Distinct magnetic ground states of R_2ZnIrO_6 (R = La and Nd) determined by neutron powder diffraction
aa r X i v : . [ c ond - m a t . s t r- e l ] F e b Distinct magnetic ground states of R ZnIrO ( R = La and Nd) determined by neutron powderdi ff raction H. Guo, ∗ C. Ritter, Y. Su, A. C. Komarek, and J. S. Gardner Neutron Science Platform, Songshan Lake Materials Laboratory, Dongguan, Guangdong 523808, China Institut Laue-Langevin, Boite Postale 156X, F-38042 Grenoble Cedex 9, France J¨ulich Centre for Neutron Science (JCNS) at Heinz Maier-Leibnitz Zentrum (MLZ),Forschungszentrum J¨ulich, Lichtenbergstrasse 1, D-85747 Garching, Germany Max-Planck-Institute for Chemical Physics of Solids, N¨othnitzer Str. 40, D-01187 Dresden, Germany (Dated: February 3, 2021)Double perovskite iridates A ZnIrO ( A = alkaline or lanthanide) show complex magnetic behaviors rangingfrom weak ferromagnetism to successive antiferromagnetic transitions. Here we report the static ( dc ) and dy-namic ( ac ) magnetic susceptibility, and neutron powder di ff raction measurements for A = La and Nd compoundsto elucidate the magnetic ground state. Below 10 K, the A = La compound is best described as canted iridiummoments in an antiferromagnet arrangement with a propagation vector k = c -axis. On the other hand, Nd ZnIrO is described well as an antiferromagnet with a propagationvector k = (1 / / T N ∼
17 K. Scattering from both the Nd and Ir magnetic sublattices were requiredto describe the data and both were found to lie almost completely within the ab -plane. Dc susceptibility revealeda bifurcation between the zero-field-cooled and field-cooled curves below ∼
13 K in Nd ZnIrO . A glassy statewas ruled out by ac susceptibility but detailed magnetic isotherms revealed the opening of the loop below 13 K.These results suggest a delicate balance exists between the Dzyaloshinskii-Moriya, crystal field schemes, and d - f interaction in this series of compounds. Double perovskite with the general formula A B’B” O forms an ordered rock-salt-like structure when the B’ and B” ions are physically and / or chemically significantly di ff erent.In general, the A site is occupied by an alkaline or lanthanidecation and B’ / B” are transition-metal elements. This struc-ture is incredibly versatile, hosting a wide variety of ionson the A , B’ , and B” sites. The interplay between crystalstructure, orbital and spin degrees of freedom, and electron-lattice coupling leads to a diverse range of physical propertiesfrom metal-insulator transition [1] to multiferroicity [2], andlarge magnetoresistance e ff ect [3] to photolysis [4]. For a sin-gle magnetic B’ / B” -site compound, the superexchange cou-pling between two nearest cations through intermediate oxy-gen takes part in the magnetic order. Generally, when the A site is occupied by a magnetic cation, the trivalent lanthanideorders at a lower temperature than the transition-metal sublat-tice [5, 6]. Recently, double perovskites with Ir ions at the B” site have attracted much attention [5, 7–13] due to the rela-tively large spin-orbit coupling (SOC) of the Ir + ion whichmay lead to novel phenomena such as SOC-driven Mott insu-lator [14] and complex magnetism [15–17]. Some A B’ IrO have been synthesized but the magnetic ground states are stillnot well understood. La MgIrO shows antiferromagnetic be-havior below T N =
12 K [7]. Isovalent substitution of Znfor Mg results in a di ff erent magnetic ground state; wherethe magnetic susceptibility increases substantially below 10K and hysteresis is observed in magnetic isothermals belowthis temperature, suggesting a ferromagnetic component to themagnetic structure [7]. More recent studies on small singlecrystals suggested a canted antiferromagnetic state with a netmoment along the crystallographic b -axis [18]. Introducingrare earth moments on the A site gives rise to complex mag-netic properties [12], and in Nd ZnIrO two transitions have been observed in the magnetic susceptibility below ∼
17 K.Preliminary neutron powder di ff raction (NPD) measurementsexist for these compounds, but a reliable model to explain thematerials with two magnetic sublattices has so far eluded sci-entists. The relatively large neutron absorption by Ir ions andtheir comparatively small magnetic moment limits the statis-tics and reduces the uniqueness of models.In this paper, we have performed NPD measurements, aswell as bulk dc and ac magnetic susceptibility measurementson polycrystalline La ZnIrO and Nd ZnIrO in order to in-vestigate the static magnetic ground state. La ZnIrO formsa canted antiferromagnetic structure where the moments areconstrained within the bc -plane with a finite ferromagneticcomponent along the long c -axis. This result di ff ers from theearlier neutron di ff raction and first principle calculations thatdid not detect a ferromagnetic component [7]. Our results aresimilar to, but also inconsistent with, the single crystal mag-netisation data that found the easy axis along the crystallo-graphic b -axis [18]. Substituting the nonmagnetic La + withthe magnetic Nd + , we found long range antiferromagneticorder below 17 K in Nd ZnIrO . Data at 1.8 K is describedwell as an antiferromaget with the moments (Nd and Ir) lyingalmost exactly within the ab -plane and with no net moment.Interestingly, only one transition was detected in both NPDand ac susceptibility measurements at T N ∼
17 K, while twoare seen in the bulk static susceptibility. The distinct groundstates of these two double perovskites reveal the delicate bal-ance between the Dzyaloshinskii-Mariya interaction and d - f superexchange interaction within these compounds.Polycrystalline samples of La ZnIrO and Nd ZnIrO wereprepared by solid state reaction method. Stoichiometricamounts of raw materials of La O , Nd O , ZnO and IrO were mixed and ground thoroughly in an agate mortar. Thepowders were then pressed into pellets and sintered between900 ◦ C and 1100 ◦ C in air for about 20 days with several inter-mediate grindings. Phase purity was checked by x-ray di ff rac-tion (XRD) and NPD measurements. Magnetization mea-surements were performed using the vibrating sample mag-netometer (MPMS3, Quantum Design). AC susceptibilitywas measured in the Physical Property Measurement Sys-tem (PPMS, Quantum Design) with an ac excitation fieldof 5 Oe. NPD measurements were performed at the In-stitut Laue-Langevin (ILL) in Grenoble, France. High Q-resolution measurements, for crystal structure determination,were performed on the D2B di ff ractometer with 1.594 Å neu-trons selected by a Ge(335) monochromator, while the D20di ff ractometer produced a higher flux beam of 2.394 Å neu-trons, albeit with lower resolution suitable for magnetic struc-ture refinements. Approximately 7-g of samples were loadedinto annular vanadium cans, to reduce the neutron flight pathwithin the sample, overcoming the high neutron absorptionfrom naturally abundant Ir. The La ZnIrO sample was mea-sured on the D20 di ff ractometer for 10.25 h at 20 K and 1.8 K.Due to the stronger magnetic signal from the Nd ZnIrO sam-ple, 2-hour measurements at (1.8, 13.5, 15.1 and 19.3) K, wereperformed. In addition, a series of 15-min measurements werecarried out while ramping the temperature at 0.1 K / / ZnIrO sample, but a small amount ofNd Ir O (about 0.1 wt %) was found in the Nd ZnIrO sam-ple. Refinement of the crystal structure confirms both samplescrystallize into a monoclinic structure with space group P / n ( B -site disorder. In La ZnIrO ,this amounts to between 6(4)% from the D2B measurementand 13(4)% from the D20 measurement, as will be discussedlater. Similar antisite disorder was found to be negligible inNd ZnIrO .The temperature dependence of the magnetic susceptibil-ity and isothermal magnetization measurements are shown inFig. 2. For La ZnIrO , the magnetic susceptibility increasesabruptly below 10 K, concomitant with a splitting between thezero-field-cooled (ZFC) and field-cooled (FC) magnetisationcurves and hysteresis in the isothermal magnetization loop,suggesting a weak ferromagnetic component in the groundstate. On the other hand, the Nd ZnIrO sample shows a morecomplicated path to the ground state. A broad peak can befound at 17 K below which long range antiferromagnetic or-der is observed, but a second magnetic phase is revealed inthe splitting between the ZFC and FC magnetization curves
20 40 60 80 100 120 (b) I obs. I cal. I diff. peak pos. Nd ZnIrO I obs. I cal. I diff. peak pos. I n t e n s it y ( a r b . u . ) ZnIrO (a) *20 40 60 80 100 120 I n t e n s it y ( a r b . u . )
2 (deg.)
FIG. 1. (Color online) Rietveld refined patterns from R ZnIrO . Thishigh resolution data was collected at room temperature on the D2Bdi ff ractometer. The lattice parameters are a = b = c = β = ◦ for R = La sample and a = b = c = β = ◦ forthe R = Nd sample. A small peak originated from the cryofurnacehas been excluded from the refinement in (b), and the asterisk marksthe peak originated from the Nd Ir O phase. below 13 K. Note that the magnitude of the susceptibility inNd ZnIrO is much smaller than that of the La ZnIrO sam-ple, although the net magnetic moment is larger (see Fig. 2(a2) and (b2)) due to the much larger Nd + moment. Thesebulk results are consistent with those reported in the literature[7, 12]. Temperature dependence of the real component of thedynamic susceptibility of Nd ZnIrO is shown in the inset ofFig. 2(b1) when measured in a zero DC field. These data re-veal no dependence on excitation frequency excluding a pos-sible spin-glass like origin to the 13 K anomaly in the staticmagnetization similar to that seen in the double perovskiteLa CoIrO by Song et al. [21]. In order to study the magneticphase below 13 K further, careful magnetic isotherms werecollected around this temperature up to 7 T. Unlike La ZnIrO (Fig. 2(a2)) where an antiferromagnet develops a ferromag-netic component and a loop opens up around the origin belowthe critical temperature, in Nd ZnIrO below the high tem-perature transition no gap is seen in the magnetisation loops, (b2)(b1)(a2)(a1) La ZnIrO Nd ZnIrO FC ( e m u / m o l O e ) T (K)
ZFC -8 -6 -4 -2 0 2 4 6 8-0.6-0.30.00.30.6
20 K M ( B / f . u . ) H (T) FC ( e m u / m o l O e ) T (K)
ZFC -8 -6 -4 -2 0 2 4 6 8-2-1012
14 K 20 K M ( B / f . u . ) H (T) I n t e n s it y ( a r b . u . ) ’ ( e m u / m o l ) T (K) -0.05 0.00 0.05-0.0020.0000.002 0 5 10 15 200.000.020.04 H C ( T ) T (K)
FIG. 2. (Color online) Temperature dependence of the magnetic sus-ceptibility and isothermal magnetization measurements for (a1, a2)La ZnIrO and (b1, b2) Nd ZnIrO . The susceptibility was mea-sured with a field of 1000 Oe. The temperature dependence of themagnetic peak intensity, measured on D20, is also shown in (b1)where the 13 K transition is absent. The inset of (b1) shows the realcomponent ( χ ′ ) of the zero field ac susceptibility with frequenciesof 13, 113, 1333, 5330 and 9918 Hz. The left inset of (b2) is anenlargement of the low field region measured at 2 (blue), 8 (cyan),10 (magenta), 12 (wine) and 14 (green) K. The right inset shows thetemperature dependence of the coercive field ( H c ). consistent with a simple antiferromagnet. However, below thesecond transition at 13 K, the loops take on a ”S” shape, typ-ical of a metamagnet and on closer inspection ( upper insetFig. 2(b2)) a small gap opens up around the origin, suggestiveof a ferromagnetic component to the ordering, similar to thatin La ZnIrO .In order to determine the underlying magnetic structure forboth compounds, we have performed NPD measurements onthe high flux D20 di ff ractometer at the ILL in the absence of amagnetic field. Atomic structure refinements were performedat 20 K in the paramagnetic state, allowing the scale factorsto be determined and fixed for subsequent magnetic structurerefinements. All measurements were performed after coolingin zero field. We first concentrate on the La ZnIrO sample,where the magnetic peaks can be indexed with a k = / Ir mixing( B -site disorder) previously observed in the crystallographicdata collected on D2B. Magnetic symmetry analysis for theIr1 moment (2 c site) and the cation disordered Ir2 momenton the Zn position (2 d site) with the monoclinic space group P / n shows that the reducible magnetic representation is de-composed into two IRs as: Γ = Γ + Γ . The basis vectors(BV) for each IR are shown in Supplementary Materials (SM)Tab. S1. The magnetic intensities can be well described by IR Γ . As can be seen from Fig. 3, a B -site ordered model resultsin a poorer fit and R B value compared to the B -site disorderedmodel. Note that in order to have the same magnitude of mag-netic moment at the Ir1 and Ir2 atoms, the degree of disorder B site disorderedR B = 4.56%(b) I n t e n s it y ( c oun t s / h ) (a) B site orderedR B = 28.6%
10 20 30 400123 I n t e n s it y ( c oun t s / h )
2 (deg.)
FIG. 3. (Color online) Rietveld refinement of the magnetic di ff rac-tion from La ZnIrO at 1.8 K. A paramagnetic data set, taken at 20 Khas been subtracted to remove the scattering from the crystal lattice.The calculated patterns are according to the IR Γ with (a) a perfectlyordered B -site model and (b) a B -site disordered model where thebest model found 13(4)% site substitution between Zn and Ir ions.Red dot: experimental data; Black curve: calculated data; Verticalbars: magnetic peak position; Blue curve: di ff erence between theexperimental and calculated data. obtained from the magnetic refinement is ∼ µ B / Ir at 1.8 K and they arelying predominately within the bc -plane with µ b = µ B and µ c = µ B at B” and µ b = µ B and µ c = µ B for the Ir that has accidentally substituted for Znon the B’ site. As can be seen, there is a canting of these spinsresulting in a net moment along the c -axis, which explainsthe ferromagnetic component observed by bulk magnetic sus-ceptibility measurements, see Fig. 2(a2) at base temperature.Single crystal magnetization measurements led Han et al . toconclude that the b -axis was the easy axis [18], which wouldcorrespond to the IR Γ (see SM Tab. S1). However, the IR Γ does not describe our neutron data, see Fig. S1 in the SM.Also incompatible with our data, Cao et al . suggested that themoments lie within the ab -plane by LDA + U calculations [7],which need to be revisited in light of our conclusions.As mention above, the static magnetic susceptibility sug-gests a di ff erent magnetic ground state for Nd ZnIrO com-pared to La ZnIrO . Additional magnetic peaks can be ob-served below T N ∼
17 K, which can be indexed with a propa-
10 20 30 40 50 60 70-250255075100125 (c)(b)
Nd, Ir orderingNd, Ir orderingNd, Ir orderingNd orderingNd orderingNd, Ir orderingR B = 9.18%Nd, Ir orderingR B = 6.02% Nd orderingT = 15.1 KT = 13.5 K I n t e n s it y ( c oun t s / h )
2 (deg.)
T = 1.8 KNd, Ir orderingR B = 5.02% (a)
20 30 40 50 60
2 (deg.)
10 20 30 40 50 60 7002550 I n t e n s it y ( c oun t s / h )
2 (deg.)
20 30 40 50 60
2 (deg.)
10 20 30 40 50 60 70-50510152025 I n t e n s it y ( c oun t s / h )
2 (deg.)
20 30 40 50 60
2 (deg.)
FIG. 4. (Color online) Rietveld magnetic refinement of the di ff er-ence pattern at various temperatures for Nd ZnIrO according to IR Γ + Γ . The left panels show the refinement results with perfectlyordered Nd and Ir sublattices, and the right panels highlights an ex-panded region of low peak intensities comparing our best model andthat with only the Nd sublattice ordering. Red dot: experimentaldata; Black curve: calculated data; Vertical bars: magnetic peak po-sition; Blue curve: di ff erence between the experimental and calcu-lated data. A small angular range around 2 Θ = ◦ was excludedfrom the fits due to contamination from the atomic scattering that didnot subtract well. gation vector k = (1 / / Θ ∼ ◦ monotonically increases as the temper-ature is lowered through T N and no anomaly is seen at 13 K,where bulk magnetic susceptibility reveals a splitting of theZFC and FC curves, see Fig. 2(b1).Symmetry analysis for the Nd ions at the 4 e site and Ir ionsat the 2 c site with the propagation vector k = (1 / / Γ Ir = Γ + Γ (1) Γ Nd = Γ + Γ + Γ + Γ (2)The corresponding BVs are listed in SM Tab. S2 and S3.As pointed out by an earlier report [12], half of the magneticmoments are not ordered for either of these IRs, and a combi-nation of the two IRs is required in order to form a fully or-dered state. In addition, if there exist a coupling between the (c)(b) Ir2Ir1 aacb cb Ir acb cb a Nd (a) (d) FIG. 5. (Color online) Magnetic structure of (a, b) La ZnIrO and(c, d) Nd ZnIrO viewed along di ff erent directions. The Ir momentshave been enlarged for clarity. Note that the smaller Ir2 moments in(a, b) is visualized taking into account the partial occupation e ff ect. two magnetic sublattices and the magnetic transition is secondorder in character [22] both sublattices should order under thesame IR. An earlier report [12] determined the ordering ofthe Nd + moments, but Vogl et al . was not able to resolve theordering of the Ir + moments due to insu ffi cient statistics. Us-ing the high flux di ff ractometer, D20, at the ILL we are ableto solve the magnetic structure of this compound which mustinclude static moments on both the Nd and Ir sublattices todescribe the data well. Consistent with the high resolutionneutron di ff raction data, no antisite disorder was required todescribe these data.As can be seen from SM Tab. S2, only one solution, i.e., Γ + Γ is possible. The best refinements with Nd and Ir sub-lattices ordering under Γ + Γ are shown in Fig. 4. Focussingon the low intensity peaks in the right panels of Fig. 4, it isclear that a model with only the Nd sublattice ordering can-not account for the peak intensities around 2 Θ ∼ ◦ . This ismore apparent close to T N where the size of both the Nd andIr moments are expected to be small and comparable, produc-ing a more significant interference e ff ect, as observed in thepyrochlore iridate Nd Ir O [16]. The refined magnetic mo-ments at 1.8 K are listed in Tab. I, and the magnetic structureis shown in Fig. 5(c,d). The moments for both Nd and Ir ionsprefer to lie within the ab -plane, in contract to the La ZnIrO case.The value of the Nd moment is much smaller than the freeion value ( g J = / J = /
2) presumably due to the crystalelectric field e ff ect as observed in similar compounds [23]. Itcan be seen in Tab. I that although both magnetic sublatticesdevelop concomitantly, the Nd moment at 15.1 K (2 K be-low the transition temperature) is only 39% of that at 1.8 K,while the Ir moment is already 72% of the low temperaturevalue by 15.1 K. This observation suggests the Ir sublatticeis the driving force of the magnetic ordering. The obtainedantiferromagnetic structure has zero net moment along anycrystallographic directions, thus, the origin of the weak ferro-magnetic component seen in static magnetic susceptibility andFC / ZFC cooled splitting measurements is still a puzzle. Thisbulk susceptibility character seems to be ubiquitous in the iri-dates. The pyrochlore iridates show all-in / all-out magneticstructure with zero net moment, whereas the static magneticsusceptibility also show the bifurcation below T N , and was at-tributed to the formation of antiferromagnetic domain walls[24–26]. One di ff erence is that the bifurcation occurs at T N for the pyrochlore iridate, while it appears at lower tempera-ture for Nd ZnIrO . Another possibility is a subtle change ofthe magnetic symmetry at this temperature which could resultin a small net moment, this was not detected in our currentstudy, but may require higher resolution, single crystal mag-netic structural refinement studies. The former scenario wouldcorrespond to a phase separation, while the latter would occurwithin one phase. Future studies using local probe would behelpful to elucidate this point.To conclude we have determined with the aid of neutronpowder di ff raction and symmetry analysis the long range mag-netic structure of La ZnIrO , albeit with approximately dis-order on the B -site. This simple k = c -axis. Further studies are needed to elucidatethe role of site disorder in these findings. Adding Nd + witha magnetic moment to the rare-earth site results in a signif-icant change to the chemistry and bulk magnetic properties.No crystallographic disorder was found in this sample, andall moments are predominantly in the ab -plane. The iridiumsublattice was found to drive the magnetic transition, but the5 d -4 f interactions are clearly important, increasing the order-ing temperature by 70%. Moreover, the single ion anisotropyof the Nd + ion probably governing the spin direction, in thiscase rotating the preferred plane from the bc -plane to the ab -plane.Double perovskites have been shown to be an excellent playground to investigate the interplay of spin-lattice and elec-tron degrees of freedom. With the wide array of chemicaldiversity many double perovskites have also been shown to betechnologically relevant. We hope this study motivates oth-ers to study Ir + based double perovskites where spin-orbitcoupling, on-site Coulomb interaction, and crystal field ener-gies are of comparable energy. For example in Nd ZnIrO theisothermal magnetization curve at 2 K has the signature of ametamagnetic transition at about 2 T which should be studiedin more detail, but this is true for many of the iridate baseddouble perovskites[5]. ∗ [email protected][1] H. Kato, T. Okuda, Y. Okimoto, Y. Tomioka, K. Oikawa,T. Kamiyama, and Y. Tokura, Phys. Rev. B , 144404 (2002). TABLE I. Moment size for the Nd and Ir ions at di ff erent tempera-tures according to the best refinement assuming both sublattices or-der under the same IR Γ + Γ . The unit of the moment is µ B . µ a µ b µ c | µ | µ a -0.47(5) -0.41(4) -0.37(5) µ b -0.53(4) -0.40(3) -0.36(3) µ c | µ | , 135 (2011).[3] K. I. Kobayashi, T. Kimura, H. Sawada, K. Terakura, andY. Tokura, Nature , 677 (1998).[4] J. Luo, J.-H. Im, M. T. Mayer, M. Schreier, M. K. Nazeerud-din, N.-G. Park, S. D. Tilley, H. J. Fan, and M. Gr¨atzel,Science , 1593 (2014).[5] X. Ding, B. Gao, E. Krenkel, C. Dawson, J. C. Eckert, S.-W.Cheong, and V. Zapf, Phys. Rev. B , 014438 (2019).[6] J. S´anchez-Ben´ıtez, M. J. Mart´ınez-Lope,J. A. Alonso, and J. L. Garc´ıa-Mu˜noz,J. Phys.: Condens. Matter , 226001 (2011).[7] G. Cao, A. Subedi, S. Calder, J. Q. Yan, J. Yi, Z. Gai, L. Poudel,D. J. Singh, M. D. Lumsden, A. D. Christianson, B. C. Sales,and D. Mandrus, Phys. Rev. B , 155136 (2013).[8] W. K. Zhu, C.-K. Lu, W. Tong, J. M. Wang, H. D. Zhou, andS. X. Zhang, Phys. Rev. B , 144408 (2015).[9] A. A. Aczel, A. M. Cook, T. J. Williams, S. Calder, A. D.Christianson, G. X. Cao, D. Mandrus, Y.-B. Kim, andA. Paramekanti, Phys. Rev. B , 214426 (2016).[10] M. Vogl, L. T. Corredor, T. Dey, R. Morrow, F. Scaravaggi,A. U. B. Wolter, S. Aswartham, S. Wurmehl, and B. B¨uchner,Phys. Rev. B , 035155 (2018).[11] B. Singh, M. Vogl, S. Wurmehl, S. Aswartham, B. B¨uchner,and P. Kumar, Phys. Rev. Research , 023162 (2020).[12] M. Vogl, R. Morrow, A. A. Aczel, R. B. Rodriguez, A. U. B.Wolter, S. Wurmehl, S. Aswartham, and B. B¨uchner,Phys. Rev. Materials , 054413 (2020).[13] Y. Gao, Z. Tian, L. Xu, M. Ashtar, Z. Wan,Z. Xia, F. Yang, S. Yuan, Y. Han, and W. Tong,J. Phys.: Condens. Matter , 465802 (2020).[14] B. J. Kim, H. Jin, S. J. Moon, J. Y. Kim, B. G. Park,C. S. Leem, J. Yu, T. W. Noh, C. Kim, S. J. Oh,J. H. Park, V. Durairaj, G. Cao, and E. Rotenberg,Phys. Rev. Lett. , 076402 (2008).[15] H. Guo, K. Matsuhira, I. Kawasaki, M. Wakeshima, Y. Hinatsu,I. Watanabe, and Z.-a. Xu, Phys. Rev. B , 060411(R) (2013).[16] H. Guo, C. Ritter, and A. C. Komarek,Phys. Rev. B , 161102(R) (2016).[17] H. Guo, C. Ritter, and A. C. Komarek,Phys. Rev. B , 144415 (2017).[18] H. Han, W. Liu, Y. Dai, Y. Gao, Z. Tian, J. Fan,S. Zhou, L. Pi, C. Zhang, L. Zhang, and Y. Zhang,Phys. Rev. B , 054403 (2018).[19] J. Rodr´ıguez-Carvajal, Physica B , 55 (1993).[20] A. Wills, Physica B , 680 (2000). [21] J. Song, B. Zhao, L. Yin, Y. Qin, J. Zhou, D. Wang, W. Song,and Y. Sun, Dalton Transactions , 11691 (2017).[22] E. Bertaut, Acta Cryst. , 217 (1968).[23] A. A. Aczel, D. E. Bugaris, J. Yeon, C. de la Cruz, H. C. zurLoye, and S. E. Nagler, Phys. Rev. B , 014413 (2013).[24] K. Matsuhira, M. Wakeshima, Y. Hinatsu, and S. Takagi, J.Phys. Soc. Jpn. , 094701 (2011).[25] E. Y. Ma, Y.-T. Cui, K. Ueda, S. Tang, K. Chen, N. Tamura,P. M. Wu, J. Fujioka, Y. Tokura, and Z.-X. Shen,Science , 538 (2015).[26] H. T. Hirose, J.-i. Yamaura, and Z. Hiroi,Sci. Rep. , 42440 (2017). SUPPLEMENTAL MATERIAL
Magnetic structure refinement for the La ZnIrO com-pound according to IR Γ is shown in Fig. S1, which wouldhave a nonzero ferromagnetic component along the b -axis. Ascan be seen, this model cannot describe the overall peak inten-sities satisfactorily, thus, resulting in a large agreement factor R B = R B is still as large as 22.0%. Moreover, the degree of B -site disorder amounts to 25%, which is in contradiction tothe D2B measurements.
10 20 30 400123 I obs. I cal. I diff. peak pos. I n t e n s it y ( c oun t s / h )
2 (deg.)La ZnIrO R B = 33.2% FIG. S1. (Color online) Magnetic structure refinement for La ZnIrO according to IR Γ . TABLE S1. Irreducible representations (IR) and the basis vectors ϕ for the Ir ions at the 2 c and 2 d sites with the space group P / n ( k = (0 0 0). Site1: (x, y, z); Site2: (-x + / /
2, -z + / ϕ Site1 Site2 Γ ϕ (1 0 0) (-1 0 0) ϕ (0 1 0) (0 1 0) ϕ (0 0 1) (0 0 -1) Γ ϕ (1 0 0) (1 0 0) ϕ (0 1 0) (0 -1 0) ϕ (0 0 1) (0 0 1)TABLE S2. Irreducible representations (IR) and the basis vectors ϕ for the Ir ions at the 2 c site with space group P / n ( k = (1 / / + / /
2, -z + / ϕ Site1 Site2 Γ ϕ (1 0 0) (0 0 0) ϕ (0 1 0) (0 0 0) ϕ (0 0 1) (0 0 0) Γ ϕ (0 0 0) (-1 0 0) ϕ (0 0 0) (0 1 0) ϕ (0 0 0) (0 0 -1)TABLE S3. Irreducible representations (IR) and the basis vectors ϕ for the Nd ions at the 4 e site with space group P / n ( k = (1 / / + / /
2, -z + / +
1, -y +
1, -z + + /
2, -y + / + / ϕ Site1 Site2 Site3 Site4 Γ ϕ (1 0 0) (0 0 0) (-1 0 0) (0 0 0) ϕ (0 1 0) (0 0 0) (0 -1 0) (0 0 0) ϕ (0 0 1) (0 0 0) (0 0 -1) (0 0 0) Γ ϕ (1 0 0) (0 0 0) (1 0 0) (0 0 0) ϕ (0 1 0) (0 0 0) (0 1 0) (0 0 0) ϕ (0 0 1) (0 0 0) (0 0 1) (0 0 0) Γ ϕ (0 0 0) (-1 0 0) (0 0 0) (1 0 0) ϕ (0 0 0) (0 1 0) (0 0 0) (0 -1 0) ϕ (0 0 0) (0 0 -1) (0 0 0) (0 0 1) Γ ϕ (0 0 0) (-1 0 0) (0 0 0) (-1 0 0) ϕ (0 0 0) (0 1 0) (0 0 0) (0 1 0) ϕ12