Journal of the fess-Common
AN ANALYSIS DEHYDRIDING
Metals, 123 (1986) 209 - 222
OF HEAT EFFECT ON THE HYDRIDING AND KINETICS OF FeTi INTERMETALLIC COMPOUND
H. S. CHUNG and JAI-YOUNG
Department of Materials Science and Engineering, Korea Advanced and Technology, P.O. Box 131, Chongryang, Seoul (Korea) (Received August 20,1984;
Institute of Science
accepted January 31,1986)
Summary Reaction rates and temperature changes in the reactor were measured during the hydridmg and dehydriding reactions of FeTi while the heat effect was reduced progressively by applying a thermal ballast technique and using a highly heat conductive copper tube reactor with the ballasts. When mixed with ballast material (manganese in this work), was hydrided FeTir.04, in the copper reactor, the heat effect was reduced significantly so that the assumption of isothermal conditions was possible; from the results an empirical rate equation was obtained. The heat effect on the rates is analysed with the aid of the empirical rate equation, which is proved to be a powerful tool for understanding the intrinsic hydriding kinetics of FeTi. Hydriding rate data, which were obtained by applying thermal ballast in the copper reactor, show parabolic dependence on the applied hydrogen pressure at low temperature and high hydrogen concentration and linear dependence at high temperature and low hydrogen concentration, indicating a transition of the kinetic mechanism with temperature and hydrogen concentration. The heat effect is more pronounced during the dehydriding reaction than the hydriding reaction. Dehydriding rate data with minimum heat effect are also presented.
1. Introduction Intrinsic kinetics of the hydriding and dehydriding reaction of hydrogen-storing intermetallic compounds are worthwhile studing as a reference to the reaction rates in practical hydrogen storage units and to check the degree of contamination by gas impurity in the hydrogen gas, as well as being interesting academically. The kinetic mechanism of the hydriding and dehydriding reaction should be investigated on the basis of intrinsic rate data. Unfortunately, hydrogenation reactions of intermetallics are accompanied by temperature changes in the reactor due to enthalpy changes of the reaction. Most hydrogen-sto~ng intermetallics absorb and desorb hydrogen very 0022-5088/86/$3.50
@ Elsevier Sequoia/Printed
in The Netherlands
fast, and thus it is difficult to obtain intrinsic rate data without the heat effect [I]. For FeTi, the 01+ p transformation generates heat of about -5500 (cu+/3) or +7000 (6 -+ Q) eal (mol Hz)-’ [Z - 41. This is enough to cause the temperature of the specimen to be changed over 200 or 300 “C in the adiabatic condition considering the heat capacities of FeTi and FeTiH to be 0.112 and 0.124 cal g-’ K-’ respectively [ 51. Rudman  recognized that intrinsic rate data cannot be obtained only by improving the heat conductivity of the reactor, and then measured the hydriding and dehydriding rates of FeTi in an elaborately designed reactor, considering heat transfer by monitoring the temperature sweep in the reactor during the reaction. The rate data was analysed by calibrating the heat effect with the aid of the following heat balance equation [ 61: -
where xP is the equilibrium hydrogen-to-metal ratio at an applied hydrogen pressure, M the amount of metal atoms, AH the enthalpy change of the reaction and F (= x/xP, x being the hydrogen-to-metal ratio in the specimen) is the reacted fraction, t is the time, h (= Ksfd) is the heat transfer factor (where X, s and d are the heat conducti~ty, heat conducting area and heat conducting length of the reactor respectively), T the temperature in the reactor, TA the ambient temperature and C is the heat capacity of the system. However, because of the difficulty in defining the accurate values of h and C, the non-uniform temperature in the reactor, complex dependence of dF/dt on both T and F and the limitation in the sensitivity of the thermocouple, an analytical approach to eqn. (I) is difficult and there is room for large errors in the calibration of the heat effect. Therefore it is doubtful if intrinsic kinetic properties can be properly obtained from the above type of approach. Goode11 and Rudman  suggested that the heat effect should be reduced effectively by mixing the sample with thermal ballast material which should be inert to hydrogen. They proposed eqn. (2) which relates the maximum temperature excursion (AT) that can occur for the adiabatic case with the amount of ballast material. AT=(-
where b is the atomic fraction of ballast. It follows from eqn. (2) that, for the cy -+ p transformation in FeTi, more than 98% ballast should be used in order to limit AT to less than 10 “C. However, the actual temperature excursion in the reactor should be less than that calculated from eqn. (2) because heat is transferred to the bath during the reaction, and so it may be possible to limit AT to smaller values by improving the heat conductivity of the reactor even with less amount of ballast than that required in the adiabatic condition. It seems impossible to eliminate the heat effect completely in the hydriding reaction of intermetallics and thus some heat effect should be
expected for a given set of experimental data and a determination should be made as to whether the assumption of isothermal conditions is reasonable or not. As a preliminary step toward the understanding of the hydriding kinetics of intermetallics, the nature of the heat effect must be understood. In the present study, the heat effect is being reduced progressively by applying thermal ballast and then by using a highly heat conductive copper tube reactor as well as the ballast. The hydriding rate and the temperature sweep in the reactor are obtained during the hydrogenation reaction. The heat effect is analysed for each test condition on the basis of both the rate data and the temperature change in the reactor and it will be observed how the kinetic data approach the intrinsic condition as the heat effect is reduced sequentially. An examination will also be made to see whether the experimental data in the copper tube reactor can be assumed to be the isothermal ones. 2. Experimental
Sample preparation procedure, experimental apparatus and testing method were the same as those in ref. 8. The compositions of the specimens were FeTi 1.o4 and FeTi. It was verified by X-ray diffraction that almost all of the sample was in the single phase range of FeTi and a small amount of oxide phase was seen by optical microscopy. PCT curves of FeTiieo4 were similiar to those of stoichiometric FeTi except that plateau pressure is somewhat lowered and the plateau is slightly sloped. The reaction rates were nearly the same for each of the specimens. Figure 1 shows the two types of reactors used in the experiment. In the stainless steel (S.S.) reactor, A, 1 g FeTiieo4 only or 1 g FeTileo4 mixed with 6 - 7 g manganese powder was hydrided and dehydrided and the temperature sweep in the reactor was measured during the reaction. Using a smaller amount of specimen (less than 1 g) causes difficulty in measuring the rate accurately and the amount of manganese used as ballast was determined so that the available reactor volume was filled with the specimen. It was found by Kim and Lee  that the activation was hindered when FeTi was mixed with certain metal powders, i.e. copper and nickel, for which the oxide stabilities are lower than that of iron; therefore manganese, which does not absorb hydrogen and whose oxide stability is higher than that of iron, was chosen as the ballast material. The possibility of any chemical influence of manganese except the reduction of the heat effect may be excluded because the hydriding reaction occurs in a relatively low temperature range after the sample has been fully activated. In order to remove impurities on the surface of manganese particles, manganese powder was mixed with the specimen after being fully baked in a vacuum at 400 “C. Kinetic experiments were done after the sample had been fully activated. The size of FeTi particles was below 200 mesh and that of manganese, -100 +200 mesh. An abrupt backflow of hydrogen gas occurs in the initial range of the dehydriding reaction and FeTi particles which are relatively
smaller and lighter than manganese particles may tend to be segregated in the upper region of the bed. Thus it is doubtful if the heat effect can be reduced further by using a larger stainless steel reactor in which the ratio of manganese to FeTi can be increased. Therefore a copper tube reactor, shown in Fig. 1, was designed in order to reduce the heat effect further. 0.8 g FeTil,04, mixed with about 30 g manganese, was hydrided in the copper reactor. It is expected that the heat effect should be reduced significantly by using the copper reactor because the path of heat transfer is shortened and the wall of the reactor ic; highly heat conductive and segregation of FeTi particles is
Fig. 1. Two types of reactor used in the experiment. The length of the copper tube is 30 cm.
suppressed. The larger the contact area of the powder bed, the better the heat conductivity of the bed. It is also expected that the heat effect may be reduced more efficiently by using smaller manganese particles or by compacting the bed. In the present study, however, the size of manganese particles is limited to -100 +200 mesh and compacting of the bed was not attempted in order to reject the possibility of the influence of mass transfer of hydrogen molecules through the bed. The hydriding and dehydriding reaction was limited to the (Y+ /Itransformation. 3. Results and discussion Figures 2 - 4 show the variations of both the reaction rate and the temperature in the reactor with hydrogen concentration in specimens for given experimental conditions in the stainless steel reactor. It is observed that the temperature in the reactor changes significantly even when a small amount of sample reacts for the sake of experimental purposes. The assumption of isothermal conditions cannot be justified even when FeTilVo4 is mixed with manganese powder in the stainless steel reactor. Since the hydriding reaction
Fig. 2. Changes of the hydriding reaction rate and the temperature in the reactor with H:M. FeTir.w reacted at 50 “C bath temperature in the S.S. reactor: 1 and 4 Pi = 35.8 atm; 2 and 0, Pi = 31.8 atm; 3 and A, Pi = 28.0 atm; 4 and 0, Pi = 24.0 atm (Pi is initially applied hydrogen pressure and Pi - PH:M= 0.5 = 5.46 atm). Fig. 3. Changes of the hydriding reaction rate and the temperature in the reactor with H:M. FeTir.w, mixed with manganese powder, reacted at 30 “C bath temperature in the S.S. reactor: 1 and 0, Pi = 31.65 atm; 2 and A, Pi = 25.4 atm; 3 and a, Pi = 19.6 atm (Pi - PH:M= 0.50 = 4.58 atm).
is exothermic and a thermally activated process, the heat generated by the reaction increases the temperature in the reactor and this results both in an increase of the mobility of the reaction and in a decrease of the driving force of the reaction by raising the plateau pressure. Thus the hydriding reaction rate may be increased or decreased by the heat effect. In the dehydriding reaction, however, the heat effect causes a decrease of both the mobility and the driving force of the reaction, and the reaction rate must be decreased drastically by the heat effect. Figure 4 shows a drastic decrease of the reaction rate in the initial range of the dehydriding reaction. This tendency agrees well with the abrupt decrease of the temperature in the reactor, and it is certain that the initial abrupt decrease of the reaction rate is at least partially due to the heat effect. Figure 5 indicates the relation between the maximum temperature change and the initial hydrogen pressure when the sample reacts in the stainless steel reactor and with manganese ballast in the copper tube reactor. The higher the initial hydrogen pressure, the faster the
with Mn I” SS reactor @ (3O’C) 00°
(3O’C) WITHM~ I” cu reactor
A 20 30 IO inttiolly opplied hydrogen
Fig. 4. Changes of the dehydriding reaction rate and the temperature in the reactor with H:M. FeTiIeW, mixed with manganese powder, reacted at 30 “C bath temperature in the S.S. reactor, the reaction started from H:M = 0.50. 1 and 4 Pi = 0.4 atm; 2 and 0, Pi = 0.99 atm; 3 and A, Pi = 1.56 atm; 4 and f Pi = 2.13 atm (PH:M=e-Pi = 0.56 atm). Fig. 5. Maximum temperature sure.
changes in the reactor vs. initially applied hydrogen pres-
reaction rate and the bigger the temperature change in the reactor. The temperature change with manganese in the copper tube reactor was detected to be very small (less than 2 “C). However it cannot be completely excluded that the temperature increase of the FeTi particle itself may be somewhat higher than that detected in the bed since it is not certain that the heat generated in the FeTi particle is transferred to the ballast sufficiently fast. In this situation, however, it is reasonable to assume that the experimental data with manganese ballast in the copper tube reactor correspond to isothermal conditions. Therefore the investigation of the behaviour of reaction rates for three experimental conditions permits us to see how the kinetic data approach the intrinsic case. Figures 6 and 7 show the dependence of the reaction rate on the applied hydrogen pressure at a constant hydrogen-to-metal ratio (H:M) at each experimental condition in the stainless steel reactor. Generally the reaction rate is linearly proportional to the applied hydrogen pressure. The hydrogen pressure at which the reaction rate becomes zero, deduced from the extrapolation of these lines, coincides well with the absorption plateau pressure of
Fig. 6. Hydrogen absorption rate us. hydrogen pressure. (FeTiIeo4, reacted in the S.S. reactor; H:M = 0.10.)
=012 A 50.C
m :: s
i L! 0
6 -e “, 00
00 0 0
c H P
’ _ 0 0.’
20 Hydroqen ~rersun
Fig. 7. Hydrogen absorption rate us. hydrogen pressure. (FeTiIeW, mixed with manganese powder, reacted in the S.S. reactor; H:M = 0.12.)
the alloy. These results are due to the competitive contributions of both the heat effect and the intrinsic kinetic properties, and so the slopes of the lines are meaningful only as parameters which can be used for comparing the reaction rates at different bath temperatures when (Pn, - Pep) is identical, where Peg is the plateau pressure when the temperature in the reactor is identical with the bath temperature. The slopes are considered as the apparent rate constants by which the heat effect can be compared for given test conditions.
Fig. 8. Hydriding reaction rate us. hydrogen pressure. (FeTir_w, mixed with manganese powder, reacted in the copper tube reactor; H:M = 0.12.)
Figures 7 and 8 show the dependences of the hydriding reaction rate on the hydrogen pressure in the copper reactor with ballast. Generally the hydriding rate is linearly proportional to the hydrogen pressure. Unlike the results in the stainless steel reactor, however, the reaction rate shows parabolic-like PHP 20
I’ 0’ :
a’ ‘5 ii
#’ 0 ,4OT
‘&4' f i +?i c )I(' +" t0.5- ;ig&) 0II 9,. o*c r OS I &?y' ,b p :A /
Fig. 9. Hydriding reaction rate vs. hydrogen pressure. (FeTir.,+ mixed with manganese powder, reacted in the copper tube reactor; H:M = 0.40.) DortFd lines represent the results of best fit among the linear regressions of rate vs. P+ PH
dependence on the hydrogen pressure at low temperature and high H:M. This phenomenon is the result of the reduction of the heat effect with manganese ballast in the copper tube reactor and represents quite closely the intrinsic kinetic properties. Figure 9 shows the dependences of the hydriding reaction rate on the hydrogen pressure at 0, 25,40 and 50 “C. At 50 “C the hydriding reaction rate is proportional to PH2. However, generally at low temperature the hydriding reaction rate is proportional to Puzl”. Nonetheless linear dependences were assumed in the whole range of the reaction and the apparent rate constants were defined as the slopes of the linear regression of the reaction rate versus Pwz. This empirical treatment does not matter for the sake of comparing the heat effect with the results in the stainless steel reactor. Figures 10 - 12 show the variation of the apparent rate constants with H:M. These figures represent how the kinetic data approach the intrinsic case as the heat effect is diminished. Figure 10 shows the abnormal tendency that the apparent rate constant is decreased as the bath temperature increases, indicating an apparently negative activation energy. This phenomenon is due to the increase of the change of plateau pressure per unit temperature change. For FeTi, the increase of the plateau pressure per unit temperature is 0.147, 0.297,0.53 and 0.716 atm “C’ at 0, 25, 50 and 65 “C respectively [ 51. Therefore the driving force of the reaction is reduced more by the heat effect at the higher temperature. When the temperature variations are very large, as is the case for Fig. 10, the reduction of the driving force due to the heat effect dominates the increase of reaction mobility at high temperature, decreasing the apparent rate constant as the bath temperature increases. Figure 11 implies that, as the heat effect is reduced, the reaction rate is decreased at 0 “C and increased at 50 “C. At 65 “C the heat effect still dominates over the intrinsic kinetic properties
H/M Fig. 10. Variations of the apparent rate constants with H:M. (FeTi,_w reacted in the S.S. reactor.)
30-c A A
A 0000 A
Fig. 11. Variations of the apparent rate constants with H:M. (FeTil.w, ganese powder, reacted in the S.S. reactor.)
mixed with man-
Fig, 12. Variations of the apparent rate constants with H:M. (FeTil.w, ganese powder, reacted in the copper tube reactor.)
mixed with man-
because the increase of the plateau pressure is drastic even for a relatively small increase of the temperature in the reactor. Comparing Fig. 11 with Fig. 12, the largest difference is shown in the middle range of the reaction. This coincides well with the fact that the heat effect is the largest in that region (Fig. 3). Considering that the results in Fig. 11 were obtained with a maximum temperature change of less than about 10 “C, it can be concluded that the results in Fig. 12 approach quite nearly
the intrinsic kinetic properties even though traces of the heat effect cannot be completely removed. Figure 13 shows that the dehydriding reaction rate with manganese in the copper tube reactor is linearly proportional to Pn,r’* at H:M = 0.20. The slopes of the lines are defined as the apparent rate constants. The activation energy, pressure dependence of the reaction rate and the empirical relation obtained from the data of Figs. 9, 12 and 13 are shown in Table 1. The experimental rate of hydriding was plotted as rate versus T at 0-c . .
AA A . .
Fig. 13. Dehydriding reaction rate us. hydrogen pressure. (Stoichiometric with manganese powder, reacted in the copper tube reactor.) (H:M)sti,l H:M = 0.20 represents the approximate middle range of the reaction.
FeTi, mixed = 0.52 and
TABLE 1 Hydriding and dehydriding kinetics of FeTi Pressu i-e dependence
Activation energy (cal (mol H&l)
Rate = A(PH2 -Peg)
Rate = A’(P,q1’2
- PH2”2) exp(-4754iRT)
aA is constant varying with H:M; P, and P,, are the applied and absorption equilibrium hydrogen pressures respectively and dis the gas constant.
Fig. 14. Variation of the hydrogen absorption rate with the temperature in the reactor at constant hydrogen pressures. (The curves originate with the equations in Table 1.) constant P,* in Fig. 14 which is assumed to be the intrinsic one as mentioned before. Figure 14 can play a useful role in the qualitative analysis of the heat effect even though its use in a quantitative analysis is suspect because the pressure dependence of reaction rate is proportional to ln(Pn2/Pe9) or (Pn,1’2 - Pe,1'2) rather than Pn, at low temperature. Generally, to find the dependence of the reaction rate on the hydrogen pressure at a constant temperature, the variation of the reaction rate must be observed in the direction of A in Fig. 14, namely, under the condition that only applied pressure is different at constant temperature. But actually the heat effect causes the observation to be done in the directions of B which are determined by the heat transfer capability of the reactor. In this condition, temperature is different as well as pressure, for example, at points C and D the temperature is different because the higher the applied pressure, the bigger the temperature change in the reactor. From Fig. 14 it can be seen that at low T, the heat effect increases the reaction rate faster than for the intrinsic case (case of B’ > A’); however, at high 2’ the heat effect decreases the reaction rate drastically (case of A” > B”) even for relatively small temperature changes. The tendencies in Figs. 10 - 12 agree well with the above arguments. Because the hydriding reaction is a thermally activated process, the hydriding reaction rate should be faster at higher temperatures than at low temperature. However, Fig. 10 shows the reverse trend. This is because, as mentioned above, at high T the heat effect decreases the reaction rate but at low T increases the reaction rate. Comparing the result of Fig. 12 with that of Fig. 11, at 0 “C the reaction rate is decreased but at 50 “C the reaction rate is increased. This
tendency is contrary to that expected by the heat effect, i.e. the result from removal of the heat effect. This is further evidence that the results in Fig. 12 should represent nearly intrinsic kinetics. From the above, considering variations of both the temperature excursions measured in the reactor during the reaction and the rate data with progressive reduction of the heat effect, it can be concluded that the rate data obtained by applying thermal ballast, manganese, in the copper tube reactor represent approximately the intrinsic kinetics of the hydrogenation of FeTi. The dependence of the hydriding rate on the hydrogen pressure and H:M, two very important matters for studying hydriding kinetics. which have been hidden by the heat effect, are revealed under the approximately intrinsic conditions. The empirical rate of the dehydriding reaction is plotted as rate uersus 2’ at constant PH, in Fig. 15. The desorption plateau pressure used here was obtained from ref. 3. In the dehydriding reaction, the heat effect causes the observation of the dependence of the reaction rate on the hydrogen pressure to be done in the directions of B rather than A in Fig. 15. From Fig. 15, it can be shown that the dehydriding reaction rate is decreased drastically by the heat effect (for example, c > d at the same pressure, the temperature is different because of reaction heat) even for the relatively small temperature changes. In the experimental temperature range, the dehydriding reaction rate is faster than the hydriding rate in the initial stage of the reaction and the temperature sweep in the reactor is somewhat bigger (Fig. 4). Fast reaction rate and a drastic temperature effect on the reaction rate make it very difficult to
Fig. 15. Variations of the dehydriding reaction rate with temperature at constant hydrogen pressure. (The curves originate with the equations in Table 1.)
obtain the intrinsic dehydriding reaction rate experimentally. The result of Linder [lo], which implied first-order-like dehydriding kinetics, must have been influenced by the heat effect.
4. Conclusion A combination of thermal ballast and a highly conductive copper tube reactor was effective for reducing the heat effect in the hydrogenation and dehydrogenation of FeTi resulting in nearly intrinsic rate data. The hydriding rate is increased by the heat effect at low temperature and decreased at high temperature. The higher the temperature, the bigger the reduction of the rate due to the heat effect and so excessively high temperature must be avoided in the study of the intrinsic kinetics. The intrinsic hydriding rate of FeTi shows parabolic-like dependence on the applied hydrogen pressure at low temperature and high hydrogen concentration and linear dependence at high temperature and low H:M, which indicates a change in the mechanism of the kinetics. In the dehydrogenation the heat effect is more drastic than it is for hydrogenation.
References 1 P. D. Goode& G. D. Sandrock and E. L. Huston, J. Less-Common Met., 73 (1980) 135. 2 M. H. Mintz, S. Vaknin, S. Biderman and Z. Hadari, J. Appl. Phys., 52 (1) (1981) 463 - 467. 3 P. D. Goode11 and G. D. Sandrock, Metallurgical studies of hydrogen storage alloys, Find Rep., April 1980 (Brookhaven National Laboratories, Contract BNL 451117-S). 4 J. J. Reilly and R. H. Wiswall, Jr., Znorg. Chem., 13 (1) (1974) 218 - 222. 5 H. Wenzl and E. Lebsanft, J. Phys. F, 10 (1980) 2147 - 2156. 6 P. S. Rudman, J. Less-Common Met., 89 (1983) 93. 7 P. D. Goode11 and P. S. Rudman, J. Less-Common Met. 89 (1983) 117. 8 C. N. Park and Jai-Young Lee, J. Less-Common Met., 91 (2) (1983) 189. 9 H. C. Kim and Jai-Young Lee, J. Less-Common Met., 105 (1985) 247 - 253. 1.O D. L. Linder, Znorg. Chem., I7 (1978) 3721 - 3722.