Proceedings of Asia-Pacific Conference on Perovskite, Organic Photovoltaics&Optoelectronics (IPEROP25)
Publication date: 17th October 2024
Optimization of a photovoltaic device is a typical example of optimizing a complicated chemical network system. Herein, two chemical networks are involved: one consists of component materials, and a series of optimizations of each component does not ensure the overall device optimization. Another network consists of a series of experimental procedures after choosing the component materials, where each is pertinent to the morphologies of the materials fabricated therein. The number of combinations in materials and experimental procedures is infinite and may be a typical example of a “combinatorial explosion”. Although a variety of impressive and successful design principles have been introduced so far and resulted in rapid and steady improvements have been achieved, we are still challenging the problem of finding “a needle in a haystack”. While continuous updates of the device fabrication processes and materials are strongly demanded in such orthodox and deductive optimizations, which may be called “a normal design”, recent progress in computational techniques seems to provide new opportunities in the system optimizations through "inverse designs" or "virtual (i.e., computational) screenings". In this presentation, we will introduce and discuss our recent efforts toward optimizing the perovskite solar cell through applications of various computational techniques, including device simulations, atomistic calculations, generative AI models, and image regression to predict device performance.
The present study has been performed in a project, JPNP21016, commissioned by the New Energy and Industrial Technology Development Organization (NEDO).