Quantum dots in Si/SiGe have long spin decoherence times, due to the low density of nuclear spins and weak coupling between nuclear and electronic spins. Because of this, they are excellent candidates for use as solid state qubits. The initial approach towards creating controllable Si/SiGe quantum dots was to fabricate them in delta doped heterostructures. We provide evidence that the delta doping layer in these heterostructures provides a parallel conduction path, which prevents one from creating controllable quantum dots. Instead, it may be more favorable to supply electrons in the 2DEG through capactive gating, instead of a delta doping layer. We therefore discuss efforts to fabricate Si/SiGe quantum dots from undoped heterostructures and the difficulties encountered. A new method for fabricating ohmics in undoped heterostructures is discussed. We also discuss parallel conduction which occurs in the Si cap layer of these undoped heterostructures, which appears to be a major obstacle towards achieving workable devices in undoped Si/SiGe heterostructures.

The potential power of quantum algorithms to solve problems that are currently computationally intractable has launched investigations into a wide range of quantum systems with the goal of developing quantum hardware. Among these candidate systems for encoding quantum bits (qubits), electron spins localized in gate-defined quantum dots are promsing. In particular, quantum dots defined in Si/SiGe heterostructures have a number of advantages, including a low defect density and a low intrinsic density of interfering nuclear spins. This work presents a series of investigations into gate-defined quantum dots formed in Si/SiGe heterostructures. A key parameter for the performance of quantum dot qubits hosted in Si is the valley splitting energy. In Si/SiGe heterostructures, controlling the valley splitting remains an outstanding challenge. Here we present measurements of valley splitting in Si/SiGe heterostructures with varied Ge concentration profiles. The measured scaling of the valley splitting with vertical electric field is compared with tight-binding simulations of heterostructures with interfacial steps, and the agreement in scaling provides evidence for the important role played by interfacial disorder in setting the valley splitting in these samples. Additionally, for gate-defined quantum dots, a key device element is the dielectric layer which mediates the electric fields between the gate electrodes and the gate-defined dots. We discuss experiments to test a novel method for generating a gate dielectric for Si/SiGe quantum dots using thermal oxidation at 700 degrees C. Another critical component towards developing quantum processors based on quantum dots is the engineering of strong and controllable inter-qubit coupling. For double quantum dot qubits with an effective charge dipole moment, a capacitive dipole-dipole interaction can generate coherent coupling between neighboring qubits. Here we also discuss investigations into the capacitive interaction between double quantum dots in a quadruple quantum dot device. This includes a demonstration of the tuning of the capacitive coupling energy over a wide range using barrier gate voltages and analysis of the dependence of that coupling energy on device geometry. Finally, we present procedures for fabricating quadruple quantum dot devices in Si/SiGe using an architecture based on overlapping self-oxidized Al gates.

Quantum computing has become a thriving field over the past several decades. Although many candidate systems exist, this dissertation will focus on quantum dots as a quantum computing implementation, specifically lateral quantum dots in silicon based heterostructures. Lateral quantum dots use trapped electrons in semiconducting heterostructures to form qubits, the basic building block of a quantum computer. There are several potential qubit implementations using quantum dots and new qubit schemes, such as the valley qubit presented in Chapter 4, are still being investigated. Many of these implementations have already been successfully demonstrated. In this sense, research into quantum dots is a maturing field, having successfully demonstrated proof of concept for multiple qubit implementations. If quantum dots are to succeed as a quantum computing platform research needs to focus on improving the qubits themselves. Decoherence and dephasing need to be improved, but also yield and reproducibility. In this work I describe experiments intended to help understand and improve the performance of lateral quantum dots. I fabricated multiple lithographically identical devices on Si/SiO2 and Si/SiGe heterostructures to compare charge noise on the two Silicon based substrates. I describe the first conclusive observation and characterization of a valley based qubit. The noise characteristics of the valley qubit are particularly attractive as it's operation is resistant to charge noise, the primary source of noise in Silicon based qubits. Finally I present the ongoing development of a novel gate architecture for lateral quantum dots. Called a hybrid architecture, this design possesses good tunability along with simple fabrication and a reduced number of total gates relative to other leading architectures; this has the potential to dramatically improve yield and scalability.

Tunneling of an electron is one of the well-studied phenomena. It also found many commercial applications including tunnel diodes. In top-gated semiconductor quantum dot spin qubits, the voltage controlled tunnel barriers formed by top gates are the fundamental building block to build qubit. Moreover, energy dependence of the tunnel rate in such barriers is well utilized in many qubit measurements including single shot read out of a spin. There have been only a few studies in energy dependent tunneling in top-gated quantum dot devices. Also there has not been much study on the tunnel barrier itself. The barrier information such as height, length, and shape of the barrier is important to understand the dot system and to understand the energy dependent tunneling, which is critical in qubit operation. In this thesis research, we studied the tunnel barrier in a top gated Si/SiGe quantum device. We measured temperature dependence of the tunneling conductance, where we determined the height of the barrier by activation energy. Using the experimentally determined height of the barrier, we developed simple, empirical two-dimensional (2D) barrier models based on molecular coherent tunneling theory. The calculated tunneling conductance well fit to the experimental conductance. Using the developed models, we determined energy dependent tunneling coefficients, which agree well with experimental values that determined from pulsed gate tunnel rate measurements, performed in a dot using similar Si/SiGe heterostructure. The results suggest that the shape of the barrier is parabolic as we expected. Finally, we compared the barrier shape with conventional 1D models to check the impact of the dimensionality.

Since being proposed almost 40 years ago, scientists across many disciplines have made great progress in the fields of quantum computation and quantum information. Instead of a classical bit (0 or 1), a quantum computer uses a two-level quantum system as a quantum bit or qubit. By controllably manipulating the quantum-mechanical properties of these qubits, a quantum computer could, for example, be used to simulate other, less well understood quantum systems, or to run certain classes of quantum algorithms that cannot be run on classical hardware. In order to build a quantum computer, certain basic requirements must be met. As with a classical computer, logic gates are necessary to controllably manipulate qubits to perform calculations. One such requirement for a universal quantum computer is a two-qubit logic gate. This is an inherently quantum mechanical gate, which has no classical analog. For example, the controlled-not two-qubit gate will perform a not operation on the target qubit if and only if the control qubit is in the one state, else it does nothing to the target qubit. In either case, the control qubit is left unchanged and unmeasured. Being able to perform this gate with high fidelity is critical to creating a quantum computer. In this dissertation, I present progress towards fabricating, characterizing, and manipulating two-qubit devices in Si/SiGe heterostructures. First, I motivate the use of quantum dot qubits hosted in Si/SiGe as a suitable platform for quantum computing. Then, I present characterization of Si/SiGe substrates and discuss fabrication of a quantum dot device. Next, I outline the electronics set up for measuring a quantum dot device in a dilution refrigerator. I then present results of two, published experiments which explore multi-qubit systems: one which demonstrates controllable tunnel coupling between a quantum dot an a nearby localized impurity, and the other which demonstrates state-conditional Landau-Zener-Stückelberg oscillations between capacitively coupled double quantum dots in a quadruple quantum dot device. Next I discuss fabrication and characterization of micromagnets for spin qubit applications. I finally conclude by discussing future research avenues towards realizing a robust, multi-qubit device in silicon.

Quantum computing - leveraging quantum phenomena to perform complex and otherwise intractable computational problems - has rapidly progressed from a theoretical aspiration to a potential reality. Currently, there are many competing approaches to the way the physical qubits (quantum bits) are built, from trapped ions, to superconducting circuits, to semiconductor quantum dots, and beyond. Here, we focus on quantum dots, where electrons or holes are confined within a semiconductor and the quantized nature of charge and spin are utilized for computation. Within the field of quantum dots, heterostructures made of silicon and silicon-germanium are especially enticing due to their low density of defects and nuclear spin. Although quantum dots are a promising avenue for quantum computation because of their intrinsically small size and similarity to classical transistors, nearly every aspect of their design, realization, and control has yet to be fully optimized.This thesis explores modifications to the heterostructure, fabrication, and measurement of Si/SiGe quantum dots in the pursuit of improved quantum dot qubits. The valley splitting in silicon quantum dots, a near degeneracy of the lowest lying energy states, is critical to the formation and performance of silicon qubits. In this work, we present several modifications to the Si/SiGe heterostructure in an effort to enhance this splitting. In particular, we investigate the effects of introducing germanium to the silicon quantum well by the inclusion of a single spike in germanium concentration or an oscillatory concentration throughout the well. We present experimental measurements of the energy spectrum arising from both modifications and, coupled with theoretical support, demonstrate enhancements to the valley splitting. Next, we present several fabrication techniques with the goal of improved quantum dot functionality and lowered charge noise, a major barrier to higher quality devices. We report a new strategy for etched-palladium fabrication and discuss the current progress. Finally, we present work towards the automation of quantum dot tuning. As quantum dot devices increase in the number of qubits, so do the number of electrostatic gates which control the device. We discuss the development of automated tuning procedures and present a procedure for the formation of well-controlled quantum dots from initial voltage settings.