LONGi has announced that its independently developed crystalline silicon–perovskite tandem solar cell has reached a conversion efficiency of 35.5%, certified by the European Solar Test Installation (ESTI). The result, presented on 14 July 2026 at the 2026 Solar and Energy Storage Innovation Conference, sets a new world record for the tandem architecture and matters because it moves the most promising route beyond single-junction silicon another step closer to manufacturable hardware.
A record built on a rapid sequence of steps
Crystalline silicon–perovskite tandems are widely treated as the mainstream route for next-generation ultra-high-efficiency cells, with a theoretical efficiency limit of up to 43%. That ceiling sits far above the Shockley–Queisser limit of 33.7% for a single-junction device, which is the thermodynamic reason the industry is willing to accept the added process complexity of a second absorber.
LONGi's own progression illustrates how quickly the route is maturing. The company's tandem team reached 33.9% in November 2023 and 34.6% in June 2024, then moved through 34.85% and 35.2% before certifying 35.5%. In May 2026 the 35.2% two-terminal result was included in the 68th edition of the Solar Cell Efficiency Tables published by the team led by Professor Martin Green at the University of New South Wales — the reference record list for the field.
From laboratory coupon to industrial dimensions
Record cells are usually small, so the more consequential numbers in LONGi's announcement of the certified 35.5% tandem result concern larger areas. Under conditions closer to industrial-scale dimensions the company reports 34.3% on 261 cm² and 32.2% on 274 cm². Tandem modules have been independently certified at 31.4% and 29.4% and are likewise included in the efficiency tables.
- Champion cell: 35.5%, certified by ESTI
- Large-area cells: 34.3% (261 cm²) and 32.2% (274 cm²)
- Modules: 31.4% and 29.4%, independently certified
- Theoretical limit for the architecture: up to 43%
The gap between the 35.5% champion device and the 31.4% module is the industrialisation problem in numerical form: interconnection, encapsulation, area scaling and uniformity all take their share. LONGi frames its response as a tiered research structure — "one generation in mass production, one in development, and one in reserve" — intended to keep iterating rather than to produce isolated laboratory peaks.
The exergy view
Sunlight is a high-exergy resource, and a single-junction cell wastes much of that quality before any electricity is produced: photons above the band gap are thermalised, and photons below it pass through unused. Both are irreversibilities, and together they are what the Shockley–Queisser limit of 33.7% expresses. A tandem addresses the loss at its source by splitting the spectrum between two absorbers whose band gaps are better matched to it, raising second-law efficiency rather than merely adding hardware. The move from 33.9% to 35.5% therefore represents a genuine reduction in exergy destruction inside the device, not a rearrangement of it. The 4.1-percentage-point gap between champion cell and certified module is where the remaining destruction now sits — in resistive, optical and area-scaling losses that industrialisation, not physics, must remove.
