AI Model for Accurate Prediction of Dielectric Constants in Materials

Abstract

Accurate prediction of dielectric constants is essential for the design and optimization of semiconductor devices, energy harvesting technologies, and sensing applications. In this presentation, I will introduce an artificial intelligence (AI) model developed to predict the dielectric constants of a wide range of materials, including traditional semiconductors, ferroelectrics, and advanced materials. Using machine learning algorithms, I have created a model that predicts dielectric constants based on material composition, crystal structure, and other key properties. The model is trained on an extensive dataset of dielectric constants obtained from high-throughput calculations and experimental data. One of the key features of our model is its ability to predict dielectric constants reliably across a broad range of frequencies, regardless of material anisotropy. This makes it applicable to a wide variety of materials with different dielectric behaviors, including those with complex frequency-dependent responses or directional variations in their dielectric properties. I will present the model’s performance, showing rigorous cross-validation and comparisons to both computational results and experimental measurements. The AI model demonstrates high accuracy, even for materials that are less studied or for novel materials like graphene and nanowires. This work highlights the potential of AI-driven models to accelerate the discovery and design of new dielectric materials, reducing the need for time-consuming experimental trials. The model is particularly valuable for applications in high-temperature environments, energy-efficient devices, and advanced sensing technologies, where dielectric properties are crucial. By enabling fast and reliable predictions, this AI model offers a powerful tool for materials discovery and device optimization in modern electronics.