See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/248134146

Deoxidation of Liquid Copper with Reducing O 2 /CH 4 Flames

Article  in  Metallurgical and Materials Transactions B · June 2010

DOI: 10.1007/s11663-010-9346-8

CITATIONS READS

4 206

2 authors, including:

Tanai Marin
M4Dynamics Inc.
19 PUBLICATIONS   133 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Copper Refining View project

Development of M4Dlib, Thermodynamic and Multiphysics Simulation View project

All content following this page was uploaded by Tanai Marin on 24 November 2015.

The user has requested enhancement of the downloaded file.

Deoxidation of Liquid Copper with Reducing O2/CH4 Flames
T. MARIN and T. UTIGARD

The rate of deoxidation of molten copper during top blowing with various reducing gases has
been investigated using thermogravimetry. It was observed that the rate of deoxidation increases
with an increasing flow rate of H2 or CO and that H2 is a more effective reducing reagent than
CO. The rate of deoxidation using methane was measured for O2/CH4 ratios from 1.5 to 2.0. As
expected, the deoxidation rate decreased with an increasing O2/CH4 feed ratio because the flame
became less reducing. For all tests, initially there is a linear decrease in mass as oxygen is
removed. However, for some experiments, after some time, a sudden acceleration in the rate of
mass loss occurs. Using video and X-ray imaging, it was found that this pattern corresponded to
gas evolution from within the molten copper. This finding can be explained by the sudden water
vapor evolution because the hydrogen dissolved in the copper reacts with the remaining oxygen,
and ‘‘boiling’’ takes place, leading to an enhanced stirring of the copper.

DOI: 10.1007/s11663-010-9346-8
Ó The Minerals, Metals & Materials Society and ASM International 2010

I. INTRODUCTION by Kikuchi et al.[6] for oxygen concentrations greater

DURING liquid copper fire refining, two main steps
are necessary to remove the impurities present in the
blister copper before casting—namely oxidation fol-
lowed by reduction. This study focuses on the latter. The
fire refining process normally is carried out in batch
mode in a reactor called an anode furnace; each step
involved in this operation can take several hours.
During this time, burners are used to maintain the
temperature inside the reactor. The objective of this
study was to determine the rate of reduction using
reducing gases flowing above the surface of the molten
copper. Industrially, at the start of reduction, the copper
may have about 0.5 to 1 pct O and, at the end, have less
than 0.1 pct.
The gaseous reduction of copper has been investi-
gated using carbon monoxide,[1–4] different gas mixtures
including CO/CO2,[5,6] H2/N2,[7] and H2/Ar,[8] as well as
reformed natural gas.[3,9,10] Using carbon monoxide,
Andreini et al.[1] injected CO into about 4 kg of molten
copper and studied the deoxidation kinetics by analyz-
ing the CO2 in the off-gas. In the composition range
from 0.005 to 0.1 wt pct, they concluded that liquid
mass transfer was the rate-limiting step. Nanda and
Geiger[2] also did a study with low oxygen concentra-
tions in copper (below 0.05 wt pct), which confirmed
that liquid mass transfer was the rate-limiting step. For
higher oxygen concentrations in the melt (0.1 to
1 wt pct), Themelis and Schmidt[4] reported that the
rate was independent of the oxygen concentration in the
liquid and that mass transfer in the gas phase was
the rate-limiting step. The same conclusion was found
T. MARIN, Research Metallurgist, is with Vale-Inco, Sheridan
Park, Mississauga, Ontario, Canada. T. UTIGARD, Professor, is with
the University of Toronto, Materials Science and Engineering,
Toronto, Ontario, Canada. Contact e-mail: utigard@ecf.utoronto.ca.
Manuscript submitted April 25, 2006.
than 0.25 wt pct. Although somewhat inconclusive, it
can be concluded that the rates are only weakly
dependent on the temperature.[1,6]
With respect to the H2 reduction of oxidized copper,
there is very little information available. Capocchi and
De Lazzari[7] injected two different H2/N2 mixtures (15
and 33 vol pct H2) into molten copper at 1170 °C and
reported that the rate of deoxidation was constant for
oxygen concentrations between 0.4 and 0.15 wt pct.
They concluded that gas-phase mass transfer was the
rate-limiting step under their conditions. Fukunaka
et al.[8] carried out a study of reducing a levitated
copper droplet in an H2/Ar mixture at a very high
temperature (1697 °C). From their results, they reported
mixed rate-controlling steps depending on the oxygen
content in the copper. They also pointed out the
importance of the dissolution and accumulation of
hydrogen in the copper on the reduction mechanism.
This accumulation process started at the beginning of
the deoxidation process.
Finally, the use of natural gas also has been studied.
The use of reformed natural gas was performed by
Andreini.[3] The study focused on low oxygen concen-
trations, and liquid mass transfer was found to be the
rate-limiting step. It was suggested by Andreini[3] that
H2 is a more effective reducing reagent than CO. Riveros
et al.[9] observed during the bubbling of natural gas
directly into molten copper that the rate of reduction
was nearly linear with respect to time. It was further
noted that the reduction rate increased with an increase
in the gas flow rate. More recently, Soltanieh and
Karimi[10] did a similar study in which natural gas was
blown into a 40-g sample of liquid copper containing
almost 1.8 wt pct of oxygen. In their case, they con-
cluded that the process initially was controlled by
diffusion followed by mixed control.
Based on this literature review, it is clear that there
is information lacking about the kinetics of copper
deoxidation using a reducing gas formed by the

METALLURGICAL AND MATERIALS TRANSACTIONS B

substoichiometric combustion of natural gas. One
important objective of this investigation was to study
the deoxidation kinetics for a ‘‘stagnant’’ bath in order
to get a better understanding of the various possible
rate-controlling steps that may occur in industrial
practice. For that purpose, an experimental thermal
gravimetric setup was developed that allowed for the
determination of continuous weight changes for samples
with about 1 to 1.5 kg of molten copper.
II. EXPERIMENTAL
The experimental apparatus used to measure the
deoxidation rates consisted of a high-capacity thermo-
gravimetric unit.[11] A MgO crucible (6.9 cm inner
diameter [ID], 12 cm height) that contained the copper
sample was placed inside an electrically heated furnace.
This crucible rested on top of a vertical Al2O3 tube
(5.7 cm outer diameter [OD], 43 cm long, filled with
insulating material) that was loaded onto a top-loading
22-kg maximum-capacity digital scale with a readability
of 0.1 g. The overall setup is shown in Figure 1. A single
Al2O3 lance (6.3 mm OD, 4 mm ID) was used to top
blow CO/N2, H2/N2, and CO/H2/N2 mixtures onto the
copper surface. For the experiments with O2/CH4
flames, two concentric lances (5.6 mm OD, 3 mm ID
for the inner tube, and 17.5 mm OD and 11 mm ID for
the outer tube) were used. In this case, O2 was blown
into the outer lance, and CH4 was blown into the inner
lance with independent control of the gas flow rate for
Fig. 1—A schematic of the experimental setup for the large-capacity
TGA.
each gas to achieve the desired ratio. The experimental
procedure was similar to that described previously.[11]
The starting oxidized copper samples were formed in the
thermo gravimetric analysis (TGA) by blowing air onto
the copper. After reaching the desired oxygen content in
the melt as given by the weight increase, the air was
flushed out with nitrogen, and then the atmosphere was
switched to the desired reducing mixture to start the
deoxidation experiment. It was found that with just
nitrogen flushing at the flow rates used in these tests,
there was no noticeable mass change. This result shows
that, although the system was not completely sealed,
inert gas flushing was sufficient to prevent air ingress.
During the experiments starting with approximately
1 wt pct oxygen in the copper, the mass of the liquid
metal would decrease continuously. The data were
sampled by a data acquisition system every 4 seconds,
which gave a very high time resolution for experiments
that lasted between 30 and 60 minutes. The experimen-
tal conditions are summarized in Table I for CO/H2/N2
gas mixtures and in Table II for O2/CH4 mixtures.
III. RESULTS AND DISCUSSION
A. CO/N2 mixtures
Figure 2 shows the mass of the dissolved oxygen in
the molten copper as a function of time for the top-
blowing 20 pct CO/80 pct N2 at 1200 °C at a total gas
flow rate of 2950 cm3/min at standard conditions.
Initially, there is a linear decrease in mass until about
0.3 wt pct oxygen. After this decrease, the rate slows,
and the oxygen concentration starts to follow an
asymptotic curve. This behavior is consistent with
previous investigations,[1–4,6] which have reported that
gaseous mass transfer is the rate-controlling step at a
high oxygen concentration in copper, and liquid mass
transfer becomes the controlling step for lower oxygen
concentrations.
The effect of the total gas flow rate is observed in
Figure 3 in which the mass removed in the linear range
is plotted against time for different flow rates of the
same CO-N2 mixture. By plotting the mass removed as a
function of time, the starting point for all curves is the
same, and the difference in the rate of reduction, given
by the slope of the curves, is better appreciated. As
shown in Figure 4, the rate of reduction increases with
the gas flow rate. These results further support that, at
high oxygen contents, gaseous mass transfer is the rate-
limiting step.
B. H2/N2 Mixtures
During reduction with H2-N2, a linear trend also is
observed for the mass change with respect to time. The
rate of reduction seems to be constant for nearly the
whole range of oxygen compositions in the molten
copper. This trend is observed in Figure 5 in which the
mass removed during the reduction of dilute liquid
Cu-O alloys at 1200 °C with 40 pct H2/60 pct N2 at
different flow rates is plotted as a function of time.
METALLURGICAL AND MATERIALS TRANSACTIONS B
 Table I. Experimental Conditions and Results for Reduction of Liquid Cu-O Alloys at 1200 °C in the Large-Capacity TGA

Gas Composition,*
vol pct
Run Cu Mass Reduction Rate** Reduction Rate** Gas Flow Rate 
# g CO H2 g/min wt pct/min cm3/min
1 1005.5 20 — 0.119 0.011 2400
2 1005.5 — 40 0.156 0.016 1120
3 1005.5 — 40 0.432 0.043 2400
4 1375.9 20 40 0.256 0.019 940
5 1368.7 20 40 0.325 0.024 940
6 1384.9 — 40 0.167 0.012 900
7 1384.1 20 — 0.077 0.0056 800
8 1384.1 20 — 0.147 0.0106 1750
9 1384.1 20 — 0.203 0.015 2950
10 1356.1 — 40 0.407 0.030 1740
11 1304.0 20 40 0.497 0.038 1500
12 1479.2 20 40 0.603 0.041 1500
*Balance nitrogen.
**During linear range of mass loss.
 
At 1 atm 20 °C.

Table II. Experimental Conditions and Results for Reduction of Liquid Copper by Top Blowing of Substoichiometric Flames

Reactants Products
Estimated Rate of
Feeding Equilibrium
Reduction**
Flow Rate, Amount, Flow Rate*
cm3/min Rate Mol/Mol CH4 cm3/min wt pct/min
Run Cu Mass Ratio, vol/vol of Reduction
# g CH4 O2 O2/CH4 wt pct/ min H2 CO H2 CO rH2 rCO rcombined
13 1331.0 1960 0 0.0 0.083 2 0 3920 0 0.165 0.000 0.165
14 1328.0 1860 2940 1.6 0.058 0.429 0.409 798 761 0.034 0.020 0.053
15 1334.1 1915 3240 1.7 0.040 0.294 0.304 563 582 0.024 0.015 0.039
16 1376.5 1879 3380 1.8 0.028 0.189 0.209 355 393 0.015 0.010 0.025
*Estimated from the equilibrium H2 and CO amount in the flame.
**Based on Eqs. [8] and [9].

The effect of the gas flow rate on the rate of reduction COðgÞ þ O ! CO2ðgÞ ½1

is observed in Figure 6 for 40 pct H2/60 pct N2 mixtures
in which the relative rate of reduction from the results of
runs 2, 3, 6, and 10 is plotted as a function of the total DGo ð1 wt pctÞ ¼ 96; 705 J/mol at 1200  C
gas flow rate. Figure 6 shows that the rate of reductions
increases linearly with an increasing gas flow rate. H2ðgÞ þ O ! H2 OðgÞ ½2

C. CO/H2/N2 Mixtures DGo ð1 wt pctÞ ¼ 103; 680 J=mol at 1200  C

When comparing the reactions for the deoxidation of
liquid copper with CO or H2, both have almost the same
thermodynamic affinity for oxygen. This trend is seen
from the change in Gibbs free energy, when considering
the dissolution of oxygen into copper and assuming
1 wt pct oxygen as the standard state. The value for the
change in Gibbs free energy was calculated from
the data provided by Sigworth and Elliott[12] for the
dissolution of oxygen into molten copper and from
the thermodynamic database from HSC software
(OUTOTEC, Espoo, Finland).[13]
Experimentally, however, it is observed that H2 is more
efficient than CO in deoxidizing the copper at 1200 °C
under otherwise the same conditions as initially sug-
gested by Andreini.[3] This trend is observed in Figure 7
in which the relative rate of reduction for both CO and
H2 is plotted on the same basis (considering flow rates of
pure CO or H2). These results support the observations
by Andreini.[3]
For high mass transfer rates and gas–liquid systems,
the penetration theory from Higbie can be applied,[14] in

METALLURGICAL AND MATERIALS TRANSACTIONS B

 Oxygen content in copper, wt% 0.8 0.018

0.7 0.016

Rate of reduction, wt%/min

0.014
0.6
0.012
0.5
0.010
0.4
0.008
0.3
0.006
0.2
0.004
0.1 0.002

0.0 0.000
0 10 20 30 40 50 60 70 80 90 100 0 1000 2000 3000
Time, min 3
Gas flow rate, cm /min

Fig. 2—The oxygen removed from the liquid copper during reduction Fig. 4—The rate of reduction as a function of the gas flow rate for
with 20 vol pct CO (balance N2) at 2950 cm3/min of total gas flow the reduction of liquid copper with a top blowing of 20 pct CO/
rate and 1200 °C as a function of time in the large-capacity TGA. 80 pct N2 mixtures at 1200 °C in the large-capacity TGA.
The dotted line shows the best fit for a linear range of mass loss.

8 1.4

7 1.2 2400 cm3/min

Mass removed, wt%

6 1.0 1749 cm3/min

Mass removed, g

5
2950 cm3/min 0.8
4
0.6
3 3
1750 cm /min
0.4 1120 cm3/min
2
0.2
1 800 cm3/min 900 cm3/min

0 0.0
0 10 20 30 0 10 20 30 40 50 60
Time, min Time, min

Fig. 3—The mass removed vs time during the reduction of liquid Fig. 5—The relative mass removed vs time during the reduction of
copper with a top blowing of 20 pct CO/ 80 pct N2 at different gas liquid copper with a top blowing of 40 pct H2/60 pct N2 gas at
flow rates at 1200 °C in the large-capacity TGA. 1200 °C at different gas flow rates (runs 23, 24, 69, and 76).

which case, the mass transfer coefficient (km) is propor- sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

tional to the square root of the diffusion coefficient km;H2 =N2 DH2 ;N2
¼ ½4

(DAB), which is expressed as follows: km;CO=N2 DCO;N2
rffiffiffiffiffiffiffiffiffi
DAB The values of the diffusion coefficients at 1200 °C can
km / ½3

t be estimated from the Chapman–Enskog theory of
ordinary diffusion in gases at low density as described by
where km is the average mass transfer coefficient, DAB is Bird et al.[14] and expressed as follows:
the diffusion coefficient in m2/s, and t is the residence rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
time in seconds during which the mass transfer takes  
place.  2  1:8583  107  T3  MW 1
þ MW
1
m A B
When mass transfer in the gas phase is the rate- DAB ¼ ½5

s r2AB  XAB;D
limiting step, which is the case for high oxygen concen-
trations in liquid copper and during the linear range of where MWA and MWB are the molecular weights of
mass loss, then the ratio in the mass transfer coefficient component A and B; rAB is given by the following
between H2/N2 and CO/H2 is given by the square root expression:
of the diffusivity ratio, as shown in the following
equation: rAB ¼ 0:5  ðrA þ rB Þ ½6

METALLURGICAL AND MATERIALS TRANSACTIONS B

 0.05 Table III. Properties of Gases Used to Estimate Diffusion
Coefficients

Rate of reduction, wt%/min

0.04 Properties

MW r e/j
0.03 Gas g/mol Å K
H2 2.016 2.915 38
CO 28.01 3.59 110
0.02
N2 28.02 3.681 91.5

0.01
From Figure 7, the following expressions are
obtained by linear fit for the experimental rate of
0.00 reduction of the dilute liquid Cu-O melts with CO/N2
0 1000 2000 3000 (rCO) and H2/N2 ðrH2 Þ as a function of the flow rate of the
Gas flow rate, cm /min
3 reducing gas in the mixtures QCO and QH2 ; respectively.
   3
wt pct cm
Fig. 6—The rate of reduction as a function of the total gas flow rate rCO ¼ 2:58  105  QCO ½8

during the deoxidation of liquid copper with 40 pct H2/60 pct N2 at min min
1200 °C in the large-capacity TGA.
   3
wt pct 5 cm
rH2 ¼ 4:22  10  QH2 ½9

min min
0.05
-5
r H2 = 4.2x10 QH2 From these equations, the experimental ratio between
Rate of reduction, wt%/min

2 H2 the rate of deoxidation for the two different gas mixtures

0.04 R = 0.97
is obtained. An increase of 1.6 times for the same flow
rate of reducing reagent is calculated when changing
0.03 from the CO/N2 to the H2/N2 gas mixture. This value
agrees very well with the estimated value of 1.87 based
on the assumptions presented previously.
0.02
It is reasonable to assume that for a CO, H2, and N2
CO gas mixture, the rate of liquid copper reduction can be
0.01 -5 estimated based on the volumetric flow rates of CO and
r CO = 2.6x10 Q CO
H2 by adding Eqs. [8] and [9] as in the following
2
R = 0.96 equation:
0.00
0 500 1000 1500 r ¼ rCO þ rH2 ½10

3
Flow rate of reducing gas, cm /min By using Eq. [10], the rate of reduction for runs 4, 5, 11,
and 12 can be estimated and compared with the
Fig. 7—A comparison of the kinetics of the reduction of liquid cop-
per between CO and H2 as a reducing reagent at 1200 °C.
measured values obtained experimentally. This trend is
observed in Figure 8. Although the experimental data
points show some scatter, the estimated values agree
within 11 to 18 pct of the measured values.
XAB,D can be found tabulated in the book by Bird A very important observation when reducing copper
et al.[14] Finally, the value of eAB/j can be calculated with CO/H2 mixtures in the 1- to 1.5-kg capacity TGA is
from the following equation: that the melt at some point suddenly starts to loose
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi weight at a much higher rate. This loss is seen clearly in
eAB eA eB
¼  ½7
Figure 9 in which the oxygen concentration is plotted as
j j j a function of time for runs 4 and 5. At some point
during the reduction, a sudden mass loss takes place,
In Table III, the values of MW, r, and e/j for H2, CO,
followed by a high rate of weight loss down to about
and N2 are given. Using these values and evaluating
0.05 wt pct O. This loss can be explained by elemental
Eq. [5] at 1200 °C, the calculated diffusion coefficients
hydrogen, which has been dissolved in the molten
are 1.08 9 103 m2/s and 3.08 9 104 m2/s for H2/N2
copper, starting to react with the dissolved oxygen,
and CO/N2, respectively. Introducing these values in
leading to water vapor formation within the bulk of the
Eq. [4] gives an estimate for the ratio of the mass
molten copper, which is expressed as follows:
transfer coefficient equal to 1.87. Therefore, it is
expected that the rate of deoxidation of the dilute O þ 2H ! H2 OðgÞ ½11

liquid Cu-O alloys with H2/N2 gas mixtures will be
approximately 1.9 times faster than with CO/N2 If the local oxygen concentration in the liquid under the
mixtures. tip of the lance is sufficiently low, then the hydrogen

METALLURGICAL AND MATERIALS TRANSACTIONS B

 Rate of reduction, wt%/min 0.05
4.E-02

Hydrogen concentration, atom%

0.04 measured pH2O = 1 atm
estimated

0.03 3.E-02

0.02
2.E-02

0.01

1.E-02
0.00
0 500 1000 1500 2000
3
Total gas flow rate, cm /min
0.E+00
0.0001 0.001 0.01 0.1 1
Fig. 8—A comparison between the measured and the estimated rates
of reduction for top blowing of CO/H2/N2 mixtures as a function of Oxygen concentration, wt%
the total gas flow rate. Also shown is the best fit for measured data
(solid line). Fig. 10—A calculated isobar of H2O = 1 atm at 1473 K for a
dilute Cu-O-H system as a function of an oxygen and hydrogen
concentration.
1.0
Oxygen concentration, wt%

0.8 “Boiling” of H2O gas

0.6

0.4

0.2
Run 4
Run 5
0.0
0 10 20 30 40 50
Time, min

Fig. 9—Oxygen content vs time for runs 4 and 5.

could dissolve locally into the melt. As shown in

Figure 10, the solubility of hydrogen increases rapidly
for oxygen contents below about 0.05 wt pct.[15] If,
locally, the oxygen content drops to below this value,
then the local hydrogen content can increase rapidly.
With the very high diffusivity of hydrogen, it then is
possible that hydrogen may diffuse into neighboring
regions where the local oxygen content has not dropped
to the same extent as that directly beneath the lance tip.
This diffusion then would lead to super saturation and a
local water vapor pressure higher than 1 atmosphere,
and ‘‘boiling’’ would commence.
In our setup, it was possible to observe the surface of
the molten metal during the experiments through a
quartz window. It was seen that at the moment of the
break in the curve of mass loss, small bubbles were
released from the liquid in the zone underneath the tip
of the lance. This release is shown in Figure 11 in which
small bubbles are observed to evolve beside the lance tip.
Fig. 11—The surface of liquid copper during the reduction with
CO/H2 mixtures after a sudden increase in the rate of mass loss.
Based on video recordings, it was confirmed that the
initiation of bubble formation corresponded to the
increasing mass loss determined by the TGA. As soon as
the bubbling started, the copper melt was stirred,
exposing new and fresh copper to the surface. With
continued stirring, H2O gas can be evolved at a pressure
less than 1 atmosphere from the surface, and the copper
sample will continue to loose mass at a very high rate.
In separate experiments using a high-temperature
X-ray imaging system, after some time of hydrogen
reduction, small gas bubbles were observed to form on
the inside of the crucible below the upper surface of the
copper melt (Figure 12). In those experiments, pure H2

METALLURGICAL AND MATERIALS TRANSACTIONS B

 Equilibrium amount of products, mol
2.0 H2

H2O
1.5

1.0
CO CO2

0.5

0.0
0.0 0.5 1.0 1.5 2.0
O2/CH4 vol ratio

Fig. 13—An equilibrium amount of gas species vs the O2/CH4 ratio

for reducing flames for 1 mol of CH4 at 1200 °C.

Fig. 12—An X-ray visualization of the liquid copper during the

reduction with 100 H2(g).

estimated rate of reduction, wt%/min

0.12

gas was used. In regard to the sudden mass loss, the
video recordings of bubbles evolving from the surface of
the copper bath in addition to the X-ray images all
demonstrate that indeed water vapor may form within
the melt.
It is not clear why this phenomenon was observed
more readily when using 20 pct CO-40 pct H2-40 pct N2
than with 40 pct H2-60 pct N2. A possible explanation is
that when 20 pct CO is used together with 40 pct H2,
less H2 is consumed by the reduction reactions because
CO also reduces the copper. Therefore, there is more H2
available to dissolve into the copper as elemental H,
building up the concentration faster and reaching the
critical activity. As the oxygen content in the copper
decreases and gas-phase mass transfer no longer limits
the reaction rate, such hydrogen build-up is feasible.
Another possible explanation may be that the stirring of
the copper surface and melt is different for the two
gases. Stirring of the copper during the experiments
could be caused by the gas jet itself or by a flow induced
by surface tension gradients. It is known that the surface
tension of copper decreases with an increasing oxygen
content. Therefore, just below the reducing lance, the
lower oxygen content will lead to a locally higher surface
tension. This tension then will lead to a Marangoni flow
from around the rim toward the center of the crucible
where the lance is located. The transport of hydrogen-
rich copper melt to other regions could be affected, and
the initiation of the boiling process also would change.
D. O2/CH4 flames
The experimental conditions and results are tabulated
in Table II for the top blowing with substoichiometric
O2/CH4. From thermodynamic characterizations of
O2/CH4 flames,[16] the equilibrium composition as a
function of the O2/CH4 ratio can be calculated as shown
in Figure 13. In this case, only H2 and CO will
0.10
0.08 rH2
r=rH2 + rCO
0.06
0.04
rCO
0.02
0.00
0.0 0.5 1.0 1.5 2.0
O 2 /CH 4 ratio, vol/vol
Fig. 14—The estimated rate of reduction based on the CO and H2
equilibrium composition for 1000 cm3 of CH4 and different O2/CH4
ratios.
participate in the deoxidation; therefore, the amount
of these species generated by a given O2/CH4 ratio must
be known. This amount has been calculated for the
ratios corresponding to those used in the experiments in
Table II for 1 mol of CH4 as a unit base.
Based on Eqs. [8] through [10], the relative rate of
reduction for a reducing O2/CH4 flame can be esti-
mated for the whole range of O2/CH4 ratios. This
estimation is shown in Figure 14 considering a flow rate
of 1000 cm3/min of CH4 at standard conditions.
For the gas flow rates corresponding to runs 13 to 16,
the resulting flow rate of H2 and CO in the products can
be calculated and the relative rate of reduction can be
estimated from Eqs. [8] through [10]. This calculation is
presented in Table II. The estimated rate of reduction r
in Table II is compared with the measured value in
Figure 15 for O2/CH4 ratios above 1.5, in which case the
agreement is excellent. However, the estimated rate for

METALLURGICAL AND MATERIALS TRANSACTIONS B

 0.08
Rate of reduction, wt%/min
0.07
0.06
0.05
0.04
0.03
0.02
0.01
0.00
1.5 1.6
Fig. 15—The measured and estimated rates of reduction of liquid dilute
Cu-O alloys with a top blowing of reducing flames at 1200 °C.
pure CH4 is almost twice the measured value. In this
case, the resulting H2 gas flow rate exceeds by almost
four times the maximum flow rate from which Eq. [9]
was fitted. It is possible then, that extrapolation at this
flow rate is no longer valid and, therefore, leads to the
overestimation of the rate of reduction. In addition,
CH4 may not decompose properly. In the case of runs 14
to 16, the resulting flow rates of both H2 and CO fall in
the range from which Eqs. [8] and [9] were obtained and
show excellent agreement with the experimental results.
IV.
For the deoxidation of copper-oxygen melts with CO/
H2/N2 mixtures, mass transfer in the gas phase is the
rate-limiting step for oxygen concentrations greater than
about 0.3 wt pct. The rate of reduction of liquid copper
by top blowing of a reducing methane flame can be
estimated from the composition of CO and H2 in the
flame. Experimentally measured rates of reduction by
the top blowing of an O2/CH4 reducing flame are in
excellent agreement with the estimated rates of reduc-
tion for the O2/CH4 ratios between 1.5 and 2.
Although CO and H2 present almost the
same potential to deoxidize liquid copper from a
View publication stats
thermodynamic point of view, it was found that under
similar experimental conditions, hydrogen is a more
Measured effective reducing reagent because of the higher diffu-
Estimated sivity of H2 in N2 than CO in N2. The participation of
dissolved hydrogen on the reduction reaction was
observed using direct visualization of the molten surface
during the reaction. For the first time, a sharp acceler-
ation of the rate of reduction caused by the homoge-
neous reaction of deoxidation within the melt has been
observed and measured.
ACKNOWLEDGMENTS
1.7 1.8 1.9 2 The authors wish to express their gratitude to
O2 /CH4 ratio, vol/vol NSERC (Canada) and to CONICYT (Chile) for pro-
viding funds to carry out this investigation.
REFERENCES
1. R.J. Andreini, J.S. Foster, and R.B. Philips: Metall. Trans. B,
1977, vol. 8, pp. 633–38.
2. C.R. Nanda and G.H. Geiger: Metall. Trans., 1971, vol. 2,
pp. 1101–06.
3. R.J. Andreini: DAI, 1980, vol. 40 (7), p. 3335.
4. N.J. Themelis and P.R. Schmidt: TMS-AIME, 1967, vol. 239 (9),
pp. 1313–18.
5. A. Kikuchi, Y. Yusa, and T. Tadaki: J. Jpn. Inst. Met., 1980,
vol. 44 (6), pp. 884–91.
6. A. Kikuchi, M. Ayusawa, T. Tadaki, and S. Maeda: J. Jpn. Inst.
Met., 1980, vol. 44 (6), pp. 665–71.
7. J.D.T. Capocchi and C.P. De Lazzari: EPD Cong. The Minerals,
Metals and Materials Society/AIME, 1997, pp. 107–18.
8. Y. Fukunaka, K. Tamura, N. Taguchi, and Z. Asaki: Metall.
Trans B, 1991, vol. 22, pp. 631–38.
CONCLUSIONS 9. G. Riveros, A. Warczok, L. Voisin, and T. Marin: Yazawa Int.
Symp. Metallurgical and Materials Processing: Principles and
Technologies, vol. II, High Temperature Metal Production,
San Diego, CA, 2003, pp. 431–44.
10. M. Soltanieh and Y. Karimi: Can. Metall. Q., 2005, vol. 44 (3),
pp. 429–34.
11. T. Marin and T. Utigard: JOM, 2005, vol. 57 (7), pp. 58–62.
12. G.K. Sigworth and J.F. Elliott: Can. Metall. Q., 1974, vol. 13 (3),
pp. 455–61.
13. HSC Chemistry, version 5.0, OUTOTEC, Espoo, Finland.
14. R.B. Bird, W.E. Stewart, and E.N Lightfoot: Transport Phenom-
ena, 2nd ed., Wiley, New York, NY, 2002.
15. T. Marin: Ph.D. Dissertation, University of Toronto, Toronto,
Ontario, 2005.
16. T. Marin, T. Utigard, A. Warczok, and G. Riveros: 5th Int. Conf.
Copper 2003, Santiago, Chile, 2003, pp. 397–412.
METALLURGICAL AND MATERIALS TRANSACTIONS B