Mechanistic insight into the influence of Al2O3 concentration on the electro-reduction of V2O3 to vanadium in molten Na3AlF6

Article abstract of DOI:10.1016/j.electacta.2018.10.155

The influence of Al₂O₃ concentration on the direct electrochemical reduction of porous vanadium sesquioxide precursors in molten Na₃AlF₆ was investigated by constant voltage electrolysis and cyclic voltammetry. The products obtained by electrolysis at various Al₂O₃ concentrations were examined by XRD and SEM. Specifically, the morphologies of the outer metal layer initially formed were characterized by SEM and EDS, also their effects on the electrolysis process were systematically studied. The results indicate that the Al₂O₃ concentration in molten Na₃AlF₆ has a significant impact on the reduction process. When 0.5 mass% Al₂O₃ is added into molten Na₃AlF₆, the reduction rate of V₂O₃ can be improved because of the appropriate increase of aluminium reductant and relatively high oxide ion transfer rate. However, a high Al₂O₃ concentration (1.0–3.0 mass%) causes the massive precipitation of aluminium at cathodic outer metal layer and further forms Al–V alloy, causing the densification of the outer metal layer and hampering the reduction reaction kinetics.

Full text of DOI:10.1016/j.electacta.2018.10.155

Accepted Manuscript
Mechanistic insight into the influence of Al O concentration on the electro-
2
3
reduction of V O to vanadium in molten Na AlF
6
2
3
3
Yapeng Kong, Jianshe Chen, Binchuan Li, Kuiren Liu, Qing Han
PII:
S0013-4686(18)32395-8
10.1016/j.electacta.2018.10.155
EA 32954
DOI:
Reference:
To appear in:
Electrochimica Acta
Received Date:
Accepted Date:
27 June 2018
25 October 2018
Please cite this article as: Yapeng Kong, Jianshe Chen, Binchuan Li, Kuiren Liu, Qing Han,
Mechanistic insight into the influence of Al O concentration on the electro-reduction of V O to
2
3
2 3
vanadium in molten Na AlF , Electrochimica Acta (2018), doi: 10.1016/j.electacta.2018.10.155
3
6
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form.
Please note that during the production process errors may be discovered which could affect the
content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT
Mechanistic insight into the influence of Al O concentration on
2
3
the electro-reduction of V O to vanadium in molten Na AlF
2
3
3
6
Yapeng Kong,a, b Jianshe Chen,a, b Binchuan Li,a, b Kuiren Liu, , Qing Han,
a, b
a, b, *
a
School of Metallurgy, Northeastern University, Shenyang 110819, P. R. China
b
Key Laboratory for Ecological Utilization of Multimetallic Mineral, Ministry of Education,
Shenyang 110819, P. R. China
Corresponding author: hanq@smm.neu.edu.cn.
Abstract
The influence of Al O concentration on the direct electrochemical reduction of porous
*
2
3
vanadium sesquioxide precursors in molten Na AlF was investigated by constant voltage
3
6
electrolysis and cyclic voltammetry. The products obtained by electrolysis at various Al O
2
3
concentrations were examined by XRD and SEM. Specifically, the morphologies of the outer
metal layer initially formed were characterized by SEM and EDS, also their effects on the
electrolysis process were systematically studied. The results indicate that the Al O
2
3
concentration in molten Na AlF has a significant impact on the reduction process. When 0.5
3
6
mass% Al O is added into molten Na AlF , the reduction rate of V O can be improved
2
3
3
6
2
3
because of the appropriate increase of aluminium reductant and relatively high oxide ion
transfer rate. However, a high Al O concentration (1.0–3.0 mass%) causes the massive
2
3
precipitation of aluminium at cathodic outer metal layer and further forms Al–V alloy,
causing the densification of the outer metal layer and hampering the reduction reaction
kinetics.
Key words: Molten salt electrolysis; Al O concentration; V O ; Vanadium; FFC Cambridge
2
3
2
3
1

ACCEPTED MANUSCRIPT
process.
1
. Introduction
Recently, the FFC (Fray-Farthing-Chen) Cambridge process for producing metal has
triggered worldwide research activities because the metal oxide can be directly reduced to
pure metal. In the process, the oxygen ions of solid oxide can be expelled from cathode and
dissolved in electrolyte then discharged at anode, leaving the pure metal at the cathode [1].
Considerable research efforts have been devoted to the synthesis of various metals and alloys
via the direct electrochemical reduction of oxide precursors. The recent development of the
FFC process has been described in several excellent reviews [2-6]. As so far, this novel
method has been successfully applied to the preparation of, in particular, high-melting-point
metals, such as titanium, chromium, niobium, and tantalum. Meanwhile, it also enables the
successful preparation of many alloys directly from relevant oxide mixtures [7-16], including
Ti–Fe, Ti–Ni, Ni–Cr, Ta–Nb, Ti–Nb–Ta–Zr, Ti–Nb–Ta–Zr–Hf, Nb–Hf–Ti, etc.
There have been attempts to prepare metal vanadium via electrochemical methods, i.e.
electrodeposition [17-18] and electro-deoxidation [19-22]. Some studies have focused on the
direct electro-reduction of vanadium oxides in molten chloride salt. R.O. Suzuki et al. [19]
firstly extended the direct electro-reduction process into the preparation of metal vanadium.
The reduction of solid V O powder was completed at 1173 K in molten CaCl –CaO and fine
2
3
2
metal vanadium powder containing 0.186 mass% oxygen was obtained. They proposed a new
concept for the direct reduction of oxide in molten CaCl , named the OS (Ono-Suzuki)
2
process. Distinguishing from the FFC process, the OS process involves the deposition of
metal calcium from CaO in molten CaCl and the calciothermic reduction of the metal oxide.
2
S. Wang et al. [20] performed the electrochemical reduction of V O in molten CaCl –NaCl–
2
3
2
CaO by using compact V O pellet as the cathode. They also explored the reaction pathway of
2
3
the reduction of vanadium sesquioxide based on time-dependent changes of the phase
composition; metal vanadium can be prepared within the processing time of 8 h. Z. Cai et al.
[
21] reported the preparation of metal vanadium by direct electrochemical reduction of V O
2 5
in molten CaCl –NaCl. The results demonstrate that the electrochemical reduction of
2
vanadium pentoxide to metal vanadium is accompanied by the decomposition of CaO and
2

ACCEPTED MANUSCRIPT
calcium thermal reduction, which act as underlying mechanism in the process.
The future looks bright for this new electrochemical process, because the simple operation,
little energy consumption and low pollution. In addition, it offers a new way for the
preparation of high-melting-point metals, alloys and nonstoichiometric compounds these are
difficult to be achieved through conventional metallurgical techniques. However, further
development of the FFC process for scaled production of metal is tardy due to the poor
current efficiency. There also still existed some major issues to be solved, one of them is the
need of long electrolysis time. For instance, as reported by C. Schwandt et al. [23] the
electrolysis preparation process of titanium metal from titanium dioxide precursors in molten
CaCl would extend over the reaction time up to several days. In order to promote the process,
2
some work so far has been done concerning the preparation of oxide precursors [24-26], the
selection of electrolyte [27-29], anode materials [29-30] and the electrolysis cell [31].
Notably, many efforts have been devoted to study the effect of oxygen ion concentration in
the electrolyte on the direct electro-reduction process [32-33]. C. Schwandt et al. [32]
reported the electro-reduction of titanium dioxide in molten calcium chloride with the
addition of calcium oxide. They found that the presence of CaO in molten CaCl can
2
accelerate the overall rate of the reduction process. The reasons they reported are that the
2
-
transfer rate of O ions is proportional to their concentration and the moderate concentration
of CaO in molten CaCl can facilitate the cathodic reduction kinetics. They also reported that
2
a high CaO concentration, e.g. 5 mol% CaO in CaCl , shows an adverse effect on the
2
reduction process. The main reason seems to be that a high concentration of CaO leading to
local saturation of CaO in the reaction region, which would clog the transmission channel
inside the cathode and terminate the reaction. R.F. Descallar-Arriesgado et al. [33] confirmed
the suitable amount of CaO for the reduction of NiO in molten CaCl is in the range of 0.5-3.0
2
mol%. They explained the influence of oxygen concentration on the process by OS
mechanism. That is, when intermediate amount CaO is initially added in molten CaCl , larger
2
amount of Ca metal reductant can generate in the cathode region than that generate in pure
CaCl . Hence, the transition rate of nickel metal from NiO can be substantially enhanced
2
through calcium thermal reduction. However, when 5.0 mol% CaO is added in molten CaCl2,
3

ACCEPTED MANUSCRIPT
the reduction is found to be incomplete. Their interpretation about the unfavorable condition
at high CaO concentration is the possibility of liquid calcium precipitation on the cathode.
That is to say, a great quantity of calcium can generate when 5.0 mol% CaO is electrolyzed in
CaCl melt and the calcium metal may start to precipitate and deposit on the outer surface of
2
the cathode. This precipitation phenomenon would hinder the transmission of oxide ions,
making the reduction ineffective.
In our early published paper [22], the electrochemical reduction of porous V O pellet,
2
3
using ammonium polyvanadate as raw material, was performed in molten Na AlF salt. Our
3
6
findings suggest that both the direct electrochemical reduction of V O and aluminothermic
2
3
reduction reaction occur during the reduction process, amongst which the latter plays a key
role. The oxygen content of the metal vanadium product can decrease to 0.218 mass% after 4
h of electrolysis. However, further effort is required to optimize the process parameters, in
particularly the proper oxygen ion concentration for the direct electro-reduction process in
molten Na AlF salt should be investigated. Because it is well established that the oxide ion
3
6
concentration plays a critical role in the FFC process basing on the numerous studies
performed in molten chloride salts. And yet no relevant study has devoted to the molten
fluoride salt electrolysis process. In addition, the results reported on the influence of oxygen
ion concentration in molten chloride salts are mostly based on theories or qualitative analyses.
It is difficult to experimentally prove the deductions due to the harsh experimental conditions
in high-temperature molten salt bath, thus systematic studies are still required.
The purpose of present study is to investigate the optimum oxygen ion concentration, i.e.
alumina concentration, in molten Na AlF salt for the electrochemical reduction of porous
3
6
vanadium sesquioxide precursors, aiming to reduce the electrolysis time and increase the
current efficiency. And the mechanism on the influence of the oxygen ion concentration in
electrolyte on the FFC process is experimentally proved.
2
2
. Experimental
.1 Preparation of electrolyte and cathode
Vanadium sesquioxide cathode was prepared using ammonium polyvanadate powder (≥
4

ACCEPTED MANUSCRIPT
9
9.5%, particle size range: 100–200 mesh, Dalian Galaxy Metal Material Co., Ltd.) as raw
material. Initial ammonium polyvanadate pellets (3.0 g, 20 mm in diameter and about 3.7 mm
in thickness) were prepared by moisturizing the powder with a few drops of deionized water
and then pressing under 30 MPa in a cylindrical mold. The obtained ammonium polyvanadate
pellets were heat-treated at 350°C for 1 h in a tube furnace under argon atmosphere
(>99.99%，1 L/min) and then reduced by coal gas (50% hydrogen, 35% methane, 10%
carbon monoxide, 5% ethylene, Shenyang Gas Co., Ltd.) at 850°C for 1 h. At last, the
reduced pellets were sintered at 1300°C for 3 h at a heating rate of 5°C/min in argon
atmosphere. The obtained oxide pellets were connected to a direct current (DC) power supply
by using nickel chromium wire (1mm in diameter) as the current lead. More detailed
information about the preparation and assembly of the vanadium sesquioxide cathode was
provided in the foregoing study [22].
The electrolyte was prepared using analytical reagent NaF and AlF (3:1 in mole ratio,
3
Sinopharm Chemical Reagent Co., Ltd.) with 0–3.0 mass% Al O (analytical reagent,
2
3
Sinopharm Chemical Reagent Co., Ltd.). The salt mixtures (about 500 g) were placed in a
high-purity graphite crucible with an inner diameter of 70 mm and depth of 100 mm. The
graphite crucible was also used as the anode in the constant voltage electrolysis experiments
or as the counter electrode in the cyclic voltammogram experiments. Prior to the experiments,
the salt mixtures were dried at 150°C for 24 h in a vacuum drying oven.
2
.2 Solubility of V O in molten Na AlF
2 3 3 6
About 100 g of electrolyte was contained in a platinum crucible (50 mm in inner diameter,
5
0 mm in depth) and heated to 1010°C in argon atmosphere. When the electrolyte was
5

ACCEPTED MANUSCRIPT
melted, the sintered vanadium sesquioxide pellet was then put into the melt. The temperature
was kept at 1010°C for 1 h and the melt was stirred every ten minutes by using a molybdenum
rod. At the end of the experiment, a sample of the melt was then withdrawn by dipping with
molybdenum rod and subsequently quenched. The content of vanadium in the taken sample
was then analyzed by Inductively Coupled Plasma (ICP, Optima 8300).
2
.3 Constant voltage electrolysis
The power supply (DP1308A, Rigol Technologies Inc.) was used to perform the constant
voltage electrolysis experiment. The graphite crucible containing the salt mixtures was placed
in a sealed stainless steel reactor with argon gas flushing into continuously. The temperature
was gradually increased and hold at 1010°C to melt the electrolyte. Then the cathode pellet
was submerged into the molten salt and a constant voltage of 3.5 V was applied to the cathode
and the graphite crucible anode. The electrolysis current was recorded every 5 seconds by
digit multimeter (34401A, Keysight Technologies). After electrolysis for 3–4 h, the cathode
was lifted out from the molten salt and cooled down to room temperature in argon
atmosphere. The product obtained after electrolysis for 4 h in molten Na AlF was polished
3
6
and analyzed using scanning electron microscopy (SEM, Ultra Plus). The products obtained
after electrolysis for 3 h in molten Na AlF with the addition of Al O were ground and then
3
6
2
3
washed in 15% AlCl solution (at 80 °C for 3 h) and distilled water, dried in vacuum and
3
characterized by X-ray diffraction (XRD, D/max-2500PC). The oxygen contents of the
products were determined by oxygen nitrogen analyzer (TC-436, LECO Corporation) and all
the oxygen content data are measured by using the whole pellet as the test sample.
2
.4 Cyclic voltammogram
6

ACCEPTED MANUSCRIPT
A three-electrode system was used for the cyclic voltammogram experiment. Molybdenum
rod (>99.95%, 5 mm in diameter, 600 mm in length) was used as the working electrode. The
high-purity graphite crucible was used as the counter electrode. A platinum wire (>99.95%,
0
.5 mm in diameter, 50 mm in length) processed into spiral shape was used as the pseudo-
2
-
reference electrode Pt/PtO /O . In order to guarantee that the working electrode kept the same
x
active area, the immersion depth of the electrode was control strictly by using a self-made
device. The experimental apparatus for the cyclic voltammogram experiment are shown in
Fig. 1. The cyclic voltammogram experiments were performed with an electrochemical
workstation (IM6e, Zahner Elektrick, Germany).
3
3
. Results and discussion
.1. Current versus time curves and characterization of products
Fig. 2(a) shows the current-time curves recorded during the electrolysis process in molten
Na AlF with the addition of 0–3.0 mass% Al O . Except for the Al O concentration, the
3
6
2
3
2
3
experimental conditions were the same for all the runs, especially for the immersion depth of
the cathode pellet. As shown in the figure, the overall current increased with the increase of
Al O concentration in general. All the plots show much higher initial current values and then
2
3
declined rapidly in the subsequent 10 min due to the diffusion of oxide ions expelled from the
cathodic vanadium sesquioxide is fast during the initial stage of the electrolysis process.
Fig. 2(a) also shows that the currents measured at the initial stage of the electrolysis process
performed in molten Na AlF with Al O are higher than that in pure molten Na AlF . It was
3
6
2
3
3
6
reported that during the direct electro-reduction process of solid oxide, the transport of oxide
ions through the electrolyte is only driven by the concentration gradient [34] and the transfer
7

ACCEPTED MANUSCRIPT
rate of oxide ions is proportional to their concentration in the electrolyte [35]. Obviously, the
addition of Al O in molten Na AlF accelerates the transmission of oxide ions from the
2
3
3
6
cathode toward to the anode and thus speeds up the reduction reaction occurred in the initial
stage of the process, leading to the higher current than that in pure molten Na AlF .
3
6
It also can be seen in Fig. 2(a) that the currents maintain steady at time-independent
constant values at the later stage of the electrolysis process, except for the electrolysis
processed in molten Na AlF without Al O . Fig. 2(b) shows the increase in termination
3
6
2
3
current as the Al O concentration increases. This termination current represents a complex
2
3
electrolysis process, involving the decomposition of alumina in electrolyte [22] and the
background current arising from the conductivity of the electrolyte [4, 23]. As reported, the
continuous addition of Al O inside the molten cryolite would decrease the conductivity of
2
3
electrolyte [36] and the decomposition potential of alumina decreases in a Nernst-type fashion
when the concentration increases. So it reasonable to concluded that the increase of Al O
2
3
concentration in the electrolyte can promote the decomposition of alumina and generate more
metal aluminium during the electrolysis process.
Fig. 3 shows the XRD patterns of the as-prepared samples electrolyzed at 3.50 V for 3 h in
molten Na AlF with 0–3.0 mass% Al O . It can be seen that the Al O concentration has a
3
6
2
3
2
3
significant influence on the electro-reduction process of V O . As shown in Fig. 3(b), the
2
3
product obtained at 0.5 mass% Al O is mainly composed of metal vanadium, indicating the
2
3
complete reduction reaction of V O . However, the cathodic product obtained after 3 h of
2
3
electrolysis in pure molten Na AlF is found to be consisted of metal vanadium and low-
3
6
valence vanadium oxide, i.e. V O. As a small amount of water always existed in the
9
8

ACCEPTED MANUSCRIPT
electrolyte, introduced by inefficient drying procedure inevitably, this would cause a relative
low oxygen ion concentration in the electrolyte. This low oxygen concentration sets up a
transport limitation for the oxide ions expelled from the cathode towards to the anode and
thus causes the sluggishness of the electro-reduction process in molten Na AlF .
3
6
When Al O concentrations were increased to 1.0–3.0 mass%, the diffraction peaks of
2
3
Al V appears in the XRD patterns shown in Fig. 3(c-f), indicating that the electrodeposition
8
5
of metal aluminum is promoted at the cathode area. It can be deduced that the produced
aluminium can reacts with V O and thus enhances the reduction. On the other hand, the
2
3
excessive aluminium has also brought along some side effects: i) alloying with the reduced
metal vanadium and forming vanadium aluminium alloy, e.g. Al V ; ii) hindering the
8
5
continuous progress of the reduction process, since some oxide phase still retained in the
products. The oxygen content values of the products are exhibited in Fig. 4. It can be seen that
the oxygen content determination results are in accord with the XRD results. The oxygen
content of the formed metal vanadium after 3 h of electrolysis in molten Na AlF with 0.5
3
6
mass% Al O was decreased to 0.207 mass% and the current efficiency was calculated to be
2
3
5
5%. When the Al O concentration continued to increase, the oxygen concentration of the
2 3
products began to increase significantly. The detailed influence mechanism about the Al O
2
3
concentration on the process will be discussed in the latter sections.
It is noteworthy to address the presence of AlV O phase in the products. In some of
2
4
published articles about the preparation and properties of the vanadium spinels, the valence
states of vanadium ions in AlV O were reported to be V(II) and V(III) [37-38] and the
2
4
similar vanadium compounds can be synthesized by solid-state reactions between vanadium
9

ACCEPTED MANUSCRIPT
oxides and other metal oxides at 900–1200°C, e.g. MnO, CoO, ZnO, MgO [39]. Thus, it is
reasonable to suggest the chemical equation for the generation of AlV O in this molten salt
2
4
system as follow:
Al O + 2VO + V O = 2AlV O
4
(1)
2
3
2
3
2
3
.2. Microstructure of products prepared at different Al O concentrations
2 3
Fig. 5(a) and (b) show the SEM images of the starting cathode V O pellet and the product
2
3
reduced in molten Na AlF for 4 h, all taken at the surface of the pellets. As shown in Fig.
3
6
5
(a), the starting cathode V O particles is consisted of irregular polyhedral particles with
2 3
uniform size of about 400–500 nm. During the electrolysis process, cathode V O showed a
2
3
good stability in molten Na AlF because of the low solubility (the solubility was measured to
3
6
be 0.19 mass%). The SEM image of the reduced sample in Fig. 5(b) shows the grain
coarsening behavior of vanadium metal. The EDS analysis result shown in Fig. 5(c) reveals
that no oxide phase is detected and the metal phase is composed of vanadium. The formed
metal phase shows a porous reticular structure. This porous metal layer provides a diffusion
pathway for the molten salt and oxide ions expelled from the inner regions. As revealed in
Fig. 5(d), the EDS analysis result shows the molten salt and oxygen filled in the pores of
metal phases.
Chen et al. [5] proposed a 3PI (three-phase interlines) model to describe the cathode
reduction kinetic process of solid oxide in molten salt. That is, the direct electrochemical
reduction of the solid oxide begins at the metal (or current collector) / metal oxide /
electrolyte interlines. At the beginning of electrolysis, the 3PI only presents next to the metal
wire. Upon providing electrons, the oxide particles next to the metal wire can be reduced to
1
0

ACCEPTED MANUSCRIPT
metal firstly and the metallized area then operates as the current collector propagating the 3PI
to the core part. This model holds when the oxide precursor is an electronic insulator.
However, since V O is an electronic conductor at high temperature (the conductivity of V O
3
2
3
2
-1
is about 4.6 S·cm at 1010°C [40]), the 2-phase interface mechanism is more applicable, i.e.
only the solid oxide phase and liquid electrolyte phase presenting at the reaction interface.
Based on the results reported by D.S.M. Vishnu et al. [41], it can be certain that the oxide
pellet can act as the electron conductor. Hence the electro-reduction begins at the whole
electroactive surface of the cathode and induces a high initial current, as illustrated in Fig.
2
(a). Subsequently, the oxide surface next to the electrolyte is converted to metal, forming a
metal layer covering the inner oxide. This initially formed metal layer must be porous to
allow the electrolyte to access the inner oxide and oxide ions can move out through the pellet,
insuring the reduction continues.
After electrolysis for 1 h in molten Na AlF , the cathodic product was cut into cross-section
3
6
and analyzed by SEM and EDS, shown in Fig. 6. As indicated in Fig. 6(a-b), the outer layer
of cathode pellet contacting with the bulk electrolyte is reduced, leaving a metal vanadium
layer. The oxygen scattered in this outer area, shown in Fig. 6(c), may be corresponding to the
oxygen dissolved in the molten salt. Fig. 6(d-f) show the EDS elemental maps of the
electrolytic composition, i.e. sodium, fluorine and aluminium elements. By SEM back-
scattering analysis, it can be seen clearly that the reduced metal layer is sufficiently porous
and the electrolyte distributes in the pores and the unreduced core. The formed porous metal
layer provides an accessible way for the electrolyte to access the oxide in the inner layer and
allows the oxide ions diffusing through the solid cathode pellet and entering the bulk
1
1

ACCEPTED MANUSCRIPT
electrolyte. Hence, there is no denying that insuring a high porosity of the formed outer metal
layer is essential for guaranteeing the continuity of solid electro-reduction process.
To gain a better understanding of the influences of Al O concentration on the
2
3
microstructure of the products and reaction kinetics, the SEM images of the samples taken at
the metal layer area are shown in Fig. 7. The samples were obtained by electrolysis for 3 h in
molten Na AlF with the addition of 0–3.0 mass% Al O . It can be seen that the outer metal
3
6
2
3
layer is composed of metal phase distributed in light zone and molten salt phase distributed in
black zone. Apparently, with the increase of Al O concentration in the electrolyte, the metal
2
3
grains were agglomerated into large blocks, which would compress the volume of the molten
salt phase inevitably. This would make the reduction process kinetically difficult because the
diffusion of oxide ions from the inner layer towards to the electrolyte is blocked by the
formed dense metal layer.
In order to obtain quantitative data about the volume ratio of molten salt phase, for each
sample more than five SEM images of the metal layer area were taken. The volume ratios of
molten salt phase were quantified by Image Pro-Plus 6.0 software. The obtained data is
displayed in Fig. 8. Obviously, the volume ratio of molten salt phase almost presents
decreased line shape with the increase of Al O concentration. That is to say, with the increase
2
3
of Al O concentration, more metal phases are generated in the outer metal layer. In the view
2
3
of reaction kinetics, the dense metal layer delays the diffusion of oxide ions and electrolyte,
slowing down the reduction process.
3
.3. Influence mechanism of Al O concentration on the reduction process
2 3
The findings mentioned above provide concrete evidences for exploring the influence of
1
2

ACCEPTED MANUSCRIPT
Al O concentration on the electro-reduction process of V O by observing and analyzing the
2
3
2
3
microstructure of the formed outer metal layer. As will be described later, detailed study
about the electrochemical behavior of aluminium ions and the growth mechanism of the metal
grains in metal layer are provided.
Fig. 9 shows the cyclic voltammograms recycled for 10 times continuously on a
molybdenum electrode in molten Na AlF with the addition of 1.0 mass% Al O . In line with
3
6
2
3
the previous findings [22], the appearance of c1/a1 couple demonstrates the electrodeposition
of aluminium ions and the reoxidation of formed aluminium metal. It can be seen that the
reduction current and oxidation current increase as the scan cycle repeats. The increase of the
reduction current is explainable. As the solubility of aluminium in molten cryolite is only 0.1–
1
.5 mass% [42], together with the delayed dissolution of aluminium in the bulk molten salt,
all these factors cause the adhesion of aluminium on the molybdenum electrode. This
phenomenon may enlarge the reaction interface on the working electrode and then more
aluminium generates, causing the increasing current as observed in Fig. 9.
The cyclic voltammograms obtained on the first scan cycle for aluminium deposition in
molten Na AlF with the addition of 0.0–3.0 mass% Al O illustrate the differences among
3
6
2
3
various Al O concentrations, as shown in Fig. 10. It can be seen that with the increase of
2
3
Al O in electrolyte, the deposition potential of aluminium ions are almost constant, but the
2
3
reduction currents increase. This indicates that larger amount of aluminium is formed per unit
time at higher Al O concentration.
2
3
According to the electrochemical measurement results and the previous studies [32-33], the
improvement of electro-reduction process at around 0.5 mass% Al O can be explained by the
2
3
1
3

ACCEPTED MANUSCRIPT
appropriate increase of aluminium reductant [33] and the relatively high transfer rate of oxide
ions [10]. When the Al O concentration was increased to 1.0–3.0 mass%, a certain amount of
2
3
oxide phase still existed in the products, as shown in Fig. 3, implying the incomplete
reduction of V O at high Al O concentration. Fig. 11(a) shows the SEM image taken at the
2
3
2
3
metal layer of the sample obtained after electrolysis for 3 h at 2.0 mass% Al O . In line with
2
3
the previous finding, the metal layer is composed of metal phase and molten salt phase. But it
is notably mentioning that the metal phase exhibits heterogeneity in chemical component,
which can be observed clearly by SEM back-scattering analysis. Fig. 11 (c-d) shows the EDS
spectrums of the metal phase. It can be seen that the external layer is composed of vanadium
aluminium alloy and the inner layer is mainly vanadium, with little aluminium. This finding is
consistent with the XRD results. The synergistic effects of the substantial increase of metal
aluminium and the decrease in aluminium solubility at high Al O concentration may trigger
2
3
liquid aluminium precipitation in the electrolyte. This inference can be proved by analyzing
the elemental distribution of the molten salt phase filled in the pores of the metal layer. As
shown in Fig. 11 (b), the EDS result shows that the content of aluminium is much higher than
that of sodium and fluorine, indicating that liquid aluminium presents in the electrolyte.
Therefore, liquid aluminium may deposit on the surface of metal vanadium initially formed in
the outer layer of cathode and form Al–V alloy. The formation of alloy can cause the growth
of metal grains, which decreases the porosity of the outer layer and forms a dense metal layer
covering the inner oxide layer, as shown in Fig. 7. Then the diffusion of oxide ions expelled
from the inner oxide towards to the bulk electrolyte is hindered, making the whole reduction
process ineffective. Moreover, the increase of metal layer area could provide a larger active
1
4

ACCEPTED MANUSCRIPT
surface area for the electrodeposition of aluminum ions. Thus, it will consume more electric
energy for the aluminium electrodeposition and cause the increase of electrolysis current, as
illustrated in Fig. 2 (a). This is supported by the EDS line-scan profile data shown in Fig. 12,
which indicates that the aluminium content within 400 μm of distance range from the cathode
pellet surface is higher evidently than that of inner layer.
4
. Conclusions
The suitable Al O concentration in molten Na AlF for the direct electro-reduction of
2
3
3
6
V O to vanadium was studied. The addition of Al O into molten Na AlF has a significant
2
3
2
3
3
6
impact on the reduction process. The optimum Al O concentration is around 0.5 mass% and
2
3
this value can improve the reduction rate of V O . The electrolysis time can be shortened to 3
2
3
h, meanwhile the oxygen content of the product could be decreased to 0.207 mass% at a
current efficiency of 55%. In contrast, adding too much Al O in the electrolyte would impede
2
3
the reduction process. The experiment proves that an abundance of aluminium can electro-
deposit at high Al O concentration and liquid aluminium would adhere on the surface of
2
3
metal vanadium in the outer layer of the cathode and further form Al–V alloy, causing the
densification of the outer metal layer. This would block the pathway for the diffusion of
electrolyte and oxide ions, making the reduction process ineffective.
1
5

ACCEPTED MANUSCRIPT
References
1] 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.
2] D.J. Fray, Novel methods for the production of titanium, Int. Mater. Rev. 53 (2008) 317-
25.
3] K.S. Mohandas, Direct electrochemical conversion of metal oxides to metal by molten salt
electrolysis: a review, Trans. Inst. Min. Metall., Sect. C 122 (2013) 195-212.
4] A.M. Abdelkader, K.T. Kilby, A. Cox, D.J. Fray, DC voltammetry of electro-deoxidation
of solid oxides, Chem. Rev. 113 (2013) 2863-2886.
5] G.Z. Chen, The FFC Cambridge process and its relevance to valorisation of ilmenite and
titanium-rich slag, Int. Mater. Rev. 124 (2015) 95-105.
6] D. Fray, C. Schwandt, Aspects of the application of electrochemistry to the extraction of
titanium and its applications, Mater. Trans. 58 (2017) 306-312.
7] J. Sure, D.S.M. Vishnu, C. Schwandt, Direct electrochemical synthesis of high-entropy
alloys from metal oxides, Appl. Mater. Today 9 (2017) 111-121.
8] D.S.M. Vishnu, N. Sanil, K.S. Monhandas, K. Nagarajan, Direct electrochemical
preparation of Nb–10Hf–1Ti alloy, Electrochim. Acta 55 (2010) 2924-2931.
9] Z. Zhou, Y. Zhang, Y. Hua, P. Dong, C. Xu, Y. Li, D. Wang, Verification of the electro-
[
[
3
[
[
[
[
[
[
[
decomposition of the CaO component in equimolar CaCl –NaCl molten salt during the direct
2
electrolysis of ilmenite in a two-terminal chamber, Electrochim. Acta 271 (2018) 490-497.
[10] Z. Zhou, Y. Hua, C. Xu, J. Li, Y. Li, Q. Zhang, Y. Zhang, W. Kuang, Synthesis of
micro-FeTi powders by direct electrochemical reduction of ilmenite in CaCl –NaCl molten
2
salt, Ionics 23 (2016) 213-221.
[11] M. Panigrahi, E. Shibata, A. Iizuka, T. Nakamura, Production of Fe–Ti alloy from mixed
ilmenite and titanium dioxide by direct electrochemical reduction in molten calcium chloride,
Electrochim. Acta 93 (2013) 143-151.
[12] B.K. Jackson, D. Inman, M. Jackson, D. Dye, R.J. Dashwood, NiTi production via the
FFC Cambridge Process: refinement of process parameters, J. Electrochem. Soc. 157 (2010)
E36-E43.
1
6

ACCEPTED MANUSCRIPT
[13] Z. Liu, H. Zhang, L. Pei, Y. Zhang, Direct electrochemical preparation of Ni-Cr alloy
from solid oxides in molten CaCl , J. Electrochem. Soc. 164 (2017) D237-D243.
2
[14] Q. Xu, C. Schwandt, D.J. Fray, Preparation of Ta-Nb alloy powder by electro-
deoxidation of Ta O /Nb O mixture in a CaCl –NaCl eutectic melt, Adv. Mater. Res. 160-
2
5
2
5
2
1
62 (2011) 1131-1135.
15] R. Enmei, T. Kikuchi, R.O. Suzuki, Production of Nb–Ti–Ni alloy in molten CaCl2,
Electrochim. Acta 100 (2013) 257-260.
16] P. Hussain, R. Bhagat, R. Dashwood, Precursor preparation for Ti-Al-V-Y alloy via FFC
Cambridge Process, Appl. Mech. Mater. 330 (2013) 62-67.
17] W. Weng, M. Wang, X. Gong, Z. Wang, Z. Guo, Thermodynamic analysis on the direct
[
[
[
preparation of metallic vanadium from NaVO by molten salt electrolysis, Chin. J. Chem.
3
Eng. 24 (2016) 671-676.
[
18] O. Watanabe, Electrolytic reduction of vanadium oxides in molten salt bath, J.
Electrochem. Soc. Jpn. 34 (1966) 91-98.
19] R.O. Suzuki, H. Ishikawa, Direct reduction of vanadium oxide in molten CaCl , Trans.
[
2
Inst. Min. Metall., Sect. C 117 (2008) 108-112.
[
20] S.L. Wang, S.C. Li, L.F. Wan, C.H. Wang, Electro-deoxidation of V O in molten
2 3
CaCl –NaCl–CaO, Int. J. Min. Met. Mater. 19 (2012) 212-216.
2
[21] Z.F. Cai, Z.M. Zhang, Z.C. Guo, H.Q. Tang, Direct electrochemical reduction of solid
vanadium oxide to metal vanadium at low temperature in molten CaCl –NaCl, Int. J. Min.
2
Met. Mater. 19 (2012) 499-505.
[
22] Y. Kong, J. Chen, B. Li, K. Liu, Q. Han, Electrochemical reduction of porous vanadium
trioxide precursors in molten fluoride salts, Electrochim. Acta 263 (2018) 490-498.
23] C. Schwandt, 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.
[
24] M. Panigrahi, A. Iizuka, E. Shibata, T. Nakamura, Electrolytic reduction of mixed (Fe,
Ti) oxide using molten calcium chloride electrolyte, J. Alloys Compd. 550 (2013) 545-552.
25] Q. Xu, L. Deng, Y. Wu, T. Ma, A study of cathode improvement for electro-deoxidation
[
1
7

ACCEPTED MANUSCRIPT
of Nb O in a eutectic CaCl –NaCl melt at 1073 K, J. Alloys Compd. 396 (2005) 288-294.
2
5
2
[
26] X. Liu, M. Hu, C. Bai, X. Lv, Effect of electrical conductivity and porosity of cathode on
electro-deoxidation process of ilmenite concentrate, Rare Metal Mat. Eng. 46 (2017) 1176-
182.
27] X. Yan, D.J. Fray, Synthesis of niobium aluminides by electro-deoxidation of oxides, J.
Alloys Compd. 486 (2009) 154-161.
28] T. Wang, H. Gao, X. Jin, H. Chen, J. Peng, G.Z. Chen, Electrolysis of solid metal sulfide
to metal and sulfur in molten NaCl–KCl, Electrochem. Commun. 13 (2011) 1492-1495.
29] M. Gibilaro, J. Pivato, L. Cassayre, L. Massot, P. Chamelot, P. Taxil, Direct
electroreduction of oxides in molten fluoride salts, Electrochim. Acta 56 (2011) 5410-5415.
30] D.S.M. Vishnu, J. Sure, K.S. Mohandas, Corrosion of high density graphite anodes
during direct electrochemical de-oxidation of solid oxides in molten CaCl medium, Carbon
1
[
[
[
[
2
9
3 (2015) 782-792.
31] M. Hu, Z. Qu, C. Bai, D. Hu, G.Z. Chen, Effect of the changed electrolytic cell on the
current efficiency in FFC Cambridge Process, Mater. Trans. 58 (2017) 322-325.
32] C. Schwandt, D.J. Fray, The electrochemical reduction of chromium sesquioxide in
molten calcium chloride under cathodic potential control, Z. Naturforsch. A 62 (2007) 655-
70.
33] R.F. Descallar-Arriesgado, N. Kobayashi, T. Kikuchi, R.O. Suzuki, Calciothermic
reduction of NiO by molten salt electrolysis of CaO in CaCl melt, Electrochim. Acta 56
[
[
6
[
2
(2011) 8422-8429.
[34] C. Schwandt, On the nature of the current and the absence of an IR-drop in an FFC-
Cambridge-type electro-deoxidation cell, Electrochim. Acta 280 (2018) 114-120.
35] C. Schwandt, Comment on the article “Effect of CaO addition on preparation of
[
ferrotitanium from ilmenite by electrochemical reduction in CaCl –NaCl molten salt” by L.
2
Xiong, Y.X. Hua, C.Y. Xu, J. Li, Y. Li, Q.B. Zhang, Z.R. Zhou, Y.D. Zhang, J.J. Ru, J.
Alloys Compd. 710 (2017) 692-694.
[
36] P. Fellner, K. Grjotheim, H. Kvande, Electrical conductivity of cryolite melts, JOM 37
1985) 29-32.
(
1
8

ACCEPTED MANUSCRIPT
[
37] Y. Cai, Z. Huang, X. Ming, C. Wang, G. Chen, Charge disproportionation in AlV O : a
2
4
first-principles study, J. Alloys Compd. 505 (2010) L23-L26.
[38] A.J. Browne, S.A.J. Kimber, J.P. Attfield, Persistent three- and four-atom orbital
molecules in the spinel AlV O , Phys. Rev. Materials 1 (2017) 0520031-0520035.
2
4
[
39] D.B. Rogers, R.J. Arnott, A. Wold, J.B. Goodenough, The preparation and properties of
some vanadium spinels, J. Phys. Chem. Solids 24 (1963) 347-360.
40] Z. Xu, J. Luo, K. Chuang, V O Anode catalyst for solid oxide fuel cell using H S-
[
2
3
2
containing hydrogen, ECS Trans. 11 (2008) 1-17.
41] D.S.M. Vishnu, N. Sanil, K.S. Mohandas, Difference in the mechanism of
[
electrochemical deoxidation of conducting and non-conducting solid oxide preforms: an
experimental demonstration with TiO and SiO pellet electrodes in CaCl melt, Res. Rev.: J.
2
2
2
Mater. Sci. 5 (2017) 55-65.
42] D.C. Lynch, M.K. Han, Influence of Al O on the reoxidation of aluminum in molten
[
2
3
cryolite, Metall. Mater. Trans. B 12 (1981) 361-370.
1
9

ACCEPTED MANUSCRIPT
Figure Captions
Fig. 1. Experimental apparatus for cyclic voltammogram.
Fig. 2. Current versus time curves (a) obtained during the electrolysis for 3 h at 1010°C under
2
3
.5 V in molten Na AlF with 0–3.0 mass% Al O , cathodic geometric surface area: 7.49 cm ,
3
6
2
3
the currents (b) recorded at the end stage of the electrolysis process as a function of Al O
2
3
concentration in the electrolyte.
Fig. 3 XRD patterns of the as-prepared samples electrolyzed at 3.50 V for 3 h at 1010°C in
molten Na AlF with different Al O concentrations; a: 0 mass% Al O , b: 0.5 mass% Al O ,
3
6
2
3
2
3
2
3
c: 1.0 mass% Al O , d: 1.5 mass% Al O , e: 2.0 mass% Al O , f: 3.0 mass% Al O . Reference
2
3
2
3
2
3
2
3
standard data from JCPDS cards: V: NO. 089-3842, V O: NO. 042-1263, Al V : NO. 071-
9
8
5
0
141, AlV O : NO. 025-0025, V O: NO. 039-0773.
2 4 2
Fig. 4 Influence of Al O concentration on the oxygen content of the products electrolyzed at
2
3
3
.50 V for 3 h.
Fig. 5. SEM images of the porous V O pellet (a) and the entirely reduced product (b)
2
3
obtained after electrolysis for 4 h at 1010°C in molten Na AlF (w(Al O ) = 0 mass%), with
3
6
2
3
the EDS spectrums (c-d) measured over the specified positions in (b).
Fig. 6. SEM image (a) and EDS elemental map images (b-f) of the cross-section of the
product obtained after electrolysis for 1 h at 1010°C in molten Na AlF (w(Al O ) = 0
3
6
2
3
mass%).
Fig. 7. SEM images of metal layer area in the samples obtained after electrolysis for 3 h at
010°C in molten Na AlF at different Al O concentrations: a: 0 mass% Al O , b: 0.5 mass%
1
3
6
2
3
2
3
Al O , c: 1.0 mass% Al O , d: 1.5 mass% Al O , e: 2.0 mass% Al O , f: 3.0 mass% Al O .
2
3
2
3
2
3
2
3
2
3
2
0

ACCEPTED MANUSCRIPT
Fig. 8. Volume ratio of the molten salt phase as a function of Al O concentrations in the
2
3
electrolyte.
Fig. 9. Cyclic voltammograms recycled for 10 times continuously on a molybdenum metal
electrode at 1010°C in molten Na AlF with 1.0 mass% Al O . Scan rate: 100 mV/s.
3
6
2
3
Fig. 10. Cyclic voltammograms obtained on the first scan on molybdenum metal electrode at
010°C in molten Na AlF with the addition of 0.0–3.0 mass% Al O . The insert shows the
1
3
6
2
3
2
-
cathodic peak currents recorded at -1.6 V vs. Pt/PtO /O . Scan rate: 100 mV/s.
x
Fig. 11. SEM image and EDS spectrums of the metal layer area of the sample obtained after
electrolysis for 3 h at 1010°C in molten Na AlF with 2.0 mass% Al O .
3
6
2
3
Fig. 12. The cross-sectional SEM image and corresponding EDS line scan along the white
line of the sample obtained after electrolysis for 3 h at 1010°C in molten Na AlF with 2.0
3
6
mass% Al O .
2
3
2
1