Tunable quantum dots in carbon nanotubes

Quantum dots (QDs), also known as artificial atoms, are small structures to which electrons can be added, one by one, into discrete energy levels. Their study in the past decade has led to a considerable understanding of atomic-like systems in the solid state.¹ Recently, the interest in QDs has shifted from their mesoscopic physics to properties related to quantum information technology, since the electron spin in a QD can serve as an elementary quantum bit, or qubit.²

To date, the material used for most QD studies has been GaAs. Although much progress has been made, this material presents intrinsic barriers to application for quantum information processing. For example, the strong spin-orbit interaction severely limits the relaxation time of electron spins in these types of QDs, and the Ga and As nuclear spins limit the decoherence time.³ Overcoming such limitations requires the investigation of alternative materials, with new properties.

Our group in Delft has chosen to explore carbon nanotubes (CNTs) as a novel system for creating high quality QDs that might be useful as elementary qubits. Since they are based on carbon, CNTs have the advantages of having no nuclear spin and very small spin-orbit interaction.

Initially, we studied QDs formed in CNT segments defined by the deposition of top metal electrodes. We formed high quality QDs, with a variety of coupling strengths, in both semiconducting and metallic CNTs. Besides the known effects of Coulomb blockade and energy-level quantization, the peculiar band structure of CNTs enabled us to measure novel types of Kondo effect, such as orbital and SU(4). A variety of materials can be used to make electrical contact to CNTs, and the first superconducting QD has been recently made by depositing superconducting electrodes on CNTs.

In all of these experiments, however, the coupling between the QD and the leads is not controlled. We have therefore developed a novel technique to define QDs, with tunable coupling strength, at arbitrary locations in CNTs (see Figure 1). The method, based on the evaporation of thin electrostatic top gates, has let us create fully tunable double quantum dots in CNTs, a necessary step for more advanced microwave-based experiments to measure the electron spin relaxation and decoherence times in CNTs.

Figure 1(a) shows an atomic-force-microscope image of a CNT contacted by Pd source and drain electrodes, which make good contact to the CNT. On top of the CNT, but separated from it by a thin insulating layer, we have formed very narrow (about 25 – 50nm) electrostatic top gates (TG_L, TG_R, TG_M) of Al. The top gates are formed by evaporating 2nm of Al, then oxidizing this layer, and finally by continuing the evaporation with, typically, 50nm of Al.

By applying voltages to the Al top gates we can locally induce an electrostatic barrier, thus forming QDs. Varying the top-gate voltage allows us to tune the barrier and explore different experimental regimes, which have different degrees of coupling between the dot and the leads. Until quite recently, such tunable barriers were only possible in semiconductor-heterostructure QDs. Traditional side gates are used to change the potential of the individual QDs.

Figure 1(b) shows the current (in color scale) as a function of the left- and right-side gate (SG_L, SG_R) for a fixed source-drain bias voltage of 1mV. The characteristic honeycomb pattern indicates that we have formed a double dot in the CNT, which is in the tunnel-coupled regime.⁴ By applying voltages on the top gates, the barriers are increased and the double dot is brought into the weakly-tunnel-coupled regime, as shown in Figure 1(c). Furthermore, the excited states of the CNT double dot are observed (lines inside the triangular regions).

These results⁵ open up the road to much exciting research. The ability to create a CNT double quantum dot with controllable barriers, together with the expected long orbital and spin relaxation times make it an ideal candidate for a qubit, the elementary building block of a quantum computer.