Phase formation, structural and microstructural characterization of novel oxynitride-perovskites synthesized by thermal ammonolysis of (Ca,Ba)MoO4 and (Ca,Ba)MoO3

Article abstract of DOI:10.1016/j.jssc.2008.05.012

Reactions of AMoO₄ and AMoO₃ (A=Ca²⁺, Ba²⁺) with ammonia were investigated at 873 K3 and to study their crystal structure. CaMo(O,N)₃ and BaMo(O,N)₃ were prepared by thermal ammonolysis of the corresponding CaMoO₃ and BaMoO₃ precursors at T=898 and 998 K, respectively. The structural parameters of the oxynitrides were obtained from Rietveld refinements of X-ray and neutron powder diffraction data. CaMo(O,N)₃ crystallizes in the GdFeO₃ distorted perovskite structure with orthorhombic space group Pbnm and a=5.5029(1) A, b=5.5546(1) A, c=7.8248(1) A as determined by X-ray powder diffraction. Its O/N content refined from the neutron diffraction data corresponds to the composition CaMoO_1.7(1)N_1.3(1). BaMo(O,N)₃ crystallizes in the cubic perovskite structure with space group Pm3m and a=4.0657(1) A as determined by X-ray powder diffraction. Transmission electron microscopy reveals a complex microstructure for both CaMoO₃ and CaMoO_1.7(1)N_1.3(1) represented by twin domains of different orientation.

Full text of DOI:10.1016/j.jssc.2008.05.012

ARTICLE IN PRESS
Journal of Solid State Chemistry 181 (2008) 2243– 2249
Contents lists available at ScienceDirect
Journal of Solid State Chemistry
journal homepage: www.elsevier.com/locate/jssc
Phase formation, structural and microstructural characterization of novel
oxynitride– perovskites synthesized by thermal ammonolysis of (Ca,Ba)MoO4
and (Ca,Ba)MoO3
a
a
b
c
c
c
a,
Ã
D. Logvinovich , M.H. Aguirre , J. Hejtmanek , R. Aguiar , S.G. Ebbinghaus , A. Reller , A. Weidenkaff
a
Solid State Chemistry and Catalysis, Empa, Ueberlandstr. 129, CH-8600 Duebendorf, Switzerland
b
Institute of Physics, ASCR, Cukrovarnick a´ 10, CZ-15263, Prague 6, Czech Republic
c
Lehrstuhl f u¨ r Festk o¨ rperchemie, Institut f u¨ r Physik, Universit a¨ t Augsburg, Universit a¨ tsstrasse 1, D-86159, Augsburg, Germany
a r t i c l e i n f o
a b s t r a c t
2
+
2+
Article history:
Reactions of AMoO4 and AMoO3 (A ¼ Ca , Ba ) with ammonia were investigated at 873 KoTo1123 K
Received 5 February 2008
Received in revised form
with the particular intention to synthesize novel oxynitride–perovskites of the general composition
AMo(O,N)
ammonolysis of the corresponding CaMoO
The structural parameters of the oxynitrides were obtained from Rietveld refinements of X-ray and
3
and to study their crystal structure. CaMo(O,N)
3 3
and BaMo(O,N) were prepared by thermal
2
6 April 2008
3
and BaMoO
3
precursors at T ¼ 898 and 998 K, respectively.
Accepted 1 May 2008
Available online 16 May 2008
neutron powder diffraction data. CaMo(O,N)
3
crystallizes in the GdFeO
3
distorted perovskite structure
Keywords:
˚
˚
˚
with orthorhombic space group Pbnm and a ¼ 5.5029(1) A, b ¼ 5.5546(1) A, c ¼ 7.8248(1) A as
Oxynitride
determined by X-ray powder diffraction. Its O/N content refined from the neutron diffraction data
Molybdate
corresponds to the composition CaMoO1.7(1)
N
1.3(1). BaMo(O,N)
˚
3
crystallizes in the cubic perovskite
Crystal structure
structure with space group Pm 3¯ m and a ¼ 4.0657(1) A as determined by X-ray powder diffraction.
Transmission electron microscopy reveals
1.3(1) represented by twin domains of different orientation.
2008 Elsevier Inc. All rights reserved.
a
complex microstructure for both CaMoO
3
and
CaMoO1.7(1)
N
&
1
. Introduction
oxides CaMoO
synthesis of the corresponding oxynitrides are available in the
literature. One reported attempt to prepare BaMo(O,N) by
thermal ammonolysis of BaMoO led to a mixture of BaMoO
and Ba Mo N [11].
The lack of Mo-containing oxynitride–perovskite phases
motivated us to investigate whether CaMo(O,N) and BaMo(O,N)
3 3
and BaMoO are known [13–16], no reports on the
3
ꢀ
The interest on N - and N
2
-containing inorganic and metal-
3
organic Mo derivatives is driven by the ability of these compounds
to catalyze molecular nitrogen reduction into ammonia [1,2].
Therefore, oxynitride–perovskites and intermediate products of
their thermal reoxidation [3] are potentially interesting materials
to be tested for this application. Oxyntrides are normally prepared
by thermally activated reactions between ternary oxides like
4
3
3
2 6
O
3
3
can be prepared by thermal ammonolysis and, if yes, to study and
compare their crystal structures with those of the corresponding
ABO
bonates mixtures (B
oxidation state and ammonia [4–7]. Oxide and oxide–carbonate
4
, A
2
B
2
O
7
(A ¼ RE, AE; B ¼ transition element) or oxides–car-
oxides CaMoO
3 3
and BaMoO .
6
+
x
O
y
ꢀA (CO ) with B-site cation in its stable
n
3
)
m
Two different series of oxides, namely AMo O and its reduced
4
4
+
2+
2+
form AMo
O
3
(A ¼ Ca , Ba ), have been chosen as precursors to
4
+
precursors were successfully applied to prepare a number of Ti
,
find out whether the structure of the starting compound has an
influence on the phase formation during the ammonolysis at
different temperatures.
5+
5+
Ta
,
Nb -containing oxynitride–perovskite phases AB(O,N)
3
[
7–10]. The only known molybdenum based oxynitride–perovs-
kite system is SrMoO3ꢀx [9,11,12]. It was prepared by thermal
ammonolysis of SrMoO with Mo in its air-stable formal oxidation
the formal oxidation
N
x
4
state +6, at T ¼ 1023 ꢀ1073 K. In SrMoO3ꢀx
N
x
2. Experimental
state of molybdenum varies between +4 and +6 depending on the
anionic composition. Interestingly, although the perovskite-type
BaMoO
17]. A 0.1 M solution of Ba(NO
was poured slowly with constant stirring into a 0.1 M solution
of Na MoO
(Riedel-de Ha a¨ n, purity 499.5%). The formed
4
was prepared according to the procedure described in
[
3 2
)
(Merck, purity 499.0%)
Ã
Corresponding author.
E-mail address: anke.weidenkaff@empa.ch (A. Weidenkaff).
2
4
0
022-4596/$ - see front matter & 2008 Elsevier Inc. All rights reserved.
doi:10.1016/j.jssc.2008.05.012

ARTICLE IN PRESS
2
244
D. Logvinovich et al. / Journal of Solid State Chemistry 181 (2008) 2243–2249
precipitate was washed several times with deionised water and
calcined at 1073 K during 4 h.
3. Results and discussion
BaMoO
3
was obtained by reduction of BaMoO
4
(1 g) with a
XRPD confirmed phase purity of all the synthesized oxides,
forming gas (5% of H
2
in N ) flow of 300 mL/min. The reduction
2
except CaMoO , for which an Mo impurity of 2.3 weight% was
3
was carried out at T ¼ 1473 K during 15 h in a tubular quartz
refined. Crystallographic parameters of CaMoO3 as obtained by
reactor with an internal diameter of 30 mm.
refinement of the XRPD data in space group Pbnm are summarized
in Tables 1 and 2 and are in good agreement with a previous ND
study of this compound [14]. The observed isotropic line broad-
ening was accounted for by refining isotropic size-strain compo-
nents (in Fullprof referred to as U and Y) of the sample intrinsic
profile. Their refined values correspond to an apparent size value
of 0.13 mm and an apparent strain value of 0.31%.
The refined lattice parameter of BaMoO3 is equal to
4.0409(1) A˚ , which is close to the value of 4.0404(3) A˚ reported
in Ref. [15]. No line broadening was found for this oxide with
respect to the used standard LaB6.
CaMoO
Stoichiometric amounts of H24Mo
and CaCO (Alfa Aesar, 499.5%) were dissolved in an aqueous
.1 M citric acid C
(Riedel-de Ha a¨ n, 499.5%) solution. The
4
was synthesized by the citrate precursor method [14].
7
N
6
O
24*4H O (Fluka, 499.0%)
2
3
0
6 8 7
H O
amount of citric acid was 3 times the total molar amount of Mo
and Ca. The obtained solution was predried at 393 K overnight and
heated up to 873 K within 12 h. The final product was then
heated from 873 K to 1073 K with a heating rate of 5 K/min,
calcined at that temperature during 2 h and cooled down to room
temperature.
CaMoO
the citrate method with forming gas (5% of H
reduction of 1 g of CaMoO was carried out at 1173 K during 12 h
with an intermediate regrinding after 4 and 8 h of the reaction)
3
was synthesized by reduction of CaMoO
4
produced by
Both CaMoO and BaMoO are of a purple-red color. In contact
with moisture they oxidize slowly (within several hours/days)
3
3
2
in N
2
). [14] The
4
into corresponding CaMoO4 and BaMoO . Therefore, they were
4
(
stored under dry nitrogen atmosphere.
under a forming gas flow of 100–300 mL/min in a quartz reactor
with an internal diameter of 30 mm. [18] Forming gas was
supplied through a quartz tube with a diameter of 5.8 mm placed
above the sample and about 2 mm from the reactor end. After the
reaction, the sample was quenched down to room temperature
within 1 min.
Reflections corresponding to Ba Mo O N [26,27] and BaMoO
4
3
2
6
2
are identified for the BaMoO4 sample reacted at 898 K for 20 h.
The BaMoO main reflections disappear upon heating the sample
4
up to 973 K. XRPD confirms Ba Mo O N to be the main phase
3
2 6 2
formed after the ammonolysis at T ¼ 973 K.
All the synthesized oxides were reacted with ammonia gas
Table 1
(
PanGas, 99.999%, 100 mL/min) at 873 KoTo1123 K. The reactions
Refinement results for CaMoO
3
and CaMoO1.7(1)N
1.3(1)
were carried out at T interval of 25 K in an Al reactor with an
2 3
O
internal diameter of 30 mm. Each sample load was 0.1 g. The
reaction time was 20 h. Ammonia was supplied by means of a
quartz tube with a diameter 5.8 mm placed above the sample.
Heating and cooling rates were 10 K/min.
Name
CaMoO
3
CaMoO1.7(1)
N
1.3(1)
CaMoO1.7(1)N
1.3(1)
Radiation
l, A˚
T, K
X-ray CuKa1/2
1.5406/1.5444
298
X-ray CuKa1/2
1.5406/1.5444
298
Neutron (HRPT, PSI)
1.494
298
Structure and phase purity of the starting and reacted powders
were studied by X-ray powder diffraction (XRPD) using a Phillips
X’Pert PRO MPD Y–Y System equipped with a linear X’Celerator
detector. Crystallographic parameters were obtained from
Rietveld refinement of the XRPD data, recorded in the 2Y
angular range of 20–1401 with a step size of 0.021 and a counting
time of 20 s/step. A proportional counter was used to detect
scattered X-rays. The reflection shape was modeled with a
Thompson–Cox–Hastings (TCH) pseudo-Voigt profile function
S,G
a, A˚
b, A˚
c, A˚
V, A˚
Pbnm
Pbnm
Pbnm
5.4499(1)
5.5811(1)
7.7791(1)
236.62(1)
4
5.5029(1)
5.5546(1)
7.8248(1)
239.19(1)
4
5.5068(1)
5.5593(1)
7.8317(2)
239.76(1)
4
3
Z
Ca
x
y
z
0.9894(7)
0.0461(3)
1/4
0.997(1)
0.0294(5)
1/4
0.994(1)
0.0334(9)
1/4
B
iso, A
˚
2
0.46(4)
4c
1.13(3)
4c
1.05(7)
4c
Site
[
19] corrected for asymmetry. [20,21] Employing that function
Occ.
1
1
1
allows to extract the samples intrinsic profile by using the known
instrumental profile and performing a line broadening analysis to
conclude about particle size and strain. [22] The instrumental
Mo
x
y
z
1/2
1/2
1/2
0
0
0
6
profile was obtained from the measurement of LaB SRM 660a
0
0
0
iso, A˚ 2
standard.
B
0.11(2)
0.90(1)
0.74(5)
Site
4b
4b
4b
Neutron diffraction (ND) data were recorded at the high-
resolution powder diffractometer for thermal neutrons [23]
located at the Swiss Spallation Neutron Source (SINQ) of the
Paul Scherrer Institute in Switzerland. Samples were placed
in cylindrical vanadium cans with 6 mm in diameter. The
measurements were performed with a neutron wavelength of
Occ.
1
1
1
O/N(1)
x
y
0.081(1)
0.475(1)
1/4
0.072(1)
0.473(2)
1/4
0.0642(6)
0.4872(7)
1/4
z
Biso, A˚ 2
0.92(7)
4c
1.13(9)
4c
0.90(8)
˚
Site
4c
l ¼ 1.494 A in the angular range of 4.5–1651 with a step size of
a
Occ.
1/0
0.56/0.44
0.56(4)/0.44(4)
0
.051. The reflection shape was modeled using a TCH pseudo-
Voigt profile function. All the refinements were performed using
Fullprof [24].
The O/N content of the different samples was measured by the
hotgas-extraction method using a LECO TC500 analyzer. Silicon
nitride and silicon oxide were used as calibration standards for
nitrogen and oxygen, respectively.
Electron diffraction (ED) studies were done by transmission
electron microscopy (TEM) using a Philips CM30 microscope
operating at 300 kV. Diffraction simulations were performed using
JEMS [25].
O/N(2)
x
y
z
0.7055(8)
0.2921(8)
0.0453(6)
0.92(7)
8d
0.703(1)
0.290(1)
0.0329(9)
1.13(9)
8d
0.7113(5)
0.2885(5)
0.0329(3)
0.90(6)
8d
Biso, A˚
2
Site
a
Occ.
1/0
0.57/0.43
0.57(3)/0.43(3)
1.97
2
w
2.13
1.71
b
p
wR
0.135
0.136
0.128
0.128
b
R
p
0.102
0.149
a
Not refined.

ARTICLE IN PRESS
D. Logvinovich et al. / Journal of Solid State Chemistry 181 (2008) 2243–2249
2245
Table 2
Selected bond distances in A˚ and angles in degrees calculated for CaMoO
and
3
a
CaMoO1.7(1)N
1.3(1)
CaMoO
3
CaMoO1.7(1)
N
1.3(1)
Mo– O/N(1)
Mo– O/N(2)
Mo– O/N(2)
x2
x2
x2
2.000(1)
2.009(4)
2.012(4)
2.007(3)
3.228(6)
2.445(6)
3.138(7)
2.371(7)
2.610(5)
2.735(5)
2.382(5)
3.407(5)
2.520(6)
153.00(6)
151.9(2)
152.27(1)
1.9909(6)
1.997(3)
1.995(3)
1.994(2)
3.061(6)
2.552(6)
3.085(8)
2.446(8)
2.708(6)
2.705(5)
2.454(5)
3.278(5)
2.592(6)
/Mo– O/NS
Ca– O/N(1)
Ca– O/N(1)
Ca– O/N(1)
Ca– O/N(1)
Ca– O/N(2)
Ca– O/N(2)
Ca– O/N(2)
Ca– O/N(2)
x2
x2
x2
x2
b
/Ca– O/NS(8 short)
Mo– O/N(1)– Mo
Mo– O/N(2)– Mo
x2
x4
159.13(3)
157.0(1)
/Mo– O/N– MoS
157.72(8)
a
Derived from the neutron data.
b
The average, calculated from the 8 shortest distances.
Some additional reflections belonging to a phase, indexed as a
cubic perovskite with the lattice parameter larger than that of
BaMoO are also resolved. These reflections are attributed to a
nitrided BaMoO (further referred to as BaMo(O,N) ). The
reflections of this phase disappear after 40 h of the ammonolysis
at T ¼ 973 K; after that only reflections of Ba Mo can be
resolved. In the XRPD pattern of the sample which was reacted
with ammonia at 1023 K Mo N and Ba Mo are the main
3
3
3
Fig. 1. Evolution of the XRPD profile with time measured during the ammonolysis
of CaMoO
3
at T ¼ 898 K. (A) Non-reacted sample; (B) The sample after the 50 h
3
2 6 2
O N
reaction (the inset shows the splitting of the main reflections due to the non-
uniform O/N distribution through the particles depth); (C) The sample after the
100 h reaction.
2
3
2 6 2
O N
identified phases. Two weak reflections, which belong to an
unidentifiable phase, appear between the (015) and (110)
reflections of Ba
main phase from BaMoO
The reaction between CaMoO
The XRPD patterns reveal the appearance of broad features
coincident with the main reflections of Mo N [28]. At 898 K CaO
and Mo N can be identified. Some additional reflections that
could not be assigned to any known phase appear at that
temperature. The XRPD pattern of the sample reacted with NH
3
Mo
2
O
6
N
2
. Thus, BaMo(O,N)
under the chosen reaction conditions.
and ammonia starts at 873 K.
3
is not formed as a
The sample batch used for the crystallographic investigation by
Rietveld refinements was obtained by ammonolysing 0.7 g of
4
4
CaMoO
3
at T ¼ 898 K during 100 h (when the main reflections
splitting disappeared and no further change in the lattice
parameters was detected by XRPD) under an ammonia
flow of 300 mL/min. After the reaction the sample was quenched
to room temperature within 1 min. Increasing the synthesis
temperature to T4898 K led to the oxynitride phase with
lower nitrogen content as deduced from the Bragg reflection
positions measured by XRPD. The explanation for this is given
further in the text.
2
2
3
at higher temperatures reveals the presence of the CaO and Mo
Thus, CaMo(O,N)
2
N.
3
cannot be prepared directly from CaMoO
4
under the chosen reaction conditions.
A noticeable interaction between CaMoO
3
(Fig. 1A) and NH
3
3
The ammonolysis of BaMoO (Fig. 2A) proceeds similar to that
starts at 898 K. The ammonolysis results in a perovskite-type
phase with cell parameters deviating from the ones of the starting
oxide. Its formation can best be monitored by inspection of the
of CaMoO . The temperature had to be raised up to 998 K and the
sample mass had to be decreased down to 0.25 g in order to
promote a faster formation of the oxynitride phase. As in case of
3
(200), (020) and (002) located roughly between 321 and 331 2y.
3
the CaMoO ammonolysis, a broadening of the perovskite-phase
During the reaction the (002) and (200) reflections are shifted to
lower angles, whereas the (020) reflection is shifted to higher
angles. This indicates an increase in a and c, but a decrease in the b
parameter of the unit cell due to the progressive enrichment of the
material with nitrogen. As can be seen from Fig. 1B a broadening
of the diffraction peaks occurs in the early stages of the
ammonolysis. A closer examination reveals the splitting of the
reflections (see the inset of Fig. 1B). At that stage the pattern can
be fitted assuming two perovskite-type phases (Pbnm) of different
lattice parameters and atomic coordinates, corresponding to
reflections and their shift to lower angles is observed. This
indicates the formation of the oxynitride–perovskite phase
3 3
BaMo(O,N) with a larger lattice constant compared to BaMoO .
However, the formation of the secondary phases is simultaneously
observed (Fig. 2B). Some of the reflections were assigned to Mo,
BaO and Ba
main impurity phase reflections, which are coincident with those
of the phase formed during the ammonolysis of BaMoO at 1023 K,
3 2 6 2
Mo O N phases. However, we could not attribute the
4
to any phase of the Ba–Mo–O–N system found in ICSD (inorganic
crystal structure database) and Pdf-2 (powder diffraction files)
databases. Therefore, we conclude that they belong to a previously
undiscovered phase of the Ba–Mo–O–N system. The broadening of
3
CaMo(O,N) samples with different anionic composition. The
kinetics of oxygen exchange with nitrogen is slow due to the low
synthesis temperature. The rate of the oxynitride phase formation
enhances with increasing temperature, increasing ammonia flow
and/or decreasing sample mass. The oxynitride phase formed
BaMo(O,N)
that while the ammonolysis of BaMoO
of the BaMo(O,N) phase, the ammonolysis of BaMoO
considerable BaMo(O,N) yield.
Both CaMo(O,N) and BaMo(O,N)
color. The materials are moisture sensitive. Therefore, they
3
reflections decreases with time. It has to be noted
leads to the minor amount
leads to a
4
3
3
(
Fig. 1C) is stable under ammonia up to T ¼ 923 K. A further
3
increase of temperature leads to the formation of Mo
impurities.
2
N and CaO
3
3
possess a dark blue

ARTICLE IN PRESS
2
246
D. Logvinovich et al. / Journal of Solid State Chemistry 181 (2008) 2243–2249
proportional to the ionic radius of the cation. Hence, within one
group of the Periodic Table the strength of the positive inductive
effect decreases with decreasing the ionic radius of the element.
2
+
2+
2+
Thus, for Ca , Sr and Ba the strength of the positive inductive
2
+
2+
2+
effect decreases with the sequence Ba 4Sr 4Ca . In the same
sequence the ability of the alkaline-earth ion to retain the Mo-ions
oxidation state decreases. Indeed, during the ammonolysis of
4
AMoO the increase of the Mo-oxidation state in the main product
phase with the increase of the A-site cation ionic radius is
observed.
Apart from the ammonia flow rate and the positive inductive
effect, other parameters such as the influence of the lattice energy
and structure distortion energy on the phase formation during the
ammonolysis have to be considered.
Apparently, AMoO
during the ammonolysis than AMoO
AMoO is higher than that of AMoO . This is the most probable
explanation for the differences in the phase formation observed
3
undergoes fewer structural changes
4
. The lattice stability of
3
4
4 3
during the ammonolysis of AMoO and AMoO (with equal A-site
cation).
The tolerance factor (t) expresses the degree of distortion of
the perovskite structure (ABX ). It is calculated by dividing the
AꢀX distance by the MꢀX distance times root square of two:
Fig. 2. Thermal ammonolysis study of BaMoO
reacted with NH
3
. (A) Pure BaMoO
3
; (B) The sample
3
3
at T ¼ 998 K during 85 h (the shift of the perovskite phase
reflections to lower angles can be seen).
hA ꢀ Xi
t ¼ pffiffiffi
2
hB ꢀ Xi
were stored under the same conditions as CaMoO
3
and
can
BaMoO
During its reactions with AMoO
3
.
The deviation of t from 1 is proportional to the structure
distortion energy, which is a part of the total energy of a
compound. The higher the distortion energy, the less favored is
the formation of the perovskite structure. Thus, different tolerance
3
and AMoO
4
oxides, NH
3
act as a nitriding and a reducing agent:
AMoO4ðsÞ þ ð2 þ 2xÞ=3NH3ðgÞ
factors of AMoO
oxynitrides can explain whether substitution of oxygen with
nitrogen leads to the less distorted and thus, more stable
3
compared to those of the corresponding
¼
AMoO3ꢀxðsÞ
AMoO3ꢀxðsÞ þ yNH3ðgÞ
AMoO3ꢀxꢀy xþyðsÞ þ yH
N
x
þ ð1 þ xÞ H
2
O
ðgÞ þ ð2 ꢀ xÞ=6N2ðgÞ ðreductionÞ
4
+
4+
4+
structure. Tolerance factors of CaMo
calculated from ionic radii of the constituent ions [32,33] are 0.95,
.98 and 1.03 correspondingly, whereas for the corresponding
3 3 3
O , SrMo O and BaMo O
¼
N
2
O
ðgÞ þ yH
2
O
ðgÞ þ ðy=2ÞH2ðgÞ ðnitridationÞ
At T 4573 K ammonia undergoes considerable dissociation :
0
2
NH3ðgÞ ¼ N2ðgÞ þ 3H2ðgÞ
6
+
6+
6+
oxynitrides, for example, CaMo ON
2 2 2
, SrMo ON and BaMo ON
4 3
The formed hydrogen can also further reduce AMoO /AMoO .
they are equal to 0.97, 1.01 and 1.05. According to these values, the
2
ꢀ
3ꢀ
Depending on the relative rates of the processes described above,
a more nitrided or more reduced product will be formed. As
mentioned in Ref. [4] the dissociation of ammonia should be
minimized before it reaches the sample’s surface to achieve the
most effective nitridation conditions. Otherwise the reduction
may dominate over the nitridation reaction. That is why it is
important to increase the ammonia flow with temperature [4].
Additionally, increasing ammonia flow promotes faster water
removal [29,30] and the renewal of active nitriding species over
the sample. These factors can account for the enhanced kinetics of
the oxynitride phase formation observed during the ammonolysis
3 3
substitution of O with N in CaMoO and SrMoO reduces the
structure distortion energy. Thus, the substitution is favorable for
the perovskite structure formation. On the other hand, for
2
ꢀ
3ꢀ
BaMoO
more distorted structure. Therefore, it can be expected that
formation of the phase pure BaMo(O,N) at ambient pressure will
3
partial substitution of O
with N
would lead to a
3
be difficult, which is in an accordance with our experiments.
It has to be noted that only the tolerance factor consideration,
which in many cases is sufficient to compare crystal structures
stability of perovskite-type oxides or fluorides (e.g. materials with
predominantly ionic type of bonding) may not be enough to
compare crystal structure stability of an oxide with that of the
corresponding oxynitride. Indeed, higher covalence of the Mo–N
bonding as compared to that of the Mo–O bonding makes the
3
of AMoO when increasing the ammonia flow. Partial dissociation
of ammonia at elevated temperature before reaching the sample’s
surface explains the observed formation of the CaMo(O,N)
oxynitride phase with lower nitrogen content.
3
6
Mo–(O,N) octahedra more flexible to distort. Even if this type of
Besides the ammonia flow rate, the structure of the starting
precursor and the ionic radius of the A-site cation as well as the
covalence of the Mo–(O/N) bonding have an influence on the
oxynitride phase formation.
distortion occurs only locally, it may sufficiently reduce structure
distortion energy.
It is worth to compare our results on the ammonolysis of
BaMoO
the formation of the perovskite–oxynitride phase BaMo(O,N)
during the ammonolysis of both BaMoO and BaMoO can be
observed, while the authors of Ref. [11] reported that they were
not able to convert BaMoO and BaMoO into BaMo(O,N) . This
4 3
and BaMoO with those reported by Liu et al. [11]. Here
A detailed study on the Mo oxidation state stability of AMoO
and AMoO
4
3
3
(A ¼ Ca, Sr, Ba) against reduction is available from the
3
4
work of Kamata et al. [13,16]. The authors reported that the
alkaline-earth cation with a larger ionic radius provides better
3
4
3
stabilization of the Mo-oxidation state in AMoO
3
and AMoO
4
. The
can be explained by the difference in the ammonia chemical
potential which originates from the different ammonia flow used
during the studies as discussed above.
influence of the A-site cation ionic radius on the Mo-oxidation
state stability was discussed in terms of the positive inductive
effect. Strength of the positive inductive effect of a cation is
related with its polarizing power [31]. The latter is inversely
So far, only few information on the crystal structure of the
3
perovskite-type CaMoO is available in literature. Based on XRPD

ARTICLE IN PRESS
D. Logvinovich et al. / Journal of Solid State Chemistry 181 (2008) 2243–2249
2247
data a monoclinic symmetry was proposed for that material [13].
In contrast, de la Calle et al. [14] suggested an orthorhombic
symmetry with the space group Pbnm based on the results of ND
data refinement. An alternative monoclinic symmetry was not yet
3 3
tested. Most of both CaMoO and CaMo(O,N) crystallites
analyzed by means of selected area ED exhibit complicated ED
patterns due to the presence of twin domains. In such a case,
different twin orientations superposed by double diffraction make
the observation of the extinctions discarding glide planes and
screw axis difficult. Here, to choose between the possible space
groups X-ray diffraction was used additionally.
Extinction conditions observed on XRPD patterns of both
CaMoO
and monoclinic (P2
O2a ꢁ O2a ꢁ 2a and O2a
3
and CaMo(O,N)
/n) unit cells with lattice parameters
(a
3
matched with orthorhombic (Pbnm),
1
p
p
p
p
ꢁ 2a
p
ꢁ O2a
p
p
¼ cubic perovskite
lattice parameter), respectively. Subsequent Rietveld refinements
using these models converged with close values of the fit quality
indicators.
ED patterns (Fig. 3 and 4) do exhibit reflections with indices
(
h00), (0k0) and (00l): h ¼ 2n+1, k ¼ 2n+1, and l ¼ 2n+1. But these
reflections are less intense compared to (h00), (0k0) and (00l):
h ¼ 2n, k ¼ 2n, l ¼ 2n, and disappear upon the sample rotation.
They are therefore to be attributed to double diffraction. This
conclusion is also confirmed by the absence of any forbidden
diffraction in the [010] zone. On both our XRPD and ED (Fig. 3A)
patterns we do not observe (0kl) reflections (with k ¼ 2n+1),
which are allowed in the P2
the space group Pbnm. From that we conclude that the true space
group for both CaMoO and CaMo(O,N) is Pbnm.
Mainly two types of twin domains were found for CaMoO
51 oriented [ 1¯ 1¯ 0] and [001] domains; (B) Domains formed by the
superposition of [010] and [100] zone axes.
For CaMo(O,N) more twin-domain types have been found: (A)
01 oriented [ 1¯ 1¯ 0] domains (Fig. 4A); (B) 901 oriented [001]
1
/n space group and are not allowed in
Fig. 4. Examples of ED patterns resulting from the superposition of different twin
domains in CaMo(O,N)
3
: (A) 901 oriented [ 1¯ 1¯ 0] domains; (B) 901 oriented [001]
domains; (C) 601 oriented [ 2¯ 01] domains; (D) 451 oriented [ 1¯ 1¯ 0] and [001]
domains.
3
3
3
: (A)
4
anionic site constrained to be fully occupied. Finally, the lattice
3
0
and the profile parameters, 2Y , the background coefficients, the
thermal displacement factors and the anionic occupancies
9
domains (Fig. 4B); (C) 601 oriented [ 2¯ 01] domains (Fig. 4C) (D) 451
oriented [ 1¯ 1¯ 0] and [001] domains (Fig. 4D).
(neutron data) were refined together. To improve the fit, d-MoN
(P 6¯ m2), [38] g-Mo N (Fm 3¯ m), [39] and CaO (Fm 3¯ m) were included
2
Formation of twin domains of these types has been reported
earlier for a number of distorted perovskite-type oxides with
in the refinements. The quantities of these secondary phases were
determined from the XRPD data (Fig. 5) refinement (the refine-
ment resulted in o1 wt % of CaO and o2 wt % of each MoN and
similar lattice parameters (in our case the mismatch for CaMoO
3
between a and b is 2.3%, whereas for CaMoO1.7(1)
N1.3(1) the
2
Mo N phases) and were kept fixed during the neutron data
mismatch is 0.9%) [34–37].
refinement. Further inspection of the ND profiles and difference
graphs revealed evanescent weak features attributed to the
presence of a minor impurity phase which could not be identified.
Therefore, the regions containing main reflections of that phase
were excluded from the refinement.
The statistics of the refinements, the visual inspection of the fit
(Figs. 5 and 6) and the refined values of the displacement
parameters indicate that the chosen model was correct. The
refined O/N content and the anionic distribution correspond to the
3
Rietveld refinements of CaMo(O,N) from both neutron and
X-ray data were carried out in space group Pbnm. In the starting
structural model the Ca:Mo ratio was set to 1:1, whereas the O:N
ratio was set to 2:1. The background was determined manually,
refined at the initial stages and fixed during the further
refinement. Displacement parameters were refined isotropically
for all the atoms. Since X-rays are not able to distinguish between
2
ꢀ
3ꢀ
O
and N , we did not attempt to refine the anionic composition
from the X-ray data. During the neutron data refinement, the
occupancy factors for oxygen and nitrogen were refined with the
composition CaMoO1.7(1)N1.3(1) with a totally disordered O/N
arrangement. The result is in good agreement with the hotgas
extraction analysis, which yielded in the composition of
CaMoO1.79(5)N1.25(2) obtained after subtracting the minor second-
ary phases contribution (taken from the refinement). The neutron
data-derived O/N content corresponds to the formal oxidation
state of Mo+5.3. Higher nitrogen content of this phase as
3
compared to that of the SrMo(O,N) phase [12] can be explained
by higher ammonia flow and lower synthesis temperature
employed during the present investigation. As a summary of the
refinements, the obtained structural parameters bond lengths and
angles are displayed in Tables 1 and 2.
According to the refinements, the average Ca–O/N distance
increases upon the substitution of oxygen with nitrogen, which is
3
ꢀ
˚
consistent with a larger ionic radius of N (1.48 A) compared to
2
ꢀ
˚
that of O (1.38 A). The average Mo–O/N distance value of
Fig. 3. Electron diffraction patterns for CaMo(O,N)
B) [010] orthorhombic zone axis.
3
taken along the (A) [100] and
˚
(
CaMoO1.7(1)N1.3(1) (1.994(2) A) resembles the one determined for

ARTICLE IN PRESS
2
248
D. Logvinovich et al. / Journal of Solid State Chemistry 181 (2008) 2243–2249
Fig. 7. Angular variation of the total full-width at half-maximum (FWHM)
determined from whole profile Rietveld refinements of the XRPD data for CaMoO
CaMoO1.7(1) 1.3(1) and the standard LaB SRM 660a.
Fig. 5. Rietveld refinement plot of the X-ray powder diffraction data for
CaMoO1.7(1) 1.3(1). Space group: Pbnm. The observed intensities, calculated profile,
3
,
N
N
6
2
difference curve and Bragg positions are shown. CaO, d-MoN and g-Mo N have
been included as a minor impurity phase in the refinement.
strain values of the latter compound are 0.14 mm and 0.29% which
are close to the values obtained for CaMoO
3
.
Size broadening on the diffractograms can be attributed to
both the crystallites size and the presence of twin domains. Small
apparent strain values obtained for CaMoO
1.3(1) confirm that both materials are chemically
homogeneous. The larger apparent size and the smaller apparent
strain values obtained for CaMoO1.7(1) 1.3(1) compared to CaMoO
3
as well as for
CaMoO1.7(1)
N
N
3
can arise either from the smaller mismatch between a and c
lattice constants for the former compound or from the different
thermal history of the samples, which would also explain the
multiplicity of different twin types observed for CaMoO1.7(1)
N
1.3(1)
by TEM.
Since no apparent reflection splitting was noticed on the XRPD
pattern of BaMo(O,N) and its main reflections were isotropically
3
broadened, we assign its space group as Pm 3¯ m. This confirms its
totally disordered anionic arrangement. XRPD reveals a crystallite
3 3
size of BaMo(O,N) (0.12 mm) similar to that of CaMo(O,N) which
was in accordance with electron microscopy studies. Because of
the unknown impurity composition and structure, we did not
Fig. 6. Rietveld refinement plot of the neutron powder diffraction data for
CaMoO1.7(1) 1.3(1). Space group: Pbnm. The observed intensities, calculated profile,
N
difference curve and Bragg positions are shown. CaO, d-MoN and g-Mo
been included as a minor impurity phase in the refinement.
2
N have
attempt a whole pattern refinement of XRPD data of BaMo(O,N)
but rather estimate the lattice constant and line broadening of
3
,
˚
BaMo(O,N)3. The obtained lattice constant (4.0657(1) A) is larger
˚
SrMoO1.89(2)N1.11(2) by ND (1.999(5) A). [12] However, partial
than the one refined from the XRPD data for BaMoO3
2
ꢀ
3ꢀ
˚
4+
substitution of O
of the average Mo–O/N distance, which is not the case for CaMoO
Table 2), where the mean Mo–O/N distance is not significantly
affected by the substitution. Moreover, the smaller average
Mo–O/N distance in CaMoO1.7(1) 1.3(1) compared to that in
SrMoO1.89(2) 1.11(2) does not completely agree with the O/N
composition of these phases. This indicates that CaMoO1.7(1)
may contain anionic/cationic vacancies. Displacement parameters
values of Ca and Mo in CaMoO1.7(1) 1.3(1) are higher than those in
CaMoO , which gives a hint for the presence of cationic vacancies
in SrMoO
3
by N
leads to an increase
(4.0409(1) A). Contrary to the partial oxidation of Mo to smaller
5
+
6+
2ꢀ
3ꢀ
Mo and Mo , a partial replacement of O with larger N will
lead to the increase of the lattice constant. From the comparison
3
(
of the lattice constant values for BaMoO
follows that it is the difference in the ionic radii of O and N
which has a major influence on the lattice constant. The larger
lattice constant of BaMo(O,N) indicates that both Mo–O/N and
Ba–O/N distances increase upon partial substitution of O with
3 3
and BaMo(O,N) it
2
ꢀ
3ꢀ
N
,
N
N
1.3(1)
3
2
ꢀ
3
ꢀ
N
N
3
in BaMoO .
3
in the former compound. Hence true formal oxidation state of Mo
in this compound must be slightly lower than that calculated from
the O/N content.
4. Conclusions
Based on the refined Mo–O/N–Mo angle values for CaMoO
1.3(1), it can be concluded that the crystal structure of
the latter compound is less distorted. This is in accordance with
3
and
The possibilities to prepare previously unreported oxynitri-
CaMoO1.7(1)
N
de–perovskite AMo(O,N)
3
(A ¼ Ca, Ba) phases by thermal ammo-
nolysis of AMoO
4
and AMoO oxide precursors were successfully
3
3ꢀ
2ꢀ
higher ionic radius of N compared to that of O
Similar to CaMoO , the reflections of CaMoO1.7(1)
broadened isotropically (Fig. 7). The refined apparent size and
.
investigated.
3
N1.3(1) are
Only the ammonolysis of CaMoO
3
and BaMoO
3
led to a
considerable yield of the desired phases. CaMoO
4
decomposes to

ARTICLE IN PRESS
D. Logvinovich et al. / Journal of Solid State Chemistry 181 (2008) 2243–2249
2249
CaO and Mo
formed as a minor phase during the ammonolysis of BaMoO
which yields mainly in Ba Mo . Thus, the crystal structure of
the starting precursor is crucial for the desired phase formation.
As the other factors influencing the phase formation during the
2
N when reacted with ammonia. BaMo(O,N)
3
is
References
4
,
O
6
N
2
[1] W.E. Newton, Philos. Trans. R. Soc. Lond. B 317 (1987) 259–277.
3
2
[
2] D. McKay, J.S.J. Hargreaves, J.L. Rico, J.L. Rivera, X.L. Sun, J. Solid State Chem.
81 (2008) 325–333.
3] L. Le Gendre, R. Marchand, Y. Laurent, J. Eur. Ceram. Soc. 17 (1997) 1813–1818.
1
[
ammonolysis of AMoO
3
and AMoO
4
the gas to solid ratio, positive
[4] R. Marchand, Y. Laurent, J. Guyader, P. L’Haridon, P. Verdier, J. Eur. Ceram. Soc.
(1991) 197–213.
8
inductive effect of the A-site cation, the lattice energy, the
structure distortion energy and Mo–N bonding covalence were
considered.
[
[
5] F. Tessier, R. Marchand, J. Solid State Chem. 171 (2000) 143–151.
6] A. Rachel, S.G. Ebbinghaus, M. Gungerich, P.J. Klar, J. Hanss, A. Weidenkaff, A.
Reller, Thermochim. Acta 438 (2005) 134–143.
[
[
7] R. Marchand, F. Pors, Y. Laurent, Ann. Chim. Fr. 16 (1991) 553–560.
8] P. Antoine, R. Marchand, Y. Laurent, C. Michel, B. Raveau, Mater. Res. Bull. 23
CaMo(O,N)
3
crystallizes in space group Pbnm and possesses a
. ND reveals a
less distorted crystal structure than CaMoO
statistical distribution of O and N among the available anionic
sites and an anionic content corresponding to the composition
3
(1988) 953–957.
2
ꢀ
3ꢀ
[9] I.D. Fawcett, K.V. Ramanujachary, M. Greenblatt, Mater. Res. Bull. 32 (1997)
1565–1570.
10] P. Maillard, F. Tessier, E. Orhan, F. Chevire, R. Marchand, Chem. Mater. 17
[
[
1.3(1)
CaMoO1.7(1)N .
(2005) 152–156.
TEM shows evidence for twin domains in both CaMoO
3
and
11] G. Liu, X. Zhao, H.A. Eick, J. Alloys Compd. 187 (1992) 145–156.
CaMoO1.7(1)
N
1.3(1), which together with the average crystallite size
[12] D. Logvinovich, R. Aguiar, R. Robert, M. Trottmann, S.G. Ebbinghaus, A. Reller,
A. Weidenkaff, J. Solid State Chem. 180 (2007) 2649–2654.
may be responsible for the line profile broadening in the XRPD
patterns. The multiplicity of different twin types measured for
[
[
13] K. Kamata, T. Nakamura, T. Sata, Chem. Lett. 4 (1975) 81–86.
14] C. de la Calle, J.A. Alonso, M. Garcia-Hernandez, V. Pomjakushin, J. Solid State
Chem. 179 (2006) 1636–1641.
3
CaMoO1.7(1)N1.3(1) compared to CaMoO can be explained on the
[
[
[
15] L.H. Brixner, J. Inorg. Nucl. Chem. 14 (1960) 225–230.
16] K. Kamata, T. Nakamura, T. Sata, Mater. Res. Bull. 10 (1975) 373–378.
17] Z. Singh, S. Dash, R. Prasad, D.D. Sood, J. Alloys Compd. 215 (1994) 303–307.
one hand by the smaller mismatch between a and c lattice
constants in the former compound or on the other hand by the
different thermal history of the compounds.
[18] D. Logvinovich, A. Borger, M. Dobeli, S.G. Ebbinghaus, A. Reller, A. Weidenkaff,
Prog. Solid State Chem. 35 (2007) 281–290.
BaMo(O,N)
completely disordered O/N arrangement as found for Ca-
1.3(1). Its determined lattice parameter and conse-
quently both Mo–O/N and Ba–O/N distances are larger than
those of the corresponding BaMoO as revealed by X-ray
diffraction. This finding is attributed to the larger ionic radius of
3
adopts the space group Pm 3¯ m. This indicates a
[
[
19] P. Thompson, D.E. Cox, J.B. Hastings, J. Appl. Crystallogr. 20 (1987) 79–83.
20] B. van Laar, W.B. Yelon, J. Appl. Crystallogr. 17 (1984) 47–54.
MoO1.7(1)
N
[21] L.W. Finger, D.E. Cox, A.P. Jephcoat, J. Appl. Crystallogr. 27 (1994) 892–900.
[
[
22] J. Rodriguez-Carvajal, T. Roisnel, Mater. Sci. Forum 443–444 (2004) 123–126.
23] P. Fischer, G. Frey, M. Koch, M. Konnecke, V. Pomjakushin, J. Schefer, R. Thut,
N. Schlumpf, R. Burge, U. Greuter, Physica B 276–278 (2000) 146–147.
3
[24] J. Rodriguez-Carvajal, Physica B 192 (1993) 55–69.
[25] P.A. Stadelmann, Ultramicroscopy 21 (1987) 131–145.
3
ꢀ
2ꢀ
N
compared to that of O
.
[
[
26] M.T. Weller, S.J. Skinner, Int. J. Inorg. Mater. 2 (2000) 463–467.
27] P. Subramanya Herle, M.S. Hegde, G.N. Subbanna, J. Mater. Chem. 7 (1997)
Thus, as future work, studies on thermal reoxidation and
catalytic properties of the synthesized phases are planned.
2
121–2125.
28] P. Ettmayer, Monatshefte Chem./Chem. Monthly 101 (1970) 127–140.
29] S.H. Elder, F.J. DiSalvo, L. Topor, A. Navrotsky, Chem. Mater.
545–1553.
30] S.J. Clarke, K.A. Hardstone, C.W. Michie, M.J. Rosseinsky, Chem. Mater. 14
2002) 2664–2669.
[31] J. Etourneau, J. Portier, F. Menil, J. Alloys Compd. 188 (1992) 1–7.
[
[
5
(1993)
1
Acknowledgments
[
(
The authors acknowledge the German Science Foundation
[
[
[
32] R.D. Shannon, C.T. Prewitt, Acta Crystallogr. B 25 (1969) 925–946.
33] R. Shannon, Acta Crystallogr. A 32 (1976) 751–767.
34] Y. Wang, R.C. Liebermann, Phys. Chem. Miner. 20 (1993) 147–158.
(DFG-SPP 1136) and Empa for the financial support and Dr. Denis
Sheptyakov (SINQ, Paul Scherrer Institute) for the technical
assistance. This work is partly based on the experiments
performed at the Swiss Spallation Neutron Source SINQ, Paul
Scherrer Institute, Villigen, Switzerland.
[35] H. Nakano, M. Kobayashi, N. Kamegashira, J. Am. Ceram. Soc. 90 (2007)
276–1279.
36] A. Weidenkaff, M.H. Aguirre, T. Lippert, U. Falke, U. Bangert, Chimia 60 (2006)
42–748.
37] M.H. Aguirre, R. Robert, D. Logvinovich, A. Weidenkaff, Inorg. Chem. 46 (2007)
744–2750.
[38] C.L. Bull, P.F. McMillan, E. Soignard, K. Leinenweber, J. Solid State Chem. 177
2004) 1488–1492.
1
[
[
7
2
Appendix A. Supplementary Materials
(
[39] T. Kawashima, E. Takayama-Muromachi, P.F. McMillan, Physica C 460–462
Supplementary data associated with this article can be found
in the online version at doi:10.1016/j.jssc.2008.05.012.
(2007) 651–652.