Sodium hydride as a powerful reducing agent for topotactic oxide deintercalation: Synthesis and characterization of the nickel(I) oxide LaNiO2

Article abstract of DOI:10.1021/ja991573i

The capability of sodium hydride as a reducing agent in oxide deintercalation reactions is explored. The Ni(III) perovskite LaNiO₃ can be reduced topotactically to LaNiO₂, isostructural with the 'infinite layer' cuprates, using solid sodium hydride in a sealed evacuated tube at 190 ≤ T/°C ≤ 210, and a similar infinite-layer phase is prepared by reduction of NdNiO₃. Structural characterization indicates the coexistence of incompletely reduced regions, with five-coordinate Ni centers due to the introduction of oxide anions between the NiO₂^3- sheets, giving samples with a refined stoichiometry of LaNiO(2.025(3)). Neutron powder diffraction and magnetization measurements indicate that the lamellar Ni(I) phase does not show the long-range antiferromagnetic ordering characteristic of isoelectronic Cu(II) oxides. This may be due either to the influence of the interlamellar oxide defect regions or to the reduced covalent mixing of Ni 3d and O 2p levels.

Full text of DOI:10.1021/ja991573i

J. Am. Chem. Soc. 1999, 121, 8843-8854
8843
Sodium Hydride as a Powerful Reducing Agent for Topotactic Oxide
Deintercalation: Synthesis and Characterization of the Nickel(I)
Oxide LaNiO2
M. A. Hayward, M. A. Green, M. J. Rosseinsky,* and J. Sloan
Contribution from the Inorganic Chemistry Laboratory, Department of Chemistry, UniVersity of Oxford,
South Parks Road, Oxford OX1 3QR, U.K.
ReceiVed May 11, 1999
Abstract: The capability of sodium hydride as a reducing agent in oxide deintercalation reactions is explored.
The Ni(III) perovskite LaNiO3 can be reduced topotactically to LaNiO2, isostructural with the “infinite layer”
cuprates, using solid sodium hydride in a sealed evacuated tube at 190 e T/°C e 210, and a similar infinite-
layer phase is prepared by reduction of NdNiO3. Structural characterization indicates the coexistence of
incompletely reduced regions, with five-coordinate Ni centers due to the introduction of oxide anions between
the NiO23 sheets, giving samples with a refined stoichiometry of LaNiO2.025(3). Neutron powder diffraction
and magnetization measurements indicate that the lamellar Ni(I) phase does not show the long-range
antiferromagnetic ordering characteristic of isoelectronic Cu(II) oxides. This may be due either to the influence
of the interlamellar oxide defect regions or to the reduced covalent mixing of Ni 3d and O 2p levels.
-
Introduction
In our present study, we have employed solid sodium hydride
NaH) as a novel reducing agent in the topotactic reduction of
the Ni(III) perovskite oxide LaNiO3 to the highly metastable
Ni(I) phase LaNiO2 and similar reduction of NdNiO3 to form
an analogous infinite-layer phase. This reaction is chosen as a
suitable one to demonstrate the scope of NaH in this type of
(
Low-temperature synthesis is a developing area of solid-state
chemistry, which allows access to metastable phases with novel
electronic and chemical features. Oxygen deintercalation from
1
complex oxides is potentially an important aspect of this field,
due to the widespread application of such phases arising from
their electronic, magnetic, and catalytic properties. A variety
of reduced oxides, particularly topotactically formed oxygen-
deficient derivatives of the perovskite structure, have been
reduction chemistry because the Ni(I) phase has not been
confirmed since the original report by Crespin et al.,1
0,11
and
other workers have made statements casting doubt on its
1
2
existence. The importance of this phase lies in the relation of
2
-8
reported.
The reagents employed to carry out these trans-
9
its electronic properties to those of the isoelectronic d Cu(II)
formations have largely been hydrogen gas or finely divided
electropositive metals, such as zirconium or aluminum, used
parents of the cuprate superconductors, and the synthesis
reported here allows us to determine the electronic properties
of this phase.
8
,9
as getters. The reported reaction temperatures for these
reductions lie above 300 °C, so there is clearly a need to develop
reagents which are effective reducing agents at lower temper-
atures. This may then give access to metastable reduced oxides
inaccessible, for thermodynamic reasons, at the higher temper-
atures required for hydrogen or getter reductions to proceed at
acceptable rates. For example, the reduction of LaNiO3 to La2-
Ni2O5 with 5% H2 diluted with N2 requires a temperature of
The use of sodium hydride as a solid-state reductant follows
1
3
the synthesis of both intermetallic hydrides and amorphous
precursors to ternary oxides with metal borohydrides in
1
4-20
solution.
Metal hydrides are used extensively as drying and
21
reducing agents in solution chemistry, but their reducing power
has been underexploited to this point in solid-state reactions.
(
10) Crespin, M.; Levitz, P.; Gatineau, L. J. Chem. Soc., Faraday Trans.
1983, 79, 1181.
11) Crespin, M.; Levitz, P.; Gatineau, L. J. Chem. Soc., Faraday Trans.
2 1983, 79, 1195.
12) Martinez-Lope, M. J.; Casais, M. T.; Alonso, J. A. J. Alloys Compd.
3
90 °C, and the preparation of La2Ni2O5 by the reduction of
2
LaNiO3 with a zirconium getter requires a temperature of at
(
4
least 400 °C.
(
(
(
1) Stein, A.; Keller, S. W.; Mallouk, T. E. Science 1993, 259, 1558.
2) Poeppelmeier, K. R.; Leonowicz, M. E.; Longo, J. M. J. Solid State
1998, 277, 109.
(13) Murphy, D. W.; Zahurak, S. M.; Vyas, B.; Thomas, M.; Badding,
Chem. 1982, 45, 71.
3) Millange, F.; Caignaert, V.; Domenges, B.; Raveau, B.; Suard, E.
Chem. Mater. 1998, 10, 1974.
4) Alonso, J. A.; Martinez-Lope, M. J. J. Chem. Soc., Dalton Trans.
M. E.; Fang, W. C. Chem. Mater. 1993, 5, 767.
(14) Manthiram, A.; Dananjay, A.; Zhu, Y. T. Chem. Mater. 1994, 6,
(
1601-1602.
(
(15) Manthiram, A.; Tsang, C. J. Electrochem. Soc. 1996, 143, L143-
1
995, 17, 2819-2824.
L145.
(
5) James, M.; Attfield, J. P. Physica C 1994, 235, 751.
(16) Tsang, C.; Kim, J.; Manthiram, A. J. Solid State Chem. 1998, 137,
(
6) Cava, R. J.; Takagi, H.; Krajewski, J. J.; Peck, W. F.; Hwang, H. Y.
28-32.
Phys. ReV. B 1993, 47, 11525.
7) Hansteen, O. H.; Fjellvag, H.; Hauback, B. C. J. Mater. Chem. 1998,
, 2089-2093.
8) Hansteen, O. H.; Fjellvag, H.; Hauback, B. C. J. Mater. Chem. 1998,
, 2081.
9) Moriga, T.; Usaka, O.; Nakabayashi, I.; Kinouchi, T.; Kikkawa, S.;
Kanamaru, F. Solid State Ionics 1995, 79, 252.
(17) Tsang, C.; Dananjay, A.; Kim, J.; Manthiram, A. Inorg. Chem. 1996,
(
35, 504-509.
8
8
(18) Tsang, C. F.; Manthiram, A. J. Mater. Chem. 1997, 7, 1003-1006.
(
(19) Tsang, C.; Lai, S. Y.; Manthiram, A. Inorg. Chem. 1997, 36, 2206-
2210.
(
(20) Zhu, Y. T.; Manthiram, A. J. Solid State Chem. 1994, 110, 187-
189.
1
0.1021/ja991573i CCC: $18.00 © 1999 American Chemical Society
Published on Web 09/11/1999

8844 J. Am. Chem. Soc., Vol. 121, No. 38, 1999
Hayward et al.
Table 1. Products of the Reaction between LaNiO
3
and NaH As Observed by Powder X-ray Diffractiona
reaction
conditions
temp/°C
reaction time
reaction products
ground
ground
ground
T < 120
120 < T < 180
180 < T < 210
2 × 1 week
4 × 3 days
3 × 3 days
no reaction
reduction to La Ni O
2 2 5
LaNiO , tetragonal phase (a ) 3.958 Å, c ) 381 Å) with anisotropically broad
2
diffraction reflections
ground
ground
separate
separate
separate
annealed
annealed
210 < T < 275
275 < T
2 × 3 days
1 × 3 days
2 × 3 days
1 × 3 days
1 × 3 days
2 × 1 days
2 × 3 days
reduction/disproportionation to La
reduction/disproportionation to La
no reaction
reduction to La Ni O
2 2 5
reduction/disproportionation to La
no change to diffraction pattern
2
2
Ni
NiO
2
O
4
5
+ Ni + La
+ Ni
2
O
3
T < 220
220 < T < 340
340 < T
T < 210
T < 210
2
Ni
2
O
5
+ Ni + La
2
O
3
decomposition/disproportionation to La
2
Ni
2
O
5
2 3
+ Ni + La O
a
The outcome of the reactions with both ground and physically separated reagents are given, and the outcome of annealing the LaNiO2+x
product in the absence of reducing agent is also presented.
-
The reduction potential of the H /H2 couple has been estimated
product was finely ground and then treated in one of two ways.
2
2
“Ambient pressure” LaNiO was synthesized by pressing the decom-
as -2.25 V, making the hydride anion one of the most
powerful reducing agents known. This thermodynamic driving
force has allowed us to perform complex metal oxide reduction
at lower temperatures than would be required for hydrogen gas
or metal “getter” processes, enabling us to isolate the Ni phase,
LaNiO2, which has been reported to be unattainable by these
3
position product into 13 mm diameter pellets at 5 tons of pressure and
then firing at 850 °C under 1 atm of flowing oxygen for two periods
of 30 h with one intermediate regrinding. “High pressure” LaNiO was
3
synthesized by pressing the decomposition product into 7 mm pellets,
which were then wrapped in gold foil and heated for two periods of
I
2
4 h at 1000 °C under a static pressure of 200 atm of oxygen in a
two methods.9
,12
4
manner similar to that described by Alonso et al. The high-pressure
apparatus used in this synthesis employs a compressor to pressurize a
sealed vessel, made from Rene-41 alloy, which is then heated to the
reaction temperature by an external furnace. The products of both
synthetic methods were analyzed by powder X-ray diffraction and found
LaNiO2 was originally reported by Crespin et al. as a
tetragonal phase observed during the hydrogen reduction of
LaNiO3 in a recirculating system.10,11 An infinite-layer structure
described by the stacking sequence NiO2 -La -NiO2
La was assigned to this phase, thus suggesting LaNiO2 is both
3-
3+
3-
-
2
5
3+
to be in good agreement with previously published patterns.
I
9
NdNiO was synthesized in a manner similar to that described for
isoelectronic (Ni d ) and isostructural with the Cu(II) parents
3
23
3
“high pressure” LaNiO . Intimately mixed Nd/Ni hydroxide sols were
of the high-Tc cuprate superconductors (e.g. Ca0.85Sr0.15CuO2 ).
The electronic properties of this reduced nickel oxide are of
importance in the search to extend high-temperature supercon-
ductivity beyond copper, and so a reliable synthetic method for
the synthesis of LaNiO2 has been pursued by a number of
In this paper we present details of the reduction
chemistry of LaNiO3 with the reducing agent NaH and a
decomposed at 200 °C in air, pressed into 7 mm pellets, and heated
for two periods of 24 h at 1000 °C under a pressure of 200 atm of
oxygen as described above. Powder X-ray diffraction patterns showed
the product to be monophasic and yielded lattice parameters in good
26
agreement with previously published values for orthorhombic NdNiO
3
.
workers.1
2,24
Reduction of LaNiO with NaH. The reduction of LaNiO
3
3
was
performed using NaH as a reducing agent. A double stoichiometric
excess of NaH (Aldrich 95%, NaOH was the only impurity observable
I
structural and magnetic characterization of the Ni phase,
LaNiO2. An analogous tetragonal NdNiO2+x phase is reported
for the first time using the same synthetic method.
by X-ray diffraction) was ground with LaNiO (typically 100 mg (4.16
3
3
mmol) of NaH with 500 mg (2.04 mmol) of LaNiO ), in a He-filled
Powder neutron diffraction and high-resolution transmission
electron microscopy show that the simple infinite-layer model
for LaNiO2 must be significantly modified by a novel defect
structure consisting of additional oxide ions present in the space
Mbraun Labmaster drybox and then sealed in an evacuated Pyrex
-
4
ampule (p < 2 × 10 Torr). The sealed reaction vessels were then
heated under conditions described in Table 1, to fully explore the
reducing behavior of the reagents. The reaction progress was monitored
via powder X-ray diffraction from well-ground reaction products, loaded
into O-ring-sealed sample holders in the drybox, and measured on a
Philips PW1710 diffractometer. When the X-ray data showed that the
reaction had reached completion, the products were then washed with
3-
between the NiO2 layers, producing both four-coordinate
square-planar and five-coordinate square-pyramidal environ-
ments for nickel.
CH
any unreacted NaH from the reaction mixture. Typically, 5 × 60 mL
of CH OH (which had been previously dried by distillation from Mg)
were transferred via cannula onto 500 mg of reaction products in a
100 mL filter Schlenk, under a nitrogen atmosphere. The suspension
was then vigorously agitated, for one minute, to allow the sodium phases
to react/dissolve before being filtered. After filtration the samples were
dried under vacuum for 2 h. The complete removal of the sodium phases
3
OH to remove the NaOH (produced as a reaction byproduct) and
Experimental Section
3
Synthesis. Starting Materials. A 5 g amount of LaNiO
synthesized via the coprecipitation of the metal ions from aqueous
3
was
9
-2
solution as described by Moriga et al. An 8.82 g (2.03 × 10 mol)
portion of La(NO ‚6H
O (Aldrich, 99.999%) and 5.93 g (2.03 × 10
mol) of Ni(NO ‚6H O (Johnson-Matthey, 99.9985%) were dissolved
in separate aliquots of 100 mL of distilled water to make solutions
-2
)
3 3
2
3
)
2
2
-
1
with concentrations of approximately 0.203 mol L . The exact metal
ion concentrations were determined by inductively coupled plasma
emission spectroscopy (to account for variable levels of hydration in
the starting materials), and the two solutions were then combined such
that a 1:1 molar ratio of La to Ni was achieved. The metal ions were
precipitated from the resulting solution with tetramethylammonium
(
21) Harwood, L. M.; Moody, C. J. Experimental Organic Chemistry -
Principles and Practice; Blackwell Scientific: Oxford, U.K., 1989.
22) Cotton, F.; Wilkinson, G. Inorganic Chemistry, 4th ed.; Wiley-
(
Interscience: New York, 1980; p 249.
(23) Siegrist, T.; Zahurak, S. M.; Murphy, D. W.; Roth, R. S. Nature
1988, 334, 231.
(
24) Moriga, T.; Usaka, O.; Nakabayashi, I.; Hirashima, Y.; Kohno, T.;
Kikkawa, S.; Kanamaru, F. Solid State Ionics 1994, 74, 211-217.
25) Garcia-Munoz, J. L.; Rodriguez-Carvajal, J.; Lacorre, P.; Torrance,
2
hydroxide (Aldrich, 25 wt % solution in H O) to form an intimately
mixed hydroxide sol. This green precipitate was immediately filtered
and washed with copious amounts of distilled water and then ethanol.
The resulting green jelly was heated on a hot plate to 200 °C to
decompose the organic coprecipitating agent. The black decomposition
(
J. B. Phys. ReV. B 1992, 46, 4414-4425.
(26) Demazeau, G.; Marbeuf, A.; Pouchard, M.; Hagenmuller, P. J. Solid
State Chem. 1971, 3, 582.

Topotactic Oxide Deintercalation
J. Am. Chem. Soc., Vol. 121, No. 38, 1999 8845
was then confirmed by X-ray powder diffraction and the chemical
techniques outlined below.
3 T) and then measuring magnetization-field isotherms. The paramag-
netic susceptibility of the sample was taken to be the slope of the linear
high-field sections of these plots. A simple subtraction of two high-
field magnetization measurements (H > 3 T) proved insufficiently
precise due to the small (∼10% in 4 T) contribution of the paramagnetic
Bulk powder samples suitable for powder neutron diffraction
-
2
experiments were prepared by reacting 5 g (2.03 × 10 mol) of
ambient pressure” LaNiO
with 980 mg (4.08 × 10 mol) of NaH at
90 °C, as described above, for five periods of 3 days with regrinding
between each heating period.
To investigate the role of solid-solid versus solid-gas reactions in
the mechanism by which NaH reduces LaNiO , reductions were
performed by loading LaNiO (500 mg, 2.04 mmol) and NaH (100
-
2
“
3
1
component to the sample magnetization. To this end a sample of LaNiO
2
3
(63 mg) prepared from high-pressure LaNiO was selected, as displaying
minimal ferromagnetic contamination (∼1.6% by mass), and cooled
in a 3 T field to 6 K. The magnetization of the sample was measured
in 10 fields up to 5 T. This procedure was repeated at 52 temperatures
from 6 to 300 K. The magnetization-field isotherms were fitted to a
linear function, with data points with large errors being excluded from
the fit. All fits had at least seven points.
An estimate of the resistivity of samples was measured by pressing
a small amount of the sample in question (about 200 mg) between two
brass pistons, within a Perspex tube of diameter 4 mm. In an attempt
to minimize undesirable contact effects, the two pistons were forced
together by means of an electrically insulated clamp, compressing the
sample and ensuring good electrical contact between all components.
The electrical resistance between the two pistons was measured, thus
allowing an estimate of the resistivity of the sample to be calculated.
The whole procedure was performed under an argon atmosphere to
prevent oxidation of samples by the air.
3
3
mg, 4.16 mmol) into separate clean Pyrex “thimbles”, under a helium
atmosphere, which were then loaded and sealed under vacuum in a
larger Pyrex vessel, such that the two reagents experienced the same
atmosphere inside the larger vessel but were not in physical contact.
Reduction reactions were also performed using ground mixtures of
NdNiO (500 mg, 1.99 mmol) and NaH (100 mg, 4.16 mmol) in the
3
temperature range 160 < T/°C < 210.
Characterization. Compositional Characterization. EDX analysis
was performed on a large number of crystallites, using a JEM-2010
microscope fitted with a windowless Si(Li) detector and operated using
the LINK “pentafet” energy-dispersive X-ray (EDX) microanalysis
system, to ensure no residual sodium remained in the samples.
Thermogravimetric and differential scanning calorimetry measure-
ments were performed using a Rheometric Scientific STA 1500 thermal
analyzer. Samples of around 50 mg were loaded rapidly into platinum
Results
-
1
crucibles and then heated at 4 °C min under flowing oxygen.
Structural Characterization. Neutron powder diffraction data were
collected on 5 g samples (which were indium-sealed in vanadium cans
under helium) on the D1B instrument at 5 and 290 K and on the D2B
instrument at 298 K at the Institut Laue Langevin, Grenoble, France.
Rietveld profile refinement was performed, on the neutron powder
diffraction data collected from the material, using the GSAS suite of
Dependence of Reaction Products on Synthetic Conditions.
The outcome of the reaction between LaNiO3 and NaH is
critically dependent on the synthetic conditions, as summarized
in Table 1. The reactions between ground mixtures of LaNiO3
and NaH show an evolution of reducing power and reaction
pathway with temperature. In the lowest temperature range (120
27
programs employing a pseudo-Voigt peak shape and a 16-term cosine
Fourier series background function.
<
T/°C < 180) mixtures of LaNiO3 and NaH yield the
4
monoclinic phase La2Ni2O5.
2
8,29
The DIFFaX program
was employed to simulate the effect of
Elevation of the temperature into the range 180 < T/°C <
stacking faults and intergrown defect layers on powder diffraction
patterns. Models were described by the stacking of defined layers
according to the transition probability Rij (the probability of layer j
being stacked on top of layer i) and the translation vector Rij (the vector
between the origin of layer i and layer j, described as a fraction of the
basis unit cell vectors).
High-resolution transmission electron microscopy (HRTEM) mea-
surements were performed using a JEOL 4000EX microscope operating
at 400 kV (point resolution ca. 1.7 Å). Finely ground samples were
2
5
10 yields a tetragonal phase (a ) 3.958 Å, c ) 3.381 Å, V )
3
2.96 Å ). This is assigned as LaNiO2 by comparison of the
lattice parameters with the only previous report of this Ni(I)
1
1
3
phase (a ) 3.966 Å, c ) 3.376 Å, V ) 53.10 Å ). Powder
X-ray diffraction measurements following the evolution of the
reduction with time at 190 °C are shown in Figure 1 (a 1:1
ratio of LaNiO3 to NaH was used in this reaction to reduce the
reaction rate and allow the observation of intermediate phases).
These data show that tetragonal LaNiO2 is formed via the
monoclinic Ni(II) phase La2Ni2O5. Direct reduction of La2Ni2O5
with NaH in the temperature range 180 < T/°C < 210 results
in the synthesis of the same tetragonal phase. In order for the
reduction of LaNiO3 to LaNiO2 to approach completion, a 2:1
stoichiometric excess of NaH and repeated regrinding is
required, particularly for large samples (5 g samples prepared
for neutron diffraction experiments required 5 × 3 days to reach
completion, as opposed to 500 mg samples, which required 3
× 3 days).
2
dispersed ultrasonically in methanol, under N , and then placed dropwise
onto lacy carbon grids (Agar, Cu 300 mesh) which were rapidly inserted
into the microscope to avoid oxidation by the air. The electron
diffraction patterns collected were indexed and d spacings calculated,
using a camera length obtained from a reference TlCl standard pattern,
with an estimated accuracy of (0.05 Å. High-resolution lattice images
were recorded close to the ideal Scherzer defocus. Simulations of these
images were calculated using the EMS suite of programs.30 Fourier
transforms of selected areas of recorded lattice images were calculated
3
1
using the Scion image program.
Magnetic and Electronic Characterization. Magnetic data were
collected, on samples contained within sealed gelatin capsules (which
had been previously loaded in the drybox), using a Quantum Design
MPMS SQUID magnetometer.
At temperatures within the range 210 < T/°C < 275,
disproportionation to La2Ni2O5, La2O3, and Ni follows the
reduction to Ni(I). Increasing the temperature to 275 < T/°C <
All samples were found to be contaminated by small (1.5-5 mass
) quantities of ferromagnetic impurity assigned as nickel metal. The
3
00 induces a different disproportionation route, to La2NiO4
%
and Ni. At higher temperatures, the presence of NaOH yields a
mixture of lanthanum hydroxides and oxyhydroxides and nickel
metal.
The reduction reactions performed with physically separated
reagents follow a different scheme. Below 220 °C there is no
observable reduction of LaNiO3. In the range 220 < T/°C <
340 LaNiO3 is reduced to La2Ni2O5. Above 340 °C dispropor-
tionation produces mixtures of La2Ni2O5, La2O3, and Ni, which
become completely reduced to La2O3 and Ni as the temperature
approaches 400 °C.
2
paramagnetic susceptibility of the bulk LaNiO material was measured
by saturating the ferromagnetic part of the signal (in fields larger than
(
27) Larson, A. C.; Von Dreele, R. B. General Structural Analysis
Systems; Los Alamos National Laboratory: Los Alamos, NM, 1994.
28) Treacy, M. M. J.; Newsam, J. M.; Deem, M. W. Proc. R. Soc.
London, Ser. A 1991, 433, 499.
29) Treacy, M. M. J.; Deem, M. W. Diffax - A Computer Program
for the Calculating of Diffraction Intensity from Faulted Crystals, version
.765, 1991.
(
(
1
(30) Stadelmann, P. A. Ultramicroscopy 1987, 21, 131.
(31) Rasband, W. Scion image; Scion Corp., 1998.

8846 J. Am. Chem. Soc., Vol. 121, No. 38, 1999
Hayward et al.
Figure 1. Evolution of the powder X-ray diffraction pattern, with time,
of a 1:1 molar mixture of intimately ground LaNiO
3
and NaH at 190
Figure 2. (a) X-ray powder diffraction pattern of products of the
°
C: (a) after 24 h, the sample is mainly La Ni O ; (b) after 36 h,
2 2 5
reaction between LaNiO
3
and NaH after three periods of 3 days at 190
reflections due to LaNiO
2
are appearing; (c) after 48 h, the sample is
Ni . The upper tick marks
correspond to the positions of the most intense peaks of La Ni and
the lower marks to those of LaNiO
°
C. Bragg reflections are indexed according to the tetragonal cell of
mainly LaNiO
2
but with traces of La
2
2 5
O
2
infinite layer LaNiO , and peaks due to NaH and NaOH are marked.
2
2 5
O
(b) X-ray powder diffraction pattern of reaction products for (a) after
2
. Sharp reflections at ∼38 and 44°
3
washing with CH OH under nitrogen. (c) EDX spectrum of the washed
are due to the aluminum sample holder.
reaction products. No peaks due to sodium are observed, demonstrating
the complete removal of the sodium-containing phases. (d) X-ray
powder diffraction pattern of the infinite-layer NdNiO2+x phase, after
washing with CH OH to remove sodium-containing phases. Note the
3
greater peak width anisotropy than is observed in LaNiO2+x phases.
The success of the CH3OH washing process at removing the
sodium-containing phases was demonstrated by elemental
analysis, X-ray diffraction, and EDX. No signals due to the
presence of sodium could be observed by either technique.
Figure 2 shows the powder X-ray diffraction patterns of the
tetragonal phase, assigned as LaNiO2, (a) before and (b) after
washing with CH3OH. Figure 2c shows a typical EDX spectrum
of the washed product, confirming complete removal of the
sodium-containing phases.
Annealing of washed LaNiO2 samples, in sealed evacuated

Topotactic Oxide Deintercalation
J. Am. Chem. Soc., Vol. 121, No. 38, 1999 8847
Figure 4. Relation between the rhombohedral structure of LaNiO
3
and
viewed along [100].
is vertical in this representation.
10,11
the infinite-layer model proposed for LaNiO
The stacking axis of LaNiO
2
2
Attempts to reduce the observed peak width anisotropy by
varying the synthetic conditions proved unsuccessful. The
temperature range over which this tetragonal phase can be
synthesized is very small, as described in Table 1, and varying
the reaction or annealing temperature within this small window
did not alter the peak width anisotropy.
Figure 3. Angular variation (2θ/deg) of the full width at half-maximum
2
height of X-ray powder diffraction reflections of LaNiO reduced from
ambient-pressure LaNiO (open symbols mark diffraction reflection of
the class 〈hk0〉 and closed symbols those of the class 〈h,k,l*0〉).
3
Reduction of more crystalline LaNiO3 prepared under high
oxygen pressure (the fwhm of the 110 reflection was 15%
smaller than for “ambient pressure” LaNiO3) gave an LaNiO2+x
product with narrower Bragg reflections (fwhm reduced by
Pyrex ampules, at temperatures below 210 °C resulted in no
observable change to the X-ray powder diffraction patterns of
these samples. Annealing at temperatures greater that 210 °C
resulted in slow but significant broadening of all reflections and,
on raising the temperature above 300 °C, a more rapid
decomposition to La2Ni2O5, La2O3, and Ni. LaNiO2 is not stable
with respect to disproportionation above 210 °C.
Magnetic measurements described above indicate the presence
of 1.5-5% nickel metal impurities by mass, in LaNiO2+x
samples. The exact amount present is heavily dependent on the
quality of the LaNiO3 starting material and is smallest in the
case of “high pressure” LaNiO3.
∼
30%). However, the highly anisotropic diffraction peak widths
observed in all previous samples were still as marked.
Structural Refinement. Rietveld profile refinement was
performed, in the space group P4/mmm, on neutron powder
diffraction data collected for LaNiO2. The starting point for these
refinements was the model proposed by Crespin et al., which
consists of infinite NiO2 layers, separated by eight-coordinate
3+
La ions. LaNiO3 is a rhombohedrally distorted cubic perovs-
2
5
kite, but it can be converted to an infinite-layer structure by
the ordered removal of one-third of the oxide ions (Figure 4).
All chemical manipulations of LaNiO2+x phases were per-
formed so as to exclude atmospheric oxygen and water.
Monitoring of LaNiO2 samples exposed to air over time, via
X-ray powder diffraction, reveals that all Bragg reflections start
to broaden after 20 min, with samples becoming almost entirely
amorphous after 4 h at room temperature. At elevated temper-
atures (above 100 °C) reoxidation to poorly crystalline LaNiO3
occurs in less than 1 h.
The large peak width anisotropy was modeled with aniso-
tropic size and strain broadening parallel and perpendicular to
the [001] axis, producing significant improvement to all
“goodness of fit” parameters. Ferromagnetically ordered nickel
metal was inserted into the refinement as a second phase.
The fit to the data, using a model with a single tetragonal
2
lanthanum nickelate phase, is unsatisfactory (ø ) 3.222).
Electron diffraction patterns obtained from the sample can be
readily assigned to the [001], [0-11], and [-111] zone axes in
the P4/mmm space group (Figure 5) and display no supercell
reflections and, thus, unambiguously confirm the space group
assignment. This suggests that the discrepancies between the
structural model and the observed data are due to microstructural
effects within the sample, which do not lead to a lowering of
symmetry.
Preparation of NdNiO2+x. The reduction of NdNiO3 with
NaH appears to occur at temperatures lower than those for
LaNiO3. No reaction was observed below 160 °C, but in the
range 160 < T/°C < 200 a tetragonal phase (a ) 3.9251(9) Å,
3
c ) 3.323(1) Å, V ) 51.195 Å ) was formed. A powder X-ray
diffraction pattern of this phase, after washing with CH3OH, is
shown in Figure 2d. By analogy with LaNiO2 this phase was
assigned as NdNiO2+x. Above 200 °C largely amorphous
reaction products were produced. The only crystalline phase
which could be identified in any of these patterns was Nd2O3.
No intermediate phase, analogous to La2Ni2O5, was observed
during reactions, even at low temperatures. The reduction
proceeds directly to the tetragonal phase described above with
no crystalline intermediates.
The simulation of the [001] zone lattice image (Figure 5d),
using the simple infinite-layer model, is in good agreement with
the observed projected potential at the calculated defocus,
implying that defects along the interlamellar axis are an
important component of the correct structural model.
Figure 6 shows a lattice image of the [100] zone axis of a
different crystallite of LaNiO2 where a defect region (b) within
the bulk structure (a) is apparent. The nature of this defect region
is revealed by the calculated electron diffraction patterns
obtained by Fourier transformation of the images of these two
areas. The inset (c) is the calculated diffraction pattern of the
ordered region (a), which can be readily indexed as the [100]
zone axis of a primitive tetragonal cell with lattice parameters
in a ratio consistent with those obtained for LaNiO2 from the
Structural Analysis and Refinement of LaNiO2. The
diffraction patterns of this tetragonal phase, collected from all
the samples synthesized within the temperature range 180 <
T/°C < 210, display highly anisotropic peak widths, as shown
in Figure 3. The sharper reflections are those from diffraction
planes of the class 〈hk0〉, and the broad peaks are from the class
where l has a nonzero value, characteristic of anisotropic particle
size, stacking disorder, or uniaxial strain in layered compounds,
as has been observed in a number of other topotactically reduced
(32) Retoux, R.; Rodriguez-Carvajal, J.; Lacorre, P. J. Solid State Chem.
3
2
oxides.
1998, 140, 307.

8848 J. Am. Chem. Soc., Vol. 121, No. 38, 1999
Hayward et al.
2
Figure 6. Lattice image of the [100] zone axis of LaNiO showing
Figure 5. Electron diffraction patterns of the (a) [001], (b) [0-11],
and (c) [-111] zone axes of LaNiO indexed in the space group P4/
ordered (a) and disordered domains (b) and Fourier transforms of the
respective regions (c, d). The Fourier transforms correspond to the
electron diffraction patterns of these region and are indexed according
2
2
mmm. (d) Lattice image of the [001] zone axis of LaNiO , including
inset (top right) simulated, within the EMS suite of programs,30 using
2
to the reciprocal space of tetragonal LaNiO .
the infinite-layer model and employing the following microscope
parameters: accelerating voltage, 400 kV; spherical aberration coef-
ficient, 0.9 nm; semi-convergence angle, 0.8 mrad; objective lens
aperture, 10 nm.
contribution to the profile. This led to a marked improvement
2
in the fit quality (ø ) 2.16).
The atomic displacement parameters of the nickel and oxygen
atoms in the second phase were significantly larger than those
bulk neutron diffraction data. The inset (d) is the calculated
diffraction pattern from the defect region (b), which also displays
a tetragonal motif with an a lattice parameter consistent with
LaNiO2. However, the c* direction exhibits extensive streaking,
indicating extensive strain or structural disorder parallel to the
2
2
of the lanthanum (UNi ) 0.106 Å , UO ) 0.051 Å , ULa )
2
0.0064 Å ) and than those of all the atoms in the majority
LaNiO2 phase (UNi ) 0.0051 Å , UO ) 0.0089 Å , ULa ) 0.0010
Å ), consistent with the increased disorder observed in the lattice
2
2
2
[
001] axis. The occurrence of such highly disordered or strained
image (Figure 6b). Components of the anisotropic displacement
ellipsoids were markedly anisotropic, with displacement parallel
to the c axis nearly 7 times greater than those in the ab plane.
This, together with the expansion along the stacking direction
of the minority phase (amaj ) 3.959 Å, cmaj ) 3.374 Å, amin )
3.943 Å, cmin ) 3.454 Å), suggests displacive disorder of the
NiO2 sheets along the stacking direction produced by an
domains may be correlated with the poor Rietveld fit with the
simple infinite-layer model and suggests that the powder
diffraction data might be more effectively modeled by introduc-
ing a second tetragonal lanthanum nickelate phase to account
for these regions of the sample.
By combination of multiple-phase Rietveld refinement tech-
niques with the DIFFaX simulation program, an investigation
into possible defects present in the sample was performed. A
second tetragonal lanthanum nickelate phase was added to the
Rietveld profile refinement, with atomic coordinates identical
with those of the first phase, and freely refined lattice
parameters, Lorentzian peak profile coefficients, and thermal
parameters. The Gaussian profile width was constrained to be
the same as in the other phases, to model the instrumental
1
interlamellar defect of residual oxide anions at the (0, 0, /2)
position. The close contacts from this defect anion to the layers
led to each interlamellar anion displacing the two adjacent NiO2
layers; therefore, additional nickel and oxygen atoms were
1
inserted into the second phase at (0, 0, z) and ( /2, 0, z′),
respectively, to model this, with the occupancies of sites
constrained such that the stoichiometry of the NiO23 layers
and the La/Ni ratio were maintained (relaxation of these
-

Topotactic Oxide Deintercalation
J. Am. Chem. Soc., Vol. 121, No. 38, 1999 8849
Table 2. Structural Data for the Final Rietveld Refinement of
a
LaNiO
2
Refined from Neutron Powder Diffraction Data
Phase 1: Space Group P4/mmm
x
y
z
Uiso/Å2
site sym
Ni
La
O
0
0.5
0.5
0
0.5
0
0
0.5
0
0.00026(6)
0.00040(3)
0.0032(8)
1 a 4/mmm
1 d 4/mmm
2 f mmm
a ) 3.9592(1) Å, c ) 3.3745(2) Å; mole fraction 59.30(3)%
Phase 2: Space Group P4/mmm
x
y
z
fraction
Uiso/Å2
site sym
Ni1
Ni2
La1
La2
O1
0
0
0
0
0.5
0.5
0
0
0.639(43)
0.180(21)
0.639(43)
0.180(21)
0.639(43)
0.180(21)
0.090(11)
0.081(7)
0.027(7)
0.0058(2)
0.0087(1)
0.061(6)
0.033(4)
0.083(6)
1 a 4/mmm
2 g 4mm
1 d 4/mmm
2 h 4mm
2 f mmm
4 i 2mm
1 b 4/mmm
0.211(10)
0.5
0.601(17)
0
0.210(15)
0.5
0.5
0.5
0.5
0.5
0
O2
O3
0
0
a ) 3.9431(8) Å, c ) 3.454(2) Å; mole fraction 23.7(2)%
Ni Metal Phase: Space Group Fm 3h m
x
y
z
Uiso/Å2
site sym
4 a m 3h m
Ni
0
0
0
0.0011(6)
a ) 3.5229(3) Å magnetic moment ) 0.568 µB mole fraction: 16.7(2)%
Goodnesss of Fit Parameters
Rwp
2.94%
2.27%
1.881
Rp
2
ø
bond
dist/Å
bond
dist/Å
Ni1-O1
Ni2-O2
Ni2-O3
La1-O1
La1-O2
La -O
1.9716(4)
1.9716(4)
2.46(4)
2.6210(7)
3.15(4)
2.7882(6)
angle/deg
179.764(11)
La2-O1
La2-O2
La1-Ni1
La1-Ni2
La2-Ni1
La -Ni
2.4075(6)
2.388(30)
3.2798(7)
3.715(30)
3.111(6)
3.095(15)
angle/deg
89.9(13)
Figure 7. Observed, calculated, and difference plots for the final
Rietveld refinement of LaNiO , from powder neutron diffraction data,
2
employing a two-phase model and disordered displacements in the
minority phase. The expanded region 45 < 2θ/deg < 90 shows the
improvement to the fit achieved by the insertion of a second disordered
phase in comparison with Figure S1 (Supporting Information).
1
3
2
2
bond
bond
O2-Ni2-O2
O2-Ni2-O3
a
Phase 1 is the majority “perfect” infinite-layer phase representing
. Phase 2
regions of the sample which have the stoichiometry LaNiO
2
constraints did not result in a change in the refined stoichiometry
of greater than 2%). The ( /2, 0, 0) in-plane oxide anion was
represents defective regions of the sample which have interlamellar
oxide ion defects which change the stoichiometry of the phase to
LaNiO2.09. Refined bond lengths and angles refer to atom labels for
phase 2 shown in Figure 8b.
1
refined to be fully occupied.
2
This modification of the model gave a ø value of 1.89 and
a physically reasonable explanation of the anisotropic broaden-
ing which dominates the observed diffraction pattern. Finally,
to relax the rather short bond length between the lanthanum
The disordered minority phase, shown in Figure 8b, can be
described as a superposition, or intergrowth, of two different
stacking sequences (labeled I and II) and is depicted more clearly
in Figure 9a. Special attention should be drawn to the lanthanum
disorder site (La2) and the large displacement parameters of the
Ni1 and O1 atoms in this minority phase. These features can be
correlated with the relaxation of what would otherwise be short
La-O bond lengths in the NiO2-La-NiO2-La matrix around
a defect NiO2-LaO-NiO2 layer, as shown in Figure 9a.
The validity of this defect model can be tested, and extra
information about the correlation between defect locations
gained, by simulating powder diffraction patterns of this
disordered intergrowth using the DIFFaX simulation program.
The program employs a recursion algorithm to calculate
kinematical diffraction intensities for crystals containing planar
1
1
1
1
position ( /2, /2, /2) and the disordered oxygen at ( /2, 0, z) (2.21
Å) in the minority phase the lanthanum site was split by the
1
1
introduction of a second site at a ( /2, /2, z) position with an
occupancy constrained in the same manner as the nickel disorder
2
site, giving a final ø value of 1.881 and a relaxed La-O bond
length of 2.38 Å.
The observed, calculated, and difference plots for the final
Rietveld refinement of LaNiO2.03 are shown in Figure 7, with
structural data and goodness of fit parameters listed in Table 2.
Refined bond lengths and ORTEP diagrams of the two refined
phases are shown in Figure 8, with bond lengths and angles for
the minority phase shown in Table 2. The respective peak
broadening contributions from particle size and structural strain
2
8,29
faults.
A faulted crystal, corresponding to intermingled stacking of
LaNiO2 infinite layers and LaNiO2+x defect layers, constrained
by the refined stoichiometry, can be constructed as shown in
Figure 10 by defining the probabilities of transitions between
the block types, relative to defined unit cell parameters, as
described in Table 3. Layer 1 corresponds to the “perfect”
stoichiometric LaNiO2 majority phase refined from the neutron
powder diffraction data. Layers 2 and 3 represent the two
in the two phases are detailed in Table S1 of the Supporting
Information. 35
(
33) Greenwood, N. N.; Earnshaw, A. Chemistry of the Elements, 2nd
ed.; Butterworth-Heinemann: Oxford, U.K., 1997; p 1149.
34) Chrzan, D. C.; Falicov, L. M.; Maclaren, J. M.; Zhang, X. G.; Gonis,
A. Phys. ReV. B 1991, 43, 9442.
35) Larson, A. C.; Von Dreele, R. B. GSAS Technical Manual; Los
Alamos National Laboratory: Los Alamos, NM, 1994.
(
(

8850 J. Am. Chem. Soc., Vol. 121, No. 38, 1999
Hayward et al.
Figure 8. (a) ORTEP plot and bond lengths for majority infinite-layer
phase LaNiO . Atoms are represented as ellipsoids of 100 times the
2
+
3+
refined value. The Ni and La cations are in square and square-
prismatic coordinations, respectively, with refined bond lengths of
Ni-O ) 4 × 1.979 60(6) Å and La-O ) 8 × 2.6010(1) Å. (b) ORTEP
plot for the minority defect LaNiO2+x phase found to be in the LaNiO
2
sample. (c) Representation of (b) as the superposition of an infinite-
layer phase (I) and a defect phase with interlamellar oxygen ions (II),
and the consequent relaxation of the atomic positions around the defect.
The La position is required to relax the short La-O distances at the
2
interface between defect and bulk regions.
Figure 10. (a) NiO -La-NiO2 and NiO -LaO-NiO stacking
2
2
2
sequences used in the DIFFaX program to simulate diffraction patterns
of the defect structure of LaNiO2.03. (b) Example of a faulted crystal
formed by stacking NiO
2
-La-NiO
2
2 2
units with NiO -LaO-NiO units
using the probabilities defined in Table 3. (c) Neutron powder
diffraction pattern of LaNiO2.03 calculated using the DIFFaX program,
employing the model in Table 3, compared with the observed D2B
data. The unfitted Bragg reflections are from the Ni metal impurity.
Describing the structure by the probabilistic stacking of these
layers directs information about transitions between NiO2-La-
NiO2 and NiO2-LaO-NiO2 stacking units and the associated
relaxation of the NiO2-La-NiO2 matrix, as shown in Figure
9
b. The description of the minority phase is enhanced by the
inclusion of “stacking uncertainty” factors for layer 2, introduc-
ing an ellipsoidal error spread for the translation vector, Rij, to
allow for the effect of internal strain arising from the insertion
of a defect NiO2-LaO-NiO2 stacking unit and the resulting
perturbation of the surrounding NiO2-LaO-NiO2 over a
number of unit cell lengths.
Localization of interlamellar defects within discrete layers
improves the agreement between the calculated and observed
patterns significantly. With randomly distributed defects the
number of NiO2-LaO1-x-NiO2 stacking units required in the
model, to achieve the experimentally measured stoichiometry,
is much greater, resulting in a much broader peak shape than
observed. It should be noted that the transition probability R33
is zero, maintaining the five-coordinate environment around Ni2
refined from the data (increasing R33 above 0.03 resulted in
significant intensity mismatches between the simulated and
observed diffraction patterns). The powder neutron diffraction
Figure 9. (a) Local structure and bond lengths in the region around a
defect NiO
2
2
-LaO-NiO stacking unit in the minority defect phase,
interpreted from the positions in lamellar LaNiO2+x refined from neutron
powder diffraction data. (b) Local structure and bond lengths in the
vicinity of the NiO
DIFFaX simulation model.
2
2
-LaO-NiO defect layers as described within the
different stacking sequences (NiO2-La-NiO2 and NiO2-LaO-
NiO2, respectively, see Figure 10) observed in the minority,
LaNiO2.09, defect phase refined from the neutron diffraction data.

Topotactic Oxide Deintercalation
J. Am. Chem. Soc., Vol. 121, No. 38, 1999 8851
Table 3. Description of Stacking Units and Transition Probabilities
Employed in the DIFFaX Simulations of Defect Structure of
LaNiO2.03
Magnetic Characterization. The analysis of the magnetic
properties of LaNiO2 has been hampered by small amounts of
ferromagnetic nickel impurities masking the behavior of the bulk
sample. Magnetic measurements on six different samples of
LaNiO2 show the figure of 5.66% Ni metal by mass for the
bulk powder neutron diffraction sample above is by far the
largest observed for any sample measured. A more typical figure
of between 1.5 and 2% is observed for most samples. As noted
above samples reduced from “high pressure” LaNiO3 contain
smaller amounts of nickel metal impurities. The size of sample
reduced appears to have no direct bearing on the amount of Ni
produced. However, maintaining the reaction temperature below
Layer 1
atom
x
y
z
Ni
La
O
0
0
0.5
0
0
0.5
0
0.5
0.5
0
O
0.5
0
R
ij
i,j
R
ij
1
1
1
,1
,2
,3
0.75
0.25
0
0
0
0
0
1
1
2
00 °C does reduce the amount of nickel impurity within
samples (500 mg aliquots of the same “high pressure” LaNiO3
sample reduced for 3 × 3 days at 180 and 205 °C, respectively,
had 1.8% and 2.6% Ni by mass).
mole fraction of layer 70.2%
Layer 2
Therefore, magnetic measurements were performed on “high
pressure” samples, which were reduced, as described above, at
atom
x
y
z
Ni
La
O
0
0
0.5
0
0
1
90 °C, to further minimize the amount of metallic nickel
impurity.
Powder neutron diffraction data collected at room temperature
0.5
0.5
0
0.51
0
O
0.5
0
on the D2B instrument at the Institut Laue Langevin, Grenoble,
France, were readily refined, as described above, without the
introduction of any ordered magnetic contribution to the
scattering of the lanthanum nickelate phases. Data were also
collected at 5 and 300 K on the D1B instrument, and neither
data set displayed any features which could not be modeled
well on the basis of nuclear scattering alone. Any attempt to
introduce ferromagnetic ordering models resulted in the mag-
nitude of the ordered moment refining to zero in both lamellar
phases.
Similarly, the introduction of antiferromagnetic ordering
models, such as that displayed by Ca0.85Sr0.15CuO2 (antiferro-
magnetically ordered sheets of CuO2 aligned such that interac-
tions between adjacent sheets are also antiferromagnetic),
yielded no ordered magnetic moment in the lanthanum nickelate
phases (the largest ordered moment consistent with the 5 K
neutron diffraction data is 0.05 µB per nickel atom).
R
ij
i,j
R
ij
B
11
B
22
B
33
2
,1
,2
,3
0.65
0.25
0.1
0
0
0
0
0
0
1.02
1.02
1.02
0
0
0
0
0
0
2
2
2
2
2
mole fraction of layer 27.02%
Layer 3
atom
x
y
z
Ni
La
O
0
0
0.5
0
0
0.5
0.5
0
0.725
0
0
O
0.5
R
ij
i,j
R
ij
3
3
3
,1
,2
,3
0
1
0
0
0
0.145
mole fraction of layer 2.7%
Measurement of the magnetic behavior of the bulk material
was performed by measuring magnetization isotherms at a range
of temperatures and then using the subtraction method described
in the Experimental Section. It should be noted that this
procedure is only valid if the nickel metal particles present in
the sample are ferromagnetic rather than superparamagnetic. The
average particle size estimated from the Rietveld profile
refinement (see Table S1 in the Supporting Information) would
appear to suggest that the nickel particle size is considerably
larger than that expected for a superparamagnet (for a blocking
temperature of 300 K the critical radius is on the order of 40
Å ), so the procedure in this case is valid. It should, however,
be noted that a wide range of nickel particle sizes may be present
in the sample, some potentially below 40 Å in radius; therefore,
the values of the paramagnetic susceptibility derived by this
method should be regarded as an upper limit for the bulk
LaNiO2.03 phase. Figure 11i shows a typical linear fit to a series
of isothermal magnetization measurements, the slope of which
is the paramagnetic susceptibility of the nonferromagnetically
saturated material in the sample. Figure 11ii shows the plot of
the fitted molar susceptibility measured by the subtraction
procedure plotted against temperature, for LaNiO2.
a
Atomic positions and translation vectors are defined relative to the
basis lattice parameters: a ) b ) 3.959 Å, c ) 3.373 Å. Uncertainty
factors (B11, 22, B33) define an error spread for the translation vector
ij in an analogous manner to the way in which Debye-Waller factors
describe the uncertainty in atomic position due to thermal motion. The
B
R
values quoted for the uncertainty factors are related to the rms
2
displacement of the layers, Uij, by Rij ) 2π a
i
*a
j
*Uij
.
pattern simulated using this model can be seen to be in good
agreement with the observed data, as shown in Figure 10c.
The average composition of the lanthanum nickelate phases
derived from the Rietveld refinement of the powder neutron
data is LaNiO2.025(0.003. The thermogravimetric data, detailed
in Figure S2 (Supporting Information) for the reoxidation of
LaNiO2+x with oxygen, along with a value of 5.66% Ni metal
by mass in the sample (obtained from the Rietveld refinement
of the powder neutron data; see Table 2), give an average
composition of LaNiO2.11(0.02.
The apparent disparity between refined and thermogravi-
metrically measured stoichiometries can be partially accounted
for by the observed air sensitivity of LaNiO2+x.
Thermogravimetric analysis of the new infinite-layer phase
NdNiO2+x is limited to the statement that the minimum value
of x is 0.04 due to the lack of accurate quantification of the
impurities present in these samples (any unaccounted for Ni
metal, and accompanying Nd2O3 present, will make x appear
smaller than its true value).
3
6
The paramagnetic susceptibility appears to be largely tem-
perature independent (rising by only 60% from 300 to 4 K)
(36) Morrish, A., H, The Physical Principles of Magnetism, 1st ed.;
Wiley: New York, 1965; p 362.

8852 J. Am. Chem. Soc., Vol. 121, No. 38, 1999
Hayward et al.
climbing again from 50 to 6 K. This temperature range can be
again be modeled by the sum of Curie-Weiss and temperature
independent terms (curve c). In this low temperature range the
temperature-independent term dominates the paramagnetic
susceptibility with only a few (∼1 mol %) weakly antiferro-
magnetically coupled spins (Θ ) -4 K) remaining.
The measured resistivity of the sample is 10((5) Ω cm at
3
00 K (the metallic phases LaNiO3 and NdNiO3 have resistivi-
-
3
38
ties on the order of 5 × 10 Ω cm at 300 K ). It should be
noted, however, that this value of the resistivity is only an
estimate and may be heavily influenced by contact effects and
the presence of metallic nickel particles.
Discussion
Sodium hydride can effect the topotactic reduction of LaNiO3
to LaNiO2, which is not possible with flowing hydrogen or metal
-
getter reagents. The large reduction potential of the H ion, in
2
2
NaH, has allowed the isolation of LaNiO2 at a temperature
below 210 °C, where most other reducing agents would not be
effective. The difficulties experienced in preparing this material
with other reagents may be attributed to its low thermal
stabilitysLaNiO2 disproportionates into La2Ni2O5, La2O3, and
Ni on annealing above 210 °C under vacuum (this is in contrast
with the only previous report of the reduction of LaNiO3 to
LaNiO2 “below 670 K” in a recirculating hydrogen/helium
10,11
system by Crespin et al.
). This lack of thermal stability may
also be the reason stoichiometric LaNiO2 is not synthesized by
this method. As reaction temperatures approach 210 °C, there
may be kinetic competition between reduction and decomposi-
tion, preventing reaction completion.
Determination of the exact mechanism by which NaH is
reducing LaNiO3, below 210 °C, will require further research,
but it does not appear to involve hydrogen or oxygen gas as
intermediates. There are three major mechanisms that could be
operating in the reduction reaction:
Figure 11. (i) Magnetization-field isotherm of LaNiO2+x at 300 K.
The inset shows the linear fit to the high-field data. øparamagnetic ) [1.013-
-
3
-1
(
9)] × 10 emu mol , intercept (saturated ferromagnetic magnetiza-
tion) 290.3(3) emu mol , goodness of fit R ) 0.9989. (ii) Fitted molar
paramagnetic susceptibility of LaNiO . The data are shown as squares
-1
2NaH(s) + LaNiO (s) f 2Na(s) + H (g) +
3
2
2
LaNiO (s) f 2Na(s) + H O(g) + LaNiO (s) f
3
2
2
and diamonds in the temperature ranges 300 > T/K > 150 and 150 >
T/K > 6, respectively. Curve fits: (a) Curie susceptibility (Θ ) 0 K)
NaOH(s) + NaH(s) + LaNiO
(s) (a)
2
1
for an S ) /
2
spin-only paramagnet; (b) Curie-Weiss + temperature-
1
independent term (C/(T - Θ) + R), S ) /
2
, Θ ) -257 K, R ) 3.7 ×
emu mol , R ) 0.97; (c) Curie-Weiss + temperature-independent
term (C/(T - Θ) + R), C ) 0.0031 (corresponding to ∼1 mol % S )
spins), Θ ) -4 K, R ) 0.0012 emu mol , R ) 0.94.
NaH(s) + LaNiO (s) f NaH(s) + LaNiO (s) +
3
2
-4
-1
10
1
/2
O (g) f NaOH(s) + LaNiO (s) (b)
2 2
1
-1
/
2
NaH(s) + LaNiO (s) f NaOH(s) + LaNiO (s) (c)
3
2
and is significantly smaller than would be expected for an S )
1
Note that these equations are only intended to distinguish
solid-solid from solid-gas mechanisms and therefore do not
explicitly contain the La2Ni2O5 intermediate seen experimen-
tally.
/
2 noninteracting paramagnet (which is indicated by curve a in
the figure). The small temperature dependence confirms our
assertion that the small particles of nickel, observed in the
powder neutron diffraction data, are behaving ferromagnetically
rather than superparamagnetically, as any superparamagnetic
contribution to the fitted susceptibility (which would not have
Martinez-Lope et al.39 reported that LaNiO3 is stable up to a
temperature of 250 °C in a 5%/95% H2/N2 atmosphere, in
agreement with our observation that LaNiO3 did not undergo
any significant reduction, on a thermogravimetric balance, after
been removed by the subtraction) would display a strong
temperature dependence.37
1
0 h at 200 °C in a 5%/95% H2/N2 atmosphere. Reduction to
The susceptibility data are best described by considering two
temperature regions; 6 < T/K < 150 and 150 < T/K < 300. At
high temperatures, 150 < T/K < 300, the susceptibility gives a
reasonable fit to the sum of a temperature independent term
La2Ni2O5 occurs, with intimately mixed NaH, in the temperature
range 120 < T/°C < 180. Reaction of physically separated
LaNiO3 and NaH, within an evacuated Pyrex tube, results in
no observable reaction below 220 °C. This evidence would
appear to discount both mechanisms a and b, which have
gaseous reaction intermediates and therefore do not require
1
and a Curie-Weiss paramagnetic term with S set equal to /2
(curve b). The Weiss constant is very large and negative (-257
K), indicating significant correlation between spins and departure
from strict Curie-Weiss behavior. The susceptibility then
reaches a plateau over the region 50 < T/K < 150 before
(
38) Blasco, J.; Castro, M.; Garcia, J. J. Phys.: Condens. Matter 1994,
6, 5875-5889.
39) Martinez-Lope, M. J.; Alonso, J. A. Eur. J. Inorg. Chem. 1995, 32,
361.
(
(37) Leslie-Pelecky, D. L.; Rieke, R. D. Chem. Mater. 1996, 8, 197.

Topotactic Oxide Deintercalation
J. Am. Chem. Soc., Vol. 121, No. 38, 1999 8853
physical contact between reagents. Reduction of LaNiO3 to La2-
Ni2O5 did occur above 220 °C, when the reagents were
physically separated, presumably via mechanism a (decomposi-
tion of NaH to Na and H2), as the decomposition temperature
of NaH is 210 °C (defined as the temperature required to
two different substructures, the “perfect” infinite-layer model
(exhibiting NiO2-La-NiO2-La stacking) and a defect phase
(exhibiting defective layers of NiO2-LaO-NiO2). The super-
position observed in the second phase models regions of the
sample where the two stacking sequences have intergrown
intimately with each other, as shown in Figure 10. That is to
40
produce a hydrogen pressure of 10 mmHg ) and getter reactions
similar to mechanism b do not occur until much higher
temperatures (reaction between LaNiO3 and a Zr getter requires
say, the minority phase consists of defective NiO -LaO-NiO2
2
blocks inserted within a relaxed NiO -La-NiO -La stacking
2
2
4
a temperature of at least 400 °C ). The products obtained by
sequence.
the use of physically separate NaH, at temperatures where a
significant H2 partial pressure is present, agree well with those
found using H2(g).
The strain parallel to the stacking direction is over 4 times
that observed parallel to the NiO2 planes in the minority phase,
due to the significant mismatch in the c lattice parameters
Additional evidence to suggest mechanism c (diffusion
between solids) is occurring with the intimately mixed reagents
is provided by the observation that the rate of reaction is affected
by the sample size (large samples react more slowly), mixing
of reagents (poorly mixed samples react more slowly), and ratio
of NaH to LaNiO3 (increasing the amount of NaH with respect
to LaNiO3 increases the reaction rate at a given temperature).
All three observations indicate physical contact between the
reagents is required for reaction, suggesting the reduction occurs
via diffusion between solids (c) rather than via a gaseous
intermediate (a, b). The key point is that LaNiO2 is not formed
by either route (flowing H2 or separated reagents within a sealed
tube) involving gas-phase hydrogen as a reducing agent.
Analysis of neutron powder diffraction and high-resolution
electron microscopy data has shown that LaNiO2+x has a defect
microstructure which requires a significant modification of the
infinite-layer model, proposed by Crespin et al., in a manner
not previously observed in the defect chemistry of infinite-layer
cuprates.
(
stacking repeat length) between the intergrowing sequences ((c-
defect(Ni-O-Ni)/c-perfect(Ni-vacancy-Ni) ) 1.46), which
results in a considerable strain in the stacking direction,
compared to the small strain within the NiO2 planes ((a-defect/
a-perfect) ) 0.996). The combination of both anisotropic particle
size and uniaxial strain leads to the highly anisotropic diffraction
peak widths observed for this phase.
The coordination around nickel in the minority phase must
be discussed in terms of two sites. (a) Site Ni1, the majority
(
exhibit square-planar coordination with bond lengths very
similar to those of the majority phase (1.972 Å as opposed to
1
62.3%) of the nickel atoms in this phase occupy this site and
.979 Å in the majority phase). (b) Site Ni2, a disordered site
displaced due to its proximity to an interlamellar oxide ion (O3).
The Ni2 position has four in-plane oxygen contacts (Ni2-O2)
of 1.972 Å, and the symmetry of the P4/mmm space group also
generates two Ni2-O3 bond lengths (1.00 and 2.46 Å). However,
Ni(I) would be expected to exhibit a large Jahn-Teller distortion
(
cf. La2CuO4); therefore, all Ni2-O3 bond lengths should be
The refined structure consists of two tetragonal phases with
similar a lattice parameters but significantly different c lattice
parameters due to the incorporation of interlamellar oxide ion
defects. The first is a highly ordered phase which has the
considerably in excess of 1.979 Å. The shorter Ni2-O3 bond
length is thus rejected as unphysical, reinforcing the picture of
the minority phase as the superposition of two constituent
phases, as shown in Figures 8 and 9. Ni2 thus adopts square-
pyramidal coordination, consistent with a large Jahn-Teller
distortion at Ni(I). The ratio Ni-Oaxial/Ni-Oequatorial (Ni2-O2/
Ni2-O3) refined for the minority lanthanum nickelate phase is
“
perfect” infinite-layer structure and is observed to exist in
crystalline domains which have far larger dimensions in the ab
plane of the material than in a direction parallel to the stacking
direction (c axis) (Table S1).
1
.247, which compares well with a value of 1.289 for La2CuO4
I
At first sight, the Ni -O bond lengths for the majority phase
43
(four- plus two-coordination) and an average value of 1.202
I
(
1.979 Å) are among the longest reported for Ni in the solid
for the Cu-O bond lengths in YBa2Cu3O7-x (square-pyramidal
coordination). All interatomic contacts to O3 should be viewed
in the light of the large rms displacement of this ion (0.28 Å).
The defective regions, therefore, correspond to the insertion of
vertex-linked NiO5 square pyramids into the stacking sequence.
state. However, it should be noted that there are very few
examples suitable for direct comparison with the infinite sheets
44
I
found here, as most reported Ni oxides display either strikingly
different coordinations, for example the linear NiO2 units in
4
1
(
K,Rb)Na2NiO2 (Ni-O ) 1.77 Å ), or significant disorder in
By analysis of the other metal-oxygen bond lengths (La1-
O2 for example) and rejection of the physically unreasonable
short contacts, it becomes clear that a model of the local
structure around a defect NiO2-LaO-NiO2 block can be
constructed, from the refined minority phase, as shown in Figure
42
the nickel coordination shell, e.g. LaSrNiO3.1 or the LnSr5-
5
I
Ni3O8 phases. When only the Ni -O distances to fully ordered
oxygen sites are considered, the refined bond length for the
stoichiometric LaNiO2 appears to be reasonable (the nondisor-
I
42
dered Ni -O bond length in LaSrNiO3.1 is 2.04 Å; nondis-
9
a. The apparent compression of the NiO2-La-NiO2 blocks
I
ordered Ni -O bond lengths in LnSr5Ni3O8 range from 1.995(7)
directly adjacent to the defect layer is exaggerated in this
scheme, however, due to the imposition of a common c lattice
parameter onto the intergrowth of the two stacking units. The
refined positions of the NiO2 layers, constructed from Ni1 and
O1, should therefore be considered a weighted average of
positions summed over the locally relaxed NiO2-La-NiO2
matrix. Evidence for this can be seen in the large displacement
factors associated with Ni1 and O1. The rms displacements of
5
Å for Ln ) Ho to 1.986(7) Å for Ln ) Y ).
The second tetragonal phase deviates significantly from the
“perfect” infinite-layer model, exhibiting disordering of the NiO2
planes driven by the presence of additional interlamellar oxide
ions which change the average stoichiometry of the phase to
LaNiO2.09(1). Closer inspection shows that the disorder in this
phase can be more accurately viewed as the superposition of
(
40) Greenwood, N. N.; Earnshaw, A. Chemistry of the Elements, 2nd
ed.; Butterworth-Heinemann: Oxford, U.K., 1997; p 66.
41) Burow, W.; Birx, J.; Bernhardt, F.; Hoppe, R. Z. Anorg. Allg. Chem.
993, 619, 923.
42) Crespin, M.; Landron, C.; Odier, P.; Bassat, J. M.; Mouron, P.;
Choisnet, J. J. Solid State Chem. 1992, 100, 281.
(43) Longo, J. M.; Raccah, P. H. J. Solid State Chem. 1973, 414, 526.
(44) David, W. I. F.; Harrison, W. T. A.; Gunn, J. M. F.; Moze, O.;
Soper, A. K.; Day, P.; Jorgensen, J. D.; Hinks, D. G.; Beno, M. A.;
Soderholm, L.; Capone, D. W.; Schuller, I. K.; Segre, C. U.; Zhang, K.;
Grace, J. D. Nature 1987, 327, 310.
(
1
(

8854 J. Am. Chem. Soc., Vol. 121, No. 38, 1999
Hayward et al.
the ions (0.28 and 0.24 Å, respectively) are large enough to
dimensional ordering being suppressed by the structural and
electronic effects of the interlamellar oxide ions.
extend the short La-O and La-Ni bonds described in Figure
9
a to the more physical values used in the DIFFaX model and
The tetragonal phase, NdNiO2+x, synthesized via the reduction
of NdNiO3 with NaH, is expected to have a defect structure
similar to that of LaNiO2 due to the highly anisotropically
broadened reflections seen in all X-ray powder diffraction
patterns of this phase and the stoichiometry measured by
thermogravimetric analysis. The lattice parameters (a ) 3.9251-
shown in Figure 9b.
It should be noted that the O3-Ni2-O3 angle, in the
disordered phase, is 179.7(2)°, indicating the NiO2 sheets are
very nearly planar. This implies that the NiO2 sheets have been
1
separated by the insertion of oxygen at (0, 0, /2) with very little
3
buckling, providing further evidence that the interlamellar oxide
ions are localized in distinct layers rather than distributed
randomly throughout the whole lattice.
(9) Å, c ) 3.323(1) Å, V ) 51.195 Å ) of this phase in
comparison with those derived from a single-phase model of
2
LaNiO2 (a ) 3.961 Å, c ) 3.381 Å, V ) 53.046 Å ) are
consistent with the reduction in ionic radius from La3 to Nd .
The thermal stability of NdNiO2+x mirrors that of LaNiO2+x,
with disproportionation into poorly crystalline products occur-
ring above 200 °C (the only phase identifiable from powder
X-ray diffraction was Nd2O3).
+
3+
The magnetic data show that all samples are contaminated
by small amounts of metallic nickel. This has been assigned to
the presence of amorphous NiO in the LaNiO3 starting material,
which could not be observed by powder X-ray diffraction but
which is readily reduced to Ni metal by NaH. The significantly
smaller amounts of nickel metal in samples reduced from “high
pressure” LaNiO3 would appear to confirm this. Measurement
of the magnetic behavior of the bulk sample provides no
evidence for any long-range magnetic order in LaNiO2. The
paramagnetic susceptibility of this Ni(I) phase, measured by
the subtraction of the ferromagnetic impurity signal, appears to
be largely temperature independent, with a value considerably
Conclusion
Sodium hydride affords the metastable Ni(I) oxide LaNiO
2
4
6
in a solid-state metathesis reduction reaction, allowing char-
acterization of the (defect) structure and electronic properties
of this phase. The predominantly lamellar structure is interrupted
by defective regions where interlamellar oxide anions aggregate,
producing five-coordinate metal centers. Electronically, the d9
Ni(I) two-dimensional sheets appear to behave quite differently
from their isoelectronic Cu(II) counterparts, with no antiferro-
1
less than that expected for an ideal S ) /2 spin-only paramagnet.
This value of the susceptibility must be considered as an upper
limit on that of bulk LaNiO2.03, in view of the possible
complications from superparamagnetic nickel particles, and thus
implies that the two-dimensional NiO2 sheets are not simply
1
magnetic long-range order, despite evidence that the S ) /2
centers are strongly coupled. Isolation of the metastable lamellar
Ni(I) phase is possible because the more powerful NaH(s)/
1
layers of uncoupled S ) /2 spins. One possible interpretation
of the small temperature dependence of the paramagnetic
susceptibility is that LaNiO2.03 is metallic. The measured
resistivity is, however, several orders of magnitude higher than
is usually found for metallic oxides. A model of strongly
NaOH(s) couple reduces LaNiO at lower temperatures than
3
the H (g)/H O(l) couple. There appears to be no reason the use
2
2
of solid hydrides cannot be extended to the topotactic reduction
of a wide range of solid-state materials. As reagents they are
easy to handle and readily available, giving access to materials
which cannot be synthesized by existing routes. In this specific
instance the low decomposition temperature of NaH (210 °C)
limits its use, as above this temperature a significant pressure
of hydrogen is present, potentially preventing the formation of
highly metastable intermediates. There are, however, a range
of other solid hydrides with much higher decomposition
1
antiferromagnetically coupled localized S ) /2 centers is thus
preferable, although the nature of the transition between regions
b and c in Figure 11ii, at 150 K, remains to be clarified, as
neutron powder diffraction data indicate the absence of long-
range magnetic order in LaNiO2.03.
The electronic properties are in contrast with those of undoped
Cu(II) oxides, which are universally found to be antiferro-
magnetically ordered. The absence of such long-range order
here, unambiguously shown by neutron powder diffraction,
has two possible causes. The slight oxidation beyond Ni(I)
in the defective regions of the structure may introduce sufficient
carriers to suppress ordering (only 3% of Cu(III) centers
are required to totally suppress antiferromagnetic order in
4
0
temperatures (LiH, ∼550 °C; CaH , ∼885 °C ) which could
2
be readily employed to perform solid-state reductions over a
wide temperature range.
Acknowledgment. We thank the donors of the Petroleum
Research Foundation, administered by the American Chemical
Society, and the EPSRC for support of this work. We thank
Dr. P. G. Radaelli for his expert assistance with the collection
of powder neutron diffraction data at the Institut Laue Langevin,
Grenoble, France.
4
5
La2-xSrxCuO4 ) throughout the bulk phase. Alternatively, the
reduced covalency between the Ni 3d and O 2p orbitals in the
nickelates significantly reduces the strong antiferromagnetic
superexchange associated with the close Cu 3d-O 2p energy
match in the cuprates. The large Weiss constants and significant
temperature-independent terms required to model the paramag-
netic susceptibility suggest that this strong spin-spin coupling
is retained, however, and that the observed behavior results from
Supporting Information Available: Figure S1, a Rietveld
profile refinement for the “single phase” fit to the neutron
powder diffraction data, Figure S2, thermogravimetric and
differential scanning calorimetric plots for the reoxidation of
LaNiO2+x, and Table S1, strain and particle sizes evaluated from
the refined profile coefficients of the two-phase fit to the neutron
powder diffraction data. This material is available free of charge
via the Internet at http://pubs.acs.org.
8
suppression of long-range order by the d centers introduced
on partial oxidation. The preferred interpretation of the neutron
and spin susceptibility data is that strong antiferromagnetic
1
coupling between the S ) /2 centers gives rise to a transition
to a short-ranged ordered or frozen state below 150 K, three-
JA991573I
(
45) Budnick, J. T.; Chamberland, B.; Yang, D. P.; Niedermayer, C.;
Golnik, A.; Recknagel, E.; Rossmanith, M.; Weidinger, A. Europhys. Lett.
988, 5, 651-656.
(46) Treece, R. E.; Gillan, E. G.; Kaner, R. B. Comments Inorg. Chem.
1995, 16, 313.
1