WorksPhysics and Astronomy
Local aging effects in PuB: Growing inhomogeneity and slow dynamics of local-field fluctuations probed by Pu NMR
Abstract
The effect of self-irradiation damage can influence many properties of a radioactive material. Actinide materials involving the decay through alpha radiation have been frequently studied using techniques such as transport, thermodynamics, and x-ray diffraction. The use of nuclear magnetic resonance (NMR) spectroscopy to study such effects, however, has seen relatively little attention. Here, we use Pu NMR to study the local influence of self-damage in a single crystal of the candidate topological insulator plutonium tetraboride (PuB). We first characterize the anisotropy of the Pu resonance in a single crystal and confirm the local axial site symmetry inferred from previous polycrystalline measurements. Aging effects are then evaluated over the timeframe of six years. We find that, though the static NMR spectra may show a slight modulation in their shape, their field-rotation pattern reveals no change in Pu local site symmetry over time, suggesting that aging has a surprisingly small impact on the spatial distribution of the static hyperfine field. By contrast, aging has a prominent impact on the NMR relaxation processes and signal intensity. Specifically, aging-induced damage manifests itself as an increase in the spin-lattice relaxation time , an increased distribution of , and a signal intensity that decreases linearly by 20 % per year. The spin-spin relaxation time in the aged sample shows a strong variation across the spectrum as well as a drastic shortening towards lower temperature, suggesting growth of slow fluctuations of the hyperfine field that are linked to radiation-damage-induced inhomogeneity and could be responsible for the signal wipeout that develops over time.
relaxation time , an increased distribution of , and a signal intensity that decreases linearly by 20 % per year. The spin-spin relaxation time in the aged sample shows a strong variation across the spectrum as well as a drastic shortening towards lower temperature, suggesting growth of slow fluctuations of the hyperfine field that are linked to radiation-damage-induced inhomogeneity and could be responsible for the signal wipeout that develops over time.
Introduction
is the most widely studied isotope of plutonium, due to its availability and experimentally manageable half-life of 24,110 years. The decay of nuclei involves both decay and spontaneous fission, the former being the dominant decay mode. In calculations for metallic -Pu, such a decay process produces a 5.1-MeV particle (helium nucleus) and an 88-keV nucleus.[1] The particle travels a rather long distance (an average of 11 m) and loses 99.9 % of its kinetic energy through collisions with electrons, causing Pu displacements with approximately 265 Frenkel pairs (defects + interstitials). The recoil nucleus travels a shorter distance (about 7.5 nm) and deposits 75% of its energy, producing 2,290 Frenkel pairs and causing nearly all the radiation damage in this process. After these decay events, most of these defects are thermally repaired but a remainder persists as defects in the lattice, which can affect material properties.[2]
Nuclear magnetic resonance (NMR) spectroscopy is a powerful local probe used across many scientific disciplines including solid-state physics, chemistry, and biology. In contrast to other spectroscopic methods, the small energy scale of the nuclear Zeeman energy probed by NMR provides high sensitivity to the low-energy electronic properties, making NMR an ideal tool to explore a material’s magnetic and electronic properties in the vicinity of the NMR-active nuclei. In actinide science, NMR measurements on the ligand atoms have been used extensively to investigate local structures and dynamics, as has been reported for example in Ga stabilized -Pu using NMR.[3, 4, 5, 6, 7] Self-irradiation damage effects have been also evaluated in several materials, including AmO by NMR[8] and Pu containing zircons (ZrSiO) by magic-angle spinning NMR.[9] In the former, the -NMR linewidth showed an appreciable broadening due to self-irradiation damage over a timeframe of a few weeks to years.
Direct NMR from the actinide atoms should have higher sensitivity to the electrons that mostly govern material’s properties and has been, therefore, anticipated to be key to understanding the electronic and structural properties of actinide-based materials. Actinide NMR, however, has been limited so far due to the extremely large hyperfine coupling between unpaired electrons and nuclear spins. The magnetic hyperfine field at the nuclear positions can be very large ( T) and cause an excess broadening of the NMR response, spreading over orders of magnitude. Further, such pronounced local fields result in extremely fast nuclear spin-lattice relaxation times ( s), making the detection of NMR signals particularly challenging. For uranium compounds, studies in the 1980’s to 2000’s detected -NMR signals in a few exceptional cases where magnetic fluctuations of the otherwise unpaired electrons are absent, such as in the antiferromagnetic states of -enriched UO[10, 11] and USb[12] as well as in the liquid state of UF.[13] Regarding plutonium compounds, the signal detection proved even more difficult, and the discovery of a -NMR signal in non-magnetic PuO crystals was only reported in 2012.[14] Additional measurements in the same report further showed that variations in the oxygen content of PuO caused a substantial reduction in the size of the NMR Knight shift by 5 – 15 %, indicating that NMR is extremely sensitive to the local environment around Pu atoms. The second observation came in 2019, in the candidate topological insulator PuB that has almost non-magnetic -electron configurations.[15, 16]
Materials and Methods
The aluminum-flux grown single crystal of used here (of about 17.4 mg and dimensions of ) is identical to the one used in Ref. [16], having the isotopic content of Pu (by weight %) of 93.96 % , 5.908 % , 0.098 % , 0.025 % , and 0.013 % . The sample loaded in a copper NMR coil was encapsulated in a Stycast 1266 epoxy cube to avoid radioactive contamination. Titanium frits with - diameter pores were installed at both ends of coil’s axis to seal the containment and to allow thermal contact with He exchange gas in our cryostat. Additional sample preparation details are given in the Supplementary Material of Ref. [16]. All the single crystal NMR measurements were performed between 90 and 1,800 days from the sample’s initially synthesis. Results on the aged single crystal reported here are consistent with those reported earlier[16] on fresh powdered samples.
During six years of the measurement period, the single-crystal sample was thermally cycled several times between room temperature ( K; for storage) and liquid temperature ( K; for measurements). See Figure S1 for the full thermal history. Sample cooling and warming speeds were in the range – Kelvin/min to minimize thermal-contraction damage to the epoxy cube and to avoid potential contamination of the cryostat. All the NMR data to investigate aging effects were collected at a base temperature – K.

Figure S1. Thermal history of the PuB single crystal used in this study. The Pu-NMR experiments at low temperature happened occasionally over in total six years, although the Pu signals were found only within the first three years. Between successive measurements, the sample was stored at room temperature in air.
-NMR measurements were performed using a commercial phase-coherent pulsed NMR spectrometer (Redstone, Tecmag Inc). The nucleus has a nuclear spin , which leads to a single NMR peak without any complication arising from the nuclear quadrupolar interaction relevant for the systems with . To obtain the NMR signal, standard spin-echo techniques were used at a fixed external magnetic field of 8.47–8.60 T with a fixed carrier frequency of 19.465 MHz. The field value was calibrated using the -NMR signal of a copper coil wound around the crystal. Spin-echo signals were recorded at a fixed radio frequency after a conventional spin-echo () pulse sequence. Typical pulse widths were s, which were sufficiently short to excite the full spectral bandwidth in our spectra. Echo signals were then converted to NMR spectra using a Fast Fourier Transformation (FFT). The resonance frequencies in this work were converted to NMR Knight shifts using a bare gyromagnetic ratio MHz/T estimated in earlier studies of .[14]
The nuclear spin-lattice relaxation time was determined from standard stretched-exponential fits to the recovery of nuclear magnetization after an inversion pulse sequence (). In this case the time-dependent nuclear magnetization can be expressed as
where is the nuclear magnetization at an equilibrium state, is the inversion fraction, and is the stretching exponent () that measures the distribution of across the spectrum.
The spin-spin relaxation time was determined from the spin-echo decay of the magnetization as a function of waiting time between two pulses. Because was strongly distributed across the NMR spectrum due to inhomogeneity and its size was orders of magnitude smaller than that of , as we discuss below, the standard approach to decouple Gaussian and exponential contributions to the echo decay was not effective. Instead, a phenomenological stretched exponential form
described the echo decay more appropriately. The exponent () accounts for the change in the character of the echo decay function at a different fraction of the spectrum.
To study the anisotropic nature of the hyperfine interaction, we used a single-axis goniometer (NMR Service GmbH), with an angular resolution of 0.1 degrees, to measure the single-crystal -NMR shift and relaxation rate as a function of rotation of the crystal with respect to an applied magnetic field.
Results
-NMR spectrum and it’s anisotropy in single crystal
Figure 1(b) shows typical -NMR spectra at K in a single crystal of taken after 90 days and 690 days from initial synthesis for the external magnetic field of 8.5-8.6 T applied along the crystalline -axis (). (In-between these two measurements, the sample was thermally cycled several times between room temperature and 4 K as well as stored at 300 K for more than a year, as shown in Figure S1. A relatively broad single peak was assigned to the nuclear spin-level transition for the unique site in the unit cell [see Figure 1(a)]. Except for an overall intensity decrease by a factor of 1.66, about which we discuss below, the two spectra exhibited almost identical line shapes, with no change in the linewidth and resonance frequency at the center of the peak, or equivalently in the value of the Knight shift defined as , where is the nuclear gyromagnetic ratio. The NMR line profile reflects the histogram of the spatial distribution of the static local magnetic (hyperfine) field at the nuclear positions, and the linewidth is proportional to the Knight shift variation across the sample. The latter distribution in a non-magnetic insulator like would primarily reflect the extent of static bonding disorder. That we observed little change in the line shape between 90 and 690 days spectra for suggests that the amount of static disorder in the bonding network has not changed significantly within the first two years of aging or has little influence on the spectra after the first 90 days for this particular field orientation, reflecting an anisotropic hyperfine coupling between the orbital electrons and the nuclei as discussed previously.[16]

Figure 1. (a) Tetragonal crystal structure of PuB₄. Grey balls are plutonium atoms (having locally orthorhombic symmetry), and pink [B(1)], green [B(2)], and blue [B(3)] balls are the three crystallographically nonequivalent boron atoms in the unit cell (given by the light grey box). The colored lines between two atoms are the bondings connecting the nearest neighbor atoms. (b) Typical NMR spectrum in a single crystal of PuB₄ under an external magnetic field of – T applied parallel to the crystalline -axis (). Diamonds are the data taken in a fresh sample (90 days from synthesis), whereas the triangles give the corresponding data taken in the identical sample 600 days later (690 days from synthesis).
To have a closer look into this point, we checked the field-orientation dependence of the single crystal spectrum after aging and compared it with the powder spectrum taken in a fresh sample.[16] Figure 2(a) and Figure 2(b) present the field rotation patterns of the single crystal -NMR spectrum at K for field rotation in the crystalline -plane [Figure 2(a)] and -plane [Figure 2(b)] taken after 640 days from the synthesis. Also shown at the bottom of Figure 2(a) is the corresponding powder -NMR spectrum of a polycrystalline sample from the same batch within the first week of its synthesis. There is an approximately single peak-like spectrum in the single crystal measurements for both rotation planes at all field orientations. The resonance line showed a sinusoidal angular dependence in the -plane rotation [see the inset of Figure 2(a)], whereas little orientation dependence in the peak position was found in the -plane rotation. The overall orientation dependence suggests that the Pu ion sits in an axially symmetric environment along the crystalline -axis and points to a dominantly axially symmetric Knight shift tensor.[17] This is consistent with our earlier consideration of local Pu-site symmetry being axial based on the line-shape analysis of the fresh-sample, powder-spectral pattern shown at the bottom of Figure 2(a).

Figure 2. Field orientation dependence of -NMR spectrum in single crystal at 4 K. The magnetic field of 8.74 T is rotated in the crystalline -plane in (a) and in the -plane in (b). The angle () is defined as the direction of measured from the crystalline -axis to -axis (-axis to -axis). The data are taken after 640 days from synthesis in (a), and after 690 days in (b). Note that the small peak-like structures seen for around the sharp central peak in (a) [labeled S1-S4 in Figure 4(b)] are absent in (b) because we used 33% smaller excitation window in the pulsed NMR measurements to predominantly excite the sharp peak and focus on the angular () dependence of its center. The bottom spectrum in (a) is the corresponding powder spectrum in a fresh polycrystalline sample from the same batch taken within the first week from synthesis (replotted from Ref. [16]). Inset in (a) shows the field orientation dependence of the Knight shift value at the sharp peak position of the spectrum in the -plane plotted as a function of the angle . Solid curve gives a sinusoidal fit to the data points (see the text).
Notably, the maximum (at ) and minimum (at ) values of the Knight shift in the aged (640-days old) single-crystal nicely matches the corresponding values in the fresh (one-week old) powder sample [derived from the peak and kink positions indicated by the arrows in Figure 2(a)].
In fact, by fitting the orientation dependence of the single-crystal spectral peak position in the -plane rotation with a form expected for axially symmetric systems,[32] [see the solid curve in the inset of Figure 2(a)], we get and , whose values are in excellent agreement with those extracted from the fresh polycrystalline powder spectrum.[16] [Here, () is the isotropic (axial) component of the Knight shift, and is the angle measured from the -axis to the -axis.]
Furthermore, the linewidth in the single-crystal roughly accounts for the broadened tail structures seen at the two ends of the powder-NMR pattern.
These findings in Figure 1 and Figure 2 provide evidence that, from the perspective of the static NMR spectrum, aging effects that have developed and accumulated over 690 days have little impact on the axial site symmetry as well as on the magnitude and anisotropy of the Knight-shift tensor. These features allow us to make a direct comparison of the spectral and shift data taken at different times from synthesis.
In Figure 3(a) we show the temperature dependence of the Knight shift for in the aged single crystal (700-730 days old) plotted in comparison with the data for taken in the same sample but in a fresher condition (within 100 days, reported earlier in Ref. [16]). A monotonic decrease in the value of is observed for the two orientations towards lower temperature until levels off at around 80 K. The relative change from the low-temperature saturated value, , shows activated behavior as highlighted in Figure 3(b). To evaluate the excitation gap[15, 16] in a most general way, we fitted with one of the commonly used forms in quantum spin systems,[18]

Figure 3. (a) Temperature dependences of Knight shift in PuB single crystal at T. The results for (circles) are taken at 100 days from synthesis (replotted from Ref. [16]), and those for (squares) are recorded at 700-730 days. Dashed lines represent the leveled off values of the shift at low temperature below 80 K, and is defined as the high temperature values measured from that saturated value for each orientation. (b) Arrhenius plot of plotted against inverse temperature. Solid lines are exponential fits to the data above 200 K with the form (see the text).
The deduced activation gap for the two orientations yield meV for and meV for . The values of the two gaps are very close to each other, suggesting that the isotropic gap has not changed after two years. This provides support for the above arguments that there is qualitatively little impact of aging on the Pu NMR spectra at K, at least after the initial 90 days, which now is justified up to room temperature. (Other empirical fitting forms often employed in the analysis of , such as as summarized in Ref. [18], do not alter the extracted gap sizes and thus do not affect this argument.)
One interesting characteristic seen in the aged sample but omitted so far is that there is a slightly anisotropic feature in the spectrum that becomes obvious if we look more carefully into the spectral details for and , as depicted in Figure 4(a) [enlarged and replotted from Figure 1(b) and Figure 2(a)]. The symmetric shape of the spectrum for changes into a more complicated and asymmetric one for , with a couple of satellite-like peaks showing up at both sides of the main peak [labeled peaks S1 to S4 in Figure 4(b)]. The presence of such structure inside of a spectrum indicates that there could be Pu nuclei sitting in slightly varying local environments. It is tempting to associate this additional line structure with radiation-damage induced alteration of the bonding network around Pu sites. However, as we don’t have the corresponding data in a fresh sample, it is not possible to conclude whether this is due to aging or some other underlying structural defects and/or impurities that were present in the unaged sample. Either way, the effect seems to be rather small and more importantly the overall axial site symmetry is preserved at the site. Therefore, to gain further insight into effects of aging, as a first approximation we omit this minor structural (or bonding) inhomogeneity and concentrate on the global nature of the spectrum, including NMR signal intensity and relaxation parameters ( and ).

Figure 4. (a) Comparisons of the Pu-NMR spectrum for and taken after 640-690 days from synthesis [replotted from Figure 2(a)]. The two curves are shifted horizontally to show them on top of each other. (b) Spectrum for . In addition to the central sharp peak (labeled M), a couple of peak-like features can be resolved in both sides of the spectrum (labeled S1 to S4). Solid curves stand for five Gaussian fits to the spectrum.
Aging effects seen by spin-lattice relaxation rate and spectral intensity
We turn to dynamic aspects of aging in the PuB single crystal. Figure 5(a) presents the temperature dependence of the spin-lattice relaxation rate determined from the fully integrated intensity of the -NMR spectrum for (recorded at 700-730 days from synthesis) and for (at 100 days, replotted from Ref. [16]). decreases rapidly upon cooling from room temperature for both field orientations akin to the Knight shift data in Figure 3(a). An Arrhenius fit in the range K for gives an activation gap meV; whereas, a fit for and K yields an 80% larger value meV, as shown in Figure 5(b).

Figure 5. (a) Temperature dependences of nuclear spin-lattice relaxation rate in PuB single crystal at T. The results for (circles) are taken at 100 days from synthesis (replotted from Ref. [16]), and those for (squares) are recorded at 700-730 days. (b) Arrhenius plot of versus inverse temperature. Solid lines are exponential fits to the data above 200 K (see the text).
Below approximately 100 K, the temperature dependence of becomes qualitatively different between the two orientations. For , tends to saturate as observed in the Knight shift , but for it increases towards lower temperature and then decreases again below 20 K.
Recalling that we do not see any notable anisotropy in the temperature dependence of (Figure 3), the quantitative difference in gap values obtained from for the two orientations (Figure 5) suggests that aging effects dominate as we are comparing results between aging after 700-days () and 100-days () of the same crystal. The different gap values extracted from and in the crystal after 100-days of aging gives a ratio of , which is close to the average 1.73 for typical spin-gap systems.[18] In contrast, after the sample ages for 700 days, this same ratio determined with is about 0.9. (See Appendix B for further discussions.)
To better understand aging effects over an extended time period, we analyze all data taken on the PuB single crystal over six years from its initial synthesis and make systematic comparisons, as summarized in Figure 6. Specifically, we focus on three -NMR quantities recorded at 5 K: (a) the full spectral signal intensity () times temperature (), [Figure 6(a)]; (b) the spin-lattice relaxation time determined from stretched exponential fits to the nuclear relaxation curve [see Eq. (1); Figure 6(b)]; and (c) the associated stretching exponent [Figure 6(c)]. Here, we note that during this time period the sample experienced a complicated thermal cycling history: it was thermally cycled multiple times and also held at intermediate temperatures for various periods at temperatures between 4 K and 300 K, kept at 4 K for about a year within the first 500 days, and stored at room temperature in air on multiple occasions, in total more than four and half years (see Fig. S1). Despite this very complicated thermal history, we can clearly see in Figure 6 a long-term underlying trend in all three NMR quantities as a function of time.

Figure 6. Direct -NMR representation of aging effect in single crystal within three years from synthesis. (a) Integrated spectral intensity times temperature. Arrows indicate three isochronal annealing tests performed at 370 and 400 days (first and second trials, respectively, in Figure 7) and at 690 days (third trial in Figure 8) from initial synthesis. No signal was detected after 1850 days. (b), (c) Nuclear spin-lattice relaxation time of the integrated intensity (b) and the corresponding stretching exponent of the nuclear relaxation curve (c). The data point at 720 days is not shown in (b) as it has different anisotropic factor than all the other points and is not directly comparable. That influence seems to be relatively small for , and thus the corresponding point is present in (c). Dashed and shaded lines are guides to the eyes.
Figure 6(a) presents the first notable aging effect, which is a continuous overall loss of the integrated spectra intensity multiplied by temperature as a function of time. (So-called and corrections are not needed here as our pulse conditions are so chosen that decays coming from and effects are negligibly small.) The value of drops on average by 20% per year so that 40% of the signal is lost after two years from synthesis, and no signal at all remains in the fifth year and later. In NMR spectroscopy, the signal intensity is proportional to the size of the nuclear magnetization that follows the Boltzmann distribution of the nuclear Zeeman states and scales inversely to the temperature (). The intensity multiplied by compensates this Boltzmann factor, so is thus proportional to the number of total nuclear spins probed by the measurements, which is usually a constant. The fact that continuously decreases over time suggests that more and more nuclear spins become invisible in our measurement as aging proceeds. This loss of the signal is not an experimental artifact because the reference signal from the copper coil wrapping the sample shows no notable time-dependent change under NMR pulse conditions. We also note that at various higher temperatures after 690 days from synthesis exhibits no dependence after making proper corrections (see Appendix § 9). This rules out any noteworthy intensity loss as a function of temperature (thermal wipeout) that often appears at low temperatures in various correlated electron systems that show magnetic ordering or glassy charge or spin freezing.[19, 20, 21, 22]
A second aging effect manifests in the remaining signal and is characterized by a continuous change in the profile of the spin-lattice relaxation curve. Namely, the relaxation curve becomes increasingly stretched with increasing time after sample synthesis. Such a stretched feature in widely emerges in solids that possess certain types of electronic and/or structural inhomogeneity[18, 19, 20, 21, 23] and is known to represent a distribution in the value of across the sample that can be well characterized by a stretched exponential fit [Eq. (1)]. For instance, for , the value of represents the full width at half maximum of the distribution on a logarithmic scale, and the value of deduced from stretched fits stands for its median (note that we define as in Eq. (1) hereafter).[24] Figure 6(b) and Figure 6(c) show the median and the exponent deduced from a stretched fit. Notably, [Figure 6(c)] continuously decreases with time and reaches at 730 days, which corresponds to an order of magnitude distribution of .[24] The median value of [Figure 6(b)] seems to be more thermal history dependent than and . Nevertheless, a systematic aging test performed at 5 K, as summarized in Figure 7, captures an increasing trend of with aging at low temperature (see Sec. III.D for details).

Figure 7. First and second isochronal annealing tests probed by NMR in single crystal within the first year from synthesis. Plotted data are (a) the integrated spectral intensity times temperature, (b) the median nuclear spin-lattice relaxation time , and (c) the stretching exponent of the corresponding nuclear relaxation curve. Between annealing tests, the sample was kept at base temperature (5 K) from day 240 to 380 and from day 400 to 440 at T for . The first annealing was then conducted at day 370 [indicated by an arrow with the label 1 in (a)], at which we raised the sample temperature to 300 K, stayed there for three hours, came back to the base temperature, and performed NMR to check the impact of annealing. We then kept the sample at 5 K for one more month and then conducted the second annealing at day 400 [indicated by the second arrow in (a)].
Before moving into the details of the changing dynamics of PuB with aging, let us first reiterate that there are two dynamical manifestations of the aging effect that is found here. The first one is the overall loss of the -NMR signal intensity upon aging, and the other one is the developing inhomogeneity in the remaining signal that shows up as a growing distribution in the relaxation time across the sample when it is held at base temperature.
The intensity loss typically might be associated, at least in part, to disorder induced by alpha decay of atoms that produces a recoil atom as well as a helium nucleus ( particle). The dominant source of radiation damage is known to come from the heavier uranium recoil atoms.[1] In elemental -Pu the recoil uranium atoms trigger a concentrated displacement damage around stopping sites by knocking off the nearby Pu atoms from their original lattice positions. Wolfer[1] estimated the accumulation rate of such displacement damage in -Pu to be 0.068 displacement per atom per year, which means each Pu atom is displaced from its equilibrium position on average in 14.7 years. This displacement rate scales inversely with the sample’s melting temperature,[1] which for -Pu is 500 K and for PuB is 2,400 K,[25] so we expect approximately 0.01 displacement per atom per year in PuB. If we use this estimated displacement rate in PuB and assume that displacements suppress the Pu signal by inducing changes in the static and dynamic local hyperfine environment (leading to an excess shift and too-fast relaxation at the displaced Pu atoms), we would expect that, without annealing the damage, only about 2 % of the signal should be lost after two years of accumulated site displacements. Instead, 40 % of the signal is lost in this period, as seen in Figure 6(a), which includes excursions to room temperature (Figure S1) where some damage should be “healed”. Though other self-damage routes must contribute to a loss of signal, it is extremely unlikely that they could account for a more than order-of-magnitude discrepancy between estimated and observed signal loss. The unexpected excess loss suggests that the consequence of radiation damage is not just simple static displacements of the atoms but has a more spatially extended dynamic influence. A growing distribution in in the spectrum together with this excess signal-intensity loss is consistent with a fluctuating hyperfine field developing and distributing non-uniformly across the sample being a contributing factor. We will discuss such dynamic aspects of aging in more detail in the following Secs. III.D and IV.
Isochronal annealing effect on intensity and
In addition to the spectral and relaxation measurements discussed so far, we have also examined the outcome of isochronal annealing of the PuB single crystal within the first two years after its synthesis.
The annealing test was conducted three times: The sample underwent its first annealing at day 370 after being kept at 5 K over approximately 140 days during which it accumulated radiation damage at low temperature. The second annealing was done at day 400 after keeping the sample at 5 K for another month. Following these two annealing tests, the sample was stored at room temperature in air for over 140 days and was used for measurements at varying temperature conditions between 4 K and 250 K to collect temperature-dependence data. The third annealing test was then conducted at day 690.
For the first two trials, we annealed the sample at K for three hours, then cooled back to 5 K and performed NMR measurements with . For the third trial, we did a systematic incremental annealing in a way such that we first stayed at 4.5 K, raised the sample temperature to a target value of (= 30, 60, 100, 150, and 250 K) and stayed there for an hour, came back to 4.5 K and did NMR measurements with , and then moved to the next higher target and repeated the sequence of hold, cool, and measure.
Figure 7 summarizes the first and second isochronal annealing test results, in which we show the full-spectral NMR signal intensity [Fig. 7(a)], the median spin-lattice relaxation time [Fig. 7(b)], and the stretching exponent [Fig. 7(c)] as a function of time. Arrows in the figure indicate the two annealing tests. Before moving into detailed influences of annealing, we reiterate that cryogenic damage accumulations are observed in these quantities. Over the time at 5 K (day 240 to 380 and day 400 to 440), we see a systematic decrease in and (within error bars) and an increase in the median value of as damage builds up. For the first annealing test all three quantities show a remarkable change after annealing that tends to reduce the accumulated damage. In fact, recovers by 10 %, the median shortens by 30 %, and increases from 0.63 to 0.68, suggesting that the effects of aging have been partially healed by isochronal annealing. The second attempt was, however, not effective, and we see only a minor influence of annealing on the NMR observables. In Fig. 8 we summarize results of the third annealing test conducted a year later. In both the spectrum Figure 8(a) and the spin-lattice relaxation curve Figure 8(b), no change was observed after annealing, suggesting that the accumulated effects of aging are becoming more difficult to remove. That we found a clear impact of annealing in the first year but no effective annealing in the second year implies that there may be a threshold of accumulated damage beyond which annealing at a of 250-300 K is only partially effective in healing damage.

Figure 8. Third isochronal annealing test after 690 days from synthesis. The sample was stored at room temperature after the first two annealing tests in Figure 7 from day 500 to day 640 and then kept at various low temperature conditions from 4 K to 250 K until performing the third annealing test at day 690 [see also Figure 6(a)]. (a) -NMR spectrum at 4.5 K at T for after isochronal annealing performed at various temperatures. Different symbols represent the spectrum recorded after keeping the sample at the annealing temperature for an hour and coming back to 4.5 K. (b) Comparison of nuclear spin-lattice relaxation () curves for full integrated spectrum in (a) before and after annealing at K.
PuB is a refractory material, with strong bonding characterized by a high melting point of nearly 2,430 K [25]. It is likely that much higher temperatures are needed to thermally heal the sample with isochronal annealing. Unfortunately, for safety reasons we were unable to reach such temperatures. Materials with similar bonding, such as actinide oxides, have demonstrated beneficial annealing at 1,500 K, which restored lattice parameters to their undamaged values [26]. In contrast, actinides with metallic bonding, such as Ga stabilized -Pu metal, show responses to annealing at relatively low temperatures [2]. Similar annealing effects at temperatures below 300 K have also been observed in irradiated metals [27]. Together, this suggests that the nature of bonding in a material determines the temperature needed to “heal” the radiation-induced damage. As we have demonstrated, actinide NMR is sensitive to isochronal annealing in a strongly bonding material like PuB in its early stages of damage accumulation, and this study encourages more controlled aging studies of plutonium compounds using direct NMR.
Microscopic inhomogeneity and fluctuating hyperfine fields probed by and
Additional insight into the effects of aging comes the variation of and across the spectrum after 640 days from synthesis. For this, we focus on these relaxation rates that are derived from a fraction of the spectral intensity obtained by integrating a small slice of spectrum (a window size of 0.02 % of ). At that time, the full spectrum already possesses approximately an order of magnitude distribution in the size of [corresponding to the stretching exponent of , see Figure 6(c)], and the full spectral intensity has dropped by 40 % from that in a fresh sample [Figure 6(a)].
Figure 9 shows the distribution of and across the spectrum for at K. Interestingly, the value of (median) does not change appreciably across the spectrum, yet at each point it showed very stretched behavior with the stretching exponent being close to 0.5. This stretched feature is exactly the same as we deduced from analysis of the full spectrum and indicates a distribution of that reflects a microscopically inhomogeneous distribution of fluctuating hyperfine fields. This heterogeneity arises irrespective of a spatial distribution of static hyperfine fields (i.e., the underlying bonding network) that determines the single-crystal spectral shape.

Figure 9. Variation of relaxation parameters in PuB₄ single crystal across -NMR spectrum after 640 days from synthesis. Circles represent the spin-lattice relaxation rate (left axis), and squares gives the spin-spin relaxation rate (right axis), both of which are recorded at 8.47 T for and are determined from the integrated spectral intensity of a small fraction of the full spectrum. The values of were obtained using a fixed stretching exponent of . Top figure shows the stretching exponent of the spin echo decay curve throughout the spectrum [see Eq. (2)].
The dephasing nature of the spin-echo signal gives further insight into this issue. in Figure 9 shows a sizable variation across the spectrum and compared to presents a very different evolution as a function of . The functional form of the spin-echo decay curve is purely exponential [i.e., in Eq. (2)] at the low-shift end of the spectrum (at %), whereas it becomes more Gaussian-like (although not quite purely Gaussian) across the main part of the spectrum with . The spin-spin relaxation process, which determines , is generally related to dephasing of the spin-echo signal due to a magnetic dipole interaction between nuclear spins and the dynamic hyperfine interaction between electrons and nuclear spins.[28] The size of the former contribution can be estimated via van Vleck’s second moment technique[17] and leads, in PuB, to temperature-independent values of msec for like spins (–) and msec for unlike spins (–). These values are remarkably smaller than the experimental values in Figure 9, indicating that the dominant contribution to spin-echo dephasing comes from the latter dynamic hyperfine interaction. As the variation profile of across the spectrum differs significantly from that of , we speculate that there is a distribution in the spectral density of fluctuating hyperfine fields that brings contrasting impacts on and relaxation mechanisms.
The temperature dependence of should offer further insights about inhomogeneity, but when we attempted to measure spin-echo decay in the period 690-700 days after synthesis, the intensity loss due to aging prevented standard measurements, except at low temperatures (Figure S2). Instead, we estimate at various temperatures in a non-standard way. Here we use the temperature dependence of the bare that was determined from full spectral intensities with a fixed waiting time between and pulses (Figure S3). The estimate relies on our observation that the -corrected at 4.2 K has the same magnitude as at 100 K. (Specifically, the correction allows us to compensate for -decaying effects [Eq. (2)] and recover the intensity for an imaginary infinite- limit). The bare shows strong temperature dependence below 60 K and at 4.2 K becomes half of its value at high temperatures (Figure S3). This suggests that the apparent decrease of the bare arises from a shorting of . Assuming a temperature-independent stretching exponent in Eq. (2) and its value at 4.2 K (), we use Eq. (2) to convert the observed intensity in Figure S3 to corresponding estimates of at different temperatures, which are dubbed hereafter (see Appendix § 9 for details). The value of is strictly speaking not identical to the true but seems to provide its reasonable representation.

Figure S2. -NMR spin echo decay curve at 4.2 K in PuB single crystal at 690 days from synthesis at the external magnetic field of 8.47 T for . The echo intensity corresponding to the full integrated spectral intensity is plotted as a function of the waiting time between and pulses in the spin echo. Solid curve gives the best fit to the data with the form , which yields the spin-spin relaxation time of and the stretching exponent of .

Figure S3. -NMR intensity times temperature plotted against temperature. Data is taken in single crystal at 690 days from synthesis at the external magnetic field of 8.47 T for . The waiting time of is used in the spin echo measurement.
Figure 10(a) presents the thus determined temperature dependence of with .
At high temperatures, is weakly temperature dependent as expected for dephasing of nuclear spins by magnetic dipole interactions. With decreasing temperature, however, increases notably below 40 K and at the base temperature reaches five times its high-temperature value. This strong temperature dependence of points to the underlying trend of the true that equally grows toward low temperatures. Note that in the same temperature range, the linewidth of the spectrum does not change significantly as a function temperature [Fig. 10(b)]. The characteristic increase of cannot be driven by the nuclear dipole interaction and combined with little change in the linewidth strongly suggests that fluctuations of the electronic hyperfine field should be responsible for shortening . Interestingly, there is no increase of in the corresponding low temperature range [inset of Fig. 10(a)]. This observation implies that there should be a fluctuating hyperfine field that does not affect the relaxation process but strongly influences the relaxation process in the aged PuB sample.
Discussion
The single crystal -NMR reveals dynamic spin fluctuations that cause a notable distribution of fluctuating hyperfine fields across the sample and that they could be equally responsible for the signal wipeout with little change in the line shape that develops over an extended time period. The shortening of seems to be critical for all of these observations and reflects a time-scale of the longitudinal component (parallel to the external field ) of the spin fluctuations near zero frequency.
The NMR spin-spin relaxation rate can be generally decomposed into nuclear and electronic contributions: . The latter dominates the relaxation process in PuB [i.e., ] as discussed in Sec. III.D. The electronic part can be further written as a sum of two distinct contributions in terms of the auto-correlation function of the fluctuating hyperfine field as[17]
The spin-spin relaxation rate thus picks up fluctuations perpendicular to at the NMR Larmor frequency () [represented by ] and is also sensitive to the longitudinal fluctuations parallel to near zero frequency [given by ].[28] The first term is known to be driven by spin-lattice relaxation processes and leads to the so-called Redfield contribution to , which for a spin system is given by .[28, 29, 30] With over the entire temperature range as seen in the inset of Figure 10(a), our data naturally suggest the growing (and ) towards low temperature to be dominantly caused by longitudinal fluctuations near zero frequency .

Figure 10. Temperature dependence of -NMR spin-spin relaxation rate (a) and full width at half maximum (FWHM) (b) in single crystal at 690 days from synthesis at 8.47 T for . The values of are determined in a non-standard way from the spin-echo decay data at 4.2 K (see Appendix § 9 for details). Dashed lines are guides to the eyes. The FWHM is determined from the linewidth of the central peak [the line M in Figure 3(a)] by performing multi Gaussian fits to the full spectrum. Inset of (a): Temperature dependence of plotted with the corresponding Redfield contribution [replotted from Figure 5(a)].
We recall that the so far discussed and is averaged over the full spectrum, and in aged PuB, strongly distributes across the spectrum at least at base temperature, as seen in Figure 9. When only a fraction of the spectrum is excited on the lower frequency side where is largest, the functionality of the echo-decay curve becomes purely exponential. Such an exponential decay is characteristic of fluctuations of the longitudinal component of the hyperfine field [31] in the fast motion regime, , and the corresponding spin-spin-relaxation rate is given by[30]
where is the correlation time of the fluctuating hyperfine field that is defined by an exponentially decaying auto-correlation function as . The increasing trend of in Figure 10(a) thus indicates a rapid increase of with decreasing temperature. Importantly, as is strongly distributed across the sample, the fluctuations should constitute a finite spectrum with a certain width. The spectral intensity loss we observe, therefore, could be due to the low frequency tale of this spectrum in which gets so large that the time scale of becomes shorter than the dead time of the measurement, resulting in the signal wipeout.
Growing slow fluctuations of such a longitudinal hyperfine field have been reported in a variety of cases. For example, similar fluctuations were observed in classical and quantum spin[32, 33, 34] and charge[35, 36] glasses, ionic and molecular motion,[31, 37, 38] charge- and spin-ordered cuprates,[19, 39, 40] frustrated[41, 42] and quantum critical[43] magnets, as well as disordered Mott insulators[44, 45] and nematic domain walls.[46, 47] In these cases, spatial inhomogeneity seems to be required for the fluctuations to develop.
For a Pu-based material like PuB, self-irradiation damage due to radioactive decay of Pu atoms[1] is the most plausible source of inhomogeneity, which can cause accumulated magnetic and non-magnetic defects upon aging.[48] In our case, -decay of (93.96 weight %, half-life 24,110 years) should be one of the main causes. This decay channel generates thousands of non-magnetic lattice defects (Frankel pairs) per decay[1] and recoil atoms as a daughter product at a rate of ppm/yrs. The -decay of (0.025 weight %, half-life 14.329 years) is another source that results in an accumulation of atoms at an even faster rate of ppm/yrs. Both of these recoil atoms can be radicals that behave as paramagnetic centers[48] that from ligand NMR are known to cause a shortening of and in turn a dynamic NMR signal wipeout[49] at nearby () nuclei due to the direct dipole-type hyperfine coupling that provides fluctuating hyperfine field at the target nuclei. However, the amount of radicals and the number of nearby atoms feeling the fluctuating field seems to be way too small to account for the observed massive intensity loss in NMR that develops as fast as %/yrs. This notable discrepancy suggests that there are some indirect hyperfine coupling processes that boost the signal wipeout.
For a metallic system, one possible origin for this discrepancy could be the indirect relaxation between radical spins and host nuclear spins that is assisted by conduction electrons. This type of indirect hyperfine interaction is a nuclear spin version of the Ruderman-Kittel-Kasuya-Yosida (RKKY) interaction. It drives an oscillatory interaction that makes the influence of radical-induced electron spin polarization spatially more extended (see for example Ref. [50]). In magnetic dilute alloys the impact of such an indirect interaction on the NMR relaxation parameters and intensity has been extensively studied and is shown to cause a sizeable shortening of and even for a tiny magnetic impurity concentration of a few hundreds of ppm or less per unit cell.[50, 51, 52, 53, 54, 55] The observed pronounced intensity loss could be, therefore, due to this sort of indirect (RKKY) interaction that enhances the wipeout effect over time.
Questions remain though. Perhaps the most immediate question is whether there are electrons within the gapped energy spectrum of PuB that would allow the indirect RKKY-type hyperfine interaction. Although there is no concrete answer to this question, our conjecture is supported by the proposition of in-gap states in PuB suggested earlier by transport and heat capacity[15] as well as NMR [16] measurements at temperatures below the energy scale of the gap (below 100 K). Certainly, in-gap states have been widely observed in a group of related narrow-gap Kondo insulators, such as SmB and YbB.[56, 57, 58] In these, impurities and defects play a role in producing such low-energy states as well as in their bulk and surface electronic properties.[59, 60]
We speculate that radioactive decay in PuB not only generates magnetic radicals but also metallic impurity states that mediate the indirect hyperfine interaction. The effects of aging in this context could be understood to appear as a shift of the fluctuation spectrum towards lower frequency that turns more and more of the sample into the wipeout regime. Similar aging effects have been previously observed in AmO by ligand O NMR.[8] Further systematic investigations with X-ray diffraction and radiochemical analysis would be useful to improve our understanding of the complex mechanism of radiation damage in this Pu-based material.
Conclusions
We have used Pu-NMR to spectroscopically characterize consequences of aging on static and dynamic properties of single-crystal PuB. Field-angle dependent measurements imply an effective tetragonal site symmetry of Pu atoms despite their sitting in an orthorhombic crystalline environment. The Pu resonance spectrum of aged PuB reveals the existence of multiple, slightly distinct Pu sites possessing very similar axial local-site symmetry. Aging has little effect on the static NMR spectrum, whereas it shows up more directly and strongly in the spectral intensity as well as in the dynamic spin-lattice relaxation () and spin-spin relaxation () profiles. Aging drives to longer values, increases its distribution across the sample, and reduces the spectral intensity by 20 % per year. There is a strong variation in over the spectrum and a notable shorting of below about 40 K, albeit with no appreciable change in . Together, these point to the development of slow fluctuations that could be responsible for wipeout of the spectral signal over the course of years. Isochronal annealing at room temperature only partially heals the aging damage in the first year and has little effect after a second year of accumulated damage, suggesting the presence of a threshold amount of damage that is healable. This work demonstrates that the actinide resonance spectroscopy can be used as a uniquely sensitive probe to study radiation damage in materials.
Acknowledgements
We thank H. Yasuoka, H. Mason, A. Altenhof, and J. Thompson for fruitful discussions and comments on aging effects and relaxation rate analyses as well as on the manuscript. The dynamic nuclear spin-lattice relaxation measurements were performed under the auspices of the U.S. Department of Energy, Office of Basic Energy Sciences, Division of Materials Science and Engineering. The spectral measurements including their rotation and aging studies were supported from the Laboratory Directed Research and Development programs and LANL Office of Experimental Sciences. We additionally thank the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences, Heavy Element Chemistry program (2020LANLE372) for funding (SAK).
Determination of temperature dependence of from intensity data
Figure S1 shows the -NMR spin-echo decay curve recorded for at base temperature (4.2 K) in a sample after 690 days from its synthesis. In Figure S3 we present the temperature dependence of the full-spectral NMR signal intensity recorded by using a fixed waiting time (= 100 s) between and pulses in the spin echo with a last delay at least ten times longer than . A nearly temperature independent is seen at higher temperature, which is expected if the number of the nuclear spins excited by the NMR pulses is preserved. With decreasing temperature, shows a strong decrease and at base temperature is only about 50 % of its higher temperature value. This clear intensity drop is not a signal wipeout: after a proper correction, based on the spin-echo decay curve in Figure S2, the value of perfectly matches the high temperature constant value. As PuB is a modestly good insulator and will not be accompanied by a skin-depth effect, this intensity decrease should represent a shortening of . Although we don’t have enough data at various temperatures, this intensity data allows us to estimate the value of , dubbed hereafter, at different temperatures by assuming that the functional form of the spin-echo decay curve [Eq. (2) in the main text] does not differ from that at 4.2 K (Figure S2):
where is the intensity at a given temperature, is the -corrected constant intensity at 4.2 K, , and is the exponent determined at 4.2 K.
Fitting forms to the temperature dependence of and
Ref. [16] reported identical results for [ data in Figure 3(a)] and [ data in Figure 5(a)] using other fitting functions and obtained bigger gap sizes of meV and meV. That work assumed a model density of states, following Caldwell et al.[57] for their analysis of the B-NMR results in SmB, and employed the so-called Korringa relation[17, 30] that connects the NMR Knight shift and via electronic density of states at the Fermi level . This analysis explicitly assumes that the deduced gap in the NMR measurements reflects the size of the band gap around (as NMR in this case probes the number thermally excited electron-hole pairs generated across the band gap). Moreover, it relies on multiple fitting parameters that control the shape of the model density of states used in that fit. Although that analysis allowed the authors to make a direct quantitative comparison of the gap sizes with those reported in SmB, it is not obvious whether a simple semiconductor-like picture is relevant to PuB, where the itinerant versus localized picture of 5f-electrons is not well understood. To avoid unwanted complications, we instead have taken a more general approach and used commonly employed exponential forms for spin-gapped systems[18] in the analysis of the spin gaps in Figure 3 and Figure 5. Despite the different approaches used to extract the gaps, the ratio of the two gap, , is almost identical, within estimated error, to that reported in Ref. [16]. This common ratio supports the validity of our approach.
References
- [1]W. G. Wolfer, Radiation Effects in Plutonium, Los Alamos Sci. 26, 274 (2000).
- [2]C. H. Booth, Yu Jiang, S. A. Medling, D. L. Wang, A. L. Costello, D. S. Schwartz, J. N. Mitchell, P. H. Tobash, E. D. Bauer, S. K. McCall, M. A. Wall, P. G. Allen, Self-irradiation damage to the local structure of plutonium and plutonium intermetallics, J. Appl. Phys. 113, 093502 (2013).DOI
- [3]N. Curro and L. Morales, Nuclear Magnetic Resonance Studies of δ-Stabilized Plutonium, Mat. Res. Soc. Symp. Proc. 802, 53 (2004).DOI
- [4]Yu. Piskunov, K. Mikhalev, A. Gerashenko, A. Pogudin, V. Ogloblichev, S. Verkhovskii, A. Tankeyev, and V. Arkhipov, Spin susceptibility of Ga-stabilized d-Pu probed by 69Ga NMR, Phys. Rev. B 71, 174410 (2005).DOI
- [5]S. Verkhovskii, Yu. Piskunov, K. Mikhalev, A. Buzlukov, V. Arkhipov, Yu. Zouev, A. Korolev, S. Lekomtsev, I. Svyatov, A. Pogudin, and V. Ogloblichev, Magnetic state of f electrons in δ-phase of Pu–Ga alloys studied by Ga NMR, J. Alloys Compd. 444–445, 288 (2007).DOI
- [6]Yu. Piskunov, K. Mikhalev, Yu. Zuev, S. Verkhovskii, V. Arkhipov, I. Svyatov, I. Podgornova, A. Shestakov, V. Ogloblichev, A. Pogudin, and A. Buzlukov, 69Ga NMR in Pu1−xGax (x < 0.01) alloy, J. Alloys Compd. 444–445, 325 (2007).DOI
- [7]Yu. Piskunov, K. Mikhalev, A. Buzlukov, A. Gerashenko, S. Verkhovskii, V. Ogloblichev, V. Arkhipov, A. Korolev, Yu. Zouev, and I. Svyatov, J. Nucl. Mater. 385, 25 (2009).
- [8]Y. Tokunaga, T. Nishi, M. Nakada, A. Itoh, H. Sakai, S. Kambe, Y. Homma, F. Honda, D. Aoki, and R. E. Walstedt, Self-radiation effects and glassy nature of magnetic transition in AmO₂ revealed by ¹⁷O-NMR, Phys. Rev. B 89, 214416 (2014).DOI
- [9]I. Farnan, H. Cho, and W. J. Weber, Quantification of actinide α-radiation damage in minerals and ceramics, Nature 445, 190 (2007).
- [10]K. Ikushima, H. Yasuoka, S. Tsutsui, M. Saeki, S. Nasu, M. Date, Observation of ²³⁵U NMR in the Antiferromagnetic State of UO₂, J. Phys. Soc. Jpn. 67, 65-66 (1998).DOI
- [11]K. Ikushima, S. Tsutsui, Y. Haga, H. Yasuoka, R. E. Walstedt, N. M. Masaki, A. Nakamura, S. Nasu, and Y. Ōnuki, First-order phase transition in UO₂: ²³⁵U and ¹⁷O NMR study, Phys. Rev. B 63, 104404 (2001).DOI
- [12]H. Kato, H. Sakai, K. Ikushima, S. Kambe, Y. Tokunaga, D. Aoki, Y. Haga, Y. Onuki, H. Yasuoka, and R. E. Walstedt, ²³⁵U NMR study of the itinerant antiferromagnet USb₂, Physica B 359–361, 1012 (2005).DOI
- [13]H. Le Bail, C. Chachaty, P. Rigny, and R. Bougon, Observation of the uranium ²³⁵ nuclear magnetic resonance signal, J. Physique Lett. 44, 1017 (1983).
- [14]H. Yasuoka, G. Koutroulakis, H. Chudo, S. Richmond, D. K. Veirs, A. I. Smith, E. D. Bauer, J. D. Thompson, G. D. Jarvinen, and D. L. Clark, Observation of ²³⁹Pu Nuclear Magnetic Resonance, Science 336, 901 (2012).DOI
- [15]H. Choi, W. Zhu, S. K. Cary, L. E. Winter, Z. Huang, R. D. McDonald, V. Mocko, B. L. Scott, P. H. Tobash, J. D. Thompson, S. A. Kozimor, E. D. Bauer, Jian-Xin Zhu, and F. Ronning, Experimental and theoretical study of topology and electronic correlations in PuB₄, Phys. Rev. B 97, 201114 (2018).DOI
- [16]A. P. Dioguardi, H. Yasuoka, S. M. Thomas, H. Sakai, S. K. Cary, S. A. Kozimor, T. E. Albrecht-Schmitt, H. C. Choi, J.-X. Zhu, J. D. Thompson, E. D. Bauer, and F. Ronning, ²³⁹Pu nuclear magnetic resonance in the candidate topological insulator PuB₄, Phys. Rev. B 99, 035104 (2019).
- [17]A. Abraham, Principles of Nuclear Magnetism, Oxford University Press, New York (1961).
- [18]M. Itoh and H. Yasuoka, Interrelation between Dynamical and Static Spin Gaps in Quantum Spin Systems, J. Phys. Soc. Jpn. 66, 334-336 (1997).DOI
- [19]N. J. Curro, P. C. Hammel, B. J. Suh, M. Hücker, B. Büchner, U. Ammerahl, and A. Revcolevschi, Inhomogeneous Low Frequency Spin Dynamics in La₁.₆₅Eu₀.₂Sr₀.₁₅CuO₄, Phys. Rev. Lett. 85, 642 (2000).pubmed.ncbi.nlm.nih.gov/10991360
- [20]V. F. Mitrović, M.-H. Julien, C. de Vaulx, M. Horvatić, C. Berthier, T. Suzuki, and K. Yamada, Similar glassy features in the ¹³⁹La NMR response of pure and disordered La₁.₈₈Sr₀.₁₂CuO₄, Phys. Rev. B 78, 014504 (2008).DOI
- [21]P. M. Singer, A. Arsenault, T. Imai, and M. Fujita, ¹³⁹La NMR investigation of the interplay between lattice, charge, and spin dynamics in the charge-ordered high-Tc cuprate La₁.₈₇₅Ba₀.₁₂₅CuO₄, Phys. Rev. B 101, 174508 (2020).DOI
- [22]A. P. Dioguardi, M. M. Lawson, B. T. Bush, J. Crocker, K. R. Shirer, D. M. Nisson, T. Kissikov, S. Ran, S. L. Bud'ko, P. C. Canfield, S. Yuan, P. L. Kuhns, A. P. Reyes, H.-J. Grafe, and N. J. Curro, NMR evidence for inhomogeneous glassy behavior driven by nematic fluctuations in iron arsenide superconductors, Phys. Rev. B 92, 165116 (2015).DOI
- [23]D. C. Johnston, S.-H. Baek, X. Zong, F. Borsa, J. Schmalian, and S. Kondo, Dynamics of Magnetic Defects in Heavy Fermion LiV₂O₄ from Stretched Exponential ⁷Li NMR Relaxation, Phys. Rev. Lett. 95, 176408 (2005).DOI
- [24]D. C. Johnston, Stretched exponential relaxation arising from a continuous sum of exponential decays, Phys. Rev. B 74, 184430 (2006).DOI
- [25]D. L. Clark, S. S. Hecker, G. D. Jarvinen, and M. P. Neu, Chapter 7, Plutonium, The Chemistry of the Actinide and Transactinide Elements (Springer, Dordrecht, 2006), pp. 813-1264.
- [26]M. Takano, M. Akabori, and Y. Arai, Annealing behavior of (Pu,Cm)O₂ lattice and bulk expansion from self-irradiation damage, J. Nucl. Mater. 414, 174 (2011).DOI
- [27]W. Schilling and K. Sonnenberg, Recovery of irradiated and quenched metals, J. Phys. F: Met. Phys. 3, 322 (1973).DOI
- [28]M. Horvatic and C. Berthier, NMR Studies of Low-Dimensional Quantum Antiferromagnets, in High Magnetic Fields: Applications in Condensed Matter Physics and Spectroscopy, C. Berthier, L. P. Lévy, and G. Martinez, Springer, Berlin, p. 191-21 (2002)
- [29]R. E. Walstedt, Spin-lattice relaxation of nuclear spin echos in metals, Phys. Rev. Lett. 19, 146 (1967); ibid 19, 816 (1967).
- [30]C. Slichter, Principles of Magnetic Resonance, Springer, Berlin (1978).
- [31]M. Takigawa and G. Saito, Evidence for the Slow Motion of Organic Molecules in (TMTSF)₂ClO₄ Extracted from the Spin Echo Decay of ⁷⁷Se NMR, J. Phys. Soc. Jpn. 55, 1233-1243 (1986).
- [32]D. E. MacLaughlin and H. Alloul, Host Nuclear Resonance in a Spin-Glass: CuMn, Phys. Rev. Lett. 36, 1158 (1976).
- [33]D. A. Levitt and R. E. Walstedt, Comments on “Host Nuclear Resonance in a Spin-Glass: CuMn”, Phys. Rev. Lett. 38, 178 (1977).
- [34]M. C. Chen and C. P. Slichter, Zero-field NMR study on a spin-glass: Iron-doped 2H—niobium diselenide, Phys. Rev. B 27, 278, (1983).DOI
- [35]R. Chiba, K. Hiraki, T. Takahashi, H. M. Yamamoto, and T. Nakamura, Extremely Slow Charge Fluctuations in the Metallic State of the Two-Dimensional Molecular Conductor θ–(BEDT–TTF)₂RbZn(SCN)₄, Phys. Rev. Lett. 93, 216405(1) (2004).pubmed.ncbi.nlm.nih.gov/15601040
- [36]T. Le, A. Pustogow, J. Wang, A. Henderson, T. Siegrist, J. A. Schlueter, and S. E. Brown, Disorder and slowing magnetic dynamics in κ-(BEDT–TTF)₂Hg(SCN)₂Br, Phys. Rev. B 102, 184417 (2020).DOI
- [37]C. León, M. L. Lucía, J. Santamaría, M. A. París, J. Sanz, and A. Várez, Electrical conductivity relaxation and nuclear magnetic resonance of Li₀.₅La₀.₅TiO₃, Phys. Rev. B 54, 184 (1996).pubmed.ncbi.nlm.nih.gov/9984245
- [38]D. Opherden, F. Bärtl, Sh. Yamamoto, Z. T. Zhang, S. Luther, S. Molatta, J. Wosnitza, M. Baenitz, I. Heinmaa, R. Stern, C. P. Landee, and H. Kühne, Freezing of molecular rotation in a paramagnetic crystal studied by ³¹P NMR, Phys. Rev. B 103, 014428 (2021).
- [39]A. W. Hunt, P. M. Singer, A. F. Cederström, and T. Imai, Glassy slowing of stripe modulation in (La,Eu,Nd)₂₋ₓ(Sr,Ba)ₓCuO₄: A ⁶³Cu and ¹³⁹La NQR study down to 350 mK, Phys. Rev. B 64, 134525 (2001).DOI
- [40]T. Wu, H. Mayaffre, S. Krämer, M. Horvatić, C. Berthier, W. N. Hardy, R. Liang, D. A. Bonn, and M.-H. Julien, Magnetic-field-induced charge-stripe order in the high-temperature superconductor YBa₂Cu₃Oᵧ, Nature 477, 191 (2011).DOI
- [41]S. Fujiyama, M. Takigawa, and S. Horii, Charge Freezing in the Zigzag Chain PrBa₂Cu₄O₈ Cuprate, Phys. Rev. Lett. 90, 147004 (2003).pubmed.ncbi.nlm.nih.gov/12731941
- [42]P. Carretta, R. Melzi, N. Papinutto, and P. Millet, Very-Low-Frequency Excitations in Frustrated Two-Dimensional S = 1/2 Heisenberg Antiferromagnets, Phys. Rev. Lett. 88, 047601 (2002).DOI
- [43]Y. Tokunaga, D. Aoki, H. Mayaffre, S. Krämer, M.-H. Julien, C. Berthier, M. Horvatić, H. Sakai, S. Kambe, and S. Araki, Reentrant Superconductivity Driven by Quantum Tricritical Fluctuations in URhGe: Evidence from ⁵⁹Co NMR in URh₀.₉Co₀.₁Ge, Phys. Rev. Lett. 114, 216401 (2015).pubmed.ncbi.nlm.nih.gov/26066447
- [44]T. Itou, E. Watanabe, S. Maegawa, A. Tajima, N. Tajima, K. Kubo, R. Kato, and K. Kanoda, Slow dynamics of electrons at a metal–Mott insulator boundary in an organic system with disorder, Sci. Adv. 3:e1601594 (2017).DOI
- [45]R. Yamamoto, T. Furukawa, K. Miyagawa, T. Sasaki, K. Kanoda, and T. Itou, Electronic Griffiths Phase in Disordered Mott-Transition Systems, Phys. Rev. Lett. 124, 046404 (2020).DOI
- [46]H. Xiao, T. Hu, A. P. Dioguardi, N. apRoberts-Warren, A. C. Shockley, J. Crocker, D. M. Nisson, Z. Viskadourakis, Xianyang Tee, I. Radulov, C. C. Almasan, N. J. Curro, and C. Panagopoulos, Evidence for filamentary superconductivity nucleated at antiphase domain walls in antiferromagnetic CaFe₂As₂, Phys. Rev. B 85, 024530 (2012).DOI
- [47]L. Bossoni, P. Carretta, W. P. Halperin, S. Oh, A. Reyes, P. Kuhns, and P. C. Canfield, Evidence of unconventional low-frequency dynamics in the normal phase of Ba(Fe₁−xRhₓ)₂As₂ iron-based superconductors, Phys. Rev. B 88, 100503(R) (2013).
- [48]T. F. Wall, S. J. Ian, M. Autillo, K. L. Nash, L. Guerin, C. Le Naour, P. Moisy, and C. Berthon, Paramagnetism of Aqueous Actinide Cations. Part I: Perchloric Acid Media, Inorg. Chem. 53, 2450–2459 (2014).DOI
- [49]J. A. Shusterman, H. E. Mason, J. Bowers, A. Bruchet, E. C. Uribe, A. B. Kersting, and H. Nitsche, Development and Testing of Diglycolamide Functionalized Mesoporous Silica for Sorption of Trivalent Actinides and Lanthanides, Appl. Mater. Interfaces 7, 20591–20599 (2015).DOI
- [50]H. Alloul, Host NMR in dilute magnetic alloys—spatial variation of spin polarization, J. Phys. F: Met. Phys. 4 1501 (1974).DOI
- [51]R. E. Levine, Nuclear Spin Lattice Relaxation in CuMn, Phys. Lett. 28A, 504 (1969).DOI
- [52]J. M. Brettell and A. J. Heeger, Nuclear Magnetic Resonance in Dilute Alloys of Mn, Fe, and Cu in Aluminum, Phys. Rev. 153, 319 (1967).DOI
- [53]N. Karnezos and J. A. Gardner, NMR Study of Near Neighbor Wipeout and Satellites in Cu Mn, AIP Conf. Proc. 10, 801 (1973).DOI
- [54]B. Giovannini, P. Pincus, G. Gladstone, and A. J. Heeger, Nuclear Relaxation in Dilute Magnetic Alloys, J. Phys. Colloques 32, C1-163-C1-171 (1971).DOI
- [55]F. Y. Fradin, Influence of Localized Rare-Earth Moments on ⁴⁵Sc Nuclear Magnetic Relaxation, Phys. Rev. Lett. 26, 1033 (1971).DOI
- [56]W. T. Fuhrman and P. Nikolić, Magnetic impurities in Kondo insulators: An application to samarium hexaboride, Phys. Rev. B 101, 245118 (2020).DOI
- [57]T. Caldwell, A. P. Reyes, W. G. Moulton, P. L. Kuhns, M. J. R. Hoch, P. Schlottmann, and Z. Fisk, High-field suppression of in-gap states in the Kondo insulator SmB₆, Phys. Rev. B 75, 075106 (2007).DOI
- [58]P. Schlottmann, Impurity bands in Kondo insulators, Phys. Rev. B 46, 998 (1992).pubmed.ncbi.nlm.nih.gov/10003287
- [59]N. Shishiuchi, Y. Kato, L. M. Vyaselev, M. Takigawa, S. Hiura, F. Iga, and T. Takabatake, Defect-induced magnetic fluctuations in YbB₁₂, J. Phys. Chem. Solids. 63, 1231 (2002).
- [60]W. Matsuura, N. Takahashi, K. Yamada, M. Kadowaki, and F. Iga, Development of In-Gap State and Related Collapse of Two-Step Energy Gap Structure by Nonmagnetic Ion Substitution in Kondo Insulator YbB₁₂, J. Phys. Soc. Jpn.90, 124705 (2021).