Abstract
Quantum states cohere and interfere. Atoms arranged imperfectly in a solid rarely display these properties. Here we demonstrate an exception in a disordered quantum magnet that divides itself into nearly isolated subsystems. We probe these coherent spin clusters by driving the system nonlinearly and measuring the resulting hole in the linear spectral response. The Fano shape of the hole encodes the incoherent lifetime as well as coherent mixing of the localized excitations. For the Ising magnet LiHo_{0.045}Y_{0.955}F_{4}, the quality factor Q for spectral holes can be as high as 100,000. We tune the dynamics by sweeping the Fano mixing parameter q through zero via the ac pump amplitude as well as a dc transverse field. The zero crossing of q is associated with a dissipationless response at the drive frequency. Identifying localized twolevel systems in a dense and disordered magnet advances the search for qubit platforms emerging from strongly interacting, manybody systems.
Similar content being viewed by others
Introduction
Localization in quantum systems remains both fundamental to science as well as to technology. It is a venerable subject, starting with the work of Anderson^{1}—whose name is associated with disorderinduced localization—and Mott^{2}, whose Mott localization transition is due to repulsion between electrons. Since these pioneering studies, similar concepts have been extended beyond electrons to a wide range of interacting quantum systems. The combined problem of manybody localization (MBL)^{3,4,5,6,7,8,9,10} persists to this day and has inspired recent numerical simulations^{11,12} as well as experiments on cold atoms^{13,14,15}. There is also practical relevance for systems of quantum devices; a notable example is the DWave processor^{16,17}, which attempts to implement adiabatic quantum computation, but may be limited as a matter of principle by localization effects.
To control these longlived and independent states, it is necessary to know how they interact with each other and with the outside world. Minimizing the interactions between coherent localized states and the continuum of states in the broader environment is an important goal for realizing an effective quantum computer^{18,19,20,21}. However, these environmental couplings are by definition weak compared with the transitions among the states contributing to the spectrum for a particular localizing environment, making it difficult to study them directly. Recently, it has been posited that many weak couplings of this sort can be probed by pumping the system into a nonlinear response regime^{22}, saturating the discrete transition associated with the coherent state, and resulting in spectral holes. In the present work, we use the Fano lineshapes of the spectral holes to characterize the coupling between the localized subsystems and both external and internal fields. Our work differs from experiments in quantum optics^{23} in that we are examining emergent degrees of freedom in a (magnetic) manybody system rather than weakly coupled (to each other) singleparticle states in atoms and semiconductor quantum dots. At the same time, it represents a major advance over our own previous activity on pumpinduced Fano resonances in the same magnetic material^{24,25} in that we uncover a remarkably simple phenomenology, including the discovery of a zero crossing for the Fano asymmetry parameter q, as a function of nonlinear drive amplitude and quantum mixing via a transverse field.
Asymmetric absorption lineshapes in atomic gases were first addressed by Ugo Fano more than 50 years ago^{26}. They arise from interference between discrete transitions in the atoms and ionization into the continuum. Now known as Fano resonances, this formalism has found wide applicability in systems ranging from photoionized gases^{27} to highT_{c} superconductors^{28} and photonic crystals^{29} to quantum wells^{30}. Their extension to spin liquids provides a means to characterize coherent spin clusters labeled in the time/frequency domain but distributed randomly in space.
The remainder of the paper is structured as follows. We begin by presenting an overview of the experimental system, the dilute Ising magnet LiHo_{x}Y_{1 − x}F_{4}, focusing on the linear and nonlinear susceptibilities at small x. We then describe a pumpprobe technique to study these clusters, using the detailed lineshape of the observed Fano resonances to track the evolution of the clusters as a function of external variables such as temperature and applied magnetic field. Most notably, we find that the interaction between the clusters and the incoherent bath of free spins can be tuned continuously to a point where the direct absorption of energy, and hence the dissipation, of the clusters goes to zero. We then discuss the implications of tuning the system into a dissipationless regime coincident with zero crossings of the Fano asymmetry parameter q and how these results offer a potential avenue for accessing manybodylocalization physics and the possible followup experiments that could explore such connections.
Results
Review of the experimental system LiHo_{x}Y_{1 − x}F_{4}
Spin clusters functioning as quantum twolevel systems form in the dilute dipolarcoupled magnet LiHo_{x}Y_{1 − x}F_{4} under the appropriate thermodynamic conditions^{25,31}. The magnetism in this family of rareearth fluorides has long been studied as a realization of the dipolecoupled S = 1/2 Ising model, with the spins carried by the Ho^{3+} ions and the Ising axis lying along the crystallographic c axis^{32,33}. The nonmagnetic Y^{3+} ions randomly occupy the same sites as the magnetic Ho^{3+} ions with probability 1− x. The hierarchy of quantum levels accounting for the chargeneutral excitations of individual Ho^{3+} ions (in the dilute limit where x ≪ 1) of this widegap insulator has been summarized recently by Matmon et al.^{34}. Most relevant for the current lowtemperature study are the groundstate doublet for the Ising spins with a crystalfieldderived 9.4 K gap to the first excited state and the hyperfine interaction between the electronic (J = 8) and nuclear (I = 7/2) spins of the Ho^{3+} ions^{32}, which yields a nuclear Zeeman ladder consisting of eight states with spacing 0.2 K between consecutive levels. Analytic solutions of the microscopic Hamiltonian^{35,36,37,38}, combined with measurements of the crystalfield parameters^{32,39}, have quantitatively connected the microscopic Hamiltonian to the longstanding effective Hamiltonian for the spin physics. Magnetic fields applied perpendicular to the Ising axis mix the groundstate doublet with the first excited state, inducing a splitting of the doublet, which in the lowfield limit scales as \({\mathrm{\Gamma }} \propto {{H}}_{\mathrm{t}}^2\) (in contrast to the Zeeman splitting \(\propto {{H}}_1\) of the ground state in a longitudinal field H_{l}) and leads to an effective transversefield Ising Hamiltonian^{40,41}:
In the pure (x = 1) limit, a classical ferromagnetic transition occurs at the Curie point T_{C} = 1.53 K dictated by the magnitude of the (predominantly) dipolar interactions J_{ij}. The quantum fluctuations induced by the transverse field disorder the Ising spins, producing a quantum critical point at Γ = 1.6 K where the Ising ferromagnetism vanishes. This zerotemperature transition is linked by a line of secondorder transitions between paramagnetic and ferromagnetic phases to zero field. The principal features of the dynamics are propagating soft magnetic modes^{42}.
Although Eq. (1), which takes no account of the nuclear spins, is an excellent starting point for understanding the physics of pure LiHoF_{4}, at temperatures below ~0.6 K the electronic and nuclear spins of the Ho^{3+} ions^{32} combine to form composite degrees of freedom with effective spins I + J, resulting in an upturn in the ferromagnet–paramagnet phase boundary for pure LiHoF_{4}^{41}. Furthermore, entanglement of the nuclear and electron spins results in an incomplete softening of the principal magnetic excitation mode at the quantum phase transition^{39,42}.
Additional physics has been revealed at holmium concentrations between x = 1 for the pure ferromagnet and the x ≪ 1 dilute ion limit^{43}. The combination of disorder, the magnetic dipole interaction, which can be ferromagnetic or antiferromagnetic (depending on the relative orientation of the preferred spin direction and the vector separating two spins), and quantum fluctuations^{44,45,46} creates a sequence of states as a function of decreasing dipole (Ho) concentration^{43}, progressing from meanfield ferromagnet^{41} to randomfield ferromagnet^{47}, to spin glass^{40,48,49,50} to spin liquid^{31}. The loss of translational symmetry arising from the dilution creates a local loss of rotational symmetry, which generates internal transverse fields. The quantum fluctuations that arise from these internal transverse fields can act to prevent complete freezing in the T → 0 limit even without an externally applied transverse field^{51}. Here we focus on the Bhatt–Lee spin liquid^{52} (originally proposed for phosphorusdoped silicon near its metalinsulator transition) for x ~ 0.05 and weak thermal coupling, characterized by a hierarchy of singlets derived from combinations of doublets (isolated spins) and triplets (ferromagnetically coupled spins). The hierarchy of singlets results in a lowfrequency susceptibility which scales as 1/T^{α} with α ≠ 1 rather than following Curie or spinglass forms. For LiHo_{0.045}Y_{0.955}F_{4}, α was experimentally measured to be 0.75, less than the Curie exponent α = 1.0 due to the compensation of spins by each other as they form singlets on cooling^{51}. The nature of the ground state for LiHo_{0.045}Y_{0.955}F_{4} has been a matter of debate, with some groups reporting a spinglass state^{49,53} and other work indicating a spinliquid state^{31,43}. The discrepancy was resolved in ref. ^{25}, which showed that by varying the strength of thermal coupling between the crystal and an external heat bath, it was possible to tune between the two limits. This controlled the rate at which lattice phonons and, subsequently, spins via spinlattice coupling could exchange energy with the external bath. As discussed below, the measurements reported here were all obtained in the weak thermalcoupling limit, which favors the quantum spin liquid.
Figure 1a shows schematically how a disordered quantum spin system such as LiHo_{x}Y_{1 − x}F_{4} at low concentrations or Si:P^{52,54} breaks into decoupled clusters and isolated spins, focusing on some clusters, whose classical ground states are ferromagnetic but which, because of small transverse fields imposed by other clusters, can be described as twolevel systems with lowenergy eigenstates that are coherent superpositions of up and down configurations ⇑ > ± ⇓ > = ↑…↑〉 ± ↓↓…↓〉^{31}. More elaborate wavefunctions and level schemes^{42,55}, taking account of, e.g., electronuclear spin mixing and also classical ground states with antiferromagnetic correlations, will not qualitatively change the physics of hole burning (see Supplementary Note 1) but will be invoked later when we discuss certain details of our data.
Nonlinear susceptibility of LiHo_{0.045}Y_{0.955}F_{4}
For LiHo_{0.045}Y_{0.955}F_{4}, both the thermal link to the heat bath^{25} and magnetic fields applied transverse to the Ising axis^{24} control the observability of localized excited states. In particular, strong coupling to a thermal bath yields the linear response expected from a spin glass^{25,49}, no observable localized excited states, and consequently no evidence for the Fano resonance is seen for strong coupling (Fig. 1d). We therefore concentrate here on weak coupling to the bath.
The first step in the measurements is to determine pump fields h_{pump} sufficient to enter the nonlinear regime. Figure 1c shows the magnitude and phase angle of the magnetization induced for a range of frequencies f_{pump} limited to a maximum of 10 Hz, to allow measurement to drive fields above 1 Oe without excessive eddycurrent heating. The data generally follow an Sshaped curve, whose inflection point moves outward with increasing frequency, in agreement with the model (Supplementary Eq. 1) where the longitudinal field couples the localized singlet ground state of a cluster to an excited state separated by a small gap^{51}; the susceptibility at higher frequencies is more sensitive to smaller clusters where the gap is larger, which can only be overcome by a higher drive field h_{pump}. For our pumpprobe experiments, we operate near to and somewhat above the inflection point; very high driving fields are avoided, because they lead to excessive heating. The effective increase in energy of the spin system can be characterized by measuring the shift in the peak frequency of χ″(f) as a result of nonlinear drive and comparing with linearresponse measurements at a range of temperatures. As reported in ref. ^{31}, a 5 Hz, 0.2 Oe pump provided energy equivalent to a 40 mK increase in temperature. This can be distinguished from purely thermal effects associated with the drive field by noting the clear change in the dissipative lineshape: the distribution of oscillators visible in χ″( f ) lacks the tails seen at equilibrium (in the absence of pumping). Although the system consists of an ensemble of spin clusters with a broad range of sizes (previous magnetization measurements found clusters ~250 spins at a 5 Hz resonant frequency), choosing a particular pump frequency selects for a set of clusters that are similarly sized and hence sorted by resonant frequency. These clusters are largely independent of each other and hence pumping at two wellseparated frequencies simultaneously results in two distinct holes in the overall spectrum^{31}.
Pumpprobe spectroscopy
We turn now to pumpprobe measurements, examples of which are shown in Fig. 2a. In these measurements, the system is driven with a strong ac pump field at frequency f_{0} and probed with a weak field at a range of frequencies Δf = f_{probe} − f_{0} around f_{0} (see Methods for details). The characteristic asymmetric shape of a Fano resonance is immediately obvious, providing a direct indication of a weak coupling between the coherent spin cluster and the continuum of surrounding spin states (Fig. 1b, c). By comparing the integrated area of the resonant response with the area of the entire linearresponse spectrum, we estimate that the fraction of spins bound in clusters resonant at the chosen drive frequency is of order 2 × 10^{−6} of the entire sample. As the temperature is increased, the amplitude of the resonant response drops and the resonance appears to broaden, with the response suppressed to 8% of its original amplitude at T = 500 mK and to below the noise floor of the measurement at 700 mK. Given that the overall linear susceptibility of LiHo_{0.045}Y_{0.955}F_{4} has a strong temperature dependence, the thermal evolution of the resonant response can be seen more clearly by normalizing it to the linear response at each temperature (determined by measuring χ″(Δf = 30 mHz)). We show in Fig. 2b spectra obtained at a series of temperatures, normalized, and then combined into a surface plot where color and height now represent the absorption for a given Δf and T. The broadening of the resonance with increasing temperature emerges clearly in this visualization and we examine it quantitatively in Fig. 2c by looking at the evolution of the linewidth in the fits to the Fano form,
where Γ is the linewidth of the resonance and q, known as the Fano parameter, characterizes the interference between the different transition pathways. The mHz scale lowtemperature limit of the linewidth suggests that the coupling between the clusters and the background spin bath is weak, and hence that the system can be considered in the framework of a twolevel system in weak contact with the environment rather than a continuous relaxation process. Quantum states with splittings substantially smaller than nominal bath temperatures are very common in solids and liquids, and indeed form the basis for various resonance (e.g., nuclear magnetic resonance (NMR)) spectroscopies, many of which rely on nonequilibrium quantumstate preparation. Reduced bath coupling during cooldown increases the T_{1} and T_{2} times associated with such quantum states, and so makes a description of the magnetic response of the system as due to a set of independent multilevel quantum systems more appropriate than a picture based on classical, thermal diffusion.
When a multilevel quantum description applies, for a fixed bath coupling (which extracts heat) and ac driving field (which inserts heat), an equilibrium with a set of fixed state occupancies will characterize the system, and to first order that equilibrium can be described by a fixed effective temperature. On the other hand, the nonequilibrium dynamics are dominated by small multilevel systems describable in terms of some generalized Bloch equation, exactly as is the case, e.g., for NMR performed even at room temperature, and are therefore quantum mechanical. The linewidth increases exponentially with T, consistent with a thermally activated process with a gap Δ = 740 mK, which is an energy well below the 9.4 K first excited crystalfield state energy but of the same order as nearestneighbor spin couplings as well as the energy difference^{34,41,56} (~750 mK) between electronuclear states with nuclear moments of 7/2 and 1/2. This suggests that hole burning is favored when Ho nuclear spins are not relaxing relative to electron spins, but does not exclude composite electronuclear wavefunctions involving several Ho ions. Indeed, the fact that several ions need to be involved can be deduced from comparison of the crossover fields (all below 1 Oe) in our ac drive data (Fig. 1c) with the step locations at multiples of 200 Oe in the magnetometry of Giraud et al.^{55}. The greatly increased density of level crossings suggested by our data follows naturally from the dipolar interactions between multiple Ho ions. Further indications that the interactions are important come from measurements where changing the thermal boundary conditions vary the state of the system^{25}. For strong coupling to an external bath (the mixing chamber of the cryostat), collective spin glass behavior obtains and the nonlinear response also changes qualitatively with no trend towards saturation. In other words, the resonances are not due to singlespin behavior but instead are consistent with a picture consisting of clusters of many spins behaving as a large effective twolevel system protected from others in a wavefunction such as that ⇑> ± ⇓> = ↑↑…↑〉 ± ↓↓…↓〉 (discussed above) at low temperatures because of a relatively large single spinflip energy, associated with either the hyperfine or electronic dipole interactions. In addition to the thermal broadening of the resonance, the lineshape asymmetry q(T) is also Tdependent, varying in approximately linear fashion for T < 0.25 K and plateauing above that point (Fig. 2d). A possible microscopic origin for this behavior is that as the temperature grows, thermally activated spin flips will occur within clusters containing antiferromagnetically coupled spins. Such spin flips will result in increased dipolar moments for the clusters and thus an increased coupling to remnant dipole moments of the other clusters forming the underlying bath, an effect seen in the temperaturedependent linewidth shown in Fig. 2c. Provided that a single constant J_{AFM} characterizes the underlying antiferromagnetic coupling, we would find that for k_{B}T ≫ J_{AFM} the probabilities that two spins are either ferromagnetically or antiferromagnetically correlated become equal and we would see a plateau in the coupling to other clusters as well as the value q. Such plateaus also can be found if there is a discrete series of antiferromagnetic couplings J_{AFM,i} and the conditions J_{AFM,i} ≪ k_{B}T ≪ J_{AFM,i + 1} are met. The discrete nature of the distribution of dipolar couplings for the LiHo_{x}Y_{1 − x}F_{4} lattice leads to the possibility that these conditions obtain and therefore the data in Fig. 2d, which extend only as far as hole burning can be seen (and q can be measured), could be a manifestation of such a plateau.
It should be noted that the freeinduction relaxation time of ~10–30 s observed previously^{31} is substantially shorter than the ~500–1000 s of the inverse linewidth of the hole uncovered in the driven pumpprobe measurements. This follows because the freeinduction decay was measured for relaxation after the strong ac drive field was turned off, whereas the linewidth here is measured in the far more weakly driven linear regime. More formally, the rotating wave approximation^{57} does not apply for the combination of strong (nonlinear) drive fields and low frequency employed for our experiments. In particular, the Rabi frequency f_{Rabi} associated with the 0.5 Oe drive field h_{pump} for the electronic (Isinglike) spin of a single Ho^{3+} ion is g_{∥}μ_{B}h_{pump} ~ 10 MHz ≫ f_{pump}, which is precisely opposite to the requirement that f_{Rabi} ≪ f_{pump} for the rotating wave approximation to hold.
Tuning of nonlinear cluster response
We now examine the effects of various tuning parameters on the cluster response, showing the measured susceptibility spectra in Figs. 3 and 4, and derived quantities in Figs. 5 and 6. We explore in Figs. 3 and 5 the effects of changing the amplitude of the pumping field. Most important is the change in sign of the Fano q: for the largest drive field (0.5 Oe), the low and highfrequency responses are enhanced and suppressed, respectively, opposite to what we see for the lower drive field. The zero crossing of q occurs at a critical h_{pump} = 0.45 Oe (Fig. 5a). The data point at the pump frequency at which pump and probederived signals cannot be distinguished are ignored for the Fano fits, because they represent the response of the highly excited (pumped) clusters and not the perturbatively mixed clusters with other resonant frequencies.
Figure 5b, c reveal clear distinctions between χ_{drive}, the total signal at f_{pump} and χ_{Fano}, the linear Fano contribution calculated from evaluation at f = f_{pump} of the fitted Fano form to data at f ≠ f_{pump}. First, \(\chi _{{\mathrm{drive}}}^\prime\) goes through a maximum at the zero crossing of q (Fig. 5c), whereas \(\chi _{{\mathrm{Fano}}}^\prime\) undergoes a decrease that looks like a rounded step. Second, when we plot the phases \(\phi = \tan ^{  1}\chi \prime\prime /\chi \prime\) (Fig. 5b), we find that although both χ_{drive} and χ_{Fano} have phase shifts that are smaller at high H_{drive}, the latter actually has a zero near the zero of q. In other words, for small linear perturbations, the Fano response is dissipationless in the limit f → f_{pump}. This result follows from Eq. 2, which gives \(\phi = \phi \left( q \right) = \tan ^{  1}\frac{{q^2}}{{1  q^2}}\), a functional form that we superpose over the experimental results in Fig. 5b. The absence of dissipation in the Fano response that describes the linear continuum at q = 0 means that hole burning is actually complete at the drive frequency: there is no continuum contribution to χ″, which remains unaffected by the drive in the limit f → f_{pump}. Significantly, the absence of dissipation coincident with the q zero crossing indicates that the clusters cannot be excited between their ground and first excited states by the external drive. Although ϕ(q) gives a rough account of the experimental phase angle as q moves away from zero, the data ultimately deviate from ϕ(q), which implies that some oscillators even with frequencies close to f_{pump} are not contributing to the Fano profile.
Varying the ac pump amplitude accesses different mixtures of the states of the localized clusters. The additional power applied to the drive solenoid also results in eddycurrentinduced heating of the copper susceptometer mount and hence some degree of conductive heating of the sample despite the lowthermalconductivity Hysol epoxy spacers holding the sample inside the susceptometer. This, as well as dissipation within the sample itself, gives rise to a higher effective temperature, with a concomitant loss of coherence. The decoherence of the resonant excitation is reflected by a measurable increase in the linewidth Γ_{r}, whose temperaturedependent evolution can be traced readily in Fig. 2c. Over the range of pump amplitudes shown in Fig. 3, the resonance linewidth increases from 1.1 to 1.8 mHz (Fig. 5d), equivalent to ~50 mK of direct thermal heating. Even while heating is clearly present, the line widths remain negligible on the scale of the drive frequency, allowing the coherent superpositions of the cluster and continuum oscillations. Their relative signs change at a critical longitudinal pump field of 0.45 Oe, thus yielding the zero crossing of q, one of the main results of our experiment.
We now take advantage of one of the key features of the Li(Ho,Y)F_{4} family, the ability to tune the microscopic Hamiltonian by applying a magnetic field transverse to the Ising axis, thereby quantummechanically mixing the single ion and cluster eigenstates^{35,40} via different matrix elements than does the ac longitudinal field. Nonetheless, application of a transverse field induces a crossover (Fig. 4) at a welldefined field of H_{t} = 3.5 kOe^{24}, similar to that seen earlier as a function of pump amplitude (Fig. 3). We plot in Fig. 6a the transverse field dependence of the Fano parameter q. This parameter changes linearly with H_{t} over most of the experimental range, showing that the external transverse field not only changes the energies of different states but also tunes their interactions with the broader spin bath environment. In particular, at the H_{t} = H_{c} = 3.5 kOe crossover field, q vanishes. At the same time, as also seen when we varied h_{pump} to obtain a zero crossing of q, there is a quadratic zero in the phase for χ_{Fano}(f_{pump}) (Fig. 6b) and a maximum in \(\chi _{{\mathrm{drive}}}^\prime\) (Fig. 6c), both of which coincide with the zero of q and can be roughly described by the function ϕ(q). The simultaneous vanishing of q and the phase (and hence the dissipation) opens the possibility for the static transverse field to be used to decouple the localized twolevel systems from the external ac field. In contrast to what we saw for the h_{pump} scan with H_{t} = 0, \(\chi _{{\mathrm{drive}}}^\prime\) and \(\chi _{{\mathrm{Fano}}}^\prime\) are nearly indistinguishable at f_{pump}. Another contrast, anticipated from the previous paragraph and visible in the comparison of Figs. 5d and 6d is that the linewidth is, to within error, H_{t} independent. The essentially constant behavior of the linewidth is a strong indication that the evolution due to the transversefieldinduced quantum fluctuations is fundamentally different from the purely classical behavior seen as a function of increasing temperature.
Discussion
It is important to consider the origin of the zero crossings of the Fano q—a key result of our experiments—in terms of the ingredients of q, which are matrix elements linking the ground and excited states of the resonantly driven spin clusters to each other as well as to the bath formed by other clusters. The measurements characterize the spectral holes inserted by a nonlinear drive field into the continuum of magnetic excitations in a dense set of interacting dipoles. The large but finite lifetime of the excitations is due to a slight mixing between these (almost perfectly) localized excitations and the continuum formed by their ensemble (see Supplementary Note 2). Considering this mixing allows for the proposal of a phenomenological form (solid blue line) for the phase angles plotted in Figs. 5b and 6b:
(obtained as Eq. 6 in Supplementary Note 2) where c = 0.12 represents the intrinsic contributions of the offdiagonal matrix element and d = 1.4 Oe incorporates the thermal effects associated with changes in pump amplitude. We note that Eq. (3) explicitly incorporates the experimental observation that the vanishing of the dissipation (and hence of φ) coincides with the zero crossing of q.
The field scale H_{c} can be connected to the microscopic properties of Li(Ho,Y)F_{4} by comparison with an exact diagonalization of the full Hamiltonian for a pair of coupled Ho^{3+} atoms^{37}, where there is a crossing of the lowest coupled electronuclear levels for nearestneighbor spins positioned in the ab plane at (100) or (010) relative to each other. The effects of these pairwise level crossings can be seen in the measured linear longitudinal susceptibility χ_{zz}(H_{t})^{24,37}. The characteristic dynamics of the pairwise susceptibility shift as a function of classical (temperature) and quantummechanical (transverse field) energy scales, moving from 9 kOe at 70 mK to 5 kOe at 150 mK. We expect the same level crossings to impact the dynamics of the larger spin clusters addressed by the pumpprobe measurements described here, with the transverse field scales reported here reduced in comparison with those obtained in the pairwise calculation due to additional interactions from furtherneighbor spins.
Given that we have not performed similar experiments for compositions x other than x = 4.5%, we can only speculate as to the possible evolution of the localized cluster behavior as a function of the holmium ion concentration. As x is increased to 16.7%, only spinglasslike behavior has been seen^{40,43,58}, suggesting that placing the system in the “antiglass” state, which is a prerequisite for hole burning, would be difficult. By x = 20%, a combination of quantum effects and randomfield physics for a more nearly ferromagnetic system changes the nonlinear response^{59}, and isolated clusters even in the limit of thermal isolation are unlikely and become impossible at the onset of longrange ferromagnetic order at x ~ 30%^{43}.
Via hole burning, we have demonstrated localization of excitations among interacting magnetic dipoles. For strong coupling to the heat bath, previous work has shown a trend towards spin freezing at low temperatures^{25,49} and the current work (Fig. 1d) shows no spectral hole burning in this strong coupling limit, corresponding to delocalized excitations. Reducing the coupling to the external thermal bath induces a transition to a spin liquid state^{25,51} where the excitations are localized to a very high degree, with Q ~ 10^{5} for the spectral hole. The small mixing between excited states of different spin clusters leads to a Fano effect, whose phenomenology for this complex system is remarkably simple, with a vanishing q coinciding with an inflection point in the ac magnetization induced by the drive field responsible for the hole burning.
Even if only a subset of the spin clusters in driven LiHo_{0.045}Y_{0.955}F_{4} display localized excited states, the material with its relatively high density of interacting spins is a promising venue to investigate MBL. As the data have shown, when driven strongly with a “pump” ac magnetic field, select clusters with excitations resonant with the pump frequency can largely decouple from the remaining clusters and can be (almost) fully described by a finite number of local spin operators. The rotating reference frame of the strong pump field suggests the possibility of considering the system in a Floquet picture^{60}, where the effective spins of the onresonant clusters form a network of corotating spins, whereas the offresonant clusters are far more weakly driven and form the continuum of excitations required for the appearance of Fano resonances. By measuring the structure of the resonances, we probe the dynamics of the localized cluster excitations and have shown that the dynamics are tunable both by varying the strength of the strong pump field and by introducing quantum mixing through a static transverse magnetic field. It has been suggested^{11,61} that MBL would be observable in such systems via detailed measurements of the energy spectra of spatially localized operators. In LiHo_{x}Y_{1 − x}F_{4}, the requisite localized (spin) operators are localized in frequency, which is likely to be equivalent to spatial localization given the broad magnetic response spectrum of this disordered magnet and the very sharp holes burnt by the nonlinear ac drive. Direct measurements of the evolution of the cluster coherence times as temperature approaches zero would be helpful to distinguish between very slow conventional quantum dynamics and actual manybody localization effects.
Methods
Highresolution magnetic pumpprobe spectroscopy
We cooled a single crystal of LiHo_{0.045}Y_{0.955}F_{4} in a helium dilution refrigerator and measured its ac magnetic susceptibility for frequencies from 1 to 2000 Hz. The thermal coupling between the crystal and the heat reservoir—the mixing chamber of the dilution refrigerator—was in the weakly coupled regime described by ref. ^{25}, thereby maximizing quantum fluctuations. We employed a pumpprobe technique^{24,25,31} to access the nonlinear response regime. In this configuration, a twofrequency ac magnetic field is applied along the Ising axis of the crystal. A strong (0.2–0.6 Oe) field h_{pump} cos (2πf_{pump}t) excites clusters at the pump frequency f_{pump}. Simultaneously, a 20 mOe probe field is swept through a range of frequencies to yield a linear response from the crystal. The net response of the crystal is then sensed by an inductive pickup coil. Due to the extremely narrow separation between the pump and probe frequencies (as small as 1 mHz), disentangling the responses required the development of a twostage lockin technique, where the combined response is passed into a commercial lockin amplifier tuned to the probe frequency, and the resulting output is sampled by a computer and passed through a software lockin detector tuned to Δf = f_{pump} − f_{probe}. It should be noted that the sampling step requires a minimum of 1/Δf seconds; all of the data reported here were sampled for a minimum of 2/Δf seconds.
Data availability
The data that support the findings of this study are available from the corresponding author upon reasonable request.
References
Anderson, P. W. Absence of diffusion in certain random lattices. Phys. Rev. 109, 1492–1505 (1958).
Mott, N. F. The basis of the electron theory of metals, with special reference to the transition metals. Proc. Phys. Soc. A 62, 416–422 (1949).
Gornyi, I. V., Mirlin, A. D. & Polyakov, D. G. Interacting electrons in disordered wires: Anderson localization and lowT transport. Phys. Rev. Lett. 95, 206603 (2005).
Basko, D. M., Aleiner, I. L. & Altshuler, B. L. Metal–insulator transition in a weakly interacting manyelectron system with localized singleparticle states. Ann. Phys. 321, 1126–1205 (2006).
Oganesyan, V. & Huse, D. A. Localization of interacting fermions at high temperature. Phys. Rev. B 75, 155111 (2007).
Polkovnikov, A., Sengupta, K., Silva, A. & Vengalattore, M. Colloquium: Nonequilibrium dynamics of closed interacting quantum systems. Rev. Mod. Phys. 83, 863–883 (2011).
Huse, D. A., Nandkishore, R., Oganesyan, V., Pal, A. & Sondhi, S. L. Localizationprotected quantum order. Phys. Rev. B 88, 014206 (2013).
Nandkishore, R. & Huse, D. A. Manybody localization and thermalization in quantum statistical mechanics. Annu. Rev. Condens. Matter Phys. 6, 15–38 (2015).
Altman, E. & Vosk, R. Universal dynamics and renormalization in manybodylocalized systems. Annu. Rev. Condens. Matter Phys. 6, 383–409 (2015).
Alet, F. & Laflorencie, N. Manybody localization: an introduction and selected topics. Comptes Rendus Phys. 19, 498–525 (2018).
Johri, S., Nandkishore, R. & Bhatt, R. N. Manybody localization in imperfectly isolated quantum systems. Phys. Rev. Lett. 114, 117401 (2015).
Khemani, V., Lim, S. P., Sheng, D. N. & Huse, D. A. Critical properties of the manybody localization transition. Phys. Rev. X 7, 021013 (2017).
Schreiber, M. et al. Observation of manybody localization of interacting fermions in a quasirandom optical lattice. Science 349, 842–845 (2015).
Choi, J.Y. et al. Exploring the manybody localization transition in two dimensions. Science 352, 1547–1552 (2016).
Bernien, H. et al. Probing manybody dynamics on a 51atom quantum simulator. Nature 551, 579–584 (2017).
Johnson, M. W. et al. Quantum annealing with manufactured spins. Nature 473, 194–198 (2011).
Harris, R. et al. Phase transitions in a programmable quantum spin glass simulator. Science 361, 162–165 (2018).
Pellizzari, T., Gardiner, S. A., Cirac, J. I. & Zoller, P. Decoherence, continuous observation, and quantum computing: a cavity QED model. Phys. Rev. Lett. 75, 3788–3791 (1995).
Byrd, M. S. & Lidar, D. A. Comprehensive encoding and decoupling solution to problems of decoherence and design in solidstate quantum computing. Phys. Rev. Lett. 89, 047901 (2002).
Tyryshkin, A. M., Lyon, S. A., Jantsch, W. & Schäffler, F. Spin manipulation of free twodimensional electrons in Si/SiGe quantum wells. Phys. Rev. Lett. 94, 126802 (2005).
Ithier, G. et al. Decoherence in a superconducting quantum bit circuit. Phys. Rev. B 72, 134519 (2005).
Kroner, M. et al. The nonlinear Fano effect. Nature 451, 311–314 (2008).
McCall, S. L. & Hahn, E. L. Selfinduced transparency. Phys. Rev. 183, 457–485 (1969).
Silevitch, D. M., Gannarelli, C. M. S., Fisher, A. J., Aeppli, G. & Rosenbaum, T. F. Quantum projection in an Ising spin liquid. Phys. Rev. Lett. 99, 057203 (2007).
Schmidt, M. A., Silevitch, D. M., Aeppli, G. & Rosenbaum, T. F. Using thermal boundary conditions to engineer the quantum state of a bulk magnet. Proc. Natl Acad. Sci. USA 111, 3689–3694 (2014).
Fano, U. Effects of configuration interaction on intensities and phase shifts. Phys. Rev. 124, 1866 (1961).
Bryant, H. C. et al. Observation of resonances near 11 eV in the photodetachment cross section of the H^{−} ion. Phys. Rev. Lett. 38, 228–230 (1977).
Thomsen, C., Cardona, M., Gegenheimer, B., Liu, R. & Simon, A. Untwinned single crystals of YBa_{2}Cu_{3}O_{7δ}: an optical investigation of the ab anisotropy. Phys. Rev. B 37, 9860–9863 (1988).
Poddubny, A. N., Rybin, M. V., Limonov, M. F. & Kivshar, Y. S. Fano interference governs wave transport in disordered systems. Nat. Comms. 3, 914 (2012).
Faist, J., Capasso, F., Sirtori, C., West, K. & Pfeiffer, L. Controlling the sign of quantum interference by tunnelling from quantum wells. Nature 390, 589–591 (1997).
Ghosh, S., Parthasarathy, R., Rosenbaum, T. F. & Aeppli, G. Coherent spin oscillations in a disordered magnet. Science 296, 2195–2198 (2002).
Hansen, P., Johansson, T. & Nevald, R. Magnetic properties of lithium rareearth fluorides: ferromagnetism in LiErF_{4} and LiHoF_{4} and crystalfield parameters at the rareearth and Li sites. Phys. Rev. B 12, 5315–5324 (1975).
Beauvillain, P., Renard, J. P., Laursen, I. & Walker, P. J. Critical behavior of the magnetic susceptibility of the uniaxial ferromagnet LiHoF_{4}. Phys. Rev. B 18, 3360–3368 (1978).
Matmon, G., Lynch, S. A., Rosenbaum, T. F., Fisher, A. J. & Aeppli, G. Optical response from terahertz to visible light of electronuclear transitions in LiYF_{4}:Ho^{3}. Phys. Rev. B 94, 205132 (2016).
Chakraborty, P. B., Henelius, P., Kjonsberg, H., Sandvik, A. W. & Girvin, S. M. Theory of the magnetic phase diagram of LiHoF_{4}. Phys. Rev. B 70, 144411 (2004).
Schechter, M. & Stamp, P. C. E. Derivation of the low T phase diagram of LiHo_{x}Y_{1−x}F_{4}: a dipolar quantum Ising magnet. Phys. Rev. B 78, 054438 (2008).
Gannarelli, C. M. S., Silevitch, D. M., Rosenbaum, T. F., Aeppli, G. & Fisher, A. J. Contribution of spin pairs to the magnetic response in a dilute dipolar ferromagnet. Phys. Rev. B 86, 014420 (2012).
McKenzie, R. D. & Stamp, P. C. E. Thermodynamics of a quantum Ising system coupled to a spin bath. Phys. Rev. B 97, 214430 (2018).
Rønnow, H. M. et al. Magnetic excitations near the quantum phase transition in the Ising ferromagnet LiHoF_{4}. Phys. Rev. B 75, 054426 (2007).
Wu, W., Ellman, B., Rosenbaum, T. F., Aeppli, G. & Reich, D. H. From classical to quantum glass. Phys. Rev. Lett. 67, 2076–2079 (1991).
Bitko, D., Rosenbaum, T. F. & Aeppli, G. Quantum critical behavior for a model magnet. Phys. Rev. Lett. 77, 940–943 (1996).
Rønnow, H. M. et al. Quantum phase transition of a magnet in a spin bath. Science 308, 389–392 (2005).
Reich, D. H. et al. Dipolar magnets and glasses: neutronscattering, dynamical, and calorimetric studies of randomly distributed Ising spins. Phys. Rev. B 42, 4631–4644 (1990).
Brooke, J., Rosenbaum, T. F. & Aeppli, G. Tunable quantum tunnelling of magnetic domain walls. Nature 413, 610–613 (2001).
Schechter, M. & Stamp, P. C. E. Significance of the hyperfine interactions in the phase diagram of LiHo_{x}Y_{1−x}F_{4}. Phys. Rev. Lett. 95, 267208 (2005).
Silevitch, D. M., Aeppli, G. & Rosenbaum, T. F. Switchable hardening of a ferromagnet at fixed temperature. Proc. Natl Acad. Sci. USA 107, 2797–2800 (2010).
Silevitch, D. M. et al. A ferromagnet in a continuously tunable random field. Nature 448, 567–570 (2007).
Tabei, S. M. A., Gingras, M. J. P., Kao, Y.J., Stasiak, P. & Fortin, J.Y. Random field effects in the transverse field Ising spinglass LiHo_{x}Y_{1−x}F_{4} magnetic material. Phys. Rev. Lett. 97, 237203 (2006).
Quilliam, J. A., Meng, S., Mugford, C. G. A. & Kycia, J. B. Evidence of spin glass dynamics in dilute LiHo_{x}Y_{1−x}F_{4}. Phys. Rev. Lett. 101, 187204 (2008).
Tam, K.M. & Gingras, M. J. P. Spinglass transition at nonzero temperature in a disordered dipolar Ising system: the case of LiHo_{x}Y_{1−x}F_{4}. Phys. Rev. Lett. 103, 087202 (2009).
Ghosh, S., Rosenbaum, T. F., Aeppli, G. & Coppersmith, S. Entangled quantum state of magnetic dipoles. Nature 425, 48–51 (2003).
Bhatt, R. N. & Lee, P. A. Scaling studies of highly disordered spin½ antiferromagnetic systems. Phys. Rev. Lett. 48, 344–347 (1982).
Biltmo, A. & Henelius, P. Unreachable glass transition in dilute dipolar magnet. Nat. Commun. 3, 857 (2012).
Paalanen, M. A., Rosenbaum, T. F., Thomas, G. A. & Bhatt, R. N. Stress tuning of the metalinsulator transition at Millikelvin temperatures. Phys. Rev. Lett. 48, 1284–1287 (1982).
Giraud, R., Wernsdorfer, W., Tkachuk, A. M., Mailly, D. & Barbara, B. Nuclear spin driven quantum relaxation in LiY_{0.998}Ho_{0.002}F_{4}. Phys. Rev. Lett. 87, 057203 (2001).
Mennenga, G., de Jongh, L. & Huiskamp, W. Field dependent specific heat study of the dipolar Ising ferromagnet LiHoF_{4}. J. Magn. Magn. Mater. 44, 59–76 (1984).
Hanggi, P. in Quantum Transport and Dissipation 249–286 (Wiley VCH Verlag GmbH, 1998).
Quilliam, J. A., Meng, S. & Kycia, J. B. Experimental phase diagram and dynamics of a dilute dipolarcoupled Ising system. Phys. Rev. B 85, 184415 (2012).
AnconaTorres, C., Silevitch, D. M., Aeppli, G. & Rosenbaum, T. F. Quantum and classical glass transitions in LiHo_{x}Y_{1−x}F_{4}. Phys. Rev. Lett. 101, 057201 (2008).
Ponte, P., Papić, Z., Huveneers, F. & Abanin, D. A. Manybody localization in periodically driven systems. Phys. Rev. Lett. 114, 140401 (2015).
Fischer, M. H., Maksymenko, M. & Altman, E. Dynamics of a manybodylocalized system coupled to a bath. Phys. Rev. Lett. 116, 160401 (2016).
Acknowledgements
We thank G. Refael, Michael Buchhold, and Markus Müller for helpful discussions. The work at Caltech was supported by US Department of Energy Basic Energy Sciences Award DESC0014866.
Author information
Authors and Affiliations
Contributions
D.M.S. and T.F.R. performed the measurements. D.M.S., C.T., G.A. and T.F.R. participated in the analysis of the data and the writing of the manuscript.
Corresponding author
Ethics declarations
Competing interests
The authors declare no competing interests.
Additional information
Peer review information: Nature Communications thanks the anonymous reviewers for their contribution to the peer review of this work.
Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary information
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Silevitch, D.M., Tang, C., Aeppli, G. et al. Tuning highQ nonlinear dynamics in a disordered quantum magnet. Nat Commun 10, 4001 (2019). https://doi.org/10.1038/s41467019119851
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41467019119851
This article is cited by

Emergence of highly coherent twolevel systems in a noisy and dense quantum network
Nature Physics (2024)

Observation of an exotic insulator to insulator transition upon electron doping the Mott insulator CeMnAsO
Nature Communications (2023)

Emergence of mesoscale quantum phase transitions in a ferromagnet
Nature (2022)
Comments
By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.