Predicting the Interface

In this thesis, a new methodology for designing, evaluating and comparing interface structures is presented. First, a history and background surrounding the search for materials over human history paints a picture of the societal impacts of materials discovery. This history concludes with an overview of the current forefronts of the exploration of materials space, with particular consideration paid to contrained systems such as surfaces, 2D materials and interfaces. It is recognised that interfaces play a key part in the functionality of devices, but their properties remain largely unknown due to difficulties in both the practical and theoretical sense. Having identified that the interface presents a wealth of opportunities for materials discovery, we then explore methods by which the properties of interfaces could be predicted from a theoretical standpoint, and arrive at the first principles method density functional theory. The development of this method from first principles is presented, along with a discussion of the difficulties in implementing such methods to constrained systems such as interfaces. A significant highlight presented is that many structural prediction methods adopt high throughput approaches, which in the case of interfaces would often result in prohibitively expensive studies. Armed with these conclusions, a method for sensibly predicting the structure of an interface is laid out, termed RAFFLE, which utilises a mixture of the aforementioned first principles techniques, alongside empirically inspired functions and machine learning. This method is applied to an increasingly complex set of materials, culminating in the study of the Si-Ge interface and the graphene-MgO-graphene nanoencapsulated interface, identifying interesting phases in the case of the latter, identifying phases present in the literature, and predicting further their nature. Two other examples are then presented of the RAFFLE method. Firstly, the consideration of transition metal dichalcogenides (TMDC) interfaces for use as photocatalytic water splitters, in which RAFFLE is applied to study the stability of the structures to substitution, identifying MoS2 to be vastly more stable to geometric reconfiguration than \pods. Secondly, the consideration ofScS2 as an ion battery material is supported by a RAFFLE search, which identified the most stable sites for intercalation and verifies that the most stable intercalated ground state structure found is a layered material suitable for cycling. These 3 cases, C-MgO, TMDC and ScS2 demonstrate the versatility of RAFFLE for various material across a multitude of applications.