Towards practical quantum computers: transmon qubit with a lifetime approaching 0.5 milliseconds

Quantum computers hold the promise of significantly accelerating computations for tasks like factorization and quantum simulation. Superconducting quantum computing (SQC), compatible with semiconductor technology, is a leading contender in this field. Recent years have witnessed breakthroughs in SQC, but a key challenge has been the relatively short coherence time of superconducting qubits compared to other quantum computing platforms like trapped ions or cold atoms. While coherence times have improved dramatically, reaching tens or hundreds of microseconds with state-of-the-art transmon designs, this is still insufficient for practical SQC systems, particularly for quantum error correction. Longer coherence times are crucial for higher gate fidelity and increased circuit depth. Transmon qubits are widely used due to their favorable coherence properties, ease of coupling, and readout. Their design involves a Josephson junction shunted by a large capacitor. Common materials include aluminum and niobium, but exploring alternative materials for improved coherence remains important. Previous research with tantalum films demonstrated improved coherence times, exceeding 0.3 ms, but utilized wet etching, which is less advantageous than dry etching for large-scale fabrication due to its lack of anisotropy and automation capabilities. This work focuses on developing a robust dry etching process for tantalum films to fabricate high-coherence transmon qubits.

The paper reviews existing literature on superconducting qubits and their materials. It highlights the advantages and disadvantages of various superconducting materials, including aluminum and niobium, which are currently popular choices due to their established fabrication technologies. It discusses previous work on tantalum qubits, noting that while tantalum has shown promise in achieving long coherence times, previous methods involved wet etching, which is less scalable than dry etching. The authors emphasize the need for a dry etching process for tantalum to enable the fabrication of large-scale quantum circuits.

The researchers optimized a dry etching process for tantalum (Ta) films to fabricate transmon qubits. They compared the performance of these qubits with those made from niobium (Nb) and aluminum (Al) using the same design and fabrication processes. The transmon design employed a single-junction, fixed-frequency approach to minimize flux noise. The circuit was minimized to reduce environmental noise, and the capacitor area was enlarged to decrease electric field density and surface loss. For fabrication, a superconducting film (Nb, Al, or Ta) was deposited on a sapphire substrate, patterned using ultraviolet lithography, and etched using either inductively coupled plasma (ICP) or reactive ion etching (RIE). Aluminum Josephson junctions were then fabricated using electron-beam evaporation. The researchers optimized the deposition conditions for Nb and Ta films, aiming for high-quality films with desirable characteristics. Two sets of optimized etching parameters for Ta films, using both ICP and RIE methods, are provided in the paper. Finally, the chips were wire-bonded to a copper submount. The paper also includes details on the characterization methods used to measure the coherence properties (T1 and T2) of the fabricated qubits, including CPMG echo experiments.

The key finding is the achievement of a transmon qubit with a T1 lifetime of 503 μs using tantalum films and a dry etching process. This significantly surpasses the lifetimes obtained for qubits fabricated with niobium and aluminum using the same methods. The average T1 lifetime for the tantalum qubits was 401 μs. The results from different tantalum chips (Ta-2, Ta-3, Ta-4) showed some variation in T1, ranging from 102 μs to 476 μs, attributed to fluctuations in the fabrication process and two-level system defects. The authors also performed a CPMG echo experiment, measuring a T2*CPMG of 557 μs for one of the tantalum qubits. Further analysis indicated that surface oxide layers might play a role in coherence times; surface treatment methods to remove oxides were suggested for potential further improvements. Experiments with a flip-chip design to enhance the metal-air interface showed a reduction in coherence times, suggesting that the metal-air interface is a significant contributor to decoherence. Finally, a 56-qubit chip fabricated with the same process showed lower coherence times than the single-qubit chips, which the authors attribute to increased environmental noise in the multi-qubit system.

The results demonstrate the effectiveness of the dry etching process for tantalum films in fabricating high-coherence transmon qubits. The significantly longer coherence times achieved with tantalum compared to niobium and aluminum highlight the potential of this material system for SQC. The superior performance is likely due to a reduction in two-level system (TLS) defects, which are a primary source of decoherence in superconducting qubits. The study's findings address the critical need for longer coherence times in practical quantum computing. The dry etching technique's scalability also holds significant implications for building large-scale quantum circuits. Further research focusing on surface treatments to minimize the impact of oxide layers could lead to even longer coherence times.

This work successfully demonstrated a dry etching process for tantalum films that yields transmon qubits with a remarkably long T1 lifetime exceeding 500 μs. This surpasses the performance of niobium and aluminum-based qubits fabricated using identical methods, highlighting the potential of tantalum as a superior material for SQC. The scalability of the dry etching technique is crucial for large-scale quantum circuit fabrication. Further optimization, including surface treatments to reduce oxide layer effects, could potentially lead to even longer coherence times, approaching the millisecond range.

The study notes variations in T1 lifetimes among different tantalum chips, indicating sensitivity to fabrication process fluctuations. The coherence times in the 56-qubit chip were lower than in single-qubit chips, highlighting the challenges of controlling environmental noise in larger-scale systems. While tantalum showed superior performance, the exact reasons for the improvement over niobium and aluminum remain to be fully elucidated, requiring further investigation. The effects of surface oxides also need further exploration.