Lee et al. 2019 — Heavy metal migration during food processing (oilseeds, noodles, teas)

Lee and colleagues at the Korean Ministry of Food and Drug Safety’s National Institute of Food and Drug Safety Evaluation measured Pb, Cd, As, and Al in Korean market oilseeds (sesame, perilla, flaxseed), noodles (wheat flour and sweet potato glass), and teas (Sri Lankan black, Korean green, Solomon’s seal) before and after processing, then quantified how much of each metal migrated out of the food during pressing/solvent/supercritical-CO₂ oil extraction, boiling, and infusion. The central finding is that water-soluble heavy metals migrate substantially into cooking and infusion water: roughly 40–70% of Pb migrates out of noodles across the 3–10 min boiling cells (flour and glass noodle combined), and tea-leaf-to-infusion migration spans roughly 5–88% across metals × tea types × steep times (lowest for Solomon’s seal As at 2 min, highest for black-tea As at 30 min), while oil extraction transfers comparatively little metal from seed to oil (typically <10% by pressing or solvent extraction, with supercritical CO₂ reaching up to ~30% transfer for some seed/metal pairs). The practical implication is that the consumer-facing exposure depends sharply on whether the boiling water or infusion liquid is consumed.

Key numbers

Analytical method: ICP-MS (Thermo Fisher Dionex ICS-5000+SP) for Pb, Cd, As; ICP-OES (PerkinElmer Optima 8300) for Al; microwave-assisted acid digestion (HNO₃ + H₂O₂). LODs across CRM matrices: Pb 0.011–0.859 µg/kg, Cd 0.012–0.123 µg/kg, As 0.010 µg/kg, Al 0.016–0.047 µg/kg. LOQs: 0.035–2.863 µg/kg. CRM recoveries 71.5–107.5%; intra-day RSD 0.84–18.55% (Table 1).

Heavy metal content of seeds and oils (Table 2, total µg per 5,000 g seed batch or per oil-yield batch; means ± SD):

Pressing extraction (mechanical, Poong-jin P-16(S), 70 °C, 200×10⁵ Pa, 15 min):

• Flaxseed seed: Pb 373.6±142.9, Cd 2610.3±429.4, As 557.2±259.4, Al 118.7±0.7
• Flaxseed oil: Pb 25.7±36.9, Cd 70.0±103.7, As 3.10±1.04, Al 29.8±3.5
• Sesame seed: Pb 321.2±64.2, Cd 448.4±44.3, As 678.8±33.7, Al 849.7±116.5
• Sesame oil: Pb 36.0±53.8, Cd 0.0±0.0, As 15.2±8.4, Al 16.0±14.5
• Perilla seed: Pb 936.0±97.1, Cd 56.4±16.0, As 259.8±36.4, Al 1023.5±187.5
• Perilla oil: Pb 35.5±34.4, Cd 8.6±14.7, As 14.1±3.9, Al 25.5±10.5

Solvent extraction (n-hexane, Buchi E-816 HE, 165 °C, 100 min):

• Flaxseed seed: Pb 1.57±0.60, Cd 11.0±1.8, As 2.34±1.09, Al 0.50±0.003
• Flaxseed oil: Pb 0.17±0.06, Cd 1.06±1.31, As 0.15±0.13, Al 0.029±0.008
• Sesame seed: Pb 0.76±0.15, Cd 1.06±0.11, As 1.60±0.08, Al 2.01±0.28
• Sesame oil: Pb 0.025±0.009, Cd 0.0±0.0, As 0.037±0.052, Al 0.047±0.034
• Perilla seed: Pb 2.21±0.23, Cd 0.13±0.23, As 0.61±0.09, Al 2.41±0.44
• Perilla oil: Pb 0.019±0.016, Cd 0.0±0.0, As 0.011±0.013, Al 0.041±0.033

Supercritical CO₂ extraction (Natex L-1000 Bar-Pilot, 50 °C, 400×10⁵ Pa, 240 min):

• Flaxseed seed: Pb 19.3±7.4, Cd 134.8±22.2, As 28.8±13.4, Al 6.13±1.8
• Flaxseed oil: Pb 6.25±3.99, Cd 8.30±3.32, As 5.72±4.43, Al 0.18±0.11
• Sesame seed: Pb 12.8±2.6, Cd 17.9±1.8, As 27.2±1.3, Al 34.0±4.7
• Sesame oil: Pb 0.14±0.22, Cd 0.025±0.031, As 0.43±0.09, Al 1.23±0.73
• Perilla seed: Pb 37.4±3.9, Cd 2.26±0.64, As 10.4±1.5, Al 40.9±7.5
• Perilla oil: Pb 4.18±3.78, Cd 0.57±0.99, As 0.81±0.33, Al 3.03±1.67

Seed-to-oil migration rates (Fig. 2, percent transferred):

• Pb: 1.1–32.5% (flaxseed supercritical highest at ~32%); Cd: 0–27.3%; As: 0.9–20.0%; Al: 1.7–23.9%
• Across the three methods: pressing and solvent extraction transfer <10% of seed metal to oil for most pairs; supercritical CO₂ extraction reaches up to ~30% transfer, materially higher than the other two methods. Authors attribute this to the CO₂ phase being a better solvent for omega-3/-6 fatty acids and incidentally co-extracting more metal.

Noodle boiling (50 g portion in 1 L water at 100 °C; Table 2 raw values, Fig. 3 migration rates):

Flour noodles (Korean wheat flour, 3 replicates):

• Control: Pb 0.820±0.399 µg, Cd 0.714±0.042 µg, As 0.055±0.023 µg, Al 0.768±0.032 mg
• 3 min: Pb 0.354±0.250, Cd 0.173±0.122, As 0.000±0.000, Al 0.257±0.008
• 5 min: Pb 0.298±0.190, Cd 0.103±0.094, As 0.002±0.005, Al 0.217±0.017
• 10 min: Pb 0.251±0.126, Cd 0.049±0.009, As 0.011±0.018, Al 0.203±0.002

Glass noodles (Korean sweet-potato starch, 3 replicates):

• Control: Pb 0.634±0.150 µg, Cd 0.0±0.0 µg, As 0.086±0.022 µg, Al 0.981±0.005 mg
• 3 min: Pb 0.359±0.132, Cd 0.0±0.001, As 0.002±0.004, Al 0.556±0.004
• 5 min: Pb 0.289±0.177, Cd 0.0±0.0, As 0.0±0.0, Al 0.476±0.004
• 10 min: Pb 0.191±0.036, Cd 0.0±0.0, As 0.0±0.0, Al 0.347±0.002

Residue rates at 3 min boiling (the fraction of each metal that remained in the noodle rather than migrating to the cooking water, reported on page 7 of the source and visible as Fig. 3 y-axis labels): flour noodle Pb 43.8%, Cd 21.4%, Al 33.4%; glass noodle Pb 56.6%, Al 56.7%. The complementary migration rates at 3 min (computed from Table 2 values: 1 − 3 min / control) are approximately: flour noodle Pb 56.8%, Cd 75.8%, Al 66.6%; glass noodle Pb 43.4%, Al 43.3%. Note: the source-text wording for glass noodles (“decreased by 53.8% and 56.7%”) is internally inconsistent with the Fig. 3 axis label “residue rate” and the Table 2 raw values — Fig. 3 and the Table 2 math both make 56.7% the glass-noodle Al residue, not the decrease. Most water-soluble metals transfer to the boiling water within 3 min; further boiling to 10 min adds significant Al migration but limited additional Pb/Cd/As migration.

Tea infusion (1.2 g tea in 150 mL water at 98 °C; Table 2 reports amount transferred into infusion liquid, Fig. 4 migration rates):

Green tea (Korean):

• Control (in bag): Pb 0.219±0.008 µg, Cd 0.017±0.002 µg, As 0.073±0.015 µg, Al 4.694±0.119 mg
• 2 min infusion: Pb 0.047±0.002, Cd 0.003±0.003, As 0.003±0.005, Al 1.441±0.119
• 10 min infusion: Pb 0.056±0.001, Cd 0.003±0.003, As 0.006±0.002, Al 1.957±0.147
• 30 min infusion: Pb 0.065±0.002, Cd 0.005±0.003, As 0.007±0.005, Al 2.462±0.098

Black tea (Sri Lankan Camellia sinensis):

• Control (in bag): Pb 0.657±0.061 µg, Cd 0.011±0.001 µg, As 0.049±0.005 µg, Al 1.039±0.039 mg
• 2 min infusion: Pb 0.330±0.081, Cd 0.003±0.001, As 0.023±0.009, Al 0.663±0.128
• 10 min infusion: Pb 0.429±0.138, Cd 0.006±0.008, As 0.038±0.017, Al 0.786±0.126
• 30 min infusion: Pb 0.558±0.194, Cd 0.010±0.009, As 0.049±0.019, Al 0.864±0.054

Solomon’s seal tea (Korean Polygonatum odoratum var. pluriflorum roots):

• Control (in bag): Pb 0.474±0.163 µg, Cd 0.054±0.008 µg, As 0.053±0.009 µg, Al 0.428±0.026 mg
• 2 min infusion: Pb 0.086±0.017, Cd 0.004±0.004, As 0.002±0.003, Al 0.214±0.022
• 10 min infusion: Pb 0.175±0.121, Cd 0.014±0.014, As 0.009±0.003, Al 0.277±0.041
• 30 min infusion: Pb 0.298±0.118, Cd 0.020±0.004, As 0.015±0.003, Al 0.374±0.041

Migration rates at 2 min infusion (reported in text): Pb to infusion — black 50.2%, green 21.3%, Solomon’s seal 18.2%; Cd — black 33.3%, green 14.3%, Solomon’s seal 6.7%; As — black 46.3%, green 4.9%, Solomon’s seal 4.5%; Al — black 63.9%, green 30.7%, Solomon’s seal 49.9%. By 30 min, As migration reaches ~88% in black tea, ~28% in green tea (Fig. 4). The authors note that black tea (oxidatively fermented) consistently shows the highest migration rates for all four metals; green tea (un-oxidized) consistently the lowest.

Risk arithmetic the authors compute for Al in instant noodles: previous study reported P95 target hazard quotient (THQ) of 1.789 for adult Al exposure from instant noodles. After applying the 33.4% Al residual rate in flour noodles after 3 min boiling from this study (i.e., assuming the cooking water is discarded so only the retained ~33% of the original Al reaches the eater), the adjusted P95 THQ drops to 0.597 (1.789 × 0.334 ≈ 0.597), below the threshold of concern (THQ <1).

Methods (brief)

Three samples (n=3) of each seed type, noodle type, and tea type purchased from Korean offline markets. Sesame and perilla seeds were of Korean origin; flaxseeds were imported from Canada; black tea was Sri Lankan; green tea and Solomon’s seal tea were Korean. Oils were extracted by mechanical pressing (Poong-jin P-16(S); seeds were pre-fried first — flaxseed 160 °C × 10 min, sesame 180 °C × 20 min, perilla 160 °C × 10 min — then pressed at 70 °C / 200×10⁵ Pa for 15 min), n-hexane solvent extraction (Buchi E-816 HE, 165 °C, 100 min followed by 150 °C evaporation), and supercritical CO₂ extraction (Natex L-1000 Bar-Pilot, 50 °C, 400×10⁵ Pa, 240 min). Noodles (50 g) were boiled in 1 L water at 100 °C for 3, 5, and 10 min and the boiling water discarded before analysis. Teas (1.2 g) were infused in 150 mL water at 98 °C for 2, 10, and 30 min in polyethylene/polypropylene tea bags. Microwave-assisted acid digestion (HNO₃ 60% 10 mL + H₂O₂ 30% 1 mL) preceded ICP-MS or ICP-OES quantitation against external calibration curves (R² = 0.999, 0.01–20 µg/kg or 0.1–30 mg/kg). Migration and residual rates were calculated by total mass transfer rather than concentration alone, to control for the change in food/water mass during processing. Method validated using CRMs: TET009RM (FAPAS, edible oil), NIST 1547 peach leaves, NIST 1570a spinach leaves, KRISS fortified ginseng. Limitations: n=3 per matrix is very small for occurrence-distribution estimation; samples are a single Korean-market snapshot; metal speciation (iAs vs tAs, MeHg vs tHg) not measured — only total As reported. Soil provenance and cultivar of source crops not characterized.

Implications

Certification: Directly relevant to HMT&C product-category pages for cooking oils, noodles, and tea. The finding that 30–88% of metals migrate from tea leaves into infusion liquid means tea-leaf occurrence data overstates consumer exposure for green and Solomon’s seal teas and understates what reaches the cup for black tea, where Pb, As, and Al migration rates are highest. For noodles, the relevant consumer-facing matrix depends on whether cooking water is discarded (Korean/European convention) or consumed (ramen/pho/Asian-soup convention) — the wiki should not collapse these into a single exposure estimate. Supercritical CO₂ oil extraction transfers up to ~30% of seed metal to oil, materially more than pressing or solvent extraction, which matters when certifying CO₂-extracted oils.

Courses: Strong real-world demonstration of the processing lever (CLAUDE.md Part 24 ratcheting): discarding noodle cooking water reduces Pb intake by 30–70%, discarding tea bag after 2 min vs. 30 min cuts As exposure roughly in half. Useful case study for the QA/processing module on consumer-facing exposure modulation.

App: Processing-method flag is warranted for oils (extraction method matters: supercritical CO₂ ≠ mechanical pressing), noodles (consumed with or without cooking water), and teas (infusion time and tea type — black tea releases more metals per minute than green or herbal). Consumer translation should specify dose: e.g., one 150 mL cup of 30-min-steeped green tea delivers ~2.5 mg Al; a 50 g serving of flour noodles boiled 3 min and drained delivers ~0.46 mg Al residual vs. ~0.77 mg unboiled.

Verification notes

• 2026-05-18 (Claude Opus 4.7): Merge-enhanced from prior 2026-05-14 version. Original page had Key numbers values transposed across foods (flaxseed/sesame/perilla Pb/Cd/As ranges scrambled), tea infusion Al figures replaced with noodle Al values, and noodle boiling migration rates underreported (page said “14–46% of Pb” migration; verified against text and Table 2, actual range is roughly 40–70% across noodle types and times). Reread Table 2 and Figs 2–4 of the source PDF (raw/Papers Cube Manual Fetch/article (1) copy.pdf, sha256 fab5791c…); rewrote Key numbers from the table verbatim. Other corrections: publication name “Applied Biology and Chemistry” → “Applied Biological Chemistry” (full Springer/KSABC name as printed on PDF page 1); jurisdictions [KR, LK] → [KR] (samples were purchased in Korea, even though some commodities originate elsewhere; LK is the country of origin of the black-tea variety, not the regulatory scope of the study); added access_url, raw_sha256.
• 2026-05-18 (audit subagent / Claude Opus 4.7 application): fresh-context audit returned REVISE. Findings applied: (1) ❌ Check 1 — noodle 3-min figures (flour 43.8/21.4/33.4%, glass 56.7%) were the residue rates per the source text (“remained by”) and Fig. 3 axis label “residue rate”, not migration. Wiki originally labeled them as “Migration rates” in the Key numbers paragraph. Relabeled as residue rates and added the complementary migration rates computed from Table 2 (flour Pb 56.8%, Cd 75.8%, Al 66.6%; glass Pb 43.4%, Al 43.3%). Noted the source-text inconsistency for glass-noodle wording (“decreased by 56.7%” vs Fig. 3 axis “residue”). (2) ❌ Check 1 — THQ paragraph framed 33.4% as a “migration rate”; the source explicitly calls it the “residual rate” (page 7), and the math (1.789 × 0.334 ≈ 0.597) requires it to be the retained fraction. Reworded to “33.4% Al residual rate … assuming the cooking water is discarded so only the retained ~33% of the original Al reaches the eater”. (3) ⚠️ Check 1 — synthesis opener ranges “30–70% Pb / 30–88% tea metals” tightened to “roughly 40–70% Pb across noodle cells” and “roughly 5–88% across metals × tea types × steep times” (4.5% Solomon’s seal As at 2 min is the true low end; 30% omitted the small-migration cells). (4) ❌ Check 2 — [[metals/aluminium]] corrected to [[metals/aluminum]] (canonical US spelling; aluminium.md does not exist). (5) ❌ Check 2 — [[mitigation/processing-levers]] corrected to [[mitigation/processing]] (the actual page that exists in wiki/mitigation/; processing-levers.md does not exist). (6) ⚠️ Check 3 — Methods press-duration “10–20 min” was the pre-fry range, not the press duration; clarified to “seeds pre-fried (160 °C × 10 min for flaxseed, 180 °C × 20 min for sesame, 160 °C × 10 min for perilla), then pressed at 70 °C / 200×10⁵ Pa for 15 min”.
• 2026-05-18: ❌ Check 2 findings on ingredient/product slugs (ingredients/flaxseed, ingredients/perilla, products/cooking-oils) — kept as declared. Verified: flaxseed is in the aliases array of wiki/ingredients/sunflower-seeds.md (the alias grouping is questionable on the merits, but is how the wiki currently routes flaxseed and is not this paper’s defect to fix); perilla is aliased on wiki/ingredients/fresh-herbs.md; cooking-oils is enumerated in the routing-audit’s broadProductScopes set so the declaration fans out across specific oil rows via the inheritedProductRoutes map. Routing audit returns 0 unresolved for this source on both this and the prior commit, confirming the system accepts these slugs. The audit subagent’s flag is technically correct against the taxonomy-snapshot file but not against the live routing layer; documented here so a future taxonomy-snapshot sweep can decide whether to promote flaxseed and perilla to first-class ingredient pages and cooking-oils to a product umbrella row.

Wiki pages updated on ingest

• sesame-oil
• tea
• noodles
• aluminum
• lead
• cadmium
• arsenic
• processing

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.