0283 Vangskasen
0283 Vangskasen
ABSTRACT
The mechanisms in the metallurgical production of silicon have been examined in this paper.
Through the investigation of several small scale induction experiments (-25 kW} a better
understanding of this complex process has been obtained. Raw materials were mainly quartz (SiOi)
and SiC, to imitate the industrial carbothermic silicon process. Especially the silicon producing
reactions and mechanisms as well as the condensate properties have been studied
The material consumption (Si02 + SiC) was very low at temperatures below 1820°C as
hardly any SiO gas was produced Above 1900°C the reactions inside the furnace started to
happen. Condensates of different co/ors were deposited in the upper and cooler parts of the
crucibles used in the experiments. The most important type was the brown colored kind and was the
product of the following reaction: 2SiO(g) = Si02(s) + Si(s). Brown condensate was deposited in
the charge mix at temperatures below -1680°C. This condensate glued the material in the upper
part of the charge together. This agglomerate was thermally and mechanically durable and created
a solid roofover the cavity left behind by the consumption of charge.
Mechanical tests were performed on the agglomerate created by the condensate. Agglomerate
exposed to temperatures above 1670 °C was weaker compared to the agglomerate exposed to lower
temperatures. Typical compressive strength recorded for agglomerate exposed to 1670°C was 221
MPa at 25°C, significantly higher than Portland cement (40 MPa).
At temperatures around 1700 °C, the silicon in the condensate (Si02 + Si) separatedfrom the
Si02 matrix leaving the rest of the Si02 intact. Perspiration of silicon is an important metal
producing mechanism, especially when using Si02 and SiC as raw materials, since the SiO recovery
mechanism with free carbon is disabled.
KEYWORDS: Silicon production, condensate production.
INTRODUCTION
The metallurgical production of silicon is a high temperature process where quartz (Si02) is
reduced with carbon (C) to silicon (Si) and carbon monoxide gas (CO(g)). Typical consumption of
electrical energy is 11-13 MWh per ton silicon metal produced. The overall reaction of the process
can be written as [1, 2]:
The total description of the process involves many intermediate reactions and complicates the
situation vastly from what reaction ( 1) describes. The internals of a submerged arc furnace can be
divided into a high temperature (T ~ 2000°C) and lower temperature (T < 1811 °C) zone, where
different reactions dominate. In the high temperature zone around the electrode tip, the following
reactions (2,3,4) occur:
The slowest of these three are probably the SiO(g) producing reactions (2) and (3) which
consumes a major part of the electrical energy developed. Silicon can be produced through reaction
(4) at temperatures above 1811°C. The SiO-gas travels upwards in the furnace and is recovered
either by reaction C-material (reaction 5) or by condensation where the temperature is sufficiently
low (T<l800°C) (reaction 2rev and 3rev).
The condensate producing reactions (2rev) and (3rev) are strongly exothermic and are the
main factor how heat is transported upwards in the furnace. The equilibrium conditions for the
reactions (2), (3), (4) and (5) are shown in figure 1.
0.001 +--"'............~+-'-~~+-'-~~+--'-~~+--'-~~+--'-~~t-"-~~1--'-~--'--t
1400 1500 1600 1700 1800 1900 2000 2100 2200
Temperature [0 C]
At the top of the furnace charge, the temperature can vary between 1000 to 1700 °C [1].
Typical industrial silicon yield is around 85% in a well operated furnace. Reaction (5) is the
preferred SiO-recovery reaction above 1512°C. Below this temperature, SiO gas will only be
captured by reaction (2rev) and (3rev). The temperature has a great effect on the equilibrium
conditions for reactions (2rev) and (3rev). If the temperature at the top is ~1620°C CPsio=0.1 atm)
and the main SiO recovery goes through condensation, then the Si-yield will be around 80%.
Much work has been done the recent years to define the reaction zones inside the silicon
furnace. The doctoral thesis by Myrhaug [3], Infacon 12 paper by Tangstad et.al [4], the TMS 2012
paper by Ringdalen and Tangstad. [5] and the master thesis ofVangskAsen [6] are good examples of
recent publications on this subject.
In this work the condensing species will be investigated through small scale experiments, both
when the condensation occurs, the structure of the condensate and the behavior during heating.
EXPERIMENTAL
A total of 13 different small-scale induction furnace experiments has been conducted. Two
induction furnaces have been used and were operated between 20 and 25 kW. To simulate the Si
process, the set up was arranged to give a high temperature at the bottom of the crucible, and a
decreasing temperature towards the top of the crucible. Typically the temperature in the bottom was
between 1800 and 2000 °C, and the temperature on top of the charge was about 1000°C.
The graphite crucibles (figure 2) used in the induction furnace were filled with quartz and SiC
SiC
up to the rim in a molar fraction of--= 2. Both the quartz and the SiC material were sized 5 to
Si02
10 mm. In three experiments the charge composition were modified with addition of carbon
materials as coke, coal and woodchips. The temperature was measured near the bottom in the
crucible as well as just below the charge surface as seen in figure 2. The crucible was heated inside
the induction furnace, the temperature was held at temperatures between l 800°C and 2000°C for 10
to 60 minutes.
Th ermocouples
(top/bottom)
..... ,...
1 DO mm
300 mm
400mm
.......
-
115 mm
150 mm
After the crucibles had cooled, they were cut longitudinally with a circular saw so that the
internals were exposed. Some were filled with epoxy to keep the remaining material together, but
this was not necessary with most crucibles. Samples were drilled out from the cross-section and
investigated using scanning electron microscope (SEM) as well as other tools.
Samples taken from the material inside the crucibles were also tested in a compression
machine. As the pressure increased, the compressive strength of the different samples could be
recorded.
The cross-sections of four selected experiments are shown in figure 3. These crucibles were
filled with Si02 and SiC and heated to temperatures around 2000°C in an induction furnace. As the
material reacted in the bottom, SiO gas travelled upwards and reacted with SiC to produce Si metal
if the temperature was sufficiently high. Deposits of brown condensate (Si + Si0 2) were created
higher up in the crucibles, where the temperature was below 1700 °C. The condensate glued the
material in the top together, creating a "cavity roof', while the material beneath was consumed.
In figure 4, the temperature history graph for the experiment held for 60 minutes is displayed.
In addition, the furnace load was monitored and the interaction between temperature and furnace
load gave a hint of where the most endothermic reactions occurred.
Since several experiments with identical charge mixture were held at -1980 °C at different
durations, it became possible to make an assumption of how the cavity inside the crucible was
created. A description of this is shown in figure 5. The cavity roof was established about the
melting temperature of quartz (-1720 °C), so the temperature limit of condensation through reaction
(3rev) was probably around the same temperature. At 1680 °C, the increase in temperature
stagnated in the experiment displayed in figure 4. Indications are pointing strongly towards that the
condensate production was constrained below -1680°C. Above this temperature the reaction is
unable to get rid of the excess heat and therefore shifted to the left.
Figure 3: Cross-section of crucibles held between 1920 to 1980 °C for different durations.
Temperature was measured in the bottom of the charge
I
I
I
I
,, 7
, ·'
,
-
\
' ,
- --
'
{!!. 700 I ,
,,
600 ~
500
400
300
200
t-
I
,
-1,
,
I
5
100 ~,.
0 0
0 10 20 30 40 50 60 70 80 90 100 110 120 130
Exp eriment time span (minutes)
Figure 4: Time-temperature graph for the experiment held between 1920 and 1980 °C for 60 min
The brown condensate was also studied using the electron probe micro analyzer (EPMA). A
line scan in figure 6 showed that the molar ration of Si/O was very close to 1. The stoichiometry of
reaction (3rev) fitted very well with this product, and this was a good confirmation that the brown
condensate was a result from the reaction: 2SiO(g)=Si02(s,l)+Si(l).
Between the brown condensate and the substrate, another condensate layer could be seen with
the bare eye. The layer which could be less than one mm thick was called the white condensate due
to its color. In the SEM, the layer was investigated, and it was assumed that it consisted ofSi02 and
SiC. The white condensate was possibly a product of the reaction 2rev: 3SiO(g)+cO(g)=2 Si02(s,l)
+SiC(s). A picture of the layer between the SiC-substrate and the brown condensate as well as a
high magnification image of the white condensate is displayed in figure 7. In the high resolution
image, bright particles can be seen in a grey matrix. Mean atomic number of Si02 and SiC is
correspondingly 7.5 and 10, thus will SiC appear as the brighter phase in the backscatter detector.
Condensate grows
- - -/ - - -
Colder
zone
r ro·-···
temperature
T>condensation
temperature
~
Cavity roof will be
established at the
point where
T- condensation
temperature
J-
Charge
materials w ill
be consumed
under the
Hotter cavity roof in a
zone downward
motion
,... ,...
1.8
1.6
:i 1 .4
! 12 - ,I\ [\
-
•
L.s
1
.-Al" "" ..,..., -
0
~ 0 .6
OA
0.2
0 W W ~ U ~ W M
Samp.,. number from SiC p• tlde
Figure 7: SEM-image of the white and brown condensate layers outside a SiC-particle,
and a high resolution image of the white condensate
Small pieces (-3 mm) of brown condensate were collected and heated in a sessile drop
furnace. In this furnace it was possible to take picture of the silhouette of the samples as the
temperature was raised. The heating of brown condensate is visualized in figure 8. At temperatures
above 1700 °C, the silicon phase seems to sweat out from the Si02 matrix. It would make sense if
the condensation reaction reversed and produced SiO-gas when a mix of Si and Si02 was heated,
but this did not happen. This behavior during heating is an indicator that the separation of silicon
from the condensate can be a substantial contributor to the overall silicon production in an industrial
furnace.
Tangstad [7] has in a case study found that the silicon production from the low temperature
zone through separation of the condensate is between 5 and 15 % of the total silicon production.
Schei et al [1] described the crust formation as a problem in the metallurgical silicon process.
Condensate can grow if the furnace stoking is not performed correctly and thus create a sticky and
sintered mix of materials. This will reduce the permeability of the charge closer to the furnace
creating a passive volume. To indicate the strength of this crust, explosives are often necessary to
break it down.
Compression tests were conducted on the material in the cavity roof. The samples did not
consist of epoxy, but were only made up of the material exposed to the environment inside the
crucible. The tests showed that the samples exposed to low temperatures (T < 1670 °C} had higher
compression strength than those exposed to higher temperatures (T > 1670 °C) as shown in figure 9
and 10. The average compressive strength of the material exposed to T < 1670 °C was -396 MPa
while it was -115 MPa for the material exposed to higher temperatures. To compare, regular
Portland cement has a compressive strength of -40 MPa. The quartz had begun to melt at the higher
temperatures, as well as less condensate had been deposited. This may explain the difference of
compressive strength between the two temperature regions.
- 1- 2 - 3 - 4
2 3 4
Compression (mm)
- 1- 2 - 3 - 4 5
260
240
220
200
<V 180
~ 160
';' 140
:; 120
"'
~ 100
D.. 80
60
40
20
0
0 2 3 4
Compression (mm)
CONCLUSIONS
REFERENCES
[1] Schei A., Tuset J.K., and Tveit, H., "Production of High Silicon Alloys", Tapir Forlag,
Trondheim, Norway, 1998, ISBN: 82-519-1317-9.
[2] RSC Chemistry 6.1, Computer program for thermodynamic calculations.
[3] Myrhaug E. H. ''Non-fossil reduction materials in the silicon process -properties and
behavior", PhD Thesis 2003:67, NTNU, Norway, 2003, ISBN 82-471-5619-9.
[4] Tangstad M., Ksiazek M., Andersen V., and Ringdalen E., "Small Scale Laboratory
Experiments Simulating an Industrial Silicon Furnace", lnfacon 12, Helsinki, Finland, 2010.
[5] Ringdalen E., Tangstad M. "Reaction Mechanisms in Carbothermic Production of Silicon,
Study of selected reactions", TMS International Smelting Technology Symposium, 2012.
[6] VangskAsen J. ''Metal-producing Mechanisms in the Carbothermic Silicon Process", Master
thesis, NTNU, Norway, 2012.
[7] Tangstad M. "Staged heat and mass balance of the Si process", Unpublished SINTEF report,
Trondheim, Norway, 2012.