Overview

This project focuses on developing novel qubit fabrication techniques with the goal of increasing coherence times. In quantum computing, the coherence time (also sometimes called the $T_1$ time), refers to the amount of time a qubit can remain in its first excited state before inevitably returning to the ground state due to interactions with the environment. To date, the primary loss mechanisms we are trying to minimize are so-called two level system (TLS) losses, which are generally introduced by surface roughness, residues, and other forms of “damage” that can naturally arise from the fabrication process.

To start, the central component in a quantum circuit is the Josephson Junction (JJ), which can be thought of as a nonlinear inductor.¹ Physically, the JJ is realized as two superconducting metal leads, with an insulating barrier in between. The idea is essentially that when the metals become superconducting, Cooper pairs² from one metal are able to tunnel through the barrier from one superconductor to the other, and we can control the rate of tunnelling which then allows us to control the qubit transitioning between the $\ket{0}$ and $\ket{1}$ state. There’s a lot of extremely interesting physics I’m skipping here, but that’s the basic idea.

Another component central to a quantum processor are the capacitor pads that the JJs are connected to. The capacitors, when combined with the inductive JJ element, form a LC resonator with a resonance frequency, and its this resonance that allows us to drive the qubit with electric signals from voltage and current sources. For this project, the capacitors are relevant because they are the component that we need to attach our JJ to in order to make a proper circuit.

In essence, here’s the problem we are trying to solve: because JJs are on the order of hundreds of nanometers thick, they cannot be patterned using photolithography – these machines don’t have the sub-micron level precision required to make the junctions, and instead we need to rely on electron beam (e-beam) lithography. Therefore, our lithography is separated into two steps: we first pattern large features like feedlines and capacitor pads, then pattern the JJs in a separate e-beam lithography step. As a result of this two-step process, a lossy oxide will grow on our capacitor pads, which we then need to remove in order to ensure an ohmic contact between the capacitor and JJ pads.

Historically, the method to remove these oxides was using an ion mill, which uses ionized Argon (or some other noble gas) to mechanically remove the oxide ensuring ohmic contact. However, this is non-ideal, since the ion mill will also damage the silicon surface by increasing its surface roughness, which can be a source of TLS loss. The goal of this project is to find alternative chemical processes that can be done to reduce the ion milling step, thereby reducing the damage to the silicon and thus eliminating one source of TLS loss.

Experimental Procedure

To investigate different chemical processes, we first fabricate an array of test structures on which we can test different chemical procedures, and evaluate their efficacy in generating ohmic contact. For us, the array looks as follows:

[INSERT IMAGE HERE]

Shaded in red are the large pads which we pattern using photolithography, and the regions in blue are the regions where we use e-beam lithography. Therefore, the region where we have both red and blue are where we are testing the ohmic contact, since these regions will have aluminum deposited from the photolithography step, and then have an additional aluminum layer from the e-beam step. Embedded in this design is also a reference channel where we deposit the entire structure in one deposition (in the photolithography step).

Once both deposition steps are complete, we then use a probe station to measure the DC resistance through the channel. The working principle is as follows: if an ohmic contact is achieved, then we should get a near-zero resistance across all structures. However, if an oxide barrier (or some other species) forms, then it would show up as some large resistance. Typically, our shorts usually measure anywhere from 10 to 20 ohms, whereas oxides will generally result in mega-ohm scale resistances, so it’s pretty easy to tell when an oxide is forming.

1. The nonlinearity of the JJ is important for reasons pertaining to driving the qubit, which I won’t go into detail here. ↩

2. These are pairs of entangled electrons that give rise to the phenomenon of superconductivity, where current can be passed through a metal with zero resistance under the right conditions. The fundamental mechanism behind their creation is given by the Bardeen-Cooper-Schrieffer (BCS) theory, which deserves a deep dive on its own, so I won’t go more into it here. What’s important for now is that it’s a fundamentally quantum phenomenon, and is the central mechanism that defines a qubit. ↩