Quantifying the effects of noise on early states of spinodal decomposition:

An investigation into the dynamics of spinodal decomposition under varying levels of noise is presented. Obtaining a clear understanding of how noise affects the short-term dynamics of this phenomenon is critical for accurate model development and subsequent efforts for morphology control. Following the well-known Cahn–Hilliard model for phase separation, we present an efficient and scalable finite element–based framework to study the dynamics of spinodal decomposition. This framework is massively parallel with good scalability up to 100k CPUs. In addition, we formulate a higher-order time scheme and discuss thermodynamically consistent numerical treatment of the added noise term through the use of a scalable, parallel random number generator. Then, we perform a high-throughput analysis of over 6400 independent simulations and examine the interplay between noise amplitude and various numerical features critical to model development (discretization, domain size, and boundary conditions). Our numerical results suggest that a proper treatment of the stochastic term is required to study the influence of noise on the initiation of phase separation. This opens avenues for microstructure control and has important implications for materials-by-design.