🔬 mesoscale physics

Coherence Limits in Interference-Based cos(2 $φ$ ) Qubits

This paper demonstrates that while parity-protected $\cos(2\phi)$ qubits can achieve millisecond relaxation times ( $T_1$ ), their coherence is fundamentally limited to microsecond-scale dephasing times ( $T_\phi$ ) due to an unavoidable trade-off between charge and flux noise, thereby establishing practical constraints on the long-term viability of this qubit architecture.

Original authors: S. Messelot, A. Leblanc, J. -S. Tettekpoe, F. Lefloch, Q. Ficheux, J. Renard, É. Dumur

Published 2026-10-08

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: S. Messelot, A. Leblanc, J. -S. Tettekpoe, F. Lefloch, Q. Ficheux, J. Renard, É. Dumur

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

In the quest to build a quantum computer, scientists are trying to harness the strange behavior of subatomic particles to solve problems that are impossible for today's machines. To do this, they need to create tiny, artificial atoms called qubits that can hold information in a fragile state of superposition. One of the most successful types of these qubits, known as the transmon, is widely used because it is relatively simple to build and can hold its information for a long time. However, even these successful qubits eventually lose their data due to two main enemies: energy decay, where the qubit simply falls out of its excited state, and dephasing, where the delicate timing of the quantum state gets scrambled by tiny fluctuations in electric or magnetic fields. To build a truly powerful computer, researchers are searching for a new kind of qubit that is naturally immune to both of these problems at once.

A promising candidate for this role is the cos(2φ) qubit, a design that relies on a clever trick of quantum interference to protect itself. Imagine a superconducting loop containing two special junctions that allow electricity to flow without resistance. In a standard setup, these junctions act like simple gates that let single pairs of electrons pass through. The cos(2φ) design, however, is engineered so that the two junctions interfere with each other in a way that blocks single electron pairs from tunneling through, allowing only pairs of pairs to pass. This creates a protective barrier that should, in theory, stop the qubit from losing energy. The idea is that by forbidding the simplest way for the system to lose energy, the qubit becomes incredibly stable. However, researchers have found that due to junction asymmetry or imperfect magnetic flux parking, the Hamiltonian retains a residual term that prevents perfect protection. Despite this, the design has been built using various materials, from superconducting circuits to tiny semiconductor wires, hoping to combine this energy protection with the ability to resist the scrambling effects of environmental noise. To date, none have managed to combine this protection with protection against dephasing.

A team of researchers from Grenoble, France, set out to test whether this theoretical promise could hold up in the real world. They did not build a new physical device for this study; instead, they created a detailed mathematical model of the circuit and ran extensive computer simulations to see how it would behave under various conditions. Their goal was to understand the fundamental limits of this design: could it truly protect against both energy loss and dephasing, or would the very features that stop energy loss create new problems? By simulating the behavior of these circuits with realistic imperfections, they mapped out exactly how the qubit would react to the inevitable noise found in any laboratory environment.

The simulations revealed a surprising and fundamental limitation. While the design successfully blocks the flow of single electron pairs, thereby protecting the qubit from losing energy, it creates a difficult trade-off when it comes to resisting dephasing. The researchers found that the circuit cannot be made insensitive to both electric charge noise and magnetic flux noise at the same time using the current interference-based architecture. If the circuit is tuned to be very resistant to electric noise, it becomes extremely sensitive to magnetic fluctuations. Conversely, if it is tuned to ignore magnetic noise, it becomes vulnerable to electric disturbances. This means that the "perfect" protection the designers hoped for is physically impossible to achieve with this specific architecture. The very mechanism that creates the energy barrier also forces the qubit to choose which type of noise it can ignore, leaving it exposed to the other.

The team discovered that the best possible outcome for these qubits, using the materials and construction methods available today, is a coherence time of only a few microseconds. This is the duration the qubit can hold its information before the noise scrambles it. While this is a significant improvement over some other designs, it falls far short of the milliseconds achieved by the standard transmon qubits currently used in quantum computers. The researchers showed that even with perfect symmetry in the circuit components and the most precise control over magnetic fields, the dephasing time remains stuck at this microsecond level. They found that the only way to push this limit higher would be to drastically change the ratio of the circuit's energy components, a feat that would require materials and fabrication techniques that do not yet exist.

Perhaps the most critical finding was that the qubit's ability to find a "sweet spot"—a specific setting where it is naturally immune to noise—depends heavily on the circuit's energy parameters. While a sweet spot exists where the qubit is insensitive to magnetic fluctuations, the researchers found that in the regime where the circuit is most resistant to charge noise (large energy ratios), this sweet spot becomes so narrow that it is effectively non-existent compared to the natural noise in an experiment. However, the study also revealed a practical solution: the qubit can still benefit from the protection of this architecture by operating slightly away from the sweet spot. By carefully balancing the circuit's settings just off this ideal point, researchers can partially balance the competing effects of charge and flux noise, extending the coherence time as much as possible within the fundamental limits of the design.

The study concludes that while the cos(2φ) qubit is a brilliant theoretical concept for stopping energy loss, the interference-based approach used to build it hits a hard wall when it comes to dephasing. The researchers suggest that this specific path may not be the right one for building scalable quantum computers, as the fundamental trade-off between charge and flux noise cannot be overcome with current technology, raising questions about the long-term potential of this approach. They point out that future progress might require entirely different designs, such as a recently proposed architecture featuring a finite junction asymmetry with specific energy properties that could, in principle, circumvent the trade-off identified in their work. For now, the work serves as a clear map of the boundaries of this technology, showing that while the qubit can be made to last longer against energy loss, it cannot be made to last long enough against the noise of the world to be useful in a large-scale computer without significant design changes.

Technical Summary: Coherence Limits in Interference-Based cos(2φ) Qubits

Problem Statement
The transmon qubit, while widely used for its simplicity and long coherence times, remains susceptible to energy relaxation ( $T_1$ ) due to its large charge matrix elements. To address this, "parity-protected" qubits based on a $\cos(2\phi)$ potential have been proposed. These systems, realized via interference between Josephson elements (e.g., Rhombus, KITE, Flowermon, or semiconductor-based junctions), theoretically suppress single Cooper pair tunneling, thereby protecting against energy relaxation. However, despite successful demonstrations of relaxation protection, no experiment has yet combined this with protection against dephasing ( $T_\phi$ ). The central problem investigated is whether the interference-based implementation of the $\cos(2\phi)$ qubit introduces fundamental trade-offs that limit coherence, specifically regarding the interplay between charge and flux noise.

Methodology
The authors investigate a general model of interference-based $\cos(2\phi)$ qubits consisting of two bi-harmonic Josephson junctions in a SQUID geometry, shunted by a capacitance.

Hamiltonian Modeling: They derive a Hamiltonian that includes both first ( $E_{J1}\cos\phi$ ) and second ( $E_{J2}\cos2\phi$ ) harmonic terms for each junction, accounting for junction asymmetry ( $d$ ) and magnetic flux offset ( $\delta\Phi$ ). This model unifies various experimental platforms (superconducting circuits, semiconductor nanowires, heterostructures) under a single theoretical framework.
Numerical Simulation: Using the scQubits and QuTiP Python packages, the authors numerically simulate the system's eigenstates, wavefunctions, and transition frequencies.
Coherence Analysis: They calculate relaxation rates ( $T_1$ ) and dephasing rates ( $T_\phi$ ) by evaluating relevant matrix elements for charge ( $\hat{n}$ ) and flux ( $\hat{\Phi}$ ) noise. They consider realistic experimental parameters, including junction asymmetries ( $d \sim 1\%$ ), flux offsets ( $\delta\Phi \sim 10^{-5}\Phi_0$ ), and thermal limits.

Key Contributions and Results

Fundamental Trade-off Between Charge and Flux Noise:
The primary finding is an unavoidable trade-off between protection against charge noise and flux noise.
- Charge Noise: Suppression of charge noise requires a large ratio of second-harmonic to charging energy ( $E_{J\Sigma 2}/E_C$ ). In this regime, the system behaves like a charge-insensitive qubit.
- Flux Noise: However, in the same regime, the sensitivity to flux noise increases. The suppression of the first harmonic (necessary for the $\cos(2\phi)$ potential) relies on destructive interference at the flux frustration point ( $\delta\Phi = 0$ ). This interference is highly sensitive to flux fluctuations.
- The Conflict: High $E_{J\Sigma 2}/E_C$ flattens charge dispersion (good for $T_\phi$ against charge noise) but makes the qubit frequency linearly dependent on flux offset (bad for $T_\phi$ against flux noise). Conversely, low $E_{J\Sigma 2}/E_C$ reduces flux sensitivity but reintroduces charge sensitivity.
Relaxation Protection ( $T_1$ ):
The study confirms that energy relaxation protection is robust. Even with finite junction asymmetry ( $d \approx 1\%$ ) and small flux offsets, the matrix elements coupling the ground and excited states to the environment remain exponentially suppressed. Consequently, $T_1$ can theoretically exceed milliseconds, limited primarily by dielectric losses or flux noise coupling, but not by the intrinsic parity protection mechanism.
Dephasing Limits ( $T_\phi$ ):
- Current Parameters: With currently achievable circuit parameters (specifically $E_{J\Sigma 2}/E_{J\Sigma 1} \approx -0.1$ , typical of most proposed implementations), the dephasing time $T_\phi$ is limited to a few microseconds. This is significantly lower than the millisecond-scale coherence times of state-of-the-art transmons.
- Optimization: The authors identify that the optimal operating point involves a ratio $E_{J\Sigma 2}/E_C \sim 10$ and biasing the flux slightly away from the exact frustration point ( $\delta\Phi \sim 10^{-5}$ ) to balance the susceptibility to both noise sources.
- Extreme Parameters: Even if the ratio $|E_{J\Sigma 2}/E_{J\Sigma 1}|$ were increased by two orders of magnitude (to $\sim 250$ , as theoretically possible in some heterostructures like the Flowermon), the maximum achievable coherence time ( $T_2$ ) would only reach approximately 20 ms. While an improvement, this requires parameters far beyond current experimental capabilities.
Low-Frequency Limits:
The paper investigates operating the qubit at very low frequencies (near zero) to mitigate dephasing. They find that in the interference-based architecture, the "sweet spot" for flux noise (where flux dispersion vanishes) shrinks exponentially as $E_{J\Sigma 2}/E_C$ increases. In realistic regimes, this sweet spot is smaller than the experimental flux noise floor, rendering the low-frequency strategy ineffective for significantly extending coherence.

Significance and Claims
The paper establishes practical limits on the coherence of interference-based $\cos(2\phi)$ qubits. The authors claim that while these qubits successfully protect against energy relaxation, the interference mechanism required to generate the $\cos(2\phi)$ potential inherently creates a vulnerability to flux noise that cannot be simultaneously eliminated with charge noise protection.

Consequently, the authors raise questions about the long-term potential of this specific approach for scalable quantum processors. They conclude that with currently accessible parameters, the dephasing time is fundamentally limited to the microsecond scale, which is "substantially below the state-of-the-art coherence times demonstrated in transmons." They suggest that future progress may require alternative design strategies, such as interference-based architectures with finite junction asymmetry where $E_{J\Sigma 2} < 0$ , to circumvent this specific trade-off, though such elements remain an open experimental challenge.

More like this