A salt that melts for a moment improves the perovskite cell layer

A solar cell loses part of its light in places you cannot see with the naked eye. A team from California and South Korea has found a way to repair those places. While the film is being heated, a liquid appears briefly between the crystal grains. Then it disappears on its own. What remains is a layer that was made into a cell with 26.5 percent efficiency.

The result was published on 8 October 2026 in Science by Jason Yoo of the Korea Research Institute of Chemical Technology, Connor Dolan and David Fenning of the University of California San Diego, and Seong Sik Shin of Sungkyunkwan University. The paper is titled "Volatile eutectics to tailor crystallization for perovskite optoelectronics". In plain words: volatile mixtures that steer how perovskites crystallize. The story reads like an investigation. There is a scene, there is a suspect, and there are pictures taken in the act.

The scene: grain boundaries

Perovskite is a material from which a thin light-absorbing film can be made. Its advantage is that it forms from a solution. You apply a liquid to glass, heat it, and the film crystallizes. According to the university's press release, this opens the way to simpler and less energy-hungry manufacturing than for silicon cells. That is why perovskites are seen as a promising next generation of solar cells.

There is a catch. Such a film is not a single crystal. It is a mosaic of very small grains that grow side by side and touch along their edges. At those contacts, the grain boundaries, defects form. According to the university's release, part of the energy captured from light is lost right there. That energy does not turn into current.

The bigger the cell, the sharper the problem. The release says the defects matter more in larger panels, because efficiency falls as the area grows. Anyone who wants to sell modules one day, not just results from laboratory plates, has to fix this.

The suspect: two salts and their shared melting point

The team added two ordinary compounds to the solution: zinc bromide and methylammonium chloride. Together they form a mixture that melts at a low temperature. Chemists call it a eutectic. Salt on an icy road works in a similar way: a mixture of salt and ice freezes at a lower temperature than ice alone.

Here the mixture does what was wanted of it. During heating it melts locally around the growing grains. For a moment there is liquid at the boundaries. The grains have room to shift and settle into a tidier arrangement. As heating continues, the methylammonium chloride evaporates. The liquid vanishes and the crystal finishes growing. The zinc stays, exactly where it is needed: at the boundaries.

The "volatile" in the title therefore means something simple. One component of the eutectic escapes from the finished film. It does not linger as a useless additive.

Pictures from the scene

A story about a liquid mixture at grain boundaries could be dismissed as a guess if nobody had photographed it. So the team used X-ray microscopy with a probe that resolves nanometre-scale features. The authors' institutions include the synchrotrons at Argonne and Brookhaven, facilities that produce very intense X-ray beams.

Several things follow from these observations. According to the authors, the eutectic really does melt locally at the grain boundaries during heating. After such treatment the film has fewer defects inside the grains. The distribution of halides, that is iodine and bromine, is more even. And zinc gathers at the grain boundaries and near the surfaces.

The editors of Science add a careful explanation: zinc "likely" forms small zinc iodide inclusions there that take the defects out of play. That is still a hypothesis, not a certainty. In the UC San Diego release the authors describe the crystallization itself as a solid, liquid, solid transition that happens at one temperature. They now want to generalize it to other mixtures.

Such pictures settle whether this is only a nice story. The team did not infer the liquid from cell results alone. They saw it at the grain boundaries, and then saw it vanish together with the evaporating salt.

Numbers from cells and LEDs

Before any cell was built, the team examined the film itself. It had a high light-emission yield after optical excitation, long-lived charge carriers and low recombination at the surface. These are all ways physicists measure how much energy a film wastes. Then came the devices. The authors report three kinds of results.

The first is small laboratory cells. Efficiency in converting light to current reached 26.5 percent. A certified measurement gave 25.9 percent. The second figure matters more, because it is confirmed outside the team's own laboratory. The difference is 0.6 percentage points. Such gaps are common, but when comparing results it is wise to stick with the lower value.

The second is a minimodule with an area above 15 square centimetres. It had an efficiency of 23.9 percent. The drop relative to the small cell is clear, but it is normal. The larger the area, the harder it is to keep the film uniform.

The third is durability. Small cells with a hole-transport layer made of the polymer PTAA started at 24.1 percent efficiency. After 1400 hours of continuous operation at 85 degrees Celsius they kept 85.8 percent of that value. That is about 58 days of non-stop light and heat.

The same film also works as a diode. The team built light-emitting diodes from it with an external quantum efficiency of 26.1 percent. Put simply: of every hundred electrons fed into the diode, about 26 come out as light. A cell and a diode are the same physics in opposite directions. In a cell, light creates charges. In a diode, charges create light. In both cases a defect in the crystal wastes energy. That is why one improved film can help both devices.

Connor Dolan said in the release that in films with this additive the electron-hole pairs live longer than reported for any other direct-bandgap semiconductor. That is his opinion, not a result someone has independently checked. He also noted that the method worked for the different perovskite compositions the team tried.

How to read these numbers

A result in an efficiency table is one thing. A product on a roof is another. It is worth separating two questions.

First: is the cell efficient? Yes, and it is good. According to the university's release, efficiencies of 24 to 26 percent are among the highest for perovskite devices. The authors show the same method in three uses: in a cell, in a minimodule and in a diode.

Second: will the cell survive twenty years on a roof? This work does not show that. The test lasted 1400 hours, a little over eight weeks. It was done on small cells, in the laboratory. Silicon panels are sold today with warranties counted in decades. The cells in this work kept about 86 percent of their initial efficiency after two months, not after years. That is not proof of durability. It is a signal that this road does not end at the first wall.

The abstract also contains no data for a full-size panel or for outdoor tests. The minimodule has an area of about a third of a credit card. A rooftop panel is about a thousand times larger.

What it changes for industry

The most interesting thing about this work is how little has to be added. Instead of a new machine or a new film composition, the authors propose an additive to the solution. That is a change easier to try in an existing laboratory than a new production line. That is my conclusion, not a result from the paper. It is not yet known how this mixture will behave in production over large areas.

The zinc that stays in the film will also have to be checked in long operation. The authors treat it as a helper, but for now we have 1400 hours of observation, not years. The team is now studying other mixtures of this kind to better understand and control crystallization.

Perovskites are developed in Europe by companies and universities, including in Poland. For them this work is first of all a recipe that can be compared with their own. The cell from California will not go onto a roof, but the method may reach laboratories looking for a way to deal with defects in large films.

I rate the technology readiness level at TRL 4. The results are confirmed in the laboratory, on small cells and on a minimodule. What is missing is a test in near-real conditions and results for modules at panel scale.

The limits of this story

First, this is one paper from one team. The efficiency results are impressive, but replication by other laboratories is only beginning.

Second, the durability test concerns small cells with one kind of hole-transport layer. From the abstract we do not know how a larger module will fare after a thousand hours at 85 degrees.

Third, the mechanism with zinc at the grain boundaries carries a question mark in the Science text. The authors showed images and measurements, but the explanation of why the defects disappear is still a hypothesis.

Fourth, we do not know what this additive costs at factory scale or how the evaporating compound will affect production. The abstract does not answer that.

The work was funded by, among others, the US Department of Energy, the US National Science Foundation and the National Research Foundation of Korea.

Despite these reservations the result is interesting for a simple reason. The authors changed not only the composition of the film but also how it matures. A liquid that appears for a moment and leaves helped in a cell, in a module and in a diode.

Bibliography

Yoo J. J., Dolan C. J., Nam S., Lee J., Kang K. M., Kang B. J., Desai R., Kumar N. M., Chakrabarti A., Kim D., Cho J. H., Chung D., Palmer J. R., Vences K. X., Deng Z. J. D., Luo Y., Lai B., Zhou T., Cai Z., Holt M. V., Kiss A. M., Cho I. S., Kim D. H., Noh J. H., Bertoni M. I., Mannodi-Kanakkithodi A., deQuilettes D. W., Fenning D. P., Shin S. S. Volatile eutectics to tailor crystallization for perovskite optoelectronics. Science 394(6820), 228–233 (2026). DOI: 10.1126/science.aeb4554
University of California San Diego. A temporary melting liquid could help build better perovskite solar cells and LEDs. Press release, EurekAlert! (8 October 2026). eurekalert.org/news-releases/1147046
Szuromi P. Editor's summary of Yoo et al. Science (2026). science.org/doi/10.1126/science.aeb4554