Crystal chemical study of the high-temperature phase R(O,F)1.93±δ with R = Tm, Yb and Lu. Professor Dr. Georg Brauer on the occasion of his 90th birthday

Article abstract of DOI:10.1016/S0925-8388(98)00586-6

The crystal chemistry of the non-stoichiometric compounds R(O,F)_x with R = Tm, Yb, and Lu with x≈1.93±0.02, which are formed by peritectoid reaction of R₂O₃ with RF₃ at ca. 900 K, ca. 950 K and 1076 K, respectively, has been studied by X-ray powder diffraction (Guinier method). Samples of the common composition R(O,F)_1.93, which were obtained by quenching from 1375±15 K to room temperature, displayed diffraction patterns of a rather complex superstructure based on the fluorite-related parent structure. The corresponding body-centered basis structure with Z = 2 was found to be orthorhombic with the following lattice parameters: R a(angstroms) b(angstroms) c(angstroms), Tm 3.7508(2) 5.3772(3) 3.8210(2), Yb 3.7341 (2) 5.3609(4) 3.8129(3), Lu 3.7008(2) 5.3371 (4)3.8170(2). This type of compound could not be found in the system Er-O-F. It is therefore concluded that its existence is restricted to the systems with R = Tm, Yb, Lu and Sc, which are characterized by the occurrence of the monoclinic, baddeleyite-type modification of stoichiometric ROF. It could be shown that for the Tm-O-F system the parent structure formula volumes of the orthorhombic vernier phase Tm(O,F)_2.11, cubic α-TmOF and orthorhombic Tm(O,F)_1.93 depend approximately linearly on composition. The question, whether monoclinic baddeleyite-related TmOF, which has been described in the literature, exists as a thermodynamically stable compound, is tentatively discussed.

Full text of DOI:10.1016/S0925-8388(98)00586-6

Journal of Alloys and Compounds 278 (1998) 165–168
L
Crystal chemical study of the high-temperature phase R(O,F)1.936d with
R5Tm, Yb and Lu
Professor Dr. Georg Brauer on the occasion of his 90th birthday
a
a
a
b,
*
¨
J.-H. Muller , T. Petzel , B. Hormann , O. Greis
aUniversity of the Federal Armed Forces, Division of Inorganic Chemistry, Holstenhofweg 85, D-22043 Hamburg, Germany
bTechnical University Hamburg-Harburg, Central Division of Electron Microscopy, Eißendorfer Straße 42, D-21071 Hamburg, Germany
Received 16 March 1998; received in revised form 9 April 1998
Abstract
The crystal chemistry of the non-stoichiometric compounds R(O,F)x with R5Tm, Yb, and Lu with x¯1.9360.02, which are formed by
peritectoid reaction of R2O3 with RF3 at ca. 900 K, ca. 950 K and 1076 K, respectively, has been studied by X-ray powder diffraction
(Guinier method). Samples of the common composition R(O,F)1.93, which were obtained by quenching from 1375615 K to room
temperature, displayed diffraction patterns of a rather complex superstructure based on the fluorite-related parent structure. The
corresponding body-centered basis structure with Z52 was found to be orthorhombic with the following lattice parameters:
˚
˚
˚
R
a (A)
b (A)
c (A)
Tm
Yb
Lu
3.7508(2)
3.7341(2)
3.7008(2)
5.3772(3)
5.3609(4)
5.3371(4)
3.8210(2)
3.8129(3)
3.8170(2)
This type of compound could not be found in the system Er–O–F. It is therefore concluded that its existence is restricted to the systems
with R5Tm, Yb, Lu and Sc, which are characterised by the occurrence of the monoclinic, baddeleyite-type modification of stoichiometric
ROF. It could be shown that for the Tm–O–F system the parent structure formula volumes of the orthorhombic vernier phase
Tm(O,F)2.11, cubic a-TmOF and orthorhombic Tm(O,F)1.93 depend approximately linearly on composition. The question, whether
monoclinic baddeleyite-related TmOF, which has been described in the literature, exists as a thermodynamically stable compound, is
tentatively discussed.
1998 Elsevier Science S.A. All rights reserved.
Keywords: Rare earth oxide fluorides; High-temperature phase; Crystal structure; X-ray powder diffraction
1. Introduction
shown that the respective stoichiometric oxide fluorides
ROF decompose into R(O,F)x and phases of fluoride-rich
compositions which in case of R5Sc is the stoichiometric
trifluoride and in the other cases are orthorhombic vernier-
type phases with x¯2.10. With the exception of
Sc(O,F)1.88, the high-temperature phases could be pre-
served by rapid quenching to room temperature. Inves-
tigation by X-ray powder diffraction so far only allowed
the tentative conclusion that these phases have anion
deficient fluorite-related superstructures. In case of
Yb(O,F)1.91 it has been assumed by Bevan [3] that this
compound is monoclinic, while Taoudi et al. [4] claim that
this compound, as well as Tm(O,F)x and Lu(O,F)x, are
Thermoanalytical and X-ray powder diffraction studies
of the binary systems R2O3 –RF3 with R5Sc, Tm, Yb and
Lu have revealed the existence of slightly non-stoichio-
metric compounds of composition R(O,F)x with average
values of x ranging from 1.88 (R5Sc) to approximately
1.93 (R5Tm, Yb, Lu) [1–5]. These compounds are formed
by a peritectoid reaction of R2O3 with ROF at tempera-
tures ranging from 141963 K (R5Sc) to 107664 K and
ca. 900 K for R5Lu and Tm, respectively. It has also been
*
Corresponding author.
0925-8388/98/$19.00
1998 Elsevier Science S.A. All rights reserved.
PII: S0925-8388(98)00586-6

¨
J.-H. Muller et al. / Journal of Alloys and Compounds 278 (1998) 165–168
166
characterised by a rhombohedral subcell with a very
complex superstructure. In this paper we present the results
of an investigation into the crystal data and the homo-
geneity regions of these phases by X-ray powder diffrac-
tion (XRPD).
liminary splitting analysis of a systematic procedure for
indexing of super-structure reflections [10] and the so
called P.M. de Wolff method [11]. The lattice parameters
of all these phases have been refined using also the
computer program LSUCRE [7].
3. Results and discussion
2. Experimental
The minimum temperatures at which R(O,F)x is formed
by the eutectoid reaction of R2O3 and RF3 with R5Tm,
Yb, Lu and Sc, and the homogeneity regions (related in all
cases to an equal quenching temperature of 1375615 K)
are summarized in Table 1. The limits of the homogeneity
regions can only be considered as approximate results
because of the uncertainty of the temperature inherent in
the quenching procedure. They represent the compositions
at which the X-ray reflections of R2O3 at the oxide-rich
side and those of the orthorhombic vernier phases at the
fluoride-rich side were definitely absent from the Guinier
patterns. Since quenching proved to be unsuccessful in
case of Sc(O,F)x, no results concerning the non-stoichiom-
etry of this compound can be given. While the quantitative
results of the DTA and DSC measurements allowed the
determination of the composition at x51.9160.01 for
R5Lu and at 1.9260.01 for Yb and Tm, this value is
1.8860.01 for R5Sc [1]. Attempts to prepare Er(O,F)x
proved to be unsuccessful, and it is hence concluded that
the existence of this type of high-temperature compound is
restricted to the rare earth elements which are character-
ized by the occurrence of the monoclinic, baddeleyite-type
modification of stoichiometric ROF. It is worth noting that
the widths of the homogeneity regions as derived approxi-
mately from the X-ray results are practically independent
of R, while the temperatures of formation increase sig-
2.1. Preparation of compounds
Samples of analytical compositions R(O,F)x with 1.85#
x#1.95 were prepared by solid-state reaction of mixtures
of the sesquioxides (stated purity 99.99%; Auer-Remy)
and the respective anhydrous trifluorides in gas-tight
platinum tubes at 1400 K for 2 days. The reaction products
were homogenised mechanically, annealed at 1425 K for 7
days and subsequently quenched to room temperature by
dropping into ice water. Prior to use the sesquioxides were
annealed at 1300 K for several hours in air in order to
destroy the hydroxides and carbonates, which usually form
during storage under ambient conditions. The anhydrous
trifluorides were prepared by dissolution of the sesquiox-
ides in hydrochloric acid, precipitation of the hydrated
fluorides with hydrofluoric acid, drying at 390 K in air and
subsequent dehydration under a gas stream of dry HF–N2
at 1000 K. The trifluorides so obtained were further
purified by distillation in a high vacuum inside a
molybdenum tube as described elsewhere in detail [6].
Quenching of samples was achieved by dropping the
platinum ampoules into ice water. The weight of the
samples did in no case change by more than 60.2 mg out
of a total weight of ca. 150 mg during the solid-state
reaction as well as during annealing and quenching, and
we therefore assumed that these preparative procedures did
not noticeably influence the original compositions.
nificantly with decreasing ionic radius of R31
.
The powder patterns (see Table 2) proved to be quite
similar. In all cases, the strong reflections were fluorite-
related, but split in such a way that the most common
symmetry reductions, tetragonal and rhombohedral, could
be excluded immediately. Therefore, a systematic analysis
of the splitting [10] had to be applied to determine the
lower symmetries. This procedure is the preliminary step
of most indexing methods for complicated superstructure
patterns derived only from X-ray powder diffraction. Our
2.2. Characterisation of compounds by X-ray powder
diffraction (XRPD)
The starting materials and the reaction products were
characterized by their X-ray powder diffraction patterns
(Guinier method, Nonius chamber, CuKa1 with l51.5406
˚
˚
A, silicon NBS 640 A with a55.4308 A as internal
calibration standard). The intensities of the X-ray reflec-
tions were determined using a Huber Guinier chamber type
632, connected to a modified Huber diffractometer system
type 644. The diffraction patterns of the starting materials
did in no case contain reflections which could not be
unambiguously attributed to the compounds in question.
Their lattice parameters, which were evaluated using the
computer program LSUCRE [7], were all found to be in
very good agreement with well established data from the
literature [8,9]. Indexing of the patterns of the quenched
high-temperature phases was performed using the pre-
Table 1
Minimum temperatures of formation and approximate widths of homo-
geneity regions of R(O,F)x
R in R(O,F)x
Tmin (K)
Approx. homogeneity
range at 1375 K
Tm
Yb
Lu
Sc
ca. 890
ca. 930a
1076
1.93–1.95
1.90–1.95
1.93–1.95
1.88–?
1420
a Not in phase equilibrium.

¨
J.-H. Muller et al. / Journal of Alloys and Compounds 278 (1998) 165–168
167
Table 2
X-ray powder diffraction data of R(O,F)1.93 with R5Tm, Yb, Lu
Tm(O,F)1.93
Yb(O,F)1.93
Lu(O,F)1.93
h
k
l
dacalc
d(obs)
I/Ib0
I/Ic0
h
k
l
dacalc
d(obs)
I/Ib0
I/Ic0
h
k
l
dacalc
d(obs)
I/Ib0
I/Ic0
5.2871
4.9368
3.1038
3.0412
2.7245
2.6671
2.6568
2.3920
2.2411
2.1348
2.1194
2.0623
1.9790
1.9081
1.8827
1.8503
1.7625
1.7330
1.6365
1.6165
1.6123
1.6035
1.5894
7
20
73
1000
807
17
149
414
47
12
83
21
27
4.9172
3.1150
3.0774
53
1000
987
71
1000
971
4.9827
3.1076
3.0649
54
1000
972
76
1000
964
52
1000
937
14
175
506
48
6
68
11
19
0
1
1
1
1
0
3.1147
3.0763
0
1
1
1
1
0
3.1072
3.0641
0
1
1
1
1
0
3.1047
3.0412
0
1
2
0
0
1
2.6886
2.6767
2.6892
2.6781
2.4099
187
507
45
171
403
49
0
1
2
0
0
1
2.6805
2.6679
2.6803
2.6683
2.3941
168
481
41
164
406
48
0
1
2
0
0
1
2.6685
2.6570
2.1553
2.1301
2.0694
1.9879
1.9106
1.8971
1.8758
1.7754
1.7549
66
15
16
82
23
24
2.1506
2.1351
2.0525
1.9915
1.9070
1.8915
1.8678
1.7721
1.7489
72
12
15
90
22
30
9
13
9
15
8
11
0
1
2
0
2
0
2
1
0
1.9105
1.8969
1.8754
141
564
137
31
160
569
159
37
0
1
2
0
2
0
2
1
0
1.9065
1.8909
1.8671
157
597
147
35
172
593
166
40
0
1
2
0
2
0
2
1
0
1.9085
1.8829
1.8504
127
528
141
32
7
7
175
91
67
174
484
141
39
11
17
168
149
72
6
9
6
9
1
0
1
2
1
3
3
1
2
1
0
1
1.6230
1.6227
1.6172
1.6066
1.6228
288
306
1
0
1
2
1
3
3
1
2
1
0
1
1.6187
1.6181
1.6119
1.6004
1.6184
295
334
1
0
1
2
1
3
3
1
2
1
0
1
1.6166
1.6125
1.6034
1.5895
1.6174
1.6067
1.5644
1.5572
1.5498
1.5381
1.4802
1.4191
1.4179
1.3999
1.3898
1.3444
1.3384
1.3049
1.2891
1.2592
1.2501
82
216
24
47
45
48
13
11
12
7
155
238
35
58
61
59
17
21
17
12
10
21
90
22
26
25
18
1.6120
1.6003
79
221
121
257
217
217
0
2
2
2
2
0
1.5573
1.5382
0
2
2
2
2
0
1.5536
1.5320
1.5538
1.5427
1.5319
1.4718
1.4167
1.3925
1.3802
1.3617
1.3404
1.3339
1.3260
1.3000
1.2852
1.2560
1.2431
1.2366
1.2309
1.2227
73
48
48
14
11
11
5
100
66
61
22
20
18
8
0
2
2
2
2
0
1.5524
1.5206
1.5525
1.5369
1.5206
1.4634
1.4087
1.3919
1.3768
1.3522
1.3343
1.3284
1.2954
1.2760
1.2492
57
53
49
15
10
13
7
103
61
56
28
16
16
11
7
10
87
23
25
35
5
4
8
4
0
2
4
0
0
2
1.3443
1.3383
17
76
18
14
15
5
0
2
4
0
0
2
1.3402
1.3339
15
74
3
21
15
18
5
32
41
48
22
92
9
28
26
30
10
54
63
93
0
2
4
0
0
2
1.3343
1.3285
10
67
23
13
16
0
1
2
1
3
3
3
2
1
1.2394
1.2344
1.2271
1.2395
1.2343
1.2270
1.2229
1.2177
1.2060
1.2012
1.1981
1.1882
27
41
40
8
30
22
16
36
23
42
69
64
19
56
25
54
49
36
0
1
2
1
3
3
3
2
1
1.2367
1.2309
1.2228
0
1
2
1
3
1
3
3
0
1
3
2
1
3
0
1.2377
1.2276
1.2157
1.2032
1.2019
1.2375
1.2277
1.2157
22
32
43
47
53
73
3
1
1
2
3
1
3
4
2
0
0
0
1
2
1
1.2178
1.2060
1.2013
1.1981
1.1883
3
1
1
2
3
1
0
4
2
0
0
3
1
2
1
1.2125
1.2032
1.1976
1.1942
1.1833
1.2125
1.2031
1.1974
1.1943
1.1831
30
26
14
41
23
55
42
22
61
36
1.2021
41
77
1
2
3
4
2
0
1
2
1
1.1924
1.1893
1.1738
1.1893
1.1737
37
23
80
35
a53.7508(2)
b55.3772(3)
c53.8210(2)
V577.064(5)
Z52
a53.7341(2)
b55.3609(4)
c53.8129
V576.329(7)
Z52
a53.7008(2)
b55.3371(4)
c53.8170(2)
V575.392(6)
Z52
a Basis structure.
b Peak height.
c Peak area.
analysis indicated a symmetry reduction from cubic face-
centred (Z54) to orthorhombic body-centred (Z52). With
this result, the full reciprocal lattice of the basis structure
could be established using the so called P.M. de Wolff
vector method [11]. The relation between parent and basis
structure is: ab¯ap0.5œ2, bb¯ap, cb¯ap0.5œ2, Vb5
0.5Vp. As far as the basis structure is concerned, all three
phases are unambiguously isostructural. There are, how-

¨
J.-H. Muller et al. / Journal of Alloys and Compounds 278 (1998) 165–168
168
ever, numerous superstructure reflections with rather weak
intensities. A comparison of the observed superstructure
reflections listed in Table 2 also indicates, that all three
phases most likely have the same superstructure geometry.
Unfortunately, we were unsuccessful in indexing the
superstructure(s) unambiguously. Obviously, the number of
superstructure reflections is too small to construct a
densely populated and reliable reciprocal lattice. The
weakness of the superstructure reflections can easily be
explained by the fact that, in the case of the title phases,
the superstructure is caused predominantly by ordered
vacancies in the anion sublattice and not by ordered
defects in the cationic sublattice, as compared with
(M,R)F21d phases with R6F38 clusters [12]. To overcome
these difficulties with too few and too weak superstructure
reflections, a further study is planned for the near future
employing high-energy electron diffraction (HEED) from
single crystals of the title phases. For the time being, all
available X-ray diffraction data of the R(O,F)1.93 phases
are summarised in Table 2.
approximately linearly dependent on x. However, the
formula volume of monoclinic baddeleyite-type TmOF
cannot be reasonably correlated with the volumes of the
fluorite-related basis structures. This observation leads one
to ask, whether or not monoclinic TmOF actually exists
under equilibrium conditions. Taoudi et al. [4] were able to
prepare this modification only by quenching of stoichio-
metric TmOF from 4508C to room temperature. We
observed that slowly cooled samples of the composition
Tm(O,F)x with 1.98#x#2.02 always yielded the rhom-
bohedral b-modification accompanied by C-Tm2O3 for
x,2.0 and by the most oxide-rich, limiting vernier-type
phase for x.2.0. However, the X-ray diffraction patterns
of quenched stoichiometric TmOF and of samples with
x$2.0 generally but weakly displayed the three strongest
reflections of the monoclinic modification beside the
dominating rhombohedral pattern. A high-temperature
Guinier diffraction study of stoichiometric TmOF under
near- equilibrium conditions showed that the b→a phase
transition was simultaneously accompanied by the forma-
tion of a minor amount of the monoclinic modification
within the temperature range .750–880 K. While these
observations could serve to explain the poor reproducibil-
ity of the DSC results relating to the b→a transition of
Within the group of R–O–F systems studied in this
work, Tm–O–F is exceptional insofar as three fluo-
rite-related phases, namely orthorhombic (vernier-
type)
Tm(O,F)2.11
(5Tm8O9F11),
orthorhombic
¨
Tm(O,F)1.9360.02 and rhombohedral b-TmOF exist as well
as cubic a-TmOF. In Fig. 1 the formula volumes of the
fluorite-related parent structures and the formula volume of
monoclinic TmOF, as described by [4], are plotted as
functions of composition. Considering the observation that
the rhombohedral (b)→cubic(a) phase transition of stoi-
TmOF as observed by Muller et al. [5], they are not
accurate enough to allow any consistent conclusion con-
cerning the conditions of state under which monoclinic
TmOF may exist as a thermodynamically stable modi-
fication.
chiometric TmOF is accompanied by a volume increase of
3
˚
0.560.05 A [2], the fluorite-related formula volumes are
References
[1] T. Petzel, F. Schneider, B. Hormann, Thermochim. Acta 276 (1996)
1.
¨
[2] T. Petzel, J.-H. Muller, B. Hormann, J. Alloys Comp. 266 (1998)
134.
[3] D.J.M. Bevan, pers. comm., 1986.
[4] A. Taoudi, J.P. Laval, B. Frit, Mater. Res. Bull. 29 (1994) 1137.
¨
[5] J.-H. Muller, T. Petzel, B. Hormann, Thermochim. Acta 298 (1997)
109.
[6] T. Petzel, V. Marx, J. Potthast, Th. Ahnen, Thermochim. Acta 194
(1992) 319.
[7] H.T. Evans, D.E. Appleman, D.S. Handworker, Computer Program
LSUCRE, Ann. Meeting Program 42, Am. Crystallogr. Assoc.,
Cambridge, MA, 1963.
[8] O. Greis, T. Petzel, Z. Anorg. Allg. Chem. 403 (1974) 1.
[9] R.S. Roth, S.J. Schneider, J. Res. Natl. Bur. Std. 64A (1960) 309.
[10] O. Greis, Monatsh. Chem. 108 (1977) 205.
[11] P.M. de Wolff, Acta Crystallogr. 10 (1957) 590.
[12] O. Greis, J.M. Haschke, Rare Earth Fluorides. In: K.A. Gschneidner,
Jr., L. Eyring (Eds.), Handbook on the Physics and Chemistry of
Rare Earths, Chap. 45, North Holland Publ., 1982, pp. 387–460.
Fig. 1. Plot of formula volumes vs. composition for compounds in the
Tm-O-F system.