Can an Ultra-Thin Layer Shield Qubits Without Adding Noise?
Cosmic rays can knock out several qubits at once. A few-atom-thick boron nitride layer might block them. This underground study checks it does no harm first.
In one paragraph
Energy from radioactivity and cosmic rays can spread through a quantum chip as phonons and disrupt several qubits at once. A layered material such as hexagonal boron nitride might block them, if it does not add loss. This preprint built eight niobium resonators on a 3 nm hBN film and eight on bare sapphire and measured them underground at Fermilab. Loss rose slightly, about 1.6 times for defect loss, but noise rose roughly an order of magnitude. Shielding itself was not tested.
How was the boron nitride layer tested?
Phonons, vibrations in a chip's substrate, can break the electron pairs in superconducting circuits, causing correlated errors across many qubits. Hexagonal boron nitride (hBN) is made of flat sheets that phonons struggle to cross, so it could act as a shield. But new materials can add defects called two-level systems that absorb energy.
The authors grew few-layer hBN, about 3 nm thick, on sapphire by molecular beam epitaxy. They built eight niobium resonators on it and eight on bare sapphire in the same batch, then cooled both chips together in a dilution refrigerator at QUIET, an underground lab at Fermilab under 100 m of rock. They measured loss versus power and temperature, and noise and frequency stability over time.
Did the boron nitride layer add loss or noise?
At low power, internal loss had a median of 4.50 × 10−6 on the hBN chip against 1.65 × 10−6 on the control, higher but within the same order of magnitude. The control was on par with state-of-the-art niobium resonators.
The temperature sweep showed two-level-system loss about 1.6 times higher with hBN (5.54 × 10−6 against 3.43 × 10−6), still far below typical deposited dielectrics.
Noise told a less flattering story. Two-level-system noise was roughly an order of magnitude larger on the hBN resonators, and over a 200-minute timestream most hBN resonators showed pronounced frequency switching while the controls stayed steady.
Why does a low-loss shielding layer matter?
Correlated errors from stray energy are a serious obstacle for quantum computers, and existing defences include radiation shielding, gap engineering and phonon traps. A thin layer built into the chip could add another option. In particle detectors built from similar resonators, blocking energy loss could improve sensitivity.
Before testing whether hBN blocks phonons, it had to be shown that it does not spoil the circuit. The authors conclude that epitaxial few-layer hBN is a low-loss dielectric compatible with superconducting circuits, and note that chemical vapour deposition can produce similar films on whole wafers.
A fair reading is that the loss penalty is small but the added noise needs attention, especially for sensors.
What are the limits of this hBN resonator study?
Read before you quote it
This is a preprint that has not been peer reviewed, and its limits are clear:
- No shielding test. Phonon blocking was not measured.
- Small comparison. One hBN chip and one control chip.
- Partial coverage. hBN was likely etched away in the resonator gaps.
- Setup difference. A warm amplifier was present only for the control measurements.
- Rough isolation. Estimating the film's own loss relied on simulations with assumptions.
The study does not show that hBN protects qubits from correlated errors, and it does not test hBN in working qubits or show whether the extra frequency switching would harm qubit performance.
- It does not show that hBN protects qubits from correlated errors.
- The proposed benefit, blocking phonons to prevent quasiparticle bursts, was not tested.
- Only one hBN chip and one control chip were compared.
- The hBN was likely etched away in the resonator gaps, so it remained only under the metal.
- Some measurement conditions differed: an extra warm amplifier was present only for the control measurements.
Takeaways
- saysAdding a few-atom-thick boron nitride film under niobium resonators raised energy loss, but it stayed within the same order of magnitude as plain sapphire.
- saysDefect noise was about an order of magnitude higher with boron nitride, with more frequency jumps over hours.
- saysThe phonon shielding the layer is meant to provide wasn't tested yet.
- explainsTwo-level systems are tiny material defects that absorb energy from superconducting circuits and add noise.
- interpretsA fair reading is that the shield passed its first test, adding little loss, but must still prove it shields.
Ask the paper
In PaperSays you can ask this paper anything, by text or by saying “Hey Aiden”. Three questions listeners asked:
What is quasiparticle poisoning in quantum computers?
Stray energy breaks superconducting electron pairs, creating quasiparticles that cause errors, often across several qubits at once. Hear the full answer in the app →
Does boron nitride add loss to superconducting circuits?
In this study, a 3 nm film added only a little loss, staying in the same order of magnitude as bare sapphire. Ask in the app →
Why was this experiment done underground?
100 m of rock shields more than 99% of cosmic ray muons, reducing outside disturbances. Ask in the app →
Briefing reviewed by an editor before publication. The paper belongs to its authors; this page summarises it and links to the original. Paper licence: CC BY 4.0. Corrections: support@papersays.com.