Concerns report: "Solvent-Free Chemical Recycling of Polyesters and Polycarbonates by Magnesium-Based Lewis Acid Catalyst" (Angew. Chem. Int. Ed., 2025; DOI: 10.1002/anie.202420688)

Academic fraud report · Geng Detector

Summary

Verdict: Questionable. This report flags serious statistical and presentational irregularities in a paper that reports catalyst yields from triplicate experiments and a 10-run catalyst recycling test, yet presents every yield as a whole integer without error bars or standard deviations. The most consequential issue is that all reported average yields in Tables 1 and 2 (e.g., 90%, 96%, 98%, 94%, 92%, 91%) are given to the unit percent despite the authors stating these are means of three independent experiments; genuine replicate-based means virtually never collapse to exact integers. The Figure 5 recycling histogram is described as unnaturally flat across 10 cycles with no error bars, which is inconsistent with expected catalyst attrition. TGA traces in Figure 2 are described as suspiciously uniform. Confidence is moderate; the strongest evidence is statistical (absence of any variability measures), while the image-level concerns are partly subjective. No fabricated figures or duplicated panels were identified.

Verdict

Questionable. The paper's quantitative yield tables and recycling data lack any measure of dispersion (SD, SEM, error bars) despite the authors explicitly stating the values are means of three independent experiments. The convergence of all reported means on whole-integer percentages is statistically implausible and constitutes the strongest single concern. Figure-level manipulations could not be conclusively demonstrated from the PDF alone.

Key findings

Unrealistic integer-only averages (Tables 1 and 2): Footnotes state [b] The average separation yield of three experiments, yet every reported mean (90%, 96%, 98%, 94%, 92%, 91%, and others) is a whole integer with no decimals, no SD, and no SEM. Triplicate gravimetric/workup means in synthetic chemistry almost invariably carry at least one decimal place. 🔴 High severity.
Figure 5 catalyst recycling plot is implausibly uniform: The bar heights across the 10 MgCl2 reuse cycles are described as forming an almost perfectly flat top edge within the stated 92–96% range, with no error bars. Real heterogeneous or homogeneous catalysts show run-to-run drift due to handling losses, weighing error, and gradual deactivation. 🔴 High severity.
Missing statistical reporting: No Mean ± SD, no Mean ± SEM, no p-values, and no error bars are provided anywhere in the quantitative tables or the recycling figure, despite n = 3 being claimed. 🟠 Medium-high severity.
Figure 2 TGA traces look over-smoothed: The TGA curves for several Lewis acid catalysts on PCL are described as visually parallel with very low background noise, raising the possibility of excessive software smoothing. 🟡 Medium severity; subjective, evidence insufficient to confirm manipulation.

Evidence highlights

DOI: 10.1002/anie.202420688
Tables 1 & 2: every value is an exact integer percentage; footnote attributes values to averages of three experiments. No dispersion metric is given anywhere in the table caption or body.
Figure 5 / Page 6: bar chart of 10-cycle MgCl2 recycling shows near-identical bar heights; no error bars are drawn; the text states "monomer recovery yields were maintained in the range of 92%–96%".
Figure 2 / Page 2: TGA weight-loss curves for different Lewis acids on PCL are described as unusually parallel and noise-free.
Methods/Results: no statistical analysis section, no replicate variance reporting, no indication that any runs were excluded.

Notes

These findings are based on the published PDF and the analytical observations summarized above; raw NMR spectra, balance printouts, and the underlying TGA files were not available for inspection, so definitive claims of fabrication cannot be made.
The integer-only mean pattern is the most defensible statistical red flag and is the issue most likely to withstand peer review or editorial scrutiny.
The Figure 2 TGA concern is suggestive but not sufficient on its own; a request for the original .csv/.xls TGA exports and the smoothing settings used in the plotting software would resolve it.
Recommended follow-up: (1) request raw replicate data and per-run yields from the corresponding author; (2) raise a PubPeer comment focused on the missing SD/SEM and the integer-mean anomaly; (3) flag the absence of error bars on Figure 5 to the journal editorial office.
This report is AI-assisted and intended for academic discussion only; final determination of misconduct requires an institutional or editorial investigation.