Structure of Zr2ON2 by neutron powder diffraction: The absence of nitride-oxide ordering

Article abstract of DOI:10.1006/jssc.1999.8372

The structure of the fluorite-related zirconium oxynitride Zr₂ON₂ has been refined against a combination of laboratory X-ray and time of flight neutron powder diffraction data. The structure is of the bixbyite (C-M₂O₃) type. The high symmetry of the powder diffraction pattern does not allow us to easily distinguish between fully ordered, fully disordered, or partially ordered anion models. However, Rietveld refinements against both high and medium resolution neutron powder diffraction data clearly favour a statistical distribution of oxide and nitride anions over a single crystallographic site which is contrary to the predictions of previous calculations. The space group is Ia3 (No. 206), with a = 10.13940(7) Aat 298 K and Z = 16. This symmetry and unit cell are retained at low temperatures (a = 10.1250(1) A at 4.5 K). Partial oxidation reveals the existence of a phase with N₂ molecules weakly bound and the stoichiometry ZrO₂(N₂)_0.028(1). We also report the isostructural hafnium oxynitride Hf₂ON₂ (Ia3 with a= 10.0692(2) A at 298 K as determined using laboratory X-ray diffraction data).

Full text of DOI:10.1006/jssc.1999.8372

Journal of Solid State Chemistry 146, 399}405 (1999)
Article ID jssc.1999.8372, available online at http://www.idealibrary.com on
Structure of Zr2ON2 by Neutron Powder Diffraction:
The Absence of Nitride}Oxide Ordering
S. J. Clarke,*ꢀꢁ C. W. Michie,- and M. J. Rosseinsky-ꢀꢁ
*School of Chemistry, University of Exeter, Stocker Road, Exeter, EX4 4QD, United Kingdom; and -Inorganic Chemistry Laboratory,
University of Oxford, South Parks Road, Oxford, OX1 3QR, United Kingdom
Received April 9, 1999; accepted May 10, 1999
distributed and the ionic conductivity is similar to that of
The structure of the 6uorite-related zirconium oxynitride
Zr2ON2 has been re5ned against a combination of laboratory
X-ray and time of 6ight neutron powder di4raction data. The
structure is of the bixbyite (C-M2O3) type. The high symmetry of
the powder di4raction pattern does not allow us to easily distin-
guish between fully ordered, fully disordered, or partially ordered
anion models. However, Rietveld re5nements against both high
and medium resolution neutron powder di4raction data clearly
favour a statistical distribution of oxide and nitride anions over
a single crystallographic site which is contrary to the predictions
of previous calculations. The space group is Ia3 (No. 206), with
yttria doped zirconia at 10003C (8). Thompson (9) suggested
that, in addition to anion/vacancy ordering, there should be
O/N ordering in zirconium oxynitrides. This has not, how-
ever, been con"rmed experimentally. Gilles et al. (2)
described three new phases: Zr ON , Zr O N , and
ꢂ
ꢂ
ꢆ ꢇ ꢃ
Zr O N . The latter two phases, designated b and bꢁ,
ꢆ ꢁꢁ ꢂ
respectively, have rhombohedral superstructures of the
#uorite structure. The b-phase, Zr O N , is built up of
ꢆ ꢇ ꢃ
Bevan clusters which are formulated Zr O N and which
ꢆ ꢇ ꢃ
contain ordered anion vacancies. The bꢁ-phase, Zr O N ,
ꢆ ꢁꢁ ꢂ
a ؍ 10.13940(7) A at 298 K and Z ؍ 16. This symmetry and unit is composed of Bevan clusters and Zr O units alternating
ꢆ ꢁꢃ
cell are retained at low temperatures (a ؍ 10.1250(1) A at 4.5 K).
Partial oxidation reveals the existence of a phase with N2 mol-
ecules weakly bound and the stoichiometry ZrO2(N2)0.028(1). We
also report the isostructural hafnium oxynitride Hf2ON2 (Ia3
with a ؍ 10.0692(2) A at 298 K as determined using laboratory
X-ray di4raction data). ( 1999 Academic Press
along the [001] direction in the hexagonal setting of the unit
cell. The Zr O clusters are related to the Bevan clusters,
ꢆ ꢁꢃ
by "lling of the anion vacancies. Lerch and co-workers
(4, 5, 10) have identi"ed a third b type phase designated bꢂ
and with stoichiometry Zr O
mal nitridation at 19003C. This phase contains four Bevan
N
obtained by carbother-
ꢆ ꢈꢉꢂ ꢄꢉꢂ
clusters stacked along the [001] direction separated by
a Zr O unit. The issue of O/N ordering in the Bevan
clusters and the incommensurate nature of some of the
INTRODUCTION
ꢆ ꢁꢃ
b-phase structures, particularly the bꢂ-phase (5), has not
been fully addressed in zirconium oxynitrides. Here we
address the question of O/N ordering in the most nitride-
rich zirconium oxynitride Zr ON , designated the c-phase
One approach to the low-temperature stabilisation of the
high-temperature form of ZrO with the #uorite structure
ꢂ
and high oxide ion mobility is aliovalent substitution of
Zrꢃ> by Yꢄ> or Caꢂ> (1). As well as stabilising the #uorite
structure, this introduces anion vacancies enhancing the
oxide ion conductivity. A second approach is to make an
anion substitution of Oꢂ\ by Nꢄ\ to make a series of
zirconium oxynitrides which will have anion vacancies ac-
ꢂ
ꢂ
which has the cubic C-M O bixbyite structure which is
ꢂ ꢄ
also #uorite related (11). Neutron di!raction is usually
ideally suited to the question of O/N ordering as the two
anions have very di!erent scattering lengths: N"9.36 fm,
O"5.81 fm. However in this case the high symmetry of the
di!raction pattern makes it di$cult to distinguish a cubic
model in Ia3 with a single anion site from an alternative
orthorhombic model in Ibca with partial or full anion order
over three sites and a metrically cubic unit cell. Our Rietveld
re"nements suggest that there is no ordering between oxide
and nitride anions. This is in contrast to the conclusion of
other workers (6) reached using the results of Madelung
energy calculations carried out on a structural model ob-
tained from X-ray powder di!raction measurements.
cording to ZrO
N
<
and represent a solid solution
ꢂ\V ꢂVꢅꢄ Vꢅꢄ
between ZrO and Zr N . This was originally explored by
ꢂ
ꢄ ꢃ
Gilles et al. (2) and has received more attention in the past
few years, particularly from Lerch and co-workers (3}6).
These materials have ordered anion vacancies at low tem-
peratures. Ohashi et al. (7) showed that the ionic conductivi-
ties are very low when there is anion/vacancy ordering. At
high temperatures, however, the vacancies are randomly
ꢁTo whom correspondence should be addressed.
399
0022-4596/99 $30.00
Copyright ( 1999 by Academic Press
All rights of reproduction in any form reserved.

400
CLARKE, MICHIE, AND ROSSEINSKY
EXPERIMENTAL
lected in the 2h range 5 to 1103 with a step size of 0.023 and a
count time of 7.7 seconds per point. High resolution X-ray
powder di!raction data were obtained using a Siemens
D5000 di!ractometer operating in Bragg}Brentano ge-
Synthesis
Synthesis was carried out using the method of Gilles et al.
(2). Low hafnia grade zirconia (ZrO , 99.99%, Aldrich
Chemical Co.) was spread along the length of a 4 cm long
ometry with monochromatic CuKa radiation (Ge(111)
ꢂ
ꢁ
monochromator) and a scintillation counter detector. Soller
alumina boat in a 22 mm internal diameter silica tube. The
tube was placed inside a split tube furnace and ammonia
slits placed before and after the sample combined with
a 0.1 mm detector slit improved the resolution so that the
peak widths were determined by the sample. The instrumen-
tal resolution in this con"guration was approximately 0.053
full width at half maximum, as determined using a silicon
standard.
(NH , 99.98%, British Oxygen Co.) used as supplied, was
ꢄ
#owed at a rate of approximately 12 dmꢄ hr\ꢁ through the
tube. The temperature was raised to between 950 and
10003C at a rate of 73C min\ꢁ for periods of between 24 and
48 hours whence the #ow tube was isolated from the ammo-
nia #ow and removed from the furnace so that the sample
cooled to room temperature in a few minutes under a static
atmosphere of ammonia. The #ow tube was constructed so
that there was no exposure of the sample to air while the
sample was hot. When cool, the material is not air-sensitive
and was handled in air. The phase purity was monitored
using X-ray powder di!raction. During the reaction with
#owing ammonia, the more nitride-rich phases form prefer-
entially at the upstream end of the sample where the partial
pressure of ammonia is greatest and the partial pressure of
water is lowest. Thus, the 2 cm long zone of sample at the
upstream end of the boat consisted of mainly the c-phase,
while the 2 cm zone at the downstream end consisted mainly
of b-phases. It was necessary to regrind and reheat the
sample 10 times in order to remove entirely the white
oxide-rich b-phases and achieve purely the lemon-yellow
c-phase along the whole length of the sample. Although Zr'4
is quite resistant to reduction under these conditions, pro-
longed treatment at temperatures greater than 10003C led
to partial reduction of the sample at the extreme upstream
Neutron Powder Diwraction
Time of #ight data were collected on 0.6 g of material using
the di!ractometer POLARIS at the ISIS Facility, Rutherford
Appleton Laboratory. We used ꢄHe tube detector banks at
35 and 1453 and the ZnS scintillation detector bank at 903 to
access an overall d-spacing range between 0.5 and 5 A. The
measurement was made for an integrated proton current at
the production target of 560 lAhr at room temperature and
288 lAhr at 4.5 K (counting times of 3 and 1.5 hours, respec-
tively). The sample was contained in a cylindrical 5 mm
diameter, thin walled vanadium can sealed under 1 atm of He
with an indium gasket. The 4.5 K data were obtained in an
&&ILL orange'' cryostat. Data were also collected at room
temperature using the high resolution powder di!ractometer
(HRPD) at ISIS on 0.6 g of material for an integrated proton
current of 70 lAhr (2.5 hours). HRPD data were collected
using the back-scattering bank (ZnS scintillation counter) at
2h"1683 in the d-spacing range 0.85 to 1.9 A. Rietveld
re"nement of X-ray and neutron data was carried out using
the general structure analysis system (GSAS) (13). We carried
out Rietveld re"nements against the HRPD data, against the
laboratory X-ray data, and against all three POLARIS banks
simultaneously. A combined re"nement of the model against
X-ray and POLARIS data simultaneously was weighted very
heavily toward the neutron data and gave a result which was
indistinguishable within error from the re"nement against
POLARIS data only.
end of the #ow and the formation of a blue-grey Zr N phase
V
(12) (x+0.9) with the rock-salt structure (lattice para-
meter"4.545(5) A) which was removed mechanically. This
restricted us to the synthesis of 0.6 g of Zr ON which was
ꢂ
ꢂ
phase pure by X-ray powder di!raction and which was used
in this study. Subsequently we found that re-oxidation of
a mixture of zirconium oxynitrides in air at 8003C results in
very "nely divided ZrO which can be transformed to
ꢂ
Zr ON at temperatures as low as 8003C under #owing
ꢂ
ꢂ
ammonia and thus without reduction of Zr'4 at the up-
stream end.
RESULTS
Chemical Analysis
Chemical analysis was carried out by combustion and by
full oxidation back to ZrO in air using a Rheometric
Scienti"c 1500H thermogravimetric balance.
Chemical analysis by combustion indicated a 12.2(1)% by
mass nitrogen content which corresponds to a stoichio-
ꢂ
metry of Zr O
ꢂ ꢁꢉꢊꢄ(ꢂ) ꢁꢉꢈꢇ(ꢂ)
indicated a composition of Zr O
ꢂ ꢁꢉꢊꢈ(ꢂ) ꢁꢉꢈꢃ(ꢂ)
N
. Thermogravimetric analysis
N
.
X-Ray Powder Diwraction
Measurements were made using a Philips PW 1050/81
di!ractometer operating in Bragg}Brentano geometry with
Structure
Fuglein et al. (6) synthesised Zr ON by solid state reac-
CuKa /a radiation and equipped with a di!racted beam
ꢁ ꢂ
ꢂ
ꢂ
monochromator. Data for Rietveld re"nement were col- tion between Zr N (99% obtained by ammonolysis of
ꢄ ꢃ

SYNTHESIS AND STRUCTURE OF Zr ON
401
ꢂ
ꢂ
ZrCl (14)) and ZrO (Alfa 99.9%) at 7003C. They re"ned
ꢃ
ꢂ
ꢂ
the structure of Zr ON using X-ray powder di!raction
ꢂ
data and con"rmed that the material is of the bixbyite type,
isostructural with Tl O (15) and a-Mn O (16). These
ꢂ ꢄ
ꢂ ꢄ
authors were unable to distinguish between nitride and
oxide experimentally, but used the results of MAPLE (17)
calculations to suggest possible reduction of symmetry from
cubic Ia3 with disordered anions to orthorhombic Ibca with
an almost metrically cubic unit cell and partially or fully
ordered anions. The authors point out the similarity of the
simulated neutron di!raction patterns of the ordered and
disordered models. This is due to the fact that if orthorhom-
bic, the cell must have axes of nearly equal length, and the
resultant peak overlap obscures di!erences in the intensities
of individual re#ections in the cubic and orthorhombic
models. Our simulations indicated that there are very small
di!erences in the di!raction patterns at high d-spacing. We
obtained an excellent agreement with all data sets using the
FIG. 1. The result of Rietveld re"nement of Zr ON against
POLARIS data from the 1453 detector bank collected at 298 K showing
the data (dots), "t (line), and di!erence pro"le.
ꢂ
ꢂ
fully disordered model in Ia3 and the results of the X-ray disordered model. The agreement factors of these re"ne-
and neutron re"nements at 298 K and the neutron re"ne- ments were also larger for the ordered model (sꢂ"3.52 and
ment at 4.5 K are presented in Table 1. The "t to the R "0.020 compared with values of 2.91 and 0.018, respec-
ꢁꢂ
POLARIS data from the 1453 data bank is shown in Fig. 1. tively, for the disordered model re"ned against the same
However, model independent (Le Bail type (18)) "ts to the data). Second, we started from a fully disordered orthor-
data were fairly similar in Ia3 and Ibca, (goodness of "t, sꢂ hombic model and allowed a degree of order to be introduc-
of 3.20 and 4.03, respectively) indicating that it might be ed by carrying out an unconstrained re"nement of the
di$cult to distinguish the two models. Starting from the scattering length of each anion site. If the re"nement was
cubic Ia3 model re"ned against the 1453 POLARIS bank carried out so that the changes in scattering length were
with a sꢂ of 2.91, we reduced the symmetry to Ibca (atom strongly damped, the re"nement was stable and the re"ned
positions in Table 2) and calculated sꢂ to be 4.05 prior to values of the scattering lengths on each of the three anion
re"nement of this model. An initial attempt to re"ne this sites had chemically sensible values. However, the large
orthorhombic model was carried out assuming full ordering standard deviation in these parameters rendered the re"ned
of the anions over the three available sites. However, re"ne- composition of Zr O
N
almost meaningless. These
ꢂ ꢁ(ꢂ) ꢂ(ꢂ)
ment of atomic positions or the lattice parameters or both standard deviations were at least two orders of magnitude
resulted in unstable, nonconvergent re"nements with stan- larger than in the disordered case. Attempts to re"ne atomic
dard deviations in the atomic positions and lattice para- positions, and subsequently the lattice parameters also, re-
meters which were between one and two orders of sulted in unstable nonconvergent re"nements with very
magnitude greater than those of the same parameters in the large standard deviations in the structural and cell para-
meters.
The atomic positions and zirconium-anion distances for
TABLE 1
the disordered Ia3 model of Zr ON re"ned against the
ꢂ
ꢂ
Results of X-Ray and Neutron Re5nements of Zr2ON2
Radiation
Instrument
Temperature/K 298
Formula
X-ray
Neutron
Neutron
Neutron
POLARIS
4.5
TABLE 2
Philips PW1050/81 POLARIS HRPD
Atomic Positions Appropriate to Ordered Anion Model in Ibca?
298
298
Zr ON
Zr ON
Ia3
Zr ON
Ia3
Zr ON
ꢂ
ꢂ
ꢂ
ꢂ
ꢂ
ꢂ
ꢂ
ꢂ
Atom
Site
x
y
z
Space group
Ia3
Ia3
Formula weight 226.5
226.5
226.5
226.5
Zrl
8a
8c
8d
0
0
0
0
a/A
Volume/Aꢄ
o /kgm\ꢄ
#!ꢀ#
No. of variables 20
10.1415(1)
1043.04(4)
5.768(1)
10.13940(7) 10.13277(4) 10.1250(1)
1042.40(2) 1040.36(1) 1037.96(4)
5.772(1)
49
Zr2a
Zr2b
Zr2c
O/Na
O/Nb
O/Nc
0.2832
ꢁ
ꢃ
ꢁ
0.2832
0
5.783(1)
26
1.15
0.122
0.0975
0.0772
5.797(1)
60
1.95
0.0113
0.0181
0.0451
ꢃ
8e
0
ꢁ
0.2832
0.093
0.129
0.357
ꢃ
16f
16f
16f
0.357
0.093
0.129
0.129
sꢂ
0.94
3.17
0.357
0.093
R
0.109
0.0765
0.0326
0.0272
0.0436
0.0212
ꢁꢂ
R
ꢂ
R(Fꢂ)
?From Ref. (6).

402
CLARKE, MICHIE, AND ROSSEINSKY
TABLE 3
tion obtained by chemical analysis. However re"nement of
Atomic Positions for Zr2ON2
the composition had a very small e!ect on sꢂ. We con"rmed
that there were no anions on the vacancy sites in this sample
(the fractional occupancy of an atom placed on the vacant
site at 0.13, 0.13, 0.13 re"ned to 0.00(2)). All three elements
are strong coherent scatterers of neutrons and the neutron
powder di!raction data have a better signal to noise ratio
than the X-ray data, especially at small d-spacing because of
a smaller fall-o! of the nuclear scattering factor with de-
creasing d-spacing. Thus, as well as being weighted much
more heavily than the X-ray data in a combined re"nement,
the neutron data gave much greater precision in the struc-
tural parameters.
Data type
Instrument
Temperature/K
X-ray
Philips PW1050
298
Neutron
POLARIS
298
Neutron
POLARIS
4.5
0
0
Zr1
x
y
z
0
0
0
0
0
0
0
100;; /Aꢂ
0.75(5)
0.2844(6)
0
0.65(1)
0.28465(4)
0
0.59(4)
0.2846(1)
0
*ꢃꢄ
Zr2
x
y
z
0.25
0.25
0.25
100;; /Aꢂ
0.44(3)
0.3583(4)
0.1304(5)
0.0968(4)
0.2(1)
0.44(1)
0.35813(3)
0.13008(3)
0.09678(3)
0.53(1)
0.33(4)
0.35806(8)
0.12987(9)
0.09675(9)
0.36(4)
*ꢃꢄ
O/N? x
y
z
A similar series of re"nements was carried out on HRPD
data. Although the counting statistics were poorer in this
case the disordered model again re"ned stably (sꢂ"1.15,
100;; /Aꢂ
*ꢃꢄ
?Oxygen and nitrogen atoms with 1/3 and 2/3 occupancy, respectively, R "0.122). The orthorhombic ordered model did not
ꢁꢂ
re"ne stably and the estimated standard deviations were
occupied a single site.
a factor of 10 larger than in the disordered case. Thus, all
our Rietveld re"nements strongly favour the disordered
three POLARIS banks simultaneously are presented in model with cubic symmetry. In addition, "tting of indi-
Tables 3 and 4, respectively. Re"nement of anisotropic dis- vidual HRPD re#ections using the program PKFIT1Vb
placement parameters for all three atoms indicated that the (20) indicated that they could all be "t using a single peak.
atoms were approximately isotropic at 298 K and 4.5 K. This is strongly indicative of the cubic disordered model
A Debye}Scherrer absorption correction using the func- especially in view of the extremely high resolution of HRPD
tional form due to Hewat (19) made a signi"cant improve- (*d/d for this sample on HRPD was a constant 1.8;10\ꢄ
ment to the "t at long time of #ight. If the cation and anion compared with a constant 4.2;10\ꢄ on POLARIS). A sim-
sites are all fully occupied, the composition must be Zr ON
ilar treatment of the high resolution D5000 X-ray data
ꢂ
ꢂ
because the Zr oxidation state is #4. We carried out an (*d/d+2;10\ꢄ) indicated that the full-width at half-max-
unconstrained re"nement of the scattering length of the imum (FWHM) of the peaks was less than the spread of the
single occupied anion site; i.e., there was no requirement centers of the re#ections contributing to a di!raction peak
that the site should be fully occupied or that the oxidation at high angle if the best orthorhombic re"nement was as-
state of Zr should be #4. This indicated a composition of sumed. In both the HRPD and D5000 high resolution
Zr O
N
, which is consistent with the composi- di!raction data, re#ections such as (044) (Fig. 2) and (004),
ꢂ ꢊꢉꢈꢈ(ꢁ) ꢂꢉꢊꢄ(ꢁ)
TABLE 4
Interatomic Distances in A and Angles in Degrees for Zr2ON2
Zr1}O/N ;6 2.1845(4)
O/N}Zr1}O/N ;3 180
O/N}Zr1}O/N ;6
O/N}Zr1}O/N ;6
99.758(8)
80.242(8)
Zr2}O/N ;2 2.1497(4)
Zr2}O/N ;2 2.1698(4)
Zr2}O/N ;2 2.2345(4)
O/N}Zr2}O/N ;1 164.50(2)
O/N}Zr2}O/N ;1 139.84(2)
O/N}Zr2}O/N ;1 114.90(3)
O/N}Zr2}O/N ;2 109.76(2)
O/N}Zr2}O/N ;2
O/N}Zr2}O/N ;2
O/N}Zr2}O/N ;2
O/N}Zr2}O/N ;2
O/N}Zr2}O/N ;2
Zr1}O/N}Zr2 ;1 124.82(2)
Zr1}O/N}Zr2 ;1 100.41(1)
Zr1}O/N}Zr2 ;1
99.36(1)
86.39(2)
79.68(1)
79.45(2)
79.25(1)
O/N}Zr1 ;1 2.1845(4)
O/N}Zr2 ;1 2.1497(4)
O/N}Zr2 ;1 2.1698(4)
O/N}Zr2 ;1 2.2345(4)
98.41(1)
FIG. 2. The (044) re#ection (d"1.792 A) measured on HRPD and
"tted to a single peak using the program PKFIT1Vb (20). This shows that
the cubic model in Ia3 is correct since no reduction in symmetry is
indicated. d-spacing/A"time of #ight/ms;0.0207 for HRPD.
Zr2}O/N}Zr2 ;1 126.51(2)
Zr2}O/N}Zr2 ;1 100.21(2)
Zr2}O/N}Zr2 ;1
99.59(1)

SYNTHESIS AND STRUCTURE OF Zr ON
403
ꢂ
ꢂ
which should be split in orthorhombic symmetry, were no
broader than the (222) re#ection which should be a single
peak in both cubic and orthorhombic symmetry. In per-
forming this analysis account was taken of the decrease in
instrumental resolution with increasing 2h for the D5000
X-ray di!ractometer which was determined using a silicon
standard. Thus, there is no direct evidence from the highest
resolution di!raction data available that the cell is orthor-
hombic rather than cubic and this is consistent with our
re"nement results. If the symmetry is cubic Ia3 then the
anions must be statistically distributed over a single site
since the orthorhombic Ibca model is the highest symmetry
one which allows ordering for the size of unit cell under
consideration, and there are no superstructure re#ections to
indicate that a larger cell should be considered.
Oxidation
FIG. 4. The reduction in symmetry of the bixbyite structure from Ia3
to R3 as a result of "lling up some of the vacant anion sites. The X-ray
powder di!raction pattern of material with composition Zr O N pre-
Thermogravimetric analysis (Fig. 3) indicates that oxida-
tion to ZrO is not complete until 12003C. The mass of the pared at 7203C (upper) is compared with that of Zr ON (lower). Indexing
sample at temperatures above 6003C is greater than that of
the "nal ZrO oxidation product. The partially oxidised
ꢆ
ꢇ ꢃ
ꢂ
ꢂ
ꢂ
refers to a hexagonal cell in R3 with a"14.3831(4) A and c"17.5889(8)
A for Zr O N and a cubic cell in Ia3 with a"10.13951(2) A for Zr ON .
ꢆ
ꢇ
ꢃ
ꢂ
ꢂ
ꢂ
The data have been normalised so that the maximum intensities of the
material, cooled from 6503C in the TGA, is a poorly crystal-
peaks at about 30.53 are equal at 40,000 counts per 30 s.
line phase with the baddeleyite (monoclinic ZrO ) structure.
ꢂ
The source of the additional mass is N molecules which are
ꢂ
weakly bonded in the structure, presumably occupying the
anion sites. The Raman spectrum (inset to Fig. 3) measured
on a Dilor Labram spectrometer at a frequency of
632.82 nm showed a fairly strong resonance at 2329(1) cm\ꢁ
Solid Solution in the c-Phase
Reaction of "nely divided ZrO with Zr ON as a cold-
pressed pellet in a sealed evacuated silica tube at between
700 and 7503C with the composition of the b-phase,
ꢂ
ꢂ
ꢂ
which is similar to the N vibrational frequency in gaseous
ꢂ
or liquid N and con"rms that N moieties are present in
ꢂ
ꢂ
Zr O N produced a single phase which was clearly related
the solid. Thermogravimetric analysis suggests a composi-
tion of ZrO (N ) at 6503C. This nitrogen retention
ꢆ ꢇ ꢃ
structurally to the c-phase Zr ON . This result is in keeping
ꢂ
ꢂ
ꢂ
ꢂ ꢊꢉꢊꢂꢇ(ꢁ)
with the phase diagrams of Ohashi (7) and Lerch (8). This
material, which is pale green in color, has the same-sized cell
as Zr ON but with lower symmetry. The symmetry of the
phenomenon has been observed previously in the oxidation
of several oxynitrides (21). We have recently found that
substitution of other metals, notably Ti, for Zr in the
bixbyite structure oxynitrides leads to a 10-fold increase in
ꢂ
ꢂ
Zr ON structure is reduced from Ia3 to R3 as a result of
ꢂ
ꢂ
"lling some of the anion vacancies. This results in some
additional weak superstructure re#ections as shown
in Fig. 4. The hexagonal lattice parameters are a"
14.3831(4) A and c"17.5889(8) A. The structural details of
this anion rich c-phase are under investigation.
the amount of N retained on oxidation (22).
ꢂ
Hf ON
ꢂ
ꢂ
The greenish-grey hafnium analogue (23) was prepared in
a similar manner to Zr ON . HfO containing 1% ZrO
ꢂ
ꢂ
ꢂ
ꢂ
(Aldrich Chemical Co.) was reacted with ammonia at 9503C
for 3 days and 9803C for 2 days with intermediate regrind-
ing. In contrast to the zirconium case, there was no evidence
for intermediate b-phases during synthesis. The structure
was re"ned from laboratory X-ray data and showed
Hf ON to be isostructural with Zr ON , but with a slight-
FIG. 3. The results of thermogravimetric analysis of Zr ON showing
ꢂ
ꢂ
that N is retained during oxidation. The N stretch measured by Raman
ꢂ
ꢂ
ꢂ
ꢂ
ꢂ
ꢂ
spectroscopy is shown as an inset.
ly smaller cell parameter of 10.0692(2) A. The re"nement is

404
CLARKE, MICHIE, AND ROSSEINSKY
TABLE 6
Atomic Positions for Hf2ON2
Data type
X-ray
Temperature/K
298
0
0
Hf1
x
y
z
0
100;; /Aꢂ
0.98(6)
0.28385(7)
0
*ꢃꢄ
Hf2
x
y
z
0.25
100;; /Aꢂ
0.52(3)
*ꢃꢄ
O/N?
x
0.3584(7)
0.131(1)
0.0954(7)
0.5 (not re"ned)
y
z
FIG. 5. The result of Rietveld re"nement of the structure of Hf ON
ꢂ
ꢂ
100;; /Aꢂ
against X-ray data at 298 K showing the data (dots), "t (line), and di!erence
pro"le. Re#ections due to the aluminum sample holder have been excluded
from the re"nement.
*ꢃꢄ
shown in Fig. 5, the re"nement results are given in Table 5, tetrahedron described by the other four anions. The iso-
and the atomic positions in Table 6. The Hf1-anion distance tropic displacement parameter ; for Zr1 at 298 K is
*ꢃꢄ
is 2.175(1) A and the three Hf2-anion distances are 2.107(1), somewhat larger than that of Zr2, but this di!erence is
2.182(1), and 2.217(1) A.
entirely in keeping with the result obtained for Tl O (15)
and presumably re#ects the slightly smaller size of the Zr2
ꢂ ꢄ
DISCUSSION
site compared to the Zrl site.
The results of MAPLE calculations (6) on the ordered
The structure of Zr ON is shown in Fig. 6. It is isostruc- Ibca and disordered Ia3 models indicated a stabilisation of
ꢂ
ꢂ
tural with a-Mn O (16) and Tl O (15). The interatomic 0.8% for the ordered model relative to the disordered
ꢂ ꢄ ꢂ ꢄ
distances and angles are listed in Table 4. There are two Zr model. Our results show, in contrast, that there appears to
sites which are shown in Fig. 7. Zr1, located at the origin, be no preference for any ordering. MAPLE calculations are
is surrounded by six anions in a distorted octahedral probably not good at distinguishing between the two mod-
arrangement all at a distance of 2.1845(4) A and has an els as they do not take into account the covalent conribu-
isotropic thermal ellipsoid. Zr2 is in a much more distorted tions to the bonding which will increase as the proportion of
environment and is surrounded by six anions in three crys- nitride anions increases. These calculations were based on
tallographically distinct pairs. Two pairs of these are dis- a structural model for Zr ON which was obtained using
ꢂ
ꢂ
posed approximately tetrahedrally about the Zr at distances X-ray powder di!raction data. The precision in the atomic
of 2.1497(4) and 2.1698(4) A. The other pair lie 2.2345(4) A
from Zr2 and are above the centres of two of the faces of the
TABLE 5
Crystallographic and Re5nement Data for Hf2ON2
Data
X-ray
Instrument
Temperature/K
Formula
Formula weight
Space group
a/A
Philips PW1050/81
298
Hf ON
400.99
ꢂ
ꢂ
Ia3
10.0692(2)
1020.90(6)
10.436(1)
19
Volume/Aꢄ
o
/kgm\ꢄ
#!ꢀ#
No. of variables
sꢂ
3.58
FIG. 6. One eighth of the unit cell of Zr ON showing the relationship
R
0.0686
0.0475
0.0266
ꢂ
ꢂ
ꢁꢂ
R
of this structure with that of #uorite. Small spheres are Zr, large spheres are
anions. The ordered anion vacancies are represented by the large open
circles.
ꢂ
R(Fꢂ)

SYNTHESIS AND STRUCTURE OF Zr ON
405
ꢂ
ꢂ
FIG. 7. The coordination geometries about the two Zr sites. Anisotropic displacement ellipsoids are shown at the 99.9% level.
positions is an order of magnitude greater for our measure-
2. J.-C. Gilles and R. Collongues, C. R. Acad. Sci. 254, 1084 (1962).
ment.
Zirconium oxynitrides are interesting in view of the high
3. M. Lerch, H. Boysen, and P. G. Radaelli, J. Phys. Chem. Solids 58, 1557
(1997).
4. M. Lerch, J. Am. Ceram. Soc. 79, 2641 (1996).
5. M. Lerch, F. Krumreich, and R. Hock, Solid State Ionics 95, 87
(1997).
6. E. Fuglein, R. Hock, and M. Lerch, Z. Anorg. Allg. Chem. 623, 304
(1997).
7. M. Ohashi, H. Yamamoto, S. Yamanaka, and M. Hattori, Mater. Res.
Bull. 28, 513 (1993).
8. M. Lerch, J. Mater. Sci. ¸ett. 17, 441 (1998).
9. D. P. Thompson, Proc. 2nd ECRS, Augsburg, 35 (1991).
10. D. Walter, M. Lerch, and W. Laqua, J. ¹hermal Anal. 48, 709
(1997).
oxide ion conductivity which is shown by the yttria and
calcia stabilised zirconias. The higher vacancy concentra-
tion in the oxynitrides could increase anion mobility, but
appears only to do so at high temperatures where both the
anions and the vacancies are disordered (7, 8). In these
circumstances the ionic conductivity is comparable to that
of the calcia and yttria stabilised materials. It is not yet
known whether zirconium oxynitrides with disordered va-
cancies can be stabilised at low temperatures. The dimen-
sion of the #uorite subcell in Zr ON is 5.07 A (half of the 11. A. F. Wells, &&Structural Inorganic Chemistry,'' 5th ed. Clarendon
ꢂ
ꢂ
Press, Oxford, 1984.
12. R. Juza, A. Gabel, H. Rabenau, and W. Klose, Z. Anorg. Allg. Chem.
329, 136 (1964).
13. A. Larson and R. B. von Dreele, &&The General Structure Analysis
System,'' Los Alamos National Laboratory, 1985.
14. M. Lerch, E. Fuglein, and J. Wrba, Z. Anorg. Allg. Chem. 622, 367
(1996).
dimension of the actual bixbyite cell). This is smaller than in
the pure oxides: Zr (24) with the #uorite struc-
Y
O
ꢊꢉꢇ ꢊꢉꢂ ꢁꢉꢈ
ture has a lattice parameter of 5.1473(1)A.
ACKNOWLEDGMENTS
15. H. H. Otto, R. Baltrusch, and H.-J. Brandt, Physica C 215, 205
(1993).
We are grateful to Drs. Richard Ibberson, Ron Smith, and Steve Hull of
ISIS for assistance with the neutron di!raction measurements and to Dr.
Amelia Fowkes for assistance with PKFIT1Vb. MJR thanks the EPSRC
for "nancial support, for access to ISIS, and for a studentship for CWM.
SJC thanks the Royal Society for a research grant and the Lloyd's of
London Tercentenary Foundation and St. Hugh's College, Oxford for
fellowships.
16. S. Geller, Acta Crystallogr. B 27, 821 (1971).
17. R. Hoppe, Angew. Chem. Int. Ed. 5, 95 (1966).
18. A. Le Bail, Mater. Res. Bull. 23, 447 (1988).
19. A. W. Hewat, Acta Crystrallogr. A 35, 248 (1979).
20. D. Sivia, unpublished.
21. L. LeGendre, R. Marchand, and Y. Laurent, J. Eur. Ceram. Soc. 17,
1813 (1997).
22. S. J. Clarke, C. W. Michie, and M. J. Rosseinsky, in preparation.
23. A. Pialoux, J. Nucl. Mater. 200, 1 (1993).
REFERENCES
1. A. R. West, &&Solid State Chemistry and Its Applications,'' p. 478. Wiley, 24. M. Yashima, S. Sasaki, M. Kakihana, Y. Yamaguchi, H. Arashi, and
Chichester, 1984. M. Yoshimura, Acta Crystallogr. B 50, 663 (1994).