the stretch that

sorters can't see

Problem statement

Fibre-to-fibre recycling — turning old clothes into new yarn instead of rags or insulation — depends on knowing what a garment is made of, and every recycler sets limits on which fibres it can accept. Elastane (spandex/Lycra) is added at low percentages (typically a few percent, often cited as 2–5 percent) to a large share of cotton and polyester garments for stretch, and even that little contaminates mechanical shredding, chemical pulping and polyester depolymerisation. The catch is that the near-infrared (NIR) scanners now being installed to automate textile sorting in Europe cannot reliably detect it: the Fashion for Good study that NIR-scanned 21 tonnes of post-consumer garments across six countries found elastane "in only 2% of this Project's cotton sample," yet concluded that "a large share of the volume classified as pure cotton in this Project most likely contains elastane," and listed among NIR's disadvantages that it is "unable to recognise low content of fibre in blends, especially elastane, which is a significant contaminant for chemical recycling." The EU JRC's 2023 assessment repeats the caveat: the share of recyclable feedstock locked in fibre mixes "is likely even higher, given that elastane may also be present in the fractions classified as 'pure' in the report due to analytical limitations." So the feedstock that recyclers are being promised as "100 percent cotton" is quietly laced with a fibre that ruins their process, and no scaled step exists to take it out.

Why this matters

The EU made separate collection of textiles mandatory from January 2025, and the JRC counts more than 8 million tonnes of used and waste textiles incinerated or landfilled annually. Fashion for Good estimated 673,000 tonnes per year of non-rewearable and low-value textiles in its six focus countries, of which about 140,000 tonnes (21 percent) were suitable feedstock for mechanical fibre-to-fibre recycling — a figure that already assumes the "pure cotton" fraction is clean. Where the study could see blends, elastane was pervasive: among cotton-rich blends (polycotton excluded), 35 percent were contaminated with elastane and cross-checks with care labels suggested that "most of the contamination claimed as 'polyamide' actually consisted of elastane." Every recycler that receives a mis-sorted bale either rejects it, downgrades the output, or suffers process upsets; every sorter that cannot certify elastane-free bales cannot command the price premium that would pay for automated sorting in the first place. The JRC frames the underlying issue as a "technological externality": brands add elastane for fit and comfort, and the cost lands on a downstream sorter or recycler that has no way to bill them for it.

What’s been tried and why it hasn’t worked

Manual sorting reads care labels, but labels are missing, worn, or wrong, and manual throughput (100–150 kg per person-hour, per the JRC citing Dahlbom et al.) cannot handle collected volumes. NIR sorting has scaled — Sysav's Siptex plant in Malmö (24,000 t/yr capacity), Coleo in A Coruña, NewRetex in Denmark, hand-held NIR at Boer Group and LSJH — and is cheap (about EUR 150,000–200,000 per optical machine per the JRC), but NIR reads only the outermost layer, struggles with dark and carbon-black-dyed fabrics, coatings and finishes, and, in the Fashion for Good study, resolved blends only up to two fibre types; a 3 percent minority component in a knit is below what commercial spectra-plus-chemometrics reliably resolve. On the process side, mechanical recyclers already recover only "<20% for cotton" as spinnable fibre from unravelling (JRC citing Duhoux et al.), and elastane fouls the machinery; chemical pre-treatments to dissolve elastane out of blends (the JRC cites Phan et al. 2023, "Analysing the potential of the selective dissolution of elastane from mixed fibre textile waste," and later work with DMSO/DBN or other solvent systems) work at lab scale but add a solvent-recovery step whose cost has not been shown to fit a low-value feedstock — and they can only be targeted if the sorter knows the elastane is there. Digital product passports would in principle carry composition data, but the garments entering sorters today were made without them and will keep arriving for a decade. Nobody has closed the loop between detection, routing and removal.

What would unlock progress

Two complementary breakthroughs would unlock the stream: a low-cost sensing method that flags low-percentage elastane at line speed (candidates: mid-infrared or Raman spectroscopy tuned to polyurethane bands, hyperspectral imaging combined with machine-learning models trained on garments of known composition, or a mechanical/optical stretch-recovery test that exploits elastane's defining property rather than its chemistry), and a cheap, scalable elastane-removal step (selective dissolution, thermal or enzymatic degradation) that recyclers can bolt on once elastane-positive bales are routed to them. Upstream, ecodesign rules that cap or label elastane in recyclable garment categories would shrink the problem over a product generation, but they do not touch the existing stock. The adjacent precedent is plastics recycling, where NIR sorters could not see carbon-black packaging until sorting moved to other detection principles and design guides pushed the market off the pigment.

Entry points for student teams

A sensing team could build a benchmark: assemble 200 garments of label-verified composition (0, 2, 5, 10 percent elastane in cotton, polyester and viscose bases; light and dark dyes), acquire NIR, MIR and hyperspectral spectra, and quantify each method's detection floor and false-negative rate — a dataset sorters and equipment makers currently lack. A materials/chemistry team could reproduce a selective-dissolution protocol on cotton-elastane knits and measure fibre-length and yield loss on the recovered cotton, then cost the solvent loop per tonne. A product-design team could prototype an inexpensive mechanical "stretch signature" test (elastic recovery under standard load) as a pre-screen at manual sorting stations. Relevant skills: spectroscopy and chemometrics, polymer chemistry, textile engineering, process costing.