When less is more: simplified models for interpretable materials design

Simplified models have long played an important role in materials modelling, from hard-particle and lattice models to coarse-grained descriptions of molecules, polymers, colloids, and frameworks. The value of simplified models is not just computational efficiency. Reducing microscopic degrees of freedom while retaining macroscopic behaviour can reveal which physical and chemical ingredients are responsible for a material's structure and properties. Simplified models can therefore occupy a middle ground between detailed simulation, which seeks to represent specific materials accurately, and abstract theory, which seeks general principles. In this Perspective, we discuss how simplified models can be used to understand emergent structure in materials. We consider the role of simplified models in deriving design principles across materials classes and their limitations. We argue that the most useful simplified models are tools for organising physical insight as they identify control variables, map regimes of behaviour and suggest which aspects of chemistry, geometry, or architecture are likely to matter for materials design. Their value for materials design therefore depends on identifying favourable regions of model space, and establishing how those regions can be accessed through chemical and synthesis conditions.