Kerr nonlinearity and three-wave mixing in superconducting resonators hosting Al-InAs weak links

AI summary

Overview

Research area: Superconducting quantum circuits / hybrid superconductor–semiconductor device physics; specifically, nonlinear microwave resonators for parametric amplification and bosonic-mode control.

Technical level: Intermediate (readers will benefit from familiarity with Josephson junctions, Andreev bound states, and resonator quality factors, but the central question is stated plainly).

Scope: A single experimental study of a NbTiN microwave resonator hosting two Al-capped InAs nanowire weak links, characterizing how the Kerr (quartic) and three-wave-mixing (cubic) nonlinearities depend on gate voltage and magnetic flux, and demonstrating three-wave-mixing amplification.

What This Paper Is About

Superconducting resonators become useful for quantum information when they are deliberately made nonlinear, but the "good" nonlinearity (cubic, enabling three-wave mixing) usually comes bundled with an unwanted "bad" one (Kerr, which limits amplifier dynamic range and degrades bosonic operations). The goal is therefore a resonator whose cubic nonlinearity is sizable while the Kerr term can be tuned to zero. This paper asks whether semiconducting weak links — Al-capped InAs nanowires — can provide that tunability through two independent knobs: electrostatic gating of junction transparency and magnetic flux through superconducting loops.

Key Contributions

1. First experimental map of ABS-induced nonlinearity versus both control parameters. The authors measure the Kerr nonlinearity K and cubic nonlinearity |g₃| as functions of junction phase (flux) and gate voltage, a dependence the paper states had previously only been discussed theoretically, with experiments limited to power-dependent state discrimination.

2. Experimental demonstration of three-wave mixing in a superconductor–semiconductor hybrid device. Pumping at twice the resonator frequency produces on-off gain exceeding 20 dB, and the authors state this is, to their knowledge, the first experimental demonstration of a nonzero cubic nonlinearity in such hybrid devices.

3. Demonstration that K can be tuned through zero with either knob. For Datasets I and II, K/2π ≈ 0 kHz near φ₁ = 0.8π and 1.2π, where |g₃| reaches approximately 35 kHz and 70 kHz respectively.

4. A double-junction configuration described by the sum of single-junction contributions. With both weak links active, biasing one junction shifts the K and g₃ curves of the other, adding a further degree of control, and this yields a 25 dB on-off gain with a −1 dB compression point of −138 dBm.

Main Findings

• Bare resonator parameters: With both junctions pinched off (Ṽ₁ = −10 V, Ṽ₂ = −8 V), the resonator has ω_r⁰/2π = 6.759 GHz, internal quality factor Q_i⁰ = 2.1 × 10⁴, and external quality factor Q_c⁰ = 2.5 × 10³.
• Flux and gate both tune the linear response: Activating one junction produces the periodic SQUID-like modulation of the resonance frequency versus flux, while the gate dependence is irregular, attributed to the opening of conduction channels of varying number and transparency. The quality factor is also modulated, with the resonator remaining overcoupled near the maximum of ω_r(I₁) and approaching critical coupling near the minimum.
• Three-wave mixing observed: With a pump at ω_p/2π = 13.518 GHz and power −56 dBm, comparing probe sweeps with and without the pump gives an on-off gain exceeding 20 dB in the degenerate condition ω_p = 2ω_s.
• Kerr nonlinearity is tunable to zero: K stays weakly negative around φ₁ = 0, 2π and reaches a maximum near φ₁ = π. Its amplitude grows from Dataset I to II and shows a discontinuous pattern in Dataset III, following the corresponding avoided crossing in δω_r.
• Cubic nonlinearity has a symmetric double-peak structure: |g₃| shows two peaks symmetric about φ = π whose amplitude and position depend on junction transparency; no gain is observed around φ₁ = π over a phase range that widens from Dataset I to III, which the authors attribute to the combination of increasing Kerr and reduced internal quality factor.
• Model agreement is qualitative: Equations based on single- and two-channel ABS models reproduce δω_r, and the approximated expressions for g₃ and K agree qualitatively with all three datasets. For Dataset III, adding the second channel improves the linear fit but worsens the K fit around φ₁ = π.
• Participation ratio: A value λ = 0.028 is estimated from material characterization and electromagnetic simulations, but the data are best described with λ as a free fit parameter, giving values from 0.020 to 0.035.
• Double-junction control: Biasing φ₁ = π shifts K(φ₂) vertically by K(φ₁ = π) = −70 kHz. Biasing JJ₁ near the |g₃(φ₁)| maxima maintains large cubic nonlinearity over a broad range of φ₂, enabling amplification over a wider frequency range with less pump power for a given gain.
• Amplifier performance in the optimized double-junction point: Maximum on-off gain of 25 dB, bandwidth of approximately 2 MHz, signal-to-noise improvement ΔSNR = 20 dB, and a −1 dB compression point of −138 dBm. The authors note that quantitative assessment is limited by the lack of a calibrated input attenuation and amplification chain.
• No Andreev molecule effects: None were observed, consistent with the junction separation of about 4 μm substantially exceeding the superconducting coherence length of hundreds of nm.
• No bare-resonator Kerr background: A power sweep with both junctions pinched off shows no power-induced frequency shift in the probed range, so the reported Kerr is not from NbTiN or Al kinetic inductance.

Methodology in Plain English

The team built a microwave resonator from NbTiN and placed an Al-capped InAs nanowire near its voltage node, then etched away a roughly 250-nm section of the aluminum shell in two places to create two weak links separated by 4 μm. Each weak link sits in a superconducting loop, so the phase across it can be set by an external flux line, and each sits above gate lines, so its transparency can be set electrostatically. Because flux and gate lines influence each other, the authors first measure and compensate that crosstalk with a linear transformation, which lets them treat the two knobs independently.

To read out the device they drive the resonator through a feedline and measure the reflection coefficient S₁₁ in a dilution cryostat at 10 mK. They extract the resonance frequency and quality factor using a circle fit. To get the Kerr nonlinearity they sweep probe power, convert power to intracavity photon number, and fit the resulting frequency shift to a form quadratic in photon number. To get the cubic nonlinearity they apply a pump tone at twice the resonance frequency, record the gain profile with and without the pump, fit a Lorentzian to that profile, and invert the gain formula to solve for g₃. The experimental data are compared against an effective model in which each weak link hosts Andreev bound states with energy depending on the superconducting gap, the phase, and a gate-tunable transparency, treated phenomenologically with an effective gap smaller than the aluminum gap and an average transparency across channels.

Why This Matters

Impact on research. Semiconductor–superconductor hybrid Josephson elements offer an electrostatic knob that tunnel-junction circuits such as the SNAIL or ATS do not have. This work shows that the same device platform that already demonstrated four-wave-mixing amplification can also supply a three-wave-mixing (cubic) nonlinearity with a Kerr term that can be nulled, and it benchmarks the result against an ABS-based model — providing both a design target and a modeling recipe for hybrid nonlinear circuits.

Real-world applications (as framed by the paper):

• Parametric amplifiers for dispersive readout of superconducting qubits, where Kerr suppression enlarges dynamic range.
• Quantum control of bosonic modes, including non-Gaussian operations that require cubic interactions.
• Beamsplitter-type coupling between microwave modes, useful for bosonic-mode quantum computing architectures.
• Continuous-variable quantum computing and engineered mode interactions in microwave circuits.

Industry relevance. Three-wave-mixing amplifiers are attractive because they need only a single pump tone at twice the signal frequency, simplifying cryogenic microwave hardware. A gate-tunable nonlinear element also means a single fabricated design could be re-tuned in situ rather than re-fabricated, which matters for scaling microwave control and readout hardware in quantum computing. The authors flag that amplifier performance is not yet quantitatively benchmarked, so near-term industrial uptake would depend on calibrated noise and compression measurements.

Future Directions

• Full fourth-order theory. The authors state that a fourth-order expansion of the Hamiltonian including the ABS pair degrees of freedom is beyond the present scope; it is needed to explain why the two-channel model worsens the K fit for Dataset III around φ₁ = π, and whether inter-channel coupling terms are responsible.
• Calibrated amplifier characterization. Noise performance and compression must be measured with a calibrated input attenuation and amplification chain; the paper explicitly says further studies are required to optimize amplifier performance and identify limiting factors.
• Shorter junction spacing. Fabricating devices with much closer junctions would allow measurement of nonlinearities induced by the Andreev molecule potential, which was not observed here given the ~4 μm separation.
• Extending the operating point. Because flux noise and the steep phase dependence of K(φ₁) near π limit the accessible range of the K(φ₂) = 0 operating point, engineering around that sensitivity is an open design question.

Target Audience

Researchers and graduate students in superconducting quantum circuits, quantum-limited microwave amplification, and hybrid superconductor–semiconductor materials. It is most useful to device physicists designing nonlinear microwave elements for qubit readout and bosonic-mode architectures, and to theorists modeling Andreev bound state contributions to resonator nonlinearity. Readers seeking a complete amplifier noise characterization will find that the paper reports it is not yet available.

Authors’ abstract

Nonlinear microwave resonators are a versatile tool in quantum information processing, enabling parametric amplification, continuous variable quantum computing, and engineered mode interactions. Many of these applications especially benefit from cubic nonlinearities enabling three-wave mixing; at the same time, they are limited by quartic nonlinearities giving rise to undesired Kerr effects. A recurrent challenge is therefore to engineer resonators with a finite cubic nonlinearity while suppressing quartic terms. Here, we investigate a superconducting resonator hosting two weak links fabricated from an aluminum-capped indium arsenide nanowire. We characterize the Kerr nonlinearity as a function of magnetic flux and gate bias, showing that it can be tuned to zero with either control parameter. Furthermore, we experimentally demonstrate three-wave mixing in a semiconductor-superconductor hybrid device, establishing nonzero cubic nonlinearity. An effective model based on Andreev bound states qualitatively captures the observed trends. Our results validate semiconductor-superconductor hybrid devices as a promising platform for tunable nonlinear superconducting circuits, with applications in parametric amplification, quantum control of bosonic modes, and engineering interactions between microwave modes.