Phase transitions and thermal properties of gadolinium trifluoride

Article abstract of DOI:10.1016/S0925-8388(99)00231-5

The temperature dependence of gadolinium trifluoride density in the solid and liquid states was determined by a gamma attenuation technique. Volume changes on solid-solid and liquid-liquid transitions were measured. The density and volume coefficient of thermal expansion of the liquid salt at the melting point are equal to 5609±25 kg m^-3 and (16.2±0.5)×10^-5 K^-1, respectively. The relative density reduction on melting equals 16.7±0.15%. The polymorphic transformation from the low-temperature orthorhombic modification (a lattice of β-YF₃ type) into the high-temperature hexagonal phase (LaF₃ type) occurs with an increase of the density (about 2.1%). The peculiarities of the thermal expansion of the solid rare earth fluorides, undergoing phase transition β-YF₃-type→LaF₃-type, are discussed.

Full text of DOI:10.1016/S0925-8388(99)00231-5

Journal of Alloys and Compounds 290 (1999) 30–33
L
Phase transitions and thermal properties of gadolinium trifluoride
*
S.V. Stankus , R.A. Khairulin, K.M. Lyapunov
Institute of Thermophysics, Siberian Branch of the Russian Academy of Sciences, Kutateladze, 2, Novosibirsk 630090, Russia
Received 30 March 1999; received in revised form 11 May 1999
Abstract
The temperature dependence of gadolinium trifluoride density in the solid and liquid states was determined by a gamma attenuation
technique. Volume changes on solid–solid and solid–liquid transitions were measured. The density and volume coefficient of thermal
expansion of the liquid salt at the melting point are equal to 5609625 kg m23 and (16.260.5)31025
K
21, respectively. The relative
density reduction on melting equals 16.760.15%. The polymorphic transformation from the low-temperature orthorhombic modification
(a lattice of b-YF3 type) into the high-temperature hexagonal phase (LaF3 type) occurs with an increase of the density (about 2.1%).
The peculiarities of the thermal expansion of the solid rare earth fluorides, undergoing phase transition b-YF3-type→LaF3-type, are
discussed.
1999 Elsevier Science S.A. All rights reserved.
Keywords: Rare earth fluoride; Density; Phase transition; Solid; Melt; Gamma method
1. Introduction
conducted up to now. The exception was the study of
Thoma and Brunton [1] where the lattice parameters of the
hexagonal phase were determined at about 1185 K. The
density of the liquid salt has likely been measured only by
Ikemiya et al. [4] by the maximum bubble pressure method
over the temperature range 1620 to 1732 K (the measure-
ment error of the density was 61.9%).
The aim of this work is the study of the thermal
properties of gadolinium trifluoride over a wide tempera-
ture range (293–1700 K) of both solid and liquid states
including the regions of phase transitions.
The gadolinium trifluoride (as well as SmF3 and EuF3)
on heating, apart from melting undergoes a solid–solid
transformation from the orthorhombic form (a lattice of
b-YF3 type) into a hexagonal one (LaF3 type) [1,2].
However, the parameters of these transitions and the
thermal properties of solid and liquid phases have not been
adequately studied. In particular, according to the data of
Thoma and Brunton [1] the temperature of the poly-
morphic transformation Tt equals 1173 K, whereas Sped-
ding et al. [2] give the value of Tt51348 K. Sobolev et al.
[3] determined the lattice parameters for the hexagonal
modification of GdF3 at 293 K (these values were mea-
sured on samples containing additions, which stabilize the
high-temperature form, with subsequent extrapolation to
zero concentration of impurities). It was found that at room
temperature the molar volume of the metastable phase of
LaF3 type was 3.9% less than that of stable modification
(b-YF3 type). From this fact the authors [3] reached the
conclusion that on heating, the polymorphic transformation
had to occur with a noticeable increase of the density.
However, the high-temperature measurements of the vol-
ume properties of solid GdF3 appear not to have been
2. Experimental details
Gadolinium trifluoride was synthesized from metallic
gadolinium (purity 99.85%, the main impurities were rare
earth metals) by first dissolving it in hydrochloric acid and
then precipitation from a solution with hydrofluoric acid.
The salt was refined by drying and remelting in a vacuum
furnace and then by vacuum distilling (the last manipula-
tion was conducted to clean the sample from the rare earth
oxides and oxyfluorides). The technique and the equipment
used for the purification of the rare earth fluorides have
been described more fully in our previous articles [5,6].
The investigation of thermal properties of GdF3 was
performed by irradiation of the samples with a narrow
beam of gamma quanta from a 137Cs source. Details of the
*Corresponding author. Tel.: 17-3832-391-541; fax: 17-3832-343-
480.
E-mail address: stankus@itp.nsc.ru (S.V. Stankus)
0925-8388/99/$ – see front matter
PII: S0925-8388(99)00231-5
1999 Elsevier Science S.A. All rights reserved.

S.V. Stankus et al. / Journal of Alloys and Compounds 290 (1999) 30–33
31
experimental equipment and procedures have been de-
scribed elsewhere [7,8]. The temperature was measured
with tungsten–rhenium thermocouples embedded in the
sample in protective molybdenum sleeves. The thermocou-
ple calibration was checked against the melting points of
pure Cu and Ni both before and after the experiments. The
deviations of measured melting temperatures from refer-
ence data did not exceed 3 K.
liquid GdF3 is free of anomalies and linearly decreases
with the temperature (K):
rm(T) 5 5609 2 0.9097(T 2 1509).
(1)
The melting and crystallization of the sample occurred
over the narrow temperature interval (DTLS ,6 K). The
temperature of the beginning of melt solidification Tf 5
150963 K was close to the literature data on the melting
point of gadolinium fluoride (1504–1505 K) [1,2].
As noted above, previously the density of liquid GdF3
was measured by Ikemiya et al. [4]. Unfortunately the
measurement data in this work are presented in graphical
form, and a numerical result is given only at 1500 K
(r557306110 kg m23). These authors [4] obtained this
value by the extrapolation of the temperature dependence
of the melt density measured over the range 1620 to 1732
K. The extrapolation of the relation (1) to 1500 K gives a
value of the density equal to 5618630 kg m23, that is the
difference of the results is not in excess of the total
measurement error.
The experiments were conducted in an atmosphere of
pure argon. The measurements of the density (r) of the
melt and high-temperature solid phase have been per-
formed with the sample placed in a molybdenum crucible.
However, once the remelted sample was cooled somewhat
below (|100 K) the point of solid–solid transition, its
progressive cracking started. This caused a distortion of
the measured temperature dependence of the density for
the low-temperature phase. To determine r(T) for the
orthorhombic form we performed further investigations
using a sample that was sintered at temperatures below the
point of solid–solid transition. Because the sintered sample
was porous, these experiments allowed us to measure the
relative density r(T)/r(293 K). To calculate r(T) we
used, as reference value r(293 K), the data on the density
from Ref. [3], determined from the lattice parameters of
the orthorhombic phase at room temperature. The differ-
ence of the results on the density of the low-temperature
modification near the point of polymorphic transformation
obtained with the sintered sample and with the remelted
one did not exceed 0.4%. This difference was less than the
measurement error evaluated by us (|0.5%).
3.2. Solid state
As observed in our experiments, the solid–solid trans-
formation proceeded over a reasonably wide temperature
range of about 30 K. However, the temperatures of the
beginning of the transformation both on heating and on
cooling of the samples were closely related (1310610 K).
This temperature was taken by us as the point of poly-
morphic transition Tt. The value of Tt is somewhat lower
than the data of Spedding et al. [2] and 130 K higher than
the value of Tt reported by Thoma and Brunton [1]. It is
possible that the very low transition temperature deter-
mined in Ref. [1] is connected with inadequate purity of
the samples investigated in this work. As noted in Ref. [9],
a minor amount of rare earth oxide impurities may lead to
effects of this kind.
The value of volume change on polymorphic trans-
formation was measured during cooling of the remelted
sample (on heating, the sample cracked in the region of
transition, causing significant measurement errors). The
relative density change on solid–solid transition drt 5
(r2 2 r1)/r2 (r2, r1 are the densities of the low- and
high-temperature phases near Tt) is found to be
22.160.3%, that is the density of the high-temperature
phase is larger than that of the low-temperature form.
The best fit of experimental data provided the following
equations for the temperature dependences of the density
of the low- and high-temperature solid phases:
3. Results and discussion
3.1. Liquid state and region of melting-crystallization
Some original data illustrating the reproducibility of the
measuring values are shown in Table 1. The measurements
of melt density were performed over a wide temperature
range (the overheating above the melting point was about
200 K) both on heating and on cooling of the sample. The
experiments have shown that the density rm (kg m23) of
Table 1
a
Measurement results of thermal properties of liquid GdF3
Regime
rL
(kg m23
bL
(3105 K21
drf
(%)
)
)
Cooling
Heating
Cooling
Weighted mean
a
561163
560863
560863
17.361.1
16.260.5
16.160.4
16.260.5
16.6660.11
16.7760.12
16.6560.09
16.760.15
r(T)5706020.1498(T2293)21.78331024(T2293)2
5609625
(293 . . . 1310 K, orthorhombic modification)
rL, bL are the density and volume coefficient of the thermal
expansion of the liquid at the melting point; drf 5 (rS 2 rL)/rS is the
relative density change on melting; rS is the density of the solid at the
melting point. A random error is listed for the measured values, and a
total error for the mean values. All are for a 0.95 confidence level.
(2)
r(T) 5 6867 2 0.6812 (T 2 1310)
(3)
(1310 . . . 1509 K, hexagonal modification)

32
S.V. Stankus et al. / Journal of Alloys and Compounds 290 (1999) 30–33
Fig. 1 shows our experimental results as well as
of the mean TECs are equal to 7.631025 K21 and 5.53
available literature data on the GdF3 density in the solid
and liquid states. Using data on the density of the phase of
LaF3 type at 293 K [3] and our results over the range 1310
to 1509 K one can evaluate the temperature dependence of
the density for this form in the metastable region (dashed
line in Fig. 1). It is noticeable that the difference of the
densities of the two modifications decreases with tempera-
ture, and near the melting point these values become very
close to each other. We determined the mean volume
coefficient of the thermal expansion (mean TEC) as
follows:
1025
K
K
21, respectively (so their difference is also 2.13
1025
21). However, because the transition temperature of
SmF3 (Tt ¯760 K [2,10]) is significantly lower than that of
GdF3, the absolute value of the relative density change on
polymorphic transformation of SmF3 (drt 522.9 to
23.6% [10,11]) is higher than that of GdF3.
As noted above, apart from SmF3 and GdF3, the solid–
solid transformation from the orthorhombic modification to
the hexagonal one is observed in europium trifluoride at
Tt 5920 K [2]. For this salt the relative difference between
the densities of the two forms at 293 K (3.4% [3]) is also
close to those of SmF3 and GdF3. One can suggest that the
mean TECs of the EuF3 solid phases will be of the same
order as those of the other two fluorides: kbl¯531025
K21 for the orthorhombic form, and kbl¯731025 K21
for the hexagonal one. These values allow one to evaluate
(in linear approximation) the temperature dependences of
the density in the solid state and the density change at the
polymorphic transformation for EuF3 (Fig. 2), whose
volume properties have not been essentially investigated
by experiment up to now. Our evaluation of the density of
the hexagonal phase is in good agreement with the value
calculated from the lattice parameters of this form de-
termined at 985 K by Thoma and Brunton [1], which can
be taken as evidence in favor of our assumptions.
1
r(293 K) 2 r(T)
]
]
kbl(T) 5
?
T 2 293
r(293 K)
so for the hexagonal phase (the range 293 K–Tf) and for
the orthorhombic phase (293 K–Tt) the values of kbl are
equal to 6.831025 K21 and 4.731025
K
21, respectively.
The difference of the mean TECs equals 2.131025
K
21. It
is of interest that the relative volume properties of solid
samarium trifluoride investigated by us previously [10] are
practically the same: the relative difference between the
densities of the metastable (LaF3 type) and stable (b-YF3
type) phases at 293 K is equal to 3.9% [3], and the values
The combination of our results and literature data allows
us to reach the following conclusions: (a) the relative
differences between the densities of stable (b-YF3 type)
and metastable (LaF3 type) forms of SmF3, EuF3, and
GdF3 at room temperature are closely related (3.4–3.9%).
Fig. 1. Temperature dependences of the GdF3 density in the solid and
liquid states. Open and solid circles are the results of this work obtained
with the remelted sample and with the sintered one, respectively: AL is
the melt; LS is the region of melting-crystallization; SB is hexagonal
phase; BC is the region of solid–solid transformation; CD is ortho-
rhombic phase; a, b are data of Ref. [3] on the density of the
orthorhombic (stable) and hexagonal (metastable) phases at room tem-
perature; c is the density of the hexagonal form at about 1185 K
calculated from the lattice parameters measured in Ref. [1]. Dashed line is
the evaluation of the temperature dependence of the density of the
hexagonal phase over the range from 293 K to the melting point.
Fig. 2. Evaluation of the temperature dependence of the EuF3 density in
the solid state. 1 is the orthorhombic form, 2 is the hexagonal form. The
solid circle is the value of the density of EuF3 hexagonal phase calculated
from the lattice parameters measured by Thoma and Brunton [1].

S.V. Stankus et al. / Journal of Alloys and Compounds 290 (1999) 30–33
33
Moreover, the relative differences between the densities of
the orthorhombic and the hypothetical (nonexistent in pure
salts) hexagonal phases of TbF3, DyF3, and HoF3 also lie
in this range [3]. (b) The mean TECs of the hexagonal
phases of the fluorides over the range from 293 K to the
smaller than the size of Gd31. Therefore, the first-order
solid–solid transitions in these salts had to occur at
temperatures much higher than 1310 K (Tt for GdF3). But
because TbF3 –HoF3 melt at 1420–1430 K, the first-order
transition is not observed in these compounds [2,9,13].
That is, TbF3, DyF3, and HoF3 conventionally remain in
the structure of b-YF3 type over the whole solid state
range. But actually, according to our evaluations [13], the
densities of the orthorhombic and hypothetical hexagonal
phases of DyF3 near the melting point are practically equal
to each other. In other words, in the vicinity of the melting
temperature the transformation in this compound (and,
probably, in TbF3 and HoF3) has already been completed
entirely by a continuous mechanism.
melting point are of the same order (|731025
K
21) and
in turn are significantly larger than those of the ortho-
rhombic modifications (|531025 21), so the difference
K
between the densities of two modifications decreases with
temperature. (c) The temperature of polymorphic trans-
formations in the SmF3 –GdF3 series rises (consequently,
the absolute value of the relative density change on
transition is diminished), as the size of cation decreases.
(d) The enthalpies of solid–solid transformations in SmF3 –
GdF3 are very small (about 10 times less than the
enthalpies of melting) [2].
To check the model of phase transformations in the
solids SmF3 –HoF3 proposed in this work, high-tempera-
ture investigations of the crystal structures of the rare earth
fluorides are required. We hope that our article will attract
interest to the research in this field.
These facts suggest that the distinctions between the
crystal structures of the orthorhombic and hexagonal forms
of the fluorides are also diminished with increasing
temperature. In fact, it has been demonstrated by
Garashina et al. [12], that the crystal lattices of b-YF3 type
and LaF3 type arise as a result of distinct distortions of a
common prototype structure (‘idealized’ lattice of LaF3
type). Therefore, the transformation from one form to the
other has a deformational mechanism and may occur over
a broad temperature range. This led us to propose that the
transformation from an orthorhombic form into a hexagon-
al one actually proceeds in two stages: first, over the
interval from 293 K to Tt the progressive deformation of
the crystal lattice of the b-YF3 type occurs; second, at the
transition temperature Tt a sharp change of lattice parame-
ters takes place. The first stage manifests itself by the fact
that the mean TEC of the orthorhombic phase is lower
when compared with that of the hexagonal one, resulting in
a decrease of the difference between the densities of two
modifications with increasing temperature. The second
stage shows itself as a first-order phase transition. As the
size of cation involved in the composition of the fluorides
of the SmF3 –GdF3 series is reduced, the value of Tt rises.
Consequently, the absolute value of the relative density
change at the first-order solid–solid transition is dimin-
ished, and the contribution of the continuous stage to the
overall process of transformation increases.
References
[1] R.E. Thoma, G.D. Brunton, Inorg. Chem. 5 (1966) 1937.
[2] F.H. Spedding, B.J. Beaudry, D.C. Henderson, J. Moorman, J.
Chem. Phys. 60 (1974) 1578.
[3] B.P. Sobolev, L.S. Garashina, P.P. Fedorov, N.L. Tkachenko, K.B.
Seiranyan, Kristallografiya 18 (1973) 751.
[4] N. Ikemiya, S. Hara, H. Maki, K. Ogino, J. Japan Inst. Metals 55
(1991) 1194.
[5] S.V. Stankus, R.A. Khairulin, P.V. Tyagel’sky, High Temp. 35 (1997)
988, (translated from Teplofizika Vysokikh Temperatur, 35 (1997)
999).
[6] R.A. Khairulin, S.V. Stankus, P.V. Tyagel’sky, High Temp. High
Press. 30 (1998) 479.
[7] S.V. Stankus, R.A. Khairulin, Teplofizika Vysokikh Temperatur 30
(1992) 487.
[8] S.V. Stankus, P.V. Tyagel’sky, Int. J. Thermophys. 15 (1994) 309.
[9] P.P. Fedorov, B.P. Sobolev, Kristallografiya 40 (1995) 315.
[10] S.V. Stankus, R.A. Khairulin, P.V. Tyagel’sky, High Temp. High
Press. 27–28 (1995/96) 493.
[11] K. Rotereau, Ph. Daniel, A. Desert, J.Y. Gesland, J. Phys. Condens.
Matter 10 (1998) 1431.
[12] L.S. Garashina, B.P. Sobolev, V.B. Aleksandrov, Yu.S. Vishnyakov,
Kristallografiya 25 (1980) 294.
[13] R.A. Khairulin, S.V. Stankus, P.V. Tyagel’sky, Inorg. Mater. 34
(1998) 742, (translated from Neorganicheskie Materialy, 34 (1998)
888).
The sizes of the cations in TbF3, DyF3, and HoF3 are