Key takeaways:
UNE’s ISA team is using AI and supercomputing to identify suitable solvents for separating silicon from PV modules
The approach replaces trial-and-error lab testing with simulations to identify effective chemical pathways
Improving silicon recovery is important as PV waste volumes grow and material value increases
Recycling of PV modules is a hot topic in the solar industry, with several companies globally operating recycling infrastructure. Their methods vary, however, as do the efficiency of the recycling and purity of the recovered materials. Currently, 95% of the PV module’s mass can be recycled. However, the second-most valuable material, silicon, cannot be recycled with more than 95% purity, according to Germany’s Solar Materials. Moreover, the silicon wafer cannot be retrieved from the module.
A research team at the Institute for Strategic Artificial Intelligence (ISA) at the University of New England (UNE) is working towards using AI and supercomputers to evaluate the potential of separating silicon with minimal contamination. The idea behind this approach is to avoid laborious trial-and-error testing and to identify the right methods and materials to achieve this. The researchers plan to replace traditional methods with AI-driven chemical quantum simulations to model the molecular behavior of solvents, propose useful molecular formulations, evaluate their efficacy, and identify new pathways. The AI is used to identify suitable molecules that can act as solvents for cleanly separating silicon wafers. This method also involves a lot of computing; however, it avoids the need to physically evaluate and test multiple materials in a laboratory.
With global PV production capacity increasing and projected to reach 1 TW/year by 2030, Australia alone is expected to generate 1 million tons of end-of-life panels by 2035. The material value is projected to be over AUD 1 billion. The institute has already started collaborating with developers in Australia who would provide panels for this project. ACEN Australia is one developer; the company’s Managing Director, David Pollington, said that it is committed to PV recycling. He added that this research will improve the effectiveness and efficiency of the recycling process.
Prof. Amir Karton, a member of the research team, said that this system can help establish an effective feedback loop between AI-driven predictions and experimental observations. This allows researchers to steer the experimental discovery of optimal recycling pathways at unprecedented speeds. He further added, “The discovery process is being driven through pairing of an AI platform based at UNE, and AUD 2.7 billion ARC-funded automated robotic laboratory shared by several institutions.”