Universal bound on microwave dissipation in superconducting circuits

This paper establishes a universal empirical scaling relation between microwave dissipation and superfluid density across diverse superconducting materials and geometries, revealing an intrinsic bulk dissipation limit caused by nonequilibrium quasiparticles trapped in disorder-induced gap variations that sets a fundamental bound on superconducting qubit coherence.

The Gist

The Big Picture: The "Perfect" Wire That Isn't Perfect

Imagine you are trying to build a super-fast, super-quiet computer using tiny circuits made of special metals that conduct electricity without any resistance (superconductors). In theory, these metals should be perfect. If you send a microwave signal (like a radio wave) through them, it should bounce around forever without losing any energy, just like a ball rolling on a perfectly frictionless track.

However, in the real world, these circuits lose energy. They get "tired" and stop working after a short time. This loss of energy is called dissipation. For quantum computers to work, we need these circuits to hold onto their energy for as long as possible.

The authors of this paper asked a simple question: Why do these "perfect" wires still lose energy, and is there a hard limit to how good they can get?

The Discovery: A Universal "Speed Limit"

The researchers gathered data from hundreds of experiments involving different types of superconducting metals (like Aluminum, Niobium, Titanium Nitride, and some very messy, disordered alloys). They looked at two main things for each experiment:

1. How much energy was lost? (Measured by something called the "Quality Factor," or $Q_i$).
2. How "stiff" was the supercurrent? (Measured by something called "superfluid density," which relates to how many electrons are working together).

When they plotted all this data on a graph, they found a surprising pattern. It looked like a giant, invisible wall. No matter what material they used or how they built the circuit, the data points never went above a specific diagonal line.

The Analogy: Imagine a highway with a strict speed limit. No matter how powerful your car is (the material), no matter how good your driver is (the engineering), you simply cannot go faster than the limit. The paper found that the "speed limit" for how long a quantum circuit can hold energy is directly tied to the material's internal "stiffness."

The Culprit: Trapped "Ghost" Particles

So, what is causing this energy loss? The paper rules out the usual suspects. Usually, scientists blame "dielectric loss," which is like friction caused by the air or the surface of the road. But the researchers found that even when they cleaned the surfaces perfectly and removed the air, the energy loss remained.

Instead, they identified the culprit as nonequilibrium quasiparticles.

The Analogy: Think of the superconductor as a crowded dance floor where everyone is holding hands and dancing in perfect unison (this is the supercurrent).

• Disorder: In some materials, the floor is uneven or has bumps (disorder).
• The Ghosts: Occasionally, a dancer gets bumped, lets go of their partner, and becomes a "ghost" (a quasiparticle).
• The Trap: Because the floor is bumpy, these ghosts get stuck in the low spots (trapped in disorder-induced gaps). They can't get back to the dance floor easily.
• The Loss: When the microwave signal tries to push the dancers, these trapped ghosts get in the way, absorbing energy and slowing the whole system down.

The paper suggests that the number of these "ghosts" is set by a universal rule based on the material's disorder. You can't just clean the surface to get rid of them; they are trapped deep inside the material's structure.

The Two Different Rules of the Road

The paper actually found two different "speed limits" depending on the shape of the circuit:

1. The "Bulk" Limit (The Material Rule):
   For 3D boxes (like hollow metal cavities) and very clean materials, the limit is set by the "ghosts" trapped inside the metal. The more disordered the metal is, the more ghosts get trapped, and the more energy is lost. This explains why some messy materials have lower performance limits than clean ones.

2. The "Floor" Limit (The Substrate Rule):
   For flat, 2D circuits (like chips sitting on a silicon wafer), there is a second, lower ceiling. Even if the metal is perfect, the circuit loses energy because of the substrate (the board it sits on).
   The Analogy: Imagine a high-performance race car (the superconductor) driving on a track. Even if the car is perfect, if the track itself is made of soft mud (the substrate), the car will sink and lose speed. The paper found that for flat chips, the "muddy track" of the silicon or sapphire substrate creates a hard limit around $Q_i \approx 10^7$, preventing them from reaching the higher limits seen in 3D boxes.

What This Means for the Future

The paper concludes that we have found an empirical ceiling for how good these circuits can get.

• If you want the absolute best performance, you need to use materials with the highest "superfluid density" (like Niobium) and build them in 3D shapes to avoid the "muddy track" of the substrate.
• We cannot simply make the surfaces cleaner to break this limit; the limit comes from the material's own internal structure and the trapped "ghosts" inside it.

In short, the universe has set a maximum score for how long these quantum circuits can "sing" before they go silent, and that score depends on the material's DNA and how it's built. To go higher, we need to change the materials or the architecture, not just polish the surface.

Technical Summary

1. Problem Statement

The development of scalable, fault-tolerant superconducting quantum processors is fundamentally limited by the coherence time ($T_1$) of qubits. While superconductors are theoretically perfect conductors, they exhibit residual energy dissipation when driven by microwave currents, even at temperatures near absolute zero.

• Current Understanding: Historically, residual dissipation has been attributed primarily to dielectric losses arising from two-level systems (TLS) at material surfaces and interfaces.
• The Paradox: A class of "strongly disordered" superconductors (e.g., granular aluminum, amorphous indium oxide, titanium nitride) exhibits high dissipation that is independent of the surrounding dielectric environment. Conversely, "clean" superconductors (e.g., Niobium, Aluminum) achieve extremely high quality factors ($Q_i$), but the fundamental limit of their intrinsic bulk dissipation remains unclear.
• The Gap: There is no unified empirical framework linking the intrinsic material properties (specifically disorder and superfluid density) to the maximum achievable coherence across different device geometries (planar resonators, 3D cavities, transmon qubits) and material types.

2. Methodology

The authors conducted a comprehensive, data-driven analysis of the internal quality factor ($Q_i$) across a vast dataset of superconducting devices reported in the literature.

• Dataset Scope: The study aggregates data from:
  • Materials: Ranging from highly disordered films (grAl, a:InO, WSi, TiN) to moderately disordered (Hf, $\beta$-Ta) and ultra-clean systems (Al, Nb, Ta).
  • Geometries: Planar resonators (coplanar waveguides, microstrips), 3D cavities, and transmon qubits.
  • Conditions: Measurements were restricted to low temperatures ($T \lesssim 0.1$ K) and low photon numbers to minimize thermal and power-dependent effects.
• Key Metric: The authors correlated $Q_i$ with the imaginary part of the complex conductivity ($\sigma_2$), which is directly proportional to the superfluid density ($n_s$) and inversely proportional to the magnetic penetration depth ($\lambda$).
  • $\sigma_2 = \frac{n_s e^2}{m^* \omega} = \frac{1}{\mu_0 \omega \lambda^2}$
  • $\sigma_2$ serves as a proxy for the degree of material disorder (lower $\sigma_2$ implies higher disorder).
• Analysis: By plotting $Q_i$ vs. $\sigma_2$ on a double-logarithmic scale, the authors identified empirical scaling laws and upper bounds that were previously obscured by focusing on specific materials or geometries.

3. Key Contributions & Results

A. Discovery of a Universal Scaling Relation

The central finding is an empirical upper bound on microwave dissipation that scales linearly with the superfluid density (or $\sigma_2$) for a wide range of materials.

• The Bound: For strongly disordered to moderately disordered materials, the maximum quality factor follows:
  $$Q_i^{\text{max}} \approx \kappa \sigma_2$$
  where $\kappa \sim (0.1 - 1) \, \Omega \cdot \text{m}$.
• Implication: This establishes a fundamental limit where the dissipation is intrinsic to the bulk material, independent of surface dielectric losses. As disorder increases (lower $\sigma_2$), the maximum achievable $Q_i$ decreases linearly.

B. Identification of the Dissipation Mechanism

The authors attribute this intrinsic bulk dissipation to nonequilibrium quasiparticles trapped in disorder-induced spatial variations of the superconducting gap.

• Mechanism: In disordered superconductors, the superconducting gap ($\Delta$) fluctuates spatially, creating shallow potential wells (sub-gap states) where quasiparticles can become trapped.
• Universal Density: The density of these trapped quasiparticles ($n_{qp}$) is set by a universal dimensionless parameter related to the coherence length ($\xi$), rather than the specific generation rate of quasiparticles.
• Theoretical Model: The authors derive that the real part of the conductivity ($\sigma_1$) due to these trapped quasiparticles leads to the observed scaling $Q_i \propto \sigma_2$. This explains why disordered materials (low $\sigma_2$) have lower $Q_i$ limits than clean ones.

C. Distinct Regimes for Clean Materials and 3D Cavities

• Clean Limit ($\sigma_2 \gtrsim 10^8$ S/m): For ultra-clean materials (like Nb and Al in 3D cavities), the scaling changes. The best-performing 3D cavities follow a steeper trend:
  $$Q_i^{\text{max}} \propto \sigma_2^{3/2}$$
  This is attributed to the small kinetic inductance fraction ($\alpha \ll 1$) in 3D cavities, where geometric inductance dominates.
• Planar vs. 3D Limit: A distinct "ceiling" appears for planar devices and transmons at $Q_i \sim 10^7$, regardless of how clean the material is. The authors identify this as a substrate loss limit (dielectric loss in silicon or sapphire), which becomes the dominant factor once bulk conductive losses are minimized.

D. Re-evaluation of Transmon Qubits

The study shows that transmon qubits made from disordered materials (e.g., grAl) are strictly limited by the bulk quasiparticle mechanism ($Q_i \sim 10^5$). However, transmons made from clean materials are limited by the substrate loss ceiling ($Q_i \sim 10^7$), suggesting that improving coherence in clean transmons requires addressing substrate dielectric losses rather than just material purity.

4. Significance and Implications

• Material Selection Guide: The paper provides a data-driven framework for selecting materials for future quantum circuits. It suggests that Niobium (Nb), having the highest $\sigma_2$ among common superconductors, offers the highest theoretical coherence ceiling, potentially outperforming Aluminum (Al) and Tantalum (Ta) by an order of magnitude if dielectric losses are eliminated.
• Redefining the Coherence Limit: The findings challenge the prevailing view that TLS is the sole dominant loss mechanism. Instead, they highlight bulk conductive losses driven by trapped quasiparticles as a fundamental, intrinsic limit for disordered superconductors.
• Pathways for Improvement:
  • For Disordered Materials: Further improvements require reducing the density of trapped quasiparticles via better gap engineering, improved filtering of stray radiation, and phonon trapping.
  • For Clean Materials/Planar Circuits: The path to higher coherence lies in substrate engineering (e.g., suspended circuits, on-chip 3D designs) to bypass the $10^7$ substrate loss limit.
• Theoretical Impact: The work bridges the gap between standard electrodynamics and the behavior of disordered superconductors, offering a microscopic explanation for the "inductive loss" observed in granular and amorphous films.

Conclusion

This paper establishes a universal empirical bound on microwave dissipation in superconducting circuits, linking the maximum coherence time directly to the material's superfluid density. It identifies trapped nonequilibrium quasiparticles as the root cause of intrinsic bulk dissipation in disordered superconductors and substrate dielectric loss as the limiting factor for clean planar devices. These insights provide a critical roadmap for optimizing materials and architectures to push the coherence limits of superconducting quantum processors toward the transformative milestone of $T_1 \sim 0.1 - 1$ second.