Chemical Recycling's Missing Number
Mechanical recycling — sorting, shredding, melting — works well for clean, separated streams of a single polymer. It works poorly for everything else: multilayer films, coloured mixtures, food-contaminated packaging, anything laminated. That residual fraction is most of plastic waste, and it is the gap chemical recycling promises to fill.
The dominant technology is pyrolysis: heating plastic without oxygen until it breaks down into an oil that can, in principle, be fed back into a steam cracker and polymerised into new plastic. If that worked at scale it would be genuinely significant.
So the question is simple. Put a tonne of mixed waste plastic in; how much new plastic comes out?
There is no agreed answer. More tellingly, the companies operating the plants decline to supply one.
The yield cascade
The confusion comes from measuring at different points in the process. Each point gives a flattering number compared to the next.
Lab, clean polymer. Pyrolysis of a single clean polymer under controlled conditions produces oil yields of roughly 70–95% by weight. These are the figures that tend to reach press releases.
Pilot, real waste. Feed in actual post-consumer mixed plastic — with its soil, food residue, paper and PVC — and oil yields fall to roughly 30–50%. Contaminants promote char formation. The rest leaves as gas and solids.
Oil is not plastic. This is the step most often skipped. Pyrolysis oil must be upgraded, separated into a naphtha-range fraction, cracked, then polymerised. Each stage loses material. Crucially, pyrolysis oil is typically blended into a steam cracker as a minority feedstock — life-cycle studies commonly model substitution rates of 5–20% — and less than half of a cracker’s output is ethylene and propylene to begin with.
Run that cascade end to end and the number collapses. ProPublica’s 2024 investigation put it at 15–20% of input plastic emerging as the ethylene and propylene that become new plastic, with 80–85% becoming diesel, methane, hydrogen and other chemicals. The estimate drew on peer-reviewed studies, industry presentations and specialists including analysts at Wood Mackenzie and Lux Research. More advanced processes were put at 25–30%; the lower range was used because it is the more common technology.
For comparison, the same investigation put mechanical recycling at 55–85% recovery as new plastic.
Why the number is missing
The gap is not a mystery of physics. It is a disclosure problem.
Asked for an industry-wide yield, the American Chemistry Council responded that too many factors affect output for any single figure to be meaningful. Individual companies declined to provide their yields or production volumes.
That is not unreasonable on its face. Feedstock composition genuinely does drive yield, and plants differ. But it has a consequence: a sector asking for regulatory recognition, public subsidy and recycled-content credit is also declining to publish the measurement that would establish whether it works.
The academic literature cannot close the gap either. Reviews of pyrolysis life-cycle assessments find that most rely on theoretical models, simulations, or laboratory experiments on clean polymers. Only a small minority examine actual contaminated post-consumer waste at pilot or industrial scale — which is the only condition that matters.
The accounting layer
Because output molecules cannot be physically traced, recycled content is assigned on paper through mass balance: a plant feeds some proportion of pyrolysis oil into a cracker and allocates a corresponding “recycled” claim across its products.
The industry’s defence here is real and deserves stating properly. Ethylene derived from pyrolysis oil is chemically identical to ethylene from fossil naphtha; once blended, the two cannot be distinguished, so some allocation rule is unavoidable. The ACC compares it to renewable electricity and fair-trade coffee, where certified input is tracked through a fungible system rather than physically followed. The analogy is sound as far as it goes.
What it does not settle is which allocation rule. Generous rules let a small quantity of pyrolysis oil justify a large volume of product labelled recycled. ProPublica reported that recycled content in finished goods is at most around 10%, with studies suggesting 2–5% in practice.
The EU has now chosen a rule
In 2026 the European Commission settled on a fuel-exempt mass balance method for chemically recycled PET in beverage bottles, approved by member states in February and adopted later in the year. Its central provision: any input that ends up as fuel, or is lost in processing, must be deducted from recycled-content claims.
This bites directly on the cascade above. Under the EU’s definition, material burned for energy was never recycled, so the large share of pyrolysis output going to fuels cannot be counted. An analysis by CE Delft estimated the fuel-exempt approach implies a plastic-to-plastic yield of around 49% — higher than ProPublica’s figure because it measures a different thing, but far below the impression left by laboratory yields.
It is also the first time EU legislation has formally recognised pyrolysis and gasification as recycling, which is why neither side is satisfied. Industry objects that the deductions are punitive. Zero Waste Europe and the Rethink Plastic Alliance argue the allocation method is wrong in principle and legitimises the technology prematurely. The Commission has signalled the methodology will serve as a template for packaging, automotive and textile rules, so the stakes run well beyond bottles.
In the United States the EPA has sought comment on removing pyrolysis from an air emissions rule — a reclassification industry says resolves regulatory uncertainty, and environmental groups say exempts the sector from stricter controls.
The commercial record
Technology arguments are eventually settled by plants that run. The record so far is thin.
Dow and Mura cancelled a 120,000-tonne-per-year facility in Germany that had been announced in 2022 and was expected to be operating by now; the partners had previously discussed adding up to 600,000 tonnes of capacity by 2030. Brightmark’s Indiana plant filed for bankruptcy in March 2025, having never reached full capacity despite public support. Oregon’s Regenyx closed in 2024 before reaching capacity. Encina cancelled a planned Pennsylvania plant.
Most striking, the US Department of Energy cancelled a $375 million award for a Texas facility in 2025, describing the project as not economically viable. That is a funder walking away, not a critic.
Campaign groups have been keeping count — Beyond Plastics puts the number of constructed US facilities at eleven — and their framing is advocacy rather than neutral analysis. But the cancellations themselves are company and government announcements, and the direction is not ambiguous.
What would settle it
A published, independently audited mass balance from a plant running real mixed post-consumer waste: tonnes in, tonnes of polymer out, energy consumed. One credible dataset would end the argument in either direction.
Until that exists, the honest position is not that chemical recycling is a fraud. Pyrolysis works as chemistry. Pyrolysis oil does displace some fossil feedstock. The ACC is right that collection and sorting infrastructure is a genuine constraint, and mechanical recycling cannot address the mixed and contaminated fraction.
The honest position is narrower and more awkward. A technology being written into law, counted toward recycled-content targets and funded from public money has not published the single number that would show how well it does the thing it is named for. One analyst’s projection, quoted by ProPublica, is that pyrolysis might at best displace around 0.2% of annual new plastic production within three years.
That is not nothing. It is also not a solution to plastic waste, and the difference matters a great deal for anyone deciding where the next decade of regulation and capital should go.
Sources: ProPublica; American Chemistry Council; CE Delft; Zero Waste Europe.