Perovskite solar's durability problem has killed better pitches than this. The material converts sunlight to electricity more efficiently than silicon, degrades faster than silicon, and has been "almost ready for commercialization" since roughly the Obama administration. Every year brings a new efficiency record and the same stability problem. Until — possibly — now.
A team from the University of Electronic Science and Technology of China (UESTC) in Chengdu and Zhejiang Jinko Solar Co. published results in Nature Energy on March 31 showing that a molecule called 2-mercaptobenzothiazole — 2-MBT, a chemical used in rubber vulcanization — can stabilize perovskite layers on industrial-grade tunnel oxide passivated contact (TOPCon) silicon wafers well enough to survive realistic operating conditions. The two-terminal monolithic tandem cell achieved a certified stabilized power conversion efficiency of 32.76 percent and retained 91 percent of its initial output after 1,700 hours of continuous operation, according to the paper.
That 1,700-hour figure is notable — but it is not a field-wide durability record. Prior lab results have exceeded this under different test conditions, including a 2021 perovskite-silicon tandem that ran past 10,000 hours under accelerated testing, per Nature Reviews Materials. The more specific point here is what the device was tested on: TOPCon silicon wafers, the standard building block of modern commercial silicon solar, rather than bespoke lab cells. Getting stability data on production-grade substrates is a different engineering problem than getting it on optimized single-junction lab cells, and that distinction is where this paper lives. The trap-assisted recombination rate — a key failure mode in perovskite cells — dropped by roughly a factor of seven, from 3.2 × 10⁵ to 4.3 × 10⁴ centimeters per second, per the same paper.
The industrial wafer context matters. Previous perovskite stability research often used thin, bespoke silicon not designed for tandem architectures — the kind of setup where you baby the cell through every deposition step. This work used TOPCon wafers, the standard building block of modern commercial silicon solar. The perovskite layer is deposited on top, and the thermal mismatch between the two is a genuine manufacturing problem. Thin wafers have lower thermal mass, so they transfer heat faster during perovskite deposition, which causes the perovskite to crystallize too quickly — leaving voids, defects, and a film that starts degrading the moment it is made. 2-MBT slows the crystallization reaction by binding to the perovskite's organic cations in two different ways simultaneously, giving the film time to form uniformly.
The molecule is not glamorous. It has been used in rubber manufacturing for decades. But the homeliness of the chemistry is the point — this is not a proprietary catalyst requiring a specialized supply chain. It is a commodity chemical applied in small quantities to an industrial process that Jinko Solar already uses at scale.
This work is not a world record. LONGi Solar holds the NREL-certified perovskite-silicon tandem efficiency record at 34.85 percent, set in April 2025, according to Fluxim's tracker. Jinko Solar itself hit 34.76 percent on TOPCon wafers in December 2025, certified by China's National PV Metrology and Testing Center — per pv-magazine. LONGi has also communicated an unconfirmed result of approximately 35 percent, reportedly certified by ESTI, though no independent certification report or NREL table update has been published, per Fluxim.
The efficiency ranking tells you something important about what this paper is actually doing. It is not chasing the record — it is betting that stability matters more than the last fraction of a percent. The authors, led by Qilin Zhou, Renjun Guo, and Yi Hou at UESTC, explicitly frame the work as addressing a barrier to manufacturing rather than a barrier to laboratory performance. The Science paper Aydin et al., 2024 mapped the same terrain: efficiency records are accumulating faster than operational lifetime data, and the gap between them is why perovskite-silicon tandems remain a research field rather than a product category.
Jinko Solar's strategy reflects this. The manufacturer recently partnered with XtalPi, an AI-for-science platform company, to build what they describe as the first fully closed-loop tandem development line combining robotic execution with AI-driven model refinement. The stated goal is to accelerate stability testing and device optimization — specifically moving past the trial-and-error workflow that has slowed the field, per pv-magazine.
What remains unknown is how 2-MBT performs under real-world conditions: humidity, outdoor thermal cycling, partial shading, and the mechanical stresses of rooftop mounting. The 1,700-hour result was measured under continuous operation in controlled conditions. Perovskite degrades differently in the field than in accelerated tests, a point made in Duan et al., Nature Reviews Materials, 2023. The field history of perovskite solar panels — as opposed to lab cells — does not yet exist, because no manufacturer has shipped them at scale.
Whether 2-MBT survives a summer on a rooftop in Arizona is a question that no peer-reviewed paper has answered. This work suggests someone is at least asking the right question.
