betaFeedback

WorksPhysics and Astronomy

Reducing TLS Losses in Superconducting Circuits Using Solid Neon

Abstract

Two level systems (TLSs) are a major source of loss in superconducting devices, limiting the performance of qubits, detectors, and resonators. These defects arise from a range of sources, including surface oxides, crystal defects, as well as fabrication residues which are often difficult to remove. Using NbTiN coplanar waveguide resonators, we show that depositing and subsequently removing a solid neon film can significantly improve device performance. Analysis of the temperature and power dependence indicates that this enhancement arises from reduced TLS dissipation, which we attribute to the redistribution and removal of surface contaminants during neon deposition and evaporation. Atomic force microscopy images of contaminated hexagonal boron nitride flakes before and after neon exposure show that neon treatment can redistribute and remove surface residues. These results demonstrate a simple, non-destructive approach for mitigating TLS losses in superconducting devices and other sensitive materials that cannot withstand aggressive cleaning processes.

Introduction

Coplanar waveguide (CPW) resonators are a fundamental component of many superconducting devices and circuits, including single-photon detectors [1], quantum limited amplifiers [2, 3] and quantum processors [4, 5]. Their performance is often limited by energy dissipation [6] which broadens the linewidth and reduces the internal quality factor QiQ_i. Excitations of environmental two-level systems (TLSs) are widely recognised as a dominant sources of loss in many superconducting circuits [7], especially at millikelvin temperatures and low excitation powers. Considerable efforts have been focussed at identifying the sources of these losses and developing ways to reduce them [8, 9, 10, 11, 12, 13].

Although the exact microscopic origin of many TLSs remains uncertain [7], they are generally associated with atomic or electronic configurations that can occupy two metastable states. A TLS can absorb energy from the resonator’s electric field and subsequently relax, transferring energy to the environment, for example via emission of a phonon [6]. The resulting dissipation depends on temperature and excitation power when TLSs become saturated as their populations and transition rates change. Some known sources of TLSs include crystal defects within the substrate material [12], native oxides on the surface of the substrate or superconductor [14], chemical adsorbates [15], and contaminants from the fabrication process such as residual resist polymers [16] or etch byproducts [17], as illustrated in Figure 1a. Changing the device geometry can reduce the electric field participation ratio of these lossy interfaces [11], however, it does not remove the underlying defects. Because TLSs possess an enormous number of degrees of freedom, they are able to contribute to loss over a wide range of resonator frequencies. Several approaches have been used to reduce TLS loss. This includes oxide removal by chemical treatments [14], oxide-growth suppression through encapsulation or passivation [18, 19], absorbate removal through high temperature annealing [15] or ultraviolet light exposure [20], and removal of fabrication residues using strong acids [8]. These methods can be effective, but some are chemically aggressive, require elevated temperatures, or are unsuitable for materials that are sensitive to processing. A treatment compatible with a wider range of superconductors and two-dimensional materials would therefore be useful.

Figure 1. Sources of TLS losses and device specifications. (a) Illustration of device material layers (not to scale) and possible sources of TLS losses. Key resonator dimensions, the central conductor width (w) and gap to the ground plane are indicated. (b) Design schematic of the devices consisting of eight numbered resonators coupled to a single feedline (blue), resonators 1-4 (red) have a narrow structure design while resonators 5-8 (green) have a wider structure design. (c) Amplitude and phase of the forward transmission coefficient (S21S_{21}) as a function for frequency for a single resonator (resonator 6). The solid lines indicate the fit to the transmission model from which the internal (QiQ_i) and coupling (QcQ_c) quality factors have been extracted.

Cryogenic noble-gas deposition offers a possible non-reactive way to modify surface contaminants. Previous experiments coating superconducting circuits with noble gases at cryogenic temperatures have observed increases in resonator quality factors [21, 22, 23]. The persistence of those changes after the noble gas is removed, and the mechanism responsible for them, remain unclear. Noble-gas films have also been proposed to affect radiation-induced quasiparticle loss [23] and device thermalisation [24].

Here we study NbTiN CPW resonators before, during and after cryogenic neon deposition. By comparing measurements from multiple cooldowns before and after the neon treatment, we observe an increase in the internal quality factor, (QiQ_i), that persists even after the neon has been removed. Fitting the power- and temperature-dependent losses to an analytical model indicates that the improvement arises from a reduction in TLS dissipation caused by the redistribution or removal of fabrication residues from the device surface, rather than changes in dielectric participation, cooling of the TLS bath, or shielding from quasiparticle-generating radiation. The observed increase in QiQ_i was then replicated on an identical chip in a different dilution refrigerator setup. To investigate the underlying mechanism, we exposed a contaminated hexagonal boron nitride flake to the same neon treatment. Atomic force microscopy images acquired before and after treatment reveal a redistribution of surface contaminants, supporting the hypothesis that the neon process removes or displaces fabrication residues responsible for excess TLS loss.

Setup and Analysis Methods

The measured devices were fabricated from NbTiN films deposited on high-resistivity silicon substrates. We observed similar behaviour across two resonator samples measured in different setups. Both devices contain eight quarter wavelength hanger resonators multiplexed to a single feedline, as shown in Figure 1b. In the main text of the paper we focus on the results from device A, with additional data from device B contained in the Supplementary Materials. In device A, resonators 1-4, (red), have frequencies spaced between 5.75 and 6.06 GHz and have a central conductor width of 6 μ\mum with a gap of 5 μ\mum, while resonators 5-8, (green), have frequencies spaced between 6.10 and 6.49 GHz and have a central conductor width (w) of 10 μ\mum with a gap of 8 μ\mum. The resonators are capacitively coupled to the feedline via a 200 μ\mum length of superconductor, with a separation distance of 35 μ\mum. The sample was bonded to a PCB and placed in a hermetically sealed copper cell which was attached to the mixing chamber of a Bluefors dilution refrigerator with a base temperature of ∼20\sim20 mK. Full details of the resonator design and fabrication are provided in the Supplementary Materials A and B.

The resonators were characterised by performing transmission spectroscopy using a commercial Vector Network Analyzer (VNA) with the output power attenuated to control the average number of photons in the resonator, nˉ\bar{n}. Figure 1c shows an example of the amplitude and phase of the transmission coefficient (S21S_{21}) for a single resonator. The device was initially characterised at 20 mK (cooldown 1) and then warmed up to room temperature. In the second cooldown, the sample was measured at 20 mK, and then characterised after neon deposition. Finally, the sample underwent a third cooldown where it was characterised both before and after neon deposition. Full details on the measurement setup and timeline are provided in Supplementary Materials C.

The neon deposition process used is similar to described in other works [22, 25]. To coat the sample with neon, all plates on the cryostat were heated to 26 K, above the triple point temperature of neon (24.556 K, [26]), before a set amount of room temperature gas was released into the cell via a capillary feedline. The neon gas condenses into liquid on the sample, where it crystallises once the temperature was reduced below the triple point. Full details of the neon deposition process are provided in Supplementary Materials D. The presence of neon on the chip was confirmed by a shift of the resonance frequency, fr=12πLCf_r = \frac{1}{2\pi\sqrt{LC}} [27], due to increased capacitance of the resonator, CC, from the dielectric neon (ϵ=1.244\epsilon= 1.244 [28]).

Figure 2. Evolution of QiQ_i at 20 mK. (a) Change in QiQ_i over different cooldowns applying and removing neon for each resonator, the dashed line connects points for each resonator for visual clarity. (b) Change in QiQ_i between initial and final measurement for both devices, each dot represents the change in QiQ_i for an individual resonator, with the gold diamond indicating the mean improvement.

The amplitude and phase response for each measurement was fit using [29]

S21(f)=Aeiαe−2πifτ(1−Ql∣Qc∣eiϕ1+2iQlf−frfr).(1)S_{21}(f) = A e^{i\alpha} e^{-2\pi i f\tau} \left(1-\frac{\frac{Q_l}{\lvert Q_c\rvert}e^{i\phi}}{1+2iQ_l\frac{f-f_r}{f_r}}\right). \tag*{(1)}

Here, AA and α\alpha are the baseline transmission amplitude and phase, ff is the probe frequency, τ\tau is the phase adjustment for the signal travel time, QlQ_l is the total, or loaded, quality factor and QcQ_c is the coupling quality factor. The phase shift due to impedance mismatches in the measurement circuit is captured by ϕ\phi. From this fit we can extract frf_r, QcQ_c and QiQ_i through the relationship that 1Qi=1Ql−1Qc\frac{1}{Q_i}=\frac{1}{Q_l}-\frac{1}{Q_c}.

We further parameterize QiQ_i using a model that represents the total loss as the sum of TLS-associated, quasiparticle associated and other contributions [8, 30],

1Qi=1QTLS(nˉ,T)+1QQP(T)+1Qother.(2)\frac{1}{Q_i}=\frac{1}{Q_{\mathrm{TLS}}(\bar n,T)}+\frac{1}{Q_{\mathrm{QP}}(T)}+\frac{1}{Q_{\mathrm{other}}}. \tag*{(2)}

Here, QTLSQ_{\mathrm{TLS}} describes the power and temperature dependent TLS contribution, QQPQ_{\mathrm{QP}} describes the temperature-dependent quasiparticle contribution, and QotherQ_{\mathrm{other}} captures all other losses which are not dependent on temperature and power.

At 20 mK20\ \mathrm{mK}, we assume the losses from quasiparticles are negligible so that 1Qi≈1QTLS(nˉ,T)+1Qother\frac{1}{Q_i}\approx\frac{1}{Q_{\mathrm{TLS}}(\bar n,T)}+\frac{1}{Q_{\mathrm{other}}}. The TLS losses can be parameterised by the model [6, 8, 30]

QTLS(nˉ,T)=QTLS,01+nˉnctanh⁡(ℏω2kBT).(3)Q_{\mathrm{TLS}}(\bar n,T)=Q_{\mathrm{TLS},0}\frac{\sqrt{1+\frac{\bar n}{n_c}}}{\tanh\left(\frac{\hbar\omega}{2k_{\mathrm{B}}T}\right)}. \tag*{(3)}

QTLS,0Q_{\mathrm{TLS},0} sets the low power, low temperature limit of TLS losses, which captures both the density of TLSs and how strongly they couple with the resonator, known as the filling factor. The number of photons needed to saturate the TLS systems, ncn_c on 1/T21/T_2 is inversely proportional to the product of the TLS relaxation (T1T_1) and dephasing (T2T_2) times, these terms describe the rate at which energy from TLSs are dissipated to the environment. We further approximate that T1∝tanh⁡(ℏω2kBT)T_1 \propto\tanh\left(\frac{\hbar\omega}{2k_{\mathrm{B}}T}\right), (T2)−1∝kBTβ1(T_2)^{-1} \propto k_{\mathrm{B}}T^{\beta_1} together fit parameter β2\beta_2 to account for the non linear response of different TLS modes to increasing number of photons [8].

To model how the QQ factor responds to temperature, we need to account for quasiparticle losses which dominate at higher temperatures. Quasiparticle losses can be parameterised by [8, 30]

QQP(T)=QQP,0eΔkBTsinh⁡(ℏω2kBT)K0(ℏω2kBT).(4)Q_{\mathrm{QP}}(T)=Q_{\mathrm{QP},0}\frac{e^{\frac{\Delta}{k_{\mathrm{B}}T}}}{\sinh\left(\frac{\hbar\omega}{2k_{\mathrm{B}}T}\right)K_0\left(\frac{\hbar\omega}{2k_{\mathrm{B}}T}\right)}. \tag*{(4)}

Here QQP0Q_{\mathrm{QP}0} is the total quasiparticle number scaling parameter, Δ\Delta is the superconducting gap, approximated as Δ≈1.764kBTc\Delta\approx1.764k_{\mathrm{B}}T_c [8], where TcT_c is the critical temperature of the superconductor, and K0K_0 is the zeroth order modified Bessel function of the second kind. Further details of the fitting procedure are contained in the Supplementary Materials E.

Power dependence of $Q_i$ at 20 mK

Figure 3. Power dependence of QiQ_i at 20 mK. QiQ_i as a function of power for (a) resonator 1, that showed a significant improvement in QiQ_i, and (b) resonator 5, representative of one that showed no increase in QiQ_i over the measurement sequence. The dashed lines shows the fit of QiQ_i from equation (2) for each cooldown.

QiQ_i IMPROVEMENT AFTER NEON DEPOSITION AND REMOVAL

fig:2a shows the evolution of QiQ_i across the different cooldowns at 20 mK20\ mK and high power (nˉ≈105\bar{n} \approx10^5) in device A. We observe a similar QiQ_i value between the initial characterisation and after neon was first deposited in cooldown 2. However, once the neon was removed from the sample by warming to room temperature, and cooled down a third time, QiQ_i increased in resonators 1, 2, 4 and 6, with the others staying constant, or decreasing slightly. For those resonators that showed an improved QiQ_i, this improvement was further enhanced by an additional neon deposition. fig:2b summarizes the change in QiQ_i for each resonator across the two samples between the initial characterisation and the final measurement. On average, resonators in device A improved to 4.3 times their initial QiQ_i, while those in device B improved to 2.4 times their initial QiQ_i.

POWER AND TEMPERATURE DEPENDENCE

To isolate the different loss mechanisms, we measured the temperature dependence of QiQ_i across a range of powers. Figure 3 compares QiQ_i as a function of photon number at 20 mKmK for resonator 1, which showed a significant improvement after neon removal, and resonator 5, which did not. Fits to the model described in Section II are shown as dashed lines in Figures 3a and 3b. For resonator 1 (Figure 3a), QiQ_i increases across the measured power range and does not saturate. While the improvement is most evident at higher powers, in the single photon regime (nn=1), which is most applicable to quantum computing applications, QiQ_i increased from 4.7×1054.7 \times10^5 to 1.4×1061.4 \times10^6, an increase of ∼190%\sim190\%. In comparison, for resonator 5 (Figure 3b) QiQ_i first slightly increases with the first neon deposition, and then decreases for the subsequent removal and neon deposition. Results

for the other resonators are shown in Supplementary Materials E and H.

Figure 4 shows the temperature and power responses of resonator 6 as neon is deposited, removed and redeposited on the sample. During the first neon deposition (Figure 4a), QiQ_i shows a weak dependence on increasing both microwave power and temperature, with smaller increases observed at high power. After the neon is removed (Figure 4b), and redeposited (Figure 4c), the temperature response of QiQ_i becomes stronger across the full range of microwave power. In contrast, resonators that showed no improvement in QiQ_i retained a flat temperature dependence, similar to Figure 4a (see Supplementary Information).

Beyond 1 K, QiQ_i decreases in all resonators as quasiparticle related losses dominate. Previous studies have found that a thin layer of hydrogen can shield superconducting resonators from incident radiation, reducing the contribution of quasiparticles to the loss [23]. By comparing QiQ_i with and without neon at 4 K, where thermal quasiparticle dissipation dominates, we conclude that neon does not significantly reduce the quasiparticle losses in the resonator by shielding the device from high energy radiation (see Supplementary Materials G).

ATTRIBUTION OF LOSSES

We fit the combined power and temperature dependent data to the model described in Section II to extract the TLS dependent losses.

Figure 5 shows the relative change in fitted QiQ_i and QTLSQ_{TLS} compared to the initial characterisation for all cooldowns and resonators for both the single photon (nˉ=1\bar{n}=1) and high power (nˉ=105\bar{n}=10^5) regimes. While the first application of neon decreases QTLSQ_{TLS} for almost all resonators, removing the neon increases QTLSQ_{TLS} well beyond the original QTLSQ_{\mathrm{TLS}} of the sample. Further addition of neon continues to increase QTLSQ_{\mathrm{TLS}}, up to a factor of 6 times in the single photon regime and more than an order of magnitude at high power. Interestingly, an increase in the fitted QTLSQ_{\mathrm{TLS}} is observed even for resonators that did not show an increase in overall QiQ_i. This indicates a reduction in the QotherQ_{\mathrm{other}} parameter between cooldowns. This decrease in QotherQ_{\mathrm{other}} could be explained by a reduction in the T1T_1 or T2T_2 times of the coupled TLSs. When the lifetimes are sufficiently short, the nˉ/nc\bar{n}/n_c term in equation (3) becomes vanishingly small, resulting in a weak power and temperature dependence that causes their loss contribution to be absorbed into the QotherQ_{\mathrm{other}} parameter, rather than QTLSQ_{\mathrm{TLS}}. The non-uniform response shows multiple dissipation mechanisms exist at the same time and suggests neon processing can locally redistribute the surface contaminants rather than uniformly reduce all losses.

Figure 5. FIG. 5: QTLS_{\mathrm{TLS}} attribution Change in QiQ_i for each resonator relative to first measurement at (a) nˉ=1\bar{n}=1 and (b) nˉ=105\bar{n}=10^5. Change in QTLSQ_{\mathrm{TLS}} for each resonator relative to first measurement at (c) nˉ=1\bar{n}=1 and (d) nˉ=105\bar{n}=10^5.

Figure 4. Fit of QiQ_i across a range of temperatures. Observed temperature dependence of QiQ_i along with the fitted model for Resonator 6 in cooldown 2 with neon (a), cooldown 3 without neon (b) and cooldown 3 with neon (c).

Given that QiQ_i and QTLSQ_{\mathrm{TLS}} improve even after the neon is removed from the device and has been thermally cycled, we conclude that the results are not from changes to the energy participation ratio due to the neon filling the gaps between the resonator central conductor and ground plane. A plausible explanation is that the increase in QTLSQ_{\mathrm{TLS}} results from the neon removing or redistributing contaminants on the surface of the chip, as illustrated in Figure 6a. We hypothesize that in the first neon deposition, fabrication residues on the top of the superconductor are pushed by the neon into the active area of the resonator, increasing the surface participation ratio of these molecules. When the sample is brought to room temperature and the neon evaporates, these contaminants are taken with it. As no systematic trend is seen for resonators with wide gaps (resonators 1-4) compared to narrow gaps (resonators 5-8), the dominant contribution of TLS is likely from those residing at the surface of the superconductor, rather than the substrate.

Figure 6. FIG. 6: Surface cleaning (a) Illustration of proposed cleaning mechanism. (b) AFM image of HBN flake contaminated with polymer particles before neon treatment. (c) Data slice of height shown by blue line before neon treatment. (d) AFM image of HBN flake after neon treatment, where small polymer contaminants from before have been moved into larger clusters. (e) Data slice of height shown by blue line after neon treatment.

To support this hypothesis, we expose a flake of hexagonal boron nitride (hBN) to frozen neon. While the hBN surface differs from the CPW surface in topology and chemistry, this provides qualitative insights into the effect of neon on surface contaminants. This flake is contaminated with polymer residues from the exfoliation process. Figure 6b–Figure 6e shows atomic force microscopy (AFM) images taken before and after the treatment which show that the polymer residues are moved over micrometer scale distances into clusters, leaving large patches of the surface free of these contaminants. This clumping of residual particles may explain the observed results as the number of TLSs on the surface of the chip have been reduced, or, moved such that coupling to the resonator was reduced. A possible mechanism for this effect is that the thin film of liquid neon adsorbs contaminants from the surface of either the substrate or metal. These colloidal molecules are then moved during either the freezing or evaporation of the neon through thermomolecular forces that repel the molecules from the phase-change front [31, 32, 33]. As particles gather on the phase-change front, they can accumulate into clusters before being encapsulated, similar to effects observed in soft matter systems [34].

Discussion and Outlook

In this paper we have demonstrated that repeated freezing and evaporation of neon from the surface of a superconducting resonator increases the resonator’s internal quality factor. Improvements of almost a factor of 2 in the single photon regime and over an order of magnitude at higher powers are observed. By analysing the temperature and power dependence of these losses

and how they change over the course of the neon depositions, we conclude that these improvements in QiQ_i are the result of decreased losses from two-level systems, opposed to a change in the energy participation ratio or from radiation-induced quasiparticles. The results are consistent with the neon processing reducing TLS losses through removal or redistribution of surface contaminants. Our conclusion is supported by AFM images of a hBN flake contaminated with polymers taken before and after neon treatment that demonstrates that residual contaminants on the surface can be moved and clumped together by crystallising and removing neon. Our results demonstrate an effective method of reducing TLS losses due to fabrication residues that could be applied to a range of superconducting or semiconducting materials that are not robust against traditional acid or plasma cleaning processes.

Acknowledgements

We thank Dafei Jin, Xinhao Li, Erika Kawakami and Ivan Grytsenko for assistance in establishing the neon experiments in our lab.

A. Palmer acknowledges the support of the CSIRO Next-Gen Program. K. Matkovic and P. Russell acknowledge the support of Sydney Quantum Academy Undergraduate Summer Scholarships, and CSIRO Next-Gen Undergraduate Scholarships. L. Delventhal acknowledges a Sydney Quantum Academy Supplementary Stipend. K. Zuo acknowledges support from the ARC Center of Excellence for Engineered Quantum Systems (CE170100009), M. Cassidy acknowledges support from a UNSW Scientia Fellowship and an Australian Research Council Discovery Early Career Research Fellowship (DE240100590). The authors acknowledge the support of the Australian Research Council Centre of Excellence for Quantum Computer Performance and Integration (CE260100009). The authors also acknowledge the facilities as well as the technical assistance of the Research and Prototype Foundry Core Research Facility at the University of Sydney, part of the NSW node of the NCRIS-enabled Australian National Fabrication Facility. The authors acknowledge the facilities and the scientific and technical assistance of Microscopy Australia at the Electron Microscope Unit (EMU) within the Mark Wainwright Analytical Centre (MWAC) at UNSW Sydney.

Supplementary Materials

Resonator Design

The sample contains 8 resonators coupled to a single feedline with differing lengths to spread the frequencies spaced between 5.7465.746 GHz and 6.4946.494 GHz. Each resonator is a meandered transmission line that has been shorted on one end with the other coupled to the feedline. Resonators 1-4 have a central conductor width of 66 µm and gap 55 µm, while resonators 5-8 have a central conductor width of 1010 µm and gap of 88 µm. The resonators are coupled capacitively to the feedline by a length of 200200 µm and the have a meander separation of 8484 µm. The feedline has a central conductor width of 1515 µm and a gap of 88 µm. A detailed image of design for resonator 1 is shown in Figure S1.

Schematic diagram of resonator 1

Figure S1. Scematic diagram of resonator 1, the coloured sections indicate where NbTiN has been etched.

Sample Preparation

A 4 inch polished, high resistivity, silicon wafer, with a thickness of 525525 µm was used as the substrate for device A. The wafer was first treated with a hydrofluoric acid (HF) dip, sputtered with a layer of 4040 nm thick NbTiN and diced into 99mm ×\times 99mm chips. The sample was then cleaned using acetone and IPA and baked at 180∘180^\circC for 5 minutes to remove any residual. The sample was coated with 200200nm of ARP6200.09 e-beam resist by spin coating (40004000 rpm for 6060s). The resist was patterned using electron beam lithography with a Elionix ELS-F125 and developed for 6060s using xylene, followed by 3:1 MIBK IPA mix for 6060s and then dried with an N2 gun. The layer of NbTiN was then etched away using reactive ion etching with 44 mTor, 100100W and 2020 SCCM of SF6 before the e-beam resist was removed using NMP heated to 80∘80^\circC for an hour, followed by 2 minutes of O2 plasma ash at 5050W.

The patterned chips were then attached to a two-layer PCB, fabricated with Rogers 3010 with an EPIG coating, using GE-varnish which was heated to 100∘100^\circC. Two surface mount SMA connectors were soldered to the PCB and connected to the feedline input and output using 1 MIL aluminium wirebonds using a Westbond 6476D wirebonder. Wirebonds were also used to connect the ground plane of the sample to the PCB ground. The sample was packaged in a copper cell which was hermetically sealed using indium with a stainless steal tube centered above the sample for neon deposition.

Measurement Setup

All measurements were taken in a dilution refrigerator with a base temperature of ∼20 mK\sim20\ \mathrm{mK}. Attenuation was placed on the input line on the 50K, 4K, still and mixing chamber (MXC) plates totalling 60dB and a LNF-LNR1_15A cryoamplifier with a gain of around 37dB was added to the output line. Depending on the power level of the specific measurement, different levels of room temperature attenuation or amplification were applied. Measurements were taken using a commercial VNA at temperatures ranging from 20 mK to 4 K. A diagram outlining the fridge setup shown in Figure S2.

Figure S2. Diagram of RF input and output setup for measurements.

A sweep taken across a wide frequency range showing all 8 resonances is shown in Figure S3.

Figure S3. VNA Wide frequency sweep.

Table I shows the timeline of the cooldowns and neon depositions. Between cooldowns the sample was removed from the fridge and removed from the sample cell.

Date
Cooldown 115 September - 16 September
Cooldown 207 February - 12 February
Cooldown 2 Neon Deposition13 February - 15 April
Cooldown 330 April - 20 May
Cooldown 3 Neon Deposition21 May - 02 June

Table I. Timeline of sample cooldowns.

Neon Deposition

To coat the chip with neon, room temperature gas was pumped into the cell via a 0.18 mm internal diameter stainless steel capillary feedline connected an external gas handling system to the cell. The gas feedline was thermally anchored to the 50 KK and still plates of the dilution refrigerator, allowing for room temperature 5N purity neon to be deposited into the system where it crystallised on top of the chip. The amount of neon gas deposited was controlled using a small volume at a controlled pressure of 2.1 bar between two solenoids which was repeatedly filled and then released into the sample cell which was warmed to ∼26\sim26 KK, allowing for the neon to liquefy and flow through the tube to the sample, this setup is similar to previous neon setups [22, 25]. The presence of neon on the chip was confirmed by a shift of the resonance frequency, fr=12πLCf_r = \frac{1}{2\pi\sqrt{LC}}, due to increased characteristic capacitance, CC, as a result of the dielectric neon.

Fitting Methodology and Results

Complete temperature and power sweeps were taken for a subset of resonators in each cooldowns. Initially, the database contained 1,934 frequency sweep measurements. Sweeps were excluded from analysis if the relative uncertainty in the fitted QiQ_i exceeded 20%, if the temperature data was missing, or, if the temperature variance during data collection was excessively high due to control system issues, accounting for approximately 19% of measurements. After excluding these data points, 1,572 QiQ_i measurements remained. An independent model was used for each resonator-cooldown combination. The data was fit in stages, with each stage providing the initial parameter estimates for the subsequent fitting stage.

To further parameterise the two-level system model shown in Equation (2), we adopt the approximations outlined in Ref. [8], where T1∝tanh⁡(ℏω2kBT)T_1 \propto\tanh\left(\frac{\hbar\omega}{2k_BT}\right) and (T2)−1∝kBTβ1(T_2)^{-1} \propto k_BT^{\beta_1}. In addition, we introduce the fit parameter β2\beta_2 to account for the non linear response of different TLS modes to increasing number of photons and DD to capture the proportionality constant. This yields the final TLS model

QTLS(nˉ,T)=QTLS,01+(nˉβ2DTβ1)tanh⁡(ℏω2kBT)tanh⁡(ℏω2kBT)(5)Q_{TLS}(\bar{n},T) = Q_{TLS,0}\frac{\sqrt{1+\left(\frac{\bar{n}^{\beta_2}}{DT^{\beta_1}}\right)\tanh\left(\frac{\hbar\omega}{2k_BT}\right)}}{\tanh\left(\frac{\hbar\omega}{2k_BT}\right)} \tag*{(5)}

All fits were completed using the lmfit Python package via weighted least squares. For resonator-cooldown combinations where the temperature sweep data was not available, all measurements were taken at 20 mKmK. QiQ_i was fit using the relationship 1Qi=1QTLS+1Qother\frac{1}{Q_i} = \frac{1}{Q_{TLS}} + \frac{1}{Q_{other}}, assuming quasiparticle losses were negligible at low temperatures. QTLSQ_{TLS} was modelled using Equation (5). The term DTβ1DT^{\beta_1} was combined into a single fit parameter, D′D', due to the degeneracy of the two fit parameters, DD and β1\beta_1. This results in a model with four fit parameters: D′D', QTLS,0Q_{TLS,0}, β2\beta_2 and QotherQ_{other}. The process to fit the data is outlined below:

  • Model 1a: Fit the simplified model, weighting points by the inverse of the uncertainty in the QiQ_i estimate.

  • Model 1b: If the QotherQ_{other} parameter is degenerate, refit the model with QotherQ_{other} fixed to the estimate provided in Model 1a, using the parameter fits from Model 1a as initial estimates.

Where temperature sweep data was available, the parameters from Equations (5) and (4) were fit using an iterative process that segmented the data into TLS dominated and quasiparticle dominated regimes. The model has seven fit parameters: DD, QTLS,0Q_{\mathrm{TLS},0}, β1\beta_1, β2\beta_2, QotherQ_{\mathrm{other}}, TcT_c and QQP,0Q_{\mathrm{QP},0}. During the fit, TcT_c was bounded between 10 and 17 KK, which is the expected superconducting critical temperature for the NbTiN film. The fitting procedure is outlined below:

  • Model 2a: Fit the model, excluding quasiparticle losses, using a dataset filtered to temperatures, T<0.8KT < 0.8K, weighting points by the inverse of the uncertainty in the QiQ_i estimate.

  • Model 2b: Using the fit values from Model 2a as initial parameter estimates, fit the model on the entire dataset, fitting only quasi particle parameters, TcT_c and QQP,0Q_{\mathrm{QP},0}. The points were weighted by the temperature to yield an initial estimate of the quasiparticle parameters that fits the high temperature regime.

  • Model 2c: Fit all model parameters on the entire dataset, using the fit values from Model 2b as initial parameter estimates. The points were weighted by the inverse of the uncertainty in QiQ_i.

  • Model 2d: If QotherQ_{\mathrm{other}} is degenerate, refit the model with QotherQ_{\mathrm{other}} fixed and using the initial estimates from Model 2c.

  • Final Selection: Select the model which yields the best fit between Model 2b and 2c/2d based on the R2R^2 metric to prevent overfitting the low temperature data.

Figure S4 shows the fits at 20 mKmK for all resonators and cooldowns.

Figure S4. QiQ_i fits for all resonators at 20 mK20\ \mathrm{mK}.

Figure S5 shows examples of the temperature and power dependent sweeps for resonators where this data was available.

Figure S5. Examples of QiQ_i fits as a function of temperature

Table II - Table IV shows the fit parameters and the uncertainties for each model, where missing uncertainties have been fixed in the fitting process outlined above.

ResonatorCooldownNeonData PointsParameterEstimateRel Uncertainty
11010QTLS,0Q_{TLS,0}6.34 E+052.65%
11010D′D'4.29 E+0140.91%
11010β2\beta_{2}2.27 E-018.14%
11010QotherQ_{other}3.00 E+07
12124QTLS,0Q_{TLS,0}3.81 E+057.90%
12124D′D'1.85 E+0036.62%
12124β2\beta_{2}1.21 E-019.56%
12124QotherQ_{other}3.00 E+07
130103QTLS,0Q_{TLS,0}1.20 E+063.39%
130103DD9.64 E+0031.70%
130103β1\beta_{1}-2.31 E-0127.90%
130103β2\beta_{2}3.71 E-015.31%
130103QotherQ_{other}1.33 E+07
130103TcT_c1.00 E+01
130103QQP,0Q_{QP,0}2.58 E+01
13186QTLS,0Q_{TLS,0}1.35 E+062.56%
13186DD8.29 E-0130.98%
13186β1\beta_{1}-4.57 E-0116.57%
13186β2\beta_{2}4.79 E-014.40%
13186QotherQ_{other}1.85 E+07
13186TcT_c1.00 E+01
13186QQP,0Q_{QP,0}4.48 E+00
21010QTLS,0Q_{TLS,0}6.40 E+052.79%
21010D′D'3.96 E+0142.97%
21010β2\beta_{2}2.22 E-018.72%
21010QotherQ_{other}3.00 E+07
22120QTLS,0Q_{TLS,0}3.28 E+0519.48%
22120D′D'9.57 E-0168.17%
22120β2\beta_{2}9.93 E-0214.85%
22120QotherQ_{other}3.00 E+07
230120QTLS,0Q_{TLS,0}8.45 E+0515.83%
230120DD2.87 E+00>100%
230120β1\beta_{1}-1.26 E-0195.29%
230120β2\beta_{2}2.07 E-0123.97%
230120TcT_c1.00 E+0111.33%
230120QQP,0Q_{QP,0}3.04 E+01>100%
230120QotherQ_{other}8.69 E+06
23144QTLS,0Q_{TLS,0}7.98 E+054.67%
23144DD7.03 E+01>100%
23144β1\beta_{1}9.75 E-0152.94%
23144β2\beta_{2}2.16 E-015.61%
23144TcT_c1.53 E+01>100%
23144QQP,0Q_{QP,0}3.27 E+01>100%
23144QotherQ_{other}1.01 E+07
31010QTLS,0Q_{TLS,0}4.12 E+0525.65%
31010D′D'1.47 E+0098.08%
31010β2\beta_{2}9.98 E-0219.83%
31010QotherQ_{other}3.00 E+07
321174QTLS,0Q_{TLS,0}4.03 E+0512.51%
321174DD4.29 E-0130.89%
321174β1\beta_{1}-2.79 E-0226.50%
321174β2\beta_{2}1.60 E-015.27%
321174QotherQ_{other}1.71 E+062.94%
321174TcT_c1.60 E+01
321174QQP,0Q_{QP,0}7.81 E+01

Table II. Fit model parameters.

ResonatorCooldownNeonData PointsParameterEstimateRel Uncertainty
61010QTLS,0Q_{TLS,0}5.06 E+054.11%
61010D′D'1.96 E+0141.67%
61010β2\beta_{2}2.05 E-018.40%
61010QotherQ_{other}3.00 E+07
621169QTLS,0Q_{TLS,0}3.80 E+0510.73%
621169DD8.10 E-0130.52%
621169β1\beta_{1}-2.04 E-0252.53%
621169β2\beta_{2}2.06 E-015.58%
621169QotherQ_{other}1.44 E+062.81%
621169TcT_c1.00 E+01
621169QQP,0Q_{QP,0}1.00 E+00
630113QTLS,0Q_{TLS,0}8.09 E+059.36%
630113DD8.66 E+0082.52%
630113β1\beta_{1}-1.05 E-0192.56%
630113β2\beta_{2}3.13 E-0115.37%
630113TcT_c1.00 E+0113.85%
630113QQP,0Q_{QP,0}2.64 E+01>100%
630113QotherQ_{other}3.93 E+06
631129QTLS,0Q_{TLS,0}7.85 E+054.87%
631129DD1.24 E+0029.25%
631129β1\beta_{1}-1.99 E-0116.94%
631129β2\beta_{2}3.07 E-017.25%
631129QotherQ_{other}8.20 E+06
631129TcT_c1.16 E+01
631129QQP,0Q_{QP,0}1.40 E+01
71010QTLS,0Q_{TLS,0}5.32 E+054.52%
71010D′D'2.52 E+0146.77%
71010β2\beta_{2}2.21 E-018.77%
71010QotherQ_{other}3.00 E+07
72120QTLS,0Q_{TLS,0}1.81 E+0552.97%
72120D′D'2.26 E-01>100%
72120β2\beta_{2}1.30 E-018.06%
72120QotherQ_{other}3.00 E+07
73013QTLS,0Q_{TLS,0}7.76 E+057.23%
73013D′D'1.21 E+0147.78%
73013β2\beta_{2}4.07 E-0113.46%
73013QotherQ_{other}5.68 E+052.94%
73126QTLS,0Q_{TLS,0}5.21 E+057.08%
73126D′D'1.19 E+0022.86%
73126β2\beta_{2}2.03 E-018.35%
73126QotherQ_{other}1.12 E+065.54%
81010QTLS,0Q_{TLS,0}5.09 E+053.30%
81010D′D'2.43 E+0137.12%
81010β2\beta_{2}2.07 E-017.41%
81010QotherQ_{other}3.00 E+07
82120QTLS,0Q_{TLS,0}2.06 E+0518.86%
82120D′D'4.56 E-0150.77%
82120β2\beta_{2}1.15 E-016.35%
82120QotherQ_{other}3.00 E+07
83013QTLS,0Q_{TLS,0}9.29 E+057.66%
83013D′D'1.71 E+0162.55%
83013β2\beta_{2}4.75 E-0116.67%
83013QotherQ_{other}3.15 E+051.55%
83118QTLS,0Q_{TLS,0}7.53 E+057.75%
83118D′D'1.58 E+0027.89%
83118β2\beta_{2}3.28 E-0112.08%
83118QotherQ_{other}4.52 E+052.28%

Table IV. Fit model parameters.

ResonatorCooldownNeonData PointsParameterEstimateRel Uncertainty
330111QTLS,0Q_{\mathrm{TLS},0}4.21 E+0514.96%
DD3.95 E-0140.03%
β1\beta_1-6.91 E-0232.01%
β2\beta_21.22 E-018.88%
QotherQ_{\mathrm{other}}1.63 E+063.62%
TcT_c1.00 E+01
QQP,0Q_{QP,0}3.68 E+01
33144QTLS,0Q_{\mathrm{TLS},0}7.23 E+0512.91%
DD5.31 E-02>100%
β1\beta_1-7.21 E-0136.69%
β2\beta_22.02 E-0111.02%
TcT_c1.53 E+01>100%
QQP,0Q_{QP,0}8.57 E+01>100%
QotherQ_{\mathrm{other}}1.11 E+06
41010QTLS,0Q_{\mathrm{TLS},0}5.02 E+055.89%
D′D'1.35 E+0157.05%
β2\beta_21.78 E-0112.73%
QotherQ_{\mathrm{other}}3.00 E+07
42119QTLS,0Q_{\mathrm{TLS},0}2.23 E+0540.35%
D′D'3.84 E-01>100%
β2\beta_21.14 E-0112.41%
QotherQ_{\mathrm{other}}3.00 E+07
43013QTLS,0Q_{\mathrm{TLS},0}9.59 E+059.86%
D′D'2.31 E+0159.00%
β2\beta_23.89 E-0116.68%
QotherQ_{\mathrm{other}}3.00 E+0713.92%
43139QTLS,0Q_{\mathrm{TLS},0}1.03 E+069.53%
D′D'3.11 E+0030.00%
β2\beta_24.63 E-0114.90%
QotherQ_{\mathrm{other}}3.00 E+078.17%
51010QTLS,0Q_{\mathrm{TLS},0}1.00 E+05>100%
D′D'2.46 E-01>100%
β2\beta_22.77 E-0293.11%
QotherQ_{\mathrm{other}}3.00 E+07
52120QTLS,0Q_{\mathrm{TLS},0}1.43 E+0529.56%
D′D'5.01 E-0187.11%
β2\beta_25.03 E-0219.25%
QotherQ_{\mathrm{other}}3.00 E+07
530116QTLS,0Q_{\mathrm{TLS},0}8.91 E+03>100%
DD1.00 E-04>100%
β1\beta_11.89 E-02>100%
β2\beta_21.24 E-0143.93%
TcT_c1.00 E+0113.69%
QQP,0Q_{QP,0}3.09 E+0150.42%
QotherQ_{\mathrm{other}}2.66 E+053.28%
53138QTLS,0Q_{\mathrm{TLS},0}8.94 E+0516.79%
DD1.88 E-01>100%
β1\beta_1-4.20 E-01>100%
β2\beta_22.51 E-0121.64%
TcT_c1.00 E+01>100%
QQP,0Q_{QP,0}1.00 E+00>100%
QotherQ_{\mathrm{other}}2.40 E+052.49%

Table III. Fit model parameters.

Uncertainty Estimates

The uncertainty in the fitted value of QTLSQ_{TLS} was estimated using σQTLS=JTΣJ\sigma_{Q_{TLS}} = \sqrt{\mathbf{J}^{T}\mathbf{\Sigma}\mathbf{J}}, where J\mathbf{J} is the Jacobian of the QTLSQ_{TLS} function and Σ\mathbf{\Sigma} is the covariance matrix of the fitted parameters. The Jacobian, J=(∂QTLS∂QTLS,0,∂QTLS∂D,∂QTLS∂β1,…)T\mathbf{J} = \left(\frac{\partial Q_{TLS}}{\partial Q_{TLS,0}},\frac{\partial Q_{TLS}}{\partial D},\frac{\partial Q_{TLS}}{\partial\beta_{1}},\ldots\right)^{T} was estimated around the point of best fit using the method of finite differences. Table V shows the uncertainty in the estimated QTLSQ_{TLS} for the initial cooldown (cooldown 1, no neon) and the final cooldown (cooldown 3, with neon) evaluated at T=20 mKT = 20\ mK and nˉ=105\bar{n} = 10^{5}.

ResonatorInitialFinal
QTLSQ_{TLS} (Million)σQTLSQTLS\frac{\sigma_{Q_{TLS}}}{Q_{TLS}}QTLSQ_{TLS} (Million)σQTLSQTLS\frac{\sigma_{Q_{TLS}}}{Q_{TLS}}QTLS(final)QTLS(initial)\frac{Q_{TLS}(\mathrm{final})}{Q_{TLS}(\mathrm{initial})}
10.730.38%9.675.54%13.28 ± 0.74
20.740.39%2.361.81%3.21 ± 0.06
30.730.34%2.563.63%3.50 ± 0.13
40.630.43%8.4936.07%13.48 ± 4.86
50.260.15%3.9437.51%15.35 ± 5.76
60.630.41%2.914.48%4.63 ± 0.21
70.650.59%1.628.31%2.48 ± 0.21
80.610.40%4.0319.99%6.58 ± 1.31

Table V. Uncertainty in QTLSQ_{TLS} estimate evaluated at T=20 mKT = 20\ mK and nˉ=105\bar{n} = 10^{5}.

Quasiparticle Loss Impact

To evaluate whether the neon film alters quasiparticle induced dissipation in the device, we compare the measured internal quality factors of the resonators with and without neon at 4 K4\ K. At this elevated temperature, thermal quasiparticle losses dominate, such that Qi≈QQP(T)Q_i \approx Q_{QP}(T). Figure S6 compares measured QiQ_i at a number of photon occupancies for resonators 1 and 6, the resonances which could be reliably measured 4 K4\ K. As we observe lower QiQ_i when the neon is present compared to the bare chip, we conclude that the presence of the neon film does not reduce the quasiparticle dissipation.

Figure S6. Internal quality factor (QiQ_i) measured at 4 K4\ K for resonators 1 and 6, with and without neon. Each dot represents an individual measurement taken across various photon occupancies, and the gold diamonds indicate the mean QiQ_i for each configuration.

Device B

A separate device was prepared using the same design, with 12 nm thick NbTiN film on a sapphire substrate. The sample was packaged into a sample cell and the process of neon deposition and evaporation was applied. Due to time constraints a full temperature and power characterisation was not conducted on this device. Comparing the fitted QiQ_i from the initial and final cooldowns, this device showed improvement in QiQ_i for 6 out of the 8 resonators on the chip. The results for each resonator are shown in Table VI.

ResonatorInitial QiQ_i (’000)Final QiQ_i (’000)Relative Change (%)
170.8150.2112.3
279.2238.0200.6
388.9213.3140.0
4153.6210.537.1
533.998.7191.0
68.348.4481.6
723.616.3-31.0
8272.0258.7-4.9

Table VI. Change in fitted QiQ_i for device B evaluated at T=20T = 20 mKmK and nˉ≈105\bar{n} \approx10^{5}.

HBN Cleaning

A separate sample was prepared using the same NbTiN film as device A, patterned with a Field Effect Transistor. A flake of hexagonal boron nitride was placed on top of the patterned chip using polycarbonate tape before being cleaned with chloroform. Residual polymer particles remained on the surface as evidenced by an AFM image taken of the surface using a scanassyst-air tip, shown in Figure 6b. The sample was then packaged into a sample cell a thin film of neon was deposited onto the sample using the process outlined in Supplementary Materials D. The neon was then removed and the sample removed from the cell and another image was taken of the sample, shown in Figure 6d. This appears to show that the contaminants were clustered together and moved over micrometer scales, leaving some areas cleaned of these contaminants.

References

  1. [1]P. K. Day, H. G. LeDuc, B. A. Mazin, A. Vayonakis, and J. Zmuidzinas, A broadband superconducting detector suitable for use in large arrays, Nature 425, 817 (2003).pubmed.ncbi.nlm.nih.gov/14574407
  2. [2]M. A. Castellanos-Beltran and K. W. Lehnert, Widely tunable parametric amplifier based on a superconducting quantum interference device array resonator, Applied Physics Letters 91, 083509 (2007).
  3. [3]B. Ho Eom, P. K. Day, H. G. LeDuc, and J. Zmuidzinas, A wideband, low-noise superconducting amplifier with high dynamic range, Nature Physics 8, 623 (2012).
  4. [4]V. E. Manucharyan, J. Koch, L. I. Glazman, and M. H. Devoret, Fluxonium: single cooper pair circuit free of charge offsets, Science 326, 113 (2009).DOI
  5. [5]A. Wallraff, D. I. Schuster, A. Blais, L. Frunzio, R.-S. Huang, J. Majer, S. Kumar, S. M. Girvin, and R. J. Schoelkopf, Strong coupling of a single photon to a superconducting qubit using circuit quantum electrodynamics, Nature 431, 162 (2004).pubmed.ncbi.nlm.nih.gov/15356625
  6. [6]C. R. H. McRae, H. Wang, J. Gao, M. Vissers, T. Brecht, A. Dunsworth, D. Pappas, and J. Mutus, Materials loss measurements using superconducting microwave resonators, Review of Scientific Instruments 91, 091101 (2020), arXiv:2006.04718 [physics].DOI
  7. [7]C. Müller, J. H. Cole, and J. Lisenfeld, Towards understanding two-level-systems in amorphous solids: insights from quantum circuits, Reports on Progress in Physics 82, 124501 (2019).
  8. [8]K. D. Crowley, R. A. McLellan, A. Dutta, N. Shumiya, A. P. M. Place, X. H. Le, Y. Gang, T. Madhavan, N. Khedkar, Y. C. Feng, E. A. Umbarkar, X. Gui, L. V. H. Rodgers, Y. Jia, M. M. Feldman, S. A. Lyon, M. Liu, R. J. Cava, A. A. Houck, and N. P. d. Leon, Disentangling losses in tantalum superconducting circuits, Physical Review X 13, 041005 (2023), arXiv:2301.07848 [quant-ph].DOI
  9. [9]P. Foshat, S. Poorgholam-khanjari, V. Seferai, H. Feng, S. Johny, O. A. Mukhanov, M. Hutchings, R. H. Hadfield, M. Weides, and K. Delfanazari, Quasiparticle dynamics in NbN superconducting microwave resonators at single photon regime (2025), arXiv:2506.17816 [quant-ph].arxiv.org/abs/2506.17816
  10. [10]J. M. Sage, V. Bolkhovsky, W. D. Oliver, B. Turek, and P. B. Welander, Study of loss in superconducting coplanar waveguide resonators, Journal of Applied Physics 109, 063915 (2011), arXiv:1010.6063 [cond-mat].DOI
  11. [11]Y. Tominaga, S. Shirai, Y. Hishida, H. Terai, and A. Noguchi, Enhancing intrinsic quality factors approaching 10 million in superconducting planar resonators via spiral geometry (2025), arXiv:2502.17901 [quant-ph].arxiv.org/abs/2502.17901
  12. [12]J. Lisenfeld, A. Bilmes, A. Megrant, R. Barends, J. Kelly, P. Klimov, G. Weiss, J. M. Martinis, and A. V. Ustinov, Electric field spectroscopy of material defects in transmon qubits, npj Quantum Information 5, 105 (2019).DOI
  13. [13]R. Lescanne, M. Villiers, T. Peronnin, A. Sarlette, M. Delbecq, B. Huard, T. Kontos, M. Mirrahimi, and Z. Leghtas, Exponential suppression of bit-flips in a qubit encoded in an oscillator, Nature Physics 16, 509 (2020).DOI
  14. [14]D. P. Lozano, M. Mongillo, X. Piao, S. Couet, D. Wan, Y. Canvel, A. M. Vadiraj, T. Ivanov, J. Verjauw, R. Acharya, J. Van Damme, F. A. Mohiyaddin, J. Jussot, P. P. Gowda, A. Pacco, B. Raes, J. Van de Vondel, I. P. Radu, B. Govoreanu, J. Swerts, A. Potočnik, and K. De Greve, Low-loss α-tantalum coplanar waveguide resonators on silicon wafers: fabrication, characterization and surface modification, Materials for Quantum Technology 4, 025801 (2024).
  15. [15]S. E. de Graaf, L. Faoro, J. Burnett, A. A. Adamyan, A. Y. Tzalenchuk, S. E. Kubatkin, T. Lindström, and A. V. Danilov, Suppression of low-frequency charge noise in superconducting resonators by surface spin desorption, Nature Communications 9, 1143 (2018).DOI
  16. [16]C. M. Quintana, A. Megrant, Z. Chen, A. Dunsworth, B. Chiaro, R. Barends, B. Campbell, Y. Chen, I.-C. Hoi, E. Jeffrey, J. Kelly, J. Y. Mutus, P. J. J. O’Malley, C. Neill, P. Roushan, D. Sank, A. Vainsencher, J. Wenner, T. C. White, A. N. Cleland, and J. M. Martinis, Characterization and reduction of microfabrication-induced decoherence in superconducting quantum circuits, Applied Physics Letters 105, 062601 (2014).DOI
  17. [17]M. Sandberg, M. R. Vissers, J. S. Kline, M. Weides, J. Gao, D. S. Wisbey, and D. P. Pappas, Etch induced microwave losses in titanium nitride superconducting resonators, Applied Physics Letters 100, 262605 (2012).DOI
  18. [18]M. Bal, A. A. Murthy, S. Zhu, F. Crisa, X. You, Z. Huang, T. Roy, J. Lee, D. v. Zanten, R. Pilipenko, I. Nekrashevich, A. Lunin, D. Bafia, Y. Krasnikova, C. J. Kopas, E. O. Lachman, D. Miller, J. Y. Mutus, M. J. Reagor, H. Cansizoglu, J. Marshall, D. P. Pappas, K. Vu, K. Yadavalli, J.-S. Oh, L. Zhou, M. J. Kramer, F. Lecocq, D. P. Goronzy, C. G. Torres-Castanedo, P. G. Pritchard, V. P. Dravid, J. M. Rondinelli, M. J. Bedzyk, M. C. Hersam, J. Zasadzinzki, J. Koch, J. A. Sauls, A. Romanenko, and A. Grassellino, Systematic improvements in transmon qubit coherence enabled by niobium surface encapsulation, npj Quantum Information 10, 43 (2024).DOI
  19. [19]H. Gupta, R. Pereira, L. Koch, N. Bruckmoser, M. Singer, B. Schoof, M. Kompatscher, S. Filipp, and M. Tornow, High temporal stability of niobium superconducting resonators by surface passivation with organophosphonate self-assembled monolayers, Communications Materials 10.1038/s43246-025-01068-8 (2026).
  20. [20]M. Mergenthaler, S. Paredes, P. Müller, C. Müller, S. Filipp, M. Sandberg, J. B. Hertzberg, V. P. Adiga, M. Brink, and A. Fuhrer, Ultralight vacuum packaging and surface cleaning for quantum devices, Review of Scientific Instruments 92, 025121 (2021).DOI
  21. [21]Y. Tian, I. Grytsenko, A. Jennings, J. Wang, H. Ikegami, X. Zhou, S. Tamate, H. Terai, H. Kutsuma, D. Jin, M. Benito, and E. Kawakami, NbTiN nanowire resonators for spin-photon coupling on solid neon (2025), arXiv:2505.24303 [cond-mat].arxiv.org/abs/2505.24303
  22. [22]K. Matkovic, P. Russell, A. Palmer, E. Helgemo, L. Delventhal, K. Zuo, K. Surse, R. Rahman, and M. C. Cassidy, Characterizing neon thin film growth with an NbTiN superconducting resonator array (2025), arXiv:2510.21029 [cond-mat].arxiv.org/abs/2510.21029
  23. [23]F. Valenti, A. N. Kanagin, A. Angerer, L. Buimaga-Iarinca, C. Morari, J. Schmidmaeyer, and I. M. Pop, Hydrogen crystals reduce dissipation in superconducting resonators, Physical Review B 109, 054503 (2024), arXiv:2306.04112 [cond-mat].DOI
  24. [24]M. Lucas, A. V. Danilov, L. V. Levitin, A. Jayaraman, A. J. Casey, L. Faoro, A. Y. Tzalenchuk, S. E. Kubatkin, J. Saunders, and S. E. de Graaf, Quantum bath suppression in a superconducting circuit by immersion cooling, Nature Communications 14, 3522 (2023).DOI
  25. [25]X. Zhou, G. Koolstra, X. Zhang, G. Yang, X. Han, B. Dizdar, X. Li, R. Divan, W. Guo, K. W. Murch, D. I. Schuster, and D. Jin, Single electrons on solid neon as a solid-state qubit platform, Nature 605, 46 (2022).DOI
  26. [26]H. Preston-Thomas, The international temperature scale of 1990 (ITS-90), Metrologia 27, 3 (1990).
  27. [27]D. M. Pozar and D. M. Pozar, Microwave engineering (Wiley, Chichester, United States, 2012).
  28. [28]S. Inui, Y. Qi, Y. Xing, C. Peretti, D. Jin, and W. Guo, On-chip levitated neon particle arrays for robust and scalable electron qubits (2025), arXiv:2512.00654 [quant-ph].arxiv.org/abs/2512.00654
  29. [29]S. Probst, F. B. Song, P. A. Bushev, A. V. Ustinov, and M. Weides, Efficient and robust analysis of complex scattering data under noise in microwave resonators, Review of Scientific Instruments 86, 024706 (2015).DOI
  30. [30]G. Moshel, S. Masis, M. Schechter, and S. Hacohen-Gourgy, Material loss model calibration for tantalum superconducting resonators (2025), arXiv:2507.12141 [cond-mat].arxiv.org/abs/2507.12141
  31. [31]A. W. Rempel and M. G. Worster, The interaction between a particle and an advancing solidification front, Journal of Crystal Growth 205, 427 (1999).DOI
  32. [32]R. Mondal and G. Kumaraswamy, Materials prepared by freezing-induced self-assembly of dispersed solutes: A review, Materials Advances 3, 3041 (2022).DOI
  33. [33]E. Rabani, D. R. Reichman, P. L. Geissler, and L. E. Brus, Drying-mediated self-assembly of nanoparticles, Nature 426, 271 (2003).pubmed.ncbi.nlm.nih.gov/14628047
  34. [34]J. G. Meijer, V. Bertin, and D. Lohse, Frozen Cheerios effect: Particle-particle interaction induced by an advancing solidification front, Physical Review Fluids 10, 034002 (2025).DOI

Paper details

vlnc:W_rzhn98rw
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents