UC Berkeley

Quasi-two-dimensional (quasi-2D) semiconductors are the subject of intense research interest as platforms for both developing atomically-thin devices and exploring novel physics. These are atomically-thin structured derived from layered materials with covalent bonding in each layer and weak coupling between layers so that individual monolayers can be easily peeled off. Changes in confinement and screening in reduced dimensions can lead to drastic changes in quasiparticle (QP) and optical properties when compared with bulk materials. In this dissertation, we use ab initio methods based on many-body perturbation theory (MBPT) to explain and predict the QP and optical properties of quasi-2D semiconductors. We describe the QP properties within the GW approximation, for single particle excitations, and the optical properties within the GW plus Bethe Salpeter equation (GW-BSE) approach, for two particle excitations, such as correlated electron-hole pairs, or excitons. We find that, in general, reduced dimensionality results in strong spatial variations in the screening of the Coulomb interaction as well as reduced total screening, allowing for strongly-bound, non-hydrogenic excitons. We then use this understanding to explore how changing the external screening environment through layering and substrate engineering can be used to tune properties of quasi-2D materials. Finally, we perform ab initio calculations of exciton dispersion in quasi-2D and show that the 2D Coulomb interaction can give rise to unusual, massless excitons. This dissertation is organized as follows:

In chapter 1, we introduce the theoretical and computational techniques used to compute the QP and optical properties of real materials.

In chapter 2, we calculate the QP bandstructure and optical spectrum of monolayer MoS2, a prototypical quasi-2D semiconductor. We find that monolayer MoS2 has very strongly-bound excitons, with binding energies that are two orders of magnitude larger than in the bulk. Moreover, we show that these excitons have excited states that do not follow the commonly used 2D hydrogenic model and that this strongly-bound, non-hydrogenic behavior arises as a consequence of fast spatial variations in the screening environment. For isolated systems, the screening is in fact diminished to zero in the limit of large distances compared to the layer thickness.

The inhomogeneous screening found in quasi-low-dimensional systems introduces significant computational challenges because very fine spatial sampling is required to capture this variation in screening. In chapter 3, we develop new computational techniques to sample reciprocal space more efficiently, resulting in more than two orders of magnitude reductions in computational time for GW and GW-BSE calculations on low-dimensional systems.

Based on our understanding of screening in MoS2, we then explore how screening from a substrate or capping layer might change the electronic and optical properties in other quasi-2D materials. In a joint experimental study, we find that a bilayer graphene substrate can renormalize the QP band gap and exciton binding energies in monolayer and few-layer transition metal dichalcogenides (TMDs) by as much as 30%. This work is discussed in chapters 4 and 5.

In chapter 6, we discuss how the substrate screening effect is even more pronounced in black phosphorene, which has a lower intrinsic screening than TMDs. Consequently, encapsulation, which is commonly used to stabilize the material under ambient conditions, can completely eliminate the presence of bound excitons in few-layer systems, changing the optical spectrum qualitatively as well as quantitatively.

Traditional implementations of the GW-BSE approach focus on excitons with zero center-of-mass momentum (Q), since these are the states probed by linear optical probes. However, exciton dispersion--energy as a function of Q--is essential to understanding properties such as exciton dynamics, lifetimes, and the formation of novel phases such as exciton condensates. In chapter 7, we implement GW-BSE for finite Q, and calculate the exciton bandstructure of MoS2. We find that, in general, the exchange interaction in 2D gives rise to a massless dispersion, while in MoS2 and other TMDs in particular, the interplay of the intervalley and intravalley exchange gives rise to a parabolic band and a massless, v-shaped band with a valley quantum phase of winding number two.