Proton exchange membrane electrolyzer technology: Challenges and advancements
This week, we have, really, the pleasure of Kathy Ayers from Nel Hydrogen to speak about proton exchange membrane electrolyser. And this is an extremely hot area with a lot of development and with a rapid scaling up at all scales and across the world as a really promising technology to help us decarbonize our industrial processes, to decarbonize, potentially, aviation and many sectors that are ready to go to decarbonize. So Kathy has been at Nel Hydrogen since 2007. She's a VP, vice president of technology.
R&D, same thing. [LAUGHTER] Vice president of R&D. And she received her PhD in chemistry from Caltech with Nate Lewis. And actually, I met her and worked with her at Energizer in a period of around 20 years ago. And we spent two or three years at Energizer, worked on, back then, alkaline-based AA or AAA batteries and really alkaline and zinc manganese oxide chemistry. And here we are, after 20 years, working on very different chemistry and very different technologies.
So keep in mind, especially for students postdoc, something we learn during our graduate studies, really it's just a way to help us learn how to solve problems, and we can essentially have opportunity to work on, take your skills and knowledge to solve world-challenging problems as we move forward. And Kathy has managed many of internally and externally-funded programs involving a range of collaborators from academia, industry, and the National Lab. She's also the lead PI, Principal Investigators, of US Department of Energy DoD benchmarking program for water-splitting technologies.
So if you have developed amazing technology in the lab, she will be the person you want to talk to to benchmark your technologies. And she has also received a number of awards over the years. She received R&D awards 2012, and the American Chemical Society Rising Star Award 2014.
She became a fellow of the Electrochemical Society in 2020. And she just received the inaugural Sustainability Award, also from the Electrochemical Society. So it's really a great pleasure that we have an Electrochemical Society meeting this week in Boston.
We have the pleasure of having Kathy here out of her extremely busy schedule. So Kathy, the floor is yours. All right. Thank you, Yang. [APPLAUSE] So thanks to Yang for inviting me while I was up here in Boston. Although, I could have made the trip separately since I'm only two hours away.
So if anybody actually wants to come see what a real-life electrolyser looks like, you're welcome to contact me and come visit Nel. So I'm going to talk a little bit about where the challenges are in scaling something from the lab to a commercial product at reasonable scale and, hopefully, tie it back together to how that still requires a lot of fundamental science as we go through. And if I have time at the end, I'll talk about the benchmarking project. I have a couple of extra slides after the conclusions, if you want to hear about those.
So this is just a safe harbor slide. Nel is publicly traded on the Norwegian Stock Exchange, so I doubt too many people know how to navigate the Norwegian stock system. But just in case, don't use anything I say as investment advice in the stock.
So some of the key points that I want to get across as I go through the talk today are, 1, just give some context around electrolysis because a lot of people think, oh, hydrogen is this new thing. Do you have any commercial products? But actually, electrolysers have a long history as a viable business. It's just increasing drastically at scale right now. Also, renewable hydrogen is required to address decarbonization. We can debate the transportation application. There's a lot of people who say, why are we even working on hydrogen for vehicles? I personally think it has a place for some of the larger vehicles.
But even if it didn't, we still need renewable hydrogen for other industrial processes. The market is expanding rapidly. But we're still not there yet in terms of cost, both operating cost and capital cost. And we'll talk about some of the things that we're doing to address that.
But there is a high potential for improvements in both areas, despite the fact that electrolysers have been around for a long time. Then I'll talk about how scaling presents a lot of different challenges that you might not necessarily-- when you think about them, they may be relatively obvious. But coming from an academic lab, they certainly weren't things that I was thinking of at the time in terms of how do you scale a material, synthesis, component manufacturing, and qualification of these new designs.
And as I said, fundamental understanding is really required to drive all of that progress. So to talk a little bit about Nel-- we have a long history in electrolyser manufacturing in both technologies. So Nel as a company makes two different electrolyser technologies. We make liquid alkaline systems in Norway.
That technology has been around since 1927, so quite an old technology. But even the PEM systems have been around since the '50s. So Proton Exchange Membrane electrolysis or PEM electrolysis was invented in the '50s at GE, largely for military and aerospace applications. It was then purchased by United Technologies, specifically what was then Hamilton Sundstrand, is now Collins Aerospace. And again, it was basically used for outer space and underwater applications.
So in this case, it wasn't used for hydrogen. It was used for oxygen generation. And I'll talk a little bit more about how that has shaped the technology and why we are where we are. In 1996, some of the employees at United Technologies decided that they saw a business for hydrogen with this technology.
And so they founded Proton OnSite-- it was then Proton Energy Systems-- and started developing systems for hydrogen generation rather than for oxygen generation specifically. And in 2017, we were purchased by Nel Hydrogen, which is an offshoot of the old Norsk Hydro, which was making the alkaline systems. So this is our two factories in Norway. This is Notodden. It's a beautiful site, but it's a historic site. So we, actually, because that's where the original 1927 electrolysers were built, we can't change anything.
We can't even repaint the building. And so there's only so much expansion they could do there. So they built a new plant in Heroya, Norway, which is capable-- will be capable of two gigawatts of alkaline electrolysis. Right now they have 500 megawatts, and they've already purchased the equipment for the second 500 megawatts that they're installing now. This is our Connecticut facility. And we can do, currently, about 50 to 100 megawatts of PEM electrolysis there.
But we are actively expanding that to 500 here. And then we'll be expanding in a separate plant. So you might wonder why I put this slide in here. These are the Nel values. I put them in here because a lot of people ask me, well, there's all this activity in hydrogen and all these different players. Why do you choose to stay at Nel? And these are really the reasons.
These are our values-- commitment to our customers and to our technology, honesty is a huge value for us. You will see that the number of players in the electrolyser market has grown substantially. When I started, there were probably four people in the PEM space with viable products. All of us have been purchased by larger companies. So Proton was purchased by Nel. Giner was purchased by Plug Power Hydrogenics by Cummins, and ITM by Linde.
But there are now over 50 companies, and a lot of them are what my CEO calls PowerPoint companies, where they don't really have anything besides a piece of paper at the moment. And there's a lot of hype. And I value the fact that, for better or for worse, Nel has not been one to go out and make a gazillion press releases about what we're doing.
And we have been more honest with our suppliers about where we really think the capacity is going to be. And they appreciate that. However, we still have the boldness.
So we are actively innovating on our products and trying to expand as rapidly as we can. So in terms of the legacy of both of these technologies, as I mentioned, the alkaline electrolysis has been around for 100 years. This is a plant that ran for almost 40 years in Norway. We talk about renewable hydrogen. This actually
was renewable hydrogen. So we're kind of going back in history. This was a plant that was operated using hydropower in Norway, making hydrogen via electrolysis. And then this was used to make ammonia through the Haber-Bosch process. If you look at our alkaline stacks today, they don't look all that different from this. They're still enormous. So a 1-megawatt, 2-megawatt stack is about 30 meters long.
But it points out that these technologies have been at scale since the '50s. And even then, we can still make quite a few improvements in electrode designs, reducing the gap. One of the reasons that alkaline electrolysis runs at such a low current density, 250 milliamps per centimeter squared or so, is because there's quite a large gap between the electrodes. So by advancing our separator materials and such, we can narrow that and get to higher current efficiency, or current density at the same efficiency.
On the PEM side, this is a fairly recent installation. This is our largest installation, 20 megawatts in Puertollano, Spain for ammonia generation also, in this case driven by solar. But as I mentioned before, PEM was really developed for oxygen generation. And so it was designed for life support. Now the requirements, when you're generating gas for life support, are quite different than when you're generating gas for an industrial process. So nobody cared how much platinum you were putting on in these cells.
Nobody cared how thick the membrane was. They just cared that it worked. And so there is a lot of margin, a lot of resistance to change as the PEM systems were developed. And therefore, there's still a lot of opportunity to take the cost out.
And we can see that through what the fuel cell side has done. So if you look at Proton Exchange Membrane fuel cells, they've come quite a bit farther down those curves and increased their manufacturing capability much more than on the electrolyser side. And that's one of the reasons that we've partnered with GM to help us accelerate that. And so we see a lot of opportunity for higher efficiency through reducing the membrane thickness as well as lower cost through manufacturing. So just some of the commercial background around these, the liquid alkaline systems.
They have the advantage of the catalysts being non-mobile metals. So they're things like nickel, cobalt. There is some platinum on the hydrogen side, but not nearly as much as in the PEM systems.
The electrolyte is quite corrosive. It's actually very similar to what Yang and I worked with at Energizer, 9 molar KOH, those kinds of things. So you do have to worry about expensive balance of plant and creep corrosion and things like that. It's also typically ambient pressure, which reduces the cost but also limits the current density. And so these are relatively low-output but high-efficiency systems.
On the PEM side, we do still have to use rare metals. We can talk, maybe during the question and answer, about iridium and it's availability. But these are very high-throughput, so these two stacks are roughly the same capacity.
This one is about 20 cubic meters, and this one's about half a cubic meter. So a PEM stack kind of stands about this high. And this just shows the stackup of components. So it's a very complex stackup of different interfaces, from carbon materials to catalyst electrode membranes, and then typically titanium pore structures on the anode for gas management.
So in terms of where we've come from on the PEM stack side, our initial products were at the 28 square centimeter scale. These were largely for the lab market. We've actually exited that market now because we're more focused on the larger systems. We sold that business off, but we still make the stacks for some of those manufacturers. When I started, we were just in this transition from the 100 to 200 square centimeter active area. I was going to replace this picture, but I didn't send the new draft.
We are transitioning this round stack to a square stack that's very similar to these platforms so that these three are kind of all very similar in form factor. But in the last few years, we've scaled this 200 to about 700 square centimeters. 100-cell stack of this size produces about 100 kilograms a day. And then our largest stack is 1,600 square centimeters, which presents its own problems in terms of handling parts and such. But the stack, at 200 cells, makes 500 kilograms a day.
So there are two of these stacks in our 2 and 1/2 megawatt unit that produce a ton of day of hydrogen. This shows, similarly, how the systems have scaled. So again, I don't have the lab unit on here, but the smallest unit we still produce today is about the size of a dishwasher. The next size up is about the size of an industrial freezer. And then we start splitting into two cabinets-- one fluids, one electrical-- and containerizing the large systems. And I just put this up here so that you can kind of see the numbers of systems that we've put out.
I didn't put the capacity chart on here, but you would see that you're really starting to see that hockey stick curve that people have been predicting for 20 years is finally here in terms of the amount of electrolysis that's out on the grid. So these are some photos of both of our factories. This is the Connecticut facility. This is the systems area.
We have a separate room where we make the cell stacks. But you can see, we actually do manufacture products, quite a lot of them in fact. And actually, business has been very busy, not only at the large scale, but even for these industrial units for other types of applications. This is the Heroya plant, where everything is automated. These are some of the plating tanks. You can see they're quite large versus the person standing here.
But it's pretty cool. You can go into the control room and see electrodes kind of moving around the plant, whether they're going through the plating baths or other welding processes, et cetera-- something that I wouldn't have thought of as an electrochemist at Caltech, doing solar cell research in cells this big. So these are some of the traditional electrolyser markets. And again, I just want to put these up here to point out that electrolysis has been a sustainable business for quite some time. Hydrogen delivery at these scales is quite expensive. And so even though electrolysers are not at the scale of what you can get natural gas for at a Haber-Bosch plant, it's still quite affordable for some of these smaller applications.
So we've sold units into the food industry for hydrogenation of oils, the glass industry for heat treating, silicon laboratory gas, as I mentioned. We still do supply stacks to the navy for life support, and other areas where you need either an inner heat treating environment or reducing environment or also as a cooling fluid for the power industry. Most gas and coal-fired power plants use hydrogen as the cooling fluid in the turbine windings.
So that's kind of the past. Where we see hydrogen going in the future-- as I mentioned, hydrogen at scale is really needed for decarbonization. I would say electrolysis is one of the few things that can address that need in the near term. There's a lot of other technologies that are being developed but are probably less mature than this. We need renewable hydrogen for the chemical industry to make materials like ammonia. And also, if you're going to try to capture CO2 and do something with it, if you're going to make it into anything else, you need hydrogen to convert it to a hydrocarbon.
And then, as Yang mentioned, there are areas where you could use hydrogen as a fuel, potentially in aviation or as a precursor to an aviation fuel, like sustainable aviation fuel. And long term energy storage-- batteries are great if you're storing for 8 to 12 hours. But if you're going longer than that, it starts to become very expensive to have that much battery capacity as opposed to just storing extra hydrogen. And this is a schematic from the Department of Energy, showing how hydrogen can bridge all these different areas, both electrical, chemical, and transportation markets.
So just to point out the magnitude of the problem-- if you look at ammonia specifically, ammonia production creates 1% of the CO2 emissions. And that's basically because of the hydrogen step. So if you're making hydrogen from natural gas, you're producing something like 9 tons of CO2 per ton of hydrogen. And this one
is the highest by far CO2 producer of any of the major chemicals. So if you look at this chart, it shows the greenhouse gas emissions in terms of megatons of CO2 and the production volumes. Ammonia actually has its own y-axis because it's higher than any of these others in here.
And so that's where we see just one need for renewable hydrogen. As I mentioned before, we also need hydrogen if we're going to recycle CO2. So you could either do that indirectly, capturing air and taking hydrogen from electrolysis to make chemicals. Or you might be able to do it directly in the stack. But it's still going to be very much like a proton exchange membrane type of stack, where you're just-- you're not producing the hydrogen electrode-- the hydrogen directly, but you're producing protons that then react.
So moving into some of the more large-scale potential markets-- I mentioned ammonia already. But also, as long as we have fossil fuels, we still need hydrogen for refining. Again, we don't need to compound the problem by having that hydrogen not be renewable.
So that can help us transition at least. And then as I mentioned, things like reducing the environment or as a reductant, hydrogen can help make steel manufacturing much cleaner. We do see a strong momentum within mobility as well. So as I mentioned, you can debate the viability of, say, passenger vehicles, fuel cell versus battery. But for these heavy-duty vehicles, it really makes a lot of sense rather than carrying a huge payload of batteries. So there's a lot of companies, including GM, Nikola, and others-- Hyundai, who are doing fuel cell-- Toyota-- based trucks.
But even beyond that, things like mining vehicles and other transport vehicles as well as buses. So again, I just wanted to point out, perfect can be the enemy of good. And so I think the reality is that we need to be deploying electrolysers at large scale today, whether they are the perfect technology or not, because it just takes so long to scale something that's at a low technology readiness level today. And we really need to start addressing the CO2 problem now, which is why things like the Infrastructure bill et cetera are so helpful.
So this was a chart we put together that just shows, if you assume that in two to three years we can be installing 100-megawatt PEM electrolysers, which is quite reasonable. I showed you. We already have 20-megawatt installations.
We're already doing 100-megawatt installations with alkaline. Then you can grow the total installed capacity relatively rapidly. But even if you make very, I would say, wildly positive assumptions-- and I think my friends in the solar fuels research area and others would agree with me-- these alternate technologies still are going to take a long time to catch up to PEM or alkaline. So if you look at electrolysers and their installation over the next decade, these two together are going to be a very large fraction of the amount of renewable hydrogen that would be deployable. It doesn't mean we shouldn't work on these advanced technologies.
But we need to start getting these out there now to start addressing the problem, because that's what's going to help in the near term. And this timeline is something that's been published by others as well. [? Della ?] [? Stilton ?] at [? Julik ?] published a similar study, where he was looking at-- he put a thermometer up on there and said that, basically, if you're starting with a new technology, it kind of takes 10 years to develop it and then another 10 years to get to full market penetration.
So we're talking 20 years to really get things out there at a substantial scale. So that's why we need electrolysis, and we need to be deploying it now. How we see the he growing-- this was a study by the Hydrogen Council. So in 2020, there were about 70 million metric tons of hydrogen production. Very little of that was from electrolysis.
I would say 1% or so. Most of it was from natural gas and fossil fuels. But if you look at, then, how the market today will likely scale up through 2050 and in applications such as transportation, but also increases in industrial feedstock, things like heat and power and energy storage, you could have an eight-time total overall capacity increase. And since only 1% of that is addressed with renewable hydrogen today, if you were going to meet all of this demand with electrolysis, the industry has to grow by 800 times. If you compare where we are versus fuel cells today, we have significantly higher material usage.
Our catalyst loadings are about 8 to 10 times where the fuel cells are. Our membranes are also about 10 times as thick. And we may not be able to get all the way down there, but we can certainly get farther than we are now. And the cell thicknesses overall are three to five times higher.
So we're using a lot more material. Our processes are much slower. And our total capacities from our plants are much lower.
So we can do-- the automotive manufacturers can do about 100 to 1,000 times as many cells per year as we can do today. So we're really trying to leverage those learnings in advancing electrolysers. This is where we see some of the potential going. So this was actually an independent assessment by the IEA. You can see, there's quite broad ranges here. But nonetheless, both for alkaline and PEM, people are predicting that the costs will go down significantly over the next 20 to 30 years.
Alkaline has a shallower curve than PEM, largely because of things like reductions in precious metal that can be achieved here. But they end at fairly similar points. And this is just a high-level roadmap of what we plan to be doing in our stack. So we predict that our stack costs are going to come down by over half within the next three years or so. On the system side-- so that's for the stack. On the system side, I think will largely be able to take cost out just by scaling the system.
So if you look, for example today, at our C Series, which is that two cabinets, we're already at cost parity with the alkaline systems at similar scale. But when you look at our megawatt KOH systems, they're quite a bit less expensive than, say, trying to build these up-- this is about a 200-kilowatt unit-- into a megawatt. And that's because-- I think it's fairly obvious-- if you try to put five cabinets together and try to get it to the same cost as this one contained unit, it's going to be a lot more expensive.
And so the system part is going to scale down in cost fairly naturally as you scale the system. And so the rest of the talk will be more about the stack. In terms of some of the material needs, one of the reasons the precious metal loadings are so high is because of the iridium side, which we use for the oxygen evolution reaction. It's difficult to find support structures that are stable at these kind of conditions. So the PEM cell is slightly acidic.
And you're operating at close to 2 volts, so there aren't a lot of materials that remain conductive and stable in those conditions. On the platinum side, we can use carbon as a support. But that would oxidize on the anode. So there's no acceptable OAR support. I would say there's also still a lack of understanding in terms of what that local pH really is and what we need to design for, and also how the ionomer is distributed throughout the catalyst layer to help with that ion transport, and how we really look at that at an electrode scale after you put it together.
Also, despite these systems being around for 25 years or 50 years even, the materials that we're using were not designed for electrolysis. They were typically adapted from other industries. The polymers, for example, we're still using Nafion.
That was developed for the chloralkali industry many, many years ago. The porous transport layers are largely metal filter materials. So they weren't optimized in terms of chemical structure, physical structure, et cetera for this application. And I think there's room.
And also, all of these different interfaces have to work together in the device. So it's a fairly complex optimization problem. However, we've seen already that we can take quite a bit of cost out of the different components. This is an example I like to use because the bipolar plate used to be the most expensive component in the stack. We had a DOE program that was about $3 million. We worked with Oak Ridge largely on this project.
And by kind of going back to the drawing board and saying let's really design a plate that has all the functionality that we want in an electrolyser but is made in the least expensive way we can think of, we actually were able to take a lot of cost out. And that's no longer the most expensive component. So we did a lot of both FEA and fluid modeling to make sure that the mechanical strength was sufficient to support the pressures in the stack and also give us even fluid distribution. We did coding development with Oak Ridge, where we're doing some accelerated testing at high pressure and high temperature to see whether exposing these coatings to hydrogen would eventually result in embrittlement of the plate, et cetera, and came up with something stable. And we were able to take about 80% of the cost out of that component. And that's really formed the basis of those two largest stack platforms.
And as I mentioned, were implemented back in the smaller ones as well. So that's what we want to do with the rest of the major components in the stack, and those are the things that we're working on. But the next several slides, I'll talk more about what the challenges are there.
So when we're trying to translate something from the lab to a product that we're making every day, just to get a sense of scale, our typical test cell is about 25 square centimeters in active area. If we made one megawatt electrolyser, that's about 270,000 square centimeters of total active area. So it's 11,000 times this little cell. And we have to basically make that same thing that many times.
If you look at the alkaline plant that I showed you before, that 130 by 5 megawatt plant, that's almost 10 million times our test cell total area. So we have to be able to do the same thing millions and millions of times. The lifetime expectations for these systems are 7 to 10 years.
So we have to know that they're going to last that long without necessarily testing them for that long. And we have to, therefore, know that our processes are robust and understand them very well so that when we send them out into the field we can predict how they're going to perform. Again, this may seem fairly obvious. But again, I've talked to other graduate students and it kind of makes their eyes go big when we think about it.
And it would have for me, too, at Caltech. If you were doing tests in the lab, and you were able to get the same result 99 out of 100 times, you'd probably think that was pretty good. I wouldn't be too upset about one failed experiment at this scale. But if you think about translating that into manufacturing, if we only had 99 out of 100 good cells, we would fail every single stack.
So that's completely not acceptable. We have to be many orders of magnitude beyond that in terms of how perfect we make these cells. And as we change things, things change from the materials all the way through the process. So even if you think about scaling your catalyst synthesis from grams to kilograms, the reaction vessels are going to have different conditions than when you were doing it in the lab.
So this would be, as an example, if you're using a process that requires heat for the reaction or even if you have heat from the reaction, the thermal distribution from a small vessel is going to be quite different from what it is in a larger vessel. You may get to the same temperatures, but the profile is not going to be the same. You're not going to have the same dwell time at the max temperature, et cetera. And so you have to, again, understand how to design these so that you get the right reaction conditions. Otherwise, you end up with a lot of differences. And so these are some of the differences that you can see.
Just as an example, in catalyst, these are all iridium oxide or iridium Ox because a lot of these have metal blacks in them as well. You can see some particles here. But you can get different shapes, different sizes.
So these are very angular. These are much smaller than these larger particles. And also, a lot of these are reacting from salts. So when you scale you can get some problems with rinsing and things like that, where you're not getting quite the same conditions. And you can get impurities in your final catalyst.
This is one example where we had calcium contamination, which is not good because these can leach into the membrane and reduce the conductivity. Similarly, on the membrane side, it's relatively easy-- some of the polymer chemists would maybe say not-- to cast a membrane at this scale. This is basically beaker chemistry. You can use small batches of ionomer. You can cast them by hand or on a small table like this. And at this scale, the drying across that is going to be relatively even.
You obviously want to be at roll-to-roll manufacturing to have a viable process. But getting from here to here can be quite challenging. We've seen it with even established polymer suppliers, where they can give us small quantities, even meters of material that work very well.
But when they actually try to make it at scale, they get problems with things like inclusions. Some of these membranes are hydrolyzed. And if you don't do that evenly, you can get spots that aren't conductive, et cetera.
And so you need to be able to make huge areas of polymers that are within plus or minus microns in thickness and are very consistent in performance. And then when you start combining these things together, so making membrane electrode assemblies, you can cast electrodes pretty easily, again, at this small scale. This type of experiment is really good for trying to get an initial idea of where the viscosity needs to be, what kinds of loadings-- catalyst loadings-- will give you the performance that you want. But translating that to a roll-to-roll process is more complicated. So just the fluid dynamics of putting a strip of ink and dragging it across here versus trying to put something through a slot dye is quite different.
And understanding how those particles move and things like that and things like surface tension et cetera are really important in translating these over. So these are some of the types of defects that you can get. You can get bubbling or mud cracking in the catalyst layer.
That can have to do with viscosity or surface tension effects. If you're coating large areas and then trying to dry them, if you have too much airflow or other things, you can get ink flow. And then you get things like striations. Or you can get things like streaking and voids. So these are all things that we want to avoid when we're moving to scale. Also, most of these electrode processes are lamination processes.
So the ion exchange membranes tend to warp and do things when you change moisture conditions. And so it's easier to print an electrode and then laminate it to a membrane than to print an ink on a membrane directly. But again, it's much easier to achieve a consistent lamination that has very uniform pressure across the entire electrode and uniform temperature when you're at these small scales and when you try to scale up. Where, if your platens aren't perfectly flat over these very large areas, you get quite a bit of difference in the total load. And that causes nonuniformities in the lamination. Moving on to the porous transport layer, our porous transport layer has to serve multiple functions.
As I mentioned, the materials we use today were not developed for this purpose. They were developed for things like filtration. But they have to both contact the catalyst layer well because it's providing the conductive support to that. They have to support the membrane, so they can't be too porous. And they also have to transport fluids.
So they have to transport water to the membrane and oxygen away from the membrane. And this development is still relatively immature. There's relatively few porous transport layer suppliers. And we're just buying what they have. This was a paper by Hubert Gasteiger a few years ago, showing that if we really want to get to these low catalyst loadings and we have these massive pores in our porous transport layer, as we reduce this thickness, you're not going to be able to contact all of the catalyst. And therefore, you probably need something like a microporous layer.
So those are some of the things that we're working on to really design that material for our purpose. In doing that, again, it has to be something that's manufacturable. We can figure out what we need by doing things like 3D printing. And really, I mean, you can make all kinds of extremely complex structures this way.
But it's probably not going to be fast enough or cheap enough to do at scale. So this helps you determine what the parameters are. But then you have to figure out how to make it at scale. And so this is an example of a PTL with a microporous layer that was actually made by chemical manufacturing in large sheets. So those are some of the challenges in actually building things.
Then there's additional challenges in qualifying and testing things. So if you're making these systems at these kinds of scales, you really have to not only think about the manufacturing of the components but also how are you going to inspect them. Because you're making them at high volume, you don't want to make a whole day's worth of parts that is going to be a lot of meters of membrane or a lot of kilograms of catalyst, and then find out it was wrong.
So you have to be able to inspect these things as they're coming off the line quickly. Visual inspection is quite slow. We've used it quite a bit in the past. Our operators are really good at finding defects.
But that means they have to have a very constant attention span. And your eyes can fatigue and things like this. So both for reasons of speed and also accuracy as you're producing a lot of these, you want to move towards more electronic methods. However, the computer doesn't necessarily know what's a defect and what's not. So you have to be able to teach it to look for certain things and ignore others.
So this is just an example of one of those porous transport layers. You can see there's not much to see when you look at it kind of far away. But an operator could find maybe an inclusion or something like that with the bare eye. This is a computer image.
Again, you can't necessarily tell too much until you colorize it. And then you start to see differences in the surface. And then finally, actually trying to qualify these things-- so as I mentioned, we can't test things for 7 to 10 years, but we still to test them for some period of time at scale before we actually start putting product out into the field. And so when we're doing screening tests, we can get through these pretty quickly. It's a fast setup, relatively low dollars to build. But when you get to the point that you're actually going to test a full-scale stack for, say, 5,000 hours, which is about eight months, it takes days just to build all the components to put together.
It's hundreds of thousands of dollars to put together. And it's also hundreds of thousands of dollars to operate because this is-- if you're talking about a megawatt stack that you want to run for 5,000 hours, even at industrial electricity rates of, say, $0.08 a kilowatt hour, that's about $400,000 in electricity alone. So we do work with the National Labs to try to test these. But it just points out the importance of understanding your materials back at this level and trying to be able to predict things with accelerated stress tests et cetera and doing detailed characterization and mechanistic studies before you get to this point so that you're pretty confident this is going to work.
So finally, what does this really mean for new invention? I gave a similar talk at UConn last week. And one of the questions was, OK, so what should university people do then? Because you're telling me all this stuff needs to be understood, and that's what industry does. And that's not really the point. There's still a large need for academics to do, to look at new ideas, new research, start developing things. But I think what I wanted to point out through all of this is that there are-- this is a schematic from DOE that shows the innovation cycle. This particular one shows an example where the industry is starting with the idea.
But this could also be academia or National Labs starting at this level and transitioning it over. But the point is that there can be multiple valleys of death. So there's the technology one. But there's also a commercialization one, which is largely what I've been talking about.
And then there are also maybe a more minor deployment one. And so we need to think about that as we're developing things and have industry and academia work together to bridge this. So this is another schematic I found online, showing that in academia we typically think about people working in these low TRL levels, developing knowledge, starting to develop technology. Industry tends to work more in the higher-TRL levels, at least when they're commercializing product. And the question is, how do we form that bridge so that this is a continuum and we're helping each other get across this valley? I've seen certain universities, where we get a call and somebody has a great idea, but they're not willing to work with us. It's just like, no, I want you to just license this, and then it's just engineering.
You guys can figure that out. Well, that hands-off approach doesn't work. We still need characterization and understanding in this middle part for that to be successful. So in conclusion, hopefully I've convinced you that electrolysis really needs to grow rapidly over the next few years.
Transitioning results from the lab to a product is a complex but feasible process. We need to think about things like how the tools, methods, and parameters are changing from the bench scale to the larger scale so that we can make that transition smoother. And we need characterization and testing all along the way to qualify new designs.
And that requires a lot of fundamentals. We've been doing a lot more looking at step-by-step through our processes, looking at all the underlying physics and chemistry around these, rather than relying on empirical data to make sure that we can innovate faster. And so, as I mentioned at the beginning, we're expanding our Wallingford facility to about 500 megawatts.
We also just announced-- this is Governor Whitmer with our CEO-- that we're going to be building a new plant in Michigan that will heavily leverage our strategic collaboration with GM. So I'm happy to answer questions. I could also talk about benchmarking if we want. But maybe we'll start with questions and see where we get.
Thank you. [APPLAUSE] Thank you Kathy for this fantastic talk. We have time for questions. [INAUDIBLE]? You said that the average expected stack life is like 7 to 10 years.
And we know that iridium supply is seriously in demand, especially if you need to scale the industry by 800 times. So I'm just wondering if people are thinking about end-of-life solutions for the stack and what happens to it. Can you recover the iridium oxide and use it again to build a new stack? I would say we wish we could do it that directly. But we do recycle our catalysts.
Typically it's a much longer life cycle. So we send out-- we ask our customers to return our stacks at end-of-life. And they get a credit for doing that when they buy a replacement stack. But then we're basically sending out our components that have precious metal on them to a third party recycler.
And then we have to buy new ore. DOE is looking at funding a fairly large consortium to look at that in more depth, and are there ways to make that a shorter lifespan. So, yes. We need to figure out a more direct way of recycling. We're also looking at how do we reduce the amount of iridium loading so that we can make that supply last longer as well.
Great. So if i may, building upon this question, so you showed this trajectory of PEM electrolysers in the coming decade or more. And we're currently already using a significant fraction of iridium supply, global supply, in this industry. Right so then how do we bridge the gap between, really, the global production or, really, the supply chain sustainability versus the growth of electrolyser technology based on-- So first, I would say I don't think electrolysers actually are using that much of the global iridium supply.
I think it's largely going to things like catalytic converters, some LEDs, crucibles for electronics, things like that. So there's, I'd say, the number I've heard is there is, like, 8 gigawatts or so of iridium available for electrolysers, which is not enough to meet these needs, but it's more than we're making today. Sorry, what was the second part of the question? [LAUGHS] How do we deal with that? How do we deal with that because the global supply of iridium production is at 4 or 8 gigaton per year. And if we consider all the projected electrolyser technologies, we're looking at something like 120 gigaton per year. So we need to get the loading down by at least a factor of 10 to start with. We need to recycle.
And then I do think that there are other applications, like catalytic converters, which will hopefully reduce in need so that the electrolysers can take more of that. Corky Mittelsteadt at Plug actually has a really good talk on this whole subject around looking at the ebbs and flows of iridium today. And if we can get down by a factor of 10, and we recycle, and we do all these things, we're still probably at about-- once this is implemented and evens out-- at about 20% of the iridium supply.
So we need to be able to hold on to 20% of the market. That's where we need to be. OK. I encourage maybe a catalyzing conversation between you and Hubert Gasteiger.
[LAUGHS] Great. Great. And then two related questions-- so there is just huge growth for electrolyser in the past decade.
And how much the price of iridium has increased over the same period? So it was actually very stable until about 2020. And then a couple of things happened. One, I think all of the announcements about electrolyser capacity influenced it. But also, there was a major smelter explosion, which reduced the supply by about half.
So that caused about a 6 to 8 times increase in the cost of iridium. That was starting to level out, and then the war in Ukraine happened, and precious metals are at a premium again. So we are assuming that we're never going to go back to those low levels but that we will level out at somewhere in between the highest it got during this period and that level.
Thank you. So-- great talk, by the way. One question I had was around the breakdown of different types of technology, then how you see them dominating in the next several years.
It sounds like, with some of these questions, such as the iridium supply, for example, it seems to me like alkaline or other technologies which don't [INAUDIBLE] might have a better chance. So if I were to ask you to look into a crystal ball, if you will, at what point do you see [? things ?] really starting to dominate in the market. Or maybe it's not-- like, what technology would you place your money on? I mean, I kind of have to be politically correct, since we make two technologies. But in all honesty, I think that there will be a mix, not only of those two but also, if anion exchange membranes get to a point that are stable enough, I think a lot of the PEM stacks will move to that technology. And there's also solid oxide.
So I think there's going to be not a one-size-fits-all solution. If you are at a hydropower plant where you have nice, steady electricity, and you're trying to off-take that with hydrogen, and you have a ton of land, alkaline electrolysis is perfectly fine. There's no reason to put a bunch of precious metal in your stacks. If you are offshore on a wind platform, then you probably want something that handles dynamic loads very easily and is very compact, like PEM.
If you're at a nuclear facility and you have a ton of excess heat, maybe solid oxide is the way to go. So I think it's very application-dependent. And we need to look at how do we bring all of these technologies up to the point that they need to be to address this need.
And that's a really great question. So for example, if you want to use hydrogen for aviation, so really, it has a different requirement in terms of high-power fuel cells, [? so it ?] might be also sort of application function-specific for electrolyser technology as well. Other questions? Have you seen a lot of different types of catalysts, like, [? skipping ?] iridium? And what's your viewpoint on that? So it depends.
I would say for PEM, it's pretty hard to get away from iridium, although Corky would tell you, yeah, but you could always use platinum. Of course, that gives you about a 300 millivolt penalty, which again, with electricity being the dominant cost, you probably don't want to do that. However, even ruthenium, under some circumstances, can be stable. So you could potentially mix those. Iridium-ruthenium blends under some circumstances and at higher efficiencies that we're at today, if we can keep the voltage down, might be stable enough-- mixing in platinum. I think going to totally non-noble metals for an acidic solution is quite a challenge.
I think you're more likely to develop a membrane on the alkaline side that allows you to go to the nickels and irons and things like that on the OAR side. That would be an interesting project for a graduate student. [LAUGHTER] Other questions? Thanks for your excellent presentation. And I have-- I have two questions. First thing is, I'm more familiar with alkaline, so I'm curious about the scaling bottlenecks of PEM, one single PEM stack, except for the manufacture of electrode and the membrane, and what size do you think may be the [? selling ?] capacity of one single PEM stack. This is the first.
And the second is, have you ever tested high varying load conditions of megawatt scale PEM stack. And if yes, how about the performance [INAUDIBLE]?? So for the first question, which was how big do I think the stacks will ever get? I think megawatt is actually a good size. You can design the balance of plant fairly well-- I mean, even chloralkali plants today, again, 100-plus-year-old technology, typically the stacks are 1, 2 megawatts, and they just put a lot of them in a plant together.
It's the fact that you can combine all the pressure vessels and things like that that gives you the lower cost and the balance of plant. If you get too much bigger than that, as I mentioned, even with our 1,600 square centimeter stack, it gets harder to handle the components. These are very-- we want to go thinner than we even are already, so they're not self-supporting anymore.
They can kind of break under their own weight. And so you have either multiple touch points, multiple operators assembling the stacks, even where we are today. So going even bigger is maybe not the most effective use of resources. In terms of dynamic operation, we haven't done too much with the megawatt stack particularly.
But the form factor and the whole stackup of the components is very, very similar to our smaller platforms, where we have done a lot of dynamic load. At the loadings we're at today, we don't really see any impact. NREL's done significant operation on our stacks that show not much higher degradation. I think the question is going to be, as we take the catalyst loading from where it is today to a tenth of that, are we going to start to see that? So that's, again, why we need to figure out, how do we simulate that, how do we test that, what is changing in the mechanism that would make it degrade or not degrade? Thank you for an excellent presentation.
Europe has announced more than 300 gigawatts of offshore wind. And there are now multiple projects of 10 gigawatts hydrogen production announced around the North Sea. And there's a lot of heat being produced with this hydrogen. What do you see, from your perspective, on this ability of supplying this heat into other areas, the temperature amounts, and maybe the tradeoff between producing the hydrogen versus maybe supplying high-quality heat from the [INAUDIBLE] independent electrolysers? It's not an area of my expertise, so I'll say that up front. Typically, our systems, the heat is fairly low grade because we're running at 60c or so.
And the stoichiometry of the water that's going into the stack is much, much higher than what's needed for a reaction. So it's actually acting as the cooling fluid for the stack. So most of our heat is just kind of escaping to the ambient atmosphere today.
I do know that there are some studies, Brookhaven included, that are trying to look at how would you use that better, either for city water heating or some other thing that doesn't require huge amounts of heat. But I don't a lot about how feasible. It is at this point. Over there, please. I thank the presentation. I have two questions, actually.
The first question is that you mentioned the use of automation and the project in Norway. I wanted to ask you, how much was that impact the CapEx eventually. And depending on how many years you have been operating it, what is the cost reduction already seen with it? The second question is that you mentioned partnerships.
And I'm just curious, like, for the partnerships which [INAUDIBLE] already looking into, are you considering industrial clusters, or are you seeing the problem of demand and [INAUDIBLE] and how are you tackling that idea? So in terms of the Heroya plant, it's been in operation for a little bit over a year. And I would say the CapEx came down by about half. But again, that's not just the fact that you're automating processes. It's the fact that you're changing the process that you're automating.
So it's a combination of looking at what you're doing today and figuring out how-- what's a better way to do that that would be automatable. And so it's a combination of those two things. But it is a quite substantial cost reduction.
In terms of partnerships, so as I mentioned, we have the partnership with GM. That's largely a very technical one, where we're taking what we believe is one of the best electrolyser stack manufacturers and one of the best PEM stack manufacturers and figuring out-- or fuel cell manufacturers-- and figuring out how can we learn from each other. We also are looking at-- I mean, we have partnerships with NREL to test our stacks. We have a an electrolyser at the ARIES campus there as well as working with them on things, like we're part of the Roll-to-Roll Consortium, looking at how do we improve roll-to-roll coatings.
We partner with different universities. We partner with different engineering and procurement companies. Because we're a relatively small company, we don't necessarily believe we are the right people to do a full project. We'll supply the electrolyser, but maybe somebody else is integrating it into the plant or building all of the infrastructure around it.
So companies like Wood, Black and Veatch, and others, we've been in contact with. So it's a whole range of things. But partnership is very important, especially for a company our size.
Two questions-- I mean, can you explain the business model of Nel, and how do you monetize as you scale up the technology? So I would say we view ourselves as an equipment supplier. And so there's been questions about, do you basically continue to own the equipment and act as just the owner operator, or do you sell the equipment? And typically we're selling the equipment. So we do not sell hydrogen, we sell electrolysers. So that's basically our philosophy. What was the second part of the question? To scale up, how you monetize because, I mean, when the technology is new, I mean, you price accordingly, according to that. So what exactly are you looking at for each of the steps and how you can get the best out of that? So as I mentioned in the beginning, we've kind of seen that at every price point of hydrogen there's been an affordable way to make that with an electrolyser.
So at smaller scale, it was quite easy to-- for example, we pretty much have the market cornered on weather balloon filling. So we sell electrolysers to remote islands because it would be extremely expensive to get either helium or hydrogen there. So we can make decent margin on those products.
Same thing with as we scale to the power plant size. That's a very small cost for them in order to get more efficiency by cooling their generators with hydrogen. And so we've pretty much been able to reduce the cost of the unit to the price point that it needs to be, as we've gone to the next level of hydrogen generation.
And that's what we're trying to do with the energy markets as well. So you have seen that in fuel cells, also in the electrolysers, over the past two decades, just tremendous development growth, scaleup. And so if we want to accelerate the pace, how we develop the technologies, make them more viable so we can use them to decarbonize, currently what is really limiting us? Is it materials? Is it capital? Is it people, workforce? Or-- Yes. [LAUGHTER] So let's say, what venture can help or what-- should we have more people? So what is limiting us? What can we do to accelerate the pace? So capital is a big one, I would say. I mean, if you look at all those companies I mentioned at the beginning-- Proton Energy Systems, Hydrogenics, Giner, I would say the general population has never heard of any of those companies.
They were all small. We were developing and we have smart people, but we're developing at the pace we can develop. We're not GM, we're not Toyota, we're not Honda. You've heard of those companies, and they've put billions into the fuel cell development.
I don't think we need to put as much investment in electrolysers because we can use their learnings, but there does need to be a substantial amount of capital put into research, testing infrastructure, all of those things. And we also need the workforce development so that we have the smart scientists who want to come work in hydrogen and figure out things like what's the best polymer, what's the best way to make a catalyst electrode, and what structure should it look like, what's the mechanical design need to look like for the stack, all of those things. Thank you. Thank you so much, Kathy, for this fantastic seminar.