The Cave in the Digester

The energy industry's problem is hydrogen sulfide. Sour gas, digester gas, refinery streams: H₂S shows up wherever organic matter breaks down without oxygen, and it is treated as pure liability. It corrodes equipment, it kills workers at concentrations a nose stops smelling, and the entire industrial relationship with it is removal. You scrub it out with amine units, you run it through a Claus train, and you convert it to low-value elemental sulfur that you are mostly relieved to be rid of. Nobody in that world asks what H₂S is for. It is a cost center with a body count.

The flavor industry's problem sits at the other end of the same molecule. Esters are the workhorses of natural flavor, and a natural label is the difference between which shelf a product sits on and at what margin. Take butyl butyrate, a fruity, pineapple-leaning ester and one of the first we ever made. Both halves of it fall out of fermentation cheaply and unarguably naturally: butyric acid from butyrate fermentation, butanol from solventogenic Clostridium. You build the ester by reacting the acid with the alcohol over an acid catalyst, a Fischer esterification, and the chemistry is trivial. The constraint is legal. Under EU flavor law a substance can only carry the word "natural" if it is produced by physical, enzymatic, or microbiological processes from material of natural origin, and where a reagent is needed, that reagent has to be natural too. A drop of synthetic sulfuric acid as your esterification catalyst and the claim is gone. So with both precursors sitting right there, cheap and natural, the only thing between them and a natural butyl butyrate was the catalyst. And that gap is not specific to one ester. A whole class of acid-catalyzed transformations, esterifications, hydrolyses, condensations, runs on a strong Brønsted acid, and the strong acid is where the natural claim breaks. You cannot reliably source a natural strong-acid catalyst at process scale. The cheap acid is synthetic, and synthetic breaks the label. Butyl butyrate is just the cleanest place to watch it happen, because both halves come off one platform and the only synthetic atom in the room is the acid.

One industry pays to destroy a sulfur compound. The other cannot find the natural acid it needs to turn cheap natural precursors into a natural product. The wall between them is the only reason both problems exist.

What follows is a system we built and ran. The precursors came off AGATE, our polyculture fermentation platform. The catalyst we had to go to a cave for.

The organism that lives on the seam

The solution was already operating, at scale, in the dark, in a cave in Tabasco, Mexico called Cueva de Villa Luz.

It is a sulfidic cave. Groundwater rich in dissolved sulfide flows through it, H₂S degasses off that water into the cave air, and where the two meet there is a third thing: snottites. The name is exactly as undignified as it sounds. They are mucus-like microbial biofilms that hang from the cave ceiling and drip, and the dominant organism in them is Acidithiobacillus thiooxidans, a sulfur-oxidizing bacterium that lives at a pH near zero. The drips are dilute sulfuric acid. The biofilm is, functionally, an acid factory built out of slime.

What A. thiooxidans does is the whole trick. It sits on the redox seam, where reduced sulfur rising from an anoxic world below meets oxygen arriving from the surface above, and it makes its living by oxidizing that H₂S to sulfuric acid. The cave does not assign the organism a niche so much as the geometry of the cave is the niche: anoxic and sulfide-rich below, oxic above, and a thin living interface between them doing the conversion.

That last sentence is the one that matters, because a geometry can be rebuilt.

Rebuilding the cave in a tank

The conventional move would be to isolate the organism, give it a clean aerobic culture, feed it sulfide, and harvest acid. That is the version that does not work, because A. thiooxidans is not the point. The interface is the point, and the interface is what we reconstructed.

We ran it inside an anaerobic digester, stratified top to bottom to match the cave. At the base, anaerobic digestion did what it does: methanogens making biogas and sulfate-reducing bacteria making, deliberately, hydrogen sulfide. That sludge zone is the cave's sulfidic groundwater. The H₂S degassed upward into the headspace, exactly as it does off the cave water.

Into that headspace went the snottite consortium, and into that headspace, against every instinct an anaerobic-digester operator has, we micro-injected oxygen. The headspace became the cave air. The oxygen feed became the cave's opening to the surface atmosphere. With reduced sulfur from below and a trace of oxygen from above, A. thiooxidans did in the vessel exactly what it does on the ceiling in Tabasco: it oxidized the H₂S to sulfuric acid.

And then the acid dripped, and we ran the reaction where it fell. A tray below the biofilm caught the runoff, the biogenic sulfuric acid carried along in the metabolite matrix it came in, and the tray was where we charged the precursors. Butyric acid and butanol, both off AGATE, met the acid film in the headspace and esterified in place. The tray is the cave floor, the place the drips pool, except we made the pool the reactor. The esterification is equilibrium-limited and the headspace is wet, so we pulled water at the reaction surface and out of the vapor by whatever the run called for, desiccant riding in the tray, molecular sieves on a recirculation loop, the specifics tuned per campaign. One vessel made the catalyst, one platform made the precursors, and the bond formed inside the living process. One waste gas walked all the way from liability to natural product without ever leaving the tank.

The knife-edge

The reason this is not a standard process is that it asks one tank to hold two biologies that want to kill each other.

The digestion below depends on a strict absence of oxygen. Methanogens and sulfate reducers are anaerobes; oxygen poisons them, and if you lose them you lose the H₂S that is the entire feedstock for the layer above. The sulfur oxidation in the headspace depends on the opposite. Starve A. thiooxidans and the biofilm dies and you get no acid. The whole apparatus lives or dies on a single control variable, the oxygen micro-injection rate into the headspace. Too much and it bleeds down into the sludge and strangles the feedstock. Too little and the acid factory goes dark.

What makes it stable rather than merely precarious is that the biofilm defends the boundary itself. The aerobic colony at the top consumes oxygen as fast as it arrives, scavenging it toward nothing before it can descend into the anaerobic zone. The layer that needs the oxygen is also the layer that protects the anaerobes from it. The oxic skin guards the anoxic body. That is precisely the arrangement the cave runs on, and it is why a system this contradictory holds together at all once the injection rate is tuned. It is not balanced by the operator. It is balanced by the organism.

Why the crudeness is the compliance

Here is the part that looks like sloppy process engineering and is actually the entire legal mechanism.

We did not purify the acid. The catalyst was used as the fermentate it arrived in, the biogenic sulfuric acid still carried in its biological matrix off the biofilm, collected by gravity and used as is. To a process chemist that reads like an unfinished step. Why not concentrate it, clean it, bring it to reagent grade? Because reagent grade is exactly what would destroy the claim. The moment you purify the acid into a defined, synthetic-equivalent reagent, you have arguably stepped outside the microbiological process that made it natural in the first place. Catch it in its matrix, never separate it from its biological origin, and the catalyst stays inside the definition. That is the first argument, and it is an argument about the reagent.

The stronger one is about the reaction. EU flavor law permits physical, enzymatic, and microbiological processes. A mineral-acid-catalyzed esterification is, read strictly, a chemical process, and the provenance of the acid does not change the category of the reaction. That is the real exposure, and purity of the reagent does not reach it. Running the esterification in the reactor does. When the bond forms in the headspace, over an acid film the organism is generating in real time, in the matrix, at the point of production, never isolated and never handed to a separate chemical unit operation, the transformation is not a chemical step fed by a microbial reagent. It is an event inside the microbiological process. The reaction is a feature of the microenvironment, not a stage bolted on downstream. That is a different claim from "the acid is natural," and it is the one that survives a strict reading of the category.

We ran this past buyers who carry the compliance liability themselves. The isolated-acid version drew a split. The in-reactor version drew none, and the line they drew was the reactor wall.

This is a habit nature has and industry does not. Living systems almost never purify. They work in matrices, in mixtures, in context, and they do their chemistry in place rather than shipping intermediates between clean rooms. The instinct to isolate the pure compound and run the reaction in a dedicated vessel is an industrial reflex, and here it is twice the wrong one.

The proof a skeptic cannot wave off

A claim like this lives or dies on verification, and there are two anchors.

The first is the regulation itself. The natural flavoring definition under EU Regulation (EC) No 1334/2008, Article 3(2)(c), covers substances obtained by physical, enzymatic, or microbiological processes from material of natural origin, reagents included. The isolated-acid route argues its reagent into that definition. The in-reactor route does not have to argue, because the esterification happens inside the microbiological process rather than alongside it, which is the case buyers accepted without a split. The provenance argument and the process-category argument answer two different objections, and the piece is built to satisfy the definition rather than skirt it.

The second anchor is physics, and it is the one a flavor chemist will actually trust, because it cannot be argued with. The industry's authenticity test for "natural" is radiocarbon. Carbon drawn from living, recently-fixed sources carries modern ¹⁴C activity; carbon from fossil substrates is dead, devoid of radiocarbon, which is treated as robust proof that a material came from petrochemistry. Butyl butyrate built from fermentation butyric acid and fermentation butanol carries a modern radiocarbon signature in every carbon it has. That is not a labeling argument. It is a measurement.

The search pattern

Notice which doors this walked through. The feedstock was the gas the energy industry fears most. The vessel was a sewage digester running on waste. The organism was cave snot, about the most dismissible thing a person could be asked to take seriously. The most hazardous, lowest-status, most repellent combination on offer, running precision natural-flavor catalysis at a pH near zero.

That is not a coincidence. It is the method. The places an industry refuses to look are precisely the places its solutions sit unclaimed, because the refusal is what keeps them uncrowded. The energy world could not see a catalyst in its poison because the poison was something to be removed. The flavor world could not find a natural acid because it was looking in supplier catalogs, not in caves. The answer required standing in the seam between two industries and being willing to pick up the thing both of them had decided was beneath them.

That seam is where this publication works. The mechanism always comes first, and the mechanism is almost always hiding somewhere the field has trained itself to look away from. The flinch is the signal. Next time, we follow it somewhere else.