Coherence Limits in Interference-Based cos(2 phi) Qubits

arXiv:2601.10209 · quant-ph, cond-mat.mes-hall · Submitted 2026-01-15 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Quantum Radio. Generated commentary on the latest quantum physics and condensed matter papers.

Kai: Today's paper: "Coherence Limits in Interference-Based cos(2 phi) Qubits".

Mira: Qubit implementations of a cos(2φ) potential using interferences between two Josephson elements in a superconducting loop can be described by the same Hamiltonian as two multiharmonic Josephson junctions in a…

Kai: First, who's behind it and why it matters.

Paper summary: Kai: So, we're looking at this paper, "Coherence Limits in Interference-Based cos(2φ) Qubits," and what I'm seeing is that they’re exploring how these parity-protected qubits actually behave when you start adding noise <ref:2601.10209#pg0,Coherence Limits in Interference-Based cos(2φ) Qubits>. It seems like the core idea is using interferences between two Josephson elements to build a qubit that resists energy relaxation through a specific symmetry, which I want to make sure we get right before we go deeper into the physics.

Mira: Right, Kai, what strikes me immediately is their focus on the fundamental trade-offs between charge and flux noise channels; they claim that even with this protection against energy relaxation, you still run into limitations when it comes to dephasing time Tφ. I'm curious if they really nail the underlying physics of those noise channels or if it's just a parameter tuning exercise.

Lev: From my end, what I look for is how these theoretical limits translate into actual hardware requirements for error correction; specifically, how robust these systems are when you try to scale them up or integrate them into a larger architecture where noise is unavoidable. I need to know if the protection they describe actually holds up under realistic noise conditions.

Kai: Exactly, Lev; the paper investigates various physical platforms, including superconducting circuits and even semiconductor weak links like Germanium and InAs nanowires, which gives us a good idea of what's currently being built versus what’s theoretically proposed. I want to hear what they found regarding those specific implementations.

Mira: That diversity in platforms is interesting because it shows the universality of the interference-based approach, regardless of whether you are using superconducting loops or these semiconductor junctions; they describe it all under a single Hamiltonian (<ref:2601.10209#pg1>). I'm wondering how sensitive those coherence limits are to the specific junction ratios they tested, like the EJΣ2 /EJΣ1 ratio which ranged from-zero point two in one case down to-zero point zero four in another (<ref:2601.10209#pg2>).

Paper summary: Lev: If the authors are finding that these ratios dictate the performance of those specific implementations, that suggests we have a clear design space for hardware engineers to aim for when building these devices, focusing on maximizing or minimizing those energy and charging energies relative to each other. I think this gives us concrete targets rather than just abstract theory.

Kai: It seems the paper points out two main figures of merit they use to assess the protection from energy relaxation, M1φ and M2φ, and they claim that for low junction asymmetry, like a d of one percent, these values can be several orders of magnitude lower than what you see in transmons (<ref:2601.10209#pg2>). I want to understand what that means in practical terms for qubit operation.

Mira: That "several orders of magnitude lower" claim is significant because it directly addresses the energy relaxation protection mechanism they introduced, which relies on encoding states with different parities and only allowing specific Cooper pair tunneling (<ref:2601.10209#pg1>). I'm interested in how those matrix elements relate to the wavefunction symmetries they discuss concerning even versus odd charge parts of the state.

Lev: If the relaxation protection is that strong, it sets a high bar for any error correction code we try to run on these systems; if the fundamental relaxation channel is suppressed this effectively, then we can focus our error correction efforts entirely on managing dephasing noise instead. But what about when you introduce that finite offset flux deltaΦ?

Kai: That's where the paper gets really interesting because it shows that for dephasing time Tφ, there's an unavoidable trade-off between charge and flux noise protection when a loop is introduced (<ref:2601.10209#pg1>). They even suggest that biasing the circuit loop away from the frustration point can help balance susceptibility to these two noise sources, leading to coherence times of several microseconds with accessible parameters.

Mira: That trade-off is the crux of their argument; it means you can't perfectly optimize for both charge noise suppression and flux noise suppression simultaneously in this setup. I wonder if this implies that achieving very long coherence times will always require sacrificing protection against one specific type of environmental noise, or if that balance point they suggest is truly achievable with existing circuit parameters.

Paper summary: Lev: If the optimal configuration is found by biasing the loop away from a specific point, that’s actually very useful information for experimentalists; it gives a direction for tuning the external magnetic flux control to maximize performance on real hardware. I think this moves us closer to designing actual working prototypes rather than just theoretical models.

Kai: So, when we look at the title of "Coherence Limits in Interference-Based cos(2φ) Qubits" and the authors—Smeselot, Leblanc, Tettekpoe, Lefloch, Ficheux, Renard and Dumur—it really highlights that this isn't just another type of qubit; it's a specific engineering solution to a known problem in superconducting circuits <ref:2601.10209#pg0,Coherence Limits in Interference-Based cos(2φ) Qubits>.

Mira: And what this work tells us is that the limitations aren't just about the inherent physics of cos(2φ) but are dictated by the coupling between charge and flux noise channels, which is a more fundamental constraint on coherence than just managing one type of noise in isolation (<ref:2601.10209#pg1>).

Lev: For error correction researchers like myself, this points toward a clear path: we should prioritize designing qubits that can operate near that optimal biasing point they identified, because those are the ones offering the best balance for achieving measurable coherence times. That's where the practical work needs to be directed.

Kai: It’s clear that these results from "Coherence Limits in Interference-Based cos(2φ) Qubits" show us exactly where the current limits of qubit performance lie when we use this interference-based approach, especially concerning how charge and flux noise interact in a loop configuration <ref:2601.10209#pg0,Coherence Limits in Interference-Based cos(2φ) Qubits>.

Mira: The implication for condensed matter theory is that understanding how symmetry dictates wavefunction structure across different physical platforms—from superconducting circuits to InAs nanowires—is key to designing resilient quantum architectures (<ref:2601.10209#pg2>).

Lev: If we can reliably map these theoretical limits onto achievable circuit parameters, then we can start designing experiments that actually test these bounds, which is the next necessary step for any error correction strategy.

Kai: We've covered a lot of ground today discussing the thesis of this paper on how parity protection works and where the current coherence bottlenecks are located in interference-based cos(2φ) qubits <ref:2601.10209#pg0,in interference-based cos(2φ) qubits>.

Conclusion: Kai: So, what's the main point you want to nail down about the title and who wrote this paper?

Mira: The core idea is that they’ve engineered a cos(2φ) potential using interferences between two Josephson elements to protect against energy relaxation through symmetry <ref:2601.10209#pg0,a cos(2φ) potential using>.

Lev: And for us, the real question is what those limits actually mean when we try to build something on a bench with noisy components.

Kai: Exactly, Mira; I want to understand how these specific physical constraints translate into something that can actually be measured in a lab setting.

Mira: The authors Smeselot and colleagues found that this interference-based approach can reduce energy relaxation matrix elements by several orders of magnitude compared to standard transmons when the junction asymmetry is low.

Lev: That level of suppression sounds promising for T1 times, but I'm concerned about the dephasing time Tφ they mentioned; what happens to coherence when you factor in that unavoidable trade-off between charge and flux noise?

Kai: That trade-off is where it gets interesting; they show that even with good protection against relaxation, the dephasing time Tφ remains limited by either charge or flux noise channels.

Mira: The paper suggests that biasing the circuit loop away from a certain frustration point helps balance these two noise sources, which allows for coherence times in the microsecond range using existing parameters.

Lev: If we can tune it to that sweet spot, then maybe we can actually get those microsecond coherence times onto a real quantum computer architecture without needing entirely new materials or components.

Kai: So, the implication is that the interference-based cos(2φ) qubit isn't just a theoretical curiosity; it’s a viable platform for achieving decent coherence if you manage the noise environment correctly <ref:2601.10209#pg0>.

Mira: It really highlights how crucial controlling both charge and flux noise channels is for realizing long-lived qubits in any superconducting circuit design.

Lev: That gives us a clear direction for error correction research; we need to focus our efforts on designing noise environments that respect those optimal biasing conditions they found.

Univ Grenoble Alpes, CNRS, Grenoble INP

quant-ph, cond-mat.mes-hall

Submitted: 2026-01-15

Updated: 2026-01-15

Comments: 19 pages, 14 figures

DOI: 10.1103/x4vn-51dj

License: http://creativecommons.org/licenses/by/4.0/

Importance score: 76/100

The gist: Qubit implementations of a cos(2φ) potential using interferences between two Josephson elements in a superconducting loop can be described by the same Hamiltonian as two multiharmonic Josephson

Key concepts

cos(2φ) Qubit
This qubit uses the shape of a cos(2φ) potential, created by interfering two Josephson elements in a SQUID. It is engineered to protect its energy states from relaxation by having wavefunctions with different parities, which prevents unwanted energy loss.
Energy Relaxation Protection
The system protects against losing energy by only allowing the coherent tunneling of Cooper pairs through specific operators that couple charge states of the same parity. This protection is highly effective when the junction asymmetry is low and the ratio of Josephson energies is high.
Charge and Flux Noise Dephasing
Coherence in this qubit is limited by two noise sources: charge noise, which affects the energy levels, and flux noise, which causes phase fluctuations. The paper finds an unavoidable trade-off where improving protection against one type of noise degrades the coherence time related to the other.
Figure of Merit (M1φ and M2φ)
These mathematical metrics are used to quantify how well the qubit protects against energy relaxation. They show that for low asymmetry, these values can be much smaller than those in transmons, indicating efficient protection from energy loss.

Terminology

Summary

Qubit implementations of a cos(2φ) potential using interferences between two Josephson elements in a superconducting loop can be described by the same Hamiltonian as two multiharmonic Josephson junctions in a SQUID geometry, revealing fundamental trade-offs between charge and flux noise dephasing channels.

How it works

The cos(2φ) qubit is engineered to protect against energy relaxation by leveraging the internal symmetry of the system, encoding ground and excited states with wave functions that have different parities. This protection is achieved by only allowing the coherent tunneling of pairs of Cooper pairs through an operator N⟩⟨N + 2+N + 2⟩⟨N, which couples charge states of the same parity. In practice, this resembles a transmon but utilizes a π-periodic circuit element described by a potential proportional to EJ2 cos(2φ), where EJ1 0.

The specific implementation involves interferences between two bi-harmonic Josephson elements in parallel, assembled in a SQUID configuration. The Hamiltonian is given by Eq. (1), which incorporates the charging energy EC and the Josephson energies for both junctions, resulting in an interference-based cos(2φ) Hamiltonian that includes first harmonic terms from practical realizations. The quantum phase operator obeys flux quantization, and the gauge choice φb = (φbA + φbB)/2 satisfies the irrotational constraint.

Implementation Platforms

The paper investigates various physical platforms capable of implementing this interference-based cos(2φ) qubit, including superconducting circuits and semiconductor weak links. The implemented circuits include:

  1. Rhombus [6, 9] with an estimated EJΣ2 /EJΣ1 ratio of-0.2.

  2. Pinhole JJ [26] with a ratio of-0.1.

  3. KITE (Low inductance) [17] with a ratio of-0.025, and KITE (High inductance) [16] with a ratio of-0.04.

  4. Semiconductor platforms such as Germanium [22], Graphene [24], and InAs nanowires [19].

  5. Heterostructures like Flowermon [27, 34] with ratios ranging from ±0.1 to ±15, and twisted cuprate van der Waals heterostructures [27].

Protection Mechanisms and Wavefunctions

The resilience of the energy relaxation protection is examined by analyzing the matrix elements involved in qubit relaxation. The paper identifies two figures of merit:

M1φ = ⟨0 cos(φb)1⟩ cos(πδΦ) + ⟨0sin(φb)1⟩d sin(πδΦ)

/M2φ = −⟨0 cos(2φb)1⟩sin(2πδΦ) + ⟨0sin(2φb)1⟩d cos(2πδΦ)

The analysis shows that for low junction asymmetry (d = 1 %), the matrix element values for these figures of merit can reach several orders of magnitude lower than transmons, ensuring an efficient protection from energy relaxation. The robustness to finite offset flux δΦ is explained by wavefunction symmetries in the charge basis: the even charge part of the wavefunction is symmetric versus charge inversion, and the odd charge part is antisymmetric for both states.

Coherence Limits and Trade-offs

The primary finding regarding coherence is that there exists a fundamental trade-off between charge and flux noise dephasing channels.

  1. For energy relaxation protection, symmetry requires low asymmetry (d) and a large EJΣ2 /EC ratio to increase the barrier height.

  2. However, for dephasing time Tφ, the introduction of a loop brings an unavoidable trade-off between charge and flux noise protection.

  3. Using currently existing circuit parameters, the qubit lifetime T1 can exceed milliseconds while the dephasing time Tφ remains limited to only a few microseconds due to either flux or charge noise.

  4. The optimal configuration for maximum coherence is found by biasing the circuit loop away from the frustration point, which helps balance susceptibility to these two noise sources, allowing coherence times of several microseconds with accessible parameters.

Operating Regimes and Optimization

The system exhibits two regimes based on magnetic flux dependence: near δΦ ∼ ±0.5 (where it is similar to a transmon) and near δΦ ∼ 0 (where the cos(2φ) behavior is dominant). The study shows that the qubit can still localize wavefunctions in different wells even with finite offset flux and asymmetry, as long as the condition for localization is met.

Improvements for AI systems

Based on the provided scientific paper, here are specific improvements for AI systems derived from the insights into coherence limits of interference-based cos(2φ) qubits:

The core takeaway is that while parity protection against energy relaxation exists, it creates an unavoidable trade-off between charge and flux noise dephasing channels, limiting coherence times to a few microseconds with current parameters.

Here are specific improvements for AI systems:

  1. Improve Noise Resilience in Quantum Machine Learning (QML) Models:

  2. Develop Noise-Aware Circuit Design for Qubit Hardware:

  3. Enhance Robustness of Quantum Algorithms Against Environmental Fluctuations:

  1. Improve Noise Resilience in Quantum Machine Learning (QML) Models:

The paper demonstrates that dephasing times are limited by noise channels (charge and flux noise), and that the protection mechanism is sensitive to finite offset flux and junction asymmetry.

The improved AI system can be a QML model trained on quantum hardware where the coherence time is explicitly modeled as a function of circuit parameters, noise amplitudes, and operating point (flux/charge).

  • Specific Improvement: Implement an adaptive error mitigation strategy that dynamically adjusts the noise-resilient encoding (e.g., switching between charge-noise protection regimes based on real-time flux/charge measurements).

  • Capability: The improved AI can perform complex variational quantum eigensolvers or quantum neural network training with significantly longer coherence times than current state-of-the-art transmons, allowing for the execution of deeper circuits and higher precision computations before decoherence limits the result.

  1. Develop Noise-Aware Circuit Design for Qubit Hardware:

The analysis identifies a critical trade-off: low charge noise protection (large energy ratio) is highly sensitive to flux noise, and vice versa. The paper suggests optimizing circuit parameters (like the ratio of Josephson energies, EJΣ2/EJΣ1) and operating points (flux offset δΦ) to balance these two competing dephasing channels.

  • Specific Improvement: Design superconducting qubit architectures with integrated control systems that actively tune the operating point away from frustration points or sweet spots based on real-time noise spectroscopy feedback.

  • Capability: The improved hardware can achieve coherence times exceeding 100 microseconds (as projected for specific high energy ratios) by balancing charge and flux noise, enabling fault-tolerant quantum operations that require longer gate sequences than currently possible.

  1. Enhance Robustness of Quantum Algorithms Against Environmental Fluctuations:

The paper shows that the residual cos(1φ) term, which breaks the ideal parity protection, is influenced by flux offset and asymmetry in ways that depend on the regime (hybridized vs. localized wavefunctions).

  • Specific Improvement: Develop quantum algorithms specifically tailored to exploit the residual physics of interference-based qubits—using algorithms that are less sensitive to small phase errors or residual single-pair tunneling terms inherent in imperfect implementations. This involves designing gates that are robust against the specific Hamiltonian perturbations identified in Appendix A and B.

  • Capability: The improved AI can execute quantum simulations or search algorithms (like QAOA) on noisy hardware with higher fidelity, as it understands the specific symmetry breaking mechanisms that limit coherence, allowing for more accurate error modeling and mitigation tailored to the cos(2φ) architecture.

Abstract

We investigate the coherence properties of parity-protected (2φ) qubits based on interferences between two Josephson elements in a superconducting loop. We show that qubit implementations of a (2φ) potential using a single loop, such as those employing semiconducting junctions, rhombus circuits, flowermon and KITE structures, can be described by the same Hamiltonian as two multi-harmonic Josephson junctions in a SQUID geometry. We find that, despite the parity protection arising from the suppression of single Cooper pair tunneling, there exists a fundamental trade-off between charge and flux noise dephasing channels. Using numerical simulations, we examine how relaxation and dephasing rates depend on external flux and circuit parameters, and we identify the best compromise for maximum coherence. With currently existing circuit parameters, the qubit lifetime T 1 can exceed milliseconds while the dephasing time T φ remains limited to only a few microseconds due to either flux or charge noise. Our findings establish practical limits on the coherence of this class of qubits and raise questions about the long-term potential of this approach.

Sources

Related papers