Direct synthesis of Mg-Ni compounds from their oxides

Article abstract of DOI:10.1016/j.jallcom.2010.05.071

A study was carried out on the synthesis of Mg-Ni compounds as well as on the extraction of pure Ni and Mg from their oxides using the method of electro-deoxidation. The oxides sintered at 1200 °C were in the form of discrete phases NiO and MgO, suitably proportioned to yield Ni, MgNi ₂, Mg₂Ni and Mg. The oxides were electrolyzed at 3.2 V in a eutectic mixture of CaCl₂-NaCl solution maintained at a constant temperature (900-600 °C), using a graphite anode. The study has shown that NiO rich mixture, MgO:NiO = 1:2, can be reduced successfully to metallic state. Some loss of Mg was apparent in the latter, with the result that the product was far from the target composition MgNi₂. The electroreduction of MgO rich mixtures was difficult to achieve. The mixture MgO:NiO = 2:1 when electrolyzed at 725 °C for 24 h, could be reduced to metallic phases only in small proportions (18 wt.%). In pure MgO, no trace of reduction was observed during the electrolysis at 600 °C. Difficulties in the electroreduction of MgO and MgO rich mixtures were partially attributed to low conductivity of MgO.

Full text of DOI:10.1016/j.jallcom.2010.05.071

Journal of Alloys and Compounds 504 (2010) 134–140
Contents lists available at ScienceDirect
Journal of Alloys and Compounds
journal homepage: www.elsevier.com/locate/jallcom
Direct synthesis of Mg–Ni compounds from their oxides
∗
˙
Serdar Tan, Kadri Aydınol, Tayfur Öztürk , Ishak Karakaya
Department of Metallurgical and Materials Engineering, Middle East Technical University, Ankara 06531, Turkey
a r t i c l e i n f o
a b s t r a c t
Article history:
A study was carried out on the synthesis of Mg–Ni compounds as well as on the extraction of pure
Ni and Mg from their oxides using the method of electro-deoxidation. The oxides sintered at 1200 ◦C
were in the form of discrete phases NiO and MgO, suitably proportioned to yield Ni, MgNi2, Mg2Ni and
Mg. The oxides were electrolyzed at 3.2 V in a eutectic mixture of CaCl2–NaCl solution maintained at a
constant temperature (900–600 ◦C), using a graphite anode. The study has shown that NiO rich mixture,
MgO:NiO = 1:2, can be reduced successfully to metallic state. Some loss of Mg was apparent in the latter,
with the result that the product was far from the target composition MgNi2. The electroreduction of
MgO rich mixtures was difficult to achieve. The mixture MgO:NiO = 2:1 when electrolyzed at 725 ◦C for
24 h, could be reduced to metallic phases only in small proportions (18 wt.%). In pure MgO, no trace of
reduction was observed during the electrolysis at 600 ◦C. Difficulties in the electroreduction of MgO and
MgO rich mixtures were partially attributed to low conductivity of MgO.
Received 13 December 2009
Received in revised form 9 May 2010
Accepted 15 May 2010
Available online 27 May 2010
Keywords:
Electroreduction
Magnesium oxide
Nickel oxide
Electrical conductivity
Mg2Ni
MgNi2
© 2010 Elsevier B.V. All rights reserved.
1. Introduction
oxides such as SiO2 was also difficult to reduce and required the
number of contacting points in the cathode to be increased [8].
Solid state electroreduction of oxides has been an area of intense
research during the last decade [1–4]. The method was used not
only to extract metals from their oxides [1] but also to synthesize
alloys and compounds from the oxide mixtures [2]. In this method,
pellets of oxide powders used as the cathode were deoxidized by
the application of a potential against graphite without decompos-
ing the electrolyte, a molten salt that is capable of dissolving and
transporting oxygen ions. During electrolysis, while oxygen ions
are removed from the pellets and transported to the anode to form
either CO and/or CO2, metallic constituents are left behind in the
cathode, which via in situ reactions may yield alloys or compounds.
The method has the advantage that when inert anodes are used,
the electrolysis leads to oxygen generation rather than CO or CO2
emission [5].
A number of factors affect the ease with which the reduction is
achieved in the electro-deoxidation process. It has been reported
that, oxides of low reduction potentials, e.g. NiO, Cr2O3, Fe2O3, are
deoxidized faster than others, e.g. rare earth oxides [6]. The ease
of reduction is also affected by the oxygen diffusivity in the solid
state. Oxides such as TiO2 and ZrO2 are reduced relatively easily in
the early stages, but as pointed out by Chen et al. [7], once the
metallic state is achieved, the rate slows down, since the diffu-
sion of oxygen is extremely slow in the metallic lattices. Insulating
Here the reduction is claimed to proceed via the propagation of
conductor–insulator–electrolyte triple interline [9,10]. Similarly,
the deoxidation of some oxides, such as MgO and Al2O3, has been
found to be complicated due to difficulties in the initial metal-
lization of the oxide pellets as well as due to the fact that the
corresponding metals have low melting points [11].
The reduction of oxide mixtures is often more complicated than
their pure counterparts. The mixtures, such as Tb2O3–NiO [12],
Fe2O3–TiO2 [13] and TiO2–NiO [14], have been successfully deox-
idized to yield the respective intermetallic compounds, i.e. TbNi5,
FeTi, and TiNi. The complication often arises because the oxides are
reduced at different rates which may give rise to separate individual
phases rather than to the compounds themselves. Yong et al. [14]
reported that the use of a higher reduction potential may reduce
the differences in the reduction rates and could thus promote the
compound formation. In this context, the pre-compounding of the
oxide mixtures as well as the optimization of the porosity in the
pellets could be beneficial to have more complete reduction of the
oxides to the required stoichiometry. The reduction of the oxide
mixtures also has its advantages. For instance, compounds, such as
Ni2MnGa [15] and Nb3Sn [16], have been successfully synthesized
in the solid state at temperatures higher than the melting points of
some of their constituent metals. Furthermore, Qui et al. [17] points
out that the reducibility of the oxide mixture could be improved
by fast reduction of one of the oxide phases and the formation of
metallic particles in situ may enhance the reduction of the other
oxide constituents.
∗
Corresponding author. Tel.: +90 312 210 5935; fax: +90 312 210 2518.
E-mail address: ozturk@metu.edu.tr (T. Öztürk).
0925-8388/$ – see front matter © 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.jallcom.2010.05.071

S. Tan et al. / Journal of Alloys and Compounds 504 (2010) 134–140
135
Fig. 2. Schematic drawing of an electrolytic cell used in the deoxidation experi-
ments.
the other (working electrodes) for electro-deoxidation, Fig. 2. Anode-to-cathode
distance in both pairs was 25 mm. Anodes were graphite rods, 13 mm of diameter
and 40 mm in length, and were connected to a stainless steel current collector. The
cathodes were stainless steel wire of 4 mm in diameter, which in the case of work-
ing electrode was connected to the oxide pellet. The reactor was sealed, allowing
electrolysis in an argon atmosphere (99.995% purity) maintained at a flow rate of
150–250 ml/min.
A eutectic mixture of CaCl2–NaCl was used as the electrolyte, which has a melt-
ing point of about 505 ◦C. Approximately 1 kg of electrolyte was used for each
experiment. The salt mixture was heated slowly to the electrolysis temperature to
reduce its moisture content. The electrolyte was further purified by pre-electrolysis
using the auxiliary electrodes at a potential of 3.0 V for a minimum of 6 h. Following
the pre-electrolysis, the auxiliary electrodes were lifted above the electrolyte, and
the electrolysis was initiated by immersing the working electrodes, i.e. oxide pellet
(cathode) and a fresh graphite rod (anode) into the salt bath. During electrolysis,
current-time data was collected by a computer connected to the power supply.
At the end of each experiment, the electrodes were disconnected from the elec-
trical supply and both the graphite anode and the reduced sample were removed
from the electrolyte by positioning them above the melt. After cooling to room tem-
perature under a continuous flow of argon gas, the reactor lid was opened and the
sample with its stainless steel connector was removed. The sample was washed in
a hot methanol–ethanol–water mixture and all undissolved material was collected
for XRD and SEM analysis.
Fig. 1. SEM images of oxide powders; (a) NiO and (b) MgO used in the experiments.
Volumetric porosities of the oxide pellets after sintering were determined from
geometric dimensions and the mass of the pellets, making use of specific density
values of the constituent oxides. Phase make-up of the samples was determined
via X-ray diffraction. The quantity of phases was determined following Rietveld
refinements of the powder diffraction patterns using the software MAUD [19].
The current work deals with synthesis of Mg–Ni compounds as
well as the extraction of pure Ni and Mg from their oxides. The study
is of particular interest since the system involves oxides, MgO and
NiO, which have widely different physical, such as electrical con-
ductivity and oxygen diffusivity, and thermodynamic properties.
Therefore, electro-deoxidation of these oxides, either on their own
or as mixtures, are of considerable significance. Moreover, Mg2Ni,
one of the target compounds has a considerable potential as hydro-
gen storage alloy [18].
3. Results and discussion
The oxides following the sintering treatment yielded pellets
with porosity values varying between 37 and 57%. An X-ray diffrac-
togram of a sintered pellet of mixed oxide MgO:NiO = 1:2 is given
in Fig. 3(a). Since MgO and NiO have the same crystal structure
and very similar lattice parameters, the peaks are very close to
one another showing considerable overlapping. Still, the peaks of
MgO and NiO are differentiated from each other, indicating that
the mixed oxides have a two-phase structure rather than a single-
phase solid solution. Considering that the sintering was carried out
at 1200 ◦C this is somewhat surprising and could be due to insuf-
ficient mixing of the oxides in the SPEX mill. This has been further
verified via SEM imaging in backscattered mode. Regions of two
different contrasts were apparent, see Fig. 3(b), in compliance with
the presence of two-phase structure. Grains in the sintered samples
had sizes (mean intercept value) of 1.9 ␮m and 3.2 ␮m for MgO and
NiO respectively.
2. Experimental
Starting materials were NiO (99%, Alfa Aeser) and MgO (99%, Merck) powders.
NiO powders were irregular in shape and 1.20 ␮m in size, Fig. 1(a). MgO particles
were much finer, i.e. sub-micron in size, Fig. 1(b). Both powders, in the as-supplied
form, were agglomerated to larger granules with average sizes, measured by a laser
diffraction technique, of D50 of 18 ␮m and 30 ␮m for NiO and MgO, respectively.
Oxide powders were mixed in fixed proportions corresponding to metallic com-
positions: Ni, MgNi2, Mg2Ni and Mg. The mixing was carried out for 30 min in a Spex
Mill at ball-to-powder ratio (B/P) of 1. The mixtures were further hand mixed with
some PVA solution and allowed to air dry for 24 h. They were then cold-compacted
under a pressure of 110 MPa to cylindrical pellets of 15 mm in diameter and 4–7 mm
in height. All pellets were heated to 1200 ◦C with a heating rate of 5 ◦C/min and
sintered for 6 h.
Deoxidation experiments were conducted in a stainless steel reactor placed in a
vertical furnace. The reactor comprises a stainless steel crucible for holding molten
electrolyte and has a lid with retractable electrodes immersible into the electrolyte.
Two sets of electrodes were available, one set (auxiliary) for pre-electrolysis and
The process of electroreduction leads to the discharging of oxy-
gen from the cathode thus leaving the metallic constituents behind.
More specifically, under the applied potential, the oxygen is ionized

136
S. Tan et al. / Journal of Alloys and Compounds 504 (2010) 134–140
Fig. 4. Current–time data collected during electro-deoxidation of NiO at 900 ◦C with
3.2 V.
and Mg2Ni, are also included in the table. Here ꢀE◦ values of the
overall reactions, in volts, were directly calculated from:
ꢀG◦
ꢀE◦ = −
nF
Here ꢀG◦ is the standard Gibbs free energy change of the reac-
tion (in Joules) at temperature T, n is the number of electrons
transferred in the balanced electrochemical reaction, and F is the
Faraday’s constant (96,500 Coulombs/gram equivalent), assuming
that the components are in pure state.
The data, given in Table 1, indicates that the reduction poten-
tial of NiO is positive and therefore it is expected to reduce more
easily. The electro-deoxidation of MgO on the other hand requires
the application of a relatively high potential. It should also be noted
that at 900 ◦C, the reactions leading to CO evolution are more favor-
able whereas at lower temperatures, e.g. 600 ◦C, the more favorable
reactions are those leading to CO2 evolution. The data in Table 1 fur-
ther shows that the choice of 3.2 V used in the current experiments
is safely below the decomposition potentials of both CaCl2 and NaCl
at 600 ◦C and 725 ◦C. This voltage is quite close to the decomposition
potentials of the electrolyte at 900 ◦C. The decomposition voltage
of NaCl given in Table 1 at 900 ◦C is in fact slightly less than 3.2 V
but this value refers to the pure state. The value calculated for NaCl
decomposition in equimolar solution of CaCl2–NaCl by using an
activity coefficient of NaCl as 0.66 [22] is 3.27 V, which is above
the current applied potential. Furthermore, the addition of IR drop,
Fig. 3. MgO:NiO = 1:2 pellet sintered at 1200 ◦C for 6 h; (a) X-ray diffractogram,
(b) SEM image recorded in back scattered electron mode indicating a two-phase
structure; NiO (bright phase) and MgO (gray phase). Black regions are pores.
(either directly from the cathode [1] or through calcium deposited
onto the cathode [3,4]), dissolved in the electrolyte and is dis-
charged at the anode (carbon) where it forms CO or CO2. Standard
reduction potentials, ꢀE◦, of oxides (MgO and NiO) for both CO and
CO2 evolution and the standard decomposition potentials of elec-
trolyte (CaCl2 and NaCl) were calculated from the thermodynamic
data given in [20,21] and are shown in Table 1. The potentials cal-
culated for oxides for the case of intermetallic formation, i.e. MgNi2
Table 1
Standard reduction potentials of MgO, NiO and CaO and decomposition potentials of CaCl2 and NaCl calculated at temperatures of 900 ◦C, 725 ◦C and 600 ◦C. The data in
parenthesis refers the potential when the metal is in liquid state. For details, see text. Standard reduction potentials of reactions yielding the intermetallics MgNi2 and Mg2Ni
are also shown. The reactions are given in the order of increasing reduction potentials at 725 ◦C.
Reaction
ꢀE◦ (V)
600 ◦
C
725 ◦
C
900 ◦
C
NiO(s) + C(s) = Ni(s) + CO(g)
+0.16
+0.20
+0.26
+0.25
+0.43
+0.33
NiO(s) + 0.5C(s) = Ni(s) + 0.5CO2
MgO(s) + 2NiO(s) + 3C(s) = MgNi2(s) + 3CO(g)
2MgO(s) + 4NiO(s) + 3C(s) = 2MgNi2(s) + 3CO2(g)
2MgO(s) + NiO(s) + 3C(s) = Mg2Ni(s,l) + 3CO(s)
4MgO(s) + 2NiO(s) + 3C(s) = 2Mg2Ni(s,l) + 3CO2(g)
MgO(s) + C(s) = Mg(s,l) + CO(g)
−0.36
−0.32
−0.98
−0.94
−1.64
−1.60
−1.83
−1.79
(−3.37)
−3.40
−0.24
−0.26
−0.86
−0.87
(−1.52)
(−1.53)
−1.71
−1.72
(−3.28)
−3.32
−0.09
−0.18
(−0.70)
(−0.78)
(−1.34)
(−1.43)
(−1.52)
(−1.63)
(−3.16)
(−3.21)
MgO(s) + 0.5C(s) = Mg(s,l) + 0.5CO2(g)
CaO(s) + C(s) = Ca(s,l) + CO(g)
CaO(s) + 0.5C(s) = Ca(s,l) + 0.5CO2(g)
NaCl(l) = Na(l) + 0.5Cl2(g)
CaCl2(l) = Ca(s,l) + Cl2

S. Tan et al. / Journal of Alloys and Compounds 504 (2010) 134–140
137
Fig. 5. NiO pellet; X-ray diffractogram of the sample (a) before and (b) after electro-
deoxidation, at 900 ◦C for 24 h (3.2 V). (c) Refers to SEM image of the reduced sample.
Fig. 6. MgO:NiO = 1:2 pellet; (a) X-ray diffractogram and (b) SEM image of powders
after electro-deoxidation for 24 h at 900 ◦C with 3.2 V.
anodic and cathodic overpotentials to the above value makes the
actual decomposition voltage even greater than 3.27 V.
shows again a particulate structure, less coarse than that obtained
from electro-deoxidation of pure NiO.
Current–time curve recorded during deoxidation of NiO pellet
at 900 ◦C is given in Fig. 4. As seen from the plot, the current shows
an initial transient rise, and over a period of 30 min it settles down
to a steady state, which then cascades down to smaller values. After
3 h, the current reduces to a residual value of 0.1 A that continues
to drop at a very slow rate until the end of the electrolysis (24 h).
X-ray diffractograms of the NiO pellet before and after electro-
deoxidation for 24 h at 900 ◦C are given in Fig. 5(a) and (b). As seen
in the diffractograms, the pellet was reduced successfully to pure
Ni. The SEM image of the pellet, given in Fig. 5(c), shows metallic
powders in a sintered state, with particle sizes ranging from 5 ␮m
to 10 ␮m. Deoxidation of NiO for 24 h at 725 ◦C and 600 ◦C was
successful in yielding pure Ni.
The oxide mixture with a target composition of MgNi2, i.e.
MgO:NiO = 1:2, was deoxidized at 900 ◦C. X-ray diffractogram of
the resulting product, given in Fig. 6(a), contained metallic con-
stituents and residual MgO, indicating that the reduction was not
yet complete. The metallic constituents were MgNi2, i.e. the tar-
get composition, and a pure Ni. Detailed analysis of XRD pattern
showed that the fractions of phases were; 50 wt.%, 44 wt.% and
6 wt.% for MgNi2, Ni and MgO, respectively. The presence of resid-
ual MgO (but not NiO) reveals that the reduction of NiO proceeded
faster than that of MgO. SEM image of the product given in Fig. 6(b)
Although the oxide mixture MgO:NiO = 1:2 after the electroly-
sis contains a certain fraction of MgO, this fraction is not enough
to convert all Ni into MgNi2. This implies that a certain fraction of
MgO either in partially or fully reduced form was lost from the cath-
ode, probably as a result of dissolution in the electrolyte. This was
checked by examining a sample taken from the electrolyte after the
electrolysis. The sample was dissolved in water and the resulting
residue was analyzed by EDS. The amount of Mg when scaled for
1 kg of electrolyte was approximately 0.09 g. This is nearly 15% of
the total Mg in the cathode.
The mixture MgO:NiO = 2:1 could not be electrolyzed at 900 ◦C,
since the target composition Mg2Ni is liquid at that temperature.
The experiments were therefore carried out at 725 ◦C, safely below
the melting point of Mg2Ni (760 ◦C). As given in Fig. 7(a), the pellet,
after 24 h of electrolysis, contained only a small fraction of metallic
phases: Mg2Ni, MgNi2 and Ni. The rest, i.e. 82 wt.% of the sam-
ple, was unreduced oxides. The final pellet was easy to grind and
SEM image of the pulverized product, given in Fig. 7(b), had a non-
metallic appearance. In the case of pure MgO, electro-deoxidation
was even more difficult. The electrolysis carried out at 600 ◦C did
not lead to any reduction. No trace of Mg could be detected in the
sample after 24 h of electrolysis at 3.2 V.

138
S. Tan et al. / Journal of Alloys and Compounds 504 (2010) 134–140
quite erratically, probably as a result of more than one reduction
reactions, e.g. reactions leading to gas evolution at anode [23]. The
experiment was quite successful for MgO:NiO = 2:1 mixture. The
pellet electrolyzed for 24 h at 600 ◦C was reduced almost fully to the
metallic state. The phases were Mg2Ni, MgNi2, Mg (OH)2 and a small
amount of residual MgO (5 wt.%). The presence of Mg (OH)2 is prob-
ably due to the reaction of metallic Mg with H2O used to separate
reduced pellet from the salt after electrolysis. The experiment with
pure MgO, however, was not as successful. The electrolysis with
5 V at 600 ◦C yielded only a trace amount of Mg2Ca and Mg(OH)2,
both adding up to about 1 wt.%, the rest being MgO. Thus, even the
application of 5 V was not enough to reduce the MgO pellet.
Returning to the results obtained with 3.2 V, it should be pointed
out that, of the four metallic phases aimed in the current work, the
ease of reduction closely parallels the values of the standard reduc-
tion potentials reported in Table 1. The reactions in the table were
listed according to their reduction potentials (at 725 ◦C). Accord-
ingly, the reductions become more and more difficult as one moves
from Ni-rich compositions to Mg rich ones.
The ease of reduction in NiO could be attributed to its low, in fact
positive, reduction voltage as well as to the other factors such as
its high conductivity (see below). Similar observations were made
in a study by Qiu et al. [24] in which 100–200 ␮g of NiO powders
were reduced successfully within a few minutes at 900 ◦C.
The oxide mixture targeted for MgNi2 was also reduced rela-
tively easily. Here, the difficulty was in the production of samples
with desired stoichiometry since there was some loss of Mg
from the cathode during the electrolysis. Still, the electrolysis of
MgO:NiO = 1:2 mixture is not unlike that of NiO, i.e. the mixture
resulted in a reduction which is nearly complete. Also, as NiO has
very low reduction potential, formation of a Ni-rich environment
in the pellet probably enhanced the reduction of MgO in the oxide
mixture. Similar arguments were put forward by Ma et al. [25] and
Yong et al. [14] in explaining the enhanced reduction of TiO2 in the
presence of Fe2O3 and NiO.
The loss of Mg during the electrolysis appears to be the major
obstacle for the desired quantity of compound formation. Condi-
tions in the pellet should be such that Mg upon reduction should
react with Ni, which had been reduced earlier in the process, form-
ing the targeted compound. Circumstances that favor compound
formation are well documented in the earlier studies. One approach
is to carry out the electrolysis under high potential. In such con-
ditions as explained by Yong et al. [14], the differences in the
reduction rate of the oxides are diminished and all elements are
extracted in the same time scale, which would allow them to react
with one another resulting in the compound formation. The other
approach is the solutionizing of the oxide mixture itself [17,24].
In this case, the metallic species being present in the same oxide
phase are close to each other, which upon reduction could react
easily with one another. The latter approach would be particularly
suitable for the current system since the oxides NiO–MgO have
complete solid solubility. Thus, it is possible to form MgO and NiO
in a single-phase structure in which Mg:Ni ratio can be adjusted to
any preselected value.
Fig. 7. MgO:NiO = 2:1 pellet; (a) X-ray diffractogram and (b) SEM image of powders
obtained after electro-deoxidation for 24 h at 725 ◦C with 3.2 V.
In an attempt to reduce the MgO rich mixtures, separate
experiments were carried out; in which the potential was raised
to 5 V. This value is well above the decomposition potential of
the electrolyte. Fig. 8 shows the current–time plot recorded for
MgO:NiO = 2:1 mixture. As seen in the figure, the current fluctuates
Current experiments show that the reduction of MgO via
electro-deoxidation is an extremely difficult task. This is true for
the mixture targeted for Mg2Ni as well as that for the pure Mg.
MgO:NiO = 2:1 mixture for instance after 24 h of electrolysis yielded
only a small fraction of metallic phases, four-fifths of the pellet still
remained as oxides. This implies that the phases in the oxide mix-
ture do not behave independently. If they were, NiO could have
been reduced to pure Ni totally, leaving behind MgO. In a pellet with
porosity in the order of 40% or so, this situation could be partially
attributed to the insufficient conductivity of the pellet.
To check this, the electrical conductivities of the oxide pellets
were measured with a two-probe technique. The results, given
Fig. 8. Current–time data collected during electro-deoxidation of MgO:NiO = 2:1
effected with 5 V at 600 ◦C.

S. Tan et al. / Journal of Alloys and Compounds 504 (2010) 134–140
139
which the reduction proceeds by the propagation of three-phase
(conductor/insulator/electrolyte) interlines into the insulator com-
pound. Obviously, this mechanism was not operative in the current
samples. The experiments carried out by pellets sandwiched
between two contacting metallic plates establishing many con-
tacts with the oxide did not lead to noticeable improvements in the
electrolysis.
It is apparent that the oxides in the current system, both MgO
and NiO, differ from most of the other oxides in the details of their
electro-deoxidation. Electroreduction of oxides such as TiO2 [26],
Nb2O5 [27] and B2O3 [28], Cr2O3 [29] and Ta2O5 [30] all involve the
formation of intermediate phases which, in the respective order,
are CaTiO3, CaNbO3, CaB2O4, CaCr2O4, CaCr2O4 and calcium tanta-
lates. Thus, the reduction in these oxides is not a direct process but
occurs through these intermediaries. This is also true for ZrO2 and
SiO2 where as mentioned by Peng et al. [31] and Yasuda et al. [32],
CaZrO3 and CaSiO3 do form during their electro-deoxidation. The
electroreduction of Al2O3 to Al [33], as has recently been reported
by Yan and Fray [34], similarly, involves the formation of calcium
aluminates. Thus, in the three-phase interline where the reduction
proceeds, it is not the starting oxide but the intermediate phase,
which is in contact with the conductor and the electrolyte. Thus, the
formation of intermediate phases such as CaSiO3, CaZrO3 or calcium
aluminates is expected to alter the conditions for electroreduction
including the local conductivity of the pellet. MgO (and also NiO)
in contrast do not form such intermediate phases. Therefore, the
insulating nature of MgO is not altered during the electrolysis.
Values of conductivity reported above refer to oxides at room
temperature. Since the electrolysis was carried out at elevated
temperatures, the applicable values are much higher than those
reported above. Conductivities of MgO and MgO:NiO = 2:1 mix-
tures were measured at temperatures up to 600 ◦C and are reported
in Fig. 9(b). The conductivity of MgO rises to 10−9 ohm−1 cm−1
,
and that of MgO:NiO = 2:1 to 10−6 ohm−1 cm−1 at 600 ◦C. The
values measured for MgO are comparable with those reported
by Lewis and Wright [35]. The conductivity of NiO on its own
was not measured in this study, but the reported values are as
high as 10−2 ohm−1 cm−1 at 600 ◦C [36]. Since NiO can be elec-
trodeoxidized quite successfully at this temperature, the value of
10−2 ohm−1 cm−1 can be taken as a rough measure of the required
conductivity. Then to reach this value in MgO, using the tempera-
ture dependence of conductivity given by Lewis and Wright [35],
the required temperature needs to be as high as 1500 ◦C.
Electrolysis at elevated temperatures may be employed for Mg
compounds of high melting points but for the current compositions
of Mg2Ni and pure Mg, which have the melting points of 760 ◦C and
650 ◦C, this approach would not be relevant. Clearly, for successful
electroreduction of MgO rich mixtures, it would be desirable to find
processing techniques that would render pellets with an acceptable
conductivity. Evidently, any processing techniques that would dis-
rupt the continuity of MgO phase would be helpful in this respect.
But to increase the conductivity to its highest possible level, it is
necessary to develop carefully tailored microstructures in the oxide
pellets. One possibility, which might be relevant for MgO:NiO = 2:1,
would be to make NiO the host phase, i.e. NiO, though small in its
proportions, could be distributed in a manner that could preserve
its connectivity while enveloping each and all MgO particles.
Fig. 9. (a) Electrical conductivities of MgO, MgO:NiO = 2:1, MgO:NiO = 1:2 and NiO
at room temperature, and (b) conductivity values as a function of temperature up
to 600 ◦C for MgO and MgO:NiO = 2:1.
in Fig. 9(a), verify that the conductivity of MgO differs drasti-
cally from that of NiO. The value for MgO at room temperature
is close to 10−14 ohm−1 cm−1, which is five order of magnitude less
than that of NiO. It is interesting to note that the conductivities of
the mixtures are also quite different. While NiO rich mixture, i.e.
MgO:NiO = 1:2 has a value which is intermediate between those of
MgO and NiO, the same does not hold true for MgO:NiO = 2:1. This
mixture has almost the same conductivity as pure MgO.
Clearly, in mixtures corresponding to Ni and MgNi2, NiO is the
continuous phase with the result that conductivity of the pellet
is close to that of NiO. Thus, the pellet in such mixtures benefits
from the high conductivity of the host phase. In MgO:NiO = 2:1
mixture, it appears that MgO is the continuous phase and the
presence of NiO which occurs as isolated islands, does not exert
a pronounced influence on the conductivity. Thus, the conductivity
of the MgO rich mixture does not differ significantly from that of
pure MgO.
4. Conclusions
It should be mentioned that with the conductivities measured
for MgO and the MgO rich mixture are not unlike those expected
for other insulating oxides such as ZrO2 and SiO2. The latter oxide,
for instance, was reduced quite successfully by Nohira et al. [8]
using the contacting electrode method, i.e. increasing the num-
ber of contacting points in the cathode. Chen and co-workers
[9,10] formalized this by three-phase interline model according to
A study was carried out on the synthesis of Mg–Ni compounds
as well as pure Ni and Mg from their oxides using the method of
electro-deoxidation. The oxides were in the form of discrete phases
NiO and MgO, 3.2 ␮m and 1.9 ␮m in size respectively, suitably
proportioned to yield Ni, MgNi2, Mg2Ni and Mg. The oxides were
electrolyzed in a eutectic mixture of CaCl2–NaCl solution main-

140
S. Tan et al. / Journal of Alloys and Compounds 504 (2010) 134–140
tained at a constant temperature (900–600 ◦C), using a graphite
anode at a potential of 3.2 V.
[7] G.Z. Chen, D.J. Fray, T.W. Farthing, Metall. Mater. Trans. B 32 (2001) 1041.
[8] T. Nohira, K. Yasuda, Y. Ito, Nat. Mater. 2 (2003) 397.
[9] X.B. Jin, P. Gao, D.H. Wang, X.H. Hu, G.Z. Chen, Angew. Chem. Int. Ed. 43 (2004)
734.
The following can be concluded from the current study:
[10] W. Xiao, X. Jin, Y. Deng, D. Wang, X. Hu, G.Z. Chen, Chem. Phys. Chem. 7 (2006)
1750.
(i) NiO with positive reduction potential can be reduced quite
successfully to a metallic state at temperatures studied in
this work. The oxide mixture MgO:NiO = 1:2 can also be elec-
trodeoxidized to a metallic state comprising the phases MgNi2
and Ni.
[11] A. Cox, D.J. Fray, Electrochem. Soc. Proceed. 19 (2002) 745.
[12] G. Qui, D. Wang, X. Jin, G.Z. Chen, Electrochim. Acta 51 (2006) 5785.
˙
[13] S. Tan, T. Örs, M.K. Aydinol, T. Öztürk, I. Karakaya, J. Alloys Compd. 475 (2009)
368.
[14] Z. Yong, M. Meng, W. Dihua, J. Kai, H. Xiaohong, J. Xianbo, G.Z. Chen, Chin. Sci.
Bull. 51 (2006) 2535.
(ii) MgO rich mixtures are difficult to reduce via electro-
deoxidation. The mixture MgO:NiO = 2:1 can be reduced to
metallic phases only in small proportions. Difficulties in the
reduction of MgO and MgO rich mixtures are attributed to low
conductivity of MgO.
[15] A.J.M. Wood, R.C. Copcutt, G.Z. Chen, D.J. Fray, Adv. Eng. Mater. 5 (2003) 650.
[16] X.Y. Yan, D.J. Fray, Adv. Func. Mater. 15 (2005) 1757.
[17] G. Qiu, D. Wang, M. Ma, X. Jin, G.Z. Chen, J. Electroanal. Electrochem. 589 (2006)
139.
[18] L. Zaluski, A. Zaluska, J.O. Stromolsen, J. Alloys Compd. 217 (1995) 245.
[19] L. Lutterotti, S. Matthies, H.R. Wenk, Int. U. Crystallogr. Comm. Pow-
der Diffr. Newslett. 21 (14) (1999), MAUD Version 2.072, 2008
http://www.ing.unitn.it/luttero/maud.
Further, the low conductivity of the latter compositions is due to
MgO being the continuous phase. It is therefore suggested that the
conductivity and thus the reducibility of the oxide mixtures would
be improved if NiO could be made the host phase.
[20] I. Barin, O. Knacke, O. Kubaschewski, Thermochemical Properties of Inorganic
Substances, 2nd ed., Springer-Verlag, Berlin, 1973.
[21] C.W. Bale, P. Chartrand, S.A. Degterov, G. Eriksson, K. Hack, R. Ben Mahfoud, J.
Melanc¸ on, A.D. Pelton, S. Petersen, Calphad 26 (2002) 189.
˙
[22] I. Karakaya, W.T. Thompson, Can. Met. Q. 25 (1986) 307.
[23] J.J. Peng, G. Li, H. Chen, D. Wang, X. Jin, G.Z. Chen, J. Electrochem. Soc. 157 (2010)
F1–F9.
Acknowledgements
[24] G. Qiu, M. Ma, D. Wang, X. Jin, X. Hu, G.Z. Chen, J. Electrochem. Soc. 152 (2005)
E328.
Support for this work was provided by BAP-03-08-
DPT.200305K120920-20 and in part by the FP6 program of
the European Commission Project (FP6-200-3-518-271) NESSHY,
which we gratefully acknowledge. The authors also thank Volkan
Kalem for his help with the conductivity measurements and Cara
Murphy Keyman for reading the manuscript.
[25] M. Ma, D. Wang, X. Hu, X. Jin, G.Z. Chen, Chem. Eur. J. 12 (2006) 5075.
[26] K. Dring, R. Bhagat, M. Jackson, R. Dashwood, D. Inman, J. Alloys Compd. 419
(2006) 103.
[27] T. Wu, W. Xiao, X. Jin, C. Liu, D. Wang, G.Z. Chen, Phys. Chem. Chem. Phys. 10
(2008) 1809.
˙
[28] T. Örs, S. Tan, T. Öztürk, I. Karakaya, J. Mater. Sci. 44 (2009) 3514.
[29] D.J.S. Hyslop, A.M. Abdelkader, A. Cox, D.J. Fray, Acta Mater. 58 (2010)
3124.
[30] Q. Songa, Q. Xua, X. Kanga, J. Dub, Z. Xib, J. Alloys Compd. 490 (2010) 241.
[31] J. Peng, K. Jiang, W. Xiao, D. Wang, X. Jin, G.Z. Chen, Chem. Mater. 20 (2008)
7274.
References
[32] K. Yasuda, T. Nohira, R. Hagiwara, Y.H. Ogata, J. Electrochem. Soc. 154 (2007)
E95.
[33] H. Xie, H. Zhang, Y. Zhai, J. Wang, C. Li, J. Mater. Sci. Technol. 25 (2009)
459.
[34] X.Y. Yan, D.J. Fray, J. Appl. Electrochem. 39 (2009) 1349.
[35] T.J. Lewis, A.J. Wright, Br. J. Appl. Phys. D 1 (1968) 441.
[36] S.P. Mitoff, J. Chem. Phys. 35 (1961) 882.
[1] G.Z. Chen, D.J. Fray, T.W. Farthing, Nature 407 (2000) 361.
[2] D. Wang, G. Qiu, X. Jin, X. Hu, G.Z. Chen, Angew. Chem. Int. Ed. 45 (2006) 2384.
[3] R.O. Suzuki, K. Tatemoto, H. Kitagawa, J. Alloys Compd. 385 (2004) 173.
[4] R.O. Suzuki, J. Phys. Chem. Solids 66 (2005) 461.
[5] C.S. Seo, S.M. Jeong, S.B. Park, J.Y. Jung, S.W. Park, S.H. Kim, J. Chem. Eng. Jpn. 39
(2006) 77.
[6] X. Kang, Q. Xu, S.J. Ma, L. Zhao, Q.S. Song, Electrochemistry 77 (2009) 663.