The goal of the proposed project is to develop a machine-learning (ML) driven computational platform for modeling vapor-based synthesis of multi-component thin-film materials. This entails a great challenge: state-of-the-art methods—including density functional theory (DFT), molecular dynamics (MD), and kinetic Monte-Carlo (kMC)—cannot simultaneously achieve the atomistic accuracy and access the time scales required for attaining the project goal. We will overcome this challenge by integrating DFT and kMC in a single computational workflow and leverage their distinct advantages through multiple layers of ML which will allow the model to self-evolve via active learning. We will deploy our computational concept for simulating growth of thin noble-metal films—comprising silver (Ag), gold (Au), and copper (Cu)—on graphene. This film-substrate combination is relevant for a wide array of devices. Moreover, the ternary Ag-Au-Cu alloy is an ideal system for unraveling—for the first time with experimentally-consistent atomic-scale precision—the way by which the interplay between thermodynamics (i.e., interaction strength between constituent species) and kinetics (i.e., the atomic migration rates) control structure formation in multi-component materials synthesized far from equilibrium. Vapor-deposited thin noble-metal films (e.g., Ag and Au) frequently serve as multifunctional contacts in graphene-based devices, whereby the resulting device performance crucially depends on film morphology. In some cases, it is important that the film wets the underlying 2D substrate forming a flat two-dimensional (2D) layer, while in other cases synthesis of three-dimensional (3D) metal nanostructures with well-defined sizes and shapes is required. Achieving the desired 2D or 3D film morphology is a non-trivial task, as noble-metal atoms exhibit a weak interaction (i.e., bond strength) with graphene, leading to uncontrolled formation of 3Dmetal agglomerates. Hence, theory-informed strategies for manipulating film morphological evolution on weakly-interacting substrates are a key component for leveraging the unique physical properties of graphene and other van der Waals materials in novel heterostruture devices. The goal of this project is to build upon the knowledge we have generated and the results we have achieved and continue modelling initial and late stages of noble-metal (Ag, Au, Cu,) film formation on model graphene surfaces. Our previously developed kMC code will be used as starting point and it will refined and augmented as detailed below: (1) We will use DFT calculations to determine surface adsorption energies and diffusion barriers of metal adatoms on graphene in for a variety of local atomic enviroments. (2) We will also employ non-equilibrium ab initio molecular dynamics simulations (modified version of VASP developed by us (Sangiovanni et al. PhysRevB 93, 094305 (2016)), to investigate atomistic processes and dynamics during the initial stages of film island nucleation and growth and improve the accuracy of KMC simulations. (3) We will describe surfactant-modified growth of noble-metal layers on weakly-interacting substrates by coupling our previous KMC algorithm with machine-learning methodologies.