+ MODEL

H O S T E D BY Available online at www.sciencedirect.com

ScienceDirect
Journal of Magnesium and Alloys xx (2015) 1e6
www.elsevier.com/journals/journal-of-magnesium-and-alloys/2213-9567

Full length article

Magnesium production by carbothermic reduction in vacuum*

Yang Tian a,b,c,d, Bao-qiang Xu a,b,c,d, Bin Yang a,b,c,d,*, Cheng-b Yang b,c, Tao Qu a,b,c,d,
Da-chun Liu a,b,c,d, Yong-nian Dai a,b,c,d
a
National State Key Laboratory of Complex Nonferrous Metal Resources Clear Utilization, Kunming University of Science and Technology, Kunming 650093,
China
b
National Engineering Laboratory for Vacuum Metallurgy, Kunming University of Science and Technology, Kunming 650093, China
c
Key Laboratory of Non-Ferrous Metals Vacuum Metallurgy of Yunnan Province, Kunming 650093, China
d
College of Metallurgy and Energy Engineering, Kunming University of Science and Technology, Kunming 650093, China
Received 1 July 2014; revised 9 April 2015; accepted 10 April 2015
Available online ▪ ▪ ▪

Abstract

In present work, we investigate production of magnesium by carbothermic reduction under vacuum conditions. The process was divided into
two parts, one is reduction process, and the other one is condensation process. The experimental results revealed that during reduction process, the
gasesolid reaction between MgO and CO was not occurred at a temperature and pressure of 1723 K and 30e100 Pa respectively. So the main
reduction reaction was MgO(s) þ C(s) ¼ Mg(g) þ CO(g) (under vacuum) and reaction type belonged to solidesolid reaction. In Condensation
Process, according to a contrast and analysis, the condensation quality of magnesium is associated with CO concentration. The resultant product
C was formed and it followed magnesium vapor condensation which prevents mutual combination of two metal droplets to forms the compact
condensation produces. Therefore, in order to compact morphological forms of magnesium crystal whiskers, we must control the technical
conditions and find the method to separate the magnesium vapor and carbon monoxide. That's the key factor to get better crystalline structure.
Copyright 2015, National Engineering Research Center for Magnesium Alloys of China, Chongqing University. Production and hosting by
Elsevier B.V. All rights reserved.

Keywords: Metal magnesium; Carbothermic reduction; Condensation; Vacuum

1. Introduction route in western countries, accounting for about 77 pct of

Generally, magnesium is produced by two main processes:
electrolysis of molten magnesium chloride and thermal
reduction of magnesia [1]. Electrolysis is the pre-dominant
*
Foundation item: Project supported by National Natural Science Foundation
of China (No. 51304095); the Personnel Training Funds of Kunming Uni-
versity of Science and Technology, China (No. 14118665) and Educational
Commission of Yunnan Province, China (No. 14051618).
* Corresponding author. College of Metallurgy and Energy engineering,
Kunming University of Science and Technology, Kunming 650093, China.
Tel.: þ86 871 65161583.
E-mail address: kgyb2005@126.com (B. Yang).
Peer review under responsibility of National Engineering Research Center
for Magnesium Alloys of China, Chongqing University.
western countries total production, however, this process is
characterized by high-energy requirements [2]. In recent
years, production of magnesium in China by the Pidgeon
process has dominated world production. In this process,
ferrosilicon reduces magnesia from calcined dolomite under
vacuum. The process suffers from high energy usage and low
productivity [3]. Our research group is exploring the carbo-
thermic reduction of magnesia area of study for many years,
and the process of carbothermic reduction and magnesium
vapor condensation were the important link to solve the
problem of industrialization. The reaction between MgO and
C may provide a new route toward the production of mag-
nesium. But for the carbothermic reduction reaction type, one
point of view was that the main reaction is solidesolid

http://dx.doi.org/10.1016/j.jma.2015.04.001.
2213-9567/Copyright 2015, National Engineering Research Center for Magnesium Alloys of China, Chongqing University. Production and hosting by Elsevier B.V. All rights reserved.

Please cite this article in press as: Y. Tian, et al., Journal of Magnesium and Alloys (2015), http://dx.doi.org/10.1016/j.jma.2015.04.001
 + MODEL
2 Y. Tian et al. / Journal of Magnesium and Alloys xx (2015) 1e6
reaction, the other was that the main reaction is gasesolid
reaction [4e6].
Qu analyzed the uniform nucleation of water from nucle-
ation rate and randomness, while investigating magnesium
vapor condensation process. Zhang theoretically studied the
gas bubble nucleation of molten metal. Zeng evaluated the
properties of metastable equilibrium, and the mechanism of
gas bubble nucleation through thermodynamics and kinetic
theory. Izmailov studied the statistical mechanics of nucle-
ation. Ring researched the reason of nano-clusters nucleation.
Leubner established nucleation and growth model. In Lüm-
men's paper, the uniform nucleation of supersaturation iron
vapors has been investigated by molecular dynamics simula-
tion method [7e15]. Therefore, analysis of the magnesium
vapor nucleation in theory is helpful to find the key factors
which affect the way of magnesium vapor condensation. In
addition the reactions of carbothermic reduction process were
discussed. In order to investigate the mechanism of the car-
bothermic reduction process in vacuum, XRD analysis was
also proposed.
2. Reduction process
2.1. Raw material
Analytical-grade magnesia and carbon were used as the raw
materials in our experiments.
2.2. Schematic diagram of vacuum furnace
The experiments were carried out in a self-made vacuum
furnace shown in Fig. 1.
2.3. Experimental procedure
The raw materials, magnesia (20 g) and carbon (12 g)
grounded to 140 mesh after drying were mixed in a molar
ratio (MgO/C ¼ 1/2). The mixed materials were compacted
into pieces of F30 mm  10 mm dimensions under 6 MPa.
The compacts were placed in a graphite crucible and heated at
the heating rate of 15 K  min 1. The system temperature
needed to be kept at 1723 K for 1e2 h under vacuum. At high
Fig. 1. Schematic diagram of vacuum furnace.
Please cite this article in press as: Y. Tian, et al., Journal of Magnesium and Alloys (2015), http://dx.doi.org/10.1016/j.jma.2015.04.001
reduction temperature, Mg vapors and CO were produced and
moved into the upper condensing towers. The temperature of
the condensing surfaces was kept between 773 and 973 K. The
reversion reaction occurred during the condensing process.
2.4. Analysis method
The crystalline phase of the resulted products was identified
by X-ray diffraction (XRD) instrument (D/max-3B) using Cu
Ka radiation with a scanning rate of 2 ( )/min which was made
by Rigaku Corporation of Japan.
The mass loss ratio (a) is defined as
M0 M1
a¼  100% ð1Þ
M0
where M0 is the mass at the start of the reduction reaction and
M1 is the reaction equilibrium mass at the end of reduction.
The term M0 M1/M1, represents the reduction degree of
magnesia available for reaction.
2.5. Analysis of experimental results
Carbothermic reduction was carried out under the following
conditions: MgO:C ¼ 1:2 (molar ratio), temperature
1073e1773 K, hold temperature time 1e2 h, system pressure
30e100 Pa, compacting pressures 6 MPa of the blend MgO.
During the thermo gravimetric reduction process, the mass
loss of the sample was monitored as a function of time. It was
calculated by formula (1). The photographs of materials
(MgO þ C) before and after reaction at 1723 K are shown in
Fig. 2. A typical mass-loss percentage vs. temperature curve of
reduction of magnesia by carbon is shown in Fig. 3.
Fig. 2 shows that the structure of briquetted materials was
changed into porous and loose after reduction reaction, which
revealed that the solidesolid reaction occurred between MgO
and C. It can be noticed that there was no sharply mass loss
detected when the temperature was below 1553 K as shown in
Fig. 3. The values below 15% indicate that there was not any
reduction reaction. When the temperature was higher than
1553 K, the sample weight changes sharply with increase of
temperature, and there was an obvious inflection point of mass
loss at 1553 K. Therefore, this temperature was considered as
initial reaction temperature. It accorded with the theoretical
calculation. It can also be seen that the maximum of mass loss
reached 75% in the experimental temperature range, the sol-
idesolid reduction reaction was completely finished under the
experimental condition.
The reaction between MgO and CO may provide a new
method toward investigation of gasesolid reaction. A stomatal
dish was added in the carbothermic reaction crucible to divide
reduction zone into two parts. The pieces of MgO were placed
on the Al2O3 flats (above the stomatal dish) in order to avoid
contacting the graphite, so that the reaction between MgO and
crucible(C) was not occurred. Under the stomatal dish,
magnesia and carbon as the starting materials were mixed in a
molar ratio (MgO/C ¼ 1/2) with addition of 5% CaF2 and
 + MODEL

Y. Tian et al. / Journal of Magnesium and Alloys xx (2015) 1e6 3

Fig. 2. Photographs of material (MgO þ C) before (a), and after reaction (b) at 1723 K.

estimated. The photograph of material (MgO þ CO) before

and after reaction at 1723 K is shown in Fig. 4. A typical
mass-loss percentage vs. temperature curve of reduction of
magnesia by CO is shown in Fig. 5.
The morphology of MgO was the same after a gasesolid
reaction, as shown in Fig. 4. No pore was appeared inside
MgO. Through the calculation of mass loss ratio, the a of
MgO is shown in Fig. 5. It showed that the value of a was less
than 20%, the result was in the range of a of pure MgO, and
no condensing product was obtained in condensation zone too.
It was accorded with the theoretical calculation of gasesolid
reaction. Therefore, the gasesolid reaction was not occurred
in the carbothermic reduction process under the condition of
10e60 Pa and 300e2000 K.

3. Condensation process
Fig. 3. Effect of reduction temperature on material (MgO þ C) weight loss.

were made into pieces under compaction pressures of 6 MPa. 3.1. The condensation of magnesium in carbothermic
When the system temperature needed to be kept at 1723 K for reduction of magnesia
more than 2 h, a mass of CO was produced by mixed MgO and
C under the stomatal dish and moved through the stomatal As reverse reaction between magnesium vapor and CO
dish to the gasesolid reaction zone (Above the stomatal dish). occurred:
The Analytical-grade magnesia could make a good contact MgðgÞ þ COðgÞ ¼ MgOðsÞ þ CðsÞ
with CO, and then the reaction between MgO and CO was

Fig. 4. Photograph of material (MgO þ CO) before (a) and after reaction (b) at 1723 K.

Please cite this article in press as: Y. Tian, et al., Journal of Magnesium and Alloys (2015), http://dx.doi.org/10.1016/j.jma.2015.04.001
 + MODEL

4 Y. Tian et al. / Journal of Magnesium and Alloys xx (2015) 1e6

Fig. 7. XRD patterns of the condensing product in carbothermic reduction.

Fig. 5. Effect of reduction temperature on material (MgO þ CO) weight loss.
The raw materials and reduction product samples are shown
in Fig. 6. Fig. 7 shows the XRD patterns of the condensing
product in carbothermic reduction, it can be seen that metal
magnesium was obtained. Meanwhile, magnesia was also
found as indicated by peaks.
The method of vacuum distillation was used to separate the
metal magnesium, MgO, and Mg2C3. In the distillation pro-
cess, the condensing product was used as raw material and
heated from room temperature to 1373 K. After distillation for
1 h or 2 h, the temperature of distillation product cooled to
room temperature and took out. After the vacuum distillation
process, MgO, Mg2C3 and carbon were stayed loaded into the
reaction crucible, high purity metal magnesium was produced
in the condensing tower, which called distillation product.
The SEM images of condensing product which were ob-
tained from condensing tower are shown in Fig. 8. The image
of condensing product (a) shows irregular shape with an
incompact structure which had a lot of small openings, where
the carbon can be found. But the distillation product (b) shows
irregular shape with a compact structure and good crystalline
feature.
As the BESECOMP image of condensing product is shown
in Fig. 9, the loose crystallization with chinky structure was
observed which is due to the carbon and magnesia, obtained
by reverse reaction. It means that the carbon and magnesia
(obtained by reverse reaction) were simultaneously in the
condensing product blocks the atomic aggregation and coac-
ervation, decreases the rate of metal magnesium nucleation
and the rate of crystal growth. Then, the different appearance
of magnesium crystal from block-like to loose-like was
obtained.
3.2. Analysis of reverse reaction ratio
The reversion reaction ratio (g) was defined as
W1
g¼  100% ð2Þ
W0

Fig. 6. The picture of raw materials (a) and reduction product (b).

Please cite this article in press as: Y. Tian, et al., Journal of Magnesium and Alloys (2015), http://dx.doi.org/10.1016/j.jma.2015.04.001
 + MODEL

Y. Tian et al. / Journal of Magnesium and Alloys xx (2015) 1e6 5

Fig. 8. SEM images of condensing product (a) and vacuum distillation product of purity magnesium (b).

Fig. 9. The BSECOMP image of condensing product.

where W0 represents the total mass of magnesium and
magnesia in the condensing product, and W1 the mass of
magnesia in distillation residual.
The reverse reaction ratio was calculated using formula (2),
as shown in Fig. 10, it is clear that the reverse reaction ratio
changed slowly with the incremental molar ratio, the reverse
reaction ratio increased with increasing reaction time. How-
ever, the peak value was less than 9%, it can be deduced that
most magnesium vapors transformed into metal magnesium,
the condensing condition of the vacuum equipment was
appropriate for the carbothermic reduction.
4. Conclusions
Fig. 10. Relationships between reversion ratio and reaction time in different
molar ratios.
condensation which prevents mutual combination of two
metal droplets to forms the compact condensation pro-
duces. That's the reason of the structure decentralized
products obtained which loose metal luster. Therefore, in
order to compact morphological forms of magnesium
crystal whiskers, we must control the technical conditions
and find the method to separate the magnesium vapor and
carbon monoxide. That's the key factor to get better
crystalline structure.

1) In reduction process, the analysis of experimental results References

revealed that the gasesolid reaction between MgO and CO
was not occurred at 1723 K temperature and 30e100 Pa
pressure. So the main reduction reaction was
MgO(s) þ C(s) ¼ Mg(g) þ CO(g) in vacuum. The reaction
type belonged to solidesolid reaction.
2) In Condensation Process, according to a contrast and
analysis, the condensation quality of magnesium is asso-
ciated with CO concentration. The resultant product C
was formed and it followed magnesium vapor
[1] Rong-ti Li, Wei Pan, Masamichi Sano, Metall. Mater. Trans. 8 (2003)
433e437.
[2] A. Froats, Light Metal, Metallurgical Society of AIME, New York, NY,
1980, pp. 969e979.
[3] S. Ramakrishnan, P. Koltun, Resour. Conserv. Recycl. 42 (2004) 49e64.
[4] M. Sakamoto, S. Akiyama, J. Mater. Sci. Lett. 16 (12) (1997)
1048e1050.
[5] Ding-cheng Mo, Metallurgical Kinetics, Central South University Press,
Changsha, 1987, pp. 173e326.

Please cite this article in press as: Y. Tian, et al., Journal of Magnesium and Alloys (2015), http://dx.doi.org/10.1016/j.jma.2015.04.001
 + MODEL

6 Y. Tian et al. / Journal of Magnesium and Alloys xx (2015) 1e6

[6] Dong Yang, Chao-wan Wu, Zhong-yi Wang, Special Metallurgical
Problems 10, China Industry Press, Beijing, 1963, pp. 58e84.
[7] Jiu-sheng Bao, Tiong-gang Liu, Eng. Mech. Mater. 32 (2) (2008) 4e7.
[8] Lan Peng, Chao Liu, J. Eng. Thermophys. 21 (6) (2000) 683e685.
[9] Kai-yang Qu, Yi Jiang, J. Phys. 49 (11) (2000) 2214e2219.
[10] Hua-wei Zhang, Yan-xiang Li, J. Phys. 56 (8) (2007) 4864e4871.
[11] Dan-Qin Zeng, J. Chongqing Univ. 18 (2) (1995) 1e8.
[12] Alexander F. Izmailov, Allan S. Myerson, Stephen Arnold, J. Cryst.
Growth 196 (1999) 234e242.
[13] Terry A. Ring, Adv. Colloid Interf. Sci. 91 (2001) 473e499.
[14] Ingo H. Leubner, Curr. Opin. Colloid Interf. Sci. 5 (2000) 151e159.
[15] N. Lümmen, T. Kraska, Aerosol Sci. 36 (2005) 1409e1426.

Please cite this article in press as: Y. Tian, et al., Journal of Magnesium and Alloys (2015), http://dx.doi.org/10.1016/j.jma.2015.04.001