Loading summary
Transcript77 lines
- [00:00]
A
What is the hidden ingredient behind every advanced chip? It's not the silicon machines or engineers. It is the world of specialty gases, my friends. These chemicals are purified to extraordinary levels, shipped across continents, are often toxic, explosive and both. To understand this foundational piece of the chip ecosystem, I'm joined by Carl Jackson, co founder and managing director of SSOT engineering and gas, who spent more than 25 years at the center of the specialty gas industry. Co host hosting the inevitable Chris Miller as well as ChinaTalk Akib Zakaria. We're going to do a 101 on the chemicals behind chips. See how China has built a world class semiconductor gas industry in just 15 years and talk about vulnerabilities in the global chemical supply chain which have been exposed by the recent war with Iran. Chris Helium, Iran.
- [00:58]
B
So Carl, tell us to start about the role of helium and the chip making process and why the hormuz shutdown has been so disruptive to the industry.
- [01:11]
C
So on the application side, helium's used by every semiconductor manufacturer in every fab, in every location in the world and it's primarily used for cooling. A lot of these semiconductor manufacturing processes are quite violent, they're quite exothermic, even though it looks very calm and silent from the outside in. So there's a lot of cooling required to keep the manufacturing process at reasonable and workable temperatures. And that's the role of helium mainly in these fabs. In terms of why the stratiformis issue has been so significant, the Ras Lafan Qatar helium production facility now produces about 15% of the world's capacity. So it was an overnight closure of the tap essentially of 15% of the world's capacity that generally operates at or around a kind of equal supply between production and demand. So this is not an example of a molecule that's got a huge overcapacity that's very easy to just take up slack, say from alternative sources and deliver alternative sources of that molecule to a fab. It's also a very difficult molecule to deliver. So you need to move this around in the kind of the world's most expensive thermos flask at minus 269 degrees Celsius. So that whole logistics supply chain is extremely complicated as well. So just turning off the ability to move those assets plus the production that comes from that, that, that site in Qatar is, is a pretty major disruption.
- [03:04]
B
And that's just one of dozens of types of gases that are used in the chip making process. Walk us through at a super high level, like what are the gases you need to make a chip.
- [03:17]
C
Yeah, so I think it's a misunderstood, it's misunderstood the role that gases play generally, I think in chip manufacturing. So when I typically talk to people about what we do and what I'm involved in and gases and semiconductors don't generally sit together, people don't know that semiconductors are basically made using gas. You look at your phone and it's made of plastic and metal. People don't see the inputs required to make that in these fabs. That is basically, as I say, all gas. So there's probably 120 odd different chemicals that go into a fab from various different suppliers. 60 of those will be kind of unique molecules. Then there's a lot of blends and a lot of different, different types of inputs. But there's probably 60 unique chemicals that go in and they'll all basically go in, in gaseous form. So it's a huge chemistry toolkit that's been developed and required to build all chips. So this is not just, you know, future advanced technology. This is every chip everywhere that needs this chemistry set.
- [04:35]
A
So you mentioned cooling as one of the applications. What are other things that the gases do they build?
- [04:43]
C
So I think everybody's seen a picture of a silicon wafer, right? This, this, this wafer that's pulled ultra high purity silicon. The most perfect structure in the world that gets delivered as a, as a building block. So imagine that, that is, you know, I always talk about this in terms of a, of a skyscraper just to tell people, try to imagine what this thing looks like. So this, this building block will arrive and that's your concrete foundations basically of a skyscraper. And then every floor, every elevator, every piece of wiring, then that builds the next 300 floors of that skyscraper, which is essentially what a chip looks like, is built using gases. So silicon, more silicon will get deposited via a gas called silane. Those deposition steps happen over and over again. So there's a repetition of how you build these structures. You then need to knock some of that structure out. So you'll need fluorine based gases to try and etch. And to try and remove some of that you need doping gases. So you need to change the electrical properties of some of these layers. You need to build the transistors in there. That needs different types of gases, different kind of processes. So I think you guys are well aware that, you know, it's an extremely complex, in some ways unbelievable manufacturing process. And it's all made available, all made possible by the use of gases. And this Underlayer, this kind of, this supply chain and this chemistry set is, I wouldn't say an unknown, but not a very well understood input, I think, into this whole process, into this industry.
- [06:35]
B
Perhaps we can talk about the categories of gas. As I understand, some are pretty straightforward, produced in bulk, others highly specialized to the chip industry. Walk us through the key categories we should think about.
- [06:48]
C
Yeah, so I mean, I would split them probably into two. So any major fab needs a huge amount of what we call ball gas. So these would be the kind of inerting gases. These are the gases that are used to create the, you know, the clean environments needed to make the chips. So that would be a nitrogen, for example, an argon, you know, things that, things that don't react or essentially don't do anything in the process apart from keeping things clean. And then you've, then you've got the specialty gases. So the air gases, these bulk gases would typically be delivered by the gas companies with an on site air separation system. So the volumes of these gases required are massive. So they dictate that you would need a plant that would be built next to your fab that would deliver these gases 24, 7. And we're talking about tens of millions of dollars of investment there by the gas company. So once they're built and they get built wherever the fab goes, then there's a relationship there with the gas company that extends over 15, 20 years because nobody then builds a competing air separation unit next to your fab. So there's a relationship there that is anchored around the bulk gases. And then on top of that you've got this other 60 or 70 different process gases that are typically then delivered in a range of different containers. So you've seen these, you know, these larger containers rolling down the highway. You've seen cylinders, there are packages that are smaller than cylinders. So they would then deliver the rest of these specialized gases and they would typically be for deposition. So putting new material onto a wafer, they would be doing iron implanting, which changes the electrical characteristics. They will be doing etching which shape and help create these shapes and devices that you need on silicon. So to say there's different gases for different processes and the toolkit's very big. And I would say a typical fab and we can get into locations, will be taking these materials basically from every geography probably with the exception of China. So if you want to talk about Taiwan as one example, they will be receiving their bulk gases from the on site supplier. But the balance of the materials that they need to run their fabs will probably come from at least another four or five countries. So extremely complicated logistics and supply chains to get that balance of, of the other 60, 70 materials needed
- [09:54]
B
in terms of the, the purity level of these gases, you know, we know that silicon wafer is extraordinarily pure silicon. How do we think about the level of purity in each of these types of gases and is a different bulk versus specialty or what's the way to distinguish the level required?
- [10:12]
C
Well, one commonality is that it's always increasing. So there's a drive on the customer side. And when customer here, I mean a TSMC or a Samsung or Global Foundries, these guys run in very sophisticated SPC and quality assurance data where they're consistently tracking everything that comes into that fab. And obviously as nodes sizes decrease and the, the complexity of these devices decrease that they want to kind of make sure that everything that goes into that manufacturing process is an import, has a higher and higher purity. So you know, at the extreme is the silicon. So the, the Silicon wafer is 11 nines, which is industry speak for almost impossible by you know, other than two companies in the world currently that can do that. It's a, almost a human impossibility. And then the gases range from parts per million, which is now just completely standard. So you know, one part in a million impurity of a, of a gas is just really the basic requirement to be a supplier of any material into this market. And then some of the gases, now not all of them are going into parts per billion and some even get into parts per trillion. And the parts per trillion is, you can, I mean it's very, very difficult to imagine this, but it's basically a heartbeat. In 32,000 years, that's a part per trillion. So for the entirety of anything that's been known to be living on this planet, it's a heartbeat within that scale of time. That's that impurity. That's how to think about how difficult that is.
- [12:21]
B
And I'm just adding up zeros in my head, but that's dramatically more pure than 100 nineteens for silicon wafer. Is that right?
- [12:29]
C
Yeah, and that's the trajectory and where that all stops, you know, I don't know. I mean it's obviously leveled out. There was a huge push in the early days of semiconductor where there was big advances made in the way that materials are purified, the way they're packaged, the way they're delivered, the way they're used. And it's obviously, you know, know it's tailed off a little bit, but they still continue to try and improve the purity of all of these materials to impossible levels. To levels really that is, you know, the difficulty is measuring the impurity, not making the product.
- [13:09]
A
Yeah. How do you measure part per trillion
- [13:14]
C
with extremely expensive analytical instruments that hardly anybody knows how to use properly. And I mean, really, it gets into, as I say, it gets into kind of ultra specialist and extremely expensive equipment that's so sensitive that it makes that whole job very difficult. So I think that ultimately will be the limitation. I mean, there's a cost implication obviously of increasing the purity of these products, but ultimately it's the metrology around it that will be cost prohibitive.
- [13:46]
A
Yeah, it just seems like a weird, like when you're that deep, it's like there's some like, you know, Schrodinger's principle going on where, like it's the, you know, measuring tool's fault more than, you know, your poor gas refiner or whatever.
- [14:00]
C
Yeah, yeah, yeah.
- [14:02]
B
Could I ask a naive question of how do you get these gases? And I guess it's different for different types of gases, but if you want a gas purified to a, you know, one part per trillion level, how does that happen?
- [14:17]
C
Well, that's, I mean, that's the extreme, Chris, as well. I mean. But you know, from a. If you want to take one example of a. I mean, the other thing to say is quite a lot of this chemistry set is also used in industrial chemistry. So one example I could pick is hf. So hydrofluoric acid, which is made in the hundreds of thousands of tons of. So that material starts life in a mine, so it starts life underground as a material called fluorospar. So it's literally dug out of the earth at no concentrations that you would recognize to go into a semiconductor fab. It's then made into an industrial grade of hf. So that isn't, again, that isn't at that point destining for a semiconductor fab. So everything's fine. And that's used, as I say, in the hundreds of thousands of ton scale industrially. And then the semiconductor guys will get involved and they'll take a small part of that volume and they'll start to process it using very different types of purification systems that have been developed around the requirements for semiconductors. So you start, you then, you know, a typical industrial grade HF will be three nines. And that's good enough for almost everything required on earth that uses HF for pharmaceuticals for all of the different industrial applications. But not for semiconductors. So that material then needs to go from that three nines up to six, seven or eight, depending on the process. And that's where the semiconductor supply chain, the people that are involved in this really add their magic. It's how do you get and maintain and then ship and have that high purity product delivered all the way to the wafer essentially through this supply chain. So at that point it becomes very specialized and you know, you need then to rely on a very small number of people in the world that can, that can do that.
- [16:37]
B
And if you get like a jar of HF at three nines and want it to be six nines, what do you physically do to that jar to get the level?
- [16:44]
C
In that case, you put it through a distillation system. So it's, I mean, it's a little bit like the petrochemical distillation system. I mean, distillation is distillation. But not all distillation systems can be set up a to handle HF&B, to get to that level of purity required for these semiconductor customers. So it's think about it as advanced distillation with other steps built into it to take out impurities that you wouldn't normally need to. So it might be distillation plus another couple of steps that, as I say, wouldn't be necessarily in industrial processes.
- [17:26]
D
And sorry, just to, just to get a sense, is the jump from 3.9 to 6.9Significantly easier than it is from 6, 9 to 99 in the sense that, you know, we think of the semiconductor industry of going from 7 nanometer to 5 nanometer as extremely difficult, a lot more difficult than from 28 to 14. Can you just give a picture of like how much money it takes to get to the next step and how much time it takes?
- [17:51]
C
Yeah, it's a good question. And it depends on the product, obviously. And some materials are easier than others to purify. I mean, the next nine from where we're at now, the next nine is extraordinarily difficult to produce and to deliver. And almost nobody's interested investing to get to that next nine because there's no demonstratable evidence normally that that next nine is actually required in the process. I mean, there's an interesting part of this where the drive from the customer side to get even more pure material delivered, consistently improving that purity isn't actually required to make the devices. So there's a drive there that's all based around math and SPC and quality data. It doesn't necessarily triangulate with what's needed on the process side. And if you understand that and you're on the production side, then you really are not that motivated to try and increase your whole production system and your supply chain to that next nine. It might give you some competitive advantage, but ultimately nobody's going to pay for it unless it's really needed on the manufacturing side.
- [19:13]
D
And you were saying earlier about, at some point the semiconductor guys get involved and say, hey, I need an extra 9. I'm willing to pay this much money for. In a sense, is the semiconductor guy in that scenario, Is it TSMC saying we need better gas, or is it the tool maker, the deposition tool maker, where the gas is used, saying, we need the better gas? What's the player here?
- [19:36]
C
Yeah, that's a good question. Again, so the industry talks about recipes. So, so when these gases are used, they're used in a recipe, just like a cooking recipe. And the, and the recipe dictates what materials you need to make a certain, you know, to make a certain type of device at certain type a certain time in the manufacturing process. And those recipes typically will be developed by the tool maker. So they will tell you what chemistry you need and at what purity. They will guarantee that you can make a particular device. So they say, okay, you know, you'll need five materials and they all need to make this purity. And if you provide that into this tool, you can make this device. So that's what TSMC will receive. They then will improve that recipe. So they've got thousands of people working on optimization of their processes and their tools. So there's a juxtaposition, you know, always between improving, optimizing, but not changing anything. Right, but that's always a force. That's this fighting here. But, but they typically will have people that are optimizing and improving, and part of that will be, okay, we need pure material. Sometimes it's just a blind. It just needs to be purer. If you say, you know, if you challenge it, we're the customer, we're telling you that it needs to be purer. And why it doesn't matter if we're going to keep buying it from you, it needs to be purer. Okay, but at the toolmaker level, they will typically, as I say, they will put performance guarantees around a particular recipe set and a particular purity for that recipe set that will work. When it gets into the hands of the guys that are actually making the devices, then it goes into a different world. And some customers don't change. Some customers are pretty, pretty active in terms of demanding, you know, better materials. But as I say, it's slowing down. The parts per trillion thing is in the really extreme, it's an outlier. And I, I don't think. Well, I'm certain that all of this kind of standard chemistry isn't going to get there. Some specialist materials, maybe for some ultra specialist devices, but generally people are not going to be demanding that. So it will stay in the parts per million. Parts per billion. But people, it's easy to forget that parts per million even is an unbelievably difficult thing to deliver for a material that is, for example, an hf. To go back to that example, HF will try and eat the cylinder that it's being delivered in. Right. So you've got this material.
- [22:34]
B
Could we dig into that? Because I think one of the fascinating parts of this is that it seems like half the materials we're talking about are either toxic or explosive or both. And so I guess A, tell us why it is that you can't make chips without all these seemingly awful chemicals. And then B, I'd love to hear more about the challenges it imposes in terms of transporting them.
- [22:57]
C
Yeah, so I often say that this, almost everything in this toolkit is lethal. So it either poisons you, it explodes instantly with contact with air. It can kill you quickly, it can kill you slowly. But the lethals here is not a bug. It's the specification. As I said before, this process, if you've been to a fab, right, and walk around a fab or you've seen pictures of a fab or videos or whatever, you see this very kind of quiet, controlled, seemingly kind of very delicate process that's going on, right. To make these chips delicate is I, you know, people that. I think that's how people envisage it. But at the kind of atomic level, it's an extremely violent process. And the violence requires the most violent chemicals and the most kind of reactive chemistries. You know, this deposition and this etching process that builds these layers. You know, you're trying to drill the equivalent of an elevator shaft down 300 stories in one go straighter than anything that's kind of ever imaginable doing it physically. So you need something that's extremely reactive in the process to allow you to be able to do that. Similarly, when you're doing this iron implant. So when you're trying to change the electrical characteristics of the silicon so that it actually does work and it's a transistor, you're firing boron or you're Firing phosphorus or you're firing arsenic, all of these lovely gases that you need to do some work. You're firing these things in at, I think, the numbers 400 km a second into the silicon wafer. So this. And these things basically disappear. This molecule disintegrates when it hits this wafer. So it's like firing a bullet into a concrete block. That bullet needs to go exactly where you need it to in that concrete block. The depth, the angle, the everything, and then it disintegrates and goes. It leaves something behind that does some work. But the process itself is extremely reactive, as I say, extremely violent. And that's why you need these chemistries, which is unfortunate, because to make them and to deliver them, as you say, safely is a little counterintuitive when you see the PAB itself running. Right? So, as I say, the chemistry set that you need because of the difficulty of the device that you're making needs to be large and it needs to contain these difficult gases. And to your point, Chris, about how are they delivered? All kind of weird and wonderful ways. And it's part of the ip, if you like, of the semiconductor industry and the people that feed into it. How do you get these products to the customer? As I say, some products are eating themselves on the way. So you make these products and they start consuming themselves. Some of the products start to consume the packages, the cylinders that they're in, and create impurities on the way. So they've got a very short shelf life. Some of them are so dangerous that you can't just ship them in a cylinder. You need to ship them in some other container that in the event that there's some damage to that container, that these molecules won't come out. So there's innovations around the handling and the safety and maintenance of the purity in this business that is never really seen. But it's a huge part of enabling how these devices are made.
- [27:08]
B
What's the most dangerous chemical used in chip manufacturing?
- [27:13]
C
There's so many. I think that the one that people are most scared of is hf. I think, because there's been a few incidents with hf, it's used in fairly large volumes. It doesn't immediately kill you. It gets into your skin and then. And then pulls the calcium out of your bones over some time and kind of kills you slowly. I mean, I have to say, in all of this, I'm not a kind of scaremonger, and I'm painting a. You know, I'm painting a picture here of what could happen if this industry wasn't as well managed as it is, but actually there are barely any incidents these days. Certainly on the fab side, there are almost no reportable incidents in a year across every fab everywhere in the world, which is absolutely unbelievable when you think about the chemistry set. Similarly, in the shipping of these materials, there's barely any incidents. On the production side, there are a few. Right.
- [28:19]
D
But.
- [28:20]
C
But nowhere near the amount that you would have, say, in any, you know, a normal industrial process. So it's probably one of the safest businesses, you know, given. Given the kind of risk profile of these materials. So, But. But I think, you know, HF's not. Not a good actor, you know, that's a difficult one to deal with. Our scene, our scene is used on every chip. Every chip requires arsene, you know, obviously arsenic. What do you call it? The housewife's poison of choice. So it's the same molecule, right? The same molecule that makes a transistor work is the one that you used to be able to buy in the 19th century to slowly kill your husband. So it's the same chemistry just doing a different job. So, yeah, so those aren't great. Silane's not great. So this, this, you know, silane is silicon in a gas form. So, you know, people figured out a long time ago, okay, we. How do you. You keep. You need to build these structures using silicon, right. Still. So you get your starting wafer, but then where do you get the silicon from to build these structures where you can't just kind of get a matchstick size of silicon and glue it on. Right. You need to somehow build it in a different way. So you need silane. Silane's great. Unless it starts to leak. And if it leaks from a cylinder, the current safety protocol for that is leave it leaking. Right. It's quite counterintuitive. Again, if a valve on a cylinder fails, the silane starts to come out, you literally get sand all over the place because it reacts with air and the hydrogen gets replaced with the oxygen, and you end up with silicon oxide, which is sand. But if you create some kind of a static or some issue that may cause that to ignite, you end up with a fireball coming out of the cylinder as opposed to a load of sand. And it's still really not very well understood why that happens, but it does. So if you are training someone to deal with a silane incident, the instruction is, if it's leaking, leave it to leak. Don't close the valve, because in closing the valve, you might Create a fire, which is worse than creating the sand. So the different, as I say, it's a, it's a big toolkit and really none of it is very friendly. You know, even nitrogen, even the most benign chemical that's used in, in large quantities is really dangerous. You know, if, if you're in a room just with nitrogen, 79% in the room now, which is fine because you've got your oxygen there, but if you start getting concentrations of nitrogen that are any higher than that, then you start having issues. And those are the most common safety issues that you'll see is a nitrogen asphyxiation issue rather than some of these more dangerous chemicals causing problems.
- [31:51]
D
Let's, let's. So we have these, you know, violent chemicals and sometimes they're delivered at purities that are completely mind warping. Can we, can we just talk about who, who are these guys? Who are the companies that are actually delivering them? Like how many of them are they and what, what differentiates one from the other?
- [32:11]
C
Yeah, yeah, so, so probably best to look at that over a time period because it's changing rapidly, really, really quickly. And the change is almost all happening in China. So if you go back maybe even just 10, 15 years before the big fund, which is the China Initiative to localize semiconductors and all of the supply chains around it, then it was, it was segregated depending on the chemistry. So Japan historically, for example, was really good with the fluorinated gases. Right. So anything that had this fluorine backbone and there's, there's probably 15 different chemicals and chemistries that go into, into semiconductors that are fluorine based. Japan was, was the place for that. The US had a, had a pretty broad range of manufacturing capabilities because that's where this industry grew up. So there was a self sufficiency there maybe 20 years ago of almost everything that was required. That's somewhat depleted now, the scale and scope required, but that kind of existed there. And all of these gases were typically made by a few Japanese guys, but then the majority were the majors. So Linde Air Products, Air Liquid, the companies that were building and putting in the air separation units were also always providing all of the balance of these other chemicals required. And it was a self funding, self financing. You know, you put the, you know, the air separation unit in and that kind of was the anchor that paid for the development of all of the rest of the materials that go in. And they had a full portfolio that they could then offer to that single customer. So almost you go back 20 years and it almost all came from one of the majors and they had, as I say, almost a kind of complete portfolio to feed these semiconductor customers. As I say, that has begun to change mostly.
- [34:31]
B
Can I ask one more question?
- [34:33]
C
Sure, yeah. Yeah.
- [34:35]
B
Before we go to the present situation, just to kind of understand what the industry looked like 20 years ago. Would it be a situation where like one intel fab would have an air separation unit from Air Liquide and then a different one would have it with Linde? Or would it be company to company relationships you generally see, or fab to fab?
- [34:55]
C
The fabs like to split risk. I mean, there's, there's low risk really now with these air separation units that, you know, they're so reliable. There's so many installations around the world. But so, you know, for example, if a TSMC were to build a new cluster of fabs, say three fabs in a new location, they would typically award an ASU per major. So, you know, Lindy would get one, maybe two, Elite Keyed would get one, maybe two. There's a very little differentiation there in terms of what the customer would get out of that. You know, obviously extremely good reliability, quality, you know, some negotiation on pricing and. But, you know, so not one major would win them all. But the split between the fabs and the awards would really depend on, you know, the situation really at that customer site or whether there's something compelling, you know, that one of the majors can offer. So it's basically split, Chris, almost evenly.
- [36:05]
B
So, Carl, you said industry's changing. Tell us what's new in the gas industry and the role of China.
- [36:12]
C
Yeah, So I don't know whether the Big Fund is something that's familiar topic for you guys, right? I guess it is. And really that's the thing that's driving the China change at this level. Right. So big fun one was kicked off, I think in 94, something like that. And it was different. We could see it was different at the time because it, you know, unlike other similar initiatives like the chip site, for example, the Big Fund was going after all levels of semiconductor manufacturing. You know, from. I won't talk to the things that I don't understand like the, you know, the design and the, you know, the tool manufacturing infrastructure, but all of that was included in the Big Fund. So not only do you get the new fabs and the new IDMs, the new people that are making the devices at the end, but you're getting the whole supply chain and they went after everything all at once. And the way that, the way that, that was kind of handled. There was a, there was a lot of money. So I think it's now up to kind of $120 billion that's been invested in that whole infrastructure, that ecosystem. And it was managed regionally as opposed to a kind of top down directive of okay, you, you guys do this over there and you guys do this over there. So there was, there was, there wasn't a kind of top down, there was a top down directive, but then it was a managed on a regional basis. And where I'm going with this is that those regions ended up competing with each other for all of the different materials in this chemistry toolkit. So let's take one product, NF3, right? NF3 is part of the fluorine family. It's again, it's used for, it's used for cleaning. So it's like the, it's like the janitor of the semiconductor world used by everybody on every process. And at the time China didn't have any NF3 capability, so it needed to import NF3. So, okay, so we need to do our own NF3 thing. So different provinces then with different leadership and different scopes started to develop in parallel a supply chain for these different materials. So there was basically as a result of that, a massive over capacity. So there was new capacity to start with. So here are new guys that are coming in that have never been active in the production of these materials in these supply chains. So it was the whole story of China. Low quality materials don't trust where these things are being made and how they're being made. So that took a long time and they started out feeding themselves. But because of the way that the Big fund was organized, they ended up with a huge overcapacity. So now you've got a situation where you've got China plus the rest of the world in terms of their ability to provide these other 60 gases. So they've got now capability to make almost everything in China independently and in parallel with everybody else in the world and at a capacity that could almost feed everywhere in the world from China. So we've ended up in this situation that is totally different to where we were 15 years ago because of the way, as I say, that this Big Fund has been A funded and B rolled out and really without any checks and balance on. Does this make any sense to put this capacity in? Well, we don't care. It doesn't matter. We do our job. And as a result of that there's now a lot of material available that can now feed and compete with this existing supply chain that had a very small number of suppliers.
- [40:40]
B
And do we see that happening? Do we see Taiwanese or South Korean firms buying gases from the Chinese supply chain?
- [40:48]
C
So Taiwan is 100% reliant on Chinese supply chains today. 100% reliant. So, Chris, you talk about missiles and I could just Change that to NF3. To use the same example, if the Chinese government decided to put an export restriction on an F3, then the Taiwanese fabs would shut down.
- [41:18]
B
Help us understand. Yeah, help us understand how we should think about how easy or hard it is to find alternative sources of supply. And I guess I ask because I think back to 2022, start of the Russia, Ukraine, war. Both countries big suppliers of neon, as I understand it. Even the Azovstal steel plant, where there was the big siege at the start of the war, was itself a supplier of ne. And I think that was a sort of a crisis for the semiconductor gas industry. But it didn't end up impacting chip production much at all, which suggests, at least for neon, there was some flexibility in the market. Is that unique to neon? Is it different from NF3? How should we think about that? Elasticity.
- [42:01]
C
Yeah, I think helium is probably the best example of the supply chain that's the most inelastic, and that's mainly because of the packages. So, you know, the most basic package to get these gases from supplier to consumer is a cylinder. You know, these cylinders are $50. And you, you know, there's, there's, there's no shortage of those. There's millions circulating around the world. The, the helium packages, as I mentioned before, they're a million dollars each and they're extremely complex. They've got a. You know, you've got 45 days to get from production to consumption before you start to lose the, the product. So that, that, that's pretty inelastic. But then on the extreme, you've now got to go back to NF3, for example, if you want some, if, you know, if you want to talk about how elastic NF3 is, you've got a producer in China. So the Chinese domestic consumption requirement for NF3 is about 8,000 tonnes. You have one producer in one province in China that's making 55,000 tons a year, right? So those guys, you better believe, are busy recruiting international sales and marketing teams and going out and trying to sell their NF3 internationally into Taiwan, into South Korea, into the US, into anywhere that can take that material, because they've Got almost zero incremental cost of selling anything that is above and of what's required domestically. So you've got real extremes on either side of this where a little, I wouldn't say the Iran situation is a minor blip, it's a pretty big one. But anything like that. If you saw something similar on an NF3 plant, which in fact we did 20, 24, NF3 plant blew up in Japan, took one offline. There was a. The kind of impact that rippled around the semiconductor world was pretty small. The existing customers needed to re qualify and re establish supply chains. But for a material that's got an abundance and is easy to deliver because the packages are available there, then those supply chains can reroute and the customer can take new material really very quickly.
- [44:44]
B
And is there a reason that, I mean, obviously subsidy over capacity. I think we understand that story. Absent that, is there a reason why you'd produce more NF3 in China than elsewhere? Energy costs or some sort of comparative advantage? Or is it solely a story of, of governments paying for it?
- [45:05]
C
I think it's a mixture, it's a blend of things. I mean, NF3 is part of this fluorine family. Fluorine comes from Florus bar. China's got 70% of the world's florist bar. So that whole supply chain is, is integrated inside of China. So they can do all of it internally themselves at an extremely low cost. So some of this is just about economies of scale. Some of it is about just ignoring how much is required domestically and building out huge capacity for some future requirement. Some of it is just an economic decision that doesn't make any sense from a Western point of view. Right. You're looking typically, if Lindy invests in a plant using the Western example, the amount of due diligence required around how profitable that plan will be, who the customers are going to be, what the ramp of that thing is going to be. It's all like a, it's a finely polished machine. I mean the Chinese example sometimes doesn't follow the same economic route. Build it. We've got a directive to build it. How big do we need it? Oh well, we need 10,000. Build 25. Okay. I mean it's that kind of math really. I mean it's no exaggeration. And then somebody in a province a thousand kilometers away is making that same decision. Because it doesn't matter. The availability of these products internally is far more important than any kind of return on investment on an individual plant. And the result of that now internationally is what we're starting to see.
- [46:57]
B
So you mentioned fluorospar, critical for a bunch of different gases. 70% mined in China, I understand, some mined in Mexico. Can you walk us through floraspar mined to gas? Because it does seem like this is one of the areas of focus in terms of China's market position.
- [47:16]
C
Yeah, yeah. So I mean it literally is dug out of the ground, you know, so it looks like a fluorospar, just looks like a rock. Right. Those, those rocks then are put into an autoclave. So just a, you know, basically a massive hot reactor. And China has got hundreds of thousands of tons of capacity of these, of these reactors. And, and in, you know, we're talking about a few thousand tons required for semiconductor. So it's an, it's a tiny proportion of what's made industrially. So yeah, you dig the rock out of the ground, it goes to a industrial HF producer. There's probably 10 of those world scale in China that then gets converted into industrial hf. And as I said before, the requirement then for the semiconductor business is to go and siphon off a tiny proportion of that and make it into ultra high purity material. So I think, you know, but the thing to come back to is these families, these chemical families. So HF is then used to make fluorine. Fluorine is then used to make these, you know, these C4, F6, not necessarily NF3, but sometimes NF3. There's a whole family of chemistry that's built off of HF or built off of fluorine that requires additional processing but ultimately comes back to that fluorospar mine.
- [48:59]
D
I'm curious if there's any cases of it going the other way. Right. I think we're pretty accustomed to the story like rare earth minerals now, fluorospar, other things where it's all coming from China and there's that great dependency. But I'm wondering if there's any cases of these 60 gases or so where it's actually. Oh, sorry, the mine's just not in China. It's actually in Peru or something and China has to import all of it. Is there any cases where the dependency goes kind of two ways in that sense?
- [49:26]
C
Yeah, I think helium's an interesting one for China. I think China is, you know, the kind of biggest gap and the biggest high value and important material for semiconductor is helium from a Chinese perspective. And the only reason that they've, they haven't got enough helium is just a, is just a geology question. Right. The, the US is extremely lucky in that it's got great geology for extracting helium. And anybody that's making LNG also has got a great capacity for making helium. So remember, you know, typically helium is, is not something that's, that's produced exactly. It's a co product from something else. So up until recently, nobody made helium or nobody mined or extracted helium. You extracted LNG and helium just happened to come along with the lng. And when helium started to get kind of high value, important applications like MRI and like semiconductor, people became more interested in the separation and the, and the production of that helium. But China has a challenge in that the amounts of helium in the LNG that they produce is extremely small. Where they produce it is basically all in the wrong place. And it's distributed around a couple of hundred different small sites. So their LNG infrastructure's not conducive to making helium. So they're still a net importer of helium. Um, that's been fixed as you know, in, in the Chinese way. They figured out a way to, to produce their own helium even with these restrictions. But it's going to take some time to develop the infrastructure that they need so that there are not a net importer. But at the moment that, that's, that's probably their biggest vulnerability, to be honest, for, for the semiconductor supply chain. But as I say, it will get fixed and you know, they're busy trying to figure that out in really smart ways. Taiwan and Korea, on the other hand, they're in a similar position where the geology doesn't allow economically for helium to be made. They will always be 100% reliant on helium, plus almost every other chemical that needs to run these fabs forever. Unless they do something that they haven't yet done in 30 years, their supply chain vulnerabilities will remain when China's fully kind of fitted out.
- [52:23]
A
So Carl, what's your, given your vantage point, how does the comparison between the Big Fund and America's Chips act look to you?
- [52:37]
C
I think at one level, which is the fab level, so reassuring the fabs. So the actual device manufacturing, it looks somewhat similar. So the CHIPS act is about reassuring the manufacturing of semiconductors, typically though with existing manufacturers. So, you know, the CHIPS act tried to and successfully got, you know, TSMC to make a new investment in Arizona. It's got Samsung to make a new investment in the US So at that level it's worked because there's going to be more domestic production of semiconductors. Where it stopped short in my view, is anything that's underneath that layer that is required to support that domestic production. So I haven't seen anything at all yet about how the supply chain and the security of the supply chain and the chemistry supply chain that feed those fabs and any new fabs and the Tesla fabric. So you know, all of these kind of domestic production requirements are being announced but without any of the supply chain underneath it. So if nothing changes there from the US perspective then there will be a total reliance on the import of almost every chemical required to build those new capacities. On the Chinese side they've done it in a different way as we mentioned before, in that they've set out to put the entire ecosystem in all at once. So it's not just new fabs built by existing suppliers, it's new companies entirely building new fabs, creating new technology. And underneath that is the full and complete supply chain to put all of the materials required to build those components in fully domesticated. So the scope and the scale of the big fund is on a, you know, it's on a much larger and I say not perfect by any stretch, but it's certainly more integrated and forward thinking than chips in its current form. That looks like it's either either ignored, not quite got to it yet, or isn't interested in the supply chain vulnerabilities underneath those new fabs that being installed.
- [55:34]
D
I think you said it just right and I think at least for me this has been like history just keeps on rhyming. In terms of Forest Bar, in terms of so many things that everything I saw, every new supply chain I look at, it's almost like the same story, copy and paste, you just swapped out words. Gallium, rare earths, there's just so many things I think it's really interesting, I think it needs to be better studied as to Western economics makes you prioritize things that are later down the value chain even though they're dependent on things that are way more upstream. And for one reason or another China has become dominant in those things, whether it's because of state policy or because of just their perceptions of secure supply chains. And then you know, that sometimes poses a threat for us in the United States or the west because you know, I might be making the thousand dollar iPhone but it's still dependent on the $1 gas or $1 mineral coming from China, which leads to some pretty disruptive or strange geopolitical topics. I'm curious what you think about this, Chris.
- [56:45]
B
So Carl, if a chip costs a hundred dollars to make, what share of that is gases?
- [56:51]
C
That's a really good question. So the numbers typically that are talked about, I think it's 10%, something like that. If you, if you want to kind of round it, it's around 10% is the bill of materials that contribute to the value of a, of a chip. So that's all 60 combined. Depends on the chip. I did some work on this maybe 10 years ago when we were trying to figure out what the value of the products that we were selling and the company that I was working for into an iPhone. So for each iPhone manufactured how much value of gas is in there, we did what's the weight of the gas in there? Right. Because this is you all you, you trying to explain that this is made with gas. It's like it can't be. It's solid. What are you talking about? So we tried to figure out how we could contextualize it a little bit more and then how much is in a Tesla or. So it's about 10%. But again, depending on the structures you're trying to make, the devices that you're trying to make. But as I say, that's one way to look at it. The other way to look at it is which one of those 60 can you live without and still manufacture your products? And the answer to that is zero. None of them. You need all of them exactly when you need them exactly at the spec that you need them. So the vulnerability is there. It's not how much you know, do you need to. What, what's this, you know, this, this kind of 10%. What's the breakdown? What's it worth? The question is what do you do if one of those is missing? How do you run your fab? And it, it looks even worse because if you break the 60 down into 10%, you might have a line item in your bill of materials that's $100,000 a year for one gas that is barely used. If you are missing $100,000 worth of gas, you can't run your fab. So the output cost, whatever you want to call it, the opportunity cost or the downside or the whatever it is, is huge in relative terms to the, the, the input gases. So you more important that you get them there and in spec than. Than what you pay for them ultimately.
- [59:26]
B
Sorry, I have one more question. This is a sign of good podcasts when we can't stop dressing. So given that one might think that you're going to stockpile a bunch of gas on site to avoid supply disruptions and I know you mentioned certain types of gases you can't actually stockpile because they start corroding the container. They' how do chip companies think about what's the right inventory to keep on site?
- [59:54]
C
Yeah, well, I have to say, you know, credit where it's due to the gas companies, typically all of those materials arrive on site. There's barely any safety stock. I mean, some of these materials you can't physically safety stock because you don't have a permit needed on the sites to store a large amount of the material. So for hf, you want to go back to hf, you can't just put a few tons of HF in the fab somewhere waiting to be used just in case there's a supply chain. You can't do it. You know, you need to order and have delivered basically what you need for that next month of production. That's all you can store. So it's, you know, there's, there's no stockpiling. You know, if you want to talk about stockpiling as in there's some disruption in the supply chain and we can cope with it. There is none. But these supply chains work miraculously. They're so complicated, but they end up with all of these materials and molecules being delivered through the customer's gates when they need them all of the time. And you could say that's part of the problem as well, in that it's become so reliable that if there is really any disruption, there's no facility in place that can cope with that disruption very well, particularly if all of these materials are coming from overseas. And I think that's what the Chinese infrastructure, that's where it's going to kind of see its power. It's not just a cost question, it's the fact that they've got multiple domestic options, all at the right capacity, all being made up the road. It might be a few thousand kilometers, but it's not across an import export barrier that needs a ship or a plane or an export restriction in place to disrupt that supply. And I think that's the vulnerability more so than anything else, I think in this business.
- [62:06]
B
Pretty ominous because it's obvious what the implications are.
- [62:10]
C
Well, as I say, for places like Taiwan, and I come back to it, I live there, I love the country. But it is the single worst location that you could pick to put semiconductor fabs. It's got no natural resources, water is an issue. It's got these geopolitical risks. They had a brilliant pioneer of that industry in Morris Chang. But if you were just to look on the globe and Drop down a semiconductor industry, you could make an argument that Taiwan is probably not a great place. Seismic Zone 4 or whatever it is, the earth moves regularly in Taiwan. Not a good place on the face of it to make these kind of, these devices. But anyway, it is what it is. But yeah, they've got some existential risks I would say. On the supply chain side still we
- [63:09]
D
talked about like 10, 15 year contracts for like a gas supplier. I don't know how healthy that seems. And like, I don't know if pure gas is going to be the next bottleneck for the next node. I don't know if that's a worry, if it's just not going to happen because the investment's not there. What do you think?
- [63:28]
C
So, I mean 15 years has become, it's become a norm and I think it's based around a sensible premise in that these plants are so reliable that the 15 year operational lifetime of these air separation plants has been proven over and over. And as I say, you know, some of these air separation units will be 50, 60 million plus. So no gas supplier is going to put one of those in with a one year contract. It's pointless. You aren't going to move it. The piping required to get these gases from the separation unit to the FAB is under the ground. All the infrastructure is permanent and built in. So there's no sense in not having that long term relationship. On the special gases side on this balancer plant is where you start to see these problems. Right? So they will be typically one year contracts, sometimes more. Then there'll be rolling contracts. Right, Here's a one year contract and if you don't do anything wrong then you might get another year and if you can improve the quality, you might get two years. But it's, it's definitely not 10 years. So these are, these are shorter contracts. The, the, the race to the bottom on pricing is in full force on almost all of these materials. They're fully commoditized, right? Very, very difficult to differentiate between these products when the customer itself is demanding that they're identical from everyone. So you, you, there's no way, right? They want silane, six nines, you, you, you, what are you going to do? You're going to deliver it in a different color bottle, right? So, so it's, it's necessary that you've got the same thing and that's great for the buyers because they're seeing no difference apart from pricing and some reputation. And what else are you guys supplying me? What kind of relationship have we got. But, but really it's difficult to remain an unchallenged incumbent. And the race to the bottom will result in the big companies, the airlocked, the Lindys, the mervs, having less and less money to spend on R and D and being less and less prepared with the next new material that's required, that's been developed in a lab in Santa Clara or in Tohoku or, or wherever the OEMs are putting together the new recipe list required for the next device. Those materials may not be ready and certainly the supply chains may not be there. You go back 15 years ago and there used to be really nicely integrated programs, tech set. There was a lot of materials that were being developed at my time in Lindi that had a very high risk of never being used. But we were okay, right? We were happy to spend the money on R and D for products that we thought would be going into the next generation of devices. And that was all part of the program. But if your portfolio is under a lot of cost pressure permanently, and that's getting worse, then you can see this kind of not a death spiral, but it's a spiral that make R and D economics extremely difficult because that's the first thing that comes out, right? You aren't going to continually spend on 10 different R& D programs when everything in the portfolio is under pricing pressure. At some point you just say, okay, I just need to compete on this level. And that's at scale and price and not necessarily doing the kind of frontier stuff. And I definitely see that as something that's happened over the next, you know, over the past kind of 10 years and I think will continue to happen. I think there'll be a disconnect there and these new material supply chains will take longer to ramp.
- [67:43]
B
I have a dumb question, Carl, if I can, can you tell us what an air separation unit is? How do they work?
- [67:50]
C
Yeah, I don't question at all. So you get different flavors of air separation units, but basically all of the components that are in air, so nitrogen, oxygen, and in some cases argon and the rare gases as we call them. So this will be neon, krypton, xenon, they're all in the air at some concentration. And the only way that you can economically get the amount of nitrogen or oxygen that a fab needs is to separate the air right next to the fab. So you will have seen these things, but you, you won't have recognized them. These things are a tall, you know, 30, 40 meter tall box that will be sat close to a fab. Basically, they just suck air in. So they've got huge compressors that suck air out of the atmosphere, cryogenically separate the air. So it'll cool it down, liquefy the air, basically, and then at different boiling points, you'll get different components of the air that you can separate and you can purify and you can deliver then straight to the customer. So the customer, typically a semiconductor customer, needs a lot of nitrogen. So nitrogen is used for, you know, inertia. It's used for, you know, to basically for environmental control. And they need a huge amount of it. So it would be physically impossible to ship that amount of nitrogen in. You have to make it on site. And the only way you can make it at those volumes is by separating the air. So, I mean, it's an unbelievable thing to do, but it's now so kind of commonplace that it's just that, oh, well, it's just an air separation, just separating the air cryogenically and fitting it into a fab 247 for 15 years with zero interruption. No quality problems. Well, you know, it's just Tuesday at a gas company that's, you know, but yeah, that happened regularly.
- [70:04]
B
So, Carl, you've mentioned that the Japanese are the experts in fluorinated gases also, you know, lead in many of the materials that go into 700 manufacturing. But the three key companies that. That have historically provided gases, air products, liquid. Linda, Us or European. Why is there not a big Japanese gas provider?
- [70:29]
C
Yeah, I mean, there are. There are large ones, but the, the sort of myopic focus of, you know, in Japan, you know, some of them have got JVs and they've done things internationally. But I think it's fair to say that they've never quite got the scale and scope of the big three that you've mentioned. As I say, I think mostly it's a domestic focus. You know, it's a Japanese company that's set up to serve a Japanese market. So I think there's a little bit of that that you don't see necessarily in the other companies. There's the. There's a portfolio mix. You know, the. Typically those. Those companies that you've just mentioned are very strong in air separation, and so that allows them to build a global footprint. Because if you need to put an air separation unit somewhere, then you need to go and physically do it, not just make it in one place and sell that material to somewhere else. So you end up with a global footprint as a result of that. And the Japanese are Not as strong I think, you know, historically with their separation. So I, I think it's a mixture of things, Chris, like, like always with, with, you know, with a complicated question. There's a, there's a, there's not one thing that I could pinpoint why there is not a Japanese equivalent of a linde. But certainly in, if you go to Japan, you know, there are dominant and very large scale industrial gas and special gas manufacturers with a high degree of self sufficiency. If you compare it for example, with a Taiwan.
- [72:16]
B
Should we turn the supply chain question? So do companies know if they're getting NF3 into their fabs, where the F is coming from?
- [72:30]
C
I would say most cases, no. I think that there's probably examples, outline examples maybe from people like tsmc. So TSMC are kind of at the extreme, I would say of their, their quality systems and their supply chain tracking. So they may know one or two steps. But typically, you know, the purchaser of those materials at FAV will be interested in the specification, how much you can provide and what the price is. They will not definitely go back one, two or even, you know, there's sometimes the six steps in these manufacturing process to understand okay, you know, at what point do my supply chain risks here converge, if at all right. Does it happen? No one's really looking at that level. Sometimes it matters, sometimes it doesn't. The reason that mostly it wouldn't is that the volumes required for most of these gases are so small compared with the industrial volumes that are being made that there's not a huge amount of risk. But where people are not looking is where these materials are coming from and whether there's any issues there in terms of those countries being able to turn off those taps via export controls, for example. So it's more so the geopolitical risks I think than, you know, than the actual supply chains that may converge into a single mine somewhere for 10 or 12 of those materials. The mine is so big, the industrial base is so big that it isn't the question of okay, I'm going to run out of capacity. It's if something geopolitically happens and I'm unable to access this, then what happens? That's, that's more of the important question.
- [74:43]
B
Seems like a highly relevant question to be asking Carl.
- [74:47]
A
What, what are these people like,
- [74:50]
B
you
- [74:50]
A
know, who gets into chemicals? How would you distinguish them from other, you know.
- [74:56]
C
Wonderful. These people in this business are wonderful. I, I mean, yeah, I mean it's, everyone's got a strange story coming into industrial Gases, I would say no one's going to school and saying, I want to be an industrial gas business. I mean, you know, there's, and there's a wide mixture of engineers and chemical engineers that kind of seem to end up on a project that may involve some gases and get interested in, you know, production methods or metrology or some element to do with this supply chain, and then they stick around for a long time. The amount of people I've met in my career that have been involved this for 20, 30, 40 years is amazing. People are not in and out of this. They come in and they stay. Don't ask me why, I'm not sure. But that seems to be a kind of universal.
- [75:57]
A
I have to ask you why. What's the appeal? We got, we got a lot of kids listening. Why should they get into industrial gases? Carl?
- [76:04]
C
Well, as I said before, I think if you, if you start to understand what these materials are required for, the, the scope of the different chemistries underneath this and the amount of different projects that you can work on is, is, is almost limitless. You're at the frontier of everything. If, if you want to look at it that way, you know, if these chemistries don't fulfill the requirements of the next device that's being manufactured, there's no device. So you can look at it as a, well, it's just a gas, right? It's a gas, is a gas. Who's interested in that? But if you want to frame it in terms of enabling everything, then you can, right? You just need a bit of imagination of how you talk about this industry, this business. If it's logistics of moving a cylinder around the world, yawn, right? But what's in that cylinder? How's it been made? What's it going to get used for and by who, and how has that kind of all happened and where's it going to go in the future? You know, from my point of view, that's pretty interesting stuff to be working on. So that's what keeps me going. It's okay. There's challenges of taming these materials. There's challenges of developing new ones. The customers that you're selling into are changing all the time. Nothing's still in this business. Nothing is the same from day to day, but on the outside it looks pretty boring, right? This is like, wow, this is. I see a gas truck going up the highway now and again. I'm not sure what's in it, but whatever. But when you get on the inside, it paints a different picture.
- [77:57]
A
Carl, thank you so much for giving us all this insight as well as an inspirational closure. I did not think that we were gonna clo. We were gonna end this podcast with.
- [78:11]
C
With a.
- [78:11]
A
With a bit of industrial gas romance, but here we are. So thank you so much for being a part of Chinatalk.
- [78:19]
C
My pleasure.