Design and Realization of High Performance Textured Lead-Free Piezoelectric Ceramics through Human-AI Collaboration

Abstract Developing multi-element Pb-free (K,Na)NbO₃ (KNN) piezoelectric compositions is constrained by non-comprehending complex doping trends as well as tedious trial and error based investigative experimentations. Here, we present a human in the loop, artificial intelligence (AI) guided materials design framework which utilizes large language models (LLMs) to extract structure - property knowledge from prior KNN chemistries to generate application specific compositions. Further, expert intervention directs experimental realization based on materials science principles and experiential knowledge, enabling more efficient discovery of targeted compositions. Using this collaborative approach, a random composition was synthesized that exhibited d 33 of 440–500 pC/N. Further enhancement was achieved through crystallographic texturing and sintering aid, enabling the designed composition to exhibit d 33 of 600–620 pC/N. This composition manifested a steady magnitude of electromechanical coupling ( k 31 and k p ) up to the Curie temperature (T c ~ 180°C). To validate practical relevance, a cantilever-based magneto-mechano-electric (MME) energy harvester was designed and evaluated at the second harmonic. The MME harvester with textured KNN composition exhibited power density ~ 705µW/cm 3 , outperforming its random counterpart as well as reported Pb-free MME designs. These results demonstrate an exceptional approach towards developing application-specific functional materials through synergy of AI-assistance and experimental optimization.