate systems, this interaction pattern cannot provide fast conversion of initial metals into the product in the course of CS in both thermal explosion and wave propagation mode. Note that this mechanism can take place in thin-film NiAl systems (at h0Al<0.3 m) even at high heating rates characteristic of the SHS wave, vT ~ 105-106 K/s. In domain IV, non-equilibrium crystallization of NiAl grains from the melt occurs (without intermediacy of a continuous solid NiAl interlayer). This brings about fast accomplishment of interaction, which is typical of CS.
69
Figure 3.8. Diagram of phase formation mechanisms for synthesis of intermetallic compound NiAl in non-isothermal conditions: domain I (below line 1 or 1') is the solidstate diffusion-controlled NiAl layer growth, or slow route typical of furnace synthesis; domain II [between lines 1 (or 1') and 3 (or 3')] combines diffusion-controlled growth of the displacing NiAl layer and precipitation of NiAl grains from melt; domain III [between lines 3 (or 3') and 2 (or 2')] is the quasi-equilibrium dissolution-precipitation route where the thickness of the intermediate NiAl film is almost constant; domain IV (above line 2 or 2') is the non-equilibrium crystallization mechanism, or fast route typical of CS. Diffusion parameters for NiAl (Table 3.1): data set No. 1 (lines 1, 2 and 3) and No. 2 (lines 1', 2' and 3'). Mark z corresponds to phase formation pattern observed in [23] and mark refers to the experimental conditions of [21] for SHS in the Ni-Al system. [Reprinted with permission from: B.B.Khina, Journal of Applied Physics, Vol. 101, No.6, 063510 (11 pp.), 2007. Copyright 2007, American Institute of Physics]
In this parametric region, models [54-59] are valid; the rate-limiting stage is either diffusion in the liquid phase [54-56] or crystallization kinetics [58,59]. As seen from Figure 3.8, at h0Al>5 m conversion of reactants into the final product in the SHS wave where the heating rate attains 105-106 K/s can occur only via this fast route. In experiments on SHS in multilayer Ni-Al foils in the TE mode [23], the half-thickness of Al layers was 56 m and the measured heating rate at temperatures above 1300 C was 104 K/s. It corresponds to domain IV were NiAl grains can form via non-equilibrium crystallization. Very fast formation of the
70
B. B. Khina
product at T> 1300 C was ascribed to precipitation of solid NiAl from a supersaturated liquid without an intermediate solid interlayer [23], which agrees with the outcome of our numerical modeling. If a large amount of melt is formed that cannot react completely to form the solid product (NiAl) during a short heating time in the SHS wave, its outflow can occur, which was observed experimentally in a thick-foil Ni-Al system [136]. ( cr ,3) Line 3 or 3' refers to the 3rd critical heating rate, v T . Region II lying between lines 1 and 3 (or between 1' and 3') corresponds to a combined mechanism when the NiAl layer, whose thickness is still increasing with time due to solid-state interdiffusion, starts displacing between the Al-base and Ni-base melts, and a certain fraction of the final product is formed via melt crystallization. This domain is rather narrow, especially for diffusion data set No. 1 (see Table 3.1), and this interaction pattern can be considered as a transient one. In domain III located between lines 3 and 2 for diffusion data set No. 1 or between lines 3' and 2' for set No. 2, the quasi-equilibrium dissolutionprecipitation mechanism takes place: a thin continuous NiAl film separating two liquid phases, whose thickness remains almost constant during heating, displaces as a whole while the final product, viz. NiAl grains, can form only via precipitation from the melt. The kinetics of product crystallization can be controlled by solid-state diffusion across the film since the diffusion coefficient in NiAl is substantially lower than in the melt. This situation is close to that considered in [67]. However, as mentioned above, model calculations in [67] were performed using fitting parameters instead of real diffusion data for phase NiAl. Besides, that model cannot give information about the product grain size, which is important for practical applications. Thus, to correctly describe this interaction route, the approach developed in this work should be combined with a relevant crystallization model. In studying SHS of NiAl in a Ni-Al powder mixture [21], the initial particle radius of Al was 50-75 m and the measured heating rate in the SHS wave was vT2700 K/s. This is close to the border between domains III and IV (Figure 3.8). Since the temperature and heating rate during SHS is non-uniform in the crosssection normal to the combustion wave propagation because of heat losses to the environment, which is especially true for small-sized samples, interaction at the Ni/Al interfaces in the specimen during heating in the SHS wave can occur both via QEDP in the presence of a thin NiAl film and via non-equilibrium crystallization of NiAl grains. Hence, the results of our modeling qualitatively agree with results of microstructural studies [21], where the product (NiAl) grains were observed to precipitate from the saturated Al-Ni liquid.
71
It should also be noted that if we use a different set of E and D0 values for diffusion in the NiAl layer or take into account the concentration dependence of interdiffusion coefficient D2 (e.g., according to [150-154]) and consider metastable phase equilibria in the Ni-Al system below the Ni3Al melting temperature, this will result in displacement of lines in Figure 3.8 (probably towards lower heating rates because the parameters listed in Table 3.1 give upper estimates of D2 values), but the general physical picture revealed above by numerical modeling most likely will not change. For complete modeling of CS in the Ni-Al thin-film system, the above kinetic approach should be combined with a heat transfer model such as developed in [177], where the heat release rate will be proportional to h2/t times the reaction enthalpy.
3.7. CONCLUSION: ROLE OF PHASE-FORMATIONMECHANISM MAPS IN THE CS OF ADVANCED MATERIALS

In this research, the validity of the quasi-equilibrium concept of solid-state diffusion-controlled product-layer growth is examined for the formation of interstitial compounds such as carbides (on the example of TiC) and substitutional phases such as intermetallics (taking NiAl as an example) in strongly nonisothermal conditions, which are intrinsic in CS. In both systems, real (i.e. experimentally known) parameter values, viz. activation energy E and preexponential factor D0, for diffusion in the product phase are used along with relevant equilibrium binary phase diagrams. For titanium carbide, spherical symmetry of a unit reaction cell is considered, which is typical of CS in particulate systems, and all the possible scenarios are examined. For nickel aluminide, the case of planar symmetry is studied, which corresponds to CS in multilayer thin films. As a final result, phase-formation mechanism maps for nonisothermal heterogeneous interaction in the above binary systems are constructed (Figures 7 and 15), which permit classifying the experimentally known reaction pathways and determining the applicability limits of existing kinetic models of CS. Besides, the effect of high heating rates, which are inherent in CS, on the interaction pattern is revealed. It is demonstrated that at high heating rates typical of CS and the characteristic reactant sizes of above 5 m for the Ti-C system and 20 m for the Ni-Al system, which are typical of SHS in powder charges, complete conversion of reactants into a solid product can occur via non-equilibrium crystallization of the refractory product grains from a saturated melt without the presence of a thin
72
B. B. Khina
solid interlayer of an equilibrium phase. This is connected with a low rate of solid-state diffusion across the product phase. Thereby the existence of nonequilibrium patterns of phase formation, which were observed experimentally and debated in literature, is proved theoretically ex contrario for certain parametric domains. In these situations (domain II in Figure 2.7 and domain IV in Figure 3.8), theoretical approaches [54-59] can be used for modeling CS. At a large characteristic size of reactants and relatively low heating rates, product formation can occur via the equilibrium solid-state diffusion-controlled route (domain I in Figures 7 and 15). However, for the particle sizes typical of CS in a powder charge mixture, this reaction pathway refers not to CS but to traditional furnace synthesis where the heating rate is low. Besides, this interaction pattern can take place during CS, i.e. at high heating rates, only for a small size of reactants: R000.5 m for the TiC system and h0Al<0.3 m in the NiAl system. In this parametric domain, equilibrium diffusion-based models such as [62-66,68,69] are valid. A quasi-equilibrium dissolution-precipitation route via the intermediacy of a continuous solid-product interlayer (the shrinking-core mechanism in the spherical symmetry) appears impossible in the Ti-C system. This is connected with a substantial difference in partial diffusion coefficients of the Ti and C atoms in the titanium carbide layer (DTi << DC), whose diffusion fluxes are responsible for the displacement of different phase boundaries; typically in metal carbides MCx DM << DC in a wide temperature range. But in the formation of substitutional compound, e.g., NiAl, whose growth is determined by the interdiffusion coefficient (i.e. can occur at both Ni/NiAl and NiAl/Al interfaces), which is typically exceeds the diffusion coefficient in refractory carbides, this interaction route is possible in a certain parametric range (see domain III in Figure 3.8). This pattern was considered in [67] implying solid-state diffusion across the intermediate NiAl layer as a rate-limiting stage of CS and using fitting parameters for calculations instead of real diffusion data for NiAl. In the formation of an interstitial compound (here TiC) in non-isothermal conditions, a criterion for the changeover from the equilibrium solid-state diffusion controlled mechanism to the non-equilibrium dissolution-precipitation route is the rupture of a primary refractory-product shell formed on the surface of a solid metal particle at melting of the latter. Hence a diagram similar to Figure 2.7 can be plotted for non-isothermal interaction in regard to CS in some other systems, such as metal-carbon, where the value of TCS exceeds the melting temperature of the metallic reactant but is below that of a non-metal, and probably for some metal-gas (e.g., nitrogen) systems.
73
In the formation of a substitutional compound (here intermetallic NiAl) in non-isothermal conditions typical of CS, the criteria for transitions from the equilibrium (solid-state diffusion controlled) interaction pattern to the quasiequilibrium (with the presence of a thin layer of solid product) mechanism and then to the non-equilibrium route (crystallization of product grains from a saturated melt) are determined by competition between the diffusion-limited growth of the product layer and its dissolution in the parent phases. Therefore, phase-formation diagrams similar to that shown in Figure 3.8 can be constructed for other binary systems forming substitutional compounds (intermetallics). Finally, it should be outlined that the development of novel advanced materials using CS and CS-related technologies and, in future prospect, creating controllable CS processes necessitates elaboration of complex models for combustion synthesis in binary and multi-component systems. Such models are supposed to include, along with heat transfer equation, adequate description of the mechanism and kinetics of heterogeneous interaction, which, as shown above, may substantially differ in different ranges of a heating rate, temperature and characteristic size of reactants. In this aspect, the maps of phase formation mechanisms constructed in this work for typical compounds produced by CS provide a better understanding of the transformation patterns in strongly non-isothermal conditions inherent in CS and permit predicting a particular interaction route which determines that or another particular structure of the product and hence the properties of the latter. This opens up a prospect of purposely altering the synthesis conditions for attaining a desirable structure, composition and thus functional properties of a target product. Farther, the developed approaches and the maps can serve as a basis for working out both new models of SHS and novel, controllable CS-based technologies for producing advanced composite materials with improved performance. It should be noted that the problem of interaction mechanism in nonisothermal conditions addressed in Chapters 3 and 4 is pertinent not only to CS but also to other processes related to synthesis of advanced materials, coatings and net-shape articles such as deposition of protective coatings, e.g., by plasma spraying, and different methods of laser-assisted direct manufacturing such as selective laser sintering (SLS) [178], laser cladding [179], selective laser melting [180], laser engineered net shaping (LENS) [181] etc., where reactive powder mixtures are used rather than alloy particles with final chemical and phase composition.
Chapter 4
ANALYSIS OF THE EFFECT OF MECHANICAL ACTIVATION ON COMBUSTION SYNTHESIS

4.1. INTRODUCTION
Mechanically activated (or assisted) SHS (MA-SHS) is currently a subject of extensive experimental investigation [182-187]. It was observed experimentally that in a number of systems preliminary MA of the charge mixture brought about a decrease in both the ignition (Tig) and combustion (TSHS) temperatures along with changes in the SHS wave velocity vSHS and an increase of the conversion degree (01); in some cases mechanical activation (MA) changed the phase composition of the final products. In the Ni-Ti and Ni-Al systems, with increasing the time of MA (tMA) in a high-speed planetary mill the values of vSHS and TSHS were found first to increase and then to decrease substantially, and the position of a maximum in curves TSHS(tMA) and vSHS(tMA) did not coincide: the former was observed at smaller tMA than the latter [187]. At certain regimes of MA, a phenomenon of purely solid state combustion (PSSC) was observed in the Ni-13 wt.% Al system during the synthesis of Ni3Al: the SHS temperature fell down below the melting point of aluminum, TSHS < Tm(Al)=933 K [186,187]. Despite extensive experimental investigation, intricate physicochemical mechanisms responsible for a strong influence of MA of a charge powder mixture on subsequent SHS are not clearly understood yet. It is known that during initial stages of mechanical activation/mechanical alloying of metal-metal or metalnonmetal systems, composite (e.g., lamellar) particles are formed due to fracturing and cold welding processes in a comminuting device used for MA such as a vibratory or planetary mill or attritor [188,189]. In regard to subsequent SHS, this increases the contact surface area between unlike substances and decreases the
76
B. B. Khina
characteristic size of solid particles/layers of the reactants. However, the reasons for the experimentally observed effects cannot be reduced to the aforesaid factors, the latter can only explain an increment of the conversion degree . An increase in the contact surface area of the reactants after MA, which reduces heat losses during combustion, along with the observed buildup of the conversion degree will eventually lead to an increase in the SHS temperature: the latter is supposed to approach the adiabatic value, TSHS Tad. As a consequence, the combustion velocity should increase continuously according to the Zeldovich and FrankKamenetskiy formula [32]. But, as mentioned above, this is not the case: after prolonged milling, both the TSHS and vSHS values were observed to decrease substantially [183-187]. In experimental work [186,187], the observed PSSC phenomenon was attributed to the influence of non-equilibrium point defects (mainly vacancies), which were accumulated in metallic reactants during intensive periodic plastic deformation (IPPD) in the course of MA, on the formation the product phase (Ni3Al) in the SHS wave. In [186,187] it was considered that the relaxation of these defects to the equilibrium concentration did not occur during heating of the activated powder mixture in the preheating zone of the SHS wave, and this effect was named reverse quenching of the nonequilibrium defects [187]. However, this explanation was presented merely on a conceptual level and was not supported by experimental investigation nor by numerical estimates [186,187]. In a series of theoretical papers [191-195], the effect of preliminary MA on SHS was attributed to a decrease of the apparent activation energy of combustion by the value of energy stored in solid reactants during MA. However, the numerical values of stored energy used in modeling were not reported [192-195]. In the authors view, not only the model equations but also the parameter values used in calculations are supposed to have a definite physical meaning and should be carefully grounded and linked to a particular SHS system. In [191], all the calculations were performed using dimensionless complexes, thus real parameter values and their intrinsic physical meaning seem vague. This casts serious doubts on the applicability of this approach to modeling MA-SHS. In studying MA-SHS in the Mo10 wt.% B system [196], the burning velocity was observed to increase after MA in comparison with a non-activated charge mixture, and the apparent activation energy of combustion, E, was found to decrease by a value of Ea ranging from 245 kJ/mol to 305 kJ/mol (depending on the method used for estimating the E value). Similar results were reported for Cr-B, Ti-C and Ti-B systems [196]. However, a physical explanation for these phenomena was not presented.
Analysis of the Effect of Mechanical Activation on Combustion Synthesis 77 In connection with the above, the objective of this chapter, according to [197,198], is (i) to perform numerical assessment of the main physicochemical factors, which may be responsible for the influence of MA on subsequent combustion synthesis, using relatively simple physicochemical estimates, (ii) to carry out critical analysis of the existing theories of MA-SHS using realistic parameter values, and (iii) to develop a physical explanation (at the current stage, on a qualitative level only) for the observed effects basing on certain results of the state-of-the-art solid-state diffusion theory, including the up-to-date concepts of diffusion processes stimulated by IPPD during MA.
4.2. STORED ENERGY IN METALS AFTER MECHANICAL ACTIVATION

4.2.1. Mechanical Activation vs. Mechanical Alloying
Mechanical activation has acquired extensive use in industry as an efficient means for increasing the yield of target products during their extraction from minerals, improving chemical reactivity of both inorganic and organic substances, synthesis of a wide range of compounds, and some other applications. However, despite a long history of experimental research and development in the field of mechanochemistry, there is still a lack of understanding of the exact nature of atomistic processes underlying physical and chemical transformations under mechanical processing of different substances [199]. Intricate interaction mechanisms that can operate during MA in inorganic systems are a subject of lively debates in literature. At present, most developed are theories based on the concept of mechanical reactive mixing in interfacial regions of reactant particles formed by fracturing during intensive ball milling [200-202]. The contact and displacement of juvenile surfaces during shear codeformation of dissimilar reactant particles bring about a local disorder (e.g., rupture of chemical bonds) in the contacting substances, and the interface reaction results then from a local reorganization of the disordered lattice involving few atomic layers. Within this concept, the main factor influencing mechanochemical transformation is a fraction of the impact energy transferred to the reactant particles during incidental ball-powder-ball or ballpowder-wall collisions in a mechanochemical reactor, and chemical affinity of the components. A certain portion of this energy can be stored in the solid particles in the form of defects/local disorder.
78
B. B. Khina
In the authors view, this semi-quantitative theory refers to a larger extent to brittle inorganic substances, where the dominating process is fracturing and hence formation of juvenile surfaces, rather than to metals. In systems based on metals whose distinctive feature is ductility, along with fracturing the most important processes are (i) plastic deformation of a metal particle as a whole, which results in the formation of a large amount of lattice defects and grain refining, and (ii) cold welding of dissimilar reactant particles. The latter brings about the formation of composite particles with nano or submicron grain size, which are composed either of interlaced lamellas of ductile reactants or small inclusions of a brittle substance in a metal matrix [188,189,203]. Then intensive deformation-driven interaction can occur inside such composite particles resulting in the formation of metastable phases such as supersaturated solid solutions, quasicrystalline and amorphous phases. This process is referred to as mechanical alloying, which is used for producing a wide range of far-from-equilibrium particulate metal-base materials with unique physical and mechanical properties, and an important role in the product formation is ascribed to deformation-enhanced solid-state diffusion [188,189,204]. From this viewpoint, mechanical activation as method of charge mixture preparation for subsequent SHS can be considered as an early stage of mechanical alloying because the former is intended primarily to attain a high contact surface area and small characteristic size of the reactants. However, as will be discussed below in more detail, the role of enhanced bulk diffusion cannot be ignored even at short-time processing in a high-energy-density device for MA such as vibratory and planetary mills. Since the existing theories of MA-SHS [191-195] are based on the concept of stored energy, it seems necessary to estimate its value for metal-base systems where this energy is associated mainly with crystal lattice defects accumulated in the course of MA. The issues connected with the stored energy in metals are important not only in studying MA and MA-SHS but also in regard to bulk nanostructured materials produced by severe plastic deformation [205].
4.2.2. Numerical Estimation of Stored Energy in Metals

It is known that during cold plastic deformation of metals only 0.5-5% of the spent energy is stored in the form of non-equilibrium defects while the main portion of the energy dissipates as heat [206]. Let us estimate the energy stored in metallic reactants due to cold work during MA (Es), which could affect the apparent activation energy of solid-state reaction during subsequent SHS due to
Analysis of the Effect of Mechanical Activation on Combustion Synthesis 79 distortion of the crystal lattice of the parent phases (pure metals), using the Ni-Al and Fe-Al systems as examples. The stored energy is assessed as Es = Eres + Epd +Ed + Egb, (4.1)
where Eres is the energy of residual elastic strain, Epd is that of point defects generated by plastic deformation during MA, Ed is the elastic energy associated with dislocations, and Egb is the energy of grain boundaries. A maximal energy that can be accumulated in a metal due to IPPD corresponds to solid-state amorphization, which is known to occur during prolonged MA in binary or multicomponent systems and, in some cases, at intensive milling of pure metal powders [188,189,207]. The enthalpy level to be reached for solid state amorphization to occur in a pure crystalline substance is Ham 0.4Hm, where Hm is the enthalpy of melting [208]. Similarly, the heat of crystallization of a binary amorphous alloy to intermetallic compound at the same composition typically constitutes about 50-70% of the equilibrium heat of fusion of the latter [209]. Therefore, the values of stored energy Es and its constituents [see formula (4.1)] should be estimated in terms of the melting enthalpy; the values of Hm for several metals used in estimations are listed in Table 4.1. Table 4.1. The parameter values used in calculations
Metal 0.2, MPa (cold worked metals at T=300 K) Al 98.1 [213] Fe () Ni Cu Cr Mo 263 [213] 247 [213] Y, GPa (cold worked metals at T=300 K) 70.8 [213] 213 [213] 216 [213] G, GPa b, nm (at 300 [42] K)
Hfv, eV
D0*, cm2/s
E*, Hm, kJ/mol kJ/mol
265 [42, 130 [42] Table 22.7] -
25.4 [214] 64.0 [214] 78.9 [214] 48.3 [42] -
0.286 0.248 0.249 0.256 -
0.76 [219] 2.25 [42] 1.40 [221] 2.0 [42] 1.70 [220] 1.27 [42] 1.17 [219- 221] 2.27 [221] 970.0 [222] 3.0 [221] 0.1 [222]
144.35 [42] 251.0 [42] 281.16 [42] 435.0 [222] 385.4 [222]
10.5 13.8 17.8 13.0 -
80
B. B. Khina
It should be outlined that the regimes of MA, which are used as a preliminary stage for SHS, are typically intended to obtain composite particles with small lamella thickness and high contact surface area of solid reactants, and the processing is terminated long before the amorphization can occur. Contrary to a popular opinion that nanocrystals do not contain dislocations, recent experimental investigations of metals subjected to IPPD have revealed the presence of dislocations inside nanograins (not only in the grain boundaries) whose density is very high. For example, in nickel specimens produced by cold rolling the dislocation density inside the nanocrystalline (NC) grains was found to reach = 4.710111.31012 cm2 [210]. For a Ni20 wt.% Fe alloy subjected to high-pressure torsion, the value of = 41012 cm2 within NC grains was reported [211]. Hence formula (4.1) is also valid for NC powder materials obtained by MA. The residual stress in materials cannot exceed the limit of elasticity. In literature on mechanical properties of materials, typically the 0.2% proof stress, 0.2, is presented. Hence an upper estimate for the energy of residual strain, Eres, per one mole of a metal can be obtained using a simple formula that corresponds to the case of an elastically strained rod [212]:
E res = 0.2 /( 2Yd) ,
(4.2)
where Y is the elastic modulus, is the molar mass and d is density. The values of 0.2 and Y for cold-worked Ni, -Fe, Al [213] and Cu [42] are listed in Table 4.1. The results of calculations by Eq. (4.2) are presented in Table 4.2. It is seen that the value of Eres is insignificant: it does not exceed 2 J/mol (for Cu) and constitute only 0.0060.015% of Hm of the metals. In severely cold-worked metals, the maximal dislocation density is ~1012 cm2. The stored energy per one mole associated with dislocations is estimated as [212] Ed = fGb2/d, (4.3)
where f=0.51 is a coefficient accounting for the type of dislocation, G is the shear modulus and b is the Burgers vector. Using reference data [42,214] listed in Table 4.1 and assuming f=1 we obtain that for Al, -Fe, Ni and Cu the maximal energy stored after IPPD, which is connected with dislocations, lies within
Analysis of the Effect of Mechanical Activation on Combustion Synthesis 81 0.20.32 kJ/mol (see Table 4.2). This constitutes about 2% of the melting enthalpy for corresponding metals, thus the value of Ed is small. Table 4.2. Estimation of stored energy in cold-worked metals
Metal Eres J/ mol 0.68 % of Hm 0.006 0.008 0.005 0.015 Ev J/mol Ed Egb % of J/ Hm mol 1.98 2.03 1.82 1.74 340 617 % of Hm 2.46 4.75 Es J/mol 1472 % of Hm 10.7
Al
Fe () 1.2 Ni Cu 0.94 1.93
% of J/mol Hm 288 [at Cv = 2.74 207.8 50Cv0(Tm)] 850 [at Cv = 6.16 280.4 50Cv0(Tm)] 82.0 [226] 0.46 324 56.4 [225] 0.43 226.1
901.4 6.9
The excess energy associated with boundaries of nanograins, which are formed in the course of IPPD during MA, is calculated as Egb = gbS, where S is the total interface area and gb is the energy per unit surface area. To roughly estimate the value of S, it can be assumed (only for the sake of simplicity) that cubical grains with size a are stacked into a simple cubic lattice. Then the total interface area in volume V is S3V/a, and per unit volume we have s=S/V=3/a. Hence the energy of grain boundaries per one mole of a metal is assessed as Egb = 3gb/(ad). (4.4)
In literature, experimental data of the energy of nanograin boundaries, gb, are reported for -Fe with the grain size a=10 nm: gb=0.16 J/m2 [215], and for nanocrystalline Cu (a=11 nm): gb=0.29 J/m2 (after 10 h milling time) [216], the latter being the maximal value observed in experiment [216]. These values are somewhat lower than the boundary energy of normal, i.e. micron-size grains in bulk metals [217]. Then from Eq. (4.4) we obtain Egb=340 J/mol for -Fe, which corresponds to about 2.5% of Hm(Fe), and for copper Egb=617 J/mol (about 4.8% of its melting enthalpy). Thus, the stored energy in metallic reactants after MA, which is associated with nonograin boundaries, is also not large. It should be noted that in a binary system A-B, nanocomposite (typically lamellar) particles formed in the course of MA contain not only A/A but A/B nanograin boundaries. The values of interface energy for such boundaries are not known in literature. However, it seems reasonable to assume that the excess energy associated with A/B boundaries, which may release during high-temperature interaction after MA
82
B. B. Khina
of starting pure reactants, is lower that that of A/A and B/B interfaces because of chemical affinity between dissimilar substances. Non-equilibrium point defects are generated during plastic deformation by multiple routes, e.g., annihilation of edge dislocations located in closely spaced slip planes and jog dragging by gliding screw components of dislocation loops (the Hirsch-Mott mechanism). The concentration of vacancies is substantially higher than that of self-interstitials [218]; the role of the latter will be analyzed below. Then we assume that Epd Ev where Ev is the stored energy associated with excess vacancies, which is expressed as
E v = H f N C , v A v
(4.5)
where Hfv is the enthalpy of vacancy formation, which is expressed in eV, NA is the Avogadro number and Cv is the vacancy concentration. The concentration of vacancies accumulated in metals in the course of MA is not reported in literature, but known are the values measured experimentally after IPPD by multi-pass equal-channel angular pressing/extrusion (ECAP/ECAE), repetitive cold rolling (RCR) and high-pressure torsion (HPT), where the conditions of straining (constrained deformation) are still more severe than in MA of powders (unconstrained deformation); the latter were estimated using the concept of Hertzian collision [223,224]. For fine-grained copper after ECAP, RCR and other deformation methods, the maximal experimentally measured vacancy concentration was Cv=5104 at the dislocation density of about 41011 cm2 (see Figure 2 in [225]). After IPPD by high-pressure torsion (HPT) the vacancy concentration in copper reached Cv=4104 (the saturation value) at dislocation density =1.21012 cm2; for Ni the corresponding value was Cv=5104 at =1.01012 cm2 [226]. These values are extremely large as compared with those in thermal equilibrium at a room temperature. However, they agree with the results of numerical simulation of deformation-enhanced interdiffusion during MA in the Cu-Al system at the conditions of IPPD corresponding to laboratory shaker mill SPEX 8000 with the vibration frequency of 20 Hz, which were obtained using an original mathematical model [227-229]. Using the Hvf values listed in Table 4.1, from Eq. (4.5) we obtain for copper Ev=56.5 J/mol (at Cv=5104 [225]), which constitutes only 0.43% of Hm(Cu), while for nickel at Cv=5104 [226] the calculated value of Ev corresponds to 0.46% of Hm(Ni) (see Table 4.2). The equilibrium concentration of vacancies is evaluated as
Analysis of the Effect of Mechanical Activation on Combustion Synthesis 83
C0 = exp[ H f N /(RT)] , v v A
(4.6)
where R is the universal gas constant. Calculation using Eq. (4.6) and the values of Hfv presented in Table 4.1 demonstrate that the concentration of vacancies accumulated in copper and nickel during IPPD at ambient temperature, which are reported in [225,226], is by the factor of 10 (for Cu) to 45 (for Ni) larger than the equilibrium value at the melting point of a corresponding metal, Cv0(Tm). Then, to make an upper-level estimate of Ev for aluminum and iron, we assume that Cv = 50Cv0(T=Tm). The results are presented in Table 4.2. It is seen that the accumulated energy associated with nonequilibrium vacancies is not large: it constitutes about 6% of the fusion enthalpy (an upper-level estimate for -Fe). The above numerical estimates have demonstrated that the stored energy associated with the defects accumulated in metals after IPPD during MA is small: Es 11% of Hm (see Table 4.2), and this value is low in comparison with a typical heat release of the reaction during SHS even in weakly exothermic systems. This outcome of our estimates agrees with the data reported for iron in regard to MA-SHS in the Fe-Si system (without any numerical calculations): Es15-20% of Hm [230]. Hence the experimentally observed substantial decrease in the apparent activation energy of MA-SHS, Ea, cannot be attributed to the stored energy of plastic deformation. For example, the value of Ea=2430 kJ/mol obtained for MA-SHS in the Mo10 wt.% B system [196] is very high: it corresponds to 6784% of the melting enthalpy of molybdenum (Hm(Mo)=35.6 kJ/mol [42]) or 104130% of that of boron (Hm(B)=23 kJ/mol [231]).
4.3. ANALYSIS OF EXISTING KINETIC THEORIES OF MA-SHS

In the experimental and modeling study of MA-SHS in the Ti-N system, the stored energy was estimated as being proportional to the broadening of X-ray diffraction (XRD) lines of titanium after mechanical milling, but the proportionality factor was not reported [190]. This led to overrated value of Es=36.6 kJ/mol [190]: it exceeds the titanium melting enthalpy, Hm(Ti)=17.5 kJ/mol [42], by the factor of 2.1. If it were really so, solid-state amorphization of titanium would inevitably occur during MA, which, in turn, would be
84
B. B. Khina
unambiguously detected by the XRD analysis. It should be noted that a similar method used in [232] for evaluating the energy stored in W and Ti during MA of mixtures 94 wt.% WC and 80 wt.% TiC gave strongly overestimated values: Es=84.7 kJ/mol and 122.9 kJ/mol, correspondingly. The former exceeds the melting enthalpy of tungsten, Hm(W)=35.2 kJ/mol [42], by the factor of 2.4 while the latter is 7 times larger than Hm(Ti) and constitutes about two-thirds of the heat release, Q, of reaction Ti+C TiC1.0, where Q = H0298(TiC1.0) = 184 kJ/mol [42]. The XRD method is widely used for measuring a change in the lattice period associated with residual elastic strain and estimating the dislocation density and other structural characteristics of a material. In submicron and NC metals obtained by severe plastic deformation during MA, distortion of XRD lines is connected with a number of effects such as (i) diminishing of the size of coherent scattering blocks (small grains or subrains, which are formed due to fragmentation and dislocation patterning during IPPD) down to nanoscale and their misorientation, (ii) lattice disorder due to defects and elastic strain, (iii) penetration of foreign atoms into grains, i.e. deformation-enhanced diffusion, and some other factors. Thus it seems incorrect to ascribe the distortion of a XRD spectrum only to the accumulated energy. It is worth noting that structural changes in materials due to IPPD in the course of MA such as formation of nanograins, high dislocation density etc. are known to have a profound effect on the physical and mechanical properties while the stored energy associated with these features is actually small, as shown above. Therefore, the aforesaid method [190,232] has brought about substantially overestimated, physically meaningless values of the stored energy of plastic deformation. An increase in heat release in the W-C and Ti-C systems after MA observed in [232] can be attributed to a rise in the conversion degree due to high contact surface area and small characteristic size of reactants along with some other factors discussed below. In a series of theoretical papers [191-195], modeling of MA-SHS was performed using the heat transfer equation with a source term associated with heat release of an exothermal reaction [similar to Eq. (1.1) in Chapter 1] and the kinetics of reaction during SHS was described by formal model /t k0exp[E/(RT)] [see Eq. (1.2)]. The basic assumption in the developed model is that the product formed during SHS can inherit structural defects (and also excess energy) from the reagents [195, p.48; 191, p.320]. However, it is not explained via which particular physical mechanism this can occur during reactions in condensed matter. Basing on this assumption, it has been postulated that after preliminary MA the apparent activation energy of combustion decreases by the
Analysis of the Effect of Mechanical Activation on Combustion Synthesis 85 value of energy stored in solid reactants: E = E0 bEEs, where E0 is the activation energy without MA and bE1 is a numerical factor. Here, for the sake of readers convenience, the notation is somewhat different from that used in [191-195]. Besides, the model is supplemented with equations accounting for partial relaxation of the stored energy upon heating in the SHS wave in the form Es,i/t = f(Es,i), iA,B, where A and B denote the reactants and Es,A, Es,B is the corresponding excess energy per unit mass of a corresponding substance. In calculations, the charge mixture was assumed to be stoichiometric, thus the total excess energy per unit mass is expressed as Es(t) = xAEs,A(t) + (1xA)Es,B(t), (4.7)
where xA is the mass fraction of A in the product [191-195]. All the calculations in [191,192-195] were performed using dimensionless complexes = RT*/E0, = cpRT*2/(E0Q), (t) = Es(t)/(RT*), i,0 = Es,i,0/(RT*), iA,B, where T* = T0 + Q/cp, (4.12) (4.8) (4.9) (4.10) (4.11)
and cp is the heat capacity, Q is the heat release of reaction without MA, T0 is the initial (room) temperature, T* is the combustion temperature without preliminary MA, i.e. T*Tad, Es,i,0 refers to the initial state (at t=0, i.e. after MA but before heating in the SHS wave) of i-th substance. Here parameters and are those traditionally used in the combustion theory [32], is the total dimensionless stored energy at any instant of time and i,0 is the dimensionless initial (after MA) excess energy per unit mass of the i-th reactant. In Eqs. (4.8)-(4.12) all the parameters, including universal gas constant R, are defined per unit mass. It should be emphasized that in [192-194] the parameter values used for calculations were not listed, thus it appears impossible to link the model and
86
B. B. Khina
results of simulation to a real system. In [191,195], the values of , and i,0 were presented and it was assumed that bE=1 and xA=0.5. Let us extract the physical parameters Tad=T*, E0 and Es,0 Es(t=0) from those data and correlate then with a particular system or MA-SHS process. Since the complexes , and in Eqs. (4.8)-(4.11) are dimensionless, E0 and Es,0 can be retrieved directly in J/mol. In numerical calculations, the following values were used: =0.15, =0.166 and A,0=B,0=2 (see Figure 1 in [195]), i.e. the excess energy per unit mass was the same for both of the reactants. Hence from Eqs. (4.7), (4.10) and (4.11) 0 (t=0) = A,0 = 2. Then, using Eqs. (4.8), (4.9) and (4.12) and assuming T0=298 K (because preheating the charge above room temperature prior to ignition was not reported in [195]), we obtain T* = Tad = 3092 K and E0=171.4 kJ/mol. These values seem reasonable and can be linked to a real SHS process. But from formula (4.10) it follows that the stored energy per mole of the reaction product is very large: Es,0=51.4 kJ/mol. As seen from Eq. (4.7), the contribution of reactants to excess energy Es,0 is proportional to their mass fractions. In this case, the obtained combustion temperature without MA, T*, is close to SHS of such refractory compounds as TiB2 (Tad=3192 K [159]), W2B (Tad=3040 K [159]) or WB (Tad=3070 K [159]). Then for MA-SHS of TiB2 the energy stored in Ti after MA will be Es,Ti,0=35.2 kJ/mol, which is extremely large: it exceeds the melting enthalpy of titanium by the factor of 2. If we take WB, where the fusion enthalpy of W substantially exceeds that of Ti [Hm(W)=35.4 kJ/mol], we have Es,W,0 = 48.5 kJ/mol = 1.37Hm(W). For W2B as a possible MA-SHS product, the stored energy in W will be still higher: Es,W,0 = 49.9 kJ/mol = 1.41Hm(W). Therefore, the value of energy stored in a metallic reactant during MA, which corresponds to dimensionless numbers , and used for calculations in [195], is strongly overrated and thus physically meaningless. In another variant (Figure 3 in [195]), simulation was performed with =0.05, =0.125 and A,0=B,0=0.5. From Eqs. (4.8)-(4.12) it follows that this data set corresponds to Tad=500 K 230 C, E0=82.6 kJ/mol and Es,0=2.06 kJ/mol. Such a low combustion temperature cannot be correlated with a real SHS or MA-SHS process in conventional systems, and may refer only to self-propagating exothermal reactions in organic systems, e.g., reaction of piperazine (Tm=112 C) with malonic acid (Tm=136 C) [15] (see Table 1.2). However, in such systems interaction starts only after melting of both of the reactants and the reaction wave propagates in a liquid phase [15], thus it is impossible to speculate about the effect of stored energy. In [191], calculations were performed with dimensionless parameters =0.08 and =0.1. It was assumed that the composition of a charge mixture corresponded
Analysis of the Effect of Mechanical Activation on Combustion Synthesis 87 to a stoichiometric one, and the other parameters were bE=1 and A,0=B,0=5 (see Figure 3 in [191]). Using the above described procedure we obtain that in this case Tad=1490 K and E0=154.9 kJ/mol, which is close to SHS of such intermetallics as Ni2Al3 and Ni3Al. The adiabatic combustion temperatures for these compounds calculated using formula (3.1) and standard thermodynamic data [110,160,233] are the following: Tad(Ni2Al3)=1469 K and Tad(Ni3Al)=1532 K. But the stored energy, according to formula (4.10), appears to be enormously large: Es,0=61.94 kJ/mol. Since in [191] it is assumed that the stored energy per unit mass is the same for both of the reactants, A,0=B,0, then contribution of each of them to total excess energy Es,0 is proportional to mass fraction. Hence, for Ni2Al3 as a possible reaction product Es,Ni,0 = 36.7 kJ/mol, which exceeds the fusion enthalpy of nickel by the factor of about 2.1, and for aluminum Es,Al,0 = 25.2 kJ/mol = 2.4Hm(Al). Thus, the parameter values used in works [191,195] appear to be physically meaningless, and hence it can be concluded that the theoretical concept [191-195] of the effect of stored energy of cold work on high-temperature exothermal reactions in condensed phases seems inapplicable to MA-SHS. In its physical meaning, a situation when stored energy can directly decrease the activation energy of a heterogeneous reaction refers to a case where at least one of the reactants is dispersed to atomic level, e.g., chemical interaction between gas molecules or ions exited by an external energy source and a solid surface.
4.4. RELAXATION OF NON-EQUILIBRIUM VACANCIES IN NON-ISOTHERMAL CONDITIONS

The excess vacancies distributed uniformly in a metal can cause the distortion of crystal lattice, which may be responsible for a change in the apparent activation energy of a heterogeneous reaction during SHS. This idea was used for interpreting (on a conceptual level only) the effects observed in MA-SHS [186,187]. However, severely cold-worked metals after MA contain a lot of edge dislocations and grain/subrain boundaries, which are known to be natural sinks for non-equilibrium point defects. Let us estimate the relaxation of excess vacancies, which were accumulated in metals during MA, in non-isothermal conditions corresponding to heating of the powder charge during SHS. The density of edge components of dislocation loops, which act as volumedistributed sinks for non-equilibrium point defects, is estimated as e=/2 [212,218]. For calculations we take =1010-1011 cm2, which is one-two order of
88
B. B. Khina
magnitude below a typical value attained in metals after MA. Average spacing between edge dislocations is L ~ e1/2, and hence the maximal distance traveled by a vacancy to a sink is Lv ~ L/2 = (2)1/2. In isothermal conditions, the characteristic diffusion length is estimated as L v ~ Dvt, where Dv is the diffusion coefficient of vacancies and t is time. For non-isothermal conditions T=T(t), this dependence is expressed in the integral form:
2
~ D v[T()]d = L2 v
0
1 . 2
(4.7)
To determine the value of Dv, the following simple expressions known in the diffusion theory can be used [219]:
D * = D* exp[ E*/(RT)] = C 0 D , 0 v v
(4.8)
where D* is the self-diffusion coefficient in a metal, which is measured at a longtime isothermal annealing, i.e. in the condition of local equilibrium of vacancies, D0* and E* are the preexponential factor and activation energy for self-diffusion, Cv0 is the equilibrium vacancy concentration defined by Eq.(4.6). To examine the relaxation of vacancies in non-isothermal conditions, we assume linear heating with a certain rate vT; this corresponds to SHS in the thermal explosion mode. Then T(t) = T0 + vTt, where T0=298 K is the initial temperature. Combining Eqs. (4.7), (4.8) and (4.6), we obtain:
1 = D* 0 2
(Tr T0)/v T
E * + H f N v A exp dt , + R(T v t) 0 T
(4.9)
where Tr is the temperature at the attainment of which the complete relaxation of vacancies occurs. The dependence of Tr on vT obtained by numerical solution of Eq.(4.9) for several metals used in MA-SHS is presented in Figure 4.1 for two values of dislocation density: =1010 and 1011 cm2; the heating rate was varied from 101 to 106 K/s. The parameter values used in calculations (Hfv, D0* and E*) for aluminum, nickel, -iron, chromium and molybdenum are listed in Table 4.1.
Analysis of the Effect of Mechanical Activation on Combustion Synthesis 89
Figure 4.1. The temperature at which complete relaxation of non-equilibrium vacancies can occur in non-isothermal conditions vs. heating rate for Al, -Fe, Ni, Cr and Mo at different dislocation densities.
It should be emphasized that the calculated values of Tr are upper-level estimates since we have taken the maximal vacancy diffusion length and lowerlevel estimates for dislocation density in severely deformed metals. Let us compare the results of calculations with experimental data on MA-SHS for different systems. In the Ni-Al system, the maximal heating rate in the SHS wave was vT=6000 K/s for TSHS900 K (below the melting point of aluminum) [186,187], hence the heating time from T0 to TSHS was 0.1 s. From Figure 4.1 it is seen that during this short time the non-equilibrium vacancies in both Al and Ni can relax completely, thus their effect on the solid-state interaction kinetics during SHS is insignificant. For MA-SHS in the Fe-Al system, TSHS=11731200 K, i.e. above Tm(Al), and vT=8001600 K/s (depending on the process conditions) [234]. Thus, as shown in Figure 4.1, complete relaxation of non-equilibrium vacancies in both Al and Fe can occur in the preheating zone of the SHS wave: here Tr < TSHS and even below the temperature of -to- transformation in pure iron, T-(Fe)= 1185 K. For molybdenum, even at high heating rates typical of SHS the relaxation of non-equilibrium vacancies can occur at temperatures much below the
90
B. B. Khina
experimentally observed combustion temperature at MA-SHS in the Mo10 wt.% B system: TSHS1550 K [196, Figure 9]. The concentration of self-interstitial atoms generated by IPPD during MA is lower than that of vacancies but they may cause a large local distortion of the crystal lattice [212,218]. However, the migration enthalpy of self-interstitials, Him0.1 eV [221,235], is substantially lower than for vacancies: cf. Hvm=0.62 eV for vacancy migration in Al and 1.35 eV for that in Ni [219]. Hence the relaxation of self-interstitials at heating in the SHS wave will occur still faster.
4.5. A POSSIBLE PHYSICAL MECHANISM OF THE EFFECT OF MECHANICAL ACTIVATION ON CS

The above estimates have demonstrated [198] that the effect on combustion synthesis of both point defects accumulated in metallic reactants during MA and the stored energy of cold work appears to be negligible. Therefore, the observed influence of preliminary MA on SHS can be connected only with intricate physicochemical processes that take place during MA itself: in particular, enhanced solid-state diffusion, which was observed experimentally and is responsible for the formation of metastable phases in the course of MA (see reviews [188,189,204] and references cited therein). However, the physical mechanism of this phenomenon has not been understood till recently and is a subject of debates in literature [236-240]. In works [227-229], a new, self-consistent model of deformation-enhanced solid-state diffusion in metals was developed for a binary system A-B and calculations were performed using the known values of equilibrium diffusion coefficients for a particular system (Al-Cu) and deformation parameters typical of a vibratory mill used for MA. It has been demonstrated that enhanced diffusion is caused by the generation of excess point defects during short-term deformation and their slow relaxation in time intervals between collisions in isothermal conditions, increase of diffusion coefficients of metal atoms due to high vacancy concentration, and, to the utmost, by the cross-effect connected with off-diagonal terms in the matrix of Onsager coefficients, i.e. the influence of the vacancy flux on the diffusion flux of lattice atoms (sometimes referred to as the inverse Kirkendall effect). This effect can bring about the formation of a supersaturated solid solution (SSS) in lamellar particles in the vicinity of the A/B interfaces within a reasonably short time of MA [228,229]. During heating in the SHS wave, the non-equilibrium vacancies disappear (as shown above) but the heat release of
Analysis of the Effect of Mechanical Activation on Combustion Synthesis 91 the exothermal reaction decreases by the value of the excess enthalpy of mixing, which is associated with SSS. Hence after a certain milling time, when the SSS formation in composite particles begins in the course of MA, the value of TSHS starts decreasing. Besides, it has been demonstrated using both kinetic [81] and thermodynamic considerations [241-244] that nucleation of a new phase (e.g., intermetallic compound ABx) in the field of a steep concentration gradient c/x (i.e. at the contact of pure metals A/B) is suppressed. It can occur only after the absolute value of the gradient, |c/x|, decreases to a certain critical value so that the width of a domain adjacent to the initial A/B interface, whose composition falls within the concentration range of compound ABx, reaches the size comparable with that of a critical nucleus, hcr ~ 10al where al is the lattice period of a new phase [241244]. Thus, enhanced diffusion during MA under the action of IPPD decreases the concentration gradient in the vicinity of the A/B boundaries in lamellar particles that have formed on the early stages of MA due to fracturing and cold welding. This, in turn, lowers the nucleation barrier for a new phase at the A/B interfaces during subsequent heating in the SHS wave, i.e. heterogeneous reaction can start at a lower temperature. The above factors probably constitute the main physical reason for the PSSC phenomenon [186,187], all the more that the formation of SSS of Al in Ni was outlined in Ref [187]. A similar effect, viz. partial dissolution of Al in Ni during MA and a strong influence of this phenomenon on the formation of NiAl and Ni3Al at subsequent heating was observed experimentally in the Ni36.5 at.% Al system [245]. Besides, in MA-SHS of FeAl-base nanocomposite powder strengthened by disperse alumina inclusions, which is used for plasma spray deposition of protective coatings, by reaction of aluminum with iron and chromium oxides, partial dissolution of initial Al in Fe2O3 and Cr2O3 particles was observed during MA, and the effect of this phenomenon on subsequent SHS was outlined [246]. The synergetic effect of the aforesaid factors may also be responsible for a difference in the position of the maximum in curves TSHS(tMA) and vSHS(tMA), which was revealed experimentally during MA-SHS in the Ni-Ti and Ni-Al systems [187]. It should be noted that these physical reasons can also be valid for systems where TSHS exceeds the melting point of a lower-melting reactant. The nuclei of a refractory product phase that can form upon heating at the A/B interfaces will act as crystallization centers at high temperatures where the dissolution-precipitation mechanism of the final product formation starts operating in the SHS wave. This will eventually lead to an increase of the conversion degree even at lower TSHS (as compared with a non-activated state) and may change the phase composition of the MA-SHS product, at least at the
92
B. B. Khina
initial stage. It should also be noted that the observed decrease in the apparent activation energy of MA-SHS [196] may be attributed (on the qualitative level) to the aforesaid mechanism. Thus, development of a comprehensive model describing MA-SHS and, particularly, the phenomenon of PSSC necessitates a deeper insight into the physical nature of both mechanical activation/alloying and nucleation of a solid phase in a field of concentration gradient in strongly non-isothermal conditions.
REFERENCES
[1] [2] [3] [4] Merzhanov, A. G.; Borovinskaya, I. P.; Shkiro, V. M. Bull. Izobr. USSR 1967, 32, 3. Merzhanov, A. G.; Borovinskaya, I.P. Dokl. Chem. 1972, 204, 429-432. Merzhanov, A. G. Russ. Chem. Bull. 1997, 46, 1-27. Merzhanov, A. G. In Combustion and Plasma Synthesis of HighTemperature Materials; Munir, Z. A.; Holt, J. B.; Eds.; VCH Publishers: New York, NY, 1990; pp 1-53. Munir, Z. A.; Anselmi-Tamburini, U. Mater. Sci. Rep. 1989, 3, 277-365. Bockowski, M. Physica B 1999, 265, 1-5. Varma, A.; Lebrat, J. P. Chem. Eng. Sci. 1992, 47, 2179-2194. Merzhanov, A. G. Ceramics Int. 1995, 21, 271-279. Merzhanov, A. G. J. Mater. Proc. Technol. 1996, 56, 222-241. Merzhanov, A. G.; Mukasyan, A. S. Solid-Flame Combustion; Torus Press: Moscow, 2007. Khusid, B. M.; Khina, B. B.; Rabinovich, O. S. Comp. Chem. Eng. 1994, 18, 1183-1195. Merzhanov, A. G.; Rogachev, A. S.; Mukasyan, A. S.; Khusid, B. M.; Borovinskaya, I. P.; Khina, B. B. J. Eng. Phys. Thermophys. 1991, 59, 809816. Borovinskaya, I. P.; Merzhanov, A. G.; Mukasyan, A. S.; Rogachev, A. S.; Khina, B. B.; Khusid, B. M. Dokl. Phys. Chem. 1992, 322, 78-83. Khusid, B. M.; Khina, B. B.; Demidkov, S. V. J. Mater. Sci. 1994, 29, 2187-2191. Khusid, B. M.; Mansurov, V. A.; Ubortsev, A. D.; Shulman, Z. P.; Merzhanov, A. G. J. Chem. Soc. Faraday Trans. 1993, 89, 727-732. Shteinberg, A. S.; Knyazik, V. A. Pure Appl. Chem. 1992, 64, 965-976.
[5] [6] [7] [8] [9] [10] [11] [12]
[13] [14] [15] [16]
94
B. B. Khina
[17] Rogachev, A. S.; Mukasyan, A. S.; Merzhanov, A. G. Dokl. Phys. Chem. 1987, 297, 1240-1243. [18] Merzhanov, A. G.; Rogachev, A. S.; Mukasyan, A. S.; Khusid, B. M. Comb. Expl. Shock Waves 1990, 26, 92-102. [19] LaSalvia, J. C.; Kim, D. K.; Lipsett, R. A.; Meyers, M. A. Metall. Mater. Trans. A 1995, 26, 3001-3009; LaSalvia, J. C.; Meyers, M. A. ibid. 1995, 26, 3011-3019. [20] Sung-Won, J.; Gi-Wook, L.; Jong-Tae, M.; Yong-Seog, K. Acta Mater. 1996, 44, 4317-4326. [21] Fan, Q.; Chai, H.; Jin, Z. Intermetallics 2001, 9, 609-619. [22] Fan, Q.; Chai, H.; Jin, Z. Intermetallics 2002, 10, 541-554. [23] Zhu, P.; Li, J. C. M.; Liu, C. T. Mater. Sci. Eng. A 2002, 329-331, 57-68. [24] Christian, J. W. The Theory of Transformations in Metals and Alloys; Pergamon Press: Oxford, UK, 1975. [25] Lyubov, B. Ya. Kinetic Theory of Phase Transformations; National Bureau of Standards: Gaithersburg, MD, 1978. [26] Moore, J. J.; Feng, H. J. Progr. Mater. Sci. 1995, 39, 275-316. [27] Makino, A. Progr. Energy Comb. Sci. 2001, 27, 1-74. [28] Rogachev, A. S.; Baras, F. Int. J. SHS 2007, 16, 141-153. [29] Mukasyan, A. S.; Rogachev, A. S. Progr. Energy Comb. Sci. 2008, 34, 377416. [30] Shkadinskii, K. G.; Khaikin, B. I.; Merzhanov, A. G. Comb. Expl. Shock Waves 1971, 7, 15-22. [31] Aldushin, A. P.; Martem'yanova, T. M.; Merzhanov, A. G.; Khaikin, B. I.; Shkadinskii, K. G. Comb. Expl. Shock Waves 1972, 8, 159-164. [32] Zeldovich, Ya. B.; Barenblatt, G. I.; Librovich, V. B.; Makhviladze, G. M. The Mathematical Theory of Combustion and Explosions; Plenum Press: New York, NY, 1985. [33] Ivleva, T. P.; Merzhanov, A. G.; Shkadinskii, K. G. Sov. Phys. Dokl. 1978, 239, 255-256. [34] Ivleva, T. P.; Merzhanov, A. G. Phys. Rev. E 2001, 64, 036218 (4 pp). [35] Astapchik, A. S.; Podvoisky, E. P.; Chebotko, I. S.; Khusid, B. M.; Merzhanov, A. G.; Khina, B. B. Phys. Rev. E 1993, 47, 319-326. [36] Khusid, B. M.; Khina, B. B.; Podvoisky, E. P.; Chebotko, I. S. Comb. Flame 1994, 99, 371-378. [37] Kirdyashkin, A. I.; Maksimov, Yu. M.; Nekrasov, E. A. Comb. Expl. Shock Waves 1981, 17, 377-379. [38] Khusid, B. M.; Khina, B. B.; Bashtovaya, E. A. Comb. Expl. Shock Waves 1991, 27, 708-714.
References
95
[39] Khusid, B. M.; Khina, B. B.; Bashtovaya, E. A.; Quang, V. Z. Comb. Expl. Shock Waves 1992, 28, 389-395. [40] Hu, G.; Zhang, L. Comp. Mater. Sci. 2008, 42, 558-563. [41] Zhang, L.; Wang, Z. Exper. Thermal Fluid Sci. 2008, 32, 1255-1263. [42] Smithells Metals Reference Book; Brandes, E. A.; Brook, G. B.; Eds.; Butterworth-Heinemann: Oxford, UK, 1992. [43] Wilson, J. R. Metall. Review 1965, 40, 381-590. [44] Hwang, S.; Mukasyan, A. S.; Varma, A. Comb. Flame 1998, 115, 354-363. [45] Varma, A.; Mukasyan, A. S.; Hwang, S. Chem. Eng. Sci. 2001, 55, 14591466. [46] Kochetov, N. A.; Rogachev, A. S.; Merzhanov, A. G. Dokl. Phys. Chem. 2003, 389, 80-82. [47] Merzhanov, A. G.; Kovalev, D. Yu.; Shkiro, V. M.; Ponomarev, V. I. Dokl. Phys. Chem. 2004, 394, 34-38. [48] Merzhanov, A. G. Dokl. Phys. Chem. 1997, 353, 135-138. [49] Rogachev, A. S.; Merzhanov, A. G. Dokl. Phys. Chem. 1999, 365, 127-130. [50] Beck, J. M.; Volpert, V. A. Physica D 2003, 182, 86-102. [51] Rabinovich, O. S.; Grinchuk, P. S.; Khina, B. B.; Belyaev, A.V. Int. J. SHS 2003, 11, 257-270. [52] Aldushin, A. P.; Khaikin, B. I. Comb. Expl. Shock Waves 1974, 10, 273283. [53] Aldushin, A. P.; Kasparyan, S. G.; Shkadinskii, K. G. In Combustion and Explosion; Stesik, L. N.; Ed.; Nauka: Moscow, 1977, pp 207-212. [54] Gennari, S.; Maglia, F.; Anselmi-Tamburini, U.; Spinolo, G. J. Phys. Chem. B 2003, 107, 732-738. [55] Gennari, S.; Maglia, F.; Anselmi-Tamburini, U.; Spinolo, G. J. Phys. Chem. B 2004, 108, 19550-19556. [56] Gennari, S.; Anselmi-Tamburini, U.; Maglia, F.; Spinolo, G.; Munir, Z. A. Acta Mater. 2006, 54, 2343-2351. [57] Locci, A. M.; Cincotti, A.; Delogu, F.; Orru, R.; Cao, G. Chem. Eng. Sci. 2004, 59, 5121-5128. [58] Locci, A. M.; Cincotti, A.; Delogu, F.; Orru, R.; Cao, G. J. Mater. Res. 2005, 20, 1257-1268. [59] Locci, A. M.; Cincotti, A.; Delogu, F.; Orru, R.; Cao, G. J. Mater. Res. 2005, 20, 1269-1277. [60] Khina, B. B.; Formanek, B.; Solpan, I. Physica B 2005, 355, 14-31. [61] Khina, B. B. Int. J. SHS 2005, 14, 21-31. [62] Nekrasov, E. A.; Smolyakov, V. K.; Maksimov, Yu. M. Comb. Expl. Shock Waves 1981, 17, 305-311.
96
B. B. Khina
[63] Nekrasov, E. A.; Smolyakov, V. K.; Maksimov, Yu. M. Comb. Expl. Shock Waves 1981, 17, 513-520. [64] Smolyakov, V. A.; Nekrasov, E. A.; Maksimov, Yu. M. Comb. Expl. Shock Waves 1982, 18, 312-315. [65] Maksimov, Yu. M.; Smolyakov, V. A.; Nekrasov, E. A. Comb. Expl. Shock Waves 1984, 20, 479-486. [66] Nekrasov, E. A.; Tkachenko, V. N.; Zakirov, A. E. Comb. Sci. Technol. 1993, 91, 207-223. [67] Biswas, A.; Roy, S. K.; Gurumurthy, K. R.; Prabhu, N.; Banerjee, S. Acta Mater. 2002, 50, 757-773. [68] Oliveira, A. A. M.; Kaviany, M. Int. J. Heat Mass Transfer 1999, 42, 10751095. [69] Vrel, D.; Lihrmann, J. M.; Tobaly, P. J. Mater. Synth. Proc. 1994, 2, 179187. [70] Armstrong, R. Comb. Sci. Technol. 1990, 71, 155-174. [71] Kidson, G. V. J. Nucl. Mater. 1961, 3, 21-29. [72] Gusak, A. M.; Gurov, K. P. Phys. Metal. Metallogr. 1982, 53, 842-847. [73] Williams, D. S.; Rapp, R. A.; Hirth, J. P. Thin Solid Films 1986 142, 47-64. [74] Baglin, J. E. E.; Poat, J. M. In Thin Films: Interdiffusion and Reactions; Poate, J. M.; Tu, K. N.; Mayer, J. W.; Eds.; J. Wiley and sons: New York, NY, 1978; chapter 9. [75] d'Heurle, F. M.; Gas, P. J. Mater. Res. 1986, 1, 205-221. [76] Dybkov, V. I. J. Mater. Sci. 1993, 28, 6371-6380; idem, ibid, 2000, 35, 1729-1736. [77] Bouche, K.; Barbier, F.; Coulet, A. Mater. Sci. Eng. A 1998, 249, 167-175. [78] Gosele, U.; Tu, K. N. J. Appl. Phys. 1982, 53, 3252-3260. [79] Desre, P. J.; Yavary, A. R. Phys. Rev. Lett. 1990, 64, 1533-1536. [80] Gusak, A. M.; Gurov, K. P. Solid State Phenom. 1992, 23-24, 117-122. [81] Khusid, B. M.; Khina, B. B. Phys. Rev. B 1991, 44, 10778-10793. [82] Atzmon, M. Metall. Trans. A 1992, 23, 49-53. [83] Farber, L.; Klinger, L.; Gotman, I. Mater. Sci. Eng. A 1998, 254, 155-165. [84] Janssen, M. M. P.; Riek, G. D. Trans. Met. Soc. AIME 1967, 239, 13721385. [85] Janssen, M. M. P. Metall. Trans. 1973, 4, 1623-1633. [86] Song, H. Y.; Wang, X. J. Mater. Sci. 1996, 31, 3281-3288. [87] Elliot, R. P. Constitution of Binary Alloys: First Supplement; McGrow-Hill: New York, NY, 1965. [88] Zenin, A. A.; Korolev, Yu. M.; Popov, V. T.; Tyurkin, Yu. V. Sov. Phys. Dokl. 1986, 31, 239-242.
References
97
[89] Samsonov, G. V.; Vinitskii, I. M. Handbook of Refractory Compounds; IFI/Plenum: New York, NY, 1980. [90] Samsonov, G. V.; Upadhaya, G. S.; Neshpor, V.S. Physical Metallurgy of Carbides; Naukova Dumka: Kiev, 1974. [91] Dergunova, V. S.; Levitskiy, Yu. V.; Shurshakov, A. N.; Kravetskiy, G. A. Interaction of Carbon with Refractory Metals; Metallurgiya: Moscow, 1974. [92] Shkiro, V. M.; Borovinskaya, I. P. Comb. Expl. Shock Waves 1976, 12, 828831. [93] Nekrasov, E. A.; Maksimov, Yu. M.; Ziatdinov, M. Kh.; Shteinberg, A. S. Comb. Expl. Shock Waves 1978, 14, 575-581. [94] Kirdyashkin, A. I.; Maksimov, Yu. M.; Merzhanov, A. G. Comb. Expl. Shock Waves 1981, 17, 591-595. [95] Kiparisov, S. S.; Levinskiy, Yu. V.; Petrov, A. P. Titanium Carbide: Synthesis, Properties, Applications; Metallurgiya: Moscow, 1987. [96] Andrievskii, R. A.; Spivak, I. I. Strength of Refractory Compounds and Materials on Their Basis: a Handbook; Metallurgiya: Chelyabinsk, 1989. [97] Sarian, S. J. Appl. Phys. 1969, 40, 3515-3520. [98] Sarian, S. J. Appl. Phys. 1968, 39, 3305-3310. [99] Sarian, S. J. Appl. Phys. 1968, 39, 5036-5041. [100] Kohlstedt, D. L.; Williams, W. S.; Woodhouse, J. B. J. Appl. Phys. 1970, 41, 4476-4484. [101] Adelsberg, L. M.; Cadoff, L. H. Trans. Metall. Soc. AIME 1967, 239, 933935. [102] Ushakov, B. F.; Zagryazkin, V. N.; Panov, A. S. Inorg. Mater. USSR 1972, 8, 1921-1925. [103] Van Loo, F. J. J.; Wakelkamp, W.; Bastin, G. F.; Metsalaar, R. Solid State Ionics 1989, 32-33, 824-832. [104] Iida, T.; Guthrie, R. I. L. The Physical Properties of Liquid Metals; Clarendon Press: Oxford, UK, 1988. [105] Lepinskih, B. M.; Kaybichev, A. V.; Savel'ev, Yu. A. Diffusion of Elements in Liquid Metals of the Iron Group; Nauka: Moscow, 1974. [106] Properties of Elements: a Handbook, vol. 1: Physical Properties; Samsonov, G. V.; Ed.; Metallurgiya: Moscow, 1976. [107] Kostikov, V. I.; Varenkov, A. N. Interaction of Metallic Melts with Carbon Materials; Metallurgiya: Moscow, 1981. [108] Frank-Kamenetskiy, D. A. Diffusion and Heat Transfer in Chemical Kinetics; Plenum Press: New York, NY, 1969.
98
B. B. Khina
[109] Tables of Physical Parameters; Kikoin, I. K.; Ed.; Atomizdat: Moscow, 1976. [110] Kubaschewski, O.; Alcock, C. B. Metallurgical Thermochemistry; Pergamon Press: Oxford, UK, 1979. [111] Lyubov, B. Ya. Theory of Crystallization in Large Volumes; Nauka: Moscow, 1975. [112] Kartashov, E. M.; Lyubov, B. Ya. Heat Transfer-Soviet Research 1976, 8 (2), 1-39. [113] Landau, L. D.; Lifshiz, E. M. Theory of Elasticity; Pergamon Press: Oxford, UK, 1986. [114] Holt, J. B.; Munir, Z. A. J. Mater. Sci. 1986, 21, 251-259. [115] Dubois, S.; Karnatak, N.; Beaufort, M. F.; Vrel, D. J. Mater. Synth. Proc. 2001, 9, 253-257. [116] Adachi, S.; Wada, T.; Mahara, T.; Miyamoto, Y.; Koizumi, M.; Yamada, O. J. Amer. Ceram. Soc. 1989, 72, 805-809. [117] Efimov, Yu. E.; Zaripov, N. G.; Bloshenko, V. N.; Bokii, V. A.; Kashtanova, A. A.; Belinskaya, U. L. Comb. Expl. Shock Waves 1992, 28, 496-499. [118] Eremeev, V. S. Diffusion and Stresses; Energoatomizdat: Moscow, 1984. [119] Shcherbakov, V. A.; Shteinberg, A. S. Dokl. Phys. Chem. 1995, 345, 275279. [120] Scher, H.; Zallen, R. J. Chem. Phys. 1970, 53, 3759-3761. [121] Stauffer, D.; Aharony, A. Introduction to Percolation Theory; Taylor and Francis: London, UK, 1994. [122] Aleksandrov, V. V.; Korchagin, M. A. Comb. Expl. Shock Waves 1987, 23, 557-564. [123] He, C.; Blanchetiere, C.; Stangle, G. C. J. Mater. Res. 1998, 13, 2269-2280. [124] Sytschev, A. E.; Merzhanov, A. G. Russ. Chem. Reviews 2004, 73, 147-159. [125] Deevi, S. C.; Sikka, V. K.; Liu, C. T. Prog. Mater. Sci. 1997, 42, 177-192. [126] Stoloff, N. S.; Liu, C. T.; Deevi, S. C. Intermetallics 2000, 8, 1313-1320. [127] Kolomytsev, P. T. High-temperature Protective Coatings for Nickel-base Alloys; Metallurgiya: Moscow, 1991. [128] Rogachev, A. S. Russ. Chem. Reviews 2008, 77, 21-37. [129] Swiston, A. J.; Hufnagel, T. C.; Weihs, T. P. Scripta Mater. 2003, 48, 15751580. [130] Wang, J.; Besnoin, E.; Knio, O. M.; Weihs, T. P. Acta Mater. 2004, 52, 5265-5274. [131] Trenkle, J. C.; Wang, J.; Weihs, T. P.; Hufnagel, T. C. Appl. Phys. Lett. 2005, 87, 153108 (3 pp).
References
99
[132] Pascal, C.; Marin-Ayral, R. M.; Tedenac, J. C. J. Alloys Compounds 2002, 337, 221-225. [133] Besnoin, E.; Cerutti, S.: Knio, O .M.; Weihs, T. P. J. Appl. Phys. 2002 92, 5474-5481. [134] Duckham, A.; Spey, S. J.; Wang, J.; Weihs, T. P. J. Appl. Phys. 2004, 96, 2336-2342. [135] Wang, J.; Besnoin, E.; Knio, O. M.; Weihs, T. P. J. Appl. Phys. 2005, 97, 114307 (7 pp). [136] Shteinberg, A. S.; Shcherbakov, V. A.; Munir, Z. A. Comb. Sci. Technol. 2001, 169, 1-24. [137] Anselmi-Tamburini, U.; Munir, Z. A. J. Appl. Phys. 1989, 66, 5039-5045. [138] Binary Alloy Phase Diagrams; Massalski, T. B.; Okamoto, H.; Subramanian, P. R.; Kacprzak, L.; Eds.; AMS: Materials Park, OH, 1990, p 183. [139] Nash, P.; Singleton, M. F.; Murray, J. L. In Phase Diagrams of Binary Nickel Alloys; Nash, P.; Ed.; ASM: Metals Park, OH, 1991; p 3. [140] Gurov, K. P.; Kartashkin, B. A.; Ugaste, U. E. Interdiffusion in Multiphase Metallic Systems; Nauka: Moscow, 1981. [141] Lapshin, O. V.; Ovcharenko, V. E. Comb. Explos. Shock Waves 1996, 32, 299-305. [142] Kovalev, O. B.; Neronov, V. A. Comb. Explos. Shock Waves 2004, 40, 172179. [143] Mann, A. B.; Gavens, A. J.; Reiss, M. E.; Van Heerden, D.; Bao, G.; Weihs, T. P. J. Appl. Phys. 1997, 82, 1178-1188. [144] Jayaraman, S.; Knio, O. M.; Mann, A. B.; Weihs, T. P. J. Appl. Phys. 1999, 86, 800-809. [145] Jayaraman, S.; Mann, A. B.; Reiss, M.;. Weihs, T. P; Knio, O. M. Comb. Flame 2001, 124, 178-194. [146] Janssen, M. M. P.; Riek, G. D. Trans. Metall. Soc. AIME 1967, 239, 13721385. [147] Janssen, M. M. P. Metall. Trans. 1973, 4, 1623-1633. [148] Hickl A. J.; Heckel, R. W. Metall. Trans. A 1975, 6, 431-440. [149] Garcia, V. H.; Mors, P. M.; Scherer, C. Acta Mater. 2000, 48, 1201-1206. [150] Watanabe, M.; Horita, Z.; Nemoto, M. Defect Diffus. Forum 1997, 143-147, 345-350. [151] Nakajima, H.; Sprengel, W.; Nonaka, K. Intermetallics 1996, 4, S17-S28. [152] Herzig, C.; Divinski, S. V.; Frank, St.; Przeorski, T. Defect Diffus. Forum 2001, 194-199, 317-336. [153] Herzig C.; Divinski, S. Intermetallics 2004, 12, 993-1003.
100
B. B. Khina
[154] Nakamura, Ryusuke; Takasawa, Koichi; Yamazaki, Yoshihiro; Iijima, Yoshiaki. Intermetallics 2002, 10, 195-204. [155] Helander, T.; Agren, J. Acta Mater. 1998, 47, 1141-1152. [156] Wei, H.; Sun, X.; Zheng, Q.; Guan, H.; Hu, Z. Acta Mater. 2004, 52, 26452651. [157] Khina, B. B. J. Appl. Phys. 2007, 101, 063510 (11 pp). [158] Khina, B. B.; Formanek, B. Int. J. SHS 2007, 16, 51-61. [159] Novikov, N. P.; Borovinskaya, I. P.; Merzhanov, A. G. In Combustion Processes in Chemical Technology and Metallurgy; Merzhanov, A. G.; Ed.; Institute of Chemical Physics: Chernogolovka, 1975; pp 174-188. [160] Barin, I.; Knacke, I.; Kubaschevski, O. Thermochemical Properties of Inorganic Substances: Supplement; Springer-Verlag: Berlin, 1977; p 490. [161] Pelekh, A. E.; Mukasyan, A. S.; Varma, A. Ind. Eng. Chem. Res. 1999, 38, 793-798. [162] Kharatyan, S. L.; Chatilyan, H. A.; Mukasyan, A. S.; Simonetti, D. A.; Varma, A. AIChE J. 2005, 51, 261-270. [163] Ma, E.; Thompson, C. V.; Clevenger, L. A. J. Appl. Phys. 1991, 69, 22112218. [164] Qiu, X.; Wang, J. Scr. Mater. 2007, 56, 1055-1058. [165] Michaelsen, C.; Lucadamo, G.; Barmak, K. J. Appl. Phys. 1996, 80, 66896698. [166] Barmak, K.; Michaelsen, C.; Lucadamo, G. J. Mater. Res. 1997, 12, 133146. [167] Edelstein, A. S.; Everett, R. K.; Richardson, G. Y.; Qadri, S. B.; Altman, E. I.; Foley, J. C.; Perepezko, J. H. J. Appl. Phys. 1994, 76, 7850-7859. [168] Trenkle, J. C.; Koerner, L. J.; Tate, M. W.; Gruner, S. M.; Weihs, T. P.; Hufnagel, T. C. Appl. Phys. Lett. 2008, 93, 081903 (3 pp). [169] Sinelnikova, V. S.; Podergin, V. A.; Rechkin, V. N. Aluminides: a Handbook; Naukova Dumka: Kiev, 1965. [170] Properties of Elements: a Handbook; Drits, M. E.; Ed.; Metallurgiya: Moscow, 1985. [171] Zinoviev, V. E. Thermophysical Properties of Metals at High Temperatures: a Handbook; Metallurgiya: Moscow, 1989. [172] Gunduz, I. E.; Fadenberger, K.; Kokonou, M.; Rebholz, C.; Doumanidis, C. C. Appl. Phys. Lett., 2008, 93, 134101 (3 pp). [173] Gunduz, I. E.; Fadenberger, K.; Kokonou, M.; Rebholz, C.; Doumanidis, C. C.; Ando, T. J. Appl. Phys. 2009, 105, 074903 (9 pp). [174] Garg, S. P.; Kale, G. B.; Patil, R. V.; Kundu, T. Intermetallics 1999, 7, 901908.
References
101
[175] Myagkov, V. G.; Bykova, L. E. Dokl. Phys. 2004, 49, 289-291. [176] Myagkov, V. G.; Zhigalov, V. S.; Bykova, L. E.; Maltsev, V. K. Tech. Phys. 1998, 43, 1189-1192. [177] Rabinovich, O. S.; Grinchuk, P. S.; Andreev, M. A.; Khina, B. B. Physica B 2007, 392, 272-280. [178] Shishkovsky, I.; Morozov, Yu.; Smurov, I. Appl. Surf. Sci. 2009, 255, 55655568. [179] Langelier, B. C.; Esmaeili, S. J. All. Comp. 2009, 482, 246-252. [180] Gu, D.; Shen, Y; Lu, Z. Mater Lett. 2009, 63, 1577-1579. [181] Unocic, R. R.; DuPont, J. N. Metall. Mater. Trans. B 2003, 34, 439-445. [182] Bernard, F.; Gaffet, E. Int. J. SHS 2001, 10, 109-132. [183] Maglia, F.; Milanese, C.; Anselmi-Tamburini U.; Doppiu, S.; Cocco, G. J. Mater. Res. 2002, 17, 1992-1999. [184] Medda, E.; Delogu, F.; Cao, G. Mater. Sci. Eng. A 2003, 361, 23-28. [185] Gras, C., Gaffet. E., Bernard, F. Intermetallics 2006, 14, 521-529. [186] Korchagin, M. A.; Grigor'eva, T. F.; Bokhonov, B. B.; Sharafutdinov, M. R.; Barinova. A. P.; Lyakhov, N. Z. Comb. Explos. Shock Waves 2003, 39, 43-50. [187] Korchagin, M. A.; Grigor'eva, T. F.; Bokhonov, B. B.; Sharafutdinov, M. R.; Barinova. A. P.; Lyakhov, N. Z. Comb. Explos. Shock Waves 2003, 39, 51-58. [188] Suryanarayana, C. Progr. Mater. Sci. 2001, 46, 1-184. [189] Zhang, D. L. Progr. Mater. Sci. 2004, 49, 537-560. [190] Lapshin, O. V. Macrokinetics of High-temperature Synthesis of Chemical Compound in the Conditions of Thermal Explosion of Powder Mixtures: D.Sc. Thesis; Tomsk State University: Tomsk, 2007. [191] Smolyakov, V. K. Comb. Explos. Shock Waves 2005, 41, 319-325. [192] Smolyakov, V. K.; Lapshin, O. V.; Boldyrev, V. V. Int. J. SHS 2007, 16, 111. [193] Smolyakov, V. K.; Lapshin, O. V.; Boldyrev, V. V. Theor. Foundations Chem. Eng. 2008, 42, 54-59. [194] Smolyakov, V. K.; Lapshin, O. V.; Boldyrev, V. V. Theor. Foundations Chem. Eng. 2008, 42, 187-196. [195] Abdulkarimova, R. G.; Ketegenov, T. A.; Mansurov, Z. A.; Lapshin, O. V.; Prokofev, V. G.; Smolyakov, V. K. Comb. Explos. Shock Waves 2009, 45, 48-58. [196] Levashov, E. A.; Kurbatkina, V. V.; Rogachev, A. S.; Kochetov, N. A. Int. J. SHS 2007, 16, 46-50. [197] Khina, B. B.; Formanek, B. Sol. State Phenom. 2008, 138, 159-164
102
B. B. Khina
[198] Khina, B. B. Int. J. SHS 2008, 17, 211-217. [199] Boldyrev, V. V. Russ. Chem. Reviews 2006, 75, 177-189. [200] Butyagin, P. J. Mater. Synth. Process. 2000, 8, 205-211. [201] Delogu, F.; Cocco, G. J. Mater. Synth. Process. 2000, 8, 271-277. [202] Delogu, F.; Mulas, G.; Schiffini, L.; Cocco, G. Mater. Sci. Eng. A 2004, 382, 280-287. [203] El-Eskandarany, M. S. Mechanical Alloying for Fabrication of Advanced Engineering Materials; Noyes Publ.: Norwich, NY, 2001. [204] Khina, B. B.; Froes, F. H. J. Metals (JOM) 1996, 48 (7), 36-38. [205] Valiev, R. Z.; Islamgaliev, R. K.; Alexandrov, I. V. Progr. Mater. Sci. 2000, 45, 103-189. [206] Bever, M. B.; Holt, D. L.; Titchener, A. L. Progr. Mater. Sci. 1973, 17, 1190. [207] Delogu, F.; Schiffini, L.; Cocco, G. Phil. Mag. A 2001, 81, 1917-1937. [208] Fecht, H. J. Mater. Sci. Eng. A 1994, 179-190, 491-494. [209] Highmore, J.; Greer, A. L. Nature 1989, 339 (6223), 363-365. [210] Wu, X. L.; Ma, E. Appl. Phys. Lett. 2006, 88, 231911 (3 pp). [211] Wang, Y. B.; Ho, J. C.; Cao, Y.; Liao, X. Z.; Li, H. Q.; Zhao, Y. H.; Lavernia, E. J.; Ringer, S. P.; Zhu, Y. T. Appl. Phys. Lett. 2009, 94, 091911 (3 pp). [212] Dieter, G. E. Mechanical Metallurgy; McGraw-Hill: New York, NY, 1986. [213] Tikhonov, L. V.; Kononenko, V. A.; Prokopenko, G. I.; Rafalovski, V. A. Structure and Properties of Metals and Alloys: A Handbook; Naukova Dumka: Kiev, 1986. [214] Frost, H. J.; Ashby, M. F. Deformation-Mechanism Maps; Pergamon Press: Oxford, UK, 1982. [215] Zhao, Y. H.; Sheng, H. W.; Lu, K. Acta Mater. 2001, 49, 365-375. [216] Zhao, Y. H.; Lu, K.; Zhang, K. Phys. Rev. B 2002, 66, 085404 (8 pp). [217] Gupta, D. Interface Sci. 2003, 11, 7-20. [218] Nabarro, F. R. N.; Basinski, Z. S.; Holt, D. B. Adv. Phys. 1964, 50, 193-323. [219] Bokshtein, B. S. Diffusion in Metals; Metallurgiya: Moscow, 1978. [220] Mehrer, H. Diffusion in Solids: Fundamentals, Methods, Materials, Diffusion-Controlled Processes; Springer-Verlag: Berlin, 2007, p 79. [221] Orlov, A. N.; Trushin, Yu. V. Energies of Point Defects in Metals; Energoatomizdat: Moscow, 1983. [222] Gupta, D. In Diffusion Processes in Advanced Technological Materials; Gupta, D.; Ed.; William Andrew, Inc.: Norwich, NY, 2005; pp 1-68. [223] Maurice, D. R.; Courtney, T. H. Metall. Trans. A 1990, 21, 289-303.
References
103
[224] Lovshenko, G. F.; Khina, B. B. Trenie Iznos [Frict. Wear] 2005, 26, 434445. [225] Ungar, T.; Schafler, E.; Hanak, P.; Bernstorff, S.; Zehetbauer, M. Mater. Sci. Eng. A 2007, 462, 398-401. [226] Setman, D.; Schafler, E.; Korznikova, E.; Zehetbauer, M. J. Mater. Sci. Eng. A 2008, 493, 116-122. [227] Khina, B. B.; Solpan, I.; Lovshenko, G. F. J. Mater. Sci. 2004, 39, 51355138. [228] Khina, B. B.; Lovshenko, F. G.; Konstantinov, V. M.; Formanek, B. Metal Phys. Advanced Technol. 2005, 27, 609-623. [229] Khina, B. B.; Formanek, B. Defect Diffus. Forum 2006, 249, 105-110. [230] Gras, C.; Gaffet, E.; Bernard, F.; Niepce, J. C. Mater. Sci. Eng. A 1999, 264, 94-107. [231] Rabinovich, V. A.; Khavin, Z. Ya. Brief Chemical Handbook; Khimiya: Leningrad, 1978; p 56. [232] Ermilov, A. G.; Safonov, V. V.; Doroshko, L. F.; Kolyakin, A. V.; Polushin, N. I. Izv. Vyssh. Uchebn. Zaved. Tsvetn. Metall. 2002, 3, 48-53. [233] Barin, I.; Knacke, O. Thermochemical Properties of Inorganic Substances; Springer-Verlag: Berlin, 1973. [234] Klein, D.; Niepce, J. C.; Charlot, F.; Gaffet, E.; Bernard, F. Acta Mater. 1999, 47, 619-629. [235] Johnson, R. A.; Lam, N. Q. Phys. Rev. B 1976, 13, 4364-4375. [236] Skakov, Yu. A. Metal Sci. Heat Treat. 2004, 46, 137-145. [237] Shtremel, M. A. Metal Sci. Heat Treat. 2004, 46, 146-147. [238] Gapontsev, V. L.; Koloskov, V. M. Metal Sci. Heat Treat. 2007, 49, 503513. [239] Skakov, Yu. A. Metal Sci. Heat Treat. 2007, 49, 514-516. [240] Shtremel, M. A. Metal Sci. Heat Treat. 2007, 49, 517-518. [241] Hodaj, F.; Gusak, A. M.; Desre, P. J. Phil. Mag. A 1998, 77, 1471-1479. [242] Gusak, A. M.; Hodaj, F.; Bogatyrev, A. O. J. Phys. Cond. Matter 2001, 13, 2767-2787. [243] Hodaj, F.; Gusak, A. M. Acta Mater. 2004, 52, 4305-4315. [244] Pasichnyy, M. O.; Schmitz, G.; Gusak, A. M.; Vovk, V. Phys. Rev. B 2005, 72, 014118 (7 pp). [245] Chung, C. Y.; Zhu, M.; Man, C. H. Intermetallics 2002, 10, 865-871. [246] Talako, T. L.; Grigor'eva, T. F.; Letsko, A. I.; Barinova, A. P.; Vitiaz, P. A.; Lyakhov, N. Z. Int. J. SHS 2009, 18, 125-132.
INDEX
A
accounting, 35, 103, 107 accuracy, 80 acid, 4, 109 activation, xiv, 5, 7, 14, 15, 23, 44, 51, 62, 69, 89, 95, 96, 97, 98, 99, 100, 106, 107, 110, 111, 115 activation energy, 5, 7, 14, 15, 23, 44, 51, 62, 69, 89, 97, 100, 106, 107, 110, 111, 115 adiabatic, 2, 6, 24, 26, 27, 28, 29, 30, 31, 34, 35, 36, 53, 54, 60, 72, 96, 109 aluminum, 9, 14, 50, 51, 54, 57, 60, 61, 67, 68, 73, 74, 75, 78, 95, 106, 110, 111, 112, 114 amorphization, 100, 102, 106 amorphous, 99, 101 amorphous phases, 99 AMS, 123 annealing, 47, 51, 55, 69, 71, 111 annihilation, 104 assessment, 97 assumptions, 13, 81 asymmetry, 76 asymptotic, 27, 34 atoms, 12, 20, 21, 22, 23, 31, 37, 38, 41, 42, 44, 69, 71, 79, 91, 107, 113 automata, 7
B
barrier, 13, 14, 114 behavior, 7, 51, 77, 79, 85 bending, 33 blocks, 107 boundary conditions, 31, 32, 37 bulk nanostructured materials, 100 burn, 6, 25, 45, 77 burning, 97
C
CAP, 105 carbide, 3, 4, 12, 17, 21, 22, 24, 28, 33, 34, 35, 37, 38, 39, 43, 44, 46, 47, 80, 89, 91 carbides, xiii, 1, 9, 12, 21, 45, 89, 91 carbon, 10, 12, 18, 20, 21, 22, 23, 31, 34, 35, 36, 37, 39, 40, 41, 42, 43, 44, 45, 91 carbon atoms, 12, 22, 23, 41, 42 casting, 1 cell, 8, 12, 18, 20, 25, 26, 27, 28, 33, 34, 40, 41, 89 chemical bonds, 98 chemical interaction, 22, 24, 44, 110 chemical reactions, xiii, 24 chemical reactivity, 98 chemical stability, 17 chromium, 111, 114
106
Index
defects, 96, 98, 99, 100, 106, 107 deformation, 32, 96, 99, 100, 104, 105, 106, 107, 113 density, 5, 23, 28, 31, 38, 39, 40, 44, 61, 62, 67, 68, 69, 99, 102, 105, 107, 110, 111, 112 deposition, 93, 114 deviation, 14 differential scanning calorimetry, 58 diffusion process, 97 diffusion rates, 63 diffusion time, 79 diffusivities, 12 diffusivity, 21, 25, 38, 69 dislocation, 102, 103, 104, 105, 107, 110, 111, 112 disorder, 98, 107 displacement, 32, 37, 38, 41, 67, 76, 79, 80, 89, 91, 98 DSC, 58 ductility, 99 dynamic viscosity, 23
cladding, 93 classes, 11 classical, 5, 15, 16, 51, 52 coatings, 93 collisions, 98, 113 combustion, xiii, 2, 3, 5, 6, 7, 8, 9, 14, 15, 16, 24, 25, 27, 54, 57, 71, 73, 74, 88, 92, 95, 96, 97, 107, 108, 109, 113 compaction, 39 competition, 58, 59, 76, 82, 92 components, 23, 49, 98, 104, 110 composition, 4, 8, 13, 15, 28, 39, 52, 56, 69, 73, 77, 92, 93, 95, 101, 109, 114 compounds, xiii, 1, 2, 3, 4, 9, 11, 16, 21, 33, 45, 89, 92, 98, 109 concentration, 13, 20, 21, 41, 42, 56, 58, 61, 62, 69, 73, 89, 96, 104, 105, 111, 113, 114, 115 condensed matter, 107 condensed media, xiii conductivity, 5 configuration, 14, 17 constitution, 120 consumption, 1, 45, 55, 67, 79, 81 continuity, 31 control, 36 conversion, 4, 5, 6, 34, 35, 36, 37, 48, 74, 76, 86, 87, 90, 95, 96, 107, 114 cooling, 4, 7, 8, 57, 73, 77 copper, 104, 105 corrosion, 49 couples, 13, 14, 55, 69 covalent, 23 critical analysis, 97 critical value, 114 crystal lattice, 43, 99, 100, 110, 113 crystalline, 12, 43, 101 crystallization, 7, 9, 10, 39, 44, 47, 60, 79, 81, 82, 86, 87, 88, 90, 92, 101, 114 crystallization kinetics, 10, 87
E
elaboration, 92 elastic deformation, 32 elasticity, 102 energy, 1, 7, 15, 97, 98, 99, 100, 102, 103, 104, 105, 106, 107, 108, 109, 110, 113 energy consumption, 1 environment, 45, 88 estimating, 16, 97, 107 evaporation, 5 evolution, 60 experimental condition, 87 extraction, 98 extrapolation, 23 extrusion, 1, 105
F D
data set, 30, 36, 46, 47, 69, 70, 71, 74, 75, 77, 78, 81, 82, 83, 84, 85, 87, 88, 109 film, 13, 14, 31, 41, 42, 43, 49, 52, 54, 55, 60, 62, 66, 67, 71, 73, 74, 77, 79, 80, 81, 86, 87, 88, 89
Index
film thickness, 41, 42, 80 films, 51, 56, 57, 73 flame, 1, 7, 8 flame propagation, 8 fluid, 32 foils, 49, 51, 53, 87 fragmentation, 107 fusion, 9, 27, 101, 106, 109, 110
107
I
impact energy, 98 impurities, 1, 5 in situ, 57 incompressible, 32 industrial, 11, 17 industrial application, 11 industry, 98 inert, 8 infinite, 28, 37 inhomogeneity, 53 initial state, 108 inorganic, 98, 99 insight, 52, 115 integration, 34 interface, 10, 11, 12, 13, 21, 22, 28, 37, 38, 41, 44, 52, 55, 56, 57, 60, 61, 66, 72, 73, 98, 103, 104, 114 interface energy, 104 intermetallic compounds, 8, 56 intermetallics, xiii, 1, 9, 10, 12, 45, 58, 89, 92, 109 interstitial, xiv, 12, 21, 37, 45, 89, 91, 113 interstitials, 105, 113 interval, 25, 59, 63 intrinsic, 6, 10, 11, 52, 89, 97 ions, 110 IPPD, 96, 97, 100, 102, 103, 105, 106, 107, 113, 114 iron, 8, 106, 111, 112, 114
G
gas, 3, 4, 5, 39, 40, 49, 91, 105, 108, 110 generation, 113 glasses, 49 graduate students, xiv grain, 1, 4, 88, 99, 100, 102, 104, 110 grain boundaries, 100, 102, 104 grains, 10, 39, 40, 41, 44, 50, 52, 54, 59, 79, 81, 86, 87, 88, 90, 92, 102, 103, 104, 107 graphite, 18, 21, 35, 39, 45 growth mechanism, 35, 47, 76 growth rate, 9, 41, 59, 66, 76
H
hardness, 17 heat capacity, 5, 27, 72, 108 heat loss, 88, 96 heat release, 5, 6, 9, 16, 24, 25, 26, 29, 31, 36, 40, 44, 53, 54, 72, 74, 89, 106, 107, 108, 113 heat removal, 54, 77 heat transfer, 5, 8, 53, 89, 92, 107 heating rate, 10, 14, 45, 46, 48, 52, 56, 57, 59, 67, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 87, 88, 89, 90, 92, 111, 112 heterogeneous, xiii, 5, 6, 8, 10, 44, 46, 50, 74, 78, 83, 85, 90, 92, 110, 114 heterogeneous systems, 10 high pressure, 39 high temperature, 4, 14, 39, 49, 52, 57, 79, 114 high-speed, 8, 95
J
justification, 57, 73
K
kinetic model, 8, 16, 73, 90 kinetics, 5, 8, 10, 11, 12, 13, 51, 52, 54, 55, 58, 62, 63, 70, 86, 87, 88, 92, 107, 112
108
Index
L
lamella, 101 lamellae, 13 lamellar, 50, 96, 104, 113, 114 laminated, 49 laser, 93 lattice, 38, 43, 80, 98, 99, 104, 107, 113, 114 law, 13, 26, 32, 57, 58, 72 lifetime, 45 linear, 27, 38, 42, 46, 62, 66, 74, 75, 78, 111 liquid metals, 7, 23, 68 liquid phase, 3, 54, 66, 68, 79, 80, 87, 88, 109 losses, 88, 96 low temperatures, 66, 69, 81 low-temperature, 14, 25 lying, 87
molar ratio, 15 mole, 102, 104, 109 molecules, 110 molybdenum, 106, 111, 112 morphology, 7 multi-component systems, 92 multilayer films, 57
N
nanocrystalline, 102, 104 nanocrystals, 102 natural, 110 nitrides, 1, 9, 12, 21 nitrogen, 91 nonequilibrium, 96 non-uniform, 53, 88 normal, 88, 104 nucleation, 13, 44, 56, 114, 115 nuclei, 14, 114 nucleus, 43, 80, 114
M
manufacturing, 49, 93 margin of error, 15 mass transfer, 5, 61, 76 matrix, 99, 113 mechanical properties, 11, 33, 99, 102, 107 mechanochemistry, 98 melting temperature, 7, 17, 29, 33, 45, 46, 50, 54, 59, 73, 75, 79, 80, 89, 91 melts, 44, 50, 57, 60, 63, 76, 79, 80, 81, 85, 88 memory, ix, 7 metal carbides, 91 metals, 7, 23, 56, 68, 74, 86, 99, 100, 101, 102, 103, 104, 105, 106, 107, 110, 111, 112, 113, 114 microstructure, 39, 40 migration, 113 minerals, 98 mixing, 98, 114 modeling, xiii, xiv, 5, 6, 7, 9, 10, 11, 14, 15, 44, 49, 50, 51, 52, 57, 62, 67, 69, 70, 71, 73, 74, 76, 80, 87, 88, 89, 90, 97, 106, 107 models, 8, 10, 13, 16, 48, 52, 86, 87, 90, 91, 92 modulus, 32, 33, 102, 103
O
observations, 39, 58 observed behavior, 76 organic, 4, 98, 109 oscillations, 8 oxide, 3 oxides, 1, 2, 114
P
parabolic, 12, 13, 58, 72 parameter, 7, 8, 12, 15, 16, 36, 51, 62, 72, 89, 97, 101, 108, 110, 111 particles, xiv, 7, 8, 9, 14, 17, 18, 22, 25, 31, 33, 35, 37, 38, 39, 40, 41, 43, 44, 45, 46, 48, 54, 93, 96, 98, 99, 101, 104, 113, 114 pathways, 12, 46, 90 patterning, 107 percolation, 8, 40 percolation theory, 40 periodic, 96
Index
permit, 90, 92 phase boundaries, 38, 51, 55, 58, 60, 61, 63, 64, 67, 72, 91 phase diagram, 10, 12, 13, 17, 19, 21, 29, 41, 50, 51, 52, 56, 59, 76, 80, 81, 89 physical and mechanical properties, 11, 99, 107 physicochemical, xiv, 95, 97, 113 planar, 9, 13, 14, 54, 60, 62, 90 planetary, 95, 96, 99 plasma, 93, 114 plastic, 96, 99, 100, 104, 106, 107 plastic deformation, 96, 99, 100, 104, 106, 107 plasticity, 32 play, 5 point defects, 96, 100, 104, 110, 113 pore, 39 pores, 40, 41 porosity, 37, 39, 40, 44 powder, 2, 49, 50, 88, 90, 93, 96, 98, 102, 110, 114 powders, 100, 105 precipitation, xiv, 9, 10, 11, 39, 41, 43, 44, 48, 50, 52, 53, 59, 76, 79, 87, 88, 91, 114 pressure, 32, 39, 40, 102, 105 propagation, 15 proportionality, 106 protective coating, 49, 93, 114
109
reaction mechanism, 8, 48, 49 reaction order, 5 reaction rate, 5 reaction zone, 6, 57, 74 reactivity, 7, 98 reagents, 107 recrystallization, 6 refining, 99 refractory, xiii, 1, 3, 4, 9, 16, 23, 33, 46, 48, 90, 91, 109, 114 regular, 65 relationships, 8, 64 relaxation, 96, 107, 110, 111, 112, 113 research and development, xiv, 98 resistance, 49, 53 rolling, 102, 105 room temperature, 53, 67, 69, 105, 109 roughness, 8
S
salt, 4 sample, 2, 5, 25, 28, 39, 40, 54 saturation, 105 scaling, 14 scattering, 107 series, 97, 107 shape, 1, 49, 80, 93 shaping, 93 shear, 32, 33, 76, 98, 103 short-term, 39, 113 simulation, 5, 52, 57, 60, 62, 66, 67, 69, 71, 76, 81, 85, 105, 108, 109 sintering, 25, 39, 41, 45, 93 sites, 41 solid phase, 9, 54, 68, 115 solid solutions, 3, 99 solid state, 95, 100 solubility, 12, 20, 28, 50, 56, 57, 58, 59, 66, 67, 69, 75, 76 spatial, 14 species, 21, 23 specific heat, 54 spectrum, 107 speed, 8, 15, 95
Q
qualitative concept, 14 quasi-equilibrium, xiii, 17, 41, 43, 45, 47, 48, 52, 54, 59, 66, 74, 87, 88, 89, 91, 92
R
race, 8 radius, 21, 22, 23, 29, 31, 36, 42, 43, 46, 88 rat, 58 reactant, xiv, 2, 3, 4, 7, 9, 10, 12, 13, 14, 44, 45, 46, 47, 48, 86, 90, 91, 98, 99, 108, 109, 114
110
Index
transfer, 5, 8, 53, 89, 92, 107 transformation, 92, 98, 112 transition, xiii, 10, 51, 52, 80, 82, 85 transitions, 92 transport, 4 tungsten, 106
spin, 6 stability, 17 stages, 66, 96, 114 standards, 118 stoichiometry, 39, 57, 67 strain, 32, 100, 102, 107 strength, 33, 49 stress, 32, 102 structural changes, 107 structural characteristics, 12, 44, 107 structural defect, 107 structure formation, 4, 50, 85 structuring, 8 subjectivity, 85 substances, 7, 11, 27, 96, 98, 99, 104 substitution, 27, 38 superalloys, 49 superconductors, 2 superimposition, 77 supply, 21 surface area, 96, 99, 101, 103, 107 surface diffusion, 20, 22, 31 symmetry, 9, 13, 17, 21, 31, 32, 42, 89, 91 synchrotron, 57 synthesis, xiii, xiv, 1, 2, 3, 4, 6, 11, 15, 16, 24, 47, 48, 74, 85, 86, 90, 92, 93, 95, 97, 98, 113
U
uniform, 4, 20, 45, 53, 88 universal gas constant, 5, 105, 108
V
vacancies, 96, 104, 105, 110, 111, 112, 113 vacuum, 39 validity, 16, 89 variation, 12, 59, 64, 67 velocity, 3, 7, 9, 14, 22, 25, 51, 57, 72, 79, 95, 96, 97 vibration, 105 viscosity, 23, 71 voids, 41 volatilization, 1
W T
technology, 10 temperature dependence, 28, 34, 35, 62, 64, 67, 68, 72 temperature gradient, 4 tensile, 32 tensile stress, 33 thermal equilibrium, 105 thermal resistance, 53 thermodynamic, 60, 109, 114 thermodynamics, 13 thin film, 41, 42, 69, 71, 90 thin films, 69, 71, 90 time increment, 63 titanates, 2 wave propagation, 1, 45, 86, 88 welding, 49, 96, 99, 114 wires, 54
X
X-ray diffraction, 106 X-ray diffraction (XRD), 106 XRD, 106, 107
Y
yield, 15, 53, 98

Combustion - Synthesisof Advanced Materials

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Combustion - Synthesisof Advanced Materials

Uploaded by

Copyright:

Available Formats

CHEMISTRY RESEARCH AND APPLICATIONS SERIES

COMBUSTION SYNTHESIS OF ADVANCED MATERIALS

CHEMISTRY RESEARCH AND APPLICATIONS SERIES

CHEMISTRY RESEARCH AND APPLICATIONS SERIES

COMBUSTION SYNTHESIS OF ADVANCED MATERIALS

Nova Science Publishers, Inc.

Copyright 2010 by Nova Science Publishers, Inc.

Published by Nova Science Publishers, Inc. New York

ADVANCES AND CHALLENGES IN MODELING COMBUSTION SYNTHESIS

Advances and Challenges in Modeling Combustion Synthesis

Figure 1.1. Schematic of the thermal structure of a combustion wave.

Advances and Challenges in Modeling Combustion Synthesis

Advances and Challenges in Modeling Combustion Synthesis

1.2. BRIEF REVIEW OF DIFFUSION-BASED KINETIC MODELS OF CS

Advances and Challenges in Modeling Combustion Synthesis

MODELING DIFFUSION-CONTROLLED FORMATION OF TIC IN THE CONDITIONS OF CS

Modeling Diffusion-Controlled Formation of TiC

2.2. SCENARIO 1: GROWTH OF S TIC CASE ON THE TITANIUM PARTICLE SURFACE

Modeling Diffusion-Controlled Formation of TiC

2.4. DIFFUSION DATA FOR TIC

1.31103 347.3 4.36104 736.4

Modeling Diffusion-Controlled Formation of TiC

2.5. TEMPERATURE OF THE REACTION CELL IN THE SHS WAVE

2.6. ADIABATIC HEAT RELEASE IN THE REACTION CELL

c p (TiC)dT + m (t) c p (C)dT +

mTi(t) c p (Ti )dT + I[Tad Tm (Ti )]H m (Ti ) , 298

Modeling Diffusion-Controlled Formation of TiC

2.7. MODELING OF TIC LAYER GROWTH ON THE TITANIUM PARTICLE SURFACE

B. B. Khina 1/2(/2) exp(/2)2 erf(/2) = q, (2.13)

2.7.2. Results of Calculations for Scenario 1

Modeling Diffusion-Controlled Formation of TiC

2.8. RUPTURE OF THE PRIMARY TIC SHELL

Modeling Diffusion-Controlled Formation of TiC

Y Y [(1)urr + 2u], = (u + urr), (1 + )(1 2) (1 + )(1 2)

B. B. Khina Y(T) = Y0 bYT exp(T0/T), G(T) = G0 bGT exp(T0/T), (2.21)

2.9. GROWTH OF A TIC LAYER ON THE SURFACE OF A SOLID CARBON PARTICLE

Modeling Diffusion-Controlled Formation of TiC

2.9.2. Results of Calculations for Scenario 2

Modeling Diffusion-Controlled Formation of TiC

2.9.3. Displacement of the C/TiC Interface in the Emptying-Core Mechanism

The initial conditions are R0(t=0) = R1(t=0) = R00. (2.27)

i (t) = D i[T()]d , i C,Ti,

2.9.4. Product Porosity in the Emptying-Core Mechanism

Modeling Diffusion-Controlled Formation of TiC

2.10. ANALYSIS OF THE SHRINKING-CORE MECHANISM IN THE TI-C SYSTEM

Modeling Diffusion-Controlled Formation of TiC

, R0(t=0) = R00, h0 = R1(t)R0(t) = const, (2.32)

2.11. PHASE-FORMATION-MECHANISM MAP FOR NONISOTHERMAL INTERACTION IN THE TI-C SYSTEM

Modeling Diffusion-Controlled Formation of TiC

Modeling Diffusion-Controlled Formation of TiC

MODELING INTERACTION KINETICS IN THE CS OF NICKEL MONOALUMINIDE

Modeling Interaction Kinetics in the CS of Nickel Monoaluminide

3.2. PHYSICAL BACKGROUND OF THE MODEL

c p (pr, s)dT + I[Tad Tm (pr)]f m H m (pr) +

Modeling Interaction Kinetics in the CS of Nickel Monoaluminide

3.2.2. Phase Composition of the Reaction Zone

Modeling Interaction Kinetics in the CS of Nickel Monoaluminide

3.3. MODEL FORMULATION FOR NIAL FORMATION IN NONISOTHERMAL CONDITIONS

Modeling Interaction Kinetics in the CS of Nickel Monoaluminide

3.3.2. Quasi-Isothermal Submodel

and (ii) at interface 2/3 [NiAl(s)/Ni(s,m)]

3.3.3. Non-Isothermal Submodel