Materials Transactions, Vol. 49, No. 8 (2008) pp. 1826 to 1829
#2008 The Japan Institute of Metals
Effect of Ar/Ar-H2 Plasma Arc Melting on Cu Purification
Jae-Won Lim1 , Min-Seuk Kim1 , N. R. Munirathnam1; *1 , Minh-Tung Le1;2 , Masahito Uchikoshi3 ,
Kouji Mimura3 , Minoru Isshiki3 , Hyuk-Chon Kwon4 and Good-Sun Choi1; *2
1
Minerals & Materials Processing Division, Korea Institute of Geoscience & Mineral Resources,
92 Gwahangno, Yuseong-gu, Daejeon 305-350, Korea
2
Department of Materials Science Engineering, Chungnam National Univeristy,
220 Gung-dong, Yuseong-gu, Daejeon 305-764, Korea
3
Institute of Multidisciplinary Research for Advanced Materials, Tohoku University, Sendai 980-8577, Japan
4
Korea Institute of Industrial Technology, 7-47 Songdo-Dong, Yeunsu-Gu, Inchen 460-840, Korea
Removal of impurities from Cu metal by Ar and Ar-20%H2 plasma arc melting (PAM) has been carried out. Several impurities such as Li,
Na, Mg, P, S, Cl, K, Ca, Zn, Pd, Pb and Bi in Cu were efficiently removed when only Ar plasma gas was used. Moreover, removal degrees for the
above mentioned impurities were significantly increased after Ar-20%H2 -PAM, especially for K, Zn and Pd. It was found that Ar-H2 PAM
showed an excellent effect to eliminate impurities with higher vapor pressures than that of Cu metal. [doi:10.2320/matertrans.MRA2008054]
(Received February 15, 2008; Accepted May 19, 2008; Published July 25, 2008)
Keywords: copper, impurity, hydrogen, plasma arc melting, glow discharge mass spectrometry
1.
Introduction
In ultra-large scale integration circuits, Cu has been used
as a substitute for Al because of its high resistance to electromigration and low resistivity. In addition, continuing decrease in feature size of microelectronic devices has
enhanced the demand for ultra-high purity copper. The
purity of Cu is also very important in fundamental research
because of the influence of trace impurities on electrical
resistivity, thermal conductivity, ductility and oxidation
kinetics.1,2) It was reported that the impurities such as Al,
S, Cr, Ni, As, Se and Bi were effectively separated during Cu
purification by anion-exchange separation and float zone
refining of copper under hydrogen atmosphere has resulted
in a remarkable decrease of S, Se, Al and Si.3,4) Despite of
these works, sufficient experimental data as well as detailed
impurity analyses on efficient purification processes have
not been reported.
Ar-H2 plasma arc melting (PAM) enabled removal of
non-metallic impurities from Fe, Mo, Ta, and V5,6) and also
metallic impurities from Zr, Nb, Mo and Hf.6–8) However,
there was no report on the purification of Cu using Ar-H2
PAM, because, it was considered difficult to refine Cu due
to its relatively high vapor pressure compared to that of
refractory metals. Therefore, in the present work, the refining
of Cu using Ar/Ar-H2 PAM and its impurity analysis using
glow discharge mass spectrometry (GDMS) were carried out
for the first time.
2.
Experimental
All experiments were conducted using a laboratory-scale
DC arc discharged type plasma torch with a maximum power
of 20 kW equipped on a stainless steel vessel (W 200 mm *1Present
address: Ultrapure Materials Division, Centre for Materials for
Electronics Technology (C-MET), IDA, Phase-III, Cherlapally, HCL
Post, Hyderabad-500051, India
*2Corresponding author, E-mail: gschoi@kigam.re.kr
L 800 mm H 180 mm) and connected directly to a 3 phase
AC power mains. A thoriated W cathode with a length of
39 mm including the nozzle (2 mm L 5 mm) orifice at the
tip was placed in the plasma torch and a specimen mounted
on the water-cooled copper crucible was utilized as an anode.
The angle of the nozzle portion from the W cathode was 15
to the normal. The system was evacuated to 27 Pa and
repeatedly flushed with Ar gas to reduce the trapped gases.
During the experimentation, 5.92 kW DC power with 160 A
and 37 V was required to ensure a stable arc at Ar-20%H2
gas pressure. A Cu specimen (W 25 mm H 0.8 mm, 31 0:5 g) was loaded on a water-cooled Cu crucible of 45 mm in
diameter and 4 mm in depth. High purity Ar (>99:9995%)
and H2 (>99:9999%) gases were mixed and introduced into
the plasma torch. The H2 content in the plasma gas was
20 vol% at atmospheric pressure. Total melting times for
each specimen were 20 min and 60 min respectively. GDMS
(VG ELEMENTAL 9000) was used for precise analysis of
impurity concentrations in Cu specimens before and after
PAM. The discharge voltage and current used to obtain
plasma were 1 kV and 3 mA, respectively. Impurity concentrations were estimated by repeating measurements several
times for each isotope. Oxygen and Nitrogen analyses were
performed by an inert gas fusion infrared absorption method
using O-N analyzer (LECO TC-436).
3.
Results and Discussions
In PAM process, the vaporization of metallic and nonmetallic impurities from the molten melt is expected to play
an important role to remove impurities. The recorded
weight loss (%) during Ar/Ar-H2 PAM was graphically
represented as a function of melting time and the result is
shown in Fig. 1. A weight loss of nearly 0.3% was found
in Ar-H2 PAM melting and this value was 6 times higher
when compared to the weight loss in Ar PAM. This was
attributed to the raise in the molten metal temperature by a
few hundred degrees due to the presence of H2 in Ar20%H2 .,7) which results in a higher vaporization rate caused
Effect of Ar/Ar-H2 Plasma Arc Melting on Cu Purification
106
0.5
Ar
Ar -20% H2
105
Vapor Pressure (Pa)
0.4
Weight Loss (%)
1827
0.3
0.2
0.1
K Na Zn Mg
Ca
Pb
Mn
Cu Al Cr Fe
104
Si
V
103
102
10
Ni, Co
1
M. P.
100
10-1
W
10-2
0.0
0
10
20
30
40
50
60
70
Melting Time, t/min
Fig. 1
Weight loss of Cu metal as a function of melting time.
by the temperature increase of the molten metal. The efforts
to measure the surface temperature of the molten copper has
proved unsuccessful. It was impossible to measure directly
the surface temperature by two color pyrometer because
of a high temperature plasma flame on the surface of the
molten metal. Therefore, an alternative method to measure
indirectly the specimen temperature using a melting point
technique was adopted. As preliminary experiments, arc
power values required for melting source of refractory
metals and their alloys were determined and the relation
between the arc powers and their melting points was used
as a reference curve to estimate the surface temperature
of specimens. These experiments showed that the metal
temperature increased with H2 content in the plasma gas at
the atmospheric pressure under a same arc power. For
example, at 5.92 kW of arc power, the temperature difference (T) of Ar plasma (TAr ) and Ar-20% hydrogen plasma
(TAr+20%H2 ) was nearly 445 5 K, respectively. On the
other hand, we could unsuccessfully measure the precise
temperature in Ar and Ar-H2 plasma arc melting due to
rapid heat losses by high thermal conductivity of Cu.
Figure 2 shows the vapor pressure of main impurities and
Cu metal as a function of temperature based on the thermochemical equations.9) The impurities with higher vapor
pressures than that of Cu were expected to be eliminated
easily from Cu metal. On the other hand, the impurities such
as Ni, Co, Cr, Fe were not separated due to closer and lower
vapor pressures than that of Cu. In addition, Al and Si
impurities in Cu molten metal cannot be separated easily due
to the formation of Al2 O3 and SiO2 inclusions in the starting
material itself.4)
GDMS analytical results of the Cu metal purified by Ar/
Ar-H2 PAM for 20 and 60 min are shown in Table 1. Other
elements in the periodic table except impurities given in
table were not detected in the Cu matrix. The results indicate
that over all purity of the Cu starting was above 99.9962%.
On the same terms, this purity was raised up to 99.9979%
after Ar-20%H2 PAM. A slight increase in oxygen concentration during PAM in Ar for 60 min was due to some
unknown contamination during sample treatment. For the
description of experimental data, the removal degree, RD(%),
of each impurity is defined by eq. (1).
10-3
500
1000
1500
2000
2500
3000
3500
Temperature, T / K
Fig. 2 Vapor pressure of main impurities and Cu metal as a function of
temperature.
RDð%Þ ¼ 100ðCi Cf Þ=Ci
ð1Þ
where Ci and Cf , are the initial and final concentrations,
respectively. Since the Li and Na impurity concentrations in
Cu specimens before and after melting were close to its
respective detection limit of GDMS analysis, it was difficult
to clarify its behavior and then these results were omitted
from the calculation of removal degree. No significant
removal of gaseous (O, N) and gas-forming (C) impurities
were found in Ar/Ar-H2 PAM. Among halogen impurities,
Cl was removed very efficiently, whereas the initial concentration of F impurity was so small that we could not find its
behavior by Ar/Ar-H2 PAM. On the other hand, Se and As
were not affected in both Ar/Ar-H2 PAM. Non-metals
separation was limited to P and S. Zhu et al. ascribed that S in
Cu could be removed by the formation of H2 S as well as the
vaporization of S.4) In metallic impurities, removal of Zn, Pb,
Bi Ca and Mg in Cu were significantly higher. Among
transition metals, Pd removal was relatively high. The others
like Ni, Co, Cr and Fe have no significant effect in Ar/Ar-H2
PAM. However, W impurity element has slightly increased
from 0.001 ppm to 0.11 and 0.14 ppm after Ar/Ar-H2
PAM. In the previous work,10) we have found that W was
contaminated from the W-2%ThO2 cathode used as a plasma
torch and it is explained that this contamination could be
suppressed by efficient cooling system in the plasma torch.
Table 2 shows the RD(%) of each impurity from Cu during
PAM with Ar (RDAr ) and Ar-H2 (RDAr+20%H2 ) gases and the
difference in the removal degree, RD ¼ RDAr+20%H2 ð%Þ RDAr ð%Þ. We classified according to the RD(%) and the
difference in the removal degree of each impurity element
from Cu during PAM. Figure 3 shows the RD(%) relation of
each impurity removed from Cu after Ar/Ar-H2 PAM. For
Ar-PAM, Cl, Ca, Bi and S show a high RD(%), where as the
RD(%) of Zn, Pb and Mg was remarkably high for Ar-H2
PAM. The figure clearly indicated, in general, a higher
removal efficiency of impurities by Ar-H2 PAM when
compared to Ar PAM. This means that there was a certain
driving force chemically induced by activated hydrogen
atoms during Ar-H2 PAM. Tanaka et al.11) predicted that
the cause for the arc constriction in plasma melting process
is not by increased thermal conductivity but by increased
1828
J.-W. Lim et al.
Table 1
Impurity concentrations (mass ppm) in the Cu metal refined by Ar/Ar-H2 PAM.
Starting material
Ar
(20 min)
Ar
(60 min)
Ar+20%H2
(20 min)
Ar+20%H2
(60 min)
Li
0.002
<0:001
<0:001
<0:001
<0:001
C
N
1.34
1.4
1.25
1.7
1.33
1.7
1.27
1.0
1.20
1.2
Impurity
O
15.0
F
15.3
33.5
11.4
6.8
0.003
0.003
0.003
0.003
0.002
Na
0.002
<0:001
<0:001
<0:001
<0:001
Mg
0.08
0.04
0.02
0.004
0.002
Al
0.24
0.23
0.24
0.28
0.26
Si
0.31
0.30
0.28
0.24
0.22
P
S
4.28
5.22
3.55
3.41
2.95
1.25
3.68
1.76
2.35
0.33
0.005
Cl
0.19
0.05
0.03
0.03
K
0.12
0.06
0.07
0.01
0.005
Ca
0.03
0.01
0.005
0.002
0.002
Cr
0.008
0.008
0.008
0.007
0.008
Mn
1.25
1.11
1.09
1.07
1.08
Fe
0.86
0.84
0.83
0.85
0.85
Ni
Co
0.67
0.002
0.68
0.004
0.67
0.002
0.66
0.002
0.66
0.002
Zn
0.19
0.13
0.07
0.005
0.003
As
0.16
0.16
0.15
0.17
0.16
Se
0.07
0.07
0.06
0.07
0.07
Pd
0.02
0.008
0.005
0.005
0.006
Ag
5.19
5.20
5.01
5.21
5.04
Sn
0.02
0.02
0.02
0.03
0.02
Sb
W
0.11
0.001
0.10
0.06
0.11
0.11
0.10
0.12
0.11
0.14
Pb
0.84
0.42
0.18
0.20
0.02
Bi
0.01
0.005
0.002
<0:001
<0:001
Purity
(mass%)
99.9962%
99.9965%
99.9950%
99.9972%
99.9979%
Purity (mass%)
(except C,N,O)
99.9980%
99.9984%
99.9987%
99.9985%
99.9989%
Table 2 RD(%) of each impurity from Cu during PAM with Ar (RDAr ) and Ar-H2 (RDAr+20%H2 ) gases and the difference in the removal
degree, RD ¼ RDAr+20%H2 ð%Þ RDAr ð%Þ.
Impurity
Mg
RD(Ar) (%)
75
Impurity
RD(Ar+20%H2 ) (%)
Impurity
Difference in RD (%),
RD ¼ RD(Ar+20%H2 ) RD(Ar)
Mg
97.5
Mg
P
31.1
P
45.1
P
14
S
Cl
76.1
84.2
S
Cl
93.7
97.4
S
Cl
17.6
13.2
K
50
K
95.8
K
45.8
Ca
83.3
Ca
93.3
Ca
10
Zn
63.2
Zn
98.4
Zn
35
Pd
75
Pd
70
Pd
5
Pb
78.6
Pb
97.6
Pb
19
Bi
80
Bi
90
Bi
10
specific heat (Cp ) and thus enthalpy (H). However, when
10% of hydrogen gas was introduced into Ar as an additive
gas, there was an increase in thermal conductivity of a factor
of 10 at 3000 K. This increased thermal conductivity causes a
22.5
cooling of the outer edge of the arc and thus a reduced arc
cross-section followed by increased magnetic pinch pressure
and increased axial plasma flow and thus the enhanced
voltage. This voltage enhancement caused by the addition of
Effect of Ar/Ar-H2 Plasma Arc Melting on Cu Purification
50
140
Ar
40
120
∆ RD
100
∆ RD (%)
30
80
20
60
10
40
0
20
-10
Removal Degree, RD (%)
Ar+20%H2
0
Mg
P
S
Cl
K
Ca
Zn
Pd
Pb
Bi
Impurities in Cu
Fig. 3 Comparison between RD(%) of main impurities in the Cu metal
refined using Ar-PAM and Ar-H2 PAM.
10% hydrogen gas probably due to the large maximum in
thermal conductivity of hydrogen at the low temperatures
when hydrogen molecules dissociate in to atoms. In addition
to the enhanced removal efficiency of Mg, P, S, Cl, K, Ca, Zn,
Pb and Bi impurities in Cu, the removal efficiency of gaseous
impurities like oxygen, carbon and nitrogen was higher when
20% hydrogen gas was introduced in to Ar. This enhancement of RD(%) was attributed to the increase of metal vapor
transfer from a melt surface to gas phase by activated
hydrogen atoms as well as the increase in temperature of
the molten metal due to the higher thermal conductivity of
hydrogen.10,12)
4.
Conclusions
The refining effect of Ar/Ar-20%H2 PAM for Cu metal
and purity evaluation using GDMS were carried out.
Excellent removal degree, RD (>90%) for the impurities
like Zn, Pb, Mg, Cl, K, S, Ca and Bi in Cu was found using
Ar-H2 PAM. The impurities like K, Zn and Pd in Cu were
found to have higher RD(%) in Ar-H2 PAM compared to
Ar-PAM. For Ar-PAM, Cl, Ca, Bi and S impurities show the
high RD(%), whereas the RD(%) of Zn, Pb and Mg was
remarkably high for Ar-H2 PAM. In addition to the enhanced
1829
removal efficiency of Mg, P, S, Cl, K, Ca, Zn, Pb and Bi
impurities in Cu, the removal efficiency of gaseous impurities
like oxygen, carbon and nitrogen was higher when 20%
hydrogen gas was introduced in to Ar-PAM. The purity of
copper increased from 99.9962% to 99.9979(%) after Ar-H2
PAM due to high RD(%) of impurities with higher vapor
pressure than that of Cu metal.
Acknowledgement
This work was supported by a grant from the fundamental
R&D program for Core Technology of Materials funded by
the Ministry of Commerce, Industry and Energy, Republic
of Korea. One of the authors, Dr. N. R. Munirathnam thanks
Brain-Pool program of Korean Federation of Science and
Technology for the financial support as visiting researcher at
KIGAM.
REFERENCES
1) S. P. Murarka, R. J. Gutmann, A. E. Kaloyeros and W. A. Lanford: Thin
Solid Films 17 (1993) 257–266.
2) C. Cagran, B. Wilthan and G. Pottlacher: Thermochim. Acta 445
(2006) 104–110.
3) M. Uchikoshi, T. Kekesi, Y. Ishikawa, K. Mimura and M. Isshiki:
Mater. Trans. 38 (1997) 1083–1088.
4) Y. Zhu, K. Mimura, Y. Ishikawa and M. Isshiki: Mater. Trans. 43
(2002) 2802–2807.
5) K. Kaneko, N. Sano and Y. Matsushita: J. Iron Steel Inst. Jpn. 62 (1976)
43–52.
6) K. Mimura and M. Nanojo: Proc. 8th Int. Symp. Plasma Chem.,
(IUPAC, Tokyo, 1987) pp. 886–1891.
7) K. Mimura, S. W. Lee and M. Isshiki: J. Alloys Compd. 221 (1995)
267–273.
8) J. W. Bae, J.-W. Lim, K. Mimura and M. Isshiki: Mater. Lett. 60 (2006)
2604–2605.
9) O. Knache, O. Kubaschewski and K. Hesselmann: Thermochemical
Properties of Inorganic Substances, (Springer-Verlag, New York,
1991) pp. 2370.
10) J.-W. Lim, K. Mimura and M. Isshiki: J. Mater. Sci. 40 (2005)
4109–4111.
11) M. Tanaka and J. J. Lowke: J. Phys. D: Appl. Phys. 40 (2007)
R1–R23.
12) D. Elanski, J.-W. Lim, K. Mimura and M. Isshiki: J. Alloys Compd.
413 (2006) 251–258.