Oxygen stoichiometry and the problem of the growth of (VO)2P2O7 single crystals

Article abstract of DOI:10.1016/S0025-5408(00)00388-3

The crystal growth of (VO)₂P₂O₇ single crystals is discussed with respect to the formation of a high-temperature nonstoichiometric phase (VO)₂P₂O_7+x. The interaction between vanadyl pyrophosphate and oxygen was investigated in the temperature region 600-900 °C. It was found that the decay of the high-temperature nonstoichiometric phase at lower temperature led to imperfections in crystals grown in an atmosphere with oxygen concentration higher than 0.2%. Two mechanisms for the decay are considered. The oxygen content of the crystals may be reduced at the surface by evolution of gaseous oxygen, whereas in the bulk the formation of an oxygen-enriched solid phase is observed.

Full text of DOI:10.1016/S0025-5408(00)00388-3

Pergamon
Materials Research Bulletin 35 (2000) 1859–1868
Oxygen stoichiometry and the problem of the growth of
(VO)2P2O7 single crystals
A.V. Prokofiev*, F. Bu¨llesfeld, W. Assmus
Physikalisches Institut, Universita¨t Frankfurt, 60054, Frankfurt, Germany
(Refereed)
Received 3 November 1999; accepted 27 December 1999
Abstract
The crystal growth of (VO)2P2O7 single crystals is discussed with respect to the formation of a
high-temperature nonstoichiometric phase (VO)2P2O7ϩx. The interaction between vanadyl pyrophos-
phate and oxygen was investigated in the temperature region 600–900°C. It was found that the decay
of the high-temperature nonstoichiometric phase at lower temperature led to imperfections in crystals
grown in an atmosphere with oxygen concentration higher than 0.2%. Two mechanisms for the decay
are considered. The oxygen content of the crystals may be reduced at the surface by evolution of
gaseous oxygen, whereas in the bulk the formation of an oxygen-enriched solid phase is observed.
© 2000 Elsevier Science Ltd. All rights reserved.
Keywords: B. Crystal growth; D. Phase equilibria; D. Microstructure; D. Defects
1. Introduction
(VO)2P2O7 (VOPO) is interesting as a low-dimensional spin system and is widely used as
a catalyst for production of maleic anhydrate. The interpretation of the magnetic properties
has been revised several times. On account of its structure [1] and properties, (VO)2P2O7 was
considered for several years as a one-dimensional (1D) system with a spin gap—either a spin
ladder or an alternating spin chain [2,3]. The ladder model was finally rejected and the
alternating chain model was accepted [4]. The existence of a second gap observed in this
earlier work was associated with the two-magnon bound state. Later, on the basis of
* Corresponding author. Guest scientist from A.F. Ioffe Physical Technical Institute, St. Petersburg 194021,
Russia.
0025-5408/00/$ – see front matter © 2000 Elsevier Science Ltd. All rights reserved.
PII: S0025-5408(00)00388-3

1860
A.V. Prokofiev et al. / Materials Research Bulletin 35 (2000) 1859–1868
monoclinic structure [1], it was found that two crystallographically different types of V–O
chains exist in VOPO prepared at ambient pressure [5,6]. This leads to an alternative
explanation of the existence of two spin gaps. Recently, a high-pressure modification of
VOPO with the crystallographically single type of the V–O chains was found. This modi-
fication has a single gap [7].
The vanadium valence is one of the key problems in the properties of (VO)2P2O7. It was
observed [8,9] that (VO)2P2O7 samples calcined in air (showing X-ray diffraction patterns
proper for the single phase (VO)2P2O7) have, after oxidation to VOPO4, a mass increase less
than that expected from the reaction
͑VO͒2P2O7 ϩ 1/2 O2 ϭ 2VOPO4
(1)
The authors proposed that the phase of the variable excessive oxygen content exists, on the
basis of the (VO)2P2O7 structure. The homogeneity range of the phase can reach the
composition (VO)2P2O7.5. A much narrower range of the oxygen content has been reported
by other authors [10,11] who investigated the problem of VOPO nonstoichiometry in
connection with the structure defectiveness.
It has been supposed [8] that the nonstoichiometric phase is a “hybrid crystal,” consisting
of microdomains of VOPO4 phase and of stoichiometric (VO)2P2O7. The two structures fit
closely, which makes the phase boundary coherent. However, electron spin resonance (ESR)
investigations have indicated [9] a tight neighborhood or pairing of the ions of both valences
in the homogeneous phase. Nuclear magnetic resonance (NMR) studies [11] have showed
that Vϩ5 in VOPO is closely associated with Vϩ4 within the single phase. The defects that
cause the oxygen nonstoichiometry are those connected with conversion of pyrophosphate
groups P2O7 to two PO4 groups [10,11]:
ϩ5
͑Vϩ4
V
xO͒ ͑P2O7͒1Ϫx͑PO4͒2x
2
1Ϫx
Up to now, the width of the homogeneity range, stability, and the microstructure of the
mixed-valence phase remain insufficiently understood. In most of the previous works,
powder VOPO samples were investigated at temperatures far lower than the melting point.
The behavior of VOPO above 800°C up to the melting point remains to be investigated. This
temperature region is of the greatest importance for the growth of VOPO crystals. In the
present work, we tried to analyze the relationship between crystal growth conditions,
physical properties, and composition of the VOPO crystals with respect to formation of the
nonstoichiometric phase. For the analysis, our earlier partially published data as well as
optical transmission spectra and phase diagrams measured in this work were used.
2. Experimental
The oxygen content in the samples was determined from the measured mass change of a
sample with a Netzsch thermal analyzer. The atmospheres were prepared either by controlled
mixing of argon and oxygen or by diluting industrially prepared calibrated mixtures of argon
and oxygen, with argon. The composition of the atmosphere in the chamber was measured
by mass spectrometry. The range of the oxygen concentration was 0.5–4 vol%. Powder

A.V. Prokofiev et al. / Materials Research Bulletin 35 (2000) 1859–1868
1861
samples of 60–80 mg in a shallow alumina crucible were heated stepwise at various
temperatures in atmospheres with fixed oxygen concentrations. The exposure at every
temperature was 10–12 h. The oxygen content was determined from the increase of the mass
of the originally stoichiometric vanadyl pyrophosphate, which was obtained by decomposi-
tion at 700°C in argon of the precursor VO(HPO4)⅐0.5H2O. The latter was prepared from
inorganic reagents according to technique described by Centi et al. [12].
Optical transmission spectra in the range of 190–900 nm were measured with Perkin-
Elmer spectrometer (Lambda 16) using polished plates cut from the single crystals. The
thickness of the plates was about 0.4 mm.
3. Results and discussion
3.1. Problems arising from the characterization of as-grown VOPO single crystals
Single crystals of (VO)2P2O7 up to centimeter-size were grown earlier from the melt in an
atmosphere with low oxygen concentration [13]. The growth procedure was based on our
preliminary studies of the stability conditions of (VO)2P2O7 just below the melting point.
X-ray diffraction of the powder samples had showed that (VO)2P2O7 remains single phase
up to the melting point if the atmosphere contains an oxygen concentration below 0.7%. On
the basis of this fact, the crystals were grown in an atmosphere with various concentration
of oxygen in the range 0.2–0.7%.
The subsequent characterization of the grown crystals gave rise to many questions. Some
of the crystals were transparent and firm, with mirror facets. However, some of the large
crystals were opaque and brittle. According to X-ray diffraction, the crystals of both types
were single phase. They did not differ in quality according to Laue pictures. Several days
after growth, large crystals formed cracks and gradually broke into small pieces. Some
inclusions of a second phase of yellow color could be seen on the crack surface of the
remaining pieces. These inclusions induced stress in the crystals that led to the crack
formation [13].
The second phase could not be detected by powder X-ray diffraction of the crystals. Even
after being concentrated by a scraping from the blue-green (VO)2P2O7 crystals, this phase
gives in the diffractograms only a few weak impurity peaks on the background of strong
VOPO peaks [13]. These impurity peaks might be related to various VOPO4 phases. The
yellow color of the phase indicates a vanadium valence of 5 for the compound. Energy-
dispersive X-ray (EDX) and wavelength-dispersive X-ray (WDX) analyses of the yellow
inclusions showed a phosphorus-to-vanadium atomic ratio equal to 1. These results strongly
suggest that this is a phase with the composition VOPO4. This phase could not be formed
during crystallization from the melt at about 900°C because of its thermal instability above
750°C [8,10,14]. We do not think that this phase was formed during the long cooling of the
furnace after the pulling of the crystals from the melt. Otherwise, the external surface of the
grown boule should be oxidized to a large degree. Nevertheless, all mirror facets remained
optically perfect, whereas inclusions were observed in the bulk.
Since our crystals were grown in atmospheres with various oxygen concentrations, it was

1862
A.V. Prokofiev et al. / Materials Research Bulletin 35 (2000) 1859–1868
natural to expect different oxygen content in the crystals. Indeed, depending on the oxygen
concentration, the vanadium oxidation state measured by thermogravimetric analysis (TGA)
was in the range 4.0–4.4. The color of the powdered samples varied from green-blue to
green. No red-brown crystals were observed, in contrast to other studies [10,15].
The oxygen content in the crystals was determined by measuring the mass increase after
oxidation to VOPO4, or the mass decrease after annealing in argon, of ground samples [13].
The annealing of the ground samples in argon changed their color from green to green-blue.
The samples for such analysis were taken from the polycrystalline part of the crystal batch.
The result of the test was assigned to the whole batch, on account of the equilibrium
conditions and temperature homogeneity of the growth space. We did not succeed in
complete dissolution of our crystals in order to analyze them with reduction–oxidation
titration according to Nakamura et al. [16].
Despite the fact that the crystals had different compositions, the powder X-ray diffraction
analysis showed the same cell dimensions. No difference in either peak positions or peak
widths was found for the crystals grown in atmospheres with different oxygen concentrations
(although a different relation between the peak intensities can be observed).
The measurements of the magnetic susceptibility and ESR investigations of the crystals
showed only a small difference between the samples of different stoichiometry. This was
strange, because the presence of non-magnetic Vϩ5 ions indicates defects exist in the
low-dimensional spin system. The influence of the defects on the magnetic properties of
(VO)2P2O7 is known. The ideal (VO)2P2O7 has the non-magnetic singlet ground state at T ϭ
0 and the magnetic susceptibility trend to 0 with T 3 0. The defects (the edge states of the
spin chains) are free paramagnetic Vϩ4 ions, and their contribution to ␹ at low temperature
increases, resulting in the increasing of the total ␹. Fig. 1 shows the difference in the
magnetic susceptibility of three defects and the ideal (VO)2P2O7 [17,2]. In the case given in
ref. 17, the various degrees of defectiveness were achieved by annealing at different
temperatures and duration in inert atmosphere of the ill-crystalline VOPO. Our crystals with
quite different among themselves stoichiometry (according to TGA) behaved like low-defect
alternating spin chain systems (Fig. 2) [18,19]. Evaluation of the low-temperature Curie
upturn of the magnetic susceptibility gives the concentration of the free Vϩ4 spins as low as
about 1.0 and 1.3% (for the crystals with the vanadium oxidation state 4.05 and 4.35, grown
in atmosphere with 0.2 and 0.65% O2, respectively).
In Fig. 3, the optical transmission spectra of three single crystals grown in two different
atmospheres are presented. Curve 1 represents a sample grown in the atmosphere with the
oxygen concentration 0.2%. The thin plate was cut from a large crystal grown from the melt.
Side by side with the large crystals grown from the melt, small needle-like (0.5–1.5 mm in
width) free standing crystals covered the surface of the crystal boule. We considered them
as grown from the gas phase (such crystals can be seen in the photo given in ref. 13). Usually
these crystals were optically perfect, whereas the melt-grown crystals looked opaque. Curve
3 shows the transmittance of the sample cut from such a small crystal grown in the same
atmosphere as the melt-grown crystal 1 (the same growth run). Crystal 2, which is also
needle-like, was grown in the atmosphere with oxygen concentration 0.65%. The thickness
of all plates was nearly the same.
The spectra of the small needle-like crystals grown in atmospheres with different oxygen

A.V. Prokofiev et al. / Materials Research Bulletin 35 (2000) 1859–1868
1863
Fig. 1. Magnetic susceptibility of (VO)2P2O7 vs. temperature (a) for three samples with different degree of
crystallinity achieved by different regimes of annealing of ill-crystalline (VO)2P2O7. The crystallinity increases
with the sample number [17]. (b) Measured data with subtracted paramagnetic term introduced by magnetic
defects [2]. The solid line is a calculation according to the alternating chain model.
concentrations (curves 2 and 3) look quite similar. On the other hand, there is a pronounced
qualitative difference between the crystals grown in the same atmosphere (curves 1 and 3).
The plates cut from the large crystals show lower transparency than those from the small
crystals. The oxygen content of the samples could not be measured because of the small size
of the plates. The vanadium oxidation state of the measured crystals was assumed from the
TGA measurements of the polycrystalline part of the batches. These were 4.35 and 4.05 for
the crystals grown in atmospheres with oxygen content of 0.65 and 0.2%, respectively.
Study of the nonstoichiometry of vanadyl pyrophosphate could help to explain these facts.
3.2. Oxidation-reduction equilibrium in the system (VO)2P2O7–VOPO4
The P-T-x diagram of the (VO)2P2O7–VOPO4 reduction–oxidation system was system-
atically investigated earlier [8,10], in the range of the oxygen partial pressure of 10Ϫ4–1 atm

1864
A.V. Prokofiev et al. / Materials Research Bulletin 35 (2000) 1859–1868
Fig. 2. Temperature dependence of the magnetic susceptibility of the VOPO single crystals with different
vanadium oxidation states, measured thermogravimetrically: (1) 4.05 [18]; (2) 4.35 [19].
at temperatures up to 790°C. However, for crystal growth, the region of higher temperature
is of more interest. As in previous work [8], we used the thermogravimetric method.
Although TGA gives only the brutto content of oxygen and not the phase composition of the
samples, it is a direct high temperature method of measuring oxygen content. The results are
presented in Fig. 4.
At 550°C, VOPO begins to oxidize to VOPO4 only if the oxygen concentration in the
atmosphere exceeds 1%, and this compound remains stable up to 700–730°C (region A,
partially shown). At 730–750°C (region B) VOPO4 reduces to a substance with a lower
vanadium oxidation state. In an atmosphere with oxygen concentration higher than 2.5% the
samples melt before 770°C. For oxygen concentrations 2.5% and lower, the situation is
different. The samples remain solid up to the temperature 880°C. In the atmosphere with 2%
oxygen VOPO4 decomposes to the stoichiometric solid (VO)2P2O7.
The region of the temperatures Ͼ 750°C (region C) is the most important. The liquid
samples continue to lose oxygen with increasing temperature. The opposite process takes
place in the solid samples. These samples absorb oxygen. The curves corresponding to 0.5,
1, and 2% oxygen show that absorption takes place irrespective of the pre-history of VOPO,
that is, whether samples were obtained from VO(HPO4)⅐0.5H2O (0.5 and 1%) or by
decomposition of VOPO4 (2%). The solid sample cooled quickly to room temperature
showed the diffraction pattern corresponding to single phase (VO)2P2O7.

A.V. Prokofiev et al. / Materials Research Bulletin 35 (2000) 1859–1868
1865
Fig. 3. Transmission optical spectra of the VOPO single crystals. (1) The plate was cut from a large single crystal
grown from the melt in atmosphere with 0.2% O2; (2) the plate was cut from a small single crystal grown in
atmosphere with 0.65% O2; (3) the plate was cut from a small single crystal grown in atmosphere with 0.2% O2.
It is the most probable that in region C the formation of the nonstoichiometric single-phase
vanadyl pyrophosphate takes place. There are no other oxygen-enriched compounds with the
ratio P/V ϭ 1 except the numerous phases with stoichiometry VOPO4. But all VOPO4 phases
have decomposed at temperature higher 750°C.
One can see that the absorption of oxygen occurs with the increasing of the temperature,
a behavior that is not typical from the point of view of thermodynamics. An increase in
temperature advances the endothermic process with positive change of the entropy. Absorp-
tion of oxygen by VOPO leads to a decrease of the entropy of the gas phase. Hence, the
entropy of the solid phase should increase sharply to provide the total positive balance of the
entropy of the process. This means an anomalous increasing of disorder (configuration
entropy) in the solid phase with increasing temperature beginning at 750°C. Such anomalous
disorder is proper for strongly nonstoichiometric phases. The behavior of the solid samples
contrasts with that of the liquid samples. The temperature increase does not lead to a drastic
disordering in liquids and they continue to evolve oxygen.
In the temperature region below 790°C, our results are in satisfactory agreement with
those previously reported [8,10]. However, our conclusion regarding the increasing of
disorder and nonstoichiometry with temperature and oxygen pressure does not agree with

1866
A.V. Prokofiev et al. / Materials Research Bulletin 35 (2000) 1859–1868
Fig. 4. Dependence of the vanadium oxidation state in VOPO on the temperature for various concentrations of
oxygen in the atmosphere. The O2 concentration is given in volume percent for every curve. Experimental points
for molten samples are marked by the letter “m.”
that of ref. 10. The authors of that study consider that the oxygen nonstoichiometry is
rigidly fixed by defects formed in the stage of the topotactic transformation of the
precursor VO(HPO4)⅐0.5H2O to (VO)2P2O7, and this cannot be changed significantly by
subsequent annealing in atmospheres with various oxygen concentrations. However, as
is seen from Fig. 4, our results are not contrary to this thesis if the temperature does not
exceed 800°C and the oxygen concentration is lower than 1%. The authors did not define
the range of the oxygen partial pressure in which they observed this invariance of the
oxygen stoichiometry.
3.3. High temperature nonstoichiometry phase, crystal quality and characterization
problems
It can be concluded from Fig. 4 that the nonstoichiometric phase (microdomain or really
homogeneous) with a rather large deviation from the stoichiometry (VO)2P2O7 does exist.
On the other hand, we did not find any direct evidence of the nonstoichiometry in the
physical properties (measured at the room and low temperatures). This leads us to the
conclusion that the nonstoichiometric phase (VO)2P2O7ϩx formed at temperatures close to
the melting point of vanadyl pyrophosphate is unstable and decomposes, losing oxygen in the
stoichiometric compound (VO)2P2O7 at lower temperatures.
There are two possible ways for the phase to become rid of excess oxygen. The first is the
evolution of the gaseous oxygen
͑VO͒2P2O7ϩx ϭ ͑VO͒2P2O7 ϩ ͑ x/2͒O2 ͑ gas–phase mechanism͒
(2)

A.V. Prokofiev et al. / Materials Research Bulletin 35 (2000) 1859–1868
1867
The second possibility, which can be realized only below 750°C, is a decay in two solid
phases, one of them having higher oxygen content than the other (disproportion)
͑VO͒2P2O7ϩx ϭ ͑1 Ϫ x͒͑VO͒2P2O7 ϩ ͑2x͒VOPO4 ͑solid-state mechanism͒ (3)
As mentioned above, we found admixtures of VOPO4 phase after the growth of our crystals.
This fact is indicative of the decomposition according to the solid-state mechanism.
The hypothesis on the decay of the high temperature (VO)2P2O7ϩx phase according to
both mechanisms explains many facts that were not understandable earlier. It gives an
explanation of the breaking of the crystals. Decomposition in stoichiometric (VO)2P2O7 and
VOPO4 (or oxygen) explains why the measurements (X-ray and magnetic) could not detect
the difference in the properties of the crystals with different stoichiometry. It now becomes
clear why the external surface of the crystal boule remains brilliant, while the inner parts of
the crystals are opaque and have inclusions of the foreign phase.
Let us consider the two possible mechanisms of the decomposition of the nonstoichio-
metric phase in more detail. The gas–phase mechanism requires a diffusion of oxygen atoms
from the inner part of the crystal to the surface. The appropriate conditions for its realization
exist only in the surface layer of the crystal, and it is this part of the crystal where it is
realized. The mechanism preserves the surface layer of the crystal from the inclusions of the
VOPO4 phase. The diffusion is quite slow at the temperatures at which the nonstoichiometric
phase begins to decompose. Taking into account the size of our crystals, oxygen cannot pass
a distance of several mm from the bulk to reach the surface. Therefore, the solid-state
mechanism takes place in the inner part of the crystals. With decreasing temperature, the
diffusivity is reduced and the precipitation of the oxygen-enriched phase VOPO4 takes place
on an increasingly local scale.
The optical spectra of different crystals can be understood from this position. The plates cut
from small free-standing crystals are optically more perfect than specimens from the bulk due to
the greater role of the free surface of the former. The inclusions of the VOPO4 phase in the plate
cut from the bulk show strong light scattering, which leads to a lowering of the transparency.
The question regarding the structure of the high-temperature nonstoichiometric phase
remains open. The single crystals, even centimeter-sized, retain their monocrystallinity after
the loss of oxygen. This indicates that the high-temperature nonstoichiometric phase has the
same or similar structure as stoichiometric (VO)2P2O7. High-temperature X-ray diffraction
study of VOPO is now in progress.
Acknowledgement
We are grateful to H. Schwenk for provision of magnetic susceptibility data. The financial
support of SFB 252 is also gratefully acknowledged.
References
[1] P.T. Nguyen, R.D. Hoffman, A.W. Sleight, Mater Res Bull 30 (1995) 1055.
[2] D.C. Johnston, J.W. Johnson, D.P. Goshorn, A.J. Jacobson, Phys Rev B 365 (1987) 219.

1868
A.V. Prokofiev et al. / Materials Research Bulletin 35 (2000) 1859–1868
[3] T. Barnes, J. Riera, Phys Rev B 50 (1994) 6817.
[4] A.W. Garret, S.E. Nagler, D.A. Tennant, B.C. Sales, T. Barnes, Phys Rev Lett 79 (1997) 745.
[5] Z. Hiroi, M. Azuma, Y. Fujishiro, T. Saito, M. Takano, J Solid State Chem 146 (1999) 369.
[6] J. Kikuchi, K. Motoya, T. Yamauchi, Y. Ueda, Phys Rev B 60 (1999) 6731.
[7] M. Azuma, T. Saito, Y. Fujishiro, Z. Hiroi, M. Takano, F. Izumi, T. Kamiyama, T. Ikeda, Y. Yarumi, K.
Kuido, Phys Rev B 60 (1999) 10145.
[8] E. Bordes, P. Courtine, J Catal 57 (1979) 236.
[9] M. Lopez Granados, J.C. Conesa, M. Fernandes-Garcia, J Catal 141 (1993) 671.
[10] P.T. Nguyen, A.W. Sleight, in: B.K. Warren, S. Ted Oyama (Eds.), Heterogeneous Hydrocarbon Oxidation,
ACS Symposium Series 638, American Chemical Society, Washington, DC, 1996, p. 236.
[11] P.T. Nguyen, A.W. Sleight, N. Roberts, W.W. Warren, J Solid State Chem 122 (1996) 259.
[12] G. Centi, F. Trifiro, G. Poli, Appl Catal 19 (1985) 225.
[13] A.V. Prokofiev, F. Bu¨llesfeld, W. Assmus, Cryst Res Technol 33 (1998) 157.
[14] G. Koyano, T. Okuhara, M. Misono, J Am Chem Soc 120 (1998) 767.
[15] J.R. Ebner, M.R. Thompson, in: R.K. Grasselli, A.W. Sleight (Eds.), Studies in Surface Science and
Catalysis, Elsevier, New York, 1992, p. 18.
[16] M. Nakamura, K. Kawai, Y. Fujiwara, J Catal 34 (1974) 345.
[17] D.C. Johnston, J.W. Johnson, J Chem Soc Chem Commun (1985) 1720.
[18] A.V. Prokofiev, F. Bu¨llesfeld, W. Assmus, H. Schwenk, D. Wichert, U. Lo¨w, B. Lu¨thi, Eur Phys J B 5
(1998) 313–316.
[19] H. Schwenk, unpublished data.