Search

Quantum computers (QC) are predicted to solve relevant problems beyond the reach of even the most powerful supercomputers today. However, to do so, their essential building blocks called quantum bits, or qubits, need to scale up in both quantity and quality. Qubits formed by the spin states of electrons or holes, captured in semiconductor quantum dots are a promising candidate for such large-scale QC. These spin qubits exhibit extremely long quantum coherence times, consist of materials compatible with semiconductor industry, and high fidelity single and two-qubit gates have been demonstrated by research labs around the world. Yet, spin qubit numbers are still in the single-digit regime, while fault-tolerant QC will probably require millions of qubits. Low yield, non-uniformities, and high charge noise are important limitations that slow down further progress. Here, we detail the advances towards spin qubit development as part of imec's quantum computing program. The work is centred around four components: integration, characterization at high and low temperature, and design and modelling. We detail a 300mm fabrication platform which combines optical and e-beam lithography for high flexibility and throughput. Room temperature test suites, including custom probecards, allow to rapidly verify fabrication, material, and design parameter impact on the qubit devices as well as collect wafer-scale data for statistical analysis. This data can then be correlated to lower temperature device behaviour. To further test qubit performance, we cool down a double quantum dot with a novel triple poly-silicon gate stack to milli-Kelvin temperatures. We demonstrate unprecedented and uniform control over single electron tunnel coupling. Tunnel coupling control is achieved for both nMOS and pMOS devices in high and lowest charge carrier occupation numbers. The measured charge sensor's charge noise is on par with state-of-the-art devices. We demonstrate first steps towards spin qubit operation. First, we show holespin readout using Pauli spin blockade and manipulation through the intrinsic hole spin-orbit interaction (SOI). Single-shot electron spin readout using the Elzerman readout allows for magnetospectroscopy, revealing a relaxation hotspot mediated by spin-valley mixing. The electron spin resonance linewidth agrees with donor bound spin qubits in natural silicon. To combine the best of electron and hole spin properties, we propose optimized micromagnet (MM) designs that allow to create an artificial SOI for electron spin manipulation. With a multi-physics modelling approach we estimate the MM induced dephasing rates can be improved up to three orders of magnitude compared to state-of-the-art. Finally, to increase measurement throughput at low temperature, we show our progress towards radio-frequency based readout which could help speed up basic qubit characterization by orders of magnitude. We propose two approaches towards impedance matching, superconducting spiral inductors and a novel matching circuit based on a hybrid coupler. Preliminary data supports basic circuit functionality. Together, these components form a closed design cycle which can be further used to optimize single qubit quality in terms of materials, fabrication, characterization and design. Next, because of the demonstrated high device yield and uniformity, we foresee a transition can soon be made towards smallscale qubit arrays. This would be an important step towards the development of quantum error corrected systems based on silicon spin qubits. status: published