High-fidelity elementary operations for superconducting quantum computers

Superconducting circuits provide one of the leading architectures for the realization of a large-scale quantum computer, with potential future applications, e.g., in quantum physics simulations, quantum chemistry, and cryptography. To provide an advantage over classical computers, a quantum computer must contain a sufficiently large register of quantum bits, i.e., qubits, while enabling accurate and fast elementary operations on the qubits, including qubit initialization, single- and two-qubit quantum logic gates, and readout of the qubit state. In current superconducting quantum computers based on transmon qubits, these elementary operations are still too error-prone to achieve a computational advantage in problems of practical relevance mainly due to decoherence arising from noise, leakage errors to non-computational states, control imperfections, and unwanted residual interactions. In this thesis, we study novel circuits and control techniques for improving the speed and fidelity of elementary operations in superconducting quantum computers, with the focus on single- and two-qubit logic gates, and the initialization of superconducting circuits. First, we theoretically propose and experimentally demonstrate a novel superconducting qubit coined the unimon that is based on a Josephson junction embedded in a gradiometric superconducting resonator. The unimon is protected against low-frequency charge noise by design, while enabling high anharmonicities of up to 750 MHz and fast single-qubit gates with a fidelity of 99.9% in our first experimental devices. Second, we investigate novel analytical control methods for suppressing leakage errors in fast single-qubit gates suffering from frequency crowding. In experiments on a transmon qubit, we demonstrate a 20-fold reduction in the leakage error of 6.25-ns single-qubit gates in comparison with previous state-of-the-art techniques. Third, we realize a long-distance tunable coupler for a scalable transmon-based architecture that enables fast two-qubit CZ gates with a fidelity of 99.8%, a low residual ZZ interaction, and sufficient space for individual Purcell filters needed for fast qubit readout. Finally, we study the use of voltage-biased normal-metal–insulator–superconductor (NIS) junctions as a tunable dissipative environment for superconducting circuits, which allows us to characterize partly cryogenic amplification chains, to study fast initialization of a superconducting resonator, and to observe a broadband Lamb shift in a superconducting resonator. In the future, the methods presented in this thesis may improve the speed and fidelity of elementary operations in superconducting quantum processors, thus bringing the realization of useful quantum computers one step closer. Some of the proposed techniques, such as the analytical control pulses for fast logic gates, are essentially agnostic of the qubit type, and may even find applications in other quantum computing architectures.