Investigation Report: 'Rational Design of High-Performance Nanozymes via Machine Learning for Smartphone-Based Quercetin Sensing' (ACS Appl. Nano Mater., DOI: 10.1021/acsanm.6c01897)

Academic fraud report · Geng Detector

Summary

Verdict: 🔴 Substantiated (实锤). This investigation identifies multiple severe inconsistencies in the paper by Sun et al. (ACS Appl. Nano Mater., 2026; DOI: 10.1021/acsanm.6c01897). First, the text (Sections 2.7 and 3.5) states that spiked quercetin concentrations in chrysanthemum extract were 0, 1, 5, 10, and 25 μM, but Table 2 internally self-consistent calculations show the actual spikes used were 1, 20, and 50 μM. Second, reported mean ± SD values in Table 2 do not match the reported RSDs in arithmetic terms (e.g., 2.632 ± 0.076 yields 2.887% RSD, not the reported 2.212%), indicating fabricated or mismatched statistics. Third, the reported Fe²⁺/Fe³⁺ ratios of 2.01:1, 1.53:1, and 2.39:1 for Fe3O4 nanozymes contradict basic chemistry (theoretical 1:2). Findings 1 and 2 are mathematically verified; Finding 3 is chemistry-based inference. Confidence is high for Findings 1–2; Finding 3 requires expert XPS verification. Authors should provide raw data and original XPS peak fittings.

Verdict

🔴 Substantiated. Two findings are mathematically confirmed; a third finding presents a strong inferential concern that requires expert review.

Key findings

F1 — Text/table data conflict (spike concentrations): Methods (Section 2.7) and Results (Section 3.5) state spiked quercetin concentrations of 0, 1, 5, 10, and 25 μM, but Table 2 reports spikes of 1, 20, and 50 μM. Internal consistency check (e.g., 23.169 − 1.690 = 21.479; 21.479 / 20 × 100% = 107.40%, matching Table 2's recovery for the 20 μM row) confirms the table was generated from 1, 20, and 50 μM.
F2 — Mathematically impossible SD/RSD values in Table 2:
1 μM spike: reported 2.632 ± 0.076; expected RSD = (0.076/2.632)×100% = 2.887%, but Table 2 reports 2.212%.
20 μM spike: reported 23.169 ± 0.217; expected RSD = 0.936%, but Table 2 reports 0.753%.
50 μM spike: reported 47.721 ± 0.088; expected RSD = 0.184%, but Table 2 reports 0.135%.
F3 — Implausible Fe²⁺/Fe³⁺ ratios for Fe3O4: XPS deconvolution yields Fe²⁺/Fe³⁺ of 2.01:1 (E10), 1.53:1 (E20), and 2.39:1 (E30), inconsistent with the stoichiometric Fe3O4 ratio of 1:2. The 2.39:1 ratio approaches FeO rather than Fe3O4.

Evidence highlights

F1 verification (worked example): Using Table 2's reported 'total detected' (23.169) and 'original content' (1.690) at the 20 μM spike level, recovery computes to 107.40%, matching the published cell. The same arithmetic pattern holds across rows, showing the table was derived from {1, 20, 50} μM, not {0, 1, 5, 10, 25} μM.
F2 verification: Direct division of reported SD by reported mean yields RSD values that do not match those listed in Table 2 for all three spiked rows.
F3 chemistry benchmark: Pure Fe3O4 has Fe²⁺:Fe³⁺ = 1:2 (≈0.5:1); reported ratios of >2:1 are chemically improbable for a material labeled Fe3O4.
DOI: 10.1021/acsanm.6c01897 (preserved).

Notes

F1 and F2 are reproducible from the numbers printed in the paper and require no specialized equipment to verify.
F3 is a strong inference rather than a direct calculation; it should be confirmed by a surface-chemistry specialist examining the raw XPS spectra and deconvolution constraints.
Authors are urged to release raw recovery-experiment data (chromatograms/sensor readouts), replicate SD/RSD tables from raw measurements, and provide original XPS fitting files.
This automated report is intended for academic discussion and is not a formal finding of misconduct; final determination rests with the journal and the authors' institution.