betaFeedback

WorksPhysics and Astronomy

Evidence of anisotropic surface state residual loss in UTe2_2 amid isotropic low-energy excitations

Abstract

We explore the pairing mechanism and possible topological properties of the unconventional superconductor UTe2_2 through their influence on both the inductive and resistive channels using resonant cavity perturbation techniques. We measure a composite surface impedance for a variety of UTe2_2 single crystals at microwave frequencies, where the inductive and resistive channels are comparable and can be treated on equal footing. By studying the composite electrodynamics for a large number of resonant modes, we are able to provide a comprehensive picture of the directional dependence of the excitations out of the ground state. We find similar power-law temperature dependencies at low-temperature for both the penetration depth and surface resistance, which in aggregate across modes, are quadratic and isotropic. On the other hand, we find a distinct and systematic anisotropy of the residual loss. The power-law results are inconsistent with a single pair of point nodes on the Fermi surface under weak scattering. We discuss the possibility of topological superconductivity as an interpretation of these isotropic power-laws in conjunction with the anisotropic residual loss.

Introduction

The underlying properties of a superconductor, such as the pairing mechanism, Fermi surface, and topological properties, can imprint themselves on the low-temperature behavior of observable quantities. The electrodynamics of such low-energy excitations out of the superconducting ground state are commonly studied through the inductive channel, such as with the magnetic penetration depth. The resistive superconducting channel, though, is far less commonly studied, especially since it is not accessible to dc or low-frequency measurements. Residual loss, however, is an important feature of exotic superconductors such as UTe2\mathrm{UTe}_2, and the resistive channel should also be sensitive to the low-energy excitations, though it is more susceptible to other influences.

The heavy Fermion superconductor UTe2\mathrm{UTe}_2 has been a hot topic for several years now as it demonstrates a confluence of exotic properties, including evidence of spin-triplet pairing, nodes in the superconducting gap, and topological surface states. During this time, the material has undergone multiple iterations of growth methods, namely chemical vapor transport (CVT) and molten salt flux (MSF). The CVT-grown samples for which superconductivity was first reported in this material had Tc∼1.6T_c \sim1.6 K [1, 2]. Since then, the transition temperature has been increased to Tc≲2T_c \lesssim2 K in these CVT-grown samples [3]. Subsequently, very high-quality samples (RRR ∼100\sim100) with even higher transition temperatures of Tc≈2.1T_c \approx2.1 K have been produced using a MSF growth method [4, 5]. The TcT_c apparently saturates for RRR ≳50\gtrsim50, though, and samples with different RRR values can have the same TcT_c [3, 4, 5]. There is considerable variation in RRR between samples across and within generations [6].

The upper critical field in UTe2\mathrm{UTe}_2 has been observed to exceed both the Pauli and orbital limits, suggesting triplet superconductivity [1]. Additionally, the expected temperature dependence of the NMR Knight shift for singlet superconductors is not observed [1, 7]. In the first generation of CVT samples, a double transition in the specific heat as well as the onset of a non-zero polar Kerr effect at TcT_c were observed, suggesting time-reversal invariance (TRS) breaking and chiral superconductivity [8]. Despite this, no evidence of TRS breaking has been seen in the subsequent generations [9, 10, 11].

Since the first generation of CVT samples, evidence for nodal superconductivity has been seen in this material using electrodynamic [12, 13, 14] and thermodynamic [1, 8, 15] measurements. Experiments to determine the specific nodal configuration have continued for the proceeding generations of samples. Based on the symmetry of the crystal structure, there are certain allowed configurations of the superconducting gap structure, called irreducible representations; however, non-symmetry-imposed nodes are still possible. For the D2hD_{2h} (orthorhombic) point group crystal symmetry of UTe2\mathrm{UTe}_2 with triplet superconductivity, the possibilities are a full gap (AuA_u) or point nodes along the aa-, bb-, cc-directions (B3uB_{3u}, B2uB_{2u}, and B1uB_{1u}, respectively). For a singlet superconducting state, there could be line nodes instead of point nodes [12, 16].

There have been many conflicting reports in the literature for the superconducting order parameter with each of the listed irreducible representations being claimed by various authors [17]. For example, thermal conductivity measurements have been performed by multiple researchers with varying results [15, 18, 19]. These experimentally deduced single-component order parameter results range from fully-gapped AuA_u [18, 19] to point nodal including B1uB_{1u} [20], B2uB_{2u} [15, 21], as well as B3uB_{3u} from STM [22].

These proposed order parameters have been single-component. For a chiral system, it is possible to have a multi-component order parameter which is a complex combination of two irreducible representations; however, the literature does not support chiral superconductivity in UTe2_2. Nonetheless, a multi-component combination of B3uB_{3u} and AuA_u has been concluded from magnetic screening measurements, albeit with caveats [14, 23, 24].

While there is no general agreement on the superconducting order parameter, thermodynamic measurements [15, 21] tend to indicate nodes along the bb-axis while surface-sensitive measurements [11, 22] tend to indicate nodes along the aa-axis. The possible existence of topological [25, 26] or other surface states [27, 28] could contribute to the differences between results obtained from surface and bulk measurements of UTe2_2. There are some measurements that conclude nodes along the cc-axis; however, due to the quasi-2D nature of the Fermi surface [29, 30], this can be difficult to experimentally realize.

In this paper, we perform surface impedance and magnetic penetration depth analyses for two MSF-grown UTe2_2 single crystals. We perform three separate measurements of one of these samples in orthogonal orientations to thoroughly probe the possible configurations of screening current flow directions, increasing the number of modes studied from 31 in Ref. [31] (which only consisted of CVT-grown samples) to 130 modes across all samples, both in this work and Ref. [31]. This collection of modes densely samples most of the space of induced-current weights, where only a small region was previously sampled. This reveals features which were previously unknown and provides a substantial level of confidence to our results. We study the low-temperature power-law behavior of the penetration depth and surface resistance and find isotropic, ∼T2\sim T^2 behaviors for both, which is inconsistent with simple point nodes under weak scattering. In contrast, we find a systematic anisotropy of the residual loss, which we discuss in the context of possible surface states.

We will first give an overview of the experiment then discuss the results of low-temperature excitations through both the penetration depth and the surface resistance. We will then present the residual loss at zero-temperature and finally discuss the interpretation of our results as a whole.

Results

We use microwave cavity perturbation techniques [13, 32, 33, 34, 35, 36, 37, 38, 39, 40] to determine the surface impedance of UTe2_2 single

crystals. These experiments have previously been performed on CVT-grown UTe2_2 samples [13, 31]. Here we measure MSF-grown samples to provide a more complete picture of the electrodynamics of this exotic material. As in Ref. [31], our measured surface impedance and the quantities which we derive from it are composite in the sense that they arise from screening currents flowing in multiple crystallographic directions. We are, however, able to use the characteristic anisotropy of the normal state dc resistivity tensor to estimate the fractions of the induced screening currents which flow in each crystallographic direction, which we call the induced-current weights wiw_i (i=a,b,ci = a, b, c). This information is also critical to determine the surface impedance from the underlying resonance data as discussed in Ref. [31]. It is clear that high-quality dc resistivity tensor data for MSF-grown samples is necessary to characterize the new MSF samples which we measure. We discuss this independently measured dc resistivity data for MSF samples and the fitting we use to determine the surface impedance in Appendix A 2.

An example of the surface impedance Zs(T)=Rs(T)+iXs(T)Z_s(T) = R_s(T) + iX_s(T) for a particular mode of one the MSF samples we study is shown in Figure 1. As with the CVT samples discussed in Refs. [13, 31], a reactance peak is observed in Xs(T)X_s(T) just below the critical temperature TcT_c. As shown in Figure 1, while the surface resistance Rs(T)R_s(T) rapidly drops as the temperature crosses below TcT_c, the surface reactance Xs(T)X_s(T) increases first into this peak before eventually decreasing. A reactance peak is broadly observed to some degree in all of our data for both CVT and MSF samples. This effect is indicative of the competition between the normal and superfluid screening responses (which act in quadrature) just below TcT_c, where the superfluid is initially less effective at screening than the normal fluid. These peaks are a sign of long quasiparticle mean free paths [40, 41, 42, 43, 44, 45]. Note that this differs from the coherence peak which occurs just below TcT_c in the real part of the complex conductivity σ1(T)\sigma_1(T) for conventional superconductors. The complex conductivity σ(T)=σ1(T)−iσ2(T)\sigma(T) = \sigma_1(T) - i\sigma_2(T), which is determined from Zs(T)Z_s(T) using the local limit relation Zs=iμ0ωσZ_s = \sqrt{\frac{i\mu_0\omega}{\sigma}}, is shown in the inset of Figure 1. Below TcT_c, the real part, which is a measure of the loss, shows an immediate monotonic decrease while the imaginary part, which represents the superfluid response in the superconducting state, increases monotonically. There are some ‘exceptional’ modes discussed in Appendix § 1.1.2 1 for which σ1(T)\sigma_1(T) monotonically increases below TcT_c or demonstrates a peak; however, these are also not ‘coherence peaks’.

Plots vs temperature of the surface resistance and surface reactance

Figure 1. Plots vs temperature in the vicinity of the superconducting transition of the surface resistance Rs(T)R_s(T) in blue and surface reactance Xs(T)X_s(T) in red, demonstrating a peak just below TcT_c, for a particular mode with a MSF-grown UTe2_2 sample. The complex conductivity is similarly plotted in the inset with real and imaginary parts shown in blue and red, respectively.

To investigate excitations out of the ground state and possible nodal structures of UTe2_2, we carefully study the temperature dependence of the magnetic penetration depth, which we determine from the complex conductivity using λ(T)=1/μ0ωσ2(T)\lambda(T) = 1/\sqrt{\mu_0\omega\sigma_2(T)}. For low temperatures (T<Tc/3T < T_c/3) we fit our penetration depth data to a power-law form given by,

λ(T)−λ0λ0=η(TTc)α,(1)\frac{\lambda(T)-\lambda_0}{\lambda_0} = \eta\left(\frac{T}{T_c}\right)^\alpha, \tag*{(1)}

which is predicted for nodal superconductors [46, 47]. From these fits, we extract the zero-temperature penetration depth λ0\lambda_0, power-law exponent α\alpha, and power-law coefficient η\eta as parameters for each measured mode in each measurement of a sample. The parameters α\alpha and η\eta have theoretical predictions for various nodal configurations on the Fermi surface [46, 47], to which we will compare our results below.

Since our measurement is sensitive to the microwave superconducting loss, we also carefully study the temperature dependence of the surface resistance. We observe a power-law temperature dependence at low-temperature,

Rs(T)−R0R0=ηR(TTc)αR,(2)\frac{R_s(T)-R_0}{R_0} = \eta_R\left(\frac{T}{T_c}\right)^{\alpha_R}, \tag*{(2)}

which we write in an analogous form to Eq. (1). We determine the zero-temperature surface resistance R0R_0, power-law exponent αR\alpha_R, and power-law coefficient ηR\eta_R as fitting parameters for each measured mode in each measurement of a sample. Unlike the power-law temperature dependences of λ\lambda, those of RsR_s do not have such well established interpretations; however, they very likely arise from the same underlying excitations out of the ground state, and R0R_0 is an important measure of the residual loss at zero-temperature. Our measurements provide an important insight into the superconducting loss as a tool to study the excitations out of the ground state.

In Ref. [31], the penetration depth and surface resistance were found to have comparable values for α\alpha and αR\alpha_R. Across modes, these demonstrated isotropic values near or less than 2. The residual loss, on the other hand, demonstrated a distinct anisotropy with larger loss for a certain region of the induced-current weight parameter space, which corresponded to the ‘exceptional’ modes demonstrating a monotonic increase in σ1(T)\sigma_1(T) with decreasing temperature. Much of the induced-current weight parameter space was un-probed, though, and these results were only for CVT-grown samples. To expand upon these results in this paper, we measure two MSF-grown UTe2_2 samples. One of these samples (23 B) was needle-shaped along the aa-direction, which is how crystals typically form for this growth method, while the other sample (28) has comparable dimensions along each of the crystallographic axes, similarly to samples of the CVT growth methods. Additionally, sample 28 has facets along the crystallographic axes, allowing it to be oriented in our measurements. We measure sample 28 three separate times under orthogonal orientations to probe as many distributions of the induced screening currents as possible. See Appendix § 1.1.1 1 for information about the growth of the samples discussed here.

Low-temperature excitations

We will now discuss the results of the power-law exponent α\alpha from fittings of the composite penetration depth data λ(T)\lambda(T) to the power-law form given by Eq. (1) for multiple UTe2_2 samples. We will additionally discuss the results of the power-law exponent αR\alpha_R from fittings of the composite surface resistance data Rs(T)R_s(T) to the power-law form given by Eq. (2). As with the CVT samples discussed in Ref. [31], we find no systematic dependence of α\alpha or αR\alpha_R on the crystallographic contributions as shown in Figure 2. The error bars shown in Figure 2 represent the uncertainty in the determination of the fit parameters corresponding to a 3% degradation of the fit quality. Contours of the α\alpha and αR\alpha_R data are shown in the waw_a-wcw_c plane with colors (indicated in the colorbars) corresponding to the value of the quantity being plotted. The contours help to visualize location in the induced current weighting space of values of α\alpha and αR\alpha_R which are much different from the rest. The large values of α\alpha and αR\alpha_R for a few modes can be seen as small contour loops with warm color tones. Unlike the residual loss, which will be discussed below, these modes do not appear to form a clear systematic dependence on the induced-current weights, rather they are isolated without consistently similar values in neighboring modes in terms of wiw_i. For most regions of the induced current weighting space, though, the contours of α\alpha and αR\alpha_R have intermediate values, indicating that the data fluctuates around these intermediate values without a particular systematic dependence on the induced current weights. The α=2\alpha= 2 and α=4\alpha= 4 predictions for currents parallel and perpendicular, respectively, to point nodes [47] are shown as horizontal planes in Figure 2a. For reference, Figure 2b also includes these predictions for the penetration depth. There is no reason a priori, though, to expect that the power-law exponent for the surface resistance should correspond to these values; however, we include these values as references since we find empirically that α\alpha and αR\alpha_R have such similar values.

Plots of penetration depth and surface resistance power-law exponents

Figure 2. Plots of (a) penetration depth power-law exponent α\alpha and (b) surface resistance power-law exponent αR\alpha_R vs current-flow contribution of each crystallographic axis (and vs mode frequency in the insets). The simple pp-wave predictions for the penetration depth of α=2\alpha= 2 with currents parallel to the point nodes and α=4\alpha= 4 for currents perpendicular to the point nodes are indicated as horizontal planes. Data for two CVT-grown UTe2_2 samples B39 and B40 is shown in orange and yellow, respectively. For the MSF-grown UTe2_2 samples, data for sample 23B is shown in magenta while data for sample 28 is shown in red, green, and blue for the cases of the aa-, bb-, and cc-axes aligned with axis of the ‘hot finger’, respectively. Contours of the data are also included in the waw_a-wcw_c plane with colors (indicated in the colorbars) corresponding to the value of the quantity being plotted.

Since the composite values of α\alpha and αR\alpha_R (as well as η\eta) show no systematic dependence on estimates of the axis weights, we present average values of these quantities for each sample as well as the standard deviation among the modes as a measure of the uncertainty of these averages in Table I. Considering these uncertainties, α\alpha and αR\alpha_R are practically equivalent; although, for sample 23B, the power-law exponent averages are considerably different, but they do have very large uncertainties. For other quantities in this table which show a systematic dependence on the crystallographic responses, we present the range of values observed for all modes measured. We compare our results with those of other researchers in Table I. The results from Ref. [13] in Table I are from composite λ(T)\lambda(T) data for a single resonant mode (f0=11.26f_0 = 11.26 GHz) of the same cavity we utilize. While the results of Ref. [14] presented in Table I are axis-resolved, these dc tunnel-diode oscillator screening measurements are not sensitive to loss, such as the surface resistance or σ1\sigma_1.

Batch/ citationGrowth methodTcT_c (K)α\alphaη\etaλ0\lambda_0 (μ\mum)αR\alpha_RR0/ω2R_0/\omega^2 (Ω\Omega ps2^2)
B39CVT1.961.70 ± 0.060.52 ± 0.170.64 − 2.011.73 ± 0.080.02 − 5.41
B40CVT1.961.72 ± 0.060.56 ± 0.111.00 − 2.421.79 ± 0.060.05 − 22.39
23BMSF2.252.34 ± 0.492.21 ± 3.680.62 − 2.003.39 ± 0.840.00 − 6.59
28 aMSF2.131.36 ± 0.470.47 ± 0.120.80 − 3.811.88 ± 0.110.00 − 26.34
28 bMSF2.132.00 ± 0.170.45 ± 0.141.03 − 2.812.05 ± 0.130.15 − 9.21
28 cMSF2.132.16 ± 0.190.48 ± 0.211.40 − 3.911.98 ± 0.122.85 − 18.48
28 (aggregate)MSF2.131.44 ± 0.500.47 ± 0.120.80 − 3.911.90 ± 0.120.00 − 26.34
S. Bae (UMD) [13]CVT1.572.110.7913.0
K. Ishihara (Tokyo) [14]CVT1.65(2.01,1.90,1.60)
CVT1.75(2.08,2.06,1.84)
MSF2.1(2.07,2.16,2.04)(1.42,0.710,2.75)

Table I. A comparison of several composite and axis-resolved electrodynamic properties for various UTe2_2 samples.

Measures of the microwave superconducting loss, such as the surface resistance RsR_s and real part of the complex conductivity σ1\sigma_1, are more difficult to interpret than inductive quantities, such as the penetration depth. Though there is a predicted intrinsic value for the residual superconducting loss in the case of line-nodes [48], such a universal value cannot be established for a general nodal structure. Additionally, Ref. [48] shows that in addition to the number of quasiparticles excited, the low-temperature behavior of the loss depends on the quasiparticle scattering rate, the scattering phase shift, and the magnetic penetration depth. Despite this, our observation of simple power-law behavior of Rs(T)R_s(T) at low temperatures motivates its efficacy as a tool to study excitations out of the ground state. Furthermore, the similarity between the penetration depth and surface resistance power-laws suggests that these are probing a common phenomenon.

Zero-temperature loss

In a BCS superconductor, the finite temperature loss at frequencies below the gap is expected to scale quadratically with frequency, Rs(T)∼ω2R_s(T) \sim\omega^2, in the absence of a residual loss mechanism. Additionally, a roughly quadratic frequency dependence has been observed in the zero-temperature surface resistance in CeCoIn5_5 [40]. For line-nodal superconductors, the universal residual surface resistance also scales quadratically with frequency, namely R0line−nodal=μ0ωλLℏω2Δmax⁡R_0^{\mathrm{line-nodal}} = \mu_0 \omega\lambda_L \frac{\hbar\omega}{2\Delta_{\max}} [48]. To provide a benchmark with which to compare our residual loss values, we estimate R0line−nodal≈8.4R_0^{\mathrm{line-nodal}} \approx8.4 mΩ\Omega at 10 GHz (corresponding to R0line−nodal/ω2=2.1 Ω ps2R_0^{\mathrm{line-nodal}}/\omega^2 = 2.1\ \Omega\ \mathrm{ps}^2) for UTe2_2, assuming Δmax⁡.=0.29\Delta_{\max.} = 0.29 meV and λL=1.5 μ\lambda_L = 1.5\ \mum. Aside from the residual loss expected from nodal quasiparticles, Andreev (or Majorana, etc.) bound states can also contribute to loss in nodal superconductors. In chiral pp-wave superconductors, bound states are also predicted to contribute to the residual loss [49, 50]. We use R0/ω2R_0/\omega^2 as a nominally frequency independent

measure of the residual loss, where R0R_0 is the zero-temperature surface resistance determined from fits to Eq. (2). This quantity relies almost directly on raw data as opposed to other frequency independent measures of the loss, such as the real part of the complex conductivity, which involve more assumptions and processing. We do not observe a clear, systematic frequency dependence of R0/ω2R_0/\omega^2 for our data. We plot R0/ω2R_0/\omega^2 vs the current-induced weights in a two-dimensional, top-down view in Figure 3 for which the value of R0/ω2R_0/\omega^2 is indicated by the color of each data point. The error bars shown in Figure 3 represent the uncertainties in the determination of wiw_i as fit parameters in Eq. (3) corresponding to a 3% degradation of the fit quality. To further accentuate regions of large residual loss in the parameter space of induced-current weights, we also include contours of the R0/ω2R_0/\omega^2 data. The estimate of the line-nodal residual loss is indicated with an arrow on the R0/ω2R_0/\omega^2 colorbar. A systematic dependence of the residual loss on the induced-current weights can be seen in Figure 3 which peaks for an arc of current flow directions indicated by the large, gray arrow with endpoints having no current in the bb-direction. The aa-/cc-axis current contributions at these endpoints are about 30/70% and 50/50%. We observe the largest frequency-independent loss at the former endpoint. The (relatively) large values of the frequency-independent loss on this arc tend to increase along the arc from the latter toward the former endpoints, as indicated by the widening gray arrow in Figure 3. The zenith of this arc has a contribution of bb-direction of about 40%. The ranges of values for R0/ω2R_0/\omega^2 across various samples are also included in the summary Table I.

Plot of nominally frequency independent zero-temperature loss $R_0/\omega^2$ versus current-flow contribution of each crystallographic axis

Figure 3. Plot of nominally frequency independent zero-temperature loss R0/ω2R_0/\omega^2 vs current-flow contribution of each crystallographic axis. Data for two CVT-grown UTe2_2 samples B39 and B40 is shown as circles and squares, respectively. For the MSF-grown UTe2_2 samples, data for sample 23B is shown as left triangles while data for sample 28 is shown as left, up, and down for the cases of the aa-, bb-, and cc-axes aligned with axis of the ‘hot finger’, respectively. The colors indicate the value of R0/ω2R_0/\omega^2 in units of Ω\Omega ps2^2. The large, gray arrow indicates the direction of increasing loss along the arc of large residual loss. Contours of the R0/ω2R_0/\omega^2 data are also included with the same color scale (indicated in the main colorbar) corresponding to the value of R0/ω2R_0/\omega^2. The estimate of the universal dd-wave loss is also marked with an arrow in this colorbar. The inset shows the contours of the fraction of TcT_c at which σ1(T)/σn\sigma_1(T)/\sigma_n has its maximum below TcT_c indicating the exceptional modes (which are blue according to the colorbar in this inset).

The contours of the fraction of TcT_c at which σ1(T)/σn\sigma_1(T)/\sigma_n has its maximum below TcT_c are shown in the inset of Figure 3 for comparison with the region of large residual loss in the parameter space of induced-current weights. This region of large residual loss does not appear to be correlated with extreme values of the other power-law parameters of λ(T)\lambda(T) and Rs(T)R_s(T), such as α\alpha, which occur for a few modes. Furthermore, the modes nearby these few modes in the induced-current weights do not show increased values of these other power-law quantities as they do for R0/ω2R_{0}/\omega^{2}. For both R0/ω2R_{0}/\omega^{2} and T/TcT/T_{c} of maximum σ1/σn\sigma_{1}/\sigma_{n} below TcT_{c}, the modes demonstrating large loss or low-temperature enhancement of σ1\sigma_{1} are contained in, what is apparently, the same distinct region of the induced current weight phase space, for which nearby modes exhibit similar behavior.

It is unlikely that two-level system (TLS) loss plays a role in the residual loss described here. See Appendix § 1.1.2 2 for further discussion of the possibility of TLS loss in this system.

Discussion

We have thoroughly probed the space of possible current flow directions through 130 resonant modes across the UTe2_2 samples we have studied. This fills out the wiw_i parameter space as can be seen in the top-down view of this space in Figure 3. Despite this, there is no systematic variation with respect to the induced-current weights of the penetration depth or surface resistance power-law exponents shown in Figure 2 to indicate the direction of a nodal symmetry under weak scattering. In the simple point nodal (pp-wave) case with weak scattering, we would expect α=2\alpha= 2 for currents along the direction of the nodes while α=4\alpha= 4 would be predicted for currents in the plane perpendicular to the nodal symmetry axis [47]. We would expect this to manifest as a distinct anisotropy between a few modes consistent with the parallel-to-nodes results and many more modes consistent with the perpendicular-to-nodes results since the space of current flow directions perpendicular to the nodal symmetry axis is higher dimensional than that along it. Similarly, for a line node under weak scattering, a characteristic anisotropy would be expected, but with α=3\alpha= 3 along the symmetry axis and α=1\alpha= 1 in the perpendicular direction (currents through the nodes) [47]. In this case, the higher power-law exponent would only correspond to a few modes. Our data, apparently, demonstrates excitations out of the ground state on the order of currents near points nodes for all current directions, without any apparent anti-nodes. This is inconsistent with the simple pp- and dd-wave cases just discussed. The case of line nodes is, additionally, particularly unlikely since the residual loss values we observe are less than our estimate of the universal dd-wave prediction for many modes. The case of strong (unitary) scattering, where the excitations at the nodes become averaged across the Fermi surface, could produce this behavior. For both the pp- and dd-wave cases under unitary scattering, α≈2\alpha\approx2 is predicted [47]. However, the high RRR values of the crystals and consistent appearance of the Xs(T)X_s(T) peaks below TcT_c argue against this interpretation. It has been proposed that multiple pairs of point nodes slightly off of one or more of the high symmetry axes could produce such an isotropic α≈2\alpha\approx2 result [14, 26]. Similar, nearly isotropic power-laws with α=1.9−2\alpha= 1.9 - 2 in PrOs4_4Sb12_{12} were attributed to multi-domain orientations of a point nodal order parameter [51].

A possible mechanism which is qualitatively consistent with our isotropic power-law exponent results shown in Figure 2 as well as the anisotropic residual loss results shown in Figure 3 (indicated by the large, gray arrow) is the presence of topological Weyl nodes. Such a scenario would require broken time reversal symmetry, though, which is not consistently supported by the literature. There are multiple conjectures, however, which seek to devise a means by which such a state could still be possible, such as a combined Kondo/spin-liquid model [52] or a mixed order parameter arising from nearly degenerate irreducible representations [53].

For a topological Weyl superconductor, it is predicted that the penetration depth power-law exponent would have an isotropic value of α=2\alpha= 2 [54]. Qualitatively, this matches our results for α\alpha shown in Figure 2a as well as those reported in Refs. [13, 14]. The observed values of α\alpha are close to this prediction in aggregate, though there are several observed values which depart from this prediction (though not systematically). Most notably, the isotropy of α\alpha is a striking feature which is difficult to explain simply with nodes. This topological model is compelling for its simplicity, though multiple nodes may still be present in the topological case. Refs. [25, 26] discuss theoretical predictions and consequences of Majorana-Dirac surface states in UTe2_2. Christiansen et al. conclude that for B2uB_{2u} and B3uB_{3u} order parameters, several non-symmetry-imposed nodes would be necessary in addition to the symmetry-imposed nodes, which is conceivable considering the isotropic ∼T2\sim T^{2} power-laws that have been observed for the penetration depth.

The second feature of a topological Weyl superconductor is that the loss is predicted to have a characteristic anisotropy, corresponding to larger loss for electric fields polarized in the direction of the Weyl nodes. This loss arises from the absorption of a photon from the gapless 2D Majorana fermion surface fluid to the bulk. This loss mechanism is predicted to contribute to the zero-temperature residual loss and to be robust to disorder [54]. The arc we have observed in the space of induced-current weights (illustrated by the large, gray arrow in Figure 3) could possibly indicate the presence of Weyl nodes along it if such a mechanism were active in the material. Such an effect is not necessarily unique to this theory; the concept of a surface to bulk photon absorption as a loss mechanism could arise from other topological scenarios, potentially one which does not require broken time reversal symmetry. Ref. [11] find a zero-energy Andreev conductance peak on the (0̄11) surface, which they conclude is consistent with a TRS preserving topological state.

Conclusions

In conclusion, we have analyzed the electrodynamic properties of both CVT- and MSF-grown UTe2_2 samples for 130 microwave-frequency modes at low temperatures using the resonant cavity perturbation measurements described in Ref. [31]. By studying so many modes with diverse structures of surface currents across multiple samples, we thoroughly probe the possible anisotropy of the properties of UTe2_2. Unexpectedly, we find similar power-law temperature dependencies at low-temperature for both the penetration depth and surface resistance, which in aggregate across modes, are quadratic. We do not observe any particular dependence of these power-law exponents on our estimates of the directions of screening current flows. We do, however, find a distinct anisotropy of the residual loss, which forms an arc in the space of induced current directions. Our power-law results are inconsistent with a single pair of point nodes under weak scattering. These isotropic power-laws in conjunction with the anisotropic residual loss are more consistent with topological Weyl superconductivity, though this would require TRS breaking. Alternatively, it is conceivable that a similar loss mechanism involving a surface to bulk photon absorption could occur for some other topological or surface state.

Acknowledgments

This work was supported by the National Science Foundation grant no. DMR2004386 and ARO/FSDL grant no. W911NF-24-1-0153 (microwave experiments), the U.S. Department of Energy award no. DE-SC-0019154 (sample characterization), the Gordon and Betty Moore Foundation’s EPiQS Initiative through grant no. GBMF9071 (materials synthesis), the National Institute of Standards and Technology, and the Maryland Quantum Materials Center. GSS was supported by NSF DMR2514293. NPB was supported by NIST. SRS acknowledges support from the National Institute of Standards and Technology Cooperative Agreement 70NANB17H301.

Data availability

The data that supports the findings of this article is openly available.

Competing interests

The authors declare no competing interests.

Author contribution

AC-J performed the microwave cavity perturbation experiments and the corresponding analyses discussed in this paper under the supervision of SMA. AC-J wrote this paper including all text, tables, and figures under the supervision of SMA, JP, and NPB. SRS grew the CVT UTe2_2 samples. AC-J prepared these samples for the microwave cavity perturbation measurements. GSS grew the MSF UTe2_2 samples and prepared them for the microwave cavity perturbation measurements under the supervision of NPB. JZD and GSS prepared MSF-grown UTe2_2 samples for axis-resolved, bar geometry electrical transport measurements, and JZD performed these measurements under the supervision of JP.

Methods

In this appendix, we will discuss the growth methods, preparation, and configurations of the samples used in these experiments (Appendix A1). We will also discuss the anisotropic dc resistivity tensor data used to perform the normal state fittings to determine the surface impedance (Appendix A2).

Materials

The UTe2_2 samples B39 and B40 discussed here and in Ref. [31] were grown using the CVT technique [2, 3]. The UTe2_2 samples 23B and 28 discussed here were grown using the MSF technique [4, 5].

The CVT-grown samples were prepared with a coarse polishing (to shape the crystals) in air followed by a fine polishing (to remove surface oxides) in a reduced oxygen (≤0.01%\le0.01\% O2_2 by volume) environment with no particular orientation of the final surfaces, as discussed in Ref. [31]. These samples were then coated in Apiezon N-grease before being returned to the ambient atmosphere. The final side lengths of these crystals were roughly 0.5−10.5 - 1 mm. We were able to determine the orientations of the MSF-growing samples by their crystal facets and an understanding of the morphology of this growth method. We used wooden picks to scrap these crystal facets while the samples were submerged in acetone in an attempt to remove the oxide layers. These samples were then dried and encased in N-grease.

The CVT-grown samples were mounted on the tip of the sapphire ‘hot finger’ with no particular orientation of the crystallographic axes, as discussed in Ref. [31]. Sample 23B was a needle-shaped sample (as is typical of this growth method) with a length of about 1.31.3 mm and a size of about 0.10.1 mm in the bcbc-plane cross-section. This sample was mounted on the side of the sapphire ‘hot finger’ rather than on its tip. Sample 28 was a bulk, roughly rectangular shaped sample with well-defined [100], [10], and [1] surfaces rather than a needle. Its side lengths were roughly 1.41.4, 0.80.8, and 0.50.5 mm along the aa-, bb-, and cc-directions, respectively. We performed three separate measurements of this sample on the tip of the ‘hot finger’ with the [100], [10], and [1] surfaces aligned with the axis of the ‘hot finger’ in turn.

Impedance fitting and anisotropic resistivity data

The surface impedance Zs=Rs+iXsZ_s = R_s + iX_s is given by

Rs=GQsampleR_s = \frac{G}{Q_{\mathrm{sample}}}

and

Xs=−2GΔf0,samplef0,tot.(T0)+X0,X_s = -2G\frac{\Delta f_{0,\mathrm{sample}}}{f_{0,\mathrm{tot.}}(T_0)} + X_0,

where the subscripts ‘sample’ and ‘tot.’ denote the resonant responses of the sample alone and the whole sample and resonator system, respectively [13, 38]. We determine the surface impedance for each mode by fitting this resonance data to,

Rs(T)Xs(T)=∑i=a,b,cwiμ0ωρi(T)/2,(3)\sqrt{R_s(T)X_s(T)} = \sum_{i=a,b,c} w_i\sqrt{\mu_0\omega\rho_i(T)/2}, \tag*{(3)}

to determine the geometry factor GG and minimum temperature reactance X0X_0 as well as estimate the direction of induced currents described by the induced-current weights wiw_i, where i=a,b,ci=a,b,c indicates the crystallographic direction [31]. This fitting relies on high-quality measurements of the dc resistivity tensor ρi\rho_i. For the CVT-grown samples B39 and B40 described in Ref. [31], Montgomery geometry resistivity data from Ref. [55] was used in this fit. We have measured bar geometry resistivity data for MSF UTe2_2 samples to describe the MSF-grown samples 23B and 28. The aa and bb bars were cut from another sample from batch 28 while the cc bar was cut from a different MSF batch with a similar RRR. We measured ρa\rho_a and ρb\rho_b for two pairs of samples. Since the bulky MSF samples (such as sample 28) tend to have lower RRR values than those of the needle-shaped ones, we have taken the lower RRR measurements of ρa\rho_a and ρb\rho_b to be most representative of our sample 28 data. The RRR values for the ρa\rho_a, ρb\rho_b, and ρc\rho_c data which we use are 145, 53, and 14, respectively.

At low-temperatures, each component of this MSF resistivity data follows a quadratic power-law ρi(T)=A(i)+B(i)T2\rho_i(T) = A^{(i)} + B^{(i)}T^2 as was observed for CVT resistivity data [55]. We model each component of our MSF resistivity data with a piecewise polynomial which crosses over between this quadratic power-law below 6 K and a fifth order polynomial ρi(T)=∑n=05Cn(i)Tn\rho_i(T) = \sum_{n=0}^{5} C_n^{(i)}T^n from 8–20 K as described in Ref. [31]. A summary of the parameters used in this model is given in Table II. We compare our MSF resistivity data to the CVT resistivity data from Ref. [55] in Figure 4. Note that the dc resistivity tensor has been previously measured for MSF-grown UTe2_2 microstructures fabricated with a focused ion beam (FIB) method [56]. Zhang, et al. found that the dc resistivity components for the aa- and bb- directions were comparable, which agrees with our resistivity measurements; however, they find a very large anisotropy between the cc-direction and the aa- and bb- directions (∼50\sim50 fold as the temperature approaches zero while we find it to be ∼10\sim10 fold). Additionally, the hierarchy of the dc resistivity tensor components for MSF samples in our measurement agrees with that of Ref. [56]; however, this is different from the CVT result, for which the magnitudes of ρa\rho_a and ρb\rho_b are flipped in this hierarchy.

ρa\rho_aρb\rho_bρc\rho_c
AA (μΩ\mu\Omega m)8.02×10−28.02 \times10^{-2}8.89×10−28.89 \times10^{-2}1.081.08
BB (μΩ\mu\Omega m/K2^2)1.28×10−21.28 \times10^{-2}1.27×10−21.27 \times10^{-2}0.1510.151
C0C_0 (μΩ\mu\Omega m)−9.01×10−2-9.01 \times10^{-2}3.363.3634.334.3
C1C_1 (μΩ\mu\Omega m/K)0.2070.207−1.46-1.46−16.7-16.7
C2C_2 (μΩ\mu\Omega m/K2^2)−6.15×10−2-6.15 \times10^{-2}0.2430.2433.243.24
C3C_3 (μΩ\mu\Omega m/K3^3)1.02×10−21.02 \times10^{-2}−1.51×10−1-1.51 \times10^{-1}−0.259-0.259
C4C_4 (μΩ\mu\Omega m/K4^4)−4.77×10−4-4.77 \times10^{-4}4.19×10−44.19 \times10^{-4}9.36×10−39.36 \times10^{-3}
C5C_5 (μΩ\mu\Omega m/K5^5)7.31×10−67.31 \times10^{-6}−4.27×10−6-4.27 \times10^{-6}−1.28×10−4-1.28 \times10^{-4}

Table II. Polynomial fit parameters used to model bar geometry resistivity data ρ(T)\rho(T) for MSF-grown UTe2_2 samples.

Resistivity tensor data for CVT-grown and MSF-grown UTe$_2$ samples

Figure 4. Resistivity tensor (decimated) data for two growth methods of UTe2_2 samples. The tensor components ρa\rho_a, ρb\rho_b, and ρc\rho_c are shown in red, green, and blue, respectively. Data for the CVT growth method from Ref. [55] is shown in the left panel while data for the MSF growth method is shown in the right panel. Our fits (which are quadratic-in-temperature up to 6 K) to each of these resistivity data sets are shown as dotted lines of the corresponding colors.

Other notes about high-loss modes

In this appendix, we will make some additional remarks about the modes which make up the arc of high residual loss. We discuss the correspondence of the modes demonstrating an ‘exceptional’ enhancement of σ1(T)\sigma_1(T) at low-temperature with this region of induced-current weight parameter space (Appendix B1). We will also consider the possibility of two-level system (TLS) loss as a contribution to the large residual loss we observe (Appendix B2).

Firstly, we note that this arc of large residual loss is a superconducting state phenomenon. The normal state loss, shown in Figure 5, has a distinctly different dependence on the induced current weights. In the normal state, the loss is expected to scale as ∼ω\sim\sqrt{\omega}, so we plot Rs2(Tc)/ωR_s^2(T_c)/\omega in Figure 5a as a nominally frequency independent measure of the (square of the) normal state loss. This shows a clear planar increase from low loss for modes with currents in the aa- and bb-directions to high loss with currents in the cc-direction. The character of this anisotropy arises from the anisotropy of the resistivity tensor. We determine a composite, effective resistivity ρeff=2RsXsμ0ω\rho_{\mathrm{eff}} = \frac{2R_sX_s}{\mu_0\omega} from our surface impedance data, which we fit to a power-law form given,

Plots of residual loss and effective resistivity versus induced-current weightings

Figure 5. Plots vs the induced-current weightings (and vs mode frequency in the insets) of (a) a nominally frequency independent measure of the normal state residual loss (squared) and (b) the zero-temperature (in the absence of superconductivity) effective resistivity (ρeff=2RsXsμ0ω\rho_{\mathrm{eff}} = \frac{2R_{\mathrm{s}}X_{\mathrm{s}}}{\mu_{0}\omega}) implied by our surface impedance data determined from normal state power-law fits. Data for two CVT-grown UTe2_2 samples B39 and B40 is shown in orange and yellow, respectively. For the MSF-grown UTe2_2 samples, data for sample 23B is shown in magenta while data for sample 28 is shown in red, green, and blue for the cases of the aa-, bb-, and cc-axes aligned with axis of the ‘hot finger’, respectively. Contours of the data are also included in the waw_a-wcw_c plane with colors (indicated in the colorbars) corresponding to the value of the quantity being plotted.

ρeff(T)−ρeff,0ρeff,0=ηeff(TTc)αeff,(4)\frac{\rho_{\mathrm{eff}}(T)-\rho_{\mathrm{eff},0}}{\rho_{\mathrm{eff},0}} = \eta_{\mathrm{eff}}\left(\frac{T}{T_c}\right)^{\alpha_{\mathrm{eff}}}, \tag*{(4)}

in the normal state. We plot the zero-temperature value ρeff,0\rho_{\mathrm{eff},0} from this fit vs the induced-current weights in Figure 5b, which demonstrates the same trend observed for Rs2(Tc)/ωR_s^2(T_c)/\omega. The values in the corners of the triangular plane formed by the data this plot imply the (axis-resolved) tensor components of ρeff,0\rho_{\mathrm{eff},0}. These are roughly consistent with the zero-temperature values from quadratic fits to the MSF dc resistivity tensor data discussed in Appendix A2, which are ρ0,i=(0.08,0.09,1) μΩ\rho_{0,i}=(0.08, 0.09, 1)\ \mu\Omega m (A in Table II). While there are some modes in these plots along the wb=0w_b=0 line which depart from the overall trend, these are isolated, unlike the clear arc of high residual loss shown in Figure 3. The frequency dependencies of the quantities in Figure 5 are shown in the insets. There is an apparent ‘tent arch’ shape to these frequency dependencies (though is less clear for ρeff,0\rho_{\mathrm{eff},0}) which peaks at 7 GHz; however, only a small portion of data (comprising the upper limit) forms this frequency dependence, as opposed to the planar dependence on the induced-current weights which involves most of the modes.

Complex conductivity and exceptional modes

We will now discuss our results for the complex conductivity σ\sigma, specifically the real part σ1\sigma_1, which is a measure of the loss; however, this requires more analysis and assumptions to determine than the surface resistance RsR_s, which relies almost directly on the data. For most modes, we observe a monotonic decrease in σ1(T)\sigma_1(T) as the temperature drops through the superconducting state, as shown in the inset of Figure 1, or a peak slightly below TcT_c which gives way to an eventual decrease at low enough temperature. For a few ‘exceptional’ modes, though, we observe a monotonic increase of σ1(T)\sigma_1(T) with decreasing temperature. This results in comparable values of σ1\sigma_1 and σ2\sigma_2 in the superconducting state for these modes, which is unconventional. This exceptional temperature dependence was also observed for a CVT sample [13].

To quantify the ‘exceptionality’ of each mode, we plot the fraction of TcT_c at which the maximum of σ1(T)\sigma_1(T) occurs vs the induced-current weights in Figure 6a. Typical modes have values near 1 in these plots while exceptional modes have values approaching 0. A region of exceptional modes emerges in the induced-current weight space which can be seen as a blue region of the contours in Figure 6a. A top-down view of these contours is also shown in the inset of Figure 3, which demonstrates the correspondence between these exceptional modes and the arc of large residual loss, illustrated by a large, gray arrow in the main figure. The residual loss can also be exhibited through σ1,0\sigma_{1,0}, the extrapolated zero-temperature value of σ1(T)\sigma_1(T), as shown in Figure 6b, for which this same region can be seen in the space of induced-current weight albeit less clearly than in the plot of R0R_0 in Figure 3. As discussed for R0R_0, we observe residual values of σ1(T)\sigma_1(T) both above and below the universal dd-wave loss σ00\sigma_{00}, indicating that the scenario of line nodes is unlikely. The insets of Figure 6 show no particular frequency dependence of either the temperature of maximum σ1(T)\sigma_1(T) below TcT_c or the residual value of σ1(T)\sigma_1(T).

Plots of temperature at maximum conductivity and residual conductivity versus induced-current weightings

Figure 6. Plots vs the induced-current weightings (and vs mode frequency in the insets) of (a) the temperature (normalized by TcT_c) at which σ1\sigma_1 attains its maximum below TcT_c and (b) the residual value of σ1\sigma_1 at zero-temperature. Data for two CVT grown UTe2_2 samples B39 and B40 is shown in orange and yellow, respectively. For the MSF grown UTe2_2 samples, data for sample 23B is shown in magenta while data for sample 28 is shown in red, green, and blue for the cases of the aa-, bb-, and cc-axes aligned with axis of the ‘hot finger’, respectively. Contours of the data are also included in the waw_a-wcw_c plane with colors (indicated in the colorbars) corresponding to the value of the quantity being plotted. Our estimate of the universal dd-wave residual loss σ00\sigma_{00} is included in (b). This value is also marked with an arrow on the colorbar for the contours.

We also note that there is a region near the aa-direction with larger values of the frequency-independent loss (R0/ω2R_0/\omega^2 or σ1,0\sigma_{1,0}) than those of the surrounding modes and for which the maximum of σ1(T)\sigma_1(T) occurs at a lower temperature; however, this region is much smaller (only two points) than the arc and also has much smaller R0/ω2R_0/\omega^2 values than those of the arc. On the other hand, for σ1,0\sigma_{1,0}, these two points have larger values than those of the rest of the arc; however, the arc of large residual loss is much less prominent in σ1,0\sigma_{1,0} to begin with.

Input power and TLS loss

The analyses performed in these experiments assume a linear response of the sample and the resonator components to the input rf fields. This means that a ratioed parameter such as S21S_{21} does not depend on the input power, and thus, quantities derived from it, such as the surface impedance, are similarly independent of input power. Nonlinearities can arise from the response of the sample (due to the intrinsic nonlinearity of super-

conductors) or external/ background sources, such as an amplifier in the measurement circuit. It is assumed that for some range of input powers (generally low powers, as most physical systems will be pushed to nonlinearity with sufficiently high applied power), the response will be linear. The input power must be carefully selected to achieve a linear response. We find that this power varies greatly between resonant modes and can also vary with temperature. We have inspected the power dependence of each mode we study from the perspective of ensuring a linear response; however, these power dependencies have other significance, such as in determining the influence

of two-level system (TLS) loss.

At low (input) powers, the transmission is noisy and may not be sufficiently prominent above the noise floor, but as the power increases, the signal will emerge from the noise floor where the output power was previously too low. Additionally, a higher input power can clean up the transmission by reducing noise in the measurement of S21S_{21}. It is, thus, favorable to use as high of a power as possible to get an adequate signal to noise ratio; however, too high of an input power will result in nonlinearity. The optimal power is then the maximum for which the response is still linear. Other components, such as amplifiers, may impose additional restrictions on acceptable input powers to avoid damage to equipment. As the input power is raised, the noise near resonance will decrease but the (true) shape of the resonance will not change until the optimal power, above which the resonance will depart from its low power behavior.

To understand at what power ∣S21∣|S_{21}| begins to change shape, we carefully observe how upper and lower extents of the ∣S21∣|S_{21}| vs frequency curve change with input power near resonance. We determine approximate curves that bound the noisy ∣S21∣|S_{21}| data as shown in Figure 7a. We determine a local mean and standard deviation of the ∣S21∣|S_{21}| data at each frequency point. This local mean is just the smoothed ∣S21∣|S_{21}| trace. The upper and lower bounds are set at two standard deviations above and below the mean, respectively. The reduction of noise due to increasing power can then be seen in Figure 7b. As the power increases from red to pink, the upper (solid) and lower (dashed) bounds move together while staying inside the bounds of the previous power, up until the dark blue curve (-13 dBm). Above this power, the upper bound begins to increase as seen in the purple and pink curves. Figure 7c plots the bounds (upper-solid, lower-dashed) on the transmission on resonance vs power for multiple temperatures. In this plot, the minimum in the upper bound (which is marked with a black triangle) indicates the optimal input power, which increases with temperature. The cyan curves, marked 1.56 K, correspond to the examples in Figure 7a,Figure 7b. The cyan upper bound of Figure 7c shows a minimum at −13 dBm, which agrees with the results of Figure 7b. These transmission measurements swept over various powers must be repeated at multiple temperatures throughout the superconducting state and into the normal state to determine the dependence of the linear regime on temperature. The purple upper bound (0.26 K) shows a minimum at −21 dBm, and the red upper bound (2.36 K) shows a minimum at −9 dBm.

Four-panel plots of transmission bounds and power dependence

Figure 7. (a) An example of the raw transmission data ∣S21∣|S_{21}| in red and estimates of the upper (solid) and lower (dashed) bounds of the data in black. (b) Evolution of the bounds (solid for upper and dashed for lower) on ∣S21∣|S_{21}| near resonance with power increasing from red to pink. (c) Bounds of ∣S21∣|S_{21}| on resonance plotted vs power for three different temperatures. Optimal powers are marked for each temperature. (d) As in (c) but in the absence of a sample. The insets of (c) and (d) show the power dependencies of the corresponding quality factors for each temperatures.

We have broadly observed the increase of the optimal power with temperature in the superconducting state for all the UTe2_2 samples we have studied. In the normal state, though, the optimal power remains mostly constant in temperature. In the absence of a sample, however, we have observed that the optimal power is fairly temperature independent. In Figure 7d, the upper bounds at different temperatures for the corresponding background run all show a minimum near −21 dBm (which is marked); whereas, over roughly the same temperature range, Figure 7c indicated a significant departure from this optimal power at different temperatures with a sample present.

Above the optimal input power, ∣S21∣|S_{21}| tends to increase (both upper and lower bounds together). This is consistent with an increase in the quality factor (decrease in loss); however, the magnitude of the transmission at a sharp resonance can still be shifted up or down by changes in broad neighboring modes without necessarily affecting the QQ of the sharp mode. The power dependence of the quality factor at various temperatures is included in the insets of Figure 7c, Figure 7d. Even though for the sample run, ∣S21∣|S_{21}| on resonance is higher than it is for the background run at low temperatures, it is clear from the insets that QQ is consistently higher for the background run than for the sample run. The quality factor does indeed tend to increase with power (reduced loss at high powers) above the optimal power, though. While such behavior is suggestive of the two-level system (TLS) loss mechanism (which can commonly be hosted in dielectrics such as the sapphire and rutile elements used in the resonator), the change in QQ due to sweeping power is much smaller than that induced by changing temperature through the superconducting transition. TLS loss is, furthermore, expected to decrease with temperature, where a crossover between high loss at low temperature and low loss at high temperatures is expected to occur at temperatures around ℏω/kB∼0.4\hbar\omega/k_{\mathrm{B}}\sim0.4 K. For the background case shown in Figure 7d, we do not observe the expected increase in QQ with respect to temperature for the TLS loss mechanism, and in Figure 7c, the loss dramatically increases with temperature due to the superconducting transition. The temperature of the sapphire ‘hot finger’ and sample (if present), which is specified in Figure 7c, Figure 7d, is higher than that of the other dielectrics in the resonator, which are nominally held at ∼100\sim100 mK. Despite this, we would still expect to see some increase of QQ with temperature due to the heat which does leak in to the resonator body and attached dielectrics. It is unlikely, then, that a TLS mechanism is responsible for the increase in QQ with power which we observe in our data for this range of input powers; furthermore, this effect produces a far smaller change in QQ than does the superconducting transition, so superconductivity should be dominant over this effect. Since this effect is observed across all samples we have studied as well as in the absence of a sample, it is very unlikely that it is due to the sample, so the background subtraction to isolate the resonant properties of the sample [13, 38] should remove any trace of this effect. The sample and background runs are performed at different powers by design to maintain a consistent linear response. The observed effect is likely due to nonlinearities of the output amplifier.

References

References

  1. [1]S. Ran, C. Eckberg, Q.-P. Ding, Y. Furukawa, T. Metz, S. R. Saha, I.-L. Liu, M. Zic, H. Kim, J. Paglione, and N. P. Butch, Nearly ferromagnetic spin-triplet superconductivity, Science 365, 684 (2019).DOI
  2. [2]S. Ran, I.-L. Liu, S. R. Saha, J. P. Prathum Saraf, and N. P. Butch, Comparison of two different synthesis methods of single crystals of superconducting uranium ditelluride, J. Vis. Exp. 173 (2021).
  3. [3]P. F. S. Rosa, A. Weiland, S. S. Fender, B. L. Scott, F. Ronning, J. D. Thompson, E. D. Bauer, and S. M. Thomas, Single thermodynamic transition at 2 K in superconducting UTe₂ single crystals, Communications Materials 3, 33 (2022).DOI
  4. [4]H. Sakai, P. Opletal, Y. Tokiwa, E. Yamamoto, Y. Tokunaga, S. Kambe, and Y. Haga, Single crystal growth of superconducting UTe₂ by molten salt flux method, Phys. Rev. Mater. 6, 073401 (2022).DOI
  5. [5]D. Aoki, Molten salt flux liquid transport method for ultra clean single crystals UTe₂, Journal of the Physical Society of Japan 93, 043703 (2024).
  6. [6]E. Svanidze, A. Leithe-Jasper, M. Schmidt, N. Zaremba, M. Krnel, Y. Prots, U. Burkhardt, M. König, R. Ramlau, B. H. Goodge, and Y. Grin, Chemical facets of superconductivity in UTe₂, Journal of the American Chemical Society 147, 35809 (2025).DOI
  7. [7]G. Nakamine, S. Kitagawa, K. Ishida, Y. Tokunaga, H. Sakai, S. Kambe, A. Nakamura, Y. Shimizu, Y. Homma, D. Li, F. Honda, and D. Aoki, Superconducting properties of heavy fermion UTe₂ revealed by ¹²⁵Te-nuclear magnetic resonance, Journal of the Physical Society of Japan 88, 113703 (2019).
  8. [8]I. M. Hayes, D. S. Wei, T. Metz, J. Zhang, Y. S. Eo, S. Ran, S. R. Saha, J. Collini, N. P. Butch, D. F. Agterberg, A. Kapitulnik, and J. Paglione, Multicomponent superconducting order parameter in UTe₂, Science 373, 797 (2021).DOI
  9. [9]M. O. Ajeesh, M. Bordelon, C. Girod, S. Mishra, F. Ronning, E. D. Bauer, B. Maiorov, J. D. Thompson, P. F. S. Rosa, and S. M. Thomas, Fate of time-reversal symmetry breaking in UTe₂, Phys. Rev. X 13, 041019 (2023).DOI
  10. [10]N. Azari, M. Yakovlev, N. Rye, S. R. Dunsiger, S. Sundar, M. M. Bordelon, S. M. Thomas, J. D. Thompson, P. F. S. Rosa, and J. E. Sonier, Absence of spontaneous magnetic fields due to time-reversal symmetry breaking in bulk superconducting UTe₂, Phys. Rev. Lett. 131, 226504 (2023).DOI
  11. [11]Q. Gu, S. Wang, J. P. Carroll, K. Zhussupbekov, C. Broyles, S. Ran, N. P. Butch, J. A. Horn, S. Saha, J. Paglione, X. Liu, J. C. S. Davis, and D.-H. Lee, Pair wave function symmetry in UTe₂ from zero-energy surface state visualization, Science 388, 938 (2025).DOI
  12. [12]T. Metz, S. Bae, S. Ran, I.-L. Liu, Y. S. Eo, W. T. Fuhrman, D. F. Agterberg, S. M. Anlagnе, N. P. Butch, and J. Paglione, Point-node gap structure of the spin-triplet superconductor UTe₂, Phys. Rev. B 100, 220504 (2019).
  13. [13]S. Bae, H. Kim, Y. S. Eo, S. Ran, I. I. Liu, W. T. Fuhrman, J. Paglione, N. P. Butch, and S. M. Anlage, Anomalous normal fluid response in a chiral superconductor UTe₂, Nature Communications 12, 2644 (2021).DOI
  14. [14]K. Ishihara, M. Roppongi, M. Kobayashi, Y. Mizukami, H. Sakai, Y. Haga, K. Hashimoto, and T. Shibauchi, Chiral superconductivity in UTe₂ probed by anisotropic low-energy excitations, Nature Communications 14 (2023).DOI
  15. [15]I. M. Hayes, T. E. Metz, C. E. Frank, S. R. Saha, N. P. Butch, V. Mishra, P. J. Hirschfeld, and J. Paglione, Robust nodal behavior in the thermal conductivity of superconducting UTe₂, Phys. Rev. X 15, 021029 (2025).
  16. [16]E. I. Blount, Symmetry properties of triplet superconductors, Phys. Rev. B 32, 2935 (1985).
  17. [17]F. Theuss, A. Shragai, G. Grissonnanche, I. M. Hayes, S. R. Saha, Y. S. Eo, A. Suarez, T. Shishidou, N. P. Butch, J. Paglione, and B. J. Ramshaw, Single-component superconductivity in UTe₂ at ambient pressure, Nature Physics 20, 1124 (2024).DOI
  18. [18]S. Suetsugu, M. Shimomura, M. Kamimura, T. Asaba, H. Asaeda, Y. Kosuge, Y. Sekino, S. Ikemori, Y. Kasahara, Y. Kohsaka, M. Lee, Y. Yanase, H. Sakai, P. Opletal, Y. Tokiwa, Y. Haga, and Y. Matsuda, Fully gapped pairing state in spin-triplet superconductor UTe₂, Science Advances 10, eadk3772 (2024).DOI
  19. [19]S. Hosoi, K. Imamura, M. M. Bordelon, E. D. Bauer, S. M. Thomas, F. Ronning, P. F. S. Rosa, R. Movshovich, I. Vekhter, and Y. Matsuda, Pseudo point nodal superconducting gap in spin-triplet UTe₂, Nature Communications 17, 8707 (2026).DOI
  20. [20]Z. Li, C. M. Moir, N. J. McKee, E. Lee-Wong, R. E. Baumbach, M. B. Maple, and Y. Liu, Observation of odd-parity superconductivity in UTe₂, Proceedings of the National Academy of Sciences 122, e2419734122 (2025).
  21. [21]I. M. Hayes, E. Fang, S. R. Saha, V. Mishra, P. J. Hirschfeld, and J. Paglione, Field-angle-resolved heat transport in UTe₂: determination of nodal positions in the superconducting order parameter (2025), arXiv:2512.13877 [cond-mat.supr-con].arxiv.org/abs/2512.13877
  22. [22]S. Wang, K. Zhussupbekov, J. P. Carroll, B. Hu, X. Liu, E. Pangburn, A. Crepieux, C. Pepin, C. Broyles, S. Ran, N. P. Butch, S. Saha, J. Paglione, C. Bena, J. C. S. Davis, and Q. Gu, Odd-parity quasiparticle interference in the superconductive surface state of UTe₂, Nature Physics 21, 1555 (2025).DOI
  23. [23]Y. Iguchi, H. Man, S. M. Thomas, F. Ronning, P. F. S. Rosa, and K. A. Moller, Microscopic imaging homogeneous and single phase superfluid density in UTe₂, Phys. Rev. Lett. 130, 196003 (2023).DOI
  24. [24]S. Kamat, J. Dans, S. Saha, A. D. Kokovin, J. Paglione, J. Schmalian, and B. J. Ramshaw, Thermodynamic discovery of tetracritically emergent multicomponent superconductivity in UTe₂ (2026), arXiv:2603.17905 [cond-mat.supr-con].
  25. [25]H. S. Rösing, M. Geier, A. Kreisel, and B. M. Andersen, Thermodynamic transitions and topology of spin-triplet superconductivity: Application to UTe₂, Phys. Rev. B 109, 054521 (2024).
  26. [26]H. Christiansen, M. Geier, B. M. Andersen, and A. Kreisel, Nodal superconducting gap structure and topological surface states of UTe₂, Phys. Rev. B 112, 054510 (2025).DOI
  27. [27]A. Aishwarya, J. May-Mann, A. Raghavan, L. Nie, M. Romanelli, S. Ran, S. R. Saha, J. Paglione, N. P. Butch, E. Fradkin, and V. Madhavan, Magnetic-field-sensitive charge density waves in the superconductor UTe₂, Nature 618, 928 (2023).DOI
  28. [28]Q. Gu, J. P. Carroll, S. Wang, S. Ran, C. Broyles, H. Siddiquee, N. P. Butch, S. R. Saha, J. Paglione, J. C. S. Davis, and X. Liu, Detection of a pair density wave state in UTe₂, Nature 618, 921 (2023).DOI
  29. [29]A. G. Eaton, T. I. Weinberger, N. J. M. Popiel, Z. Wu, A. J. Hickey, A. Cabala, J. Pospíšil, J. Prokleška, T. Haidamak, G. Bastien, P. Opletal, H. Sakai, Y. Haga, R. Nowell, S. M. Benjamin, V. Sechovský, G. G. Lonzarich, F. M. Grosche, and M. Vališka, Quasi-2d fermi surface in the anomalous superconductor UTe₂, Nature Communications 7, 223 (2024).
  30. [30]C. Broyles, Z. Rehfuss, H. Siddiquee, J. A. Zhu, K. Zheng, M. Nikolo, D. Graf, J. Singleton, and S. Ran, Revealing a 3D fermi surface pocket and electron-hole tunneling in UTe₂ with quantum oscillations, Phys. Rev. Lett. 131, 036501 (2023).
  31. [31]A. Carlton-Jones, A. Suarez, Y.-S. Eo, I. M. Hayes, S. R. Saha, J. Paglione, N. P. Butch, and S. M. Anlage, Revealing isotropic abundant low-energy excitations in UTe₂ through complex microwave surface impedance, Phys. Rev. B 112, 014519 (2025).
  32. [32]R. A. Waldron, The Theory of Waveguides and Cavities (Maclaren & Sons Ltd., London, 1967).
  33. [33]S. Sridhar and W. L. Kennedy, Novel technique to measure the microwave response of high T_c superconductors between 4.2 and 200 K, Review of Scientific Instruments 59, 531 (1988).DOI
  34. [34]D. L. Rubin, K. Green, J. Gruschus, J. Kirchguessner, D. Moffat, H. Padamsee, J. Sears, Q. S. Shu, L. F. Schneemeyer, and J. V. Waszczak, Observation of a narrow superconducting transition at 6 GHz in crystals of YBa₂Cu₃O₇, Physical Review B 38, 6538 (1988).pubmed.ncbi.nlm.nih.gov/9945326
  35. [35]D. A. Bonn, D. C. Morgan, and W. N. Hardy, Split-ring resonators for measuring microwave surface-resistance of oxide superconductors, Review of Scientific Instruments 62, 1819 (1991).
  36. [36]S. M. Anlage and D. H. Wu, Magnetic penetration depth measurements in cuprate superconductors, Journal of Superconductivity 5, 395 (1992).DOI
  37. [37]W. N. Hardy, D. A. Bonn, D. C. Morgan, R. X. Liang, and K. Zhang, Precision-measurements of the temperature-dependence of lambda in YBa₂Cu₃O₆.₉₅ - strong evidence for nodes in the gap function, Physical Review Letters 70, 3999 (1993).pubmed.ncbi.nlm.nih.gov/10054019
  38. [38]J. Mao, D. H. Wu, J. L. Peng, R. L. Greene, and S. M. Anlage, Anisotropic surface impedance of YBa₂Cu₃O₇−δ single-crystals, Physical Review B 51, 3316 (1995).pubmed.ncbi.nlm.nih.gov/9979138
  39. [39]M. R. Trunin, Temperature dependence of microwave surface impedance in high-Tc single crystals: Experimental and theoretical aspects, Journal of Superconductivity 11, 381 (1998).DOI
  40. [40]R. J. Ormeno, A. Sibley, C. E. Gough, S. Sebastian, and I. R. Fisher, Microwave conductivity and penetration depth in the heavy fermion superconductor CeCoIn₅, Physical Review Letters 88, 047005 (2002).pubmed.ncbi.nlm.nih.gov/11801159
  41. [41]M. A. Hein, R. J. Ormeno, and C. E. Gough, High-frequency electrodynamic response of strongly anisotropic clean normal and superconducting metals, Phys. Rev. B 64, 024529 (2001).DOI
  42. [42]R. J. Ormeno, M. A. Hein, T. L. Barraclough, A. Sibley, C. E. Gough, Z. Q. Mao, S. Nishizaki, and Y. Maeno, Electrodynamic response of Sr₂RuO₄, Physical Review B 74, 092504 (2006).DOI
  43. [43]P. J. Baker, R. J. Ormeno, C. E. Gough, Z. Q. Mao, S. Nishizaki, and Y. Maeno, Microwave surface impedance measurements of Sr₂RuO₄: The effect of impurities, Physical Review B 80, 115126 (2009).DOI
  44. [44]P. J. Baker, R. J. Ormeno, C. E. Gough, Y. Matsushita, and I. R. Fisher, Microwave surface impedance measurements of Tl₂Pb₁−xTe: A proposed negative-U induced superconductor, Physical Review B 81, 064506 (2010).
  45. [45]G. Ghigo, D. Torsello, R. Gerbaldo, L. Gozzelino, F. Laviano, and T. Tamegai, Effects of heavy-ion irradiation on the microwave surface impedance of (Ba₁−xKx)Fe₂As₂ single crystals, Superconductor Science and Technology 31, 034006 (2018).
  46. [46]F. Gross, B. S. Chandrasekhar, D. Einzel, K. Andres, P. J. Hirschfeld, H. R. Ott, J. E. Beuers, Z. Fisk, and J. L. Smith, Anomalous temperature dependence of the magnetic field penetration depth in superconducting UBe₁₃, Zeitschrift für Physik B Condensed Matter 64, 175 (1986).DOI
  47. [47]R. A. Klemm, K. Scharnberg, D. Walker, and C. T. Rieck, Electromagnetic response of unconventional superconductors, Zeitschrift für Physik B Condensed Matter 72, 139 (1988).
  48. [48]P. J. Hirschfeld, W. O. Putikka, and D. J. Scalapino, Microwave conductivity of d-wave superconductors, Physical Review Letters 71, 3705 (1993).pubmed.ncbi.nlm.nih.gov/10055051
  49. [49]M. Matsumoto and M. Sigrist, Quasiparticle states near the surface and the domain wall in a pₓ ± ipᵧ-wave superconductor, Journal of the Physical Society of Japan 68, 994 (1999).
  50. [50]S. V. Bakurskiy, Y. V. Fominov, A. F. Shevchun, Y. Asano, Y. Tanaka, M. Y. Kupriyanov, A. A. Golubov, M. R. Trunin, H. Kashiwaya, S. Kashiwaya, and Y. Maeno, Local impedance on a rough surface of a chiral p-wave superconductor, Physical Review B 98, 134508 (2018).DOI
  51. [51]E. M. Chia, M. B. Salamon, H. Sugawara, and H. Sato, Probing the superconducting gap symmetry of PrOs₄Sb₁₂: A penetration depth study, Physical Review Letters 91, 247003 (2003).pubmed.ncbi.nlm.nih.gov/14683149
  52. [52]Z. Zhuang and P. Coleman, Topological superconductivity in a spin-orbit coupled kondo lattice, Physical Review B 110, 235418 (2024).DOI
  53. [53]H. C. Choi, S. H. Lee, and B.-J. Yang, Correlated normal state fermiology and topological superconductivity in UTe₂, Communications Physics 7, 273 (2024).arxiv.org/abs/2206.04876
  54. [54]T. C. Wu, H. K. Pal, and M. S. Foster, Topological anomalous skin effect in Weyl superconductors, Physical Review B 103, 104517 (2021).DOI
  55. [55]Y. S. Eo, S. Liu, R. Saha, H. Kim, S. Ran, J. A. Horn, H. Hodovanets, J. Collini, T. Metz, W. T. Fuhrman, A. H. Nevidomskyy, J. D. Denlinger, N. P. Butch, M. S. Fuhrer, L. A. Wray, and J. Paglione, c-axis transport in UTe₂: Evidence of three-dimensional conductivity component, Physical Review B 106, L060505 (2022).DOI
  56. [56]L. Zhang, C. Guo, M. Graf, C. Putzke, M. BordeIon, E. D. Bauer, S. M. Thomas, F. Ronning, P. F. S. Rosa, and P. J. W. Moll, Dimensionality of the reinforced superconductivity in UTe₂, Nature Communications 16, 10308 (2025).DOI

Paper details

vlnc:W_npcsfu2d
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