Measurement of traces of heavy metals in cosmetic raw materials and finished products according to ISO/DIS 21392:2021 using triple quadrupole ICP-MS

Moriceau and Naëls 2025 — ICP-MS heavy-metals measurement in cosmetic raw materials and finished products per ISO/DIS 21392:2021

This Thermo Fisher Scientific application note (AN003602, January 2025) demonstrates a triple-quadrupole ICP-MS workflow on a Thermo Scientific iCAP MTX instrument for quantifying 19 trace elements in cosmetic products following ISO/DIS 21392:2021. Seventeen finished cosmetics from the French and European market and eight cosmetic raw materials were digested by microwave (Milestone ultraWAVE SRC) and analyzed in both KED and TQ-O₂ modes. The note’s primary contribution is methodological — showing that TQ-O₂ mode eliminates tungsten-on-mercury (¹⁸⁴W¹⁶O¹H⁺ on ²⁰¹Hg) and zirconium-on-cadmium (⁹⁴Zr¹⁶O¹H⁺ on ¹¹¹Cd) polyatomic interferences that KED mode cannot fully suppress — but it also reports concentrations for seven aggregated finished-product categories and ten raw materials, useful as occurrence context.

Key numbers

• Study design: 27 samples total — 17 finished products + 10 raw materials (the paper’s narrative says “eight” raw materials, but Tables 2 and 6 enumerate ten; the ten-row tabulation is treated as authoritative). Sample preparation per ISO/DIS 21392:2021: 200 mg sample digested with 1 mL H₂O + 5 mL HNO₃ + 1 mL HCl, digestate transferred to 50 mL with ultrapure water then 4× further dilution to 200 mL final volume in 2.5% HNO₃ / 0.5% HCl; total dilution factor ×1000 (p. 3 Table 3; p. 2 narrative).
• Method quantification limits (MQL, mg·kg⁻¹), Table 4 (p. 4): Pb 0.004, Cd 0.010, As 0.031, Hg 0.017, Cr 0.041, Ni 0.115, Sb 0.010, Al 1.118, Sn 0.047, Co 0.002, Tl 0.002, Se 0.718, Be ≤0.001, Mn 0.005, Fe 0.149, Zn 0.232, Ti 1.169, Pt 0.023, W 0.101.
• Acquisition mode: Pb, Co, Ni, Sn, Sb, Be, Al, Fe, Tl, Pt, W used SQ-KED; Cr, Mn, As, Se, Cd, Zn, Ti, Hg used TQ-O₂ with on-mass measurement after O₂ reaction (Table 4 p. 4).
• EU regulatory ceilings cited (Regulation EC 1223/2009, p. 2): Cr 10, Pb 10, As 5, Cd 5, Hg 1 mg·kg⁻¹.
• Finished-product concentrations (Table 5, p. 5), reported as the range across samples in each aggregated category, mg·kg⁻¹:
  • Nail varnish (3 varnish samples): Pb 0.17-0.64; Cd <MQL-0.01; As <MQL; Hg <MQL; Cr 0.17-0.29; Ni 0.12-0.20; Sb <MQL-0.02; Sn 0.02-0.25; Al 667-1,079; Ti 10.4-364; Fe 256-352; Mn 1.09-49.6; Zn 11.5-46.4; Co 0.04-0.06.
  • Balm and lipstick (lip balm + 1 other solid lip product, n unstated for the category): Pb <MQL-0.86; Cd <MQL-0.04; As <MQL-0.09; Hg <MQL; Cr 0.01-0.37; Ni 0.01-0.33; Sb <MQL-0.08; Sn 0.01-18.1; Al <MQL-311; Ti 0.01-332; Fe 0.01-138; Mn 0.01-5.01; Zn <MQL-54.8; Co 0.01-0.09.
  • Perfumes (perfume mist, hair perfume, eau de cologne, solid perfume, perfume, eau de toilette, eau de parfum): Pb 0.01-5.59; Cd <MQL; As <MQL; Hg <MQL; Cr <MQL-21.6; Ni <MQL-20.5; Sb <MQL-0.01; Sn <MQL-0.23; Al 0.08-33.0; Ti <MQL-6.44; Fe <MQL-89.4; Mn 0.01-2.14; Zn 0.03-76.8; Co <MQL-0.39.
  • Toothpaste (n=1): Pb 0.85; Cd <MQL; As <MQL; Hg <MQL; Cr 0.65; Ni <MQL; Sb <MQL; Sn 0.77; Al 31.6; Ti 4.77; Fe 15.0; Mn 0.26; Zn 27.1; Co 0.04.
  • Micellar water (n=1): Pb 0.17; Cd <MQL; As <MQL; Hg <MQL; Cr <MQL; Ni <MQL; Sb <MQL; Sn <MQL; Al <MQL; Ti 0.55; Fe <MQL; Mn 0.07; Zn 11.0; Co <MQL.
  • Cream and gel (scrub cream, hand cream, pink clay mask, exfoliating cleansing gel): Pb 0.27-2.30; Cd <MQL; As <MQL; Hg <MQL; Cr <MQL-8.00; Ni <MQL; Sb <MQL; Sn <MQL-0.67; Al <MQL-11,344; Ti 1.62-308; Fe <MQL-955; Mn 0.02-3.80; Zn 13.2-1,677; Co <MQL-0.37.
  • Nourishing body milk (n=1): Pb 0.19; Cd <MQL; As <MQL; Hg <MQL; Cr <MQL; Ni <MQL; Sb <MQL; Sn <MQL; Al <MQL; Ti 1.42; Fe <MQL; Mn 0.03; Zn 16.3; Co <MQL.
• Notable finished-product highs: perfume Pb max 5.59 mg·kg⁻¹ (within EU 10 mg·kg⁻¹ ceiling); cream/gel Al max 11,344 mg·kg⁻¹ and Zn max 1,677 mg·kg⁻¹ (no EU ceiling for either); perfume Cr max 21.6 mg·kg⁻¹ (above the EU 10 mg·kg⁻¹ ceiling); perfume Ni max 20.5 mg·kg⁻¹; balm/lipstick Sn max 18.1 mg·kg⁻¹.
• Raw-material concentrations (Table 6, p. 5), mg·kg⁻¹ unless noted as % w/w; digestion incomplete for 9 of 10 raw materials (Ophiopogon japonicus root the only complete digestion):
  • Iron oxide mica: Pb 4.63; Cd <MQL; As 0.21; Hg <MQL; Cr 8.18; Ni 11.3; Sb 0.16; Sn 56.0; Al 5.8% (w/w); Ti 258; Fe 18.0% (w/w); Mn 165; Zn 97.9; Co 3.37; W 14.0; Be 6.58; Tl 1.58.
  • Titanium dioxide (anastase form): Pb 0.66; Cd <MQL; As <MQL; Hg <MQL; Cr <MQL; Ni 0.11; Al 14.1; Ti 212; Fe 15.6; Mn 0.07; Zn 0.93.
  • Iron oxide mica / aluminum oxide blend: Pb 5.51; Cd <MQL; As 0.14; Hg <MQL; Cr 5.84; Ni 2.69; Sn 21.8; Al 5.2% (w/w); Ti 196; Fe 16.8% (w/w); Mn 28.4; Zn 15.8; Co 3.09; W 10.9; Be 3.74; Tl 0.55.
  • Hollow silica sphere: Pb 2.78; Cd <MQL; As <MQL; Hg <MQL; Cr 0.28; Ni 0.26; Al 6.17; Ti 2.71; Fe 17.8; Mn 0.17; Zn 42.8; Sb 0.03.
  • Ophiopogon japonicus root extract (complete digestion): Pb 0.05; Cd 0.04; As 0.62; Hg <MQL; Cr 0.14; Ni 0.32; Al 23.8; Ti 0.63; Fe 11.5; Mn 17.2; Zn 5.12; Co 0.08; Sb 0.02.
  • White powder A: Pb 4.20; Cd <MQL; As 0.09; Hg <MQL; Cr 3.50; Ni 1.95; Al 8.68; Ti 194; Fe 26.2; Mn 0.60; Zn 24.9; Co 0.26; Sn 0.05.
  • White powder B: Pb 4.08; Cd <MQL; As <MQL; Hg <MQL; Cr 5.01; Ni 2.63; Al 7.00; Ti 180; Fe 36.2; Mn 0.78; Zn 22.8; Co 0.12.
  • Titanium dioxide (rutile form): Pb 2.26; Cd <MQL; As <MQL; Hg <MQL; Cr 0.50; Ni 0.15; Al 0.1% (w/w); Ti 246; Fe 7.90; Mn 1.23; Zn 32.8; Pt 0.11.
  • Titanium dioxide / iron oxide / silica blend: Pb 4.35; Cd <MQL; As <MQL; Hg <MQL; Cr 15.7; Ni 17.8; Sn 160.8; Al 3.2% (w/w); Ti 249; Fe 10.7% (w/w); Mn 376; Zn 39.8; Co 5.12; Be 0.82; Tl 0.27.
  • Mica pellets coated with iron and titanium: Pb 5.96; Cd <MQL; As <MQL; Hg <MQL; Cr 3.04; Ni 4.62; Sn 5.02; Al 1.8% (w/w); Ti 422; Fe 19.4% (w/w); Mn 851; Zn 36.0; Co 3.94; Be 0.64; Tl 0.19.
• Tungsten-on-mercury interference demonstration (Table 8, p. 6): in iron oxide mica (W = 14.0 mg·kg⁻¹) KED-mode ²⁰¹Hg = 33.48 mg·kg⁻¹ but TQ-O₂ mode <0.017 mg·kg⁻¹ (the MQL); in iron/aluminum oxide (W = 10.9 mg·kg⁻¹) KED-mode ²⁰¹Hg = 31.49 mg·kg⁻¹ but TQ-O₂ <0.017 mg·kg⁻¹. The KED-mode values exceed the EU 1 mg·kg⁻¹ Hg ceiling by ~30×; TQ-O₂ reveals no Hg is actually present.
• Zirconium-on-cadmium interference demonstration (Figure 6, p. 8): zero added Zr → ¹¹¹Cd KED 0; +1 µg·g⁻¹ Zr → ¹¹¹Cd KED 0.54 µg·g⁻¹ (false positive), TQ-O₂ ~0; +10 µg·g⁻¹ Zr → ¹¹¹Cd KED 5.20 µg·g⁻¹ (false positive), TQ-O₂ 0.06 µg·g⁻¹.
• Spike recovery (Figure 7, p. 9): per-element recovery across 16 finished-product matrices ranged 60-105% for the seven measured isotopes (⁵²Cr, ⁵⁹Co, ⁶⁰Ni, ⁷⁵As, ¹¹¹Cd, ¹²¹Sb, ²⁰⁸Pb); the source narrative claims all within the 70-150% ISO/DIS 21392:2021 acceptance window, but the data table shows three sub-70% values on a single sample, Hair perfume (⁵²Cr 67%, ⁵⁹Co 69%, ⁶⁰Ni 60%); all other 15 samples are within the 70% floor for all seven isotopes.
• Internal standard stability over >120 samples / >8 h: 80-110% recovery for all five internal standards (⁷³Ge, ¹⁰³Rh, ¹¹⁵In, ¹⁷⁵Lu, ¹⁹³Ir) in both KED and TQ-O₂ modes (Figure 1, p. 4), well within the 70-130% ISO/DIS 21392:2021 robustness window.

Methods (brief)

Instrument: Thermo Scientific iCAP MTX triple-quadrupole ICP-MS with iSC-65 autosampler, iCAP MX Series Nebulizer, Ni-tipped sample and skimmer cones, PLUS torch, cyclonic quartz spray chamber. Operated with Argon Gas Dilution (AGD level-5, ~5× automatic dilution) to manage variable total dissolved solids across cosmetic matrices. Two acquisition modes: SQ-KED (collision, helium 4.91 mL·min⁻¹) and TQ-O₂ (reaction, 0.31 mL·min⁻¹) on the same run. RF power 1,550 W; 3 replicates per sample (Table 1, p. 2).

Sample preparation per ISO/DIS 21392:2021: microwave-assisted acid digestion (Milestone ultraWAVE Single Reaction Chamber system); reagents dispensed by Milestone EasyFILL (1 mL H₂O + 5 mL HNO₃ + 1 mL HCl per 200 mg sample); 200 °C plateau for 30 min. Note that complete digestion was not achieved for 9 of 10 raw materials (silica/mica/oxide deposits persisted); for those samples the reported concentrations represent the acid-leachable fraction rather than total elemental content. Digestate diluted to 50 mL with ultrapure water, then 4× further dilution and 0.45 µm membrane filtration before injection. Internal standards (¹⁰³Rh for Cr/Co/Ni/As/Cd/Sb; ¹⁷⁵Lu for Pb per the ISO standard, plus ⁷³Ge/¹¹⁵In/¹⁹³Ir added by the authors for robustness check) introduced online at 1,000 µg·L⁻¹ before nebulization.

Speciation discipline: 75As measured as ⁷⁵As¹⁶O⁺ product ion in TQ-O₂ — this is total arsenic, no inorganic/organic speciation performed (recorded as tAs). Mercury measured on ²⁰¹Hg in TQ-O₂ mode — this is total mercury, no methylmercury speciation (recorded as tHg). Chromium measured on ⁵²Cr¹⁶O⁺ product ion — this is total chromium, no Cr-VI / Cr-III separation (recorded as Cr). Tin measured on ¹¹⁸Sn in SQ-KED — total tin, no organotin speciation (recorded as Sn).

Calibration: standards 0.5-10 µg·L⁻¹ in matched acid matrix (2.5% HNO₃ / 0.5% HCl) for trace elements; up to 2,500 µg·L⁻¹ for the high-concentration elements (Al, Ti, Fe). R² ≥0.9964 for all 19 analytes (W is the lowest at R² 0.9964; Be is 0.9986; Al, Ti, Zn, Sn at R² >0.9999; the rest fall between; Table 4 p. 4). Accuracy validated by spike-recovery on the 16 finished-product matrices (Figure 7, p. 9). LOD and MQL reported in Table 4.

Limitations: (1) Sample selection is convenience, not representative — 25 samples drawn from the authors’ lab to demonstrate workflow; not designed as an occurrence survey. (2) Incomplete digestion for raw materials means raw-material values are minimum estimates of total content. (3) Speciation is absent for As, Hg, Cr, Sn. (4) The note is a vendor publication promoting Thermo Scientific instruments; reported figures of merit reflect the demonstration laboratory’s optimized conditions.

Implications

• Certification (HMTc): Method-validation reference for the ISO/DIS 21392:2021 analytical approach now widely adopted in cosmetic regulatory labs. The demonstration that KED-mode Hg in tungsten-containing raw materials produces >30× false-positive concentrations is operationally important — labs relying on KED-only quadrupole ICP-MS for cosmetics analysis may be over-reporting Hg in mica- and tungsten-bearing finished products. Auditors of childrens-makeup, childrens-nail-polish, makeup-foundation-powders-blush, and other mica-pigment-using product categories should ask whether the lab used reaction-cell mode for Hg. The reported finished-product concentrations (Table 5) are occurrence context only, not population-level evidence (n=1 per category for toothpaste, micellar water, body milk; n=17 total across all categories).
• Courses: Useful methods-module reading for a cosmetic-QA / analytical-chemistry course unit on triple-quadrupole ICP-MS interference removal, particularly for instructors covering W/Mo/Zr-based polyatomic interferences on Hg and Cd.
• App: Not relevant to ingredient contamination_profile (cosmetics are not in the food-ingredient taxonomy).

Wiki pages this source may touch

• lead
• cadmium
• mercury-total
• arsenic-total
• chromium
• nickel
• antimony
• aluminum
• tin
• toothpaste
• lipstick

Verification notes

• Source type recorded as industry-application-note and evidence tier C: this is a vendor application note from Thermo Fisher Scientific (AN003602-EN 0225S, ©2025), not a peer-reviewed paper. The note’s purpose is to demonstrate the iCAP MTX instrument’s compliance with ISO/DIS 21392:2021. Numerical values are usable as occurrence context with caveats noted in Methods (sample selection is convenience; raw-material digestion incomplete for 9 of 10 raw materials).
• DOI is null: vendor application notes typically do not carry DOIs. The publisher reference is the document ID “AN003602-EN 0225S” and the publication channel thermofisher.com/icp-ms.
• Speciation discipline (CLAUDE.md Part 14): paper measures total As (no iAs/tAs split → tAs), total Hg (no MeHg split → tHg), total Cr (no Cr-VI speciation → Cr), total Sn (no organotin speciation → Sn). Frontmatter metals: reflects this.
• Brand firewall (CLAUDE.md Part 12): paper does NOT name any cosmetic-product brands for the sampled products (samples described by product form only: “Scrub cream”, “Eau de cologne”, “Perfume mist”, etc.). The instrument and reagent vendor names retained per Verification Checklist Exception 2 (scientific-method vendor names): Thermo Scientific iCAP MTX, iSC-65, Qtegra ISDS Software, Hawk Consumables Assistant; Milestone EasyFILL, ultraWAVE SRC. These are method components, not branded contamination subjects.
• Wiki/HMTc firewall (CLAUDE.md Part 2): Implications section reports method-validation relevance to certification auditing without proposing thresholds. EU Regulation 1223/2009 ceilings (10/10/5/5/1 mg·kg⁻¹) are quoted from the source, not synthesized against HMTc.
• Product routing: paper covers a mix of finished cosmetic categories (nail varnish, lipstick/balm, perfumes, toothpaste, micellar water, creams/gels, body milk). The current taxonomy has wiki pages for lipstick and toothpaste; the other product categories sampled (perfumes, nail varnish for adults, body milks, cleansers, micellar water, hand creams) do not yet have wiki pages. Per skill stop-condition rules, do NOT create new product pages here (Karen’s Step 0 Lock workflow handles those); the source routes to the two existing slugs. Adult-cosmetic categories sampled may warrant taxonomy expansion in a future Step 0 Lock pass; flagged here for triage. The seven raw-material categories (iron oxide mica, titanium dioxide variants, mica pellets, hollow silica, white powders, ophiopogon japonicus root) are cosmetic raw materials and are not in the food-ingredient taxonomy.
• Children’s-product relevance: this paper sits in the babycare_03 folder but it is NOT a children’s-products study — the samples are adult French/European cosmetics (perfumes, eaux de toilette, lipstick, nail varnish, hand cream, toothpaste, etc.). Folder placement is a Manual-Fetch curation artifact, likely because raw-material categories (iron oxide mica, mica pellets, titanium dioxide) overlap with children’s-cosmetics pigment supply chains. No claims about children’s products are made on this source page.
• Tungsten-on-Hg false-positive finding (Table 8, p. 6) is the methodologically significant result: any KED-only ICP-MS lab analyzing mica-pigment-bearing cosmetics (which is most lipsticks, nail polishes, eye shadows, mineral makeup, and many children’s makeup products) is at risk of reporting falsely-high Hg values. This is captured in the Implications section as a cross-reference for HMTc auditors of mica-bearing product categories without proposing a threshold.
• Sample size attribution: paper aggregates results by product category (Table 5 columns: “Nail varnish”, “Balm and lipstick”, “Perfumes”, etc.) without stating per-category n. Total finished-product n=17 is stated; per-category n inferred from Table 2 sample list (Nail varnish n=3 — Varnish, Rose varnish, Red varnish; Balm and lipstick n=2 — Lip balm, Solid perfume [the latter may not be lip — ambiguous]; Perfumes n includes Hair perfume, Eau de Cologne, Perfume mist, Perfume, Eau de toilette, Solid perfume; Toothpaste n=1; Micellar water n=1; Cream and gel n=4 — Scrub cream, Hand cream, Pink clay mask, Exfoliating cleansing gel; Body milk n=1). Per-category sample counts are inferred and may not be definitive; treated as approximate.
• Source-internal contradiction on raw-material count (flagged by audit subagent 2026-05-18, verified against source): paper narrative on p. 2 says “eight samples were raw materials,” but Table 2 lists 10 raw-material rows and Table 6 has 10 data columns. The tabular enumeration is treated as authoritative; sample_n recorded as 27 (17 finished + 10 raw materials), not the narrative’s 25.
• Audit subagent (2026-05-18) flagged Figure 7 spike-recovery attribution as ❌ definite error — verified against PDF p. 9 data table: my original entry transposed columns 6 (Hair perfume) and 7 (Eau de cologne). Corrected: the three sub-70% values are all on Hair perfume (⁵²Cr 67%, ⁵⁹Co 69%, ⁶⁰Ni 60%); Eau de cologne values are ⁵²Cr 80%, ⁵⁹Co 74%, ⁶⁰Ni 87%, ⁷⁵As 78%, ¹¹¹Cd 74%, ¹²¹Sb 77%, ²⁰⁸Pb 88% — all above the 70% floor.
• Audit subagent (2026-05-18) flagged R² claim “R² >0.9986” as ⚠️ inaccurate — verified against Table 4: W has R² 0.9964 which is below 0.9986; Be is exactly 0.9986. Methods section corrected to “R² ≥0.9964 for all 19 analytes.”
• Audit subagent (2026-05-18) flagged “60-105% for the four toxic targets (Cr, Co, Ni, As, Cd, Sb, Pb)” as ⚠️ internally inconsistent (says four, lists seven) — corrected to “60-105% for the seven measured isotopes.”
• Audit subagent (2026-05-18) flagged makeup-foundation-powders-blush as an “invented slug” not in the taxonomy snapshot — VERIFIED FALSE POSITIVE: the slug exists as a real wiki product page at wiki/products/makeup-foundation-powders-blush.md; the docs/gpt-collaboration/taxonomy-snapshot.md file is stale (generated 2026-05-17 from commit e442cbe and does not include the page). No change to wiki page; this is a taxonomy-snapshot-refresh hygiene issue, not a slug error.
• Audit subagent (2026-05-18) flagged 50 mL (Methods narrative) vs 200 mL (Table 3) as a ⚠️ source-internal contradiction — VERIFIED FALSE POSITIVE: the two figures are arithmetically consistent (50 mL initial transfer × 4 further dilution = 200 mL final volume per Table 3). Key numbers and Methods sections reworded to make the relationship explicit (“digestate transferred to 50 mL with ultrapure water then 4× further dilution to 200 mL final volume”).
• Audit subagent (2026-05-18) flagged W and Pt missing from frontmatter metals despite being measured/reported — applied: added W and Pt to the metals array.

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.

Commit	Date	Description
ce3e07c	2026-05-28	activation | Vercel DATACITE env slots set, curators.md filled with founder entry + six scoped reviewer invitations, peer-review onboarding playbook drafted
51400b9	2026-05-28	audit-queue: gasparik2017-wild-boar-slovakia-metals audited-revised