Thermodynamics of Synthetic Fuels

Hydrocarbons fueled industrial civilization to the abundance of the modern age. Can we make them cheaply without killing ourselves? Yes. Here's how.

There is no single greater reason for life in the modern world being better than the feudal era than humanity’s ability to harness hydrocarbons. The scale of civilization and prosperity of its citizens depends on the abundance and availability of energy more than any other commodity, and the industrial revolution was made possible by harnessing the energy stored in hydrocarbons to allow industry to flourish. In short, there are no energy-poor rich countries, a statement which justified putting in one of my favorite plots of all time.

There’s a downside, however: burning hydrocarbons releases CO2, and the scientific consensus is that excess CO2 emissions is driving climate change and presents a peril to humanity.

What if there was a way to produce hydrocarbons in a way that doesn’t release any new net CO2 into the atmosphere? What if instead, we use the atmosphere as a giant carbon battery - sucking out CO2, producing hydrocarbons, burning them for energy, releasing them back into the atmosphere where they are re-distributed after which they can be captured once again and turned into my hydrocarbons, indefinitely? This is exactly the promise of synthetic fuels, and what I’ll explain in detail.

Currently only 20% of the world’s final energy is delivered in the form of electricity with the remainder directly fuelled by hydrocarbons like oil, gas, and coal. In 1990 the split was more like 15% to 85%, so while electrification has progressed it has since 2020 largely plateaued at 20%. We are still a hydrocarbon society living in a hydrocarbon age.

Even of all the electricity produced the majority still comes from hydrocarbons. In 1990 this was around 65% and in 2023 it is now 60%. Wind and solar have grown quickly but only add 13% of total electricity generation, and nuclear and hydropower another 25%. Despite humanity’s efforts to decarbonize there are a few big advantages to hydrocarbons that will keep them relevant for many years to come:

1. Power Density: industrial vehicles like ships, airplanes, and trucks need megawatts of power on-demand and are sensitive to the weight of large battery systems. Electric passenger vehicle cars make sense, but we are decades away from a battery-powered long-distance airplane.

2. Storage: Hydrocarbons can sit idle for years without appreciable losses in the energy stored, in contrast to batteries which naturally self-discharge

3. Transport: Losses from moving electricity through the grid over long distances nets out to 10-15% of all the electricity produced, while pushing natural gas through a pipe is 99% efficient.

4. Ease of Use and Implementation: All of the above points would hold equally true for nuclear fuels, which are about a million times more energy dense than hydrocarbons, but using nuclear fuels is extremely expensive due to cooling, radiative shielding, and regulatory overhead, driving nuclear power production mostly into large-scale multi-reactor sites. This may change over time with regulatory adjustments in the US, however nuclear powered vehicles are an unlikely remote possibility until we crack electrostatic inertial confinement fusion.

A simpler way to put this is: there will never be a battery powered supersonic jet fighter. Ever. If you want huge on-demand power with minimal weight, from a source that can be transported with minimal losses and stored indefinitely - you need hydrocarbons.

Of all the hydrocarbons humanity uses there is one that is head and shoulders above the rest when it comes to amenability to synthetic production, and also wins out over most others in terms of energy density: methane, or CH4, in its natural form known as natural gas. Carbon and hydrogen are extremely abundant elements on Earth and readily available in the atmosphere, taking carbon from CO2 and hydrogen from H2O. How’s it made?

Energy Production Pathways to Methane

The usual process for producing methane is via CO2 hydrogenation via the Sabatier reaction, or less commonly via CO methanation. The primary chemical equations are:

CO₂ Methanation (Sabatier reaction):
``CO₂ + 4 H₂ → CH₄ + 2 H₂O`` ΔH° ≈ –164 kJ/mol​

.

CO Methanation:
``CO + 3 H₂ → CH₄ + H₂O`` ΔH° ≈ –206 kJ/mol

The negative delta-H means these reactions release heat, and if you count the number of molecules on both sides you’ll see there are fewer products than reactants. This means these reactions are more favorable at lower temperatures and higher pressures which shifts the equilibrium towards CH4. In practice the usual catalyst is something like Nickel and CO is produced as an intermediary byproduct that then combines with H2 at high concentrations to produce methane. Above 500 C the reaction starts to proceed in reverse, and methane breaks down into CO and H2, and while the reaction will run spontaneously at room temperature with a nearly 100% conversion rate to methane, to get an industrially relevant output you need a catalyst like Nickel or another noble metal, and there is a threshold temperature of around 200C after which the reaction drastically increases. This means there’s an optimal reaction temperature of around 300-350C.

Unfortunately you can’t usually get your hydrogen for free, you have to breakdown water vapor and for the Sabatier process this means splitting 4 H2O into 4H2 which requires 967 kJ, then the Sabatier process kicks out 164 kJ for a net input of 803 kJ per mole of CH4. The net energy required to produce one mole of CH4 (803kJ) is therefore almost exactly the energy stored by one mole of CH4 and released when it's burned (802 kJ). This is why producing synthetic gas is using the atmosphere as a ‘carbon battery’ - the energy you put into methane synthesis is exactly the energy you get out when you burn it. You can also just phrase it as using methane itself as the battery - the energy that goes in is the energy that goes out.

An important thing to note is the Sabatier reaction which takes H2 and CO2 to produce CH4 or methane can achieve around 98% conversion per pass if designed well. Often a small slip of CO and H2 remains which can be recycled. The selectivity of this process for CO/CO2 is nearly 100%, meaning essentially all CO2 is converted directly into CH4, and unwanted carbon byproducts can be prevented from forming by running with an excess of H2 and keeping temperatures well controlled.

Of course, any conversion of energy between storage formats has associated losses, and no thermal engine is perfectly efficient (this is a hard rule of thermodynamics via the Carnot cycle). There are more than a few ways of producing H2 from water and reacting it with CO2 to produce methane, so let’s dive in to see what the tradeoffs are and if there is an ‘ideal’ method of synthetic gas production from thermodynamic first-principles. Any approach to CH4 generation has two three steps:

1. Hydrogen Gas Production: We need H2 of high purity and lots of it. There are many ways to get hydrogen. Splitting water into its constituent components of Oxygen and Hydrogen is called electrolysis and there’s a few methods: Proton-Exchange Membrane electrolysis, Alkaline electrolysis, high-temperature steam catalyzed electrolysis. All of these methods use electricity as part of their process. There’s also ways of chemically splitting water: Reverse-Water Gas Shift reaction (RWGS), Fischer-Tropsch which is electrolysis and partial RWGS, and coal gasification

2. CO2 capture: We also need to suck a lot of CO2 out of the atmosphere and capture it directly. There are three general approaches: cheap, horrendous materials like alkaline hydroxide or calcium carbonate that are not very energy efficient; moderately expensive, nice to handle materials that are moderately energy efficient like amine sorbents; and then really expensive, nice to handle and high-energy efficient materials like Metal-Oxide Frameworks or Covalent Organic Frameworks. (Most people go for the cheap, horrible to handle alkalines).

3. Mix em up into Methane: There’s really only one approach here and thats the Sabatier process. Good ol Sabatier.

Here’s the simple takeaway from the synthetic fuel production landscape in general: the primary variable factor among companies competing in this space is where they get their hydrogen gas from. Some people want to use solar panels and PEMs membrane or alkaline electrolysis, because they are betting on solar as being incredibly cheap in the future. Others are betting on using lots of heat, because heat is cheap from geothermal or fission sources, and mixing it either with electricity in heat-catalyzed electrolysis or using a heat-catalyzed chemical reaction to get H2. There’s a lot less variability in CO2 capture since most people use the same ‘materially cheap, horrible to handle, low-energy efficiency’ alkali hydroxides or calcium carbonates. But, there are better options on the horizon which I explain below.

So far we’ve covered the Sabatier process itself, which is how you mix CO2 and H2 to get sweet sweet hydrocarbons. Let’s first look at the trade-offs of different CO2 capture methods before getting into the far messier tangle of H2 gas production trade-offs.

Direct Air Capture of CO2

Fortunately for synthetic fuel production and plant growth, and unfortunate for the current consensus on global climate change the atmosphere is increasingly abundant in CO2. Capturing it into a concentrated or high purity form that can be used for synthetic fuel production typically means just running large fans that drive a huge volume of air through filters that contain CO2 absorbing compounds, for example alkali or alkaline-Earth hydroxides like potassium hydroxide, sodium hydroxide or calcium hydroxide. These are relatively inexpensive raw materials with well-understood chemistry, however they have high energy requirements to release the captured CO2 and are corrosive so require specialized handling and infrastructure. These are currently the most cost-effective solutions, but others exist: a more expensive but lower energy requirement would be solid adsorbents like amines and functionalized polymers. Polyethyleneimine, amino-functionalized resins, solid amine absorptions, all have higher up-front capital costs but lower operational costs in energy usage. Companies like Carbon Engineering use hydroxide-based aqueous solutions with a calcium carbonate cycle for cost-effective large-scale DAC, in contrast to Climeworks which uses solid amine-based sorbents that are better efficiency and use less energy and so have better long-term cost effectiveness.

Alkaline sorbents capture CO2 through ionic chemical reactions, so there is a distinct chemical reaction that takes place favorable under certain conditions which produces a carbonate or bicarbonate salt. To reverse the reaction and release the CO2 you just need to reverse the conditions of the chemical reaction by applying heat and the compounds that have been formed dissociate and the reaction proceeds in reverse. For example, using sodium hydroxide first you run the reaction:

CO2​ + 2NaOH → Na2​CO3​ + H2​O

To run the reaction in reverse you just take the N2CO3 and apply heat, dissociating it into Na2O and CO2:

Na2​CO3​ + heat → Na2​O + CO2​

Sounds great: you capture CO2 from the atmosphere using relatively abundant and cheap adsorbents, apply some heat, run the reaction in reverse and out pops your CO2 and the original alkaline sorbent can be recycled with the H2O released in the first step to form more NaOH, but there’s a catch - the compounds formed are via an ionic bond, meaning an electron is transferred from one atom to another, and so there is a fairly strong electrical attraction between them. This means the chemical bond is quite strong and so requires a lot of energy to dissociate: running the reaction in reverse requires temperatures anywhere from 100C - 900C with calcium carbonates requiring the highest temperatures.

The reason amine-based sorbents are cheaper to run in terms of energy is that the chemical bond formed between them is covalent, meaning two atoms share a single electron. A weaker bond means less energy to separate it, therefore the temperature required to dissociate the CO2-capturing compound is more like 80 - 120C. Isn’t that nice? The energy game is always a game of trade-offs: alkali compounds might be cheaper to produce but they’re corrosive and waste handling is tough and requires more energy. Amine sorbents aren’t corrosive, they’re more stable and so easier to handle, require less heat but then are more expensive to produce in the first place.

Is there a better approach for Direct Air Capture in the future? Yes, and one that looks promising although is still nascent and so not yet perfected - it’s still expensive but is both chemically stable, and so not corrosive or difficult to handle, and also has low-to-moderate energy requirements for regeneration. These are known as Metal-Organic or Covalent-Organic ‘Frameworks’ (MOFs and COFs). You can think of these like crystalline ‘foams’ with extremely high surface areas made by connecting metal ions or clusters, like zinc, copper, or aluminum with organic molecules called ‘linkers.’ Metal centers act like nodes or junctions and organic linkers connect these nodes into an overall structure. CO2 is adsorbed into pores in the ‘foam’ lattice by weak intermolecular attractive forces known as van der Waals forces, which emerge due to electromagnetic dipole attractions at a distance. These are not covalent or ionic chemical bonds but simply coulombic attractive forces more like static electricity than a chemical bond, so the adsorbed CO2 is relatively easy to release from a ‘saturated’ lattice by heating it up at around 90-120C or applying vacuum pressure. The downside is these are still pretty expensive to produce, but, if this technology can be brought to a better level of maturity it’ll win out over both amine sorbent and alkali compounds both for ease-of-use and low energy requirements. In short, MOFs and COFs are the ‘future solution’ in need of better refinement to bring the manufacturing cost down, at which point they are the clear win overall.

Okay, let’s summarize all this in a nice table so we can see the trade-offs and keep them in mind as we go through the different pathways to producing synthetic hydrocarbons:

Important Note: CO2 Capture from Industrial Sources

Direct air capture usually means building a big facility with tons of fans that drive massive volumes of air across alkaline or amine adsorbents to capture CO2. But, you might be thinking, the actual concentration of CO2 in the atmosphere is quite small, and most of the emissions come from concentrated locations, like factories with big smoke stacks. For example, one kilogram of CO2 is produced for every kilogram of cement, and cement plant waste exhaust is somewhere between 14-33% CO2 by volume; natural gas processing more like 50%, and processes like ammonia production up to 90% CO2 by volume. So, why put our expensive amine sorbent CO2 extractors out in the open where CO2 is at most 500 parts per million, or four orders of magnitude less concentration than a natural gas processing plant (0.0005% versus 50%).

For capturing CO2 directly at a point source like a large refinery or factory you end up with a cost of CO2 of around $0.02 - 0.05 cents per kg, for a high purity output like ammonia or natural gas. For a lower purity output like cement this ranges more like $0.03 - $0.12 cents per kg of CO2. Compare this to a Direct Air Capture facility sucking in regular atmosphere which gets a price of around $0.25 - $0.60 per kg of CO2.

Capturing CO2 at a point-source uses the exact same methods as a general DAC unit out in the open, but gets 10x reduction in price per kg. This is a huge win. The lower the concentration of CO2 in your airstream, the more sorbent/solvent you need to process the air volume, and the more energy consumed just running giant fans to drive those massive air volumes.

Net-Net, the cheapest CO2 is from alkaline-based sorbents at high-concentration CO2 point sources, like ammonia production facilities.

You Require more Hydrogen Gas

Like I mentioned earlier the biggest variability in approaches to producing synthetic fuels is where an approach gets their hydrogen from. This is a big tangled mess of different approaches but I’ll do my best to simplify them as much as possible, explain the chemistry and physics, and give some intuition for what the trade-offs are in terms of energy efficiency and economics. But first, another table:

PEM Electrolysis

Now we have our CO2. Great. We’re halfway there - all that’s left is getting some H2 and mixing it up over some nice and hot nickel catalysts and boom, we’re cooking with gas. Or rather, we’re cooking synthetic gas. The first approach we’ll consider is to use electricity to drive a Proton-Exchange Membrane (PEM) electrolysis unit and then combine CO2 concentrated out of the atmosphere by one of the methods listed above into a Sabatier reaction to produce CH4 and H2O, venting excess heat along the way. Hypothetically you can do this using any kind of electricity really, but for reasons I’ll explain later solar is rapidly approaching the cheapest per-watt electricity around and since you can transport synthetic hydrocarbons with extremely low-losses through an already enormous network of pipes designed explicitly to transport methane, we might as well take solar as the preferred electrical input.

The easiest way to understand how PEM works is actually just consider it as a hydrogen fuel cell running in reverse - H2 gas flows through tiny channels and reacts at an anode, usually containing platinum, which splits it up into constituent protons and electrons. Now we need to get those protons somewhere useful, and that’s where the Proton Exchange Membrane lives up to its name. The protons are drawn into the PEM made of a material like nafion which has hydrophilic sulfionic acid groups which only allow protons to pass through, and is insulating to electrons. It’s basically a “proton attractor” and “electron impeder” and the attraction is enough to strip the protons away from the attractive charge of the electrons. How does this actually work? It’s pretty neat - there are channels inside the nafion material that are filled with a single-file line of water molecules, and sulfionic acid in the material itself. What happens is, the sulfionic acid is negatively charged, and so attracts protons off the water molecules on one side of the channel and then passes them along the line of water molecules inside the channel bucket-brigade style in a process known as the Grotthuss mechanism. This ‘hopping’ mechanism proceeds fairly quickly and only lets protons through - the electrons are actually repelled by the negative sulfionic ions and so stay on one side. Think of this nafion material as a “proton sieve.”

If you just left it at that, the protons and electrons would ‘pile up’ until the voltage potential across the membrane equalled the attractive force of the sulfionic groups doing the splitting - but that would make a lame short-lived fuel cell. Instead, you flow oxygen through at the cathode side where it reacts with the accumulated hydrogen to form water, and then draw the electrons out through a wire as electrical current. Voila - feed in H2 gas which gets split into protons and electrons by a platinum catalyst, protons hop along the line water molecules inside a nafion channel drawn in by the negatively charged sulfionic groups, and on the other side they react with oxygen to form water, and on the other side you have an accumulation of electrons which you can collect as usable electric current.

For synthetic gas production a PEMs unit just works in reverse: you flow water in one side, and then the same PEM material like nafion draws the protons in. Instead of producing electrical energy, however, this now consumes electrical energy which is ‘used up’ splitting the water into hydrogen and and oxygen. The simple intuition here is that burning hydrogen in the presence of oxygen releases energy and produces water, this is actually why its called hydrogen - hydro + gen means ‘water producing,’ and so running this reaction in reverse naturally consumes energy.

The reaction taking place here is simply:

2H2​O → O2​ + 4H+ + 4e−

Since only the protons (H+) can pass through the membranes, you have O2 and electrons accumulating on one side. You collect up those electrons through an external circuit and then pass them to the other side where the 4H+ is accumulating, letting them recombine into our regular and friendly electrically neutral H2. The byproducts are O2 gas and H2 gas and all it takes is electricity to run.

Levelized Cost of Hydrogen using PEMS + PV

The way to get intuition on solar PV electrolysis for the cost of hydrogen production is this: it takes ~50 kWh of electricity per kg of H2 produced by an electrolyzer running at ~60% efficiency, so every additional cent per kilowatt hour adds 50 cents to the cost of a kg of hydrogen. If power costs 5 cents per kWh, the electrical cost is $2.5 per kg of hydrogen, and given the typical cost of electrolyzer of around $1,000/kW amortized over its lifetime gives another $1.2 - $2 per kg of H2. The net-net is, it's very difficult to get hydrogen down to less than $1/kg, even if the electricity was free. For now, this means PV-driven electrolysis of water into hydrogen gas keeps the cost per kilogram of H2 somewhere between $4 - $7 per kilogram where the variation has as much to do with geographic availability and cost of PV power as it does the scale of plant implementation. If you had large-scale solar with 3 cents per kWh it brings the levelized cost of hydrogen down to $3/kg.

PEMs are ideal to use with PV power since they can ramp up quickly and so operate flexibly during the day whenever the solar power is available. However, in terms of energy efficiency they can operate around 65 - 75% efficiency when it comes to converting electrical energy into the embodied energy of hydrogen gas. There has been one trial program to run PEM units off the Three Mile Island nuclear reactor, so people do pair them up with a stable baseload power source, but generally speaking PV is the cheapest-per-watt power around and the easy ramp-up of PEMs units make them a natural fit. PEM is a very mature technology and it looks like by 2030 it may overall be 30% cheaper; the downside is they require platinum-group metal catalysts which puts in a price floor.

Alkaline Electrolysis

PEM electrolysis is “nice” - it's a self-contained unit, its relatively modern technology, you just supply gas feedstocks and electricity and it delivers hydrogen gas at modest pressure and price, perfect for downstream reactions to produce methane. Alkaline electrolysis is the ‘cheap and dirty’ method - not dirty in the sense that it emits more CO2, but rather dirty in the sense that they use extremely concentrated (5 Molar, if you know what that means) alkaline materials like potassium hydroxide or sodium hydroxide which are extremely corrosive and toxic and so difficult to handle. Puncturing one of these units basically unleashes a pressurized stream of pure evil that will destroy anything it touches and eat through metal. Yeah, it's bad, but it's also the most mature and cheapest method of separating hydrogen gas from water using just electricity and operating at low temperatures.

The way it works is the standard ‘battery’ setup where you have an alkaline solution that conducts ions between two electrodes, the cathode and the anode. If you remember highschool chemistry, reduction occurs at the cathode and oxidation occurs at the anode - just like a PEM unit is a hydrogen fuel cell running in reverse, alkaline electrolysis is like a battery running in reverse. When you apply a voltage, negatively charged ions move towards the positive anode, and negative charged ions move towards the cathode. The reaction of interest is the reduction reaction at the cathode where water is split up into H2 gas and hydroxide, and on the other side you get that unwanted, unloved oxygen gas. In this setup the potassium and sodium ions aren’t actually consumed in any meaningful way, they just act as charge carriers and they are far, far cheaper than the more expensive platinum-group catalysts you need for PEM units.

Levelized Cost of Hydrogen using Alkaline Electrolysis

Akaline units are generally a lot cheaper than PEM, around $500 - $900 per kW versus $1000 - $1200 per kW. They usually run around 60-70% energy efficient which isn’t as high as PEM units, and more importantly for use with solar, they are not as good at ramping up and ramping down as electrical power availability varies with the daily solar cycle. In fact, a lot of designs require operating at least 20% for stable operation, usually to limit ‘bad’ chemical reactions occurring at the electrodes that degrade overall system performance and efficiency. On the plus side they can last a lot longer than PEM units, >90,000 hours versus 20,000 to 60,000. Similar to PEM units about half the levelized cost of hydrogen ends up being the cost of electricity supplied, although as electricity gets cheaper and cheaper alkaline ends up winning out over PEMs because the up-front capital costs is 30-50% cheaper, e.g. if you had $10 / MWh power available 24/7, the levelized cost of hydrogen would be around $0.75 / kg for alkaline units versus $1.12/kg for PEM operating at 100 MW scale.

The major downside with alkaline units versus PEM is issues with cycling them when connected to the cheapest source of power - PV solar. Basically you need to either keep them warm and keep the salt mixture circulating; thermally cycling the electrodes induces mechanical stress and loosens the seals, and also reduces conductivity of the electrolyte solution which reduces performance efficiency. Shutdowns can also accelerate electrode corrosion and form oxide-layers on the electrodes which also hurts efficiency. Basically the net-net is, alkaline electrolyzers are cheap, reliable, fairly efficient, good at being deployed at mass scale, filled with extremely concentrated corrosive liquid death, and want to be run at fairly constant operating conditions to stay nice and happy. Their unit economics are slightly better than PEM since they’re cheaper to mass-produce, but, because of their physical chemistry aren’t as good at operating with intermittent solar power, which is the cheapest electricity source by far, so a large-scale installation either has to have some on-site battery storage to keep them running (driving up costs) or else use a different source of grid power (also driving up costs).

So far we have one up-front moderately more expensive option for making hydrogen that’s ideal to use with the cheapest electrical energy, PEM, along with a moderately less expensive option that's not ideal for using with the cheapest electrical energy, Alkaline electrolysis. Look at all these trade-offs. Welcome to the world of energy, where the trade-offs never end. Shouldn’t there be a significantly more up-front expensive option that can use even less electrical energy, perhaps, by subsidizing it with the absolute rock-bottom cheapest energy of all, heat? Well let me tell you, there sure is, and it’s called Solid Oxide Cell Electrolysis.

High-Temperature Solid Oxide Cell Electrolysis

How do we use cheap heat energy to subsidize the electrical power it takes to split water into hydrogen and oxygen? To understand this we need to understand perhaps the single most important equation in all of thermodynamics, the Gibbs free energy equation which is:

ΔG = ΔH − TΔS

At standard operating temperatures the Gibbs free energy of the water electrolysis reaction is 285 kJ per mole, the enthalpy is 237 kJ per mole, and the entropy is 48 kJ per mole. The enthalpy is the energy associated with actually breaking the chemical bonds, so that doesn’t change, but the temperature certainly does! The positive or negative signs on the quantities G, H, and S help us understand if the reaction wants to proceed spontaneously or not: reactions that release heat (ΔH < 0) tend to proceed favorably, similarly if the entropy increases by a reaction (ΔS > 0) it’ll also tend to proceed favorably, meaning you don’t have to force it. Whether the reaction occurs spontaneously by itself, meaning no external energy input is required, is determined by whether ΔG < 0 - if Gibbs free energy is zero, it’ll happen on its own. If it's positive it takes energy so you’re rolling the ball up hill, so to speak.

Now, splitting water takes a lot of energy at normal room temperature because ΔG is a pretty big positive number: 285kJ per mole works out to 2.26 megajoules per kilogram. If we crank up the temperature term though, ΔG gets smaller and smaller - enthalpy doesn’t change since the energy required to split the chemical bonds remains constant. The hotter we get, the less energy it’ll take to break up water and release free hydrogen gas. If you go to 700 - 900 C, Gibbs free energy is negative 180 kJ / mole instead of positive 237 kJ per mole, and water will just break apart on its own.

This does take more raw energy as measured in kJ than electricity driven electrolysis, around 4 times as much (200 kWh per kg in heat compared to 50 kWh per kg in electricity) but, thermal energy is a lot cheaper than electrical energy: 2 - 3 cents per kWh compared to 5 - 7 cents per kWh of electricity. If you look at the math there you’ll quickly see, however, that we’re using 4 times as much energy that's only 2x cheaper - that’s just dumb. So, instead of using tons of raw heat by itself, the usual route here is to reach temperatures of around 700 - 900 C and then use some electrical energy, where the high temperatures are basically just ‘subsidizing’ the amount of electrical energy you need to use. The right balance of heat and electrical energy depends on the specifics of the chemistry you’re using to catalyze the reaction.

Solid Oxide Electrolysis Cells (SOEC) are a great example of this approach - taking relatively cheap and abundant thermal heat energy and using it to greatly reduce the electrical energy it takes to break apart water. In a sense they operate in an opposite fashion to PEM electrolysis because they conduct oxygen ions instead of protons, and since these operate at 700 - 900 C they use superheated steam rather than liquid water. The catalyst material in SOEC is usually yttria-stabilized zirconia (YSC) which acts as the ion-conducting layer, pulling O2- ions away from water molecules and liberating the H2 gas.

In an SOEC cell, steam enters into the cathode where oxygen is stripped away by the catalyst leaving behind H2 gas and free electrons.

Cathode Reaction:

H2​O + 2e− → H2​ + O2−

Then, the O2- is transported through the ceramic layers towards the anode, where it is recombined with the freed electrons to prevent the build up of electrical charge imbalance that would prevent the reaction from continuing.

Anode Reaction:

O2− → (½)​O2 ​+ 2e−

As I mentioned earlier, operating this at higher temperature reduces the electrical energy required and also increases the ionic conductivity of the oxygen-conducting ceramic. In contrast, in PEM membrane the electrolyte resistance is a major efficiency bottleneck. The net-net here is this: PEM electrolysis requires ~60 kWh / kg of H2 while SOEC requires only 30-40 kWh / kg of H2 and can operate at up to 80 - 90% overall energy efficiency, compared to the more conventional 60 - 70% PEM or alkaline electrolyzer efficiency. The downside is that SOEC materials are a lot more expensive and so cost around $2,000 - $3,000 / kW compared to PEM which are more like $1,200 per kW or $500 - $1,000 - so, you’re trading CAPEX for OPEX in that regard. On the plus side, the US Department of Energy is targeting $500 / kW for commercial-scale SOEC by 20230 - right now, SOEC are not really produced at large scale, but mass-manufacturing can both bring down system cost and improve operating lifetime to make this a far more competitive approach than PEM electrolysis, especially when you consider the availability of cheat reliable heat, either from fission nuclear plants, geothermal plants, or industrial waste heat processes.

Levelized Cost of Hydrogen using SOEC

Now for the brass tacks. The downside with SOEC is they’re more expensive up-front, as an added bonus they also last a lot fewer operating hours than either PEM or alkaline electrolyzers, only 20,000 - 30,000 hours. Operating at high temperatures is tough for any material and it shows. Right now there are no real large-scale SOEC units operating so it's hard to get a firm LOH cost, but the ballpark number is around $6 - 9 per kg of H2. Even in the limit of almost free energy this is a far cry from the PEM or Alkaline target costs of around $0.75 to $1.00 per kg of hydrogen. Since energy isn’t free, and in a lot of industrial applications there is waste heat that is almost free, and SOEC has 20-30% lower energy electrical requirements than other electrolyzer methods, if we can scale up SOEC to bring the cost per kW to the DoE expected future cost of $500 per kW, SOEC can knock it out of the park.

There’s a silver bullet for SOEC to redeem itself in the production of synthetic gas, however, beyond just waiting for production to scale up and bring about the juggernaut might of industry and its learning curve. Remember the goal isn’t just to get as much hydrogen as cheaply as possible, but use it to make methane. If you’re using PEM or alkaline you’re basically committed to getting your CO2 through a regular direct-air capture method, which is a whole separate unit with its own chemistry and power requirements.

The magic of SOEC units for making synthetic gas is it can produce CO and H2 at once by feeding it H2O and CO2 at the same time. This can directly feed into the Fischer-Tropsch process which produces not just methane but a host of other hydrocarbons as well. This is something entirely different from the Sabatier process used for methane production so I’ll describe it briefly here.

Fischer-Tropsch Process and Bottom-Up vs Top-Down Hydrocarbon Production

Recall at the beginning of this article I described there's two pathways to synthetic fuel production, the Sabatier process which takes CO2, and also a more general ‘methanation’ process that uses just CO. Methanation is the general term but the more specific process here is known as Fischer-Tropsch. Unlike the Sabatier process this produces more than just methane, but depending on the operating conditions and catalysts can produce a whole host of different long-chain hydrocarbons like diesel, gasoline, and jet fuel. The general equation just for reference is:

(2n+1)H₂ + nCO → Cn​H2(n+2)​ + nH₂O

Here the ‘n’ can just be any integer and the reaction will be balanced, and, you can see by increasing the ‘n’ you can get longer and longer-chain hydrocarbons. Methane is actually quite short as far as hydrocarbons go, just one C and 4 H, Ethane is C2H6, Propane is C3H8, Butane C4H10, and so on. Gasoline has somewhere between 5 to 12 carbon atoms, kerosene (jet fuel) somewhere between 9 to 16, diesel more like 12 to 20, lubricating oils 20 to 34 carbons, and so on. The longer the hydrocarbon chain is, the less flammable it is, the lower the vaporization pressure, and generally the less useful it is as fuel and gradually transitions into something that is stable at atmospheric pressure and used to grease the wheels.

Crude oil itself is a mixture of a huge number of different hydrocarbons where fractional distillation is used to separate out the different chain-length hydrocarbons by their boiling point: at the top of the distillation stack you have the lightest hydrocarbons and at the bottom, the heaviest. Breaking down large hydrocarbons into smaller ones to make usable fuels is therefore known as a ‘top down’ process.

In contrast, the Fischer-Tropsch is “bottoms-up” - you’re starting off with just raw CO and H2 and then building up heavier and heavier hydrocarbons. How this works is, CO and H2 molecules adsorb onto the surface of a catalyst like iron or cobalt, where the CO dissociates which lets the carbon bind with hydrogen. In this way hydrocarbons grow step-wise, adding one carbon at a time by repeated reactions of the surface carbon species with hydrogen. This process overall is exothermic, so releases heat, and the exact operating temperature and pressure determines whether you’re producing lighter or heavier hydrocarbons with low temperatures and high pressures producing longer hydrocarbon chains and vice-versa for shorter chains.

The sweet-spot when using the Fischer-Tropsch process with SOEC units is that they both run in the same temperature range of 200-300C.

So far we’ve covered PEM electrolysis, Alkaline electrolysis, SOEC and its connection to the Fischer-Tropsch process for producing longer-chain hydrocarbons. All of these are fairly ‘high-tech’ in some sense - they involve catalysts, specialized material design, separation of purified gas streams, and so on. Since this is the energy space where there’s tons of endless trade-offs, shouldn’t there also be some ridiculously simple, dirty, cheap, smooth-brained method of getting abundant H2? Perhaps using one of the dirtiest forms of energy known to man? Perhaps even the one thing that kicked off this whole industrial revolution anyway?

Well you bet there is, and it's called…

Coal Gassification

Coal has everything you’d look for in a fuel source: tons of hydrocarbons, ridiculously cheap and easy to mine, except for that part where it emits a ton of CO2. Unsurprisingly based on it chemistry we can very well extract H2 from coal, and along with it CO and CO2 as well - everything we need to make synthetic fuels via the Sabatier of Fisher-Tropsch process for longer hydrocarbons. The method here is simple: partially oxidize coal in a high-temperature, high pressure environment which you can think of like ‘partial combustion’ - its not completely burning, but it is dissociating into molecular compounds in the presence of oxygen.

The main reaction to start with is partial oxidation of coal

C + (½)​O2​ → CO (ΔH<0; exothermic)

Some of the leftover heat is then re-used in the next step called steam gasification, where we combine hot water steam and carbon to make CO and H2, also known as syngas:

C + H2​O → CO + H2 ​(ΔH≈+131 kJ/mol; endothermic)

Now, if we want to use the Fischer-Tropsch process we’re basically done there - we have H2, and CO, and from there can produce hydrocarbons. If we’re after CO2 for use in a Sabatier reaction we do one more step known as the Water-Gas Shift reaction to maximize hydrogen yield:

CO + H2​O → CO2 ​+ H2 ​(ΔH≈−41 kJ/mol; exothermic)

A well-designed coal gasification process uses the exothermic heat from oxidation to help run the downstream reactions and usually runs at around 1,200 - 1,600 C. If we just did the first two reactions and looking at the energy content of the syngas we’re running at around 60-70% energy efficiency; if include the third step we’re down to 50-60% efficiency.

If all we were after is the H2 gas then the production costs from coal gasification are staggeringly cheap compared to the previous methods: $1.0 - 1.5 per kg of H2. However, if our eventual goal was to both emit less net CO2 to the atmosphere and also capture CO2 to make synthetic fuels, we need carbon capture methods directly attached to this process. Unfortunately this pushes the cots up to around $2.0 - 3.0+ per kg of H2, depending on the CO2 sequestration chemistry used (e.g. alkaline versus amine sorbents), and to get towards the lower-end of that scale requires considerable up-front investments into a very large facility.

Summary of Thermodynamics of Synthetic Fuel Production

We’ve covered a lot of ground in the quest to squeeze sweet sweet hydrocarbon fuel out of the atmosphere and water, and all starting from first-principles and exploring the raw thermodynamics along the way. Let’s summarize these approaches to help them glue into our memory:

Low-Temperature Electrolysis + Sabatier

Think of these as overall the friendlist, cheapest-currently, and most modular approaches. They use relatively mature and well understood technology like alkaline or PEM electrolysis and direct air capture methods to get H2 out of water, and CO2 out of the atmosphere. In terms of raw energy efficiency these aren’t the overall best, around 50-60%

High Temperature SOEC + Sabatier

SOEC units are more expensive up-front, but, they let us subsidize the price of electricity - which ends up being about half the cost per kg of H2 using PEM or Alkaline electrolysis - using far cheaper heat energy. These run around 65-75% efficiency but even go up to 80% overall energy efficiency with careful heat integration, for example in a geothermal or fission implementation. The downside is the more complex materials used in SOEC are more expensive and have a shorter lifespan.

Other Approaches

Of course there is more than just those methods covered. Previously I mentioned the Water Gas Shift reaction where you CO and H2O and get CO2 and H2 - you can also run that reaction in reverse and then methanate the products to get CH4 and water. This is a lot less energy efficient than the methods mentioned above, but, can produce some pretty cheap methane at the end of the day.

Lets summarize everything we’ve learned so far in one giant table and include both the overall energy efficiency and the eventual cost per kg of CH4.

Table N: Another Tables of things, whatever (@andercot)

We can note the huge range in price between these methods: PEM Electrolysis + Sabatier, ideal for use with the absolute cheapest source of energy around: PV solar, at around $4-10/kg of CH4, all the way down to the cheapest, dirtiest, highest-emission coal gasification of around $2-4/kg. There’s a huge variability, and this document has glossed over most of the more relevant engineering design details and tradeoffs when actually designing and implementing a production facility. Most importantly is that the availability of key inputs - things like waste heat, or cheap land for PV solar, or point sources for CO2 capture like factories, and not to mention the regionally variable byzantine regulatory environment of subsidies, tax breaks, and so on, make for a very complicated energy landscape when it comes to synthetic fuels production.

The main takeaway is this: Synthetic Fuel production is a lot more sensitive to operating costs than capital costs, but currently the highest-efficiency electrolyzer methods which have the lowest opex, also have the highest capex.

With all of that in mind let's take a look at a few companies operating in this space and see what their specific design approach now that we’re equipped to make some well-informed technical commentary.

Companies in the Space

Syntholene (SOEC + Point Source Capture CO2 + Geothermal)

Dan Sutton’s Syntholene is relatively new on the scene of synthetic fuel production but they are adopting a fairly strong approach by going all-out to find the absolute cost-floor in terms of operating expenditure for the raw inputs. SOEC has the lowest energy usage, geothermal heat is the cheapest heat source by a large margin, and coupled with point-source capture CO2 also gets the cheapest CO2. SOEC requires temperatures of around 750-850 C and requires 20-30% less energy than Alkaline or PEMs electrolyzers by subsidizing it with much cheaper on-demand heat energy. Let’s work through some example numbers for a 100MW installation.

Electrolyzing water via SOEC follows this reaction:

H2O → H2 + (½)O2

Which takes ~285kJ/mole. The theoretical minimum electrical energy required at 850 C is 200 kJ per mole, and the remaining balance of energy can be taken up as heat for 85 kJ/mole. In terms of the more familiar kg thats 35 kWh / kg H2 of electricity and 15 kWh / kg of H2 as heat. So, a 100 MW SOEC facility operating at 80% efficiency would end up with 80 MW of energy going into this reaction, divided up to 65 MW of electrical energy and 15 MW of thermal energy. If you had geothermal heat integration you can use it for both the SOEC process as well as the CO2 capture regeneration, since if you recall the CO2 adsorbing chemicals can be re-used after releasing the captured CO2 by an application of heat.

Now for the CO2 production and capture - take a typical cement plant that might produce 400,000 tons per year, and for each ton of cement produced it releases 0.8 tons of CO2. Calcium carbonate CO2 capture takes around 1.5GJ / ton in the form of heat, so running all year long we need 32 MW continuous heat energy for capture of 400,000 tons of CO2.

Lets ballpark some numbers here both in capital expenditures and operational costs to guess at an eventual price for synthetic fuel, keeping in mind that current SOEC units run around $2,500 / kW:

Capital Costs

100 MW SOEC System with 80 MW H2 ~ $250 M

Geothermal Plant with 30 MW Thermal ~ $50 M

CO2 capture plant of 400,000 tons per year ~ $100 M

Methanation reactor ~ $6.2 M

Total Capex: ~ $ 400 M

Operating Costs:

Electricity capture for SOEC, 35 kWh per kg H2 for 100,000 kg per day @ $20 / MWh ~ $25 M / year

CO2 capture, $50 / ton and we need 2kg CO2 per kg H2 is $100 per ton of H2 ~ $ 4M / year

General maintenance and operating costs assume 4% of CAPEX per year, ~ $16 M / year

Total Opex ~ $ 45 M / year

If we assume a 20 year plant life and 10% discount rate we net out to a levelized cost of H2 at $2.3 / kg, which is pretty good, but then again consider this is a fairly large plant size and so should achieve good economies of scale. How does this net out to synthetic fuel production? It takes 2 kg of H2 per kg of CH4 (count the H’s). Methanation reactors cost around $500 - $1000 per kW and operate at 250 - 400 C, but since we have geothermal on-site we can basically assume we get that heat for free. Based on our rate of H2 production and that 1 MW of methanation capacity produces 250 kg of CH4 per hour, we need 8.3 MW of methanation.

We end up producing 18,250 tons of CH4 per year for a levelized cost of $5.50 per kg of CH4, or $5.28 per MMBtu, which is actually pretty close to fossil gas prices in the US of around $3 - 6 / MMBtu and fairly in line with the summary table above.

There’s a silver bullet waiting for SOEC, though, which is if the DoE projection is correct and the capital up front costs can be brought down to $500 / kW from $2500 / kW, we reduce the CAPEX of the SOEC units by 5x, and they are by far the largest line item in the list. Assuming everything else stays the same - cost of geothermal, electrical energy prices, etc, we get a price of $3.60 per kg CH4 or $3.45 / MMBtu which would actually undercut natural gas sources, especially in Europe. If electricity costs also drop to $10/MWh, methane goes to $3 / MMBtu.

This is the big advantage of the SOEC + Geothermal + Point-source capture - you are running the absolute minimum in terms of operational costs, and the primary capital cost is expected to get 5x cheaper in the years to come. If that pans out then this route to synthetic gas is directly competitive with fossil fuel sources. Now, an additional comment: Syntholene claimed in a direct communication they think they can actually bring their levelized cost of hydrogen down from $2.30 / kg to $1.00 / kg, which cuts the direct cost to CH4 by more than half. This would bring the cost of CH4 below natural sources, making it a clear market winner.

Terraform Industries (Solar PV + Alkaline Electrolysis)

Casey Handmar’s Terraform Industries is by all accounts an exceptional founder: great twitter presence, super active blog, reads ancient scrolls by eye, and most importantly publishes very transparent and well-written white papers describing the exact technical and economic tradeoffs used in his approach which can be found here. His premise for using solar power is simple: production of solar generating capacity in 2023 is 450 GW and increasing at 40-50% year over year. This is a phenomenal rate of growth, and the learning rate or decrease in production cost per every doubling in manufactured capacity has recently inflected from a long-run average of 23% to around 44% in the last few years. The net-net of that growth rate and learning rate is that installed solar cost per GWh of energy produced will be reduced by 2x every 2-3 years going forward into the future. The current trend means that overall costs could be 10x cheaper than the current cheapest costs without requiring any particular miracles in underlying technology beyond the steady improvements we already see.

This all sounds great - but is there a nicely illustrated diagram we can look at to remind us of how everything fits together? Yes. Yes there is.

How about an even bigger diagram with a futuristic looking city connecting to a methane distribution system powered by solar photovoltaic cells? Also yes.

Okay, enough illustrations. How about economics? Solar PV in the US costs around $1000 per kW to install; electrolyzer is around $900 - $1200 per kW. The Sabatier methanation reaction and balance of plant (compressors, heat exchangers) add cost but typically less than the electrolyzer itself, around a few hundred dollars per kW of CH4 output. For example, one study found ~$230 / kW for methanation in a large Swiss power-to-methane project. CO2 is a major upfront cost if you use direct air capture at around $500/ton CO2 equipment cost, but, an alternative is to source CO2 from cheaper industrial waste streams or else biogas. These are just commonly available numbers online, Terraform Industries actually has more aggressive targets.

Terraform is taking the approach of ‘brutally inefficient electrolyzer’ and so uses alkaline-based electrolyzer, which are cheaper than PEMs but less efficient which is okay - they have tons of dirt cheap solar power. They’re going all-out on solar panels in the desert, and the projected cost curve on solar panels does indeed make them seem like they’ll become brutally cheap in the near future. A ‘high-tech’ electrolyzer would operate at high pressures and temperatures to maximize efficiency; Terraform is going for atmospheric pressure and venting oxygen to the environment. Their long-term cost target is $20/kW which is a solid 60x below the usual market price, operating at 50% efficiency.

Assuming they can hit these unit economics on electrolyzer scale, and $10/MWh of installed solar capacity (which as of 2025 is around $35/MWh), their target marginal price of hydrogen production is a staggeringly-low $0.78/kg. This is a very speculative number and far below current cost estimates for solar-driven electrolysis, which is more around $4-$6 per kg. A 2020 study showed the levelized cost of hydrogen using solar PV at $5.5-6.1/kg H2.

At the end of the day society needs cheap power. What’s Terraforms projected cost? At today’s cost they would produce CH4 at about $28-30 per million BTU (MMBtu), and natural gas by comparison is more like $3-5/MMBtu. However if they can get to mass-produced electrolyzer modules at just $30k/MW and ultra-cheap solar of $0.005-0.001/kWh, the levelized cost of CH4 drops to $4-5 per MMBtu. This does assume <$1/kg H2 as well.

Notably this is actually higher than the levelized cost of methane predicted for the at-scale SOEC + Geothermal + Point Source capture; even though they get electricity as cheaply as possible, and have cheap electrolyzers, those electrolyzers end up consuming more electricity than SOEC since they’re both less efficient, and, don’t subsidize the power with heat. But most importantly, the cost of CO2 via direct-air-capture versus point-source capture is about 6x higher.

Synthetic fuel game is all about minimizing the levelized cost per kg of CH4 produced, so its pretty sensitive to the raw costs of inputs. Super cheap H2 is a win, but not if you have to get relatively expensive CO2.

Big Take-Aways

After spending a few weeks researching this topic, here’s the number one takeaway I’ve been able to piece together:

Synthetic Fuels are 100% the important to the future of hydrocarbon production and can be done at-scale in industrially relevant quantities using well-understood technologies, BUT, to beat fossil fuels economically and without massive government subsidies the correct approach will grind down OPEX above all else.

That’s really it at the end of the day. Of course, grinding down OPEX means getting the CO2 and H2 for as cheap as humanly possible, and the biggest cost factor in both of those is the price of energy. If you want to Direct Air Capture from the atmosphere your CO2 on a per-ton basis is going to be 5x the cost of a Point Source Capture at an individual factory, but, you can locate it anywhere. If you want to cut the electrical energy consumption it takes to split H2 from water, you can use thermally boosted SOEC cells to get higher efficiency, but the up-front CAPEX is going to be higher and they have a shorter lifetime.

My personal bet is the biggest thing to flip in the next 5 years is where people will get their H2 from, and the argument is simple: PEM and Alkaline electrolyzers might see a 50% price drop and marginal, e.g. 10% improvement in efficiency, meanwhile SOEC cells are already 80% efficient and could see a price drop of 4-5x at which point they become the clear winner for H2 production. This means having a steady and reliable source of cheap heat becomes the winner, so facilities like geothermal energy or fission are necessary.

Now, where do we get the cheapest heat around? The breakdown of radioactive elements. You can either dig those up and put them in a fission reactor, or dig down and pipe up the heat as in geothermal. For a modern high temperature gas-cooled reactor the cost per kilowatt-hour thermal for steam is around 1.7 cents. Enhanced geothermal heat can cost 2 cents per kWh-thermal, and the DoE wants to reduce the cost of geothermal by 95% by 2035.

If you think this through a bit, large industrial facilities often have a ton of both waste heat, and waste vented CO2, and transporting hydrocarbons is extremely easy given the incredibly mature infrastructure the entire global industrial economy is built around. Fuel is much more price sensitive than it is location sensitive since shipping costs are already rock-bottom low. Factories producing cement and ammonia might start having synthetic fuel production facilities built adjoining them all over the world to take advantage of the natural byproducts of industrial processes.

Anyway you slice it, the market for synthetic fuels, especially for aviation fuels, is going to absolutely explode in the years to come due to government mandated fuel targets imposed mostly by Europe. Like any other industrial process, as the economies of scale are brought to bear on the synthetic fuel economy as a whole the price will keep going down and down over time in the race to the most important thing for civilizational abundance: cheap, reliable, accessible, on-demand high density energy.