The past year saw great progress and activity within all of the Centre’s programs. The development of spin-based qubits in the Si:P materials system is the key focus of the Centre’s solid-state programs here at UNSW.

Silicon-based Si:P Qubits: Single Atom Device Fabrication using Scanning Probes and Measurement

Figure 1: (a) STM image (3.5 V sample bias) of an oxidised Si(001) surface obtained by 5 min atomic O–exposure; (b): Effect of surface oxidation on the scanning tunneling spectroscopy I-V curves showing an increase in the bandgap; (c) and (d) cross-sectional TEM micrographs of 25nm-thick SiO2 layer grown at room temperature on a Si(001) substrate.

Over the past year the Atomic Fabrication Group at UNSW have overcome one of the final fabrication hurdles to realising gated qubit architectures – the development of a low temperature, ultra-high vacuum (UHV) compatible dielectric – silicon dioxide. This allows us to electrically gate devices so that we may control single electron transfer events. To achieve this we have developed the growth of a uniform silicon dioxide layer, shown in Figure 1. This final hurdle completes the 8-stage atomically precise fabrication strategy outlined in 2001 for the fabrication of atomically precise devices in silicon.

Figure 2: (a) I-V characteristics as a function of gate voltage at 4.2K for a MOSFET with 25nm UHV-SiO2 grown at room temperature. (b) For comparison the same measurements for a MOSFET with 10 nm conventional thermal silicon dioxide grown at 900°C. Both sets of curves show turn-on, linear and saturation regimes.

Following this we have developed and incorporated this silicon dioxide layer into a UHV-compatible STM-based metal oxide semiconductor field effect transistor (MOSFET) process. Such a process allows us to assess the electrical quality of the oxide grown and to develop the worlds smallest MOSFET. The oxide is grown at room temperature, using a plasma source that provides a beam of neutral oxygen atoms. From the mobility versus carrier density we are able to extract an interface trap density of ~4×10¹¹ cm^-2 for our low temperature oxide, which is excellent for a completely room temperature process.

Figure 3: (a) Filled state STM image of an 8nm x 57nm wire patterned into a hydrogen resist on the Si(100) surface. (b) Four terminal DC measurements at 4 K of the wire demonstrates ohmic behaviour with a resistivity of 24x10-6 Wm-1.

In parallel we have assessed the quality of source-drain leads (that will be used for direct current measurements through quantum dot structures for spin and charge detection) by fabricating the narrowest conducting wires in silicon that still exhibit ohmic behaviour. Four-terminal electrical measurements of an 8nm (wide) x 57nm (long) wire was shown to have a resistivity of ~24x10^-6Wm^-1, which is the lowest resistivity for a doped silicon wire reported to date. This highlights the uniqueness of the atomically precise fabrication route, where we can separate the conducting regions from surface and interface states with atomic precision.

Finally new insights have arisen from studying time dependent reaction sequences in STM images taken of a P atom during its transition from a component of a gas phase molecule to a dopant in the surface. In collaboration with the modelling group at the University of Sydney, Kinetic Monte Carlo (KMC) simulations of these reaction processes have been used so that we may observe how a P atom is liberated from its ‘parent’ PH₃ molecule undergoing successive dissociation reactions until it incorporates into the surface. Our work on understanding PH₃-Si(001) sets a new benchmark for the study of adsorbate-surface systems, based on combining knowledge from STM experiments, density functional theory calculations and kinetic Monte Carlo simulations.

Figure 4: Step-wise PH3-Dissociation: (a)-(c) STM images showing the same 20Å x 32Å area of the Si(001) surface after PH3 dosing, obtained with a sample bias of –2.0 V. Surface models are presented in the right column. (a) shows a PH2+H adsorbed on a silicon dimer. (b) Around 4 minutes later the PH2+H dissociates to form PH+2H and dissociates further (c) to form P+3H after another ~70 minutes. The P+3H feature is stable at room temperature and was not seen to change during the next 235 minutes of imaging.

Silicon-based Si:P Qubits: Top-Down Device Fabrication and Measurement

Figure 5: Microwave spectroscopy of a 4-P-atom device, probed with an rf-SET, that exhibited a single charge transfer event. (a) Cross-sectional schematic of the 4-P-atom device geometry. (b) Colour plot of SET signal as a function of gate voltage and applied microwave frequency. The dotted lines are a fit to theory. (c) Energy levels of the (P3-P)2+ system.	Figure 6: Nano-Schottky reservoir devices for readout of spin qubits. (a) Schematic of a nano-Schottky reservoir in close proximity to a single P donor. (b) Pulse train applied for pulsed-voltage spectroscopy. (c) Tunnel rates for reservoir emptying vs. amplitude and offset of reservoir potential. (d) SET response as the Schottky reservoir is swept downwards in voltage for varying magnetic field.

Prior to our development of nanoscale silicide Schottky reservoirs in 2005, spin-state readout was a considerable hurdle and so during 2004-06 effort was focused on the fabrication and measurement of Si:P charge qubits (in a P-P⁺ charge configuration), since charge-state readout was conveniently accessible using the Centre’s rf-SET technology. During this period we have refined the process of counted single ion implantation, allowing known numbers of P atoms to be implanted with nearly 100% confidence. These devices enabled the extraction of the phonon-limited charge-state relaxation time to be obtained, while a more complex 4-P-atom device was used to determine, for the first time, the energy levels of a donor-molecule system in silicon, using microwave spectroscopy.

The 4-P-atom system was configured by implanting four phosphorus atoms through two nanoscale apertures, individually detected by on-chip single-ion detectors (Figure 5a). Using a nearby radio-frequency single-electron transistor (rf-SET) we observed a single charge transfer event between components of the 4-P-atom system. Microwave spectroscopy performed on the device exhibited an intriguing stacked-diamond characteristic (Figure 5b). Fitting lines to the main features in the data showed that they originate from two points on the gate voltage axis with slopes that follow a 1/n law. This is consistent with multi-photon excitation between sets of energy levels that cross at the gate voltages from where the features originate. A similar set of lines and origins can be identified for some of the weaker features in the data. With this, an energy level spectrum can be derived that contains two intersecting sets of three energy levels (Figure 5c) and is consistent with resonant multi-photon excitation between charge configuration states in a (P₃-P)²⁺ system (i.e. P₃ “molecule” and single P, 50 nm apart, with only two electrons).

Two of the most demanding experimental challenges for the development of a Si:P quantum computer are single spin measurement and the determination of spin lifetimes for donor states. For this purpose our teams at UNSW have developed nanoscale PtSi Schottky contacts (Figure 6a) for use as electron reservoirs, which allow us to controllably add and remove electrons to or from implanted P atoms. Pulsed voltage spectroscopy measurements were performed on implanted clusters of 50, 20 and 5 P-atoms, involving the application of a pulse train (Figure 6b) to the nano-Schottky reservoir, forcing electrons to tunnel back and forth between the reservoir and the cluster. An rf-SET is used to observe these minute changes in charge distribution, from which tunnel rates can be derived. The abrupt changes in tunnel rates that we observe are believed to be a signature of excited states of the implanted clusters, which can be considered as complex P-in-Si molecules. Work is now underway to apply the pulsed-voltage spectroscopy technique to counted single-donor samples, to map out the energy levels of a single P atom in Si.

Figure 7: Silicon quantum dot device. (a) Schematic of a Si quantum dot structure, using double-layer Al gates to induce an electron layer in intrinsic Si. (b) Conductance of annealed and non-annealed quantum dot devices as a function of gate voltage. (c) ‘Coulomb diamonds’ for a quantum dot containing ~30 electrons. (d) ‘Coulomb diamonds’ for a quantum dot in the few-electron regime, showing evidence of quantum-confined states.

Silicon quantum dots are an interesting system for the study of few electron physics, have the potential to be used as qubits, and are compatible with CMOS technology. Towards this end we have improved the stability of our Si quantum dots using a forming gas anneal for the reduction of charge traps. A new gate isolation technique has been developed allowing the fabrication of tuneable barrier gates for controllably defining Si quantum dots. The gates can be tuned to either strongly or weakly couple the dot to its leads and the top gate is used to control the number of electrons in the dot. The quality of the data shown in Figure 7 holds promise for the possibility of single spin manipulation in silicon and the use of the dot as a single electron transistor. The extension of this technology to the triple-quantum-dot regime will provide an experimental testing round for the study of coherent transport by adiabatic passage (CTAP).