3D Scanning Photocurrent Nanoscopy and Applications

After graphene ignited the field of two-dimensional (2D) van der Waals (vdW) materials, other 2D materials, such as transition metal dichalcogenides (TMDs), have emerged as attractive integration elements for novel heterostructures and offered extraordinary optoelectronic application potential. In addition, plasmonic resonances typically play a unique role in efficiently harvesting incident light due to their ability to manipulate electromagnetic waves on a deep-subwavelength scale. These capabilities are utilized in nano-electronic devices to further enhance their exceptional performances, such as high speed, high sensitivity, and low power consumption. To exploit the full potential of a variety of photodetector/photoconversion devices, it is crucial to clarify the complex nature of the photoresponse mechanisms at the nanoscale. In addition, a rapid and high-throughput characterization method for plasmonic metal nanoparticles is required for application design.This thesis covers two of my most significant contributions: 1. Photocurrent mechanisms in a van der Waals metal-semiconductor junction In this part, we report the photocurrent generation mechanism distribution in a Schottky barrier photodiode consisting of a defect-free molybdenum disulfide-gold (MoS2-Au) interface. We find that the in-plane PTE current generated by the near fields of a nanoscale light source can be four-folds stronger than the PV current, which is the working horse of a conventional Schottky barrier photodiode. Furthermore, we show that the spatial variation of the PTE effect in amplitude and polarity can be controlled through nanoscale heat management, by adding a thin hexagonal boron nitride layer (h-BN, 5 nm) onto the MoS2. This study points out a promising way for high-efficiency integrated photodetectors. 2. Machine learning-enhanced rapid size quantification for nanoparticles we develop a rapid optical characterization method that integrates through-focus scanning microscopy (TSOM) with machine learning (ML) to directly measure the size of individual nanoparticles and the size distribution of a nanoparticle ensemble. With silver nanocubes (AgNC) as a model system, we achieved an estimation error of 5% and a root mean square error (RMSE) of 8.2 nm on individual particles, as well as an error of 1.6 % on the average particle size and 0.4 nm on the size distribution on particle ensembles. The single particle level measurement allows the measurement of complex mixtures with broad size distributions, which have size-dependent features smeared in other rapid measurements such as UV-vis absorption spectroscopy. This work describes a ML-assisted single particle-level optical characterization for the size information and can be potentially extended to more complicated nano-materials such as anisotropic and dielectric nanoparticles