This paper was converted on www.awesomepapers.org from LaTeX by an anonymous user.
Want to know more? Visit the Converter page.

Anisotropic magnetic excitations from single-chirality antiferromagnetic state in Ca-kapellasite

Y. Ihara yihara@phys.sci.hokudai.ac.jp Department of Physics, Faculty of Science, Hokkaido University, Sapporo 060-0810, Japan    K. Arashima Department of Condensed Mattter Physics, Graduate School of Science, Hokkaido University, Sapporo 060-0810, Japan    H. Yoshida Department of Physics, Faculty of Science, Hokkaido University, Sapporo 060-0810, Japan    M. Hirata Institute for Materials Research, Tohoku University, Sendai 980-8577, Japan    T. Sasaki Institute for Materials Research, Tohoku University, Sendai 980-8577, Japan
Abstract

We present a 35Cl NMR study for spin S=1/2S=1/2 perfect kagome antiferromagnet Ca-kapellasite (CaCu3(OH)6Cl2{}_{2}\cdot0.6H2O) with a magnetic transition at T=7.2T^{\ast}=7.2 K. The static magnetic structure in the ground state has been determined to be a chirality-ordered Q=0Q=0 state, which is selected by a finite Dzyaloshinskii-Moriya interaction. The low-energy magnetic excitations in the ordered state are investigated by the nuclear spin-lattice relaxation rate measurement. We detect a weakly temperature dependent contribution in the magnetic fluctuations perpendicular to the kagome plane in addition to the dispersive spin-wave contribution in the kagome plane. The low-energy magnetic excitations from the coplanar spin structure are attributed to the zero mode originating from the flat band in this kagome antiferromagnet.

The ground state of quantum magnets with spin S=1/2S=1/2 has been intensively discussed because the quantum fluctuations destabilize the canonical magnetic ordering and give rise to a new state of matter known as the spin liquid state. balents-Nature464 This prominent quantum effect emerges when the magnetic ordering is suppressed at very low temperatures by the competing interactions among several spins. One commonly recognized example is the geometrical frustration effect for antiferromagnetically interacting spins on a triangular lattice. Magnets with triangle-based spin configuration such as kagome and pyrochlore networks for two-dimensional (2-D) and three-dimensional (3-D) structures are also candidates for experimental realization of the spin liquid state. In fact, the spin liquid state has been observed in herbertsmithite (2-D) shores-JACS127 ; norman-RMP88 ; han-Nature492 ; imai-PRL100 ; fu-Science350 ; jeong-PRL107 and Yb2Ti2O7 (3-D). ross-PRX1 ; chang-natcom3 ; tokiwa-natcom7 A recently found new candidate for a spin liquid material is the Kitaev ferromagnet, for which theory suggests that the spin-orbit coupled anisotropic frustration leads to a spin liquid ground state. kitaev-anphy321 ; nasu-NatPhy12 In the spin liquid state the spin-spin interactions extend to the distance, yet the spins maintain the dynamics. As a result the elemental excitation can be fractionalized and fermionic spinon excitations were observed in α\alpha-RuCl3 sandilands-PRL114 ; kasahara-Nature559 and Ir based magnets singh-PRL108 ; kitagawa-Nature554 . In the case of Yb2Ti2O7 the monopole excitations were observed. tokiwa-natcom7 Experimental observation of fractionalized magnetic excitations is not only striking evidence for the spin liquid state, but also a characteristic potentially applicable to the quantum computation. Although fractional magnetic excitations were predicted only for the spin liquid state without any magnetic ordering, in kagome antiferromagnets low-energy magnetic excitations can be generated even in the magnetically ordered state because of the macroscopic degeneracy left in the ground state.

Magnetic excitations in the ordered state are generally understood as the bosonic magnons which can be perfectly explained by spin-wave theory. yildirim-PRB73 ; matan-PRL96 In the kagome antiferromagnet, however, a flat band known as the zero mode appears at zero energy in the magnon dispersion for a coplanar magnetic state chalker-PRL68 ; harris-PRB45 and its effect on the magnetic properties have not been revealed because the flat band can infinitely populate low-energy excitations. chernyshev-PRB92 As the zero mode is the spin excitations perpendicular to the kagome plane, we could interpret that the spin component perpendicular to the kagome plane is disordered, while the in-plane components are ordered. Experimental input is crucially important to advance understanding about low-energy excitations in S=1/2S=1/2 kagome antiferromagnets, but so far no experimental results have been reported due to the lack of suitable materials that show a regular magnetic ordering on the perfect kagome network.

Refer to caption
Figure 1: (a) Magnetic kagome network of Ca-kapellasite. In the kepellasite family JdJ_{d} across the hexagon is finite in addition to the nearest and next nearest interactions J1J_{1} and J2J_{2}. The short blue arrows are the 𝑫\bm{D} vectors for Dzyaloshinskii-Moriya interaction. Spin configuration for upward and downward 𝑫\bm{D} vectors are shown in (b) and (c). Depending on the 𝑫\bm{D} vector directions, positive (b) and negative (c) chirality spin configurations will be selected.

In this study, we focus on a 2-D kagome antiferromagnet Ca-kapellasite (CaCu3(OH)6Cl2{}_{2}\cdot0.6H2O), yoshida-JPSJ86 which has a magnetically perfect kagome network of S=1/2S=1/2 Cu spins. The magnetism in Ca-kapellasite is understood on the basis of the J1J2JdJ_{1}-J_{2}-J_{d} model, fak-PRL109 ; kermarrec-PRB90 ; boldrin-PRB91 ; bieri-PRB92 ; iqbal-PRB92 in which the long range interaction across the hexagon JdJ_{d} is finite in addition to the nearest and next nearest interactions J1J_{1} and J2J_{2}. [Fig. 1 (a)] Within the family of kapellasite materials colman-CM20 ; colman-CM22 , Ca-kapellasite is the most interesting because of the antiferromagnetic J1J_{1}. The ground states for antiferromagnetic J1J_{1} were theoretically investigated only from the classical limit without quantum fluctuation effects. messio-PRB83 In the phase diagram 3×3\sqrt{3}\times\sqrt{3} and Q=0Q=0 ground states reside very close to each other when JdJ_{d} is small, which is the case for Ca-kapellasite. The ground state in the real material will be selected by further interactions, such as Dzyaloshinskii-Moriya (DM) interaction 𝑫𝑺i×𝑺j\bm{D}\cdot\bm{S}_{i}\times\bm{S}_{j} which is intrinsic for the kagome network. elhajal-PRB66 The DM vector 𝑫\bm{D} perpendicular to the kagome plane stabilizes the coplanar magnetic state [Figs. 1 (b), (c)] opening a gap for the zero mode. chernyshev-PRB92

Previous bulk property measurements yoshida-JPSJ86 revealed a weak magnetic ordering at T=7.2T^{\ast}=7.2 K from the peaks in the in-plane magnetization and the heat capacity. From the 35Cl NMR experiment for a powdered sample a peak in the spin-lattice relaxation rate 1/T11/T_{1} at TT^{\ast} and spectrum broadening below TT^{\ast} were observed as the evidence for static magnetic ordering. ihara-PRB96 In the ordered state, the dispersive magnon excitations were observed as the T2T^{2} term in the heat capacity and the T3T^{3} term in 1/T11/T_{1}. In addition to these magnon contributions Ca-kapellasite has nontrivial magnetic excitations which introduce TT-linear temperature dependence both in the heat capacity and 1/T11/T_{1}. As this temperature dependence is reminiscent of a Fermi liquid state in conductors, one would invoke a fermionic nature for the low-energy excitations.

To identify the origin of nontrivial magnetic excitations, we performed 35Cl NMR experiments on crystalline samples. From the NMR spectra and 1/T11/T_{1} measurements in field applied parallel (H||H_{||}) and perpendicular (HH_{\perp}) to the kagome plane, we found that the ordered state is highly anisotropic due to the coplanar magnetic structure. Moreover, we revealed that the magnetic excitations are also anisotropic and that the weakly temperature dependent contribution was observed only in the cc axis component. We discuss the zero mode in kagome antiferromagnets as a possible origin of the orientation ordered magnetic excitations.

The 35Cl NMR experiment was performed for a single crystal with a typical dimension of 3×2×0.53\times 2\times 0.5 mm3. The sample was mounted on a single axis rotator, with which the field direction can be tuned from the aa axis (H||)H_{||}) to the cc axis (H(H_{\perp}). The NMR spectra were obtained by a single Fourier transformation at fixed field when the linewidth is narrow at high temperatures, and by recombining several Fourier transformations during the field sweep for the broad spectra at low temperatures. The peak frequency/field and their standard error were estimated by fitting the obtained spectrum with a Gaussian function. For the high field NMR experiment, we used the high TcT_{c} superconducting magnet in the Institute for Materials Research, Tohoku University.

Refer to caption
Figure 2: 35Cl NMR shift measured in field perpendicular (blue balls) and parallel (red triangles) to the kagome plane. Open (filled) symbols are Knight shift determined by the field (frequency) sweep NMR spectra. Consistent results were obtained for both methods at the overlapping temperatures. The solid lines are the temperature dependence of uniform susceptibility yoshida-JPSJ86 multiplied by the hyperfine coupling constants. Note that the vertical axis has negative sign because of the negative hyperfine coupling constants. Inset shows the Cl sites. Crystallographically equivalent Cl sites are located above and below the Cu triangles.

Figure 2 shows the temperature dependence of the Knight shift KK. Temperature-independent term was subtracted from the total shift. The solid lines in Fig. 2 are the uniform susceptibility χ\chi multiplied by the hyperfine coupling constants for H||H_{||} and HH_{\perp}, which are A||=267(3)A_{||}=-267(3) mT/μB\mu_{B} and A=370(3)A_{\perp}=-370(3) mT/μB\mu_{B} sup . At high temperatures above 3030 K χ\chi follows KK as expected for a paramagnetic state. At low temperatures KK deviates from χ\chi due to the formation of short-range ordering structure. The in-plane Knight shift K||K_{||} goes down to zero around TT^{\ast}, suggesting a 120120 degrees structure in the ordered state, for which the transferred hyperfine fields are canceled out at the Cl sites on the trigonal axis. (inset of Fig. 2) Remarkably, the out-of-plane Knight shift KK_{\perp} levels off around TT^{\ast} and remains finite even in the ordered state. We can exclude any extrinsic magnetization from impurities and defects as the source of finite KK_{\perp} because such contributions would not shift the peak positions but increase the linewidth of the NMR spectrum. The finite KK_{\perp} evidences that the cc-axis susceptibility is finite and thus the low-energy magnetic excitations survive partially in the antiferromagnetically ordered state. We will discuss the dynamics of this low-energy mode later.

Refer to caption
Figure 3: Temperature dependence of 35Cl NMR spectra obtained in (a) H||H_{||} and (b) HH_{\perp}. Spectra at each temperature are vertically shifted for visibility. Clear spectrum broadening was observed only in H||H_{||}, which leads us to suggest a coplanar magnetic structure. (c) Temperature dependence of the full width at half maximum (FWHM). The onset of spectrum broadening is slightly higher than TT^{\ast} for H||H_{||}.

Anisotropic behavior was also observed in the linewidth. As shown in Fig. 3 (a), (b), the emergent line broadening associated with magnetic transition was observed only in H||H_{||}. The small increase in the linewidth for HH_{\perp} further confirms that the finite cc-axis susceptibility does not originate from magnetic impurities. In the temperature dependence of full width at the half maximum [FWHM, Fig. 3 (c)] the onset of the line broadening was observed slightly above TT^{\ast} in H||H_{||}. The precursor to the magnetic transition is ascribed to the growth in the coherent length for interactiong spins. In the ordered state a broad spectrum without apparent structure was observed. In contrast, in HH_{\perp} magnetic line broadening at the lowest temperature is approximately 1212 mT, which is 1/81/8 of that in H||H_{||}. The finite line broadening in HH_{\perp} can be ascribed to a field misalignment of less than 77 degrees. From these results we can conclude that the internal fields at the Cl site orient parallel to the kagome plane, and its amplitude is distributed.

To explain the reduction of K||K_{||} from 30 K to TT^{\ast} we assume a coplanar 120120 degrees spin structure, where all spins are in the kagome plane. Then, we can consider two different spin configurations; the positive- and negative-chirality spin configurations, depending on the direction of the spin rotation as shown in Fig. 1 (b), (c). In the Q=0Q=0 magnetic structure on the kagome network all the Cu triangles possess the same chirality, whereas in the 3×3\sqrt{3}\times\sqrt{3} state both chirality are arranged alternatively. As the transferred hyperfine fields are canceled for both spin configurations, the spectrum broadening in H||H_{||} is caused purely by dipole fields. The dipole fields at the Cl site have only the cc-axis component for the positive chirality, and in-plane component for the negative chirality. The details of the dipole field calculation are given in the supplemental material. sup As our experimental result unveils that the internal fields appear only along the kagome plane, we conclude that all the spins form the negative-chirality spin configuration. Thus, we suggest that the negative-chirality Q=0Q=0 magnetic structure is realized in Ca-kapellasite.

The chirality of spins on a triangle is selected by the DM interaction, namely the 𝑫\bm{D} vector perpendicular to the triangular plane, elhajal-PRB66 whereas the in-plane 𝑫\bm{D} component introduces a non-coplanar spin structure. The coplanar and negative-chirality magnetic structure suggests that 𝑫\bm{D} is perpendicular to the kagome plane and pointing down as Fig. 1 (c). The thermal Hall effect was also consistently explained by the negative-chirality Q=0Q=0 state. doki-PRL121 The negative-chirality Q=0Q=0 state was also suggested for a sister compound Cd-kapellasite okuma-PRB95 , and YCu3(OH)6Cl3. zorko-PRB100 In Cd-kapellasite the abrupt increase in susceptibility above the magnetic ordering temperature was explained as the precursor to the Q=0Q=0 magnetic state from symmetry considerations. A similar upturn was also observed in Ca-kapellasite. yoshida-JPSJ86

It should be noted that the Q=0Q=0 state has a freedom of global spin rotation. If the spin directions were locked to a certain direction with respect to the external field orientation, only two kinds of antiparallel dipole fields would be generated at the Cl site, which results in a two-peak NMR spectrum. From our results, however, a broad NMR spectrum without any structure was observed as shown in Fig. 3 (a). We thus suggest that Q=0Q=0 state forms a domain to induce distributed internal fields at the Cl sites.

Refer to caption
Figure 4: Temperature dependence of 1/T11/T_{1} in (a) H||H_{||} and (b) HH_{\perp}. For both figures filled (open) symbols are the result obtained in 5.65.6 T (2424 T). TT^{\ast} at each field is determined by the peaks in HH_{\perp} and pointed by arrows. The dashed lines indicate the slope for TT-linear and T3T^{3} temperature dependence. In HH_{\perp} 1/T11/T_{1} shows a peak at TT^{\ast} and T3T^{3} behavior was observed below TT^{\ast}. The temperature dependence is weaker in H||H_{||}, namely at low fields and low temperatures.

Next we measured the nuclear spin-lattice relaxation rate 1/T11/T_{1} to explore the low-energy magnetic excitations in the ordered state. As shown in Fig. 4, 1/T11/T_{1} becomes anisotropic below TT^{\ast}. In HH_{\perp} 1/T11/T_{1} shows a peak at TT^{\ast} due to the critical fluctuations near the magnetic phase transition and a power-law behavior was observed in the ordered state. The exponent close to 33 is explained by the linear spin-wave theory for the Q=0Q=0 state, whereas in H||H_{||} the peak at TT^{\ast} is suppressed and the temperature dependence below TT^{\ast} is weaker than T3T^{3}. We suggest that the weak temperature dependence is due to the additional contribution in H||H_{||}, which corresponds to the TT-linear contribution observed in the powder sample and also in the heat capacity measurement. ihara-PRB96 ; yoshida-JPSJ86

The nuclear spins excited by the rf fields are relaxed by the local magnetic fluctuations perpendicular to the external field direction. The T3T^{3} temperature dependence observed in HH_{\perp} strongly indicates that spin fluctuations in the kagome plane are perfectly understood as the dispersive magnon excitations, while the spin fluctuations along the cc direction have larger weight at lower energy, as sensed by 1/T11/T_{1} in H||H_{||}. We address the zero mode as the origin of the low-energy excitations perpendicular to kagome plane. The finite and temperature independent cc-axis susceptibility is also consistently explained by the zero mode. We should note that the flat band is lifted by the DM interaction, which results in a gapped excitations. The power-law temperature dependence was observed down to 22 K in Ca-kapellasite because the gap by DM interaction is suppressed by J2J_{2} and JdJ_{d} which introduce dispersion to the flat band. chernyshev-PRB92 ; matan-PRL96

In Fig. 4, 1/T11/T_{1} measured at 2424 T is shown together with the result at low field. In HH_{\perp} we observed a peak in 1/T11/T_{1} at TT^{\ast} and T3T^{3} temperature dependence below TT^{\ast} even at 2424 T. The peak temperature increases to 99 K in 2424 T probably due to the weakly ferromagnetic nature of Q=0Q=0 state. To reveal the origin of the increase in TT^{\ast} in high magnetic fields, further experiments are necessary. In H||H_{||} the peak at TT^{\ast} is strongly suppressed, and the T3T^{3} behavior appears in a narrow temperature range just below TT^{\ast}. Then the deviation from T3T^{3} behavior was observed below 55 K. From these results we suggest that the zero mode is more strongly suppressed by fields compared with the dispersive spin-wave contribution because of the small energy scale relevant to the flat band. The heat capacity measurement also suggests that the TT-linear contribution is suppressed by fields of about 1010 T. yoshida-JPSJ86

Refer to caption
Figure 5: Real space images of zero modes in (a) 3×3\sqrt{3}\times\sqrt{3} and (b) Q=0Q=0 structures. The spin excitations are localized around a hexagon in 3×3\sqrt{3}\times\sqrt{3} structure, while they form a chain in Q=0Q=0 structure.

The zero mode from the 3×3\sqrt{3}\times\sqrt{3} state is understood as the local excitations of six spins surrounding one hexagon. harris-PRB45 [Fig. 5 (a)] Such local excitation without any long range propagation will contribute to a temperature independent 1/T11/T_{1} as in the case of free spins. In contrast, the zero mode from the Q=0Q=0 state forms a chain structure [Fig. 5 (b)], which would introduce a power-law temperature dependence defined by the evolution of the coherence length. We propose that the low-energy excitations in Ca-kapellasite can be understood as the superposed one-dimensional chains weakly connected by long range interactions J2J_{2}, JdJ_{d}. Further experiment is required to directly observe the low-energy magnon dispersion, and possibly additional continuum excitations originating from weakly dispersing low-energy band.

To summarize, we found the anisotropic spectrum broadening in the ordered state of Ca-kapellasite, which is consistently explained as the negative-chirality Q=0Q=0 state. From the 1/T11/T_{1} measurement in H||H_{||} and HH_{\perp} we conclude that the in-plane component of the low-energy magnetic excitations are dominated by the spin-wave contribution. Moreover, we unveiled weaker temperature dependence for the magnetic excitations along the cc axis. These nontrivial excitations are consistent with the finite susceptibility along cc axis and also TT-linear contribution to the heat capacity. We suggest that these excitations can be generated from the flat band in the kagome antiferromagnet when it obtains a weak dispersion due to long range interactions.

Acknowledgements.
We would like to acknowledge Jun Ohara and Kazuki Iida for fruitful discussions. Part of this work was performed at the High Field Laboratory for Superconducting Materials, Institute for Materials Research, Tohoku University (Project Nos. 17H0046, 18H0015, 19H0037). This study was partly supported by the Grant-in-Aid for Young Scientist B Grant Number JP15K17686 and JSPS KAKENHI Grant Numbers JP18K03529, JP19H01832, and JP19H01846.

References

  • (1) L. Balents, Nature 464, 199-208 (2010).
  • (2) M. P. Shores, E. A. Nytko, B. M. Bartlett, D. G. Nocera, J. Am. Chem. Soc. 127, 13462 (2005).
  • (3) T.-H. Han, J. S. Helton, S. Chu, D. G. Nocera, J. A. Rodriguez-Rivera, C. Broholm, and Y. S. Lee, Nature 492, 406-410 (2012).
  • (4) M. R. Norman, Rev. Mod. Phys. 88, 041002 (2016).
  • (5) T. Imai, E. A. Nytko, B. M. Bartlett, M. P. Shores, and D. G. Nocera, Phys. Rev. Lett. 100, 077203 (2008).
  • (6) M. Fu, T. Imai, T.-H. Han, and Y.-S. Lee, Science 350, 655 (2015).
  • (7) M. Jeong, F. Bert, P. Mendels, F. Duc, J. C. Trombe, M. A. de Vries, and A. Harrison, Phys. Rev. Lett. 107, 237201 (2011).
  • (8) K. A. Ross, L. Savary, B. D. Gaulin, and L. Balents, Phys. Rev. X 1, 021002 (2011).
  • (9) L.-J. Chang, S. Onoda, Y. Su, Y.-J.  Kao, K.-D. Tsuei, Y. Yasui, K. Kakurai, and M. R. Lees, Nat. Commun. 3, 992 (2012).
  • (10) Y. Tokiwa, T. Yamashita, M. Udagawa, S. Kittaka, T. Sakakibara, D. Terazawa, Y. Shimoyama, T. Terashima, Y. Yasui, and Y. Matsuda, Nat. Commun. 7, 10807 (2016).
  • (11) A. Kitaev, Ann. Phys. 321, 2-111 (2006).
  • (12) J. Nasu, J. Knolle, D. L. Kovrizhin, Y. Motome, R. Moessner, Nat. Phys. 12 912 (2016).
  • (13) L. J. Sandilands, Y. Tian, K. W. Plumb, Y.-J. Kim, and K. S. Burch, Phys. Rev. Lett. 114, 147201 (2015).
  • (14) Y. Kasahara, T. Ohnishi, Y. Mizukami, O. Tanaka, S. Ma, K. Sugii, N. Kurita, H. Tanaka, J. Nasu, Y. Motome, T. Shibauchi, and Y. Matsuda, Nature 559, 227 (2018).
  • (15) Y. Singh, S. Manni, J. Reuther, T. Berlijn, R. Thomale, W. Ku, S. Trebst, and P. Gegenwart, Phys. Rev. Lett. 108, 127203 (2012).
  • (16) K. Kitagawa, T. Takayama, Y. Matsumoto, A. Kato, R. Takano, Y. Kishimoto, S. Bette, R. Dinnebier, G. Jackeli, and H. Takagi, Nature 554, 341 (2018).
  • (17) T. Yildirim, and A. B. Harris, Phys. Rev. B 73, 214446 (2006).
  • (18) K. Matan, D. Grohol, D. G. Nocera, T. Yildirim, A. B. Harris, S. H. Lee, S. E. Nagler, and Y. S. Lee, Phys. Rev. Lett. 96, 247201 (2006).
  • (19) J. T. Chalker, P. C. W. Holdsworth, and E. F. Shender, Phys. Rev. Lett 68, 855 (1992).
  • (20) A. B. Harris, C. Kallin, and A. J. Berlinsky, Phys. Rev. B 45, 2899 (1992).
  • (21) A. L. Chernyshev, and M. E. Zhitomirsky, Phys. Rev. B 92, 144415 (2015).
  • (22) H. Yoshida, N. Noguchi, Y. Matsushita, Y. Ishii, Y. Ihara, M. Oda, H. Okabe, S. Yamashita, Y. Nakazawa, A. Takata, T. Kida, Y. Narumi, and M. Hagiwara, J. Phys. Soc. Jpn. 86, 033704 (2017).
  • (23) B. Fak, E. Kermarrec, L. Messio, B. Bernu, C. Lhuillier, F. Bert, P. Mendels, B. Koteswararao, F. Bouquet, J. Ollivier, A. D. Hillier, A. Amato, R. H. Colman and A. S. Wills, Phys. Rev. Lett. 109, 037208 (2012).
  • (24) E. Kermarrec, A. Zorko, F. Bert, R. H. Colman, B. Koteswararao, F. Bouquet, P. Bonville, A. Hillier, A. Amato, J. van Tol, A. Ozarowski, A. S. Wills, and P. Mendels, Phys. Rev. B 90, 205103 (2014).
  • (25) D. Boldrin, B. Fak, M. Enderle, S. Bieri, J. Ollivier, S. Rols, P. Manuel, and A. S. Wills, Phys. Rev. B 91, 220408(R) (2015).
  • (26) S. Bieri, L. Messio, B. Bernu, and C. Lhuillier, Phys. Rev. B 92, 060407(R) (2015).
  • (27) Y. Iqbal, H. O. Jeschke, J. Reuther, R. Valentí, I. I. Mazin, M. Greiter, and R. Thomale, Phys. Rev. B 92, 220404(R) (2015).
  • (28) R. H. Colman, C. Ritter, and A. S. Wills, Chem. Mater. 20, 6897 (2008).
  • (29) R. H. Colman, R. Sinclair, and A. S. Wills, Chem. Mater. 22, 5774 (2010).
  • (30) L. Messio, C. Lhuillier, and G. Misguich, Phys. Rev. B 83, 184401 (2011).
  • (31) M. Elhajal, B. Canals, and C. Lacroix, Phys. Rev. B 66, 014422 (2002).
  • (32) Y. Ihara, T. Sasaki, N. Noguchi, Y. Ishii, M. Oda, and H. Yoshida, Phys. Rev. B 96, 180409(R) (2017).
  • (33) See Supplemental Material [url].
  • (34) H. Doki, M. Akazawa, H.-Y. Lee, J. H. Han, K. Sugii, M. Shimozawa, N. Kawashima, M. Oda, H. Yoshida, and M. Yamashita, Phys. Rev. Lett. 121, 097203 (2018).
  • (35) R. Okuma, T. Yajima, D. Nishio-Hamane, T. Okubo, and Z. Hiroi, Phys. Rev. B 95, 094427 (2017).
  • (36) A. Zorko, M. Pregelj, M. Gomilšek, M. Klanjšek, O. Zaharko, W. Sun, and J. -X. Mi, Phys. Rev. B 100, 144420 (2019).