One of the hardware challenges in realizing a fault-tolerant universal quantum computer is that a large number of physical qubits are required to implement error-correcting codes, and in the case of superconducting qubits, the number is said to be as many as 10⁸ with a typical error rate of 0.1%.
In this research and development theme, to solve this problem, we aim to improve the coherence time of the qubits, develop manufacturing technology of high-quality quantum bits to reduce the error rate, and contribute to reducing the number of quantum bits needed for the realization of fault-tolerant universal quantum computers. In addition, we will develop quantum bit manufacturing technology using optical lithography and multi-layer processes in order to improve the productivity and the uniformity of quantum bits for future large-scale circuits.
We will also conduct exploratory research on bosonic codes, which are expected to enable fault-tolerant quantum computation with fewer physical qubits than those required in the current mainstream of the surface code, to identify their possibilities and promising schemes.

① Development of high-coherence superconducting qubits

Recent studies revealed that dielectric loss at the interface layers such as that between the substrate and the superconducting metal is one of the main sources of decoherence in the superconducting qubits. We try to realize high-quality Nb-based transmon qubits through the optimization of the fabrication processes, particularly paying attention to those interfaces. By clarifying the fabrication process which has large impact on the qubit lifetime (T₁), we aim to establish the fabrication recipe to produce high-quality qubits in a reproducible way.
Recently, we focused on the niobium etching process and attempted to optimize conditions such as the gas species and pressure used in etching. The figure at the bottom left shows an example of a cross-sectional view of the electrode after etching using SF₆ as the etching gas. We achieved a relatively smooth Si surface with minimal over-etching. Using these conditions, we fabricated a transmon qubit and evaluated its T₁; as shown in the figure at the bottom right, we obtained values averaging over 200 us. We speculate that this is due to the improved smoothness of the Si surface and the suppression of contaminants adhering to or remaining on the Si surface and the sides of the Nb superconducting electrodes, both of which resulted from the optimized etching conditions.

② Research and development of bosonic codes using superconducting resonators

Error-correcting codes, known as bosonic codes, are a method for protecting quantum information from errors by utilizing the degrees of freedom found in the energy levels of resonators. Because these energy levels are theoretically infinite, bosonic code can potentially reduce the number of physical qubits required compared to conventional error-correcting codes. In our efforts to improve the Q factor of a coaxial resonator (cavity) made of niobium, which is a key component of a bosonic qubit, we improved surface treatment of the cavity. As a result, we achieved a world-leading Q factor of 3 × 10⁹ and confirmed that this performance remains stable under thermal cycling. Furthermore, we built a bosonic qubit consisting of a three-dimensional memory cavity, an auxiliary qubit, and a readout resonator. By using automatic differentiation to generate optimized control pulses, we successfully encoded the logical states $\mid 0_L⟩$ and $\mid 1_L⟩$ of the binomial (1,1) code and performed quantum gate operations. In this work, we achieved fidelity exceeding 80–90%.