Perovskite-type SrTi1-xNbx(O,N)3 compounds: Synthesis, crystal structure and optical properties

Article abstract of DOI:10.1016/j.jssc.2011.02.017

The synthesis, crystal structure, thermal stability and absorbance spectra of perovskite-type oxynitrides with the general formula SrTi _1-xNb_x(O,N)₃ (x=0.05, 0.10, 0.20, 0.50, 0.80, 0.90, 0.95) have been investigated. Oxide samples were prepared by a polymerized complex synthesis route and post-treated under ammonia at 850 °C for 24 h to substitute nitrogen for oxygen. Synchrotron X-ray powder diffraction (XRD) evidenced that the mixed oxide phases were all transformed into oxynitrides with perovskite-type structure during a thermal ammonolysis. SrTi _1-xNb_x(O,N)₃ with compositions x≤0.80 crystallized in a cubic and samples with x≥0.90 in a tetragonal structure. The Rietveld refinement indicated a continuous enlargement of the lattice parameters towards higher niobium content of the samples. Thermogravimetric analysis (TGA) and hotgas extraction revealed the dependence of the nitrogen incorporation upon the degree of niobium substitution. It showed that more nitrogen was detected in the samples with higher niobium content. Furthermore, TGA disclosed stability for all oxynitrides at T≤400 °C. Diffuse reflectance spectroscopy indicated a continuous decrease of the band gap's width from 3.24 eV (SrTi_0.95Nb_0.05 (O,N)₃) to 1.82 eV (SrTi_0.05Nb_0.95(O,N)₃) caused by the increasing amount of nitrogen towards the latter composition.

Full text of DOI:10.1016/j.jssc.2011.02.017

Journal of Solid State Chemistry 184 (2011) 929–936
Contents lists available at ScienceDirect
Journal of Solid State Chemistry
journal homepage: www.elsevier.com/locate/jssc
Perovskite-type SrTi1 Nb (O,N) compounds: Synthesis, crystal structure
ꢀx
x
3
and optical properties
1
n
Alexandra Maegli, Songhak Yoon, Eugenio Otal, Lassi Karvonen, Peter Mandaliev , Anke Weidenkaff
Empa – Swiss Federal Laboratories for Materials Science and Technology, Solid State Chemistry and Catalysis, Ueberlandstr. 129, CH-8600 Duebendorf, Switzerland
a r t i c l e i n f o
a b s t r a c t
Article history:
The synthesis, crystal structure, thermal stability and absorbance spectra of perovskite-type oxynitrides
Received 20 December 2010
Received in revised form
with the general formula SrTi1 Nb (O,N) (x¼0.05, 0.10, 0.20, 0.50, 0.80, 0.90, 0.95) have been
ꢀx
x
3
investigated. Oxide samples were prepared by a polymerized complex synthesis route and post-treated
under ammonia at 850 1C for 24 h to substitute nitrogen for oxygen. Synchrotron X-ray powder
diffraction (XRD) evidenced that the mixed oxide phases were all transformed into oxynitrides with
1
4 February 2011
Accepted 16 February 2011
Available online 26 February 2011
perovskite-type structure during a thermal ammonolysis. SrTi1ꢀxNb
x
(O,N)
3
with compositions xr0.80
Keywords:
crystallized in a cubic and samples with xZ0.90 in a tetragonal structure. The Rietveld refinement
indicated a continuous enlargement of the lattice parameters towards higher niobium content of the
samples. Thermogravimetric analysis (TGA) and hotgas extraction revealed the dependence of the
nitrogen incorporation upon the degree of niobium substitution. It showed that more nitrogen was
detected in the samples with higher niobium content. Furthermore, TGA disclosed stability for all
oxynitrides at Tr400 1C. Diffuse reflectance spectroscopy indicated a continuous decrease of the band
Oxynitrides
Perovskites
X-ray diffraction
Rietveld refinement
Band gap
gap’s width from 3.24 eV (SrTi0.95Nb0.05 (O,N)
3
) to 1.82 eV (SrTi0.05Nb0.95(O,N)
3
) caused by the
increasing amount of nitrogen towards the latter composition.
&
2011 Elsevier Inc. All rights reserved.
1
. Introduction
number of anions (e.g. BaTaO3.5-BaTaON
2
[12]) and by forming
oxygen vacancies (SrTiO -SrTiO3ꢀ [13]).
3
d x
N
0
0
Perovskites tolerate substitutions with a wide range of elements
In a perovskite-type d transition-metal oxide (with d such as
4
þ
5þ
5þ
because of their flexible crystal structure. Cationic substitutions in
ABX perovskite-type compounds are well explored and provide a
Ti , Ta and Nb ), the top of the valence band is mainly formed
by oxygen 2p (O 2p) interactions, whereas the bottom of the
conduction band consists of interactions between the transition-
metal dt2g and O 2p orbitals [14]. The purpose of nitrogen insertion
into the oxide lattice is to modify the electronic structure of the
pristine oxide primarily at the valence band [15]. Since nitrogen
3
large class of functional materials used for diverse applications
depending on their specific composition [1–3]. Anionic substitution
in perovskites is comparatively less investigated. Oxygen is most
likely replaced by nitrogen and fluorine, leading to oxynitrides and
oxyfluorides [4] and the synthesis of oxynitrides has been success-
fully accomplished in the past [5–10]. The resulting increase
(
w(N)¼3.04) is less electronegative than oxygen (w(O)¼3.44), the
nitrogen 2p (N 2p) levels are energetically located above the O 2p
levels and mixing of N 2p and O 2p levels therefore leads to a
narrower band gap. Hence, oxynitrides may absorb visible light and
are consequently often coloured [16,17]. Ch e´ vire et al. [10] reported
3
ꢀ
of negative charge (N ) compared to the parent oxide can be
balanced by substituting with a higher valence A- or B-site cation
(
e.g. BaTiO
3
-LaTiO
2
N [10]), by changing the oxidation state of the
-SrMoO2.5 0.5 [11]), by reducing the
transition metal (e.g. SrMoO
4
N
a progressive colour and band-gap change in the system LaTiO
N–ATiO
(A¼Sr, Ba) depending on the nitrogen content. Further-
more, it has recently been shown that nitrogen substitution can
diminish the band gap byꢁ1.6 eV from 3.3 eV for LaNbO down to
.7 eV for LaNbON [8]. Nevertheless, besides modelling the band
gap with nitrogen, structural distortions and the electronegativity
2
3
n
4
Corresponding author.
E-mail addresses: Alexandra.Maegli@empa.ch (A. Maegli),
1
2
Songhak.Yoon@empa.ch (S. Yoon), Eugenio.Otal@empa.ch (E. Otal),
Lassi.Karvonen@empa.ch (L. Karvonen),
w
of the B-site cation can also change the electronic configuration near
the Fermi level [14,16,17].
Because of their variable colours, oxynitrides have a potential
as non-toxic pigments whereby they could replace the currently
used pigments that contain dangerous elements (Cd, Pb, Cr, Hg,
Petar.Mandaliev@env.ethz.ch (P. Mandaliev),
Anke.Weidenkaff@empa.ch (A. Weidenkaff).
1
Present address: ETH – Swiss Federal Institute of Technology, Institute of
Biogeochemistry and Pollutant Dynamics, Universitaetsstr. 16, CH-8092 Zuerich,
Switzerland.
0022-4596/$ - see front matter & 2011 Elsevier Inc. All rights reserved.
doi:10.1016/j.jssc.2011.02.017

9
30
A. Maegli et al. / Journal of Solid State Chemistry 184 (2011) 929–936
etc.) [16,18]. It was demonstrated that depending on the cationic
composition, Ca1ꢀxLa (1þx) obtain colours from red to
black residues were ground with an agate pestle and heated in an
x
TaO(2ꢀx)
N
alumina crucible at 1000 1C for 6 h with a heating ramp of
ꢀ
1
yellow. Therefore it could substitute cadmium sulphoselenide
pigments, which are widely used nowadays [17]. Furthermore,
the possibility to tailor the band gap promotes oxynitrides as
interesting candidates in the photocatalytical and photoelectro-
chemical water-splitting reaction [5,6,19]. A shortcoming of
most oxide semiconductors used as photocatalysts is their
wide band gap which prohibits them from absorbing solar radiation
10 1C min resulting in white oxide powders.
2.2. Thermal ammonolysis
The as-synthesized oxide precursors were transformed into
oxynitrides via thermal ammonolysis, a reaction between an oxide
precursor and ammonia (NH ) at elevated temperatures [6,7,12].
Batches of 1.5 g oxide powder were mixed with potassium chloride
ꢁ
l4400 nm [20]. Because the photon flux is highest in the range of
the visible light (400 nmolo700 nm), oxynitrides may provide an
3
(
KCl) in the molar ratio oxide/KCl¼1/3. KCl acted as flux to enhance
efficient solar energy to hydrogen conversion-rate. Another require-
ment for a material used in the photocatalytic approach is that it
should be defect-free (e.g. no anionic vacancies) in order to prevent
the phase formation and crystallinity of the oxynitride. The reaction
was carried out in an alumina cavity-reactor. The air inside the
cavity-reactor was evacuated twice before filling it to atmospheric
pressure with argon (Ar) (Messer, Z99.998%). Afterwards, the
samples were heated with 10 1C min
this temperature, Ar was replaced by NH
electron–hole recombination [20]. Nitridation of SrTiO
3
has been
shown to evoke undesired oxygen vacancies but a diminishment of
ꢀ
1
3
þ
3ꢀ
under Ar up to 750 1C. At
(Messer, Z99.98%) and
3
these was achieved by co-substituting La and N in SrTiO [21].
5
þ
3ꢀ
Recently, the effect of co-‘‘doping’’ SrTiO
3
with Nb
and N
has
3
further heated until reaching the reaction temperature of 850 1C.
The reaction was completed after 24 h and the oxynitride samples
were cooled down under Ar. The flow rate during the synthesis was
been studied and a change of the band gap and the photocatalytic
activity with the amount of the niobium ‘‘dopant’’ was observed [22].
3
The cross substitution of SrTiO with nitrogen and niobium leads to
ꢀ
1
200 mL min
3
for NH and Ar. Thereafter, the powders were
the system SrTi1ꢀxNb (O,N) and the reaction can be generalized as
x
3
washed with deionized water to eliminate all traces of KCl. The
complete removal of the flux was confirmed by testing the super-
5
þ
þxTi4þ þxO2ꢀ
SrTiO
3
þxNb þxN3ꢀ-SrTi1ꢀxNb
x
O
3ꢀx
N
x
(1)
3
natant water with an aqueous solution of AgNO . Finally, all
3 2
The two end compounds of the system, SrTiO and SrNbO N
powders were dried at 130 1C for approximately 3 h.
crystallize in the cubic space group Pm-3m [23] and in a distorted
perovskite structure, respectively. A tetragonal structure with
space group I4/mcm has been suggested for SrNbO
2
N [9,24].
2.3. Characterisation
A recent work that combined neutron with electron diffraction
postulated that besides the rotation about the c-axis also an
Lab source X-ray diffraction (XRD) measurements were
performed for all oxides, oxynitrides and reoxidized oxynitrides.
The instrument was a PANanalytical X’Pert PRO system with
X ꢀC elerator detector operating in Bragg–Brentano geometry
2
anionic ordering might be present in SrNbO N, thus lowering the
symmetry to the monoclinic space group I112/m [25].
In the present work, the structural changes between the cubic
SrTiO
3
and the distorted SrNbO
2
N were investigated. Additionally,
1
(y/2y) using Cu Ka (45 kV and 40 mA) and a Ge (1 1 1) incident
the influence of these structural modifications on the band gap
was analyzed. Furthermore, a progressive change of the nitrogen
amount in the anionic lattice was expected, which in turn would
influence the evolution of the band gap and colour throughout the
solid solution.
monochromator.
Synchrotron XRD experiments were performed for all oxides
and oxynitrides at the MS Powder beamline at the Swiss Light
Source. The incident X-rays were monochromatized to the wave-
˚
length of 0.44 A. Quartz capillaries (Hilgenberg GmbH) with a
nominal diameter of 0.33 mm were filled with the powders and
sealed with capillary wax. The diffraction patterns were scanned
from 0.11 to 1161 with a step size of 0.0041 in 2y. XRD patterns of
2
. Experimental
the oxynitrides were analysed by Rietveld refinement using the
software FullProf [27]. Thompson–Cox–Hastings pseudo-Voigt [28]
was chosen as a profile function and the refined range was from
2.1. Oxide precursors
1
01 to 501 in 2
y.
A series of seven oxide precursor with the general composition
Thermogravimetric analysis (TGA) was used to study the
SrTi1ꢀxNb
x
O
3þ0.5x (x¼0.05, 0.10, 0.20, 0.50, 0.80. 0.90, 0.95) was
thermal reoxidation and stability in air as well as the nitrogen
content of the oxynitrides. The experiments were performed with
a NETZSCH STA 409 CD thermobalance coupled to a NETZSCH
QMS 403 C Aeolos mass spectrometer. Around 50 mg of each
prepared by a Pechini-type polymerized complex route with citric
acid (CA) as chelating and ethylene glycol (EG) as polymerizing
agents. The molar ratio of the cations (strontium, titanium and
niobium), EG and CA was according to Sr/Ti/Nb/EG/CA¼1/1ꢀx/x/
ꢀ
1
oxynitride was heated in an alumina crucible with 10 1C min
up to 1500 1C in synthetic air with a flow of 50 mL min
40/10 where x stands for the desired substitution level for a specific
ꢀ
1
.
composition [26].
Additionally, to check the gas species absorbed and released at
each temperature, coupled mass spectrometry (MS) was used for
4
Titanium isopropoxide (Ti(O-iPr) , Sigma-Aldrich, 99.999%) was
added to EG (Merck, Z99.5%), therein CA (Alfa Aesar, Z99%) was
appended. Strontium nitrate (Sr(NO , Merck, Z99%) was dissolved
in deionized water and niobium pentachloride (NbCl , Sigma-
Aldrich, Z99%) in EtOH (Merck, Z99.5%). Both solutions were
poured into the mixture of EG, CA and Ti(O-iPr) . In the following,
this mix was stirred for 4 h under reflux at 70 1C to promote the
complete dissolution of CA and the formation of metal–citrate
complexes. Thereafter, the viscous solution was heated at 120 1C
for 4 h in order to continue the polymer formation by an esterifica-
tion between CA and EG. Afterwards, the gel was slowly heated up
to 300 1C to decompose the organic matrix. Finally, the obtained
2
certain specific measurements with an Ar/O atmosphere (volume
3 2
)
ratio of 80/20). The other experimental parameters were the same
as those of the TGA measurements. The nitrogen stoichiometry (y)
5
was evaluated from the weight gain
experiment according to Ref. [6]:
Dm measured during the
4
2
D
m
MðABOxþ3y=2Þ
y ¼
ꢃ
ð2Þ
½
ð3=2ÞMðO
2
ÞꢀMðN
2
Þꢂ mðABOxþ3y=2Þ
where M(O ) is the molar mass of molecular oxygen, M(N
2
2
) is the
molar mass of molecular nitrogen, M(ABOxþ3y/2) is the molar

A. Maegli et al. / Journal of Solid State Chemistry 184 (2011) 929–936
931
mass of the formed oxide and m(ABOxþ3y/2) is the mass of the
formed oxide.
x¼0.05–0.50 crystallized in space group Pm-3m, sample x¼0.80
crystallized in space group R3m and samples x¼0.90 and 0.95
possessed a layered-perovskite structure (Pna21) as shown in
Fig. 1. After nitrogen substitution, the diffraction pattern indi-
cated perovskite-type structure for all seven compositions, thus
Hotgas extraction using a LECO TC500 analyzer provided the
O/N content of the samples x¼0.20, 0.50, 0.80 for SrTi1ꢀx
x 3 2
Nb (O,N) oxynitrides. Nitrogen was measured as N by thermal
conductivity and oxygen as CO
2
by infrared radiation.
x 3
offering a SrTi1ꢀxNb (O,N) solid solution (Fig. 2). Oxynitride
UV-visible diffuse reflectance spectra were collected with a
UV-3600 Shimadzu UV–vis–NIR spectrophotometer with an inte-
grating sphere over a spectral range of 300–1500 nm (4.13–0.83 eV),
measuring in reflectance mode. The powders were pressed into
pellets of about 200 mg each. Only for the relatively dark oxynitride
samples with x¼0.20 and 0.50 had a minor impurity phase.
The results of the Rietveld analysis done for the oxynitrides,
i.e. crystallographic data as well as agreement factors are given in
Table 1. The cubic space group Pm-3m was chosen as model for
samples x¼0.05–0.80, according to the end member SrTiO
3
. For
sample x¼0.20, BaSO
4
powder was added in the ratio 1/9¼sample
samples x¼0.90 and 0.95, the tetragonal space group
I4/mcm [9,24] was used as a first approach because peak splitting
in x¼0.90 and 0.95 in the higher order reflections was observed
/
BaSO . The diffuse reflectance was transformed into absorbance
4
according to Ref. [29]
(Fig. 3). In a second approach, these two samples were refined in
A ¼ RmaxꢀRcd
ð3Þ
the monoclinic space group I112/m [25] which allows the possi-
bility of anionic ordering. The refinement with the monoclinic
structure did not show an improvement of the agreement factor.
Yet, with our X-ray data we could neither exclude nor postulate
an anionic ordering. The tetragonal I4/mcm evolves from the cubic
Pm-3m by an octahedral rotation about the cubic [0 0 1] axis with
where A is the absorbance, Rmax is the maximum value of the diffuse
reflectance and Rcd is the diffuse reflectance measured over a certain
wavelength interval.
Shapiro’s method [30] was chosen to approximate the optical
g
band gap E . The minimal absorbance A calculated with Eq. (3) is
0
0
ꢀ
set to zero and the band gap wavelength is determined by
extrapolating the long-wavelength edge (the onset) of the absor-
bance to this zero line.
out-of-phase rotations of successive planes (a a c in Glazer
notation [31]). The structural effect of octahedral tilting is to
double some or all of the cubic unit cell axes, which gives rise to
the splitting of reflections and appearance of superlattice reflec-
tions in the XRD pattern [32]. The reflection splitting for x¼0.95 is
apparent, whereas it is less pronounced in x¼0.90 (Fig. 3). Thus it
is expected that the octahedral tilting in sample x¼0.90 is smaller
than in sample x¼0.95. This is plausible since Goldschmidt’s
tolerance factor t [32] deviates slightly but increasingly from
unity moving towards the niobium-rich side of the system. The
enlargement of the unit cell parameters with increasing niobium
3
. Results and discussion
3.1. X-ray diffraction and Rietveld refinement
The diffraction pattern of the oxides revealed mixed crystal
phases for the different niobium substitution levels. Samples
Fig. 1. Synchrotron XRD pattern of oxide samples x¼0.05 (Pm-3m), x¼0.80 (R3m) and x¼0.95 (Pna21); l
¼0.44 A˚ .
Fig. 2. Synchrotron XRD pattern of oxynitride samples x¼0.05–0.95 with perovskite-type structure, ‘o’ assigns impurity phase; l
¼0.44 A˚ .

9
32
A. Maegli et al. / Journal of Solid State Chemistry 184 (2011) 929–936
Table 1
Summary of Rietveld refinement of oxynitride samples x¼0.05–0.95.
Nb content x
0.05
0.10
0.20
0.50
0.80
0.90
0.95
Space group
Pm-3m
Pm-3m
Pm-3m
Pm-3m
Pm-3m
I4/mcm
I4/mcm
a ( A˚ )
3.8983(0)
3.9046(9)
3.9149(8)
3.9636(6)
3.9989(0)
5.6720(6)
5.6794(7)
b ( A˚ )
c ( A˚ )
Volume ( A˚ 3
8
.0473(9)
8.0609(0)
260.014
59.241
59.542
60.005
62.271
63.947
258.902
)
Sr
x
0
0
0
0
0
0
0
y
0
0
0
0
0
½
½
z
0
0
0
0
0
¼
¼
site
1a
1a
1a
1a
1a
4b
4b
a
occ
1
1
1
1
1
1
1
iso ( A˚ 2
0.55582
0.47607
0.47194
1.06445
0.99909
0.84100
0.82022
U
)
Ti/Nb
x
½
½
½
½
½
0
0
y
½
½
½
½
½
0
0
z
½
½
½
½
½
½
½
site
occ
1b
1b
1b
1b
1b
4c
4c
0.95/0.05
0.36354
0.90/0.10
0.33362
0.80/0.20
0.33272
0.50/0.50
0.31537
0.20/0.80
0.49357
0.10/0.90
0.63103
0.05/0.95
0.48134
iso ( A˚ 2 b
U
)
)
)
O/N(1)
x
0
0
0
0
0
0
0
y
½
½
½
½
½
0
0
z
½
½
½
½
½
¼
¼
site
occ
3c
3c
3c
3c
3c
4a
4a
2.94/0.06
0.71559
2.92/0.08
0.66620
2.79/0.21
0.69350
2.55/0.45
1.70249
2.40/0.60
2.33547
0.71/0.29
1.48502
0.70/0.30
1.36128
iso ( A˚ 2
U
O/N(2)
x
0.7701(7)
0.2701(7)
0
0.7723(2)
0.2723(2)
0
y
z
site
occ
8h
8h
1.41/0.59
1.68651
1.41/0.59
1.24993
iso ( A˚ 2
U
oB–X–B4 (deg.)
(3x) 180
(3x) 180
(3x) 180
(3x) 180
(3x) 180
(2x) 170.8
(2x) 169.8
(1x) 180
1.52
(1x) 180
2
x
1.29
4.23
3.44
2.28
4.80
3.86
3.94
6.71
5.47
4.14
6.87
5.25
3.10
4.84
3.88
1.84
5.23
3.85
R
R
wp
p
4.76
3.86
For samples x¼0.05–0.80 the space group Pm-3m was used. The samples x¼0.90 and 0.95 were refined in the tetragonal space group I4/mcm.
a
Cationic occupancy according to the chosen substitution level; anionic substitution level calculated from TGA and hotgas extraction.
b
Constraints B(Ti)¼B(Nb).
refinement (Table 1). Similar to our results, Sulaeman et al. [22]
reported on an enlargement of the unit cell parameters with
increasing amount of niobium. In short, the crystal structure
between SrTiO
of the unit cell towards the niobium-rich side of the solid solution
Fig. 4). An octahedral rotation (with possible anion order) is
3 2
and SrNbO N evolves with a linear enlargement
(
initiated between samples 0.80oxo0.90, changing the symme-
try from cubic to tetragonal (monoclinic). As an example, the
refinement plot of samples x¼0.05 and 0.95 are presented in
Fig. 5.
3.2. Thermal stability, reoxidation behaviour and nitrogen content
All oxynitrides in our system showed thermal stability in air
up to Tꢁ400 1C. Above this temperature, the release of nitrogen
and uptake of oxygen lead to reoxidation of the compound. The
seven samples can be divided into three groups according to their
characteristic reoxidation behaviour.
Fig. 3. Synchrotron XRD pattern of oxynitride in samples x¼0.90 and 0.95
showing splitting of higher order reflections. The split tetragonal (I4/mcm) 116,
Group I (x¼0.90 and 0.95) formed a reoxidation intermediate
assigned as B in Fig. 6(a). Characteristic for this intermediate (B) is
the mass gain which is higher than the transformation of an
oxynitride into its corresponding oxide would suggest. Similar
behaviour was also observed previously [8,12,16] and it was
3
32 and 420 reflections correspond to the cubic (Pm-3m) 310 reflection;
l
¼0.44 A˚ .
4
þ
5þ
˚
˚
content (ionic radii r(Ti )¼0.61 A and r(Nb )¼0.64 A) is
detectable as a shift of the reflections towards lower 2
(
y angles
Fig. 2) and this observation could be confirmed by Rietveld

A. Maegli et al. / Journal of Solid State Chemistry 184 (2011) 929–936
933
p
Fig. 4. Enlargement of the unit cell parameter a in the oxynitride samples
crystallizing in the cubic perovskite-type structure (x¼0.05–0.80). The plot
illustrates the existence of a solid solution.
Fig. 6. (a) TGA plots of group
I
(x¼0.90 and 0.95) with A: oxynitride,
B: intermediate, C: oxide (b) MS data of x¼0.90.
release of the bonded nitrogen, finally leading to
oxidized phase.
a fully
To determine the nature of the evolved gases, the MS-signals of
x¼0.90 were tracked (Fig. 6(b)). In the range of 500 1CoTo600 1C,
there is a positive peak in the MS-signal of N
2
(m/z¼28) and
a negative peak in the MS-signal of O
2
(m/z¼32). It illustrates the
release of N and uptake of O , mirroring the sharp mass increase in
2
2
the TGA plot (between stage A and B). The mass drop between stage
B and C (Tꢁ1100 1C) is again accompanied by a small peak in
the MS-signal of N
2
(m/z¼28), which is caused by the complete
2
release of the remaining N . XRD patterns of the reoxidized
products revealed layered perovskite-structures similar to their oxide
precursors.
Group II (x¼0.20, 0.50 and 0.80) samples have in common that
they do not form a stable oxide phase upon heating (Fig. 7(a)).
Samples x¼0.20 and 0.80 possess also an intermediate stage
(
B) but by further temperature increase, the samples’ weight
slowly continue to decrease (C’). Sample x¼0.50 differs most from
all the others because it did not form an intermediate phase and
uninterruptedly lost weight after reaching its maximum.
The MS-signals of sample x¼0.20 were investigated (Fig. 7(b))
and revealed at Tꢁ600 1C a peak for N
2
(m/z¼28). Similar to
is released while forming the intermediate stage.
Group I, N
2
Various smaller signals follow in the range 900 1CoTo1300 1C
suggesting that nitrogen further evolves without reaching a
stability plateau. XRD indicated that only sample x¼0.20 trans-
formed into a cubic structure with space group Pm-3m, the other
two compounds consisted of various oxide phases.
Fig. 5. Rietveld refinement of samples (a) x¼0.05 (Pm-3m) and (b) x¼0.95 I4/
mcm.
Group III (x¼0.05 and 0.10) exhibited the smallest mass
change of all samples but they also consisted of the three distinct
postulated that the intermediate (B) is a partially oxidized phase
that retains dinitrogen weakly bonded in the oxide matrix [33].
Above 1100 1C, the sharp mass decrease originates from the
stages (Fig. 8(a)). However, the mass difference
B and C was only small.
Dm between stage

9
34
A. Maegli et al. / Journal of Solid State Chemistry 184 (2011) 929–936
Fig. 9. Dependence of nitrogen content upon niobium substitution. The circles
represent the experimentally determined nitrogen content y. The dotted line
5
þ
indicates the expected nitrogen values y according to the co-substitution of Nb
3
ꢀ
and N
(Eq. (1)).
In sample x¼0.10 (Fig. 8(b)), no release of N
2
could be
monitored, but the signal of NO (m/z¼30) was observed that is
correlated to the nitrogen release between A and B. All reoxidized
compounds had a cubic structure with space group Pm-3m.
The small weight loss at low temperatures (To500 1C) in some of
the samples (stage A in Fig. 7(a) and Fig. 8(a)) could be attributed to
2 3
the release of H O and NH , which was monitored by the MS-signals
of the mentioned molecules m/z¼18 and m/z¼17, respectively. Such
behaviour has been previously reported elsewhere [16].
Fig. 7. (a) TGA plots of group II (x¼0.20, 0.50 and 0.80) with A: oxynitride,
B: intermediate, C’: unstable phase (b) MS data of x¼0.20.
TGA was additionally used to determine the nitrogen stoichio-
metry (y) using Eq. (2). Hotgas extraction was considered for
group II samples. It was not possible to apply Eq. (2) for group II
samples because stable oxide phases were not formed and they
decomposed into various phases. Fig. 9 indicates the dependence
of nitrogen content upon niobium substitution. The dotted line
represents the theoretically expected nitrogen stoichiometry y.
For most of the samples (except x¼0.05 and 0.20) the experi-
mentally determined nitrogen is slightly below the theoretically
expected value (according to Eq. (1)). A possible explanation is
3
þ
3þ/
that some of the cations are in a reduced state (as Ti and Nb
4
þ
)
and/or that the anionic lattice is oxygen deficient.
3.3. Colour, absorbance spectra and optical band gap
After the ammonia treatment, the oxynitrides underwent a
gradual colour change from bluish/greenish (x¼0.05) to reddish/
brownish (x¼0.95). The absorbance of visible light giving the
individual hue to the samples illustrates that their electronic
structure has been changed compared to the white, insulating
pristine oxides (Table 2).
The absorbance in the oxynitrides is illustrated in Fig. 10.
Samples x¼0.05, 0.10 and 0.20 feature a shoulder at around
375–450 nm. Liu et al. [13,34] reported that this shoulder can be
attributed to isolated N 2p levels, which are energetically located
above the O 2p levels within the band gap of the material. With
increasing nitrogen content, the shoulders disappear because of
complete mixing between N 2p and O 2p levels. The band gap
E
g
is approximated by the absorption minima that originate from
valence- to conduction-band electron-transitions (interband tran-
sitions). The band gap E progressively decreases from 3.24 eV
x¼0.05) to 1.82 eV (x¼0.95). These values are in good agreement
with other reported values for the band gap in SrTiO (E
3.2–3.3 eV) [35–37] and in SrNbO N (E
¼1.8 eV) [38]. Yet, our
g
(
Fig. 8. (a) TGA plots of group III (x¼0.05 and 0.10) with A: oxynitride,
B: intermediate, C: oxide (b) MS data of x¼0.10.
3
g
¼
2
g

A. Maegli et al. / Journal of Solid State Chemistry 184 (2011) 929–936
935
Table 2
Oxynitride samples x¼0.05–0.95 with selected bond angles and electronegativities
w.
Oxynitride
Sample colour
B–X–B bond
angles
B-site cation
a
sample x
average w
0.05
0.10
0.20
0.50
0.80
0.90
0.95
1801
1801
1801
1801
1801
1
1
1
1
1
.543
.546
.552
.570
.588
1
1
70.81
69.81
1.594
1.597
a
The electronegativity
w for the B-site cation was averaged by its Ti/Nb ratio.
g
Fig. 11. Band gap energy E vs. nitrogen content (y).
Each sample features an absorbance after the band gap E
towards longer wavelengths as shown in Fig. 10. Likewise to our
samples, an absorption of wavelengths longer than E was reported
by Liu et al. [34], who attributed it to the presence of oxygen
vacancies in nitrogen doped SrTiO . Also Kim et al. [9] and
Siritanaratkul et al. [38] detected a similar absorption in SrNbO N,
BaNbO N and CaNbO N. They correlated it with the introduction of
charge carriers into the conduction band, which in turn reflects
g
g
3
2
2
2
5
þ
2
a partial reduction of Nb . Also for absorption spectra of TiO an
increasing background absorbance towards longer wavelengths was
reported and it was attributed to free carriers injected by reduced
titanium ions [39]. The reason for the absorbance in our samples
cannot be identified unambiguously because oxygen vacancies as
well as partial reduction of the B-site cations are conceivable.
Fig. 11 demonstrates that the nearly linear trend of the mixed N
2
0
p and O 2p levels is interrupted by the compounds x¼0.50 and
.80, which possess a comparatively small band gap. This observa-
tion can be explained if besides nitrogen content, the electronega-
tivity of the B-site cation and the crystal symmetry are taken into
account (Table 2). Throughout the solid solution, the titanium cation
¼1.54) is increasingly replaced by the more electronegative
niobium cation (
w
(w
w
¼1.6). A more electronegative transition metal
cation has its t2g orbitals closer to the anionic p orbitals leading to
more covalent bonds and a narrower band gap [14,16]. Therefore,
with progressive niobium content, the band gap E
diminished by the larger nitrogen abundance but additionally by the
increasing electronegativity of the B-site cation. This effect is
g
is not only
w
observable in samples x¼0.50 and 0.80. Yet, there is no continuation
of this trend for x¼0.90 and 0.95 as it would be suggested by the
even more increasing electronegativity w. Crystal structure investi-
gations revealed that the solid solution transforms from Pm-3m to
I4/mcm between x¼0.80 and 0.90. B–X–B bond angles of 1801
provide the best orbital overlap and therefore strongest covalence
for bonds between N/O p orbitals and metal d orbitals. Distortions to
bond angles (1801ꢀ
j
) reduce the overlap and covalent character,
wider [14,16]. Rietveld refinement
thus making the band gap E
g
provided B–X–B bond angles of 170.81 and 169.81 for x¼0.90 and
Fig. 10. Absorbance spectra of oxynitride samples (a) x¼0.05–0.20 and (b) x¼
.50–0.95. The spectra were obtained from UV–vis diffuse reflectance measurements.
0.95, respectively. Consequently, in those samples, the decreasing
0
crystal symmetry (increase of E
electronegativity (decrease of E
the band gap E predominantly on nitrogen content.
g
) is countervailing the increasing
w
g
), thus providing a dependence of
g
results differ slightly from the ones obtained by Sulaeman
et al. [22]. As they reported, the nitrogen content did not change
much with increasing niobium ‘‘doping’’ and as a consequence the
4. Conclusions
g
band gap E did not continously decrease. The differences in the
synthesis route (microwave assisted solvothermal synthesis by
Sulaeman et al. [22]) might lead to the observed discrepancies.
Perovskite-type structures with seven different compositions
of SrTi1ꢀxNb (O,N) were successfully prepared via thermal
x
3

9
36
A. Maegli et al. / Journal of Solid State Chemistry 184 (2011) 929–936
ammonolysis of oxide precursors. The crystal structure/symmetry
of the oxynitride series changed from cubic Pm-3m to tetragonal
I4/mcm by initiation of an octahedral tilting between x¼0.80 and
[8] D. Logvinovich, S.G. Ebbinghaus, A. Reller, I. Marozau, D. Ferri, A. Weidenkaff,
Z. Anorg. Allg. Chem. 636 (2010) 905–912.
[9] Y.-I. Kim, P.M. Woodward, K.Z. Baba-Kishi, C.W. Tai, Chem. Mater. 16 (2004)
1267–1276.
0
.90. The tilting increased from x¼0.90 towards x¼0.95 and
[10] F. Chevir e´ , F. Tessier, R. Marchand, Eur. J. Inorg. Chem. 2006 (2006)
1223–1230.
the refined cell parameters showed a continuous enlargement
towards higher niobium and nitrogen content.
[
11] I.D. Fawcett, K.V. Ramanujachary, M. Greenblatt, Mater. Res. Bull. 32 (1997)
565–1570.
1
The amount of nitrogen insertion could be adjusted by
appropriate co-substitution with niobium. TGA and hotgas extrac-
tion revealed that rising niobium content favoured nitrogen
incorporation.
[
12] A. Rachel, S.G. Ebbinghaus, M. G u¨ ngerich, P.J. Klar, J. Hanss, A. Weidenkaff,
A. Reller, Thermochim. Acta 438 (2005) 134–143.
13] C.M. Liu, X.T. Zu, W.L. Zhou, J. Phys. D 40 (2007) 7318–7322.
14] H.W. Eng, P.W. Barnes, B.M. Auer, P.M. Woodward, J. Solid State Chem. 175
[
[
(2003) 94–109.
Nitrogen insertion could lower the energy gap E
valence and conduction band into the range of visible light as
shown by the change from white oxides into coloured oxynitride
g
between
[15] C.M. Fang, E. Orhan, G.A. de Wijs, H.T. Hintzen, R.A. de Groot, R. Marchand,
J.Y. Saillard, G. de With, J. Mater. Chem. 11 (2001) 1248–1252.
[16] R. Aguiar, D. Logvinovich, A. Weidenkaff, A. Rachel, A. Reller, S.G. Ebbinghaus,
Dyes Pigm. 76 (2008) 70–75.
powders. The width of E
g
was correlated with the nitrogen
[17] M. Jansen, H.P. Letschert, Nature 404 (2000) 980–982.
[
[
18] E. Guenther, M. Jansen, Mater. Res. Bull. 36 (2001) 1399–1405.
19] A. Kasahara, K. Nukumizu, G. Hitoki, T. Takata, J.N. Kondo, M. Hara,
H. Kobayashi, K. Domen, J. Phys. Chem. A 106 (2002) 6750–6753.
20] R. van de Krol, J. Schoonman, in: K. Hanjalic, R. van de Krol, A. Lekic (Eds.),
Sustainable Energy Technologies: Options and Prospects, Dordrecht, 2008,
pp. 121–142.
21] M. Miyauchi, M. Takashio, H. Tobimatsu, Langmuir 20 (2003) 232–236.
22] U. Sulaeman, S. Yin, T. Sato, Adv. Sci. Tech. 63 (2010) 52–57.
23] Y.A. Abramov, V.G. Tsirelson, V.E. Zavodnik, S.A. Ivanov, I.D. Brown, Acta
Crystallogr. B 51 (1995) 942–951.
amount and it diminished from 3.24 eV down to 1.82 eV with
increasing nitrogen (and niobium) content. Other factors besides
the nitrogen presence, e.g. the electronegativity
w
and symmetry
[
changes also influenced the band gap E
g
and explained some of
the non-linear variations.
[
[
[
Acknowledgments
[
[
24] S.G. Ebbinghaus, A. Weidenkaff, A. Rachel, A. Reller, Acta Crystallogr. C 60
(2004) i91–i93.
The authors thank Dr. Fabia Gozzi of MS beamline at the Swiss
Light Source for experimental assistance. We are also grateful to
Renato Figi from the Laboratory for Analytical Chemistry (Empa)
for support in hotgas extraction and to Dr. Yaroslav Romanyuk
from Laboratory for Thin Films and Photovoltaics (Empa) for
assistance in diffuse reflectance measurements. This research
has been funded by the European Commission (Project NanoPEC,
Contract number 227179).
25] M. Yang, J. Oro-Sole, J.A. Rodgers, A.B. Jorge, A. Fuertes, J.P. Attfield, Nat.
Chem. 3 (2011) 47–52.
[26] K. Masato, O. Toru, A. Momoko, N. Yoshiyuki, Y. Masatomo, M. Yoshimura,
J. Sol–Gel Sci. Technol. 12 (1998) 95–109.
[
[
27] J. Rodriguez-Carvajal, Physica B 192 (1993) 55–69.
28] P. Thompson, D.E. Cox, J.B. Hastings, J. Appl. Cryst. 20 (1987) 79–83.
[29] A.B. Murphy, Sol. Energy Mater. Sol. Cells 91 (2007) 1326–1337.
[
[
[
30] I.P. Shapiro, Opt. Spektrosk. 4 (1958) 256.
31] A. Glazer, Acta Crystallogr. B 28 (1972) 3384–3392.
32] R.H. Mitchell, Perovskites: Modern and ancient, Almaz Press Inc., Thunder
Bay, Ontario, 2002, pp. 13–73.
[
[
33] F. Tessier, L. Le Gendre, F. Chevir e´ , R. Marchand, A. Navrotsky, Chem. Mater.
References
17 (2005) 3570–3574.
34] C.M. Liu, Z.L. Yuan, Y.F. Ju, X. Xiang, X.T. Zu, W.L. Zhou, Can. J. Phys. 86 (2008)
1345–1348.
[35] M. Miyauchi, A. Nakajima, A. Fujishima, K. Hashimoto, T. Watanabe, Chem.
Mater. 12 (2000) 3–5.
[36] J.A. Noland, Phys. Rev. 94 (1954) 724.
[37] Y.Y. Mi, Z. Yu, S.J. Wang, X.Y. Gao, A.T.S. Wee, C.K. Ong, C.H.A. Huan, J. Appl.
Phys 101 (2007) 063708-1–063708-5.
[
[
[
[
[
1] J.G. Bednorz, K.A. M u¨ ller, Z. Phys. B 64 (1986) 189–193.
2] R.E. Cohen, Nature 358 (1992) 136–138.
3] M. Uehara, S. Mori, C.H. Chen, S.W. Cheong, Nature 399 (1999) 560–563.
4] Y.-I. Kim, P.M. Woodward, J. Solid State Chem. 180 (2007) 3224–3233.
5] R. Aguiar, A. Kalytta, A. Reller, A. Weidenkaff, S.G. Ebbinghaus, J. Mater. Chem.
1
8 (2008) 4260–4265.
[
[
6] S.G. Ebbinghaus, H.-P. Abicht, R. Dronskowski, T. M u¨ ller, A. Reller,
A. Weidenkaff, Prog. Solid State Chem. 37 (2009) 173–205.
7] D. Logvinovich, M.H. Aguirre, J. Hejtmanek, R. Aguiar, S.G. Ebbinghaus,
A. Reller, A. Weidenkaff, J. Solid State Chem. 181 (2008) 2243–2249.
[38] B. Siritanaratkul, K. Maeda, T. Hisatomi, K. Domen, ChemSusChem 4 (2011)
74–78.
[39] T. Sekiya, K. Ichimura, M. Igarashi, S. Kurita, J. Phys. Chem. Solids 61 (2002)
1237–1242.