Single-Crystal Silicon Manufacturing

This Idea, if one only paid attention to the Title/Description, has been thoroughly half-baked for a couple of decades or so. The problem is, it is STILL only half-baked today! It needs to become fully baked! So, I wish to offer a suggestion or two that may help.

As you may know, silicon is at the heart of the electronics industry these days. See the "Silicon" link for more detailed information about the element, and about certain siliceous minerals such as quartz. Very pure silicon (impurities of one part per billion) is the starting point for both integrated circuits and solar cells. I will call both "chips" here. In both cases it is known that if the initial silicon is preprocessed into a single large crystal, the final chips will be more efficient at what they are expected to do (integrated circuits may not function at all, if anything less than single-crystal silicon is used).

The standard means of creating single-crystal silicon is to start with a crucible of molten silicon, touch a "seed crystal" to the surface of the liquid, and withdraw the seed SLOOOOOWWWLY. As the molten silicon cools and solidifies where it touches the seed, the regularly ordered layout of atoms within that silicon crystal provides a template that encourages growth of the crystal. The art of doing this seed-pulling trick has improved over the years, such that today we regularly make long cylinders of single-crystal silcon, that are 30 centimeters (1 foot) in diameter. They are quite expensive, as you might expect -- but WORSE, these cylinders have to be sliced up like a sausage into thin flat discs, using diamond saws, and a large percentage of the material becomes immediately wasted as sawdust. Sure, the dust can be recycled, but it would be better if flat single-crystal silicon could be directly grown. Such has been a goal for years, as already stated.

The second link describes some of the attempts that have been made over those years of trying to grow large flat single crystals of silicon. Did you know that when molten silicon solidifies it expands (like water expands when it turns to ice)? It is a quite rare phenomenon in materials science (the third and last substance having that property that I personally know of, at this writing, is Element 31, gallium, which melts in your hand -- see link -- and has to be shipped in rubber flasks). Any engineering of the liquid-to-solid transition of silicon needs to take that expansion into account!

Anyway, I have two suggestions. First, there ARE some existing techniques which can produce flat sheets of solid silicon from the liquid state (two are the "dendritic web" and "edge defined film-fed" techniques), but they yield "polycrystalline" and not single-crystal silicon. However, there is an additional process that I am recommending be used ON those sheets, called "zone refining". It works by melting and recrystallizing a substance, along its length, to purify it (crystal formation is a process that naturally tends to exclude impurities). The purification aspect is not necessarily needed for the silicon under discussion here (it is already very pure), but the recrystallization thing, if done slowly enough, could be just the ticket to convert flat polysilicon.... The link leads to a drawing showing a cylinder of silicon being zone-refined, but there is nothing to prevent the notion from being used on a flat sheet.

My other suggestion involves borrowing from the glass-making industry. You might be surprised at how many centuries went by before a process was found for mass producing quality sheets of flat glass (not til 1952!). The link shows how ingredients of glass are melted on a "bed" filled with molten tin. The two substances do not mix, and so the glass can be drawn off cleanly (and flatly). Well, considering that the glass industry is currently processing these silicon compounds at high temperatures in ways very similar to what we want here (yielding lots of flat product), I do wonder why I haven't encountered this suggestion before. Perhaps the answer will be found in the following problem.

See, silicon is much more chemically reactive stuff than glass. Molten tin will almost certainly become tin silicide if you tried using it to melt pure silicon! Still, there are plenty of substances in the world besides tin, and perhaps one of them can be found that will do the trick. It merely needs to be chemically inert with respect to silicon, and denser than silicon (so that the molten silicon will float on top and can be drawn off). It doesn't necessarily have to be liquid also, but if not, it needs to be polished very flat, and most definitely needs to be something that molten silicon can't dissolve (like a block of sugar underneath melted ice). As a starting point for research along these lines, I might suggest sapphire (aluminum oxide) for the polished-solid "bed", and barium fluoride for the liquid "bed" -- but I don't know enough to accurately predict the outcome. The idea is still half-baked, after all! I just hope it won't stay that way...