Textile collection and sorting

Every chemical route for recycling textiles works on one polymer class: PET hydrolysis needs polyester, cellulose dissolution needs cotton or viscose. The bottleneck in circularity therefore sits upstream of the reactor — in whether a stream of used garments can be separated into compositionally uniform fractions at a speed and cost compatible with the price of the material recovered.

The physics of identification

Industrial sorting rests on near-infrared spectroscopy, roughly 1000 to 2500 nm. What lies in that window is not fundamental bond vibrations but their overtones and combination bands, chiefly of C–H, O–H and N–H. Cellulose with its abundant hydroxyls, polyamide with its amide bond, and polyester with its aromatic ring and ester group give distinguishable spectral fingerprints. A belt runs under a sensor line, each point’s spectrum is matched against a trained model, and an air jet ejects the item into the correct bin.

The method’s limits shape everything else. It is a surface technique, seeing a few hundred micrometres, so it reports the face fabric and not the lining, the fill or the sewing thread. Carbon black, which colours most dark textiles, absorbs across nearly the whole range and extinguishes the returned signal — black garments have traditionally fallen into the unidentified stream, and getting past that requires other approaches, from mid-infrared to hyperspectral imaging. Blends are recognised as blends, but quantifying the ratio from a spectrum is markedly less reliable than detecting the components, and it is the ratio that decides whether a batch is usable.

The contamination threshold

Elastane is the instructive case. It is added at a few per cent for fit, distributed as fine filaments throughout the fabric, and cannot be removed mechanically. For cellulose dissolution and for enzymatic hydrolysis it survives as an undissolved residue that blinds filters and degrades product quality; on the polyester route it injects urethane chemistry into the monomer stream. Reactive dyes, flame-retardant and water-repellent finishes act similarly, as do hard points — zips, rivets, underwires — which must be cut out before anything enters a reactor.

Sorting is consequently not binary. In practice it distributes a stream among several destinations: resale as clothing, which almost always wins on any impact balance; fibre-to-fibre for a narrow clean fraction; mechanical shredding to short staple for nonwovens and insulation, which is downcycling but genuinely useful; and residue. The economics are unforgiving, because revenue from a sorted fraction competes with the price of virgin polyester, and disassembling a garment by hand still costs more than buying new yarn.

The regulatory frame

The revised EU Waste Framework Directive obliged member states to set up separate collection of textiles from 1 January 2025. That changes the input to the problem, not its solution: mandatory collection guarantees volume, not homogeneity, and without sorting capacity growing ahead of it the collected material simply accumulates. Hence the counter-pressure on garment design — mono-material fabrics, removable trim, and a digital product passport that would state composition rather than leaving it to be inferred from a spectrum.