⚛️ quantum physics

TLS loss in superconducting resonators on epitaxial hBN thin films

This study demonstrates that integrating epitaxial hexagonal boron nitride (hBN) thin films into niobium superconducting resonators significantly reduces dielectric loss and mitigates phonon-mediated errors, offering a promising materials-driven strategy to enhance the performance of quantum computers and sensors.

Original authors: J. Christiansen-Salameh, D. J. Temples, R. Linehan, S. Sussman, Aloyse Maille, F. Abouzahr, L. Alexis, K. Anyang, D. Bowring, G. Bratrud, E. Celi, A. H. Colón Cesaní, R. Khatiwada, D. Molenaar, O. Pee

Published 2026-10-12

📖 3 min read🧠 Deep dive

CC BY 4.0

Original authors: J. Christiansen-Salameh, D. J. Temples, R. Linehan, S. Sussman, Aloyse Maille, F. Abouzahr, L. Alexis, K. Anyang, D. Bowring, G. Bratrud, E. Celi, A. H. Colón Cesaní, R. Khatiwada, D. Molenaar, O. Peek, A. Rodriguez, O. Seidel, D. Schlom, D. Sotir, S. Uemura, Y. Wu, E. Figueroa-Feliciano, D. Baxter, Z. Tian

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

Quantum computers and ultra-sensitive sensors rely on circuits made of superconducting materials, which conduct electricity with zero resistance when cooled to temperatures near absolute zero. However, these delicate machines are vulnerable to invisible disturbances. High-energy particles from space or natural radioactivity can strike the solid material holding the circuit, creating a shower of sound-like vibrations called phonons. These vibrations travel through the material and, if they reach the superconducting layer, can break the pairs of electrons that carry the current. This creates a burst of unwanted energy that scrambles the computer's calculations or blinds the sensor. To stop this, scientists have long sought a way to block these vibrations from crossing the boundary between the substrate and the superconductor, but adding a new layer of material to do so often introduces its own problems, such as absorbing energy and ruining the circuit's performance.

A team of researchers set out to test a specific solution: inserting a thin sheet of hexagonal boron nitride, a two-dimensional material, between the superconductor and the substrate. This material is known for its ability to reflect and scatter sound waves, potentially acting as a shield against the vibrations that cause errors. The challenge was to prove that this shield could be added without introducing a new source of noise or energy loss that would defeat the purpose. The researchers grew a few layers of this material directly onto a sapphire base using a precise, high-vacuum technique, then placed a thin film of niobium, a superconducting metal, on top. They built tiny resonators, which are like tuning forks for radio waves, on these new structures and compared them side-by-side with identical resonators built directly on bare sapphire without the extra layer.

The measurements revealed that the new material did not significantly harm the circuit's performance. The resonators with the boron nitride layer maintained a very low level of internal energy loss, comparable to the best existing superconducting circuits. In the quietest operating conditions, where the circuit holds just a single packet of energy, the loss remained on the order of one part in a million. This finding is crucial because it suggests that the material can be integrated into the standard stack of components used in quantum devices without degrading their quality. The researchers also observed that the new resonators were slightly more sensitive to random fluctuations in their frequency over time compared to the control devices, indicating that the interface between the layers still holds some microscopic defects that cause minor instability. However, the overall loss remained low enough to be useful.

The study confirms that this specific type of two-dimensional material can serve as a functional barrier against the phonons that plague quantum systems. By blocking these vibrations, the material could prevent the correlated errors that currently limit the size and reliability of quantum computers. The same principle could also improve quantum sensors by ensuring that energy from particle impacts is collected efficiently rather than lost to the surrounding packaging. While the current method of growing the material is complex, the results suggest that future devices could use similar layers to protect their sensitive components, paving the way for more robust and scalable quantum technology. The work demonstrates that it is possible to engineer the interface between materials to control how energy moves, offering a practical path forward for building the next generation of quantum machines.

Technical Summary: TLS Loss in Superconducting Resonators on Epitaxial hBN Thin Films

Problem Statement
Superconducting circuits, essential for quantum computing and sensing, are susceptible to excess quasiparticle (QP) bursts driven by non-equilibrium phonon excitations. In computing applications, these "quasiparticle poisoning" events cause correlated decoherence across multiple qubits. In sensing applications, such as Kinetic Inductance Phonon-Mediated (KIPM) detectors, phonon loss to passive films or packaging limits collection efficiency. The authors propose inserting a two-dimensional (2D) material, specifically hexagonal boron nitride (hBN), at the superconductor-substrate interface. Due to hBN's weak interlayer Van der Waals forces, it is hypothesized to reflect and diffuse athermal phonons, thereby suppressing phonon-mediated QP bursts in computing or blocking energy loss in sensing. However, introducing new dielectric layers risks increasing dielectric loss via Two-Level Systems (TLS), which degrade resonator quality factors ( $Q$ ) and detector sensitivity. The challenge is to integrate hBN without introducing significant TLS loss.

Methodology
The study fabricates and characterizes niobium (Nb) coplanar waveguide (CPW) resonators on two types of substrates:

hBN Chip: Few-layer, epitaxial hBN thin films grown via Molecular Beam Epitaxy (MBE) on c-plane sapphire substrates. The growth utilized laser substrate heating to ensure film purity and crystallinity, resulting in wrinkle-free films with ~3 nm total thickness and 0.3 nm RMS roughness.
Control Chip: Identical Nb resonators fabricated directly on bare sapphire substrates.

Both chips were processed in the same batch to minimize variations in Nb film purity and thickness. The devices were measured simultaneously in the Quantum Underground Instrumentation Experimental Testbed (QUIET) at Fermi National Accelerator Laboratory, a deep-underground facility shielding >99% of cosmic ray muons.

The experimental characterization included:

Power Sweeps: Measuring internal loss ( $\delta_i$ ) as a function of average photon number ( $\langle n \rangle$ ) to probe TLS saturation.
Temperature Sweeps: Analyzing fractional frequency shifts ( $\delta f/f_0$ ) from 15 mK to 1.1 K to extract zero-temperature TLS loss ( $F\delta^0_{TLS}$ ) using a model based on TLS physics and Mattis-Bardeen theory.
Time Dynamics: Recording time-streams of resonance frequency and dissipation to analyze noise power spectral densities (PSD) and Allan deviations, distinguishing between white noise and TLS-induced fluctuations.
Finite Element Method (FEM) Simulations: Calculating participation ratios to estimate the intrinsic dielectric loss of the hBN interfaces and thin film.

Key Results

Low-Power Internal Loss: The hBN-integrated resonators demonstrated low-power internal loss tangents ( $\delta^{LP}_i$ ) on the order of $10^{-6}$ . The median $\delta^{LP}_i$ for hBN resonators was $4.50 \times 10^{-6}$ , compared to $1.65 \times 10^{-6}$ for control resonators. While the hBN devices showed slightly higher loss and greater variance, they remained within the same order of magnitude as state-of-the-art Nb resonators.
Temperature-Dependent Loss: Fitting the temperature-dependent frequency shift yielded a median effective TLS loss ( $F\delta^0_{TLS}$ ) of $5.54 \times 10^{-6}$ for hBN resonators and $3.43 \times 10^{-6}$ for controls. The presence of hBN increased the effective loss by a factor of approximately 1.6, but the values remained consistent with high-quality superconducting circuits.
Time Dynamics and Noise: The hBN resonators exhibited increased TLS-driven temporal fluctuations compared to controls. Specifically, the TLS noise component in the frequency quadrature was approximately 1.5 orders of magnitude larger for hBN resonators. Long-duration time-streams (hours) revealed pronounced frequency switching in hBN devices, whereas control devices remained steady.
FEM Analysis: Simulations estimated the intrinsic loss of the hBN interfaces and thin film ( $\delta_{MS}$ ) to be approximately $5.69 \times 10^{-3}$ , comparable to the estimated loss of the metal-substrate interface in the control device ( $5.67 \times 10^{-3}$ ). This suggests the hBN layer itself does not introduce a catastrophic increase in material loss, though interface defects may contribute to the observed variance.

Significance and Claims
The authors conclude that epitaxial, few-layer hBN is a low-loss dielectric compatible with superconducting circuits. The research represents a "first step" toward a materials-driven strategy to eliminate phonon-mediated correlated errors in solid-state quantum systems. The study demonstrates that hBN can be integrated into the conventional superconducting qubit stack with "minimal degradation" of the quality factor.

The paper posits that this integration leverages the thermal properties of 2D materials to prevent athermal phonons from transmitting across the superconductor-substrate interface. While the work focuses on the feasibility of the material stack, the authors suggest that if the phonon-blocking mechanism functions as hypothesized, it could protect devices from correlated decoherence in computing and maximize phonon collection efficiency in quantum sensors by blocking energy loss into passive films. The authors note that while MBE was used here, Chemical Vapor Deposition (CVD) offers a scalable alternative for wafer-scale integration.

More like this