ISSN 1061-3862, International Journal of Self-Propagating High-Temperature Synthesis, 2007, Vol. 16, No. 3, pp. 141153.

Allerton Press, Inc., 2007.

Models of SHS: An Overview

A. S. Rogacheva and F. Barasb
a Institute

of Structural Macrokinetics and Materials Science, Russian Academy of Sciences, Chernogolovka, Moscow, 142432 Russia b Institute Carnot de Bourgogne, UMR 5209 CNRS, Universit de Bourgogne, France e-mail: rogachev@ism.ac.ru
Received July 20, 2007

AbstractThe theoretical models of SHS based on lamellar or cellular approximations of the heterogeneous reactive media are comparatively analyzed. It is shown that the ratio of the reaction time to the characteristic time of heat transfer between particles is a decisive parameter for the combustion wave propagation. When the time of reaction is shorter than the time of heat exchange, the combustion occurs in a discrete mode; in the opposite case, a quasi-homogeneous combustion mode occurs. Development of the discrete cellular model does not discard the quasi-homogeneous approach but markedly extends the scope of combustion theory. This extension enables explanation of many old and new experimental results that could not be rationalized within the framework of homogeneous theory. Key words: SHS, combustion synthesis, mathematical models, quasi-homogeneous and discrete modes, ame propagation, stability PACS numbers: 81.20.n, 81.20.Ka DOI: 10.3103/S1061386207030077

1. INTRODUCTION As is known, Self-propagating High temperature Synthesis (SHS) is based on propagation of reaction wave through heterogeneous mixture of solid reactants [1, 2]. Microstructure of heterogeneous reaction mixture inuences combustion process in two ways. First, heterogeneous nature of the medium affects the kinetics of exothermal chemical reaction and, therefore, the rate of heat release. Second, variations in thermal conductivity intrinsic to heterogeneous mixtures may affect the heat ow from the hot combustion products toward fresh cold reactants, which is an important stage of the wave propagation mechanism. The inuence of heterogeneity on the reaction kinetics was thoroughly studied by combustion theory. It was shown that retardation of the reaction by solid products must result in deformation of temperature prole of the combustion wave in such a way, that thickness of the reaction zone increases and leading temperature of the process becomes markedly lower than a maximum combustion temperature. In contrast to the Zeldovich model for combustion of premixed gases, the so-called narrow-zone combustion mode, the heterogeneous process with strong retardation of the reaction was termed the broad-zone combustion mode [3, 4]. The lamellar model used to describe a heterogeneous microstructure of the medium [57] assumes that the combustion front propagates parallel to the layers of reactants, as shown in Fig. 1. Further, it was supposed that the reaction rate is controlled by solid-state diffusion, while the thermal diffusivity of reaction mixture has a value typical of the metal bulk.

Under these conditions, thermal relaxation between layers occurs much faster than the equating of concentrations by means of diffusion, and temperature variations along the combustion front (i.e. normal to layers) are negligible. Therefore, fast thermal relaxation allows approximation of the lamellar heterogeneous medium by a homogeneous medium with averaged thermal properties (thermal conductivity, heat capacity, etc.). This approach creates a base for a quasi-homogeneous model of the heterogeneous combustion. Thus, the quasi-homogeneous model takes into account the effect

Reactant A

Reactant B Product ABx Direction of the combustion front propagation

Fig. 1. Layered quasi-homogeneous model.

141

142
Reactive layer Inert layer Inert layer Direction of the combustion front propagation

ROGACHEV, BARAS

in the most pronounced form. More sophisticated twodimensional models involving different types of microstructure irregularities such as random pores etc. will not be considered here. The relevant information about 2D models can be found e.g. in review [2]. 2. MULTILAYER MODEL 2.1. Formulation of the Problem In order to take into account thermal heterogeneity of the reactive medium, a multi-layer microstructure shown in Fig. 2 was proposed as a model of real heterogeneous structure. Distinctions of this structure from that in Fig.1 are not only in different orientation of the layers but also in different chemical nature. If in the model with layers parallel to direction of combustion front propagation (Fig. 1) the layers composed of pure reactants (layers of fuel and layers of oxidizer), the multi-layer structure with perpendicular orientations of the layers regarding the combustion propagation direction (Fig. 2) is composed of alternating combustible and inert layer. Each combustible layer contains fuel and oxidizer, and combustion can occur within the layer without necessity of diffusional mass transfer between layers (as it takes place in the model of Fig.1). The thermophysical parameters of inert layers, such as heat capacity, density, thermal conductivity, differ from those of reactant layer. The inert layers may be solid or gaseous, and no combustion can happen inside these layers. Mathematical analysis of the model [28] has shown that two limiting regimes of combustion can be realized in such a system: (1) quasi-homogeneous for thin layers and (2) relay-race for thick layers. Development of this approach led to non-stationary mathematical model of thermal ame propagation in multilayer heterogeneous system [29] to be described below. The layer microstructure presented in Fig. 2 is used in this model under the assumption that heat transfer proceed via thermal conductivity. Equation of heat conduction in homogeneous medium is applied for combustible and inert layer as follows. For a combustible layer, (i 1)(d + ) < < id + (i 1), where i is the layer number, d the thickness of combustible layer, the thickness of inert layer, and the spatial coordinate, we obtain: 1 - ----- = ------- + -- ------, 2
2

Fig. 2. Layered heterogeneous model.

of microstructure on the reaction rate but ignores microstructure as regards to heat transfer in the medium. It is interesting to note that recently this approach has provided good results in modeling of combustion waves in multilayer nano-foils (cf. [8, 9]). However, applicability of the quasi-homogeneous model to porous powder mixtures does not seem indisputable. Recent experimental results shake foundations of the quasi-homogeneous model, at least for combustion of powder mixtures. It has been shown that solid-state diffusion is not a necessary stage of the heterogeneous combustion reactions. In many cases, mass transfer occurs through ow of melts, diffusion of liquid or grain boundary, and other processes that are much faster than diffusion in bulk solids (cf. [1018]). On the other hand, the thermal conductivity of reactive powder mixtures was found to be much lower than that of bulk metals, thus indicating that contacts between solid particles work as barriers for heat uxes in the mixture [19, 20]. As a result, duration of the reaction in some local volume of the reactive mixture may become comparable or even shorter than the time of thermal relaxation in this volume. If it happens, it can result in signicant thermal heterogeneity of the combustion front, because the heat released during heterogeneous reaction has not enough time for spreading out of the local reaction region and cause over-heating of this region relative to surrounding medium. Experimental observations conrm the existence of thermal heterogeneity of the combustion waves in powder mixtures on the microscopic scale [15, 2127]. This phenomenon was termed the micro-heterogeneous, or scintillating, combustion mode [24]. Several theoretical models were suggested to describe the micro-heterogeneous combustion waves. In this overview, the basic principles of two modelsa layered model and cellular oneare discussed from the viewpoint of adequate description of experimental data. Our analysis is restricted by one-dimensional models in order to reveal the physical background of the models

Product

(1)

------ = ( 1 ) exp --------------- = F ( , ), 1 + 1(1 ) exp ( /(1 + ) ), F(, ) = 0, 1.

Vol. 16

(2)

0 1,

(3)

INTERNATIONAL JOURNAL OF SELF-PROPAGATING HIGH-TEMPERATURE SYNTHESIS

No. 3

2007

MODELS OF SHS: AN OVERVIEW

143

For an inert layer, id + (i 1) < < i(d + ), we have: c ----- = ------- . 2

2

At the boundaries between layers, the temperatures and heat uxes are assumed to be equal:
R

(4)

= I,

----

= . ---- I

(7)

Dimensionless variables and parameters are dened in a common way adopted in the theory of homogeneous combustion: dimensionless temperature = t ( T T )E * ------------------------ ; dimensionless time = ---- , t = 2 t RT * * 2 RT c R exp ( E/RT ) * *---------- ---- ------------------------------ ; dimensionless spatial coordinate E Q k R x = ----- , x = -----------t ; is the degree of conversion cR R * x * (dimensionless ratio of the mass of products to the net 2 RT c R * mass of the system); parameter = ---------- ---- characterE Q izes the ratio of the reaction to preheated zone width; RT * parameter = ---------- characterizes temperature sensitivE cI I ity of the reaction rate; parameters c = ----------- and = cR R I ----- are the ratios of heat capacities of unit volume (c) R and thermal conductivity () of the inert (subscript I) and reactive (subscript R) layers, respectively. Dimension variables are the following: T is temperature, T is the characteristic temperature of the process (see below), E the activation energy of exothermal chemical reaction, R universal gas constant; cR, cI the specic heat capacity of the combustible and inert layers (respectively), R, I the density of the combustible and inert layers, Q the heat of reaction, k the pre-exponential factor in the Arrhenius function. Adiabatic combustion temperature Tad is usually used as the characteristic temperature of the process T. However, a heterogeneous structure of this model gives rise to two adiabatic temperatures. The temperature of adiabatic combustion inside the reactive (combustible) layer Q (1) T ad = T 0 + ---cR (5)

The model has been further developed to introduce heat losses and radiative heat transfer [30, 31], however, these questions are beyond the scope of our analysis. Thus, we can conclude that the above model describes a continuum with periodically changing properties. It involves two homogeneous mathematical models, (1)(3) and (4); each model is based on heat conduction equation in active or inert layers. Thermal contact between layers is very good [conditions (7)]. It should be also noted that the model is one-dimensional, so that the layers (Fig. 2) can be substituted with straight segments, parallel to the direction of combustion. Initial conditions are uniform for all layers: = 0, > 0: = 0 , = 0 ( T = T 0 ). (8) Combustion is initiated locally by a hot wall, i.e. temperature at the boundary of the rst layer is assumed to have some constant value (ignition temperature): > 0, = 0, = ign . (9) Computer modeling of initiation and propagation of the combustion is performed until the combustion wave achieves a stationary regime. 2.2. Results of Computing Calculations according to model (1)(4), (7), (8) provided temperature proles and instantaneous propagation velocity as function of time [29, 30]. Figure 3 illustrates the results obtained for thin layers. Preheated zone covers many reactive and inert layers, which allows introducing some effective thermal properties and averaged kinetics, i.e., considering the medium as roughly homogeneous. This regime can be identied as quasi-homogeneous combustion. Note that the instantaneous velocity undergoes strong periodical oscillations even in the quasi-homogeneous combustion mode (Fig. 3b). With increasing of layer thickness, the combustion undergoes transition to the relay-race regime, as illustrated in Fig. 4. This regime can be described as alternating periods of fast movement of the combustion front inside the reactive layer and slow spreading of heat when the front achieves the boundary of inert layer. When a sufcient amount of heat is transferred from one reactive layer to another through the inert intermediate layer, the next reactive layer is ignited, and the periodic process starts again. The dependence of averaged combustion velocity on the layer thickness d (at xed d/ ratio) shown in Fig. 5 was reported in [29]. Evidently, this dependence does not agree with predictions of the quasi-homogeneous theory, where combusVol. 16 No. 3 2007

must be used when the layers are thick (T0 is initial temperature of reactants), while the averaged adiabatic temperature Q R d (2) T ad = T 0 + --------------------------------cR R d + cI I (6)

(taking into account thermal dilution of the system by inert layers) should be used in case of thin layers.

INTERNATIONAL JOURNAL OF SELF-PROPAGATING HIGH-TEMPERATURE SYNTHESIS

144 0 1 2 3 4 5 0 0.2 0.1 0 400 20 (a)

*

ROGACHEV, BARAS 2 0 2 40 (b) 6 0 60 80 4 1 200 2 400 3 600 4 800 1000 1200

450

500

550

600

650

700

Fig. 4. Temperature proles predicted by heterogeneous layered model: relay-race regime [29].

Fig. 3. Temperature proles (a) and local combustion velocity (b) predicted by heterogeneous layered model: quasihomogeneous regime [29].

tion velocity must decrease monotonically with increasing d, due to the retardation of chemical reaction. However, as mentioned above, the chemical nature of layers is different in the quasi-homogeneous (Fig. 1) and layer model (Fig. 2); therefore, there is no real controversy. According to the layer model, the rate of reaction inside the combustible layer does not depend on the layer thickness. Three segments can be identied in the plot presented in Fig. 5. When the layers are thin, combustion velocity only weakly depends on d, because temperature exchange between the reactive and inert layer is so fast that the system is thermally homogeneous. Maximum combustion temperature is (2) equal to T ad [see Eq. (6)]. With increasing layer thickness, thermal inuence of inert layers on the combustible ones weakens, and a maximum temperature inside (1) the combustible layer rises up to a value of T ad [Eq. (5)]. With increasing combustion temperature, the local and average combustion velocity grow. Further growth of layer thickness leads to a drop in the average velocity, because combustion velocity inside the reactive layers does not increase, while the time of heat transfer from one reactive layer to another increases. 2.3. Relation to Experimental Results Real objects which comply with the multilayer model are the stacks of thin reactive disks separated by gaps as reported in [32]. Scheme of the experiments is presented in Fig. 6. Disks with diameters D = 6, 10, 12, and 16 mm and thickness h 80 m were pressed from different reactive mixtures wetted with a 1% ethanol

solution of poly(vinyl butyral). The pressed disks were dried in an air at 150C for at least 1 h, and, in some experiments, annealed in vacuum (101 Pa) at a temperature not higher than 600C for 24 h. Then the disks were placed equidistantly in a frame consisting of a Ushaped boron nitride holder and ledge-forming tungsten wires 100 m in diameter. Thus, each disk lied on two wires and did not contact with the others. After ignition of the rst disk with an electrically heated coil, the combustion propagated sequentially to the other disks owing to heat transfer. Temperature of the disks during combustion was measured with tungsten-rhenium thermocouples (thickness smaller than 25 m). Up to 20 discs were used in each experiment. First series of experiments were done with the disks pressed of Nb powder, and the combustion was realized in the nitrogen atmosphere (reaction Nb + N2) [32]. Quasihomogeneous and relay-race regimes were detected in these experiments. In the quasi-homogeneous regime,
V 0.09 0.08 0.07 0.06 0.05 0.04 e2 e1 e1 lnd

Fig. 5. Average burning velocity V as a function of reactive layer thickness d [29]. Vol. 16 No. 3 2007

INTERNATIONAL JOURNAL OF SELF-PROPAGATING HIGH-TEMPERATURE SYNTHESIS

MODELS OF SHS: AN OVERVIEW 1 2 3 4 2.0 5 1.5 1.0 0.5 h = 0.38 mm = 0.20 mm TTi m T, 103 C 2.5 Tad

145

Fig. 6. Experimental setup for testing the layer model: (1) initiating coil, (2) reacting disk, (3) wire ledges, (4) thermocouples, (5) video camera [33].

0.1

0.2

0.3

0.4 t, s

reaction zone involves several Nb disks reacting simultaneously, while in the relay-race regime there was a time interval between burning out of the disk and ignition of the next one. However, when gasless mixture 3Ti + 3Si was used for preparation of the disks, no quasi-homogeneous regime was found, even when the gap between disks was decreased down to 0.2 mm [33]. Moreover, when the disks were stacked without gaps, and compressing stress up to 90 kPa was applied to improve contacts between layers, combustion still occurred in the relay-race regime [34]! Therefore, these experiments lead to a conclusion that fast-burning powder mixture exhibits the relay-race (discrete) behavior even without separating into layers. The results of thermocouple measurements shown in Fig. 7 are in a good qualitative agreement with the results of computer modeling for the relay-race regime. The existence of thermal heterogeneity (shortly mentioned in the Introduction), conrmed by high speed micro-video recording [15, 2127] is in qualitative agreement with the results of layer model, however, this kind of heterogeneity can be caused by many different reasons, without relation to the layer microstructure. Quantitative comparison of the computing results with experiments on combustion of powder mixtures has not been done up to now, because the model does not provide quantitative dependences that can be directly compared with the available experimental data. The dependence of average combustion velocity on the layer thickness (Fig. 6) can be compared with experimental results on combustion in the TiC [35] and Ti Si [36] systems, where the combustion velocity was found to increase with increasing size of Ti particles. At the same time, explanation of such regularity can be done without the hypothesis of layer microstructure based on either the evolution of impurity gases [35] or the shape of metal particles [36].

Fig. 7. Thermorgams of relay-race combustion in the model layered system of Ti/Si disks [33].

On one hand, the layered model has no contradiction with the available experimental results, but on the other, the grounds for comparison and justication of the model are very limited. Another problem of the layered model is that the equations of homogeneous heat conduction (1) and (4) are assumed to be applicable for any region of the medium. A situation, when this condition is not required, will be discussed below. 3. CELLULAR MODEL 3.1. Formulation of the Problem As is known from experiment, the heat and mass transfer at the scale of reactant particles have complicated mechanisms that cannot be described by simple diffusion-type equations. For example, in the mixtures Ni + Al, Ti + Si, etc., where the combustion temperature exceeds the melting point of reactants, the reaction occurs through coalescence of liquid drops [16, 17]. Processes of coalescence of drops, mixing of melts (including turbulent ows, diffusion, etc.), reaction, and heat evolution take place simultaneously. In the systems where only one reactant melts (such as Ti + C) fast spreading of the molten reactant over the solid one, draw of solid particles into the melt, fragmentation of the solid reactant plays a signicant role in the mass and heat transfer at the microscopic scale [1012, 37]. To date, there are no comprehensive mathematical models describing such a complex behavior, while the attempt to describe them with the equation of homogeneous heat transfer can lead to wrong results. In our opinion, a more fruitful approach is to enclose the interacting particles into some small volume, the socalled reaction cell, and try to describe separately processes inside the cells and interaction between cells. This approach underlies the cellular model of heterogeVol. 16 No. 3 2007

INTERNATIONAL JOURNAL OF SELF-PROPAGATING HIGH-TEMPERATURE SYNTHESIS

146

ROGACHEV, BARAS

i3
Ti 3

i2
Ti 2

i1

i+1

i+2

i+3

Ti 1

The state of the ith reaction cell at any time is uniquely determined by two parameters: the dimensionless temperature i of the ith cell and the degree of conversion i . Since the temperature inside a cell is uniform, heat balance for every cell can be described by the following system of dimensionless equations: i ------- = i 1 2 i + i + 1 + -------i , i ------- = ( i, i ), where Ti T0 i = ----------------TC T0 the dimensionless temperature of the i-th cell, S = ------- t the dimensionless time, cM (14) (11) (12)

Ti

Ti + 1

Ti + 2

Ti + 3

Fig. 8. One-dimensional cellular model.

neous combustion. Let us present a powdered mixture as a medium consisting of identical reaction cells, each containing a fuel and an oxidant, mixed in a stoichiometric ratio. Let us also assume that relaxation of temperature inside each cell occurs much faster than heat transfer between the cells, and the temperature inside each cell is uniform. This very important assumption is based on the fact that the thermal conductivity across the interface between metal particles in the initial mixture is much lower than the thermal conductivity of each metal particle. This conclusion, drawn from studies on the microstructure, thermal conductivity, and electrical conductivity of actual mixtures (cf., [2, 12, 19, 20, 26, 27]), is explained by the small contact area of metal particles and by the presence of oxide layers and other contaminations at their surface. For most gasless compositions, the size of metal particles (Ti, Zr, etc.) is much greater than the size of nonmetal particles (C, B, Si, etc.). Next, metal particles form a continuous skeleton, while the nely dispersed nonmetal reagent occupies pores between them. In this case, one metal particle with attached nonmetal particles can be treated as a reaction cell. In some systems, such as Ni + Al or Ti + Ni, the particle size of reagents is close. In this case, two neighboring particles of dissimilar reagents, located in one particle layer with respect to the position of the combustion front, can be treated as a reaction cell. The specic microstructure of reaction cell depends on the phase composition of the mixture, particle size/shape, porosity and other factors. It should be emphasized, however, that in the present work we consider such a combustion model in which heat is transferred exclusively by thermal conductivity of condensed phase (heat transfer through the gas phase in the voids and radiative heat transfer are ignored). Moreover, since we deal here only with one-dimensional models, let us focus on the linear chain of reaction cells shown in Fig. 8 [38]. Heat exchange between neighboring cells occurs by the Newton law: q = S ( T i T i + 1 ), (10) where q is the heat ux from cell i to cell i + 1, the coefcient of heat exchange between cells, S the surface area of contacts between cells, Ti , Ti + 1 the current temperature of cells i and i + 1, respectively.

(13)

m i = 1 -----i the degree of conversion for the ith cell, (15) m0 (i , i ) the dimensionless kinetic function. (16) Dimensional variables and parameters are as follows: Ti is the current temperature of the cell, T0 the iniQ tial temperature, TC = T0 + ------- the adiabatic combuscM tion temperature [compare with (5)], c the specic heat capacity, M the mass of cell, t time, mi the current mass of fuel inside the ith cell, m0 the initial mass of fuel inside one cell. The kinetic function can be approximated by different expressions. Two cases were investigated up to now: step-like function [39] and the Arrhenius kinetic function Ar for the zero order reaction [38]. In the latter case, this function can be written as follows: 1 p exp ----------------------------- , Ar 0 + Ar C i ( i, i ) = 0, = 1. i 0 i < 1;

(17)

Expression (17) contains the following three dimensionless parameters: cM p = ------- k0, where k0 is a pre-exponential factor in the S Arrhenius function; RT 0 Ar 0 = ---------; E R(T C T 0) Ar C = -------------------------- . E
Vol. 16 No. 3 2007

INTERNATIONAL JOURNAL OF SELF-PROPAGATING HIGH-TEMPERATURE SYNTHESIS

MODELS OF SHS: AN OVERVIEW

147 (a) q

It is worth noting that all the parameters have a clear physical meanings directly related to experimental conditions. The parameter p shows the ratio of the characteristic heat-transfer and reaction times. Obviously, for small p, quasi-homogeneous combustion modes (in which the heating and reaction zones contain many particles) should be expected to occur; for large p, the quasi-homogeneous approach is inapplicable. In experiments, parameter p can be varied by changing the specimen density or particle size. The parameter Ar0 is the Arrhenius number for the initial temperature of the specimen; its value can be changed upon preheating the green mixture. The parameter ArC is close in magnitude to the Arrhenius number for the combustion temperature (if TC T0), and its value can be varied by diluting the mixture with an inert substance or by changing the fuel-to-oxidant ratio. Main features of the combustion processes have been studied within the frames of the cellular model, depending of the values of three governing parameters. 3.2. Results of Computing As a rule, value of Ar0 is assumed to be small (0.01 or less), which indicates that the reaction rate can be neglected at initial temperature (increase in Ar0 leads to thermal explosion mode of the process). Therefore, behavior of the combustion wave is determined by two parameters, ArC and p. Computer simulation has shown that small values of p leads to a quasi-homogeneous thermal structure of the combustion wave while large p, to the relay-race mode, as illustrated by Fig. 9 [38]. This result is quite similar to the conclusions inferred from the layer model (Section 2), and, therefore, can not serve as a criterion for applicability of different models. In order to construct a map of different combustion regimes, computer simulations of the combustion process were carried out for wide ranges of parameters ArC and p. Ratio of the time of reaction and time of heat transfer between cells is responsible for realization of quasi-homogeneous or relay-race mechanisms of the ame spreading. It is difcult, however, to distinguish a pure time of heat exchange, because process of heat exchange takes place before, during, and after reaction in the combustion wave. For this reason, we calculated the ratio of the reaction time at adiabatic combustion temperature, tR to the time of shifting tS, e.g. the time interval when combustion front shifts at one reaction cell. The value of tS includes both preheating and reaction time. Quasi-homogeneous mode of combustion occurs when (it means that many cells are involved in the reaction zone), while the micro-heterogeneous, or relay-race mode can be expected when tR/tS 1. Calculated values of tR/tS are shown in Fig. 10. A decrease in ArC (at xed p) inevitably results in a loss of stability of the combustion wave, and oscillations appear. The dashed curve shows the boundary of stability: steady-state propagation is possible only on the

, , q = 104(d/d) 1.0 1 0.8 0.6 0.4 0.2 0 0 5 10 15 , , q = 5(d/d) 1.2 0.8 0.4 0 0 1 2 1

20 (b)

25

30

35

5 /scale

Fig. 9. Thermal structure of the combustion wave in terms of the cellular model: (a) quasi-homogeneous mode and (b) discrete mode [38].

right-hand side of this curve. Approximate boundary between the relay race and quasi-homogeneous regime can be accepted as the condition tR/tS 1; this boundary is marked by horizontal dotted line. Finally, in the combustion processes, the Arrhenius number ArC possesses values from 0.05 to 0.2 (as it is known from experiments), these limits are denoted by two dash-dot vertical lines. Thus, region abcd in Fig. 10 corresponds to the quasi-homogeneous while adefg, to the relay-race steady combustion. Oscillating regimes can arise also in the quasi-homogeneous or relay-race modes. Figure 11 demonstrates a boundary between the regions of steady-state regime (when the combustion propagation velocity is constant) and oscillating regime in the coordinates ArCp. It is worth noting that critical value of ArC, when the combustion front loses its stability, decreases with increasing p. Therefore, steady-state combustion in the relay-race mode can be realized at lower Arrhenius numbers than the quasi-homogeneous steady combustion. This trend can be seen also in Fig. 10. Similar results have also been reported in [38]. Thus, computer calculations using cellular models have outlined regions of common combustion modes, which gave a basis for comparison of these results with a wide range of the available experimental data.
Vol. 16 No. 3 2007

INTERNATIONAL JOURNAL OF SELF-PROPAGATING HIGH-TEMPERATURE SYNTHESIS

148 tR/tS 101

ROGACHEV, BARAS b c

Quasi-homogeneous pulsating combustion

Quasi-homogeneous steady-state combustion

100

a
Relay-race pulsating combustion

d
p = 109 p = 102 p = 108 p = 107 p = 106 p = 105 p = 104 p = 103 p = 101 p = 100 p = 101

101

g
Relay-race steady-state combustion

f 101

e 100 Arc

Fig. 10. Map of combustion regimes in terms of the cellular model.

3.3. Relation to Experimental Results The existence of hot regions in the combustion front with a local temperature higher than average Tad was conrmed by numerous experimental data [15, 2127]. However, these over-adiabatic regions can appear due to different reasons, for example, as a result of thermal instability (oscillations) of the combustion wave. A discrete model requires that the hot spots must directly correlate with metal particle size. This correlation was experimentally proved in [26, 27]. Statistical treatment of microscopic high-speed video recordings has shown that the ratio of the time of reaction in one reaction cell (e.g., lifetime of hot spot) to the time of combustion front movement over the distance of one reaction cell is
Arc 0.16 0.14 0.12 0.10 0.08 0.06 0.04 0.02 0 0 2 4 6 log(p) 8 0.20 0.18 0.16 0.14 0.12 0.10 0.08 0.06 0.04 0.02 0 10

Quasi-homogeneous mode

Relay-race mode

Steady-state combustion

0.050.1. Comparison of these values with Fig. 10 shows that they are within the region of discrete combustion mode. It is interesting to note that recent experiments with mechanically activated (pre-structured) SHS mixtures open up new possibilities for investigating micro-heterogeneous mechanism of gasless combustion [40]. Mechanical treatment of two or more ductile reactants (i.e., Ni and Al) in a planetary mill results in formation of poreless composite particles, which contain all ne dispersed reactants in a stoichiometric ratio. Thus, each composite particle plays a role of reaction cell, and correlation between size of these particles and front movements have been demonstrated [40]. To date, more than 20 measurements of the effective activation energy E, together with combustion temperature TC, have been reported for different powder mixtures. Using these data, it is easy to calculate a value of R(T C T 0) ArC = -------------------------- or, more traditionally, the values of E RT C RT C = -------------------------- and = --------- . These values allow comE(T C T 0) E parison of a large group of experimental results with predictions of the quasi-homogeneous and discrete models. There are two criteria for the existence of steady-state combustion suggested in the framework of quasi-homogeneous model. Analytically derived criterion [41, 42]: 1 -- < 2 + 5 means that steady-state combustion is possible if > 0.236. The criterion of Shkadinskii, Khaikin, and Merzhanov calculated by computer simulation [43], 9.1 2.5 > 1,
Vol. 16 No. 3 2007
2

Oscillating combustion

Fig. 11. The boundary line between steady-state and oscillating combustion, as calculated in terms of the cellular model.

INTERNATIONAL JOURNAL OF SELF-PROPAGATING HIGH-TEMPERATURE SYNTHESIS

MODELS OF SHS: AN OVERVIEW Effective kinetic parameters of gasless combustion as derived from the dependence of burning velocity on temperature Reaction Ti + C TiC Ti + 0.5C TiC0.5 Ti + 0.5C TiC0.5 Ti + C TiC Ti + C TiC Ti + C + Ni TiC + Ni (25 wt %) Ti + 2B TiB2 Ti + 2B TiB2 Ti + B TiB Zr + 2B ZrB2 Zr + 1.5B + xTiB2 ZrB2 + Zr + TiB2 Hf + 2B Hf + 2B 5Ti + 3Si Ni + Al Ni + Al Ni + Al Ni + Al Co + Al 3Cu + Al 2Cu + Al Zr + 3Al Zr + 2Al Ti + Co HfB2 HfB2 Ti5Si3 NiAl NiAl NiAl NiAl CoAl Cu3Al Cu2Al CuAl3 CuAl3 TiCo Temperature range, K 27503200 23002600 26902970 26002711 27112940 20002500 24403000 19003170 21002400 21253125 26002900 27003570 24003000 18002400 13001600 16001850 16001850 14201600 14701750 11501250 11501250 12501800 12701800 14001700 Eeff, kJ/mol 251 318 42 117 33 364 25 124 31 133 50 318 302 230 310 134 13 (149 13) 398 390 126 140 76 75 134 129 180 146 90 70 188 Green porosity, % 3862 45 45 50 50 50 ~70 40 40 ~70 4060 ~70 40 35 40 40 40 40 40 ~40 ~40 ~40 ~40 ~40 RT C y = -------------------------E(T C T 0) 0.1020.117 0.0730.077 0.2350.264 0.0670.070 0.2040.220 0.1470.178 0.0730.087 0.0620.096 0.0890.099 0.0660.093 0.1820.201 (0.1640.180) 0.0630.081 0.0580.071 0.1430.181 0.1000.117 0.2150.242 0.2180.245 0.1120.122 0.1190.136 0.0720.076 0.0890.094 0.1520.200 0.1970.257 0.0790.091
2

149

Refs. [44] [45] [45] [46] [46] [46] [47] [48] [48] [47] [49] [47] [50] [51] [52, 54] [52, 54] [52, 54] [52, 54] [52, 54] [52, 54] [52, 54] [54] [54] [53]

can be reduced to > 0.145 (taking into account that TC T0 ----------------- 0.850.9 for the most typical combustion TC conditions). The values of derived from the available experimental data are listed in table. Evidently, less than a half of the systems (marked grey) t the requirements of stability suggested by quasi-homogeneous theory. Steady-state combustion of all other systems, including such traditional mixtures as Ti + C, Ti + B, Ni + Al, etc., cannot exist in the framework of quasi-homogeneous approach. On the other hand, comparison of the data presented in table with predictions of discrete model (Figs. 10 and 11) clearly show that all these data t the criteria of the cellular model, because this model allows steady combustion at ArC down to 0.05 ( 0.06). Another interesting question is a behavior of the combustion front in the vicinity of steady/unsteady boundary. Recent experimental investigation for the Ti + xSi mixtures (x = 0.32.0) showed that the struc-

ture of the combustion wave near the limit with excess metal is quite different as compared to the limit for lack of the metal. The difference is illustrated in Fig. 12. When instability occurs due to excess of metal, continuous network of metal (Ti) particles provides effective heat transfer. Thus, the reaction started in the preheated layer can spread quickly over the entire cross section of the sample, as shown in Fig. 12a. This type of instability is close to the idealized one-dimensional oscillations predicted by quasi-homogeneous theory. At the opposite limit, when steady-state combustion drops down because of excess of silicon, instability proves itself in the form of hot spots, as shown in Fig. 12b. This behavior relates probably to disintegration of metal particles network (heat ow penetration limit). Difference of the steady combustion limits shows an important role of reaction mixture microstructure on the stability of combustion wave. This phenomenon should be studied more thoroughly. Thus, the cellular model not only correlates with microscopic observations of combustion wave, but also
Vol. 16 No. 3 2007

INTERNATIONAL JOURNAL OF SELF-PROPAGATING HIGH-TEMPERATURE SYNTHESIS

150

ROGACHEV, BARAS

(a)

(b)

Fig. 12. Combustion of Ti + xSi samples in the vicinity of the steady/unsteady boundary: (a) Ti + 0.32Si and (b) Ti + 1.4 Si. Bright pellets in the left side of the pictures are igniting pellets (Ti + 0.6Si).

demonstrates good agreement with the most experimental data concerning global characteristics of the process. 4. DISCUSSION AND CONCLUSIONS Detailed comparison of three considered above general approaches to describing a combustion synthesis process reveals difference in basic assumptions of the models, but it also reveals similarities. Thus, quasihomogeneous model assumes that differential heat transfer equation T T c ------ = -------- + QF ( T , ), 2 t x
2

(18)

[where F(T, ) = ------ and Q has a dimension of J/m3] t can be applied for any point of the reaction medium at any arbitrary moment of time. If we adopt the dimensionless temperature dened by (13), take into account that Q/c = TC T0, and quantize Eq. (18) for numerical computation, we obtain 1 - -------i = ----- -------- ( i 1 2 i + i + 1 ) + -------i . 2 t c x t (19)

Note that x is the spatial interval used in the quantized heat transfer equation, e.g., spatial step of the cal2 cx culation grid, and -------------- has a dimension of time and characterizes the time of thermal relaxation within interval x. Thus, if we introduce the dimensionless time as = ------------------- t, 2 c ( x ) (20)

Equation (19) will take exactly the form of Eq. (11). This fact often causes some certain misunderstanding, when the cellular model is treated as a coarse version of the homogeneous model. Basic difference between

these two approaches is hidden in the physical meaning of variables and parameters. In the homogeneous description, the thermal diffusivity /c is an intrinsic parameter while the spatial step x depends on the quantization, and the homogeneous limit is supposed to be recovered for x 0. In the cellular model, the parameter S/cM is xed by the heat transfer characteristics at the contact between cells and is directly related to the microstructure of the material. As we have seen in the previous section, the dimensional parameter p given by the ratio between this heat transfer parameter and the pre-exponential factor plays a crucial role in the development of the relay race mechanism and has no analogous in the homogeneous description. Quasihomogeneous model assumes a smooth temperature prole, as shown in Fig. 13a. We take quantized values of temperature in points A, B, C, etc. which correspond to the points of calculation grid. Distance between points x can be as small as necessary, decreasing of this distance leads to better approximation of the temperature prole by discrete sequence of numbers, as it is evident from Fig. 12a. Cellular model is based on the assumptions that the temperature prole is step-like, due to large difference between high thermal conductivity inside the cell and low heat conduction of contacts between cells. Calculating grid is strictly linked to dimension of the cell, L, and temperature values are taken in points A, B, C, D, as shown in Fig. 13b. Note that, in the cellular model, term 2i of equation (11) is a sum of two equal temperatures taken in points B and C. If we try to decrease a spatial step of the calculation grid (small triangles in Fig. 13), we have to change completely the system of equations, and to write different equations for regions inside cells and for contacts. In some special cases, this procedure can lead to the layer model (1)(4), however, as it was discussed above, processes inside cells and contacts can not be described by simple heat transfer equations, which is one of the reasons to create the cellular model. Layer model can be converted in the cellular one if we assume that thickness of inert layer is much smaller than reacVol. 16 No. 3 2007

INTERNATIONAL JOURNAL OF SELF-PROPAGATING HIGH-TEMPERATURE SYNTHESIS

MODELS OF SHS: AN OVERVIEW (a)

151

i 1

i B x (b) i 1 A i B L i C i + 1 D

i + 1 C

balance between heat transfer and heat evolution, because decreasing of heat exchange time (between cells) can be compensated by decreasing of reaction time (inside cell). Only in case when the rate of heat evolution is small compared to the rate of heat exchange, the cellular model gives the same results as the quasi-homogeneous one. Thus, the cellular model does not discard the quasi-homogeneous approach, it rather extends the applicability limits of the model. This extension enables explanation of many old and new experimental results that could not be rationalized within the framework of homogeneous theory. ACKNOWLEDGMENTS This work was supported by the Russian Foundation for Basic Research (project no. 05-03-32407) and by the Joint FrenchRussian Program of Scientic Cooperation (grants PICS 3462RFBR 06-03-22000). REFERENCES
1. Merzhanov, A.G. and Borovinskaya, I.P., Self-Propagating High-Temperature Synthesis of Inorganic Compounds, Dokl. Akad. Nauk SSSR, 1972, vol. 204, no. 2, pp. 366369. 2. Varma, A., Rogachev, A.S., Mukasyan, A.S., and Hwang, S., Combustion Synthesis of Advanced Materials: Principles and Applications, Adv. Chem. Eng., 1998, vol. 24, pp. 79226. 3. Merzhanov, A.G., The Theory of Stable Homogeneous Combustion of Condensed Substances, Combust. Flame, 1969, vol. 13, pp. 143156. 4. Aldushin, A.P., Merzhanov, A.G., and Khaikin, B.I., Some Peculiarities of Condensed System Combustion with Refractory Reaction Products, Dokl. Akad. Nauk SSSR, 1972, vol. 204, no. 5, pp. 11391142. 5. Aldushin, A.P. and Khaikin, B.I., On Combustion Theory of Mixed Systems Forming Condensed Reaction Products, Fiz. Goreniya Vzryva, 1974, vol. 10, no. 3, pp. 313323. 6. Khaikin, B.I., On Combustion Processes Theory in Heterogeneous Condensed Media, in Protsessy Goreniya v Khimicheskoi Tekhnologii i Metallurgii (Combustion Processes in Chemical Technology and Metallurgy), Chernogolovka, 1975, pp. 227244. 7. Hard, A.P. and Phung, P.V., Propagation of Gasless Reactions in Solids. I. Analytical Study of Exothermic Intermetallic Reaction Rates, Combust. Flame, 1973, vol. 21, no. 1, pp. 7789. 8. Barbee, T.W. and Weihs, T., Ignitable Heterogeneous Stratied Structure for the Propagating of an Internal Exothermic Chemical Reaction Along an Expanding Wavefront and Method of Making Same, US Patent 5 538 795, 1996. 9. Mann, A.B., Gavens, A.J., Reiss, M.E., van Heerden, D., Bao, G., and Weihs, T.P., Modeling and Characterizing the Propagation Velocity of Exothermic Reactions in Multilayer Foils, J. Appl. Phys., 1997, vol. 82, no. 3, pp. 11781188.
Vol. 16 No. 3 2007

Fig. 13. Presumed temperature proles in relation to calculation grid: (a) quasi-homogeneous model and (b) cellular model.

tion layer, while thermal conductivity of the reaction layer is much higher. Taking into account all these arguments, we still have to answer the question, why, analyzing the same equation (11), homogeneous and cellular model yields different results. The point is that the quasi-homogeneous model, looking for stationary steady-state solution of the problem, avoids situations when reaction heat evolution occurs faster than conductive heat transfer. If this happens, the quasi-homogeneous model allows decreasing x down to arbitrary small values, which results in the rise of temperature gradient and, therefore, heat ow jumps up. Doing this, we restore balance between heat evolution rate and rate of heat transfer from the reaction zone into preheating zone. Cellular model does not allow decreasing spatial step below dimension of the cell L (as a result, overheated hot spots appear in the combustion front); instead, it gives possibility to study steady-state propagation of the combustion wave in forms of sequence of thermal explosions in the cells. Another evident question is: what if we decrease the size of the cell, L, shall we come to the quasi-homogeneous model? An answer depends on effect of L on the reaction rate. The time of heterogeneous reaction may decrease with decreasing L, in this case a decrease in L not necessarily restore a

INTERNATIONAL JOURNAL OF SELF-PROPAGATING HIGH-TEMPERATURE SYNTHESIS

152

ROGACHEV, BARAS Reaction Waves in Heterogeneous Media, Proc. Natl. Acad. Sci. USA, 1998, vol. 95, pp. 11 05311 058. Mukasyan, A.S., Rogachev, A.S., and Varma, A., Microstructural Mechanism of Combustion in Heterogeneous Reaction Media, Proc. Combust. Inst., 2000, vol. 28, pp. 14131419. Kochetov, N.A., Rogachev, A.S., and Merzhanov, A.G., Causes of Thermal Heterogeneity of SHS Wave, Dokl. Akad. Nauk, 2003, vol. 389, no. 1, pp. 6567. Mukasyan, A.S., Rogachev, A.S., Mercedes, M., and Varma, A., Microstructural Correlations between Reaction Medium and Combustion Wave Propagation in Heterogeneous Systems, Chem. Eng. Sci., 2004, vol. 59, pp. 50995105. Merzhanov, A.G., Propagation of Solid Flame in a Model Heterogeneous System, Dokl. Ross. Akad. Nauk, 1997, vol. 353, pp. 504507. Krishenik, P.M., Merzhanov, A.G., and Shkadinskii, K.G., Unsteady-State Conditions for Conversion of Multilayer Heterogeneous Systems, Fiz. Goreniya Vzryva, 2002, vol. 38, no. 3, pp. 7079. Krishenik, P.M., Merzhanov, A.G., and Shkadinskii, K.G., Conditions for Frontal Transformation of High-Energy Structured Heterogeneous Systems, Fiz. Goreniya Vzryva, 2005, vol. 41, no. 2, pp. 5161. Krishenik, P.M. and Shkadinskii, K.G., Stability of Thermal Front with Heat Conductivity Dependent on Temperature, Int. J. SHS, 2005, vol. 13, no. 4, pp. 253261. Vadchenko, S.G. and Merzhanov, A.G., Heterogeneous Model of Flame Propagation, Dokl. Akad. Nauk, 1997, vol. 352, no. 4, pp. 487489. Vadchenko, S.G., Gasless Combustion of a Model Multilayer System: Combustion of Disks with a Gap, Fiz. Goreniya Vzryva, 2001, vol. 37, no. 2, pp. 4250. Vadchenko, S.G., Gasless Combustion of a Model Multilayer System: Combustion of Disks without Gaps, Fiz. Goreniya Vzryva, 2002, vol. 38, no. 1, pp. 4953. Seplyarskii, B.S. and Vadchenko, S.G., Role of Convective Heat Transfer in Gasless Combustion by the Example of Combustion of the TiC System, Dokl. Phys. Chem., 2004, vol. 398, no. 1, pp. 203207. Kochetov, N.A., Rogachev, A.S., Emelyanov, A.N., Illarionova, E.V., and Shkiro, V.M., Microstructure of Heterogeneous Mixtures for Gasless Combustion, Fiz. Goreniya Vzryva, 2004, vol. 40, no. 5, pp. 7480. Doronin, V.N., Itin, V.I., and Barelko, V.V., Non-Thermal Self-Activation Mechanism of Interaction Process for Solid Reagents Mixture in Combustion Wave, Dokl. Akad. Nauk SSSR, 1986, vol. 286, no. 5, pp. 11551159. Rogachev, A.S., Microheterogeneous Mechanism of Gasless Combustion, Combust. Explos. Shock Waves, 2003, vol. 39, no. 2, pp. 150158 (Engl. transl. of Fiz. Goreniya Vzryva). Rogachev, A.S. and Merzhanov, A.G., Theory of RelayRace Propagation of a Combustion Wave in Heterogeneous Systems, Dokl. Akad. Nauk, 1999, vol. 365, no. 6, pp. 788791. Rogachev, A.S., Kochetov, N.A., Kurbatkina, V.V., Levashov, E.A., Grinchuk, P.S., Rabinovich, O.S., Sachkova, N.V., and Bernar, F., Microstructural Aspects of Gasless Combustion of Mechanically Activated Mixtures. I. High-Speed Microscale Video Recording of a
Vol. 16 No. 3 2007

10. Shkiro, V.M. and Borovinskaya, I.P., Capillary Flow of Liquid Metal at Burning of the TitaniumCarbon Mixtures, Fiz. Goreniya Vzryva, 1976, vol. 12, no. 6, pp. 945948. 11. Nekrasov, E.A., Maksimov, Yu.M., Ziatdinov, M.Kh., and Shteinberg, A.S., Capillary Spreading Inuence on the Combustion Wave Propagation in Gasless Systems, Fiz. Goreniya Vzryva, 1978, vol. 14, no. 5, pp. 2633. 12. Rogachev, A.S., Mukasyan, A.S., and Merzhanov, A.G., Structural Transformations at Gasless Combustion of TitaniumCarbon and TitaniumBoron Systems, Dokl. Akad. Nauk SSSR, 1987, vol. 297, no. 6, pp. 14251428. 13. Knyazik, V.A., Merzhanov, A.G., and Shteinberg, A.S., On Combustion Mechanism of TitaniumCarbon System, Dokl. Akad. Nauk SSSR, 1988, vol. 301, no. 4, pp. 899902. 14. Merzhanov, A.G., Rogachev, A.S., Mukasyan, A.S., and Khusid, B.M., Makrokinetics of Structure Transformations during Gasless Combustion of TitaniumCarbon Powder Mixtures, Fiz. Goreniya Vzryva, 1990, vol. 26, no. 1, pp. 104114. 15. Merzhanov, A.G. and Rogachev, A.S., Structural Macrokinetics of SHS Processes, Pure Appl. Chem., 1992, vol. 64, no. 7, pp. 941953. 16. Rogachev, A.S., Varma, A., and Merzhanov, A.G., The Mechanism of Self-Propagating High-Temperature Synthesis of Nickel Aluminides. I. Formation of Product Microstructure in a Combustion Wave, Int. J. SHS, 1993, vol. 2, no. 1, pp. 2538. 17. Rogachev, A.S., Shugaev, V.A., Khomenko, I.O., Varma, A., and Kachelmyer, C.R., On the Mechanism of Structure Formation during Combustion Synthesis of Titanium Silicides, Combust. Sci. Technol., 1995, vol. 109, pp. 5370. 18. Shteinberg, A.S., Shcherbakov, V.A., and Munir, Z.A., Kinetics of Combustion in the Layered NiAl System, Combust. Sci. Technol., 2001, vol. 24, pp. 124. 19. Aleksandrov, V.V., Gruzdev, V.A., and Kovalenko, Yu.A., Thermal Conduction of Some SHS Systems Based on Aluminum, Fiz. Goreniya Vzryva, 1985, vol. 21, no. 1, pp. 98104. 20. Emelyanov, A.N., Shkiro V.M., Rogachev A.S., and Kochetov, N.A., Thermal Conductivity of Powder Mixtures for Gasless Combustion, Izv. Vyssh. Uchebn. Zaved., Tsvet. Met., 2005, no.1, pp. 6063. 21. Levashov, E.A., Bogatov, Yu.V., and Milovidov, A.A., Macrokinetics and Mechanism of a Self-Propagating High-Temperature Synthesis Process in TitaniumCarbon-Based Systems, Fiz. Goreniya Vzryva, 1991, vol. 27, no. 1, pp. 8893. 22. Merzhanov, A.G., Mukasyan, A.S., Rogachev, A.S., Sytschev, A.E., Hwang, S., and Varma, A., Microstructure of the Combustion Front in Heterogeneous Gasless Systems: Combustion in the 5Ti + 3Si System, Comb. Explos. Shock Waves, 1996, vol. 32, no. 6, pp. 6881 (Engl. transl. of Fiz. Goreniya Vzryva). 23. Rogachev, A.S., Mukasyan, A.S., and Varma, A., Microstructure of Self-Propagating Waves of Exothermal Reactions in Heterogeneous Media, Dokl. Ross. Akad. Nauk, 1999, vol. 366, no. 6, pp. 777780. 24. Varma, A., Rogachev, A.S., Mukasyan, A.S., and Hwang, S., Complex Behavior of Self-Propagating

25.

26. 27.

28. 29.

30.

31. 32. 33. 34. 35.

36.

37.

38.

39.

40.

INTERNATIONAL JOURNAL OF SELF-PROPAGATING HIGH-TEMPERATURE SYNTHESIS

MODELS OF SHS: AN OVERVIEW Ni + Al Composition, Fiz. Goreniya Vzryva, 2006, vol. 42, no. 4, pp. 6170. Machviladze, G.M. and Novozilov, B.V., Zh. Prikl. Mekh. Tekh. Fiz., 1971, vol. 5, no. 1, p. 51. Matkowsky, B.J. and Sivashinsky, G.I., SIAM J. Appl. Math., 1978, vol. 35, p. 465. Shkadinskii, K.G., Khaikin, B.I., and Merzhanov, A.G., Propagation of a Pulsating Exothermic Reaction Front in the Condensed Phase, Combust. Expl. Shock Waves, 1971, vol. 7, no. 1, pp. 1522. Shkiro, V.M. and Borovinskaya, I.P., Combustion Regularities of TitaniumCarbon Mixtures, in Protsessy goreniya v khimicheskoi tekhnologii i metallurgii (Combustion Processes in Chemical Technology and Metallurgy), Merzhanov, A.G., Ed., Chernogolovka: Izd. Inst. Chem. Phys., 1975, pp. 253258. Kirdyashkin, A.I., Maksimov, Yu.M., and Nekrasov, E.A., Mechanism of Titanium Interaction with Carbon in Combustion Wave, Fiz. Goreniya Vzryva, 1981, vol. 17, no. 4, pp. 3336. Dunmead, S.D., Readey, D.W., Semler, C.E., and Holt, J.B., Kinetics of Combustion Synthesis in the TiC and TiCNi Systems, J. Am. Ceram. Soc., 1989, vol. 72, pp. 23182324. Borovinskaya, I.P., Merzhanov, A.G., Novikov, N.P., and Filonenko, A.K., Gasless Combustion of Powder Mixtures of Transition Metals with Boron, Fiz. Goreniya Vzryva, 1974, vol. 10, no. 1, pp. 415.

153

41. 42. 43.

44.

45.

46.

47.

48. Azatyan, T.S., Maltsev, V.M., Merzhanov, A.G., and Seleznev, V.A., Propagation Mechanism of Combustion Wave in TitaniumBoron Mixtures, Fiz. Goreniya Vzryva, 1980, vol. 16, no. 2, pp. 3742. 49. Dunmead, S.D., Munir, Z.A., and Holt, J.B., Temperature Prole Analysis in Combustion Synthesis: II. Experimental Observations, J. Am. Ceram. Soc., 1992, vol. 75, no. 1, pp. 180188. 50. Andreev, V.A., Maltsev, V.M., and Seleznev, V.A., Investigation of HafniumBoron Combustion by Optical Pyrometry, Fiz. Goreniya Vzryva, 1980, vol. 16, no. 4, pp. 1823. 51. Azatyan, T.S., Maltsev, V.M., Merzhanov, A.G., and Seleznev, V.A., Some Peculiarities of TitaniumSilicon Mixtures Burning, Fiz. Goreniya Vzryva, 1979, vol. 15, no. 1, pp. 4349. 52. Naiborodenko, Yu.S. and Itin, V.I., Gasless Combustion of Powder Mixtures of Different Metals. I. Peculiarities and Combustion Mechanism, Fiz. Goreniya Vzryva, 1975, vol. 11, no. 3, pp. 343353. 53. Itin, V.I., Bratchikov, A.D., and Lepinskikh, A.V., Phase Transition at Combustion of CopperAluminum Powder Mixtures, Fiz. Goreniya Vzryva, 1981, vol. 17, no. 5, pp. 3134. 54. Itin, V.I. and Naiborodenko, Yu.S., Vysokotemperaturnyi sintez intermetallicheskikh soedinenii (High-Temperature Synthesis of Intermetallic Compounds), Tomsk: Izd. Tomsk. Univ., 1989.

INTERNATIONAL JOURNAL OF SELF-PROPAGATING HIGH-TEMPERATURE SYNTHESIS

Vol. 16

No. 3

2007