Anomalous nuclear spin coherence in superconducting Nb 3 Sn Gan Zhai, William P. Halperin, Arneil P. Reyes, Sam Posen, Chiara Tarantini, Manish Mandal, David C. Larbalestier Abstract We have investigated the normal and superconducting states of the technologically important compound Nb 3 Sn using 93 Nb nuclear magnetic resonance. From spin-lattice relaxation we find strong suppression of the zero-temperature superconducting order parameter by magnetic field. We have identified an anomalously large electron-nuclear exchange interaction from spin-spin relaxation measurements, an order of magnitude beyond that of the dipole-dipole interaction, and thereby sensitive to vortex dynamics and vortex pinning. I Introduction Nb 3 Sn is a superconductor with high critical temperature T_{c}(\sim 18\, K) and high critical field H_{c2}(\sim 30\, T). Zhou et al. 2011 ; Zhou 2011 ; Mandal et al. 2025 Its unique physical properties have led to applications in superconducting magnet technology, especially for achieving very high magnetic fields. Little is known about this compound from the microscopic perspective provided by nuclear magnetic resonance, 93 Nb (NMR). In our previous brief work, we surveyed the 93 Nb NMR properties of Nb 3 Sn in both normal and superconducting states. Zhai et al. 2024 Here we report detailed measurements of the spin-lattice, i.e. longitudinal, relaxation rate ( T_{1}^{-1} ) at low temperatures to provide a measure of the average amplitude of the superconducting order parameter and its suppression by magnetic field. However, the focus in the present work, is the temperature dependence of the spin-spin, i.e. transverse, relaxation rate ( T_{2}^{-1} ). Previous NMR studies of superconducting compounds, most notably in high- T_{c} superconductors, Recchia et al. 1997 ; Bachman et al. 1998 has related T_{2} to vortex dynamics. This is also possible for Nb 3 Sn owing to its anomalously high nuclear spin coherence. The measure of nuclear spin coherence is expresssed by the relaxation time T_{2} which is determined by magnetic field fluctuations at relatively low frequencies, of order kHz, in the direction of the steady magnetic field. In metallic materials magnetic field fluctuations can couple to the nuclear spin through the hyperfine interaction via conduction electrons. In special cases this Ruderman-Kittel indirect exchange interaction (RKKY) Ruderman and Kittel 1954 between like nuclear spins reduces the nuclear dipole-dipole interaction leading to substantial narrowing of the NMR spectrum; often described as being similar to extreme motional narrowing in liquids. Anderson and Weiss 1953 ; Slichter 2013 The prime example in a metal is 195 Pt NMR in pure platinum. Walstedt et al. 1962 We discovered that Nb 3 Sn is in the same class of such systems with an unusually long T_{2} , perhaps not surprising since niobium has a single, non-zero spin, isotope 93 Nb ( I=9/2 ), a requirement for this phenomenon. We discuss this later, and, most importantly, the effect of the superconducting state on the RKKY interaction. II Sample preparation and characterization The high-quality powder sample was produced in the Applied Superconductivity Center of the National High Magnetic Field Laboratory following a process of high-energy ball milling, initial cold isostatic press densification, and final hot isostatic press reaction and densification Zhou 2011 . In this work we have two samples with slightly different preparation process. The parameters of the two samples are listed in Table 1 . According to X-ray diffraction (XRD) characterization, Nb 3 Sn has a typical cubic A15 crystal structure, shown in Fig. 1 . The Sn atoms sit on the corners and the body-centered positions of a unit cell, while the Nb atoms occupy three orthogonal chains on the faces of the cube. Figure 1: Crystal structure of stoichiometric Nb 3 Sn. Sample Composition Heat Treatment H_{c2}(0) T_{c} at 0 field Weight No.1 27% Sn 1800 °C 28.9 T 17.4 K 20 mg No.2 25% Sn 1600 °C 30.2 T 18.0 K 220 mg Table 1: A summary of Nb 3 Sn powder samples used for NMR measurements; also Supplementary Materials. III Experiment The NMR experiments were performed at the National High Magnetic Field Laboratory with magnetic fields varying from 3.7 T to 15 T. The superconducting transition temperature T_{c} at different fields was identified from heat capacity measurement, which is consistent with the detuning of the NMR coil at onset of superconductivity. The T_{2}^{-1} was measured by a 90^{\circ}-180^{\circ} pulse sequence, known as the Hahn echo Hahn 1950 , with varying time interval \tau between the two pulses. The T_{2}^{-1} was obtained by fitting the recovery curve, which will be discussed later. The T_{1}^{-1} is measured using a 90^{\circ}-\tau-90^{\circ}-\tau_{0}-180^{\circ} pulse sequence, where the Hahn echo is acquired with a fixed \tau_{0} from a pure Lorentzian recovery. IV Results IV.1 Longitudinal Relaxation The longitudinal relaxation rate T_{1}^{-1} was measured at the central transition in the superconducting state at 3.7 T, 7.5 T, 11.3 T and 15.0 T respectively in Sample 1. It was found to be the same at satellite frequencies. The T_{1}^{-1} was obtained by fitting the recovery curve M(t) to a stretched ”Master equation” Suter et al. 1998 with \beta\sim 0.9 for the five pairs of quadrupolar satellites: \begin{split}M(t)=M_{0}-&M_{0}[\frac{1}{165}\mathrm{exp}(-t/T_{1})^{\beta}+\frac{24}{715}\mathrm{exp}(-6t/T_{1})^{\beta}\\ &+\frac{6}{65}\mathrm{exp}(-15t/T_{1})^{\beta}+\frac{1568}{7293}\mathrm{exp}(-28t/T_{1})^{\beta}\\ &+\frac{7938}{12155}\mathrm{exp}(-45t/T_{1})^{\beta}]\end{split} (1) Figure 2: 93 Nb longitudinal relaxation rate T_{1}^{-1} in the superconducting state of the Nb 3 Sn powder (Sample 1.). The lines are the fitting results using Eq. 2 and Eq. 3 at different magnetic fields. The slope approximation in the low temperature region represents the average energy gap at zero temperature ( \Delta(0) ). Inset: \Delta(0)/k_{B}T_{c} vs. external field where \Delta(0) comes from the fitting, the dashed line is a guide to the eye. The temperature dependence of T_{1}^{-1} in the low temperature region is plotted in Fig. 2 . The data are fitted by Eq. 2 , identifying the dependence of T_{1}^{-1} on the superconducting energy gap \Delta(T) , and Eq. 3 , illustrating the temperature dependence of the gap with strong-coupling taken into account Halperin and Varoquaux 1990 : T_{1}^{-1}\propto\mathrm{exp}(-\Delta(T)/k_{B}T) (2) \Delta(T)=\Delta(0)\mathrm{tanh}\{\frac{\pi k_{B}T_{c}}{\Delta(0)}[\frac{2}{3}(\frac{T_{c}}{T}-1)\frac{\Delta C}{C}]^{1/2}\}, (3) where the heat capacity jump \Delta C/C incorporates strong-coupling in a BCS energy gap. The energy gap in the zero temperature limit \Delta(0) was obtained, taking the temperature dependence of T_{1}^{-1} and \Delta C/C from heat capacity measurements. This is plotted in the inset of Fig. 2 as a function of T_{c}/T . We observe a clear decrease in the energy gap as a function of the external magnetic field, indicating suppression of the order parameter by field. This suppression behavior is also seen in V 3 Si from STM measurements. Ding et al. 2023 The \Delta(0) at 15 T is even lower than the BCS weak coupling value 1.76\,k_{B}T_{c} . In this temperature range there is no evidence of a linear temperature term associated with vortex cores. Fetter and Hohenberg 1968 IV.2 Transverse Relaxation Figure 3: Temperature dependence of 93 Nb transverse relaxation rate T_{2}^{-1} of the Nb 3 Sn powder in various fields (Sample 1.). (a) Relaxation rate T_{2}^{-1} vs. Temperature. The lines are guides to the eye. Inset: T_{2}^{-1} Jump at T_{c} in different fields. (b) The same relaxation rate T_{2}^{-1} vs. T/T_{c}(H) . Plotting versus temperature and magnetic field clearly shows common aspects in both the superconducting and normal states in this sample. The dashed lines in the normal state are linear fits and in the superconducting state a guide to the eye. The extrapolations to T=0 are notable. The transverse relaxation rate T_{2}^{-1} was measured at the central transition in both the normal and superconducting states at various fields. For Sample 1, this is shown in Fig. 3 . The T_{2}^{-1} was obtained from a stretched exponential fit to measured relaxation curves. In Fig. 3 (a,b), T_{2}^{-1} exhibits a linear dependence on temperature in the normal state, despite the fact that the nuclear dipole-dipole interaction is temperature-independent. However, in addition to the dipole-dipole interaction, T_{2}^{-1} is also constrained by the lifetime of the excited nuclear spin, referred to as the ”Redfield effect” Redfield 1957 ; Slichter 2013 , which explicitly takes into account the dependence of T_{2}^{-1} on T_{1}^{-1} . From the Korringa law, T_{1}^{-1} is linear in the normal state, consequently this linear behavior of T_{2meas}^{-1} strongly suggests a Redfield effect expressed as, T_{2meas}^{-1}=T_{2}^{-1}+\kappa\cdot T_{1i}^{-1} (4) where T_{2meas}^{-1} is the measured echo response and \kappa\cdot T_{1i}^{-1} represents the Redfield term. The rate T_{1i}^{-1} is the initial component of the energy relaxation over the same time scale as T_{2}^{-1} , corresponding to the first-order approximation in Eq. Master since T_{2}\,<<\,T_{1} by several orders of magnitude. Although important, this requirement has been ignored in previous NMR work with quadrupolar nuclei in superconducting compounds. Recchia et al. 1997 ; Bachman et al. 1998 In Fig. 3 (a) the normal state T_{2}^{-1} is field independent, consistent with expectation of magnetic-field-independence of the dipole-dipole interaction as well as of the density of electronic states that determine T_{1}^{-1} . In the superconducting state, there is clearly a jump of T_{2}^{-1} at T_{c} , shown in Fig. 3 (b). The magnitude of the jump is then measured and plotted in the inset of Fig. 3 (a). The fact that the jump scales with the external field is a clear indication of vortex dynamics, since the vortex density is proportional to the field. In the superconducting state, we find a common linear dependence of T_{2}^{-1} on temperature in Fig. 3 (b) which means that it scales with magnetic field being plotted versus T/T_{c} (H) (Supplementary Materials). Consequently, we also attribute this field dependence to vortex dynamics. Figure 4: Temperature dependence of 93 Nb transverse relaxation rate T_{2}^{-1} of the Nb 3 Sn powder at 9.43 T from Sample 2. Data are shown as dots with error bars less than dot size. The green and red data are the Lorentzian and Gaussian components of T_{2}^{-1} respectively, achieved from fitting the recovery curve with Eq. 5 . The Lorentzian component is further separated into a Redfield component (purple curve) and a dipolar component (blue curve). The Redfield component was calculated from the measurement of T_{1i} plotted in the inset. With the larger Sample 2 we investigated T_{2}^{-1} behavior in more detail, measuring both T_{1i}^{-1} , inset to Fig. 4 , and T_{2}^{-1} at 9.43 T with a significant enhancement in the signal-to-noise ratio of the NMR signal, M(t) allowing separation of the T_{2}^{-1} relaxation into two components from the combined function Bachman et al. 1998 , M(t)=M_{0}e^{-\frac{t}{T_{2e}}}e^{-(\frac{t}{T_{2g}})^{2}} (5) where T_{2e} (green data) and T_{2g} (blue data) are the Lorentzian and Gaussian components of T_{2}^{-1} respectively. For these data, the Lorentzian Redfield component of T_{2meas}^{-1} was determined from T_{1i}^{-1} and Eq. 4 with \kappa=0.8 (purple curve). The latter reflects the constraint of temperature independence of T_{2}^{-1} in the normal state, already evident in the extrapolations in Fig. 3 (b). Consequently, the Gaussian dipolar contribution to T_{2}^{-1} is the red data in Fig. 4 . After subtraction of the Redfield term, the Lorentzian dipolar contribution is given by the blue curve. In the normal state, the Gaussian component of T_{2}^{-1} is approximately temperature independent \sim 1 ms -1 , and much smaller than expected based on the square root of the second moment of the classical dipolar field, at a Nb nucleus in Nb 3 Sn (indicated as j ), with the following equation Abragam 1961 : \overline{\Delta\omega^{2}}=M_{2}=(\frac{\mu_{0}}{4\pi})^{2}\frac{3}{4}\gamma^{4}\hbar^{2}I(I+1)\sum_{k}\frac{(1-3\cos^{2}\theta)^{2}}{r_{jk}^{6}}, (6) where \gamma is the gyromagnetic ratio, {\bf r}_{jk} is the separation vector of two nuclei, j and k , at an angle \theta to the magnetic field, summed over all nuclei k in the Nb 3 Sn structure. From the calculation, we find M_{2}^{1/2}=20\, ms -1 , more than an order of magnitude larger than the experimental data shown in Fig. 3 and Fig. 4 . This is evidence of a very strong indirect exchange interaction between Nb nuclei Slichter 2013 reducing the dipolar T_{2}^{-1} , and modifying the spectral NMR lineshape to be a mixture of Gaussian and Lorentzian character. Yu et al. 1980 In the superconducting state, the Redfield component is gapped out in the low temperature limit following the behavior of T_{1i} , measured independently and shown in the inset to Fig. 4 . The dipolar component remains non-zero in the low temperature limit as expected, consistent with the low temperature limit of Sample 1. The Gaussian component decreases toward zero as the temperature decreases, and the Lorentzian component dominates. We attribute this to a modification of the indirect exchange interaction in the superconducting state. This new result may be important for understanding the nature of competition between superconductivity and magnetic RKKY interactions in metallic systems. Comparing the T_{2}^{-1} result between Sample 1 and Sample 2, we found that Sample 2 does not exhibit a jump at T_{c} . Furthermore, the temperature dependence in the superconducting state also appears to be missing a contribution from vortex dynamics. This suggests that vortices are pinned in Sample 2 in contrast to Sample 1, at least on the kHz time scale. From the sample parameters, we know that Sample 2 has a higher critical field and critical temperature (Supplementary Materials) and is closer to being stoichiometric. Comparisons with other superconducting properties of the two samples is beyond the scope of this work. V Conclusion In summary, the superconducting order parameter amplitude is strongly suppressed by magnetic field in Nb 3 Sn as determined from NMR longitudinal relaxation measurements. From transverse relaxation we found an anomalously high nuclear spin coherence, which we attribute to a strong indirect RKKY exchange interaction between Nb nuclei. We discovered that this interaction is modified in the superconducting state. We also identified vortex dynamics and vortex pinning that is sample dependent. VI