The Reaction In The Middle

Entrade put two hundred wood-gasification machines into buildings across the UK and Europe. I could draw the front end of those plants from memory. Feedstock handling, the reactor, the flare. Every project had a beautiful front end. The back end is where the money was supposed to be, and the back end is one reaction. Carbon monoxide plus water goes to carbon dioxide plus hydrogen. The enthalpy change is minus 41.2 kilojoules per mole. That is arithmetic a sixteen-year-old can do. It is also the step that decides whether the other ninety per cent of the plant was worth building. The probability that the middle step is your binding constraint is not zero. Almost everyone prices it at zero, because it looks solved. It has looked solved for a long time, which is exactly why nobody has gone back and re-engineered it. Nobody re-engineers the middle. They buy it off a shelf, size it from a data sheet, and hope.

The reaction is exothermic and equilibrium-limited, and those two words set your ceiling before a single piece of steel is ordered. The equilibrium constant runs about 200 at 200 degrees Celsius, about 12 at 400 degrees, and about 1 at 800. Read that backwards. The hotter you go, the less the reaction wants to convert. You do not get fast kinetics and a high equilibrium yield in the same bed. You pick one and buy the other back with a second bed. Pressure does almost nothing. There is no change in mole count across the reaction, so raising pressure moves nothing. Operators still run high pressure, because throughput is throughput. So the industry splits the job. A high-temperature stage at 310 to 450 degrees takes carbon monoxide from 10 to 15 per cent down to 2 to 4 per cent, roughly 87 per cent conversion. Interstage cooling recovers the heat and drops the gas. A low-temperature stage at 190 to 250 degrees finishes the job at about 99 per cent total conversion. Two beds, one exothermic reaction, and the size of every downstream compressor in the plant already decided. Activation energies tell the same story: 80 to 100 kilojoules per mole on the iron high-temperature catalyst, 50 to 70 on the copper low-temperature one.

Why the two temperatures. Because the reaction happens by two different routes, and the routes want different metals. On iron it is the redox route, the Mars-van Krevelen cycle. Carbon monoxide takes oxygen off the surface. Steam puts it back. The catalyst is not a spectator, it is a reagent that keeps getting rebuilt: magnetite reduced, then reoxidised, over and over. On copper and platinum it is the associative route. Carbon monoxide meets a hydroxyl, forms formate or carboxyl, and the intermediate falls apart into carbon dioxide and hydrogen. Density functional work published by Gokhale in 2008 and Grabow in 2008 put the carboxyl pathway above the formate pathway on both copper and platinum surfaces. That is the kind of detail that decides which metal goes in which bed. Same reaction. Different chemistry. Different catalyst families. Different failure modes.

The commercial high-temperature catalyst is iron oxide, chromium oxide and copper oxide at 88, 9 and 3 per cent by weight. The working phase is magnetite, formed in situ. Chromium holds the surface apart and stops the magnetite sintering. Copper lowers the temperature at which the whole thing wakes up. That chromium is now a regulatory problem, and the shelf is moving. The first commercial chromium-free high-temperature catalyst is a zinc-spinel formulation sold as SK-501 Flex. A second product line, ShiftMax 120 HCF, is marketed as ultra-low hexavalent chromium. Low-temperature shift is copper, zinc oxide and alumina, roughly 40 to 60, 20 to 30, and 10 to 20 per cent. It takes carbon monoxide from 2 or 3 per cent down to about 1,000 parts per million, which is a factor of twenty or thirty and the only reason the stage exists. It has one fatal flaw. It dies at parts-per-billion hydrogen sulphide, and chlorine kills it faster. On a clean natural gas feed that is acceptable. On anything that came out of a gasifier, it is a liability you carry for the life of the plant. Where sulphur is present the industry reaches for cobalt and molybdenum on alumina with a potassium promoter: 250 to 500 degrees, steam-to-carbon as low as 0.3 to 0.5, and a requirement to keep more than 50 parts per million of hydrogen sulphide in the gas. Strip the sulphur out and you have also switched the catalyst off.

Nobody re-engineers the middle. They buy it off a shelf and hope.

A gasifier does not hand you a clean feed. The syngas comes off at 15 to 25 per cent carbon monoxide, 10 to 20 per cent hydrogen, 10 to 20 per cent carbon dioxide and 2 to 10 per cent methane. It also carries 50 to 500 parts per million of hydrogen sulphide and 0.5 to 10 grams of tar per normal cubic metre, plus particulates, alkali metals and hydrogen chloride. Now put that against the shelf. The copper low-temperature catalyst is poisoned outright. The precious-metal options tolerate sulphur to about 100 parts per million and cost more than the rest of the plant. The iron high-temperature catalyst works, and does nothing about the tar. The good news is that poisoning is not always permanent. Dou showed in 2019 that hydrogen sulphide and hydrogen chloride poisoning of iron shift catalysts is fully restorable once the poison is removed. Kraussler ran a tar-rich biomass syngas train for 2,250 hours in 2016, which answers whether this works over a season rather than an afternoon. Corella measured 140 grams of hydrogen per kilogram of biomass in 2008. So the chemistry is not the risk. The middle step is a known quantity. That is the entire point.

If you want to know where this goes, read the papers rather than the catalogues. Platinum on ceria reaches a turnover frequency of 0.85 per second at 400 degrees and holds 100 hours at 100 parts per million of hydrogen sulphide. A single-atom platinum on alpha molybdenum carbide system, published in Nature in 2021, is active at 40 degrees Celsius. Forty. Those are lab conditions and lab lifetimes, and they are still the direction of travel. The non-precious work is closer to a plant. A facet-controlled molybdenum carbide nanorod reached about 100 per cent conversion at 175 degrees. A copper-cobalt oxide ran 100 per cent at 250 degrees and a space velocity of 84,000 per hour. A cerium-cobalt oxide held 95 per cent at 240 degrees for more than 3,000 minutes. A nickel-yttrium oxide went past 90 per cent at 250 degrees for more than 100 hours. Meanwhile the reactor is being rebuilt rather than the catalyst. Sorption-enhanced shift captures carbon dioxide in the same bed and reports better than 95 per cent capture with hydrogen purity above 95 per cent in a single step. Palladium membrane reactors pull hydrogen out as it forms and drag the equilibrium with them; a 2025 techno-economic study put the cost of hydrogen about 10 per cent lower, under 2.2 dollars a kilogram. Plasma-assisted shift went from 6 to 96 per cent conversion at 150 degrees with the plasma on. Electrochemical shift runs at room temperature, above 99.99 per cent hydrogen, stable for 475 hours. Photocatalytic shift is still under 9 per cent solar-to-hydrogen. The distance between 175 degrees in a paper and 400 degrees in a plant is not a distance in chemistry. It is a distance in who paid for the pilot.

Here is what I would do with any of this, and none of it depends on which metal wins. Price the middle step before you price anything else. Take the equilibrium constant at your operating temperature. Take the conversion your two beds can honestly deliver. Multiply the four or five per cent of carbon monoxide that slips through by the cost of the separation you now need. That number moves your hydrogen cost further than the feedstock contract does, and it is a number you chose. Most projects I look at have a front end drawn to three decimal places and a middle step drawn as a box. The probability that a project dies on the chemistry it put five slides behind is low. The probability that it dies on the box labelled shift is not zero, and it is being priced at zero by nearly everyone in the room. Middle steps do not get glamorous. They get expensive.