silicon for the chemical and solar industry XIII

which will vary from one customer to the other. Further refining may be done after

tapping, if needed. This is normally decided upon receiving the analysis of samples

taken shortly before the ladle is full. Stable furnace operation and tapping makes the

refining much easier.

Suggested literature on tapping is [1] and [3] and on refining of silicon [1] and [4].

Process gases, silica fumes and energy recovery

Almost all free carbon that is added in the raw materials will leave the furnace as

CO(g). Only a small amount of it will be dissolved in the liquid silicon. A small

fraction of the CO(g) will come out of the tap hole, but most of it will leave through

the charge top together with some SiO(g) produced in the hot zone and water vapour,

volatiles and other gases driven off from the carbon materials as they are heated.

These very hot gases mix with air sucked in through the charging gates above the

charge and they will burn to CO

(g) and fine SiO

particles called silica fume or silica

dust as they are sucked into the filter unit where the silica is collected. The silica is

sold, mostly to producers of concrete. Typically 10-20% of the quartz ends up as

silica, depending on the furnace operation.

Modern furnaces normally have energy recovery from the hot process gases before

filtering out the silica. Producing hot water for district heating is more energy efficient

than producing electric energy, but not all plants have a demand for district heating

nearby.

Basic overall chemical reactions

In the carbothermic silicon process we use carbon at high temperatures, in the order of

2000 C, to remove the two strongly bonded oxygen atoms from the silicon in quartz

(SiO

) according to Eq. 1.

SiO

(l) + 2C(s) = Si(l) + 2CO(g)

(1)

If we have exactly 2 moles of chemically active carbon (fix-C) for every mole of

quartz, we say that the

carbon balance

is 100%.

However, several other chemical species are present at these high temperatures,

most notably SiO(g), CO(g) and SiC(s), giving the more realistic Eq. 2 which is

discussed in much more detail for example in [1] than it is here.

a SiO

(l) + b SiC(s) + c C(s) = x Si(l) + y SiO(g) + z CO(g) + u SiC(s)

(2)

This equation allows both for net production of silicon carbide in periods where we

add more carbon that we should have done (overcoked process, u>0 and b=0) or net

consumption of SiC that has been deposited earlier if we add less carbon than we

should have done (undercoked process, b>0, u=0).

The special case b=0 and u=0 means a process where SiC is neither consumed nor

produced overall. Eq. 2 then transforms to Eq. 2.a when the parameter ‘a’ in Eq. 2 is

assigned the value 1 and x is the silicon yield (the fraction of Si in the quartz in the

raw materials that comes out of the tap hole as liquid silicon).

SiO

(l) + (1+x) C(s) = x Si(l) + (1-x) SiO(g) + (1+x) CO(g)

(2.a)

This is the situation we are aiming for where the carbon balance is optimal and good

furnace operation may prevail for a long time if everything else is also optimal.