Accelerated Discovery of Sustainable Materials for the Semiconductor Industry
Advances in memory technologies, novel compute architectures, artificial intelligence, quantum computing, robotic automation, and cloud technologies are poised to further accelerate discovery, drive profound transformation, and help enable a sustainable future. However, can these same technologies allow the semiconductor industry to address its own key sustainability challenges to ensure a sustainable future for computing?
This talk will update some recent efforts that we had previously reported at the 2021 Strategic Materials Conference on of discovering more sustainable cationic onium compounds for use as photoacid generators [1-3]. Since then, the world has seen the explosive development of AI, and in particular, large language models and other so-called foundation models. At the same time, our work has shifted to focus on tackling the challenge of accelerating the discovery of replacement alternatives for per- and polyfluoroalkyl substances (PFAS). In this talk, we will discuss the application of tuned chemistry foundation models to that task of predicting environmental hazard properties [4] and generative workflows [5] to design new fluorine-free superacid moieties. These will be discussed in the context of building new tools and workflows to enable efficient Human-AI Co-creation [6,7].
1) Pyzer-Knapp, E.O.; Pitera, J.W.; Staar, P.W.J.; Takeda, S.; Laino, T.; Sanders, D.P.; Sexton, J.; Smith, J.R.;Curioni, A. "Accelerating materials discovery using artificial intelligence, high performance computing and robotics," npj Comput Mater 2022, 8, 84. https://doi.org/10.1038/s41524-022-00765-z
2) https://www.research.ibm.com/science/photoresist/
3) Hoffman, S.C.; Chenthamarakshan, V.; Zubarev, D.Y.; Sanders, D.P.; Das, P. “Sample-Efficient Generation of Novel Photo-acid Generator Molecules using a Deep Generative Model,” NeurIPS, Deep Generative Models and Downstream Applications, 2021. https://doi.org/10.48550/arXiv.2112.01625
4) Soares, E.; Brazil, E. V.; Gutierrez, K. F. A.; Cerqueira, R.; Sanders, D.; Schmidt, K.; Zubarev, D. "Beyond Chemical Language: A Multimodal Approach to Enhance Molecular Property Prediction," Poster, AI4Science Workshop, 37th Conference on Neural Information Processing Systems (NeurIPS), New Orleans, LA, United States, (2023). https://doi.org/10.48550/arXiv.2306.14919
5) Soares, E.; Cipcigan, F.; Zubarev, D.; Brazil, E.V.; “A Framework for Toxic PFAS Replacement based on GFlowNet and Chemical Foundation Model,” Poster, AI4Science Workshop, 37th Conference on Neural Information Processing Systems (NeurIPS), New Orleans, LA, United States, (2023). https://openreview.net/forum?id=oIMhS5n0Qz
6) Zubarev, D.; Mendes, C.R.; Brazil, E.V.; Cerqueria, R.; Schmidt, K.; Segura, V.; Ferreira, J.J.; Sanders, D.P. “Toward Human-AI Co-creation to Accelerate Material Discovery,” Poster, AI4Science Workshop, 36th Conference on Neural Information Processing Systems (NeurIPS 2022), New Orleans, LA USA, 2 Dec 2022. https://doi.org/10.48550/arXiv.2211.04257
7) Ferreira, J.A.; Segura, V.; Souza, J.G.R.; Barbosa, G.D.J.; Gallas, J.; Cerqueira, R.; Zubarev, D. “Human-AI Co-Creation Approach to Find Forever Chemicals Replacements,” In GenAICHI Workshop at CHI 2023, April 28, 2023, Virtual Workshop. https://doi.org/10.48550/arXiv.2304.05389
BIOGRAPHY
Dmitry Zubarev, Ph.D. is a theoretical/computational chemist leading Accelerated Materials Discovery effort at IBM Research - Almaden. He received PhD from Utah State University and worked as a post-doctoral researcher at UC Berkeley and Harvard University on a range of problems, including chemical bonding models, computational photoelectron spectroscopy and X-ray absorption spectroscopy of molecules, stochastic projection methods in electronic structure theory, and uncertainty quantification in quantum chemistry. As a postdoctoral fellow with Simons Collaboration on the Origins of Life, he developed methods of computational reconstruction of prebiological reaction networks. His current efforts include development of AI for end-to-end discovery of materials with focus on agent-centric AI systems, and multi-agent discovery systems.