Electrodeposition of tungsten from ZnCl2-NaCl-KCl-KF-WO3 melt and investigation on tungsten species in the melt

Article abstract of DOI:10.1016/j.electacta.2009.10.021

The electrodeposition of tungsten in ZnCl₂-NaCl-KCl-KF-WO₃ melt at 250 °C was further studied to obtain a thicker deposit. In the ordinary electrolysis at 0.08 V vs. Zn(II)/Zn, the current density decreased from 1.2 mA cm^-2 to 0.3 mA cm^-2 in 6 h. A thickness of the obtained tungsten layer was 2.1 μm and the estimated current efficiency was 93%. A supernatant salt and a bottom salt were sampled after 6 h from the melting and were analyzed by ICP-AES and XRD. It was found that the soluble tungsten species slowly changes to insoluble ones in the melt. The soluble species was suggested to be WO₃F^- anion. One of the insoluble species was confirmed to be ZnWO₄ and the other one was suggested to be K₂WO₂F₄. Electrodeposition was carried out under the same condition as above except for the intermittent addition of WO₃ every 2 h. The current density was kept at the initial value and the thickness was 4.2 μm. The intermittent addition of WO₃ was confirmed to be effective to obtain a thicker tungsten film.

Full text of DOI:10.1016/j.electacta.2009.10.021

Electrochimica Acta 55 (2010) 1278–1281
Contents lists available at ScienceDirect
Electrochimica Acta
journal homepage: www.elsevier.com/locate/electacta
Electrodeposition of tungsten from ZnCl2–NaCl–KCl–KF–WO3 melt and
investigation on tungsten species in the melt
Koji Nittaa,b,∗, Toshiyuki Nohiraa,∗∗,1, Rika Hagiwaraa, Masatoshi Majimab, Shinji Inazawab
a Department of Fundamental Energy Science, Graduate School of Energy Science, Kyoto University, Sakyo-ku, Kyoto 606-8501, Japan
b Electronics & Materials R&D Labs, Sumitomo Electric Industries, Ltd., 1-1-3, Shimaya, Konohana-ku, Osaka 554-0024, Japan
a r t i c l e i n f o
a b s t r a c t
The electrodeposition of tungsten in ZnCl2–NaCl–KCl–KF–WO3 melt at 250 ◦C was further studied to
obtain a thicker deposit. In the ordinary electrolysis at 0.08 V vs. Zn(II)/Zn, the current density decreased
from 1.2 mA cm−2 to 0.3 mA cm−2 in 6 h. A thickness of the obtained tungsten layer was 2.1 ␮m and
the estimated current efficiency was 93%. A supernatant salt and a bottom salt were sampled after 6 h
from the melting and were analyzed by ICP-AES and XRD. It was found that the soluble tungsten species
slowly changes to insoluble ones in the melt. The soluble species was suggested to be WO3F− anion. One
of the insoluble species was confirmed to be ZnWO4 and the other one was suggested to be K2WO2F4.
Electrodeposition was carried out under the same condition as above except for the intermittent addition
of WO3 every 2 h. The current density was kept at the initial value and the thickness was 4.2 ␮m. The
intermittent addition of WO3 was confirmed to be effective to obtain a thicker tungsten film.
© 2009 Elsevier Ltd. All rights reserved.
Article history:
Received 31 May 2009
Received in revised form 7 October 2009
Accepted 8 October 2009
Available online 10 November 2009
Keywords:
Molten salt
ZnCl2–NaCl–KCl
Electrodeposition
Tungsten
Tungsten oxide
1. Introduction
temperature range in which the resist is not deformed. We have
already reported that electrodeposition of tungsten is possible in
Micro-parts for micro-electro-mechanical systems (MEMS) are
becoming increasingly important, and the lithographie galvanofor-
mung abformung (LIGA) process for producing such parts is
attracting attention [1]. The LIGA process is already used in practical
applications [2,3], and holds great promise for the future. The cur-
rent LIGA process, however, requires electroforming in an aqueous
solution, and thus the applicable materials are restricted to cop-
per, gold, nickel, Ni–Fe alloys and Ni–Co alloys [4–9]. Nevertheless,
the strength and durability of the materials used in the micro-parts
are expected to be improved, wherein refractory metals (especially
tungsten) have the greatest potential.
Although the electrodeposition of tungsten from an aqueous
solution is extremely difficult, Senderoff and Mellors obtained elec-
trodeposits of refractory metals from alkali metal fluoride melts
at 700–850 ◦C [10,11], and Katagiri et al. successfully electrode-
posited metallic tungsten from ZnBr2–NaBr [12] and ZnCl2–NaCl
[13] at 350–450 ◦C. However, the resist sheet for the LIGA pro-
cess deteriorates at such high temperatures (350 ◦C or more). We
have therefore been investigating tungsten electrodeposition in a
a ZnCl2–NaCl–KCl melt (mp 203 ◦C) at 250 ◦C using WCl4 as the
tungsten ion source, and that the addition of KF to the melt gives
the smoother deposits probably owing to the change of ligand
structure of the tungsten complex ions in the melt [14]. Further-
more, we reported that the usage of WO3 as a tungsten ion source
gives smoother and denser deposits, and that it is possible to coat
conventional LIGA parts with tungsten [15]. In addition to the
higher quality of the deposit, WO3 has another advantage of non-
volatility compared with WCl4 that is volatile at 250 ◦C. However,
the decrease of current during a long-term electrodeposition was
found for the ZnCl2–NaCl–KCl–KF–WO3 melt, which was a major
obstacle in obtaining thicker tungsten deposits. So, in this study,
we further studied the electrodeposition of tungsten and especially
focused on tungsten species in the melt. The mechanism of the
change of ionic species is discussed and a method to maintain the
electrodeposition longer time is proposed.
2. Experimental
2.1. Materials and apparatus
All the chemicals were anhydrous reagent grade. ZnCl2 (99.9%,
Wako Pure Chemical Industries, Ltd.) was dried in a furnace under
vacuum at 130 ◦C for three days or more. NaCl, KCl, KF (99% each,
Wako Pure Chemical Industries, Ltd.) and WO3 (99.5%, Wako Pure
Chemical Industries, Ltd.) were dried in a furnace under vacuum
∗
Corresponding author. Tel.: +81 6 6466 5637; fax: +81 6 6466 5683.
Corresponding author. Tel.: +81 75 753 4817; fax: +81 75 753 5906.
E-mail addresses: nitta-koji@sei.co.jp (K. Nitta), nohira@energy.kyoto-u.ac.jp
∗∗
(T. Nohira).
1
ISE member.
0013-4686/$ – see front matter © 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.electacta.2009.10.021

K. Nitta et al. / Electrochimica Acta 55 (2010) 1278–1281
1279
at 400 ◦C for three days or more. The chemicals were thoroughly
mixed (ZnCl2:NaCl:KCl:KF:WO3 = 0.6:0.2:0.2:0.04:0.0054, in mole
fraction) in an alumina crucible on a heating plate in a glove box
filled with argon. The salts were melted at 250 ◦C. The working elec-
trode was a nickel plate (99.9%, 10 × 10 × 0.2 mm, Nilaco Corp.).
The nickel plate was electrochemically polished in sulfuric acid
beforehand. The counter electrode was a glassy carbon rod (3 mm
in diameter, GC-20, Tokai Carbon Co., Ltd.). A zinc wire (99.99%,
0.5 mm in diameter, Nilaco Corp.) immersed in the melt was used
as the reference electrode. A chromel–alumel thermocouple was
used for the temperature measurement.
2.2. Electrolysis and evaluation of electrodeposits
The cyclic voltammetry and potentiostatic electrolysis were car-
ried out using an electrochemical measurement system (Hokuto
Denko Co., Ltd., HSV-100) in a glove box filled with argon. A
schematic drawing of experimental apparatus has been shown
elsewhere [14]. After the electrolysis, electrodes were immersed
in distilled water to remove adherent salts. In order to evaluate the
thickness of the film obtained, the cross-sectional view of deposits
was observed by field emission scanning electron microscopy (FE-
SEM, Hitachi S-800).
Fig. 1. Cyclic voltammograms for nickel electrodes in the ZnCl2–NaCl–KCl–KF–WO3
melt at 250 ◦C. Solid curve: immediately after melting. Dashed curve: after poten-
tiostatic electrolysis at 0.08 V vs. Zn (II)/Zn for 6 h. Scan rate: 0.05 V s−1
.
2.3. Analysis of the salts
The supernatant salt and the bottom salt containing precipitate
were sampled after 6 h had passed from the melting. These sam-
ples were scraped from the cooled and solidified melt. The bottom
salt was a mixture of the precipitate and the solidified melt. The
compositions of Zn, K, Na and W in these samples were measured
by inductively coupled plasma atomic emission spectrometry (ICP-
AES, Thermo Fisher Scientific Inc., iCAP6500 DUO). The composition
of F was measured by ion chromatography (IC, Dionex Corp., ICS-
3000). In order to identify the bottom salt, it was sealed into a
polyethylene bag under argon and analyzed by X-ray diffractome-
try (XRD, Rigaku Industrial Co., Ltd., RINT1500).
3. Results and discussion
3.1. Cyclic voltammetry, potentiostatic electrolysis and analysis
of electrodeposited tungsten
Fig. 2. A change of current density during the potentiostatic electrolysis at 0.08 V
A cyclic voltammogram (CV) was measured immediately after
the salts (ZnCl2–NaCl–KCl–KF–WO3) were melted at 250 ◦C, which
is shown as a solid curve in Fig. 1. Cathodic currents were observed
at more negative than 0.45 V vs. Zn (II)/Zn. According to our
previous study [15], these cathodic currents correspond to the
electrodeposition of tungsten. In the same melt, potentiostatic elec-
trolysis was conducted for a nickel electrode at 0.08 V for 6 h. It has
been clarified that electrodeposition of tungsten occurs without the
codeposition of zinc at this potential [15]. Fig. 2 shows a change of
current density during the electrolysis. In the first five minutes, cur-
rent density decreased rapidly to 1.0 mA cm−2. Then, it increased
slightly from 1.0 mA cm−2 to 1.2 mA cm−2 in 30 min. After that, it
decreasedgraduallyto0.3 mA cm−2 in 6 h. Sincethe decreaseofcur-
rent was observed, a CV was measured again after the potentiostatic
electrolysis, which is shown as a dashed curve in Fig. 1. The current
attributed to electrodeposition of tungsten decreased and the cur-
rent assigned to the nickel–zinc alloy formation was observed at
more negative than 0.05 V [14]. This cathodic current of nickel–zinc
alloy formation was observed in the melt without WO3 [15]. There-
fore, it is considered that the rate of tungsten electrodeposition
became slow, and that the nickel–zinc alloy formation proceeded
before the surface of electrode was covered with tungsten. Fig. 3
vs. Zn(II)/Zn for 6 h in the ZnCl2–NaCl–KCl–KF–WO3 melt at 250 ◦C.
Fig. 3. A cross-sectional SEM image of electrodeposited tungsten by the potentio-
static electrolysis in the ZnCl2–NaCl–KCl–KF–WO3 melt at 0.08 V vs. Zn(II)/Zn for 6 h
at 250 ◦C.

1280
K. Nitta et al. / Electrochimica Acta 55 (2010) 1278–1281
Table 1
Compositions for the designed salt, the supernatant salt and the bottom salt includ-
ing a precipitate. The salt was sampled after 6 h from the melting.
Na
K
Zn
W
F
Designed salt/mol ratio
Supernatant salt/mol ratio
Bottom salt/mol ratio
20.0
19.7
18.5
24.0
23.9
22.2
60.0
60.0
60.0
0.54
0.06
7.2
4.0
3.9
5.2
salt is lower compared with the designed value. On the other hand,
the tungsten content in the bottom salt is higher than the designed
one. It is also seen that the composition of F in the bottom salt
is high. From these results, the following three facts were found:
(i) A soluble tungsten species is generated upon melting. (ii) The
soluble species slowly changes to insoluble tungsten species. (iii)
The insoluble species contains fluorine. It is also confirmed that the
decrease of current during the electrodeposition was caused by the
decrease of soluble tungsten species in the melt.
To investigate the tungsten species in the bottom salt, the sam-
pled salt was analyzed by XRD. The obtained spectrum is shown in
Fig. 5. Peaks at 22.7◦, 23.9◦ and 36.1◦ are assigned to a polyethy-
lene bag that was used to seal the hygroscopic salt. Other peaks are
mostly assigned to ZnWO4, but several peaks remain unidentified.
These unidentified peaks may be due to a tungsten compound of
which crystal structure has not been reported yet. Moreover, no
peaks of ZnCl2, NaCl and KCl are observed, although they are the
main components of the melt. It is likely that the solidified salt
exists in an amorphous phase. In our previous study [16], the ideal
glass transition temperature of the ZnCl2–NaCl–KCl eutectic melt
was determined to be 10 ◦C from VTF plots of viscosity and ionic
conductivity, which is consistent with the result of XRD. Such glass
forming behavior of the melt shoul−d be related−with the formation
of chlorozincate anions like Zn2Cl5 and ZnCl3 [17].
Fig. 4. (a.) The appearance of the ZnCl2–NaCl–KCl–KF–WO3 melt immediately after
melted at 250 ◦C. (b.) The appearance of the ZnCl2–NaCl–KCl–KF–WO3 melt after 6 h
from melting at 250 ◦C (without electrolysis).
shows a cross-sectional SEM image of the electrodeposited tung-
sten. The thickness is approximately 2.1 ␮m. The current efficiency
is calculated to be 93% from the thickness of the deposit, assuming
that the electrodeposition proceeds via a six-electron reduction as
follows.
W(VI) + 6e = W
(1)
Although the accuracy of this value is not very high due to the error
in estimating the thickness, it is reasonable to consider that the
electrodeposition proceeds by a six-electron reduction.
3.3. Discussion on the change of the tungsten species
The reasons for the decrease and the increase of current are dis-
cussed. Firstly, it is considered that the rapid decrease in the first
five minutes is mainly due to the decrease of tungsten species or the
increase of oxide ions in the vicinity of working electrode. The slight
current increase in the next half an hour indicates the increase of
bulk concentration of W(VI) ion. It is assumed that the bulk concen-
tration of W(VI) ion reached the highest value after about 0.5 h from
the melting due to a time lag of WO3 dissolution. On the other hand,
the moderate current decrease in 6 h is presumed to be caused by
the decrease of bulk concentration of W(VI) ion. Since the amount
of W(VI) ion consumed by the electrodeposition is calculated to be
only 0.36% of the total amount of WO3 in the melt, the decrease of
bulk concentration of W(VI) ion was not caused by the electrode-
position. Furthermore, the decrease of W(VI) ions was not caused
by volatilization of tungsten compounds because very little subli-
mate was observed on the top of the vessel and tungsten was not
detected from the sublimate by XPS analysis. Thus, it is considered
that the decrease of W(VI) ions is caused by the transformation of
the electrochemically active tungsten species to inactive ones.
In our previous study [15], it has been found that F− ions are
necessary to dissolve WO3 in the melt. Thus, the soluble tung-
sten species, which is generated immediately after melting, should
contain F− ions. According to the literature [18], many tungsten
compounds containing chlorine or fluorine are volatile at 250 ◦C.
On the other hand, the tungsten species in this system is not volatile
as described in section 3.1. Thus, it is considered that the soluble
tungsten species is WO3F− anion.
WO3 + F− = WO3F−
(2)
3.2. The analysis of the melts by ICP-AES, IC and XRD
The appearance of the melt immediately after melted at 250 ◦C
is shown in Fig. 4-a. The melt was yellow and not transparent. After
6 h without electrolysis, the melt became transparent, as shown in
Fig. 4-b, and a gray precipitate was observed in the bottom of the
crucible. Then, the compositions of Zn, K, Na and W in the bottom
salt and the supernatant salt were analyzed by ICP-AES, and those
of F were analyzed by IC. These results are listed in Table 1. The
data has been standardized so that the composition of Zn becomes
60.0 for all cases. The composition of tungsten in the supernatant
Fig. 5. An XRD spectrum of the bottom salt of the ZnCl2–NaCl–KCl–KF–WO3 melt
after 6 h from melting at 250 ◦C.

K. Nitta et al. / Electrochimica Acta 55 (2010) 1278–1281
1281
Fig. 7. A cross-sectional SEM image of the electrodeposited tungsten by the poten-
tiostatic electrolysis in the ZnCl2–NaCl–KCl–KF–WO3 melt at 0.08 V vs. Zn(II)/Zn for
6 h with the intermitted addition of WO3 every 2 h at 250 ◦C.
Fig. 6. A change of current density during the potentiostatic electrolysis in the
ZnCl2–NaCl–KCl–KF–WO3 melt at 0.08 V vs. Zn(II)/Zn for 6 h with the intermitted
addition of WO3 every 2 h at 250 ◦C.
4. Conclusions
A cause of the current decay was revealed for the electrodepo-
sition of tungsten in the ZnCl2–NaCl–KCl–KF–WO3 melt at 250 ◦C.
It was found that the soluble tungsten species slowly changes to
insoluble ones. The soluble species was suggested to be WO3F−
anion. One of the insoluble species was confirmed to be ZnWO4
and the other one was suggested to be K2WO2F4. The intermittent
addition of WO3 every 2 h was confirmed to be effective to main-
tain the current density at a high level, which enables to obtain a
thicker tungsten film.
Then, a reaction of tungsten electrodeposition is written as
WO3F− + 6e− = W + 3O2− + F−
(3)
On the other hand, it is predicted that WO3F− anions slowly
change into insoluble ZnWO4. Considering that both zinc and oxy-
gen are necessary to transform WO3F− to ZnWO4, and that the
precipitate should also contain fluorine, the following reaction is
suggested.
Zn2Cl5− + 2WO3F− + 2K+ + 2F−
Acknowledgments
= ZnWO4 + K2WO2F4 + ZnCl3− + 2Cl−
(4)
Here, the unidentified peaks in Fig. 5 are considered to cor-
respond to K2WO2F4. The moderate decrease of current during
the potentiostatic electrolysis can be explained by the decrease of
WO3F− anion according to reaction (4).
This study was partly supported by Industrial Technology
Research Grant Program in 2003 from the New Energy and Indus-
trial Technology Development Organization (NEDO) of Japan.
References
3.4. A method to maintain the electrodeposition for a longer
period
[1] Y. Hirata, Nucl. Instrum. Methods B 208 (2003) 21.
[2] T. Haga, K. Okada, J. Yorita, Y. Hirata, S. Shimada, ICEP Proceedings, 2002, p. 421.
[3] Y. Hirata, H. Nakaishi, T. Numazawa, H. Takada, IEEE Ultrasonics Symposium
Proceedings 2, 1997, p. 877.
[4] E.W. Becker, W. Ehrfeld, P. Hagmann, A. Maner, D. Münchmeyer, Microelectron.
Eng. 4 (1986) 35.
[5] A. Rogner, J. Eicher, D. Münchmeyer, R.-P. Peters, J. Mohr, J. Micromech. Micro-
eng. 2 (1992) 133.
[6] J. Hormes, J. Göttert, K. Lian, Y. Desta, L. Jian, Nucl. Instrum. Methods Phys. Res.
B 199 (2003) 332.
From the above discussion, the slow transformation of soluble
tungsten species to insoluble ones was suggested to be the main
reason for the current decrease. However, according to the current
decay curve in Fig. 2, the current density is relatively high in the ini-
tial 2 h. So, it is expected that the electrodeposition proceeds for a
longer period by adding WO3 every 2 h. To confirm this, electrode-
position was conducted under the same condition as Fig. 2 except
for the addition of WO3 every 2 h. The amount of WO3 in every
addition was the same as the initial amount. The current response
is shown in Fig. 6 together with the case of no intermittent addition.
A sharp increase of the current is observed at every addition and the
average current is maintained at the same level as the initial 2 h.
Fig. 7 shows a cross-sectional view of the electrodeposited tung-
sten. The thickness of tungsten layer was approximately 4.2 ␮m and
the current efficiency was calculated to be 90%. So, it is confirmed
that the intermittent addition of WO3 is effective to maintain the
current density at a high level for a long period, which enables us
to obtain a thicker tungsten film.
[7] L. Singleton, J. Photopolym. Sci. Technol. 16 (2003) 413.
[8] S. Sugiyama, Proc. SPIE 5062 (2003) 821.
[9] D. Golodnitsky, N. Gudin, G. Volyanyuk, Plat. Surf. Finish. 2 (1998) 65.
[10] S. Senderoff, G.W. Mellors, Science 153 (1966) 1475.
[11] S. Senderoff, G.W. Mellors, J. Electrochem. Soc. 114 (1967) 586.
[12] A. Katagiri, M. Suzuki, Z. Takehara, J. Electrochem. Soc. 138 (1991) 767.
[13] M. Masuda, H. Takenishi, A. Katagiri, J. Electrochem. Soc. 148 (2001) C59.
[14] H. Nakajima, T. Nohira, R. Hagiwara, Electrochem. Solid-State Lett. 8 (7) (2005)
C91.
[15] H. Nakajima, T. Nohira, R. Hagiwara, K. Nitta, S. Inazawa, K. Okada, Electrochim.
Acta 53 (2007) 24.
[16] K. Nitta, T. Nohira, R. Hagiwara, M. Majima, S. Inazawa, Electrochim. Acta 54
(2009) 4898.
[17] J.J. Watkins, E.C. Ashby, Inorg. Chem. 13 (1974) 2350.
[18] J. Canterford, R. Colton, Halides of the Transition Elements: Halides of the sec-
ond and third Row Transition Metals, vol. 208, Wiley, New York, 1968.