Here’s an idea you see spreading across the internet every so often: that all semiconductor and solar PV manufacturing depends on extremely pure quartz from the town of Spruce Pine, North Carolina.
Fascinating article. I found it especially noteworthy that while the mines are in the US, they're run by foreign companies. Supply webs are complicated, something I think people talking about onshoring critical production don't fully appreciate.
Next, I want to know more about how quartz underground gets turned into crucibles. I was envisioning chunks of quartz getting turned on a lathe. I can't see how that would work. What's the actual process? If it involves casting the quartz, what do you melt _that_ in? A superquartz crucible? Is it quartz elephants all the way down?
The image of a cross-section of a double-layer crucible is actually illustrating a combination of mechanical properties, not relative purity. See https://www.sciencedirect.com/science/article/pii/S2666539522001043. This paper also answers Pete Smoot's question rather exhaustively.
It's both: lower purity quartz gets used for the outer, bubble-containing layer, while higher-purity quartz gets used for the bubble free layer. From the paper: "There is also another option, in which sands of different purities are mixed. In this case, the sand with the lowest purity will be used for the formation of BC-layer, while the sand with the highest purity will be used for the formation of the BF-layer." An executive from a CZ manufacturer confirmed to me that this is how it's done.
Great article! I stayed in Spruce Pine a few years ago to hike Mount Mitchell. It’s a charming town in a beautiful landscape, and I hope they recover quickly from the recent floods. Visit if you get a chance, and bring home a chunk of quartz as a souvenir!
Until the hurricanes hit it wasn’t exactly clear that spruce pine’s quartz supply was not going to affect semi manufacturing. In hindsight, it seems like we’re ok. I’m glad it is so.
But due to the secretive nature of this industry, lack of easily available info on what inventory levels are, and what volumes of quartz are provided from other suppliers, it sure seemed like there could be a problem.
I wonder what inventories of quartz from Spruce Pine are kept later in the supply chain (either as raw material or inventories of crucibles). In addition to that providing a buffer for temporary issues at Spruce Pine, that could give the time for other sources to scale up if some extended outage occurred at Spruce Pine. In general, inventories can be a reasonable way to avoid single source risks in a supply chain when the reason behind the single source decision is price differential rather than availability at all.
Hi Brian, thanks so much for this article! I clicked through some of the links you attached on foam frothing, and although I read quite a few Wikipedia articles, they were a bit incomprehensible. I wonder if it might make sense for you to write an article on the history of ore purification techniques. It seems like a relevant topic for the blog, and I think could be pretty interesting.
I wonder WHY only 6in diameter can be drawn via the CZ process and how much R&D would be required to increase this to 12in?
But also I believe to once have read that CZ was used originally only after the FZ-process to remove the last impurities. Then metallurgical grade Si was tried to purify with CZ to solar grade. But can CZ really nowadays already be used to get from metallurgical grade all the way to electronic grade Si?
My guess is the limits of the RF heating ring as shown in the picture. It can heat to the center of an ingot with a 3" radius but not if the radius is 6". There are probably all sorts of RF energy limits and the matter of quartz heat conductivity leading to a gradient.
I wonder if microgravity float zone processing of silicon could support larger wafers. This was looked at ages ago, with the conclusion it would be close to profitable at $100/lb. to orbit (in 1976 dollars). Starship promises much better than this, adjusted for inflation.
Fascinating article. I found it especially noteworthy that while the mines are in the US, they're run by foreign companies. Supply webs are complicated, something I think people talking about onshoring critical production don't fully appreciate.
Next, I want to know more about how quartz underground gets turned into crucibles. I was envisioning chunks of quartz getting turned on a lathe. I can't see how that would work. What's the actual process? If it involves casting the quartz, what do you melt _that_ in? A superquartz crucible? Is it quartz elephants all the way down?
The image of a cross-section of a double-layer crucible is actually illustrating a combination of mechanical properties, not relative purity. See https://www.sciencedirect.com/science/article/pii/S2666539522001043. This paper also answers Pete Smoot's question rather exhaustively.
It's both: lower purity quartz gets used for the outer, bubble-containing layer, while higher-purity quartz gets used for the bubble free layer. From the paper: "There is also another option, in which sands of different purities are mixed. In this case, the sand with the lowest purity will be used for the formation of BC-layer, while the sand with the highest purity will be used for the formation of the BF-layer." An executive from a CZ manufacturer confirmed to me that this is how it's done.
Makes sense. I appreciate your precision (why I learn so much from your posts).
Great writeup! Lot's of new information even for a 33 year analog IC designer.
Adding an addendum, 8 Part Thread, from SemiAnalysis (which I found only on X, but couldn't trace to their notes on substack): https://x.com/SemiAnalysis_/status/1840871017746698617?t=HGYzXa3-NYcEfwPdDC9tmw&s=19
Great article! I stayed in Spruce Pine a few years ago to hike Mount Mitchell. It’s a charming town in a beautiful landscape, and I hope they recover quickly from the recent floods. Visit if you get a chance, and bring home a chunk of quartz as a souvenir!
Excellent technical article. Learned a lot
Until the hurricanes hit it wasn’t exactly clear that spruce pine’s quartz supply was not going to affect semi manufacturing. In hindsight, it seems like we’re ok. I’m glad it is so.
But due to the secretive nature of this industry, lack of easily available info on what inventory levels are, and what volumes of quartz are provided from other suppliers, it sure seemed like there could be a problem.
I wonder what inventories of quartz from Spruce Pine are kept later in the supply chain (either as raw material or inventories of crucibles). In addition to that providing a buffer for temporary issues at Spruce Pine, that could give the time for other sources to scale up if some extended outage occurred at Spruce Pine. In general, inventories can be a reasonable way to avoid single source risks in a supply chain when the reason behind the single source decision is price differential rather than availability at all.
Hi Brian, thanks so much for this article! I clicked through some of the links you attached on foam frothing, and although I read quite a few Wikipedia articles, they were a bit incomprehensible. I wonder if it might make sense for you to write an article on the history of ore purification techniques. It seems like a relevant topic for the blog, and I think could be pretty interesting.
Seconded; ore purification has come up a few times and I don't really understand how it works (or why it's so environmentally destructive).
I'll go one further; I'll commit to an annual membership if you do!
I wonder WHY only 6in diameter can be drawn via the CZ process and how much R&D would be required to increase this to 12in?
But also I believe to once have read that CZ was used originally only after the FZ-process to remove the last impurities. Then metallurgical grade Si was tried to purify with CZ to solar grade. But can CZ really nowadays already be used to get from metallurgical grade all the way to electronic grade Si?
My guess is the limits of the RF heating ring as shown in the picture. It can heat to the center of an ingot with a 3" radius but not if the radius is 6". There are probably all sorts of RF energy limits and the matter of quartz heat conductivity leading to a gradient.
I wonder if microgravity float zone processing of silicon could support larger wafers. This was looked at ages ago, with the conclusion it would be close to profitable at $100/lb. to orbit (in 1976 dollars). Starship promises much better than this, adjusted for inflation.