Kjølholt et al. 2015 — Chemical substances in car safety seats and other textile products for children (Danish EPA Survey No. 135)

This 2015 Danish Environmental Protection Agency consumer-products survey (Report No. 135) commissioned from COWI A/S and Teknologisk Institut examined whether car safety seats, baby slings, and baby mattresses sold on the Danish market contain chemical substances that may pose a risk to children aged 0–3 years. Thirty products were purchased and analysed; sub-samples of the car-safety-seat textiles were screened by wavelength-dispersive X-ray fluorescence (WD-XRF) for the elements tin and bromine as indicators of organotin compounds and brominated flame retardants, while quantitative analyses across all three product groups targeted phosphorus-based flame retardants (TCEP, TCPP, TDCP, TPP), formaldehyde, isocyanates (TDI, MDI), the phthalate DIDP, azo-dye-derived aromatic amines, and four classes of brominated flame retardant (PBB, PBDE, HBCD, TBBPA). For the Heavy Metal Index corpus, the relevant analyte is tin (Sn): tin was not detected (<20 mg/kg) in 13 of the 14 XRF-screened car-safety-seat textile sub-samples, with a single positive sample (A3B) at 65 mg/kg. The authors concluded that “tin was not found in significant quantities in any of the samples” and elected not to follow up with quantitative organotin speciation. Bromine, by contrast, was detected at ≥20 mg/kg in 9 of the 13 samples reported with values up to 27,000 mg/kg; bromine is outside HMI’s metal scope but is summarised here for context because it drove the report’s downstream brominated-flame-retardant quantitation. None of the four named brominated flame retardants (PBB, PBDE, HBCD, TBBPA) accounted for the high bromine signal at quantitative analysis, leading the authors to infer the presence of polymeric brominated flame retardants not covered by the analytical method.

Key numbers

Tin (Sn) by WD-XRF screening of car-safety-seat textiles (Table 5-4, p.60)

• Method: Philips PW2400 / UNIQUANT ver. 5.49 wavelength-dispersive X-ray fluorescence (WD-XRF) on textile sub-samples of car safety seats only; baby slings and baby mattresses were not XRF-screened.
• Limit of detection (LOD): 20 mg/kg for both Sn and Br.
• Method variability: 5–15 % RSD.
• Samples screened (n = 14): A1, A2A, A3A, A3B, A4, A5A, A6A, A6B, A7A, A8A, A8B, A29, A30 — all textile sub-samples. Note: the report’s text on p.60 states “9 out of the 13 samples” for bromine detections, treating one of the 14 textile sub-samples as a paired-textile duplicate; the Sn table itself lists 14 rows.
• Tin results, mg/kg dry textile:
  • <20 mg/kg (below LOD): 13 of 14 sub-samples — A1, A2A, A3A, A4, A5A, A6A, A6B, A7A, A8A, A8B, A29, A30 (and a second sub-sample at LOD).
  • 65 mg/kg: A3B (single positive).
• Author interpretation (p.60): “tin was not found in significant quantities in any of the samples. Therefore, it was considered unnecessary to carry out specific analyses for organotin compounds.” No quantitative organotin speciation was performed in this study.

Bromine (Br) by WD-XRF (context only; out of HMI metal scope)

• Br results, mg/kg dry textile (Table 5-4, p.60): A1 5,700; A2A 220; A3A 170; A3B 410; A4 450; A5A <20; A6A 430; A6B <20; A7A <20; A8A <20; A8B 290; A29 27,000; A30 90.
• 9 of 13 reporting textiles ≥20 mg/kg Br; samples A1, A3B, A4, A6A, A29 selected for follow-up quantitative brominated-flame-retardant analysis (p.61).
• Subsequent quantitation (Table 5-12, p.71) returned only trace concentrations of named brominated flame retardants: 2,4,4’-TriBDE (BDE-28) up to 0.14 mg/kg (A6A); HBCD up to 1.2 mg/kg (A3B, single duplicate determination only); sum-PBB all below LODs of 0.02–0.04 mg/kg; sum-other-PBDE all below LODs of 0.05–0.08 mg/kg; TBBPA all below LODs of 0.01–0.04 mg/kg, across the five samples. The authors conclude the high Br signal is most plausibly explained by polymeric brominated flame retardants outside the analytical scope (p.72).

Context: non-metal analytes (recorded for scope; not HMI-actionable)

• Phosphorus-based flame retardants (TCEP, TCPP, TDCP, TPP) were detected at functional-use concentrations (>10,000 mg/kg = >1 % by weight) in eight sub-samples — five car seats (A3A TDCP 20,300; A3B TDCP 21,100; A4 TDCP 42,600; A5A TDCP 31,500; A8A TCPP 18,100), two baby slings (B12A TCPP 11,200; B18A TCPP 16,300 and TDCP 13,000), one baby mattress (M24B TDCP 89,700 ≈ 8.97 % by weight; Table 5-8, p.65–66).
• Phthalate DIDP detected at functional concentration only in one car-seat textile (A6B): di-isodecyl-phthalate isomer 1 at 41,000 mg/kg and isomer 2 at 350,000 mg/kg; combined DIDP ≈ 390,000 mg/kg (39 % by weight; Table 5-10, p.68), exceeding REACH childcare-articles provisions; reported to the Danish Chemicals Inspectorate.
• Formaldehyde: not detected (LOD 2–5 mg/kg) in any of 37 textile sub-samples analysed (p.67).
• Isocyanates (2,4-TDI, 2,6-TDI, MDI): all <1 mg/kg in the 14 foam sub-samples analysed quantitatively; highest single value 0.91 mg/kg 2,4-TDI in foam from baby mattress M24B (Table 5-9, p.68). The GC/MS screening at 10–940 mg/kg was attributed to method artefact, not actual content.
• Azo-dye-derived aromatic amines: one car-seat textile (A1) contained 50 mg/kg 4-methyl-m-phenylene diamine (2,4-TDA, CAS 95-80-7), exceeding the REACH Annex XVII 30 mg/kg limit; reported to the Danish Chemicals Inspectorate (Table 5-11, p.70).

Methods (brief)

The survey selected products for chemical analysis via questionnaire to 11 Danish distributors (five returned) and a literature review of prior Nordic / German studies. Thirty products were purchased: 11 car safety seats (sample codes A1–A8, A29, A30 with sub-samples labelled by material), 10 baby slings (B9–B18 with sub-samples), and 10 baby mattresses (M19–M28 with sub-samples). Sub-samples were selected by material type — textiles, foam, plastic, rubber — focused on parts that come into prolonged or direct skin contact with the child. The HMI-relevant analytical track is the X-ray screening (Section 5.2.2, p.60): WD-XRF on 14 textile sub-samples of car safety seats, primary objective the elemental detection of tin (as organotin-compound indicator) and bromine (as brominated-flame-retardant indicator), LOD 20 mg/kg, method variability 5–15 % RSD, instrument Philips PW2400 with UNIQUANT ver. 5.49. No quantitative inorganic analyses were performed — the screening for Sn was sufficient to rule out further organotin work, and the screening for Br informed the selection of textiles for brominated-flame-retardant quantitation. Baby slings and baby mattresses were not XRF-screened; the report’s rationale (p.59) is that organotin compounds were “considered primarily to occur within this product group [car safety seats], if at all”. Migration testing (Section 5.4, p.72) targeted phosphorus-based flame retardants only and is therefore outside HMI’s metal scope.

Limitations relevant to HMI: (1) the 20 mg/kg LOD is high for tin in textile substrates — a substantial fraction of children’s textiles may carry organotin residues from biocide treatment or PVC stabilisers at the µg/kg–single-mg/kg range that WD-XRF would not detect; (2) only car-safety-seat textiles were screened, so the dataset says nothing about Sn in baby slings or baby mattresses despite both being included in the broader survey; (3) the single Sn positive (A3B, 65 mg/kg) was not speciated, so it is unknown whether the detected tin is inorganic or organotin; (4) sample size (n=14) is small for population-level inference.

Implications

Certification: This is one of very few European regulatory surveys to directly probe heavy metals in car-safety-seat textiles using elemental screening. The dominant literature occurrence finding — 13 of 14 samples <20 mg/kg Sn by WD-XRF — is regulatory-grade negative evidence that detectable tin in car-safety-seat textiles is uncommon on the Danish market (and, by import-origin extension on p.7, on the broader EU and East-Asian export market) at the 20 mg/kg WD-XRF LOD. The single positive (A3B, 65 mg/kg) shows that a fraction of car-seat textiles do carry detectable Sn at this LOD. The literature occurrence record has three structural gaps relevant to certification: (a) the 20 mg/kg LOD is high relative to migration-relevant exposure, so the dataset does not constrain Sn occurrence at the µg/kg–single-mg/kg range; (b) the report did not speciate the A3B positive, so the literature cannot distinguish inorganic tin from organotin at this concentration; (c) the survey did not analyse the foam and plastic shell components where organotin PVC stabilisers would more plausibly occur. These are gaps in the published evidence base, not threshold recommendations.

Courses: Useful as an example of how a regulatory survey designs an elemental-screening tier ahead of expensive speciation work, and how negative XRF screening (Sn <LOD) is a defensible reason to skip downstream organotin speciation — provided the LOD is documented. The report is also a clean illustration of cross-element proxy logic (Br as a brominated-FR indicator) and its limits (high Br but no detectable named BFRs at quantitation).

App: Not directly app-actionable — the wiki’s consumer app is food-and-personal-care focused. Car-safety-seat textiles are out of scope for ingredient-list / nutrition-facts inference. The negative Sn finding may be useful as background context if the app expands into durable-goods.

Microbiome: Not applicable. The exposure pathway is dermal / incidental oral via mouthing of fabric, not dietary; no microbiome-relevant findings.

Verification notes

• Author/editor list: the title page (p.2) lists seven editors split between COWI A/S (Kjølholt, Warming, Lassen, Mikkelsen, Brinch) and Teknologisk Institut (Nielsen, Jakobsen). The report does not designate a corresponding author; Kjølholt is listed first and is used as the cite-key anchor.
• ISBN: 978-87-93352-07-0 (p.2). No DOI is assigned.
• Sn detection discrepancy: the report text on p.60 says “9 out of the 13 samples” for bromine detections while Table 5-4 lists 14 textile rows. The discrepancy reflects the report treating one row pair (A3A/A3B from a single car seat) as a paired-textile observation in narrative but as two separate analytical units in the table. The Sn data is taken from the table as the canonical analytical record (14 textile rows, 1 positive at 65 mg/kg).
• Brand-firewall: the report anonymises products as alphanumeric codes (A1, B9A, M19A, etc.) and does not name brand names; no brand-firewall edits required.
• Jurisdictions: the survey is Danish-market, but import data on p.7 (75–80 % non-EU origin, majority China; remaining 20–25 % EU, mostly Germany and Italy) means the underlying product population is internationally sourced. Frontmatter records DK (publication jurisdiction) and EU (regulatory framework: REACH, Annex XVII referenced throughout).
• Near-duplicates: no near-duplicates identified in the existing HMI corpus via cite-key, DOI, or raw_handle grep. Stringer 2001 (stringer2001-greenpeace-pvc-children-products) covers a partly overlapping product space (PVC children’s products including a UK high-chair cushion) but is methodologically distinct (extraction + speciation rather than XRF) and reports tin at non-LOD-floor concentrations.
• Audit subagent 2026-06-01 (REVISE verdict): flagged three findings. (1) “four car seats” undercount in the P-FR >10,000 mg/kg summary — verified against Table 5-8 p.65–66; A3A, A3B, A4, A5A, A8A all carry one P-FR species >10,000 mg/kg, giving five car seats and eight sub-samples total. Corrected to “five car seats” / “eight sub-samples” with per-sample concentrations. (2) sum-PBDE LOD generalisation “all <0.03 mg/kg” was too tight — verified against Table 5-12 p.71; sum-other-PBDE LODs range 0.05–0.08 mg/kg across the five samples, PBB 0.02–0.04, TBBPA 0.01–0.04. Corrected to per-class LOD ranges. (3) Implications/Certification paragraph crossed into HMTc-program framing (“permissive prior”, “does not justify removing Sn from the analyte panel”) and imported an external Cd-in-PVC 100 mg/kg analogue not discussed in the source. Reframed as literature-occurrence findings with three explicit evidence-base gaps; external analogue removed.

Wiki pages updated on ingest

• tin
• car-seats

Page history

The five most recent substantive edits to this page. The full version history lives in git; when DOI minting comes online (see schema docs), each entry below will also link to a version-pinned DataCite DOI.