The speciation of antimony ions in the high temperature glass

melts is a subject of growing interest. One of the reasons is the

elucidation of the fining process, i.e., the removal of bubbles
originating from raw materials as well as from included air.
Fining is believed to be achieved by evolution of oxygen gases
(bubbles) released during the reduction of fining ions such as
As
5+
, Sb
5+
or SO4
2
(SO3).
16
Because of the highly toxic nature
of arsenic, the latter two are used mainly now. The qualitative
explanation of fining by antimony ions is shown in Fig. 1. At
the first step of melting of the glasses, foam is present in glass
melts, because of the decomposition of starting glass-materials
and the contamination of vapor-phase gas. The big bubbles rise
freely in the glass melts and move out to the vapor-phase. On
the contrary, the small bubbles remain in the glass melts. At a
high temperature, redox equilibria of fining ions such as
antimony ions in glass melts shifts to the reduced direction and
evolution of oxygen gas supports the rise of the small bubbles.
When the temperature of the melts was lowered at the glass-
forming process, the redox equilibria shifts to the oxidized
direction. Any oxygen gas which remained becomes
incorporated into the glass network. Now it becomes important
to deduce why antimony ions have been used as effective
refining ions. The temperature dependence of the redox
equilibria of antimony ions in glass melts must be throughly
examined. Generally, the equilibrium concentration of Sb
5+
and
Sb
3+
was measured by chemical analysis of the quenched
glass.
7,8
However, possible deviation of redox equilibria may
occur during cooling the melts. Thus, in situ determination of
the redox equilibrium in high temperature melts has been
required. Electrochemical methods such as voltammetry have
been successfully applied to the study of the redox equilibria of
multivalent ions in glass melts.
919
The half wave potential, E1/2,
can be related to the equilibrium concentration ratio between
redox pair ions through Nernsts equation. E1/2 values of
reduction of Sb
5+
in silicate and borate melts were examined by
a differential pulse voltammetry (DPV).
15
E1/2 shifted to the
45 ANALYTICAL SCIENCES JANUARY 2001, VOL. 17
2001 The Japan Society for Analytical Chemistry
Voltammetric Studies of Antimony Ions in Soda-lime-silica
Glass Melts up to 1873 K
Hiroshi YAMASHITA,

Shigeru YAMAGUCHI, Ryuichi NISHIMURA, and Takashi MAEKAWA

Department of Applied Chemistry, Faculty of Engineering, Ehime University,
3 Bunkyo-cho, Matsuyama, Ehime 7908577, Japan
The half wave potential of reduction of Sb
5+
in 16Na2O10CaO74SiO2 glass melts was examined by differential pulse
voltammetry up to 1873 K. The half wave potential shifted to the positive direction with an increase in temperature. The
results indicate that the equilibrium of Sb
5+
/Sb
3+
shifted to negative direction with an increase in temperature. The half
wave potential shifted to positive direction (48 mV at 1473 K) when the atmosphere over the melts changed from pure
oxygen gas to air, in agreement with the theoretical prediction. The reversibility of Pt:O2 reference electrode is
confirmed.
(Received June 20, 2000; Accepted September 29, 2000)
Fig. 1 Qualitative explanation of fining by antimony ions (see text).

To whom correspondence should be addressed.

negative direction with an increase in content of sodium oxide
and shifted to the positive direction with an increase in
temperature; thus Sb
3+
/Sb
5+
equilibrium shifted to the oxidized
direction with an increase in the basicity or with a decrease in
temperature. The substitution of Na2O by CaO shifted E1/2 to
the positive direction. Soda-lime-silica glass is one of the
fundamental glasses in commercial uses. It was argued that
CaO is a weak basic oxide compared to Na2O.
15
A square wave
voltammetry (SWV) study of reduction of antimony ions has
also been published.
11
However, in the previous studies
including our results, the data in high temperature region
beyond 1500 K were not presented. In order to see the fining
phenomena deeply, it is necessary to obtain more precise data
about E1/2. In this research, E1/2 values of the melts of soda-
lime-silica system, i.e., 16Na2O10CaO74SiO2 in mol ratio,
were determined by DPV. The usefulness of antimony ions in
fining regents will also be seen.
Experimental
Electrode reaction
The DPV uses three electrodes: i.e., a reference electrode, a
working one and a counter one, in which current(i)potential(E)
curves are recorded during the electrolysis.
15
Potential is
measured with respect to that of a reference electrode, which
corresponds to the following reaction:
(1/2)O2 + 2e = O
2
. (1)
On the working electrode, the reductions of solute metal ion
take place:
Ox + ne = Red. (2)
In the case of the reduction of Sb
5+
, Eq. (2) is replaced by
Sb
5+
+ 2e = Sb
3+
. (3)
Thus, the overall reaction can be written by
Sb
5+
+ O
2
= Sb
3+
+ (1/2)O2 (4)
or
1/2(Sb2O5)in glass = 1/2(Sb2O3)in glass + 1/2O2. (4)
The voltage, E, the difference between working and reference
electrode potentials, is related to a standard potential, E
0
, or a
formal potential, E
0
:
E = E
0
+ (RT/2F) ln a(Sb
5+
)/a(Sb
3+
) + (RT/nF) ln a(O
2
)
(RT/4F) ln p(O2)
= E
0
+ (RT/nF) ln([Sb
5+
]/[Sb
3+
]). (5)
Here, a(Sb
5+
), a(Sb
3+
) and a(O
2
) refer to activities of Sb
5+
, Sb
3+
,
and O
2
, respectively. The bracket, [i] means concentration of
ion i. The formal potential E
0
includes all the terms other than
the concentration ratio of the redox ion pair and depends on the
basicities of the solvent melts, a(O
2
), and the oxygen pressure
over the melts. The relationship of E1/2 and E
0
is the following:
E1/2 = E
0
+ (RT/2F) ln[D(Sb
3+
)/D(Sb
5+
)]
1/2
. (6)
D(Sb
3+
) and D(Sb
5+
) are diffusion coefficients of Sb
3+
and Sb
5+
,
respectively. If the diffusion coefficient of Sb
5+
is assumed to
be the same as that of Sb
3+
, E1/2 determined by DPV gives the
formal potential directly.
Apparatus and reagents
The desired mol numbers of sodium carbonate, calcium
carbonate and SiO2 and 0.5 mol % per solvent glass of Sb2O3
were melted at 1273 K in a platinum crucible. Care must be
paid in order to set this crucible into the high temperature
reaction tube, because many bubbles are evolved within the
melt and the melt would drop away from the crucible by a rapid
heating. All experiments were conducted in a furnace under air
or 1.013 10
5
Pa oxygen gas. The flow rate of both the gases
was 200 300 ml/min. The cell assembly was the same as that
shown in a previous paper.
15
Figure 2 shows a schematic view
of a reaction cell. Three electrodes were inserted into a liquid in
a platinum crucible. A working electrode made of a coiled
platinum wire, 0.4 mm in diameter, was well immersed into the
liquid. A reference electrode of fold platinum wire was half
46 ANALYTICAL SCIENCES JANUARY 2001, VOL. 17
Fig. 2 Schematic cell assembly for voltammetry.
Fig. 3 DPV voltammogram of 16Na2O10CaO74SiO20.5Sb2O3 melt
for various pulse heights (E) at 1473 K and p(O2) = 1.013 10
5
Pa.
immersed on the surface of the liquid; thus, a reversible O2/O
2
reaction could be expected. The platinum crucible was used as
a counter electrode. This cell assembly with three electrodes
was fixed to alumina rods and hung on the middle part of a
reaction tube. Oxygen gas or air flowed from the bottom of the
reaction tube. The three electrodes were connected to a pulse
polarographic analyzer (Yanaco P-1100). After measurement of
DPV, several selected glasses were dissolved in HF aqueous
solution and subjected to an inducting coupled plasma (ICP)
(Perkin Elmer, Optima 3000) analysis operated at 206.8 nm.
The peak potentials of DPV depend on the pulse height (E).
In DPV, the peak potential, Ep, is related to the half wave
potential as:
Ep = E1/2 + E/2, (7)
where E is the pulse height (see below). For infinitely small
pulses, the peak potential occurs at the polarographic half wave
potential.
20,21
The influence of the pulse height on the
voltammogram was examined first. Figure 3 represents the
DPV voltammogram for various pulse heights. As E
increases, the peak potential moved to the anodic direction
(more positive direction) for a cathodic sweep. Table 1 shows
the relation between the peak potential and the half wave
potential. One sees a good relation between them. In order to
avoid the disturbance around the working electrode, the pulse
height was set to 10 mV. The duration of the pulse, that of the
scan rate and the time between pulses were set to 50 ms, 10
mV/s and 100 ms, respectively. In usual voltammetry, the
potential sweep is initiated from zero volt. However, in the
present case, the peak potential is located around 0 V, so that
the initial potential must be set to a positive value. The
difference of voltammograms between the cathodic sweep from
0.3 V and the anodic sweep from 0 V was examined. Figure 4
represents the two curves. No difference could be found by two
different sweeps. Thus, in the present experiment, the
voltammograms were measured only by the cathodic sweep.
Results and Discussion
Reaction on the working electrode
Figure 5(a) represents a typical voltammogram of the melt
containing 0.5 mol per cent Sb2O3 at 1473 K. By cathodic
sweep, two reduction peaks are observed. Peak 1 is somewhat
ambiguous due to overlap of the current in the solvent melt. By
subtracting the current due to the solvent (dotted line), the peak
comes to be seen more clear. Figure 5(b) shows the theoretical
curves for a two-electron step and a three-electron step,
respectively.
22
Thus, the first step reduction on the working
47 ANALYTICAL SCIENCES JANUARY 2001, VOL. 17
Table 1 Relation between the peak potential and the half-
wave potential
E/mV Ep/mV (E/2)/mV (EpE/2)/mV
50 287 25 312
20 303 10 303
10 309 5 314
5 311 2.5 314
Fig. 4 Typical voltammogram of 16Na2O10CaO74SiO20.5Sb2O3
melt measured by the difference of sweep direction at 1430 K and
p(O2) = 2.127 10
4
Pa.
Fig. 5 Typical voltammogram of 16Na2O10CaO74SiO2 melt at
1473 K and p(O2) = 1.013 10
5
Pa. (a): solid line, containing 0.5
mol% Sb2O3; dotted line, Sb2O3-free. (b): solid line, after subtraction
of Sb2O3-free melt from containing Sb2O3 one; dotted line, theoretical
curve.
electrode can be written by Eq. (3), that is n equals 2, whereas
the second step reduction expressed by n = 3 corresponds to the
following reaction:
Sb
3+
+ 3e = Sb
0
(second step). (8)
Above about 1673 K, the peak 2 split into two as shown in Fig.
6. The additional peak was seen when the oxidized or reduced
ions adsorbed on the electrode. At 1873 K, the peak intensity of
the additional peak decreased with time and appreciable
amounts of noise were seen in iE curves. This phenomenon
may be the evaporation of metallic antimony deposited on the
electrode. Unfortunately, at present, a more detailed discussion
can not be made. Possible formation of platinum + antimony
alloy should also be considered. Table 2 shows the ICP results
of the total residual antimony content in selected quenched
glasses after the DPV measurements of the corresponding melts
were carried out. The remaining total antimony contents
decrease with an increase in melting temperature. However, the
half wave potential did not change with the antimony contents.
Pt:O2/electrode
Figure 7 represents the variation of the potential of peak 1
with the oxygen pressure over the melt. By changing the
oxygen pressure from pure oxygen to air, the peak potential
shifts gradually to the positive direction and finally reaches a
constant value. When the atmosphere changes again to oxygen
gas, the peak potential then shifts to the negative direction.
When the oxygen pressure was changed, the reference electrode
potential is changed. On the other hand, the working electrode
potential when a(Sb
5+
) = a(Sb
3+
) does not change, so that the
half wave potential under any oxygen pressure shifts to the
positive direction according to the following equation:
E1/2* = E1/2[p(O2) = 1.013 10
5
Pa] (RT/4F) ln p(O2). (9)
E1/2* is the half wave potential of Eq. (4) under any oxygen
pressure. The peak potential of the peak 1 shifts to positive
direction with a decrease in the oxygen pressure over the melts.
The difference of voltages between under pure oxygen and air is
calculated to be 49 mV at 1473 K. The calculated value
coincides with that expected from Eq. (9). The experimental
value was 48 mV. The response of the peak potential of peak 2
is also seen. However, the difference is somewhat small (36
mV) compared to that of peak 1. As cited above, this may be
due to an irreversible reaction on the working electrode, that is,
deposition of metallic antimony. For peak 1, it is possible to
calculate E1/2 of any oxygen pressure, if E1/2 under 1.013 10
5
Pa of oxygen pressure is known. The reversibility of the
oxygen reference electrode in glass melts was discussed.
23
The
Pt:O2/electrode, the present ones, took longer times to reach
equilibrium than the Pt:ZrO2/electrode, although the former had
the good point of being stable in high temperature oxide melts.
It is argued that the Pt:O2/electrode functioned as an oxygen gas
reference electrode in the DPV used here.
Half wave potential and fining
The temperature dependence of the voltammogram is shown
in Fig. 6 under 1.013 10
5
Pa of O2 gas. Here, the background
currents due to solvents are subtracted from the measured ones.
In Table 3, the E1/2 values of the peaks 1 and 2 are listed, such
values are determined by a least-squares method with a second
order equation. The number of the bracket means the measured
times. The error was expressed by t-distribution of 95%
certainty. In Fig. 8, the relation between E1/2 and the
48 ANALYTICAL SCIENCES JANUARY 2001, VOL. 17
Fig. 6 Temperature dependence of the voltammogram of
16Na2O10CaO74SiO20.5Sb2O3 melt under 1.013 10
5
Pa of O2
gas.
Table 2 ICP results of the total antimony content remaining in
selected glasses after the measurements of DPV
Melting
temperature/K
Melting time/
h
Antimony content,
mol%
Remaining
percentage
1377 132 0.497 99.4
1473 95 0.489 97.8
1773 36 0.286 57.2
1873 21 0.247 49.4
Fig. 7 Variation of the potential of peak 1 with the oxygen pressure
over the melt.
temperature is plotted. The solid lines are fitted by the least-
squares method. E1/2 increases linearly with an increase in
temperature as
E1/2/V = 4.27 10
4
T 0.60 (under air)
E1/2/V = 6.28 10
4
T 0.79 (under 1.013 10
5
Pa oxygen).
The dotted lines are estimated from Eq. (9) based on the line of
E1/2 of 1.013 10
5
Pa oxygen pressure.
[Sb
5+
]/[Sb
3+
] concentration ratios derived from Eq. (6) are
shown in Fig. 9. The Sb
5+
/Sb
3+
equilibrium shifts toward more
reduced state with an increase in temperature. When the
temperature is raised, Sb
5+
converted to Sb
3+
with release of
oxygen gas as shown in Fig. 1 and Eq. (4). If the evaporation of
the total antimony ions (0.5 mol %) in the melts is neglected in
the process, the change of evolved and disappeared gas
quantities is maximum at the temperature where [Sb
5+
] = [Sb
3+
]
(i.e. around 1200 K), as shown in Fig. 1. Imagawa et al.
measured the quantities of the oxygen gas by a mass
spectroscopic analysis. It increases with an increase in
temperature and the evolution finished in the vicinity of 1673
K.
24
The oxygen gas is evolved markedly at around 1573 K.
The evolution increased with an increase of Sb2O3 content.
They concluded that the oxygen evolution is conducted by the
refining reaction: Sb2O5 Sb2O3 + O2 which occurred during
the heating period. Effective evolution of oxygen bubbles
supports the rise of bubbles originated from starting materials as
shown in Fig. 1.
In glass melts, antimony ions are present as complex ions
accompanying a certain number of oxide ions, because the
[Sb
5+
]/[Sb
3+
] ratio increases with an increase of basicity, i.e.
[O
2
]. Thus, the notations of SbOn
(2n5)
and SbOm
(2m3)
are
preferable to free Sb
5+
and Sb
3+
.
25
In order for the above ratio to
increase with an increase in the basicity, n>m must be followed.
When the temperature is lowered, the equilibrium shifted to the
oxidized direction, so that the remaining oxygen (small bubbles)
reacted with Sb
5+
to form Sb
5+
-complex and dissolved stable in
melts. The bubbles may be eliminated from the melts and
fining can be established.
Conclusion
The half wave potential of reduction of Sb
5+
in
16Na2O10CaO74SiO2 glass melts was examined by
differential pulse voltammetry at various temperatures. The
49 ANALYTICAL SCIENCES JANUARY 2001, VOL. 17
Table 3 Relation between the temperature and the E1/2
T/K
E1/2/mV (peak 1) E1/2/mV (peak 2)
O2
(p(O2) = 1.013 10
5
Pa)
O2
(p(O2) = 1.013 10
5
Pa)
Air
(p(O2) = 2.127 10
4
Pa)
Air
(p(O2) = 2.127 10
4
Pa)
1322 452[5] 35647[5]
1377 3318[6] 34520[6]
1430 6829[5] 104[1] 33120[5] 306[1]
1473 9315[10] 12812[4] 32510[10] 29714[4]
1519 11028[6] 31414[6]
1549 1229[5] 30825[5]
1573 1528[6] 205[2] 30510[6] 264[2]
1673 18412[5] 252[1] 26711[5] 220[1]
1773 21822[5] 22015[5]
The figures in the brackets are the number of measurements.
Fig. 8 Relation between E1/2 and the temperature. Fig. 9 Relation between [Sb
3+
]/[Sb
5+
] concentration ratios and the
temperature.
half wave potential shifts to the positive direction with an
increase in temperature, which corresponds to the increase of
Sb
3+
. Antimony ions may become an active refining agent like
arsenic ions. The half wave potential moves to the positive side
(48 mV at 1473 K) when the atmosphere changes from pure
oxygen gas to air, in agreement with the theoretical prediction.
Acknowledgements
This work has been made under the Research and Development
Program for International Standards for Supporting New
Industries funded by New Energy and Industrial Technology
Development Organization (NEDO). The work has also in part
been supported by a Grant-in-Aid for Scientific Research (No.
09450327) from the Ministry of Education, Science, Sports and
Culture, Japan.
References
1. G. E. Rindone, Glass Industry, 1957, 38, 489, 516, 526, 528.
2. G. E. Rindone, Glass Industry, 1957, 38, 561, 576.
3. M. Cable, Glass Tech., 1961, 2, 151.
4. M. Cable and M. A. Haroon, Glass Tech., 1970, 11, 48.
5. R. Pyare, S. P. Singh, A. Singh, and P. Nath, Phys. Chem.
Glasses, 1982, 23, 158.
6. M. C. Weinberg, J. Non-Cryst. Solids, 1986, 87, 376.
7. W. D. Johnston, J. Am. Cer. Soc., 1965, 48, 184.
8. D. M. Krol and P. J. Rommers, Glass Tech., 1984, 25, 115.
9. K. Takahashi and Y. Miura, J. Non-Cryst. Solids, 1986, 80,
11.
10. T. Yokokawa, K. Kawamura, and S. Denzumi, Tr.
Electrochem., 1992, 1, 71.
11. C. Rssel and E. Freude, Phys. Chem. Glasses, 1989, 30,
62.
12. C. Rssel, J. Non-Cryst. Solids, 1990, 119, 303.
13. C. Rssel and G. Sprachmann, J. Non-Cryst. Solids, 1991,
127, 197.
14. O. Clauen and C. Rssel, Glastech. Ber. Glass Sci.
Technol., 1997, 70, 231.
15. M. Yokozeki, T. Moriyasu, H. Yamashita, and T.
Maekawa, J. Non-Cryst. Solids, 1996, 202, 241.
16. M. Nakashima, H. Yamashita, and T. Maekawa, J. Non-
Cryst. Solids, 1998, 223, 133.
17. H. Yamashita, S. Shimaoka, S. Yamaguchi, and T.
Maekawa, J. Cer. Soc. Jpn., 1999, 107, 385.
18. H. Yamashita, S. Yamaguchi, M. Yokozeki, M.
Nakashima, and T. Maekawa, J. Cer. Soc. Jpn., 1999, 107,
895.
19. S. Gerlach, O. Clauen, and C. Rssel, J. Non-Cryst.
Solids, 1998, 226, 11.
20. E. P. Parry and R. A. Osteryoung, Anal. Chem., 1965, 37,
1634.
21. J. H. Christie, J. Osteryoung, and R. A. Osteryoung, Anal.
Chem., 1973, 45, 210.
22. J. Osteryoung and J. J. ODea, in Electrochemical
Chemistry, ed. A. J. Bard, 1986, Dekker, New York, 14,
209.
23. T. Tran and M. P. Brungs, Phys. Chem. Glasses, 1980, 21,
133.
24. H. Imagawa, M. Aoyagi, K. Saitoh, and S. Uchiyama,
Glastech. Ber. Glass Sci. Tech., 1995, 68C2, 217.
25. H. Hirashima, T. Yoshida, and R. Brckner, Glasstech.
Ber., 1988, 61, 283.
50 ANALYTICAL SCIENCES JANUARY 2001, VOL. 17