ISSN 1063-7745, Crystallography Reports, 2008, Vol. 53, No. 7, pp. 1112–1118. © Pleiades Publishing, Inc., 2008.
STRUCTURE
OF INORGANIC COMPOUNDS
Transformation of the Corundum Structure upon HighTemperature Reduction
A. Ja. Dan’ko, M. A. Rom, N. S. Sidelnikova, S. V. Nizhankovskiy,
A. T. Budnikov, L. A. Grin’, and Kh. Sh-o. Kaltaev
Scientific and Technological Corporation “Institute for Single Crystals,” National Academy of Sciences of Ukraine,
pr. Lenina 60, Kharkov, 61001 Ukraine
e-mail: danko@isc.kharkov.ua
Received November 1, 2006
Abstract—This paper reports on the results of investigations into the transformation of the corundum structure
upon reducing annealing at high temperatures in the range from 1700 to 2050°C. It is established that the reduction results in the transformation of corundum into new phases with a lower oxygen content, including the phase
with a spinel structure. These structures are assumed to be stabilized by anion vacancies. A model of the crystal
structure of the spinel phase is proposed. This model provides an adequate description of the compound
obtained in the experiment.
PACS numbers: 61.66.-f
DOI: 10.1134/S1063774508070031
At present, the only well-studied oxide that exists in
the condensed state in the aluminum–oxygen system is
aluminum oxide Al2O3. The data available in the literature on the existence of oxide compounds of aluminum
with the lowest valence in the condensed state are very
scarce. Filonenko et al. [1] were the first to report on the
formation of the AlAl2O4 phase (the alumina spinel)
upon the carbothermal reduction of corundum. There
are only few publications containing information on the
preparation of some compounds (including the aforementioned phase) in the aluminum–oxygen system;
however, up to now, special investigations in this direction have not been performed. The present study continues a series of investigations [2–4] concerning the
transformation of corundum into phases with a different symmetry upon reducing annealing at high temperatures.
and reducing additives (CO, H2, and CO + H2). In the
case when the chamber was filled with argon, the formation of the reducing medium in the chamber
occurred spontaneously as a result of the interaction of
adsorbed oxygen and water vapors with carbon–graphite materials. The concentration of the reducing component in this case did not exceed 5%. The concentration
of the reducing component (CO + H2) in the gaseous
medium was determined with the use of a Kristall
2000M gas chromatograph. The duration of annealing
was varied from 2 to 10 h. The temperature of the samples was controlled using a Marathon MRISCSF integrated infrared pyrometer. The X-ray powder diffraction analysis was carried out on a DRON-1.5 diffractometer (Cu K α1, 2 radiation, pyrolytic graphite(002)
monochromator, θ–2θ scan mode). The thickness of the
polycrystalline zone formed on a sapphire substrate
was determined according to standard techniques [2].
OBJECTS AND METHODS OF INVESTIGATION
RESULTS AND DISCUSSION
For our investigations, we used plane-parallel sapphire samples and layers of finely dispersed α-Al2O3
powders (with particle sizes of ~15–25 µm) in which
the content of majority impurities was less than or equal
to 10 ppm. Annealing was performed in a furnace
equipped with carbon–graphite heat screens [5] at temperatures in the range from 1700 to 2050°C. After the
preliminary vacuum technical preparation was accomplished and the residual pressure (~0.1 Torr) was
attained, the chamber was filled up to a pressure of
800 Torr either with argon or with a mixture of argon
The X-ray powder diffraction analysis revealed that,
upon the reducing annealing, the near-surface layer of
the single-crystal samples transforms into a polycrystalline layer with a phase composition different from
that of the original material (Fig. 1). The thickness of
the polycrystalline zone formed on the sapphire substrate varies in the range from 1 to 80 µm. Figure 2
shows typical X-ray diffraction patterns of the transformed surface polycrystalline layers, which were
formed as a result of the reduction of corundum at temperatures ranging from 1700 to 1900°C. The X-ray dif-
INTRODUCTION
1112
TRANSFORMATION OF THE CORUNDUM STRUCTURE
fraction lines observed for one of the phases (Fig. 2a)
correspond to the cubic structure (space group Fd3m).
The unit cell parameter of this phase varies from sample to sample in the range 7.933–7.948 Å, which is
somewhat different from the data reported by
Filonenko et al. [1] and close to the value obtained by
Vert et al. [6].
All the diffraction lines observed for the other
revealed phase (the H phase) (Fig. 2b) correspond to the
hexagonal structure, which can be assigned to either the
space group P63/mmc or a space group with similar
symmetry elements. As in the case of the spinel phase,
the estimated lattice parameters of this structure differ
for different samples and have the following values: a ~
3.107–3.115 Å and c ~ 4.984–4.985 Å. In the surface
layer, the phase composition varies from a mixture of
phases (Fig. 2b) with a high content of the H phase
(>90%) to an almost complete (within the limits of the
sensitivity of the X-ray powder diffraction analysis)
transformation of the material into the spinel phase
(Fig. 2a). Annealing of thin layers (with a thickness of
~0.5 mm) composed of α-Al2O3 powders leads to similar phase transformations.
Both phases are stable under normal conditions. The
observation of the synthesized compounds during a
period of several years did not reveal any variation. The
X-ray diffraction pattern of the transformed surface
layer remains unchanged after annealing at a temperature of ~700–800°C for 10 h in air. Insignificant
changes in the X-ray diffraction pattern of the transformed surface layer are observed after annealing under
similar conditions at a temperature of 1000°C, whereas
the annealing at a temperature of ~1300°C leads to a
complete transformation of both phases into corundum.
This transformation is accompanied by an increase in
the weight of the sample. Under the assumption that the
revealed increase in the weight of the sample occurs as
a result of oxygen absorption, we can evaluate the composition of new phases. It has been established that the
phase transition of the material of the transformed layer
on the surface of the sapphire substrate to corundum is
attended by the absorption of ~0.15–0.25 oxygen atoms
per molecule of the corundum thus formed. According
to these estimates, the compound formed on the sapphire surface has the composition Al2.8–2.9O4. The transformation into corundum upon oxidation of the powders is
accompanied by a more considerable absorption of oxygen. In this case, the material absorbs ~0.36–0.58 oxygen
atoms per molecule of the newly formed corundum.
From the aforesaid, it follows that, prior to oxidation,
the initial compound had the composition Al3O3.95–3.63.
The observed difference, most likely, can be explained
by the fact that the transformed layer on the sapphire
surface consists of phases with a low oxygen content
and partially oxidizes at the expense of oxygen diffusing into this layer from sapphire.
Further investigations included the construction of
the models describing the crystal structures of the
CRYSTALLOGRAPHY REPORTS
Vol. 53
No. 7
2008
1
1113
2
10 mm
Fig. 1. Illustration of the transformation of the sapphire surface due to the reducing annealing: (1) the initial polished
sapphire sample and (2) the sapphire sample with a transformed near-surface layer ~30 µm thick after annealing.
revealed phases, the calculations of the model X-ray
diffraction patterns, and comparison of the results
obtained from these calculations with the experimental
data. The construction and analysis of the models of the
crystal structures and the calculation of the model
X-ray diffraction patterns were carried out with the
Diamond 2.1e [7] and POWDER CELL (Version 2.3)
[8] software packages. The solution and simulation of
the crystal structures were performed using the X-ray
diffraction patterns of the reduced powders.
When constructing the model of the spinel structure,
we carried out a comparative analysis of the X-ray diffraction patterns calculated for the structure of the compound with a variable composition (space group Fd3m,
Fig. 3) for all diffraction lines within the 2θ range from
10° to 150°. The theoretical X-ray diffraction patterns
were calculated by varying the coordinate of the oxygen atom at the 32e symmetry position in the range
0.3600–0.3875 for different occupancies of the symmetry positions of aluminum (8a and 16d) and oxygen
(32e), which corresponded to the composition of the
compound in the Al2O3–Al3O4–Al3O4 – x region. The
temperature factor in the calculations was ignored (B = 0).
The results of the calculations were compared with the
experimental X-ray diffraction pattern of the spinel
phase. The X-ray diffraction characteristics of this
phase are listed in Table 2. The unit cell parameter of
the spinel structure was determined to be 7.9437 Å.
According to the amount of absorbed oxygen, the composition of the compound under investigation was estimated as Al3O3.76. The R factor served as a criterion for
reliability of the model. In the simulation of the structure, good agreement was achieved for the model with
the atomic coordinates presented in Table 1. The dependences of the R factor on the occupancy of the 32e symmetry position according to the results obtained from
the calculations of the X-ray diffraction patterns for the
1114
DAN’KO et al.
I, counts/s
4000
800
0
800
3200
Sp
(664)
(931)
(840)
(660)(822)
(751)(555)
(800)
(533)
(620)
(531)
400
0
H
Sp
(844)
H
1200
(b)
(422)
(333)(511)
(222)
(111)
(400)
(220)
2000
(444)
(711)(551)
(642)
(553)(731)
(440)
(311)
(844)
(a)
Sp
400
Sp
Sp
Sp
Sp
Sp
Sp
Sp
Sp Sp
Sp
Sp
H
Sp
H
Sp
H H
0
10
30
50
Sp
H
H Sp
H Sp
H
H
H H
Sp
Sp Sp Sp
H H Sp H SpH
H
70
Sp
90
110
130
H
150
2θ, deg
Fig. 2. X-ray diffraction patterns of the polycrystalline layer on the (0001) surface of the sapphire sample: (a) spinel and (b) spinel + H
phase with the 〈001〉 texture (Cu K α radiation).
1, 2
c
a
b
8a—Al
16d—Al
32e—O
Fig. 3. A model of the crystal structure of the spinel phase according to the calculations performed with the Diamond 2.1e program [7].
compound of variable composition Al3O4 – x in comparison with the experimental X-ray diffraction pattern are
shown in Fig. 4. As can be seen from this figure, the
best agreement is observed for the occupancies in the
range 0.88–0.96, which corresponds to the compound
of the composition Al3O3.52–3.84; i.e., the structure of the
compound contains anion vacancies. It should also be
noted that there is a good agreement with the experi-
mental results for the compound of the composition
Al3O3.76.
The results of the calculations of the theoretical
X-ray diffraction pattern for the occupancy equal to
0.94 (Al3O3.76) (I calcd) in comparison with the experimental data (I exp) and the quantity |∆I | = |I exp – I calcd | are
presented in Table 2. According to the calculation with
the inclusion of all 22 observed diffraction lines, the R
CRYSTALLOGRAPHY REPORTS
Vol. 53
No. 7
2008
TRANSFORMATION OF THE CORUNDUM STRUCTURE
factor is approximately equal to 0.11, whereas the calculation performed without regard to the diffraction
lines observed in the range 110°–150° (where the temperature factor is more pronounced) gives R ≈ 0.07.
Therefore, the proposed model provides an adequate
description of the compound obtained in the experiment.
The revealed H phase cannot be identified as one of
the known phases Al2O3 with a hexagonal structure (β,
χ, ε). A similar phase with a hexagonal lattice and the
unit cell parameters a = 3.11 Å and c = 4.97 Å in the
products of the reduction of corundum was observed
simultaneously with the spinel phase in the experiments
performed by Vert et al. [6]. Upon the condensation of
sublimates of reduced α-Al2O3, Beletskiœ and Rapoport
[9] also obtained a compound (supposedly of the composition Al2O) with a lattice assigned to one of the
4
4
space groups D3h, C 6v , or D 6h with the unit cell
parameters a = 3.1 ± 0.01 Å, c = 4.99 ± 0.01 Å, and
c/a = 1.61. A more detailed analysis of the crystal structure and the X-ray diffraction characteristics of the synthesized compound were not reported by those authors.
In our investigations, we failed to establish, within the
limits of experimental error, any significant dependence of the amount of absorbed oxygen on the ratio
between the content of the spinel phase and the content
of the H phase in the products of corundum reduction
R
0.14
Al3O3.76
Al3O4
0.13
1115
Table 1. Atomic coordinates in the model crystal structure
of the spinel phase (space group Fd3m)
Atom
Position
Al
Al
O
8a
16d
32e
Coordinates
x
y
z
0
0.625
0.38
0
0.625
0.38
0
0.625
0.38
Occupancy
1
1
0.88–0.96
which would indicate a noticeable difference in their
structural formulas. Therefore, we cannot argue that the
H phase has the composition Al2O; however, from a
comparison of all the data obtained in our studies with
those presented in [6, 9], it can be assumed that, quite
possibly, we are dealing here with the same structure.
This assumption is also supported by the fact that
needlelike formations similar to those observed in [9]
(Fig. 5) exist in the transformed surface layer containing the H phase with a clearly pronounced 〈001〉 texture
(Fig. 2b).
The averaged X-ray diffraction characteristics of the
H phase are presented in Table 3. To the best of our
knowledge, data for the Al–O compound with the
obtained set of interplanar distances are not available in
the reference literature. A similar phase Al2O3 with
close values of the unit cell parameters (a ~ 3.10 Å and
c ~ 4.99 Å) is found in the reference book of X-ray diffraction analysis [10]; however, more detailed data are
not reported.
In our calculations, the hypothetical structure of the
compound with a variable composition (Table 4) was
considered as an approximate model of the H phase.
The occupancies of the 2a symmetry position, which
corresponded to the compositions Al2O3 and Al2O2.667
1
0.12
0.11
0.10
0.09
0.08
2
0.07
0.84
0.88
0.92
40 µm
0.96
1.00
Occupancy
Fig. 4. Dependences of the R factor on the occupancy of the
32e symmetry position according to the (1) calculation
with the inclusion of all the observed diffraction lines
and (2) experimental data for diffraction lines in the
range 2θ < 110°.
CRYSTALLOGRAPHY REPORTS
Vol. 53
No. 7
Fig. 5. Micrograph (in reflected light) of the (0001) surface
of the sapphire sample after the reducing annealing at a high
temperature of ~1800°C for 5 h. The surface layer with a
thickness of ~10 µm contains the spinel phase (~30%) and
the H phase (~70%) with the 〈001〉 texture.
2008
1116
DAN’KO et al.
Table 2. X-ray diffraction characteristics of the spinel phase
No.
d, Å
hkl
I calcd
I exp
|∆I|
No.
d, Å
hkl
I calcd
I exp
|∆I|
1
2
3
4
5
*
6
7
4.5863
2.8085
2.3951
2.2931
1.9859
1.8224
1.6215
1.5288
0.15
3.07
0
2.7
4.35
*
0.86
3.53
15
*
*
16
0.993
0.9705
0.9633
0.9362
0.9173
13.7
2.17
77.5
*
2.2
3.8
12.5
*
6
*
2.49
*
2.2
0.55
1.15
*
1.16
*
*
18
19
0.9112
0.8881
0.8719
*
9.6
1.9
*
4.12
1.04
*
20
21
22
*
0.8667
0.8468
0.8327
0.8108
0.7984
11.76
0
0
1.24
2.27
12.71
3.16
0.7
13.72
0.71
0.15
0.01
2.3
19.25
64.83
0.52
0.49
0
0.66
*
*
0.99
17
800
733
644
660
822
751
555
662
840
911
753
842
664
931
844
933
755
771
11.1
*
*
4.5
1.4043
1.3427
1.324
1.256
1.2114
1.1976
1.1466
1.1123
*
1.5
15
51
*
*
0.8
4.25
13.83
*
13
14
1.0615
1.0342
11.55
27.93
100
2
54.35
0.27
10.86
6.29
32.94
75.01
1.59
0
4.35
11.35
0.72
7.16
0.09
0.95
5.6
8.9
11.41
11.7
31
100
4.7
50
*
10
35.7
8
*
9
10
11
*
12
*
111
220
311
222
400
331
422
333
511
440
531
442
620
533
622
444
711
551
642
553
731
6.3
16.5
0.7
3.81
* Peaks with a low intensity were disregarded in the calculations.
Table 3. X-ray diffraction characteristics of the H phase
No.
d, Å
hkl
I calcd
I exp
|∆I|
No.
d, Å
hkl
I calcd
I exp
|∆I|
1
2
3
4
5
6
7
8
9
10
11
12
13
2.695
2.494
2.371
1.830
1.556
1.415
1.348
1.320
1.301
1.247
1.186
1.132
1.047
100
002
101
102
110
103
200
112
201
004
202
104
203
100
49
47
13
60
5
12
33
3
8
5
13
1
100
64
44
32
27
11
18
40
20
20
10
26
18
0
15
3
19
33
6
6
7
17
12
5
13
17
14
15
16
*
*
17
18
*
19
20
21
*
1.019
0.998
0.973
0.943
0.936
0.915
0.898
0.884
0.869
0.845
0.831
0.802
210
211
114
212
105
204
300
301
213
302
006
205
10
2
24
7
1
9
12
0
2
15
3
2
12
15
9
*
*
2
2
*
2
3
9
*
2
13
15
*
*
7
10
*
0
12
6
*
* Peaks were disregarded in the calculations.
(Al3O4), were equal to 0.500 and 0.333, respectively.
The temperature factor in these calculations was
ignored (B = 0). Table 3 presents the results obtained
from the comparative analysis of the experimental (I exp)
and calculated (I calcd) X-ray diffraction patterns for this
structure with the unit cell parameters a ~ 3.112 Å and
c ~ 4.988 Å, when the occupancy of the 2a symmetry
position corresponds to the composition α-Al2O3, and
CRYSTALLOGRAPHY REPORTS
Vol. 53
No. 7
2008
TRANSFORMATION OF THE CORUNDUM STRUCTURE
1300
I, counts/s
1000
δ
Sp
500
δ Sp
δ
Sp
δ
H
Sp
X
0
10
H
Sp
δ
X
H
Sp
Mo
30
1117
Sp
δ
Sp
X
X
50
δ
H
X
70
2θ, deg
Fig. 6. X-ray diffraction pattern of the crystallized reduced melt. Designations: Sp is the spinel, H is the H phase, δ is the δ phase
[11], Mo is molybdenum (the container material), and X denotes additional unknown diffraction lines.
the quantities |∆I| = |I exp – I calcd |. All the principal diffraction lines of the model and experimental structures
coincide; however, satisfactory quantitative agreement
between the calculated and experimental data is not
observed (the R factor exceeds 0.3). Although the
approximate model under consideration does not provide an adequate description of the obtained structure,
it can be used for a qualitative description of the
observed corundum–spinel phase transformations [4].
The number of oxygen-deficient structures formed
upon reduction of corundum is not exhausted by the
two aforementioned structures. After the reduction at
temperatures above 1950°C, including the reduction of
the melt, the situation is significantly complicated and
the products of reduction are multiphase systems of
more complex composition. In this case, the X-ray diffraction patters contain not only the lines associated
with the corresponding spinel phases and the H phase
but also a number of other additional lines. Some of
these additional lines can be assigned to the compounds
known from the reference literature, whereas the other
lines cannot yet be uniquely identified. As an example,
Fig. 6 shows the X-ray diffraction pattern of the system
formed as a result of the reduction of the α-Al2O3 melt
in the medium containing 10% hydrogen. It can be seen
from this figure that, upon reduction, corundum completely transforms into other phases. Apart from the
lines that can be assigned to the aforementioned phases,
Table 4. Atomic coordinates in the model crystal structure
of the H phase (space group P63/mmc)
Atom
Position
Al
O
O
2b
2c
2a
Coordinates
x
y
z
0
0.3333
0
0
0.6667
0
0.25
0.25
0
CRYSTALLOGRAPHY REPORTS
Vol. 53
Occupancy
1
1
0.5–0.333
No. 7
2008
as well as to the container material (molybdenum), the
X-ray diffraction pattern exhibits lines that can be
attributed to the δ phase [11] and also additional
unknown diffraction lines (denoted as X in the X-ray
diffraction pattern).
CONCLUSIONS
Thus, the results obtained in this study have demonstrated that the high-temperature reduction of α-Al2O3
is accompanied by the transformation of corundum into
new phases with a lower oxygen content. It is obvious
that these phases, in particular, the spinel phase (Fig. 3,
Table 1), with a structure in which aluminum ions are
distributed over both octahedral and tetrahedral positions, can exist only in the presence of stabilizers. A
vacancy formed in the structure of aluminum oxide
owing to the reduction “localizes” two excess electrons.
Therefore, the vacancies can decrease the valence of
the surrounding cations and play the role of stabilizers
of the structures of these new phases, much as water
and different impurity ions (for example, Li+, Na+, and
K+) are stabilizers of the structures of the metastable
modifications of aluminum oxide Al2O3 [12].
REFERENCES
1. N. E. Filonenko, I. V. Lavrov, O. V. Andreeva, and
R. L. Pevzner, Dokl. Akad. Nauk SSSR 115 (3), 583
(1957).
2. N. S. Sidelnikova, M. A. Rom, A. Ya. Danko, et al.,
Funct. Mater. 11 (1), 26 (2004).
3. A. Ya. Dan’ko, M. A. Rom, N. S. Sidel’nikova, et al.,
Poverkhnost, No. 11, 89 (2005).
4. A. Ya. Dan’ko, M. A. Rom, N. S. Sidelnikova, et al.,
Funct. Mater. 12 (4), 725 (2005).
5. A. Ya. Dan’ko, N. S. Sidel’nikova, G. T. Adonkin, et al.,
Funct. Mater. 4 (1), 92 (1997).
1118
DAN’KO et al.
6. Zh. L. Vert, M. V. Kamantsev, V. I. Kudryavtsev, and
M. I. Sokhor, Dokl. Akad. Nauk SSSR 116 (5), 834
(1957).
7. G. Bergerhoff, M. Berndt, and K. Brandenburg, J. Res.
Natl. Inst. Stand. Technol. 101, 221 (1996).
8. W. Kraus and G. Nolze, J. Appl. Crystallogr. 29, 301
(1996).
9. M. S. Beletskiœ and M. B. Rapoport, Dokl. Akad. Nauk
SSSR 80 (5), 751 (1951).
10. L. I. Mirkin, Reference Book on X-Ray Diffraction Analysis of Polycrystals (Fizmatgiz, Moscow, 1961), p. 626
[in Russian].
11. ASTM 16-0394 (American Society for Testing Materials,
West Conshohocken, PA, United States).
12. A. A. Khanamirova, Alumina and Methods for Decreasing the Content of Impurities in It (ArmSSR Academy of
Sciences, Yerevan, Soviet Union, 1983), pp. 40–62 [in
Russian].
Translated by O. Borovik-Romanova
CRYSTALLOGRAPHY REPORTS
Vol. 53
No. 7
2008