Résumé
2D materials have shown fascinating fundamental physics and exciting technological prospects. Many applications rely on the ability to conduct electrons efficiently. At room temperature, this is mostly limited by the scattering of electrons by phonons. A clear understanding of transport mechanisms, along with predictive ab initio simulations, are then key to design performant and energy-efficient devices.The prototypical 2D transport setup is an electrostatically-doped semiconducting layer encapsulated by a dielectric material. We will discuss 3 major challenges to model electron-phonon scattering in this system: dimensionality, electrostatic doping and interlayer couplings. The first 2 concern the proper simulation of electron-phonon interactions in single 2D layers. The third one accounts for the additional scattering mechanisms between layers in van der Waals heterostructures. Combining solutions to all 3 challenges paves the way towards a full description of electron-phonon scattering and electronic transport in 2D heterostructures.Reduced dimensionality has strong, universal impact on all materials, notably via changes in the long wavelength behaviour of the Fröhlich interaction. Electrostatic doping refers the ubiquitous use of field effects to induce charges in 2D materials and tune their Fermi level. The added carriers screen electron-phonon scattering. We will see how to include this effect in doped semiconductors. Usually, it tends to decrease scattering, but we will also discuss a counter-intuitive enhancement mechanism in multi-valley materials like transition-metal dichalcogenides. Interlayer couplings refers the remote interactions between electrons and phonons in different layers. They are very too costly to simulate directly in first-principles codes due to the size of the associated supercells. We will discuss a newly developed model to compute long-range Coulomb interactions between electrons and polar excitations including phonons and plasmons in van der Waals heterostructures, at a reasonable relatively negligible computational cost.