Plastic is not one material, and waste is not one feed
Polyethylene is a concentrated hydrocarbon. PET stores valuable aromatic and glycol building blocks. Nylon stores amide-linked monomers. Cross-linked polyurethane was designed not to flow when heated. Their recovery routes cannot be collapsed into one recycling rate.
The scale of the open loop
These figures are the 2019 baseline year of the OECD Global Plastics Outlook: still the last complete global material balance, not a current-year estimate (Measured). Context for the present: UNEP puts current production above 400 Mt a year (2024); the COVID-19 dip took use to 435 Mt in 2020, down 2.2%; secondary plastics supplied just 6% of feedstock, and 22 Mt leaked to the environment in 2019 (Measured). Dating the baseline honestly is what makes the projections below legible.
The long-run record is not a recycling curve catching production. Resin and fibre output rose from about 2 Mt in 1950 to 380 Mt in 2015, while the global system still returned only a small minority of discarded material.
Recoverability is polymer-specific
| Stream | Loop | Best available route | Binding constraint |
|---|---|---|---|
| Clear PET | Strong | Bottle-to-bottle remelt; selective depolymerisation after colour or degradation | Collection, colour, PVC, adhesives, food-grade proof |
| Natural HDPE | Strong | Wash, sort, remelt, and return to rigid products | Grade separation, odour, pigments, prior contents |
| Rigid PP | Improving | Mechanical recycling; solvent purification for difficult colour and contamination | Grade heterogeneity, additives, consistent feed |
| Clean LDPE film | Conditional | Wash and remelt concentrated commercial streams | Low density, dirt, inks, household collection |
| PVC pipe | Dedicated | Sector-specific mechanical loop | Legacy additives and chlorine contamination |
| Polyurethane | Weak | Formulation-specific solvolysis and partial polyol recovery | Cross-linking and heterogeneous formulations |
| Multilayer film | Weakest | Selective dissolution or layer-specific chemistry | The product architecture itself |
Every stage acts on what survived the last
Collection, sorting, conversion, and quality retention multiply. Four 90% stages do not return 90% of the feed; they return 65.6%.
Follow one tonne through the system
Each stage acts on what survived the one before it. A heroic reactor cannot recover material that was never collected or turn off-spec output into virgin-equivalent resin.
Overall circular productivity is 33.0%. At these settings, 670 kg is lost, downgraded, or stranded by unused capacity.
A technically recyclable polymer inside an uncollected, dirty, multilayer package is not a recyclable system.
The global balance, written as a tonne
OECD estimated that about 156 kg of each tonne of 2019 waste was collected for recycling. After roughly 40% became residue inside the nominal recycling stream, around 94 kg reached recycled output. At the other extreme, an illustrative bottle loop with 90% collection, 95% sorting, 90% processing, and 95% quality retention returns 731 kg. The polymer did not change. The system did.
Mechanical recycling is the benchmark loop
A kilogram of sorted HDPE already contains polyethylene molecules. Washing and remelting them is a smaller transformation than cracking them into oil, purifying the oil, making ethylene, and polymerising it again.
Remelt what is clean
Clear PET, natural HDPE, rigid PP, and concentrated clean film preserve the most embodied chemistry at the lowest conversion burden.
Constraint: purity and specificationPurify without breaking
Selective dissolution can remove colour, odour, and contaminants from a known polymer while retaining its molecular backbone.
Constraint: solvent recovery and throughputReset selective bonds
PET hydrolysis, glycolysis, and methanolysis exploit ester chemistry to recover reusable building blocks; enzymes can make cleavage unusually selective.
Constraint: reaction rate, pretreatment, monomer purificationCrack only the residual
Pyrolysis can accept polyolefin-rich mixtures that mechanical systems reject, but its oil still has to traverse refinery and polymer infrastructure.
Constraint: carbon yield, heteroatoms, energy, utilisationSorting is becoming information infrastructure
Near-infrared sensors identify many polymers; cameras recognise product shape and contamination; hyperspectral systems add wavelength bands; digital watermarks and product passports can carry declared composition. These tools make waste more legible, but recognition is only the first rung.
- Object
What did this used to be?
A vision system recognises a shampoo bottle or food tray and helps a robotic or pneumatic sorter act on it.
- Polymer
What is the visible resin?
Spectroscopy distinguishes PET, PE, PP, PVC, and other families at stated speed, particle size, purity, and recovery.
- Specification
Can this make the next product?
Grade, additives, molecular state, colour, odour, and prior contact determine whether a bale is actually certified feedstock.
There is no scientifically clean global “sorting accuracy curve.” Useful metrics state purity at a specified recovery, belt throughput, false ejection, particle size, and final saleable yield.
Concentrated carbon with a reverse-logistics bill
PE and PP are about 85.7% carbon by ideal repeat-unit mass; PS is about 92.3%. Waste is rich ore. But virgin feedstock arrives uniform, continuous, and certified, while discarded plastic pays for collection, transport, sorting, cleaning, purification, and every lost kilogram.
| Cost layer | Mechanical loop | Molecular loop | Lever |
|---|---|---|---|
| Collection | Can dominate dispersed articles | The same unsolved first mile | Deposits, EPR, dense commercial streams |
| Sorting | Distinctive monomaterial articles | Controls chlorine, metals, and heteroatoms | Optical sensing and product identity |
| Conversion | Washing, remelting, filtration | Reactor, catalyst or solvent, heat | Selectivity and heat integration |
| Purification | Deodorisation and grade blending | Oil upgrading or monomer separation | Feed purity before the plant |
| Quality | Near parity for excellent loops | Potentially virgin-equivalent after purification | Verified substitution, not output mass |
Where the loop is still stuck
Collection is binary
Material that is littered, burned, or landfilled cannot be recovered by a better plant. Dense deposits and commercial streams begin with an advantage chemistry cannot manufacture.
Mixture creates the entropy bill
A bale of known clear bottles preserves more value than the same polymers laminated, printed, filled, used for food, and shredded into a municipal stream.
Additives are inherited
Plasticisers, pigments, stabilisers, metals, PFAS, adhesives, and non-intentionally added substances travel with the nominal resin and can block high-value reuse.
Quality is state-dependent
Heat, UV, oxygen, moisture, and shear alter polymer chains. There is no universal number of cycles: specification and process history decide.
Chemical reset loses carbon
Pyrolysis makes oil, gas, and char, not polymer. Purification, cracking, and repolymerisation add energy and lose more feed before plastic returns.
Some chemistry is too diffuse
Films and foams may be valuable per kilogram but contain too few kilograms per truck, sorting hour, or square metre of plant.
Who is building what
Enzymatic depolymerization, solvent dissolution, supercritical hydrothermal liquefaction, and AI optical sorting restore resin identity. Search the record, or filter by recycling process.
CarbiosEnzymatic PET RecyclingEngineered bacterial PETase enzymes depolymerizing post-consumer PET plastics and polyester textiles in aqueous media at mild temperatures (70°C) into PTA and MEG monomers
- Reported evidence
- Operating industrial demonstration plant in Clermont-Ferrand; broke ground on world's first commercial-scale enzymatic recycling plant (50k t/yr) in Longlaville, France.
- Announced next step
- Full commercial licensing of biological depolymerization to textile and packaging manufacturers globally.
- Unresolved risk
- Enzyme production costs, sensitivity to heavy PVC or nylon contamination, and energy required to concentrate dilute monomer solutions.
PureCycle TechnologiesSolvent Polypropylene PurificationSupercritical alkane solvent extraction dissolving post-consumer polypropylene (PP), washing out colors, odorants, and chemical additives without breaking polymer chains
- Reported evidence
- Commercial-scale facility operating in Ironton, Ohio, producing FDA letters of no objection (LNO) for food-contact recycled ultra-pure resin.
- Announced next step
- Scaling commercial production across North America and Europe, closing the loop on previously unrecyclable colored PP containers.
- Unresolved risk
- Solvent recovery losses, operational downtime during mechanical filtration of fine solids, and volatile organic compound (VOC) safety management.
AgilyxPolystyrene PyrolysisThermal catalytic pyrolysis decomposing post-consumer polystyrene waste back into high-purity virgin-equivalent styrene monomer (PS-to-styrene)
- Reported evidence
- Commercial joint-venture plant (Regenyx) in Oregon processed tens of millions of pounds of polystyrene waste into certified circular monomer.
- Announced next step
- Deploying large-scale chemical recycling facilities in Europe and North America in partnership with major petrochemical producers.
- Unresolved risk
- Halogenated flame retardant contaminants in post-consumer polystyrene foam and electrical/thermal process energy intensity.
Mura TechnologyHydroPRS HydrothermalHydrothermal plastic recycling using supercritical water (high pressure, >374°C) to crack mixed unrecyclable flexible films and laminates into hydrocarbon feedstocks
- Reported evidence
- Commissioned first commercial-scale plant at Wilton, UK; partnered with Dow Chemical to scale global feedstock processing capacity.
- Announced next step
- Multi-hundred-thousand-tonne annual processing capacity supplying circular feedstocks directly to chemical steam crackers.
- Unresolved risk
- High operating pressure vessel capital costs, corrosion under supercritical aqueous conditions, and heavy mineral ash disposal.
Eastman ChemicalMethanolysis Polyester RenewalCommercial-scale chemical methanolysis breaking hard-to-recycle mixed polyester waste down into dimethyl terephthalate (DMT) and ethylene glycol
- Reported evidence
- Completed and started up world's largest material-to-material molecular recycling facility in Kingsport, Tennessee, processing up to 110,000 t/yr.
- Announced next step
- Constructing additional molecular recycling facilities in France and Texas to supply circular specialty plastics and packaging.
- Unresolved risk
- High capital expenditure ($1B+ per plant) and economic sensitivity to virgin PET prices driven by fossil paraxylene markets.
AMP RoboticsAMP Cortex & AI VisionHigh-speed delta robots guided by deep-learning computer vision identifying resin types, packaging shapes, brands, and food-grade status on conveyor belts
- Reported evidence
- Hundreds of systems deployed in Materials Recovery Facilities (MRFs) globally, performing up to 80 picks per minute to purify post-consumer bales.
- Announced next step
- Fully automated secondary sortation facilities producing high-spec single-polymer bales without manual manual sorters.
- Unresolved risk
- Mechanical maintenance on suction grippers under abrasive dust, and rapid changes in packaging film formats confounding vision classifiers.
TOMRA RecyclingAUTOSORT Optical SorterNear-infrared (NIR) spectroscopy, visual spectrometers, and deep learning algorithms identifying polymer chemical signatures at multi-ton-per-hour conveyor speeds
- Reported evidence
- Global standard across municipal recycling facilities for separating PET, HDPE, PP, PVC, and black plastics using mid-infrared sensors.
- Announced next step
- Near-100% flake purity sorters allowing direct bottle-to-bottle food-grade mechanical reprocessing.
- Unresolved risk
- Multi-layer barrier packaging (e.g. EVOH or aluminum foils in flexible pouches) showing conflicting spectral signatures.
UN Global Plastics Treaty / Ellen MacArthurGlobal Plastics Policy FrameworkIntergovernmental negotiations mandating standardized mass-balance transparency, recycled content minimums, and elimination of hazardous polymer additives
- Reported evidence
- Drafting legally binding global treaty under UNEP auspices; tracking corporate progress through the Global Commitment reports.
- Announced next step
- Legally binding international treaty capping primary plastic production and mandating closed-loop design standards.
- Unresolved risk
- Diplomatic deadlocks between petrochemical-producing nations and high-ambition coalitions over production caps versus waste management.
Chemical recycling mass-balance accounting allocates recycled credits across complex product slates; physical closed-loop mechanical recycling yields depend strictly on scrap sort purity.
An optimistic view, with conditions
Waste streams become specified industrial feedstocks
The plausible future is not one universal recycling machine. It is a portfolio that keeps valuable streams separate, makes their identity legible, and applies the least destructive recovery route compatible with each feed.
Close the easy loops
Deposit systems, source separation, design-for-recycling, better washing, and mechanical processing expand PET and HDPE substitution.
Purification earns uptime
Solvent and molecular plants prove multi-year output near nameplate, with transparent yields, energy, and qualifying end markets.
Products carry memory
Material identity and formulation follow an article through use, reducing the expensive analytical work needed to turn waste back into specification.
Track output, not intention
- CapturedThe discarded article enters the recovery system.
- SeparatedPolymer identity and contaminants meet a stated feed specification.
- ConvertedA high share of input becomes saleable polymer or building block.
- SubstitutedOutput replaces the intended virgin grade near one for one.
- RepeatedThe same-grade loop survives multiple cycles with bounded virgin make-up.
- CommercialA plant sustains ordinary utilisation and all-in cost without exceptional support.
Collection can work; quality and price still decide the outlet
The global bottleneck starts before a sorting line. OECD’s 2019 mass balance finds 353 Mt of plastic waste, 55 Mt collected for recycling (15.6%), 22 Mt rejected as processing residue, and about 9% ultimately recycled. The first lost fraction is collection and product design; the second is sorting, contamination and process yield. Short-lived packaging dominates immediate waste, while construction and automotive polymers emerge years or decades after manufacture and follow a different stock-flow curve.
High collection is possible for a narrow, valuable stream: Norway’s deposit operator reports 92.4% of plastic bottles returned for deposit and 98.2% collected by all routes in 2025. Those are bottle collection measures, not a national all-plastics recycling rate or proof that every returned bottle becomes a new bottle. A deposit aligns incentives and preserves material identity; mixed flexible packaging has neither advantage.
Recycling also competes with virgin resin. Plastics Recyclers Europe reports 13.5 Mt of European installed recycling capacity in 2024 and almost 1 Mt lost to facility closures across three years, citing cheap virgin imports among the pressures. This is an industry association’s capacity account, not audited tonnes recycled. Chemical recycling requires the same distinction: announced nameplate, accepted waste input and saleable polymer output should never be conflated. Energy or fuel produced from plastic does not count as material recycling under the European Commission’s fuel-use-excluded accounting rule.
Policy is moving despite stalled global negotiations. UNEP reports that INC-5.2 adjourned in August 2025 without consensus on a treaty text. The EU’s Packaging and Packaging Waste Regulation began phased application in August 2026, with recycled-content and recyclability provisions scheduled later. These rules create demand for specified secondary resin; they do not by themselves guarantee that collection, processing yield or food-contact quality will meet it.
Sources, method, and boundaries
Production series with different scopes are not spliced into one line. The 2019 OECD balance is treated as a dated baseline, not a current-year estimate, and UNEP's production figure is kept as a separate series. “Collected for recycling” is separated from final output. Chemical-recycling nameplate capacity is separated from annual production. Carbon recovered as fuel is not counted as polymer circularity.
