Electrolyze everything!

Volts

A startup called Mattiq is using a combination of nanotechnology, AI, and electrolysis to produce novel materials that could eventually substitute for carbon-intensive materials in sectors from chemicals to plastics to fuels. In this episode, CEO Jeff Erhardt and I dig into the technological and business details.

(PDF transcript)

(Active transcript)

Text transcript:

David Roberts

By now, most energy nerds agree that a) we're going to need a bunch of hydrogen to complete the clean-energy transition, but b) the vast bulk of today's hydrogen is made using fossil fuels, ergo c) we need a rapid, wholesale shift to the only truly clean way to make hydrogen, which is electrolysis.

What is electrolysis, though? Have you ever really thought about it? Because folks … I've been thinking about it.

Subscribe now

Electrolysis is a form of electrochemistry, which is the science of using electricity to cause or control chemical reactions. In the case of hydrogen, an electrolyzer uses electricity to split water into its constituent parts, hydrogen and oxygen. If the electricity is carbon-free, then so is the resulting “green” hydrogen.

But especially attentive Volts listeners will have noticed that electrolysis keeps popping up. It is central to the story of how Sublime makes carbon-free concrete. It is central to the story of how Boston Metal makes carbon-free steel. It is central to the story of how Aqua Metals recovers materials in battery recycling.

Jeff Erhardt

It seems that electrolysis can do more than just split hydrogen from oxygen. It can, if tuned properly and given the right feedstocks, split just about anything from just about anything. It can even produce entirely novel materials that help solve problems in plastics, chemicals, and fuels.

Share

At least that is the value thesis of a startup called Mattiq, which was co-founded by two researchers at Northwestern University: Andrey Ivankin, a world-leading expert in nanofabrication technology, and Chad Mirkin, a nanoscience pioneer who just this month won the prestigious Kavli Prize in Nanoscience for his work on spherical nucleic acids (SNAs).

I'm going to talk with current Mattiq CEO and director Jeff Erhardt about the technologies that enable Mattiq to develop new materials, what kinds of problems those materials might solve, and whether electrifying everything, via electrolysis, might just win after all, in which case I get to say ha ha I told you so.

With no further ado, Jeff Erhardt of Mattiq, welcome to Volts. Thank you so much for coming.

Jeff Erhardt

Thank you, David. That was a fantastic introduction. Glad to be here.

David Roberts

Jeff, I'm going into this conversation, relative to my normal pods, I would say, extremely intrigued, but somewhat under-informed, since the public information available about Mattiq is quite limited and it's somewhat opaque exactly what you're doing. So, I want to get into that. But I think the root of it is you're designing new materials. I mean, that is the — tell me the elevator pitch for what you're doing.

Jeff Erhardt

You teed it up very well in your introduction, which is, if we look at the big picture, you mentioned two out of the three what people think of as the so-called "hard to abate" sectors: steel, cement, and then the big third one being the production of chemicals and fuels.

David Roberts

Yes.

Jeff Erhardt

And broadly speaking, there's a golden opportunity to, as you've said many times, electrify everything. And there's absolutely the potential to do that within the third pillar that you mentioned, which is the production of those chemicals and fuels. If we take a step back and look at or think about what's really going on, what we're really doing is we're just using a different form of energy. Energy is required to drive any material transformation. And those material transformations or the energy for those historically has come from heat, and it's come from heat that's come from burning things.

And so, with all of those processes that you've mentioned, both in steel and cement and now in chemicals, there is the opportunity to instead use electricity as the energy. Specifically, within the world of chemicals and fuels, we think of, or we sometimes say, using electrons as the reagent to drive that reaction. So, what we think of writ large is the opportunity to decarbonize the production of chemicals and fuels using electricity under this umbrella of going from heat powered by fossil fuel to clean electrons to be able to drive those reactions. So, the question becomes, where has that been proven?

How do you do that, and how do you make it work across a broad array of applications? That is effectively what we are solving.

David Roberts

Right. So, you're using electrolyzers to make new materials, to substitute for materials that have traditionally been made with fossil fuels. That's the root here.

Jeff Erhardt

That is the root here. That's exactly right.

David Roberts

Right. So, I want to talk about all the pieces of that. So, there's the nanoscience piece, there's the artificial intelligence piece, and there's the electrolyzer piece.

Jeff Erhardt

Yep.

David Roberts

And I want to talk about all those. But maybe let's start with nanoscience, since your founder — by the way, congratulations — just won that incredibly, incredibly prestigious — it's like the equivalent of the Nobel, basically, for this area — for his work. And he, Mirkin, I mean — listeners should go look him up, he's a remarkable dude — he's generated new materials that are useful in bioscience and medicine and on and on and on. But he also is using nanoscience for clean energy, God bless him. So talk about the nanoscience piece of this.

Jeff Erhardt

Yeah. So, if we take a step back, and again, thank you for that. And I think we, as a company and myself, are incredibly lucky to be associated with people like that and the science that they bring to the world, in particular, that cross-disciplinary science.

David Roberts

Right.

Jeff Erhardt

So, the way I would tee up this discussion, I'd maybe start at the high level, which is to talk about these electrolyzers. Your listeners, and you've done quite a bit of good reporting and conversations around the use of electrolyzers for so-called "green" hydrogen production, that is, using that technology. Sometimes I think of it as a magic black box to split water into hydrogen and oxygen. And how does that work? So, you've got this magic black box. It's got two plates called an anode, a cathode, but in the middle, it's got a membrane. And on that membrane, in the case of green hydrogen production, is a catalyst that is a unique material that drives or transforms a chemical reaction.

David Roberts

When exposed to electricity?

Jeff Erhardt

Exactly. And in the case of hydrogen production, that is a very, very rare earth element called iridium. Extremely rare. And especially as the production or the growth of the green hydrogen market has exploded over the past couple years, the cost of that iridium has exploded with it.

David Roberts

And so, there's just something unique about iridium that creates this very specific chemical reaction that busts hydrogen and oxygen apart from one another.

Jeff Erhardt

That's exactly right. It's kind of this magic material. So, you could then ask the question, "Well, gee, can I identify, using those things that you talked about, using nanotechnology and AI, in the context of these electrolyzer systems, start to identify alternatives for that very rare and somewhat magical material?" And then, if I'm able to do that, can I take the learning that I've done to identify those alternative materials in the context of these very complex electrochemical systems? And can I redesign and redeploy them? Can I change the catalyst material? Can I change the membrane? Can I change the configuration?

Can I change the input instead of water? Can I have different feedstock inputs to then be able to open up the clean production of this vast array of chemicals using this clean electricity instead of the traditional fossil fuel heat that you kicked off our show with? And so, that was exactly the inspiration for what we've done. And if we go backwards in time, it was sort of Chad's idea and inspiration from following and driving innovation and discovery in that biotechnology world, where, if you think of how that industry has been transformed over the past 20 years, what did they do?

David Roberts

They charted the genome.

Jeff Erhardt

Exactly. They went from one-at-a-time experimentation by really smart people, and taking ten plus years to bring a new pharmaceutical to market, to doing exactly what you said, charting or mapping the genome, and bringing in this idea of the language that I use, thinking of massively parallel synthesis of, in this case, proteins on a chip, to be able to drive rapid discovery and accelerate the pipeline for how you bring drugs to market. And it was exactly that inspiration that led Chad to say, "Well, what if I could do the same thing? But instead of for proteins on a chip, do that for inorganic nanomaterials?"

David Roberts

I have to pause you here, because proteins on a chip is — I need a little bit more on that. How do you put proteins on a chip or anything?

Jeff Erhardt

Yeah, so certainly I'm not an expert in that, since I'm not a biologist.

David Roberts

That's the nanoscience that we're talking about.

Jeff Erhardt

That's the nanoscience side of it. What I will point you to, and point your readers to, which is very timely. Just this week, there was an article on the front page of the New York Times, by Steve Lohr, talking about the latest advancement in this. And the title of that article was something along the lines of how AI is revolutionizing drug discovery. It was a story of a company in California called Terray. The quote that they talk about in there and that they refer to is this idea that they have created millions of individual wells on a chip, that they can characterize these interactions and then be able to feed an AI system, to be able to drive their drug development.

So, in terms of the exact technology, how they do that, that's not my area of expertise. But you can read and think about that same concept of taking a small chip, creating little miniaturized versions of the end systems that you're trying to experiment around, using it to generate massive amounts of data, using that to feed AI systems, and then eventually putting it through a development pipeline to radically transform something. What companies like that are doing is using it to transform drug discovery. What we have done is exactly the same thing, but using it to transform the world of materials discovery, that is, materials that can act as catalysts to drive a chemical reaction, then change and really develop and accelerate a pipeline, not for the development of new drugs, but for the development of clean chemicals.

David Roberts

So, on the chip, then would just be tiny, tiny little samples of different kinds of materials, like new combinations of molecules.

Jeff Erhardt

That's exactly right.

David Roberts

And then, so what are you doing? You're putting all those on a chip, and then the AI — what is the AI doing with all those materials? Basically, just like, analyzing their characteristics?

Jeff Erhardt

Yeah. So, what we are doing is exactly what you said, David. We have the ability on a chip the size of your thumbnail to synthesize, to create, to combine different elements in arbitrary combinations into unique, discrete nanoparticles. So, imagine taking three elements that maybe have never been combined before. I'll make something up.

David Roberts

Right. I was going to say, some of these, at least, are materials that have not previously existed in the world. They are genuinely new combinations of molecules, basically.

Jeff Erhardt

They are new combinations. That's exactly right. We are not synthesizing new elements per se, but what we are doing is we are combining them in unique ways. And what's fascinating about the world of nanotechnology and the physics that surround it is that when different elements or different atoms come together in different combinations, their behavior changes. When you make them small or when you make them thin, their behavior changes. So that's the foundation or the fundamental sort of aha moment that we're exploiting, which is if you were to try to do this experimentation and combine new combinations of elements, as an individual scientist, to look at and think about all the possible combinations on the periodic table, combining them in two elements together, three elements together, four elements together, the search space becomes so big, you can't do it.

David Roberts

Yeah, it verges on infinity, it seems like the number of combinations.

Jeff Erhardt

It verges on infinity. But what we now have the ability to do is to do that in millions or hundreds of millions of different combinations on a single chip all at once.

David Roberts

So, maybe the piece that I don't fully grasp is, once all these materials are on the chip, presumably very small quantities of these millions of materials on this chip, we want to analyze those materials to determine their characteristics, like how well they hold up to heat or, you know, whatever.

Jeff Erhardt

That's right.

David Roberts

How do you do that without physical testing? Like, what is the AI doing to the chip that allows it to figure out what their characteristics are?

Jeff Erhardt

Great question. Well, let's be very clear. You can't do it without physical testing. So, that's the second piece of the magic that we've developed. I like to use the word that we have mimicked and miniaturized the real-world systems into which these materials are eventually integrated to physically characterize their behavior. So, in our case, and we kicked this off in our broad discussion here, is thinking about electrochemistry and electrolyzers. So, you can think of what we have developed as a company is a miniaturized or a nano electrolyzer to be able to collect information about how those materials drive and behave around different chemical reactions.

David Roberts

So how long does it take to feed millions of different material samples through a tiny little nano electrolyzer? And what on earth does that process look like?

Jeff Erhardt

Yeah, so what that process looks like: So now, imagine having this chip with these nanomaterials on it and this miniaturized version of an electrolyzer, and you can think about it scanning or rastering across this chip and collecting information, collecting information around how active or how effective is this material in driving a particular reaction. How selective is it? How can you point that chemical reaction towards an output A versus an output B? And you care about how durable or how reliable is it over time? And we collect all of those things. And so to do this across millions of different combinations and candidates on one of these chips takes us on the order of magnitude one day.

David Roberts

So, on a nanoscale, there are actual particles going into an actual tiny little electrolyzer.

Jeff Erhardt

Yep.

David Roberts

And being electrolyzed, super small, super fast.

Jeff Erhardt

That's right.

David Roberts

That's... I mean, I know nanoscience is mind-blowing, but that's just —

Jeff Erhardt

It's just mind-blowing. And so think about, use the analogy to the DNA, and think about how long it used to take companies to do DNA sequencing, how expensive that was.

David Roberts

Right.

Jeff Erhardt

It was millions or hundreds of thousands of dollars. It took weeks. We're now able to do one of these experiments in a day, costing hundreds of dollars.

David Roberts

Yeah.

Jeff Erhardt

And what that allows us the ability to do then — so you asked, are we doing physical testing? The answer is yes, we do that physical testing for what we call the functional behavior of these materials to drive electrochemical reactions. But we also do structural testing to understand what is their atomic structure, what is the actual element, are they oxides? Are they pure metallic, etcetera? And then what we do is we aggregate that all together to just like what was being written about in the world of drug discovery. We use that effectively as a data factory, to be able to feed AI, sort through that, understand and train that based on real data, pick out good candidates.

And then, here's the really important thing, David: There is always the question in nanotechnology or materials discovery, how do things behave at the small level, at the nano level in the lab, relative to how they —

David Roberts

Yes, this was my next question: Like, is the behavior of these materials at a nanoscale reliably an indicator of how they behave at a macro scale? Like, is that a one to one correlation, or is it a guess? Or like, what do we know about that?

Jeff Erhardt

Yeah, that's the most important question, and that's what we have then built and what we are doing as a company. So the really important thing to understand is exactly what you just asked. I come from a background, and my prior two companies were in the data analytics, machine learning, and AI spaces. And so I still have lots of friends there. I hear lots of interest, as you can imagine, in everything we hear about these days, is AI. AI for this, AI for everything. And so there's lots of people, including some of my friends, thinking about starting what is an AI-based materials discovery company.

And I think the challenge there is really twofold. First of all, just like we were talking about those AI systems that we hear about today, things like ChatGPT, those weren't algorithms that were created in a vacuum based upon some expert.

David Roberts

They need data.

Jeff Erhardt

They need data to feed them. And in the world of material science, people are trying to solve that problem and create the AI for these material systems on the data. So, that's problem number one.

David Roberts

All the history of human science, don't we have a lot, don't we already have a lot of data about materials and how they behave?

Jeff Erhardt

Relatively small. So, let me give you an interesting benchmark on that. So, one of the early studies that we did — we kicked off this discussion talking about hydrogen production and that very rare earth element called iridium that's used to drive that reaction — so when we were first launching the company, we wanted to go solve that problem because it's one of the bottlenecks for the production and scaling of green hydrogen. And so, to give you some numbers on that, we mined the literature, and over the course of the last 30 years, we were able to identify roughly 1000 different material combinations that scientists tried to study to identify a good alternative to iridium for the hydrogen production. And they couldn't really find anything.

David Roberts

Because previously, just to make this clear, previously, before you had the nanoscience and the AI, you would have to create those materials one at a time in a lab and then test them one at a time in a lab. So, the whole thing is time-consuming.

Jeff Erhardt

That's exactly right. So, think about 1,000 experiments over 30 years. Over a six-week period, we studied more than three orders of magnitude more than that to identify a gigantic portfolio and suite of combinations of elements that perform as well or better than iridium in that reaction.

David Roberts

So you're putting a bunch of sort of candidate replacements, right? You're looking for a material that behaves like iridium in all the relevant ways.

Jeff Erhardt

That's right.

David Roberts

You generate millions of potential iridium alternatives, put them on this chip, run them through this nano-electrolyzer, and the ones that show promise, then you do macro testing.

Jeff Erhardt

We do macro testing. We scale them up into so-called bulk materials, and then we integrate them into real-world, full-size electrolyzers. And let's link it back to the question you asked: It's not a guarantee that something at the nano level or at the AI prediction level behaves that way when you scale it up and you integrate it in a real device. Understanding that correlation, that pipeline, and using those large amounts of data across all those scales to understand why something behaves similarly at the nano level as it does in a real device, or it doesn't, is incredibly challenging, but also incredibly powerful.

And that's the foundation that then, in turn, gives us to not just study a catalyst at the nano level, or an algorithm or a hydrogen catalyst, but rather start to build an understanding of these complete and very complex electrochemical systems and open up a world of possibilities going far beyond hydrogen.

David Roberts

So, I'm guessing when you say these things produce information for the AI to chew over, all these little nano samples being fed through the little nano electrolyzer, that's the data that the AI is working with. And then presumably, you teach the AI about what works at the macro level. And presumably, over time, the AI is learning and generalizing. And so the next time you're aiming for a material, maybe you have a smaller target, like you're — the AI has a better sense of how to get started and like, what types of materials to use for that.

Jeff Erhardt

I couldn't have said it any better. That's exactly right. So, imagine a pipeline going from that very small nano level where we're collecting those miniature, those nano size electrolyzer data. We're collecting information about crystal structure, about elemental composition. We're understanding and collecting what happens when we scale those up.

David Roberts

AI is learning about physics, right? I mean, it's like learning about material, not even specific material or specific uses. It's just kind of learning about how materials behave. Right? I mean, in general.

Jeff Erhardt

That's exactly right. And then you can start to use that to do two things. You can start to be more efficient in how you optimize or develop a given set of materials or chemical reactions. But then you can also start to extrapolate and think about, how can I use what I learned in one reaction space and use it to start, to get a jump start on how I'm going to develop and optimize a new reaction.

David Roberts

Give us a sense of the timescale there. Going after iridium is one of the first things you're doing. I take it as a test, a demonstration of what you're capable of. And I do sort of like the, I mean, there's something a little poetic about electrolyzing materials that will in turn make electrolyzers better. You're sort of like leveraging your electrolyzer up. But, how long did that take to — because you announced a few months ago you found a suite of materials that are plausible iridium substitutes that are cheaper and more abundant, et cetera, et cetera.

How long did that take and sort of how much do you envision that time being compacted as you learn more?

Jeff Erhardt

Yeah, so again, we can use analogies to the world of drug development and think about how long it has historically taken to bring a new treatment, a new drug, or pharmaceutical to the marketplace. And then think about what happened, for example, with the COVID vaccines, and how much more rapidly that development process was able to be done, given all of that historical work that was collected. People didn't know what COVID was, but they knew what other viruses and the associated vaccines were. And then look at how fast those were able to be brought to the marketplace.

David Roberts

Was AI involved in that? I never really looked into it.

Jeff Erhardt

Yeah, I'm not an expert to say whether AI, as we would define it today, was involved, but certainly large-scale databases of historical information to be able to apply math on top of to be able to accelerate that process was absolutely done. So, what I don't want to do is upset somebody by saying this was or was not AI, since I was not involved in it. But yes, there was absolutely a data plus math-driven acceleration process. And so, you can bring that exact same thing here. So, think about the way that new chemicals are brought to market today, one at a time experimentation, just like you said, slow, very challenging industrial qualification processes.

And we're going through that right now. So, as you noted, we announced that suite of materials that we had identified and proven out in our facilities earlier this year. What we've been working on now is working with our industrial partners, the existing players within that space, who are experts at large-scale manufacturing and distribution of those to go through. And let's come back to the understanding cross-scale correlations. What we now need to do is understand how things scale from our labs to their manufacturing processes, to qualify them with end customers and drive them into the marketplace.

But think about the analogy of what we're doing. We're driving towards shortening this process for bringing new, clean chemistry to market from a decade, like in the pharmaceutical industry 20 years ago, and dramatically shortening that by leveraging everything that we were just talking about.

David Roberts

So how long did the iridium project take? Was that years? Was it months?

Jeff Erhardt

So, we did, within our facility and our company, we studied, if you remember, I talked about, we studied something on the order of 6 million different combinations of novel materials relative to the 1000 that were done historically. We did that over a period of about six weeks.

David Roberts

Huh.

Jeff Erhardt

And then, over the subsequent two quarters, we scaled those up and we picked the best candidates out of there. At this point, we've done more than, sampled more than 50 different combinations, scaled those up, tested them in our own electrolyzers, and that was about two quarters' worth of work.

David Roberts

So, you've put the iridium substitute in electrolyzers and run the electrolyzers, and it works.

Jeff Erhardt

It works, exactly. And we are able to demonstrate better performance, equal durability at greater than 80% cost reduction.

David Roberts

Crazy. And is this one of these brand-new materials in the world? Like, is this one of these novel materials, the ones that you're using?

Jeff Erhardt

They are, yeah. So again, they come from different combinations of things on the periodic table, but many that we have done have never been identified or discussed in the literature. And the reason for that is it's very low cost for us to just try things. There's no downside.

David Roberts

So when you come up with a material, do you have to give it a name?

Jeff Erhardt

Good question. I'm not the creative one around here.

David Roberts

Does someone give it a name?

Jeff Erhardt

Somebody will give it a name. Exactly.

David Roberts

I mean, you're going to need a system for that if you're cranking out materials.

Jeff Erhardt

We'll use ChatGPT.

David Roberts

Yeah, exactly. All right, so we can see how it works for iridium, and this will presumably make electrolyzers cheaper and then spread them. So, I guess the first thing that comes to people's minds when they hear this, the first thing that's been on my mind the whole time I've heard it, is just, this sounds like godlike power. This sounds like "let there be light" type of things. Like "let there be new materials". I just wonder, is all of physical reality just writable now? What are the limits to, like, if I just came to you and said, "I need a material that has x, y, and z characteristics," you can just whip it up?

Like, is there any material you can't make in an electrolyzer? Or is it like a universal material generator? I don't even know how to ask this question properly, but like —

Jeff Erhardt

You are. So I'll unpack your question into a couple of different ways. And so, like many times, materials and words can be used to mean multiple different things. So let's separate sort of the materials synthesis and discovery of these things that we're talking about around iridium replacements, those nanoparticles that can act as catalysts to drive a reaction and in turn produce a "different" material that is a chemical output. So let's be clear on defining our terms there. But to your question today, can you come to me and say, give me a specification and I will define for you a material with these set of characteristics?

No, that's not what we're trying to do today. We are trying to be much more focused. And the reason for that has to do with what we kicked off the top of the show with, is there is this very big gap or this goal between the intrinsic property of a material that can be discovered in the lab or using an algorithm and how it behaves once that material is scaled and integrated into the system in which it will eventually operate. So, I think that's really important as we potentially get sucked up into the hype of materials, AI being able to do what you just described.

With that said, I'll now speak out of the other side of my mouth and say, that is exactly the vision that we're driving towards, which is as we go through and we study not just hydrogen, an alternative catalyst to produce hydrogen using electricity, but move on to this other massive family of chemicals, and then more importantly, other electrochemical systems, and we start to study those systematically, start to generate a bigger corpus of data across many different use cases that the AI can be trained on top of. Then we can start to do what you said, which is start to do what I like to think of as synthetic discovery, which is being able to specify, to be able to mine, to be able to go in and say, "All right, I would like to produce," and let's talk about what this pod is about, the electrification of everything, the electrification of chemicals, production. "I would like a specialty chemical with these characteristics."

David Roberts

Right.

Jeff Erhardt

Right. What is the input, what is the operating condition? What is the architecture of that electrolyzer? And then, very importantly, what is the unique composition and structure of that catalyst material that allows me to produce that efficiently using electricity that doesn't degrade over time, that isn't poisoned, etcetera, and that we are absolutely going to be able to do.

David Roberts

But, like, within any limits, like, you could just, you know, like, I mean, presumably, like, for any given material, there's some set of molecules and some catalyst that can produce it. I mean, is there any material you couldn't have produced —

Jeff Erhardt

Oh, absolutely.

David Roberts

in an electrolyzer? I mean, is there any material you couldn't produce?

Jeff Erhardt

Oh, absolutely.

David Roberts

in an electrolyzer?

Jeff Erhardt

When I used to run industrial AI at my former employer, General Electric, I used to say, "I'm the most boring machine learning or sort of AI person you'll talk to, because I never like to say, 'Oh, nothing is impossible.'" And here's the reason why: ultimately, even if you could dream these things up, if you could say, "This could be theoretically possible," unlike the digital world, in the physical world, eventually we have to be able to make this stuff. We have to be able to manufacture it. It has to be practical, it has to be cost-effective.

It has to be feasible. And so, no, I don't want to say or set expectations that we could magically synthesize and create anything that we want, but that's, again, the whole game. That's why we think about, and why I think it's so important to have these real data generation factories, the ability to truly synthesize things at massively parallel at the nanoscale, but then, more importantly, go across those scales and understand real-world behavior.

David Roberts

Let me ask a question about electrolyzers, since they are central to all this, and they are, you know, if you buy this, which I do, they're going to be extremely core to our industrial society pretty soon. So, I think we all need to understand them better. Like, if I have one electrolyzer to make hydrogen and another electrolyzer to make this iridium substitute, how different are those two electrolyzers? What parts are interchangeable? To what extent is that electrolyzer bespoke? Or, to what extent can electrolyzers be sort of used in a general way to produce lots of different materials?

How much is the electrolyzer designed around its purpose? I guess, is my question.

Jeff Erhardt

Yeah, so I think let's call it moderately. Let's — exactly right, so, like everything else, yeah, I could say, if I was a lawyer, I would say, "It depends." But, you know, that joke aside, they still have common components.

David Roberts

Right.

Jeff Erhardt

What are you doing? Right. You're trying to take an input that is a feedstock, and you're trying to transform that input in the context of these magic materials called catalysts into something else driven by electricity.

David Roberts

Right.

Jeff Erhardt

Okay, that is the fundamental basis that is true. Now, there are components in there. There's engineering that needs to be done around how these systems work. Some types of electrolyzers have membranes, some do not. Some work in acidic type of environments, some work in alkaline environments. And so the answer is that the general concept is the same. But the subtleties about how do you understand those interactions between all those different components and the variables therein is incredibly complicated. Our view on the world, the bet we're making, is the beauty of what's going on in the hydrogen space today, is because there's so much effort, so much investment going in, that they're starting to push the boundaries down the learning curve of how do we improve that general —

David Roberts

That's what's motivating my question is just like, how much modular, large scale, repeatable manufacturing is possible here?

Jeff Erhardt

Yeah, I think the answer around those is substantial. And that is the bet that we are making as a company, is that if you think of different architectures that can be used hydrogen, whether it's a so-called PEM type of architecture, or an AEM type of architecture. For us, we're agnostic around both of those because as we just spent the last 15 minutes or so talking about, our goal is to effectively abstract those specific architectures and to be able to test and study many different combinations again, both at the nano level and at the lab scale all at once. And so, having all of those being driven down the learning and improvement curve is goodness.

And it's going to be a benefit for something I think we're all interested in, which is how do we clean up this industry writ large?

David Roberts

Right. Well, part of what motivated your people to reach out is that kind of the weird thing happening in green hydrogen right now is that the hype is a little bit outrunning the reality. And we're on the verge of having a lot more electrolyzers than we actually have green hydrogen customers. We have a bit of an electrolyzer oversupply problem. And so part of what motivated my question is just, if you have a bunch of electrolyzers that are meant for making green hydrogen, could you theoretically just take those off the shelf and make something else with them?

Jeff Erhardt

Yeah, we wouldn't take them off the shelf per se. They would need to be reconfigured. But the idea would be to take that manufacturing capability, to take those components that are standardized and more generic. Then, how can you redeploy that by optimizing other components? How can you change a membrane? How can you change a catalyst? How can you change, again, that input feedstock and leverage that broader manufacturing capability that has been put in place that is driving this hydrogen market? But as you said, the hydrogen market has other bottlenecks that are preventing its ability to ramp.

So, exactly what you just said is one of the sort of macro-scale bets that we're making as a company, which is these are being invested in, they're being driven down the learning curve. The hydrogen market has uncertainty when it's going to ramp. We're going to take advantage of that. We're going to learn from them. And to the extent that we can redeploy that learning, that manufacturing capacity at scale, we are absolutely going to.

David Roberts

So, let's talk turkey here. Presumably, somewhere you have a business plan where you have a sequence of materials or markets you're going to go after, and the mind sort of boggles. So, let's just talk maybe about a few of the specific items on your to-do list, like what materials do you want to make substitutes for in the next, I don't know, five years?

Jeff Erhardt

Yeah. So, let's first start from the top down, and let's talk about the — if we want to clean up the broad production of chemicals and fuels, there are three major pathways that you can do that. The first one is that you can keep an existing thermochemical process and you can provide electric heat to drive those reactions.

David Roberts

Yeah.

Jeff Erhardt

The second pathway you can go down is the biological path or the synthetic biology path. So, that's companies like LanzaTech, who we share a building with, driving fantastic innovation in that space. And then the third is what we've just been talking about, that is the direct electrification, that is using electricity, electrochemistry to produce those. And our view broadly, and there's actually been some other people publishing some very nice papers around this, is that as these three technologies start to mature, there are certain core platform chemicals that are most advantageous to do in electrified heat versus synthetic biology versus direct electrification.

And it's one of the most important things that we and our peer companies have to do is start to educate the marketplace about what those pathways are and where they are best applied.

David Roberts

So you don't think this is a situation where the market will eventually settle into one of those? You think there's room for all three of those?

Jeff Erhardt

We do because we think there's certain chemicals that are best done via a biological route. There are certain other chemicals that are still best done via a thermal route, and certain other ones that are best done electrochemically. Exactly right. Not that — so to answer your question directly, no, I don't think any one of those is going to be a winner take all. And the question is, how do we start to make sure that the incumbent players and the startups driving these new technologies start to figure that out and apply the right resources in the right areas?

David Roberts

Is there any, I mean, I'm just a liberal arts major, so my pretty shallow understanding, but is there any way to characterize on a general level what kinds of chemicals are best suited to what type of synthesis? Or is it just a case by case type?

Jeff Erhardt

I think that really, it's really a case-by-case basis, and that requires the very deep scientists who understand organic chemistry to analyze those and understand why. But the good news is the smart people are starting to do that. And like I said, there was a very interesting paper by one of our advisors recently laying that out, and it lays out very nicely here are the different ones that are electro-privileged versus thermal-privileged versus bio-privileged, etcetera.

David Roberts

I guess just part of me thinks that if part of what you're developing is this super intelligence that has the matrix of all the universe's materials and their interactions in its head, it just seems like over time, you're going to figure out ways to do that with your machine. You know what? You know what I mean? Eventually. So what's on your list then? Like, what's after iridium and hydrogen?

Jeff Erhardt

So, obviously, hydrogen and helping to relieve one of the bottlenecks in that market was the first thing that we wanted to do. But after that, what we've really focused on is developing a roadmap aligned with exactly what you just said. And what I mean by that is twofold. So, the first is, we believe, and one of our north stars as a company, that if we are going to succeed in driving the decarbonization of this industry, we have to do so in a way that is economically viable and competitive relative to the status quo.

David Roberts

Right.

Jeff Erhardt

So, the first thing that we've done in our near-term roadmap is do exactly that. We took that list of this massive universe of different chemical transformations that are beneficial or that are so-called electroprivileged. And what we did is we did a giant techno-economic analysis to try to understand how difficult is the development around those. What is their potential for being economically viable? And then, what decarbonization and other benefits do they have for the world? And we created basically a prioritization matrix, where the next things on our roadmap are those that we believe provide the best economic viability, the least technical risk, and the most environmental and sustainability benefits.

David Roberts

Bang for the buck.

Jeff Erhardt

Bang for the buck. So, what are some examples of that? The examples of that are some that are in a family of what we call organic oxidations. One example is, for example, the production of acetic acid. That is a platform chemical that's used all over the place. What we can do and what we're working on is leveraging ethanol and particularly bioethanol. So, I guess one part that we didn't kick off with, I'm currently sitting in Chicago. We're a spin-out from Northwestern. We are in the heart of bioethanol production and clean energy up here in the upper Midwest.

And that gives us the ability to use electricity and electrolysis, just like we've been talking about, to start with that as a feedstock and make this major platform chemical more sustainably and more inexpensively.

David Roberts

How is acetic acid currently made? Like, what's the environmental benefit of switching?

Jeff Erhardt

Yeah, yeah. So it's currently something a little bit technical. It's called the Cativa process. It effectively combines carbon monoxide and methanol is the way that it's done. So, using traditional sort of petrochemical processing. And so, trying to do that to use renewable CO, carbon monoxide plus renewable methanol would really be prohibitively expensive. And so, this is an example of doing that, or trying to use sustainable or more circular feedstocks to produce that wouldn't work using the status quo. Which is why this is an example of using electricity and electrochemistry to drive that reaction is such a benefit in this particular example.

So, that's one key example, and there are other families like that. So, acetic acid, acrylic acid, things like that, they're boring, they don't sound very interesting, but they're used all over the place and they are major contributors to the carbon emissions around the world.

David Roberts

So, what about plastics? Everybody wants to know about plastics. Like, this is sort of like, I spent almost my entire career dodging this topic, mainly because, as you say, it's both sort of depressing and a little boring.

Jeff Erhardt

Yeah, that's exactly right.

David Roberts

When you contemplate what's possible, so conceivably, could you replace all plastics that currently come out of, you know, petrochemical production or some subset of plastics?

Jeff Erhardt

All is a strong word, but absolutely. So, let's give a specific example. PET is used all over the place. That's your plastic water bottles, that's packaging, synthetic textiles. There's an analog to that called PEF. And where this becomes really powerful is there's potential for that PEF to reduce the overall life cycle emissions of those by more than 75%. So, think about that. It's really powerful and really interesting. If you can take this, those are not going to go away. We'd all like to think, "Oh, we're going to live in a world without plastics." I try to avoid them as much as possible, but it's simply not going to happen.

They exist and they're everywhere because they're very useful. But if we can start to do that, come up with these analogs that can make them cleaner but also easier to recycle, so we can drive towards a more circular economy, that starts to become very powerful. And so today, that's all done. And this PEF has been proven; it exists, but it's done via thermal methods, thermal chemistry. We've got high energy costs, high carbon emissions, we've got the ability to do that using electrochemistry using something called converting what's called HMF into FDCA and then allowing us to scale those into this clean PEF production to drive this decarbonization of these plastics that are used everywhere as well as make them more circular and more recyclable.

David Roberts

Yeah, I mean, decarbonizing PET would be, in and of itself, just enormous.

Jeff Erhardt

Exactly.

David Roberts

Such a huge thing.

Jeff Erhardt

That's right.

David Roberts

So, we haven't talked about cost or price, and I'm guessing that there's not easy generalizations to make. But like, you know, just for PEF, as an example, can you make PEF cheaper than the alternative process for making PEF? Can you make PEF cheaper than PET? Because, you know, fossil fuels are notoriously cheap. And like fossil fuel processes, petrochemical, chemical, and plastic production is notoriously cheap. So there's not a lot of margins here. So, like, what's the what, what can you say in general about the cost of synthesizing via electrolysis versus petrochemicals?

Jeff Erhardt

Yeah, so I'd say a couple of things. So, you said it exactly right, David, which is there's no single answer, but let's talk about some general answers and some macro trends. So, the first one is exactly what we said earlier, which is they're not all the same. And so again, our thesis as a company is to start with, to learn, to debug, to optimize, to start to develop that massive database of understanding on those ones that are the most economically viable on a standalone basis. That's why we're starting with these things that we talked about earlier, the organic oxidations into things like acetic acid, as opposed to some examples that are interesting, important. On the opposite end of the spectrum are things like CO2 conversions, taking CO2 and converting it into say, something like ethylene. Very interesting. And that has a great circularity story. But splitting CO2 is incredibly energy intensive. And from our math in the near term, we don't see any way to do that on an economically viable basis without subsidies or a carbon tax in place. And so, our goal is to start, to optimize, to develop, to drive these overall systems down the cost curve, number one.

David Roberts

One, to ladder up, basically. Ladder up to these.

Jeff Erhardt

Ladder up. Exactly right. Such that then, as the overall electrolyzer technology matures, as our technology improves, as energy costs ideally come down, which I think many people would say is the macro trend over time with more renewables coming online or potentially coming back.

David Roberts

Literally, everybody I talk to on this pod is banking on that happening. So, let's cross our fingers.

Jeff Erhardt

I'm not standing alone on that one. But what that allows us to do is to bring all of those things together and then come back and start to think about some potential moonshot bets, and not just these platforms. Good and important. But what might some of those moonshot bets be? I could throw one or two of them at you.

David Roberts

Sure.

Jeff Erhardt

Which are very interesting, tell some adjacent stories. So, let's talk about — I like talking about hydrogen peroxide, because people don't think about it. But the other one also is ammonia, which is also very interesting.

David Roberts

Ammonia. Yes, I just recorded a pod on shipping emissions, and ammonia played a big role there.

Jeff Erhardt

Yeah, they're thinking about using ammonia as a fuel. Right. As is Japan. So, Japan has energy challenges of its own. They are in the process, and some people debate whether this is good or bad, but starting to convert their traditional power plants to be ammonia-fired power plants. But what's the point? Ammonia is still used for all kinds of other things, and the way that ammonia is produced today is highly centralized. So, ammonia is one of the most important platform chemicals in the world, in particular for its use as a precursor to a vast array of fertilizers like urea.

The status quo is that ammonia is produced in world-scale facilities by transforming hydrogen and nitrogen through what is known as the Haber-Bosch process. This manufacturing process is over 100 years old. It's both very established and very cost-effective, but also extremely energy and emissions intensive, accounting for nearly 2% of global emissions in its direct production.

David Roberts

So, Japan doesn't want to import a bunch of embedded carbon, which is what they would be doing now if they were switching over to today's ammonia.

Jeff Erhardt

Correct. Exactly right. And so, in addition to that, just to put the punchline on this story, shipping ammonia is expensive and dangerous. What if we could imagine a world where we could leverage the modular and distributed nature of electrolyzers to produce this critical chemical directly where it will be consumed using clean inputs and clean electricity? This is a world-changing moonshot that today is not economically viable. Competing with this incumbent status quo is extremely challenging. However, leveraging everything we've just been talking about, advancements in combinatorial materials discovery, computational chemistry, systems design, and AI, this dream starts to come much closer to being realizable. And that is exactly the mission that we are on.

David Roberts

Right. I mean, why wouldn't you? Like, when you set out to do iridium, you didn't set out to make iridium. You set out to make a material that would do what iridium does.

Jeff Erhardt

Correct.

David Roberts

Why not approach ammonia that way? Like, why not just try to think of what would be a good fuel, like what's a good performing fuel molecule?

Jeff Erhardt

No, that's exactly right. And so that's the challenge. With that said, it's also back to the question that you asked, which is, okay, what does it mean to displace these incumbent solutions? This is a fully mature, fully optimized —

David Roberts

Low margin.

Jeff Erhardt

Exactly. That's been in production for almost 100 years. And so that's why I said this is a moonshot bet. That is the ability to produce ammonia out of a renewable feedstock using renewable electricity. That's a moonshot.

David Roberts

But at least in shipping, you have considerable policy winds at your back. And like, the industry itself wants to —

Jeff Erhardt

That's right.

David Roberts

wants something low carbon. So, you know, that's kind of like an open door you're pushing on there to some extent.

Jeff Erhardt

That's right. So, it's, I mean, the point is, you know, think of that as being, you know, want to reduce nitrogen into its end product. Right. You could say the same thing about hydrogen peroxide. Right. In the way that's produced. So, how is hydrogen peroxide produced today? Same thing. Massive centralized facility. Where is it used? It's used in bleaching agents amongst many, many other things. The problem is, shipping hydrogen peroxide is incredibly dangerous. So, what do they do? They produce it in a centralized facility. They dilute it back with 90% water and then they reconcentrate it when it's going to be used.

David Roberts

That's so insane. It adds — so it's like adding 90% weight to what you're shipping.

Jeff Erhardt

That's right.

David Roberts

That's wild.

Jeff Erhardt

That's right. Exactly right. So, the way to think about these things, and again, those are two very non-trivial processes to change and to get right. But you ask the question, why does this matter? It matters not just because we care about new and interesting technology. It matters not just because we're going to do direct decarbonization of these by producing the electricity. It matters not just because we're going to drive towards a more circular economy, but it matters because there are all these other ancillary benefits. So again, think about the, what was this West Palestine train derailment?

I just saw another article about this. Not only did it nearly destroy that town, but now they've been finding the chemicals from that train derailment in adjacent states.

David Roberts

Predictably.

Jeff Erhardt

That's what we have the ability to prevent.

David Roberts

I don't want to let this pass without highlighting it. You say you could produce these things on site. Like, you could produce ammonia on site in a shipyard, I guess, presumably, like where ships refuel.

Jeff Erhardt

Sure.

David Roberts

Like when you say produce on site, is that like a shipping container? Is that a washing machine? Is that a, you know, a small warehouse? How big is one of these things? Or what's the size range? How small could you make it? How big could you make it?

Jeff Erhardt

Yes, and yes. You said the size range. And I don't mean to be completely flippant in that answer, because it's a very, very important point, which is one of the benefits of these systems of electrolyzers is that they are inherently more modular. And because they are more modular, we can do distributed production at smaller scales, without having to build a billion-dollar factory to gain the economies of scale in the production.

David Roberts

Right. This is so important. The reason those Haber-Bosch things and etc. are so big is that you get economies of scale. The bigger your facility, the cheaper your per unit cost, but that does not apply here. Like a unit coming out of a washing machine-sized electrolyzer is the same unit cost as one coming out of a giant industrial-sized one.

Jeff Erhardt

That's right. And of course, you know, different processes will be slightly different. It's not monolithic, but yes, that's exactly the right concept.

David Roberts

And so in your mind, do you see a path to carbon-free plastics and chemicals, or is the horizon still foggy?

Jeff Erhardt

Yeah. So, let's clarify your question. Do I see a pathway to a decarbonized production process for chemicals and fuels? Yes, I can see that.

David Roberts

Right.

Jeff Erhardt

Do I see a pathway to create more circularity of carbon within chemicals and fuels? Yes, I can see that. Do I see a pathway where carbon is not a part of those chemicals and fuels? The answer is no.

David Roberts

Right.

Jeff Erhardt

Carbon is an important part of life. " Carbon is life." That was "Breaking Bad," wasn't it?

David Roberts

It's a pretty handy molecule when you think about it. Yeah. Like, you could, I mean, you could even conceivably be synthesizing, generating carbon-based, I mean, like carbon fiber or all this, like carbon, like high tensile strength materials like that. Like, are those on the, you know, everybody's all excited about carbon composites these days for absolutely super light cars, super light planes. You can theoretically synthesize that too.

Jeff Erhardt

That's exactly right. I am not, and we are not doing that. But a friend of mine has a very interesting company called Dexmat that is doing exactly that. They are using carbon nanotubes to produce new materials like that, that have greater strength and lighter weight than the status of things like steel and titanium. Fantastically interesting. And those get used all over the place. Absolutely.

David Roberts

I mean, you think about how that could change the automotive space, aviation, I mean, building. Material science, always, whenever I talk about it, it always blows my mind. Cause like I said earlier, it brings me back to this, like, feels like this God-like power that we could just, like, create new materials in the world is just a little bit mind-blowing. So maybe as, like, a wrap-up question, it sounds like electrolysis is, like, the world's ultimate multi-tool. Like, you can literally use it to make almost anything if you find the right feedstock and the right catalyst, basically.

So I guess what a lot of people are going to be thinking is, like, we've had electrolysis for a while.

Jeff Erhardt

We have.

David Roberts

I think we've had electrolysis of hydrogen for centuries. If we have these sort of black boxes, as you say, that are theoretically capable of producing anything, why are we just now clueing in? Why are we just now getting that these can be a major part of our industrial life, sort of technologies that have made this only now possible?

Jeff Erhardt

Yeah, so I think you said it exactly right. Again, let's just come back to the very beginning of what we were talking about, which is the transformation of materials. Any kind of material happens via energy. And historically, that energy has come from heat. And what the world has been doing over the past more than a decade has been going through an energy transition. So think about the way energy is produced transitioning towards clean energy, whether solar or wind. But then what did that allow us to do? It allowed us to come in and start thinking about how do we transform transportation and have electric vehicles?

How do we electrify other aspects within our life? And so I think the answer is, it's not that the technology per se is necessarily new or novel, but rather it's that the world is waking up to: We need to fight and we need to drive this broader climate challenge that we are all facing. And how do we leverage technology to do so under the umbrella of let's electrify everything?

David Roberts

I feel like, and you would agree with this, I bet, that the transition from combustion to electricity seems more and more to me like it's inevitable on some time scale, regardless, just because electricity is so much better. It's just like it's just a better: more controllable, more precise, more — It's just a better form of energy. So this would be happening regardless. It's just like everything with climate, it's just about speed, it's just about pushing.

Jeff Erhardt

That's exactly right. And it's about changing an incumbent industry.

David Roberts

Yeah.

Jeff Erhardt

And so changing consumer industries, relatively straightforward. Software, certainly, that moves at a very fast pace.

David Roberts

Right.

Jeff Erhardt

Changing the automotive industry. Yeah, that's been a little bit harder.

David Roberts

Right. Steel, concrete, you get into chemical commodities. These are the stodgiest of stodgy industries.

Jeff Erhardt

That's exactly right. But they're going to start to happen. And the reason they're going to happen is because it has all the benefits that we just spent the last 45 minutes talking about. Cleaner, more efficient, more modular. And importantly, like you just said, the ability to open up new end products that are more finely tailored, that didn't previously exist.

David Roberts

It's such a familiar list of benefits. You know, so many different technologies I talk about, and they all are sort of riding a lot of those same waves. Right. A lot of those same curves.

Jeff Erhardt

Exactly.

David Roberts

Well, this is awesome. And I think, you know, I just think people need to — I mean, you've probably done a lot of this thinking already, but just sitting and thinking about what it means for us to be able to basically generate materials that behave the way we want them to behave, rather than be stuck using the materials that we find in our natural environment. That is such a fundamental — you know, making that fast and digital and replicable and scaling that process up is such a fundamental change for human beings on the planet. Like, it really is mind-blowing to me.

Jeff Erhardt

You said it exactly right. We can open up a whole new universe of possibilities.

David Roberts

Amazing. All right, well, Jeff, thank you so much for coming and talking this through with us. And I can't wait to see, like, you know, what happens with electrolyzers in the next five years.

Jeff Erhardt

Thank you very much. It was fantastic. I enjoyed it very much.

David Roberts

Thank you for listening to Volts. It takes a village to make this podcast work. Shout out, especially to my super producer, Kyle McDonald, who makes my guests and I sound smart every week. And it is all supported entirely by listeners like you. So if you value conversations like this, please consider joining our community of paid subscribers at volts.wtf. Or leaving a nice review or telling a friend about Volts, or all three. Thanks so much and I'll see you next time.