betaFeedback

WorksPhysics and Astronomy

Scaling behavior and giant field-enhancement of the thermal conductivity in the honeycomb antiferromagnet BaCo2(AsO4)2

Abstract

The layered honeycomb material BaCo2_2(AsO4_4)2_2 (BCAO) is of topical interest because its magnetic state is related to that of the Kitaev magnet αα-RuCl3_3. Using thermal transport to probe how magnetic excitations interact with phonons in the magnetically disordered regime, we have uncovered an unusually large enhancement of the thermal conductivity κxxκ_{xx} in an in-plane magnetic field H{\bf H}. Just above the Néel temperature TNT_{\rm N}, a field of 13 T increases κxxκ_{xx} by a factor ∼211\sim 211, much larger than reported previously in any magnetic insulator. Interestingly, κxx(H,T)κ_{xx}(H,T) exhibits a scaling behavior in the entire magnetically disordered region that surrounds the ordered zigzag state. The ratio Δκxx(H,T)/κxx(13,T)Δκ_{xx}(H,T)/κ_{xx}(13,T), measured throughout the disordered region, collapses to a one-parameter scaling function exp(−1/gx){\rm exp}(-1/gx) (where x=μBB/kBTx = μ_{\rm B}B/k_{\rm B}T and gg is a constant).

Introduction

The family of layered honeycomb spin-12\frac{1}{2} magnetic materials has attracted considerable interest. In particular, the ruthenate α\alpha-RuCl3_3 with local moments derived from 4d4d orbitals in Ru has been identified [1, 2] as the closest proximate to the Kitaev Hamiltonian HK\mathcal{H}_K [3]. The low-temperature thermal transport properties have played a central role in illuminating the extent to which predictions based on HK\mathcal{H}_K agree with experiment [4, 5, 6, 7, 8]. Interest has extended to related layered, honeycomb spin-12\frac{1}{2} materials. In the cobaltate BaCo2_2(AsO4_4)2_2 (BCAO), the local moments based on Co-3d orbitals undergo a sharp transition to a zig-zag state at the Néel temperature TN=5.4T_\mathrm{N}=5.4 K [9, 10, 11]. In its magnetic, thermal and crystallographic properties, BCAO shares many features with α\alpha-RuCl3_3. A number of recent reports have compared their magnetic and transport properties [8, 12, 13, 14, 15, 16, 17, 18, 19, 20].

In the case of BCAO, the theoretical and experimental investigations appear to favor a J1J_1-J3J_3, XXZXXZ Hamiltonian instead of the Kitaev Hamiltonian [13, 14, 15, 16, 17, 18]. Below TNT_\mathrm{N}, the spins in BCAO exhibit a helical magnetic texture in the abab-plane. The spins, nominally aligned along the bb-axis, exhibit a spiral structure with axis parallel to a∗a^*. In an in-plane magnetic field H, the spins are polarized in a ferrimagnetic, collinear pattern above a field Hc1∼0.2H_{c1}\sim0.2 T. Long-range magnetic order is suppressed at the critical field Hc2=0.5H_{c2}=0.5 T [9][11][13]. The transition at the lower field Hc1H_{c1} has been investigated by magnetization and terahertz spectroscopy experiments [12, 13, 19]. A recent study combining angular dependence of magnetization and dilatometry suggests an intermediate phase appears before the spins are fully polarized [18]. Another thermal transport experiment measured down to subKelvin temperatures supports the KQSL picture, yet other theoretical proposal attribute it to a Dirac spin liquid phase [19, 21].

Our work here focuses on the magnetically disordered regime which covers a large fraction of the HH-TT plane. We report an unusual scaling behavior of the thermal conductivity κxx(H,T)\kappa_{xx}(H,T) in BCAO observed throughout the magnetically disordered region below 10 K. In zero HH, strong inelastic scattering of phonons from spin excitations severely suppresses phonon conduction. In an in-plane magnetic field H, κxx(H,T)\kappa_{xx}(H,T) increases dramatically following a one-parameter scaling function. Just above TNT_\mathrm{N}, the field-induced enhancement factor (>200>200) exceeds that previously reported in all magnetic insulators to our knowledge.

Underlying the giant field-enhancement of κxx\kappa_{xx} is the intense scattering of phonons by the spins which leads to “localization” of the phonons in zero HH. Our results suggest that the non-analytic behavior of κxx(H,T)\kappa_{xx}(H,T) in the limit H→0H\to0, as well as the scaling behavior over a large fraction of the magnetically disordered region, may well be signature hallmarks of strong spin-phonon coupling.

In several topical areas, e.g. spin qubits based on NV centers in diamond [22], spin-phonon coupling is a key topic of interest. The scaling behavior uncovered here suggests an incisive way to probe these couplings in magnetic materials.

Experiment

To measure the thermal conductivity tensor κij\kappa_{ij} of the BCAO crystal, we heat sunk 3 RuO2_2 thermometers (matched and calibrated) to its edges using stycast and gold wires (Fig. 1a). One short edge of the crystal is heat sunk to the bath (brass substrate) while the opposite edge is attached to a heater via gold wires. The sample space is kept at a pressure less than 5×10−55\times10^{-5} mbar. We label as x^\hat{x} the common axis of the heat current density J and field H.

Typically, we measure the field profile of κxx(H,T)\kappa_{xx}(H,T) keeping the bath temperature TbathT_{\mathrm{bath}} regulated at a set point as HH is swept from 0 to 13 T and back. At low TT, the applied heat current invariably leads to a finite shift δT=T−Tbath\delta T = T - T_{\mathrm{bath}} between the sample’s temperature TT and TbathT_{\mathrm{bath}} (the shift is usually nearly HH independent). Here, however, the anomalously large (211-fold) increase in κxx(H,T)\kappa_{xx}(H,T) causes δT(H,T)\delta T(H,T) to be strongly field dependent. In all field profiles reported here, we have recorded δT(H,T)\delta T(H,T) to correct for these shifts. We also adopted steps to avoid errors arising from large magnetocaloric effects which lead to long relaxation times for both the sample and bath to reach equilibrium whenever HH undergoes a step-change (Secs. 1, 3 and 4 of Supplement [23]). Finally, the tendency of the crystal to fracture under strain exerted by magnetic torque restricts measurements so far to in-plane H\mathbf{H}.

Motivated by the observation of an unusual planar thermal Hall conductivity κxy\kappa_{xy} in the Kitaev magnet α\alpha-RuCl3_3 [6, 7], we have searched for a planar thermal Hall signal in BCAO focusing on the disordered region close to the ordered phase. Despite a careful search we did not detect a finite κxy\kappa_{xy} signal in BCAO within our resolution (See Sec. 7 of Supplement [23]).

Field and temperature profiles of thermal conductivity

The thermal conductivity κxx(H,T)\kappa_{xx}(H,T) of BCAO displays a rich set of features as a function of temperature TT and an in-plane magnetic field H\mathbf{H}. In the absence of H\mathbf{H}, κ(T)\kappa(T) (≡κxx(0,T)\equiv\kappa_{xx}(0,T), plotted as blue symbols in Figure 1c) slowly decreases to an anomalously low value (∼0.6\sim0.6 W/m K) as TT is lowered to 5.4 K. At the Néel temperature TNT_{\mathrm{N}} (5.4 K), the spins undergo a phase transition to an antiferromagnetic state with zig-zag order. In the ordered state, κ\kappa rises to a broad peak that reflects contributions from phonons and magnons in the ordered state.

Schematic of the crystal, zigzag phase, thermal conductivity curves, and phase diagram

Figure 1. Panel (a): Schematic of the crystal attached to the heater, 3 thermometers and the bath by gold wires. The applied H\mathbf{H} direction is indicated. Panel (b): The two-up, two-down zigzag phase inferred from neutron scattering [11]. Panel (c): Comparison of the zero-HH κ(T)\kappa(T) (square blue symbols, multiplied by a factor of 30) with κxx\kappa_{xx} vs. TT measured in an in-plane 13-T magnetic field (red symbols). The faint blue curve passing through the blue symbols is an interpolation supplemented by measurements in zero HH in which TT is continuously varied. The red curve is a fit to the 13-T data using the standard Callaway expression. More details can be found in the Supplement [23]. Panel (d): The inferred phase diagram of BCAO in the HH-TT plane. The magnetically ordered zigzag state is shaded red and coral. Throughout the disordered region (shaded green), κxx(H,T)\kappa_{xx}(H,T) obeys a one-parameter scaling behavior. Deviation from this scaling occurs in the blue region. The fields Hc1H_{\mathrm{c1}} and Hc2H_{\mathrm{c2}} are obtained from data presented in Figure 2a.

In our experiment, we observe that long-range order is abruptly suppressed by an in-plane field H∼0.5H \sim0.5 T, consistent with previous studies [12, 13, 15, 20]. As described below, phonons remain strongly coupled to spin excitations throughout the disordered region surrounding the ordered state (green region in Figure 1d). The strong coupling is steadily weakened as HH increases. Eventually, at large HH (13 T), scattering of phonons by the spins becomes negligible, and κxx\kappa_{xx} attains large values (red symbols in Panel c). The field enhancement, expressed by the ratio κxx(H,T)/κ(T)\kappa_{xx}(H,T)/\kappa(T) (∼211\sim211 just above TNT_{\mathrm{N}}) exceeds that in all other magnetic insulator reported to date. Our key result is that, over the entire magnetically disordered region (green region in Panel d), κxx(H,T)\kappa_{xx}(H,T) obeys a one-parameter scaling function.

fig:2a displays field profiles of κxx\kappa_{xx} within the ordered state with TT fixed at the values indicated. At the lowest TT (0.49 K), the profile of κxx\kappa_{xx} is virtually flat. When TT is raised (1.64 to 2.75 K) the field profiles feature a dome that is abruptly truncated on the large-HH side by a vertical cliff as HH approaches Hc2H_{\mathrm{c2}} (0.5 T), and long-range order is suppressed. On the low HH side, the dome displays a notch that corresponds to the lower critical field Hc1H_{\mathrm{c1}} (∼0.12\sim0.12 T). In previous reports, Hc1H_{\mathrm{c1}} was interpreted as a transition between either an incommensurate spiral state or a double zigzag pattern to the cAFM state [11, 20]. Unlike in Refs. [12, 19], we observe only a single sharp transition within the field interval 0.1–0.2 T. In the disordered phase above Hc2H_{c2}, κxx\kappa_{xx} increases steeply with HH. This is a small fraction of the scaling pattern uncovered below.

Figure 2b shows the field profiles of κxx\kappa_{xx} with the HH axis extended to 11 T (the profiles in the ordered phase are now reduced to small mesas visible below 0.5 T). At each TT (e.g. 3 K), κxx\kappa_{xx} increases from a small value (∼1\sim1 W/mK) before slowly saturating to a maximum value (∼42\sim42 W/mK) as HH increases to 13 T. All the profiles appear to follow the same functional form, with a pattern amenable to simple analysis.

Profiles of $\kappa_{xx}$ versus field and temperature

Figure 2. Profiles of κxx\kappa_{xx} vs. field and temperature. Panel (a): Field profiles of κxx(H,T)\kappa_{xx}(H,T) at low HH (μ0H<1.2\mu_0H < 1.2 T). In each curve, TT is fixed at the values indicated. Curves in the temperature interval 1.6<T<2.81.6 < T < 2.8 K display a vertical cliff that signals the abrupt collapse of long-range magnetic order as HH approaches Hc2=0.5H_{c2}=0.5 T. At low HH, a second sharp feature Hc1H_{c1} (∼0.1\sim0.1 T) is observed. Panel (b): Field profiles of κxx\kappa_{xx} extended to 11 T, with TT fixed at the values indicated. At each TT, κxx\kappa_{xx} initially increases exponentially with HH. At large HH, it gradually saturates to its high-field limit. Profiles shown in Panel (a) are visible as the small mesa below 0.5 T. Panel (c): Curves of κxx(H,T)\kappa_{xx}(H,T) plotted versus TT with HH fixed at the values indicated. The curve κxx(13,T)\kappa_{xx}(13,T) at the maximum applied HH (13 T) serves as an envelope that caps the maximum value attained at each temperature TT. Each curve (e.g. the one at H=4H=4 T) initially tracks closely κxx(13,T)\kappa_{xx}(13,T) as TT is raised above 1 K, but deviates away from κxx(13,T)\kappa_{xx}(13,T) as TT is raised (at ∼2.5\sim2.5 K). Curves taken at larger HH (e.g. 10 T) deviate from κxx(13,T)\kappa_{xx}(13,T) at a higher temperature (4 K). This pattern reflects the one-parameter scaling behavior discussed below. Note the anomalously low values of κxx\kappa_{xx} in the curve at 0.6 T. Panel (d): First hints of scaling behavior above TNT_N. When we plot the field-induced change Δκxx(H,T)\Delta\kappa_{xx}(H,T) vs. the scaling variable μ0H/T\mu_0H/T, we find that all data points above TNT_N converge to the same curve when the scaling variable μ0H/T\mu_0H/T falls below ∼0.4\sim0.4 T/K. The temperature of each data point is given by the color scale.

To this end, we turn to the orthogonal cuts in which κxx(H,T)\kappa_{xx}(H,T) is plotted vs. TT with HH fixed (Figure 2c). At each fixed value of HH, the curve of κxx\kappa_{xx} vs. TT is now non-monotonic, exhibiting a broad maximum at the temperature Tp(H)T_p(H). As shown in the inset, Tp(H)T_p(H) increases roughly linearly with HH. This suggests that, in the disordered region, κxx\kappa_{xx} depends on HH solely through the scaling variable

x=μBBkBT,(1)x = \frac{\mu_B B}{k_B T}, \tag*{(1)}

where μB\mu_B and kBk_B are, respectively, the Bohr magneton and the Boltzmann constant, and B=μ0HB = \mu_0 H is the magnetic flux density with μ0\mu_0 the vacuum permeability.

To test this preliminary hint of scaling behavior, we plot in Figure 2d the field-induced change Δκxx(H,T)=κxx(H,T)−κxx(0,T)\Delta\kappa_{xx}(H,T) = \kappa_{xx}(H,T) - \kappa_{xx}(0,T) versus xx in log-log scale for TT between 5.4 and 9.5 K. We find that the data merge to define a scaling curve that extends over nearly 3 decades in Δκxx(H,T)\Delta\kappa_{xx}(H,T).

Another important feature in Figure 2c is the role of κxx(13,T)\kappa_{xx}(13,T) (black squares) relative to κxx\kappa_{xx} measured at lower fields throughout the disordered region. Combining the perspectives afforded by the fixed-TT and fixed-HH cuts, we infer that when TT is fixed (e.g. at 4 K), κxx\kappa_{xx} increases towards κxx(13,T)\kappa_{xx}(13,T) but does not exceed it.

Physically, this implies that, in the high-field limit, phonon conduction asymptotes to the regime in which ℓ\ell, restricted by scattering from crystalline disorder only, becomes independent of TT. In this limit, the TT dependence observed in κxx(H,T)=13cgs(T)vsℓ\kappa_{xx}(H,T) = \frac{1}{3}c_{\mathrm{gs}}(T)v_s\ell arises from the heat capacity cgs(T)c_{\mathrm{gs}}(T) of the spin-phonon system (vsv_s is the group velocity of the acoustic branch). As inferred from the field profiles in Figure 2b, κxx(13,T)\kappa_{xx}(13,T) is a good approximation to this high-field limit below T=6T = 6 K. The approximation is less satisfactory above 6 K as saturation is approached at increasingly large HH. The dashed curve in Figure 2c is a fit to κxx(13,T)\kappa_{xx}(13,T) using the Callaway model in this limit (see Sec. 2 of Supplement [23]).

Scaling behavior

The functional form of the scaling curve in Figure 2d becomes evident when we plot Δκxx\Delta\kappa_{xx} vs. 1/x1/x in semilog plot. As shown in Figure 3a, all the data above 5.4 K collapse onto the straight line corresponding to the function Δκxx∼e−1/gx\Delta\kappa_{xx} \sim e^{-1/gx} with gg a constant.

Three-panel plots showing the scaling behavior of thermal conductivity

Figure 3. Uncovering the scaling behavior. (Panel a) The analytical form of the scaling behavior implied in Figure 2d becomes apparent in a semilog plot of Δκxx\Delta\kappa_{xx} vs. 1/x1/x with xx defined in Eq. (1). All data points taken above TNT_N fall on the dashed line given by Eqs. (2) and (1). The equation of the fit (dashed line) is ln⁡Δκxx=−0.628x−1+5.476\ln\Delta\kappa_{xx}=-0.628x^{-1}+5.476, which gives g=1.6g=1.6. The scatter as x→0x\to0 reflects uncertainties in Δκxx\Delta\kappa_{xx}. (Panel b) Finding the missing pre-factor in the scaling form. At temperatures below 5.4 K, the data points κxx(H,T)\kappa_{xx}(H,T) segregate into two distinct groups when plotted vs. 1/x1/x. In the ordered phase (H<Hc2H<H_{c2}), the data follow nearly flat trajectories. By contrast, data in the disordered region (H>Hc2H>H_{c2}) fall on diagonal curves that match the dashed line in Panel (a), except that they are systematically displaced downwards as T→0T\to0. This suggests that κxx\kappa_{xx} should be divided by a TT-dependent factor TβT^\beta, which we approximate by the measured curve κxx(13,T)\kappa_{xx}(13,T). (Panel c): Semilog plots of κxx/κxx(13,T)\kappa_{xx}/\kappa_{xx}(13,T) vs. 1/x1/x for T<5.4T<5.4 K (upper panel) and T>5.4T>5.4 K (lower). Once κxx\kappa_{xx} is divided by κxx(13,T)\kappa_{xx}(13,T), the diagonal curves in Panel (b) collapse to a single curve, which represents the universal curve given in Eq. (2). The data above 5.4 K also collapse onto the same curve (lower panel). Hence, over the disordered region (green region in Figure 1d), scaling behavior prevails. Departure from scaling becomes observable when 1/x1/x decreases below ∼3\sim3 (dark blue region in Figure 1d).

The simple scaling based on xx accounts only for the HH dependence of Δκxx(T,H)\Delta\kappa_{xx}(T,H). The further dependence on TT requires more care. This becomes apparent if we attempt to apply the same analysis to the data below 5.4 K in the disordered region (H>Hc2H > H_{c2}).

The semilog plot of κxx\kappa_{xx} vs. 1/x1/x shown in Figure 3b reveals that the data segregate into two distinct groups. Data in the ordered phase (not relevant to scaling) follow flat trajectories that are nearly independent of xx. By contrast, the data in the disordered region fall along diagonal curves that mimic the scaling function except that they are now displaced vertically by a factor of the form TβT^\beta with exponent 2<β<32 < \beta< 3, which we infer arises from the heat capacity cgs(T)c_{\mathrm{gs}}(T). From the discussion on Figure 2c, we anticipate that the offsets can be cancelled if we divide κxx(H,T)\kappa_{xx}(H,T) by κxx(13,T)\kappa_{xx}(13,T).

This inference is confirmed when we replot κxx(H,T)/κxx(13,T)\kappa_{xx}(H,T)/\kappa_{xx}(13,T) vs. 1/x1/x in Figure 3c (upper panel). Now all data throughout the disordered region below 5.4 K collapse onto one scaling curve. A similar plot shows that data above 5.4 K also collapse onto the same curve (lower panel). As noted above, deviation from scaling appears when xx exceeds ∼0.3\sim0.3.

Hence, throughout the disordered region below and above 5.4 K, the ratio Δκxx(H,T)/κxx(13,T)\Delta\kappa_{xx}(H,T)/\kappa_{xx}(13,T) is described by the one-parameter scaling function

Δκxx(H,T)κxx(13,T)=Aexp⁡[−1gx](2)\frac{\Delta\kappa_{xx}(H,T)}{\kappa_{xx}(13,T)} = A\exp\left[-\frac{1}{gx}\right] \tag*{(2)}

where xx is given by Eq. (1) and AA and gg are constants.

The function e−1/gxe^{-1/gx} implies that Δκxx\Delta\kappa_{xx} initially increases from zero at a very slow rate (the non-analytic limit H→0H \to0 requires all derivatives to vanish). As discussed, the ultra-slow initial growth may be seen in Figure 2a. However, once BB rises above 1 Tesla, Δκxx\Delta\kappa_{xx} undergoes an exponential increase before ultimately saturating to a very large value consistent with phonons in a crystal with a low density of lattice and spin disorder.

We sketch the physical picture suggested by our results. The single-parameter scaling form constrains the candidate scenarios for describing the field-enhanced κxx\kappa_{xx}. First, the anomalously low value of κ(T)\kappa(T) in the fluctuating interval 5.4<T<105.4 < T < 10 K at H=0H = 0 implies strong spin-phonon coupling which leads to repeated absorption or emission of phonons accompanied by spin flips. Secondly, the non-analytic nature in Eq. (2) in the limit x→0x \to0 implies that, in weak HH, deviation of Δκxx(H,T)\Delta\kappa_{xx}(H,T) from zero is initially nearly unresolvable until HH exceeds a field scale of ∼1\sim1 T. Then κxx\kappa_{xx} increases exponentially. This strongly non-linear increase is not amenable to a perturbative treatment. (We also point out that Eq. (2) is very different from the familiar expression e−Δb/kBTe^{-\Delta_b/k_B T} describing thermal activation of charge carriers across a band gap Δb\Delta_b.)

Throughout the disordered region, we infer that the phonons and spin excitations constitute a tightly coupled system in which ℓ0\ell_0 (the phonon mean-free path at H=0H = 0) is limited to less than 30 nm by intense scattering. Although the scattering may be elastic or inelastic, the scaling form implies that inelastic processes are dominant in the disordered region. By analogy with experiments on the honeycomb system α\alpha-RuCl3_3 [24, 25, 26, 27] we may expect that spin excitations in the disordered state exhibit a broad, featureless spectrum that is gapless in zero HH. The low-lying spin excitations interact strongly with phonons at a select population of sites, which absorb and emit phonons. At temperature TT, the phonon energy distribution peaks sharply at ℏω≃kBT\hbar\omega\simeq k_B T.

An applied Zeeman field opens a gap Δ=gμBH\Delta= g\mu_B H in the spin spectrum where gg is the effective gyromagnetic parameter for the low-lying spin excitations. Phonon annihilation is blocked when the gap exceeds kBTk_BT, i.e.

gμBB>kBT.(3)g\mu_B B > k_B T. \tag*{(3)}

In effect, the identified absorption sites act as escape hatches that are propped open by HH. Rendering these sites transparent to phonons allows them to escape the trapping well. Despite their small population, these sites play an outsize role in determining the initial growth of κxx(H,T)\kappa_{xx}(H,T).

When HH is weak, the average spacing between escape hatches far exceeds ℓ0\ell_0. Each liberated phonon does not travel far before becoming retrapped. Hence the initial increase in κxx\kappa_{xx} is nearly unresolvable, as shown in Figure 2a and Figure 2b, and implied by the non-analyticity of e−T/Be^{-T/B} in the limit B→0B\to0. Crucially, by Eq. (3), the field at which phonons escape at the identified sites is proportional to TT. Hence, the field-induced increase in ℓ\ell automatically satisfies the scaling in Eq. (2).

Increasing HH weakens the amplitude of the fluctuating magnetic order by increasing the polarization of spins, which causes the population of escape hatches to pro-

hatches far exceeds ℓ0\ell_0. Each liberated phonon does not travel far before becoming retrapped. Hence the initial increase in κxx\kappa_{xx} is nearly unresolvable, as shown in Figure 2a and Figure 2b, and implied by the non-analyticity of e−T/Be^{-T/B} in the limit B→0B\to0. Crucially, by Eq. (3), the field at which phonons escape at the identified sites is proportional to TT. Hence, the field-induced increase in ℓ\ell automatically satisfies the scaling in Eq. (2).

Increasing HH weakens the amplitude of the fluctuating magnetic order by increasing the polarization of spins, which causes the population of escape hatches to proliferate. In analogy with percolation conductance, the largest enhancement of ℓ\ell occurs when connected paths form skeletal spines that span the crystal. When HH exceeds ∼1\sim1 T, rapid proliferation of the hatches enables the connectivity of phonon propagation paths to exceed the percolation limit, and κxx\kappa_{xx} undergoes the exponential growth observed.

Comparison with other magnetic insulators

Experiments showing large field-induced enhancement of κxx\kappa_{xx} have been reported in several layered magnetic insulators. We briefly survey the results beginning with the largest previously reported (Figure 4 and Table S1 in Supplement [23]). Zhao et al. found that, in Co3_3V2_2O8_8, κxx\kappa_{xx} increases 100 fold in a field of 14 T [28] at 7.5 K. They interpret the increase as caused by a series of magnetic phase transitions occurring in a narrow window (5 transitions between 6 and 12 K). While there are fewer known phase transitions in the case of BCAO, recent dilatometry experiment indeed suggests more intermediate phases [18]. One difference we want to point out though is that BCAO has only one critical temperature, and it has been relatively robust even when applied pressure [15]. More importantly, the field-induced enhancement persists up to as high as 13 T well beyond the critical field Hc2=0.5H_{c2}=0.5 T, and the scaling behavior also covers a wide regime in the phase diagram surrounding the ordering states. As a result, a succession of phase transitions doesn’t seem to be the whole story, though intense spin fluctuations do seem to be one of the key components.

Ratio of increase in thermal conductivity under field

Figure 4. Comparison of the field enhancement of κxx\kappa_{xx} (expressed as a ratio) with the ratio reported for other magnetic insulators[28, 29, 30, 31, 32, 33, 34]. BCAO is marked with a red star. In the ratio for HoMnO3∗_3^* the denominator is κxx(2.2T)\kappa_{xx}(2.2\mathrm{T}). For RuCl3∗∗_3^{**}, the denominator is κxx(7.5T)\kappa_{xx}(7.5\mathrm{T}). See Sec. 6 in Supplement [23] for detailed information.

A recent experiment on another honeycomb Kitaev magnet candidate Na2_2Co2_2TeO6_6 found that κxx\kappa_{xx} increases by a factor of 70 in a field of 16 T [29]. The authors attribute this intense scattering to the abundance of magnetic excitations above a triple-Q ground state. The authors do not report any scaling behavior in the observed κxx\kappa_{xx} vs. HH.

In the Kitaev magnet α\alpha-RuCl3_3, κxx\kappa_{xx} has been reported to increase 13-fold compared to the value at 7 T, the critical field of its zig-zag state [30]. The ratio is smaller if compared to its zero field value (See Table 1 in the Supplement [23]). The authors are able to fit κab\kappa_{ab} with a modified Callaway model, adding a relaxation term that takes into account the resonance between phonon and magnetic gap, whose size is linear to the applied field. Motivated by this study, We also tried fitting our data at finite field with the modified Callaway model, with spin-phonon resonance terms. However, we have not succeeded in fitting our data with physically meaningful parameters (the exception is the curve at 13 T where ℓ\ell is limited by disorder scattering). In fact, the non-analyticity inherent in the H→0H \to0 limit in Eq.3 rules out a Callaway-type description in the scaling regime in BCAO. Because it is not possible to reproduce the non-analyticity in a Callaway-type formulation, our conclusion is that the empirical fits used in α\alpha-RuCl3_3 cannot be applied to BCAO.


Acknowledgement The thermal transport experiments were supported by the US Department of Energy (Basic Energy Sciences) under contract DE-SC0017863 to N.P.O. An award from the U.S. National Science Foundation (DMR 2011750) supported the materials growth (R.J.C.) and acquisition of cryogenic equipment (N.P.O.). N.P.O. acknowledges support from the Gordon and Betty Moore Foundation EPiQS initiative (grant GBMF9466).

Author contributions J.H., P.C. and N.P.O. conceptualized the experiment. J.H. performed all measurements with initial guidance from P.C. Data in an earlier experiment by T.G. provided substantial insights. The present data were analyzed by J.H. and N.P.O. The crystals were grown and characterized by R.Z. and R.J.C. The manuscript was written by J.H. and N.P.O. with input from all authors.

Competing financial interests

The authors declare no competing financial interests.

Additional Information

Supplementary Materials is available in the online version of the paper.

Correspondence and requests for materials should be addressed to J.H. ([email protected]) or N.P.O. ([email protected]).

References

  1. [1]Banerjee, A., Bridges, C.A., Yan, J.Q., Aczel, A.A., Li, L., Stone, M.B., Granroth, G.E., Lumsden, M.D., Yiu, Y., Knolle, J. and Bhattacharjee, S., Proximate Kitaev quantum spin liquid behaviour in a honeycomb magnet. Nature Materials. 15, 733-740 (2016).DOI
  2. [2]Jackeli, G. and Khaliullin, G. Mott Insulators in the Strong Spin-Orbit Coupling Limit: From Heisenberg to a Quantum Compass and Kitaev Models. Phys. Rev. Lett. 102, 017205 (2009,1).pubmed.ncbi.nlm.nih.gov/19257237
  3. [3]Kitaev, A. Anyons in an exactly solved model and beyond. Annals Of Physics. 321, 2-111 (2006), January Special Issue.DOI
  4. [4]Leahy, I., Pocs, C., Siegfried, P., Graf, D., Do, S., Choi, K., Normand, B. and Lee, M. Anomalous thermal conductivity and magnetic torque response in the honeycomb magnet α-RuCl₃. Physical Review Letters. 118, 187203 (2017).DOI
  5. [5]Czajka, P., Gao, T., Hirschberger, M., Lampen-Kelley, P., Banerjee, A., Yan, J., Mandrus, D., Nagler, S. and Ong, N. Oscillations of the thermal conductivity in the spin-liquid state of α-RuCl₃. Nature Physics. 17, 915-919 (2021).DOI
  6. [6]Kasahara, Y., Ohnishi, T., Mizukami, Y., Tanaka, O., Ma, S., Sugii, K., Kurita, N., Tanaka, H., Nasu, J., Motome, Y. and Others, Majorana quantization and half-integer thermal quantum Hall effect in a Kitaev spin liquid. Nature. 559, 227-231 (2018).
  7. [7]Czajka, P., Gao, T., Hirschberger, M., Lampen-Kelley, P., Banerjee, A., Quirk, N., Mandrus, D., Nagler, S. and Ong, N. Planar thermal Hall effect of topological bosons in the Kitaev magnet α-RuCl₃. Nature Materials. 22, 36-41 (2023).DOI
  8. [8]Liu, H. and Khaliullin, G. Pseudospin exchange interactions in d⁷ cobalt compounds: Possible realization of the Kitaev model. Phys. Rev. B. 97, 014407 (2018,1).DOI
  9. [9]Regnault, L., Burlet, P. and Rossat-Mignod, J. Magnetic ordering in a planar X - Y model: BaCo₂(AsO₄)₂. Physica B+C. 86-88 pp. 660-662 (1977).DOI
  10. [10]Regnault, L., Rossat-Mignod, J., Villain, J. and Combarieu, A. Specific heat of the quasi 2D-XY helimagnets BaCo₂(AsO₄)₂ and BaCo₂(PO₄)₂. Journal De Physique Colloques. 39, C6-759-C6-761 (1978
  11. [11]Regnault, L., Boullier, C. and Lorenzo, J. Polarized-neutron investigation of magnetic ordering and spin dynamics in BaCo₂(AsO₄)₂ frustrated honeycomb-lattice magnet. Helyion. 4, e00507 (2018).DOI
  12. [12]Zhong, R., Gao, T., Ong, N. and Robert J. Cava Weak-field induced nonmagnetic state in a Co-based honeycomb. Science Advances. 6, eaay6953 (2020).
  13. [13]Halloran, T., Desrochers, F., Zhang, E., Chen, T., Chern, L., Xu, Z., Winn, B., Graves-Brook, M., Stone, M., Kolesnikov, A. and Others, Geometrical frustration versus Kitaev interactions in BaCo₂(AsO₄)₂. Proceedings Of The National Academy Of Sciences. 120, e2215509119 (2023).DOI
  14. [14]Maksimov, P., Ushakov, A., Pchelkinа, Z., Li, Y., Winter, S. and Streltsov, S. Ab initio guided minimal model for the “Kitaev” material BaCo₂(AsO₄)₂: Importance of direct hopping, third-neighbor exchange, and quantum fluctuations. Physical Review B. 106, 165131 (2022).DOI
  15. [15]Huyan, S., Schmidt, J., Gati, E., Zhong, R., Cava, R., Canfield, P. and Bud’Ko, S. Hydrostatic pressure effect on the Co-based honeycomb magnet BaCo₂(AsO₄)₂. Physical Review B. 105, 184431 (2022)DOI
  16. [16]Das, S., Voleti, S., Saha-Dasgupta, T. and Paramekanti, A. XY magnetism, Kitaev exchange, and long-range frustration in the J_eff = 1/2 honeycomb cobaltates. Physical Review B. 104, 134425 (2021)DOI
  17. [17]Winter, S. Magnetic couplings in edge-sharing high-spin d 7 compounds. Journal Of Physics: Materials. 5, 045003 (2022)DOI
  18. [18]Mukherjee, P., Shen, B., Erdmann, S., Jesche, A., Kaiser, J., Baral, P., Zaharko, O., Gegenwart, P. and Tsirlin, A. Intermediate field-induced phase of the honeycomb magnet BaCo₂(AsO₄)₂. ArXiv Preprint arXiv:2403.04466. (2024)
  19. [19]Tu, C., Dai, D., Zhang, X., Zhao, C., Jin, X., Gao, B., Dai, P. and Li, S. Evidence for gapless quantum spin liquid in a honeycomb lattice. ArXiv Preprint arXiv:2212.07322. (2022).arxiv.org/abs/2212.07322
  20. [20]Zhang, X., Xu, Y., Halloran, T., Zhong, R., Broholm, C., Cava, R., Drichko, N. and Armitage, N. A magnetic continuum in the cobalt-based honeycomb magnet BaCo₂(AsO₄)₂. Nature Materials. 22, 58-63 (2023).DOI
  21. [21]Bose, A., Routh, M., Voleti, S., Saha, S., Kumar, M., Saha-Dasgupta, T. and Paramekanti, A. Proximate Dirac spin liquid in the honeycomb lattice J₁-J₃ XXZ model: Numerical study and application to cobaltates. Physical Review B. 108, 174422 (2023)DOI
  22. [22]Wolfowicz, G., Heremans, F., Anderson, C., Kanai, S., Seo, H., Galli, A., Galli, G. and Awschalom, D. Quantum guidelines for solid-state spin defects. Nature Reviews Materials. 6, 906-925 (2021).DOI
  23. [23]See Supplemental Information.
  24. [24]Ponomaryov, A., Schulze, E., Wosnitza, J., Lampen-Kelley, P., Banerjee, A., Yan, J., Bridges, C., Mandrus, D., Nagler, S., Kozheukh, A. and Zvyagin, S. Unconventional spin dynamics in the honeycomb-lattice material α-RuCl₃: High-field electron spin resonance studies. Physical Review B. 96, 241107 (2017).DOI
  25. [25]Ponomaryov, A., Zviagina, L., Wosnitza, J., Lampen-Kelley, P., Banerjee, A., Yan, J., Bridges, C., Mandrus, D., Nagler, S. and Zvyagin, S. Nature of magnetic excitations in the high-field phase of α-RuCl₃. Physical Review Letters. 125, 037202 (2020).DOI
  26. [26]Wellm, C., Zeisner, J., Alfonsov, A., Wolter, A., Roslova, M., Isaeva, A., Doert, T., Voyta, M., Büchner, B. and Kataev, V. Signatures of low-energy fractionalized excitations in α-RuCl₃ from field-dependent microwave absorption. Physical Review B. 98, 184408 (2018).DOI
  27. [27]Balz, C., Lampen-Kelley, P., Banerjee, A., Yan, J., Lu, Z., Hu, X., Yadav, S., Takano, Y., Liu, Y., Tennant, D., Lumsden, M., Mandrus, D. and Nagler, S. Finite field regime for a quantum spin liquid in α-RuCl₃. Physical Review B. 100, 060405 (2019).DOI
  28. [28]Zhao, X., Wu, J., Zhao, Z., He, Z., Song, J., Zhao, J., Liu, X., Sun, X. and Li, X. Heat switch effect in an antiferromagnetic insulator Co₃V₂O₈. Applied Physics Letters. 108, 242405 (2016).DOI
  29. [29]Hong, X., Gillig, M., Yao, W., Janssen, L., Kocsis, V., Gass, S., Li, Y., Wolter, A., Büchner, B. and Hess, C. Phonon thermal transport shaped by strong spin-phonon scattering in a Kitaev material Na₂Co₂TeO₆. Npj Quantum Materials. 9, 18 (2024).
  30. [30]Hentrich, R., Wolter, A., Zotos, X., Brenig, W., Nowak, D., Isaeva, A., Doert, T., Banerjee, A., Lampen-Kelley, P., Mandrus, D., Nagler, S., Sears, J., Kim, Y., Büchner, B. and Hess, C. Unusual Phonon Heat Transport in α-RuCl₃: Strong Spin-Phonon Scattering and Field-Induced Spin Gap. Phys. Rev. Lett. 120, 117204 (2018).DOI
  31. [31]Li, Q., Zhao, Z., Fan, C., Zhang, F., Zhou, H., Zhao, X. and Sun, X. Phonon-glass-like behavior of magnetic origin in single-crystal Tb₂Ti₂O₇. Phys. Rev. B. 87, 214408 (2013,6).DOI
  32. [32]Wang, X., Fan, C., Zhao, Z., Tao, W., Liu, X., Ke, W., Zhao, X. and Sun, X. Large magnetothermal conductivity of HoMnO₃ single crystals and its relation to the magnetic-field-induced transitions of magnetic structure. Phys. Rev. B. 82, 094405 (2010,9).DOI
  33. [33]Berggold, M., Baier, J., Meier, D., Mydosh, J., Lorenz, T., Hermberger, J., Balbashov, A., Aliouane, N. and Argyriou, D. Anomalous thermal expansion and strong damping of the thermal conductivity of NdMnO₃ and TbMnO₃ due to 4f crystal-field excitations. Physical Review B—Condensed Matter And Materials Physics. 76, 094418 (2007).DOI
  34. [34]Pocs, C., Leahy, I., Zheng, H., Cao, G., Choi, E., Do, S., Choi, K., Normand, B. and Lee, M. Giant thermal magnetoconductivity in CrCl₃ and a general model for spin-phonon scattering. Phys. Rev. Res. 2, 013059 (2020).DOI

Paper details

vlnc:W_85jwh2sb
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 9

downloads · 0

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

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents