Effect of electrolysis potential on reduction of solid silicon dioxide in molten CaCl2

Article abstract of DOI:10.1016/j.jpcs.2004.06.037

Electrochemical reduction of solid SiO₂ by using a contacting electrode method was investigated in molten CaCl₂ at 1123 K. The samples were prepared by potentiostatic electrolysis at 0.35-1.30 V (vs. Ca ²⁺/Ca) for 1 h. From the results of XRD, SEM and EPMA, it was confirmed that SiO₂ was electrochemically reduced to Si at 1.25 V or more negative potential, which agreed with the thermodynamic calculation. The current-time curves during electrolysis and the cross-sectional SEM measurements of the samples clearly showed that reduction rate is higher at more negative potential. In addition, Si-Ca alloy formation was confirmed at 0.35 V.

Full text of DOI:10.1016/j.jpcs.2004.06.037

Journal of Physics and Chemistry of Solids 66 (2005) 443–447
www.elsevier.com/locate/jpcs
Effect of electrolysis potential on reduction of solid silicon dioxide
in molten CaCl2
*
Kouji Yasuda, Toshiyuki Nohira , Yasuhiko Ito
Department of Fundamental Energy Science, Graduate School of Energy Science, Kyoto University, Sakyo-ku, Kyoto 606-8501, Japan
Accepted 4 June 2004
Abstract
Electrochemical reduction of solid SiO2 by using a contacting electrode method was investigated in molten CaCl2 at 1123 K. The samples
were prepared by potentiostatic electrolysis at 0.35–1.30 V (vs. Ca2C/Ca) for 1 h. From the results of XRD, SEM and EPMA, it was
confirmed that SiO2 was electrochemically reduced to Si at 1.25 V or more negative potential, which agreed with the thermodynamic
calculation. The current–time curves during electrolysis and the cross-sectional SEM measurements of the samples clearly showed that
reduction rate is higher at more negative potential. In addition, Si–Ca alloy formation was confirmed at 0.35 V.
q 2004 Elsevier Ltd. All rights reserved.
Keywords: A. Oxides; C. Electron microscopy; C. X-ray diffraction; D. Phase transitions
1. Introduction
succeeded in taking place the electrochemical reaction by
using the ‘SiO2 contacting electrode’, in which a Mo wire
directly contacts with SiO2. In the previous study, a high
purity quartz plate (15 mm!5 mm!1 mm, total metal
impurities !1 ppm) was successfully reduced to Si with a
considerably high rate. The reduction mechanism was
explained as follows [2]. When the electrode potential of
the Mo wire is more negative than the reduction potential of
SiO2, the corresponding oxygen fugacity at the electrode
becomes low enough to reduce SiO2. Here, the relation
between the electrode potential, E, and the oxygen fugacity,
Silicon is one of the most important technical materials
due to the eminent and ever increasing importance of
electronics in technology and everyday life. Moreover,
clean electricity produced by Si solar cells is highly
expected as new energy that can substitute for fossil energy
in the future. Since the demands for high purity Si,
especially for solar grade Si, are steadily increasing, the
development of inexpensive production method of high
purity Si has significant meaning. By the conventional
carbothermal reduction of SiO2 [1], the produced Si
inevitably includes impurities such as SiC. Therefore, the
produced Si is needed to be purified by distillation after
gasification to silane gases (SiH4, SiHCl3, etc.). Since the
purification procedure spends much energy and cost, it has
been strongly required to develop a new production process
of high purity, especially solar grade, Si without distillation.
Recently, we found that solid SiO2 can be reduced
electrochemically in molten CaCl2 at 1123 K [2]. Although
solid SiO2 is a good insulator (the specific electric resistance
is in the range of 108–109 U cm at 1123 K) [3], we have
fO , is expressed as
2
RT
E Z E0 C
lnðfO1=2=aO2K
Þ
(1)
2
2F
where E0 is the standard potential, F is the Faraday constant
and aO2K is the activity of O2K ion in molten salt. Then, the
oxygen in the SiO2 near the three-phase interface between
Mo wire, SiO2 and molten salt moves to the interface due to
difference in oxygen chemical potential. The electron-
transfer occurs at the Mo wire in the interface. Since the
resistivity of pure Si is considerably low at the experimental
temperature (about 2!10K2 U cm at 1123 K [4]), the
formed Si can serve as a new conducting medium. As
the results, new three-phase interfaces are formed
* Corresponding author. Tel.: C81 75 753 4817; fax: C81 75 753 5906.
E-mail address: nohira@energy.kyoto-u.ac.jp (T. Nohira).
0022-3697/$ - see front matter q 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jpcs.2004.06.037

444
K. Yasuda et al. / Journal of Physics and Chemistry of Solids 66 (2005) 443–447
and the reduction proceeds continuously toward bulk SiO2.
The total reaction is written as
SiO2 C4eKðthrough Mo or SiÞ Z Si C2O2K
(2)
This novel method might open up a new route for the
high purity Si production. When only oxygen can be
removed from high purity SiO2, Si with similar purity is
produced. This method is especially promising for econ-
omical production of solar grade Si, since SiO2 of solar
grade purity is inexpensive [5]. In order to assess the
possibility of this method, it is important to investigate
the reduction rate and the existence of side reactions. The
reduction rate should be closely related to fO at the three-
2
phase interface, which is controlled by E, as seen from the
Eq. (1). For instance, assuming the initial concentration
level of O2K ion in the melt to be 0.01 mol% and using the
reported DG0f ðCaOð1ÞÞ and DGf0ðSiO2ðsÞÞ [6], reduction of
SiO2 is expected to occur at more negative than 1.21 V (vs.
Ca2C/Ca). Furthermore, it was confirmed that Si–Ca alloy
was formed at negative potential region as a side reaction
[2]. In the present study, therefore, potential dependences of
the reaction and the reduction rate of solid SiO2 in molten
CaCl2 at 1123 K were investigated in detail.
2. Experimental
Fig. 1. Photograph of a SiO2 contacting electrode.
Three hundred grams of CaCl2 (reagent grade, Wako
Pure Chemical Co., Ltd) was contained in a glassy carbon
crucible (10 cm depth, 9 cm inner diameter and 0.2 cm wall
thickness, Tokai Carbon Co., Ltd) and was kept under
vacuum at 473 K for 72 h and at 773 K for 24 h to
remove water. All the experiments were performed under a
dry Ar atmosphere in a sealed stainless steel holder kept at
1123 K.
in distilled water. The compositions of the samples were
analyzed by EPMA (E-MAX ENERGY EX-200, Horiba
Corp.).
3. Results and discussion
A SiO2 contacting electrode shown in Fig. 1 was
prepared by equally winding a Mo wire (f 0.2 mm,
99.95%, Nilaco Corp.) as a current lead, about ten times
around a quartz glass plate (15 mm!5 mm!1 mm, total
metal impurities !1 ppm; NP grade, Tosoh Quartz
Corp.). The counter electrode was a graphite rod
(5 mm!5 mm!20 mm, Tokai Carbon Co., Ltd). The
reference electrode was an AgC/Ag electrode, which was
prepared by immersion of an Ag wire (f 1.0 mm, 99.99%,
Nilaco Corp.) in molten CaCl2 containing 2 mol% AgCl
(99.5%, Wako Pure Chemical Co., Ltd) set in a porous
mullite tube (f 6.0 mm, 3Al2O3$2SiO2, HB grade,
Nikkato Corp.). The potential of this electrode was
calibrated with a reference to that of a Ca2C/Ca electrode
prepared by electrodepositing Ca metal on a Mo wire. All
potentials in this report are given with a reference to this
Ca2C/Ca electrode potential.
3.1. Current–time curves
In our previous study [2], the cyclic voltammogram for a
SiO2 contacting electrode suggested that the electrochemi-
cal reduction of SiO2 occurs at more negative potential than
approximately 1.3 V (vs. Ca2C/Ca), and the reduction to Si
was confirmed at 1.1 V or more negative potential by XRD,
SEM and EPMA. In the present study, to investigate the
reduction between 1.1 and 1.3 V, potentiostatic electrolysis
was conducted at (a) 1.30, (b) 1.25, (c) 1.20, and (d) 1.10 V
for 1 h. Since the reduction rate was considerably higher at
more negative potentials [2], potentiostatic electrolysis was
also carried out at (e) 1.00 and (f) 0.70 V. Moreover,
electrolysis was also conducted at (g) 0.35 V, at which
Si–Ca alloy formation has been confirmed [2], to further
investigate the alloy formation reaction.
Samples were analyzed by XRD (Multiflex 2kW, Rigaku
Fig. 2 shows current–time curves during the electrolysis,
where the cathodic current values are directly correspond to
the reduction rates. At 1.30 V, only residual current was
˚
Corp., Cu Ka line, 1.5418 A) and observed by SEM
(S-2600H, Hitachi) after removing a Mo wire and washing

K. Yasuda et al. / Journal of Physics and Chemistry of Solids 66 (2005) 443–447
445
Fig. 2. Current–time curves during the potentiostatic electrolysis of SiO2 contacting electrodes at (a) 1.30, (b) 1.25, (c) 1.20, (d) 1.10, (e) 1.00, (f) 0.70 and
(g) 0.35 V in molten CaCl2 at 1123 K.
observed, suggesting that no reduction occurred. At all
other potentials (0.35–1.25 V), cathodic currents were
observed and were obviously larger at more negative
potentials. For 1.25, 1.20, 1.10 and 1.00 V, cathodic current
increased with the electrolysis time, indicating an enlarge-
ment of the reaction interface. For 0.70 and 0.35 V, the
currents increased in the beginning but decreased after-
wards. This behavior is explained as follows: the current
increases as long as the reaction interface enlarges,
however, after the whole surface is reduced to Si, the
reduction rate becomes low because the reaction proceeds
only the inner direction.
the reaction proceeded considerably inside and the samples
became highly brittle.
Fig. 4 shows the XRD patterns of the samples. The
samples at 0.70–1.20 V were all identified as Si. In the case
of 0.35 V, the sample was identified as CaSi and CaSi2.
Considering that the Si–Ca alloy formation was not
completed for 1 h, it is thought that CaSi exist at the
surface and CaSi2 exists inside of the sample. As expected
from appearance in Fig. 3, no peak was observed for
the samples at 1.25 and 1.30 V. To further analyze of the
latter two samples, SEM measurements were performed at
the contacting point of the Mo wire. As shown in Fig. 5,
the trace of reduction was found for the sample at 1.25 V.
EPMA analysis at the reaction zone also showed that Si was
the only detected element. It was thus confirmed that SiO2 is
reduced to Si at 1.25 V or more negative potential.
These results almost agree with the thermodynamic
consideration in the introduction section that reduction of
SiO2 is expected to occur at more negative than 1.21 V.
The slight difference may be explained by the difference in
the initial concentration level of O2K ion and/or the smaller
activity coefficient of O2K ion than unity.
3.2. Sample analysis
Fig. 3 shows the appearances of the samples after
electrolysis. In the case of 1.30 V, no change was observed.
For the samples at 1.00–1.25 V, original transparency
apparently changed to dark brown or black. Area of the
color change clearly increased as the electrolysis potential
became negative. For the samples at 0.70 and 0.35 V,
Fig. 3. Photographs of the samples obtained by potentiostatic electrolysis at (a) 1.30, (b) 1.25, (c) 1.20, (d) 1.10, (e) 1.00, (f) 0.70 and (g) 0.35 V for 1 h in
molten CaCl2 at 1123 K.

446
K. Yasuda et al. / Journal of Physics and Chemistry of Solids 66 (2005) 443–447
Fig. 4. XRD patterns of the samples obtained by potentiostatic electrolysis at (a) 1.30, (b) 1.25, (c) 1.20, (d) 1.10, (e) 1.00, (f) 0.70 and (g) 0.35 V for 1 h in
molten CaCl2 at 1123 K, and XRD patterns of (A) Si (JCPDS no. 27-1402), (B) CaSi2 (JCPDS no. 01-1276) and (C) CaSi (JCPDS no. 26-0324).
3.3. Reduction rate
aO2K Z10K4 (anion fraction). Since the difference of oxygen
chemical potential between the three-phase interface and the
bulk SiO2 is larger at more negative potential, reduction rate
becomes higher at more negative potential. The observed
reduction rate, e.g., approximately 200 mm for 1 h at 1.00 V,
seems to be sufficiently high for such a preliminary
construction of the SiO2 contacting electrode, consisting
of a dense quartz plate and a simple Mo wire. If powdery or
porous SiO2 are contacted with a fine wire mesh, the
reduction rate will become high enough for a practical
application. Moreover, current efficiency was calculated as
90% for the reduction at 1.00 V, estimating the amount of Si
and SiO2 from the cross-sectional SEM measurement. This
result also indicates the potential availability of the new
reduction method from a viewpoint of energy efficiency.
Concerning the impurities in the produced Si, it has been
To clarify the effect of electrolysis potential on reduction
rate, cross-sections of the samples were observed by SEM.
As shown in Fig. 6, the reduced Si layers were porous for all
cases. For the samples at 1.10 and 1.20 V, the observed
layer was near the contacting point of Mo wire. The
mechanism of porous formation has already been described
in our previous paper [2]. The reduced layer apparently
became thicker as the electrolysis potential being more
negative. These results prove that the reduction rate is
higher at more negative potential in both surface and inner
directions. This is explained by that the fugacity of oxygen
at the three-phase interface is controlled by the electrode
potential as seen from Eq. (1); fO Z8:3!10K34 atm
2
at 1.20 V and fO Z8:4!10K43 atm at 0.70 V, assuming
2
Fig. 5. SEM images of the contacting zone of the samples obtained by potentiostatic electrolysis at (a) 1.30 and (b) 1.25 V for 1 h in molten CaCl2 at 1123 K.

K. Yasuda et al. / Journal of Physics and Chemistry of Solids 66 (2005) 443–447
447
Fig. 6. Cross-sectional SEM images of the samples obtained by potentiostatic electrolysis at (c) 1.20, (d) 1.10, (e) 1.00, (f) 0.70 and (g) 0.35 V for 1 h in molten
CaCl2 at 1123 K.
confirmed that the impurities other than oxygen was below
detection limit of EPMA. To discuss the impurity level of
solar grade Si, further microanalysis by emission spectro-
graphy is being conducted.
Acknowledgements
This research was partially supported by the Ministry of
Education, Culture, Sports, Science and Technology of
Japan, Grant-in-Aid for Scientific Research. The authors
thank Kinki Invention Center for financial support.
4. Conclusion
References
The effect of electrode potential on reduction of solid
SiO2 has been investigated in molten CaCl2 at 1123 K. SiO2
is reduced to Si between 0.70 (vs. Ca2C/Ca) and 1.25 V,
while CaSi and CaSi2 are formed at 0.35 V. The reduction
rate of SiO2 is higher at more negative potential. The
reduction rate, approximately 200 mm for 1 h at 1.00 V, in
spite of using a dense quartz plate. The impurities other than
oxygen for the produced Si are below the detection limit of
EPMA.
[1] W. Zulehner, B. Elvers, S. Hawkins, W. Russey, G. Schulz (Eds.),
Ullmann’s Encyclopedia of Industrial Chemistry vol. A23, fifth ed.,
VCH Verlagsgesellschaft mbH, Weinheim, 1993, pp. 721–748.
[2] T. Nohira, K. Yasuda, Y. Ito, Nat. Mater. 2 (2003) 397–401.
[3] J.K. Srivastava, M. Prasad, J. Electrochem. Soc. 132 (1985) 955–963.
[4] L.C. Burton, A.H. Madjid, Phys. Rev. 185 (1969) 1127–1132.
[5] M. Bessho, K. Umehara, T. Takaura, T. Nishiyama, H. Shingu, J. Min.
Mater. Process. Inst. Jpn 117 (2001) 736–742.
[6] O. Kubaschewski, C.B. Alcock, P.J. Spencer (Eds.), Materials
Thermochemistry, sixth ed., Pergamon, Oxford, 1993.