Waste-to-food process turns PET into protein

Researchers at Southern Illinois University Carbondale are developing a process that converts compounds derived from PET plastic and agricultural waste into protein-rich biomass and other potential food ingredients using engineered yeasts.

The work, presented at the American Chemical Society’s Fall 2026 meeting, has produced a prototype food known as µBites. The researchers have also engineered microorganisms capable of making compounds including vanillin and beta-carotene, extending the project beyond bulk microbial protein towards flavour, colour, and nutritional ingredients.

The process begins with material that conventional food fermentation would normally never encounter. PET bottles, discarded corn stalks and leaves, and other biomass are subjected to oxidative hydrothermal dissolution, a proprietary SIU process that uses water and oxygen at elevated temperature and pressure to break resistant material into smaller compounds that microorganisms can metabolise.

Those compounds then become feedstock for programmed yeasts, including strains based on conventional baker’s yeast. Rather than fermenting sugar or starch supplied as a conventional food-grade raw material, the microorganisms use carbon recovered from the processed waste stream and convert it into new biological material.

The project therefore separates the manufacturing challenge into two distinct operations. The first has to break down highly stable polymers and fibrous biomass sufficiently to make their carbon accessible; the second has to run a controlled fermentation capable of producing useful compounds from the resulting mixture.

SIU’s researchers report outputs including proteins, fats, and organic acids. One engineered yeast strain can make vanillin from plant biomass, while another converts ethylene glycol derived from PET into beta-carotene, the orange-red pigment that can act as a precursor to vitamin A.

The µBite provides a tangible demonstration of the process, but it should not be mistaken for a commercial food launch. Protein-rich microbial biomass is combined with separately added fibre, starch, and sweetener before being extruded through a 3D food printer and cooked into the finished prototype.

ACS says the work includes a 32-step safety process and that current data indicate the µBites are safe to eat, although the team is still awaiting institutional approval for formal taste testing. That distinction is important because demonstrating a safe laboratory prototype is a very different proposition from qualifying an ingredient for routine production and sale.

PET is attractive as a carbon source precisely because it is difficult to deal with. The polymer is designed to remain stable through bottling, storage, transport, and use, so converting it into something biologically accessible requires substantial processing before fermentation can begin.

A commercial food route would also have to control contaminants carried with real waste. Recovered PET can contain labels, adhesives, inks, closures, multilayer materials, other polymers, and residues from its previous use. Agricultural residues vary with crop variety, season, storage, moisture, soil contamination, and handling.

Those variables would become incoming-material specifications in any industrial system. Manufacturers would need validated separation and analytical steps capable of proving that unwanted compounds had been removed before the fermentation stream entered food production, followed by equally rigorous controls around the organisms and final biomass.

Fermentation itself introduces familiar scale-up constraints. Temperature, pH, oxygen transfer, nutrient balance, agitation, contamination control, residence time, and strain stability determine yield, while downstream processing has to recover and stabilise the product without destroying the properties the microorganism was selected to produce.

The economics will depend heavily on concentration. A fermentation that produces useful protein or speciality compounds in a dilute broth can become dominated by separation and drying costs, while the hydrothermal front end adds its own energy, pressure-vessel, and materials-handling requirements.

Producing several outputs could improve that equation. Bulk microbial protein may have relatively modest value per kilogram, whereas flavour compounds, pigments, vitamins, or other functional ingredients can command substantially higher prices. A process able to allocate different waste-derived carbon streams towards different outputs could therefore have more commercial flexibility than one designed solely to make protein.

The project’s origins in NASA’s Deep Space Food Challenge explain some of its unusual design priorities. Long-duration missions have limited resupply and limited waste-disposal options, creating a strong incentive to recover carbon already present in packaging and other materials rather than continually importing fresh agricultural feedstocks.

Terrestrial food manufacturing presents a different calculation. Waste is easier to remove, conventional ingredients are readily available, and any process involving post-consumer plastic would face intense regulatory and consumer scrutiny. The Earth-based case therefore depends less on necessity and more on whether waste-derived carbon can compete technically and economically with established fermentation feedstocks.

The researchers ultimately want microorganisms to produce more of the components currently added separately to the µBites, including starch, fibre, and sweetener. Achieving that would move the work closer to an integrated production platform, although it would also increase the complexity of controlling product composition.

The most significant development is not the novelty of a biscuit linked to a plastic bottle. It is the attempt to turn waste carbon into a defined fermentation raw material. Industrial relevance will begin when that feedstock can be produced consistently enough for food manufacturing to treat it less like waste and more like an ingredient specification.