Dordevic et al. 2019 — Aluminum leaching into food from household aluminum foil during baking, Czech Republic

The authors quantified aluminum migration from five commercially available household aluminum foils into eleven food types (Atlantic salmon, mackerel, duck breasts with and without skin, cheese Hermelín, fresh tomato, fresh paprika, Carlsbad dumplings, pork roast, pork neck, chicken breasts, chicken thighs) during a standardised 40-minute baking step at 220°C, with and without a typical olive-oil/salt/black-pepper marinade. Aluminum was measured in homogenised cooked food by both ICP-MS and electrothermal-atomization AAS, against inert-packaging controls and raw-food controls, with six replicates per sample. The study pairs the experimental work with a consumer-preference survey (n=784) on aluminum-foil use in Czech households, but the contamination findings — not the survey — are the wiki-relevant content.

Key numbers

Aluminum content is reported in both dry-mass and wet-mass bases (mg/kg = ppm); wet-mass values, given alongside, are the consumer-relevant figures. The “C-foil” series is unmarinated food baked in foil A–E; the “M-foil” series is marinated food baked in foil A–E. “C” and “CM” are control samples baked without foil contact (inert packaging). All ± values are ±SD over six replicates.

• Marinated salmon (Salmo salar) baked in foil A: 41.86 ± 0.56 mg/kg dry mass; 20.95 ± 0.18 mg/kg wet mass. ICP-MS. Highest among unmarinated salmon foils: foil A 1.31 ± 0.02 mg/kg dry; salmon baked in inert packaging (CIP, no foil) was below the AAS limit of detection (Table 2; p. 3351).
• Marinated mackerel (Scomber scombrus) baked in foil A: 49.34 ± 0.44 mg/kg dry mass; 13.42 ± 0.24 mg/kg wet mass. Unmarinated mackerel control (no foil): 1.17 ± 0.01 mg/kg dry. Highest unmarinated mackerel was foil C at 4.66 ± 0.10 mg/kg dry (Table 2; p. 3351).
• Marinated duck breast without skin baked in foil E (M-foil E): 117.26 ± 1.37 mg/kg dry mass; 45.18 ± 0.01 mg/kg wet mass. This is the single highest aluminum concentration recorded in the study. Highest with-skin marinated counterpart was foil B (MS-foil B): 51.71 ± 0.73 mg/kg dry; 22.79 ± 0.13 mg/kg wet — the paper uses this value as its narrative example for the skin-as-barrier finding (p. 3354). For reference, MS-foil E with-skin marinated was 38.39 ± 0.29 mg/kg dry; 14.79 ± 0.00 mg/kg wet (Table 3, p. 3352). Table 3 is laid out as two side-by-side column-groups: the left “Duck breasts” group reports the unmarinated series, the right “Duck breasts—marinade” group reports the marinated series; within each group the C/CIP/C-foil rows are without-skin samples and the CS/CIPS/CS-foil (unmarinated) and CMS/MIPS/MS-foil (marinated) rows are with-skin samples. The abstract reports the headline 117.26 ± 1.37 value as “g/kg” — that is a typesetting error; Table 3 and the discussion body both state mg/kg.
• Marinated pork roast in foil E (highest pork-roast value): 15.87 ± 0.07 mg/kg dry mass; 6.90 ± 0.08 mg/kg wet mass (Table 4; p. 3352).
• Marinated pork neck in foil E: 4.91 ± 0.06 mg/kg dry; 1.74 ± 0.00 mg/kg wet (Table 4).
• Marinated chicken breasts in foil C: 5.25 ± 1.48 mg/kg dry; 1.70 ± 0.04 mg/kg wet (Table 4).
• Marinated chicken legs in foil D: 3.6 ± 0.2 mg/kg dry; 1.24 ± 0.03 mg/kg wet (Table 4).
• Marinated tomato in foil A: 7.78 ± 0.18 mg/kg dry; 0.37 ± 0.02 mg/kg wet (Table 4). Tomato raw mass (no foil): 1.70 ± 0.02 mg/kg dry; 0.08 ± 0.00 mg/kg wet.
• Marinated paprika (red pepper) in foil E: 1.32 ± 0.03 mg/kg dry; 0.10 ± 0.00 mg/kg wet (Table 4).
• Marinated cheese Hermelín in foil E: 4.46 ± 0.12 mg/kg dry; 1.68 ± 0.02 mg/kg wet (Table 4).
• Marinated Carlsbad dumpling in foil E: 1.88 ± 0.42 mg/kg dry; 0.90 ± 0.01 mg/kg wet (Table 4).
• Magnitude effect: the authors report that aluminum content in marinated mackerel and salmon increased up to 20 times after baking in foil relative to the no-foil controls, and the body text states an increase “up to 40 times” was measured across the dataset (Abstract; §3 first paragraph, p. 3353).
• Foil-vs-marinade interaction: not marinated samples baked in foil consistently showed lower aluminum content than the marinated counterparts in the same foil; significant differences (p<.05) between marinated and not-marinated samples confirm pH/salt-driven dissolution of the foil per the Al₂O₃ + 6H⁺ → 2Al³⁺ + 3H₂O mechanism reported in the discussion (p. 3356).
• Background-exposure context cited from the literature (not measured here): EFSA TWI 1 mg Al/kg body weight/week (EFSA, 2008); WHO 1989 weekly intake limit 7 mg/kg bw; estimated dietary aluminum exposure in European countries 1.6–13 mg/day (Bratakos et al. 2012, p. 3358); per-kg-body-mass dietary aluminum exposure 0.2–1.5 mg/kg in a 60 kg adult attributed by the paper to EFSA 2008 (p. 3358; time-basis qualifier not stated by the source for this figure).

Methods (brief)

Aluminum was determined by two methods on the same homogenised, lyophilised samples: ICP-MS on an Agilent 7700x with an ASX-520 autosampler (No-Gas mode; ²⁷Al monitored with ⁴⁵Sc internal standard; RF 1550 W; nebuliser flow 1.07 L/min; dwell 300 ms ²⁷Al, 100 ms ⁴⁵Sc; six replicates), and electrothermal-atomization AAS on a ContrAA 700 (Analytik Jena AG) with Mg(NO₃)₂·6H₂O matrix modifier (pyrolysis 1,500°C; atomization 2,400°C; λ = 396.1520 nm; calibration 0–100 ppb from a 1 g/L Analytika aluminum standard). Microwave digestion used MLS 1200 Mega closed-vessel (Milestone) for ICP-MS digestion (2 mL HNO₃ + 1 mL H₂O₂, multi-step power program from 250 W to 600 W; final volume 15 mL ultrapure water; storage −20°C) and an ETHOS SEL (Milestone) with 67% Analpure HNO₃ + 30% H₂O₂ for the AAS arm (two-stage decomposition to 200°C / 1,000 W max). Foils were independently characterised by portable X-ray fluorescence (NITON XL3t GOLDD+ Thermo Fisher Scientific; 50 keV and 15 keV beams) to confirm that the five market foils differ in alloy composition. Salt was determined by Mohr titration; lipid by Soxtec 2055 Soxhlet (FOSS Tecator); protein by Kjeltec 2300 (FOSS Tecator); dry matter gravimetrically per ČSN ISO 1442:1997 (Binder FD 53). Baking conditions: 40 min at 220°C in a GARB-IN Professional Ovens Model 23 GM UMI. Vacuum-packed, frozen homogenates were thawed and analysed sequentially. ICP-MS values are systematically higher than AAS values for the same samples, consistent with prior literature on Al determination by the two techniques.

Speciation note: the paper reports total aluminum only; no speciation is performed and none is needed for the food-contact-leaching question being asked.

Implications

Certification: Direct evidence on Al migration from food-contact aluminum foil into the wrapped food during baking, in the foils/wraps product row. The highest measured wet-mass value (45 mg/kg in a marinated duck breast, foil E) is the upper end of the range observed in this study; the authors themselves conclude on p. 3354 that “the aluminum contents found among investigated samples are not alarming, though the increase was measured up to 40 times” relative to raw food. Cooking-method context for the certification program: marinated and acidic foods baked in direct foil contact at ≥220°C are the high-leaching condition documented here; sugar-containing marinades reduce leaching (Joshi et al. 2003, cited p. 3357) while olive-oil/salt marinades, as used here, do not. The contamination pathway is the foil itself, not the food ingredient, so HMTc reads this paper as foils/wraps-row evidence with food-vehicle context, not as ingredient-occurrence evidence on salmon, duck, etc.

Courses: A high-signal teaching example for the food-contact-materials module: same food, same recipe, same oven, foil-vs-no-foil controls, with measurable Al migration that scales with marinade acidity/salinity, foil alloy composition, and baking temperature relative to the literature comparison at 180°C (Ertl & Goessler 2018) and 250°C (Turhan 2006). The Al₂O₃ + 6H⁺ → 2Al³⁺ + 3H₂O surface-chemistry mechanism is usable as the explanatory backbone. Skin as a barrier — duck breast with skin retained roughly half the aluminum of without-skin under the same foil — is a small but didactic finding for the meat-cooking lesson.

App: The aluminum contamination_profile cell for the foils/wraps product row receives a direct evidence contribution here. Ingredient-page Al cells for fish, chicken, poultry, and tomato should treat the values reported in this paper as cooking-method context (Al from foil contact during baking) rather than as intrinsic-ingredient occurrence. Only two foods in this study have intrinsic-ingredient (raw) Al values reported: raw tomato 1.70 ± 0.02 mg/kg dry / 0.08 ± 0.00 mg/kg wet, and raw Karlovarsky knedlík 2.31 ± 0.02 mg/kg dry / 1.07 ± 0.02 mg/kg wet (Table 4, p. 3352); the salmon, mackerel, duck, pork, and chicken raw columns are reported as ”–” (not measured). The app should warn on marinated foods baked in direct aluminum-foil contact, not on the ingredients themselves.

Verification notes

• Brand-firewall (Part 12): the paper names five specific aluminum-foil product brands (foil A–E) by manufacturer and retail name in §2.2 and names specific food-retail brands (Tesco Stores, Ocean48 s.r.o., Cut chicken breasts/legs, Maso Uzeniny Polička, EFES Czech Republic, etc.) as the source of the food samples in §2.1. Brand attribution has been stripped from this page in line with the Part 12 strict reading locked 2026-05-17: only the food-form descriptor and the foil-letter code (foil A, foil B, …) are retained, and the foil-letter codes are de-identified to their A–E labels exactly as the paper uses them. The five foils differ statistically in alloy composition (XRF/PCA in Figure 3c), which is the wiki-relevant point. Brand-by-brand foil rankings or food-by-food retail-brand attribution are intentionally excluded.
• Methods-vendor exception (Part 12 Exception 2): instrument vendor and model names (“Agilent 7700x ICP-MS”, “ContrAA 700 (Analytik Jena AG)”, “MLS 1200 Mega (Milestone)”, “ETHOS SEL (Milestone)”, “Soxtec 2055 (FOSS Tecator)”, “Kjeltec 2300 (FOSS Tecator)”, “NITON XL3t GOLDD+ (Thermo Fisher Scientific)”, “GARB-IN Model 23 GM UMI”, “Analytika aluminum standard 1 g/L”, “Analpure HNO₃ 67%”, “Binder FD 53”) are retained per scientific-reproducibility carve-out.
• Abstract typo: the abstract reports duck-breast aluminum content as “117.26 ± 1.37 g/kg”; Tables 3 and the discussion body both report “117.26 ± 1.37 mg/kg”. The wiki page uses mg/kg per the tables (the body-text value, not the abstract). Flagged here so a later reader does not propagate the typo.
• Dry/wet basis: all numbers above are pulled with explicit dry-mass / wet-mass labels. Wet mass is the consumer-relevant basis for cooked foods; dry mass is the basis that survives any reconstitution analysis. Both are preserved.
• Survey content (§2.6, §3.1, Figure 4): the n=784 consumer-preference survey is descriptive of Czech consumer behaviour around aluminum foil. The wiki page deliberately summarises only the contamination data and does not propagate the survey’s demographic chi-squared findings; those are not heavy-metal-occurrence evidence and would not change any wiki claim.
• Matrices vocabulary: cooked-meat (3 sources), baked-food (this is the first), and food-contact-materials (6 sources) are corpus-convention matrices labels rather than entries in the system-prompt’s enumerated matrices vocabulary. fish is a canonical matrix. All four are retained.
• Audit subagent (2026-06-02, general-purpose fresh context) verdict REVISE; 3 findings ❌ all independently re-verified against Table 3 and Table 4 of the source and applied: (1) with-skin marinated duck breast foil E values 51.71/22.79 were misattributed — Table 3 MS-foil B = 51.71 ± 0.73 dry / 22.79 ± 0.13 wet (which is the paper’s narrative with-skin example on p. 3354 and the highest with-skin marinated value), while MS-foil E = 38.39 ± 0.29 dry / 14.79 ± 0.00 wet; corrected to report both, with the foil-B value as the paper’s example and the foil-E value for comparability with the headline 117.26 figure; (2) “raw salmon 0.54 mg/kg wet” was invented — Table 4 reports raw values only for tomato and Karlovarsky knedlík (other rows are ”–”), and the 0.54 figure is actually C-foil A salmon wet-mass; corrected to remove the raw-salmon claim and to enumerate only the two raw values the source actually reports; (3) the Table 3 structural-description parenthetical described the panel layout backwards — Table 3’s left “Duck breasts” panel is unmarinated and its right “Duck breasts—marinade” panel is marinated, with the within-skin split carried by the C/CS row prefixes; rewrote the parenthetical. Two ⚠️ findings also applied: (4) the “45× EFSA TWI” framing in Implications was wiki-derived arithmetic the paper did not perform — replaced with the paper’s own published characterisation (“not alarming”); (5) the EFSA 2008 0.2–1.5 mg/kg body-mass figure was muddled into the Bratakos 2012 1.6–13 mg/day attribution — re-split the two attributions as the source page (p. 3358) actually carries them, and added a “time-basis qualifier not stated” note. Zero findings rejected.

Wiki pages updated on ingest

• aluminum
• food-packaging-foils-wraps
• fish
• chicken
• poultry
• tomato

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.