The phase diagram YF3-GdF3

Article abstract of DOI:10.1016/j.materresbull.2007.04.004

The binary phase diagram YF₃-GdF₃ was studied by differential scanning calorimetry (DSC). Yttrium fluoride and gadolinium fluoride show complete miscibility in all three phases (orthorhombic room temperature phase, trigonal or hexagonal high temperature phase, liquid). The transformations between room temperature and high temperature phases are of first order and occur at 1338.6 K (YF₃) or 1174.8 K (GdF₃). Melting points are 1403.1 K (YF₃) or 1525.7 K (GdF₃), respectively. The c_p (T) curve of GdF₃ shows a λ shaped local maximum at 1333 K that might be related to a further solid phase transformation of second order.

Full text of DOI:10.1016/j.materresbull.2007.04.004

Materials Research Bulletin 43 (2008) 676–681
www.elsevier.com/locate/matresbu
The phase diagram YF3–GdF3
b
a
b
D. Klimm a, , I.M. Ranieri , R. Bertram , S.L. Baldochi
*
a Institute for Crystal Growth, Max-Born-Str. 2, 12489 Berlin, Germany
b Center for Lasers and Applications, Inst. Pesquisas Energeticas & Nucl., CP 11049, Butanta˜ 05422-970, Sa˜o Paulo, SP, Brazil
Received 12 December 2006; received in revised form 28 March 2007; accepted 2 April 2007
Available online 8 April 2007
Abstract
The binary phase diagram YF3–GdF3 was studied by differential scanning calorimetry (DSC). Yttrium fluoride and gadolinium
fluoride show complete miscibility in all three phases (orthorhombic room temperature phase, trigonal or hexagonal high
temperature phase, liquid). The transformations between room temperature and high temperature phases are of first order and
occur at 1338.6 K (YF3) or 1174.8 K (GdF3). Melting points are 1403.1 K (YF3) or 1525.7 K (GdF3), respectively. The cpðTÞ curve
of GdF3 shows a l shaped local maximum at 1333 K that might be related to a further solid phase transformation of second order.
# 2007 Elsevier Ltd. All rights reserved.
PACS : 65.40.Ba; 81.30.Dz; 81.70.Pg
Keywords: A. Fluorides; C. Differential scanning calorimetry (DSC); D. Phase equilibria
1. Introduction
Most fluorides REF3 (RE = rare earth metal) show dimorphism. Depending on temperature T and on the RE, the
¯
¯
REF3 exist in an orthorhombic Pnma (b-YF3), a cubic Pm3m, a hexagonal P63cm (LaF3, tysonite), or a trigonal P3c1
structure, respectively. Sometimes the space symmetry groups P63=mmc and P6322 have been reported. The REF3
with smaller RE3+ (starting about at Sm3+ with an octahedral radius rSm ¼ 109:8 pm) are reported to undergo a
reconstructive phase transformation upon heating [1]. Yttrium (rY ¼ 104:0 pm) behaves here like a small rare earth
element. This phase transformation has been discussed controversially and seems to depend on the content of OHꢀ
impurities within the material [2–5]. The transformation is not observed for large RE (La3+, Ce3+, Pr3+, Nd3+) and bulk
single crystals of LaF3, CeF3, PrF3, and NdF3 can be grown from the melt without cracking [6,7].
If the radii of two RE3+are not too different, they can replace each other in crystals and doping is possible. Up to now,
most of the quasi binary systems between different REF3 have not been investigated, although some of them are
interesting examples for scintillator applications [7]. Moreover, some RE doped scheelite type crystals of the type
LiREF4 (e.g. LiYF4) are used for lasers[8]. Such compoundsare known to exist for thesmaller RE3+(startingabout atRE
= Eu). For RE3+with RE ¼ Eu, Gd, Tb, Dy, or Ho, respectively, the LiREF4 melt incongruently. For LiYF4–LiErF4 both
end members melt congruently and the system is quasi-binary [9]. Other compounds like LiGdF4 that display
* Corresponding author. Tel.: +49 30 6392 3024; fax: +49 30 6392 3003.
E-mail address: klimm@ikz-berlin.de (D. Klimm).
URL: www.ikz-berlin.de
0025-5408/$ – see front matter # 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.materresbull.2007.04.004

D. Klimm et al. / Materials Research Bulletin 43 (2008) 676–681
677
incongruent melting behavior may not be regarded as end members spanning quasi-binary phase diagrams. Hence,
detailed knowledge on the ternary system LiF–YF3–GdF3 is necessary for a complete description of the section LiYF4–
LiGdF4.
Besides the rim system LiF–GdF3 with one intermediate phase LiGdF4 melting incongruently [8], the rim system
LiF–YF3 with one intermediate compound LiYF4 melting congruently was reported previously [10]. Some contrary
reports claiming incongruent melting of LiYF4 may result from the small difference ( ꢁ 1 mol%) between this
compound and the neighboring eutectic LiYF4/YF3. The third rim system YF3–GdF3 is the subject of this study.
2. Experimental
YF3 and GdF3 are available from different commercial suppliers (e.g. Alfa Aesar, Aldrich) with nominal purities up
to 99.99%. Such chemicals were studied by simultaneous differential scanning calorimetry/thermogravimetry (DSC/
TG) as described later in this study and considerable mass loss up to several percent, together with irregular melting
(and sometimes even phase transformation) peaks was observed. One can assume that these effects are a result of
contamination by adsorbed water that reacts with the rare earth (RE) fluoride
2REF3 þ 3H2O @ RE2O3 þ 6HF "
(1)
under mass loss and contamination of the remaining REF3 with RE2O3. Considerably better thermoanalytic results
could be obtained from fluorides that were prepared from commercial Y2O3 and Gd2O3 (5N purity) by fluoridation.
The oxides were placed in a platinum boat inside a platinum tube. They were slowly heated in a stream of argon gas
ꢂ
(White Martins, purity 99.995%) and HF gas (Matheson Products, purity 99.99%) up to 850 C. This process is
described in detail in references [11,12]. Conversion rates around 99.96% of the theoretical value calculated for the
reactions Y2O3 þ 6 HF ! 2 YF3 þ 3 H2O or Gd2O3 þ 6 HF ! 2 GdF3 þ 3 H2O, respectively, were measured by
comparing the masses prior to and after the fluoridation process. If such pure REF3 were intentionally mixed with
RE2O3, irregular DSC peaks, but no mass loss was observed.
Thermoanalytic measurements were performed with a NETZSCH STA 449C ‘‘Jupiter’’ using a standard DSC/TG
sample carrier (thermocouples type S) and heating/cooling rates of ꢃ10 K/min to ꢁ 50 K above the fusion point Tf of
ꢂ
the samples, i.e. up to 1200–1350 C. The pure substances YF3 and GdF3 were investigated additionally with a cp-
DSC/TG sample carrier that allows a higher accuracy sufficient for measurements of the specific heat capacity cpðTÞ.
The sample carriers were calibrated for T and sensitivity at the phase transformation points of BaCO3 and at the
melting points of Zn, Au, and Ni. Sample powders (typically 20–50 mg) were placed in graphite DSC crucibles with
lid. Graphite crucibles were used for two reasons: (1) graphite is not wetted by the molten fluorides. Thus the melt
forms a single almost spherical drop (diameter d ꢁ 2 mm). The small surface reduces contamination and the small
volume enhances effective mixing of the sample. (2) Graphite removes residual water from the atmosphere very
ꢂ
effectively for T ꢄ 1000 C) [13]. For the cp measurements a commercial (NETZSCH) cp standard made of sapphire
(thickness 0.75 mm, mass 63.73 mg) was used. The specific heat capacity was calculated following the ASTM E 1269
method of the NETZSCH ‘‘Proteus’’ analysis software that compares curves from three subsequent DSC runs: (1)
empty crucible, (2) crucible with sapphire standard, (3) crucible with sample.
Solid solutions GdxY1ꢀxF3 were obtained by weighing appropriate quantities of YF3 and GdF3 powder directly into
the crucible. Mixing was performed by a first heating run above the melting point. Usually DSC curves from the
second heating run were used for analysis. This method of mixing proved to be superior to a preceding treatment e.g. in
a mortar, as the fluorides are known to be highly sensitive to traces of water in the atmosphere. The sample powders
were evacuated prior to heating to ꢁ 6 ꢅ 10ꢀ2 Pa within the vacuum-tight thermal analyzer to remove adsorbed water.
During the measurements a flow of 20 ml/min Ar (99.999% purity) was maintained.
3. Results
Fig. 1 shows the DSC curves measured with the cp-DSC/TG sample carrier for YF3 and GdF3 and a DSC curve for
one intermediate composition measured with a standard DSC/TG sample carrier. All curves show a first peak that is
due to the solid phase transformation of the substances and a second peak due to melting. The fusion point Tf of a pure
substance is the extrapolated onset of the melting peak [14], similarly the temperature of a first order solid phase
transformation Tt can be obtained from Tonset of the transformation peak.

678
D. Klimm et al. / Materials Research Bulletin 43 (2008) 676–681
Fig. 1. DSC heating curves obtained from the second heating of YF3, GdF3, and of a solid solution GdxY1ꢀxF3 (x ¼ 0:8090) showing the subsequent
phase transformation (t) and fusion (f) peaks. The inset shows the GdF3 curve for 850 ꢂC ꢆ T ꢆ 1150 ꢂC with larger scaling for DSC.
Mixed crystals do not show a sharp Tf but a melting range that is defined by the difference between liquidus and
solidus in the x ꢀ T phase diagram. The solidus can again easily be obtained from Tonset, but the determination of Tliq
from DSC data is not so straightforward. In this work it was found that the expression
Tliq ¼ Tsol þ DTðxÞ ꢀ DTð0Þ
(2)
gives reasonable values, if DTðxÞ ¼ Toffset ꢀ Tonset is the width of the melting peak at the intermediate composition
0 < x < 1 and DTð0Þ is the weighted (for x) width of the melting peaks for the pure end members.
From the thermodynamic point of view it makes no difference whether melting (sol $ liq) or another first order
phase transformation (high-T $ low-T) takes place. Hence, in analogy with the 2-phase-region high-T /liquid upon
melting, a 2-phase-region with the low-T and the high-T phase will occur. The phase boundaries can be determined
from the width of the phase transformation peak similar to (2). The lower boundaries of the two 2-phase-regions in the
YF3–GdF3 phase diagram (Fig. 2) could be obtained from the TonsetðxÞ of the phase transformation and melting peaks,
with typically ꢃ5 K reproducibility. The higher boundaries were obtained from the width of the peaks (2).
4. Discussion
The experimental points (circles and crosses in Fig. 2) show the expected topology of the phase diagram with
extended regions for the low-T, high-T, and liquid solution phases. All three phases GdxY1ꢀxF3 show complete
Fig. 2. Experimental points for the phase diagram YF3–GdF3 (circles: extrapolated onsets, crosses: calculated from peak width using (2) together
with a thermodynamic assessment).

D. Klimm et al. / Materials Research Bulletin 43 (2008) 676–681
679
Fig. 3. Gibbs free energy G for GdxY1ꢀxF3 in the three phases low-T (at T ¼ 1150 K), high-T (at 1370 K), and liquid (at 1550 K). Arrows indicate
the minima.
solubility for 0 ꢆ x ꢆ 1, but deviations from ideal solutions are obvious. The minimum of liquid around x ꢁ 0:25 and
the maximum of low-T around x ꢁ 0:10 lead to a narrow high-T region for small x (only ꢁ 20–30 K). Actually, the
width of the DSC peaks (cf. Fig. 1) exceeds the width of the high-T region for small x. The phase transformation and
melting peaks overlap partially leading to larger experimental errors for 0:1 < x < 0:4. Moreover, in this composition
range the phase transformation and melting processes are extended over a broader T range due to the wide 2-phase
regions. The wide 2-phase regions may lead to segregation and thus to different Tp and Tf in different parts of the
sample.
A thermodynamic assessment of the phase diagram was performed with the ChemSage [15] PC program. The SUBI
model (independent mixing of cations and anions) was used for both solid phases and RKMP (Redlich–Kister–
Muggianu) polynoms were used for liquid [15,16]. The agreement between the assessment results (lines in Fig. 2) and
the experimental data is fairly good, as the remaining differences do not exceed the experimental scatter and as the
main features (minima and maxima of the liquid and low-T regions) are reproduced well.
With the thermodynamic data file that was obtained by assessment one can calculate the Gibbs free energy G for the
three solution phases. Fig. 3 shows GðxÞ for such T, where only one phase is stable for all x. For all phases GðxÞ has a
minimum for intermediate x, but for high-T this minimum is most distant from the center near x ¼ 2=3, indicating the
pronounced stability of high-T at large x.
Surprisingly, even the fundamental thermodynamic data for the pure end members of the phase-diagram were
found to be different from values contained in thermodynamic databases. The ChemSage program [15] that was
available for this study relies on the compilation by Barin [17]. The Barin data are compared with the present results in
Table 1. For YF3 the enthalpy of phase transformation DHp has approximately the same magnitude as the enthalpy of
fusion DHf. However, Tp differs by 11 K and Tf by 25 K. Basically one could regard the Barin values as more reliable,
as these data are thoroughly cross-checked. Otherwise one should keep in mind that the present results were obtained
by measurements with two different sample holders that were calibrated repeatedly prior to and after the
measurements with YF3 and GdF3. One calibration point was the melting of gold at Tf ¼ 1336 K [18] which is very
close to Tp of YF3, thus ensuring high reliability near this point.
Discrepancies are much more pronounced for Tp of GdF3: Barin and this work differ by more than 170 K! This
difference cannot be explained by simple experimental errors. The DSC curve in Fig. 1 clearly shows the
ꢂ
transformation peak at 901.7 C (1174.8 K); one must assume that the Barin data are erroneous in this point.
ꢂ
However, a small endothermal effect around 1060 C (1333 K) is also visible in Fig. 1. It seems possible that this
Table 1
Comparison of solid state phase transformation and fusion data for YF3 and GdF3 (this work and Barin [17])
YF3
GdF3
This work
Barin
This work
Barin
Tp (K)
1338.6
21.138
1403.1
29.789
1350.0
32.468
1428.0
27.970
1174.8
9.195
1348.0
6.004
DHp (kJ/mol)
Tf (K)
DHf (kJ/mol)
1525.7
67.612
1505.0
52.426

680
D. Klimm et al. / Materials Research Bulletin 43 (2008) 676–681
Table 2
Fitting parameters for the cp expression (3)
c3 ꢅ 103
c4=107
T-range (K)
c1
c2
c5
Phase
YF3
300–1246.9
922.13
133.35
138.60
ꢀ14.041
ꢀ1.3084
ꢀ1.1815
2.1107
Low-T
High-T
Liquid
1247.0–1403.2
1403.3–1600.0
0
0
0
0
0
0
0
0
GdF3
300–1154.7
ꢀ163.59
3120.88
785.92
0.4797
ꢀ5.1795
ꢀ0.9768
0
ꢀ0.1826
2.2615
0.3996
0
1.11374
0
0
0
0
Low-T
High-T
High-T
Liquid
1154.8–1333.0
1333.1–1525.8
1525.9–1600.0
0
0
0
176.73
small effect was considered to be the result of a phase transformation by Barin. Indeed, the l shaped maximum of
cpðTÞ might be the result of a further and not yet reported second order solid phase transformation of GdF3.
The specific heat capacity cp for the three phases of YF3 and GdF3 were fitted to equations
c4
T2
cp ¼ c1 þ c2 T þ c3 T2 þ þ c5 T ln T
(3)
with fitting parameters ci that are given in Table 2.
Fig. 4 compares the Barin cp data for GdF3 with the values that were obtained in this work with powder samples. At
least for the low-T phase the values are similar around 100 J/mol K, but the present values depend stronger on T. For
high-Tand liquid Barin gives only constant values of 130.855 or 127.821 J/mol K, respectively. In this work, the liquid
data scatter around 177 J/mol K and the high-T data show a l-shaped maximum at 1333 K that is responsible for the
flat minimum of the GdF3 DSC curve in Fig. 1. This l peak might indicate an additional phase transformation (second
order) without apparent latent heat. In Table 2 the l shaped peak of cpðTÞ visible in Fig. 4 is described by two sets of
ciði ¼ 1; 2; 3Þ parameters for the high-T phase of GdF3.
The Barin compilation [17] is based on experimental data for 10 different REF3 that were obtained by Spedding
et al. [19] using DTA (for Tp and Tf) and drop calorimetry (for cp). No original DTA curves were reported by these
authors and thus the comparison with the present and with Thoma’s [20] contradicting results is not possible. It should
be noted, however, that Spedding et al. measured oxygen contaminations of their different REF3 samples ranging from
1 to > 300 ppm. Unfortunately, only for GdF3 no oxygen content data were given.
Fig. 4. Specific heat capacity cp of GdF3: comparison of data from Barin [17] and from this work.

D. Klimm et al. / Materials Research Bulletin 43 (2008) 676–681
681
Different rare earth oxyfluorides REOF and RE4O3F6 can undergo phase transformations that are marked by DTA
signals with peak temperatures (no onsets are reported) down to 613 ꢂ C (GdOF) or 560 ꢂ C (YOF) [21–23]. The high
conversion rates ꢇ 99:9% that could be obtained during our REF3 preparation process assures that such oxyfluoride
contamination is not relevant in the samples used in this study.
5. Conclusions
Like most other rare earth fluorides, GdF3and YF3 undergo upon heating a first order solid phase transformation
before melting. GdF3 and YF3 show complete miscibility in all three phases low-T, high-T, and liquid and
corresponding ‘‘one phase regions’’ can be found in the binary phase diagram. The one phase regions are separated by
two phase regions low-T /high-T and high-T /liquid.
ꢂ
In this paper, the first order phase transformation of GdF3 was found to occur at 902 C, in contrast to some
ꢂ
published data [15,17] that report a much higher transformation temperature (1075 C), but in agreement with a recent
paper by some of us [8]. It should be noted, however, that the present value is in close agreement with the older results
of Thoma and Brunton [20]. These authors prepared GdF3 with < 300 ppm oxygen content from commercial
Gd2O3 by hydrofluorination with ammonium bifluoride NH4F ꢈHF. Thoma and Brunton determined the phase
transformation by X-ray diffraction within high-vacuum better 4:5 ꢅ 10ꢀ4 Pa. This technique measures directly the
changing crystal structure and is not influenced by parasitic thermal effects that might result from chemical reactions
between sample and impurities. Therefore, the determination of the orthorhombic/hexagonal phase transformation is
expected to be highly reliable in [20].
Acknowledgements
The authors acknowledge financial support by the Deutscher Akademischer Austauschdienst (DAAD, PROBRAL
´
grant D/05/30364) and by the Coordenac¸a˜o de Aperfeic¸oamento de Pessoal de Nıvel Superior (CAPES, PROBRAL
grant Nr. 246/06).
References
[1] B.P. Sobolev, I.D. Ratnikova, P.P. Fedorov, Mater. Res. Bull. 11 (1976) 999–1004.
[2] R.C. Pastor, M. Robinson, Mater. Res. Bull. 9 (1974) 569–578.
[3] O. Greis, M.S.R. Cader, Thermochim. Acta 87 (1985) 145–150.
[4] S. Stankus, R.A. Khairulin, K.M. Lyapunov, J. Alloys Compd. 290 (1999) 30–33.
[5] K. Lyapunov, A. Baginskii, S. Stankus, J. Alloys Compd. 372 (2004) 7–9.
[6] W. Korczak, P. Mikołajczak, J. Cryst. Growth 61 (1983) 601–605.
´
[7] K. Shimamura, E.G. Vıllora, S. Nakakita, M. Nikl, N. Ichinose, J. Cryst. Growth 264 (2004) 208–215.
[8] I.M. Ranieri, A.H.A. Bressiani, S.P. Morato, S.L. Baldochi, J. Alloys Compd. 379 (2004) 95–98.
[9] J.S. Abell, I.R. Harris, B. Cockayne, J. Mater. Sci. 12 (1977) 670–674.
[10] I.R. Harris, H. Safi, N.A. Smith, M. Altunbas, B. Cockayne, J.G. Plant, J. Mater. Sci. 18 (1983) 1235–1243.
[11] H. Guggenheim, J. Appl. Phys. 34 (1963) 2482–2485;
see also: R. Bougon, J. Ehretsmann, J. Portier, A. Tressaud, P. Hagenmuller, Preparative Methods in Solid State Chemistry, Academic Press,
New York, 1972, p. 401.
[12] I.M. Ranieri, Ph.D. thesis, IPEN/University of Sa˜o Paulo (2001).
[13] D. Klimm, Thermochim. Acta 339 (1999) 111–116.
¨
[14] G. Hohne, W. Hemminger, H.-J. Flammersheim, Differential Scanning Calorimetry, Springer, Berlin Heidelberg, 1996.
[15] GTT Technologies, Kaiserstr. 100, 52134 Herzogenrath, Germany, ChemSage 4.20, http://www.gtt-technologies.de/(2000).
˚
[16] M. Hillert, B. Jansson, B. Sundman, J. Agren, Metall. Trans. A 16 (1985) 261–266.
[17] I. Barin, Thermodynamic Data of Pure Substances, VCH, Weinheim, 1993.
[18] GTT Technologies, Kaiserstr. 100, 52134 Herzogenrath, Germany, FactSage 5.4.1, http://www.factsage.com/ (2006).
[19] F.H. Spedding, B.J. Beaudry, D.C. Herderson, J. Moormann, J. Chem. Phys. 60 (1974) 1578–1588.
[20] R.E. Thoma, G.D. Brunton, Inorg. Chem. 5 (1966) 1937–1939.
[21] B.P. Sobolev, P.P. Fedorov, D.B. Shteynberg, B.V. Sinitsyn, G.S. Shakhkalamian, J. Solid State Chem. 17 (1976) 191–199.
[22] K. Niihara, S. Yajima, Bull. Chem. Soc. Japan 44 (1971) 643–648.
[23] K. Niihara, S. Yajima, Bull. Chem. Soc. Japan 45 (1972) 20–23.