circular-economy · chemistry · family: the wrong ruler
sorting forthe wrong bromine
E-waste plastic sorters measure total bromine because they cannot see the banned flame retardants — so recyclable plastic can be rejected too
Problem statement
Plastics from discarded electronics (housings of TVs, monitors, small appliances) contain brominated flame retardants (BFRs). A subset of these — the polybrominated diphenyl ethers and HBCDD — are listed as persistent organic pollutants (POPs), and EU law forbids recycling waste above a low-POP-content limit, so recyclers must separate POP-containing plastic and destroy it. The catch is that recyclers cannot detect the regulated molecules on a sorting line; the only practical online measurement is total bromine (by X-ray transmission or fluorescence), so sorting lines work on a proxy. Hennebert & Filella's French field data (n=4,283 bromine measurements, 98 BFR analyses across four WEEE plants) show how loose that proxy is for newer products: regulated substances accounted for up to 86% of total bromine in "old" waste (small appliances, CRTs), 30–50% in flat screens, and "a mean of only 8% in recent products (2009–2013)." The 2026 Van Melkebeke paper states the consequence: because not all brominated flame retardants are strictly regulated (TBBPA, for example, is not listed as a POP), using total bromine as a surrogate "may lead to an overestimation of POP-BFRs," and it cites work showing that portable XRF has incorrectly identified WEEE samples as exceeding the low-POP-content limit. Neither the Hennebert & Filella abstract nor the 2026 paper gives a figure for how much recyclable plastic is rejected this way or for how that amount is changing.
Why this matters
WEEE plastics are roughly a fifth of e-waste by mass and formal WEEE recycling rates lag at around 25 wt%; every tonne of flame-retarded polystyrene or ABS mis-sorted to the "high-Br" fraction is incinerated instead of recycled, and every tonne mis-sorted the other way carries POPs into "an unwanted global loop of brominated substances … via the international recycling of plastic scrap." Regulatory limits are only tightening (the EU low-POP-content limit for the sum of listed PBDEs in waste was cut from 1,000 mg/kg to 500 mg/kg from 10 June 2023 by Regulation (EU) 2022/2400, and falls further to 350 mg/kg from 30 December 2025 and 200 mg/kg from 30 December 2027; more flame retardants are also being added to restriction lists); Van Melkebeke et al. report that the expected reduction of these thresholds "causes concern" in the WEEE plastics recycling industry, whose stakeholders point to the constraints of existing sorting technologies. The 2026 pyrolysis-GC-MS study of a working WEEE polystyrene recycling plant found total bromine up to 2,755 ± 430 mg/kg in the density fraction meant to collect halogenated plastics, TBBPA (a non-POP BFR) as the most abundant of the five retardants it measured at up to 1,534 ± 244 mg/kg, and a loss of bromine during extrusion suggesting that "brominated degradation products might escape from the extrusion lines" — a release the authors leave to future research.
What’s been tried and why it hasn’t worked
Handheld XRF on whole equipment and online X-ray transmission after shredding (threshold ~2,000 mg/kg total Br) are what the four French plants in the Hennebert & Filella study used, and they do reduce mean bromine in the "low-Br" stream and bring decaBDE below the restricted-use limit; density separation isolates most halogenated plastics. But none of these identifies the regulated molecules; Hennebert & Filella conclude that "regulated substances are a minority of all the brominated substances … the only practical way to sort is to measure total bromine on-line" — a workaround, not a solution. Laboratory speciation (GC-MS after extraction, or the pyrolysis-GC-MS method published in 2026 that can quantify five flame retardants at once) can distinguish POP-BFRs from permitted ones, but conventional GC-MS needs time-consuming extraction steps and pyrolysis-GC-MS "does not support real-time classification," and even the 2026 authors note "challenges remain due to potential interferences of reaction products." Bromine is heterogeneously distributed even within one device (of 347 equipment items, 46% had at least one brominated part, 39% had all parts brominated, and 15% had none). A product-side restriction on decaBDE (2017) is also in place, and Regulation (EU) 2022/2400 cites declining PBDE concentrations in certain waste as a result of existing restrictions on placing PBDEs on the market and using them.
What would unlock progress
A line-speed method that discriminates POP-BFRs from non-restricted BFRs — or a cheap rapid speciation screen usable on bales or flake batches — would let recyclers keep the permitted fraction. Candidate techniques exist (Van Melkebeke et al. list Raman, near-, short-wave and mid-infrared spectroscopy, laser-induced plasma spectroscopy and hyperspectral imaging alongside XRF; statistical inference from product-type and age classifiers combined with total Br is another route), but the same paper notes that many of these techniques still use total bromine as the surrogate, and it suggests pairing a real-time bromine screen with pyrolysis-GC-MS to identify the specific compounds. A second unlock is regulatory: a sorting standard that accepts a validated proxy-plus-classifier decision rule instead of a bare total-bromine cutoff.
Entry points for student teams
A team could assemble a reference set of WEEE plastic flakes with lab-speciated BFR content (partnering with a recycler or using published tables; Van Melkebeke et al. state their data are available on request) and test whether spectroscopic methods (FTIR/Raman/hyperspectral) or a classifier on product-type plus total-Br can predict POP-BFR content above/below the legal limit; the output is a confusion matrix against the current total-Br rule and an estimate of recyclable material saved. A second team could model, from published bromine distributions by product category and age, how the over-rejection rate of total-Br sorting evolves as legacy CRTs leave the stream. Skills: analytical chemistry, spectroscopy, machine learning, waste-regulation analysis.
Genes
Genes marked “+” were added in a later review.
Structural cousins: stuck the same way, in other fields
Sources
"WEEE plastic sorting for bromine essential to enforce EU regulation," P. Hennebert & M. Filella, *Waste Management* 71 (2018), doi:10.1016/j.wasman.2017.09.031, (abstract read via Europe PMC), accessed 2026-08-17; "Identification and quantification of (brominated) flame retardants during mechanical recycling of polystyrene from WEEE by means of pyrolysis-GC-MS," M. Van Melkebeke et al., *Waste Management* (2026), doi:10.1016/j.wasman.2025.115260 (abstract read via Europe PMC), accessed 2026-08-17 go to source ↗
Working notes
Notes made while writing and checking this brief, shown as written. They use working shorthand.
Flagged [NEEDS DEEPER SOURCING]: only the abstracts of the two peer-reviewed papers were accessible at intake (publisher pages returned 403); all quantitative claims above are taken verbatim from those abstracts and the low-POP-content limit trajectory was corrected at verification to the amending regulation's published schedule (Regulation (EU) 2022/2400 amending Annex IV of Regulation (EU) 2019/1021: 500 mg/kg from 10 June 2023, 350 mg/kg from 30 December 2025, 200 mg/kg from 30 December 2027), sourced from EUR-Lex summaries rather than a fetched consolidated Annex IV text — confirm against the consolidated Annex before publication. The e-waste plastics ~20% and ~25 wt% recycling-rate figures come from the 2024 Polymers review (Achilias et al.) and 2026 Van Melkebeke abstracts respectively. `failure:regulatory-mismatch` (wrong-measurement-basis sub-pattern) is the precise fit: the rule restricts specific molecules that the only feasible line measurement cannot see. `temporal:worsening` was considered because the mismatch has a quantified trajectory (regulated share of total Br 86% → 30–50% → 8% by product age) and would pass requirements (1) and (2), but the resulting over-rejection is not itself quantified in the accessible sources, so `static` is used pending deeper sourcing. `constraint:coordination` not applicable. Related collection briefs: `circular-ewaste-disassembly-optimization-gap`, `circular-ewaste-ree-selective-recovery` and `labor-ewaste-informal-recycler-health-exposure` cover other e-waste facets; this brief is the distinct plastics-sorting/regulatory-proxy problem.
Source type: Research paper limitations/gap statement (field-data paper concluding that the only practical sorting basis is a proxy).
Verified at intake 2026-08-17: gate (net) + adversarial source check (abstracts only for both papers; [NEEDS DEEPER SOURCING] retained) + contested-tag second coding.
Reconciliation 2026-10-01: Fact repair after reading the whole write-up against every source it names. Fetched 2026-10-01. Van Melkebeke et al. was read in full this time, from the Ghent University repository copy of the published article (https://biblio.ugent.be/publication/01KD2K0ZABP45Q9D922HQCCMYE, 27-page PDF; Waste Management 210 (2026) 115260). Hennebert & Filella could be read only as its abstract (Europe PMC record for PMID 29030119, with the citation checked at https://api.crossref.org/works/10.1016/j.wasman.2017.09.031); the publisher page is blocked, the University of Geneva repository copy is restricted to that university, and the INERIS HAL record (ineris-01853528) holds a notice with no file. The Polymers review named above (Achilias, Charitopoulou and Ciprioti, Polymers 16(17), 2024) was read in full through Europe PMC (PMC11398207). Regulation (EU) 2022/2400 was read in full from an Internet Archive copy of the EUR-Lex page (web.archive.org/web/20240331011835/eur-lex.europa.eu/legal-content/EN/TXT/HTML/?uri=CELEX:32022R2400). (1) The long title ended "So Ever More Recyclable Plastic Is Rejected." No source measures the amount of recyclable plastic rejected or says it is growing; the 2026 paper says only that total bromine as a surrogate "may lead to an overestimation of POP-BFRs" and that moving beyond it may prevent unnecessary diversion to incineration or landfill. The title now ends "So Recyclable Plastic Can Be Rejected Too." (2) The Problem Statement said "the entire industry sorts on a proxy," that the field data "show why that proxy is deteriorating," and that sorting "increasingly rejects" permitted plastic while the regulated fraction "shrinks toward the noise floor of the measurement." None of this is in the sources. The passage now gives what the 2026 paper says about overestimation and about portable XRF wrongly flagging samples, and says plainly that neither the Hennebert & Filella abstract nor the 2026 paper gives a figure for the rejected amount or its trend. (3) Tighter limits were said to push "the total-bromine threshold lower and the over-rejection higher." Not in the sources; the sentence now reports the 2026 paper's statement that the expected reductions cause concern in the recycling industry. (4) The extrusion finding was said to mean "a worker-exposure dimension." The paper does not mention workers; it says degradation products might escape from extrusion lines and leaves this to future research. The sentence now says that. (5) Handheld XRF and on-line X-ray transmission were called "the industrial standard." The abstract says only that the four French sites used them; the sentence now says that. The word "because" in front of the Hennebert & Filella quotation was removed, since the abstract joins the two statements with "and." (6) Laboratory speciation "takes hours per sample": no source gives a time. The 2026 paper calls conventional extraction time-consuming and says pyrolysis-GC-MS "does not support real-time classification"; the sentence now says that. "Developed in 2026" now reads "published in 2026." (7) Removed as not in any source: that whole-item and flake decisions "both misclassify"; that product restrictions "do nothing for the decades of legacy stock" and "perversely" make sorting less accurate (the paragraph now gives the abstract's three shares of the 347 equipment items, says that the 2017 decaBDE restriction is in place, and keeps only the part of the original claim that a named source carries: Regulation (EU) 2022/2400, recital 6, cites "the declining concentrations of PBDEs in certain waste, resulting from existing limitations on the placing on the market and use of PBDEs"); that "nobody has validated" spectroscopic methods against speciated reference data; and the comparison with watermark-based sorting of plastic packaging. The list of candidate techniques now follows the 2026 paper's own list. (8) The entry point's "published datasets" now notes that the 2026 authors state their data are available on request. Confirmed in the sources and left as written: both citations on the Source line; n=4,283 and n=98, four sites in France; the 86%, 30–50% and 8% shares and the quoted words around them; 46% of items with at least one brominated part; the roughly 2,000 mg/kg sorting threshold; decaBDE brought below the restricted-use limit; the "unwanted global loop" quotation; 2,755 ± 430 and 1,534 ± 244 mg/kg; five flame retardants; "around 25 wt%"; plastics "almost 20%" of WEEE (the review's abstract; its body text and the 2026 paper both say about 30%); the earlier 1,000 mg/kg limit (2026 paper); and the regulation's 500, 350 and 200 mg/kg steps with their dates, which are now confirmed in the regulation text itself. Could not be checked: anything in Hennebert & Filella beyond its abstract. A second reader went over these repairs against the same sources on the same day. One repair was undone: the first repair had changed "is incinerated instead of recycled" to "typically landfilled or incinerated," but the 2026 paper also states that disposal of the isolated POP material "is typically done by means of incineration," so the original wording is supported and stands as written. One sentence was extended with the regulation's recital on declining PBDE concentrations (above), and two sentences were tightened (the scope of the "neither paper" statement, and an inference about recent products that the first repair had added and no source states). Genome tags untouched.