Full text of DOI:10.1111/j.1151-2916.2002.tb00320.x

J. Am. Ceram. Soc., 85 [6] 1610–12 (2002)
journal
Methods of Purification of Zirconium Tetrafluoride for
Fluorozirconate Glass
Douglas R. MacFarlane, Peter J. Newman, and Alice Voelkel
School of Chemistry, Monash University, Clayton, Victoria 3800, Australia
High-purity zirconium tetrafluoride (ZrF ) has been produced
II. Experimental Procedure
4
using sublimation and distillation techniques. The transition-
metal impurities have been determined in these purified
samples, and it has been found that distillation is the preferred
method of purification using a number of criteria. A detailed
description of the apparatus used to produce this material is
also given.
(
1) Apparatus
All purification steps were conducted in a specially built
drybox. This had the considerable advantage that purified material
was never exposed to the atmosphere. The drybox atmosphere was
constantly purged with dry nitrogen obtained from liquid-nitrogen
boil-off. Purification was conducted in a quartz reactor fitted with
a glass lid (Fig. 1). Two aluminum rings, bolted together, held the
®
glass lid to the flange of the reactor. A Viton (DuPont, Wilming-
I. Introduction
ton, DE) O-ring between the two glass surfaces acted as a seal
between them. Viton was chosen because of its superior heat
resistance. All operations were conducted in a flowing atmosphere
of purified helium. Helium was chosen because the level of water
T IS now well established that, to produce transparent, crystal-
free fluoride glass, special attention has to be paid to decreasing
I
8
the level of optically absorbent impurities (e.g., transition-metal
cations) and to decreasing the level of impurities that can cause
extrinsic losses, (e.g., scattering from crystallites). In addition, we
have discovered that stable, crystal-free fluoride glasses containing
very high levels of lanthanide fluorides (required for microchip
laser applications) can be prepared only using extremely low
oxide-containing zirconium tetrafluoride (ZrF4). Commercially
available ZrF frequently contains sufficient oxide/water contam-
ination to limit the incorporation of high levels of, for instance,
ErF in a fluorozirconate glass. Therefore, further purification of
and oxygen could be decreased to ppb levels. This was accom-
plished by first passing the gas through a U-tube filled with
molecular sieves (4 A) at 77 K and then passing the gas through a
column filled with activated manganese oxide (IOT-2-HP, Agilent
Technologies, Palo Alto, CA). A 0.5 ␮m filter was used to remove
any particulate matter from the gas stream.
1
–3
Vitreous carbon (Sigridur type G, Hochtemperatur-Werkstoffe
GmbH, Andermach, Germany) crucibles were used for all exper-
iments. Vitreous carbon was chosen because it is nonreactive
toward fluoride melts, and it couples very effectively with the field
generated by an induction furnace. Initial experiments on a small
scale were conducted using a pair of GAK3 crucibles (the volume
of each crucible was 67 mL) held mouth-to-mouth using a special
quartz assembly (Fig. 1). In latter experiments, it was possible to
scale up the purification (see below), and, in this case, the lower
vessel was a GAK10 crucible (volume of 116 mL). The top vessel
was a GAZ13 crucible (volume of 150 mL). To hold these two
crucibles together, a special grooved ring was fabricated from
vitreous carbon.
In both cases, the lower crucible was heated using induction
heating. The busbars of a 7 kW induction furnace (Radyne,
Milwaukee, WI) operating at 450 kHz were sealed through the
back of the drybox, and the jacket was positioned in place such that
the single coil of the induction furnace surrounded the lower
crucible. This arrangement ensured that only the lower crucible
was heated and that a temperature gradient was automatically
established. To combat high temperatures occurring in the drybox
during processing, a water-cooled copper heat shield surrounded
the quartz reactor, and a refrigeration unit was installed in the
drybox.
4
3
commercially available ZrF is necessary if these glasses are to
4
become useful in applications such as optically pumped single-
longitudinal-mode rare-earth-doped laser systems operating in the
eye-safe region. This communication describes in detail the
methods we have adopted to prepare such starting materials for
glass synthesis.
4
Ewing and Sommers have published a comprehensive review
on the purification and analysis of metal fluorides. Sublimation of
ZrF4 is the most-cited method in the literature. It is a simple
method that decreases transition elements and improves the anhy-
drous and oxide quality of ZrF . However, another less-cited
4
technique involves distillation of ZrF from a ZrF –BaF melt.
4
4
2
5
–7
This procedure was first reported by Robinson
cerned with purifying highly contaminated ZrF . Robinson low-
ered the level of iron from 320 to 4 ppm, whereas simple
sublimation lowered the iron to ϳ30 ppm. This communication
and was con-
4
describes the purification of ZrF using sublimation and distilla-
4
tion techniques but starting here with a much purer commercial
product as a feed stock. Therefore, the question is whether either
or both of these techniques reach their thermodynamic limit
around the ppm level or can continue to offer a route to purifica-
tion into the 100 ppb range.
(2) Reagents
A single 1 kg batch of ZrF (Batch No. 28039, Morita Kakagu,
4
Osaka, Japan) was used for all studies. Other chemicals used were
BaF (Fluorotran grade, BDH Chemicals, Pty., Ltd., Poole, U.K.),
2
NaF (Fluorotran grade, BDH), and zirconium metal (99.99%,
Cerac, Inc., Milwaukee, WI).
E. Snitzer—contributing editor
(3) Sublimation of ZrF4
Initially, 20 g of ZrF was placed in the lower crucible. The
apparatus was assembled, and the atmosphere inside the reactor
Manuscript No. 187520. Received August 14, 2001; approved February 28, 2002.
Supported by the Australian Research Council.
4
1
610

June 2002
Communications of the American Ceramic Society
1611
III. Results
Purified ZrF was prepared either by sublimation or by distil-
4
lation from a ZrF –BaF melt using the apparatus described above.
4
2
Correct placement of the bottom crucible in the coil of the
induction furnace consistently resulted in an easily removed
dendritic polycrystalline material depositing in the upper crucible.
Yields for either sublimation or distillation were similar.
To compare the cationic purity of these products, the transition-
metal impurities were monitored. This involved the development
of a complex methodology of dissolution of ZrF , complexation of
4
the transition metals, and then extraction and analysis using
8
graphite-furnace AAS techniques. We concentrated on measuring
the levels of iron, nickel, and copper. Table I gives the values of
these elements in two samples of unprocessed ZrF , two different
4
samples of sublimed ZrF , and five different samples of distilled
4
ZrF4.
In addition to transition-metal analysis, a number of other
physicochemical measurements were made on the distilled ZrF to
4
further ascertain the degree of purity of this material compared
with commercially available material. The distilled product was
subjected to XRD, FTIR, and SIMS analysis.
(
1) Samples of ZrF were analyzed using powder XRD, and
4
11
the pattern compared with a library pattern. There was found to
be a very good match with Powder Diffraction File 33–1480,
which is for the ␤ form of ZrF . Commercial samples showed
4
additional small peaks.
(2) Using SIMS, Charles Evans and Associates found the
levels of oxygen and carbon in the distilled ZrF to be at or below
4
the instrumental background, i.e., carbon Ͻ0.18 ppm and oxygen
Ͻ0.47 ppm. This is a substantial improvement over the results for
the commercial ZrF , which showed carbon between 0.8 and 2.0
4
ppm and oxygen between 5 and 15 ppm. Special precautions were
taken to seal all samples from atmospheric moisture before
dispatch to the United States for SIMS evaluation so that valid
comparisons could be made.
1
0
Fig. 1. Quartz reactor complete with GAZ6 vitreous carbon crucibles.
Coil of the induction furnace is not shown for clarity.
(3) DRIFTS is a sensitive probe for OH contamination.
Figure 2 illustrates the DRIFTS spectra of the Morita ZrF starting
4
material and distilled ZrF . Figure 2 shows that the peaks at 3400
4
Ϫ1
and 1600 cm
due to OH are very much diminished in the
purified sample.
replaced with dry helium. A flow (50 mL/min) of helium was
Because the distillation method was adopted as the superior
technique, it was then possible to examine another refinement of
this method. This is the electromotive series displacement (ESD)
maintained at all times during the purification. The ZrF was first
dried at low temperature (to remove any volatiles), and then the
temperature was increased to 850°–875°C for 1 h. The yield was
4
1
2
procedure described by Pastor and Robinson. ESD relies on the
plating out of reducible transition-metal impurities on a finely
dispersed electropositive material, e.g., zirconium particles. How-
ever, when finely divided zirconium metal was added to a
ZrF –BaF melt and then distilled, a material was obtained that
1
0 g. This method was subsequently scaled up using the larger
crucibles, and as much as 100 g of ZrF was sublimed to give ϳ50
4
g of purified material in 3 h.
4
2
(
4) Distillation of ZrF –BaF Melts
4 2
was quite different from previously purified ZrF samples. The
4
A similar procedure to that described above was used to distill
ZrF –BaF melts. A mixture of 100 g of ZrF and 25 g of BaF
polycrystalline solid in the upper crucible varied in color from
brown to gray, and the crucibles were coated with a black metallic
4
2
4
2
7
(
81:19 mol% ZrF :BaF ) was placed in the lower GAK10 vitreous
4 2
substance. These observations were not recorded by Robinson,
carbon crucible. This mixture was first dried at low temperature,
and then the temperature was increased to 850°–875°C and held at
this temperature for 3 h. A distillation rate of 15 g/h was achieved.
In some experiments, the distillation was allowed to proceed for
another 1 or 2 h. However, the production rate decreased to Ͻ10
g/h in the 4th and 5th hour. In some experiments, pure zirconium
metal (2wt%) was added to the melt before distillation.
and this method was abandoned.
One problem encountered concerned the substantial mass of
melt left after the distillation was complete (ϳ55 g ZrF and 25 g
4
Table I. Transition-Metal Impurities in ZrF Samples
4
Impurity (ppm)
ZrF
4
sample
No.
Iron (Ϯ0.04)
Nickel (Ϯ0.02)
Copper (Ϯ0.02)
(5) Analysis
Trace-metal analysis was performed using graphite-furnace
Commercial
Sublimation
Distillation
1
2
1.06
1.20
0.53
0.46
0.18
0.22
atomic absorption spectroscopy (AAS; Model SpectrAA 400 and
9
Model GTA 96 graphite tube atomizer, Varian, Palo Alto, CA).
1
2
0.53
0.44
0.25
0.20
0.10
0.06
Oxygen and carbon were measured using secondary ion mass
spectroscopy (SIMS; by Charles Evans and Associates, Sunnyvale,
CA). Diffuse reflectance FTIR spectra (DRIFTS) were measured
1
2
3
4
5
0.18
0.26
0.12
0.14
0.24
0.21
0.27
0.21
0.20
0.21
0.10
0.11
0.12
0.10
0.12
(
Telstra Research Laboratories, Melbourne, Australia), as previ-
10
ously described. Spectra were recorded in transmission mode.
Powder X-ray diffractometry (XRD; Scintag, Santa Clara, CA)
studies were conducted.

1
612
Communications of the American Ceramic Society
Vol. 85, No. 6
(2) The presence of a liquid is much more favorable for
chemical homogeneity and heat transfer. In sublimation, thermal
contact between the solid and the crucible may be variable.
Whereas the yields for both processes were comparable, the yield
for the distillation process did decrease during the 4th and 5th
hours. The reason for this was that, as the distillation proceeded,
the melt became barium-rich and higher temperatures were re-
quired to continue distilling ZrF from the melt.
4
(3) The melt functions to physically wet oxide particles,
preventing them from contaminating the final product. In subli-
mation, an impurity-rich oxide residue results, and it is possible for
this oxide impurity to become entrained in the vapor.
(4) Because a melt is produced, it is possible to use ESD. This
is not possible using simple sublimation techniques. However, our
results show that ESD using zirconium metal produces a variable
material that is not suitable for glass synthesis.
Fig. 2. DRIFTS spectra of Morita ZrF (upper curve) and distilled ZrF
4
4
(lower curve). (Spectra are offset with respect to each other for clarity).
BaF ). To maximize the yield of purified ZrF , we investigated the
V. Conclusion
2
4
effect of adding further aliquots of impure ZrF4 to the melt
remaining after distillation and then redistilling this mixture. This
procedure was repeated three times and a sample of ZrF4 was
analyzed from each distillation. Although the impurity level of the
Purification of ZrF4 by distillation of ZrF –BaF melts was
4
2
achieved. The purified material was extremely dry and had low
levels of carbon, oxygen, and transition metals. These results
showed that distillation was an effective method and was superior
to the frequently reported method of sublimation in the literature.
Using this material, we were able to report the highest level of
melt increased with each addition of ZrF , no increase in
4
transition-metal impurities was found in any of the samples of
purified ZrF ; i.e., all four samples returned similar levels of iron,
4
1
–3
Er(III) doping in fluorozirconate glasses.
nickel, and copper.
References
IV. Discussion
1
V. K. Bogdanov, W. E. K. Gibbs, D. J. Booth, J. S. Javorniczky, P. J. Newman,
and D. R. MacFarlane, “Fluorescence from Highly Doped Erbium Fluorozirconate
Glasses Pumped at 800 nm,” Optics Commun., 132, 73–76 (1996).
This research program has established that distillation is the
method of choice for purification of ZrF for several reasons.
4
2
(
1) The analytical values given in Table I show that sublima-
tion decreases the iron, nickel, and copper impurities by a factor of
, while distillation decreases the iron by a factor of 8–10. The
D. R. MacFarlane, J. S. Javorniczky, P. J. Newman, V. K. Bogdanov, D. J. Booth,
and W. E. K. Gibbs, “High-Er(III)-Content ZBN Glasses for Microchip Laser
Applications,” J. Non-Cryst. Solids, 213 & 214, 158 (1995).
2
3
J. S. Javorniczky, “Rare-Earth-Containing Fluoride Glasses for Optoelectronic
substantial decrease in the iron impurity resulting from distillation
is due to two factors. First, the melt decreases the sublimation
Devices”; Ph. D. Thesis. Monash University, Clayton, Australia, 1997.
K. J. Ewing and J. A. Sommers, “Purification and Analysis of Metal Fluorides”;
pp. 142–208 in Fluoride Glass Fiber Optics. Edited by I. D. Aggarwal and G. Lu.
4
3
ϩ
2ϩ
tendency of Fe and Fe because of a vapor pressure decrease
of their fluorides as a consequence of Raoult’s law. Second, there
is a tendency of FeF to fluorinate ZrO (Eq. (1)) to form Fe O .
Academic Press, Boston, MA, 1991.
5
M. Robinson, “Preparation and Purification of Fluoride Glass Starting Materials,”
3
2
2
3
Mater. Sci. Forum, 5, 19–34 (1985).
6
ZrO is produced by the reaction of water with ZrF (Eq. (2)) or by
M. Robinson, “Processing and Purification Techniques of Heavy-Metal Fluoride
2
4
1
3
Glass (HMFG),” J. Cryst. Growth, 75, 184–94 (1986).
the reaction of BaO with ZrF (Eq. (4)). (BaO is the hydrolysis
4
7
M. Robinson, “High-Purity Components for Fluorozirconate Glass Optical Fi-
product of BaF2 and water (Eq. (3))). The advantage of this
reaction scheme is that Fe O is far less volatile than FeF . This
bers”; pp. 11–25 in Halide Glasses for Infrared Optics. Edited by R. M. Almeida.
2
3
3
Martinus Nijhoff, Dordrecht, The Netherlands, 1987.
8
scheme is analogous to the decrease of iron in sublimation by the
D. F. Shriver and M. Drezdzon, The Manipulation of Air-Sensitive Compounds;
1
4
pp. 82–83. Wiley, New York, 1986.
deliberate addition of oxides, first reported by Tatsuno.
9
P. J. Newman, A. T. Voelkel, and D. R. MacFarlane, “Analysis of Fe, Cu, Ni, and
Co in Fluoride Glasses and Their Precursors,” J. Non-Cryst. Solids, 184, 324–28
4
2
ZrO ϩ FeF 3 ZrF ϩ Fe O
(1)
(2)
2
3
4
2
3
(1995).
3
3
1
0
C. Byrne and G. Rosman, “FTIR Spectroscopy of Fluoride Powders, Glasses, and
Fibres”; pp. 8.59–8.64 in Extended Abstracts of 7th International Symposium on
Halide Glasses (Lorne, Australia, 1991). (Available from the Conference Chair,
d.macfarlane@sci.monash.edu.au.)
ZrF ϩ 2H O 3 ZrO ϩ 4HF
4
2
2
1
1
BaF ϩ H O 3 BaO ϩ 2HF
(3)
(4)
Powder Diffraction File No. 33–1480. International Centre for Diffraction Data,
Newtown Square, PA, 1987.
2
2
1
2
ZrF ϩ 2BaO 3 ZrO ϩ 2BaF
2
R. C. Pastor and M. Robinson, “Method for Preparing Ultra-Pure Zirconium and
Hafnium Tetrafluorides,” U.S. Pat. No. 4 578 252, 1985.
4
2
1
3
K. Fujiura, S. Sakaguchi, Y. Ohishi, and Y. Terunuma, “Formation Reaction of
Scatterers in ZrF -Based Fluoride Fibers,” J. Am. Ceram. Soc., 71 [5] 460–64
1988).
One important consequence of this reaction scheme is that multiple
distillations are therefore possible, with no increase in impurity
level. This is advantageous for producing large amounts of
material with little waste.
ZrO
2
4
(
14
T. Tatsuno, “Synthesis of Iron-Impurity-Free ZrF
181–86 (1987).
4
,” Mater. Sci. Forum, 19– 20,
Ⅺ