Geng Academic Fraud Report: Preparation and Corrosion Resistance of Copper Phthalocyanine Disulfonic Acid-Modified Polyaniline

Academic fraud report · Geng Detector

Summary

Verdict: Questionable (yellow). The strongest finding is a clear internal data inconsistency between the abstract/conclusions and Table 1 plus Section 3.3 regarding salt-spray corrosion timing. The abstract claims that CuPANI coating showed only minor blistering and 8% efflorescence (white bloom) at 12.5 d, but Table 1 records a rating of '3' (4% paled) at 12.5 d, with rating '4' (8% paled) only appearing at 17.5 d and 20 d. Section 3.3 likewise states 4% and 8% corrosion at 15 and 20 d. Image-related findings (Figure 6 background similarity, copy-move detections in img-001/img-003, noise-variance anomalies in img-000/img-003) are flagged but have plausible benign explanations (multi-panel layout, identical substrate/process, journal recompression). Automated Bayesian synthesis indicates ≥99.9% posterior probability of fraud, but this relies on an uncalibrated 5% prior and conditional-independence assumptions; confidence is limited and original images/authentic records are required for confirmation.

Verdict

Questionable (yellow). One substantive textual/numerical inconsistency is confirmed; two image-related flags carry plausible benign explanations and remain unverified.

Key findings

Internal data inconsistency (Abstract/Conclusions vs Table 1, Section 3.3) — Severity: 🟠 (orange); Status: ✅ Confirmed
Suspected background reuse in Figure 6 metallographic micrographs (EBPANI vs CuPANI at 3 d, 10 d, 15 d) — Severity: 🟡; Status: ⚠️ Benign explanations plausible (multi-panel axes, identical Q235 substrate and preparation)
Noise-variance anomalies in Figure 1, Figure 7, Figure 8 instrument-exported spectra/curves — Severity: 🟡; Status: ⚠️ Benign explanations plausible (journal recompression, illumination gradients in scans); single ⚠️ insufficient to convict

Evidence highlights

Abstract states: "中性盐雾实验12.5d，CuPANI涂层只出现少量气泡和8%的白霜" (at 12.5 d neutral salt spray, CuPANI coating showed only slight blistering and 8% efflorescence).
Conclusion item (3) repeats the same 12.5 d / 8% figure.
Table 1 footnote: "3: 4% paled, 4: 8% paled". Table 1 records CuPANI rating = 3 at 12.5 d; rating 4 (8%) first appears at 17.5 d and 20 d.
Section 3.3 states: "15 and 20 d 后也只有 4% 和 8% 的腐蚀" (only 4% and 8% corrosion at 15 and 20 d), contradicting the abstract.
img-001.jpg: copy-move detection — 59 matched block pairs at offset (56, 0).
img-003.jpg: copy-move detection — 13 matched block pairs at offset (8, 40); PRNU weak similarity vs img-000 (NCC=0.016, PCE=13.4).
img-000.jpg noise-variance CV = 1.93 (1768 valid noise blocks); img-003.jpg CV = 1.22 (2520 valid noise blocks).
Statistical flags (auto-synthesis): last-digit distribution highly non-uniform (χ²=71.3, p=0.0000), adjacent-last-digit deviation 7.1σ; Benford joint test deviates seriously; 3/3 SD values are integers; BF ≈ 1.54×10¹⁶ (decisive) under a 5% prior with conditional-independence assumption.

Notes

DOI: 10.11902/1005.4537.2015.149 — Journal: 中国腐蚀与防护学报 (Journal of Chinese Society for Corrosion and Protection); preprint archive: ChinaXiv 202303.10444v1 (March 2023); publication year: 2016.
Authors: 李小宝, 徐利, 刘怡宏, 廖婷, 李明田, 崔学军.
The machine copy-move detector itself notes that regularly repeating elements (multi-panel axes, borders, scale bars, lane stripes) can produce same-offset matches; visual review states the micrograph background consistency could also arise from photographing the same specimen at different time points.
Automated Bayesian posterior (≥99.9%) is explicitly labeled as not constituting a fraud determination; the 5% prior is fixed and uncalibrated, and conditional independence is assumed, which can overestimate combined evidence strength. Treat numerical posterior as indicative only.
Recommended next actions (from original report): contact authors for clarification of the abstract–Table 1 discrepancy and original salt-spray records; request unprocessed original Figure 6 micrographs; consider PubPeer comment depending on author response.