Provenance Record

Well-structured 871-word News piece covering three distinct April 2026 perovskite milestones: the Nanchang cesium-doping paper (Nature Energy, certified 26.61%), the CGTN-reported 27.98% Chinese record, and the Halocell/Sofab Inks stability announcement. The 'What We Don't Know' section is appropriately skeptical about the uncertified 27.98% claim.

TechXplore (April 18): fetched — Jiacheng He and Zhao Guo, Nanchang University ✓; cesium 4-(diphenylphosphino)benzoate compound ✓; certified 26.61% / uncertified 26.91% ✓; 95% retention after 1,500 hours at 85°C under ISOS-L-2 protocol ✓; alpha-phase stabilization ✓; two-step fabrication ✓; Nature Energy DOI confirmed ✓. Note: fetched via Archive.org fallback (original returned 200 per automated check). CGTN (April 3): fetched — 27.98% efficiency ✓; 'first time this technology has outperformed all single-junction silicon solar cell[s] in laboratory testing' ✓; no institution/company named ✓; no certification details ✓. SolarQuarter (April 8): fetched — Halocell Energy and Sofab Inks ✓; 1,300 hours ✓; nearly 100% normalized efficiency ✓; 1,000 lux / 85% humidity / 65°C test conditions ✓; ~20% loss for conventional after 1,000 hours ✓; roll-to-roll manufacturing ✓; IoT sensors, drones, space ✓. GreenFuelJournal: fetched — silicon SQ limit ~32% ✓; commercial modules 24.5–28.6% ✓; $1.68B–$7.02B market by 2030 ✓; mass production 2027–2029 ✓; Oxford PV tandem panels to US utilities in late 2024 ✓. Note on perovskite 31% ceiling: GreenFuelJournal does not explicitly state '31%' for perovskite single-junction SQ limit; it discusses silicon's 32% limit in the context of tandems exceeding it. The 31% figure is scientifically accepted (based on perovskite's ~1.55–1.65 eV bandgap SQ calculation) and the soft attribution phrase 'as noted by industry analysts tracking the field' is appropriate. This is a minor residual issue from Round 1; the two factually wrong figures (silicon 29%→32%, commercial modules 20–22%→24.5–28.6%) were both corrected.

Both Round 1 corrections applied: (1) Silicon SQ limit corrected from 29% to 32% ✓; (2) Commercial module range corrected from 20–22% to 24.5–28.6% ✓. The 31% perovskite single-junction ceiling figure is retained with soft attribution and is scientifically defensible. All other claims verified against sources. The CGTN 27.98% claim is appropriately caveated in 'What We Don't Know.'

Round 2 re-review. Both Round 1 corrections applied and verified. All four sources accessible and claims confirmed. Ready for publication.