Kaya et al. 2024 — Heavy metals in commercial Turkish liquid milk and its packaging (JETAS)

This Turkish study quantifies ten elements (Al, As, Cd, Co, Cu, Fe, Pb, Mn, Ni, Zn) in 10 commercial liquid milk samples sold in Turkish markets and in their corresponding aseptic packaging materials using microwave-acid digestion followed by ICP-OES, then performs Pearson correlation analysis between milk and packaging concentrations for each element. The headline observation is that packaging concentrations exceed milk concentrations by one to five orders of magnitude across all detected elements, with aluminium foil–laminate packaging carrying 51,267–71,601 mg/kg Al, 31.6–42.4 mg/kg As, and 22.9–26.7 mg/kg Pb at the packaging-matrix level. The data are useful for packaging-matrix occurrence in aseptic dairy cartons and for the question of whether packaging composition correlates with milk concentrations. Important methodological caveat: most milk values reported sit at or below the instrument’s stated LOD/LOQ for the element (see Methods and Verification notes).

Key numbers

All values in mg/kg, as reported by the authors. Milk and packaging concentrations are arithmetic means of triplicates ± SD across 10 commercial samples; ranges below are the min–max of those sample means (Table 5).

Milk samples (T1–T10):

• Al: 1.219–2.578
• As (total): 1.078–1.522 — all sample means below the instrument LOD of 3.358 mg/kg for As; values reported anyway by the authors
• Cd: 0.219–0.239 — just above the instrument LOD of 0.211 mg/kg; all below the uniform LOQ of 120 mg/kg
• Co: 0.081–0.393 — T1 is an outlier at 0.393; the other nine samples span 0.081–0.129. (The abstract erroneously prints “Co (0.219–0.239)” using the Cd values; Table 5 is authoritative.)
• Cu: not detected in any milk sample (below LOD 0.689 mg/kg)
• Fe: 0.862–81.440 — T1 is an extreme outlier at 81.44; the other nine span 0.862–6.294
• Pb: 1.784–2.170 — barely above the instrument LOD of 1.771 mg/kg; all below the uniform LOQ of 120 mg/kg
• Mn: 0.040–1.884 — T1 outlier at 1.884; the other nine span 0.040–0.078
• Ni: 0.782–9.758 — T1 outlier at 9.758; the other nine span 0.782–0.941, all below the instrument LOD of 1.074 mg/kg
• Zn: 1.814–2.522

Packaging materials (T1P–T10P), aseptic multi-layer cartons (polyethylene film / laminated cardboard / aluminium foil construction per §1):

• Al: 51,267.00–71,601.25 (dominant packaging element, consistent with aluminium-foil-laminate barrier)
• As: 31.626–42.371
• Cd: 1.035–1.209
• Co: 0.769–1.167 (the running text cites T5P 0.775 as the lowest, but Table 5 shows T8P 0.769 as the minimum)
• Cu: 4.921–44.839
• Fe: 259.615–463.182
• Pb: 22.925–26.668 (the abstract prints 24.386–26.668; Table 5 shows T8P 22.925 as the minimum)
• Mn: 4.301–59.599
• Ni: 6.065–7.943
• Zn: 5.324–8.763

Pearson correlation between milk and packaging concentrations across the 10 paired samples (Table 6):

• Al: r = −0.742 (high; inverse association)
• As: r = +0.556 (moderate; positive association)
• Cd: r = +0.232 (very weak)
• Co: r = −0.089 (very weak)
• Cu: not computed (milk Cu below LOD)
• Fe: r = +0.149 (very weak)
• Pb: r = −0.281 (weak; inverse)
• Mn: r = +0.175 (very weak)
• Ni: r = +0.242 (very weak)
• Zn: r = +0.401 (weak; positive)

Analytical performance (Table 3, ICP-OES SpectroBlue II):

• Wavelengths (nm): Al 167.078; As 189.042; Cd 214.438; Co 228.616; Cu 324.754; Fe 259.941; Pb 220.353; Mn 257.611; Ni 231.604; Zn 334.502
• LOD (mg/kg): Al 0.139; As 3.358; Cd 0.211; Co 0.732; Cu 0.689; Fe 0.972; Pb 1.771; Mn 0.356; Ni 1.074; Zn 0.261
• LOQ (mg/kg): 120.000 for all elements except Zn (90.000)
• r² ≥ 0.99945 for all calibration curves; %RSD across triplicates ≤ 6.565% (Table 4)

Regulatory comparisons stated by the authors:

• Turkish Food Codex Pb limit in milk: 0.02 mg/kg. All 10 milk Pb means (1.784–2.170 mg/kg) exceed this limit by roughly two orders of magnitude as reported.
• Turkish Food Codex Zn limit in food: 5 mg/kg. Milk Zn (1.814–2.522 mg/kg) below limit.
• FAO/WHO PTWI for Pb cited as 1.2 µg/kg bw/day; no exposure calculation performed in the paper.
• BFAO/WHO tolerable Al cited as 1 mg/kg bw/week.

Methods (brief)

Sample preparation: 10 brands of commercial liquid cow milk and 10 corresponding aseptic carton packaging materials were collected from Turkish retail. Microwave-acid digestion was performed on a CEM MARS6 system in Teflon vessels with 10 mL 97% nitric acid, then made up to 25 mL with ultra-pure water. Sample masses: 0.25 mL for milk and 0.1 g for packaging. Digestion programs: 200 °C max for milk (15 min ramp, 15 min standby, 1400 W max) and 210 °C max for packaging (20 min ramp, 15 min standby, 1400 W max).

Elemental analysis: Spectro Blue II ICP-OES, cyclonic spray chamber, axial/radial dual view, quartz torch, argon (99.9%) at 6–8 bar main pressure, plasma power 1400 W, 0.8 L/min nebulizer flow, 13 L/min coolant flow, 3 replicates per sample. Calibration via Merck multielement standard IV (1000 ppm) prepared as a 5–6 point curve. The 10 analytes were determined in the same run.

Statistics: SPSS Pearson correlation between paired milk and packaging concentrations for each element. No internal reference material (CRM) or spike-recovery is reported; the only QC parameters described are LOD, LOQ, calibration r², and %RSD across triplicates. Speciation: total elemental analysis only — As is reported as total As (tAs), not speciated; no Hg measured.

Manuscript timeline: received 24 November 2023, accepted 28 May 2024, published online 30 August 2024. Funding: TUBITAK 2209-A grant 1919B012215735. Journal JETAS (e-ISSN 2548-0391) is a Turkish open-access engineering technology journal.

Limitations relevant to citing these data:

• n = 10 brands from one university’s procurement is a small, geographically narrow sample.
• For most analytes in milk, the reported sample means lie at or below the instrument’s stated LOD (As, Pb, Ni for 9/10 samples) and well below the uniform LOQ of 120 mg/kg. Quantitative interpretation of these milk values below the authors’ own quantitation threshold is not supportable; the values are usable as detection-presence indicators only.
• No certified reference material (e.g., NIST milk powder) is reported; no spike-recovery; no procedural blank values disclosed.
• The abstract contains two internal inconsistencies relative to Table 5 (Co milk range mis-labelled; Pb packaging minimum mis-cited). Table 5 is the authoritative numerical record.
• Pb concentrations reported in milk (~2 mg/kg) are 80–100× the Turkish Food Codex limit and ~100× the FAO/Codex limit for milk (0.02 mg/kg). The authors do not flag this as anomalous; for context, prior Turkish and European raw/pasteurised milk surveys typically report Pb at µg/kg-level concentrations. This paper’s milk Pb numbers should be treated as the authors’ reported values pending CRM/spike-recovery confirmation, not as a population estimate.

Implications

Packaging-matrix occurrence: this is one of relatively few datasets providing direct measurements of trace and toxic elements in finished aseptic carton material (polyethylene/cardboard/aluminium-foil laminate) used for commercial liquid milk. The Al, As, Pb, Cd, and Ni packaging-matrix values are usable as occurrence inputs for any analysis that needs concentrations in the packaging substrate itself, distinct from migration potential into the contained product.

Packaging vs product correlation: of the ten paired-correlation analyses, only Al shows a high-magnitude relationship (r = −0.742, inverse) between packaging composition and milk concentration; As shows a moderate positive correlation (r = +0.556); the remaining elements are weak or very weak. The high inverse Al correlation is the authors’ headline statistical finding and warrants caution in interpretation — the sign and magnitude are based on n = 10 sample means without adjustment.

Migration framing: this is a static-concentration study, not a migration study. The packaging values describe what is present in the packaging matrix, not what enters the product under storage conditions. Migration assessment would require contact-time experiments at controlled pH/temperature, which this paper does not perform.

Cochrane usability: the dataset is admissible for packaging-matrix occurrence at the documented quality tier (no CRM, no spike-recovery, small n) and is admissible for milk-matrix detection-presence indicators only, not quantitative milk-matrix concentration estimates, because most milk values lie below the authors’ own LOQ.

Verification notes

Merge-enhance pass 2026-05-18 against the PDF at raw/Papers Cube Manual Fetch/10.30931-jetas.1395566-3557768.pdf (DOI 10.30931/jetas.1395566). Changes relative to the prior 2026-05-14 revision:

• Milk Co range corrected from 0.219–0.239 (the abstract’s mis-labelled value, which is actually Cd) to 0.081–0.393 (Table 5 actual Co milk column). Documented the abstract’s paper-internal error.
• Milk Cd range added (was previously “not reported in milk ranges summary”): 0.219–0.239 mg/kg per Table 5, confirmed by §3 body text (“cadmium levels in milk and packaging samples were determined as 0.219–1.209 mg/kg”).
• Milk Cu added as “not detected” (was previously absent from the milk list). All milk samples reported as N.D. for Cu in Table 5 and confirmed in §3 body text.
• Packaging Pb range corrected from 24.386–26.668 (the abstract’s mis-stated minimum) to 22.925–26.668 (Table 5 actual; T8P 22.925 is the true minimum, not T6P 24.386).
• Packaging Co range refined from 0.775–1.167 to 0.769–1.167 (Table 5 shows T8P 0.769 as the minimum; the running text mis-cites T5P 0.775).
• Units note rewritten. The earlier note hedged that values “may represent µg/kg or mg/L.” The paper consistently reports mg/kg and the authors’ discussion compares directly to mg/kg literature values. The real interpretive caveat — most milk values fall at or below the instrument LOD/LOQ — is now stated directly in the limitations.
• Analytical detail expanded to include wavelengths, microwave digestion program, and explicit absence of CRM/spike-recovery QC.
• Pearson correlation table added (Table 6) — milk-vs-packaging r values for each element.
• Implications rewritten to remove HMTc-directional framing (“HMT&C should address…”, “courses module…”). Replaced with packaging-matrix occurrence framing and Cochrane usability statement, per Part 2 firewall.
• Author diacritics restored (Zurnacı, Şener).
• Legacy section heading ## Wiki pages updated on ingest replaced with current convention ## Wiki pages this source may touch.
• Speciation: metals: [Pb, tAs, Cd, Ni, Al] retained — the paper measures total elemental As (no speciation), so tAs is correct.

Fresh-context audit (2026-05-18, general-purpose subagent) verdict: REVISE.

• ⚠️ Audit finding: products: [] on a food paper flagged as a Phase 1 frontmatter defect. Verified false positive. Per CLAUDE.md Part 18 lint, the “missing products field on food/personal-care paper” rule fires only when matrices: contains product-shaped slugs (i.e., slugs in the products taxonomy). For this paper, matrices: [milk, food-packaging, packaging-material] — none of those are product-taxonomy slugs (verified against the taxonomy snapshot at docs/gpt-collaboration/taxonomy-snapshot.md). The convention for milk papers is products: [] with ingredient routing via milk-and-dairy — established by other milk papers (e.g., abdelnaby2022-lead-cadmium-milk.md) that share the identical pattern and produce the same advisory-only severity entry in data/evidence/routing_malformed.csv. The current products taxonomy does not include a “fluid milk in aseptic carton” product page; milk is intentionally represented at the ingredient level. No frontmatter change applied.
• ⚠️ Audit borderline note on matrices duplication (food-packaging vs packaging-material). Retained both: food-packaging flags the paper as a food-contact-material study; packaging-material flags the data as in-substrate concentrations on the packaging itself. They route distinct intents.
• Checks 1, 3, 4, 5 all ✅ from the audit, including independent verification of every Key numbers value against Table 5.

Wiki pages this source may touch

• milk-and-dairy
• lead
• arsenic-total
• cadmium
• aluminum
• nickel

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.