Seifi 2026 — Iranian children’s-toy survey for Pb, Cr, tHg, tAs, Cd and phthalates/BPA/PBDEs across four material categories

This single-author peer-reviewed study in the first issue of Environmental Health and Pollution Research surveys 120 children’s toys (30 each of plastic, painted wood, rubber, and plush) sampled across three Iranian cities in January–June 2024 for five heavy metals (Pb, Cd, tHg, tAs, Cr) by ICP-MS and for phthalate plasticizers (DEHP, DBP, BBP), bisphenol A, and brominated flame retardants (PBDEs) by GC-MS. The principal heavy-metal findings are that painted-wood toys carried the highest mean concentrations of Pb (212.4 ± 45.1 mg/kg), Cr (102.3 ± 25.7 mg/kg), and Hg (4.5 ± 1.4 mg/kg), with painted-wood Cr and Hg group means exceeding the EU EN-71-3 reference values cited in the same tables (60 and 2 mg/kg, respectively) and the painted-wood Pb mean approaching but not exceeding the cited 160 mg/kg Pb reference. Plush toys consistently exhibited the lowest mean concentrations across all five metals. Stated non-compliance rates were 23.3% (Pb), 30.0% (Cr), and 6.7% (Hg) in the painted-wood subset; the paper reports 0% non-compliance in the plastic subset for any heavy metal. Kruskal-Wallis testing returned p = 0.003 for Pb (painted wood > plastic, rubber, plush; plastic > plush) and p = 0.057 for Cd (no significant pairwise differences). The author concludes that painted-wood toys present the highest heavy-metal exposure risk in the Iranian retail market and calls for the Institute of Standards and Industrial Research of Iran (ISIRI) to strengthen enforcement of ISIRI 6243 and ISIRI 10166 on rubber and painted-wood toys.

Key numbers

All concentrations are reported in mg/kg as group mean ± SD (n = 30 per material category). The paper describes two sample-preparation protocols — (a) EN-71-3 simulated-gastric acid extraction (0.07 mol/L HCl, 2 h at 37 °C) for migratable elements, and (b) closed-vessel microwave-assisted HNO₃/H₂O₂ digestion of 0.5 g material for total elemental analysis — but does not state which protocol produced the Table 1 values, and the body text characterises the Table 1 values against “EU migration limits” (the EN-71-3 framing) without disambiguating. Both extractables and total-content interpretations are textually supported; this ambiguity is documented in Verification notes.

Heavy-metal concentrations by toy material (Table 1, p. 44; n = 30 per category; mg/kg, mean ± SD)

Body-text emphasis (p. 45): painted-wood toys exhibited the highest mean Pb (212.4 ± 45.1 mg/kg) and Cr (102.3 ± 25.7 mg/kg), “both exceeding the EU migration limits (160 mg/kg for lead and 60 mg/kg for chromium)”; mercury “also surpassed its limit (2 mg/kg) in some painted wooden samples.” Plastic toys “contained significant amounts of lead (145.2 ± 38.6 mg/kg) and chromium (85.6 ± 20.4 mg/kg), but their average lead content remained within the EU limit.” Plush toys “consistently demonstrated the lowest concentrations across all tested heavy metals.”

Regulatory non-compliance rates (Table 3, p. 46; samples exceeding cited EU limits)

Statistical comparison across material categories (Table 4, p. 46; Kruskal-Wallis with Dunn’s post-hoc, Bonferroni-corrected)

Organic-compound concentrations by toy material (Table 2, p. 45; n = 30 per category; mg/kg, mean ± SD)

Reported for context — these are not heavy-metal data but support the paper’s overall material-stratification finding.

Hazard-quotient summary

The paper states (Conclusion, p. 47) that “Quantitative health risk assessment confirmed these concerns, calculating Hazard Quotients (HQ) exceeding 1 for lead in painted wood toys and DEHP in rubber toys, indicating a potential for adverse health effects under reasonable exposure scenarios.” Numerical HQ values, the exposure-parameter inputs used, and the reference doses applied are not tabulated; the HQ > 1 conclusion is stated only as a qualitative summary.

Methods (brief)

Sampling (Section 2.1, p. 44). Stratified random sampling of 120 children’s toys from major retail chains, toy stores, online marketplaces, and informal vendors/bazaars in Tehran, Shiraz, and Tabriz between January 2024 and June 2024. Four material strata (plastic, painted wood, rubber, plush) with n = 30 per stratum. Inclusion criteria: newly purchased; explicitly marketed for children under 12; prioritizing mouthing or prolonged-skin-contact items (teethers, dolls, squeeze toys, building blocks). Exclusion criteria: “phthalate-free” or “BPA-free” labels; educational/scientific kits not intended for routine play.

Sample preparation (Section 2.3, p. 44). Two protocols applied to each sample: (a) EN-71-3 acid extraction — sample immersion in 0.07 mol/L HCl for 2 h at 37 °C to simulate gastric conditions and assess migratable elements; (b) microwave-assisted digestion — HNO₃ + H₂O₂ mixture applied to 0.5 g of toy material in a closed-vessel microwave system for total elemental analysis. The paper does not explicitly state which extract was used for the Table 1 mean ± SD values or for the Table 3 non-compliance rates.

Instrumental analysis (Section 2.3). Heavy metals (Pb, Cd, Hg, Cr, As, Sb, Zn, Ni, Cu, Mn) quantified by ICP-MS; only Pb, Cd, Hg, Cr, As are tabulated. Chromium speciation (Cr(III) vs Cr(VI)) assessed by HPLC-ICP-MS; speciated Cr values are not reported in the result tables, only total Cr. Organic compounds (phthalates DEHP/DBP/BBP, bisphenol A, brominated flame retardants/PBDEs) quantified by GC-MS. The paper does not report LODs, LOQs, reference-material identifiers, or recovery percentages.

Speciation. The metals tabulated as Pb, Cr, Hg, As, Cd are total elemental concentrations from ICP-MS without speciation, despite the methods section’s mention of HPLC-ICP-MS for Cr(III)/Cr(VI). No iAs/tAs distinction; no MeHg/tHg distinction. The frontmatter records this as metals: [Pb, Cd, tHg, tAs, Cr] per CLAUDE.md Part 14 (when a paper does not separate iAs vs tAs the page uses tAs; when it does not separate MeHg vs tHg it uses tHg; when total Cr without Cr-VI it uses Cr).

Statistical analysis (Section 2.4, p. 44). SPSS v26. Descriptive statistics (mean, SD, range) per compound × material. Normality assessed by Kolmogorov-Smirnov. For non-normally distributed data, Mann-Whitney U and Kruskal-Wallis tests across material categories. Post-hoc pairwise testing by Dunn’s test with Bonferroni correction. Principal Component Analysis (PCA) “employed to identify clustering patterns and potential sources of contamination” — PCA results are not reported in the manuscript. Significance set at p < 0.05.

Authorship. Single author: Mehran Seifi, Student Research Committee, Research Institute for Health Development, Kurdistan University of Medical Sciences, Sanandaj, Iran. Corresponding author email mehran.seifi72@gmail.com on the title page. Received 20 October 2025; revised 01 January 2026; accepted 01 January 2026; published 02 January 2026. AI Use Declaration: ChatGPT used for language translation; the author retains full responsibility for the manuscript content.

Implications

• Direction of finding (HMTc Cat 21 routing). The paper’s material-stratified design reproduces a pattern reported in the toy-contamination literature corpus: painted-wood toys carry higher Pb/Cr exposure risk than plush, plastic, or rubber toys; rubber toys carry higher phthalate-plasticizer exposure risk than plush, painted-wood, or plastic toys; plush toys consistently exhibit the lowest chemical burden across analyte panels. For HMTc Cat 21 (Children’s Toys, Arts, and Crafts) Row 1 (toys-painted, general toys with paint/surface coatings), Row 6 (toys-squeeze), and Row 3 (toys-stuffed-bean-bag), this paper contributes direction-of-finding evidence for the material-stratification approach, with the caveat that the absolute point estimates carry the data-integrity uncertainty documented in Verification notes (mean ± SD vs reported non-compliance rates not internally consistent under a normal-distribution assumption).

• Iranian retail-market baseline. This is one of a small number of corpus sources reporting heavy-metal contamination in children’s toys in the Iranian retail market (Tehran, Shiraz, Tabriz, 2024). The paper’s sampling frame deliberately spans formal channels (retail chains, toy stores), online marketplaces, and informal channels (street/bazaar vendors), making the dataset informative for the import-and-informal-trade segment of the Iranian children’s-toy market that is harder to surveil through customs-only data.

• App. Per-material-category mean ± SD distributions for Pb, Cd, tHg, tAs, Cr in painted-wood, plastic, rubber, and plush toys are direct input for any future app feature estimating contamination likelihood by toy material. The Cd group means range 7.8–9.1 mg/kg across the three non-plush categories (with no significant pairwise differences by Kruskal-Wallis, p = 0.057) and 1.2 mg/kg for plush — a useful background-Cd-floor data point for the Cat 21 baseline.

• Courses. Worked example for (a) the EN-71-3 simulated-gastric acid-extraction protocol vs total-elemental microwave digestion, and the EN-71-3 limits’ migration-vs-total-content framing that the paper itself does not disambiguate; (b) the limits of group-mean comparison against migration limits when non-compliance is the regulatory test (the paper reports group means alongside non-compliance rates without reconciling the two when mean ± 1 SD straddles the limit); (c) Kruskal-Wallis + Dunn’s with Bonferroni as the non-parametric workhorse for material-stratified contamination surveys when the Kolmogorov-Smirnov test rejects normality. The paper-internal data-integrity inconsistencies (see Verification notes) make this a useful teaching case for the broader question of how a wiki source page handles a peer-reviewed paper whose internal arithmetic does not fully reconcile.

• Microbiome. Not addressed by this source.

Wiki pages this source may touch

• lead
• chromium
• mercury
• arsenic
• cadmium
• toys-painted
• toys-squeeze
• toys-stuffed-bean-bag
• pacifiers-and-sucking-teething-aids

Verification notes

• 2026-06-01 fresh ingest (Claude Opus 4.7, autonomous v2.0 manual-fetch skill, NEW path). Three identity checks against wiki/sources/ returned no match: DOI 10.22034/ehpr.2026.554453.1004 not present; raw_handle: MFK_11-2024-article-238039 not present; cite-key collision check for seifi returned no existing pages. Source page written from scratch.
• Source identification. Mehran Seifi (sole author), “Investigation of the concentration of chemical compounds in toys,” Environmental Health and Pollution Research, 2026, 1(1), 43–48, DOI 10.22034/ehpr.2026.554453.1004, published 02 January 2026, open-access CC BY 4.0 per the licence stamp on the final page. The journal is in its first issue (Volume 1, Issue 1, January 2026) and the title page prints a placeholder ISSN (××××-××××) suggesting the ISSN has not yet been assigned at time of publication. The publisher imprint is “Greens Trust.” Journal website cited as www.ehpr.ir.
• Source-tier rationale. evidence_tier: B. Per CLAUDE.md Part 13, A-tier is reserved for peer-reviewed primary studies in established journals, government/agency reports, and authoritative scientific opinions. This paper has multiple quality flags that argue against A-tier classification: (i) brand-new journal (Vol 1 Issue 1) with placeholder ISSN, no established editorial track record, no indexing in PubMed/Scopus/Web of Science verifiable at ingest time; (ii) single-author study from a Student Research Committee, with no co-author oversight on data analysis or arithmetic; (iii) methods documentation is brief — no LODs, LOQs, recovery percentages, certified reference materials, or instrument identifiers reported; (iv) Cr speciation method (HPLC-ICP-MS) mentioned but no speciated values reported; (v) PCA “employed to identify clustering patterns” with no PCA results reported; (vi) AI translation use declared on the manuscript itself; (vii) multiple paper-internal data-integrity inconsistencies between Table 1 (mean ± SD) and Table 3 (% non-compliance) — when mean ± 1 SD straddles the EU limit, the reported non-compliance rate is mathematically much lower than the implied normal-distribution rate (documented in detail below). B-tier reflects that the paper’s direction-of-finding (painted-wood > plastic > rubber > plush for Pb; rubber > plastic > painted-wood > plush for DEHP) is supported by the data presented, while the absolute point estimates carry more uncertainty than the SDs alone suggest. The paper is informative for HMTc Cat 21 routing (where it contributes material-stratification evidence consistent with the corpus pattern) but should not be the sole anchor for any specific threshold value.
• Source-type rationale. source_type: peer-reviewed. The journal is presented as peer-reviewed (paper type “Research Paper,” dates for Received/Revised/Accepted/Published consistent with a peer-review cycle, AI Use Declaration present). The B-tier rather than A-tier classification reflects the journal’s nascence and the data-integrity caveats, not a different source-type classification.
• License rationale. license: CC BY 4.0: the final page carries the Creative Commons logo and the wording ”© Authors, Published by Environmental Health and Pollution Research. This is an open-access article distributed under the CC BY (license: http//creativecommons.org.licenses.by.4.0).” Treated as standard CC BY 4.0.
• Frontmatter products: field. Four slugs selected from the 2026-05-18 taxonomy snapshot that best match the paper’s stratification and inclusion criteria:
  • toys-painted — direct evidence for the painted-wood material category (n=30); the paper’s strongest finding (highest Pb, Cr, Hg means; 23.3% Pb non-compliance; 30.0% Cr non-compliance) maps directly to HMTc Cat 21 Row 1 (“General toys with paint/surface coatings”);
  • toys-squeeze — direct evidence for the rubber material category (n=30); the paper’s inclusion criteria explicitly named “squeeze toys” as a prioritized item type, and the rubber subset is where the paper’s organic-compound findings (highest DEHP, BPA, PBDEs) cluster; maps to HMTc Cat 21 Row 6;
  • toys-stuffed-bean-bag — direct evidence for the plush material category (n=30); plush is the closest semantic match for HMTc Cat 21 Row 3 (“Stuffed toys and bean bag-type toys”); the paper’s plush subset is consistently the lowest-burden category across all analytes, an A-tier-quality contribution were the rest of the paper’s quality controls more thorough;
  • pacifiers-and-sucking-teething-aids — broad-context evidence; the paper’s inclusion criteria named “teethers” as a prioritized item type, and the EN-71-3 simulated-gastric acid extraction protocol is the standard test for mouthing-pathway migration; however, the paper does not separately report teether-form values within the four material strata, so this slug is broad-context rather than direct-evidence. Routing layer will classify accordingly.
  • The plastic-toy subset (n=30) has no general “plastic toys” product slug in the current taxonomy. The plastic-subset data is therefore not separately routed; the Cat 21 ratifiability of the plastic-stratum values would require a generic-plastic-toys slug that does not yet exist. Surfaced for Karen’s awareness; not actioned in this ingest per Phase 1 broad-scope discipline. Per CLAUDE.md Part 5b, the routing layer fans broad scopes out as needed; over-specifying the products field is the failure mode to avoid here.
• Frontmatter ingredients: [] is correct — children’s-toy consumer product survey; no food ingredients in scope.
• Frontmatter matrices: [] is correct — no toy-material slug exists in the current matrices vocabulary (which is food/water/soil/cosmetic-personal-care biased). Consistent with sibling toy sources igweze2020-china-toys-nigeria-pb-cd-as, fowles2021-lead-toys, bfr2009-lead-cadmium-toys, cpsc1997-pvc-children-products, and johnson2012-cpsc-astm-f963-status which also leave matrices empty. The products field carries the routing signal.
• Frontmatter metals: [Pb, Cd, tHg, tAs, Cr] per CLAUDE.md Part 14 speciation discipline:
  • Pb — total elemental Pb measured by ICP-MS; no organolead speciation question;
  • Cd — total elemental Cd by ICP-MS;
  • tHg — total Hg by ICP-MS; the paper does not distinguish methylmercury from inorganic mercury, so tHg is the correct slug (not MeHg and not Hg ambiguous);
  • tAs — total As by ICP-MS; the paper does not distinguish inorganic from organic arsenic, so tAs is the correct slug (not iAs);
  • Cr — total Cr by ICP-MS; despite the methods section’s mention of HPLC-ICP-MS for Cr(III)/Cr(VI) speciation, no speciated Cr values are reported in Tables 1–4 or in the body text. Frontmatter therefore records total Cr only (Cr, not Cr-VI). The methods-vs-results inconsistency on Cr speciation is documented in the Methods (brief) section without being silently corrected.
• Frontmatter jurisdictions: [IR, EU] — Iran is the sampling jurisdiction (sampling in Tehran, Shiraz, Tabriz, January–June 2024; Conclusion calls for ISIRI 6243 and ISIRI 10166 enforcement). EU is the comparator regulatory framework (EN-71-3 migration limits for heavy metals in Tables 1 and 3; Toy Safety Directive TSD limits for phthalates in Table 2; REACH and the EU Toy Safety Directive referenced in Discussion). The paper does not separately identify country of toy manufacture for the 120-sample set, so no CN/manufacturer-origin jurisdiction is included.
• Brand firewall (Part 12, strict reading locked 2026-05-17). Two named online marketplaces appear in the Sampling section: “Digikala and Bamilo.” Per Part 12 Exception 1, these are named as retail-channel platforms in a methods context (specifying the sampling frame, not attributing a contamination value to a particular vendor’s products), and the wiki page reproduces the source’s own framing. No per-platform or per-brand contamination values are reported by the source or by the wiki page. The three Iranian cities (Tehran, Shiraz, Tabriz) are geographic sampling-frame locations, not brand attributions. The paper names no individual toy manufacturers, no individual brand product lines, and no product trade names within the 120-sample dataset. Cleanest brand-firewall posture in the toys subcorpus alongside the fowles2021/bfr2009/cpsc1997 regulatory sources.
• Wiki/HMTc firewall (Part 2). No HMTc threshold proposals; no consumer-audience risk advisories; no cross-source synthesis statements added on the wiki side. The Implications section reports the paper’s direction-of-finding as input to HMTc Cat 21 routing without proposing specific threshold values; the standards-setting decision belongs to the Standards Workbench (per Part 19), not the wiki. The paper’s own conclusion (ISIRI 6243/10166 enforcement recommendation) is reported descriptively as the author’s own conclusion.
• Speciation-method-vs-results inconsistency (paper-internal; not silently corrected). Methods Section 2.3 (p. 44) explicitly says “chromium speciation (Cr(III) vs. Cr(VI)) was assessed using HPLC-ICP-MS.” No Cr(III) vs Cr(VI) values are reported in any results table or in the body text. Tables 1, 3, and 4 report only total Cr. The wiki page records total Cr (Cr in frontmatter metals: array) and notes the speciation-method-vs-results inconsistency in the Methods (brief) section as a paper-side gap, not a wiki defect.
• Data-integrity inconsistency between Table 1 (Mean ± SD) and Table 3 (% non-compliance) — documented, not silently corrected. Under any reasonable distribution model (normal, log-normal, etc.) the reported group means ± SDs imply substantially higher exceedance rates than Table 3 reports. Most striking cases:
  • Painted Wood Cr. Mean 102.3 ± 25.7 mg/kg, EU limit cited as 60. Under a normal-distribution assumption, P(Cr > 60) = P(Z > (60-102.3)/25.7) = P(Z > -1.65) ≈ 95%. Table 3 reports 9/30 = 30.0% non-compliance. The reported group mean alone exceeds the limit by 70%; the SD does not credibly produce only 30% exceedance.
  • Painted Wood Hg. Mean 4.5 ± 1.4 mg/kg, EU limit 2. P(Hg > 2) ≈ P(Z > -1.79) ≈ 96%. Table 3 reports 2/30 = 6.7% non-compliance.
  • Plastic Cr. Mean 85.6 ± 20.4 mg/kg, EU limit 60. P(Cr > 60) ≈ P(Z > -1.25) ≈ 89%. Table 3 reports 0/30 = 0% non-compliance for plastic Cr (the “All Other Combos” row implicitly covers this).
  • Plastic Hg. Mean 3.2 ± 1.1, EU limit 2. P(Hg > 2) ≈ P(Z > -1.09) ≈ 86%. Table 3 implies 0%.
  • Painted Wood DEHP. Mean 890 ± 275 mg/kg, EU limit 1000. P(DEHP > 1000) ≈ P(Z > 0.40) ≈ 34%. Table 3 implies 0%.
  • Painted Wood Pb. Mean 212.4 ± 45.1, EU limit 160. P(Pb > 160) ≈ P(Z > -1.16) ≈ 88%. Table 3 reports 7/30 = 23.3% non-compliance. The cleanest of the inconsistencies — Pb non-compliance rate is at least in the right order of magnitude relative to the mean/SD, but still lower than a normal-distribution model predicts.
  • Possible reconciliations the paper itself does not offer. (a) Table 1 might report total-elemental microwave-digest values while Table 3 might report EN-71-3 acid-migration values, with migration consistently lower than total content — this is plausible given the paper’s two-protocol methods, but the paper’s body text frames the Table 1 means as exceeding “the EU migration limits” (the EN-71-3 framing), which would mean Table 1 itself is migration values, foreclosing the reconciliation. (b) The distribution may be far from normal (e.g., a small number of high-tail samples drive the mean, with most samples below the limit) — but this is not what mean ± SD reporting conventionally implies, and the paper does not report medians or percentiles to test the alternative. (c) Arithmetic error or column-label confusion in the manuscript. The wiki page transcribes both Tables 1 and 3 verbatim and flags the inconsistency here without silently harmonizing, consistent with the igweze2020 precedent of reporting source-internal inconsistencies in Verification notes rather than smoothing them on the wiki side.
• Methods inconsistency: Sb/Zn/Ni/Cu/Mn analyte panel. Methods Section 2.3 lists ICP-MS quantification of “Pb, Cd, Hg, Cr, As, Sb, Zn, Ni, Cu, Mn” but Tables 1, 3, and 4 report data for only Pb, Cd, Hg, Cr, As. No Sb, Zn, Ni, Cu, or Mn values are presented anywhere in the paper. The wiki page records metals: [Pb, Cd, tHg, tAs, Cr] (the reported analytes only); the unreported analytes are noted here but not added to the metals array.
• Statistical-method-vs-results inconsistency: PCA. Methods Section 2.4 (p. 44) states “Multivariate analysis (Principal Component Analysis, PCA) was employed to identify clustering patterns and potential sources of contamination.” No PCA results, score plots, loading matrices, or variance-explained values are reported anywhere in the paper. Noted in Methods (brief) without silent correction.
• HQ values not tabulated. The Conclusion (p. 47) states “Quantitative health risk assessment confirmed these concerns, calculating Hazard Quotients (HQ) exceeding 1 for lead in painted wood toys and DEHP in rubber toys, indicating a potential for adverse health effects under reasonable exposure scenarios.” No HQ values, no exposure-parameter table, no reference-dose table, and no exposure-scenario assumptions are reported in the manuscript. The HQ > 1 conclusion is therefore an author-asserted qualitative claim that the wiki page reports descriptively without numerical verification.
• Reference-list discipline. Three references in the bibliography lack a year of publication: Häkkinen (2010) is cited inline but the bibliography entry shows “Control of chemicals in articles. (2010). avialable at: http://hdl.handle.net/10138/37989” with “avialable” as a typo; Burgos et al. (2024) is cited inline in Section 1 but the bibliography entry’s date placement is non-standard; Lulei (2008) appears as a bibliography entry without a fully resolved venue. These are reference-list-quality observations consistent with the B-tier classification and the AI-translation disclosure.
• Folder context vs paper scope. The PDF lives under _extracted_infantcontact_02_Teethers_Pacifiers/02_Teethers_Pacifiers/ in the Kimi corruption-issue raw tree. The paper itself is a general children’s-toys survey, not a teether/pacifier-specific study — its 120-sample dataset is stratified by material (plastic, painted wood, rubber, plush), not by toy form, and teethers are named only in the inclusion-criteria language as one of several mouthing-pathway exemplars alongside dolls, squeeze toys, and building blocks. The pacifiers-and-sucking-teething-aids product slug is included in routing because mouthing/sucking is one of the central exposure pathways the EN-71-3 acid-extraction protocol is designed to model, not because the folder is named “Teethers/Pacifiers.” Same folder-vs-scope handling as fowles2021-lead-toys.
• Raw integrity. raw_sha256 = 317e288023ca46e3093e3c95bb3ef3268886bb4be1064edba3e7a2e27ac9d074 verified by shasum -a 256 against the file at raw_path on 2026-06-01.
• Near-duplicates. None identified in the corpus. Sibling toys-survey sources in the corpus (igweze2020-china-toys-nigeria-pb-cd-as, fowles2021-lead-toys, bfr2009-lead-cadmium-toys, cpsc1997-pvc-children-products, johnson2012-cpsc-astm-f963-status) address related questions but from different national-agency, regulatory, or country-of-sale frames and do not duplicate this paper’s Iranian-retail-market dataset.

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.