11. Domain-Informed Bayesian Hierarchical Modeling of Nanowire Growth at Multiple Scales

This chapter delves into the complexities of nanowire (NW) growth processes and presents a innovative approach to predict growth at multiple scales. The authors introduce a domain-informed Bayesian hierarchical modeling framework that integrates available data and physical knowledge, addressing the challenges posed by scale effects and local variability. The model consists of two main components: NW morphology, which represents the overall growth trend characterized by growth kinetics, and local variability, which captures the less understood area-specific variations. The authors employ an intrinsic Gaussian Markov random field (IGMRF) to model local variability, separating it from the growth kinetics in the morphology component. The chapter provides case studies illustrating the NW growth process model at both coarse and fine scales, demonstrating the effectiveness of the proposed methodology. By addressing the limitations of current deterministic kinetics models and the challenges of nanomanufacturing quality control, this chapter offers a comprehensive strategy for modeling and controlling nanowire growth processes, ultimately aiming to improve process yield and productivity.