J. Chem. SOC., Faraday Trans. 2, 1983, 79, 1195-1203

Reduced Forms of LaNi03 Perovskite Part 2.-X-ray Structure of LaNiOz and Extended X-ray Absorption Fine Structure

Study: Local Environment of Monovalent Nickel

BY PIERRE LEVITZ, MICHEL CRESPIN AND LUCIEN GATINEAU* Centre National de La Recherche Scientifique, Centre de Recherche sur les Solides a Organisation Cristalline Imparfaite, 1~ rue de la Fkrollerie, 45045 Orlkans, France? and L.U.R.E., Centre National de la Recherche Scientifique, Universitk Paris-Sud, Batiment 209,91405 Orsay, France.

Received 18th November, 1982

One of the products of the reduction of LaNiO, is LaNiO,, containing monovalent nickel. This valence state of nickel has scarcely been observed in mineral compounds. X-ray data from powder samples and extended X-ray absorption fine structure studies (EXAFS) allow one to define a tetragonal structure with lanthanum at the centre of a prism composed of oxygen atoms having a square base and nickel in the centre of the square. The Ni-0 distance (nearest neighbour) was found to be 1.983 A for this monovalent nickel.

~~ ~ ~ ~~

Reduction of LaNi03 perovskite in flowing hydrogen between 570 and 670 K yields a new phase, LaNiO2, which is stable in air. The reaction conditions and the valence state of nickel (+1) in this compound have been discussed in the pre- ceding paper.' It was shown that the powder diagram of LaNi02 can be indexed as a tetragonal lattice with the following parameters: a = b = 3.966 f 0.0@1 A, c = 3.376*0.001 A.

The present contribution aims to determine the structure of LaNi02 from the reflexion intensities measured on a powder diagram and from the interatomic distances found by EXAFS oscillations recorded in the domain of the K absorption edge of nickel.

CRYSTAL STRUCTURE BY X-RAY DIFFRACTION

EXPERIMENTAL CONDITIONS

The intensities of the 20 most intense reflexions of a powdered sample were measured by a diffractometer using filtered Cu K , radiation. The reflexions were recorded by step-counting and the area calculated after elimination of the back- ground (diffusion, fluorescence etc.).

The specific mass of LaNi02 was measured by a method derived from Archi- medes' law. The finely divided sample was degassed and impregnated afterwards by condensing a liquid from the gas phase. All air between the particles was eliminated by this method. The specific mass measured this way was 7.13 g ~ m - ~ , while the calculated specific mass was 7.18 g ~ m - ~ . From these results it can be inferred that there is only one formula per unit cell (2 = 1).

STRUCTURE

The observed structure factors were determined from the integrated intensities after correction for the Lorentz polarization factor and the multiplicity factor. The

1195

Publ

ishe

d on

01

Janu

ary

1983

. Dow

nloa

ded

by U

nive

rsity

of

Mic

higa

n L

ibra

ry o

n 24

/10/

2014

18:

21:2

3.

View Article Online / Journal Homepage / Table of Contents for this issue

1196 REDUCED FORMS OF LaNi03 PEROVSKITE

Table 1. Calculated and observed structure factors of LaNiO,

010 001 110 011 111 200 120 02 1 002 121 012 112 220 030 22 1 022 130 03 1 122 131

27.7 14.2 53.2 63.8 35.0 68.1 23.7 16.0 62.7 52.4 23.1 43.2 56.2 22.1 16.3 53.2 40.3 45.8 21.7 26.8

24.5 12.2 55.9 59.5 37.0 72.6 24.9 20.8 63.1 54.3 24.3 42.6 55.0 24.9 22.8 44.7 38.6 46.2 19.9 22.3

3.2 2.0 2.7 4.3 2.0 4.5 1.2 4.8 0.4 1.9 1.2 0.6 1.2 2.7 6.5 8.5 1.7 0.4 1.8 4.5

real and imaginary dispersion corrections for nickel and lanthanum have been accounted for in the calculated structure factors.

The four-fold symmetry axis suggests, in a projection on the 001 plane, that the nickel is located at the cell origin, the lanthanum in the centre of the face and the oxygen atoms in the middle of the square edges. The accuracy of this assumption can be verified for 7 hkO reflections only (table l), and for this condition the R factor has poor statistical significance. Nevertheless, one can attempt to resolve the three-dimensional structure by analogy with the LaNi03 perovskite structure, from which the reduced compound is derived, and by steric considerations. Then the simplest hypothesis (fig. 1) consists of lacing the nickel at the origin of the cell (0, 0, 0), the lanthanum at the centre ($,%,$) and the oxygen atoms in the middle of the basal square edges ($, 0,O and 0, $, 0). With these positions the residual factor calculated from the 20 recorded reflexions is

P

Attempts were made to improve this result, assuming that the oxygen atoms are displaced along the z axis, that the cell is doubled along this axis and that the main axis is a four-fold screw axis. Displacements of the oxygen atoms by as much as 0.2 A do not give any amelioration of R, and larger displacements increase it. The value of R (0.07) is relatively high in a structure where only 4 atomic positions must be determined. In the measured intensities, however, an important amount of reflexions are weak (see Part 1). Consequently the corresponding observed structure factors are relatively inaccurate.

Publ

ishe

d on

01

Janu

ary

1983

. Dow

nloa

ded

by U

nive

rsity

of

Mic

higa

n L

ibra

ry o

n 24

/10/

2014

18:

21:2

3.

View Article Online

P. LEVITZ, M. CRESPIN AND L. GATINEAU 1197

* I 2 0

La Ni

0

Fig. 1. Proposed structure of LaNi02.

INTERATOMIC DISTANCES AND COORDINATION

Eight oxygens are coordinated around the lanthanum atom, which is located in the centre of a square-based prism. Four oxygens are coordinated around the nickel atom and the nickel is located in the centre of a square. From the proposed structure, the bond lengths of La-0 and Ni-0 are 2.604 and 1.983 A, respec- tively.

The La-0 distance in the LaNi03 perovskite is 2.713 A. In this compound every lanthanum atom is surrounded by 12 oxygen atoms and the shorter distance observed in LaNiOz can be explained by a decrease in the coordination number.

The case of the Ni-0 distance is more complicated due to the unusual +1 oxidation state of the nickel and the square configuration of the oxygen atoms around it. In NiO oxide, the Ni-0 distance is 2.085 A and the coordination number is 6. The gain of one electron must increase the Ni-0 distance by ca. 0.1 A. According to Shannon2 the change of coordination number from 6 to 4 causes an average decrease of the interatomic distances of 0.15 A. Shannon reports a decrease of 0.20 A between the octa;hedral and square coordination of Ni2+. Consequently the interatomic distance Ni -0 can be expected to lie between 2.035 and 1.985 A:

2.085 A + 0.1 A- (0.15 or 0.20 A) = 2.035-1.985 A.

The Ni-0 distance of 1.983 A in the proposed structure is at the lower limit of this calculation, and the hypothetical displacement of oxygen atoms along the t axis does not give rise to a significant increase in the Ni-0 distance.

The interpretation of the X-ray data is unsatisfying. Relatively poor agreement is found between the calculated and observed structure factor, and the Ni-0 distance is at the limit obtained by a rough calculation. Therefore it seemed desirable to confirm these results by the use of EXAFS oscillations near the region of the K absorption edge of nickel to allow a determination of the interatomic distances around this element.

EXAFS

EXPERIMENTAL CONDITIONS

The absorption spectra were recorded at L.U.R.E. using the synchrotron radi- ation of the D.C.I. ring. The experimental device has been described in detail by Raoux et d3 For this work, the energy selection was carried out by means of a

Publ

ishe

d on

01

Janu

ary

1983

. Dow

nloa

ded

by U

nive

rsity

of

Mic

higa

n L

ibra

ry o

n 24

/10/

2014

18:

21:2

3.

View Article Online

1198 REDUCED FORMS OF LaNi03 PEROVSKITE

channel-cut monochromator using the 220 reflexion of silicon. The energy calibra- tion was based on the maximum of the K edge of metallic copper at 8991.2 eV. The EXAFS spectra were recorded between 8000 and 9500eV, increasing the energy step-wise by increments of 1.5 eV (Ni K edge = 8335 eV).

The samples were prepared by drying an LaNi02 suspension in a mixture of polystyrene and xylene on a sheet of mylar.

EXAFS ANALYSIS

The first stage in the interpretation of an EXAFS spectrum is to isolate the EXAFS modulation x ( k ) from the average variation of the absorption coefficient p ( E ) . The background removal is carried out by using polynomial functions that fit the variation of p ( E ) in the low-energy domain preceding the absorption edge. The same method is used to remove the smooth part of p ( E ) in the high-energy domain. This procedure has been described in detail by Lee et aL4

A first Fourier transform of x ( k ) yields a pseudo-radial distribution in real space around the nickel atom. The position of every peak of this function is systematically shifted to a value smaller than the radius of the associated coordina- tion shell. This effect can be explained by the general form of the electronic phase shift, which is not a constant function of k.435

For each peak of the pseudo-radial distribution a second Fourier transform allows the isolation of the partial EXAFS oscillations of the associated coordination shell. The zero values of these functions lead to the calculation of the characteristic distance for each shell.

For each shell the nature of the surrounding atom must be defined and a function 4 ( k ) must be chosen for the electronic phase shift, such as

2k,R +4(kn)=2.rr(n0+n) (1) where R is the unknown radius of the shell concerned and the k , are zero values of the partial EXAFS oscillations.

Two procedures can be, proposed to choose 4 ( k ) . The first consists of using the theoretical functions as defined by Teo and Lee;6 the second, which is referred to as the chemical-transferability technique, consists of calculating 4 ( k ) for an identical pair of atoms in a well known structure.

In eqn (1) no is an undefined value which cannot be derived from experiment. Therefore R must be calculated by using an algorithm of modulo 27r. The resolution of eqn (1) is obtained by adjusting the threshold energy Eo in such a way that in the final step of the calculation, the function R ( k ) will be as constant as possible. The successive steps of this mathematical treatment of the EXAFS oscillations have been given by Teo and Lee4 and Fontaine et a1.’

RESULTS

OXIDATION STATE OF NICKEL IN LaNi02

The radial distribution in nickel metal (fig. 2) is characterized by a strong peak at an ‘apparent’ distance of 2.15 A for the first coordination shell. For nickel in LaNi02 [fig. 3(b)] this region is characterized by a minimum. This simple observa- tion confirms the conclusion of the preceding paper’ concerning the absence of metallic nickel at this stage of the reduction. Indirectly this confirms the existence of monovalent Ni in LaNi02.

Publ

ishe

d on

01

Janu

ary

1983

. Dow

nloa

ded

by U

nive

rsity

of

Mic

higa

n L

ibra

ry o

n 24

/10/

2014

18:

21:2

3.

View Article Online

P. LEVITZ, M. CRESPIN AND L. GATINEAU

1

2

1199

0 2 L 6 8 10 RIA

Fig. 2. Fourier-transform magnitude, 9, of k-weighted EXAFS for a nickel metal foil.

1.5

1 .o

0.5

- 57 55 y o

Fig. 3. Fourier-transform magnitude, 9, of k-weighted EXAFS for ( a ) LaNiO, and (6) LaNiOz: (i) first coordination shell, Ni -0; (ii) second coordination shell, Ni-La; (iii) third

coordination shell, Ni -0 -Ni.

Publ

ishe

d on

01

Janu

ary

1983

. Dow

nloa

ded

by U

nive

rsity

of

Mic

higa

n L

ibra

ry o

n 24

/10/

2014

18:

21:2

3.

View Article Online

1200 REDUCED FORMS OF LaNi03 PEROVSKITE

INDEXATION OF THE RADIAL DISTRIBUTION

The radial distributions around the nickel atoms in LaNi03 and LaNiO2 as obtained by EXAFS oscillations are shown in fig. 3. Comparing the general profile of both distributions, a striking resemblance is observed for the first three coordina- tion spheres.

In LaNi03 [fig. 3 ( a ) ] the first peak is due to the nearest neighbours, consisting of 6 oxygen atoms at a distance of 1.918 A. The second coordination shell around nickel is composed of 8 Ni-La pairs at an average distance of 3.323A. The presence of two peaks [fig. 3(a ) ] corresponding to the La-Ni distance is due to the k-dependent structure of the backscattering amplitude of La. The distaqces between the two maxima in the backscattering amplitude of La is ca. 6 A- as calculated by Teo and Lee.6 This value permits one to calculate the splitting, AR, of the radial distribution in R space:

AR =0.5 A. This value corresponds well with the experimental value shown in fig. 3(a ) .

The fourth peak is due to a colinear Ni-Ni pair separated by an oxygen atom. This can be visualized by the notation Ni-0-Ni. The corresponding coordination shell is composed of 6 atoms at a distance of 3.837 A.

The radial distributions of LaNi03 and LaNiOz being very similar (fig. 3) , the peaks of LaNi02 can be indexed using the data obtained for LaNiO3. The succession of coordination shells observed in this way corresponds well with the structure described above (fig. 1). Nevertheless, this indexation does not take into account the shell corresponding to Ni-Ni distances along the z axis (3.376 A). The corres- ponding peak and the double Ni-La peak are certainly superimposed, but the contribution of the Ni-Ni shell is very weak. Indeed there are only 2 nickel atoms in this shell, while there are 8 in the Ni-La shell. Moreover, the backscattering of a heavy atom like lanthanum is much more important than the backscattering of nickel.

CALCULATION OF THE DISTANCE BETWEEN NICKEL AND NEIGHBOURING ATOMS IN LaNi02

FIRST COORDINATION SHELL: Ni-0 The theoretical relations found by Teo and Lee6 were applied to calculytion of

the phase shifts. The window in k space used covers the range 7-15A- . This window was used for the Fourier transform in order to eliminate inelastic processes, atomic relaxation and the dynamic screening.8 The function x ( k ) is multiplied bx k 3 before carrying out the Fourier transform, as was suggested by Teo and Lee for light-backscattering atoms. This procedure was verified by calculating the first shell of nickel in two well known structures, LaNi03 and NiO. The results of these calculations are given in table 2. They are in good agreement with the crystallo- graphic data. The Ni'-0 distance in LaNi02 calculated by this method is 1.985 A. SECOND COORDINATION SHELL: Ni-La

As in the case of the first coordination shell, the phase shift calculated by Teo and Lee6 was used. The various parameters used to treat the experimental values are given in table 3. As for the first coordination shell, this procedure has been

Publ

ishe

d on

01

Janu

ary

1983

. Dow

nloa

ded

by U

nive

rsity

of

Mic

higa

n L

ibra

ry o

n 24

/10/

2014

18:

21:2

3.

View Article Online

P. LEVITZ, M. CRESPIN AND L. GATINEAU 1201

Table 2. Ni-0 distances from EXAFS"

R / A

compound X-ray EXAFS

LaNiO, 1.918 1.94 NiO 2.092 2.085 LaNiOz 1.983 1.985

" Window of Fourier transform in k space is 7-15 A; multiplication factor of x ( k ) is /c3.

Table 3. Ni-La distances from EXAFS"

R I A

compound X-ray EXAFS

} 3.35 LaNiO,

LaNiO, 3.270 3.24

6 atoms at R = 3.335 2 atoms at R = 3.284

a Window of Fourier transform in k space is 4-15 A; multiplication factor of x ( k ) is k .

verified for LaNi03. Table 3 shows that Ni-La distance calculated from the EXAFS spectra is slightly larger than the one deduced from structural calculations.' This phenomenon is probably due to a poor definition of the phase shift. Noguera' has clearly shown that, as the interatomic distance increases, the phase shifts calculated by Teo and Lee become more inaccurate. Moreover, the value of 3.24 A found for the Ni-La distance in LaNiOz is affected by a weak contribution of the Ni-Ni distance along the z axis, which could not be eliminated from our calculations.

THIRD COORDINATION SHELL: Ni-0-Ni In this coordination shell the electron path is perturbed by the oxygen atom.

A theoretical electric phase shift is not available to account for this effect in the case of a relatively distant coordination sphere (>3 A). By chemical transferability it is possible to calculate an empirical phase from the Ni-0-Ni coordination in NiO (4.183 A). The radial distribution of the EXAFS oscillations of NiO (fig. 4) allows an isolation of the corresponding peak. The distance in LaNiOz calculated this way for a window in k space between 7 and 15 A-1 is 3.98 A.

DISCUSSION OF THE EXAFS RESULTS

The EXAFS method is a relatively recent one and the accuracy of these results (the interatomic distances around a given element) can be subject to dispute. The methods used in this paper permit one to verify the unknown distance by comparison with a well known structure where the nickel atoms are in the same environment. Under these conditions the accuracy can be estimated as k0.02 A.

Publ

ishe

d on

01

Janu

ary

1983

. Dow

nloa

ded

by U

nive

rsity

of

Mic

higa

n L

ibra

ry o

n 24

/10/

2014

18:

21:2

3.

View Article Online

1202 REDUCED FORMS OF LaNi03 PEROVSKITE

2

0 2 4 6 8 1 0 1 2

R I A Fig. 4. Fourier-transform magnitude, 9, of k-weighted EXAFS of NiO, coordination shell: ( a ) Ni-0, (6) Ni-Ni, ( c ) Ni-0, ( d ) Ni-0-Ni, ( e ) Ni-0, ( f ) Ni-Ni and ( R ) Ni-Ni.

Table 4. Comparison of X-ray and EXAFS data for LaNiO,

results of X-ray results of EXAFS separation structure/A determination/ A

Ni -0 1.983 Ni-La 3.270

Ni-0-Ni 3.966

1.985 3.24, 3.975

Knowing the unit-cell parameters of LaNi02 (a = 3.966 A and c = 3.376 A) the Ni-La distance, 3.24 A, must be half the unit-cell diagonal and the Ni-0 distance, 1.98 A, corresponds to half the basal square edge. Considering these two results, a structure can be proposed where Ni is located at the corners of the unit cell, La in the centre and the oxygen atoms in the middle of the basal edges. The method of calculating the Ni-0-Ni distance using chemical transferability from NiO confirms the position of the oxygen atom in a line between two Ni atoms or at least close to this position.

CONCLUSION

Table 4 summarizes the results obtained by the two different methods used to elucidate the LaNiOz structure. The localization of the Ni and La atoms is unam- biguous, but the accuracy of the two methods cannot determine exactly the position of the oxygen atoms. The uncertainty in the positions depicted in fig. 1 does not exceed 0.2 A along the z axis. Therefore the proposed pattern must be considered as an idealized model, and the real configuration of the structure can only be determined when data from LaNiOz single crystals become available. Nevertheless, using the idealized model, the linear configuration of the Ni atoms along the z axis allows one to predict interesting magnetic properties for this new compound.

Publ

ishe

d on

01

Janu

ary

1983

. Dow

nloa

ded

by U

nive

rsity

of

Mic

higa

n L

ibra

ry o

n 24

/10/

2014

18:

21:2

3.

View Article Online

P. LEVITZ, M. CRESPIN AND L. GATINEAU 1203

We are grateful for fruitful discussions and helpful assistance from Prof. J. Petiau, Prof. J. Calas and Dr H. Nijs. We thank the scientific staff at L.U.R.E. (Orsay) for cooperation and assistance.

M. Crespin, P. Levitz and L. Gatineau, J. Chem. SOC., Faraday Trans. 2, 1983,79,1181.

D. Raoux, J. Petiau, P Bondot, G. Calas, A. Fontaine, P. Lagarde, P. Levitz, G. Loupias and A. Sadoc, Rev. Phys. App l . , 1980,15, 1079. P. A. Lee, P. H. Citrin, P. Eisenberger and B. M. Kincaid, Rev. Mod. Phys., 1981, 53, 763. S. H. Hunter, Ph.D. Thesis (Stanford University, California, 1977). B. K. Teo and P. A. Lee, J. Am. Chem. SOC., 1979,101, 2815.

40, 17. C . Noguera, Thkse d’Etat (UniversitC Paris-Sud, Orsay, 1981).

’ S. D. Shannon, Acta Crystallogr., Sect. A, 1976,32, 751.

’ A. Fontaine, P. Lagarde, A. Naudon, D. Raoux and D. Spanjaard, Philos. Mug., Sect. B, 1979,

(PAPER 2/1940)

Publ

ishe

d on

01

Janu

ary

1983

. Dow

nloa

ded

by U

nive

rsity

of

Mic

higa

n L

ibra

ry o

n 24

/10/

2014

18:

21:2

3.

View Article Online