Why Recycling Fails Without Product Data

Recycling technology has improved substantially over two decades. Near-infrared sorters identify polymer types at speed, eddy current separators pull non-ferrous metals from a moving stream, and optical systems sort by colour and shape reliably.

Recycling rates have not improved proportionally, and the reason is that sorting answers a narrower question than recycling requires. A sorter can tell you a fragment is polypropylene. It cannot tell you which additives, flame retardants, pigments or coatings are in it, and those decide whether the output is usable.

What sorting can and cannot see

Only the first row is solved. Every other row is a question the sorter cannot answer and the recycler must therefore assume, and assumptions in this setting are always conservative because the cost of contaminating a batch exceeds the value of recovering one item.

Why conservative routing means downward

A recycler producing recyclate to a specification carries liability for what they sell. If an unidentified input might carry a restricted substance, accepting it risks the whole output batch, which is worth far more than the input.

The rational response is to route uncertain material to an application where the uncertainty does not matter — a lower-grade product, a filler, or energy recovery. That is downcycling, and it is a sensible commercial decision rather than a failure of will.

The four identification failures

1. Additive uncertainty: the polymer is known, its formulation is not.
2. Restricted substance uncertainty: presence cannot be excluded, so the material is treated as if present.
3. Layer uncertainty: a coating or laminate is invisible to sorting and contaminates the melt.
4. Provenance uncertainty: the material’s history is unknown, so food or medical grade reuse is excluded.

The fourth is the sharpest constraint in packaging. Recycled polymer cannot re-enter food contact applications without provenance, which is why recycled content in food packaging remains difficult regardless of how much material is collected.

What data actually changes the decision

Not a general declaration that a product is recyclable. A recycler needs specific facts at the moment of sorting: what the item is made of by component, which substances of concern are present and where, and how the layers separate.

That is a small amount of information and it is held by the manufacturer. It is also information that must arrive in a machine-readable form at line speed, which is why publishing it as a document satisfies an obligation without changing an outcome.

Why design and data have to move together

Perfect information about a badly designed product does not make it recyclable. A bonded multi-material laminate is unrecyclable whether or not its composition is documented, because no process separates the layers economically.

Equally, a well designed monomaterial product is downcycled if nobody can confirm what it is. The two failures are independent, which is why the ESPR pairs design requirements with information requirements rather than relying on either alone.

Where collection fits

None of this matters for material that is never collected, and collection remains the largest single loss in most streams. A perfectly documented, perfectly designed product discarded in general waste is recycled at exactly the same rate as a badly designed one.

Product data does not fix collection. It determines the outcome for the fraction that is collected, and it makes the loss measurable, which matters because recovery rates calculated on collected material systematically overstate system performance.

What about chemical recycling?

Chemical recycling tolerates contamination better than mechanical recycling, which is its genuine advantage. It breaks polymers back to feedstock, so additives and pigments matter less than they do in a melt process.

It is more energy intensive, capacity is limited, and it still requires knowing that restricted substances are absent. It relaxes the composition constraint rather than removing it, and treating it as a reason not to solve the data problem is a mistake.

Who actually pays for the uncertainty?

The recycler absorbs it first, in lower output value and rejected batches. It then travels forward as a price: recyclate with uncertain provenance sells at a discount to virgin material, which is why recycled content targets are expensive to meet even where material is physically available.

That discount is the market pricing information risk, and it is the mechanism by which a manufacturer’s missing composition data becomes another manufacturer’s higher input cost. The party creating the uncertainty is not the party paying for it, which is the standard shape of this problem.

Producer responsibility fees are the correction, and eco-modulation is the sharper form of it. Where fees vary with recyclability, the cost of unrecyclable design returns to whoever chose it, which is the only arrangement under which the design decision is made with the full cost visible.

What changes when identity is per item?

Batch-level composition data covers most recycling decisions, because a sorter is handling material rather than individual objects. Item-level identity adds something different: history.

For a container that may have held a non-food substance, or a component that may have been repaired with a non-original part, the question is not what the product was made of but what has happened to it since. That is the constraint on closed-loop reuse in food and medical applications, and only a per-item record answers it.

Does any of this apply to metals?

Less acutely, which is why metal recycling rates are far higher than polymer rates. Steel and aluminium can be identified by composition analysis of the melt itself, so an unknown input is characterised after the fact rather than guessed at beforehand.

The exception is contamination that cannot be removed. Copper in steel and iron in aluminium accumulate through repeated recycling and cannot be economically separated, which slowly degrades the whole scrap pool rather than one batch. Composition data at product level is how that accumulation is managed rather than merely observed.

What a manufacturer should publish

1. Material composition by component, not a single figure for the product.
2. Substances of concern with their location, so a treatment route can be chosen.
3. Separability: which parts come apart, and how.
4. Coatings, laminates and surface treatments, which sorting cannot detect.
5. Whether recyclate is suitable for closed-loop return to the same application.

The fifth item is the one almost nobody publishes and the one a recycler most wants, because it determines whether their output has a high-value buyer or a low-value one. It is also a claim that requires evidence, which is precisely what a passport is for.