Spelling suggestions: "subject:"astraining engineering"" "subject:"a.training engineering""
1 |
Characterizing and evaluating 2D material properties using spectroscopic methods and machine learningChen, Zhuofa 23 May 2022 (has links)
Atomically thin two-dimensional (2D) materials come in all necessary flavors to make semiconductor devices: conductors, semiconductors, and insulators. Graphene, transition metal dichalcogenides (TMDCs), and hexagonal boron nitride (hBN) are the quintessential building blocks. The van der Waals nature of the bonds in 2D films allows the ability to stack materials to achieve novel properties because of their exceptional mechanical, electronic, and optical properties and interactions, which enables various applications of 2D materials in transistors, biosensors, light-emitting devices, and photodetectors. Spectroscopic measurements such as Raman and photoluminescence (PL) reveal a wealth of information since 2D materials are affected by their environment and other local perturbations, e.g., strain and charge doping. My research focused on developing efficient and noninvasive optical methods to evaluate and characterize the properties of 2D materials. In particular, we investigated strain-tunable properties, the effects and signature of charge doping, and the environmental screening properties of graphene and TMDCs.
Identifying the charge density and impurities in graphene is vital for graphene-based applications, which require high-quality graphene. I developed an effective optical method to determine the doping level and the local charge density variations in graphene before any fabrication process. This method differentiates charge density variations in graphene via the Raman 2D peak asymmetry that manifests at low charge 1-25 × 1010 cm-2. We explore the effect of charge inhomogeneity, "charge puddles", within the laser spot using simulated Raman 2D spectra, revealing a different signature for large or small charge puddles. Our work provides a simple and noninvasive optical method for estimating the doping level, local charge density variation, and transport properties of graphene, with up to two orders of magnitude higher precision than previously reported optical methods.
Strain is another crucial factor that significantly impacts the properties of 2D materials. We studied the charge distribution and radiative efficiency of excitonic complexes in strained monolayer TMDCs, especially WSe2. Straining and electrostatic gating are combined to investigate the dynamics of quasi-particles in WSe2. We found that negative trions accumulate while positive trion emission is near zero, indicating that both conduction and valence bands are bent downwards in the strained area. Finite element analysis of strain distribution and density functional theory calculations of band structures of WSe2 support the experimental results. Hence, localized strain allows locally separating electrons and holes in WSe2 and manipulating light-matter interaction for applications in novel strained-engineered optoelectronics.
I applied machine learning and deep learning techniques to improve the efficiency and accuracy of data processing and analysis since traditional methods require domain expertise and have the potential to introduce artifacts. I categorized the wealth of information and data by applying machine learning to spectroscopic information to separate different influences, e.g., strain, charge doping, and dielectric environment. We developed deep learning models to classify graphene Raman spectra according to different charge densities and dielectric environments. To improve the accuracy and generalization of all models, we use data augmentation through additive noise and peak shifting. Using a convolutional neural net (CNN) model, we demonstrated the spectra classification with 99% accuracy. Our approach has the potential for fast and reliable estimation of graphene doping levels and dielectric environments. The proposed model paves the way for achieving efficient analytical tools to evaluate the properties of graphene. / 2022-11-23T00:00:00Z
|
Page generated in 0.1531 seconds