Slashdot Mirror
Silicon Seduced From Silica
Roland Piquepaille writes "Making silicon is an expensive process, which conventionally involves carbothermal reduction, in which the oxygen is removed from silica by a heterogeneous-homogeneous reaction sequence at approximately 1,700 C. Now, Japanese researchers have developed a new technique which uses electricity to remove the oxygen from silica. Their technique is based on the immersion of silica in a bath of molten calcium chloride salt at 850 C, which should reduce the costs of making silicon -- and other elements, like zirconium. Check this column for a summary or read this article from Nature for additional details."
If it's not obvious from the definition, silica is silicon dioxide, SiO2. It is the primary ingredient for making glass. IConsider it as purified sand - all the impurities that color the SiO2 have been removed.
Unfortunately, many of the conventional sand deposits (Floridian sand not exempted, I think) contains only a small amount of silica, making the refining process prohibitively expensive. This process might be a little cheaper, but proportionally speaking they're still going to do better with dredged volcanic "supersand".
Try not. Do or do not, there is no try.
-- Dr. Spock, stardate 2822-3.
not related to silicon, but i like to point that out. in case people are looking for uses for zirconium =).
for those that thought about it - no spectra is not good enough for space elevator. only 3GPa tensile strength (steel about .25 for cheap ones and 5 for REALLY good ones). space elevator needs ~62GPa. nanotubes ~150GPa theoretical.
okay. end rant.
My life in the land of the rising sun.
btw - this kinda shows how bs was bush's little thing about saddam using ALUMINUM tubes for reactors.
My life in the land of the rising sun.
Therefore the collected silicon mus be remelted, drawn, cleaned, sliced into tiny placks, etched, washed and polished. However this is also has to be done with silicon obtained in other ways. Nowadays there are machines who can perform most of these procedures in one run.
A short explanation of this can be found here
I don't know how much the raw silicon costs, but I suspect that most of the cost of the wafers comes from this month-long crystal growth and planarization. Good (ie, very flat) 200mm silicon wafers for semiconductor production can cost up to $1000 each, although they are probably much cheaper now due to lack of demand. Many processes also don't require the flattest wafers and so one can get by with wafers that cost a small fraction of that.
This new process then can mean a lot more cheap solar cells. Imagine like all available roof areas being covered, down to the top of all cars.
I'm more interested in the increase of performance of the solar cells and this means moving away from Silicon based technology.
If I have to have a 600sq ft array of Si cells to generate 1/5th the power needs of my home being able to buy them cheaper doesnt help me. I can't have 5 20X30 foot panels on my property because I live in the city.
The most interesting is this process might make the procesing of other materials used in the more efficient solar electric panels either cheaper or easier (thus making it cheaper)
Although with Si panels themselves... if everyone had 1 panel on their home in this northern midwest town tied to the grid it would make a significant difference in the electrical supply andthe amount of coal we burn every day to make that electricity...
Although not anywhere as much as simply changing every lamp in your home from Incandesant and regular Flourescent to Compact flourescent.
Do not look at laser with remaining good eye.
Hmm, last I checked Spectra 1000 has tensile modulus of 172GPa (older version, Spectra 900 is at 117GPa or so) and steel averages around 210GPa... I wouldn't call that a huge difference. certainly more than half as you claimed.
.97 (water at 1) and steel 7.8 or so...
impressive especially considering spectra has specific weight of
I mean, I can't imagine that if cost wasn't a consideration, any places where you wouldn't want a lighter material vs. the heavier one (except that polyethylene is not good with fire, so car engines are out).
but i digress. steel is cheap. but damn, as far as materials go, spectra is about the sexiest we got right now (that's mass-producable, anyway).
My life in the land of the rising sun.
Actually, most volcanic sands don't have much quartz at all, and here is why. If it has been derived from a basaltic volcano (ie Hawaii), the composition of the sand will be extremely high in mafic (very silica poor) minerals since the source magma was low in silica. Then, if you are talking about a volcano whose melt composition is closer to the felsic (silica rich...so much so that you get quartz precipitating) side, these are typically very explosive volcanos that produce lots and lots of fine grained ash but no lava flows to weather from. What you *really* want is a sand eroding from an exposed granite. You get great big fat quartz crystals, and feldspars that turn to clay very quickly. And that's just if you want to find a loose sand that will be quartz rich. What I would do is actually get a hold of some mining rights out in the Southwest US somewhere and start a quarry operation on all the excellent quartz sandstone they've got.
Project Steve
The first stage of silicon refinement takes silica and makes it silicon to about 75% pure. The second stage takes the silicon to 99% pure. But this stage involves making silicon into a gas. Electricity is used to charge a small, very pure seed crystal so that the gaseous silicon deposits on it.
I think why it took them so longer to do it with solid silicon is the fact that aluminum is a metal and silicon is a semiconductor. There are details which allows them to bridge this gap that are probably not explained in the article.
Well, there's spam egg sausage and spam, that's not got much spam in it.
WOW, a subject where I seem to be the first "expert" to post.
I work in the semiconductor industry. (actually, for one of he largest producers of semiconductor-grade silicon in the world,) and I'm intimately familiar with the process to turn silicon from sand into wafers for chip manufacture. At my work, we are the middle step. I'll explain:
Semiconductor-grade silicon is ultra-pure silicon metal (I mean parts-per-billion atomic purity.) All the semi-grade Si in the world is produced in approximately the same way.
Silica (sand) is reduced to "metallurgical grade" silicon (~99.5% pure) in an arc furnace process, the sand is melted with a reducing agent (often carbon), and the molten metal is poured off. (this is a very cool process. The smelter has a hole in the bottom that is allowed to freeze shut with Si, and when they're ready to pour, someone shoots out the Si plug with a shotgun. Cool job)
This metallurgical grade Si is sold to intermediate producers who grind it to a fine powder, and react it with gaseous HCl in fluid bed reactors to generate chlorosilanes (H3SiCl, H2SiCl2, HSiCl3, SiCl4.) These chlorosilanes are then distilled to very high purity ~99.999% or more.
The chlorosilanes (different ones for different manufacturers) are then used in the Siemens process to produce semiconductor-grade POLYCRYSTALLINE silicon. The process works by Chemical Vapor Deposition. Ultra-pure silicon rods are placed in a reactor in an inverted U shape, and each end of the U is connected to an electrical circuit. The atmosphere inside the reactor is purged of all gasses and then chlorosilane vapors are introduced. Huge amounts of electricity are used to heat the U circuits to incandescence (imagen a 600 megawatt lightbulb) and the ultra-pure chlorosilanes decompose into Si and HCl at the surface of the rods.
The problem with the silicon, at this point, is that it's polycrystalline, not single crystal. In order to produce proper IC's, the crystal structure of the silicon must be perfect 1,1,1 crystal. Polycrystalline silicon (a.k.a. poly) is a random oriented growth where the crystal structures of many crstals have grown together. The poly is reduced in size and sent to a crystal pulling facility (wafer fab) where it is used in the Czoralski process for making wafers.
The CZ process consists of melting a large amount of poly, then dipping in a "seed" crystal. This perfect single-crystal specimen is "dipped" into the molten silicon while being rotated. The seed is then carefully "pulled" upwards while rotating, and the resulting ingot grows in diameter based on the pull speed, and several other factors (300mm is current state of the art.)
Once the pull is completed, a ~1000+ kg log of single crystal silicon is made, and is ready for final processing to wafers. The tapered ends are removed (top and tail) and the "log" is shaved down perfectly round and to the proper diameter. Diamond impregnated wire saws are used to slice the log into wafers, the wafers are lapped and polished, and they are ready to have IC's printed on them. (some are further processed, but you get the gist.
HTH
GM