Defect properties and ionic transport in layered compounds

HAL Id: jpa-00220003

https://hal.archives-ouvertes.fr/jpa-00220003

Submitted on 1 Jan 1980

HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished 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.

Defect properties and ionic transport in layered compounds

J. Walker, C. Catlow

To cite this version:

J. Walker, C. Catlow. Defect properties and ionic transport in layered compounds. Journal de

Physique Colloques, 1980, 41 (C6), pp.C6-32-C6-34. �10.1051/jphyscol:1980608�. �jpa-00220003�

JOURNAL DE PHYSIQUE

Colloque C6, suppliment au no 7, Tome 41, Juillet 1980, page C6-32

Defect properties and ionic transport in layered compounds

J. R. Walker and C. R. A. Catlow

Department of Chemistry, University College London, 20 Gordon Street, London WClH OAJ, U.K.

Rhumb. - Nous prksentons les rksultats des calculs de la structure et les mkcanismes de migration dans les composks superioniques P-A1,0, et Li3N.

Abstract. - We present the results of calculations of the structure and ionic conduction mechanisms in the super- ionic compounds P-A1203 and Li3N.

1.

The structure and properties of tion techniques and the PLUTO simulation package, layer-structure materials are of importance throughout so that the equilibrium condition, dielectric and elastic the field of solid state research. constants are adequately simulated. The parameters This is particularly true in the studies of solids which

derived in this study for these crystals are given in exhibit large ionic conductivities. There are two major table I.

problems which are fundamental to all the studies on

superionic conductors. The first problem is structural Table I.

Short range parameters.

and concerns the distribution and degree of clustering

of the conducting ions

the second is mechanistic and Interaction A i j (eV) Bij (A) Ci, (A6)

concerns the followed by the migrating ions.

We examine both problems in this paper with the aid of computer simulation techniques. Two important superionic conductors, namely Li3N and /?-A1203 are considered.

Our studies upon /?-A1,03 are mainly concerned with the structural properties while those on Li,N are concerned with structure and migration.

2.

Our studies employ the PLUTO-HADES suite of lattice simulation pro- grams whose methods and applicability are described elsewhere [I] and also during these conference pro- ceedings [2].

The essential feature of the simulations, which determines the reliability of the calculations, is the description of each ion as existing within a potential field which is decomposed into a long range Madelung part and a short range term

the latter is considered as the sum of Buckingham pair potentials

F j

Aij exp(rij/Bij)

Cij/r: Vions

where rij is the interionic distance.

A shell model [3] of ionic polarization is included in the simulation as this greatly reduces the problems of polarization catastrophy which follows from the use of the point polarizable ion model. For the further details of the interionic potential, the reader is referred to another paper at this conference [2].

The adjustable parameters in the simulation are varied, with the aid of non-empirical ionic interac-

(t) Denved from separate study on Na,O.

(*) Derived from separate study on AI,O,.

These derived parameters are then used with the benefit of the HADES I11 computer program to calculate the properties of the defective lattice. The defect is introduced into the lattice and the surround- ing ions are relaxed to zero force. This method must be employed as the crystal relaxation energ may be large compared with the unrelaxed defeci energy.

In order to calculate energies of activation, an ion is moved from its lattice site and the energy of the defect so produced is calculated as a function of the path of the ion. Even amongst superionic conductors the process of lattice relaxation is rapid compared with that of ionic motion and so the lattice relaxation must be taken into account.

In order to undertake the studies on /?-alumina a development of the PLUTO package was employed which enabled the site potentials, and hence the lattice energies, of crystals with very large unit cells to be calculated.

Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphyscol:1980608

DEFECT PROPERTIES AND IONIC TRANSPORT I N LAYERED COMPOUNDS C6-33

3. Lithium nitride. - The layer structured com- pound Li3N has a hexagonal crystal structitre with space group D&(P,/mmm)

the nitrogen ions occupy the 1 ~(000) sites and the lithium ions occupying the 1 b(00i) and 2 c(i%O) (330) sites, see figure 1.

Fig. I. - The structure of Li,N

The structure may, therefore, be considered as consisting of hexagonally-coordinated Li2N layers with additional lithium in the interlayer positions between the two nitrogen ions [4, 51.

An additional feature of the X-ray data concerning this compound is that the nitrogen appears to be in the previously unreported N3- ionic charge state.

This is of some importance to our simulation as we model the properties of the crystal with the ionic model. In addition we shall suggest below that this unusual charge state, which is stabilized by the close proximity of the interplanar lithium ions, may be responsible for the kinetic stability of this compound, in spite of a very low formation enthalpy.

The energies of formation of simple defects obtained by our simulations are given in table 11. It may be seen from this table that Frenkel pairs have by far the lowest formation energies. Indeed the Frenkel pair formation energy which involves the abstraction

Table

Li,N.

Li+ (plane) Frenkel pairs 0.1 eV

Li+ (c-axis) Frenkel pairs 2.2 eV

N3- Frenkel pairs 16.2 eV

Schottky (Lit from Li2N layer) 9.3 eV Schottky (Li

from Li+ layer) 15.9 eV

of lithium from the Li2N plane is low enough to imply that thermal generation of vacancies may be important even at room temperature

a significant result in view of the very high conductivity of the compound.

Furthermore, it may be seen that the removal of lithium from the interplanar gap is an energetically unfavourable process' as the two triply-negative nitrogen ions which were adjacent to the lithium are now repelled by their own ionic charge and also by the effective negative charge of the lithium vacancy.

It is this large potential barrier which must be over- come for nitrogen molecules to be formed and for the compound to decompose.

We find that by far the most favoured transport method in this material is that of vacancy migration both in the Li2N plane and perpendicular to it, which in both cases involves no interstitial sites. We find that the results of our simulations give activation energies of 0.1 and 0.5 eV respectively parallel and perpendicular to the plane

results which compare very satisfactorily with the experimental results of 0.290 and 0.490 eV respectively [6].

The presence of divalent substitutionals N 2 + , 0'-, NH2- in place of N3- would be to increase the Li vacancy population by way of charge compensation and hence increase the ionic conductivity. The lithium vacancy so produced would have a negative charge and thus may be attracted to the effective positive charge of the substitutional. Thus the increase in conductivity with substitutional population would be very dependent upon the degree of binding of the substitutional-vacancy pairs. Our simulations show

binding energy of approximately 0.9 and 1.1 eV (respectively for Li vacancies in the plane and at the bridging sites), for N2- substitutionals. In view of this we do not expect the presence of extrinsic defects to play an important role in the conduction mechanism as any lithium vacancies which are produced by this method are bound to the substitutionals which create them.

In summary, therefore, we conclude that

duction in lithium nitride takes place by vacancy migration, the vacancies being largely produced by thermal generation of Frenkel pairs. Both factors may be attributed to the unusual layer structure of this compound.

4.

B-alumina.

We present the preliminary results-

of calculations into the structure and conductivity

of sodium 8-alumina. Considerable research has

been undertaken into the properties of this compound

due to the low barrier to ionic conductivity and its

favourable thermal stability. The approximate struc-

ture of P-A120, is well known

it comprises layers

of spinel-structured alumina between which layers of

highly mobile sodium ions are sandwiched. The

interlayer gap between the spinel blocks is rather

large and the structural strength of the compound is

C6-34 J. R. WALKER A N D C. R. A. CATLOW

due to the oxygen ions which exist in the interlayer gap, and serve to fix the relative positions of adjacent blocks of alumina. The sodium ions, therefore, exist in a shallow potential well, and are thus able to move relatively easily in response to external influences.

When a defect is introduced into the conduction plane the sodium ions are able to relax distances which are large compared to those exhibited by other oxide materials. In this way the electrostatic repulsions which are associated with defect introduction may be reduced substantially in this compound. The structure of p-alumina therefore favours the production of defects in the conduction plane. Our simulations show that these arguments are equally valid for both intersti- tial sodium and oxygen; and we believe that the reason that sodium B-alumina is normally non-stoi- chiometric with a large excess of sodium in the conduc- tion plane is bec'ause the presence of these defects is stabilized as described above. The nature of the coun- terion to the excess sodium would appear, to be additional oxygen ions in the conduction plane

the formation of defects in the spinel blocks is energeti- cally unfavourable compared to the sodium conduc- tion plane. Furthermore, we find that the interstitial oxygen is present in the complex defect structure first proposed by Roth

In this structure the aluminium ions in the spinel block which are immediately above and below the interstitial oxygen move out of their spinel positions in order to become adjacent to the

oxygen ions, while still remaining outside of the conduction plane. In this way the overall energy of the defect cluster is reduced compared to that of the isolated oxygen ion. Our simulations show that the clustered defect is more stable by 1.0 eV compared with the isolated interstitial oxygen and thus it is to be expected that the vast majority of the interstitial 02- will exist in this clustered state.

Our results show that the difference in the poten- tials between the different sites in the conduction plane is small and corresponds to a range of 0.2 eV/

sodium ion. We predict that a t e.g. 500 OC that the majority of the sodium ions will be on the Beevers- Ross sites but with a significant occupancy of the other sites.

To summarize, our results first show that the unusually high degree of non-stoichiometry which this compound exhibits is due to the ease with which defects may be added to the conduction plane. Second- ly, we have confirmed that the counterion to the excess sodium would appear to be interstitial oxygen in the conduction plane with the structure originally suggested by Roth.

Finally we suggest that the majority of the sodium ions in the conduction plane are at the Beevers-Ross sites but with significant population of the other planar sites, the relative populations of the other sites being highly temperature dependent in view of the small difference in the site potentials.

References

[ I ] NOKGETT, M . J., AERE R7650, United Kingdom A ~ o r n ~ c [5] RABENAU, A. and SCHULZ, H., J. less-common Met. 50 (1976)

Energy Authority. 155.

[2] CATLOW, C. R. A,, These proceedings. [6] VON ALPEN, U. et al., Appl. Phys. Lett. 30 (1977) 621.

[3] DICK, B. G. and OVERHAUSER, A. W., Phys. Rev. 112 (1958) 90. [7] ROTH, W. L. et al., in Superionic Conductors (Plenum Press) [4] ZINTL, E. and BRAUER, G., 2. Elektrochem. 41 (1935) 102. 1976, p. 223.