One of the thorny issues that I grappled with in writing the book was how to review the staggering number of ways in which sand is important in our lives - in the end, the diversity was so overwhelming that I resorted (as readers of the book will know) to an alphabetic medley approach; interestingly, it would seem to be a chapter that is often highlighted in reviews and comments as either highly entertaining or awkward. In my ruminations about how to weave all these disparate topics in to a narrative, one idea had occurred to me: given that an immediate and recurring (if subconscious) encounter we all have with quartz sand is every occasion that we use an electronic device, why not tell the story, the journey, of a sand grain from its geological origins to the silicon chip in the computer on which I was writing?
"Sand to chips - a biography" - seemingly a simple task, this proved anything but. To begin with, given the need for various but rigorous levels of extreme purity in silicon-based manufacturing, just any old sand won't do - the raw material has to be as pure as possible to begin with. As I wrote:
First of all, the public is understandably more interested in the technology of making the silicon chip than in its provenance in a sandbank by a river somewhere. Second, the absolute volume of semiconductor-grade silicon is only a small fraction of all silicon production. But more important, a small number of companies around the world dominate the technology and the market, and while their literature and websites go into considerable and helpful detail on their products, the location and nature of the raw materials seem to be of “strategic value,” and thus an industrial secret. I sought the help of the U.S. Geological Survey, which produces comprehensive annual reports on silica and silicon (as well as all other industrial minerals), noting that statistics pertaining to semiconductor-grade silicon metal were often excluded or “withheld to avoid disclosing company proprietary data.” The survey staff were, as always, extremely helpful, but were themselves perplexed that such an apparently simple question was not simple to answer. They put me on the trail of a number of sources, but telephone and e-mail inquiries did not shed a great deal of light. What I could deduce is this: the common source of silica for manufacturing the high-purity grades of silicon used for, among other things, silicon chips is not loose, unconsolidated sand from a beach or river sandbank, but sand that has already been ultrapurified by nature: quartzite.
Quartzite is a rock that was originally a silica sandstone; it has been so deeply buried in the Earth’s crust, cooked by such high temperatures and pressures, that many of the impurities have been distilled out and the sand grains completely annealed and welded together. Hit quartzite with a hammer and it rings like a bell because of its hardness, purity, and uniformity. Hit it hard enough and it breaks across the ghosts of the grains, not around them. Quartzite can be well over 99 percent pure silica. Grind it up to a powder of a consistent grain size and it’s a good starting point for making silicon.
The geological kitchens of the Appalachians were good at making quartzites, and my research (including phone calls to sand and gravel suppliers) led me to one possible source:
The Coosa River, for example, originates close to the border of Georgia and Tennessee and, together with its tributaries, drains a large and geologically diverse area of Appalachian rocks, including quartzites. It crosses from Georgia into Alabama, where it joins the Tallapoosa to become the Alabama River. As it does so, it enters the broad coastal plain, slows down, and dumps its sediment load. Large volumes of aggregates for all kinds of industrial purposes are extracted in this part of the country from the sediments of the Coosa and other rivers, but among those everyday aggregates are pebbles of great value—pure quartzites from the kitchens of the Appalachians. And those, I believe, are one source of raw material for high-grade silicon. So it’s true that computer chips are made from sand—but sand that was first deposited several hundred million years ago.
But the confidence level that I (and the USGS) had in this being a definitive example led me to abandon the sand grain biography idea. Then, earlier this week, the whole issue arose again: BBC Radio in the UK had a programme (it is the BBC, hence the extra "me") called "Chips with Everything" and an associated article titled "Silicon Valley's Secret Recipe." The thirty-minute broadcast is well worth listening to, for the "Silicon Valley" referred to is in the Blue Ridge Mountains of North Carolina, around the small community of Spruce Pine which calls itself the Mineral City - for good reason. The programme is culturally fascinating but also highly informative with respect to silicon chip technology - and I learnt something that I had missed in the opacity of my earlier research.
The reason the Spruce Pine calls itself the Mineral City is that in the surrounding hills is a treasure trove of precious and highly valuable industrial minerals - including some of the purest quartz in the world. These minerals, exploited since long before the ancestors of today's residents arrived, are contained in igneous rocks, solidified from molten material deep below the earth's surface; but these are special, rocks that are made up of unusually large and unusually pure, crystals - rocks called pegmatites. Below is a photograph of one of these (thanks to Callan at NOVA Geoblog), with large feldspar crystals embedded in quartz. It was long thought that these unusually obese crystals formed because the cooling rate of the molten rock was extremely slow, but this apparently logical model has been replaced by recent research that suggests that the cooling rate was unusually rapid, faster than the rate of crystallisation could keep up with.
The hills around Spruce Pine have been described as the most valuable strategic acreage on the planet. Sixty-eight-year-old Ira Thomas, a ninth-generation Spruce Pine miner who used to dig up aquamarines and prospect for mica as a child justifies this description: “Because the world runs on computers, we all know that now. And if we locked the gates to Mitchell County they could not make any more computers.” As a consequence, the quartz pegmatite mines are protected by security akin to the Pentagon or the Bank of England. Spruce Pine is billed as the world's only source of this uniquely pure quartz - this is geologically probably not quite true, but it is by far the world's major supplier of a raw material that sells for $50,000 a ton.
But so what was my revelation? Simply that the Spruce Pine quartz is valuable not as the source of silicon for chips, but as the material for making the crucibles and other bits of kit needed to process electronic-grade silicon. Silicon can be extracted from any quartz relatively easily (you can do it, albeit somewhat dangerously since silicon is an excitable element, in your kitchen from beach sand); but to process it to the level of purity needed, then ultra-pure equipment is essential - and it's made from Spruce Pine quartz. The level of silicon purity for a chip is extraordinary: electronics-grade silicon has to be at least 99.99999 percent pure - referred to in the trade as the “seven nines” - and often it’s more nines than that. In general, we are talking of one lonely atom of something that is not silicon among billions of silicon companions. Or, as a Dow Corning scientist, put it:
Imagine stringing tennis balls from the Earth to the moon, and wanting them all to be yellow. He said this would take about 5.91 billion tennis balls. For the color coding to be of semiconductor quality, "you could only tolerate two, maybe three that were orange," Lane said. For solar cells, which are slightly less demanding, Lane said, "You could tolerate five or six orange balls."
Following up on the BBC material, I came across a useful series of articles on Geek.com, which, in turn, led me to a press kit put out by Intel, titled "From Sand to Silicon: the Making of a Chip." And there, to my delight, at the head of the document is an image of a pile of sand. One of the Intel images is at the head of this post, and in the document is accompanied by the description:
With about 25% (mass) Silicon is –after Oxygen –the second most frequent chemical element in the earth’s crust. Sand –especially Quartz -has high percentages of Silicon in the form of Silicon dioxide (SiO2) and is the base ingredient for semiconductor manufacturing. [my highlighting]
The silicon is purified in a series of technologically sophisticated (and highly secret) steps; to the right of the pile of sand is melted silicon, and then a "Mono-crystal Silicon Ingot .... produced from Electronic Grade Silicon. One ingot weights about 100 kilograms (=220 pounds) and has a Silicon purity of 99.9999%." (Why six, not seven, nines?)
So there we have it - sand, or potentially sand in its quartzite incarnation, is the source of raw silicon for electronic chips. But I am no further forward on identifying geologically which sands - or sandstones, or quartzites - are the sources. The quest continues - if anyone can shed light on this, I would be hugely grateful: I can provide an update for the paperback edition of Sand!
[My profound thanks to Pete Modreski at the USGS for his continuing help throughout this quest. And there are other sources of high purity quartz in the world, among them from Norway: the NGU, the Norwegian Geological Survey, has an excellent web page, in English, on high purity quartz - http://www.ngu.no/no/hm/Georessurser/industrimineraler/Kvarts-og-kvartsitt/Hoy-ren-kvarts/]