Academic Integrity Review Report: High-quality NiOx nanoparticles synthesized via low temperature chemical precipitation method for high-performance inverted perovskite photovoltaics

Academic fraud report · Geng Detector

Summary

This report evaluates a 2026 Nature Communications article (DOI: 10.1038/s41467-026-73911-6) by Ziyuan Liu et al. on NiOx nanoparticles for inverted perovskite photovoltaics. The overall verdict is highly suspicious (orange/high-suspicion tier). The most serious finding is an unresolved numerical contradiction in the stability data for the 1.55 eV device after 2,100 h of MPPT operation: the Abstract and Discussion report 91.5% retention, while the Results section reports 91.90%. A secondary concern is the implausibly close alignment between two independent aging tests—91.5% after MPPT and 91.4% after 85°C/85%RH—suggesting possible data curation. A minor arithmetic inconsistency was also identified in a 14 cm² module PCE (claimed 23.18% vs. calculated 23.21% from reported Voc, Jsc, and FF). The narrative framing of a simple ice-bath precursor treatment producing comprehensive performance improvements is flagged as atypical. Image-level analysis was not possible from the available text. Confidence is moderate; definitive conclusions require raw J–V curves, stability logs, and figure-level inspection.

Verdict

🟠 Highly suspicious. Multiple internal inconsistencies and narrative concerns warrant formal verification of the underlying experimental data and figures.

Key findings

Internal data contradiction (severity: 🔴 critical). The retention value for the 1.55 eV device after 2,100 h MPPT is reported as 91.5% in the Abstract and Discussion but 91.90% in the Results section.
Implausibly symmetric aging metrics (severity: 🟠 high). Two independent aging protocols yield near-identical retention: 91.5% (MPPT, 2,100 h) vs. 91.4% (85°C/85% RH, 2,000 h), a 0.1% gap.
Module PCE arithmetic mismatch (severity: 🟡 medium). Reported module values: Voc = 6.57 V, Jsc = 4.55 mA cm⁻², FF = 77.61%. Computed PCE = 23.21%, whereas the paper reports 23.18% (Δ = 0.03%).
Narrative red flag (severity: 🟠 high). A single low-temperature (0°C / ice-bath) precursor step is credited with simultaneously resolving dispersion, SAM coverage, energy-level alignment, defect density, and buried-interface stability, raising efficiency from 26.31% to 27.37% (certified 27.09%).
Image analysis not performed (severity: 🟡 medium). Figures 1–5 (SEM, TEM, XPS, J–V) could not be examined at the pixel level from the available source.

Evidence highlights

MPPT 2,100 h retention contradiction: Abstract — "The devices retain 91.5% efficiency after 2,100 h of continuous operation..."; Results — "The LT-NiOx-based devices with 1.55 eV ... could maintain 91.90% of their initial PCEs ... after 2,100 h"; Discussion — "The 1.55 eV target PSCs maintained 91.5% of the initial PCEs after aging at MPPT for 2,100 h."
Symmetric aging values: 91.5% (MPPT, 2,100 h) and 91.4% (85°C/85% RH, 2,000 h).
Module arithmetic: 6.57 × 4.55 × 0.7761 = 23.21 vs. reported 23.18%.
Efficiency jump narrative: 26.31% → 27.37% (certified 27.09%) attributed to ice-bath LTCP process.
DOI: 10.1038/s41467-026-73911-6.

Notes

This assessment is based on text extracted from the article; no figures were inspected at the pixel level, so image-based duplication or splicing cannot be confirmed or ruled out.
The 0.03% module PCE discrepancy is small and could in principle arise from rounding, but combined with the larger 91.5%/91.90% discrepancy it weakens overall confidence in the reported numerics.
The "ice-bath solves everything" framing is a qualitative concern, not proof of misconduct; it should motivate scrutiny of the experimental design and reproducibility.
Recommended actions: (1) request raw, untrimmed J–V data and MPPT/aging logs; (2) reconcile the 91.5%/91.90% discrepancy with a corrected erratum or author response; (3) flag the manuscript on PubPeer; (4) monitor related publications from the same group for similar narrative patterns.
Limitations: AI-assisted analysis; final determination of misconduct requires investigation by qualified institutions.