Integrity review of 'Recyclable cyclic bio-based acrylic polymer via pairwise monomer enchainment by a trifunctional Lewis pair' (Nature Chemistry, 2023)

Academic fraud report · Geng Detector

Summary

Verdict: essentially clean, with minor points requiring attention. This report reviews a 2023 Nature Chemistry paper by Song, He, Zhang, Gilsdorf and Chen on a recyclable cyclic bio-based acrylic polymer produced via Lewis-pair polymerization. No definitive evidence of data fabrication or image manipulation was identified. Internal cross-checks were favorable: molecular-weight data in Figure 2a are internally consistent (Mw/Mn reproduces the reported Đ across all five entries, with rounding to the last decimal), and the equipment referenced (Bruker Avance II 500 NMR, Bruker APEX-II CCD diffractometer, TA Q50/Q20 thermal analyzers, Waters 1515 HPLC) was all commercially available well before the 2021–2022 experimental window. Two minor concerns remain. First, several plots (MALDI-TOF fits, zero-order polymerization kinetics, TGA overlays) appear unusually 'textbook-perfect,' with R² values ≥ 0.986 and minimal scatter, which may reflect either exceptional experimental skill or selective presentation. Second, Đ for the [M]/[I] = 1200:1 sample is reported as 1.87 in Figure 2a but 1.88 in Figure 6b; this is consistent with normal inter-run variation but warrants a small author clarification. No institutional escalation is recommended at this time.

Verdict

Essentially clean (✅) with two minor points (🟡). No evidence of fabrication, image manipulation, or methodological fraud was found. Internal arithmetic and equipment provenance both check out. Minor concerns about idealized data presentation and a trivial Đ discrepancy (1.87 vs 1.88) are noted but do not warrant journal referral.

Key findings

Internal math self-consistency (Figure 2a): Reverse-computation of Đ = Mw/Mn reproduces every reported value within rounding tolerance.
Entry 1: 348 / 1.50 = 232 → matches reported Mn = 232.
Entry 2: 508 / 1.61 = 315.5 → rounds to 316 (matches).
Entry 3: 544 / 1.58 = 344.3 → rounds to 344 (matches).
Entry 4: 735 / 1.92 = 382.8 → rounds to 383 (matches).
Entry 5: 755 / 1.87 = 403.7 → rounds to 403 (matches).
Equipment plausibility (Methods, pp. 374–376): All instrumentation (Bruker Avance II 500, Bruker APEX-II CCD, TA Q50/Q20, Waters 1515 HPLC) was commercially available years before the 2021–2022 experiments.
Suspiciously 'textbook' data presentation (Figures 2c, 2d, 4a, 4b, 6b, 6c): MALDI-TOF fits near R² = 1, zero-order kinetics with R² > 0.986 and no stray points, and TGA mass-loss curves from different-Mn samples that essentially overlap. Plausible for a highly skilled lab, but raises the possibility of over-smoothing or curated selection.
Minor numeric discrepancy (Figure 2a vs Figure 6b): Đ for the [M]/[I] = 1200:1 sample (Mn = 403) is reported as 1.87 in Figure 2a and 1.88 in Figure 6b. Likely typographic or re-aggregation related; not, by itself, indicative of misconduct.

Evidence highlights

DOI: 10.1038/s41557-022-01097-7
Journal: Nature Chemistry, Volume 15, March 2023 (online 28 November 2022).
Pages consulted: Figure 2a (p. 368), Figure 6b (p. 373), Methods (pp. 374–376).
Key numeric anchors (Figure 2a): Mw/Mn/Đ triplets 348/232/1.50; 508/316/1.61; 544/344/1.58; 735/383/1.92; 755/403/1.87.
Conflicting figure caption (Figure 6b): Đ listed as 1.88 for the Mw = 755 sample.

Notes

The unusually low scatter in kinetics, MS fits, and TGA overlays should be checked against the Supplementary Information (raw, unsmoothed spectra and replicate curves) to rule out selective presentation. No public SI anomaly has been confirmed here.
The 1.87 ↔ 1.88 Đ discrepancy is within typical GPC run-to-run variance but should ideally be reconciled by the authors (e.g., via a corrigendum).
No institutional escalation is recommended based on the present evidence; substantive misconduct would require additional proof beyond the 'overly tidy data' signal.