Content
Citation(s) & Footnote(s)
Book(s)Full review and provenance
Ancuța Chetrariu and Adriana Dabija (2021), “Spent Grain from Malt Whisky: Assessment of the Phenolic Compounds,” Molecules26,3236. DOI10.3390/molecules26113236. All20 supplied pages read, including all4 tables/every coefficient and cell, equations1–8, all5 figures (36 response-surface panels plus process diagram and FTIR spectra), and52 references. The9,290,943-byte original is preserved and uploaded internally. CC BY4.0 is stated in the article; Figure1 credits a separate source, so retain attribution and check its reuse basis. Exact Proton/mirror byte identity and browser rendering remain unverified.
What was tested
Draff came from Alexandrion Group in Romania. It was frozen, dried at50C for26h, ground and sieved; only the200–300µm fraction was tested (p15). This is a selected dried malt-whisky by-product, not whiskey liquid, fresh draff as delivered, bourbon stillage or all distillery waste. Source-batch replication and variation across sites/crops are not established.
Two methanol/water extraction methods were compared: high-speed Ultra-Turrax(UT) homogenization for30–90seconds and ultrasound(US) at30–70C for10–30minutes. Each used0.3–0.9g dry material in30mL solvent, with40–80% methanol. The variable named liquid-to-solid ratio is actually entered as sample mass at fixed volume:0.3g means100mL/g and0.9g means33.3mL/g, so increasing that factor reduces the liquid-to-solid ratio.
TPC was measured by Folin–Ciocalteau in gallic-acid equivalents; TFC by a quercetin-equivalent colorimetric assay; DPPH by in-vitro radical inhibition; selected individual compounds by HPLC-DAD; functional groups by FTIR. Replicate analytical measurements are not independent distillery batches.
Findings, with denominators retained
The reported ranges are0.62–1.76mg GAE/g dry SG for UT total phenolics and0.57–2.11 for US; flavonoids0.60–1.67mg QE/g for UT and0.50–1.63 for US; DPPH25.88–79.58% and27.49–78.30%, respectively. These are extraction/assay outcomes, not concentrations in whisky or demonstrated health benefits.
Table4(p14) gives seven detected acids for UT and three for US. p-Coumaric acid is20.4±1.72 versus14.0±1.14mg/100g dry SG; vanillic acid3.1±0.8 versus10.0±1.03; rosmarinic6.5±0.96 versus4.0±0.76. Protocatechuic0.7±0.05, p-hydroxybenzoic1.1±0.06, caffeic0.7±0.03 and chlorogenic5.4±1.1 occur in the UT column and are non-detected in US. Non-detected does not mean chemically absent without detection limits. UT's seven reported concentrations sum37.9mg/100g, US's three28.0; p-coumaric is53.8% and50.0% of those sums respectively. Thus it is largest in both, but exactly half rather than a strict majority of the US listed sum; US vanillic exceeds rosmarinic despite the generic narrative ranking.
Folin totals are not an inventory of individually identified phenols: the authors themselves discuss other reducing compounds responding to the assay(p13). HPLC identification uses retention-time standards and absorbance, with calibration R²>0.99; chromatograms, recovery, LOD/LOQ and full calibration data are not supplied. FTIR shows overlapping functional-group bands, not independent identification of each phenolic acid.
Model and internal-consistency audit
The major limitation is that reported “optimal” responses cannot consistently be reproduced from the printed models.
Using Table1's coded levels and all coefficients of Equations1–6/Table2 at the Table3 predicted settings gives UT TPC1.7225, TFC1.5083 and DPPH78.1030, versus Table3's1.40,1.36 and54.75. US gives1.4151,0.8370 and74.1713, versus1.44,1.29 and74.13. Rounded coefficients may explain small deviations, but not the large UT DPPH or US TFC discrepancies. The reported predictions largely equal the intercepts at the center point, although the listed settings are not centered. These are reviewer calculations saved in model-audit.json, not corrected experimental observations. Table3's less-than5% agreement between its own predicted and verified values does not resolve that inconsistency.
Additional issues with precise locators:
- Table2 UT TPC lack-of-fit p=.0126 is significant at.05, despite broad model-validity language. Significant model F tests do not prove adequate fit. Text on pp5/8 claims p<.0001 for TPC/TFC, whereas Table2 reports .0002/.001 and .0005/.0006. Adjusted R² falls to.7578–.8881 across the models.
- Table1 labels US time in seconds, while methods, graphs and Table3 consistently use minutes. Table3 X1–X4 rows for UT have sample mass and speed in the reverse order from Table1; audit follows the physical units and Table1 definitions.
- The design is called both Box–Behnken and full factorial. The paper supplies factor levels and fitted coefficients, but not the complete run-level response table or raw residuals, despite saying data are in the article. No supplementary appendix is included or identified in this supplied copy.
- Figure2 has12 panels of TPC; Figure3 has12 of TFC; Figure4 has12 of DPPH. These are model surfaces, not36 independent validation experiments. UT factor order is speed,mass,methanol,time; US is mass,methanol,time,temperature. Narrative discussion of US on p11 incorrectly includes stirring rate.
- The p8 statement that UT gives slightly better phenolic yield conflicts with the larger US reported maximum and Table3 verified TPC. The methods differ in more than mechanical treatment, so generalized superiority is not isolated.
- Conventional-method flavonoid and DPPH results on p5 are labeled mg GAE/g, inconsistent with the QE and percent-inhibition assays elsewhere. The quoted water-bath DPPH number57.89 also exceeds UT's verified54.77, undermining blanket claims of superiority. Do not silently convert these units.
- Moisture Equation7 on p15 subtracts container-plus-dry-sample mass from empty-container mass, yielding a negative numerator with the printed definitions. The reported5.04% cannot follow that literal formula.
- DPPH methods call distilled water the blank and divide by blank absorbance; it is unclear whether this means a reagent control with water replacing extract. The wording is insufficient for a validated assay protocol. TFC calibration R²=.95946 is also materially below the HPLC claim and needs separate consideration.
- FTIR prose first assigns3000–2850cm−1 to C–H, then assigns2921cm−1 to C=O; the internal inconsistency warrants checking rather than copying. A missing detectable band is not categorical proof of a bond's absence.
- Reference32 is Barker but the text also cites it for del Río (listed as33); the patents claim cites31, a whisky history/stock-cycle article. Market rankings, company-first claims and legal descriptions are background assertions, not current verification.
Critical assessment and Academy applications
The strongest contribution is a bounded demonstration that this prepared draff contains recoverable compounds and that extraction conditions change assay response and profile. Methanolic extraction does not itself establish a usable food ingredient. No food trial, residual-solvent assessment, sensory acceptance, shelf-life study, clinical benefit, energy balance, life-cycle assessment or economic evaluation was conducted. The conclusion's economic/environmental benefits are proposals, with further food research explicitly requested.
Use privately in a sustainability research companion: show where draff leaves malt-whisky mashing, then compare feed use, recovery and other pathways as alternatives requiring evidence. Do not imply malt draff and bourbon post-distillation grain residue are identical. A proposed learner exercise can distinguish composition, recoverability, product suitability, net environmental impact and commercial value. Another can ask why a smooth optimization surface and high R² are not enough when validation numbers conflict.
Cross-source synthesis: La Grange-Nel2003's engineered yeast improves one mechanism but has slower/lower-yield process performance; this study similarly shows that recoverable compounds do not establish viable recovery economics. Agu2006 illustrates the importance of treatment and measurement denominators. Connect through the existing Zettel “Process interventions require multi-objective validation.” Proposed uses are not edits to live course pages.
