Karavoltsos et al. 2020 — Copper complexation and 16 trace metals in Greek vinegars

This study applied differential pulse anodic stripping voltammetry (DPASV) to characterize copper complexing capacity in 43 vinegars retailed in Greece, and used ICP-MS to quantify 16 trace elements (Al, As, Ba, Cd, Co, Cr, Cs, Cu, Fe, Mn, Ni, Pb, Rb, Sr, V, Zn). Balsamic vinegars consistently showed higher concentrations of Al, Cr, Fe, Mn, Rb, Co, Sr, and Cs compared to common vinegars, reflecting their production from cooked concentrated grape must. Of the toxicologically significant metals, Pb concentrations reached 86 µg/L in one balsamic red vinegar with honey, and two samples (5% of total) exceeded the California state maximum-allowed Pb intake equivalent of 34 µg/L; total As and Cd were lower but detectable.

Key numbers

All concentrations reported as µg/L (= ppb for the as-sold liquid vinegar matrix). Sixteen-element ICP-MS on a Thermo Scientific iCAP Qc. Reported values from Table 1 (BR/BRH/BW/WR/WW/F subgroup medians and ranges) and Figure 2 (full subgroup distributions).

Organic-matter context (Table 1):

• Organic carbon (OC) ranged 17–278 mg/L overall. Balsamic median 152 mg/L vs. common median 30 mg/L (Mann–Whitney p < 0.001; ≈5-fold). Common vinegars never exceeded 50 mg/L OC; balsamic vinegars never fell below 90 mg/L OC.
• pH ranged 2.2–3.8 in common vinegars (median 2.8); 2.6–3.3 in balsamic vinegars (median 3.0).

Cu complexing capacity (L_T) and total Cu:

• L_T (DPASV, copper-complexing ligands) ranged 0.05 µM (sample F2) to 52 µM (sample BR6) across the 43 vinegars.
• Median L_T 14 µM in balsamic vs. 0.86 µM in common (Mann–Whitney p < 0.001). Within balsamic, red-grape balsamic median 16 µM and red-grape balsamic with honey 13 µM were the highest; fruit-vinegar median 0.54 µM was the lowest.
• Median log K_app 7.3 for both balsamic and common (p > 0.05); white-grape balsamic 8.0 (slightly elevated, suggesting distinct ligand chemistry).
• Total Cu ranged <0.01 to 4.0 µM across the 43 vinegars. 30% of white-grape balsamic samples had total Cu > 1.0 µM.
• In 21% of samples (9 of 43), total Cu exceeded the corresponding L_T, indicating free or weakly complexed labile Cu; ~90% of those samples were common vinegars (high acidity + low OC favors labile Cu).
• L_T/OC density (nmol Cu per mg organic carbon) was highest in red-grape balsamic (median 122) and balsamic-with-honey (76), substantially exceeding earlier studies of Greek herbal infusions (45), coffee brews (34), and beers (7.3) measured by the same DPASV method (Supplemental Table S1).

Total arsenic (tAs, paper does not speciate):

• Range across all vinegars: <0.05 to 26 µg/L (the 26 µg/L maximum was in a white-grape balsamic).
• Median common 2.6 µg/L; median balsamic 4.2 µg/L (Mann–Whitney p = 0.095, not significant).
• Maximum daily tAs intake at 15 mL serving = 0.40 µg, ≈300× lower than the JECFA-withdrawn PTDI for inorganic As of 126 µg/day for a 60 kg adult; the corresponding BMRL0.5 (lower limit on benchmark dose for 0.5% lung-cancer response) is 180 µg/day.
• Speciation limitation: paper reports total As (tAs) only; iAs/organic As speciation not performed.

Pb (lead):

• Range across all vinegars: <0.05 to 86 µg/L (the maximum was in a balsamic red vinegar with honey). Second-highest 63 µg/L in a balsamic red.
• Median common 6.0 µg/L; median balsamic 9.8 µg/L (Mann–Whitney p = 0.654, not significant).
• The California-state maximum allowable Pb intake (0.5 µg/day) corresponds to 34 µg/L vinegar at a 15 mL daily serving; 2 of 43 samples (5%) exceeded this threshold. Both exceedances were balsamic (one BR, one BRH).
• The paper attributes the dominant Pb source to anthropogenic inputs (chemicals used during production and materials used in storage/transport), per Pb-isotope evidence from Ndung’u et al. 2011 [reference 6].

Cd (cadmium):

• Range across all vinegars: 0.03 to 2.3 µg/L (the 2.3 µg/L maximum was an aged balsamic from red grapes).
• Median common 0.30 µg/L; median balsamic 0.10 µg/L (Mann–Whitney p = 0.046, common slightly higher).
• All samples well below the JECFA PTDI for Cd of 50 µg/day for a 60 kg adult, which corresponds to 3.3 mg/L (3,300 µg/L) vinegar at a 15 mL daily serving.

Al (aluminum):

• Median balsamic 4,657 µg/L (4.66 mg/L); median common 768 µg/L (0.77 mg/L); Mann–Whitney p < 0.001.

Cr (total chromium, not speciated):

• Median balsamic 44 µg/L; median common 17 µg/L; Mann–Whitney p < 0.001.
• Reported as total Cr; Cr-VI not measured.

Other elements with significant balsamic > common differences (Mann–Whitney; from Figure 2 box plots and §2.2):

• Fe: balsamic median 4,394 µg/L vs. common 782 µg/L (p < 0.001).
• Mn: balsamic 1,055 µg/L vs. common 357 µg/L (p < 0.001).
• Rb: balsamic 1,053 µg/L vs. common 318 µg/L (p < 0.001).
• Co: balsamic 4.0 µg/L vs. common 1.4 µg/L (p = 0.004).
• Sr: balsamic 555 µg/L vs. common 329 µg/L (p = 0.004).
• Cs: balsamic 1.9 µg/L vs. common 0.60 µg/L (p = 0.004).

Elements without significant subgroup differences (p > 0.05): Ba, Cu, V, Ni, Zn. Ni medians in Figure 2 sit broadly in the 10–30 µg/L range across subgroups.

Indicative enrichment factors (IEF; element/Fe ratio in vinegar normalized to upper continental crust per Rudnick & Gao 2003):

• ≈100 for Rb, Zn, Cd.
• ≈50 for As.
• 10–25 for Cu, Mn, Pb, Cs.
• <10 for remaining elements.
• High IEFs for Cd, Rb attributed to phosphate-fertilizer carry-over; high Zn IEF attributed to Zn-containing fertilizers and pesticide use in viniculture; high Cu IEF attributed to Cu-based pesticide use in viniculture.

Methods

ICP-MS (Thermo Scientific iCAP Qc, single-collision-cell mode with kinetic energy discrimination using pure He) was used to determine all 16 trace elements after HNO₃ (65% supra-pure, Merck) / H₂O₂ (30%) digestion of vinegar samples. Internal standardization with ⁴⁵Sc and ¹⁰³Rh corrected matrix-induced signal suppression and instrumental drift. Measurements in triplicate; precision repeatability (n = 5) <15% for all analytes. LODs (µg/L): Co 0.02, Cs 0.02, Cd 0.03, As 0.05, Pb 0.05, V 0.05, Ba 0.10, Cr 0.10, Rb 0.10, Sr 0.10, Mn 0.20, Cu 0.40, Ni 0.40, Zn 1.0, Al 2.0, Fe 3.0. For statistical calculations, values <LOD were assigned LOD/2. Quality assurance via spike-recovery (77.8–107.7% across the panel) and two CRMs: ERM-CD281 (rye grass) and BCR No 279 (trace elements in sea lettuce Ulva lactuca); recoveries within ±20% for As, Cd, Cr, Cu, Mn, Ni, Pb, Zn.

Copper complexing capacity (L_T) and apparent stability constant (log K_app) were determined by DPASV on a µAutolab type III (Eco-Chemie, Utrecht, Netherlands) with a three-electrode cell (663 VA Stand, Metrohm) using a static mercury drop electrode (SMDE) as working electrode, Ag/AgCl (3 M KCl) reference, and carbon rod auxiliary. Samples diluted 10× in Milli-Q water with 3 M NaCl supporting electrolyte, then standard-additions Cu titration; 15-min equilibration between additions. L_T and K_app extracted via the Ružić–van den Berg linear transformation assuming 1:1 metal:ligand complexation. Repeatability (n = 5) <10%.

Organic carbon determined by high-temperature catalytic oxidation (Shimadzu TOC-5000A). pH measured with a Jenway 3310 meter.

Statistical analysis: non-parametric Mann–Whitney U (two-group comparisons) and Kruskal–Wallis (six-subgroup comparisons) because Kolmogorov–Smirnov and Shapiro–Wilk rejected normality at p < 0.05. Principal component analysis was performed separately for common and balsamic subsets after exclusion of elements with measure-of-sampling-adequacy <0.5; KMO statistic 0.86 (common) and 0.85 (balsamic). Software: PASW Statistics v24 (SPSS Inc, Chicago, IL).

Limitations relevant to HMTc analyte coverage: total arsenic only (no iAs speciation); total chromium only (no Cr-VI speciation); mercury not measured (no cold-vapor or atomic fluorescence accessory described); tin and antimony not measured.

Implications

Certification: This is an A-tier source for Pb, total As, Cd, Al, and total Cr occurrence in Greek retail vinegars across both common (wine, fruit) and balsamic subcategories. The balsamic subgroup is consistently the higher-concentration subgroup for Pb, Al, Cr, Fe, Mn, Rb, Co, Sr, and Cs; common vinegars are slightly higher for Cd. The Pb data establish that 5% of the Greek balsamic sample set crossed a California-state intake-equivalent threshold of 34 µg/L. The paper does not measure inorganic As, MeHg, Cr-VI, Ni speciation, Sn, or Sb — gaps for HMTc-panel coverage in vinegar that other sources must fill.

Courses: Useful illustration of how organic-matter chemistry shapes metal bioaccessibility in acidic food matrices. The L_T/OC density comparison across beverage categories (vinegar, herbal infusion, coffee, beer) is a self-contained teaching example. Demonstrates that balsamic-style concentration during production transfers metals from grape juice into the finished product at multiples not seen in wine vinegars.

App: For Greek retail vinegars in this sample set — common vinegars: Pb median 6 µg/L, range <0.05–63; total As median 2.6 µg/L, range <0.05–26; Cd median 0.30 µg/L, range 0.03–1.8. Balsamic vinegars: Pb median 9.8 µg/L, range <0.05–86; total As median 4.2 µg/L, range <0.05–26; Cd median 0.10 µg/L, range 0.03–2.3. All values are for the as-sold liquid product.

Microbiome: Not applicable.

Verification notes

Page was originally drafted under the legacy manual-fetch-kimi handle and last updated 2026-05-14. Merge-enhanced on 2026-05-26 against the source PDF to:

• Replace legacy raw_handle: manual-fetch-kimi with the canonical MFK_copper-complexing-capacity-and-trace-metal-content handle.
• Correct the truncated raw_path (prior path was missing the space after “Kimi” and truncated the filename mid-word).
• Change As → tAs in metals: because the paper reports only total As; iAs is not speciated (CLAUDE.md Part 14).
• Expand metals: to all 16 elements the paper actually measures by ICP-MS, for documentary fidelity. The routing layer maps only HMTc-relevant abbreviations (tAs, Pb, Cd, Al, Cr, Ni, Cu) into product-row rows; the remainder (Fe, Mn, Zn, Co, Ba, Cs, Rb, Sr, V) are recorded for the page’s documentary completeness, consistent with the convention used on the sibling vinegar paper ozbek2016-mip-oes-turkish-vinegars.
• Simplify matrices: from the prior over-granular [balsamic-red-vinegar, balsamic-white-vinegar, wine-red-vinegar, wine-white-vinegar, apple-cider-vinegar, fruit-vinegar] to the broadest accurate set [balsamic-vinegar, wine-vinegar, apple-cider-vinegar, fruit-vinegar] per the broad-scope-frontmatter rule in docs/gpt-collaboration/system-prompt.md. Subgroup detail (BR, BRH, BW, WR, WW, F) remains in the body, which is where it belongs.
• Add wine-vinegar and apple-cider-vinegar to ingredients: (the WR and WW samples are wine vinegars; the apple-fruit samples in F are apple-cider vinegars).
• Correct the As toxicology comparison in Key numbers: the prior version wrote “BMDL0.5 of 3 µg/day,” which conflated two distinct quantities and used the wrong abbreviation. The source compares the 0.40 µg/day exposure from a 15 mL vinegar serving to the JECFA-withdrawn PTDI for inorganic As of 126 µg/day (the “≈300× lower” comparison the paper makes), with the BMRL0.5 of 180 µg/day cited separately as the toxicological basis for withdrawing the PTDI. Corrected wording and values per source page 8.
• Correct the Cd PTDI wording: the prior version wrote “PTDI of 3.3 mg/L for 60 kg person,” which conflated the daily intake (50 µg/day) with the concentration equivalent (3.3 mg/L at 15 mL serving). Rephrased to make both quantities explicit per source page 9.
• Replace the imprecise “California Prop 65 threshold of 34 µg/L for balsamic vinegar” with the source’s actual claim: California-state maximum allowable Pb intake of 0.5 µg/day, which translates to 34 µg/L vinegar at a 15 mL serving. The source does not say “Proposition 65” specifically (the 0.5 µg/day figure is the Prop 65 MADL for Pb, but the paper does not invoke Prop 65 by name), and the 34 µg/L is not balsamic-specific (it is a general vinegar threshold; both samples that exceeded it happened to be balsamic).
• Add the organic-carbon and pH subgroup data (OC range 17–278 mg/L; balsamic median 152 vs. common 30 mg/L; pH ranges) that the source treats as load-bearing context for the L_T results but the prior version omitted.
• Add the total Cu range (<0.01–4.0 µM) which was missing from Key numbers; the prior version conflated total Cu with L_T (complexing capacity).
• Expand the “other elements with significant balsamic > common differences” block to report Fe, Mn, Rb, Co, Sr, Cs medians from source page 9; the prior version named only Al, Cr, and Ni.
• Add the L_T/OC density inter-beverage comparison (red-grape balsamic 122 nmol Cu mg⁻¹ C vs. herbal infusions 45, coffee brews 34, beers 7.3), which is one of the paper’s headline findings and was absent.
• Remove the legacy ## Wiki pages updated on ingest heading and replace with the current ## Verification notes structure.
• License field corrected from “CC BY” to the source-document’s stated “CC BY 4.0”.

Auto-audit subagent (2026-05-26) verdict REVISE; Checks 1, 3, 4, 5 all clean. The sole ⚠️ flagged fruit-vinegar in matrices: as not present in the taxonomy snapshot. Verified against the corpus: the matrices vocabulary is documented in docs/gpt-collaboration/system-prompt.md as “bare strings; broad scope” with the rule that the snapshot’s “common matrices” list is illustrative and category-specific extensions are accepted (the routing layer treats matrices as a free-text controlled-by-convention field, not a closed enumeration). The corpus already contains ~18 distinct vinegar-prefixed matrices slugs across other source pages (e.g., apple-vinegar, grape-vinegar, pomegranate-vinegar, fig-vinegar, homemade-vinegar, date-vinegar, sherry-vinegar, malt-vinegar, balsamic-reduction, balsamic-glaze, etc.). fruit-vinegar is the source paper’s own subgroup label (the “F” subset comprising 3 apple + 1 pomegranate + 1 sea buckthorn) and the routing audit accepts the source with no routing_malformed or routing_unresolved entry. Finding documented as a false positive; no page change applied.

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.