Investigation of electrochemical reduction of GeO2 to Ge in molten CaCl2-NaCl

Article abstract of DOI:10.1016/j.electacta.2014.09.107

Electrochemical reduction of solid GeO₂ has been investigated in the mixed CaCl₂-NaCl melt at 1023 K for developing a more efficient process for preparation of Ge. Cyclic voltammetry and potentiostatic electrolysis were applied to study the GeO₂-loaded metallic cavity electrode. In addition, porous GeO₂ pellets were reduced by potentiostatic and constant cell voltage electrolysis with a graphite anode, and the electrolysis products were analyzed by powder X-ray diffraction, scanning electron microscopy and energy-dispersive X-ray spectrometry, focusing on understanding the reduction mechanism and the impact of electrode potential on the product purity. It was found that the reduction of GeO₂ to Ge occurred at a potential of about -0.50 V (vs. Ag/Ag⁺), but generating various calcium germanates simultaneously, whose reduction was a little more difficult and needed a potential more negative than -1.00 V. However, if the cathode potential exceeded -1.60 V, Ca (or Na) - Ge intermetallic compounds might form. These results gave an appropriate potential range between -1.10 and -1.40 V for the production of pure germanium. Rapid electrolysis of GeO₂ to pure Ge has been realized at a cell voltage of 2.5 V with a current efficiency of about 92%.

Full text of DOI:10.1016/j.electacta.2014.09.107

Electrochimica Acta 147 (2014) 352–359
Contents lists available at ScienceDirect
Electrochimica Acta
journal homepage: www.elsevier.com/locate/electacta
Investigation of electrochemical reduction of GeO2 to Ge in molten
CaCl2-NaCl
Liangbin Rong a, Rui He a, Zhiyong Wang a, Junjun Peng a, Xianbo Jin a,
,
*
George Z. Chen a,b
a
College of Chemistry and Molecular Sciences, Wuhan University, Wuhan, 430072, PR China
b
Department of Chemical and Environmental Engineering, and Energy and Sustainability Research Division, Faculty of Engineering, University of Nottingham,
Nottingham, NG7 2RD, UK
A R T I C L E I N F O
A B S T R A C T
Article history:
Received 18 May 2014
Received in revised form 17 September 2014
Accepted 22 September 2014
Available online 28 September 2014
Electrochemical reduction of solid GeO2 has been investigated in the mixed CaCl2-NaCl melt at 1023 K for
developing a more efficient process for preparation of Ge. Cyclic voltammetry and potentiostatic
electrolysis were applied to study the GeO2-loaded metallic cavity electrode. In addition, porous GeO2
pellets were reduced by potentiostatic and constant cell voltage electrolysis with a graphite anode, and
the electrolysis products were analyzed by powder X-ray diffraction, scanning electron microscopy and
energy-dispersive X-ray spectrometry, focusing on understanding the reduction mechanism and the
impact of electrode potential on the product purity. It was found that the reduction of GeO2 to Ge
occurred at a potential of about -0.50 V (vs. Ag/Ag+), but generating various calcium germanates
simultaneously, whose reduction was a little more difficult and needed a potential more negative than
-1.00 V. However, if the cathode potential exceeded -1.60 V, Ca (or Na) - Ge intermetallic compounds
might form. These results gave an appropriate potential range between -1.10 and -1.40 V for the
production of pure germanium. Rapid electrolysis of GeO2 to pure Ge has been realized at a cell voltage of
2.5 V with a current efficiency of about 92%.
Keywords:
Solid state reduction
Calcium germanates
GeO2
Ge
Molten salt
ã 2014 Elsevier Ltd. All rights reserved.
1. Introduction
Electro-reduction of solid oxides on the cathode in molten salts
at temperatures below 1173 K has attracted extensive attention in
Germanium (Ge) has attracted much attention for its electronic
and optoelectronic applications as semiconductors and infrared
optical components because of its larger excitonic Bohr radius and
higher electron and hole mobility than Si.[1–3] Besides, the
development of preparative techniques today allow more flexibili-
ty in manufacturing nanostructured germanium and its alloys
[4–7], which are applied in negative electrode materials for
lithium-ion batteries, optics fibers for long-distance communica-
tion applications, solar cells, catalysts and so on [8–11].
The conventional processes for preparation of Ge involve using
the tannic acid to extract germanium concentrate from the
minerals and then thermally reducing the obtained germanium
oxides or sulfides to the pure metal by hydrogen or carbon.
However, the volatile intermediate GeO or GeS species in these
thermal processes might lead to elemental loss and environmental
issues [12].
production of metals and alloys, and some progresses have been
reviewed recently [13–15]. In particular, preparation of Si from
SiO2 by has been reported, and nanostructured Si can be
successfully achieved [16–24]. Because Ge and Si have similar
properties, the same technique may be utilized to obtain Ge
nanowires [25]. However, the kinetics of the electro-reduction
process and methods for elimination of some or all possible
impurities should be likewise investigated.
The purpose of this work is to illustrate the mechanism of
electro-reduction of solid GeO2 to Ge in mixed CaCl2-NaCl melt at
1023 K. Cyclic voltammetry and potentiostatic electrolysis as well
as constant cell voltage electrolysis of the GeO2 loaded metallic
(Mo) cavity electrode (MCE) and GeO2 pellets were implemented,
intending to elaborate the influence of the cathode potential on
purity of the Ge product.
2. Experimental
Anhydrous CaCl2 (600 g, purity:>96 wt%, content of Mg and
* Corresponding author. Tel.: +86 27 68756319; fax: +86 27 68756319.
E-mail address: xbjin@whu.edu.cn (X. Jin).
alkali metals: ꢀ 0.3 wt%) and NaCl (291 g) were mixed at
http://dx.doi.org/10.1016/j.electacta.2014.09.107
0013-4686/ã 2014 Elsevier Ltd. All rights reserved.

L. Rong et al. / Electrochimica Acta 147 (2014) 352–359
353
equi-molar ratio and dried at 573 K before melting at 1023 K in a
graphite crucible placed in a stainless steel reactor under argon
protection. Pre-electrolysis was performed at 2.6 V for 10 h to
remove redox-active impurities (such as Mg and Zn) in the molten
salt between a nickel foil cathode and a graphite rod anode.
Cyclic voltammograms (CVs) and potentiostatic electrolysis of
solid GeO2 in mixed CaCl2-NaCl melt were recorded using the
metallic (Mo) cavity electrode (MCE) and a computer-assisted
IVIUM Electrochemical System (Ivium Technologies BV Company).
A small amount of GeO2 power (>99.999%, Nonfemet, Zhongjin
Lingnan Shaoguan Smelter) was manually loaded into the double-
cavity of the MCE (the inset in Fig. 1), which was then inserted into
the molten salt as working electrode. The graphite counter
electrode and an Ag/Ag+ reference electrode (10 mol% AgCl,
45 mol% NaCl and 45% mol% KCl, sealed in a ceramic tube) were
engaged in the electrochemical experiments.
-0.80 V can be regarded as reduction coupled with the anodic peak
A0. While other cathodic and anodic peaks remained distinct, CVs
within different potential ranges were recorded to recognize the
redox pairs. As Fig. 1B indicates definitely C1 and A1 to be redox
pair, a comparison between Fig. 1A and C suggests that C2/A2, C3/
A3 and C4/A4 are redox couples respectively.
Thermodynamic calculations were carried out to help under-
stand the reduction of GeO2,
GeO2 = Ge + O2 (g) (
D
G = 382.9 kJ/mol at 1023 K)
(1)
CaO = 2Ca + O2 (g) (DG = 1056.4 kJ - mol at 1023 K)
(2)
According to Reactions (1) and (2), the theoretical decomposi-
tion voltages of GeO2 and CaO at 1023 K are 0.992 and 2.737 V
respectively. The difference between the two suggesting that the
reduction of GeO2 to Ge could occur at a potential at least 1.745 V
more positive than that of Ca formation, which happens at
potentials around -2.30 V as shown in Fig. 1A. Thus, peak C0 in
Fig. 1A with an onset potential of -0.50 V is most likely due to the
reduction of GeO2 to Ge.
Both peak currents of C1 and A1 (shown in Fig. 1A) decreased
along with the increasing the number of cycles. Based on previous
studies [26], it was deemed that oxygen ionization showed
irreversibility on continuously recorded CVs due to the irreversible
migration of oxygen ions from the solid cathode to the molten salt.
Also, in a single cycle the cathodic current always exceeded the
anodic one, indicating the irrecoverable loss of oxygen ions which
are needed for the re-oxidation. Therefore, C1 can be ascribed to
another oxygen ionization process in addition to C0. Ge has a
divalent oxide GeO which however could not be the intermediate
product for C1 taking the below thermodynamic calculation into
account,
For potentiostatic electrolysis and two-electrode constant cell
voltage electrolysis, the GeO2 powder was pressed (8 MPa) into
cylindrical porous pellets (ꢀ13.3 mm in diameter, ca.1.4 mm in
thickness, 0.6 g in mass, ꢀ 49% in porosity), which was then dried
and sandwiched between two nickel foam sheets to form
assembled cathodes. After being electrolyzed for 4 h at several
potentials (-0.80
ꢀ
-1.70 V) or constant cell voltage (2.5 V)
controlled by a high accuracy battery tester (Shenzhen Neware
Electronic Ltd. China), the pellets were lifted and cooled above the
molten salt before being washed in distilled water and 0.5 mol/L
HCl. The fine collected products were then dried and characterized
by an X-ray diffraction (SHIMADZU Lab XRD-6000 with Cu Ka1,
scanning rate of 6ꢁ/min), scanning electron microscopy (FEI Sirion
Field Emission Gun SEM system) together with energy-dispersive
X-ray analysis (EDX), and transmission electron microscopy (JEM-
2100, JEOL Ltd.). The oxygen content was determined by inert gas
fusion oxygen analysis (RO-416DR, LECO, USA).
2Geo = 2Ge + O2(g) (
D
G = 270.5 kJ/mol at 1023 K)
(3)
3. Results and discussion
It can be found from Reactions (1) and (3) that the reduction of
GeO would happen at a potential 291 mV more positive than GeO2,
thus, GeO should be electrochemically unstable at the reduction
potentials of GeO2 (-0.50 V or more negative). Since its onset
potential was as negative as -1.05 V, peak C1 was suggested to be
the reduction of calcium germanates as will be discussed later in
detail.
Redox couples C2/A2 and C3/A3 were highly reversible in
comparison with C1/A1. Their cathodic and anodic currents
increased successively in the initial cycles. These reduction
currents were unlikely to be due to the oxygen ionization
processes. It was thus thought that all the GeO2 could have been
reduced to metallic Ge in the potential region of C1. To verify this
Fig. 1A presents the consecutive CVs of MCE-GeO2 electrode in
the mixed CaCl2-NaCl melt at 1023 K. The CV of the blank MCE
(dash line) revealed that the electro-reduction of Ca2+/Na+ and re-
dissolution of Na/Ca took place at potentials around -2.30 V near
the negative scan limit. Thermodynamically, the formation of Na
and Ca could occur at adjacent potentials considering the
theoretical decomposition voltages of CaCl2 and NaCl are
3.326 and 3.284 V at 1023 K respectively. Compared with the
blank, the CVs of MCE-GeO2 exhibited mainly five reduction steps
in the first cycle, showing an onset potential of around -0.50 V
(solid line). In the subsequent cycles, the broad reduction peaks
C0 disappeared. The small but notable peaks at potential about
Fig. 1. (A) Cyclic voltammogram (CV) of the blank MCE (dashed line) and consecutive CVs of the GeO2 powder loaded MCE. The inset image shows the metallic (Mo) cavity
electrode (MCE). (B) and (C), the second cycle CV of the GeO2 powder loaded MCE in ranges of -0.3 ꢀ -1.35 and -0.3 ꢀ -1.95 V respectively. All CVs were recorded in the mixed
CaCl2-NaCl melt at 1023 K and 50 mV .
sÀ1
Á

354
L. Rong et al. / Electrochimica Acta 147 (2014) 352–359
assumption, potentiostatic electrolysis of MCE-GeO2 electrodes
was carried out at -1.40 V (vs. Ag/Ag+) for several periods of time
and the electro-reduction products in the cavities were investi-
gated by SEM and EDX after washing in distilled water.
As shown in Fig. 2A, the original particles of the loaded GeO2 in
the MCE were about 10 mm in sizes with aggregated granules. After
Typical current-time plots are shown in Fig. 4A. At -0.80 V (before
peak C1), it can be seen that the GeO2 underwent a small cathodic
current (40ꢀ300 mA) passed through the oxide pellet with a
reduction charge of about 1000 C after taking away the back-
ground. This corresponds to only partial reduction of the 0.6 g
pellet because the theoretical charge needed for the complete
reduction is about 2210 C. However, peaks of GeO2 disappeared on
the XRD pattern (Fig. 4B), which showed a mixture of Ge and
calcium germanates (CaGeO3, CaGe4O9, and Ca3GeO5). The SEM
image of the -0.80 V product after washing in water is presented in
Fig. 5A, which reveals nanofibers and nanoparticles similar to
those in Fig. 2B and 2 C. The existence of calcium germanates in the
-0.80 V product was confirmed by the inserted EDX spectrum in
Fig. 5A. Since calcium germanates can dissolve into hydrochloric
acid but germanium will not, the -0.80 V product was further
washed in the 0.5 mol/L HCl solution. It was found that most of the
nanofibers were dissolved, and the remaining products were
mainly Ge particles as revealed by the SEM and EDX analyses
shown in Fig. 5B.
At -1.1 V, the initial reduction current increased to about 3 A
(Fig. 4A), and then leveled off in about 20 min, suggesting a very
fast step. Then the current decreased slowly in the rest of the
electrolysis time, and the reduction charge reached about 2200 C in
the 4-h electrolysis. Both the XRD (Fig. 4B) and EDX analyses
(Fig. 5C) suggested that all the electrolytic products at potentials
ranging from -1.10 to -1.40 V were pure Ge, indicating that the GeO2
and calcium germanates could be reduced to Ge at these potentials.
In addition, during washing these samples in water, no gas bubbles
were observed, indicating no Ca-Ge or Na-Ge compounds formed
at these potentials. The Ge samples were principally composed of
Ge nanoparticles with primary particle sizes of about 100 nm
(Fig. 5C). It should be pointed out that the morphology of the
only 5 s electrolysis at -1.40 V, the morphology changed greatly to a
mixture of nanoparticles and nanofibers. The oxygen content
deceased to about 4.9 wt. %, suggesting removal of about 88% of O.
The nanofibers in the 15 s electrolytic product became wider and
shorter, and the oxygen content was about 2.9 wt. %. EDX analysis
showed the existence of Ca element in these products. Formation
of calcium metallates happened quite often during the electrolysis
of many other metal oxides [14]. These nanofibers disappeared in
the 60 s product, which was confirmed to be pure Ge consisting of
predominantly nanoparticles by the SEM and EDX analyses. These
results confirmed that peak C1 was the reduction of any
intermediates to Ge.
Peaks C2 to C4 were then suggested to be the formation of the
Ca (or Na)-Ge intermetallic compounds. The successive increase of
their currents in the early potential cycles could be linked to the
gradual reduction of GeO2 to Ge until completion. As can be seen
from the Ca-Ge and Na-Ge binary phase diagrams [27] (Fig. 3),
many compounds such as Ge4Na, GeNa, GeNa3, Ge2Ca, GeCa, GeCa2
might form during the cathodic polarization of Ge. Thus, proper
reduction potentials of GeO2 should be determined to avoid the
formation of these intermetallic compounds, which will be focused
in the following discussion.
According to the above analyses and discussion of the CVs of the
MCE-GeO2, potentiostatic electrolysis of GeO2 pellets were
conducted at several potentials ranging from -0.80 to -1.70 V for
4 h to further understand the steps in the reduction of GeO2.
Fig. 2. The SEM images of the products from potentiostatic electrolysis of the GeO2 powder loaded MCE in the CaCl2-NaCl melt at -1.4 V (vs. Ag/Ag+) and 1023 K for different
times: (A) 0 s; (B) 5 s; (C) 15 s; (D) 60 s. The insets show the respective EDX spectrums.

L. Rong et al. / Electrochimica Acta 147 (2014) 352–359
355
Fig. 3. The Na-Ge (A) and Ca-Ge (B) binary phase diagrams [27].
metallic Ge generated in this work at 1023 K was in great
differences from that prepared by electrolysis of GeO2 at 873 K
in the equi-molar CaCl2-NaCl melt, which was reported to be
consisting of mainly nanowires [25]. Similarly, depending on the
electrolysis conditions, either nanoparticles or nanowires of Si
could be prepared by electrolysis of solid SiO2 in the CaCl2 based
melts [16–24]. It was found that Si nanowires would become
dominant in the product upon electrolysis of porous pellet of SiO2
particularly with nanoparticles of the oxide [22], addition of
nanoparticles of Au as catalyst [23], and/or application of a low
overpotential for slow reduction [24]. While other conditions were
likely very similar, it may be speculated that the morphology
difference in the Ge products from these two temperatures was
mainly due to the difference in the reduction speed of the solid
cathode (4 h for 0.6 g GeO2 in this work vs. ꢀ12 h at 873 K for 0.5 g
GeO2 in the previous work [25]). Generally, a slower reduction
speed is more favorable for formation of fewer nuclei and oriented
growth of crystals.
determined. As shown in Fig. 4A, the initial reduction current at
-1.7 V was much greater than that at -1.1 V, obviously due to the
increased overpotential. However, after that, another current
plateau between 20 and 90 min could be observed. At the same
time, the total cathodic charge increased to about 3700 C, which
was far beyond that needed theoretically for the reduction of GeO2
to Ge. As revealed by XRD (Fig. 4B), the direct electrolysis product
at -1.70 V before any washing contained mainly Ca0.33Ge and some
salts from the CaCl2-NaCl melt. Some CaO as the existing form of
the generated O2À ions in the melt could also be found [29]. The
Ca0.33Ge phase matched up with the eutectoid containing 76 at. %
of Ge in the Ge-Ca phase diagram (Fig. 3B). During washing in
water, reaction between the Ca0.33Ge and water occurred as
evidenced by the generation of gas bubbles, leading to an
amorphous product containing Ca, Ge and O as revealed by the
SEM and EDX analyses in Fig. 5D. These results suggested again
that peak C2 on the CVs in Fig. 1 corresponded to the formation of
Ca-Ge intermetallic compounds, thus the potential for the electro-
reduction of GeO2 to pure Ge should be controlled in a range
between -1.00 V and -1.40 V.
Ge and Si are both in the C group (or group 14) of the Periodical
Table, and can form various intermetallic compounds with Ca and
Na. An optimized potential range for the preparation of pure Si has
been suggested [28]. To prepare pure Ge from GeO2 with minimum
contamination of Ca (or Na), the upper limit potential needs to be
According to the above discussion, the mechanism of the
reduction of GeO2 to Ge during electrolysis of the solid GeO2 in the
CaCl2-NaCl melt was suggested as below,
Fig. 4. (A) Typical current-time curves recorded during the potentiostatic electrolysis of GeO2 pellets (ca. 0.6 g) in the CaCl2-NaCl melt at the indicated potentials (vs. Ag/Ag+)
and 1023 K for 4 h; (B) XRD patterns of GeO2 and the 4-hour electrolysis products obtained at the indicated potentials. All the samples were washed in water if not specified.

356
L. Rong et al. / Electrochimica Acta 147 (2014) 352–359
Fig. 5. SEM images of the products from the potentiostatic electrolysis of porous GeO2 pellets in the CaCl2-NaCl melt for 4 h at 1023 K and different potentials (vs. Ag/Ag+): (A)
and (B), -0.80 V; (C), -1.40 V; (D) -1.70 V. And the SEM (E) and TEM (F) images of the electrolysis product of a GeO2 pellet at cell voltage of 2.5 V for 4 h. The insets show the
corresponding EDX spectrums. The samples were washed in distilled water except for (B), which was observed after washing in HCl solution.
GeO2 + 4e- = Ge + 2O2-
(M1)
(M2a)
(M2b)
(M3)
precipitation upon saturation (Reaction M2b). It is worth noting
thatformationofCa-M-O (M= the studiedmetal, suchasTi, Zr, Hf,Cr,
Ta, Nb) metallates during electrolysis of various metal oxides
have been reported previously [31–39]. One suggested mechanism
of formation of the composite oxide was the electrochemical
insertion of Ca2+ into the metal oxide, usually resulting in a partial
reduction in case that the valence of the metal element (such as Ti)
may vary continuously [31]. Since the valence of Ge in the calcium
germanates is basically +4, the reaction between the unreduced
GeO2, the O2À generated from reaction M1 and Ca2+ from the melt
was suggested to be responsible for the formation of calcium
germanates in this work (Reaction M2), which was supported by the
following thermodynamic calculation taking CaGeO3 as an example,
xCa2+ + xO2- + yGeO2 = (CaO)x(GeO2)y
xCaO + yGeO2 = (CaO)x(GeO2)y
CaO)x(GeO2)y + 4ye- = Ge + xCa2+ + (x+2y)O2-
The O2À ions generated from Reaction M1 and M3 could be
dissociative if in low concentration (Reaction M2a) or CaO

L. Rong et al. / Electrochimica Acta 147 (2014) 352–359
357
the reaction between the GeO2 and the CaO impurity due to the
hydrolysis of CaCl2. Slight dissolution of GeO2 could not be
excluded, however, the solubility of GeO2 in the melt should be
very low considering that the pellet maintained its original shape
perfectly after the 2-hour immersion.
It is claimed previously that no experimental proof of formation
of germanates as the intermediate products during electrolysis of
the GeO2 pellets in the CaCl2-NaCl melt at the 873 K [25]. However,
some small but remarkable XRD peaks of the electrolysis products
that might be evidences of calcium germanates have been
overlooked, and no elemental analysis of the partially reduced
samples has been carried out. In this work, it was found that the
XRD peaks of the calcium germanates were much weaker than
those of GeO2 and Ge, although their content could be significant
according to the EDX analysis.
Constant cell voltage electrolysis of GeO2 pellets to mimic a
traditional two-electrolysis process was also studied at a cell
voltage of 2.5 V. This voltage was selected according to the above
results of CV and potentiostatic electrolysis considering that it
should be significantly lower than the decomposition voltage of
the melt (about 3.3 V) to avoid the formation of Ca-Ge or Na-Ge
compounds, but should be sufficiently high to ensure a relatively
fast electrolysis by overcoming the possible ohmic and concentra-
tion polarizations in the assembled cell. It should be pointed out
that besides the electrolysis current, the cell voltage should be also
highly dependent on the cell configuration. In this study, the 0.6 g
GeO2 pellets (1.33 cm in diameter) were electrolyzed against a
graphite rod (1.8 cm in diameter) with about 3 cm immersed into
the melt, and the distance between the cathode and anode was
about 3 cm.
The typical current-time plot recorded during the electrolysis
process was presented in Fig. 7A. As shown in the enlarged insert,
the current responding to the suddenly imposed 2.5 V voltage
shows charging of double layer at initial, after which the reduction
current climbed then decreased forming a reduction current peak
as commonly observed in previous electrolysis of various solid
oxides [41,42], which has been rationalized by the dynamic metal/
metal oxide/electrolyte three phase electrochemical interline (3PI)
theories [30]. In brief, the electro-reduction took place initially at
the current collector (foam nickel)/metal oxide (GeO2)/electrolyte
(molten salts) three phase interlines. Although the electronic
contact between the Ni and the pellet might be limited, the
reduction generated porous metal, the conjunctive unreduced
oxide and the electrolyte in the pores of the electrode could
Fig. 6. XRD patterns of GeO2 and GeO2+CaO after immersion of their pellets in
CaCl2-NaCl melt at 1023 K for 2 h.
GeO2 + CaO = CaGeO3
(D
G = -82.6 kJ/mol at 1023 K)
(4)
Reaction (4) indicates that the formation of CaGeO3 should be
spontaneous at the studied temperature. However, if combining
Reaction (M1) and (M2a), and also taking CaGeO3 as an example,
there is,
3GeO2 + 2Ca2+ + 4e- = 2CaGeO3 + Ge
(5)
Reaction (5) could indicate another mechanism for the
formation of Ca-M-O composites by redistribution of the O in
the solid state [37–39]. Theoretically, if one considers that the
origin of CaO in Reaction (4) results from the reduction of GeO2 via
Reaction (M1) in molten CaCl2, there is no thermodynamic
difference for CaGeO3 to be formed via either Reaction (4) or
(5), and it is also difficult to distinguish them experimentally.
Nevertheless, the feasibility of Reaction (4) has been verified by a
direct reaction between GeO2 and CaO in the CaCl2-NaCl melt at
1023 K. As shown in Fig. 6, after immersion of a CaO + GeO2 (1:1 in
molar ratio) pellet in the melt for about 2 h, the originally strong
XRD peaks of GeO2 disappeared. Instead, the XRD pattern showed a
mixture of various germanates, in accordance with the multiphase
equilibrium feature of a CaO-GeO2 system [40]. In contrast, after
immersion of a pure GeO2 pellet in the melt, the resultant was still
GeO2 as revealed by XRD analysis shown in Fig. 6. A few amount of
calcium germanates did have formed, which is thought to be due to
Fig. 7. (A) Current-time plot recorded during potentiostatic electrolysis of the GeO2 pellet at a cell voltage of 2.5 V. (B) XRD patterns of electrolytic samples obtained from
different stages of electrolysis.

358
L. Rong et al. / Electrochimica Acta 147 (2014) 352–359
construct the new 3PIs. The transverse expansion of such 3PIs on
the pellet surface caused the initial increase of the cathodic
current, which then decreased due to the increasing ohmic and
mass transfer polarization of the reduction in the depth direction.
The total electrolytic current-time curve of the 2.5 V electrolysis
showed basically two reduction steps (Fig. 7A). The first one was
very quick as reflected by the large current with a maximum of
around 1700 mA, which declined to about 200 mA in about 20 min.
The following platform current with some slight fluctuations
electrolysis of GeO2 that could proceed quickly at -1.1 V, the
dissolved Ca (about 10À12 in activity) would be neglectable
considering that the Ca with an activity of 1 would only form at
about -2.3 V from Fig. 1. In addition, the low cell voltage and quick
electrolysis speed were both helpful to achieve a high current
efficiency. On the other hand, the discharge of O2À ions at the
graphite anode resulted in CO2 emission and some carbonate that
dissolved into the melt. The carbonate ions would move to the
cathode and reduce to C with O2À regenerated, which could also
decrease the current efficiency and cause carbon contamination
to the metallic product [33]. Fortunately, while the relatively low
cell voltage for the GeO2 electrolysis would somewhat suppress
these shuttle reactions, the fact that Ge forms no any alloy with C
would help to alleviate the carbon contamination during
electrolysis and/or remove the carbon from the Ge product later,
and it is worth noting that Ge has been produced by carbothermic
reduction traditionally [12].
lasted about 200 min suggesting
a relatively slower second
reduction step. To make a good understanding, the reduction
products generated at different stages of electrolysis were
characterized by XRD. It can be seen from Fig. 7B that the
electrolysis impelled the gradual decrease of intensity of the XRD
peaks of GeO2, which disappeared in about 30 min. As expected,
accompanying with the formation of Ge, various calcium
germanates (CaGeO3, Ca3GeO5, CaGe2O5 and Ca2Ge7O16) emerged
immediately after imposing the 2.5 V voltage. The subsequent
reduction of germanates were much tardy. Even after 180-min
electrolysis, diffraction peaks of CaGeO3 were still perceivable in
the XRD pattern of the product. This phenomenon was different
from those report on the electrolysis of solid SiO2, where CaSiO3
was seldom found in a partially reduced sample [16–24]. However,
the CaSiO3 was stable only at the very beginning reduction
potential of SiO2, and could be reduced to Si at a slightly larger
overpotential [21,24].
The results from the constant cell voltage electrolysis at 2.5 V
were in well agreement with the above suggested reduction
mechanism (reactions M1 to M3). In addition, it was demonstrated
that the reduction of calcium germanates to Ge may be the rate
determining step for the complete reduction of a GeO2 pellet.
Nevertheless, the 4-hour electrolysis has led to a full reduction as
confirmed by the EDX analysis shown in Fig. 5E, which displayed
almost no any X-ray peak of other element but those of Ge,
suggesting a content less than 0.5 wt. % (the limit of detection of
EDX analysis) of any impurity.
The SEM in Fig. 5E suggests nodule morphology of the
generated Ge powders, consisting of both micrometer and
nanometer particles. The TEM inspection (Fig. 5F) revealed that
the primary particle size of the nanometer particles was around
100 nm. The existence of large particles of Ge was thought to be
due to the sintering of Ge at the electrolysis temperature (1023 K)
considering that the melting point of Ge is only about 1211 K.
The mass of the metallic product from the 0.6 g GeO2 pellet was
about 0.4 g, indicating an elemental recovery of Ge of about 96.8%.
The little loss might be mainly due to the difficulties in collecting
Ge nanoparticles during the processing. It is speculated that unlike
a traditional hydrogen thermo-reduction process where GeO
would form and escape away, this electrolytic method can provide
much stronger driven force for the reduction of any Ge containing
species to Ge, thus higher yield could be achieved. The metallic
product remained stable in air, and the oxygen content in the
sample after shelving for about 6 months was determined by inert
gas fusion oxygen analysis to be about 0.71 wt. %, which might
mainly arise from the surface oxidation of the Ge nanoparticles.
According to the current-time plot shown in Fig. 7a, the actual
charge passed for the reduction of 0.6 g GeO2 to Ge (requires
ꢀ2210 C theoretically) during the 4-h electrolysis was about
2400 C, indicating a current efficiency of about 92%. This current
efficiency was significantly higher than those reported for the
electrolysis of other refractory metals, such as TiO2 (15ꢀ32% for
dense pellets and 54% for TiO2 pellets with an optimized porosity)
[43,44]. It has been recognized that the loss of current efficiency
during the electrolysis could be mainly attributed to the
generation of dissolved Ca at the cathode and the electronic
conductivity of the molten salts [45,46]. However, for the
7. Conclusions
The direct electro-reduction of solid GeO2 in the equimolar
CaCl2-NaCl melt was studied at 1023 K and pure Ge can be
prepared at a potential range between -1.10 V and -1.40 V (vs. Ag/
Ag+). Cyclic voltammetry, potentiostatic and constant cell voltage
electrolysis together with XRD, EDX analyses and SEM observation
suggested that the mechanism for the reduction of solid GeO2
includes (1) electrochemical reduction of GeO2 to Ge; (2) chemical
formation of calcium germanates ((CaO)x(GeO2)y) and (3) electro-
chemical reduction of (CaO)x(GeO2)y to Ge. Although the first step
can start at a potential (-0.5 V vs. Ag/Ag+) about 1.8 V more positive
than that of cathodic decomposition of the electrolyte, the
reduction released O2À together with the Ca2+ from the electrolyte
react immediately with remaining GeO2, generating the germa-
nates whose reduction occurs at potentials more negative than
-1.0 V with a relatively slow speed. An increased polarization is
expected to speed up the reduction of the germanates. However,
potentials exceeding -1.60 V will lead to the formation of Ca-Ge or
Na-Ge intermetallic compounds. Based on these understandings,
rapid electrolysis of GeO2 to pure Ge has been carried out at a cell
voltage of 2.5 V. A current efficiency as high as 92% and an
elemental recovery of about 96.8% were achieved. The electrolytic
Ge exhibits a mixture of micrometer particles and sub-micrometer
nodular particles with the primary particle sizes of around 100 nm.
These findings promise a new approach for the metallurgy of GeO2,
as well as preparation of nanometer Ge powders.
Acknowledgements
This work is supported by NSFCs (21173161, 20973130), the
863 project (2009AA03Z503), the MOE Program (NCET-11-0397),
the Fundamental Research Funds for the Central Universities, and
the Large-scale Instrument and Equipment Sharing Foundation of
Wuhan University.
References
[1] J. Andzane, N. Petkov, A.I. Livshits, J.J. Boland, J.D. Holmes, D. Erts, Two-
Terminal Nanoelectromechanical Devices Based on Germanium Nanowires,
Nano Lett. 9 (2009) 1824–1829.
[2] L. Li, X. Fang, H.G. Chew, F. Zheng, T.H. Liew, X. Xu, Y. Zhang, S. Pan, G. Li, L.
Zhang, Crystallinity-Controlled Germanium Nanowire Arrays: Potential Field
Emitters, Adv. Funct. Mater. 18 (2008) 1080–1088.
[3] R.P. Prasankumar, S. Choi, S.A. Trugman, S.T. Picraux, A.J. Taylor, Ultrafast
Electron and Hole Dynamics in Germanium Nanowires, Nano Lett. 8 (2008)
1619–1624.
[4] H.C. Wu, T.C. Hou, Y.L. Chueh, L.J. Chen, H.T. Chiu, C.Y. Lee, One-Dimensional
Germanium Nanostructures-Formation and Their Electron Field Emission
Properties, Nanotechnology 21 (2010) 455601.

L. Rong et al. / Electrochimica Acta 147 (2014) 352–359
359
[5] V.Y. Aleshkin, A.A. Dubinov, Direct Band Ge and Ge/InGaAs Quantum Wells in
GaAs, J. Appl. Phys. 109 (123) (2011) 107.
[27] H. Landolt, R. Börnstein, Phase Equilibria Crystallographic and Thermody-
namic Data of Binary Alloys, Springer-Verlag, 1993.
[6] A. Alkhatib, A. Nayfeh, A Complete Physical Germanium-on-Silicon Quantum
[28] W. Xiao, X. Jin, Y. Deng, D. Wang, G.Z. Chen, Rationalisation and Optimisation of
Solid State Electro-Reduction of SiO2 to Si in Molten CaCl2 in Accordance with
Dynamic Three-Phase Interlines Based Voltammetry, J. Electroanal. Chem. 639
(2010) 130–140.
[29] R. Bhagat, D. Dye, S.L. Raghunathan, R.J. Talling, D. Inman, B.K. Jackson, K.K. Rao,
R.J. Dashwood, In Situ Synchrotron Diffraction of the Electrochemical
Reduction Pathway of TiO2, Acta Mater. 58 (2010) 5057–5062.
[30] W. Xiao, X.B. Jin, Y. Deng, D.H. Wang, X.H. Hu, G.Z. Chen, Electrochemically
Driven Three-Phase Interlines into Insulator Compounds: Electroreduction of
Solid SiO2 in Molten CaCl2, Chem. Phys. Chem. 7 (2006) 1750–1758.
[31] K. Jiang, X.H. Hu, M. Ma, D.H. Wang, G.H. Qiu, X.B. Jin, G.Z. Chen,
Perovskitization-assisted Electrochemical Reduction of Solid TiO2 in Molten
CaCl2, Angew. Chem. Int. Ed. 45 (2006) 428–432.
Dot Self-Assembly Process, Sci. Rep. 3 (2013) 2099.
[7] S.K. Ray, R.K. Das, S. Manna, A. Dhar, Structural and Optical Properties of
Germanium Nanostructures on Si(100) and Embedded in High-K Oxides,
Nanoscale Res. Lett. 6 (2011) 224.
[8] R. Pillarisetty, Academic and Industry Research Progress in Germanium
Nanodevices, Nature 479 (2011) 324–328.
[9] D.D. Vaughn 2nd, R.E. Schaak, Synthesis, Properties and Applications of
Colloidal Germanium and Germanium-Based Nanomaterials, Chem. Soc. Rev.
42 (2013) 2861–2879.
[10] M.H. Park, Y. Cho, K. Kim, J. Kim, M. Liu, J. Cho, Germanium Nanotubes Prepared
by Using the Kirkendall Effect as Anodes for High-Rate Lithium Batteries,
Angew. Chem. Int. Ed. Engl. 50 (2011) 9647–9650.
[11] V. Sorianello, L. Colace, C.M. Ragliano, D. Fulgoni, L. Nash, G. Assanto,
Germanium-on-Glass Solar Cells Fabrication and Characterization, Opt. Mater.
Express 3 (2013) 216–228.
[32] Q. Song, Q. Xu, X. Kang, J. Du, Z. Xi, Mechanistic Insight of Electrochemical
Reduction of Ta2O5 to Tantalum in a Eutectic CaCl2-NaCl Molten Salt, J. Alloys
Compd. 490 (2010) 241–246.
[12] R.R. Moskalyk, Review of Germanium Processing Worldwide, Miner. Eng. 17
(2004) 393–402.
[33] T. Wu, X.B. Jin, W. Xiao, X.H. Hu, D.H. Wang, G.Z. Chen, Thin Pellets: Fast
Electrochemical Preparation of Capacitor Tantalum Powder, Chem. Mater. 19
(2007) 153–160.
[34] E. Gordo, G.Z. Chen, D.J. Fray, Toward Optimisation of Electrolytic Reduction of
Solid Chromium Oxide to Chromium Powder in Molten Chloride Salts,
Electrochim. Acta 49 (2004) 2195–2208.
[35] D.T.L. Alexander, C. Schwandt, D.J. Fray, Microstructural Kinetics of Phase
Transformations During Electrochemical Reduction of Titanium Dioxide in
Molten Calcium Chloride, Acta Mater. 54 (2006) 2933–2944.
[36] A.M. Abdelkader, D.J. Fray, Electro-deoxidation of Hafnium Dioxide and
Niobia-doped Hafnium Dioxide in Molten Calcium Chloride, Electrochim. Acta
64 (2012) 10–16.
[13] G.Z. Chen, D.J. Fray, T.W. Farthing, Direct electrochemical reduction of titanium
dioxide to titanium in molten calcium chloride, Nature 407 (2000) 361–364.
[14] A.M. Abdelkader, K. Tripuraneni Kilby, A. Cox, D.J. Fray, DC Voltammetry of
Rlectro-deoxidation of Solid Oxides, Chem. Rev. 113 (2013) 2863–2886.
[15] W. Xiao, D. Wang, The Electrochemical Reduction Processes of Solid
Compounds in High Temperature Molten Salts, Chem. Soc. Rev. 43 (2014)
3215–3228.
[16] X. Jin, P. Gao, D. Wang, X. Hu, G.Z. Chen, Electrochemical Preparation of Silicon
and Its Alloys from Solid Oxides in Molten Calcium Chloride, Angew. Chem. Int.
Ed. 43 (2004) 733–736.
[17] T. Nohira, K. Yasuda, Y. Ito, Pinpoint and Bulk Electrochemical Reduction of
Insulating Silicon Dioxide to Silicon, Nat. Mater. 2 (2003) 397–401.
[37] K.S. Mohandas, D.J. Fray, Electrochemical Deoxidation of Solid Zirconium
Dioxide in Molten Calcium Chloride, Metall. Mater. Trans. B 40B (2009) 685–
699.
[38] C. Schwandta, D.T.L. Alexander, D.J. Fray, The Electro-deoxidation of Porous
Titanium Dioxide Precursors in Molten Calcium Chloride under Cathodic
Potential Control, Electrochim. Acta 54 (2009) 3819–3829.
_
ꢀ
[18] E. Ergül, I. Karakaya, M. Erdogan, Electrochemical Decomposition of SiO2
Pellets to Form Silicon in Molten Salts, J. Alloys Compd. 509 (2011) 899–903.
[19] J. Zhao, J. Li, P. Ying, W. Zhang, L. Meng, C. Li, Facile Synthesis of Freestanding Si
Nanowire Arrays by One-step Template-free Electro-deoxidation of SiO2 in a
Molten Salt, Chem Commun. 49 (2013) 4477–4479.
[39] D. Sri Maha Vishnu, N. Sanil, L. Shakila, G. Panneerselvam, R. Sudha, K.S.
[20] E. Juzeliunas, A. Cox, D.J. Fray, Electro-deoxidation of Thin Silica Layer in
Molten Salt-globular Structures with Effective Light Absorbance, Electrochim.
Acta 68 (2012) 123.
Mohandas, K. Nagarajan, A Study of the Reaction Pathways during
Electrochemical Reduction of Dense Nb2O5 Pellets in molten CaCl2 Medium,
Electrochim. Acta 100 (2013) 51–62.
[21] K. Yasuda, T. Nohira, R. Hagiwara, Y.H. Ogata, Diagrammatic Representation of
Direct Electrolytic Reduction of SiO2 in Molten CaCl2, J. Electrochem. Soc . 154
(2007) E95.
[22] J. Yang, S. Lu, S.X. Kan Zhang, J. Du, Electrochemical Preparation of Silicon
Nanowires from Nanometre Silica in Molten Calcium Chloride, Chem.
Commun. (2009) 3273–3275.
[23] Y. Nishimura, T. Nohira, K. Kobayashi, R. Hagiwara, Formation of Si Nanowires
by Direct Electrolytic Reduction of Porous SiO2 Pellets in Molten CaCl2, J.
Electrochem. Soc. 158 (2011) E55.
[40] M.a. SA3CG, N. Zhang, Germanates and Silicon germanates, Hydrometal-
lurgy of China 1 (1985) 50–56.
[41] X.Y. Yan, D.J. Fray, Electrochemical Studies on Reduction of Solid Nb2O5 in
Molten CaCl2-NaCl Eutectic - I. Factors Affecting Electrodeoxidation of solid
Nb2O5 to Niobium, J. Electrochem. Soc. 152 (2005) D12–D21.
[42] D.J.S. Hyslop, A.M. Abdelkader, A. Cox, D.J. Fray, Electrochemical Synthesis of a
Biomedically Important Co-Cr alloy, Acta Mater. 58 (2010) 3124–3130.
[43] D.T.L. Alexander, C. Schwandt, D.J. Fray, The Electro-deoxidation of Dense
Titanium Dioxide Precursors in Molten Calcium Chloride Giving
a New
[24] W. Xiao, X. Jin, G.Z. Chen, Up-scalable and Controllable Electrolytic Production
of Photo-responsive Nanostructured Silicon, J. Mater. Chem. A 1 (2013) 10243.
[25] H. Yin, W. Xiao, X. Mao, W. Wei, H. Zhu, D. Wang, Template-free Electrosyn-
thesis of Crystalline Germanium Nanowires from Solid Germanium Oxide in
Molten CaCl2-NaCl, Electrochim. Acta 102 (2013) 369–374.
[26] H.P. Gao, X.B. Jin, S.W. Zou, F.Z. Ling, J.J. Peng, Z.Y. Wang, G.Z. Chen, Liquid
Diffusion of the Instantaneously Released Oxygen Ion in the Electrolytic
Porous Fe from Solid Fe2O3 in Molten CaCl2, Electrochim. Acta 107 (2013)
261–268.
Reaction Pathway, Electrochim. Acta 56 (2011) 3286–3295.
[44] W. Li, X.B. Jin, F.L. Huang, G.Z. Chen, Metal-to-oxide Molar Volume Ratio: The
Overlooked Barrier to Solid-state Electroreduction and green Bypass
a
Through Recyclable NH4HCO3, Angew. Chem. Int. Ed. 49 (2010) 3203–3206.
[45] G.M. Haarberg, K.S. Osen, R.J. Heus, J.J. Egan, Electronic Conduction and Eletron
MobilitiesinMoltenNaCl-NaSolutions,J.Electrochem.Soc.137(1990)1777–2781.
[46] U. Stohr, W. Freyland, Intervalence Charge Transfer and Electronic Transport in
Molten Salts Containing Tantalun and Niobium Complexes of Mixed Valency,
Phys. Chem. Chem. Phys. 1 (1999) 4383–4387.