We fabricate and characterize superconducting through-silicon vias and electrodes suitable for superconducting quantum processors. We measure internal quality factors of a million fortest resonators excited at single-photon levels, when vias are used to stitch ground planes on the front and back sides of the wafer. This resonator performance is on par with the state of the art for silicon-based planar solutions, despite the presence of vias. Via stitching of ground planes is an important enabling technology for increasing the physical size of quantum processor chips, and is a first step toward more complex quantum devices with three-dimensional integration.
Superconducting devices, based on the Cooper pairing of electrons, are of outstanding importance in existing and emergent technologies, ranging from radiation detectors to quantum computers.Their performance is limited by spurious broken Cooper pairs also known as quasiparticle excitations. In state-of-the-art devices, the time-averaged number of quasiparticles can be on the order of one. However, realizing a superconductor with no excitations remains an outstanding challenge. Here, we experimentally demonstrate a superconductor completely free of quasiparticles up to seconds. The quasiparticle number on a mesoscopic superconductor is monitored in real time by measuring the charge tunneling to a normal metal contact. Quiet excitation-free periods are interrupted by random-in-time events, where one or several Cooper pairs break, followed by a burst of charge tunneling within a millisecond. Our results vindicate the opportunity to operate devices without quasiparticles with potentially improved performance. In addition, our present experiment probes the origins of nonequilibrium quasiparticles in it; the decay of the Cooper pair breaking rate over several weeks following the initial cooldown rules out processes arising from cosmic or long-lived radioactive sources.
Radiation sensors based on the heating effect of the absorbed radiation are typically relatively simple to operate and flexible in terms of the input frequency. Consequently, they arewidely applied, for example, in gas detection, security, THz imaging, astrophysical observations, and medical applications. A new spectrum of important applications is currently emerging from quantum technology and especially from electrical circuits behaving quantum mechanically. This circuit quantum electrodynamics (cQED) has given rise to unprecedented single-photon detectors and a quantum computer supreme to the classical supercomputers in a certain task. Thermal sensors are appealing in enhancing these devices since they are not plagued by quantum noise and are smaller, simpler, and consume about six orders of magnitude less power than the commonly used traveling-wave parametric amplifiers. However, despite great progress in the speed and noise levels of thermal sensors, no bolometer to date has proven fast and sensitive enough to provide advantages in cQED. Here, we experimentally demonstrate a bolometer surpassing this threshold with a noise equivalent power of 30zW/Hz−−−√ on par with the current record while providing two-orders of magnitude shorter thermal time constant of 500 ns. Importantly, both of these characteristic numbers have been measured directly from the same device, which implies a faithful estimation of the calorimetric energy resolution of a single 30-GHz photon. These improvements stem from the utilization of a graphene monolayer as the active material with extremely low specific heat. The minimum demonstrated time constant of 200 ns falls greatly below the state-of-the-art dephasing times of roughly 100 {\mu}s for superconducting qubits and meets the timescales of contemporary readout schemes thus enabling the utilization of thermal detectors in cQED.
Quantum data is susceptible to decoherence induced by the environment and to errors in the hardware processing it. A future fault-tolerant quantum computer will use quantum error correction(QEC) to actively protect against both. In the smallest QEC codes, the information in one logical qubit is encoded in a two-dimensional subspace of a larger Hilbert space of multiple physical qubits. For each code, a set of non-demolition multi-qubit measurements, termed stabilizers, can discretize and signal physical qubit errors without collapsing the encoded information. Experimental demonstrations of QEC to date, using nuclear magnetic resonance, trapped ions, photons, superconducting qubits, and NV centers in diamond, have circumvented stabilizers at the cost of decoding at the end of a QEC cycle. This decoding leaves the quantum information vulnerable to physical qubit errors until re-encoding, violating a basic requirement for fault tolerance. Using a five-qubit superconducting processor, we realize the two parity measurements comprising the stabilizers of the three-qubit repetition code protecting one logical qubit from physical bit-flip errors. We construct these stabilizers as parallelized indirect measurements using ancillary qubits, and evidence their non-demolition character by generating three-qubit entanglement from superposition states. We demonstrate stabilizer-based quantum error detection (QED) by subjecting a logical qubit to coherent and incoherent bit-flip errors on its constituent physical qubits. While increased physical qubit coherence times and shorter QED blocks are required to actively safeguard quantum information, this demonstration is a critical step toward larger codes based on multiple parity measurements.
A challenge for scaling up quantum processors using frequency-crowded, weakly anharmonic qubits is to drive individual qubits without causing leakage into non-computational levels ofthe others, while also minimizing the number of control lines. To address this, we implement single-qubit Wah-Wah control in a circuit QED processor with a single feedline for all transmon qubits, operating at the maximum gate speed achievable given the frequency crowding. Randomized benchmarking and quantum process tomography confirm alternating qubit control with ≤1 average error per computational step and decoherence-limited idling of one qubit while driving another with a Wah-Wah pulse train.