Surface area stability of lanthanum-doped ceria

  • Published on

  • View

  • Download


<ul><li><p>Solid State Ionics 63-65 (1993) 781-785 North-Holland </p><p>SOLID STATE IOIIICS </p><p>Surface area stability of lanthanum-doped ceria </p><p>M. P i jo lat , M. Pr in, M. Soustel le D~partement de Chimie Physique, Ecole Nationale Supbrieure des Mines, 158 Cours Fauriel, 42023 Saint-Etienne Cbdex 2, France </p><p>O. Touret and P. Nor t ie r Rhbne-Poulenc Recherches, C.R.A., 93308 Aubervilliers, France </p><p>High surface area ceria was doped with lanthanum ions and calcined at 900 K under various partial pressures of oxygen and water vapour. The kinetic rates ofcrystallite size increase were observed to vary as p~l/6. By addition of 0.5 and 1% of lanthanum ions, the rates were lowered, and then significantly decreased for 10% (cat.). A model of this process involved association between lanthanum ions and oxygen vacancies in order to explain the better stability of doped ceria. </p><p>1. Introduction </p><p>High surface area ceria is known to enhance the redox properties and the thermal resistance of cat- alysts used in the control of emissions from motor vehicles. Stabil ization of the surface area of many oxides like transition aluminas, t itanium dioxide and zirconia, may be achieved by the incorporation of additives. </p><p>In a previous study [ 1,2 ] we have shown that the textural evolution of a cerium dioxide powder was the result of two distinct processes: a rapid loss in microporosity and a crystallite size increase. This ar- ticle reports the kinetic study of the loss in surface area of undoped and lanthanum-doped ceria, due to crystallite size increase. The addit ion of lanthanum is achieved in order to stabilize the surface area, as it was already observed with y-alumina [3-6] , an- atase [7 ] and zirconia [8 ]. </p><p>2. Experimental </p><p>2.1. Samples </p><p>Cerium dioxide powder was supplied by Rh6ne- Poulenc. The major impurit ies were La 3+ ions (0.15% cat.), nitrates (0.005%) and chlorides </p><p>(0.001%). Its initial surface area, as measured from BET method was found to be about 100 m2.g -1, but more precise determinations [ 1,2 ] showed that the initial powder was microporous, with about 72 m2.g - I micropore surface area; the crystallite sur- face area was about 63 m2.g - l, to which corresponds a mean diameter equal to 13.7 nm as obtained from X-ray line broadening. </p><p>Lanthanum ions were added to the powder ac- cording to the incipient wet method from nitrate so- lutions. Final concentrations in lanthanum ions were 0.5, 1, 2 and 10% (cat.). </p><p>2.2. CaMnat ions </p><p>A small quantity of powder was calcined at fixed partial pressures in water vapour and oxygen. The temperature was chosen equal to 900 K for all ex- perimental conditions. </p><p>The doped samples were calcined at 900 K under flowing gases in a Pyrox B80 furnace, which allowed to anneal all the samples which were doped at var- ious lanthanum concentrations; this procedure was followed in order to avoid reproducibility errors. The good agreement between the two devices was verified. </p><p>0167-2738/93/$ 06.00 1993 Elsevier Science Publishers B.V. All rights reserved. </p></li><li><p>"782 M. Pijolat et al. / Surface area stability of La-doped ceria </p><p>2.3. Characterizations </p><p>In order to characterize the lanthanum-doped samples, the lattice parameter of ceria was measured from X-ray diffraction. With a lanthanum content equal to 10% (cat.), a value of 0.5417 nm instead of 0.5410 nm for the undoped sample was found. This suggested that lanthanum ions were located at cation sites of the lattice. </p><p>The mean crystallite size was obtained from X-ray diffraction line broadening by analysing the 111 reflection. </p><p>3. Results </p><p>3. I. Undoped ceria </p><p>value of n was optimized in order to give a good fit for all the experimental conditions. </p><p>The kinetic rate of crystallite growth dD/dt could then be readily obtained from the previous equation, as a function of (1 +At) n-~. Then the following re- lation could be obtained which relates the rate dD/ dt as a function of D: </p><p>dD [D ,~ 1/~ /2) </p><p>dt </p><p>Thus for a given value of D, it was possible to ob- tain the variation of the kinetic rate (that is ofA ) as a function of the oxygen partial pressure, as illus- trated in fig. 2. A linear regression was used to obtain the power law. Fig. 3 shows the kinetic rate varied linearly as a function of P~2 I/6. </p><p>No influence of water vapor could be evidenced [2,9]. </p><p>It should be noticed that the undoped samples of ceria contained 0.15% (cat.) of lanthanum ions as the main impurity (cf. 2.1.). </p><p>Fig. 1 shows the variation of the mean diameter of ceria crystallites (D) at 900 K versus time for var- ious partial pressures of oxygen. For each curve, it was possible to determine the parameter "A" such as: </p><p>/ )=Do(1 +At)", (1) </p><p>in which Do is the initial diameter for CeO 2 crystal- lites (Do= 13.7 nm) and n is equal to 0.085. This </p><p>3.2. Lanthanum-doped ceria </p><p>In fig. 4 the experimental rate as a function of lan- thanum concentration has been reported. The dotted line in fig. 4 corresponds to 0.13 kPa whereas the solid line corresponds to 13.33 kPa of oxygen in the calcination atmosphere. As it can be seen, the ad- dition of lanthanum ions slowed down the crystallite size increase. The inhibiting effect was the same one </p><p>40 - Rate ( nm.h- 1 ) </p><p>D] (nm) 24 ] 30- </p><p>22 </p><p>1 8 2 0 ~ ~ ~ f ~ ) 2 : ~ 14nm </p><p>1() 15 nm 16 </p><p>I 1 ~ 22.7 16nm 14~ 12~ ~!63 12 I 4 ~ 0.271 0 10 20 30 </p><p>0 2 4 6 8 10 PO 2 ( k Pa ) Time (hours) Fig. 2. Kinetic rate of crystallite growth versus oxygen pressure </p><p>Fig. I. Mean crystallite diameter of ceria versus time at 900 K. obtained for various mean crystallites. </p></li><li><p>M. Pijolat et al. / Surface area stability of La-doped ceria 783 </p><p>40 </p><p>30 </p><p>20 </p><p>10 </p><p>Rate (nm h -1) . / 14nm </p><p>/ j 15nm </p><p>/ J . i t - , / J ~ 1 8 n m </p><p>20nm 0.0 0.1 0.2 0.3 0.4 </p><p>PO2-1/6 ( Pa -1/6) </p><p>Fig. 3. Kinetic rate of crystallite growth versus p~l/6. </p><p>40- </p><p>30- </p><p>20- </p><p>10- </p><p>0 </p><p>0 </p><p>Rate ( nm.h -1 ) </p><p>I </p><p>2 I I I I </p><p>4 6 8 10 </p><p>Lanthanum content ( % ) </p><p>Fig. 4. Kinetic rate of crystallite growth in doped ceria versus lanthanum content: Po2 = 0.13 kPa (--) and 13.3 kPa (---). </p><p>for 0.5 and 1% (cat.). At higher lanthanum concen- trations, the rate was much more decreased than for 0.5 and 1%, and it reached nearly zero for 10% (cat.) of lanthanum ions. </p><p>It must be noticed that for the doped powders the inhibiting influence of oxygen partial pressure was no more observed (cf. dotted and solid lines of fig. 4). </p><p>4. In terpretat ion and d iscuss ion </p><p>4.1. Model l ing </p><p>It is well known that ceria accepts rather large con- centrations of foreign ions to form solid solutions in </p><p>which the incorporated ion takes place at a cation site of the lattice. This appears to occur in the case of lanthanum ion addition and thus the point defects to consider are those noted Vo, Cede, Late and (La, Vo) , with Kr6ger's notation. The last one occurs from the association equilibrium between La~ and Vo'. Its existence has been assumed since similar as- sociations were observed in ceria doped with yttrium and scandium ions [ 10,11 ]. </p><p>A model based on previous ones which were set- tled for several oxides like TiOz anatase [ 12 ], 7-A1203 [6] and ZrO2 [13] in order to account for accel- erating effect of water vapor can be proposed. In ta- ble 1 the six elementary steps of the model: ( 1 ) ox- ygen adsorption on convex surfaces of ceria crystallites, (2) oxygen diffusion, (3) vacancy cre- ation in the neck region between crystallites, (4) ox- ygen desorption in the neck region, (5) cerium dif- fusion and (6) vacancy annihilation at convex surfaces have been reported. The process of crystal- lite-size increase involves two reactional zones, noted "R &gt; 0" and "R &lt; 0", which differ from the sign of their curvature radius. </p><p>The calculation of the theoretical rate law for crys- tallite size increase can be done using the following approximations: </p><p>- ideal solid solution of defects in ceria (all activ- ity coefficients will be assumed equal to unity), </p><p>-Brouwer ' s approximation which applies to the charge balance of the crystal, which is given by: </p><p>2[Vo'] = [Cebe] +4[V~] ; </p><p>- rate-limiting step, which consists of four possi- ble steps (if one considers that equilibrium between ceria and gaseous oxygen is rapidly achieved due to the redox properties of this oxide): ambipolar dif- </p><p>Table 1 Modelling of the crystallite growth in ceria process. </p><p>Elementary step Kr6ger's notation </p><p>1 02 + Vo + 2Ce~: = O~ + 2Ce~:~ (R&gt;0) 2 V~'~</p></li><li><p>784 M. Pijolat et al. / Surface area stability of La-doped ceria </p><p>fusion of oxygen vacancies and electrons (step (2) ), vacancy creation (step (3)), cerium diffusion (step (5) ) and vacancy annihilation (step (6)). The de- tails of the calculations will be given in a further pub- lication, and have already been reported in ref. [9 ]. </p><p>In the case of undoped ceria, the results are as follows. </p><p>- F o r rate-limiting step (3), the theoretical rate does not depend on the oxygen partial pressure </p><p>(/)o2); - for rate-limiting step (2), the rate follows the law </p><p>given below: </p><p>"0 "O 3 jo Je </p><p>rc~ 4j o +j0, (3) </p><p>in which jo and jo are the normal diffusion current of oxygen vacancies and electrons respectively in the absence of electric field. In the Brouwer's approxi- mation where 2 [V o ] = [C~e ], the normal diffusion currents jo and jo take such forms that the rate be- comes proportional to P~21/6. In the other Brouwer's approximation [Vo]=2[V~] , one obtains the expression of eq. (3) with: </p><p>j o = ~- 21/3K~/3 </p><p>and </p><p>2-1/6K~ l/6 K ~ jo /2p~l /4, </p><p>in which Do and De are the diffusion coefficients of oxygen and electrons respectively, l is the diffusion length and/3 and K are respectively the equilibrium constants for step (3) and for equilibrium between ceria and oxygen. Two limiting cases are thus pos- sible: either the rate does not depend on Po2, or it varies like P6~/4: </p><p>- For rate-limiting step ( 5 ), the theoretical rate is ! , m/3 when 2[V~;]= [Cece], it is a proportional to 1o2 </p><p>constant when [Vo] =2[V~'~ ] - for rate-limiting step (6), the rate is found to be </p><p>constant. Let us consider now the lanthanum-doped ceria </p><p>for which only the defects containing lanthanum ions are assumed to be involved in the charge balance equation, which corresponds to [ (La, Vo) ] = [La~e]. </p><p>This leads to a theoretical rate that does not de- pend on the lanthanum content. It is proportional to K3KZv if cerium diffusion (step (5)) is rate-limit- ing (KAv is the equilibrium constant for the associ- ation between La[.~ and Vo ); for step (2) as rate- limiting, one obtains the expression of eq. (3) with: </p><p>jo = _D~ KAV ~ </p><p>and </p><p>De I',/2KI/2D-I/4 jo= l , x A V ' 0 2 , </p><p>for which limiting cases correspond to Kxv ~ , or to the expression IUI/2Ic'I/2D- 1/4 </p><p>xx JxAV IO2 . </p><p>4.2. Comparison between theoretical and experimental rates </p><p>The experimental rate of undoped ceria varies as p~/6 which appears to be in good agreement with a rate-limiting step of ambipolar diffusion of oxygen vacancies and electrons, in Brouwer's case 2 [Vo] = ICe,e]. However this cannot be consistent with the physical reality since the lanthanum ions that are already present in the undoped ceria must not be neglected. Thus the experimental results obtained for the undoped sample have to be discussed on the ba- sis of lanthanum-doped ceria. By taking in consid- eration eq. (3) which gives the theoretical rate, it is possible to get a variation as P021/6 if the two dif- fusion currents jo and j0 are of the same amplitude. For partial pressures of oxygen equal to 0.13 and 13.33 kPa, and assuming that the ratio D r-- , /r~ l-.'1/2/g1/2 O''AV/*-'e . . . . AV is equal to 1.5, we obtain the results displayed in fig. 5, which indicate an excel- lent agreement between experimental and theoreti- cal rates. </p><p>In the case of lanthanum-doped ceria, the results about the low concentrations in lanthanum ions will first be discussed: the rate remains constant from 0.5 to 1%, whereas the oxygen partial pressure has no influence at all. This appears to be in agreement with the same rate-limiting step as above if the rate law proportional to KXv ~ is considered (jo </p></li><li><p>M. Pijolat et al. / Surface area stability of La-doped ceria 785 </p><p>Rate (nm.h-1) </p><p>20, ~ o experiment </p><p>10- J - model </p><p>! </p><p>0 0.2 0.4 </p><p>p O21/~ pa -1/6 ) </p><p>Fig. 5. Comparison between experimental and predicted rates of crystallite growth in undoped ceria. </p><p>process is modified as it becomes proportional to DoK~,v ~ . This inhibiting effect of lanthanum ions ad- dition may thus be understood by a decreasing value of Do due to stronger interactions. These would also induce a departure from the ideality of the solid so- lution. In consequence the activity coefficients would vary with lanthanum concentration, which could ex- plain the experimental results, due to the term KT~v 1 </p><p>When the lanthanum concentration is raised to 10%, the same interpretation still applies though it predicts a rate independent of this concentration, be- cause of the same reasons as above. However, it is difficult to quantify the variations of the theoretical rate. Since La 3+ ion is larger than Ce 3+ ion (0.12 nm and 0.10 nm respectively) [ 14] it is probable that diffusion processes will be strongly inhibited at very large concentrations in lanthanum. </p><p>This can be easily verified by doping ceria with y3+ ions which exhibit the same charge and the same ionic radius as cerium ions. The experimental results [9 ], which will be reported in the next article, show that this interpretation is valid since the rate takes nearly constant values between 0.5 and 10% of added y3+ ions. </p><p>5. Conc lus ions </p><p>The experimental rate of variation of ceria surface due to the crystallite size increase, depends on ox- ygen partial pressure as p~/6 . For pure ceria (with- </p><p>out lanthanum ions) the proposed model explains such a variation by means of ambipolar diffusion of oxygen and electrons as the rate-limiting step. </p><p>For lanthanum-doped ceria, the relevant point de- fects are La~+ and (La, Vo)' . The same rate-limiting step as above is in agreement with the experimental results: </p><p>- at low lanthanum concentrations (0.5-1%), the decrease in the rate can be interpreted by a decrease in the diffusion coefficient of oxygen as well as an effect of the activity coefficients of La~e, Vo" and (La, Vo)" species since the solid solution may no longer be regarded as ideal; </p><p>- at large lanthanum concentrations (2-10%) the lanthanum ions, larger than cerium ions, consider- ably slow down the diffusion process. </p><p>It has been shown that for lanthanum-doped ceria no effect of oxygen pressure has been revealed. This may be of significant importance for further appli- cations of lanthanum-stabilized ceria. However a large amount of lanthanum is necessary (about 5 to 10%) in order to achieve a significant thermal sta- bility. This could result in a disadvantageous phase segregation of La203 in the case of long-term thermal aging. </p><p>References </p><p>[ 1 ] M. Pijolat, M. Prin, M. Soustelle and O. Touret, J. Chim. Phys., submitted for publication. </p><p>[2] M. Pijolat, M. Prin, M. Soustelle and P...</p></li></ul>


View more >