IAN LANG ELECTRONICS
There are certain fields of scientific endeavour in which the harvests reaped by one field have been sown entirely by those in another. In physics there is no better example of this than the methods of manufacturing integrated circuits, developed over the latter half of the 20th Century and which owes far more to chemistry than to either physicists or electrical engineers.
With that in mind, let us begin our studies with an investigation into the substrate (this being merely a term for the base) of an integrated circuit, which is silicon. The same is the starting point for most semiconductors and is not found naturally in its pure form but more usually as silicon dioxide (SiO2), also known as silica (from the Latin silex).
In order to be use to us in electronic applications Sio2 needs to be de-oxidised and turned into pure silicon. The first step in this process is to heat the material in an electric furnace. This is because of the fact that when silica melts and re-solidifies, the last part to solidify is that part that contains the most impurities. Still, the silicon produced is only 98% pure, and is known as metallurgical grade silicon. We need the silicon to be pure to one part in 10,000,000,000- i.e. a 99.999999999999% purity.
An array of chemical methods is now open to us to re-purify the silicon. Usually it is converted into a silicon compound over which Trichlorosilane is blown at high temperature; resulting in a decomposure into high purity silicon.
The Du Pont Method uses silicon and zinc vapours at 950 degrees centigrade, and produces ultra-pure silicon, but is costly and produces vast quantities of waste that is difficult to clear from the apparatus. Consequently this method is rarely used commercially.
We are still not at the end of the story here, for now we need to produce a material of singular crystalline form. Such a material has a crystal structure whereby at any part of the lattice the structure will be the same as at any other part.
There are two common methods, the first being the float zone method in which the crystal symmetry is far superior but the crystals are small, is used wherever extremely high specifications are required and as such the resultant products are more costly. A far cheaper method used for standard components is the Czochralski method and this is the subject of our next study.
The Manufacture of Integrated Circuits.