Group of companies "MINERAL"

You are here: Home » Our Publications » Carbothermic synthesis of silicon in 150 kW DC arc furnace for solar applications
Sunday, 25 Feb 2018

Carbothermic synthesis of silicon in 150 kW DC arc furnace for solar applications

E-mail Print

1Silicon for the Chemical and Solar Industry IX
Oslo, Norway, June 23-26, 2008




M.A.Arkhipov, A.B.Dubovskiy, A.A.Reu, V.A.Mukhanov, S.A.Smirnova
Mineral Ltd., 1, Institutskaya st., Alexandrov, Vladimir Region,  601650 Russia


Reaction ability of high purity raw materials in briquette could be improved by using inorganic catalyst that can be added in amount 2-4 weight percents in briquette. Maximum output of silicon was 15 kg per batch in 150 kW DC arc furnace. Sintered silicon with purity 99.97% (B=1 ppm, P=8 ppm) was recrystallized by horizontal directional solidification in vacuum and 75% of obtained ingot had purity 99.998%.


Alternative methods for solar silicon manufacturing are strongly required by industry for fulfilment huge government solar energy programs approved in many countries and because of stable growth of oil and gas prices. Traditional methods of solar silicon synthesis that use chlorosilanes are not capable to provide enough quantity of material in coming 10 years because of relatively long new plant setup time (about 4 years if all machines available) and shortage in delivery important components and units (some machinery suppliers are completely booked for over 6 years). According to declared plans suppliers of solar silicon that use chlorosilanes maximum are capable to double production in 5 years when possible demand require 5-10 times growth.

Two main alternative approaches for manufacturing solar silicon from metallurgical silicon exists:

1.Alternative carbon source: using thermal black or activated carbon or rice hulls that can be briquetted with quartz powder. Typical result is silicon 99.9-99.99% purity where boron may be in the range 1-10 ppm depends on purity of raw materials [1]. Further purification can be done by acid treatment, vacuum treatment of melt and recrystallization.

2.Traditional carbon source: using charcoal or coke of improved purity. In this case maker is capable to reach purity of metallurgical silicon 99.5-99.7% with boron content 9 ppm. Further purification to solar grade could be done by gas sparging [2], plasma treatment [3], acid treatment and vacuum treatment of melt [4] and recrystallization.

Alternative carbon source provide better purity of metallurgical silicon but reaction ability is lower and structural characteristics of briquette are different from traditional for carbothermic process charcoal or coke. Traditional carbon source with improved purity is well compatible with carbothermic process but  more purification stages are required.

Raw materials reaction ability

One of the main challenges on the way of using alternative carbon source and high purity quartz is low reaction ability due to high purity and diffusivity of SiO gas in briquette different from diffusivity in charcoal or coke that make process far from ideal for arc furnace. Ideal thermal process presuppose SiC formation in low temperature outer reaction zone and it conversion into Si in high temperature inner reaction zone [5]. But such ideal process is possible to organize if arc furnace use as raw material 2N-3N purity quartz and charcoal or coke.

We did make briquettes of thermal black and quartz powder and the main problem we found was low reaction ability of briquette components: SiC formation took place at temperatures about 2000 centigrade and great portion of carbon (over 50%) remained not reacted before briquette collapse. Reaction ability of briquettes was checked in Ar atmosphere in temperature range 1500-2100 centigrade. Quartz powder in briquette had purity 5N (B,P< 0.1 ppm) and size 100-400 microns and carbon was granulated thermal black nanocrystals with purity 3N8 (B,P <1 ppm).

For efficient  using briquettes in arc furnace we had to find a way to decrease temperature of SiC formation and make transformation rate close to 100%. We used inorganic catalyst of SiC reaction that was added in quantity 2%-4% by weight into briquette. It let to obtain complete transformation of C into SiC at temperature 1850 centigrade.

We did check furnace characteristics for molar ratios SiO2/2C and SiO2/3C in briquette. Finally we choose SiO2/3C ratio as optimum for briquette because for the case SiO2/2C we found that after SiC formation remaining quartz react with SiC as follows in outer zone of furnace:

2SiO2+SiC=3SiO+CO (1)

We also found that decreasing of briquette density from 1.5 g/cm3 to 0.8 g/cm3 has positive influence on reaction ability of materials inside. Briquettes had size in the range 15 mm – 60 mm. We found that optimum briquette size for 150 kW DC arc furnace is 15-20 mm. Briquettes were mixed with natural quartz (lascas) of purity 4N5 and size ranges 3-20 mm and 20-60 mm. Both size ranges of quartz are applicable for 150 kW DC arc furnace.

DC arc furnace and carbothermic process

In our experimental work we used 150 kW DC arc furnace designed and assembled by Quartz Palitra Ltd. (see Figure 1). We used 200 mm diameter graphite electrode and distance from electrode walls to oxide lining walls was 200 mm. Arc furnace reacting zone had 600 mm depth. Graphite electrode served as cathode and graphite crucible 600 mm inner diameter served as anode. Silicon was removed through taphole after taphole is opened and furnace was bended.  DC arc was switched on after resistive heating of inner reaction zone up to the temperature 1600 centigrade. Silicon formation is started at temperatures about 1900 centigrade (taphole opening let to remove about 0.5 kg of Si). Optimum production yield was found in temperature range 2100-2300 centigrade in inner reaction zone. Further increasing of arc furnace temperature lead to decrease yield due to boiling of silicon- big portion of silicon may be evaporated. Upon reach of silicon boiling point (2355 centigrade) thermocouples demonstrate stabilization of temperature that let us to estimate temperature ranges for processes inside furnace.

Sintering was done on following regimes of arc: 1) U=38 V, I=3400 A for temperatures below 2200 centigrade; 2) U=50 V, I=3000 A for temperatures above 2200 centigrade.

Figure 1: DC arc furnace.

Without catalyst typical output of furnace was 3-5 kg per batch. If catalyst applied output was 8.5-15 kg. Time of cycle between taphole openings varied in the range 3-5 hours. Maximum output was reached for SiO2/1.8C ratio of quartz/carbon for mixed briquettes and quartz (see Figure 2). Carbothermic silicon was solidified as disk of up to 50 cm in diameter and up to 5 cm height. Maximum output was reached for SiO2/1.8C ratio of mixed briquettes and quartz. The process is practically insensitive to size of quartz ore but very sensitive to size of briquette- optimum size: 15-20 mm.

Purity level and horizontal directional solidification

Purity of carbothermic silicon was controlled by mass spectrometer and are varied in the range 99.96-99.985% depends on ratio SiO2/C for charge (here we not counted SiC impurities). Increasing of quartz content to the value corresponded to ratio SiO2/1.6C lead to increase of electrode consumption and higher impurities content. Elements brought to furnace by inorganic catalyst were partially removed during carbothermic process with off gases and completely removed in vacuum during recrystallization. Carbothermic silicon have B=0.5-1 ppm and P=5-8 ppm.  Main impurities for 99.97% carbothermic silicon are: Ca=40 ppm, Fe=65 ppm, Al=80 ppm, Ti=15 ppm.

Figure 2: Solidified silicon.

Carbothermic silicon was etched in HCl to remove surface contamination and then it was recrystallized in horizontal directional solidification machine (see Figure 3). We used quartz crucible with dimensions L=270 mm, W=70 mm, H=35 mm supported by graphite foam. Crucible was moved in horizontal direction through heating zone with rate 30 mm/hour. Heating was done by current that flow through graphite plates located above and below moving crucible. Absence of vertical temperature gradients make horizontal directional solidification ideal process for settling of solid particles in melt. After one recrystallization in vacuum main impurities on length of 75% from ingot length was as follows: Ca=3 ppm, Fe=10 ppm, Al=3 ppm, Ti=0.5 ppm. Boron has not been removed significantly but phosphor was reduced to value 0.01 ppm due to vacuum (pressure = 0.1 Pa). Electrical resistivity of carbothermic silicon and recrystallized silicon was the same: 30-35 Ohmx cm. It can be interpreted that main contribution in electrical resistivity are done by grain boundaries and structural defects. To improve electrical resistivity necessary to reduce recrystallization rate and use seed to obtain single crystal.

Figure 3: Horizontal directional solidification machine.


  1. Inorganic catalyst added to briquette help to increase production yield of arc furnace by improving reaction ability for high purity raw materials. Chemical elements brought by the catalyst in furnace are partially removed in carbothermic process and completely removed by first recrystallization in vacuum.
  2. Chemical composition of briquette SiO2/3C is better then SiO2/2C because not reacted quartz in SiO2/2C briquette may react in outer zone of arc furnace with SiC and the product of reaction (SiO gas) to be lost with off gases.
  3. Horizontal directional solidification of carbothermic silicon is efficient process for silicon purification because of absence vertical temperature gradients that help for settling solid inclusions.



  1. Vishu D. Dosaj, Lee P. Hunt, Method for producing solar-cell-grade silicon, US Patent#4247528, 1981
  2. Roger F. Clark, Michael G. Mauk, Robert B. Hall, Allen M. Barnett, Method for purifying silicon, US Patent#6632413, 2003
  3. Maher I. Boulos, Purification of metallurgical grade silicon, US Patent#4379777, 1983
  4. Josef Dietl, Michael Wohlschlager, Process for purifying metallurgical grade silicon, US Patent#4304763, 1981
  5. Johan Kr. Tuset, The carbothermic silicon process- possible improvements? Silicon for the Chemical Industry V, Trondheim, Norway, 2000, p.9-22.

Add comment

Security code Refresh