Wednesday 19 April 2017, 14:30, Prof.dr. G. Berkhoff-zaal
Chris Spruijtenburg is a PhD student in the MESA+ research group NanoElectronics. His promoter is Wilfred van der Wiel.
Imperfection - holes and defects in silicon quantum dots
Spurred on by the promise of a quantum ``speed-up'' for certain classes of computational problems, interest in quantum computation has seen a great surge over the past decades. Many approaches have been tried, with a couple of systems vying for a place as a scalable universal quantum computer. One of these approaches is confining carriers to such a small region in a semiconductor, that quantum properties become observable. These ``quantum dots'' have their best properties in a material with little to no hyperfine-interaction. To this end, silicon has been at the forefront as a very interesting material, as it can be isotopically purified to contain a very low concentration of non-zero nuclear spin isotopes. This has garnered a lot of success in achieving long spin lifetimes.
Work thusfar has primarily been focused on electrons as the carriers of spin and charge. However, spin and charge can also be carried by holes, which can be typified as the absence of a single electron in a sea of electrons. In this thesis then, we focus on holes.
The first step in creating quantum dots in semiconductors is the necessity of control over the confinement potential. However, defects, coming about by disorder or other fabricational issues are of great influence on quantum dot formation. This particular thesis concerns itself with the properties of hole quantum dots under the influence of, and the charge transport through, these defects. We explore the origin of these defects, and devise methods of eliminating them.
Chapter 2 concerns the theoretical concepts governing quantum dot transport measurements. We posit single-electron tunneling and the constant interaction model. We discuss what the band diagram in silicon tells us about the behaviour of holes and deduce how the spin-orbit interaction leads to heavy-hole and light-hole characteristics. For holes in quantum wells we find that holes are predicted to have an anisotropic g-tensor. We then discuss the theoretical concepts underlying point defects such as Pb centers.
Chapter 3 examines the entire heterostructure, as it is used in creating electrostatically defined quantum dots. We relate three properties of these heterostructures to the performance of quantum dots: materials, interfaces, and morphology. We find that the ability to passivate defects at the Si/SiO2 interface hinges primarily on controlling the dewetting at higher temperatures. We introduce an ALD-grown Al2O3 layer, which strikes two birds with one stone. The layer both provides hydrogen for the annealing process and prevents dewetting. We also find that by supplanting Al by Pd as the electrode material, the formation of an interfacial layer between the metal and the oxide is prevented. This can be seen in TEM images. The amount of two-level fluctuators, present in glassy systems accompanying this interface layer, should therefore also be diminished. We further show that the introduction of Al2O3 introduces a layer of negative fixed charge, which can be eliminated or controlled by exposure to a UV-ozone oxidation process.
Chapter 4 shows that we can indeed make a hole quantum dot in intrinsic silicon, by demonstrating single-hole tunneling through a two-dimensional hole gas. We show resonant tunneling features of Coulomb islands highly coupled to single controlling gates. The Coulomb islands are caused by local potential fluctuations, most likely due to impurities or charge traps in the SiO2, or at the interface of Si and SiO2. Silicon is known for being extremely sensitive to disorder, owing to the large effective mass of the charge carriers, which is even higher for holes than for electrons.
Chapter 5 sets out to improve upon these results and we investigate defects in the heterostructure. Using Al2O3 grown by atomic-layer deposition in an annealing process, we passivate the majority of the electrically active defects. Using this oxide, and the passivating properties of the hydrogen contained therein, we are able to create intentional quantum dots of at least 180 nm. These quantum dots show many charge transitions, indicating the low level of disorder in the devices. The versatility of our ambipolar architecture allows us to investigate the amphoteric behaviour of a remaining charge defect, which has states ~10 meV above the valence band, and below the conduction band.
Chapter 6 extends on our new-found ability to anneal defects and create low-disorder devices. In this chapter, the spin-orbit interaction starts to play a more important role, as it is responsible for the prediction that hole spins are expected to be preferentially oriented in 2D quantum wells. We focus on the behaviour of the g-tensor in a rotating magnetic field, and extract the anisotropy of the g-tensor. We do this by studying the Zeeman splitting in a magnetic field capable of rotating 360° over all 3 degrees of freedom. Using two methods of fitting the data to our model, we extract similar anisotropies for the g-tensor (g* » 2.2, g* » 4), indicating that the g-factor is roughly twice as strong out-of-plane than in-plane. This is consistent with the prediction that light-holes and heavy-holes are oriented preferentially in a 2D quantum well.
In chapter 7 a proof-of-principle of a single-layer depletion-mode hole quantum dot is demonstrated. The depletion architecture has several advantages, such as optical accessibility and having areas of the device free from potentially damaging exposure to electron beam lithography. The negative fixed charge in ALD-grown Al2O3 enables the operation in depletion-mode by inducing a 2D hole gas. Equal coupling of charge transitions to adjacent gates indicates that an intentional quantum dot is formed in between them. We are then able to tune the single quantum dot to a double-quantum dot regime by using an interdot coupling gate. Characterization of the charge-offset stability of this device indicates that the device is extremely stable, with the charge-offset stability having an upper bound at Q0 = 0.04e and a lower bound of Q0 = 0.005e. This compares favourably to previously known results for all-Si based devices (Q0 < 0.01e) and Al-based devices (Q0 < 0.15e).
While holes are promising candidates for quantum computing, several hurdles have to be overcome in order to exploit their properties. The Si/SiO2 interface is known to have many types of defects. These have to be dealt with or avoided altogether. This can be done by annealing, or by switching to a different insulator material. In this regard the Si/SiGe system appears to be a good candidate. These, and certainly many other improvements which this thesis has touched upon, could lead to one day possibly making a truly scalable quantum technology based on silicon.