LONGi tandem cell hits 34% with a ZrO2 interlayer
Nikita LutsenkoFounder & CEO, SOLTECH
Soochow University and LONGi report a 34.0% perovskite-silicon tandem cell with a ZrO2 interlayer, 2.014 V certified Voc and 84% efficiency left after 2,000 hours.
A perovskite-silicon tandem solar cell has reached 34.0% efficiency at Soochow University in China with manufacturer LONGi, using a zirconia nanoparticle interlayer to passivate the buried interface. The device recorded an independently certified open-circuit voltage of 2.014 V and retained 84% of its starting efficiency after 2,000 hours, pv magazine reported on 10 September 2026.
What the 34% perovskite-silicon tandem cell measured
| Parameter | Value | Source |
| Power conversion efficiency | 34.0% | Science Bulletin, via pv magazine |
| Open-circuit voltage, in-house | 1.997 V | Science Bulletin, via pv magazine |
| Open-circuit voltage, certified | 2.014 V | Science Bulletin, via pv magazine |
| Short-circuit current density | 20.36 mA/cm2 | Science Bulletin, via pv magazine |
| Fill factor | 83.62% | Science Bulletin, via pv magazine |
| Stability | 84% of initial efficiency after 2,000 h | Science Bulletin, via pv magazine |
| Control cell without the interlayer | 70% after 1,000 h | Science Bulletin, via pv magazine |
| Key material | Monoclinic ZrO2 nanoparticles, bandgap about 5.49 eV | Science Bulletin, via pv magazine |
| Cell architecture | c-Si bottom cell, inverted p-i-n perovskite top cell | Science Bulletin, via pv magazine |
| Current world record, same cell type | 35.5%, LONGi, certified July 2026 by ESTI | pv magazine, 15 July 2026 |
How the zirconia interlayer works
The problem being solved is the buried interface. In an inverted perovskite cell the hole-transport layer is a self-assembled monolayer one molecule thick, sitting directly on a transparent conductive oxide. That oxide is defective, the monolayer does not cover it perfectly, and the gaps become leakage paths and recombination sites that cost voltage.
The team inserted discrete zirconia nanoparticles into that gap rather than a continuous film. A continuous dielectric would insulate the contact; a discontinuous, nanoporous one shields the worst of the oxide while leaving exposed monolayer regions to extract holes. Diluting the precursor controls how much of the surface the nanoparticles cover.
Two secondary effects matter. Zirconia has a high dielectric constant, which reduces charge accumulation at the interface and suppresses hysteresis. X-ray photoelectron spectroscopy also showed covalent Zr-O-P bonds forming, anchoring the monolayer to the surface at both ends - the "dual-anchored" design of the paper's title. As LONGi researcher He Bo put it to pv magazine, the layer serves as "a nanoscale localized contact".
Why stability is the number to read, not efficiency
Perovskite efficiency records have arrived at a steady pace for a decade; durable perovskite devices have not. The comparison in this paper is the useful one: 84% of initial efficiency after 2,000 hours for the treated cell, against a control that fell to 70% after only 1,000 hours. The interfacial fix improves lifetime more dramatically than it improves output.
Even so, 2,000 hours is about 83 days. A module sold in Dubai carries a 25 to 30 year performance warranty, which is roughly 110,000 hours of daylight operation in a climate where module back-sheet temperatures routinely pass 70C. Laboratory stability tests and Gulf field conditions are not yet the same conversation. Our note on lab records versus the roof covers why the gap is systematic rather than incidental.
The SOLTECH view
Our position is that this result changes the research agenda, not the procurement list. The headline that will travel is "34%", and it is the least interesting number in the paper. LONGi's own certified record is already 35.5%, so 34.0% is not a record at all - it is a demonstration that a specific interface fix buys voltage and lifetime at the same time, which is the harder trick.
For a villa owner in the UAE, the correct response is to do nothing. There is no tandem module to buy, there will not be one for several years, and waiting for one costs you the tariff savings of every summer in between. The modules we specify today are silicon, mostly TOPCon, and the useful comparison is between the products on the panel options we actually install, not between a shipping product and a laboratory coupon.
For an EPC contractor, the signal worth tracking is where LONGi puts its capital. Perovskite work sitting in a university partnership is research; the same work appearing in a pilot line announcement is a manufacturing decision. Until that second step happens, tandem has no bearing on module pricing or lead times, which are being set by trade policy and freight - see our note on Gulf import disruption.
For an investor, we would be cautious about any model that assumes tandem modules reach commercial supply inside this decade at Gulf-relevant volumes. Nothing in this paper supports that, and the published stability test is two orders of magnitude short of a warranty period. The reasonable planning assumption remains incremental silicon gains, which is also the assumption behind our investor economics.
What we would want to see next, and do not have: the same interlayer applied to a full-area cell rather than a small laboratory device, and a damp-heat test at 85C. Perovskites fail on heat and moisture, and the Gulf supplies both in quantity.
Sources: pv magazine - LONGi, Soochow University unveil 34.0% perovskite-silicon tandem solar cell, 10 September 2026, Science Bulletin - Nanoscale interfacial scaffold enables perovskite/silicon tandems with 34% efficiency and an open-circuit voltage over 2.01 V, pv magazine - LONGi sets new world record with 35.5% efficient perovskite-silicon tandem cell, 15 July 2026.