Accelerating mean-field phase diagram construction of multicomponent block copolymer blends

Blending block copolymers (BCPs) with homopolymers has been shown to yield novel phase behaviors and properties, such as bicontinuous microemulsions and "bricks-and-mortar" mesophases. Accurately computing phase diagrams of BCP blends remains a significant challenge due to the complex coupling of macrophase and microphase separations. Traditional approaches like the double-tangent construction or self-consistent field theory (SCFT) calculation in grand canonical ensemble are computationally expensive or difficult to implement. We present a novel surrogate-based method to optimize the phase diagram construction by treating SCFT as a black-box function for free energy evaluation. Our method leverages the gradient of free energy to improve surrogate accuracy and is versatile, working with various numerical methods. This approach significantly reduces SCFT evaluations, offering a more efficient strategy for computing phase diagrams in multicomponent polymer systems.