Đurđić et al. 2021 — Lead content and lead isotope ratios in Serbian red wines

Đurđić and colleagues measured Pb content and the full lead isotope ratio (LIR) pattern (²⁰⁶Pb/²⁰⁷Pb, ²⁰⁸Pb/²⁰⁶Pb, ²⁰⁶Pb/²⁰⁴Pb, ²⁰⁷Pb/²⁰⁴Pb, ²⁰⁸Pb/²⁰⁴Pb) in 59 Serbian red wines from four regions (Vojvodina, Belgrade, Central Serbia, South Serbia) by inductively coupled plasma quadrupole mass spectrometry (ICP-QMS), with thallium isotopes ²⁰³Tl/²⁰⁵Tl (NIST SRM 997) as internal standard and NIST SRM 981 for external mass-bias correction. Per-region mean Pb content ranged from 44.4 ± 29.6 µg/L (Belgrade) to 65.0 ± 32.6 µg/L (Vojvodina); the single-sample range across all 59 wines was 11.50 ± 0.05 µg/L (Central Serbia) to 126.10 ± 0.09 µg/L (South Serbia). All wines fell below the International Code of Oenological Practices (OIV, 2019) maximum of 150 µg/L Pb for wines produced from 2007 onward. Principal component analysis on the four radiogenic LIRs showed complete overlap of the four regions on the three-component score plot, so LIR pattern alone was not sufficient to discriminate Serbian wines by geographic origin in this dataset. The authors attribute the lack of discrimination to multiple anthropogenic Pb inputs (leaded gasoline residual signal, Fe-Mn metallurgy, Pb smelter emissions) overlying the geogenic signal and to the vinification process itself contributing Pb of variable isotopic composition.

Key numbers

All Pb-content values are average ± standard deviation in µg/L unless noted. Min/Max are single-sample extremes within the region. Sample sizes are n=12 (Vojvodina), n=22 (Belgrade), n=14 (Central Serbia), n=11 (South Serbia); total n=59 red wine samples. Per-sample analytical relative standard deviations (RSDs) for the isotope ratios are reported separately under Methods: 0.10–0.15% across analyzed isotopes, with 0.32% for ²⁰⁴Pb (the least abundant Pb isotope). Pb content was measured at 1:10 dilution; LIRs at 1:10 or 1:25 dilution depending on Pb concentration. Tl internal standard 2 µg/L; HNO₃ matrix 2% v/v.

Pb content per region (Table 2; µg/L):

• Vojvodina (n=12): mean 65.0 ± 32.6; min 14.9; max 117.6; median 56.4
• Belgrade (n=22): mean 44.4 ± 29.6; min 12.6; max 100.3; median 26.3
• Central Serbia (n=14): mean 48.8 ± 35.1; min 11.5; max 114.5; median 39.6
• South Serbia (n=11): mean 58.0 ± 44.8; min 14.5; max 126.1; median 47.4

The text reports the single overall lowest as 11.50 ± 0.05 µg/L (Central Serbia) and the single overall highest as 126.10 ± 0.09 µg/L (South Serbia). Regional rank by mean Pb content was Vojvodina > South Serbia > Central Serbia > Belgrade; by median Pb content was Vojvodina > South Serbia > Central Serbia > Belgrade. A Mann–Whitney U / Kruskal–Wallis non-parametric test was performed but the paper states “this non-parametric test demonstrates if there are significant differences between the regions” without reporting per-pair p-values in the printed Results.

Regulatory comparison (OIV International Code of Oenological Practices, 2019):

• OIV maximum Pb content in wines produced since 2007: 150 µg/L
• All 59 Serbian samples (range 11.5 – 126.1 µg/L) were below the OIV ceiling
• The single highest individual sample (126.1 µg/L, South Serbia) reached 84% of the OIV ceiling

Lead isotope ratios per region (Table 2; average ± SD between samples):

• Vojvodina: ²⁰⁶Pb/²⁰⁴Pb 18.3 ± 0.2 (17.9–18.5); ²⁰⁸Pb/²⁰⁶Pb 2.13 ± 0.03 (2.10–2.16); ²⁰⁷Pb/²⁰⁴Pb 16.0 ± 0.2 (15.7–16.3); ²⁰⁸Pb/²⁰⁴Pb 38.9 ± 0.5 (38.1–39.6); ²⁰⁶Pb/²⁰⁷Pb 1.145 ± 0.002 (1.111–1.168)
• Belgrade: ²⁰⁶Pb/²⁰⁴Pb 18.3 ± 0.4 (17.6–18.6); ²⁰⁸Pb/²⁰⁶Pb 2.14 ± 0.03 (2.10–2.17); ²⁰⁷Pb/²⁰⁴Pb 16.0 ± 0.4 (15.6–16.2); ²⁰⁸Pb/²⁰⁴Pb 38.7 ± 0.7 (37.4–39.8); ²⁰⁶Pb/²⁰⁷Pb 1.141 ± 0.003 (1.112–1.170)
• Central Serbia: ²⁰⁶Pb/²⁰⁴Pb 18.3 ± 0.4 (17.5–18.6); ²⁰⁸Pb/²⁰⁶Pb 2.14 ± 0.03 (2.10–2.17); ²⁰⁷Pb/²⁰⁴Pb 16.0 ± 0.3 (15.6–16.3); ²⁰⁸Pb/²⁰⁴Pb 39.0 ± 0.7 (37.6–39.9); ²⁰⁶Pb/²⁰⁷Pb 1.145 ± 0.002 (1.112–1.171)
• South Serbia: ²⁰⁶Pb/²⁰⁴Pb 18.3 ± 0.3 (17.9–18.6); ²⁰⁸Pb/²⁰⁶Pb 2.14 ± 0.02 (2.10–2.17); ²⁰⁷Pb/²⁰⁴Pb 16.0 ± 0.3 (15.6–16.2); ²⁰⁸Pb/²⁰⁴Pb 39.1 ± 0.5 (38.0–39.9); ²⁰⁶Pb/²⁰⁷Pb 1.145 ± 0.003 (1.116–1.171)

Whole-corpus ranges from the Discussion: ²⁰⁸Pb/²⁰⁶Pb 2.0760 – 2.1758; ²⁰⁶Pb/²⁰⁷Pb 1.1109 – 1.1898; ²⁰⁶Pb/²⁰⁴Pb 17.53 – 18.67. These overlap published ranges for European red wines and fall within wide ranges reported for Australian red wines (Kristensen et al. 2016: ²⁰⁸Pb/²⁰⁶Pb 2.0753 – 2.2013; ²⁰⁶Pb/²⁰⁴Pb 14.45 – 19.20). The Serbian ²⁰⁶Pb/²⁰⁷Pb range is slightly higher than the Australian range (1.0700 – 1.6900) reported by Kristensen.

Method validation against NIST SRM 981 (Table 1; n=6 replicates):

• ²⁰⁷Pb/²⁰⁶Pb: certified 0.91464 ± 0.00033; external (NIST 981) correction 0.9146 ± 0.0008; internal (NIST 997 Tl) correction 0.9191 ± 0.0007
• ²⁰⁸Pb/²⁰⁶Pb: certified 2.1681 ± 0.0008; external 2.1679 ± 0.0009; internal 2.1449 ± 0.0008
• ²⁰⁴Pb/²⁰⁶Pb: certified 0.059042 ± 0.000037; external 0.058 ± 0.003; internal 0.057 ± 0.002

External (NIST SRM 981 bracketing) standardization showed the best agreement with certified values and was applied to all wine samples; internal Tl standardization showed systematic bias relative to certified ²⁰⁷Pb/²⁰⁶Pb and ²⁰⁸Pb/²⁰⁶Pb in this matrix.

PCA discrimination outcome (Figure 4): Three principal components (PC1 66.46%, PC2 24.90%, PC3 8.65%; cumulative 100%) on the four LIRs (²⁰⁸Pb/²⁰⁶Pb, ²⁰⁶Pb/²⁰⁷Pb, ²⁰⁶Pb/²⁰⁴Pb, ²⁰⁷Pb/²⁰⁴Pb) yielded complete overlap of all four Serbian regions on the score plot. Loading plot showed grouping of the four LIRs but no region-discriminating structure. The authors conclude that, in the presence of multiple overlapping Pb sources, single-region pollution would have to be the dominant influence for LIRs alone to enable origin assignment.

Anthropogenic versus geogenic mixing (Figure 6): A three-isotope ²⁰⁸Pb/²⁰⁶Pb vs ²⁰⁶Pb/²⁰⁷Pb mixing line between published geogenic Pb (Teutsch et al. 2001) and European petrol (Ettler et al. 2004; Monna et al. 1995) ran close to the Serbian wine averages; the four regional averages cluster near the midpoint of the geogenic↔petrol line, consistent with a mixed natural-plus-anthropogenic isotopic signature. Pearson correlation between ²⁰⁶Pb/²⁰⁷Pb and 1/C_Pb across the four regions ranged from r = 0.05 (South Serbia) to r = 0.18 (Central Serbia) — none statistically significant — indicating multiple coexisting Pb sources rather than a single dominant pollutant.

Methods (brief)

Site and samples: 59 red wines from four Serbian administrative regions (Vojvodina, Belgrade, Central Serbia, South Serbia); all wines in glass bottles with cork stopper. No vintage, producer or grape-cultivar metadata are provided.

Instrumentation: ICP-MS iCAP Q (Thermo Scientific Xseries 2, United Kingdom) operating as a quadrupole mass spectrometer (ICP-QMS), equipped with flat pole collision reaction cell, micro-concentric nebulizer, nickel cones and peristaltic sample delivery pump, running in quantitative analysis mode under Qtegra Instrument Control Software. A cooled spray chamber was used to minimize ethanol-matrix effects. Detector run in pulse mode; working range up to 40 µg/L Pb.

Reagents and reference materials: NIST SRM 981 Common Lead Isotopic standard (NIST, USA) for mass-bias correction by external standardization; NIST SRM 997 Thallium isotopic standard for internal standardization (²⁰³Tl/²⁰⁵Tl). Mercury stock solution 1000 ± 4 µg/mL (TraceCERT, Fluka, Dorset, UK) used to study ²⁰⁴Hg isobaric interference on ²⁰⁴Pb. All chemicals analytical grade (Merck, Darmstadt, Germany). Glassware soaked 12 h in 10% HNO₃, rinsed with ultra-pure water from a Millipore Simplicity 185 system with dual UV filters (185 and 254 nm).

Sample preparation: For Pb content, wines diluted 1:10 with ultra-pure water and analyzed directly; HNO₃ added to final 2% v/v. For LIR analysis, dilution factor 10× or 25× depending on Pb concentration. Tl added to 2 µg/L as internal standard. Blanks prepared with the same matrix (2% v/v HNO₃, 2 µg/L Tl); blank correction applied per sample.

Instrumental optimization: Standard solution of 5.12 µg/L Pb spiked with 2 µg/L Tl measured at six points per peak with five replicates and varied dwell times. Dwell times optimized for lowest RSD: 100 ms for ²⁰⁴Pb, 5 ms for ²⁰⁸Pb, ²⁰⁶Pb, ²⁰⁷Pb, ²⁰³Tl and ²⁰⁵Tl 25 ms. Dead time optimized to 40 ns by linear fitting of ²⁰⁴Pb/²⁰⁸Pb vs Pb concentration (10–30 µg/L). RSDs across analyzed isotopes 0.10–0.15%, with 0.32% for ²⁰⁴Pb. ²⁰⁴Hg isobaric interference correction applied on ²⁰⁴Pb measurements.

Quantification: Per-sample LIRs corrected for mass bias via NIST SRM 981 bracketing sequences (sample sandwiched between two runs of 5.12 µg/L Pb standard); linear potential-algorithm bias factor K applied. Final LIR values reported are post-correction.

Statistics: Descriptive statistics, correlation analysis and Mann–Whitney U test in NCSS (Hintze 2001). Principal component analysis (PCA) by singular value decomposition with 0.95 confidence level for Q and T2 Hotelling outlier limits; correlation matrix; components with eigenvalues > 1 retained. PCA implemented in PLS_Toolbox v6.2.1 for MATLAB 7.12.0 (R2011a).

Limitations explicitly visible in the source: (i) only Pb is measured — no other heavy metals (Cd, As, Hg, Ni, Al, Sn, Sb, Cr or Cr-VI) reported; (ii) all wines are red — white, rosé and dessert wines are absent; (iii) no vintage, producer or grape-cultivar information is reported, so vintage-effect and cultivar-effect confounders cannot be controlled; (iv) PCA showed complete overlap of regions, so the central hypothesis (LIRs as origin marker) was not supported in this dataset; (v) the paper does not state how the wines were sourced (purchased from retail, sampled at wineries, donated) or whether they were commercial finished wines or wineries’ tank samples; (vi) per-sample raw data are not tabulated, only per-region averages and ranges; (vii) the ICP-QMS approach is documented (per Vanhaecke 2009; Almeida and Vasconcelos 1999; Larcher 2003) as having lower precision for LIRs than TIMS or sector-field multicollector ICP-MS, which the authors acknowledge.

Evidence Fitness

This source contributes 59 red-wine Pb-content values from four Serbian regions, with documented ICP-QMS methodology, NIST SRM 981 bracketing for mass-bias correction, and full disclosure of the dilution, dwell-time and dead-time optimization. It is the first published LIR dataset for Serbian wines. Per-region means and ranges are tabulated; per-sample values are not. The single-analyte focus (Pb only) limits its contribution to multi-metal product-category synthesis, but it anchors a meaningful Serbian point on the European red-wine Pb-content distribution at means 44.4–65.0 µg/L (region) and an absolute range 11.5–126.1 µg/L (across the corpus). It can support discovery, geographic context for Serbian terroir under documented mixed anthropogenic Pb pressure (residual leaded-gasoline signal, Fe-Mn metallurgy, Pb smelter), a methods anchor (ICP-QMS LIR determination with NIST SRM 981 bracketing and NIST SRM 997 Tl internal standard), and an illustration that LIR alone is insufficient for origin discrimination when multiple anthropogenic sources mix. Reported public evidence label: Context only.

Implications

• Certification: contributes red-wine Pb occurrence points (region means 44.4–65.0 µg/L; full-corpus range 11.5–126.1 µg/L; n=59) and a 0.84-of-OIV-ceiling worst-case single sample for fermented-beverages-non-tea-based for Pb. All samples were below the OIV 150 µg/L Pb ceiling; the highest sample reached 84% of the ceiling. Sample size of 59 is large enough to anchor a Serbian-wine occurrence row but the per-sample values are not tabulated, which limits pooled-percentile work to per-region means and ranges.
• Courses: useful as a worked example of why a single isotopic marker is insufficient for origin discrimination when multiple anthropogenic Pb sources mix. Also illustrates the analytical discipline required for ICP-QMS LIR work — NIST SRM 981 bracketing, ²⁰⁴Hg interference correction, dead-time optimization, and the distinction between external (NIST 981) and internal (NIST 997 Tl) mass-bias correction performance.
• App: contributes Pb occurrence range for Serbian red wine (region means 44.4–65.0 µg/L; corpus 11.5–126.1 µg/L; n=59).

Provenance notes

Open-access article published 4 October 2021 in Frontiers in Chemistry (vol 9, article 746695) under the Creative Commons Attribution License (CC BY). Funded by the Ministry of Education, Science and Technological Development of the Republic of Serbia (contracts 451-03-9/2021-14/200168 and 451-03-68/2020-14/200161). Authors declare no commercial or financial conflicts. Accessed via the Manual Fetch Discovery autopilot.

Evidence tier set to B. The journal is peer-reviewed and the methods section is unusually transparent on instrumental optimization (dwell time, dead time, per-isotope RSD, ²⁰⁴Hg interference correction, NIST SRM 981 bracketing). What disqualifies it from A on the HMI scale is the single-analyte (Pb-only) scope, the absence of vintage / cultivar / producer metadata, the non-tabulation of per-sample values (only per-region averages and ranges are printed), the unsourced provenance of the wine samples themselves, and the authors’ own acknowledgement that ICP-QMS LIR precision is inherently lower than TIMS or sector-field multicollector ICP-MS for origin assignment. The negative central result (PCA failed to discriminate regions) is itself defensibly reported.

Wiki pages this source may touch

• lead
• grapes
• fermented-beverages-non-tea-based

Verification notes

No brand names appear in the paper or on this page. The paper does not name any commercial wine label, producer, vineyard or grape cultivar; selection criterion was Serbian geographic origin across four administrative regions. The four region names (Vojvodina, Belgrade, Central Serbia, South Serbia) are Serbian administrative regions, not brand names. Part 12 brand-firewall does not apply.

Instrument and reagent references — Thermo Scientific Xseries 2 ICP-MS iCAP Q under Qtegra software; NIST SRM 981 and NIST SRM 997 reference materials; Millipore Simplicity 185 ultra-pure water system; Merck reagents; TraceCERT/Fluka Hg stock; NCSS statistical software (Hintze 2001); PLS_Toolbox / MATLAB — are reproduced under the methods-vendor exception (Part 12, Exception 2: scientific reproducibility requires naming the instrument doing the measuring, the reference material, the digestion equipment and the statistical software).

The metals frontmatter list contains only Pb, the single analyte measured. Thallium isotopes (²⁰³Tl, ²⁰⁵Tl) appear in the body only as the NIST SRM 997 internal standard for mass-bias correction; Tl is not a measured analyte in the wine samples and does not enter the metals frontmatter. Mercury appears only as ²⁰⁴Hg in the context of isobaric interference correction on the ²⁰⁴Pb measurement; Hg is not a measured analyte and does not enter the metals frontmatter.

Pb content values in Table 2 are printed in µg/L. The paper uses µg/L consistently across Table 2, the body text and the box-and-whisker plot (Figure 2). The OIV reference is 150 µg/L (OIV International Code of Oenological Practices, 2019, p. 5 of the source). The wet-weight basis is implicit (wine is a liquid measured as-is at 1:10 dilution); the source does not report dry-weight equivalents.

The four LIR ratios used in PCA are ²⁰⁸Pb/²⁰⁶Pb, ²⁰⁶Pb/²⁰⁷Pb, ²⁰⁶Pb/²⁰⁴Pb, and ²⁰⁷Pb/²⁰⁴Pb (the four radiogenic ratios). ²⁰⁸Pb/²⁰⁴Pb is reported in Table 2 but is not listed as one of the four PCA inputs in the source (page 5 text: “²⁰⁸Pb/²⁰⁴Pb, ²⁰⁷Pb/²⁰⁴Pb, and ²⁰⁶Pb/²⁰⁴Pb were subjected to” with ²⁰⁶Pb/²⁰⁷Pb added in the next sentence — total four ratios).

The figure-3 inset reports literature LIR ranges including “Romania/Moldavia, n=13 [3]” with ²⁰⁸Pb/²⁰⁶Pb 1.9961–2.0920 and ²⁰⁶Pb/²⁰⁷Pb 1.1540–1.2220, attributed by the authors to suspected coal-combustion influence. These are literature comparisons, not Serbian-wine measurements, and are summarized in the body but not extracted to Key numbers.

The Pearson r values in Figure 7A (²⁰⁶Pb/²⁰⁷Pb vs 1/C_Pb) are Belgrade r=0.15, Central Serbia r=0.18, Vojvodina r=0.17, South Serbia r=0.05; the body text page 8 prints “from 0.05 to 0.18” matching the Belgrade/Central Serbia/Vojvodina/South Serbia spread.

Provisional product scaffolds were not required: the source touches fermented-beverages-non-tea-based, which already exists in the taxonomy. No new ingredient pages are proposed; grapes is the only ingredient slug and already exists.

Audit subagent (2026-06-06) flagged a Check 5 ⚠️ in the original Implications/Certification bullet: a wiki-side cross-source comparison to bora2015-vineyard-wine-heavy-metals-romania (“wider margin to OIV than the bora2015 micro-vinified wines”) that the Đurđić paper itself does not make. Verified against the source — the paper compares the Serbian wines to European/Australian/Romanian-Moldavian wines from the LIR literature only, not to the Bora 2015 Cu/Pb wine values. The cross-source synthesis was added by Claude during drafting and violates Part 2 (synthesis is a separate Part 9 workflow, not a per-source contribution). Comparison stripped; the within-paper OIV-ratio framing retained. All other audit checks ✅ clean (Checks 1, 2, 3, 4); Check 5 has no other concerns.

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.