Frank Talk About Hydrogen Electrolysis

One of the areas we are looking very closely at is Hydrogen.  I thought I’d share some of the perspectives and open questions we’ve gathered. To start, I’ve been looking at hydrogen a long time. Been waiting for my hydrogen fuel celled Ford F150 for two decades. Still waiting. Back in the day – defined as 18 years ago or so, I headed the commercialization efforts at Lynntech Labs, one of top electrochemical R&D shops in the country. We spun out a startup test & measurement for fuel cells, electrolyzers and catalysts, Fideris, whose products are still on the market today. We came really, really close on a stealth startup code-named XPL Power, using regenerative PEM electrolyzers and fuel cells to make grid connected long term back up power systems for telecommunications and military applications, working with Flextronics, which launched its cleantech and fuel cell effort around this project. Regenerative systems are two way fuel cells and electrolyzers – run forward they electrolyze hydrogen from water, run backwards, they make electricity from hydrogen. Coolest technology imaginable. The Lynntech team had developed them for aerospace and defense applications. We did not spinout the project, despite having what we believed was the cheapest, efficient, and most reliable PEM stack technology in the market and full scale stack prototypes running. And having customers. And having solved manufacturing. The issue was storage and balance of system. We couldn’t get a technical solution for BOS and H2 storage at the time that made technical and economic sense in an actual product that our customers wanted to buy.

Despite all the talk about hydrogen, we should level set.  Making hydrogen is not an issue.  Making it cheap is not an issue.  Making it at scale is not an issue. We can reform low cost high volume hydrogen from natural gas and hydrocarbons all day long.

So what’s the issue?

Making it cheap, available, distributed, and Co2 friendly. And finding something fun and profitable to do with it that isn’t better done with something else.

The issues underneath have been manifold. Well, a history lesson may be in order. Back in the day we had the Hydrogen Economy, with billions spent in R&D, and headlined by a Republican energy President, George W Bush. The talk was not just about carbon and climate but also energy efficiency, energy security, fuel cells for cars, and distributed power. Which meant scaling H2 production down at point of use, or transporting it. The main path for making hydrogen today is the same as then – reforming hydrocarbon feedstock. That doesn’t scale down well economically for point of use (everybody and their brother tried – and while still a few pathways of interest, it’s hard technically, and easier to do inside a high temperature fuel cell itself). And transporting means better tanks, or hydrides, or hydrogen pipelines, and runs smack into issues like hydrogen does not like to be stored, isn’t particularly energy dense compared to hydrocarbons, and is an escape artist and tears up steel equipment. Then natural gas prices spiked and wrecked business cases. None of the PEM fuel cell products reached their cost /life targets well enough so demand never materialized, the same issues we saw on balance of system and storage bit everyone, and the industry for transport moved on to be electric like Tesla, while the industry focused on stationary distributed power got outcompeted by solar on cost and scale. Oh, and reforming means hydrocarbon based feedstock like natural gas, which is not as popular in energy transition circles (they color code it and discriminate against it now as blue hydrogen instead of the sexy cool kid green hydrogen), and still makes CO2, albeit a fairly easy to capture already separated stream, which has to be dealt with for carbon goals.

But there were two competing paths to make hydrogen, reforming hydrocarbons, and electrolyzing water. The second, electrolyzers, is hot again and was always a great potential solution to those issues if you wanted CO2 friendly point of use hydrogen. And investors are now pouring money at levels unheard of during the Hydrogen Economy era. The issues with electrolyzers however, still have to be addressed.

Cost. Alkaline electrolyzers have been commercially available for decades, and PEM versions for 10-20 years now. We make hydrogen on site, we send them into space. But they haven’t been particularly cheap. Part of the issue with alkaline is they just didn’t have much pathway to get cheaper because they’re big. And cost is often weight. And PEMs struggled to get cheaper because they’re finicky, and have expensive materials. Of course the other issue was their main opex input is electricity, which was traditionally more expensive per energy unit than gas, as gas was the marginal fuel cost for electricity. At high electricity costs, electrolyzers are at best a niche.

Scale. Electrochemical devices like fuel cells, PV, batteries and electrolyzers tend not to scale up well. Or more precisely, they tend to work better modularized, and so scale up a bit more linear, depending on mass manufacturing volume growth over time to bring down unit costs, so at any point in time, a large scale unit and small scale unit don’t cost a heck of a lot different, compared to process technology in refineries. And as we learned from batteries and solar, when you’re playing with a few tens or hundreds of MW/year production, you really can’t get costs out.

Reliability. To make money, they need to last. And run. In a real operating environment. To match a needed application. For alkaline, life and performance were somewhat known commodities, for PEM a decade ago, we just didn’t have the components, or life times yet. The run thing is a bit of a double edged sword. Simple math. If they don’t run 247, they don’t make hydrogen. The less hydrogen they make, the higher the unit cost of the hydrogen is. If they don’t last, the higher the cost per unit is. Now part of the cost, reliability curve is a function of the membrane, and we were always in search of a better, cheaper, reliable membrane, and the stack and balance of systems to support it, each with their own reliability curves.

Operating range. Electrolyzers traditionally have been big, or finicky. And don’t like to be told what to do. See below, operating range issues abound.

Storage. The other issue is storage. Once you make hydrogen, you’ve got to store it somewhere, or have it on demand. Storage costs money and energy. And the combination of stack, storage and operating range needs dramatically impact the cost and performance of the balance of system and stack.

5) What the hell do you do with it? While we love the idea of hydrogen, hydrogen is not an end product in most applications. It is a fuel for power, that needs a fuel cell or an engine designed for it. Or a carrier of energy or building block of chemicals or feedstock for products. But it’s a carrier and feedstock that doesn’t really like to be stored or shipped. The use cases are highly sensitive to price, and location, and duty cycle. The best scale and location to use it aren’t necessarily at the best scale and location to make it. And until the green hydrogen revolution and problem of what to do with low cost marginal renewables, and how to decarbonize things without gas came along, electrolyzers were largely left to play second fiddle to reformers, useful only if you had a PEM fuel cell and no gas. Such a chicken and egg conundrum we faced two decades ago, and why at the time electrolyzers were the proverbially red headed step child of fuel cells and waiting on better storage. The number of times I’ve been asked recently, what are the applications for green hydrogen by people that should probably be the ones answering that question? A bit headscratching.

What’s new?

Well, a few things:

Free Power. OK, not quite free, but renewables are orders of magnitude cheaper now. And in a world where the marginal power costs are zero, and average costs are low, electrolysis is now a very different game. You couldn’t fight both both high capital costs and high power costs. Today, interest rates are on the floor, power costs are on the floor, and beating down capital costs completes the triangle on the $/kg of hydrogen from electrolysis.

Life and Performance. PEMs and alkalines both have large and old installed bases now. The body of field and life data is just fundamentally better. Yes, you can probably get enough life out of them to build a business case with a decent amortization curve. Not something you could bet on back then. I recall sitting at a presentation by Mike Binder, who owned the DOD’s fuel cell fleet back in the day, on field performance. He had a few of everybody’s units. DOD was the biggest customer in the market. And he showed off his actual real life honest to goodness field performance data. All of it. It was awful. Not like close, but like your beta doesn’t work, dog doesn’t hunt awful. So far off manufacturer’s claims it wasn’t even bothering reconciling. With a myriad of failure mechanisms. To be fair, I’ve not seen a Mike Binder Come to Jesus equivalent for electrolyzers recently.

Gig Scale. This may be the gamechanger. 18 years ago we thought in MW terms. Now, on the back of solar and lithium ion batteries, we think in GW terms. There may simply be no commercial pathway for electrolyzers without GW scale production facilities. But while people are now talking GW terms. the biggest actual installs are still measured in MW, akin to utility scale solar c. 2005/6.

Blue v Green Debate. This debate needs its own article. Suffice it to say the buzz and funding around taking the free power above an using electrolysis to split water into hydrogen is out of sight right now. The real push around electrolyzers is they are far and away the easiest and closest path to making hydrogen without making greenhouse gas emissions – if powered by carbon free power (renewables or nuclear). And that has breathed new life into the Green Dragon of Electrolysis.

Tech. There are a few new widgets out there. One example is AEM membranes. Anion exchange membrane. A variant on the old alkaline FC and electrolyzer systems, started to get significant work done about a decade ago. A number of corporate and university labs and startups have been working on the area. Evonik, W7, Ionomr, Origen, Alchemr (the last two were in our Cleantech.org prize competition, Origen making it to the semi-finals). Basic claim for AEM is PEM sizes and cost curve potential, with alkaline reliability. We’ll see. Super interesting. But this tech is awfully young.

What’s not new?

Upfront Cost. Electrolyzer stack costs still pretty high. A quick look at a couple of recent installation announcements indicates best available PEM systems will still run you $2-$3k/nominal kw installed, an order of magnitude off where they need to be. And the Alkaline versions which have been fairly reliable, are awfully large, and might still cost in the $1K/ range to buy. NREL did a really good technical and cost comparison a couple of years ago. Of course, these are lagging, not leading indicators. Is a next gen platform like AEM needed? Maybe. I will say. One of my former scientists made the offhand comment to me recently that the conventional membrane technology could probably come down and order of magnitude in price at scale. And we are probably well beyond the day when bespoke components from test systems to inverters to gas and control systems were needed. Will gig scale demand bring these costs down enough?

Flexible operating envelope. Both alkaline and PEM still have the same basic operating limits they did back then. Alkalines don’t love being turned on and off and up and down. They’re still big. PEMs need everything just right to last, which is easier to do if they cost more, have more sophisticated balance of systems, and are babied. To quote the founder of Lynntech back in the day: “take a PEM system, if you can control humidity and temperature perfectly, it could run forever, but you can’t, so it won’t”. If you need on demand hydrogen, or you’re making 100% green hydrogen, you must use 100% green electrons, you prefer a flexible stack technology, small balance of system, and on demand low cost green electrons, and a low cost storage system. Of course, this feeds back to cost, as you make money on the capital and install costs running flat out, but your application usually doesn’t need a steady stream at the same pressure and volume that your electrolyzer likes to produce, and your power costs are your only major opex, and those, especially your green power costs, are volatile and cost money to smooth. The constraints model is still a bit of a unicorn.

Storage. Still haven’t seen anything new. Better tanks? Sure. Does that matter? Not sure. Compression costs money, and still getting H2 in and out of a tank or a hydride storage media costs energy, equipment, and money.

Nafion. OMG. 20 years later and we’re still looking for a Nafion killer better membrane? Just shoot me now.

Bottom line, we’ve been able to do all of this for decades. You just had to pay up for it. Now we’re dusting off old business plans left and right. So here are the final “so what” questions:

Which customers are willing to pay for green hydrogen now and how much will they pay? At today’s renewable power costs, can the current alkaline or PEM electrolyzer technology get the per unit costs down with just gig scale manufacturing like PV and lithium ion batteries did? Is there something else new interesting in hydrogen production that can change the equation and outrace that coming scale, or are the new ideas just the thin film and cellulosic ethanol equivalent plays of today? Do the issues including storage that held it back before still matter, or will they too fade before the gig scale equation?

For further reading:

Really neat startup, Altroleum, that applied to our competition, and publishes an index of hydrogen production costs.

The best primer on hydrogen is probably still the DOE Hydrogen Road Map.

Or check out the library of the hundreds of articles on hydrogen from the Cleantech.org research library from Cleantech Blog and the UFTO Notes multi-client study.