J O U R N A L O F C H E M I S T R Y Materials Lithium loss kinetics from polycrystalline LixNi1-xO at high temperatures Ermete Antolini ENEA C.R. Casaccia, Via Aguillarese 301, I-00060 Santa Maria di Galeria, Roma, Italy Received 29th July 1997, Accepted 21st September 1998 Lithium loss kinetics from polycrystalline LixNi1-xO has been investigated in the temperature range 900–1500 °C by measuring the time and temperature dependence of the weight and lattice constant change of the samples.At 900 °C the rate of Li2O evaporation was controlled by lithium ion diVusion in LixNi1-xO. An initial region of fast diVusion followed by a region of slower diVusion was observed in the thermogravimetric measurements. This observation can be interpreted as the rapid diVusion of lithium ions along the grain boundaries and subsequent diVusion into the bulk of the grain.A plot of the logarithm of lithium loss following 2 h of thermal treatment at diVerent temperatures vs. the reciprocal of absolute temperature consisted of two straight lines, the slope depending on the activation energy of the process and the change of slope occurred at 1300 °C. This behaviour suggests that up to 1300 °C lithium ion diVusion from the bulk to the surface of LixNi1-xO particles was the rate-determining step.Above 1300 °C, instead, the evaporation process depended on the demixing reaction of Li2O(g) and NiO(s) at the surface of the particles. Pure, stoichiometric nickel oxide is a compound with cubic detect, and for x>0.30 the change of crystal structure during the evaporation process would complicate the interpretation NaCl-type structure.Reaction of Li2O and NiO in the presence of the results. of oxygen gives rise to the formation of LixNi2+1-2xNi3+xO solid solution, where the oxidation state of Ni partially changes from +2 to +3.1,2 The unit cell slightly decreases with increasing lithium content as a consequence of the diVerence Experimental in the ionic radius between Ni2+ and Ni3+ ions.For lithium Lithium nickel oxide solid solution was prepared by solid state atomic fraction x>0.31 a rhombohedral distortion of the cubic reaction of Ni and Li2CO3. Nickel powder (INCO 255) and structure by lithium nickel ordering on alternate (111) planes lithium carbonate (Merck 5671) in the nominal lithium atomic takes place, giving rise to a hexagonal simmetry.2–4 LixNi1-xO fraction x=0.30 were used as starting materials.This powder solid solutions are used as a cathode material in molten mixture was submitted to the following thermal treatments in carbonate fuel cells.5 Moreover LixNi1-xO has been studied air: (i) heating from room temperature to 900 °C at with the aim of developing humidity ceramic sensors.6 Studies 1.0 K min-1; (ii) isothermal treatment at 900 °C up to 40h; on the synthesis of LixNi1-xO with low (x0.23)7,8 and high (iii) heating at 3.0 K min-1 from 900 °C to the maximum (x0.30)9 lithium content by solid state reaction of a Ni and temperature in the range 1000–1500 °C; (iv) isothermal treat- Li2CO3 powder mixture indicated that solid solution occurs ment at the maximum temperature for 2 h and (v) cooling to in two steps: (i) first, formation of LixNi1-xO with x higher room temperature at 10 K min-1.Moreover two samples were than the nominal value at the grain surface, then (ii) diVusion treated for diVerent times at 1200 and 1400 °C, respectively. of lithium ions from the surface to the bulk of the particle.Thermogravimetric measurements were performed using a Few studies have dealt with Li2O evaporation from Du Pont 2000 thermal analysis system equipped with a 951 LixNi1-xO. A recent work of Sata10 on the vaporization of TGA module. lithium oxide from LixNi1-xO solid solution at temperatures XRD patterns were collected at room temperature after up to 700 °C stated that Li2O2 formation and its diVusion rate quenching of the samples on a Philips PW 1729 powder in the specimen might be related to the rate-determining step diVractometer equipped with a 1771 vertical goniometer using in the vaporization process.Above 1000 °C, Iida found that filtered Cu-Ka radiation. the evaporation of lithium oxide from the solid solution is diVusion-controlled and governed by a parabolic law.11 The distribution of remaining lithium ions in the solid solution Results and discussion following the loss of some Li2O was also investigated.Sata LixNi1-xO was formed during the dynamic step up to 900 °C found that the lithium concentration in the specimen decreased of the thermal treatment. As lithium loss takes place during linearly from the surface to the interior along the specimen the process of formation of LixNi1-xO from Ni/Li2CO3 mix- thickness.10 Azzoni et al.12 observed the coexistence of substitures, 14 we determined the lithium atomic fraction at the tutional solid solution NiO type and ordered solid solution beginning of the isothermal step at 900 °C from both X-ray LiNiO2 type structures with diVerent lithium content following diVraction (XRD) and thermogravimetric measurements.Li2O evaporation from ordered solid solutions (x>0.30). From XRD measurements, the following relation between Moreover, Berbenni et al.13 revealed structural and microlithium atomic fraction x and lattice constant a/A° was used:15 structural changes of LixNi1-xO solid solution during Li2O evaporation at 800 °C. x=(4.1748-a)/0.17756 (1) The aim of this work is to better understand the process of Li2O evaporation from the solid solution at high temperatures From the weight change, we utilised the following relation:16 in the range 900 to 1500 °C.The nominal lithium content of the composition investigated was x=0.30, as for x<0.30 the x=[A(1+Dm/m0)-MNiO]/[A(1+Dm/m0)-(MNi-MLi)] (2) final amount of lithium would be very low and diYcult to J.Mater. Chem., 1998, 8, 2783–2786 2783with A=MNi+xn MLi2CO3/2(1-xn) (3) where Dm/m0 is the weight change of the samples, xn is nominal lithium atomic fraction, and MNiO, MNi, MLi and MLi2CO3 are the molecular weights of the compounds. The value of x obtained was 0.260 from XRD measurements (a= 4.1286 A° ) and 0.265 from thermogravimetric measurements (Dm/m0=0.113), i.e.the same within experimental error. The Li2O evaporation reaction is given by eqn. (4). LixNi1-xOAa LiyNi1-yO+b Li2O+(b/2)O2 (4) where a=(1-x)/(1-y) and b=(x-y)/2(1-y) for x>y. This reaction takes place in two steps: (i) lithium ion diVusion to the grain surface and (ii) a demixing reaction of Fig. 2 Log–log plot of fractional lithium loss from the solid solution vs. thermal treatment time at 900 °C.Li2O(g) and NiO(s) at the surface of the particles. The basic phenomenological mass transport relation, governing a solid state diVusion process like lithium ion diVusion in LixNi1-xO, grain from the bulk to the lithium-poor boundaries. DiVusion is that of Fick: in polycrystalline solids is known to occur along grain bound- J=-D grad C (5) aries more rapidly than through the interior of the crystals.Atkinson found that Ni diVusion in NiO was enhanced at where J is the number of atoms crossing a unit area in unit grain boundaries with respect to lattice diVusion and that the time, C is the concentration of the mobile species, and the faster diVusion pathway had the lower activation energies.19 constant D is the chemical diVusion coeYcient. In solids atoms The fast grain boundary diVusion is caused by the segregation adopt reasonably well defined positions.Mass transport occurs of point defects to the core region where they have higher by atoms making transitions between these positions in such concentration and higher mobility than in the lattice.20 As a way that the time of transit is much less than the residence long as the rate of grain-boundary diVusion is greater than time at any particular position.Thus, diVusion can be thought that of lattice diVusion at all temperatures, we can assume of as occurring by particle hopping in a random way on a that in the diVusion controlled region not only at 900 °C, lattice of sites distributed in space.17 The hopping event but at all temperatures, lattice diVusion is initiated after between sites involves the particle crossing an energy barrier, the boundaries are lithium-poor owing to lithium loss by the necessary energy coming from thermal fluctuations with a grain-boundary diVusion. probability described by the Boltzmann distribution.Hence, XRD measurements confirm the results of thermogravi- the diVusion process is thermally activated and the diVusion metric measurements.Fig. 3 shows the 204 reflections of the coeYcient has the Arrhenius form: solid solution following diVerent thermal treatment times at D=D0 exp (-Ea/RT) (6) 900 °C. The reflection shifts towards lower angles with time, owing to lithium loss. From 0 to 3 h, we detect a broadening where Ea is the activation energy of diVusion.The evaporation of the peak, attributed to a lithium concentration gradient of kinetics can be expressed as: outer and inner parts of the grain, as a consequence of fast Cev=ktn (7) lithium loss at the boundaries. The result of the Rietveld refinement procedure indicated the presence of three solid where Cev is the amount of evaporated species, k=k0exp solutions with x=0.26, 0.24 and 0.22, respectively.A sharpen- (-Ea/RT) is the rate constant, t is the thermal treatment time ing of the reflection then occurs between 3 and 20 h, related and the exponent n is related to the reaction mechanism. to homogenisation of the solid solution owing to the diVusion DiVusion-controlled evaporation is described by n=0.5 to of lithium ions from the bulk to the boundaries of the particles. 0.8.18 Fig. 1 and 2 show normal and log–log plots of fractional The transition point between grain-boundary and lattice lithium loss from the solid solution vs. thermal treatment time diVusion indicates that about 30% of lithium in the solid at 900 °C, respectively. As can be seen in Fig. 2, the log–log solution is present at the grain boundaries or at layers adjacent plot breaks into two straight lines with diVerent slopes.The to the grain boundaries. slopes of these lines were 0.75 and 0.48, respectively. In both After 40 h of isothermal treatment at 900 °C, the lithium cases, a diVusion-controlled process is occurring. This result atomic fraction x remaining in LixNi1-xO is 0.123. Then, these can be interpreted in terms of rapid lithium diVusion along specimens were thermally treated at various temperatures for the grain boundaries, followed by lithium diVusion into the Fig. 3 DiVractometric traces of the 204 reflection of LixNi1-xO Fig. 1 Dependence of fractional lithium loss from the solid solution on thermal treatment time at 900 °C. following diVerent times of thermal treatment at 900 °C. 2784 J. Mater. Chem., 1998, 8, 2783–2786Fig. 6 Log–log plot of lithium loss a from the solid solution vs. Fig. 4 The logarithm of lithium loss a from the solid solution as a function of the reciprocal of absolute temperature. thermal treatment time at 1200 and 1400 °C. 2 h. From the relation: Cev=k0 exp(-Ea/RT) tn (7) plotting ln Cev vs. 1/T at constant time, we can determine the activation energy of the process.Fig. 4 shows the dependence of the logarithm of lithium loss a=mLiev/m0Li, (where mLiev is the amount of evaporated lithium, in this case after 2 h of thermal treatment, and m0Li is total lithium amount at the beginning of isothermal treatment) on the reciprocal of absolute temperature. A change in the slope of the plot is observed at 1300 °C.This feature is related to the change of the ratedetermining process. It can be inferred that at high temperatures the rate-determining step is the demixing reaction at the surface of the grain. To confirm this result we have evaluated the time-dependence of lithium loss at 1200 °C (diVusioncontrolled region) and 1400 °C (demixing reaction-controlled Fig. 7 Dependence of the logarithm of remaining lithium (1-a) in region), respectively.Fig. 5 shows a plot of lithium loss a(= the solid solution on thermal treatment time at 1200 and 1400 °C. mLiev/m0Li) vs. time of samples thermally treated at 1200 and 1400 °C. There are two pathways by which the rate of respectively. We have determined the correlation parameters evaporation of Li2O can occur at high temperature.(r2) for the experimental data and theoretical equations, the (a) DiVusion control: the relation of lithium loss to time is slopes of the plots shown in Fig. 6 (the slope is the value of given by eqn. (8): n) and 7 (the slope is the value of k) and their estimated a=ktn (8) standard errors. As can be inferred from the values reported in Table 1, at 1200 °C the best fit is shown according to the (b) Surface reaction: Li2O evaporation at the grain surface diVusion process, whereas at 1400 °C the best fit is shown follows first order kinetics [eqn.(9)]: according to the demixing reaction at the grain surface. d(1-a)/dt=k (1-a) (9) The activation energy for the diVusion process was 179 kJ mol-1. From the slope of ln[-ln(1-a)] vs. 1/T, the Thus, the relation between lithium loss a and isothermal heat of evaporation of Li2O from LixNi1-xO was evaluated treatment time fits a first-order rate law [eqn.(10)]: as 92 kJ mol-1. a=1 -exp(-kt ) (10) Fig. 6 and 7 show log–log plots of lithium loss (a) vs. time Conclusions (diVusion control ) and the plot of the logarithm of remaining In the present study it was observed that lithium loss from lithium (1-a) in the solid solution vs.time (surface reaction), LixNi1-xO with x=0.26 at 900 °C initially occurs by grainboundary diVusion of lithium ions in the solid solution. Then, for long times and temperatures up to 1300 °C, the ratedetermining step was lattice diVusion of lithium ions into LixNi1-xO. Above 1300 °C, a demixing reaction of Li2O(g) and NiO(s) at the grain surfaces was the rate-controlling step.Table 1 Correlation parameters (r2) of experimental data and theoretical equations, k and n values, and their estimated standard errors Temperature/ °C log a=log k+n log t log (1-a)=-kt 1200 r2=0.991 r2=0.985 n=0.65 Dn/n=0.063 k=0.014 h-1 Dk/k=0.071 1400 r2=0.978 r2=0.996 Fig. 5 Dependence of lithium loss a from the solid solution on thermal n=0.60 Dn/n=0.103 k=0.179 h-1 Dk/k=0.042 treatment time at 1200 ($) and 1400 °C (().J. Mater. Chem., 1998, 8, 2783–2786 278510 T. Sata, Ceram. Int., 1998, 24, 53. References 11 Y. Iida, J. Am. Ceram. Soc., 1960, 43, 117. 12 C. B. Azzoni, A. Paleari, V. Massarotti, M. Bini and D. Capsoni, 1 S. Van Houten, J. Phys. Chem. Solids, 1960, 17, 7. Phys. Rev. B, 1996, 53, 703. 2 J. B. Goodenough, D. G. Wickham and W. J. Croft, J. Phys. 13 V. Berbenni, V. Massarotti, D. Capsoni, R. Riccardi, A. Marini Chem. Solids, 1958, 5, 107. and E. Antolini, Solid State Ionics, 1991, 48, 101. 3 I. J. Pickering, J. T. Lewandowski, A. J. Jacobson and 14 E. Antolini, Mater. Lett., 1993, 16, 286. J. A. Goldstone, Solid State Ionics, 1992, 53–56, 405. 15 E. Antolini, M. Leonini, V. Massarotti, A. Marini, V. Berbenni 4 R. Stoyanova and E. Zhecheva, J. Solid State Chem., 1994, 108, and D. Capsoni, Solid State Ionics, 1990, 39, 251. 211. 16 A. Marini, V. Massarotti, V. Berbenni, D. Capsoni, R. Riccardi, 5 Fuel Cells Handbook, ed. A. J. Appleby and F. R. Foulkes, Van E. Antolini and B. Passalacqua, Solid State Ionics, 1991, 45, 143. Nostrand Reinhold, New York, 1990. 17 A. Atkinson, Mater. Sci. Technol., 1994, 11, 303. 6 T. Sato, C. Hsien-Chang, T. Endo and M. Shimada, J. Mater. Sci. 18 T. Sata and T. Yokoyama, Yogyo Kyokaishi, 1973, 81, 170. Lett., 1986, 5, 552. 19 A. Atkinson, Adv. Ceram., 1987, 23, 3. 7 E. Antolini, J. Mater. Sci. Lett., 1993, 12, 1947. 20 D. M. DuVy, J. Phys. C, 1986, 19, 4393. 8 E. Antolini, A. Marini, V. Berbenni, V. Massarotti, D. Capsoni and R. Riccardi, Solid State Ionics, 1992, 57, 217. 9 E. Antolini and M. Ferretti, Mater. Lett., 1997, 30, 59. Paper 8/05948I 2786 J. Mater. Chem., 1998, 8, 2783–2786