Geng Report: Color and genesis of californite from Pakistan: insights from μ-XRF mapping, optical spectra and X-ray photoelectron spectroscopy — Highly Suspicious

Academic fraud report · Geng Detector

Summary

This AI-assisted forensic review rates the 2020 Scientific Reports paper (DOI: 10.1038/s41598-019-57186-0) as HIGHLY SUSPICIOUS of image/data integrity problems. The strongest finding is Figure 3: μ-XRF elemental maps allegedly from four different samples (FY-6, FG-3, FB-4, FO-1) of different elements show nearly identical texture and spot patterns (FY-6 Ti vs FG-3 Mg; FY-6 Cr vs FB-4 Mn), corroborated by copy-move forensics across multiple image files. Figure 6 XPS spectra from nine analysis points show unnaturally repetitive noise and fit profiles, though standardized fitting protocols (Shirley background, G:L = 60:40) offer partial benign explanation pending raw data. Additional confirmed internal inconsistencies: Table 1 EPMA values contradict the text (TiO₂, V₂O₃, Cr₂O₃ groupings), and a stoichiometrically impossible MnO→Mn³⁺ conversion (0.25 wt.% MnO cannot yield 0.34 wt.% Mn³⁺; expected ~0.19). A last-digit test on Table 1 was not used as evidence (test preconditions unmet for EPMA data). Confidence is high on Figure 3 reuse; other findings require raw data verification. Final determination requires institutional/journal investigation.

Verdict

Highly Suspicious (🔴) — AI-assisted forensic review rates this paper (DOI: 10.1038/s41598-019-57186-0, *Scientific Reports*, 2020, 10:285) as highly likely involving image reuse and data integrity problems. The figure-level evidence (Figure 3 cross-sample elemental map duplication) is difficult to explain benignly; other findings warrant author response and raw-data verification. Final determination requires official investigation.

Key findings

Figure 3 elemental map reuse (confirmed, 🔴): Sample FY-6's Ti map is visually near-identical to sample FG-3's Mg map, and FY-6's Cr map to FB-4's Mn map — allegedly independent measurements from different samples and different elements. Copy-move forensics (I2, I4, I6, I9, I11, I13) detected same-offset block matches in multiple image files, consistent with reuse. Overlap in data regions cannot be explained by multi-panel layout artifacts.
Figure 6 XPS spectra abnormal consistency (⚠️ pending verification): Nine Fe 2p₃/₂ panels (a–i) show unnaturally repetitive scatter, baseline noise and fit contours across samples (e.g., FY-6G vs FG-3G, FB-4Y vs FO-1YG). Partial benign explanation exists: the paper states all XPS data used identical processing (Shirley background, Gaussian:Lorentz = 60:40); raw spectra are needed.
Copy-move and noise-variance anomalies (⚠️ suspicious): All six examined embedded image files (img-000 to img-005) triggered copy-move detection with varying offsets; img-000 additionally showed severe noise-variance unevenness (anomalous blocks 0/2387, CV = 2.16). Layout artifacts may explain some matches, but batch positivity across all images corroborates Findings 1–2.
EPMA text–table contradiction (confirmed, 🟡): Text claims MgO distributions mirror Cr₂O₃ "as do TiO₂, V₂O₃ and NiO," but Table 1 contradicts this (FG-3G TiO₂ = 0.00 < FG-3Y = 0.06; V₂O₃: FY-6G = 0.09 vs FG-3G = 0.00). The green-site Cr₂O₃ list "0.41, 0.09, 0.17 and 0.26 wt.%" includes FG-3G's 0.09, yet the yellowish-green value 0.07 (FO-1YG) is excluded despite near-identical magnitude — selective narration.
MnO→Mn³⁺ conversion error (confirmed, 🟡): "0.25 wt.% MnO (Mn³⁺ 0.34 wt.%)" is stoichiometrically impossible: 0.25 × (54.94/70.94) ≈ 0.19 wt.%, not 0.34.
Last-digit anomaly (not used as evidence, 🟡): Odd/even ratio 40/189 in Table 1 deviates severely from expectation, but fixed two-decimal reporting and pervasive sub-detection-limit "0.00" values invalidate the test's preconditions for EPMA data.
Publication date placeholder (recorded, not misconduct): "Published: xx xx xxxx" remains unfilled in the PDF.

Evidence highlights

[I11] img-004.png: copy-move detection, offset (0,32), 35 matched block pairs (logLR = 1.46; 9.3% of combined evidence)
[I13] img-005.png: copy-move, offset (32,0), 14 matched pairs (logLR = 1.46; 9%)
[I6] img-002.jpg: copy-move, offset (8,24), 20 matched pairs (logLR = 1.46; 8%)
[I2] img-000.jpg: copy-move, offset (0,32), 9 matched pairs (logLR = 1.46; 6.9%); noise-variance CV = 2.16 (anomalous blocks 0/2387)
Bayesian synthesis: posterior ≥ 99.9% (prior 5%), 95% CI [100%, 100%], BF = 1.26e+08; conditional independence assumption may overstate correlated evidence; 4 clues excluded after review (benign explanations or inapplicable tests)
Automated verbatim checks confirmed all quoted EPMA values against Table 1; the "missing" Figure 3 caption flagged by machine review was a PDF text-layer character-mojibake artifact (">e" for "The"), and the finding was retained.

Notes

Recommended actions per the report: request original Figure 3 map data files and Figure 6 raw XPS spectra from authors; raise on PubPeer; (not yet done) report to *Scientific Reports* editors and to the authors' institution (China University of Geosciences, Beijing).
Findings 2, 3, and 6 retain documented benign explanations and should not be treated as established misconduct absent raw data.
This report is AI-assisted, for academic discussion only; final integrity determinations require official investigation by the journal or institution. Misclassification (false positives/negatives) is possible.