GPU-Accelerated Three-Dimensional Polymer Density-Functional Theory Calculations of Block Copolymer Self-Assembly*

Owing to its well-developed numerical techniques, polymer self-consistent field theory (SCFT) has been widely applied to various inhomogeneous polymeric systems. In particular, it can be considered as the most successful molecular-level theory for block copolymer (BCP) self-assembly. Its inherent mean-field approximation, however, neglects both the fluctuations and correlations in the system. While the latter can be captured by polymer density-functional theories (DFTs), their applications have been hindered by the inefficient numerical techniques used. For example, 3D DFT calculations of BCP self-assembly have never been reported. Here we report such calculations, accelerated by the massive parallelization of GPU, for a DPD model of conformationally asymmetric diblock copolymer melts¹ using a variant² of the CMS polymer DFT. Direct comparisons with our recent SCF calculations¹ of the same model quantitatively reveal the effects of correlations on the stability of Frank-Kasper phases neglected by SCFT.