Investigation of the electrochemical reduction of Na2Ti3O7 in CaCl2 molten salt

Article abstract of DOI:10.1016/j.electacta.2019.06.072

Sodium titanate (Na₂Ti₃O₇), as an intermediate product for producing TiO₂ through alkaline process, was used as precursor to prepare Ti metal successfully by FFC Cambridge Process. For the aim to gain insight into the electro-reduction mechanism, the sintered Na₂Ti₃O₇ pellets(～1.83 mm thinkness, open porosity ～20%) were electrolysed using them as cathodes against graphite counter electrode in the molten CaCl₂. The experiments were carried out at 900 °C and the applied voltage was 3.1V. Partially reduced samples were prepared by terminating the reduction process after different electrolysis times. The obtained samples were characterised by means of X-ray diffraction analysis, SEM and EDS. The results show that Na₂Ti₃O₇ reacts easily with molten CaCl₂ as 2Na₂Ti₃O₇ + 2CaCl₂ → Ca₂Ti₂O₆ + 4TiO₂ + 4NaCl and Ca₂Ti₂O₆ → 2CaTiO₃. The electrochemical reduction of sodium titanate proceeds via sequential formation of CaTiO₃, titanium sub-oxides (such as Ti₄O₇, Ti₃O₅, Ti₂O₃ and TiO), CaTi₂O₄, Ti-O solid solution and Ti. The whole reduction can be divided into three stages: the first stage is that Ca²⁺ ions from electrolyte are inserted into Na₂Ti₃O₇ particles leading to the formation of titanium sub-oxides and calcium titanates(CaTiO₃ and CaTi₂O₄); the second stage is that calcium titanates are reduced into Ti-O solid solution from outside to inside of the pellets; the third stage is that the formed Ti-O solid solution is further deoxidised to form Ti metal.

Full text of DOI:10.1016/j.electacta.2019.06.072

Electrochimica Acta 318 (2019) 236e243
Contents lists available at ScienceDirect
Electrochimica Acta
journal homepage: www.elsevier.com/locate/electacta
Investigation of the electrochemical reduction of Na2Ti3O7 in CaCl2
molten salt
Kejia Liu, Yaowu Wang* , Yuezhong Di, Jianping Peng
Department of Metallurgy, Northeastern University, Shenyang, PR China
a r t i c l e i n f o
a b s t r a c t
Article history:
Sodium titanate (Na2Ti3O7), as an intermediate product for producing TiO2 through alkaline process, was
used as precursor to prepare Ti metal successfully by FFC Cambridge Process. For the aim to gain insight
into the electro-reduction mechanism, the sintered Na2Ti3O7 pellets(~1.83 mm thinkness, open porosity
~20%) were electrolysed using them as cathodes against graphite counter electrode in the molten CaCl2.
The experiments were carried out at 900 ꢀC and the applied voltage was 3.1V. Partially reduced samples
were prepared by terminating the reduction process after different electrolysis times. The obtained
samples were characterised by means of X-ray diffraction analysis, SEM and EDS. The results show that
Na2Ti3O7 reacts easily with molten CaCl2 as 2Na2Ti3O7 þ 2CaCl2 / Ca2Ti2O6 þ 4TiO2 þ 4NaCl and
Ca2Ti2O6 / 2CaTiO3. The electrochemical reduction of sodium titanate proceeds via sequential forma-
tion of CaTiO3, titanium sub-oxides (such as Ti4O7, Ti3O5, Ti2O3 and TiO), CaTi2O4, Ti-O solid solution and
Ti. The whole reduction can be divided into three stages: the first stage is that Ca2þ ions from electrolyte
are inserted into Na2Ti3O7 particles leading to the formation of titanium sub-oxides and calcium tita-
nates(CaTiO3 and CaTi2O4); the second stage is that calcium titanates are reduced into Ti-O solid solution
from outside to inside of the pellets; the third stage is that the formed Ti-O solid solution is further
deoxidised to form Ti metal.
Received 5 March 2019
Received in revised form
10 June 2019
Accepted 12 June 2019
Available online 14 June 2019
Keywords:
Electrolysis
Sodium titanate
Calcium chloride
Titanium
© 2019 Elsevier Ltd. All rights reserved.
1. Introduction
to Kroll process for producing Ti metal, because the transportation
and storage of TiO2 is easier than TiCl4. However, the production of
As a promising method for preparing Ti metal, FFC Cambssridge
Process has been studied for about twenty years [1e11]. Its main
mechanism is oxygen ionisation of cathode oxides in CaCl2 molten
salt under the driving of a voltage that is below the decomposition
potential of CaCl2 but higher than reduction potential of titanium
oxides. The ionised oxygen passes into electrolyte from oxide pel-
lets and discharged on graphite anode to evolve CO and CO2 gas.
The whole electro-deoxidation process could be summarized as
follows:
TiO2 by the traditional sulphate process or chloride process is still a
highly energy-wasting and high-pollution process. Besides, high
purity TiO2 would be more expensive than TiCl4 [12]. That is
because that, in the pigment industry, TiCl4 is the precursor to
produce TiO2. Meanwhile, TiO2 produced via the traditional sul-
phate process is priced in the market at similar prices to that of
chloride route pigment, which means the TiO2 produced via the
traditional sulphate process could be not cheaper. In recent years, a
novel process for preparing TiO2 was invented, called alkaline
process [13e16]. In the process, ilmenite is decomposed in
concentrated KOH or NaOH solution into potassium or sodium ti-
tanates which can be converted into pigment grade titanium di-
oxide after further treatments such as acid leaching and calcination.
In the process, potassium or sodium titanates, as the intermediate
products, are produced more easily than TiO2. Therefore, using
sodium titanate (Na2Ti3O7) to replace TiO2 as precursor to prepare
Ti metal by FFC Cambridge process was investigated in this paper.
Cathode: TiO2 þ 4eꢁ / Ti þ 2O2ꢁ
(1)
Anode: xO2ꢁ þ C / COx þ 2xeꢁ (x ¼ 1 or 2) (2)
Using TiO2 as the precursor is one of the advantages comparing
* Corresponding author.
E-mail
addresses:
wangyaowu513@126.com,
wangyw@smm.neu.edu.cn
(Y. Wang).
https://doi.org/10.1016/j.electacta.2019.06.072
0013-4686/© 2019 Elsevier Ltd. All rights reserved.

K. Liu et al. / Electrochimica Acta 318 (2019) 236e243
237
2. Experiment
applying a potential promotes reaction (2). As the reaction time
went on, the diffraction peaks of NaCl became weaker and weaker,
which indicates the pellets have a good diffusion channel for ions.
The theoretical decomposition potential of CaCl2 at 900 ꢀC is
3.137V, which is calculated from thermodynamic data according to
Nernst equation [17]. In experiments, the observed decomposition
potential of CaCl2 agrees well with the calculated value. Fig. 2a
shows the typical current vs. time (I-t) curve during the electro-
chemical reduction experiments performed with a Na2Ti3O7 pellet
as the cathode and a graphite rod as the anode under an applied
voltage of 3.1V. The current jumped to about 3A in the moment the
voltage was applied and then decreased sharply in 1 min, which
could be attributed to the double layer charging and the accumu-
lation of adatoms on the electrode. In the next minutes, the current
increased quickly again to about 3A. According to 3PIs (three-phase
interlines) propagation models in initial phase of electro-
deoxidation, that could be due to the expansion of the 3PIs [18].
After passing through the maximum value, the current decayed
with time. Until about 4 h later, the current levelled out at about
0.7A. The decreasing again of the current occurred after about 16 h.
Finally the current decreased to about 0.3A in 22 h. Curves of cur-
rent against time for an experiment are closely related to the rate of
reaction, with a faster rate corresponding to a higher current. And
the rate of reaction could be again related with the phase compo-
sitions of electrode. For the experiments here, the sudden change of
the current tends to imply the possible change of the phase com-
positions of cathode pellets. On the I-t curve, two peaks (A, B), two
inflection points (C and D) and two shoulders (E and F) appeared
successively. Thereby these time points were chose as the termi-
nating time points of polarisation.
Table 1 lists the experiment conditions and the corresponding
results. After about 6min of polarisation, the sample (S1) was
found to be predominantly composed of CaTiO3 and Ti2O3. Magneli
phases, such as Ti4O7 and Ti3O5, were not detected here. But they
were detected in the sample quenched after 10min of polarisation
when cutting down the voltage to 2.8V, as shown in Fig. S2 (in the
supplementary materials). Prolonging the polarisation time to
about 10min, the sample (S2) consisted of CaTiO3, Ti2O3 and little
TiO; to ~37min, the sample (S3) was mainly comprised of CaTiO3,
Ti-O solid solution (Ti2O) and little CaTi2O4; and to 4 h, the sample
(S4) consisted mainly of Ti-O solid solution (Ti3O) and little CaTiO3.
Fig. 2c shows the XRD patterns of the partially reduced samples.
During around 6e37min of polarisation, CaTiO3 was much more
than other phases and decreased slightly with electrolysis times;
titanium suboxides gradually disappeared; Ti-O solid solution and
CaTi2O4 began to form. During around 0.6e4 h of polarisation,
CaTi2O4 disappeared completely, CaTiO3 diminished dramatically
and Ti-O solid solution (Ti3O) increased substantially. After about
16 h of polarisation, the sample (S5) was reduced into Ti metal but
the oxygen content was up to 1.6 wt%. Prolonging the polarisation
time to 22 h, the oxygen content of the sample (S6) decreased to
about 0.6 wt%. At the same time, it was found that the pellets
became thinner and thinner with the step-by-step reduction of the
pellets.
Anhydrous CaCl2 (10005861 Sinopharm) was dried for 4 h at
300 ꢀC under vacuum before use. The CaO content of the dried salt
was estimated by acid-base titration method and was found to be
about 0.045% wt%. Sodium titanite (98 wt%, Na2Ti3O7, Shanghai
Dian Yang Industry co. Ltd) powder was pressed into pellets under
the axial pressure of 40 MPa. Then these pellets were sintered for
4 h at 900 ꢀC. The sintered pellets (3.00 0.03g, 28.7 mm diameter
and 1.83 mm thickness), whose porosity was measured by Mercury
Porosimeter and is about 20%, were attached to a molybdenum
wire (2 mm diameter, 700 mm long) that was employed as the
current collector of cathode. A high-purity graphite rod (10 mm
diameter, 80 mm long) was used as anode, which was connected
with a stainless steel collector. Alumina crucibles (80 mm inner
diameter and 120 mm high) were used as containers for molten
CaCl2. 500g dried CaCl2 salt was taken in an alumina crucible and
placed in a leak tight reactor vessel. Before heated, the vessel was
evacuated to 10Pa and then was replenished with high-purity Ar
gas. This process was repeated twice. Then Ar gas passed contin-
uously through the reactor vessel at 2L/min until the end of each
experiment. The vessel was heated by silicon carbide rods. When
the salt was heated to 900 ꢀC, the graphite anode and the sodium
titanate pellet were inserted into molten salt through electrode
holes at the furnace cover. The applied voltage between the cathode
and anode was 3.1V.
The partially reduced samples were obtained after different
electrolysis times. After each experiment, the sample was moved
out from molten salt and cooled down rapidly at argon atmosphere.
Then they were rinsed with flowing tap water to remove the so-
lidified molten salt clinging to the surface of samples and were
dried in vacuum oven at room temperature. The microstructure and
constituents of samples were analysed by SEM and EDS (S-4800,
Hitachi). The phase compositions of the partially reduced samples
were analysed by X-ray diffraction analyser (X'Pert Pro
MPDDY2094, PANalytical B.V with CuKa radiation). The content of
oxygen in final product was tested by oxygen/nitrogen analyser
(TC500 LECO).
3. Results
Fig. 1a presents the XRD spectrum of the precursor material
after sintering for 4 h at 900 ꢀC. The main phase is Na2Ti3O7. The
SEM image (in Fig. 1b) shows the Na2Ti3O7 present rod-like parti-
cles ranging from 1 mm to 3 mm in length. Fig. 1d is the sectional
SEM image of a sodium titanite (Na2Ti3O7) pellet after 4 h of sub-
merging in molten CaCl2 at 900 ꢀC. The particles were still rod-like,
but the phase compositions of the pellets had become a mixture of
TiO2, CaTiO3 and little Ca2Ti2O6 as shown in Fig. 1c. At the same
time, it was observed that the space between the particles was
filled with electrolyte. Fig. S1 (in the supplementary materials)
shows the X-ray diffraction patterns of Na2Ti3O7 pellets submerg-
ing in CaCl2 melted salt for different times at 900 ꢀC. Na2Ti3O7 re-
acts with molten CaCl2 easily to form Ca2Ti2O6, TiO2 and NaCl in
~30min. And then in the next 3.5 h, Ca2Ti2O6 was transformed into
CaTiO3 completely. The involved reactions are as follows.
Fig. 2b shows the SEM images of the cross sections of S1-S6. The
samples began to delaminate after 6min of polarisation. The outer
layer is the reduced region and the core is the less reduced region.
As the polarisation time went on, the outer layer grew thick and the
core shrunk progressively. The pellets were gradually reduced from
outside to inside. After 4 h of polarisation, the reduced region was
found to be porous and sintered slightly, and the less reduced re-
gion was compact. During 4e22 h of polarisation, the pellet was
sintered fast and finally became a compact metal sheet of about
0.7 mm in thickness.
2Na2Ti3O7 þ 2CaCl2 / Ca2Ti2O6 þ 4TiO2 þ 4NaCl
Ca2Ti2O6 / 2CaTiO3
(3)
(4)
When a voltage of 3.1V was applied between a Na2Ti3O7 cathode
and a graphite anode in molten CaCl2, the Na2Ti3O7 pellet trans-
formed into a mixture of CaTiO3, Ti2O3 and little Ca2Ti2O6 in ~ 6min
as shown in Fig. 2c(S1). From the experiments, it was found that

238
K. Liu et al. / Electrochimica Acta 318 (2019) 236e243
Fig. 1. (a) X-ray diffraction spectrum of the Na2Ti3O7 pellets after sintering in air at 900 ꢀC for 4 h, showing Na2Ti3O7 ((A), ICDD: 00-031-1329), little Na2Ti5O11 ((B), ICDD: 00-011-
0289) and Na4Ti5O12 ((C), ICDD: 01-075-2497); (b) SEM image of the Na2Ti3O7 pellets after sintering in air at 900 ꢀC for 4 h; (c) X-ray diffraction spectrum of the Na2Ti3O7 pellets
after submerging in molten CaCl2 at 900 ꢀC for 4 h, showing TiO2 ((A), ICDD: 01-073-2224), CaTiO3 ((B), ICDD: 01-076-2400), little Ca2Ti2O6 ((C), ICDD: 00-040-0103); and the non-
ascribed peaks mainly corresponds to CaCl2$4H2O and CaCl2$6H2O). (d) the sectional SEM image of the Na2Ti3O7 pellets after submerging in molten CaCl2 at 900 ꢀC for 4 h
4. Discussion
etc. [3,7]. But compared the reduction with TiO2, more CaTiO3
formed during the reduction process of Na2Ti3O7. Some CaTiO3 was
reduced according to reaction (7)e(9). The rest was reduced
directly into Ti-O solid solution as reaction (11). The proof is that
only CaTiO3 and Ti-O solid solution were detected in the sample
(S4) after 4 h of polarisation. From these reactions, the whole
reduction process of Na2Ti3O7 pellets can be divided into three
stages. The first stage is the transformation of Na2Ti3O7 to calcium
titanates (CaTiO3 and CaTi2O4) and titanium sub-oxides (Ti4O7,
Ti3O5, Ti2O3 and TiO) and occurred during ~30min of polarisation.
The involved are reaction (3)e(7). This stage is mainly the insertion
of Ca2þ into pellets and no CaO or O2ꢁ ions are released, which
means the amount of oxygen in pellet during the first is changeless.
This stage processed very quickly. The second stage is the trans-
formation of calcium titanates and TiO formed in first stage to Ti-O
solid solution and occurred during 0.2 ~ 4 h of polarisation. During
this stage, amounts of CaO are released and reaction (8)e(11) are
involved. From the I-t curves in Fig. 2, the reduction slowed down
gradually with polarisation time during the second stage. The third
stage is the deoxidation of Ti-O solid solution and processed during
about 4e22 h of polarisation. Reaction (12) is involved. This stage
processed very slowly. At the same time, it was found that the
thicknesses of the pellets just decreased by about 2% in the first two
stages of the reduction. But during around 4e22 h of polarisation,
the thickness of the pellets decreased to about 40% of the initial.
Combined with the XRD analysis, the sinter occurred mainly during
the transformation of Ti-O solid solution to Ti metal. From our
unreported data, though the sample was reduced into Ti6O (Ti-O
From the XRD analysis of the partially reduced samples, the
possible reactions occurred on cathodes during polarisation are as
followed [3,7].
5Na2Ti3O7 þ 7Ca2þ þ 4eꢁ / 2Ti4O7 þ 7CaTiO3 þ 10Naþ
4Ti4O7 þ Ca2þ þ 2eꢁ / 5Ti3O5 þ CaTiO3
3Ti3O5 þ Ca2þ þ 2eꢁ / 4Ti2O3 þ CaTiO3
2Ti2O3 þCa2þ þ 2eꢁ/ 3TiO þ CaTiO3
CaTiO3 þTiO / CaTi2O4
(5)
(6)
(7)
(8)
(9)
CaTi2O4 þ Ca2þþ 2eꢁ / 2TiO þ 2CaO
(10)
(11)
(12)
(13)
(14)
TiO þ (1-
d
)Ca2þ þ 2(1-
d
)eꢁ / Ti[O]d þ (1-
d)CaO
CaTi2O4 þ (3-2
d
)Ca2þ þ 2(3-2
d
)eꢁ / 2Ti[O]d þ (4-2
d)CaO
CaTiO3 þ (2-
d
)Ca2þ þ 2(2-
d
)eꢁ / Ti[O]d þ (3-
d)CaO
Ti[O]d þ
d
Ca2þ þ 2
d
eꢁ / Ti þ
dCaO
The reduction process is similar with that of TiO2 the electro-
reduction mechanism of which was determined by Schwandt, C.

K. Liu et al. / Electrochimica Acta 318 (2019) 236e243
239
Fig. 2. (a) Current vs. time curve of an electro-deoxidation experiment under a voltage of 3.1V at 900 ꢀC. Electrolyte: calcium chlotide; cathode: Sodium titanate pellets (3 0.05g);
anode: graphite rod (10 mm diameter, 80 mm long); anode-to-cathode distance: 6 cm; atmosphere: Ar gas; (b) the sectional SEM images of the samples quenched after different
times of electrolysis, showing the decreasing thicknesses with the extension of electrolysis times: the sample (S1) after 6min of electrolysis: 1.83 mm thickness; the sample (S2)
after 10min of electrolysis: 1.83 mm thickness; the sample (S3) after 37min of electrolysis: 1.83 mm thickness; the sample (S4) after 4 h of electrolysis: 1.79 mm thickness; the
sample (S5) after 16 h of electrolysis: 0.9 mm thickness; and the sample (S6) after 22 h of electrolysis: 0.7 mm thickness; (c) X-ray diffraction spectrums of the samples quenched
after different times of electrolysis, the sample (S1) after 6 min of electrolysis showing CaTiO3 ((A), ICDD:01-076-2400), Ti2O3((B), ICDD: 01-071-1053) and Ca2Ti2O6 ((C), ICDD: 00-
040-0103); the sample (S2) after 10 min of electrolysis showing CaTiO3 ((A), ICDD:01-076-2400), Ti2O3((B), ICDD: 01-071-1053) and TiO ((C), ICDD: 00-008-0117); the sample (S3)
after 37 min of electrolysis showing CaTiO3 ((A), ICDD:01-076-2400), Ti2O ((B), ICDD: 01-073-1116) and CaTi2O4 ((C), ICDD: 01-074-1952); the sample (S4) after 4 h of electrolysis
showing CaTiO3 ((A), ICDD:01-076-2400) and Ti3O ((B), ICDD: 01-073-1583); the sample (S5) after 16 h of electrolysis showing Ti ((A), ICDD: 01-089-4893) and TiC ((B), ICDD: 03-
065-0242); and the sample (S6) after 24 h of electrolysis showing Ti ((A), ICDD: 01-089-4893).
Table 1
A lists the experiment conditions and the corresponding results.
Test No.
Porisity
Voltage
Time
Results
Thickness of the reduced pellets (mm)
Phases(the size of the particles*)
S1
S2
S3
S4
S5
S6
S7
23%
3.1V
6 min
10min
37min
4 h
16 h
22 h
1.8
1.8
1.8
1.75
0.9
0.7
CaTiO3(5.8 nm), Ti2O3, Ca2Ti2O6
CaTiO3(28.6 nm), Ti2O3, TiO
CaTiO3(41.5 nm), Ti2O(44.6 nm), CaTi2O4
Ti3O(50.2 nm), CaTiO3
Ti(20.3 nm), TiC
Ti(23.8 nm)
2.8V
10min
CaTiO3(31.0 nm), Ti4O7, Ti3O5
* The particle sizes were estimated from the XRD patterns according to Scherrer equation.
solid solution containing about 5% oxygen) under a more positive
potential at 900 ꢀC, it still was not sintered visually. That indicates
that the Ti-O solid solution with a high content of oxygen (more
than 5 wt%) is hard to be sintered on macro point.
CaCl2 at 900 ꢀC is about 12.6 wt% [19]. Excess CaO will precipitate
and occupies the pores in pellets. That will block up the channel for
electrolyte (molten CaCl2) entering the inner of the pellets. Ac-
cording to the reaction (8)e(11), the lack of Ca2þ will slow down the
reduction. That might be the reason leading to a lower reduction
rate in second stage. Fig. 3c shows the SEM image of the outer layer
of S4 and Fig. 3d(II) shows the XRD pattern of the out layer of S4,
which indicated the outer layer was reduced intoTi-O solid solution
(Ti3O). From the SEM image, the finer Ti-O solid solution particles
were sintering together. Meanwhile, some macrovoid form.
Fig. 2b(S4) also shows that the out layer of S4 presents a porous
structure on a macroscale. Those indicated that the electrolyte had
a good diffusion channel in the out layer of the samples during
polarisation.
Fig. 3a shows the SEM image of the core of S3. From the EDS
analysis, the cracked region is the solidified electrolyte and the fine
particle region is mainly composed of calcium titanates and/or Ti-O
solid solution. Fig. 3b shows the SEM image of the core of S4. The
EDS analysis indicates that there are amounts of CaO, which agrees
well with the result of the XRD analysis (Fig. 3d(I)) obtained from
the core of S4 that was tested without the treatment of washing or
pickling. By comparison, the CaCl2 (which was converted to CaClOH
because of its famous water absorption) is very little. So much CaO
remaining in the inner layer of S4 indicates that the reduction rate
of calcium titanates is much higher than the diffusion rate of CaO
from inside to outside of the pellets. The phase diagram of CaO-
CaCl2, as shown in Fig. S3, displays the solubility of CaO in molten
With the prolongation of the polarisation time to 16 h, the pellet
was reduced and sintered into a compact metal sheet of about
0.9 mm thickness. The thickness decreased by about 50%. Fig. 4aed

240
K. Liu et al. / Electrochimica Acta 318 (2019) 236e243
Fig. 3. (a) The sectional SEM image of the core of the sample (S3) quenched after 37 min of electrolysis, Zone I corresponding to the solidified electrolyte and Zone II corresponding
to a mixture of calcium titanates and/or Ti-O solid solution; (b) the sectional SEM image of the core of the sample (S4) quenched after 4 h of electrolysis, showing amounts of CaO;
(c) the sectional SEM image of the out layer of the sample (S4) quenched after 4 h of electrolysis, showing Ti-O solid solution and macropore; (d) the X-ray diffraction spectrums of
the core and the out layer of the sample (S4) quenched after 4 h of electrolysis, the core (I) showing CaO ((A), ICDD: 01-075-0264), CaTiO3 ((B), ICDD: 01-078-1013), Ti2O ((C), ICDD:
01-073-1116) and CaClOH ((D), ICDD: 00-036-0983), the non-ascribed peaks corresponding to Ca(OH)2; the out layer (II) showing Ti3O ((E), ICDD; 01-076-1644).
Fig. 4. (a) The sectional SEM image of the outermost layer of the samples quenched after 16 h of electrolysis; (b) the sectional SEM image of the secondary outer layer of the sample
quenched after 16 h of electrolysis; (c) the sectional SEM image of the tertiary outer layer of the sample quenched after 16 h of electrolysis; (d) the sectional SEM image of the core of
the sample quenched after 16 h of electrolysis; (e) the sectional SEM images of the sample retrieved after 22 h of electrolysis, showing Ti.

K. Liu et al. / Electrochimica Acta 318 (2019) 236e243
241
show the SEM images of S5 from outside to inside. The sample has a
sponge-like microstructure. From inside to outside of the sample,
the particles were sintered more and more severely and the oxygen
content was lower and lower. That proves that the higher the ox-
ygen content of Ti-O solid solution particles is, the harder it is to be
sintered. From Fig. 2b, the thickness of the pellet decreased further
about 0.7 mm after about 22 h of polarisation. At this time, the
sample became very compact as shown in Fig. 4e. The EDS analysis
indicates the pellet has completely transformed into Ti metal.
From the above discussion, a schematic of the progress of
electrochemical reduction occurring on Na2Ti3O7 pellets placed in
CaCl2 melt has been depicted in Fig. 5. Driven by a potential which
is close but positive to calcium metal deposition, Na2Ti3O7 pellets
can be reduced into Ti metal step by step. The reduction could be
divided into three stages. In the first stage of electrolysis, Ca2þ ions
from electrolyte are inserted into Na2Ti3O7 particles leading to the
formation of CaeTieO compounds and titanium sub-oxides. The
involved are reaction (3)e(7) whose potentials are more positive.
At the same time, there are enough Ca2þ ions in the pellets for these
reactions in this stage. Thus this process goes very fast. On
continued polarisation, the formed calcium titanates and titanium
sub-oxides (the main is TiO) are reduced into Ti-O solid solution
from outside to inside. Meanwhile, amounts of CaO are released.
The formed CaO could not move out timely but precipitates inside
the pellets. That blocks up the channel for electrolyte entering the
inner of the pellets, leading to the lack of Ca2þ ions for reaction
(8)e(11). Thus the reduction at second stage goes slowly. During
the last stage, the formed Ti-O solid solution is further deoxidised
to form Ti metal particles which will be sintered together quickly.
From the experiments, this is the most time-consuming process.
The main reason might be that the diffusion of oxygen in Ti-O solid
solution is very slow.
pellet and it could be described by the Cottrell equation (Eq. (13))
[20].
ꢀ
ꢁ
1=2
D
i ¼ nFCA
(15)
pt
where, i, current; n, number of electrons involved; F, Faraday con-
stant; D, diffusion coefficient; C, concentration of species in the bulk
and t, time. Fig. 6a shows the i-tꢁ1/2 curves of four electrolysis ex-
periments which correspond to Test.S3-S6 shown in Table 1. The
curves have similar trends and can be divided into three zones. The
zone I corresponds to the double layer charging and the accumu-
lation of adatoms on the electrode which occurred in 1 min after
potential switch; the zone II corresponds to the expansion of the
3PIs; and the zone III corresponds to the reduction of CaeTieO
compounds into Ti-O solid solution, the rate of which is controlled
by mass transfer of Ca2þ ions. The relationship between i and tꢁ1/2
in zone III is approximatively linear, which indicates that the sec-
ond stage can be described by the Cottrell equation. After linear
fitting towards the curves in zone III, the slopes can be obtained and
are about 43, 48, 47 and 55 respectively.
In addition, if considering the expansion of the 3PIs in zone II as
a nucleation and growth process of Ti or Ti-O solid solution, the
electro-deoxidation could be explained be Scharifker-Hill model
which is an effective diagnostic tool for the evaluation of the
dependence of current on time for nucleation with diffusion-
controlled growth [21,22]. According to the models, there are two
limiting nucleation mechanisms which are given as follows:
For the instantaneous nucleation:
ꢀ
ꢁ
2
i
im
1:9542
t=tm
¼
f1 ꢁ exp½ ꢁ 1:2564ðt=tmÞꢂg2
(16)
From the above, the second stage is an electrochemical process
controlled by the mass transfer of O2ꢁ ions to the outside of the
For the progressive nucleation:
Fig. 5. A schematic showing the progress of electro-deoxidation of Na2Ti3O7 pellets in molten CaCl2. At first, Na2Ti3O7 particles are decomposed into calcium titanates (including
CaTiO3, Ca2Ti2O6 and CaTi2O4) and titanium suboxides (Ti4O7, Ti3O5, Ti2O3 and TiO). Then all titanium suboxides and slight amounts of calcium titanates are reduced into Ti-O solid
solution. Meanwhile slight amounts of CaO are released. Next, all the calcium titanates were reduced into Ti-O solid solution. Meanwhile amounts of CaO form in the pellets, which
could block up the channel for electrolyte entering the inner of the pellets and influence the reduction rate. The last step is that the Ti-O solid solution is further deoxidised into Ti
metal. Meanwhile the pellets are sintered and shrink greatly.

242
K. Liu et al. / Electrochimica Acta 318 (2019) 236e243
ꢀ
ꢁ
2
n
h
io
2
i
im
1:2254
t=tm
¼
1 ꢁ exp ꢁ2:3367ðt=tmÞ2
(17)
where im and tm are the maximum coordinates of i transient. Fig. 6b
show the (i/im)2 - t/tm curves of the four experiments, comparing
with the curves calculated from Eqs. (14) and (15). The four
experimental curves of the electro-deoxidation have similar trends
and follow mostly the progressive nucleation, which agrees well
with C. Osarinmwian's study [22].
Fig. 6c shows the (i/im)2 - t/tm curves of Expt.3 when t/tm is in
20e160 range. When t/tm is greater than 30, which corresponds to
the third stage of the electro-deoxidation, the curve of the exper-
iment deviates obviously from those calculated from Eqs. (14) and
(15). That is mainly due to the electro-deposition of Ca metal and C
on the surface of cathode and the low diffusion rate of O in Ti-O
solid solution.
5. Conclusion
In present paper, Ti metal was prepared successfully using the
Na2Ti3O7 pellets as precursor through FFC Cambridge Process. After
22 h of polarisation at 900 ꢀC, a compact Ti metal sheet was ob-
tained. Its oxygen content was about 0.6 wt%. The reduction pro-
ceeded via sequential formation of CaTiO3, titanium sub-oxides
(such as Ti4O7, Ti3O5, Ti2O3 and TiO), CaTi2O4, Ti-O solid solution
and Ti. During the reduction, the pellets sintered seriously, which
occurs mainly during the transformation of Ti-O solid solution to Ti
metal. Besides, it was found that Ti-O solid solution containing
more than 5% oxygen is hard to be sintered, which is advantageous
to the out diffusion of CaO during the second stage of reduction. In
spite of this, the second stage is one of the control steps of the
reduction because of the low diffusion rate of CaO in pellets. That
might be improved by increasing the porosity of precursors.
Another control step during the reduction is the transformation of
Ti-O solid solution to Ti metal. That is due to the low diffusion rate
of oxygen in Ti-O solid solution.
Acknowledgements
The authors sincerely acknowledge financial support from the
National Natural Science Foundation of China (51674076).
Appendix A. Supplementary data
Supplementary data related to this article can be found at
https://doi.org/10.1016/j.electacta.2019.06.072.
References
[1] G.Z. Chen, D.J. Fray, T.W. Farthing, Direct electrochemical reduction of tita-
nium dioxide to titanium in molten calcium chloride, Nature 407 (2000) 361.
[2] K. Dring, R. Dashwood, D. Inman, Voltammetry of titanium dioxide in molten
calcium chloride at 900 ꢀC, J. Electrochem. Soc. 152 (2005) E104.
[3] C. Schwandt, D.J. Fray, Determination of the kinetic pathway in the electro-
chemical reduction of titanium dioxide in molten calcium chloride, Electro-
chim. Acta 51 (2005) 66.
[4] K. Jiang, X. Hu, M. Ma, D. Wang, G. Qiu, X. Jin, G.Z. Chen, Perovskitization"-
assisted electrochemical reduction of solid TiO2 in molten CaCl2, Angew Chem.
Int. Ed. Engl. 45 (2006) 428.
Fig. 6. (a) Current vs. inverse square root of polarisation time curves of four elec-
trolysis experiments which correspond to Test.S3-S6 shown in Table 1; (b) Normalized
2
transients (i/im
)
vs. t/tm curves when t/tm is in 0e10 range which corresponds to the
first two stages of an electro-deoxidation experiment. The full lines correspond to the
curves obtained from the four electrolysis experiments; the black dot dash line cor-
responds to the calculated curve for the instantaneous nucleation and diffusion-
limited growth according to Eq. (14); and the red dot dash line corresponds to the
calculated curve for progressive nucleation and diffusion-limited growth according to
[5] M. Ma, D. Wang, W. Wang, X. Hu, X. Jin, G.Z. Chen, Extraction of titanium from
different titania precursors by the FFC Cambridge process, J. Alloy. Comp. 420
(2006) 37.
2
Eq. (15). (c) Normalized transients (i/im
)
vs. t/tm curves when t/tm is in 10e160 range
which corresponds to the last stages of an electro-deoxidation experiment. The full line
correspond to the curve obtained from Expt.3; the black dot dash line corresponds to
the calculated curve for the instantaneous nucleation and diffusion-limited growth
according to Eq. (14); and the red dot dash line corresponds to the calculated curve for
progressive nucleation and diffusion-limited growth according to Eq. (15). (For inter-
pretation of the references to colour in this figure legend, the reader is referred to the
Web version of this article.)
ꢀ
[6] R.L. Centeno-Sanchez, D.J. Fray, G.Z. Chen, Study on the reduction of highly
porous TiO2 precursors and thin TiO2 layers by the FFC-Cambridge process,
J. Mater. Sci. 42 (2007) 7494.
[7] C. Schwandt, D.T.L. Alexander, D.J. Fray, The electro-deoxidation of porous
titanium dioxide precursors in molten calcium chloride under cathodic po-
tential control, Electrochim. Acta 54 (2009) 3819.

K. Liu et al. / Electrochimica Acta 318 (2019) 236e243
243
[8] K. Dring, Direct electrochemical reduction of titanium dioxide in molten salts,
Key Eng. Mater. 436 (2010) 27.
[9] D.T.L. Alexander, C. Schwandt, D.J. Fray, The electro-deoxidation of dense ti-
tanium dioxide precursors in molten calcium chloride giving a new reaction
pathway, Electrochim. Acta 56 (2011) 3286.
[10] X.C. Lang, H.W. Xie, X.Y. Zou, P.H. Kim, Y.C. Zhai, Investigation on direct
electrolytic reduction of the CaTiO3 compounds in molten CaCl2-NaCl for the
production of Ti, Adv. Mater. Res. 284 (2011) 282e286.
[11] D.S. Maha Vishnu, N. Sanil, L. Shakila, R. Sudha, K.S. Mohandas, K. Nagarajan,
Electrochemical reduction of TiO2 powders in molten calcium chloride, Elec-
trochim. Acta 159 (2015) 124.
[15] A.A. Nayl, N.S. Awwad, H.F. Aly, Kinetics of acid leaching of ilmenite decom-
posed by KOH Part 2. Leaching by H2SO4 and C2H2O4, J. Hazard Mater. 168
(2009) 793.
[16] T. Xue, L. Wang, T. Qi, J. Chu, J. Qu, C. Liu, Decomposition kinetics of titanium
slag in sodium hydroxide system, Hydrometallurgy 95 (2009) 22.
[17] A. Roine, HSC Chemistry, Version 6.0, Outokumpu Research Oy, Pori, Finland.
[18] G.Z. Chen, E. Gordo, D.J. Fray, Direct electrolytic preparation of chromium
powder, Metall. Mater. Trans. B Process Metall. Mater. Process. Sci. 35 (2004)
223.
[19] D.A. Wenz, I. Johnson, R.D. Wolson, CaCl2-rich region of the CaCl2-CaF2-CaO
system, J. Chem. Eng. Data 14 (1969) 250.
[12] D.S. van Vuuren, A critical evaluation of processes to produce primary tita-
nium, J. S. Afr. Inst. Min. Metall 109 (2009) 455.
[20] A.M. Abdelkader, K.T. Kilby, A. Cox, D.J. Fray, DC voltammetry of electro-
deoxidation of solid oxides, Chem. Rev. 113 (2013) 2863.
[13] Y. Liu, T. Qi, J. Chu, Q. Tong, Y. Zhang, Decomposition of ilmenite by concen-
trated KOH solution under atmospheric pressure, Int. J. Miner. Process. 81
(2006) 79.
[14] A.A. Nayl, H.F. Aly, Acid leaching of ilmenite decomposed by KOH, Hydro-
metallurgy 97 (2009) 86.
[21] B. Scharifker, G. Hills, Theoretical and experimental studies of multiple
nucleation, Electrochim. Acta 28 (1983) 879.
[22] C. Osarinmwian, I.M. Mellor, E.P.L. Roberts, Electro-deoxidation modelling of
titanium dioxide to titanium, Electrochim. Acta 209 (2016) 95.