Tags

Gas ChromatographyFlavor ChemistrySensory and Flavor

Legacy Single Excerpt

Legacy Associated Permanent Note

Content

Citation(s) & Footnote(s)

Book(s)

Whiskey Knowledge Databases › Literature Notes

Scope and core summary

Full review of the 19-page article and seven-page supplement: a model-based extraction comparison and demonstration on three commercial products. This is not comprehensive quantitative, sensory or classification validation.

Author argument

Barnes and colleagues argue that a model-spirit matrix can be used to compare extraction techniques and optimize a practical headspace method; under their conditions HS-SPME offered the best compromise of efficiency, time, cost, and environmental burden and differentiated commercial spirit categories.

Researcher synthesis

For American whiskey research, the paper clarifies what an accessible analytical workflow can and cannot do. It can profile and compare volatile patterns or batches; it does not by itself identify sensory causation, quality, or consumer preference.

Evidence assessment

Primary analytical-method study with model-mixture optimization and application to commercial samples. Strong for method selection and operating parameters.

Limitations and open questions

Limitations: Only a small commercial sample set was used; extraction conditions bias what is recovered; identified compounds are not automatically aroma-active; the paper is not a guide to blend formulation.

Open questions: Can this workflow be adapted to consumer-scale blend trials through a partner laboratory while retaining sensory-panel correlation?

Connected records

• Contributors: 8
• Verified Excerpts: 3
• Citations: 1
• Zettels: 1 existing linked record (relation preserved)

Completion record

September 27 coverage audit supersedes the earlier extraction-based record: all 19 main pages and all seven recovered supplementary pages read, with table/figure inspection and a detailed critical review appended below.

Full-source review — Barnes et al. (2022)

Coverage and identity

Read the complete 19-page Foods 11, 3358 article, including methods, all four tables (every Table 4 row and footnote), both figures and every Figure 2 panel, conclusions, funding/conflict statements and 44 reference entries. Visually inspected main PDF pages 5, 7–10 and 12–16. Recovered the official seven-page supplementary PDF through Europe PMC PMC9656916 and read every page: Tables S1–S5, Figures S1–S3 and all 35 reference entries; visually inspected supplementary pages 1–4. Reference entries were read; this does not claim that every cited publication was independently reviewed. Main mirror original preserved; Proton filename/size match remains provisional until its bytes can be hashed.

Contribution

This is a useful method-selection study, not a sensory quality trial or validated brand-authentication service. A 24-compound model allows five extraction approaches to be compared before applying a selected HS-SPME method to one gin, one rum and one whiskey. The whiskey is Glenfiddich 14 Bourbon Barrel Reserve: Scotch matured in bourbon barrels, not American bourbon. Samples are named historical study products; repeated injections do not supply independent brand or production replication.

Pages 1–6: design and representativeness

The model draws on 13 publications and 87 commercial samples, with approximately 300 compounds across 15 classes. Individual model concentrations derive from differing numbers of reported values; selecting minima/maxima and excluding nondetects does not produce a population-weighted typical spirit. The model pH of 7.9 differs from gin 7.1, whiskey 4.3 and rum 4.1. Acidification improved recovery of some acidic/polar targets, but the authors avoided it because of possible hydrolysis. That choice limits transfer from model to real matrices.

Table 1 includes thresholds measured in different solvents despite its water heading; diethyl succinate, triethoxypropane and the two whiskey lactones have footnoted ethanol matrices. These are not measured odor-activity values for the tested spirits. Furfural/sotolon concentrations in the table span about 19,403-fold, slightly below the prose claim of more than 20,000.

The methods vary sample preparation, extraction media and sometimes chromatographic platform. Table 2 sample/solvent totals are not presented on an unambiguously consistent basis with the methods. Its qualitative quantitative-analysis ratings should not be treated as measured accuracy or precision; the footnote uses peak-area bins. Equipment cost is a historical 2022 estimate, not a current operating budget.

Pages 7–10: optimization and tradeoffs

LLE detected 23 of the 24 model compounds; SPE, SBSE and HSSE detected 22, while SPME detected 21. SPME was selected as an operational compromise involving sample volume, solvents, handling and equipment, not because it detected the most compounds. The SPE discussion's statement that only nerolidol was missing needs to be read against the 22/24 count and sotolon difficulty.

The selected DVB/CAR/PDMS headspace method uses diluted 10% ethanol, 10% NaCl, 35°C and 60 minutes. Figure 1 tests other temperatures around the chosen compromise; it does not independently show a 35°C bar. Figure 2 shows different time responses: furfural approaches a plateau, alpha-pinene declines with longer extraction, and heavier targets continue increasing. Sixty minutes is a compromise, not equilibrium for every analyte. Competition is an interpretation of the decline, not a separately established mechanism.

The reported 0.2-hour handling time does not mean a 12-minute complete analysis. Extraction alone is 60 minutes, and the main GC program is approximately 107.5 minutes before other instrument steps. The one-way ANOVA description lacks enough reported test details to convert every plotted difference into a demonstrated significance claim.

Pages 10–16: commercial results and numerical audit

Table 4 reports relative total-ion peak-area categories, not absolute concentrations, odor intensity, consumer liking or causal contributions. The authors report 215 detected compounds, 27 unidentified and 188 identified. An audit of the publisher XML reproduces 214 table rows, with one rum row combining two compounds. Counting that row as two explains the 215 total and rum's 59 rather than 58 rows. Direct row presence gives gin 130 and whiskey 75. The narrative listing 75 for rum and 59 for whiskey appears transposed relative to the table.

The row audit gives 120 gin-only rows, 33 whiskey-only rows, and 16 rum-only rows (17 compounds if the combined row counts twice), versus prose 121 and 18 for gin and rum. Thirty-nine rows are shared by rum and whiskey. Four rows occur in all three: ethyl hexanoate, ethyl hexadecanoate, nonanal and limonene. These internal discrepancies should travel with any reused count.

Library matching and retention indices support provisional compound assignments with unequal certainty. Some experimental/reference RI pairs differ substantially, for example geranyl acetate 1381/1282; no uniform acceptance cutoff resolves all such differences. Unknowns classified as terpenoids are not exact molecular identifications. An unresolved cis-whiskey-lactone peak is not proof of its absence.

The results cannot establish general category superiority or stronger analytical coverage than Zhao et al. merely by comparing compound counts across different samples and procedures. The three commercial samples supply an illustration, not an independently validated category classifier. No GC-olfactometry or sensory panel establishes which detected compounds drive the samples' aromas. Trace allyl isothiocyanate plus a historical sugarcane-preservation patent is not evidence of that rum producer's actual practices.

Supplement: all seven pages

Tables S1–S4 compile selected historical extraction conditions: 10 LLE entries, eight SPE entries, 14 SPME entries and three SBSE entries. They document methodological diversity rather than a systematic meta-analysis or contemporary laboratory recommendation. The SPME row citing Zhao reports 2% salt; the independently reviewed original describes 10%, another reason to consult the original before copying a protocol. Some reference entries repeat the same underlying study, so the 35 numbered entries are not 35 independent experiments.

Figures S1–S3 are liquid-injection TIC chromatograms, with 1 µL and 1:10 split for gin/whiskey versus splitless rum. They are not directly comparable measures of between-product concentration. Their x axes lack visible retention-time ticks and the y axes differ. Gin has a dense terpene-rich pattern; whiskey and rum have fewer dominant peaks. All five labeled peaks in each trace were checked against captions. The dominant liquid-injection peaks differ from the HS-SPME relative pattern, reinforcing extraction dependence rather than invalidating either automatically.

Table S5 supplies ion lists for 11 unclassified unknowns, including relative fragment abundances. These are identification leads, not established identities or sensory descriptions. All entries and abundances were read; no molecular structures were invented.

Critical assessment and funding

Strong for demonstrating extraction selectivity and designing a practical screening workflow. Limited for validated quantification, generalization to American whiskey, flavor causation, product authenticity and budgets. Three technical replicates do not replace independent samples. The model's pH, coelution and under-recovery of polar/furan/phenolic targets matter. Comte de Grasse supplied gin and has author/resource involvement; the article reports ANRT CIFRE 2019/0119 support and declares no conflict or funder influence. Record both facts without inferring misconduct.

Linked ideas and proposed Academy applications

1. A laboratory request should name the decision first: broad screening, a target concentration, a suspected defect or sensory causation require different validation. Proposed internal brief: ask the laboratory for matrix-matched recovery, blanks, precision, calibration, detection/quantification limits and the compounds its preparation under-recovers.
2. Build a teaching comparison: a missing chromatographic peak can mean absent, below detection, coeluted or poorly extracted. A peak-area ranking is not an aroma ranking.
3. Link to Zhao's six-spirit comparison for category and calibration caution, Poisson/Schieberle for recombination/omission evidence, and Haug for the separate step of sensory association/classification. Zhao's unresolved internal-standard discrepancy should not silently become a trusted quantitative input to a model assembled from prior studies.
4. A proposed small-distillery cost exercise should distinguish capital, staff handling, extraction time, instrument throughput and ongoing validation. Do not copy the paper's historical €30,000 estimate into a current business plan.

Existing evidence EXT-1989 and EXT-1990 remains supported within its condition-specific wording. EXT-1991 is narrowed to these three historical commercial samples, with reported identification totals and their internal counting caveats. Existing Zettel relationship is preserved.

Verification boundaries

Main and supplement content/visual review complete. Source and Literature Note are updated in place, and the recovered supplement is attached internally. API read-back is recorded separately; no browser-rendering verification is claimed. No public lesson or distillery profile was changed.