"Today, subduction dominates the Earth’s appearance: it drives plate tectonics, and plays a dominant role in continental crust formation. If and how subduction operated 2.5-4 billion years ago, in the Archaean, is debated, primarily on the basis of the sparse Archaean geological record. It seems likely that some form of subduction occurred at least by the late Archaean, but may well have looked different from today’s. A proper understanding of this Archaean ‘subduction’ is essential, since so many processes are likely to depend on it.Observations of the geological (mostly isotope-geochemical) record have provided an invaluable window to peer into the Archaean world. But inferred Archaean geodynamics from these observations are non-unique. Various models fit the same data within uncertainty, and often lack a firm physical basis. To overcome these shortcomings, I propose a novel, forward approach of predicting synthetic geochemical fingerprints from numerical, geodynamically consistent physical models, and comparing those with geochemical observations. This will be used to constrain and better understand the two most pressing questions in Earth sciences: How did plate tectonics evolve, and how did continents form? In particular, this project aims to:1) assess quantitatively the geodynamical and geochemical viability of intermittent plate tectonics;2) test the various proposed models for the formation of Archaean continental crust;Comparison of calculated synthetic geochemistry (e.g. Re-Os data, rare-Earth element data) from geodynamical models with available datasets will provide powerful diagnostics to distinguish viable models.In addition, this work will also directly relevant for the evolution of the Earth’s surface, and to the differences with the other terrestrial planets. Finally, there are potential economic benefits, since the world’s largest mineral deposits (e.g. gold) occur in Archaean terrains and have been associated to subduction."