Preparation and characterization of high-purity metal fluorides for photonic applications

Article abstract of DOI:10.1111/j.1551-2916.2011.04641.x

We combine chelate-assisted solvent extraction (CASE) and hot hydrogen fluoride gas treatment to enable a general method for the preparation of high-purity binary metal fluorides. The fluorozirconate glass ZBLANI:Yb ³⁺ (ZrF₄-BaF₂-LaF₃-AlF ₃-NaF-InF₃-YbF₃), a solid-state laser-cooling material, is used as a test case to quantitatively assess the effectiveness of the purification method. The reduction of transition-metal and oxygen-based impurities is quantified directly by inductively coupled plasma mass spectrometry (ICP-MS) and indirectly by laser-induced cooling, respectively. The concentrations of Cu, Fe, Co, Ni, V, Cr, Mn, and Zn impurities in the ZrCl ₂O precursor solution were measured individually by ICP-MS at various stages of the purification process. CASE was found to reduce the total transition-metal concentration from 72500 to ～100 ppb. Laser cooling was most efficient in ZBLANI:Yb³⁺ glass fabricated from CASE-purified metal fluoride precursors, confirming the results of the ICP-MS analysis and demonstrating the effectiveness of the purification methods in a finished optical material. High-purity metal fluorides prepared by the methods presented herein will enable new high-performance optical materials for solid-state optical refrigerators, crystals for vacuum ultraviolet (VUV) spectroscopy of the Thorium-229 nucleus, VUV optics, fibers, and thin-film coatings.

Full text of DOI:10.1111/j.1551-2916.2011.04641.x

J. Am. Ceram. Soc., 94 [9] 2896–2901 (2011)
DOI: 10.1111/j.1551-2916.2011.04641.x
© 2011 The American Ceramic Society
ournal
J
Preparation and Characterization of High-Purity Metal
Fluorides for Photonic Applications*
Wendy M. Patterson,‡,† Peter C. Stark,§ Thomas M. Yoshida,§Mansoor Sheik-Bahae,‡
and Markus P. Hehlen¶
‡Department of Physics and Astronomy, University of New Mexico, Albuquerque, New Mexico, 87131
§Chemistry Division, Los Alamos National Laboratory, Mailstop J964, Los Alamos, New Mexico, 87545
¶Materials Science & Technology Division, Los Alamos National Laboratory, Mailstop E549,
Los Alamos, New Mexico, 87545
We combine chelate-assisted solvent extraction (CASE) and
hot hydrogen fluoride gas treatment to enable a general method
for the preparation of high-purity binary metal fluorides. The
fluorozirconate glass ZBLANI:Yb3+ (ZrF4–BaF2–LaF3–AlF3–
NaF–InF3–YbF3), a solid-state laser-cooling material, is used
as a test case to quantitatively assess the effectiveness of the
purification method. The reduction of transition-metal and oxy-
gen-based impurities is quantified directly by inductively cou-
pled plasma mass spectrometry (ICP-MS) and indirectly by
laser-induced cooling, respectively. The concentrations of Cu,
Fe, Co, Ni, V, Cr, Mn, and Zn impurities in the ZrCl2O pre-
cursor solution were measured individually by ICP-MS at vari-
ous stages of the purification process. CASE was found to
reduce the total transition-metal concentration from 72500 to
~100 ppb. Laser cooling was most efficient in ZBLANI:Yb3+
glass fabricated from CASE-purified metal fluoride precursors,
confirming the results of the ICP-MS analysis and demonstrat-
ing the effectiveness of the purification methods in a finished
optical material. High-purity metal fluorides prepared by the
methods presented herein will enable new high-performance
optical materials for solid-state optical refrigerators, crystals
for vacuum ultraviolet (VUV) spectroscopy of the Thorium-229
nucleus, VUV optics, fibers, and thin-film coatings.
transfer, even at trace concentrations.15 Likewise, oxygen-
based impurities such as metal oxides, oxyfluorides, hydroxyl
ions (OH-), and water can degrade material performance by
introducing UV absorptions16,17 and high-energy vibrational
modes that enhance multi-phonon relaxation rates of excited
states.18–23 The slow growth of fluoride crystals can reduce
the concentration of some of these impurities in the final
material. Fluoride glasses on the other hand are more suscep-
tible to incorporation of impurities because a glass generally
contains all of the species that were present in the melt from
which it was quenched.
Solid-state laser coolers are a class of photonic materials
with particularly demanding purity requirements. Laser cool-
ing removes thermal energy from a solid by anti-Stokes
fluorescence and thereby reduces its temperature. A compre-
hensive introduction to solid-state laser refrigeration has
been given elsewhere.11 Impurities can introduce several
undesirable processes that cause internal heating and there-
fore degrade the laser-cooling performance. The primary
quenching mechanism is by non-radiative energy transfer from
the excited laser-cooling ion (e.g. Yb3+) to transition-metal
impurities (such as Cu2+, Fe2+, Co2+, and Ni2+) which
subsequently decay non-radiatively and thereby produce
undesired internal heating of the solid.11,15 Furthermore,
impurities with high-energy vibrational modes (such as OHÀ
and H2O) can quench the excited state of the laser-cooling
ion via multi-phonon relaxation. Finally, some impurities can
also directly absorb at the pump wavelength causing heating
in the form of background absorption. Reducing transition-
metal and oxygen-based impurities suppresses these unde-
sired impurity-induced decay channels and thus improves
the laser-cooling performance. Estimates for Yb3+-doped
fluorozirconate glass ZBLAN (ZrF4–BaF2–LaF3–AlF3–NaF)
indicate that transition-metal and OH- impurities in excess
of ~100 parts-per-billion (ppb) begin to compromise laser-
cooling performance.15 Other applications have similarly
stringent purity requirements.12 The preparation of high-
purity binary fluoride starting materials is therefore critical
toward enabling the desired performance of many fluoride
optical materials.
I. Introduction
LUORIDE crystals and glasses are of interest to a wide
F
range of photonic applications including bulk optics for
vacuum ultraviolet (VUV) photolithography,1 optical fibers
for high power or long-haul transmission,2,3 host materials
for lasers, and optical amplifiers,2,4–7 scintillators,8,9 up-con-
version phosphors,10 solid-state optical refrigerators,11 and
emerging crystals for optical spectroscopy of the Thorium-
229 nucleus.12 These applications take advantage of the high
bandgap energy (>6 eV) and/or the low optical phonon ener-
gies (<500 cmÀ1) of fluorides. The attractive intrinsic proper-
ties of fluorides, however, can be substantially degraded by
the presence of impurities. Transition-metal ions have optical
transitions in the ultraviolet (UV) and visible spectral
regions13,14 that lead to undesired background absorption as
well as non-radiative quenching of excited states via energy
There currently exists no single comprehensive method for
the purification and synthesis of ultra-pure metal fluorides.
The chemical inertness and relatively high melting tempera-
tures of fluorides prevent the use of many of the standard
purification methods. Some successes have been achieved by
sublimation and distillation of ZrF4 as well as zone refin-
ing,24,25 but these methods are not applicable to many of the
other metal fluorides of interest. Reactive gas atmospheres of
CF4, NF3, SF6, Cl2, or CC14 have been used during fluoride
melting to reduce oxygen-based impurities,26,27 but these
molecules themselves often incorporate the final glass and
J. Ballato—contributing editor
Manuscript No. 29422. Received March 08, 2011; approved April 21, 2011.
*Based in part on the thesis submitted by W. M. Patterson for the PhD degree in
Optical Science and Engineering, University of New Mexico, Albuquerque, NM 87131,
(2009)
†Author to whom correspondence should be addressed. e-mail: wendy5@unm.edu
2896

September 2011
Preparation of High-Purity Metal Fluorides
2897
can be detrimental in many applications. Vapor-phase
processes appear attractive due to the success of chemical
vapor deposition (CVD) in purifying silica for fiber technol-
ogy. Although some studies have obtained encouraging
results,28–30 the preparation of bulk quantities of fluoride
glasses by CVD is difficult because of the high evaporation
temperature of many binary fluorides. Therefore, a compre-
hensive purification strategy must target the chemistry before
the formation of the metal fluoride. One such method is
chelate-assisted solvent extraction (CASE). In this process,
an organic chelate is introduced to bind to undesired metal
ions in an aqueous phase and to transfer the resulting metal-
chelate complexes into a second organic phase. CASE is
widely used in analytical chemistry to pre-concentrate heavy
metals for trace analysis.31–36 Alternatively, CASE can
remove metal-ion impurities from an aqueous solution of a
desired metal ion and thus achieve purification. Ling et al.
have used the latter approach with ammonium-pyrrolidine-
dithio-carbamate (APDC) as the chelate and methyl-isobutyl-
ketone (MIBK) as the organic phase to reduce Co, Ni, and
Cu to <5 ppb, and Fe to <10 ppb in zirconium solutions.36
The goal of the present study is to develop APDC/MIBK
as a single chelate/organic solvent system suitable for the
individual purification of all components of the laser-cooling
glass ZBLANI:Yb3+(ZrF4–BaF2–LaF3–AlF3–NaF–InF3–YbF3).
The ability to perform purification of all respective metal
fluoride precursors with a single chelate/solvent system mini-
mizes the development effort and greatly simplifies process-
ing. Specifically, we (1) expand CASE to the purification of
main group elements (Zr, Ba, Al, Na, In) and the rare earths
(La, Yb), (2) advance the method to include the precipita-
tion, drying, and fluorination of the resulting binary metal
fluorides, and (3) quantitatively verify the impact of reduced
impurities on the performance of a ZBLANI:Yb3+ optical
refrigerator material synthesized from the purified binary
metal fluorides.
(HNO3, 69%), and hydrochloric acid (HCl, 33%) were
doubly distilled and trace-metal grade (GFS Chemicals,
Powell, OH). Ultra-high purity (UHP) water (Inorganic Ven-
tures, Christiansburg, VA) and electronic grade 4-Methyl-2-
pentanone (MIBK; Acros Organics, Geel, Belgium) were
used. All processing occurred in a fume hood inside a Class
100 clean room. Acid evaporations were performed in a dis-
tillation still constructed from perfluoroalkoxy (PFA) resin,
with the evaporation side being heated to 190°C by a temper-
ature-controlled heating jacket. All vessels and utensils were
cleaned by leaching them in dilute nitric acid at elevated tem-
peratures for several days before use. A 1% solution of
APDC chelate (Acros Organics) in UHP water was prepared
daily by dissolving 1 g of APDC in 100 mL of water and fil-
tering through a < 0.2 lm pore size nylon syringe filter. A
buffer solution was prepared by dissolving 50 g of ammo-
nium acetate (Fluka, Sigma-Aldrich, St. Louis, MO;
99.995%) in 100 mL of UHP water and adding acetic acid
to adjust the pH to ~4.5.
(2) Chelate-Assisted Solvent Extraction (CASE) and
Precipitation of Metal Fluoride
Processing details are summarized in Table I. The general
strategy involved dissolving the starting material in acid or
UHP water, buffering, and filtering through a < 0.2 lm pore
size Nylon syringe filter into a separatory funnel to carry out
CASE at room temperature. A quantity of 5 mL of 1%
APDC solution was added and the mixture was briefly
shaken. A quantity of 35 mL of MIBK was added, and the
two-phase system was shaken vigorously for 1 min, and the
aqueous phase was collected after the system had equili-
brated for 10 min. This extraction procedure was repeated
three times, the final aqueous phase was collected in a poly-
tetrafluoroethylene (PTFE) beaker, and HF acid was added
to precipitate the metal fluoride. To remove residual buffer,
excess acid, and non-fluoride water-soluble compounds,
30 mL of UHP water was added to the metal fluoride precip-
itate, the slurry was sonicated for 1 h, and the water was
decanted. This washing procedure was repeated three times,
and the metal fluoride was dried in a PTFE beaker at 190°C.
In the case of Zr, the initial ZrCl2O∙8H2O solution was
boiled and water was added continuously (350 mL) to raise
II. Experimental Procedure
(1) General Procedures
Concentrated hydrofluoric acid and hot hydrogen fluoride
gas are toxic and can pose a severe health hazard if handled
improperly. Hydrofluoric acid (HF, 49%), nitric acid
Table I. Precursors for the Melt Synthesis of ZBLANI:Yb3+ Glass. The First Column Shows the Vendor-Quoted Cationic Purity
of Commercially Available Metal Fluorides. The Subsequent Columns Show the Details of the Chelate-Assisted Solvent Extraction
(CASE) and Metal Fluoride Precipitation Steps Described in Sections II(1) and II(2). Note that for Ba and La, In, Yb the
Respective Chloride and Nitrate is First Formed Before Dissolving in UHP Water for Subsequent CASE
Quantity (g)
CASE starting material,
vendor, purity (%)
pH for
CASE
Starting
material
Fluoride, purity (%)
Dissolution of starting material
Buffer
HF
ZrF4, 99.5%
ZrCl2O∙8H2O, Alfa
Aesar, 99.9985%
BaCO3, Strem
Chemicals, 99.999%
La2O3, Metall Rare
Earth Ltd, 99.99993%
AlCl3∙6H2O, Alfa
Aesar, 99.99956%
Na2CO3,
Sigma-Aldrich, 99.997%
Indium metal,
Alfa Aesar, 99.9999%
Yb2O3, Metall Rare
Earth Ltd, 99.99992%
68 g UHP water
2.5
3.8
3.3
2.8
5.0
2.5
3.6
41
60
28.0
BaF2, 99.99%
LaF3, 99.9%
AlF3, 99.9%
NaF, 99.5%
InF3, 99%
(1) 30 g HCl (slow)
(2) 34.5 g UHP water
(1) 38 g HNO3
(2) 27 g UHP water
27 g UHP water
10
33
18.5
50
42.5
21.4
40.6
40.5
25.4
21.3
2.9
6.0
7.0
7.7
2.5
32 g UHP water
73
(1) 100 g of HNO3, (slow)
(2) 32 g UHP water
(1) 35.5 g HNO3
25
YbF3, 99.9%
21.3
(2) 22 g UHP water
Alfa Aesar, Ward Hill, MA.
Strem Chemicals, Newburyport, MA.
Metall Rare Earth Ltd, New Territories, Hongkong, China.
Sigma-Aldrich, St. Louis, MO.

2898
Journal of the American Ceramic Society—Patterson et al.
Vol. 94, No. 9
the pH to ~1 to reduce the amount of buffer needed. Both
Zr and In precipitated as the respective ammonium fluoride
compounds, which were thermally decomposed in glassy-
carbon beakers to ZrF4·H2O (at 370°C) and InF3 (at 460°C),
respectively. Ba was first converted to the chloride and then
dissolved in water. La, Yb, and In were first converted to the
respective nitrate and then dissolved in water. In the case of
La, slow, drop-wise addition of HF acid under stirring
(550 rpm, PTFE stir bar) favored the precipitation of LaF3
over the competing precipitation of ammonium compounds.
mond saw, and all sides were polished. The two parallel sides
designated to receive the pump laser beam were further pol-
ished by a wet process. Al2O3 particles (sizes from 12 to
0.05 lm) were suspended in emulsions of water-free ethylene
glycol and glycerin, adjusting the viscosity by the amount of
glycol. Polishing slurries were prepared daily, and the wax
mounted samples were polished with slurries of decreasing
particle size on fine-textured, non-abrasive polishing cloths.
Samples were sonicated in optical grade methanol for several
hours and cleaned with lens paper before use.
(3) Drying of Metal Fluorides in Hydrogen Fluoride (HF)
Gas
(5) Sample Characterization
A high-resolution magnetic sector inductively coupled plasma
mass spectrometer (ICP-MS, Thermo Electron Corporation
Element II) was used to measure trace metals at various
stages during the CASE process. The ICP-MS had low parts-
per-trillion (ppt) sensitivity with linearity over nine orders of
magnitude. Samples were collected from the Zr4+ solution
immediately before CASE, after each of the CASE steps, and
from the UPH water. All samples except the UPH water
were digested to reduce interference of the ammonium ace-
tate buffer with the ICP-MS. Digestion involved boiling of
the sample in 1:2 HCl:HNO3 until near evaporation, cooling,
addition of hydrogen peroxide, and evaporation to ~1 mL.
This was repeated twice and followed by 20-fold dilution
with UHP water. The instrument was calibrated using certi-
fied 63Cu, 56Fe, 59Co, 58Ni, 60Ni, 51V, 52Cr, 55Mn, and 64Zn
standards. The total concentration of each metal was
obtained by scaling the measured isotope by its terrestrial
abundance. A quantity of 1 ppb of 115In was added to each
sample as an internal standard to eliminate sample matrix
effects and to account for variations in background readings.
The phase purity of each metal fluoride was verified by a
commercial X-ray diffractometer (XRD, Rigaku, The Wood-
lands, TX). Two-band differential luminescence thermometry
(TBDLT)39,40 was used to measure local laser-induced cool-
ing or heating in ZBLANI:Yb3+ samples.
Precipitated metal fluorides were dried and fluorinated in
HF gas at 300°C using a custom apparatus adapted from
Burkhalter et al.37 and Kramer et al.38 The fluorides were
contained in glassy-carbon boats, inserted into a glassy-
carbon-lined Inconel Alloy 600 steel reaction tube, and placed
in a horizontal tube furnace. HF gas was generated by KHF2
decomposition in a separate vertical tube furnace. 25.6 g of
HF gas was produced in 19 h by decomposing 100 g of previ-
ously dried KHF2 (dried at 190°C for >60 h in vacuum) via
heating to 390°C at 135 K/h, to 440°C at 4.5 K/h, and to
480°C at 8.3 K/h. A mass-flow controller (Aalborg, Orange-
burg, NY) maintained a constant flow rate of semiconductor
grade argon (Ar) to provide an estimated 7 vol% average HF
concentration. Ar/HF exhaust gas was bubbled through a cal-
cium hydroxide solution to remove HF by precipitation of
CaF2.
(4) Glass Formation and Sample Preparation
Table II summarizes the nominal compositions of the
ZBLANI:Yb3+ glasses prepared in this study. The dried
metal fluorides were combined, and glass melting occurred in
a covered glassy-carbon crucible inside an argon dry box
which was controlled to <0.1 ppm of O2 and <0.6 ppm H2O
and was equipped with HEPA filters. A practical limit for
the ZBLANI:Yb3+ melting temperature is set by the volatili-
zation rate of ZrF4, which sublimes at 600°C. To minimize
sublimation of ZrF4 from the batch of precursor fluoride
powders before melting occurred, the batch was inserted into
a preheated (750°C) furnace to effect rapid melting. The
batch was melted for 5 h and cooled to 550°C at 10 K/min
before the melt was cast into a platinum mold at room
temperature. The glass was annealed by heating to just below
the glass transition temperature of 250°C at 2 K/min, hold-
ing for 1 h, and cooling to room temperature at 0.2 K/min.
Sample 1 was fabricated from commercial metal fluoride
precursors that first required drying and fluorination in hot
HF gas to reduce oxide impurities that otherwise caused
significant crystallization. The annealed glass was cut into a
rectangular parallelepiped with an ethylene glycol-cooled dia-
III. Results and Discussion
(1) Binary Metal Fluoride Synthesis Strategy
Table I summarizes the starting materials used for the prepa-
ration of ZBLANI:Yb3+ glass by a melting process. The
commercial binary fluorides ZrF4, BaF2, LaF3, AlF3, NaF,
InF3, and YbF3 have substantial transition-metal and oxy-
gen-based contaminations. A ZBLANI:Yb3+ glass produced
from these precursors would contain several thousands of
ppm of transition-metal contamination, which is unaccept-
able in comparison to the target of ~100 ppb. As shown in
Table I, metal oxide, carbonate, chloride, and elemental pre-
cursors on the other hand are commercially available in sig-
nificantly higher purity and can serve as precursors for the
synthesis of binary metal fluorides. But without further puri-
Table II. Summary of the Yb3+-Doped ZBLANI Samples Synthesized and Characterized in this Study.
The Typical Batch Size was 4 g
Sample
No.
Yb3+
ZBLANI composition
(mol%)
(mol%)
Description and preparation notes
1
1
54–21–3.5–3.5–16.5–0.5
Produced from commercial metal fluoride precursors without further purification.
Some bulk scattering.
2
3
1
1
54–21–3.5–3.5–16–1
53–20–3–3–17.5–2.5
First generation CASE process. Some bulk scattering.
Second generation CASE process. Improved melting/casting scheme.
Excellent optical quality.
4
1
53–20–3–3–17.5–2.5
Third generation CASE process.
Lower HF gas drying temperature. Excellent optical quality.
5
6
7
2
1
2
53–20–2–3–17.5–2.5
53–20–3–3–17.5–2.5
Unknown
Same as sample 4 but with 2% Yb3+ doping.
Same as sample 4 but ZrF4 purified by sublimation rather than solvent extraction.
Commercial ZBLAN:Yb3+ sample from IPG Photonics.

September 2011
Preparation of High-Purity Metal Fluorides
2899
Table III. Transition-Metal Concentration (ppb) in a ZrCl2O Solution Before, During, and After Purification by CASE.
These Data are also Illustrated in Fig. 1.
Cu
Fe
Co
Ni
V
Cr
Mn
Zn
Total
Before CASE
0
0
0
0
0
51600
711
177
7130
143
0
32.8
5.07
0.695
1.10
0
2140
93.9
17.5
18.6
0.147
3820
7640
454
0
33.9
0
72540
After 1st CASE step
After 2nd CASE step
After 3rd CASE step
Ultra-high purity (UHP) H2O
3.09
0.078
0
26.4
15.1
2.56
1436
116
115
83.0
58.5
0
0
0
0
0
0.147
fication, even these precursors would result in a ZBLANI:
Yb3+ with several tens of ppm of transition-metal contami-
nation, still exceeding the impurity target range by 100–1000
fold. Clearly, additional purification is required for a sensi-
tive application such as solid-state laser cooling.
Our strategy for preparing ultra-pure metal fluorides con-
sists of (1) removing transition-metal impurities from an
aqueous phase by CASE, (2) precipitation of the metal fluo-
ride with hydrofluoric acid, (3) removing residual oxygen-
based impurities (such as metal oxides, oxyfluorides, OHÀ,
and water) by drying and fluorination in hot hydrogen fluo-
ride (HF) gas, followed by (4) synthesizing the glass in a
water and oxygen-free clean environment. The aqueous phase
in the CASE process is favorable because it enables the use
of high quality starting materials such as metal oxides, car-
bonates, chlorides, and metals that can be dissolved in acid
or water. But aqueous processing implies the challenge of
subsequently removing residual water and oxides from pre-
cipitated “wet” binary metal fluorides. Therefore, the second
step of drying and fluorination in hot HF gas is essential as
it enables the preceding aqueous CASE process. Together,
CASE and drying/fluorination in HF gas constitute a general
purification method that is applicable to the synthesis of a
wide range of metal fluorides.
(2) Effectiveness of Chelate-Assisted Solvent Extraction
The effectiveness of removing transition-metal impurities
from the aqueous metal-ion solutions was studied quantita-
tively by ICP-MS. The set of transition metals Cu, Fe, Co,
Ni, V, Cr, Mn, and Zn was chosen based on the expected
detrimental effects of some of these impurities on the effi-
ciency of Yb3+-doped laser-cooling materials.15 The ICP-MS
study was carried out for zirconium because ZrF4 amounts
to more than half of the final ZBLANI glass composition
(see Table II). The UHP water, used extensively in the CASE
process, was also analyzed for transition-metal impurities,
both to verify its purity and to establish a point of reference
for the ICP-MS measurements. None of the eight different
transition-metal ions was detected in the UHP water within
the detection limits (low ppt) of the ICP-MS. We also ana-
lyzed the doubly distilled acids, the buffer solution, and the
hydrogen peroxide (used in ICP-MS sample preparation).
Transition-metal concentrations in these reagents were found
to be negligible, indicating that any significant transition-
metal impurities detected in the zirconium sample solutions
were introduced by the ZrCl2O∙8H2O starting material.
The results of the ICP-MS study are summarized in
Table III and illustrated in Fig. 1, which shows the concen-
tration of each of the eight transition-metal impurities (on a
log-scale, in ppb) at various stages in the CASE purification
of the zirconium solution. No copper was found in any of
the samples, and it is omitted from Fig. 1. The “Before
CASE” sample represents the filtered and buffered aqueous
ZrCl2O solution taken directly before the first CASE step.
This initial ZrCl2O solution was found to be substantially
contaminated, having a total of ~72500 ppb of the eight dif-
ferent transition metals probed by the ICP-MS, of which
71% was Fe. The remaining three samples shown in Fig. 1
were taken from the aqueous phase after each of the three
Fig. 1. Effectiveness of chelate-assisted solvent extraction (CASE)
in purifying a ZrCl2O solution as measured by ICP-MS (see Table
III). The solution was analyzed before CASE, and after each of the
three successive CASE steps (see Section II). No Cu was detected in
any of the solutions within the detection limit of the ICP-MS (low
ppt).
successive solvent extraction steps. The concentration of
most of the eight transition metals decreased for each solvent
extraction step. The first CASE step reduced the total con-
centration of the eight transition metals from ~72500 to
~1400 ppb, with the residual impurities consisting primarily
of not only Fe, but also Zn and Cr. The second CASE step
further reduced the total impurity concentration from ~1400
to ~100 ppb. The third CASE step did not reduce impurity
levels any further and appeared to be ineffective. The extrac-
tion coefficients (percentage of impurity removed per extrac-
tion step) for the first CASE step were found to be 99.3%
(Mn), 98.7% (Fe), 98.3% (Co), 98.0% (Ni), 95.6% (Cr),
94.1% (Zn), and 84.5% (V). These values are consistent with
earlier reports of solvent extraction using the APDC chelate
in a water/MIBK two-phase system.31,32,34 These extraction
coefficients also illustrate that at least two extraction steps
are needed to reduce the initial concentration by 1000-fold.
The extraction coefficients were generally lower for the sec-
ond and third extraction steps (where applicable), which may
be due to recontamination of the solution partially offsetting

2900
Journal of the American Ceramic Society—Patterson et al.
Vol. 94, No. 9
the effect of the chelate at impurity concentrations <50 ppb.
This is seen for Zn, Cr, and V (Table III), for which a slight
increase in the concentration after the third extraction step
was observed.
High-purity hydrofluoric acid was added to the aqueous
solution after the third CASE step to precipitate (NH4)3ZrF6
which was then thermally decomposed to ZrF4·H2O [see Sec-
tion II(2)]. The residual transition-metal impurities in the
final zirconium solution (see Fig. 1) are therefore assumed to
be representative of the purity in the final fluoride ZrF4·H2O
before HF gas drying/fluorination. The measured impurity
level of ~100 ppb compares favorably with the range of 10–
100 ppb required for efficient optical refrigerator materials.15
(3) Drying and Fluorination in HF Gas
Purification of an aqueous phase by CASE is effective in
reducing transition metals; however, it yields metal fluorides
with residual oxygen-based impurities which can be equally
detrimental in many applications, including solid-state
optical refrigeration. Oxygen-based impurities can increase
multi-phonon relaxation, increase the tendency of glass to
crystallize, increase light scattering, and decrease chemical
durability.41 Treatment of metal fluorides in hot HF gas can
remove oxygen impurities by two mechanisms: (1) surface
water and water of crystallization evaporate from the fluo-
rides or from oxyfluorides and oxides as they are heated and
(2) remaining oxides, hydroxides, and oxyfluorides, which are
thermodynamically less stable than the respective fluoride,
react with HF to form the metal fluoride by releasing water
into the gas phase.
Samples 1–3 (see Table II) were made from fluorides
which were processed in HF gas at 550°C.37,38 We found that
metal fluorides exposed to HF at this temperature consis-
tently had a grayish tint, and a brownish gelatinous residue
emerged from the output port of the drying vessel during the
HF process. These residues were likely a result of corrosion
of the Inconel Alloy 600 parts and the glassy-carbon liner of
the apparatus. We have found that lowering the drying tem-
perature to 300°C eliminated these corrosion residues and
resulted in white metal fluoride powder. This lower HF pro-
cess temperature is also suggested by the results of Kwon
et al., which have shown that finely powdered CeO2, Nd2O3,
and SrO can be quantitatively converted to the respective flu-
orides by exposure to 33 vol% HF in argon at 300°C in less
than 1 h.42 We, therefore expect the drying/fluorination in
the apparatus described herein [19 h at 300°C in ~7 vol%
HF in argon; see Section II(3)] to be highly effective in
removing oxygen-based impurities.
Fig. 2. Solid-state laser-cooling performance of the samples
prepared in this study (Table III). The TBDLT parameter, ϑ,
(measured at 300 K) is proportional to the laser-induced change in
internal sample temperature: a positive and negative ϑ corresponds
to laser-induced heating and cooling, respectively. Samples 2–4
exhibited laser cooling, indicating that the CASE purification in
combination with drying and fluorination in hot HF gas were
successful in significantly reducing transition-metal and oxygen-based
impurities.
insufficient purity to enable laser cooling; this sample showed
substantial laser-induced heating (ϑ > 0) even at room tem-
perature. The use of CASE purification combined with dry-
ing and fluorination in hot HF gas provided a substantial
improvement of the laser-cooling performance. Samples 2, 3,
and 4 represent various stages of the development of the
CASE/HF process, and they all achieved laser-induced cool-
ing (ϑ > 0). Note that sample 4, which was prepared from
metal fluorides that were treated in HF gas at 300°C instead
of 550°C [see Section III(3)], had the best laser-cooling per-
formance of all samples prepared in this study. The Yb3+
concentration was increased from 1 to 2 mol% in sample 5,
which resulted in substantial laser-induced heating. The rate
of energy migration among Yb3+ ions increases with increas-
ing Yb3+ concentration, and some of the excitations can find
impurity sites where non-radiative relaxation takes place.
This process is more efficient in the 2% Yb3+-doped sample
5 compared to the 1% Yb3+-doped sample 4. The fact that
sample 5 showed substantial heating is direct evidence that
some transition-metal and/or oxygen-based impurities are
still present in the material. Also note that the use of ZrF4
purified by sublimation (provided by the University of Bern;
sample 6) resulted in a glass that cooled, albeit not as much
as the best sample fabricated from precursors purified by the
CASE process. This shows that the CASE/HF process is
more effective in removing impurities from ZrF4 compared
to sublimation. Sample 7 is a commercial ZBLAN:Yb3+
fluoride glass that was obtained from IPG Photonics. It was
selected as the exceptional top performer from several dozen
nominally identical commercial ZBLAN:Yb3+ batches that
varied widely in their laser-cooling performance (both heat-
ing and cooling were observed) and, by inference, aggregate
impurity concentration. This substantial batch-to-batch per-
formance variation in commercial ZBLAN:Yb3+ samples
illustrates the sensitivity of the laser-cooling efficiency to
impurities. Sample 4 showed the best laser-cooling perfor-
mance of the samples produced by the CASE/HF process in
this study. Whereas it did not outperform the best commer-
cial sample 6, we note that the CASE/HF process yielded
consistent batch-to-batch performance. Therefore, further
optimization of the baseline process presented herein is
expected to result in materials of both superior and consis-
tent performance.
(4) Effect on Solid-State Laser-Cooling Performance
The laser-cooling performance can be measured by two-band
differential luminescence thermometry (TBDLT), a non-con-
tact technique that monitors the luminescence from a sample
and deduces laser-induced internal temperature changes from
changes in the intensity distribution of the luminescence spec-
trum as the pump laser is cycled on and off.39,40 In practice,
the Yb3+ luminescence is split into two spectral bands. As
the pump laser is turned on, the relative band intensity
changes due to a change in the local temperature (laser-
induced cooling or heating), and it approaches a new equilib-
rium value according to a power law with exponent ϑ.40 This
TBDLT parameter ϑ is a measure of the laser-induced inter-
nal temperature change, where ϑ > 0 and ϑ > 0 indicate
internal heating and cooling, respectively. Therefore, ϑ can
be used as an indirect measure of the aggregate impurity con-
centration.
Figure 2 is a quantitative demonstration of the enhance-
ment of laser-cooling performance by applying the purifica-
tion methods described in this study. Sample 1 clearly
illustrates that even the best commercial metal fluorides have

September 2011
Preparation of High-Purity Metal Fluorides
2901
14S. Mitachi, T. Miyashita, and T. Manabe, “Preparation of Fluoride Opti-
cal Fibres for Transmission in the Mid-Infrared,” Phys. Chem. Glasses, 23,
196–201 (1982).
IV. Conclusions
A comprehensive purification method suited for all of the
ZBLANI:Yb3+ glass constituents was demonstrated for the
first time. We showed that CASE using APDC as a chelate
and MIBK/water as a two-phase system is effective in remov-
ing problematic transition-metal impurities from a ZrCl2O
solution. The transition-metal contamination was reduced by
almost 1000-fold, and a residual transition-metal concentra-
tion of ~100 ppb was achieved. Subsequent drying and fluori-
nation of the resulting metal fluorides in hot hydrogen
fluoride gas proved effective in removing residual oxygen-
based impurities, as confirmed by laser-cooling measure-
ments. This work is focused on optimization of CASE with
the APDC chelate, finding other chelates/solvent systems
with improved characteristics, and further developing the
process of drying and fluorination in HF gas. The suite of
processes presented herein is applicable to the preparation of
a wide range of ultra-pure binary fluorides, and it establishes
a foundation for the growth of a variety of fluoride crystals
and glasses for demanding photonic applications.
15M. P. Hehlen, R. I. Epstein, and H. Inoue, “Model of Laser Cooling in the
Yb3+-Doped Fluorozirconate Glass ZBLAN,” Phys. Rev. B, 75, 144302 (2007).
16E. O. Gbogi, K.-H. Chung, and C. T. Moynihan, “Surface and Bulk OH-
Infrared Absorption in ZrF4- and HfF4-Based Glasses,” J. Am. Ceram. Soc.,
64 C-51–3 (1981).
17P. W. France, S. F. Carter, J. R. Williams, and K. J. Beales, “OH-Abs-
orption in Fluoride Glass Infrared Fibres,” Electron. Lett., 20, 607–8 (1984).
18L. J. B. Vacha, C. T. Moynihan, B. B. Harbison, K. Cadien, R. Moss-
adegh, and P. C. Schultz, “Hermetic Coatings for Bulk Fluoride Glasses and
Fibres,” Int. J. Opt. Sensors, 2, 297–304 (1987).
19H. Poignant, pp. 35–56 in Halide Glasses for Infrared Fiber Optics, Edited
by R. M. Almeida. Nijhoff, Amsterdam, 1987.
20P. Kaiser, A. R. Tynes, H. W. Astle, A. D. Pearson, W. G. French, R. E.
Jaeger, and A. H. Cherin, “Spectral Losses of Unclad Vitreous Silica and
Soda-Lime-Silicate Fibers,” J. Opt. Soc. Am., 63, 1141–8 (1973).
+
21P. W. France, S. F. Carter, and J. R. Williams, “NH4 Absorption in
Fluoride Glass Infrared Fibers,” J. Am. Ceram. Soc., 67, C243–4 (1984).
22M. Poulain and M. Saad, “Absorption Loss Due to Complex Anions in
Fluorozirconate Glasses,” J. Lightwave. Tech., 2, 599–602 (1984).
23J. M. Jewell, J. Coon, and J. E. Shelby, “The Extinction Coefficient for
CO2 Dissolved in a Heavy-Metal Fluoride Glass,” Mat. Sci. Forum., 32–33,
421–6 (1988).
24D. R. MacFarlane, P. J. Newman, A. Voelkel, and E. Snitzer, “Methods
of Purification of Zirconium Tetrafluoride for Fluorozirconate Glass,” J. Am.
Ceram. Soc., 85, 1610–2 (2002).
25B. E. Kinsman and R. Hanney, “Preparation and Purification of Metal
Fluorides for Crystals and Glasses,” Adv. Materials for Opt. and Elect., 5,
109–15 (1995).
Acknowledgments
We thank Dr. Karl Kramer at the Department of Chemistry and Biochemis-
try, University of Bern, Switzerland, for his assistance with the design of the
hydrogen fluoride gas drying apparatus and for providing the sublimated ZrF4
used for the synthesis of Sample 6. We gratefully acknowledge the support of
the Air Force Office of Scientific Research under the Multidisciplinary Univer-
sity Research Initiative (MURI) program.
26A. M. Mailhot, A. Elyamani, and R. E. Riman, “Reactive Atmosphere
Synthesis of Sol-Gel Heavy Metal Fluoride Glasses,” J. Mater. Res., 7,
1534–40 (1992).
27C. T. Moynihan, M. G. Drexhage, B. Bendow, M. Saleh Boulos, K. P.
Quinlan, K. H. Chung, and E. Gbogi, “Composition Dependence of Infrared
Edge Absorption in ZrF4 and HfF4 Based Glasses,” Mater. Res. Bull., 16,
25–30 (1981).
28M. Jardin, J. Guery, and C. Jacoboni, “Preparation in Vapour State
of Fluoride Glass Components by a Chemical Vapour Deposition Process,”
J. Non-Cryst. Solids, 184, 204–8 (1995).
References
1T. M. Bloomstein, M. W. Horn, M. Rothschild, R. R. Kunz, S. T. Palmacci,
and R. B. Goodman, “Lithography with 157 nm Lasers,” J. Vac. Sci. Technol.
B, 15, 2112–6 (1997).
29Y. Nishida, K. Fujiura, H. Sato, S. Sugawara, K. Kobayashi, and S. Ta-
kahashi, “Preparation of ZBLAN Fluoride Glass Particles by Chemical Vapor
Deposition Process,” Jpn. J. Appl. Phys. Part 2, 31, L1692–4 (1992).
30K. Fujiura, Y. Ohishi, M. Fujiki, T. Kanamori, and S. Takahashi,
“Process for the Preparation of Fluoride Glass and Process for the Prepara-
tion of Optical Fiber Preform Using the Fluoride Glass,” U.S. Patent, 5 [071]
460 (1991).
2S. Sudo, “Progress in Optical Fiber Amplifiers”; pp. 19–21 in Current
Trends in Optical Amplifiers and their Applications, Edited by T. P. Lee. World
Scientific, NJ, 1996.
3T. G. Brown, “Optical Fibers and Fiber-Optic Communications”; pp. 1–49
in Fiber Optics Handbook: Fiber, Devices and Systems for Optical Communica-
tions, Edited by M. Bass and E. W. van Stryland. McGraw-Hill, New York,
2002.
31H. Malissa and E. Schoffmann, “Uber die Verwendung von substituierten
Dithiocarbamaten in der Mikroanalyse,” Mikrochim. Acta, 1, 187–202 (1955).
32M. Arnac and G. Verboom, “Solubility Product Constants of Some Diva-
lent Metal Ions with Ammonium Pyrrolidine Dithiocarbamate,” Anal. Chem.,
46, 2059–61 (1974).
4S. Bedo, M. Pollnau, W. Luthy, and H. P. Weber, “Saturation of the
2.71 lm Laser Output in Erbium-Doped ZBLAN Fibers,” Opt. Commun.,
116, 81–6 (1995).
5X. Zhu and R. Jain, “Numerical Analysis and Experimental Results of
High-Power Er/Pr:ZBLAN 2.7 lm Fiber Lasers with Different Pumping
Designs,” Appl. Optics., 45, 7118–25 (2006).
33D. Bertrand, J. Guery, and C. Jacoboni, “Fe,Co,Ni,Cu Trace Metal Anal-
ysis in ZBLAN Fluoride Glasses,” J. Non-Cryst. Solids, 161, 32–5 (1993).
34R. R. Brooks, M. Hoashi, S. M. Wilson, and R. Q. Zhang, “Extraction
into Methyl Isobutyl Ketone of Metal Complexes with Ammonium Pyrrolidine
Dithio-Carbamate Formed in Strongly Acidic Media,” Anal. Chim. Acta, 217,
165–70 (1989).
6T. Sakamoto, M. Shimizu, T. Kanamori, Y. Terunuma, Y. Ohishi, M.
Yamada, and S. Sudo, “1.4-lm-Band Gain Characteristics of a Tm-Ho-Doped
ZBLYAN Fiber Amplifier Pumped in the 0.8-lm Band,” IEEE Photonics
Technol. Lett., 7, 983–5 (1995).
35J. D. Kinrade and J. C. van Loon, “Solvent Extraction for Use with
Flame Atomic Absorption Spectrometry,” Anal. Chem., 46, 1894–8 (1974).
36Z. Ling, Z. Chengshan, D. Gaoxian, and W. Kangkang, “ZrOCl2 for
Fluoride Glass Preparation,” J. Non-Cryst. Solids, 140, 331–4 (1992).
37R. Burkhalter, I. Dohnke, and J. Hulliger, “Growing of Bulk Crystals and
Structuring Waveguides of Fluoride Materials for Laser Applications,” Prog.
Cryst. Growth Charact. Mater., 42, 1–64 (2001).
7B. Pedersen, W. J. Miniscalco, and R. S. Quimby, “Optimization of Pr3+
-
ZBLAN Fiber Amplifiers,” IEEE Photonics Technol. Lett., 4 446–8 (1992).
8D. F. Anderson, “Cerium Fluoride – A Scintillator for High-Rate Applica-
tions,” Nucl. Instrum. Meth. Phys. Res. A, 287, 606–12 (1990).
9E. Auffray, D. Bouttet, I. Dafinei, J. Fay, P. Lecoq, J. A. Mares, M.
Martini, G. Maze, F. Meinardi, and B. Moine, “Cerium Doped Heavy Metal
Fluoride Glasses, a Possible Alternative for Electromagnetic Calorimetry,”
Instrum. Meth. Phys. Res. A, 380, 524–36 (1996).
38K. W. Kramer, D. Biner, G. Frei, H. U. Gudel, M. P. Hehlen, and S. R.
Luthi, “Hexagonal Sodium Yttrium Fluoride Based Green and Blue Emitting
Upconversion Phosphors,” Chem. Mater., 16, 1244–51 (2004).
39W. M. Patterson, D. V. Seletskiy, M. Sheik-Bahae, R. I. Epstein, and M. P.
Hehlen, “Measurement of Solid-State Optical Refrigeration by Two-Band Dif-
ferential Luminescence Thermometry,” J. Opt. Soc. Am. B, 27, 611–8 (2010).
40W. M. Patterson, M. Sheik-Bahae, R. I. Epstein, and M. P. Hehlen,
“Model of Laser-Induced Temperature Changes in Solid-State Optical Refrig-
erators,” J. Appl. Phys., 107, 063108 (2010).
10J. Freek Suijver, “Upconversion Phosphors”; pp. 133–75 in Luminescence:
From Theory to Applications, Edited by C. Ronda. Wiley-VCH, Weinheim,
2008.
11R. I. Epstein and M. Sheik-Bahae, “Optical Refrigeration in Solids: Fun-
damentals and Overview”; pp. 1–32 in Optical Refrigeration. Science and
Applications of Laser Cooling of Solids, Edited by R. I. Epstein and M. Sheik-
Bahae. Wiley, Weinheim, 2009.
12W. G. Rellergert, D. DeMille, R. R. Greco, M. P. Hehlen, J. R. Torger-
son, and E. R. Hudson, “Constraining the Evolution of the Fundamental
Constants with a Solid-State Optical Frequency Reference Based on the
41M. Poulain, J. Lucas, and P. Brun, “Fluorated Glass from Zirconium Tet-
rafluoride – Optical Properties of a Doped Glass in Nd3+,” Mat. Res. Bull.,
10, 243–6 (1975).
Th-229 Nucleus,” Phys. Rev. Lett., 104, 200802 (2010).
42S. W. Kwon, E. H. Kim, B. G. Ahn, J. H. Yoo, and H. G. Ahn, “Fluori-
nation of Metals and Metal Oxides by Gas-Solid Reaction,” J. Ind. Eng.
13Y. Ohishi, S. Mitachi, T. Kanamori, and T. Manabe, “Optical Absorption
of 3d Transition-Metal and Rare-Earth Elements in Zirconium Fluoride
Glasses,” Phys. Chem. Glasses, 24, 135–40 (1983).
Chem., 8, 477–82 (2002).
h