Researchers at Seoul National University have developed an artificial intelligence system capable of transforming materials that are difficult to synthesize into alternative structures that can be produced experimentally, potentially accelerating the development of advanced materials for electronics and energy applications. The study was published in the Journal of the American Chemical Society.
The AI materials redesign framework uses large language models (LLMs) to modify predicted crystal structures that are impractical for laboratory synthesis, converting them into new configurations that retain desirable properties while improving the feasibility of materials synthesis.
This approach targets a major challenge in modern materials research. Computational screening methods can identify large numbers of promising candidates, but many predicted materials remain experimentally inaccessible due to synthesis constraints. AI materials redesign offers a strategy to recover these candidates rather than discarding them.
Crystal structure design enables practical materials synthesis
The system, called SynCry, represents crystal structures as invertible textual descriptions and iteratively refines them to improve synthesizability through a process described as “learn and redesign.” This method enables the AI to explore new pathways for crystal structure design that preserve functional properties while increasing the likelihood of successful materials synthesis.
By shifting from prediction to transformation, the technology moves artificial intelligence closer to practical laboratory applications. Instead of identifying only theoretical materials, AI materials redesign can generate experimentally viable candidates that researchers can evaluate in laboratory settings.
Validation shows AI can produce experimentally viable materials
The researchers initially achieved 514 successful structure transformations and, after iterative refinement, redesigned 3,395 materials into synthesizable forms. Among the top 100 redesigned structures, 34 matched materials that had already been experimentally synthesized and reported in scientific literature, despite not being included in the model’s training dataset.
These results suggest the AI materials redesign approach can produce genuinely new viable materials rather than simply reproducing known examples.
Professor Yousung Jung, who led the research, emphasized the significance of the work. “This study is the first to demonstrate that AI can directly redesign new materials starting from structures that are difficult to synthesize. We plan to expand this work to a wider range of material systems and larger datasets, ultimately developing a practical tool for discovering new materials.”
Implications for semiconductor and energy materials synthesis
The researchers expect the AI materials redesign technology to accelerate the development of next-generation semiconductor and high-efficiency battery materials by recovering candidates previously discarded due to synthesis challenges.
Improved crystal-structure design workflows could shorten development timelines and expand the pool of viable materials available for experimental validation, thereby supporting laboratory research programs focused on functional materials discovery.
What this means for laboratory workflows
For laboratory managers and research organizations, AI materials redesign could influence several aspects of experimental workflows:
- Prioritization of candidate materials for synthesis
- Resource allocation for experimental validation
- Integration of computational tools with laboratory research
- Collaboration between computational and experimental teams
- Optimization of crystal structure design processes
As artificial intelligence tools continue to evolve, their role in bridging the gap between computational prediction and laboratory materials synthesis may become increasingly important for accelerating scientific discovery.
This article was created with the assistance of Generative AI and has undergone editorial review before publishing.
