Ndung’u et al. 2004 — Lead in vinegar by ICP-MS and GFAAS: sample-preparation evaluation

This study measured lead in 59 commercial vinegars purchased from California retail using both inductively coupled plasma mass spectrometry (ICP-MS) and graphite furnace atomic absorption spectrometry (GFAAS), with the primary objective of identifying the most reliable sample-preparation protocol. Direct analysis without digestion produced erroneously high and poorly reproducible results on both instruments because of organic-matrix interferences; nitric acid digestion brought the two instrumental methods into close agreement (R = 0.997, slope 0.94), and nitric acid digestion with UV photolysis was identified as the fastest and most complete preparation when hydrogen-peroxide reagent-blank Pb is acceptable. Balsamic vinegars — particularly traditionally barrel-aged varieties — carried substantially higher lead than other vinegar types, with the highest single sample at 307 µg/L and the lead source attributed to production and storage processes rather than the agricultural inputs.

Key numbers

All concentrations are reported as µg/L of as-sold vinegar (= ppb for the aqueous matrix). Means and ranges below come from Table 4 (p. 4) unless otherwise noted; the all-vinegar summary statistic is reported on the same table’s bottom row.

Balsamic vinegar (n = 52), Table 4 sub-groups:

• Younger balsamic (n = 36): range 15–68 µg/L, mean 44 µg/L, SD 15.
• Mid-tier balsamic (n = 12): range 73–110 µg/L, mean 85 µg/L, SD 12.
• High-tier balsamic (n = 2): range 174–179 µg/L, mean 173 µg/L, SD 4.
• Aged balsamic-1 (n = 1): 276 µg/L.
• Aged balsamic-2 (n = 1): 307 µg/L.

Aggregate balsamic statistic (text, p. 5): range 14.9–307 µg/L, mean 68 ± 56 µg/L. The two aged-balsamic single-sample values dominate the high end; the highest (307 ± 19.5 µg/L, n = 6 replicate digestions) was reportedly produced by traditional Aceto Balsamico di Modena methods involving 25-year storage in vintage wood barrels. The text additionally notes a balsamic at 257 µg/L “aged for 18 years” (p. 5); this 257 µg/L value is not the Table 4 aged-1 row (276 µg/L), and the paper does not reconcile the two figures — both are reported here as the source presents them.

Other vinegar types (Table 4 single-sample or small-n rows):

• Wine vinegar (n = 4): range 36–62 µg/L, mean 50 µg/L, SD 12.
• Garlic vinegar (n = 1): 15 µg/L.
• Apple cider vinegar (n = 1): 6.6 µg/L.
• Rice vinegar (n = 1): 19 µg/L.

All vinegars combined (n = 59): range 7–307 µg/L, mean 64 ± 54 µg/L.

Method performance:

• GFAAS method detection limit (after nitric acid + heating-block digestion): 0.12 µg/L (mean blank 0.03 µg/L, n = 4; SD 0.04).
• Spike recovery on 6 different vinegars: 96 ± 5 %.
• NIST 1640 SRM (trace metals in natural waters, certified 27.89 ± 0.14 µg/L Pb) recovery: 97.4 ± 1.3 %.
• Relative standard deviation for duplicate analyses: < 8 %.
• Cross-instrument agreement after nitric acid digestion: R = 0.997, slope m = 0.94 (simple linear regression of GFAAS vs. ICP-MS on the same digests).
• Reagent contamination: TMG (trace-metal-grade) hydrogen peroxide carried ~15 µg/L Pb, comparable to the lead concentration in the lower-end vinegars and limiting UV photolysis with H₂O₂ for samples below ~50 µg/L Pb.

Effect of digestion on apparent Pb (Table 3, p. 3): four balsamic vinegars analyzed both ways. Direct dilution gave significantly higher and less precise values than nitric acid digestion on both instruments (paired t-test, P ≤ 0.05). For example, Balsamic-1: GFAAS direct dilution 595 µg/L (RSD 18 %) vs. acid-digested 319 µg/L (RSD 9 %); ICP-MS direct dilution 447 µg/L (RSD 7 %) vs. acid-digested 306 µg/L (RSD 6 %).

Methods (brief)

ICP-MS: ThermoFinnigan Element magnetic-sector high-resolution ICP-MS, Glass Expansion Conikal nebulizer, Scott-type double-pass spray chamber (cooled to 10 °C), standard nickel cones. ²⁰⁸Pb measured at low resolution (r = 300) with ²⁰⁹Bi as internal standard; RF power 1250 W; 200 scans per acquisition (Table 1, p. 2).

GFAAS: Perkin-Elmer SIMAA 6000 with Zeeman background correction and AS72 auto-sampler. End-capped, transversely heated pyrocoated graphite tubes with integrated L’vov platform. Lead electrodeless discharge lamp at 283.3 nm, lamp current 450 mA. Chemical modifier: 0.05 mg NH₄H₂PO₄ + 0.003 mg Mg(NO₃)₂ per 5 µL. The Perkin-Elmer manufacturer maximum ashing/atomization is 400/1400 °C (§4.3, p. 4); with the chemical modifier, the authors initially optimized at 800/1400 °C against NIST 1640 (digested natural-water SRM), but that program produced low recoveries on spiked vinegar digests because of a sharp absorbance drop between 700 and 800 °C in vinegar digests that was not seen in the SRM (Fig. 1, p. 5). The authors attribute this to relatively labile, volatile organolead compound(s) in vinegar digests and recommend a lowered program of 600 °C ashing / 1300 °C atomization specifically for vinegar (Conclusions, p. 6). Table 2 (p. 3) tabulates the working GFAAS program as 700 / 1400 °C; this is an intermediate working setting rather than the paper’s final published recommendation, and the paper does not reconcile the two.

Three sample-preparation paths compared:

1. Simple aqueous dilution (no digestion): produced poor precision and erroneously high values on both instruments; not recommended.
2. Nitric acid heating-block digestion: 0.5–1.0 g vinegar weighed into Teflon vessels, 10 mL TMG nitric acid, initial 50 °C for 2–3 h, then 90 °C to dryness, reconstituted in 1 M TMG nitric acid. Effective, clean, but slow.
3. Nitric acid + 30 % H₂O₂ UV photolysis: 0.5 g vinegar, 1 mL TMG nitric acid, 0.5 mL 30 % H₂O₂, custom PTFE digestion cups with quartz caps under a 1200 W medium-pressure Hg vapor lamp (9.2 ± 0.4 mW cm⁻² monitored throughout). Fastest and most complete; UV photolysis works for vinegar but the H₂O₂ reagent blank (~15 µg/L Pb) limits applicability to vinegars > 50 µg/L Pb. Cleaner H₂O₂ would be needed for low-concentration samples.

Recommended protocol: nitric acid digestion with UV photolysis where matrix-Pb permits; nitric acid heating-block digestion otherwise. ICP-MS or GFAAS detection; for GFAAS, use 600 °C ash / 1300 °C atomization rather than manufacturer defaults.

Contamination control: HEPA-filtered Class 100 laboratory; Teflon vessels cleaned in 8 M TMG HCl overnight then ≥ 8 h hot TMG HNO₃; all plastic ware double-bagged in trace-metal-clean Zip-loc bags. Sample batches included three method blanks, three spiked analytical samples, and three reference materials.

Implications

Certification: Establishes that balsamic vinegar — and particularly traditional barrel-aged balsamic — sits at the high end of the vinegar-category Pb distribution, with single-sample values reaching 307 µg/L. The Pb source is attributed to vinification contact materials (wood-barrel metal fittings, alloy welds in production equipment) by analogy to wine-vinification studies, not to the grape or the soil. This supports treating balsamic vinegar as a distinct sub-row from other vinegar types in any HMTc certification scheme. The non-balsamic vinegars in this dataset (wine, apple cider, garlic, rice) sit well below 100 µg/L Pb, with the apple cider, garlic, and rice single-sample values all under 20 µg/L.

Courses: Useful teaching case for analytical method validation in acidic food matrices. The organolead volatility finding (sharp absorbance drop 700–800 °C in vinegar digests but not in the NIST 1640 water SRM) is a self-contained illustration of why matrix-matched method optimization matters even when an SRM appears to validate the method. Also illustrates the trade-off between UV photolysis speed and reagent-blank contamination limits.

App: For US-retail vinegar in this sample set — balsamic Pb mean 68 ± 56 µg/L with extreme aged samples to 307 µg/L; wine vinegar Pb mean 50 µg/L; apple cider, garlic, rice single-sample Pb all under 20 µg/L. Vinegar serving sizes are small (typically 5–15 mL for dressing use), so even the 307 µg/L extreme corresponds to ≤ 4.6 µg Pb per serving; daily-intake context for app users should account for typical low-volume use rather than treating vinegar concentrations as drink-equivalent.

Microbiome: Not applicable; the paper does not address gut-microbial endpoints.

Verification notes

Page was originally drafted under the legacy manual-fetch-kimi raw_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_determination-of-lead-in-vinegar-by-icp-ms-and-gfa handle, matching the convention used on sibling vinegar source karavoltsos2020-copper-trace-metals-vinegars-greece.
• Correct truncated raw_path (prior value cut off at “evaluation .pdf”; corrected to the full filename “…evaluation of different sample preparation pro.pdf”).
• Correct three transposed single-sample Pb values in Key numbers. Prior version reported Garlic vinegar (n=1): 276 µg/L, Apple cider vinegar (n=1): 15 µg/L, and Rice vinegar (n=1): 6.6 µg/L. Source Table 4 (p. 4) actually shows Garlic = 15 µg/L, Apple = 6.6 µg/L, Rice = 19 µg/L. The prior page had pulled the wrong column or scrambled rows; 276 is the Table 4 Balsamic (aged)-1 row, which was separately missing from the page. All four values corrected.
• Correct misread n in Key numbers. Prior version reported Balsamic group 3 (n=4, older/traditional): range 174–179 µg/L, mean 173 µg/L. The “4” was the standard-deviation column in Table 4, not the sample count. Correct n is 2.
• Add the two aged-balsamic single-sample rows (276 and 307 µg/L) that the prior version omitted, and note the paper-internal discrepancy between Table 4’s aged-1 = 276 µg/L and the text’s “another balsamic … 257 µg/L … aged for 18 years” without inventing a reconciliation.
• Expand Table 3 cross-method comparison (direct dilution vs. nitric acid digested, for four balsamics on both instruments) in Key numbers; prior version omitted the magnitude of the dilution-vs.-digestion bias and only mentioned it qualitatively.
• Expand the Methods section to capture the instrumental detail (RF power, isotopes monitored, internal standard, modifier composition, contamination control) that supports the A-tier classification.
• Add garlic-vinegar to matrices: (the source measured one); prior version omitted it.
• Remove the legacy ## Wiki pages updated on ingest heading per the current Part 6 template; downstream-page list lives in the routing audit, not on the source page.
• Restate the µg/L = ppb equivalence explicitly in Key numbers for the aqueous matrix per CLAUDE.md Part 14.
• License kept as “All rights reserved” — Talanta (Elsevier) 2004 paywall paper, no open-access notice on the PDF.

Auto-audit subagent (2026-05-26) verdict REVISE; Checks 1, 3, 4, 5 clean. Check 2 ⚠️ flagged the matrices: / ingredients: asymmetry — matrices contains rice-vinegar and garlic-vinegar (which the paper measures as n = 1 each) while ingredients: does not list them. Verified against the source: 1 rice and 1 garlic vinegar are reported in §3.3 and Table 4. Verified against the corpus: no wiki/ingredients/rice-vinegar.md or wiki/ingredients/garlic-vinegar.md page exists, and Part 10’s ingredient-promotion threshold is 5 papers treating the sub-variant as a distinct commodity. With n = 1 each in this paper alone, neither slug crosses the threshold for an ingredient page. Resolution applied:

• Keep rice-vinegar and garlic-vinegar in matrices: (descriptive of what was measured).
• Keep them out of ingredients: (no extant page to route to; both are folded under the generic vinegar coverage).
• Routing audit confirms 0 unresolved / 0 malformed for this source after the change, so the asymmetry does not produce broken fan-out.

This is the same broad-scope-matrices, narrow-ingredients pattern documented on the sibling vinegar source karavoltsos2020-copper-trace-metals-vinegars-greece for its fruit-vinegar matrix.

Self-caught during audit-application (not flagged by the subagent): the Methods section initially misattributed the GFAAS recommended program. Corrected to attribute 600 / 1300 °C to the paper’s Conclusions section (p. 6) rather than Table 2 (Table 2 shows 700 / 1400, an intermediate working program). Also corrected the manufacturer’s recommended maximum from “800 / 1400” to “400 / 1400” per §4.3 (p. 4); the 800 / 1400 figure was the authors’ first-pass optimization against the NIST 1640 SRM, not the manufacturer default.

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.