The proposed goal of this proposal is to explore the Ge quantum dot superlattices for thermoelectric device applications. Over 1-year period, we have worked on this project extensively and achieved the following research results.

It is estimated that about 25% of all electrical energy is used for lighting purposes. Light-emitting diode (LED)-based lighting technologies can be a factor of nearly 10 times more efficient than incandescent lighting and a factor of about 2-3 times more efficient than fluorescent lighting technologies. It is predicted by Department of Energy (DOE) that by 2025 the use of solid-state lighting will reduce national energy consumption for lighting by 29%. The cumulative energy savings from 2006 to 2025 would result in more than 125 billion dollars of savings to consumer electricity bills. LEDs based on Gallium Nitride (GaN) materials are currently widely developed to conserve electrical energy in numerous lighting applications, such as headlights of vehicles, traffic lights, outdoor displays, and some indoor lighting. The trend towards higher brightness, lower operating current, and longer lifetime has increased the needs for developing alternative next-generation LEDs. Zinc Oxide (ZnO), as a wide bandgap semiconductor material, has long been believed to be a suitable candidate for high-efficient light emitting applications, due to its superior intrinsic properties over GaN. Our research on solving impurity-doping problems in ZnO offers an effective way towards the fabrication of useful LEDs for solid-state lighting applications.

Optoelectronics devices based on ZnO thin films and nanostructures are discussed in this dissertation. A ZnO homojunction LED was demonstrated. Sb-doped p-type ZnO and Ga-doped n-type ZnO on Si (100) substrate were used for the LED device. After achieving ohmic contacts on both types of ZnO, the device showed rectifying current-voltage (I-V) characteristics. Under forward bias, the device successfully showed ultraviolet emissions. The emission properties were analyzed and the emission was confirmed to come from ZnO near band edge emissions. Further analysis showed that the emission mainly comes from the p-type layer of the device. A ZnO ultraviolet laser diode was fabricated and demonstrated. The device consists of Sb-doped p-type ZnO layer and Ga-doped n-type ZnO layer. In between p-layer and n-layer, a thin MgZnO/ZnO/MgZnO quantum well structure was inserted. In this device, random lasing mechanism plays an important role. When the diode was biased, the generation of light was enhanced by the carrier localization effect from the quantum well. The light was scattered between the ZnO random grain boundaries. Since the scattering effect can be so intense that some of the light can return to its original place to form close travel loop, as "random laser cavity". As long as the gain can overcome loss from scattering and material loss, lasing action can be demonstrated. An improved ZnO LED device was grown and characterized. The device grown on c-plane sapphire substrate can favor ultimate device applications due to the improved crystal quality of ZnO and the possibility of getting single crystallinity. A double heterostructure (MgZnO/ZnO/MgZnO) was also inserted in between p-layer and n-layer of the device to enhance the light output. The device showed much enhanced output power of 457 nW, which is two orders stronger than the LED fabricated on Si substrate. The optimization of high quality ZnO thin film on c-plane sapphire substrate was discussed. The devices in chapter two, three and four utilized Si or sapphire substrate, and are all in polycrystalline nature. To solve this problem and get the basis of high output power LEDs and lasers, single crystalline, two dimensional surface ZnO thin films were grown in chapter five. MgO/ZnO double buffer layers were used to accommodate the lattice mismatch. MgO thickness was found to be very important in achieving good ZnO thin film. An optimized growth also yields low background electron concentration and high mobility, which can enable future high quality p-type ZnO engineering. Our research was also expanded from ZnO thin films to ZnO nanostructures. The purpose of chapter six is to demonstrate a ZnO nanowire laser. ZnO nanowires are an excellent cavity and itself is a great gain material. We expanded Sb-doped p-type ZnO from thin films to ZnO nanowires. A p-type ZnO nanowire/n-type ZnO thin film p-n junction was achieved. The device showed lasing action when injection current was larger than ~50 mA. The lasing mechanism and gain/feedback were also discussed in detail.

Ternary alloy CdZnO and MgZnO thin films were obtained by radio frequency-assisted (RF) molecular beam epitaxy (MBE). The films were characterized by photoluminescence, X-ray diffraction, UV-visible absorption, scanning electron microscopy and Hall coefficient measurement. When Cadmium or Magnesium was co-deposited with Zinc under Oxygen ambient, the bandgap of ZnO based samples were tuned in the range of 2.25-4.76eV which were confirmed by band edge luminescence and Tauc plot. To overcome the shortage of unstable p-type ZnO semiconductor and simplify the complexity fabrication procedure for optical applications, a semiconductor-metal-semiconductor (MSM) device structure was developed. Electrically driven random laser devices based on the MSM structure were demonstrated by both experimental and simulation results, suggesting the ZnO material with excitonic properties under room temperature a potential competitive candidate in UV optical applications.

Doping in zinc oxide (ZnO) thin films were discussed in this dissertation.The optimizations of undoped ZnO thin film growth using molecular-beam epitaxy (MBE) were discussed. The effect of the oxygen ECR plasma power on the growth rate, structural, electrical, and optical properties of the ZnO thin films were studied. It was found that larger ECR power leads to higher growth rate, better crystallinity, lower electron carrier concentration, larger resistivity, and smaller density of non-radiative luminescence centers in the ZnO thin films. Low-temperature photoluminescence (PL) measurements were carried out in undoped and Ga-doped ZnO thin films grown by molecular-beam epitaxy. As the carrier concentration increases from 1.8 × 1018 to 1.8 ×1020 cm-3, the dominant PL line at 9 K changes from I1 (3.368 - 3.371 eV), to IDA (3.317 -3.321 eV), and finally to I8 (3.359 eV). The dominance of I1, due to ionized-donor bound excitons, is unexpected in n-type samples, but is shown to be consistent with the temperature-dependent Hall fitting results. We also show that IDA has characteristics of a donor-acceptor-pair transition, and use a detailed, quantitative analysis to argue that it arises from GaZn donors paired with Zn-vacancy (VZn) acceptors. In this analysis, the GaZn0/+ energy is well-known from two-electron satellite transitions, and the VZn0/- energy

is taken from a recent theoretical calculation. Typical behaviors of Sb-doped p-type ZnO

were presented. The Sb doping mechanisms and preference in ZnO were discussed.

Diluted magnetic semiconducting ZnO:Co thin films with above room-temperature TC

were prepared. Transmission electron microscopy and x-ray diffraction studies indicate

the ZnO:Co thin films are free of secondary phases. The magnetization of the ZnO:Co

thin films shows a free electron carrier concentration dependence, which increases

dramatically when the free electron carrier concentration exceeds ~1019 cm-3, indicating a

carrier-mediated mechanism for ferromagnetism. The anomalous Hall effect is observed

in the ZnO:Co thin films. The anomalous Hall coefficient and its dependence on

longitudinal resistivity were analyzed. The presence of a side-jump contribution further

supports an intrinsic origin for ferromagnetism in ZnO:Co thin films. These observations

together with the magnetic anisotropy and magnetoresistance results, supports an intrinsic

carrier-mediated mechanism for ferromagnetic exchange in ZnO:Co diluted magnetic

semiconductor materials. Well-above room temperature and electron-concentration

dependent ferromagnetism was observed in n-type ZnO:Mn films, indicating long-range

ferromagnetic order. Magnetic anisotropy was also observed in these ZnO:Mn films,

which is another indication for intrinsic ferromagnetism. The electron-mediated

ferromagnetism in n-type ZnO:Mn contradicts the existing theory that the magnetic

exchange in ZnO:Mn materials is mediated by holes. Microstructural studies using transmission electron microscopy were performed on a ZnO:Mn diluted magnetic

semiconductor thin film. The high-resolution imaging and electron diffraction reveal that

the ZnO:Mn thin film has a high structual quality and is free of clustering/segregated

phases. High-angle annular dark field imaging and x-ray diffraction patterns further

support the absence of phase segregation in the film. Magnetotransport was studied on

the ZnO:Mn samples, and from these measurements, the temperature dependence of the

resistivity and magnetoresistance, electron carrier concentration, and anomalous Hall

coefficient of the sample is discussed. The anomalous Hall coefficient depends on the

resistivity, and from this relation, the presence of the quadratic dependence term supports

the intrinsic spin-obit origin of the anomalous Hall effect in the ZnO:Mn thin film.

Two dimensional (2D) hexagonal boron nitride (h-BN) continues to attract tremendous research interest across the globe thanks to their unique optical, electronic, and mechanical properties. However, the controllable synthesis of large size single crystal h-BN is still challenging to realize the technological potential. In this thesis, we focus on growth and characterization of h-BN by molecular beam epitaxy (MBE). By exploring the role of interstitial carbon in transition metal substrate, we have achieved a big step forward toward controllable synthesis of large-size single crystal h-BN.

In the first project, i.e., Chapter 2, we carried out a systematic study of hexagonal boron nitride/graphene (h-BN/G) heterostructure growth by introducing high incorporation of carbon (C) source on a heated cobalt (Co) foil substrate followed by boron and nitrogen sources in a molecular beam epitaxy system. With the increase of C incorporation in Co, three distinct regions of h-BN/G heterostructures were observed from region (1) where the C saturation is not attained at the growth temperature (900 °C) and G is grown only by precipitation during cooling process to form a “G network” underneath the h-BN film; to region (2) where the Co substrate is just saturated by C atoms at the growth temperature and a part of G growth occurs isothermally to form G islands and another part by precipitation, resulting in a non-uniform h-BN/G film; and to region (3) where a continuous layered G structure is formed at the growth temperature and precipitated C atoms add additional G layers to the system, leading to a uniform h-BN/G film. We realized that the top h-BN film has the largest effect on the growth of underneath G layers in region 1. It is also found that in all three h-BN/G heterostructure growth regions, a 3-hrs h-BN growth at 900 ºC leads to h-BN film with a thickness of 1~2 nm, regardless of the underneath G layers’ thickness or morphology. Growth time and growth temperature effects have been also studied.

In the second project, i.e., Chapter 3, we demonstrate that the dissolution of carbon

into cobalt (Co) and nickel (Ni) substrates can facilitate the growth of h-BN and attain

large-area 2D homogeneity. The morphology of the h-BN film can be controlled from 2D

layer-plus-3D islands to homogeneous 2D few-layers by tuning the carbon interstitial

concentration in the Co substrate through a carburization process prior to the h-BN growth

step. Comprehensive characterizations were performed to evaluate structural, electrical,

optical, and dielectric properties of these samples. Single-crystal h-BN flakes with an edge

length of ∼600 μm were demonstrated on carburized Ni. An average breakdown electric

field of 9 MV/cm was achieved for an as-grown continuous 3-layer h-BN on carburized Co. Density functional theory calculations reveal that the interstitial carbon atoms can increase the adsorption energy of B and N atoms on the Co(111) surface and decrease the diffusion activation energy and, in turn, promote the nucleation and growth of 2D h-BN.

In the third project, i.e., Chapter 4, we designed a synthetic approach, in which secondary recrystallized Ni (100) substrates underwent a carburization process, followed by the growth of h-BN in a molecular beam epitaxy system. The h-BN growth dynamics were studied by tuning different growth parameters including the substrate temperature, and the boron and nitrogen source flux ratio. With assistance from density functional theory calculations, we rationalized the role of interstitial C atoms in promoting h-BN growth by enhancing the catalytic effect of the transition metal, which lowers the nucleation activation energy barrier. Through the control of the growth parameters, a single-crystal h-BN monolayer domain as large as 1.4 mm in edge length was achieved. In addition, a high-quality, continuous, large-area h-BN single-layer film with a breakdown electric field of 9.75 MV/cm was demonstrated. The high value of the breakdown electric field suggests that single-layer h-BN has extraordinary dielectric strength for high-performance 2D electronics applications.

In the fourth project, i.e., Chapter 5, we present a study of h-BN adlayer growth and provide a strategy towards eliminating these adlayers for the precise control of the number of 2D layers. By varying the growth parameters such as substrate property, nitrogen source composition, and substrate carburization time, we found that the adlayer growth can be controlled by controlling the nucleation and intercalation processes, which is achieved by engineering the defects and impurities on substrate and the activeness of the h-BN edges. While crystallographic defects and impurities stimulate the multilayer nucleation process, activated edge tends to turn off the intercalation process by reducing the probability of precursors penetrating into the interface. We have achieved the growth of a large-area adlayer-free single-layer h-BN film.

Tunnel oxide thickness scaling is encountering problem for next generation flash memory device. With tunnel oxide shrinking down, the stress-induced leakage current becomes more serious leading to degraded retention. To overcome this, flash memory with discrete-traps as floating gate was proposed, including nitride charging and nanocrystal storage. The first report on nanocrystal memory is Si nanocrystal memory. After that, tremendous efforts have been put on using semiconductor or metal nanocrystals for memory application. However, for Si, although long retention time has been demonstrated, the long retention is believed to be related to charge storage on defect levels. Those levels are distributed throughout the whole band gap of Si and are not thermally stable, resulting in compromised device reliability. For metals, on the other hand, the band gap is negligible, therefore the issue of defect level charging is effectively avoided. However, the problem with metal nanocrystal memory lies in the reaction or inter-diffusion between metal and tunnel oxide underneath at high-temperature process step, such as source/drain dopant activation annealing for MOSFET. The reaction or diffusion of metal atoms generates weak points in the tunnel oxide, which degrades the device retention.

To solve the dilemma between long retention and scaled tunnel oxide, in this work, engineered floating gate, such as hybrid-nanocrystal and new materials nanocrystals that are compatible with current Si technology, was proposed and good memory performance was demonstrated. Chapter 1 introduces the conventional flash memory, including the operation principle, architectures, challenge in next generation flash memory and some new technologies to address this issue in conventional flash. Chapter 2 describes the methodologies used in this thesis work, including nanocrystal growth methods, nanocrystal characterization techniques, and nanocrystal memory fabrication and device characterization. Chapter 3 discusses the flash memory with Ge/Si hetero-nanocrystals as floating gate. Type-II band alignment between Ge and Si makes Ge/Si hetero-nanocrystal good for long time hole storage than Si nanocrystal memory. In addition to p-MOS, we also developed n-MOS memory with CoSi2-coated Si nanocrystals as floating gate in the Chapter 4. The Fermi-level of CoSi2 locates around the midgap of Si so that the device is good for both electrons and holes trapping. Furthermore, the quantum well formed between CoSi2 and SiO2 is deeper than that of Si and SiO2, leading to an elongated retention time. The Si nanocrystal underneath the CoSi2 effectively prevents the charges stored in CoSi2 from leaking back to the channel. In order to lower the thermal budget to make silicide nanocrystals, vapor-solid-solid (VSS) growth mode was employed and NiSi2 nanocrystsals were synthesized as presented in Chapter 5. The long retention of NiSi2 nanocrystal memory is also benefited from the deep quantum well between NiSi2 and SiO2. For next generation flash memory, more uniform nanocrystals with higher dot density, 1012cm-2, is necessary to meet the requirement of 22nm technology and beyond. In Chapter 6, we developed PtSi nanocrystals using the similar synthesis technique as NiSi2. The nanocrystal density is enhanced from 3×1011cm-2 to 1.5×1012cm-2.

In short, engineering the nano-floating gate by replacing Si nanocrystals with hybrid nanocrystals and silicide nanocrystals benefits the device retention time. These new memories also exhibit faster programming and erasing speeds. The enhanced memory performance makes the devices fit for next generation memory with further scaled tunnel oxide.

Transition metal doped ZnO has been proposed to be a Diluted Magnetic Semiconductor with room temperature ferromagnetism. High quality Mn doped ZnO was grown on R-sapphire substrate. Results support intrinsic ferromagnetism, while µB/ion number was found to be larger than maximum permitted by Hund's law. Saturation strength and coercivity field was found to be manipulated by varying Mn doping concentration. Mn/Ag co-doping was found to alter saturation strength and coercivity field as well. Ag was investigated as substitution dopant to achieve high quality p-type ZnO. With optimization of growth parameters, phase segregation was suppressed and p-type was observed. Besides being p-type, Ag doped ZnO was found to exhibit room temperature ferromagnetism. This is the first high quality demonstration of Ag doped ZnO as DMS.

Low-dimensional materials continue to attract and involve the minds and labor of scientists and engineers around the world as they are offering a set of fascinating properties which can be controlled by composition, size, and morphology. In this thesis, our focus is to explore the growth and characterization of some of such low-dimensional materials with molecular beam epitaxy (MBE) method. A technologically relevant example of low-dimensional systems is two-dimensional (2D) materials including graphene (G) and h-BN. To date, it is still challenging to reliably grow high-quality uniform 2D h-BN and h-BN/G heterostructures in a wafer scale mainly due to their complicated growth process.

In this first project, i.e., Chapter 2, we perform a systematic study of h-BN/G heterostructure growth on cobalt (Co) foil substrate in an MBE system and investigate the growth mechanisms of individual G and h-BN layers in the structure. We demonstrate with the increase of C incorporation in Co, three distinct h-BN/G growth regions can be observed: (1) the C saturation is not attained at the growth temperature (900 °C) and G is grown only by precipitation during cooling process to form a “G network” underneath the h-BN film; (2) the Co substrate is just saturated by C atoms at the growth temperature and a part of G growth occurs isothermally to form G islands and another part by precipitation, resulting in a non-uniform h-BN/G film; and region (3): a continuous layered G structure is formed at the growth temperature and precipitated C atoms add additional G layers to the system, leading to a uniform h-BN/G film. We show that in all three growth regions, a 3-hrs h-BN growth at 900 ºC leads to h-BN film with a thickness of 1~2 nm, regardless of the underneath G layers’ thickness or morphology.

Following the growth of h-BN/G heterostructures, in the next project, i.e, Chapter 3, we demonstrate that the dissolution of C atoms into heated Co substrate can also facilitate the growth of 2D h-BN and alter its morphology from 2D layer-plus-3D islands to homogeneous 2D few-layers. A high breakdown electric field of 12.5 MV/cm was achieved for a continuous 3-layer h-BN. Density functional theory calculations reveal that the interstitial C atoms can increase the adsorption of B and N atoms on the Co (111) surface, and in turn, promote the growth of 2D h-BN.

In the last project, i.e., Chapter 4, we discuss the growth and characterization of another low-dimensional form of BN family materials, namely, cubic boron nitride nanodots (c-BN NDs) which offers a variety of novel opportunities in battery, biology, deep ultraviolet light emitting diodes, sensors, filters, and other optoelectronic applications. To date, the attempts towards producing c-BN NDs were mainly performed under extreme high-temperature/high-pressure conditions and resulted in c-BN NDs with micrometer sizes, a mixture of different BN phases, and containing process-related impurities/contaminants. To enhance device performance for those applications by taking advantage of size effect, pure, sub-100 nm c-BN NDs are necessary. In this chapter, we demonstrate the self-assembled growth of sub-100 nm c-BN NDs on Co and Ni foil substrates by plasma-assisted MBE for the first time. We found that the nucleation, formation, and morphological properties of c-BN NDs can be closely correlated with the nature of substrate including catalysis effect, lattice-mismatch-induced strain, and roughness, and growth conditions, in particular, growth time and growth temperature. The mean lateral size of c-BN NDs on cobalt scales from 175 nm to 77 nm with the growth time. The growth mechanism of c-BN NDs on metal substrates is concluded to be Volmer-Weber (VW) mode. A simplified two-dimensional numerical modeling shows that the elastic strain energy plays a key role in determining the total formation energy of c-BN NDs on metals.

Because of its unique properties, such as extremely high mobility, high mechanical strength, good optical transparency and high chemical stability, graphene has attracted vast interests in nanotechnology communities, condensed matter physics, and chemistry. The method of synthesizing large-area graphene with good quality is critical for its potential applications. This dissertation reports a new synthesis method of using a home built thermal cracker enhanced gas source molecular beam epitaxy (GSMBE) system. Chapter 2 discusses graphene growth using nickel substrate in the GSMBE system. Hydrocarbon gas molecules were broken by thermal cracker at very high temperature of 1200°C and then impinged on a nickel substrate. High-quality, large-area graphene films were achieved at 800°C, and this was confirmed by both Raman spectroscopy and transmission electron microscopy. A rapid cooling rate was not required for few-layer graphene growth in this method, and a high-percentage single layer and bilayer graphene films were grown by controlling the growth time. The results suggest that in this method, carbon atoms migrate on the nickel surface and bond with each other to form graphene. Few-layer graphene is formed by subsequent growth of carbon layers on top of existing graphene layers. This is completely different from graphene formation through carbon dissolving in nickel and then precipitating from the nickel during rapid substrate cooling in the chemical vapor deposition method. Chapter 3 further discusses using cobalt as graphene growth substrate. Growth conditions including growth temperature and growth time play important roles in the resulting morphology of as-grown films. A narrow growth time window was found for different growth temperatures. Carbon absorption and desorption phenomena were responsible for temperature-dependent and growth time-dependent graphene morphology. Fast cooling rate was not required in this process due to direct growth mechanism under atomic carbon growth condition. Large-area graphene films with high single-layer and bi-layer coverage of 93% were confirmed by Raman spectroscopy and transmission electron microscopy.

Graphene based flash memory was demonstrated by using nickel nanocrystals as storage nodes in chapter 4. First, the graphene channel with a dimension of a 20 µm × 5 µm was acquired by photolithography and oxygen plasma etching. Then, the gate stack formed by the deposition of tunneling oxide, nickel nanocrystal and block oxide. On/off operation of the transistor memory was acquired by static pulse response measurement. The memory window of the device was found up to be 23.1 V by back gate sweep. This memory effect is attributed to charging/discharging of nanocrystals. Furthermore, excellent retention and endurance performance were achieved. Chapter 5 demonstrated a memory capacitor with an embedded graphene nano dots structure. Graphene nano dots were successfully fabricated by using nickel nano crystals as etching masks. By tuning the etching parameters, graphene dots with the size of 10 nm to 100 nm can be acquired. Raman spectra confirms the defects and the edges effect of nano dots. A memory capacitor using a SiO₂/graphene dots/Al₂O₃ sandwich structure was fabricated. A hysteresis from the C-V sweep proves the memory capability of the as-fabricated capacitor.