AI cracks the code for the next generation of solar energy

The increase in global energy demands push the limits of solar technology. Sweden scientists have now made a major step towards unlocking the potential of haloged perovskites.

Global demand for electricity is climbing at a rapid rate, which makes it essential to find lasting ways to meet future needs. A possible solution lies in the development of advanced solar cell materials which are much more effective than those used today. These new materials could be made so thin and flexible that they could cover everything, from smartphones to whole buildings.

Researchers at Chalmers University of Technology In Sweden, recently progressed to fight against one of the most promising and confusing options: haloged perovskites. By combining computer simulations with automatic learningThey start to disentangle the complex behavior of these materials.

According to the International Energy Agency, electricity already represents 20% of global energy consumption. Over the next 25 years, this share should exceed 50%, emphasizing the urgency of developing cleaner and more efficient energy technologies.

“To meet demand, there is an important and growing need for new environmental and effective energy conversion methods, such as more effective solar cells. Our results are essential to design and control one of the most promising solar cell methods for optimal use.

Promising materials for effective solar cells

The materials in a group called Perovskites halogenures are considered to be the most promising to produce profitable, flexible and light solar cells and optorelectronic devices such as LED bulbs, as they absorb and emit extremely effective light. However, perovskite materials can deteriorate quickly and know how to best use them requires a more in -depth understanding of the reason why it happens and how the materials work.

Scientists have long struggled to understand a particular material within the group, a crystalline compound called lead iodide of Formidinium. It has exceptional optoelectronic properties. The greater use of the material has been hampered by its instability, but this can be resolved by mixing two types of haloged perovskites. However, more knowledge is necessary on both types so that researchers can better control the mixture.

The key to the design and control of materials

A research group at Chalmers can now provide a detailed account of a large phase of the equipment which has previously been difficult to explain by experiences alone. Understanding this phase is the key to be able to design and control both this material and the mixtures according to it. The study was recently published in Journal of the American Chemical Society.

“The low temperature phase of this material has long been a missing part of the research puzzle and we have now settled a fundamental question on the structure of this phase,” explains Chalmers Sangita Dutta.

Automatic learning contributed to the breakthrough

The expertise of researchers lies in the construction of precise models of different materials in computer simulations. This allows them to test the materials by exposing them to different scenarios and these are experienced experimentally.

Nevertheless, the modeling of the materials of the family of perovskite halogenures is delicate, because the capture and decoding of their properties requires powerful supercomputers and long simulation times.

“By combining our standard methods with automatic learning, we are now able to perform simulations that are thousands of times longer than before. And our models can now contain millions of atoms instead of hundreds, which brings them closer to the real world, “explains Dutta.

Laboratory observations correspond to the simulations

The researchers identified the structure of the formidinium lead iodide at low temperature. They could also see that the formididinium molecules are stuck in a semi-stable condition while the material cools. To ensure that their study models reflect reality, they have collaborated with experimental researchers University of Birmingham. They cooled the material to – 200 ° C to ensure that their experiences corresponded to the simulations.

“We hope that the ideas we have drawn from simulations can contribute to modeling and analyzing the complex materials of Perovskite in the future,” explains Erik Frasson, in the Chalmers Physics department.

Reference: “Reveals the low temperature phase of FAPBI3 using a machine -learned potential” by Sangita Dutta, Erik Frasson, Tobias Hainer, Benjamin M. Gallant, Dominik J. Kubicki, Paul Erhart and Julia Wiktor, August 14, 2025, Journal of the American Chemical Society.
DOI: 10.1021 / jacs. 5C05265

Research was supported by the Swedish Strategic Research Foundation, the Swedish Energy Agency, the Swedish Research Council, the European Research Council, the Knut and Alice Wallenberg Foundation and the Nano advance field at the Chalmers University of Technology. The calculations were facilitated by the resources of the national academic infrastructure for the Supercalculculculculculculcul (NISS) in the C3SE.

Never miss a breakthrough: join the Scitechdaily newsletter.