PHYSICO CHEMICAL PROPERTIES OF NICKEL OXIDE AT HIGH TEMPERATURE

Transcription

1 PHYSICO CHEMICAL PROPERTIES OF NICKEL OXIDE AT HIGH TEMPERATURE R. Farhi, G. Petot-Ervas To cite this version: R. Farhi, G. Petot-Ervas. PHYSICO CHEMICAL PROPERTIES OF NICKEL OXIDE AT HIGH TEMPERATURE. Journal de Physique Colloques, 1976, 37 (C7), pp.c7-438-c < /jphyscol: >. <jpa > HAL Id: jpa Submitted on 1 Jan 1976 HAL is a multi-disciplinary open access archive for the deposit and dissemination of scientific research documents, whether they are published or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers. L archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.

2 C7-438 JOURNAL DE PHYSIQUE Colloque C7, supplkment au no 12, Tome 37, Dkcembre 1976 PHYSIC0 CHEMICAL PROPERTIES OF NICKEL OXIDE AT HIGH TEMPERATURE R. FARHI and G. PETOT-ERVAS C. N. R. S., Laboratoire PMTM, Avenue J. B. ClBment, Villetaneuse, France Rhum6. - Une etude de I'kart A la stoechiometrie dans I'oxyde de nickel pulvkrulent a Ctk effectuk par titrage coulomktrique, dans une gamme de pressions d'oxygkne s'ktendant de 0,012 & 0,21 atrn et aux deux temperatures de 8810 et 896 OC. Les rbultats semblent montrer l'existence, dans le domaine Btudie, de lacunes de Nickel ionisks une fois comrne defaut prkdominant. Le coefficient de diffusion chimique dans l'oxyde de nickel monocristallin a 6t6 determine par la m6thode de la conductivitk klectrique entre et OC, et pour des pressions d'oxygkne sy6tendant depo, = 10-4 atrn &pol = 1 atm. De plus, le comportement anormal de la conductivite klectrique a kt6 explique par la presence simultank, dans le domaine Btudie, de lacunes de Nickel une fois et deux fois ioniskes. Un modble thermodynamique est propose sur cette base. Abstract. - A study of thedeviation from stoichiometry by coulometric titration has been performed on nickel oxide powder, from po, = atrn to po2 = 0.21 atm, at 881 and 896 OC. The results are in agreement with the existence of singly ionized nickel vacancy as the predominant defect in the investigated range. Chemical diffusion coefficient in nickel oxide single crystals has been determined by the electrical conductivity technique, between and OC, from pol = 10-4 atrn topo, = 1 atm. Moreover, the anomalous behaviour of the electrical conductivity has been explained by the simultaneous presence of singly and doubly ionized nickel vacancies in the investigated range. A thermodynamic model is proposed on this basis. A study of point defects in nickel oxide has been made using two different techniques. First, coulometric titration has given us results concerning the variations of the deviation from stoichiometry at moderately high temperatures. On the other hand, electrical conductivity measurements have been performed in order to obtain chemical diffusion coefficients and conductivity values at high temperatures. Using these later results, a point defect model has been built which gives informations about the simultaneous presence of singly and doubly ionized nickel vacancies. 1. Coulometric titration. - The oxygen pressure in equilibrium with non stoichiometric nickel oxide has been measured by means of electrochemical cells of the type : Pt, ~o,,,olzr02, 15 % CaOl~o Z,, Pt. The crucible containing the nickel oxide powder is separated from the reference gas by a tight pyrex seal, as shown on figure 1. When the thermodynamic equilibrium is reached, FIG Electrochemical cell for the coulometric titrations : the measured electromotive force is expressed by the 1) reference gas ; 2) Pt/~t-Rh thermocouple ; 3) alumina rod ; Nernst relation : 4) silica container ; 5) silica tube ; 6) painted Pt electrodes ; 7) stabilized zirconia electrolyte ; 8) quartz crucible ; 9) pyrex (1) seal ; 10) Pt electrical junctions ; 11) alumina radiation shield ; 12) oxide specimen ; 13) alumina crucible. Article published online by EDP Sciences and available at

3 NICKEL OXIDE AT HIGH TEMPERATURE C7-439 The imposed variations of the sample stoichiometry are controled by coulometric titration. Then, the variation Ay of the stoichiometry can be expressed as follows : 2. Electrical conductivity measurements. - The used technique is a four-probes classical one. Figures 3 and 4 respectively show the sample device and the whole assembly. The total amount of impurities in the used single crystals has been found to be less than 20 and 45 ppm respectively before and after the runs. with : M : molecular weight of the oxide, m : sample's weight, i : current passing through the cell during the time t, F : Faraday's constant, : number of moles of oxygen in the gaseous phase above the sample. n : depends on the ionization degree of nickel vacancies. Measurements have been performed at 8810 and 896 OC, in the oxygen pressure range from to 0.21 atm. The results are reported on figure 2. It can be seen that the value n = 4 fits the results better than the value n = 6. Thus, it can be assumed that, in the investigated range, singly ionized nickel vacancy is the predominant defect. FIG Conductivity sample : two arrangements have been used, but we have verified that the case (a) is the better, because equipotentials are always perpendicular to the sample axis. This is not the case for (b). (The hatched areas are'pt coatings.) FIG Variations of the deviation from stoichiometry versus -pacrel,. (relation (2)) for n = 4 and n = 6 and for the two ~nvestigated temperatures, 881 and 896 oc. On the other hand, the proportionality constant k* has been calculated from the slopes of the straight lines obtained (see Fig. 2), and consequently, the departure from stoichiometry has been found to be : FIG Conductivity cell assembly : (a) feedthrough ; (b) gas inlet and outlet ; (c) Pt leads ; (d) alumina tube ; (e) water refrigeration ; Cf)furnace ; (g) alumina radiation shields ; (h) alumina sheath ; (i) Pt, Pt-Rh thermocouple ; (j) NiO single crystal.

4 C7-440 R. FARHI AND G. PETOT-ERVAS 2.1 DIFFUSION RESULTS. - In a first step, chemical diffusion coefficients in nickel oxide between and 1400 OC have been determined by investigating the conductivity variations of the sample as a function of time, when the equilibrium conditions of the oxide are changed. The diffusion equations for a brick-shaped sample are given by : 0- - 'Jm 3 m C exp-(2n+l)'x 'Jo - 'Jm n=o (2 n x-dt 7c2 - m 1 7c2 - exp -(2m + I)~-- Dt n2 m=o (2 m + 1)' L~ 1 7c2 5 exp -(2 p + 1)'- Dt p=o (2 p + 1)2 l2 (4) where H, L and I are the dimensions of the sample, 5 is the chemical diffusion coefficient, t is the time, oo is the conductivity at time t = 0 (when the equilibrium conditions are changed), o, is the conductivity at time t = c~ (when the new equilibrium state is reached). For a sufficiently large time, the following approximation is made : L) 5t]. 'J - = (5) [- "2 (H"P 12 'Jo - 'Jm (5) Thus, the chemical diffusion coefficient can be directly determined by plotting the logarithm of the conductivity variation versus time. Chemical diffusion coefficients are shown on figure 5, as a function of 1/T. The activation energy for the vacancy migration has been found to be AH,,, cal.mo1-l. of the oxygen pressure are reported on figure 6. The value of n, defined as Log Poz(atm) varies with oxygen pressure and temperature, and we shall explain this phenomenon later. The same behaviour is observed for the conductivity activation energies, as can be seen on figure 7 : the activation energies are function of temperature and oxygen pressure, and the variations are important mainly at low oxygen pressures. 2.2 CONDUCTIVITY RESULTS. - Measurements have been performed between and O C from to 1 atm of oxygen. The results as a function These considerations can be explained by assuming the simultaneous presence of singly and doubly ionized nickel vacancies. The classical equations for non stoichiometric oxides where predominant defects are metal vacancies are given by : 112 O2 =$ VAi + O0 -t h. KV, AH: (6) Vki + V;Gi + h* K,, AH: (7)

5 NICKEL OXIDE AT HIGH TEMPERATURE C7-441 K,, K, are the equilibrium constants for the formation of singly and doubly ionized vacancies, respectively, and AH; and AH: are the corresponding enthalpies of formation. The electroneutrality condition must take into account the two types of vacancies : and T (relations (10) and (ll)), it can be seen that E, (and consequently AH,) varies with po2 and T. The same calculus can be made for n. The following expressions are obtained : By care of simplicity, we have introduced a new variable, k, which is the ratio of the concentrations of the two types of vacancies : It can be demonstrated that the derivatives of k with respect to po2 and T are given by the following expressions : -- an - - a In Po, k (1 + 2 k) (1 + 3 k) ' (18) It may be remarked that, in relation (16), the limiting conditions are respected (n = 6 when k + oo and n = 4 when k = 0). These two cases correspond to the predominance of doubly and singly ionized nickel vacancy, respectively. If we define E, by : E a In o =-= 8-1 a - T - R AHo, (1 2) R : gas constant, AHo : activation enthalpy for conductivity. We can calculate as functions of k, provided that the mobility of holes does not vary with p,,. The following expressions are derived : Log Po,(atm) FIG Comparison of the experimental values with the analytical representation of log o vs log~~, : experimental points ; - analytical representation. The thinner lines represent the conductivity behaviour if n were constant and equal to n(0) (i. e, the value of n for po2 = 1 atm). The value n(0) has been experimentaly determined by small step measurements performed in the rangepo, = 10-1 to 1 atm. A H : - ~ A H, O k(2k+1) a In Po, 2R 4(1 + 3 k)3 ae, = - (15) AH: is the activation enthalpy for the mobility of holes (I). Taking into account the variations of k with po2 Log PO, (atm) (1) In this calculus, the conduction mechanism has been supposed to be of the small polaron hopping type, with a mobility FIG k values and relative concentrations of VLi as funcgiven by : p = p,~ exp (-AH?/RT). tions of Po, and 7.

6 C7-442 R. FARHI AND G. PETOT-ERVAS By integrating relation (18) with respect to In p,,, diagram gives the proportion of one type of defect we obtained the analytical representation of log o as a (for example Vhi) with respect to the other. It can be function of logp,, for a given temperature. We have seen on this figure that, at low temperatures and relaticompared in figure 8 the experimental values with this vely high oxygen pressures, singly ionized nickel analytical representation. vacancy is the predominant defect. This qualitative In conclusion to this work, we have drawn in result kin good agreement with those of coulometric figure 9 the iso-k lines as a function of p,, and T. This titration. DISCUSSION C. R. A. CATLOW. - I am doubtful as to the adequacy of the very simple models you have used to describe defect clustering in non-stoichiometric NiO. Our calculations suggest that clustering is far more extensive in such oxides (a result supported by e. g. the experimental diffraction data on Fe,-, 0). Moreover we find that the binding energy of the larger vacancy aggregates is so much greater than that for the simple clusters that the former should not be significant over any concentration range. The good fit you obtained of your data to simple cluster models does not, I believe, prove the existence of such models ; in your analysis the simple clustering could simulate the effect of the actual defect interactions which are, I believe, far more complex. R. FARHI and G. PETOT-ERVAS. - Nickel oxide presents very small deviations from stoichiometry and, contrary to Fe,-, 0 you have mentioned, there is no experimental evidence for vacancy clustering. For this reason, we have neglected in our calculations the possibility of existence of vacancy clusters or aggregates. Our model only deals with non-associated vacancies, but with two different ionization states. However, if, in nickel oxide, vacancy complexes can be neglected, this is not the case for other transition metal oxides (like Fe,-, 0) where deviations from stoichiometry are far more important. But our model does not describe defect clustering, and consequently it cannot be applied to such oxides.