Event
Ph.D. Dissertation Defense: Ziyang Xiao
Friday, October 26, 2018
10:00 a.m.
ERF 1205 (IREAP)
Maria Hoo
301 405 3681
mch@umd.edu
ANNOUNCEMENT: Ph.D. Dissertation Defense
Name: Ziyang Xiao
Committee:
Professor Neil Goldsman, Chair/Advisor
Professor Agis Iliadis
Professor Pamela Abshire
Professor Kevin Daniels
Professor Aris Christou, Dean's representative
Date/Time: Friday, October 26th, 2018 at 10:00am
Location: ERF 1205 (IREAP)
Title: Modeling Key Issues in Post Silicon Semiconductors: Germanium and Gallium Nitride
Abstract:
We are rapidly approaching the end of the semiconductor roadmap with respect to silicon. To continue its growth, the semiconductor industry is therefore looking into new materials. Two primary materials that are of interest for continued semiconductor development are germanium (Ge) and gallium nitride (GaN). Ge is of interest as a replacement for silicon, in an effort to increase electronics performance because of its high mobility and its ability to grow a native oxide. In addition, Ge is of interest because of its potential use for economical CMOS-based short wave infrared (SWIR) imaging systems. GaN is a nascent wide bandgap semiconductor and has many potential applications in high power electronics and ultraviolet imaging systems. In this thesis, the key material properties and applications of these two “end of the roadmap” semiconductors are explored.
Ge is an indirect bandgap material with an indirect bandgap of 0.66eV. This bandgap value corresponds to a wavelength of 1.88µm, which lies in the infrared range. The Ge material itself is also compatible with the standard Si CMOS process technology. Because of these advantages, Ge is considered a candidate for the application of photo detecting in the SWIR range. Apart from the indirect bandgap of 0.66eV, Ge also has a direct bandgap of 0.8eV. From early research, the relatively small offset between the indirect and direct bandgap can be inverted either by applying strain or alloying with tin.
GaN is a binary direct wide bandgap material with a direct bandgap of 3.4eV. It has a high breakdown voltage, and relatively high saturation velocity and carrier mobility. These properties give GaN an advantage in the realm of high power application. GaN can also form a heterostructure with AlGaN, which can form a 2D electron gas (2DEG) layer at the interface without intentionally doping either material. The 2DEG layer has an even higher mobility compared to the mobility of the bulk GaN, which allows the heterostructure to be utilized for the design of high electron mobility transistors (HEMTs). The formation of the 2DEG layer also gives rise to 2D potential well confinement at the heterostructure interface. The width of the potential well is only a few nanometers, making this interface electron gas subject to quantum confinement along the direction perpendicular to the interface. The detailed structure of the potential well is determined by the fraction of aluminum (Al) in AlGaN, as well as the applied voltage across the heterostructure.
The first set of goals for this research is to investigate how the bandstructure of Ge changes: Part (1) with the applied strain, and Part (2) with alloyed tin (Sn). The empirical pseudopotential method (EPM) was utilized for the band structure calculation, together with the rules for strain translation for the investigation of Part (1). In Part (1), simulation results give the optimal orientation for different types of applied strain and also thoroughly map the influence of strain applied on any arbitrary orientations. It also reveals that for biaxial strain, there exists another orientation that is more robust against misalignment with respect to the originally desired orientation than the optimal plane, with little compromise of bandgap and slightly higher requirements for the sufficient strain. For Part (2), EPM was combined with perturbation theory for the inclusion of the influence of the Sn atoms in the Ge lattice. A new and computationally inexpensive method is developed during the research. Simulation results agree significantly when compared to reported experimental measurements, indicating the capability of the method.
The second set of goals is to investigate the electron transport properties of the 2DEG layer at the interface of GaN HEMT and related power transistors. The potential well is approximated and quantified by a triangular potential well and the carrier sheet density is kept the same during the approximation. Thorough simulations are conducted by calculating the band alignment of the heterostructure with different structural configurations. A fixed correlation between the carrier sheet density and the shape of the potential well (slope of the triangular potential well and the height of the well) is revealed. This correlation is used as the input for the Monte Carlo (MC) simulation. The changes to the mobility of the electrons at the 2DEG layer with changing interface potential well shape are investigated and statistics of drift velocity, electron energy, and valley occupation are collected. Mobility information is also extracted and compares favorably with reported experimental measurements.
Ge is an indirect bandgap material with an indirect bandgap of 0.66eV. This bandgap value corresponds to a wavelength of 1.88µm, which lies in the infrared range. The Ge material itself is also compatible with the standard Si CMOS process technology. Because of these advantages, Ge is considered a candidate for the application of photo detecting in the SWIR range. Apart from the indirect bandgap of 0.66eV, Ge also has a direct bandgap of 0.8eV. From early research, the relatively small offset between the indirect and direct bandgap can be inverted either by applying strain or alloying with tin.
GaN is a binary direct wide bandgap material with a direct bandgap of 3.4eV. It has a high breakdown voltage, and relatively high saturation velocity and carrier mobility. These properties give GaN an advantage in the realm of high power application. GaN can also form a heterostructure with AlGaN, which can form a 2D electron gas (2DEG) layer at the interface without intentionally doping either material. The 2DEG layer has an even higher mobility compared to the mobility of the bulk GaN, which allows the heterostructure to be utilized for the design of high electron mobility transistors (HEMTs). The formation of the 2DEG layer also gives rise to 2D potential well confinement at the heterostructure interface. The width of the potential well is only a few nanometers, making this interface electron gas subject to quantum confinement along the direction perpendicular to the interface. The detailed structure of the potential well is determined by the fraction of aluminum (Al) in AlGaN, as well as the applied voltage across the heterostructure.
The first set of goals for this research is to investigate how the bandstructure of Ge changes: Part (1) with the applied strain, and Part (2) with alloyed tin (Sn). The empirical pseudopotential method (EPM) was utilized for the band structure calculation, together with the rules for strain translation for the investigation of Part (1). In Part (1), simulation results give the optimal orientation for different types of applied strain and also thoroughly map the influence of strain applied on any arbitrary orientations. It also reveals that for biaxial strain, there exists another orientation that is more robust against misalignment with respect to the originally desired orientation than the optimal plane, with little compromise of bandgap and slightly higher requirements for the sufficient strain. For Part (2), EPM was combined with perturbation theory for the inclusion of the influence of the Sn atoms in the Ge lattice. A new and computationally inexpensive method is developed during the research. Simulation results agree significantly when compared to reported experimental measurements, indicating the capability of the method.
The second set of goals is to investigate the electron transport properties of the 2DEG layer at the interface of GaN HEMT and related power transistors. The potential well is approximated and quantified by a triangular potential well and the carrier sheet density is kept the same during the approximation. Thorough simulations are conducted by calculating the band alignment of the heterostructure with different structural configurations. A fixed correlation between the carrier sheet density and the shape of the potential well (slope of the triangular potential well and the height of the well) is revealed. This correlation is used as the input for the Monte Carlo (MC) simulation. The changes to the mobility of the electrons at the 2DEG layer with changing interface potential well shape are investigated and statistics of drift velocity, electron energy, and valley occupation are collected. Mobility information is also extracted and compares favorably with reported experimental measurements.
