Hydrothermal processing is the industrial half of river cleanup, and it has spent twenty years waiting for something cheap to eat.
There is a technology that turns wet garbage into crude oil in about twenty minutes. It has been demonstrated at pilot scale on four continents. It works on sewage sludge, manure, food waste, mixed plastics, wood, and pond scum. The chemistry is settled. The engineering is understood. And it is nowhere.
Ask why and you get a lot of answers about capital costs and permitting and offtake agreements, all of which are true and none of which is the real problem. The real problem is arithmetic. A hydrothermal plant is a machine that eats biomass. Biomass costs money — to grow, to collect, to haul, to store. Build a plant that eats wood chips and you are now in competition with every paper mill and pellet stove within a hundred miles. Build one that eats municipal sludge and you are limited to whatever the sewer authority produces, which is a fixed number nobody can increase. Every hydrothermal venture of the last two decades has run headlong into the same wall: the process is fine, the feedstock is the business, and nobody has a feedstock that is both enormous and free.
The River Refugium Project is a feedstock.
That is the whole argument, and the rest of this piece is just the explanation.
What happens when you squeeze hot water
Start with the part everyone finds intimidating and get it over with.
Water at room temperature is a liquid. Heat it past 212°F and it becomes steam. Everybody knows this. What most people never learn is that the boiling point is a function of pressure — hold water under enough pressure and it will not boil no matter how hot you make it. Push far enough along that line, past about 705°F and 218 atmospheres, and water stops being either a liquid or a gas and becomes something else: a supercritical fluid. It moves and diffuses like a gas. It dissolves things like a liquid. And its chemical personality flips. Ordinary water is polar — that is why it dissolves salt and refuses to mix with oil. Supercritical water is closer to nonpolar. It dissolves oil and starts rejecting salt. It also becomes a mild acid and a mild base at the same time, which means it will attack chemical bonds it would never touch at the kitchen sink.
Hydrothermal processing lives in and just below that zone. You do not need to go all the way supercritical to get the effect; the useful chemistry starts well before, and the industry mostly operates in the subcritical range where water is still technically a liquid and already behaving strangely.
Now add the second condition: no oxygen. This is the part that trips people up, because heat plus organic matter sounds like fire. Fire is oxidation — carbon and hydrogen grabbing oxygen from the air and dumping the energy as flame, leaving you CO₂ and ash. Take the oxygen away and heat has nothing to burn with. The energy still goes in, the bonds still break, but the pieces have nowhere to go except back into each other. Long-chain molecules come apart into shorter ones. Cell walls rupture. Proteins hydrolyze into amino acids and then into smaller fragments. Fats split into fatty acids. Starches and sugars fall apart into acids and ring compounds. Lignin — the tough stuff that makes wood woody — cracks into phenolic fragments.
Nothing is destroyed. Everything is disassembled. You put a plant in one end and get its constituent chemistry out the other, sorted by density into an oil layer, a water layer, a gas, and a solid.
That is the entire trick. Heat, pressure, no air, and enough time. The rest is plumbing.
The two settings on the dial: HTC and HTL
Hydrothermal processing is not one process. It is a temperature dial, and where you set it determines what comes out.
Hydrothermal carbonization (HTC) is the low setting — roughly 350 to 480°F, moderate pressure, and a long dwell time measured in hours. At this severity the material does not liquefy. It condenses. Oxygen and hydrogen leave as water and CO₂, carbon stays behind, and you end up with a black, friable solid called hydrochar. You are running the coal-formation process on fast forward: geology in an afternoon.
Hydrothermal liquefaction (HTL) is the high setting — roughly 570 to 700°F, 150 to 250 atmospheres, and a dwell time measured in minutes rather than hours. Here the material does liquefy. The output is biocrude: a dark, viscous oil chemically similar enough to petroleum that refineries can co-process it. Biocrude from duckweed typically runs 32 to 36 MJ/kg in energy density, against roughly 42 to 44 for conventional crude. You are at about eighty percent of petroleum before any upgrading at all.
Push past HTL and you get hydrothermal gasification, where everything cracks down to methane, hydrogen, and CO₂. Useful as a polishing step, and we will come back to it.
The important thing is that both HTC and HTL are wet processes. Neither one requires drying the feedstock, and that single fact is what separates hydrothermal from pyrolysis and every other thermochemical route. Pyrolysis needs dry input, and drying wet biomass eats more energy than the biomass contains. That is the wall that killed twenty years of algae-to-fuel ventures. Hydrothermal walks around it, because the water is not a contaminant to be removed. The water is the reaction medium.
So when do you use which? The answer tracks the feedstock.
Soft, wet, low-lignin, protein- and lipid-rich material — microalgae, duckweed, azolla, filamentous mats, sludge, manure — goes to HTL. This material liquefies cleanly because it has no structural lignin to fight you. HTL is the right answer for everything the RRP grows in the water column.
Woody, fibrous, high-lignin, high-ash material — harvested reeds, cattail, willow, wood, crop residue, the greenhouse’s own spent biomass, and the coarse organic debris a river carries — goes to HTC. Lignin resists liquefaction. It condenses beautifully. Trying to run wood through HTL gets you low oil yields and a lot of char you did not want; running it through HTC gets you char you did want, and a dry, stable, storable, sellable carbon product out of a wet input.
One pond system, two reactors, and the sorting rule is simply: does it liquefy, or does it condense?
What actually comes out
This is where the argument stops being about fuel.
The biocrude is the headline, and it deserves to be — biocrude from wastewater-grown duckweed runs 30 to 35 percent of dry input mass, with better than 80 percent of the feedstock’s energy carried into the oil. That alone would justify the reactor. But a hydrothermal plant is a separations machine, and every stream leaving it is worth something to somebody.
The oil phase gives you biocrude for hydrotreating into diesel and jet-range fuel. It also carries fatty acids, phenolics from any lignin in the mix, and a suite of nitrogen heterocycles — pyrroles, pyridines, indoles. The nitrogen compounds are currently treated as a nuisance, because they poison hydrotreating catalysts and have to be stripped out. They are also, considered from a different angle, a nitrogen-chemical feedstock that arrived free in a stream we were going to throw away. That reframing has not happened yet in the industry. It should.
The aqueous phase is the sleeper. This is the water that leaves the reactor carrying everything that stayed dissolved, and it is loaded: acetic acid, formic acid, lactic and propionic and levulinic acid, sugars, furfural and HMF, phenolics, and — critically — ammonia nitrogen and potassium. In HTC, the process water runs even richer in sugars and organic acids, because you never pushed the severity high enough to crack them. These are platform chemicals. Furfural and HMF in particular are the feedstocks a whole class of bio-based plastics and solvents wants and cannot get cheaply. And the ammonia and potassium are fertilizer, recoverable by stripping or electrodialysis, or simply recycled straight back to the grow beds as plant food.
The solid phase is where phosphorus and metals concentrate. Phosphorus is the one to watch. It is finite, it is mined, it is geopolitically concentrated, and the United States dumps a colossal quantity of it into the Gulf of Mexico every year in the form of agricultural runoff. Aquatic plants take it up as they grow; hydrothermal processing puts it into a solid ash fraction where it can be recovered as struvite or phosphate salt. We are not making phosphorus. We are catching phosphorus that was already bought, already mined, already paid for once, on its way to becoming a dead zone.
Metals come out the same door. Duckweed and algae are documented hyperaccumulators — they concentrate cadmium, lead, zinc, copper, manganese, and iron out of dilute water simply by living in it. In a region like ours, where acid mine drainage is the defining water problem and the drainage carries iron, manganese, aluminum, and a measurable rare earth fraction, the biomass is doing hydrometallurgical concentration for free and the reactor is doing the ashing. Whether that pencils out at commodity prices is an open question. That it works chemically is not.
The gas phase is mostly CO₂, which goes straight back into the algae raceways, because that is what algae eat.
And there is one more that is only starting to be understood: hydrothermal conditions destroy things that do not otherwise break. Fluorinated compounds — the PFAS family — are being defeated in the lab under hydrothermal alkaline conditions in the same temperature range HTL already runs. Microplastics depolymerize; several HTL ventures deliberately feed mixed waste plastic as co-feedstock. A process built to make fuel turns out to also be one of the few things that reliably takes apart the molecules we have no other answer for. That is a public health case sitting inside an energy case, and neither one has been made loudly enough.
Sum it: fuel, fertilizer, phosphorus, industrial chemicals, carbon, recovered metals, destroyed persistent pollutants — and water leaving cleaner than it arrived.
Walking the RRP loop
Here is how the pieces connect. The full document suite lives at cernunnosfoundation.com/rrp, but the flow is simple enough to state in a paragraph.
Intake. Pull from the river, the outfall, the storm drain, the reservoir, the drainage ditch — anywhere the water is dirty. The RRP has never been fussy about the source, because the entire premise is that pollution is a raw material. There is no shortage of it and no competition to acquire it.
The refugium. Water enters a managed wetland analog — the swamp, built on purpose. Solids settle. Bacterial communities work the nitrogen cycle. Filter feeders and mollusks handle suspended load. Buffering happens. What comes out the far side is not clean water; it is prepared water — nutrient-balanced, biologically stabilized, and ready to grow things aggressively. This stage is doing what a natural bayou does at the mouth of a river, in a footprint you can control and instrument.
The grow stage. Prepared water flows into duckweed lagoons and algae raceways. Duckweed can double its biomass in about two days under good conditions. Neither crop needs soil, arable land, or fresh water, and neither competes with the food supply. They eat exactly what we are trying to remove: nitrogen, phosphorus, potassium, dissolved metals. Every pound of biomass grown is a pound of pollution taken out of the water and moved into a form you can pick up.
Harvest and prep. Skim continuously — these are crops with no season and no harvest date, just a standing stock you keep cropping. Fresh duckweed is 92 to 95 percent water, and continuous HTL reactors want a pumpable slurry in the 15 to 25 percent solids range, so a mechanical screen or belt press does the first-stage dewatering and sends the pressed water straight back to the lagoons where it came from. Then the wet mulcher, and yes, it helps: HTL reactors want particle size down around a couple of millimeters for pumpability and even heat transfer. Duckweed is soft and needs little. Filamentous algae mats and any coarse plant material need real maceration or they will bridge in the feed line. The mulcher is cheap insurance on the single most failure-prone component in the plant.
The reactor. Slurry gets pressurized, preheated against the outgoing product stream — heat recovery is the difference between a plant that works and a plant that goes bankrupt — and held at temperature for its dwell time. Soft aquatic biomass goes to HTL. Woody and fibrous material harvested elsewhere in the system goes to HTC. Separation follows: oil off the top, gas off the top of that, aqueous phase in the middle, solids at the bottom.
The returns. CO₂ back to the raceways. Aqueous phase back to the grow beds as fertilizer, or through a gasification polishing step first if the organic load is too high for the crop to tolerate. Waste heat back into the greenhouse. Solids to phosphorus and metals recovery. Biocrude and hydrochar out the gate as product.
The discharge. Water goes back to the river having been stripped of the nutrient load that was going to feed a dead zone eight hundred miles downstream.
It is a loop, and every arrow in it is either a pollutant becoming a product or a product becoming an input somewhere else. That is not an environmental program with a revenue afterthought bolted on. It is an industrial facility whose raw material happens to be the thing everybody else is paying to dispose of.
What is actually hard
Credibility requires saying this part.
High-pressure slurry pumping is genuinely difficult and it is where pilot plants fail. Reactor materials face corrosion from organic acids and chlorides at temperature, which drives you toward expensive alloys. Biocrude carries too much nitrogen and oxygen to drop into a pipeline without hydrotreating, and hydrotreating is a refinery operation, not a farm operation. Aqueous phase recycling accumulates nitrogen compounds and phenolics over repeated cycles, which degrades oil quality if you do not manage it. Heat integration has to be near-perfect or the energy balance collapses.
None of these are unsolved. All of them are unsolved at commercial scale on a continuous basis with a specific feedstock, and that is the gap the RRP is positioned to close — because the feedstock is consistent, local, unlimited, and free, which is the one thing no HTL developer has ever had.
The scale
Here is the part that is difficult to hold in your head.
Every year, American agriculture buys nitrogen and phosphorus fertilizer that was mined, manufactured with natural gas, shipped, and spread. A large fraction of it never feeds a plant. It runs off into a ditch, into a creek, into a river, and eventually into the Gulf of Mexico, where it feeds an algal bloom that dies, decomposes, strips the oxygen from the water column, and kills a section of ocean roughly the size of a small state. We pay for that fertilizer once when we make it, again when we spread it, and a third time in lost fisheries and lost tourism when it arrives at the coast.
The nutrients in the river are not a problem with an unfortunate cost. They are a resource with a negative price, in a volume no purchasing department could ever assemble, delivered continuously by gravity to any point on the bank we choose to stand.
We can grow fuel by cleaning our water. The plants that clean it are the fastest-growing plants on earth. The technology that converts them has been sitting in a demonstration facility for fifteen years waiting for somebody to hand it something cheap enough to justify the reactor.
We know how far the chemistry goes on algae and duckweed. We do not yet know how far it can be pushed — which compounds are recoverable at what severity, how much of the periodic table comes out of a contaminated watershed, what the aqueous phase is worth when somebody finally builds a separations train for it instead of treating it as effluent. That is a decade of good work for people who like problems.
But we do not have to know the ceiling to start. We know the floor, and the floor is fuel, fertilizer, carbon, and a cleaner river.
It is that simple. The scale is what should keep you up at night.
