Symmetry-oriented enumeration method for molecular crystal structure prediction and its application

Predicting the thermodynamically stable arrangements of molecular building blocks is crucial for understanding the properties and behaviors of crystalline molecular compounds. Several molecular structure prediction methods have been proposed, leading to successful applications. However, some evolutionary approaches that involve direct sampling within the infinite atomic coordinate space, without symmetry-preserving constraints, tend to suffer from biased sampling, where structures with $P1$ symmetry become disproportionately dominant. To overcome this limitation, we propose a method that utilizes a finite set of symmetrically inequivalent Wyckoff encodes to sparsely represent the search space of crystalline molecular compounds, thereby improving the coverage of the search space. This method can be used to predict the molecular crystal structures with varying numbers of molecules in the asymmetric unit, from fractional to multiple integers. The applications of this method to three-dimensional molecular crystals of ammonia dihydrate, the ${\mathrm{NH}}_{3}\text{\ensuremath{-}}{\mathrm{H}}_{2}\mathrm{O}\text{\ensuremath{-}}{\mathrm{CH}}_{4}$ mixture, and two-dimensional layered ices revealed a series of energetically quasidegenerate structures with diverse crystal packing motifs. This method expands the scope and utility of molecular building blocks for designing stable molecular crystals, providing a robust framework for achieving extensive tunability in molecular materials design.