PhD Thesis Defense: Interface Engineering of Magnetism in Two-Dimensional van der Waals Heterostructures

Feynman’s 1959 lecture There’s Plenty of Room at the Bottom imagined a world where we manipulate matter and information atom by atom. In the decades that followed, semiconductor device miniaturization and heterostructure engineering largely filled that “room”, culminating in Kroemer’s 2000 Nobel lecture dictum that “the interface is the device”. The mechanical exfoliation of graphene in 2004 then extended this interface-centric view into truly two-dimensional (2D) van der Waals (vdW) crystals, where electronic and magnetic properties are defined by stacks of atomically sharp interfaces. Building on this 2D materials landscape, this thesis investigates how interfaces in vdW heterostructures can be engineered to tune and electrically read out magnetism in two dimensions. Using the metallic ferromagnet Fe3GeTe2 (FGT) as a model system, we develop two interface knobs. The first is a chemical route, in which metal phthalocyanine (MPc) overlayers program exchange bias via their spin state and bonding. The second is an electro-ionic route, in which a native oxide layer (O-FGT) in all-solid-state FGT/O-FGT/hBN stacks yields gate-tunable exchange bias through oxygen-ion migration and enables deterministic, voltage-controlled magnetization reversal. As the field evolves, a quarter of a century after Kroemer’s dictum it is pushing beyond the view of the interface as a static boundary and asking how its very nature can be redesigned. Twistronics has shown that two layers need not simply meet but can be rotated into new electronic and magnetic phases, with superconductivity in magic-angle graphene as a striking example. In this spirit, we use the semiconducting magnet CrSBr (CSB) in twisted hBN/CSB(1L)/hBN heterostructures to go beyond the notion of a single ferromagnetic monolayer. We demonstrate a sub-monolayer spin-valve effect whose internal sublayer magnetization can be read out all-electrically via a second-order nonlinear response.