An interview with Matthew Putman, the CEO of Nanotronics, on how to fight COVID-19 by installing far-UVC disinfection lamps in indoor spaces. Edited for clarity and length.
Alyssa Vance:
Hey Matthew. So to start off with, could you tell us a bit about your background and what your experience is in technology?
Matthew Putman:
My experience in technology goes back probably as far as I can remember. I’ve been active since I was about eight years old, when my father opened a business that put personal computers on factory floors, to replace what were analog devices. At the time, personal computers were just getting popular in homes. So he was putting them onto factory floors. And then we started making instrumentation to link to them. I had this sort of digital day job, or night job, after school, being able to work with these instruments, as we started building a garage business.
I learned how to run specific gravity tests, I learned how to do these statistical calculations, I learned a little bit about computers at a very young age. In high school and college, I got interested in music and the arts, and that really came back in my 20’s, while staying interested in science especially. But I was watching the family business get bigger. I ended up being very involved with that company, and working in a university lab eventually, and started a lab doing nanotechnology research. Trying to create some kind of utopian dream that I had as a kid, from watching Star Trek when I was very young to reading Eric Drexler when I was in high school.
I was thinking that technology should be moving into these areas that make life radically better. I started a company called Nanotronics, which I’m the CEO of now. Where the whole idea is to scale technologies that are difficult to scale, or too expensive to scale. And to have something like a new type of foundational, exponential growth in technology, rather than just growth in applications. We were speaking about Moore’s Law; I think it’s needed to develop infrastructural technology, foundational technology, not just those things that ride above it. Eventually those things that ride above it start to become mundane, and it’s one more step extracting us from anything physical.
Alyssa Vance:
Could you talk a little bit about what a UV light is and how it might work to kill viruses, such as the COVID-19 virus?
Matthew Putman:
This is something that’s really interested me for a long time, long before COVID. I remember always hearing negative things about UV, and rightly so. You wear sunscreen to keep UV from damaging your skin and creating cancer. But at the same time, it works as a disinfectant. People started exploring these different wavelengths of ultraviolet light. As you get to smaller wavelengths of light, you get beyond the visible spectrum and into the ultraviolet spectrum; as you move toward shorter wavelengths, you get UVA, then B, then C. All of the things we’re talking about are in what we would call the UVC range.
Now UVC has been deployed, not just in LEDs but in mercury bulbs, for disinfecting surfaces for many years. Mercury bulbs that are around the 254 nanometer wavelength are turned on in operating rooms before patients and doctors are there, to disinfect surfaces. That same wavelength now is being used in solid state devices, as LEDs.
Alyssa Vance:
This is what YouTube called the “flesh burning death lamps”.
Matthew Putman:
Yeah. That’s one way to put it, I suppose. It’s flesh burning, but it was also a bacterial and viral disinfectant, too. But that’s the accessible, fairly inexpensive to make version of UV LEDs. You really don’t want to look at it, it’ll damage your eyes. You don’t want it to hit your skin, it will damage your skin. It’s only used for surface disinfection. What I’m more interested in is a farther, deeper UVC which is in the 220, maybe under 230 nanometer range.
Alyssa Vance:
That’s a shorter wavelength?
Matthew Putman:
Yes, it’s a shorter wavelength. Farther into the UV spectrum. When you get into that part of the spectrum, you get some incredible advantages. It isn’t the first thing, what YouTube calls … What did you say? Death?
Alyssa Vance:
Flesh burning death lamp.
Matthew Putman:
Right. It’s no longer a flesh burning death lamp, it’s something that will not damage the skin or the eyes. More importantly for COVID or other airborne pathogens, it’s not designed strictly for disinfecting surfaces. The more research we see, the more that comes out, we see that the aerosols, the airborne nature of COVID are much more dangerous than what exists on surfaces. The big goal would be to produce this wavelength of UV LEDs and to have them in our normal lighting systems, so we can start getting back to life and feel safe.
Alyssa Vance:
These are LEDs? These are the semiconductor, light emitting diodes that you find in electronics and so on?
Matthew Putman:
Exactly. No different at all. You’ll remember that there have been challenges with those over the years. Somebody won the Nobel Prize for inventing the blue LED, for instance. It took a lot of time to get to a point where you could produce a good blue LED. But LEDs themselves are so ubiquitous in our life right now that we know how to make them, and we use them everywhere. The special thing about this particular wavelength is that it requires a different type of material. They’re usually aluminum nitride. There are some other possibilities, like aluminum oxide, or gallium oxide, for instance, that could do it. But aluminum nitride is the most well tested and the most used in very small environments.
But it presents some challenges to make, it’s expensive to make. However, the LED itself is something anyone would completely recognize. You would plug it in the same way.
Alyssa Vance:
LED technology has really come very far. When I was a kid, 15 years ago, I was wondering, could you make a laser gun like they have in video games? And I concluded that no, you can’t, because there’s no way you could get enough power into that small of a space, that someone could carry around with them. But there’s a guy on YouTube, Styropyro, who’s now actually done that. The LEDs have gotten so good that he took a radar gun and converted it into a laser gun, which puts out about 100 watts of light. That’s actually pretty dangerous.
Matthew Putman:
Oh my God.
Alyssa Vance:
You could totally use that as a weapon.
Matthew Putman:
That’s funny. No, it’s true that so many people underestimate it. It sounds like you didn’t. And I didn’t, but not from any sense of knowing anything. I’ve always … I remember being at the World Science Festival in its first year, oh God, this must have been 15 years ago. I was sitting at a table next to a Nobel Laureate physicist, so somebody much more educated, much smarter than me. We were talking about vertical farming. I was incredibly interested in vertical farming. I still am, actually. He was saying that this wasn’t a feasible thing to scale, because you couldn’t have LEDs that were strong enough, that had enough power. Now, that’s certainly not true.
It was one of those arguments where I said, “That’s not true.” But I had absolutely no good reason for why. I luckily turned out to be right. I know a little bit more about the production of LEDs now, I’ve spent a lot of time with that.
Alyssa Vance:
If you have these UV lights being used in everyday environments like homes and offices and transportation, obviously, they’d have to be much safer than current UVC lamps. What kind of safety testing has been done? Has someone tried putting them in a house or in a building, and been around them for eight hours a day for four or eight weeks, and found that there has been no skin burning or other ill effects?
Matthew Putman:
Yes. We can put links in your blog to some of the research. I’m not going to be able to quote too many names off the top of my head, but this isn’t a new technology. There was recently some studies done out of Columbia University that have gotten a lot of attention lately, in the last month. But this has existed for a long time, and there have been trials. The safety and efficacy of it isn’t the thing that’s in question.
Safety has been proven out, and it hasn’t been proven out just last week. This is a rather old technology by now. It just takes commitment, a small amount of investment, and the convergence of some new technologies. Some things that I’m involved with, with using closed loop AI systems, using deep reinforcement learning to be able to make corrections in the manufacturing process. Which is where my technology comes in, where things I’m involved with come in. But the thing about the safety issue is, there’s no reason anybody should take my word for it. It’s a very well known thing if you get to that wavelength.
Alyssa Vance:
What does the availability of these devices look like right now? Can you go on Amazon and buy one? Are they being produced on a large scale? Could a business go to Alibaba, and order 10,000 of them for their office?
Matthew Putman:
There are a few companies in the United States, but not for any reasonable price, and not in stock. Any single device that you would buy would be, I don’t know, somewhere around $1,000. This is not something where most people could go and spend a million dollars to take care of an entire hospital, for instance. That’s not due to the material. The material is aluminum nitride, the two most abundant things on our planet – aluminum and nitrogen are close to the top. But it’s very hard to process.
They are available, but probably not available in the quantities where, if money disappeared tomorrow, there would be enough to go around. There’s some good news here, though, it’s made with the same type of reactors that other LEDs are made with. Companies like Cree, companies like Samsung, very large companies that make LEDs could use their reactors to make these types of safe UV lamps. The production capacity is there, the supply right now is not. Because the price is so high, it doesn’t make a great business offering at this point.
Alyssa Vance:
In order to bring the price down, do we need a new technological breakthrough? Or is it just a question of building the factory at a large enough scale to make each unit cheap?
Matthew Putman:
It’s a new technology. But the technology, strangely, it’s a software technology. This is where the work that I do comes in. (When I say “I”, I mean people on my team that actually know how to write these models. If we get too deeply technical about this, you’ll lose me.) But the main point is that they’re grown in something called a MOCVD reactor, which is a vapor deposition reactor. These are very common reactors, as I said, used for any wide-bandgap material. LEDs, but also power inverters and sensors, a lot of other things that are wide-bandgap semiconductors.
You can think of this as an additive process. You’re taking a crystal, you’re growing a crystal on top of another crystal with the gases in the deposition chamber. The problem with using aluminum nitride in the type of volume that we’re talking about is that during that process, cracks and other types of defects start to appear in the crystal. Now that’s not new, that’s something that’s happened through the history of semiconductors, having defects that propagate. But this has to be grown at extremely high temperatures, and it’s very touchy.
Getting over those hurdles has been why yields have been especially low. If you have something that’s 50% yield, then you’re losing a lot of material and have to charge a lot more. We use something that we call AIPC, artificial intelligence process control. Instead of setting conditions for your tests – normally you’d set how much of the gases you want, what the temperatures should be, what the flow rates should be – we take all of the controllers that already exist on the reactors, but we let something called an RL agent, a reinforcement learning agent, watch the process.
Because it’s an additive process, when a defect starts to emerge, it can make corrections in the next part of the process in order to fix it. It’s analogous to, and uses a lot of the same types of algorithms as AlphaGo would use, to win the game of making an ideal UVC LED. Regardless of what moves were made in the middle. We assume that there will be problems. This is hard to do. But an RL agent that’s being rewarded for getting it right, and penalized for mistakes along the way, can take actions that should win the game and improve the yields.
Alyssa Vance:
How can you tell when a microscopic defect starts appearing in one of these materials?
Matthew Putman:
There’s a whole other set of tools that can be used. You’re taking sensor data all the time. There’s time series data, we’re monitoring things as they go, so you see how things change. You might see some strange dips in temperature, or some strange loss of flow. But you have these validating steps, and those could be using spectroscopy, or using microscopy. Then there’s a whole other set of AI tools that go with that, which are a bit simpler. Using things like convolutional neural networks to do classification of the defects in real time. You’re doing imaging, whether it’s through optical or other means – see a certain type of defect, notice it’s a defect, put it into a class. That system then, due to the memory being created in a series of AI operators that we run, knows historically the best ways to correct those classes of defects.
Alyssa Vance:
As we’ve seen with everything in this pandemic, even relatively simple manufacturing becomes complicated when you have to do things at an enormous scale, for all of the seven billion people who might become infected. Derek Lowe, over at In the Pipeline, recently had an article on the glass vials that are used to contain vaccines – you make the vaccine, you put it in the vial, then you ship it off to the doctor or hospital or whoever. Right now, even if there were a vaccine tomorrow, there aren’t enough glass vials to put them in. Because even though making glass vials is a very established technology, no one has a factory that’s large enough to pump out seven billion of them, or however many we’re going to need. If we did try to scale up this LED manufacturing process, so it could cover every office and factory, and stop the pandemic, what sort of bottlenecks would you run into?
Matthew Putman:
Well, this goes beyond UVC, as you mentioned. What’s been noticed, what’s been really apparent throughout the COVID crisis, is what a disaster our supply chains are. We have taken even the most simple of devices, and created a supply chain that slows things down. And it adds potential variation in the process, potential failure points, and a length of time to get things to market that is completely unnecessary with modern technology, such as the technology I mentioned before.
To build what you need, where you need it, is something that needs to be taken into account, rather than a stockpile of potentially useful things. Distributed factories… there are certainly different steps of how this could be done. You could think about converting a regular LED factory that’s already a large factory, they already have the right machines, then add on the software. There’s plenty of aluminum and nitrogen in the world, and you can make it in North Carolina. That’s one way that’s probably the simpler way to go about it, rather than to take the 50 other steps that would normally go into getting an LED.
COVID is a wonderful chance, if we don’t squander it, to try to finally converge, and make these things that people thought would need to happen one day happen now. I used to say when we started Nanotronics, that my dream was to be the next big, important company to be built in a dorm room, that would not be a social network, but a factory. I believe that there could be big steps towards this in just this particular technology. An MOCVD reactor is a couple hundred thousand dollars. Certainly not free. The software’s another couple hundred. But for a fairly small amount of money, you could actually have a factory that is close to the size of a dorm room making quite a bit of UVC LEDs. Now that’s not going to happen tomorrow. It could, though.
You have to first of all ask, “What is physically possible?” “What is proven out to be an engineering possibility?” That could happen. What’s more likely to happen, and should happen, is that all of those reactors in the world that are making other LEDs could make the UVC LEDs, and could do it in a distributed way. New factories should be funded, new companies should be funded. It’s an underserved space right now that has the potential to be ubiquitous. There’s a great opportunity.
We even get discouraged by the fact that there isn’t anything in our lives that we can imagine making, without having 100 different companies and suppliers build it.
Alyssa Vance:
If all of the factories that are using these reactors to make light bulbs and electronics, and all sorts of other components, switched to using them to make UVC LEDs, that would be enough to supply everyone?
Matthew Putman:
For sure. I mean, it would be a fraction of that. The same reactors build 15 different chips, in the chip set of an iPhone. You wouldn’t expect them to convert everything, but you could make profitable businesses out of each of those reactors, and then buy more reactors. Certainly, the capacity is out there right now.
Alyssa Vance:
That’s what they did during World War Two.
Matthew Putman:
Yeah.
Alyssa Vance:
A lot of consumer production simply stopped, and they said, “For the duration of the war, we will make no cars, we will make no bicycles, we will make no typewriters, we switched it all over to war materiel.”
Matthew Putman:
Right. I’m trying to figure out if that was a different will of the people, an ability to sacrifice to fight, or knowing that it was something different to fight for than we have right now. Or if people are intimidated by the technology that we’re discussing. I did an interview with Stephen Wolfram a couple weeks ago, on a podcast that I have. At the end of the podcast, he started asking about AIPC and Nanotronics – he was saying that an iPhone, isn’t it so complicated that even with all of the money that Apple has, if they wanted to make an entire iPhone in the United States or in one factory, it couldn’t be done?
I challenged that idea. I think that if Apple took a war-like switching things over mentality, like you were talking about in World War Two, it’s not a technological challenge. It’s just being able to say yes, this is important and should be done. But it’s talked about so little that some of the smartest people in the world – like Wolfram who I respect, I think he’s one of the most inspiring, intelligent people around – didn’t know that.
Alyssa Vance:
I read on your Medium post that you tried to do a pilot project with the MTA (which runs the New York subways), but it hadn’t gotten launched yet. The MTA, in transit circles, is famous for being this big government agency that overspends, and has projects come in years late, and so on. If you think about doing a pilot project with some other company or agency, the first thing that comes to mind is those meat packing plants. Because they’re big centers for infection, they have lots of people packed close together. It’s hard to shut them down, because then no one could eat meat. And you can’t work from home, so there isn’t much alternative.
Matthew Putman:
Excellent idea. I love this conversation, because yeah, I’ve thought about partnering with airports, I’ve thought about partnering with clinics… meat packing is an excellent idea. And you’re also right that when you think about big government – every prejudice I had had dealing with big government, as being difficult, bureaucratic, overspending… every prejudice I had turned out to be true. I think, certainly, the MTA isn’t alone. But yeah, that’s a great idea. I will start searching that out, thank you.
Alyssa Vance:
Thank you.
Matthew Putman:
For the biz dev opportunity.
Alyssa Vance:
Back when the pandemic was getting rolling during February and March, I had in my head that… Trump wanted to deny it for a long time, but I thought that the pandemic would be so bad, there’d be so many deaths, so many hospitalizations, that he’d essentially be forced to let someone like Scott Gottlieb come in and establish a sane policy. I don’t think that’s happened yet, and I don’t see any good prospects for it happening soon. Have you thought about going outside the US and working with different national governments, like Taiwan and New Zealand and other countries that have handled it better?
Matthew Putman:
Yes, yes I have. And there’s no reason to be nationalistic about saving peoples’ lives. Whoever will do it and support it, we should be working with. Yes, I’m speaking to other governments and large companies in other countries. I feel like it’s a shame and a wasted opportunity if the United States continues to have this embarrassing, heart-breaking, frustrating rise in infections. Policy and investment throughout many institutions has been very poor, and continues to be. So yes, I am speaking to other places. This isn’t a war against another nation we’re dealing with, so we all benefit from it.
Alyssa Vance:
Where should someone go if they want to read more about this technology? Maybe there’s someone who has worked with LEDs in the past, who might be interested in getting one of these factories set up?
Matthew Putman:
In the blog post that I wrote, I have this link to reference this new work coming out of Columbia. But there’s three different US-based companies that I’ve been working with that do the technology. There’s three university programs I work with. Yeah, there’s research that goes back quite a ways. From Santa Barbara to Korea. So this is work that has been done quite a bit. We’re publishing a paper with a company in North Carolina where we did some pilot demonstrations, too.
Alyssa Vance:
You said that you’ve been – in general, not just for COVID – excited about manufacturing, and building new manufacturing companies, in the same way that we’ve had Facebook and Twitter. What are some of the opportunities in manufacturing that you’re most excited about?
Matthew Putman:
Yeah. I do want to point out that I’ve gone through these waves of being incredibly optimistic that the urgency of COVID will take some of the technologies that have been waiting, and push them out into the world – being very optimistic about it to being disappointed, like with this government story about UVC. I think there are tentacles coming out of all of these things. If you think of wide-bandgap materials, once you’re making UVC LEDs in this way, you’re also making new photovoltaics in this way, or you’re making new types of packaged power devices.
But for COVID specifically, we’re working with the largest maker of genome sequencing equipment. Then throughout their supply chain, we’re able to increase yields of genome sequencing to the point where it becomes extremely inexpensive to do – and then you can design not just vaccines, but therapeutics, and eventually individually tailored ones. The road to personalized medicine can be shortened by decades, but the will and the financing go into it because of COVID. That’s something that really excited me. I consider that manufacturing, and all these things are building something.
Alyssa Vance:
So that’s working with gene sequencers?
Matthew Putman:
Yeah. And their supply chain, the vaccine makers that they work with. To improve yields, make things less expensive. For vaccine research, and from testing through vaccine research and therapeutics, especially when you’re dealing with these RNA vaccine delivery systems.
Alyssa Vance:
If you look at light manufacturing – the types of things that you’d find in a Walmart, like clothes and electronics – we’ve seen huge decreases in price, and increases in efficiency. Most of the things in a Walmart are cheap. Even an ordinary person can buy most of them with an hour of labor, two hours of labor. And many people now have too much stuff, rather than a shortage of stuff. But a lot of things in heavier manufacturing remain very expensive. Like airplanes are obviously very expensive, skyscrapers are expensive, bridges are expensive. These things still cost hundreds of millions of dollars. Do you have any thoughts about how you might use technology to make these large manufacturing systems more affordable?
Matthew Putman:
You point this out, the commodification of these things – there is real ubiquity and affordability of these things, and when I was a kid, there wouldn’t have been. None of those things are novel – so you take an airplane that’s just like an airplane that we currently have, but make it lighter and more fuel efficient, that’s generally just improving material manufacturing. It’s easier to improve material, injection molding of plastics. The same thing can be true for doing a layup of carbon fiber, or to actually incorporate new types of nano fillers. That’s not where it gets incredibly exciting, though.
New battery technology is very interesting and exciting, including solid state batteries. Making things less expensive, I think that you get this and you’ve already pointed it out, it’s not always linked in people’s minds to… lower prices with things that are radically better and exponentially better, yeah. I don’t think it’s exactly intuitive to people, even though it should be. Being that we don’t see it.
The way that we can build things, by using techniques like I discussed earlier, is by reducing the length of these supply chains. By being able to take the software that was developed on other platforms, on the platforms that we need to improve. Funny enough, you can take an NVIDIA GPU, where deep learning models have been created and are being deployed, and use those to make the models that were created for the application, and use those to make the next generation chips. Then you’re going to start seeing something that’s really, radically different. The type of change that we would expect and want.
Alyssa Vance:
In terms of novel technology, what are some things that you might buy that aren’t being manufactured, or aren’t available on the market yet? What do you imagine in the glorious transhumanist future?
Matthew Putman:
It’s funny because we can now look at our daily lives, and see where the same failures have occurred, and use our imaginations in the way that probably you and I always have been. We see the failures in the medical establishment. We don’t have any type of personalized medicine, and that could be either completely biochemically based, but more usually with some type of nano-controllers that can seek out pathogens or can bind to different DNA sites. The idea of having telemedicine needs to turn from talking to a doctor over Zoom, to looking at your own genome and printing out a therapeutic for you right at that time, and just having your doctor consult with you if you have any questions.
You know, that’s basic. These are things that we would all love right now, and are much more possible than people think. There will be something fairly amazing, if any of these new types of vaccines work in a fairly short amount of time. I think that there will be a renewed hope that things are possible. Anyway, that’s one of them. Not to be old fashioned about this, but I do think that, in nanotechnology, it’s still where you gather the greatest amount of abundance, and abundance meaning abundance of life as well. I focus on miniaturization, not just of a chip, but a miniaturization of a factory. Then I leave it to your imagination. And everybody else’s imagination to build what we want to build.
Alyssa Vance:
I just ordered a whole genome sequencing, I sent off the tube last week. When I get the data back, do you have any ideas on what I should do with it?
Matthew Putman:
I have a feeling that actually you, personally, and a group of friends could figure out how to combine that with your own medical records and write some AI algorithms around it. But I don’t know. That’s really problematic, that I even hesitated, right? Isn’t that crazy that you even asked the question, and that I don’t have an answer to that? If I would’ve thought that you and I can just get a full genome sequence as easily as we can, and we’re not sure what to do with it? It’s amazing to me. I have no great insight.
Alyssa Vance:
I can do all kinds of studies, but I need to have more than one. Like, one genotype and one phenotype by itself doesn’t get you very much.
Matthew Putman:
I know. I mean, that’s why I’m an old fashioned factory guy who just wants to make genome sequencing so cheap that everybody has their full genome sequence, and then you’ll have plenty of data and everything to work with. I’ll focus on that goal. Because this is just like building a computer and having applications on top of it. There’ll be so many people with applications. Let’s make sure that it’s nearly free for everybody to get their genome sequence.
Alyssa Vance:
Cool. That’s all the questions I had, is there anything else that you want to add?
Matthew Putman:
No, thank you. Was this okay for you?
Alyssa Vance:
Yes. This was awesome, thanks for doing this.
The Rationalist Conspiracy 