Incorporation of Jahn–Teller Cu2+ Ions into Magnetoelectric

Oct 27, 2015 - Polycrystalline samples of Mn1–xCuxWO4 (x ≤ 0.5) have been ... For a more comprehensive list of citations to this article, users ar...
1 downloads 0 Views 3MB Size
Article pubs.acs.org/IC

Incorporation of Jahn−Teller Cu2+ Ions into Magnetoelectric Multiferroic MnWO4: Structural, Magnetic, and Dielectric Permittivity Properties of Mn1−xCuxWO4 (x ≤ 0.25) Pascaline Patureau,† Michael̈ Josse,‡ Rémi Dessapt,† Jean-Yves Mevellec,† Florence Porcher,§ Mario Maglione,‡ Philippe Deniard,† and Christophe Payen*,† †

Institut des Matériaux Jean Rouxel (IMN), Université de Nantes, CNRS UMR 6502, 2 rue de la Houssinière, F-44300 Nantes, France ‡ CNRS, ICMCB UPR 9048, F-33600 Pessac, France § Laboratoire Léon Brillouin, CEA Saclay, CNRS UMR12, F-91191 Gif-sur-Yvette, France S Supporting Information *

ABSTRACT: Polycrystalline samples of Mn1−xCuxWO4 (x ≤ 0.5) have been prepared by a solid-state synthesis as well as from a citrate synthesis at moderate temperature (850 °C). The goal is to study changes in the structural, magnetic, and dielectric properties of magnetoelectric type-II multiferroic MnWO4 caused by replacing Jahn−Teller-inactive Mn2+ (d5, S = 5/2) ions with Jahn−Teller-active Cu2+ (d9, S = 1/2) ions. Combination of techniques including scanning electron microscopy, powder X-ray and neutron diffraction, and Raman spectroscopy demonstrates that the polycrystalline samples with low copper content 0 ≤ x ≤ 0.25 are solid solution that forms in the monoclinic P2/c space group. Rietveld analyses indicate that Cu atoms substitutes for Mn atoms at the Mn crystallographic site of the MnWO4 structure and suggest random distributions of Jahn−Teller-distorted CuO6 octahedra in the solid solution. Magnetic susceptibility reveals that only 5% of Cu substitution suppresses the nonpolar collinear AF1 antiferromagnetic structure observed in pure MnWO4. Type-II multiferroicity survives a weak Cu substitution rate (x < 0.15). Multiferroic transition temperature and Néel temperature increase as the amount of Cu increases. New trends in some of the magnetic properties and in dielectric behaviors are observed for x = 0.20 and 0.25. Careful analysis of the magnetic susceptibility reveals that the incorporation of Cu into MnWO4 strengthens the overall antiferromagnetic interaction and reduces the magnetic frustration.



INTRODUCTION Magnetoelectric multiferroics exhibit simultaneous spin and electric dipole orders and display coupling between these orders. These materials are interesting in that they show novel effects and pose new scientific questions. Technological opportunities may also arise from their useful physical properties. Single-phase transition metal compounds in which ferroelectricity appears in conjunction with a noncollinear spiral magnetic order are a peculiar class of multiferroics.1 In these compounds, both the magnetic and electric dipole orders occur at the same transition temperature and these multiferroics are classified as type-II.2 Within the spin-spiral theory,3,4 the magnetoelectric coupling in these materials is described by P ∝ Aeij × (Si × Sj). Here P is the electric polarization, A is a constant proportional to the spin−orbit coupling and superexchange interactions, and eij is the unit vector connecting the nearest neighbor spins Si and Sj. This study is concerned with MnWO4, which is a prominent example of a type-II magnetically induced ferroelectric.5−7 Its monoclinic (space-group P2/c with β ≈ 91°) crystal structure is made up of infinite zigzag chains, running parallel to the c© XXXX American Chemical Society

direction, of either edge-sharing MnO6 octahedra or edgesharing WO6 octahedra (Figure 1).8,9 The distorted MnO6 and WO6 octahedra contain Jahn−Teller-inactive Mn2+ (d5) ions with spin S = 5/2 and quenched orbital angular moment ⟨L⟩ = 0 and nonmagnetic W6+ (d0) ions, respectively. Three different antiferromagnetic structures termed AF1, AF2, and AF3 have been observed by neutron diffraction in zero magnetic field.9 The AF3 structure below the Néel temperature, TN = 13.5 K, is an incommensurate (IC) sinusoidally modulated antiferromagnetic order with no electric polarization. The AF2 structure between TM1 ≈ 7.5 K and TM2 = 12.5 K is an IC helical spiral that propagates along the MnO4 chains direction. This AF2 state displays a small spontaneous electrical polarization along the b axis with ferroelectric transition temperatures TFE1 and TFE2 that are equal to the magnetic phase-transition temperatures TM1 and TM2, respectively.5,6 The nonpolar AF1 structure below TM1 has a commensurate collinear up−up− down−down spin arrangement along both the a and c axis. In Received: June 25, 2015

A

DOI: 10.1021/acs.inorgchem.5b01416 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry

et al.18 reported measurements of antiferromagnetic susceptibility, ferroelectric polarization, and specific heat on single crystals of Cu-doped MnWO4. These crystals were grown in a floating zone optical furnace from polycrystalline feed rods of Mn1−xCuxWO4 having the nominal compositions x = 0.05 and 0.10. However, the structural, paramagnetic, and dielectric permittivity properties were not investigated, and no chemical analysis was reported for these crystals. In this work, several series of polycrystalline samples of monoclinic Mn1−xCuxWO4 (x ≤ 0.25) solid solution were prepared by two different synthesis methods, both employing moderate annealing temperatures to avoid Cu(II) reduction to Cu(I). The structural effects of Cu concentration x in Mn1−xCuxWO4 were investigated by diffraction techniques and Raman spectroscopy. Magnetic susceptibility data were also analyzed. Dielectric permittivity measurements were taken at zero and nonzero magnetic fields to reveal possible ferroelectric transitions or new magnetoelectric behaviors in these substituted compounds. After initial submission of this manuscript, the authors became aware of a related investigation of polycrystalline Mn1−xCuxWO4 with x ≤ 0.19 by Kumar et al.19 This latter work is based on specific heat, magnetization, dielectric permittivity, and neutron diffraction data. Our results will be compared with and discussed in light of the data reported by Kumar et al.



EXPERIMENTAL SECTION

Synthesis and Sample Preparation. Powders of Mn1−xCuxWO4 were prepared using two different methods, namely a traditional solidstate reaction and a citrate synthesis. Attempts to prepare phase-pure samples were performed for nominal Cu contents x ≤ 0.5. In the conventional solid-state method, stoichiometric amounts of MnO, CuO, and WO3 powder materials were mechanically ground and mixed. The resulting mixtures were pressed into pellets and heated at 850 °C for 30 h in air with intermittent mechanical regrindings. The highest temperature used during the solid-state reaction, i.e., 850 °C, was chosen because both MnWO4 and CuWO4 have been previously prepared in powder forms at this temperature in air.20,13 For the citrate synthesis, appropriate amounts of manganese sulfate, copper acetate, and ammonium metatungstate were dissolved in citric acid aqueous solutions. pH was adjusted to stabilize Cu(II) species and prevent precipitation. The solutions were stirred and heated until the water evaporated. The dried solid mixture was then heated at 750 °C for 3 h in air. Dense pellets that were used for dielectric permittivity measurements were fabricated with the powders of Mn1−xCuxWO4 that were prepared by both synthesis routes. High-speed ball milling was used for particle size reduction, and pellets were made by pressing the resulting fine powders in a cylindrical die at room temperature. For x > 0, dense pellets could be obtained by heating at the same temperature that was used for the solid-state reaction, namely 850 °C, for 24 h in air, making it possible to compare the properties of the powders and the dense pellets. For pure MnWO4, pellets were sintered at 1100 °C for 2 h in air, as previously described.13 All pellets were slowly cooled in the furnace at the end of the heat treatment. Physicochemical Characterization. Thermogravimetric (TG) measurements were performed in flowing air. Finely ground powders were heated up to 1175 °C at 3 K/min. Semiquantitative elemental analyses of dense pellets were performed by energy-dispersive X-ray spectroscopy (EDS) in a scanning electron microscope (SEM) chamber. For each sample, spectra were obtained at 15 kV for different positions on the flat pellet surface. X-ray diffraction patterns of powders were collected at room temperature on a Bruker D8 Advance instrument using monochromatic CuK‑L3 (λ = 1.540598 Å) Xrays and a LynxEye detector. For the samples heated at 1175 °C and then cooled at 50 K/min in the TG apparatus, XRD data were acquired using an INEL diffractometer operating with a 120 °CPS detector and monochromatic CuK‑L3 radiation. Neutron powder

Figure 1. Monoclinic P2/c crystal structure of MnWO4, viewed perpendicular to the bc plane (top) and perpendicular to the ab plane (bottom). Mn atoms are drawn in purple, W in gray, and O in red. The triclinic P1̅ crystal structure of CuWO4 is shown in Figure S1.

AF3 and AF1, the magnetic moments of Mn2+ align along the easy axis of magnetization that lies in the ac plane forming an angle of ≈35° with the a axis.9 In multiferroic spin-spiral AF2 an additional component of magnetic moment exists along the b axis due to the energy gain caused by competing (frustrated) isotropic spin exchange interactions. Because the magnetic structure plays a key role in determining the magnetoelectric properties, several experimental works have aimed at modulating the magnetic interactions and the magnetic anisotropy through partial substitutions of divalent transition metal cations (e.g., Fe2+, Co2+, Ni2+, Zn2+) for Mn2+ ions.10−13 Cu2+ incorporation into MnWO4 could also be quite interesting because octahedral Cu2+ ions in oxide structures are prone to undergo a Jahn−Teller (JT) distortion. JT distortions are an interesting tool in the field of functional oxides because they enable peculiar relationships between structural and electronic properties. Clearly, as a consequence of the JT activity of Cu2+, the crystal structure of CuWO4 (Figure S1, Supporting Information) is distorted with respect to monoclinic wolframite MWO4 (M = Mn, Fe, Co, Ni, Zn) structure to the lower symmetry space group P1.̅ 14 All CuO6 octahedra possess the same pseudotetragonal geometry, with two opposite long Cu− O bonds and four short Cu−O bonds. Long-range alignment of the long Cu−O bonds and orbital ordering occur in the roomtemperature structure. The magnetic properties of CuWO4 are also different from those of the other members of the wolframite MWO4.15,16 CuWO4 is the only one to show lowdimensional magnetic behavior.17 No magnetoelectric properties have been reported for CuWO4. Very recently, K.C. Liang B

DOI: 10.1021/acs.inorgchem.5b01416 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry

Mn1−xCuxWO4 samples that were heated at 750 or 850 °C, see below. Also, some reflections became asymmetric for x = 0.25. Furthermore, SEM images and EDS analyses demonstrated that these samples were inhomogeneous. Both samples contained areas that appear to have melted. Structure. Room-temperature XRD data were first analyzed using the Le Bail method and the monoclinic P2/c structural model of MnWO4 (ICSD no. 67907). The patterns showed no extra reflection or peak splitting for xCI ≤ 0.2 in the samples obtained by the citrate method and for xSS ≤ 0.25 in those obtained by the solid-state route. Nevertheless, for all singlephase samples with x ≥ 0.15, some but not all Bragg reflections showed noticeable line-broadening not related to any anisotropic crystallite size effect, regardless of the synthesis method. Because these apparent line-broadenings could be due to a symmetry lowering leading to atypical peak profiles caused by overlapping, the patterns for all xCI ≤ 0.20 and xSS ≤ 0.25 were also refined using the triclinic P1̅ structural model of CuWO4 (ICSD no. 163954). However, the obtained Le Bail refinements did not show any improvement, suggesting that the average structure can be described by the monoclinic P2/c space group in all the samples xCI ≤ 0.20 and xSS ≤ 0.25. Furthermore, as can be seen in Figure S3 (Supporting Information), the anomalous shapes of the broadened peaks were successfully handled by using the P2/c structural model of MnWO4 and Stephens’s phenomenological model of anisotropic microstrain broadening.27 Figure S4 (Supporting Information) shows the lattice parameters and cell volume as a function of the nominal Cu concentration for all the nondensified as well as densified samples. These findings are in agreement with the very recent results reported by Kumar et al.19 As expected from the Vegard’s law, a and c lattice parameters and cell volume decrease as x increases, because Cu2+ ion is smaller in size than Mn2+ (ionic radii of Cu2+ and high-spin Mn2+ ions in an octahedral site are 0.73 and 0.83 Å, respectively). We noted that the x-dependence of b parameter, although being weak, does not follow the same behavior. Such an increase in b parameter is not observed when Mn2+ is replaced with smaller Jahn−Teller-inactive Fe2+, Zn2+, or Mg2+ ions.28,13 In these three latter cases, a, b, and c lattice parameters all decrease when increasing the amount of dopant. In contrast, a weak linear increase in b parameter is noted for monoclinic Zn1−yCuyWO4 compounds when y increases from 0 to 0.20.29 In these latter compounds, Cu2+/Zn2+ substitutions occur in ZnWO4 which has the same P2/c structure as MnWO4. Because of the small difference in Mn and Cu atomic numbers, Rietveld refinements of laboratory X-ray diffraction data did not allow an accurate determination of Mn/Cu fractional site occupancy of the Wyckoff 2f (1/2, y, 1/4) position in the MnWO4 P2/c structural model. On the contrary, there is a large difference between the Mn and Cu coherent neutron scattering lengths (bcoh(Mn) = −3.73 fm, and bcoh(Cu) = +7.72 fm). Combined Rietveld refinements were performed on both XRD and NPD patterns for all the Mn1−xCuxWO4 (0.05 ≤ x ≤ 0.25) compounds. As a representative example the observed, calculated, and difference patterns for a x = 0.15 sample are given in Figure S5, Supporting Information. Once again, there was no evidence for a symmetry lowering of the unit cell content. For each substitution rate, the refined Cu/Mn occupancy on the 2f site was in excellent agreement with the nominal composition. In addition, because of the large neutron scattering length of

diffraction (NPD) data were collected at 300 K on 2-g powder samples using the 3T2 instrument (incoming wavelength of λ ≈ 1.2247 Å) at the Laboratoire Léon Brillouin (CEA Saclay-CNRS). Le Bail and Rietveld analyses of the XRD and NPD data were performed using JANA 200621 and the Cheary-Coelho fundamental approach for XRD profile parameters.22,23 Room temperature Raman spectra were collected under microscope in backward scattering configuration using a Renishaw InVia spectrometer and the 514 nm line of an argon ion laser. The laser probe diameter at the sample surface was approximately 2 μm. The laser power was reduced to about 0.1 mW to avoid sample heating. A grating of 2400 grooves/mm was used to get a resolution of ≈1 cm−1. Two edge filters were used to filter the Rayleigh line intensity, allowing measurements above 100 cm−1. A Quantum Design MPMS-XL7 was used to collect DC magnetization data. Zero-field-cooled (ZFC) and field-cooled (FC) magnetization measurements were taken from 2 to 300 K in an applied field of μ0H = 0.1 T. Data were corrected for the diamagnetism of the sample holder as well as for core diamagnetism using Pascal’s constants.24 Dielectric measurements were performed on dense pellets (≈6 or 8 mm diameter, ≈1 mm thick) using an HP4194a impedance bridge. Samples were loaded into a Quantum Design Physical Properties Measurement System (PPMS). Measurements were taken in the frequency ( f) range of 102−106 kHz and in the temperature range of 2−20 K.



RESULTS AND DISCUSSION Sample Purity, Compositional Analysis, and Thermal Stability. Judging from XRD and Raman measurements, phase pure samples of Mn1−xCuxWO4 crystallizing with the wolframite P2/c crystal symmetry could only be prepared in a reproducible manner for nominal xSS ≤ 0.25 when using the solid-state (SS) reaction, and for nominal xCI ≤ 0.20 when employing the citrate (CI) method. One out of two attempts to prepare samples with nominal xSS = 0.30 resulted in mixtures of a wolframite P2/c monoclinic phase and at least one unknown phase. For nominal xSS > 0.30, XRD patterns indicated that the compounds form in the triclinic P1̅ space group as CuWO4 does. Further studies are needed to better characterize these triclinic phases and are beyond the scope of the present study focused on Cu incorporation into multiferroic MnWO4. Phase pure dense pellets of Mn1−xCuxWO4 with x ≤ 0.25 could be fabricated by heating compacted powders as described in the previous section. The cation stoichiometries were measured by EDS on flat pellet surfaces. Within the experimental accuracy of a few percent, the results agreed with the nominal starting concentrations of the metal atoms (i.e., Mn, W, and Cu). Because of this agreement, the nominal compositions are hereafter employed to designate the compositions of the samples. TG analyses of nondensified samples with x = 0.15 and 0.25 indicated that mass loss occurs above ≈1020 °C and ≈1060 °C for x = 0.15 and 0.25, respectively (Figure S2, Supporting Information). For both samples, the weight loss measured at 1175 °C is smaller but comparable to the calculated mass loss for the following reduction reaction Mn1−xCuxWO4 → Mn1−xCuxWO4−x/2 + x/2 O2. Note that CuWO4 starts to decompose in air (weight loss) at 940 °C due to the Cu(II) reduction to Cu(I)25 and that MnWO4 is stable in air up to its congruent melting point (Tm ≈ 1300 °C).26 XRD patterns of the x = 0.15 and 0.25 samples heated at 1175 °C in the TG apparatus displayed broad Bragg reflections consistent with the presence of a phase having the wolframite P2/c crystal structure. Some additional unidentified peaks also showed up in the patterns. The refined lattice parameters of the wolframite-like phase were, however, clearly inconsistent with the lattice parameter trend observed along the series of C

DOI: 10.1021/acs.inorgchem.5b01416 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry

The strain induced by the JT distortion could cause the peculiar shapes of some of the Bragg reflections in the XRD patterns for x ≥ 0.15. As can be seen in Figure S6 (Supporting Information), each of the three pairwise equivalent Mn/Cu− O bond distances in the low doping region, x ≤ 0.25, corresponds to two different unique distances in pure CuWO4. One of the two equivalent Mn/Cu−O2 distances, which increases with x, and one of the two equivalent Mn/Cu−O1 distances correspond to the two pseudotetragonally elongated Cu−O bonds in pure CuWO4. Another clue for the distinct surroundings of Cu2+ and Mn2+ comes from the distortion 6 parameter, defined as31 σ = (∑i = 1(RM−O − ⟨RM−O⟩)2)1/2 where RM−O are the six Mn/Cu−O distances of the distorted (Mn/ Cu)O6 octahedra and ⟨RM−O⟩ is the average Mn/Cu−O bond length. As can be seen in Figure S7 (Supporting Information), σ(x) has increased by more than 30% in the x = 0.25 sample compared to pure MnWO4 for which σ ≈ 0.18 Å. In pure CuWO4 the distortion parameter amounts to ≈0.49 Å. We also performed Raman scattering at room temperature to further investigate the crystal structures. According to grouptheory analysis, 18 Raman-active vibrational modes (8 Ag and 10 Bg) are expected for MnWO4 in the paramagnetic phase above TN.32 Figure 3 shows the Raman data taken at room

oxygen (bcoh(O) = +5.803 fm), NPD data allowed accurate refinements of both positions and site occupancies for the oxygen atoms. There was no oxygen off-stoichiometry on O1 and O2 sites. Refinement results also showed no evidence for anomalous atomic displacement parameters for all atoms. As for the oxygen atoms, this observation is in contrast to what one would expect in a flexible structure such as Zn1−xCuxTiO4 inverse spinel, where it was shown that an increase in the O atomic displacement parameters is induced by the JT distortion of CuO6 octahedra.30 The quality of the combined Rietveld refinements allowed an accurate determination of the M−O distances for each compound. The calculated metal-to-oxygen distances are presented in Figure 2. Whereas the substitution rate does not

Figure 3. Room-temperature Raman spectra of Mn1−xCuxWO4 polycrystalline samples (x ≤ 0.25). The inset shows the spectra in the region where four bands (observed at 258, 272, 294, and 327 cm−1 in MnWO4) that shift monotonically to higher wavenumbers as the Cu concentration increases are present.

Figure 2. Variations of the metal-to-oxygen distances as a function of x for Mn1−xCuxWO4 with x ≤ 0.25. A given distance is circled within the MnWO4 unit cell scheme using the same color as that of the symbol used in the plot. Mn/Cu−O1 distances are represented by blue squares, and the two different Mn/Cu−O2 distances are represented by red circles and green stars. The two different W−O1 distances are represented by black and magenta reversed triangles whereas W−O2 distances are represented by orange diamonds. Error bars are smaller than the size of the symbols.

temperature on the samples that were analyzed by XRD and NPD. All of the 17 modes expected in the actual experimental wavenumber range of 100−1000 cm−1 for pure MnWO4 (the 18th mode is located below 100 cm−1)32,33 were observed for all Cu concentrations in two independent series of samples. A gradual weakening of Raman band intensities with increasing Cu content was observed, consistent with the gradual color change of the samples induced by increasing Cu concentration. For each Cu concentration, several measurements were performed on different areas of the powder sample showing large differences in relative intensities from one area to another and making it impossible to analyze relative intensities. This observation is associated with the fact that Raman modes are strongly polarized32,33 and is due to the large particle size ≥0.5 μm. There is no additional band or band splitting in substituted

affect W−O and Mn/Cu−O1 distances, there are noticeable changes in the two Mn/Cu−O2 distances. The fact that two out of three pairwise equivalent Mn/Cu−O distances do not decrease with x, in contrast to what would be expected from the difference in ionic radii between Cu2+ and high-spin Mn2+, suggests that the coordination of JT-active Cu2+ and JT-inactive Mn2+ differ. A JT effect is likely to be the driving force for the excess elongation evident in one of the pairwise Mn/Cu−O2 bonds and for the anomalous behavior of b parameter as well. D

DOI: 10.1021/acs.inorgchem.5b01416 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry

Mn2+ cation resides on a highly deformed octahedral site with a C2 point symmetry, so the distortion of diluted CuO6 octahedra in MnWO4 may be an additional distortion superimposed on the already existing local distortion with no decrease in local point symmetry. In the case where all MO6 octahedra have the C2 point symmetry whether they contain Mn2+ or Cu2+ for all monoclinic compositions, the change in space-group symmetry around x = 0.3 can be viewed as a sharp crossover point where the CuO6 octahedra all transform to pseudotetragonally elongated JT-distorted octahedra with long-range alignment of the long Cu−O bonds. It is worth noting that the distortion at the Mn site in pure MnWO4 is not due to a crystal field stabilization for Mn2+ because this ion has a d5 electronic configuration. This MnO6 distortion is likely to be due to the nature of the 3D stacking of the WO4 chains with highly covalent W−O chemical bonds. The chains accommodate the smaller Cu2+ ions by rotating the rigid WO6 octahedra along their shared O−O edges so that these chains gradually shrink along the c axis when the amount of Cu increases. This shrinkage effect can be seen in Figure S8 (Supporting Information), which shows the doping-induced decrease in one selected O−O distance within a WO4 chain. On the other hand, O−O distances between two nearest neighbor WO4 chains are little affected by Cu doping. Magnetic Susceptibility. Figure 4a shows the ZFC magnetic susceptibility χ(T) for a series of samples of Mn1−xCuxWO4 in the temperature range in which pure

compounds. On increasing the Cu concentration up to x = 0.25, six bands shift to higher wavenumbers, five bands shift to lower frequencies, and six modes do not shift within the experimental uncertainty (≈1 cm−1). This indicates that the laser power did not heat the samples and tells us that Cu doping is not equivalent to applying external pressure.34 The three upper energy modes observed for pure MnWO4 at 885, 774, and 698 cm−1 display wavenumber shifts smaller than 2.0 cm−1 over the whole composition range x ≤ 0.25. Because these bands are due to internal WO6 stretching modes32,33 this confirms that the highly covalent W−O chemical bonds are not significantly affected by low doping. As can be seen in Figure 3, four modes that involve vibrations of Mn−O bonds in MnWO433,35 are monotonically shifted toward higher energies as the copper concentration increases (these modes are observed at 258, 272, 294, and 327 cm−1 in MnWO4). These shifts may reflect the difference in bonding interaction between Mn and O and Cu and O because Cu−O bonds are expected to be more covalent than Mn−O bonds on a general basis. Finally, we noted that Raman peaks are slightly broader in substituted compounds than in pure MnWO4. Broadening of Raman modes in solid solutions are likely to be mainly due to substitutional disorder.30 Lattice imperfections (defects) may also broaden Raman lineshapes.30 We have fitted lineshapes of several modes that involve vibrations of Mn−O bonds in MnWO4 and have found that these modes have Gaussian lineshapes as expected in the case of substitutional disorder with no structural defects.30 Judging from XRD, NPD, and Raman measurements, one cannot say that all CuO6 or MnO6 octahedra in each of the Mn1−xCuxWO4 (x < 0.30) compounds have the same distortion as in pure CuWO4 or MnWO4, respectively. The degree of distortion and the symmetry of a CuO6 octahedron may depend on the presence or absence of CuO6 neighbors adjacent to this CuO6 entity; the same applies for the MnO6 octahedron. Assuming a random Cu distribution within the MnO4 chains running along the c axis, the probability of one Cu2+ having all (two) JT-active Cu2+ nearest neighbors is x2. The probability of having only one Cu2+ neighbor is 2x(1 − x), and the probability of having all (two) JT-inactive Mn2+ neighbors is (1 − x)2. At x = 0.3, i.e., the composition around which one observed the lowering of the averaged lattice symmetry from P2/c to P1,̅ almost one in three Mn sites has a JT-active ion and 51% of Cu2+ ions have one or two JT-active nearest neighbors. Assuming that all CuO6 octahedra are locally pseudotetragonally JT distorted for x < 0.3, the transition at x = 0.3 can be viewed as a transition to cooperative JT distortions with longrange periodicity that results in a triclinic structure. From a structural point of view, our study is reminiscent of early work regarding Zn1−yCuyWO4 solid solution.29,36 Room-temperature XRD and NPD patterns were analyzed using either the wolframite P2/c model of ZnWO4 for low Cu content, y < 0.22, or the P1̅ structural model of CuWO4 beyond y = 0.22 (for samples annealed at 600 °C). The change in symmetry of the averaged lattice at about y = 0.22 results in the splitting of certain monoclinic Bragg reflections. A Cu EXAFS study suggested that all CuO6 octahedra maintain their pseudotetragonally elongated JT distortion in the monoclinic domain for y > 0.1.37 For lower Cu contents, however, the Cu environment switches to the same axially compressed octahedral geometry as Zn2+ has in pure ZnWO4. These distortion effects for y < 0.22 are purely local effects which random distribution preserves the P2/c space group of the average structure. In pure MnWO4, the

Figure 4. (a) Magnetic susceptibility χ versus temperature T of polycrystalline samples of Mn1−xCuxWO4 (x ≤ 0.25) at low temperatures. Data were acquired under a magnetic field of 1000 Oe. Vertical lines denote magnetic phase-transition temperatures for pure MnWO4. (b) Dependence of χ(T)/χ(TN) on T/TN for the same data sets. Néel temperatures, TN, were estimated from the d(Tχ)/dT versus T curves. E

DOI: 10.1021/acs.inorgchem.5b01416 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry MnWO4 undergoes its three magnetic phase transitions at TN = 13.5 K, TM2 = 12.5 K, and TM1 ≈ 7.5 K. For each substituted sample, one does not observe any difference between the data measured under ZFC and that measured under FC conditions, ruling out a doping-induced spin-glass or cluster-glass behavior. As can be seen in Figure 4a, the marked change in slope in χ(T), which occurs at the lowest phase transition temperature TM1 in pure MnWO4, is absent in all doped samples. This indicates that only 5% of Cu substitution suppress the collinear up−up−down−down AF1 phase, consistent with the very recent results reported by Kumar et al.19 From our magnetic susceptibility data of samples x = 0.05, 0.1, 0.15, and 0.20, however, two magnetic ordering temperatures are discernible. The magnetic phase transitions temperatures TN(x) and TM2(x), defined by sharp maxima in d(Tχ)/dT vs T curves (not shown), are plotted as a function of x in Figure S9, Supporting Information. It is apparent that Cu doping increases the Néel temperature TN(x). These findings are consistent with the magnetic phase diagram that was recently obtained from specific heat and neutron diffraction data of samples with x ≤ 0.19.19 For x = 0.25, only one magnetic transition at TN(0.25) ≈ 20 K can be reasonably estimated from the d(χT)/dT vs T curve (only a very broad maximum occurs in the d(Tχ)/dT vs T curve around 6 K). Figure 4b presents the normalized susceptibility χ(T)/χ(TN) as a function of the normalized temperature T/TN. The normalized susceptibilities of doped samples with x ≤ 0.15 are comparable and are nearly the same as that in MnWO4 for temperature higher than TM1, suggesting that these compounds have similar noncollinear AF2-like magnetic structures. This observation is fully consistent with the IC magnetic structures recently determined for x = 0.05 and x = 0.1 below TM2.19 These magnetic structures were found to be similar to that of AF2 in MnWO4. A glance at Figure 4b also suggests that x = 0.2 and x = 0.25 compounds do not have the same magnetic structures at low temperatures for T/TN < 0.75. In a recent report based on neutron diffraction data of a sample x = 0.19, a commensurate collinear magnetic structure and a IC AF2 structure were found above and below TM2 = 11.5 K, respectively.19 Overall magnetization measurements indicate that one can distinguish the most doped x = 0.20 and 0.25 samples from the lightly doped x ≤ 0.15 compositions. We can also note that all changes in slope in χ(T) at TN(x) are sharp so that TN(x) are well-defined (to within 0.05−0.1 K). This discounts the presence of large clustering effects or of phase segregation in the samples. We now turn to the paramagnetic susceptibilities above TN(x). The inverse susceptibilities 1/χ(T) are shown in Figure S10, Supporting Information. A very good fit of the 1/χ data above 100 K to the Curie−Weiss law, χ(T) = C/(T − θ), is found for all of the samples (solid lines in Figure S10). The resulting values for the Curie constant, C, and the Weiss temperature, θ, are plotted as a function of x in Figure 5. The values for the molar Curie constant C(x) are reliable because C(x) is well described by the relationship C(x) = (1 − x) C(Mn2+) + xC(Cu2+), where C(Mn2+) = 4.4 cm3 K Mn-mol−1 and C(Cu2+) = 0.47 cm3 K Cu-mol−1 (solid blue line in Figure 5). The former value of C(Mn2+) is the spin-only value expected for S = 5/2 and g = 2.0, while the latter value of C(Cu2+) was calculated using S = 1/2 and the g value determined by ESR for Cu-doped ZnWO4, g = 2.24.38 For all of the samples, the fitted value of θ is negative and its magnitude |θ| is smaller than the minimum temperature T = 100 K used in

Figure 5. Variation of the Curie constant, C, and Weiss temperature, θ, as a function of Cu concentration, x, for polycrystalline Mn1−xCuxWO4 (x ≤ 0.25). C and θ were extracted from fits of the inverse susceptibilities above 100 K. The proper vertical scale for the C data set is indicated by the arrow. The solid blue line drawn in the C(x) plot refers to the relationship C(x) = (1 − x)C(Mn2+) + xC(Cu2+) described in the text. The solid red line drawn in the θ(x) plot guides the eye. The dashed line reflects the variation of the molecular field theory Weiss temperature, θ = (S(S + 1)/3kB)) ∑iziJi, that would be solely due to the x-dependence of the S(S + 1) term.

the Curie−Weiss fit. Therefore, and given that the fitted values for C(x) are reasonable, the values of θ (x) obtained from this fit are meaningful and indicate the presence of dominant AFM interactions in all of the samples. As can be seen in Figure 5, the absolute Weiss temperature |θ(x)| decreases monotonically as the amount of Cu increases with the exception of x = 0.10, which seems anomalous in this series. It is interesting to compare these experimental Weiss temperatures to the wellknown molecular field theory (MFT) prediction θ = (S(S + 1)/ 3kB)) ∑iziJi (Ji is the exchange coupling between a central spin and the zi spins linked by Ji). It is clear that partial replacement of the S = 5/2 Mn2+ ions with Cu2+ ions having smaller S = 1/2 spins can explain the decrease in experimental |θ(x)|. As illustrated in Figure 5, the decrease in experimental |θ(x)| is, however, slower than the decrease in MFT Weiss|θ| that would be solely due to the doping-induced decrease in averaged S(S + 1) term (i.e., assuming no change in the linear combination of different exchange couplings ΣiziJi). This strongly suggests that the introduction of Cu2+ ions strengthens the overall magnetic interactions. Unlike TN(x) and Tm2(x), there is no abrupt change in θ(x) between x = 0.15 and x = 0.20, suggesting that changes in the magnetic topology or in the magnetic anisotropy impact the x-dependence of magnetic phase transitions TN(x) and TM2(x) as x increases from 0.15 to 0.20. It is also interesting to calculate the Néel temperature reduction factor f = |θ|/TN. This factor slowly decreases from 5.2 to 4.5 when x increases from 0 to 0.15, and then decreases to 3.8 and 3.3 for x = 0.2 and 0.25, respectively. In pure MnWO4, the Weiss temperature θ is a linear combination of more than 10 different short-range as well as long-range spin exchange couplings.39−41 Although MnWO4 is not a simple AFM where only near neighbor interactions are significant, the |θ|/TN ratio has been considered as a frustration index (the higher the index, the higher the frustration) because the principal interactions in MnWO4 are antiferromagnetic. Accordingly, it is tempting to think that incorporation of extra Mn−Cu and Cu−Cu spin F

DOI: 10.1021/acs.inorgchem.5b01416 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry

A microscopic understanding of the effect of Cu doping on the magnetic properties of MnWO4 is quite complicated because the extra (short-range or long-range) Cu−Cu and Mn−Cu magnetic couplings and the magnetic anisotropy for the Cu2+ ions are not known. In pure MnWO4, the magnetic anisotropy is essentially due to a single-ion anisotropy (SIA) of the −DS2iz type where D is the SIA parameter and Siz denotes the spin component along the z easy axis of magnetization. In collinear (nonpolar) AF1 and AF3 the positive SIA, D/kB ≈ +1 K,41 causes the pinning of the magnetic moment along the z easy axis in the ac plane. The SIA is however sufficiently small to allow deviations of the magnetic moments toward the b direction in the noncollinear helical AF2 state to minimize the isotropic exchange energy. In this view, the ferroelectric AF2 state can be considered as the result of competing (frustrated) short-range and long-range isotropic exchange interactions in the presence of SIA.40,41 Chemical doping modifies the delicate balance between isotropic exchange interactions and magnetic anisotropy. Replacing Mn2+ with Fe2+ (S = 2, d6) ions, for instance, results in the complete loss of the helical AF2 spin structure and hence of the ferroelectric order.10 In contrast, a small Co concentration stabilizes the polar spiral AF2 at the expense of the collinear AF1.11 A recent theoretical work pointed out that the small difference TN − Tm2 ≈ 0.8 K in MnWO4 is a consequence of the small SIA of Mn2+ ions.44 In polycrystalline samples of Mn1−xCuxWO4, the difference TN − Tm2 remains unchanged up to x = 0.15, suggesting that the introduction of Cu2+ ions does not significantly change the overall magnetic anisotropy for x ≤ 0.15. Observed changes in the x-dependences of TN and TM2 for x > 0.15 may be explained by either the effect of the increasing numbers of extra Cu−Cu and Cu−Mn spin exchange interactions, which reduce the frustration, or by a sharp change in the magnetic anisotropy. For Cu2+ ions, the general belief is that these ions do not have SIA because they are spin-1/2 ions and that the magnetic anisotropy is essentially due to anisotropic exchange. A very recent theoretical work showed, however, that Cu2+ ions can have SIA.45 In the collinear AF structure of CuWO4, the Cu2+ magnetic moments are aligned along the axis of elongation of the JT-distorted CuO6 octahedra, indicating the importance of SIA.46 It should also be noted that the presence of inversion centers in the middle of nearest-neighbors Cu−Cu pairs implies that the Dzyaloshinskii−Moriya (DM) interaction, which is an important anisotropy term in many Cu2+ compounds, vanishes by symmetry in CuWO4. In pure MnWO4 it has been suggested that a DM term that involves two next-nearest neigbors Mn2+ is an additional source of spin spirality and of ferroelectric activity because this DM term is comparable in energy to the SIA.47,48 Addition of Cu2+ or of other dopants as well will also affect this DM interaction and hence the magnetoelectric properties. Dielectric Properties. Figure S11 (Supporting Information) and Figure 7 show representative temperature-dependent data for the capacitance in zero magnetic field and in a magnetic field of 9 T, respectively. We also measured the loss factor, but the signals could hardly be separated from the background and thus are not shown. Anyhow, the very low loss index (tgδ < 0.1%) indicates that the samples are highly resistive at low temperature. At zero magnetic field, the capacitance of pure MnWO4 (Figure S11) shows features that are consistent with published data, namely a sharp peak at TFE2(x = 0) ≈ 12.5 K and a broad step-like feature that corresponds to the stepwise transition previously seen at TFE1(x = 0) ≈ 7.5 K on single crystals.5 For our samples x = 0.05, 0.1,

exchange interactions into MnWO4 reduces the magnetic frustration. In pure CuWO4, the |θ|/TN ratio is associated with the low dimensionality of the spin exchange lattice and not with magnetic frustration. Taking the value of θ ≈ −60 K resulting from a fit to the susceptibility data well above the temperature at which the broad susceptibility maximum occurs,42 one calculates |θ|/TN ≈ 2.5. Finally, we examine how the inverse susceptibility deviates from strict linearity at low temperature because deviation from Curie−Weiss law can give information about the magnetic topology and the nature of the extra Mn−Cu and Cu−Cu spin exchange interactions. For this purpose, we have plotted in Figure 6 normalized inverse susceptibility data against the

Figure 6. Normalized inverse magnetic susceptibility as a function of normalized temperature for polycrystalline samples of Mn1−xCuxWO4 (x ≤ 0.25). The nature of the normalization is described in the text. The solid line represents Curie−Weiss paramagnetism.

temperature normalized by the experimental absolute Weiss temperature, T/|θ|. This type of plot is obtained by rearranging the Curie-law, 1/χ = (T + |θ|)/C, to (C/(χ|θ|)) − 1 = T/|θ|.43 It allows a comparison of compounds that have different doping levels because the susceptibility of all samples is scaled by both the strength of the magnetic interactions, |θ|, and the magnitude of the paramagnetic moments, C. The solid straight line drawn in the plot represents the Curie−Weiss behaviors observed for T > 100 K, i.e., for T/|θ| > 1, in all of the samples. Negative deviations from this Curie−Weiss line correspond to samples with uncompensated AF (ferrimagnetic) spin exchange interactions, whereas positive deviations indicate the presence of fully compensated AF interactions. It is seen from Figure 6 that the data obey the high-temperature Curie−Weiss law even for temperatures significantly lower than |θ|, a hallmark of magnetic frustration. Deviations from Curie−Weiss paramagnetism are all positive, indicating that there is no uncompensated magnetic interactions in the substituted samples. Furthermore, the deviations are comparable and much less than what is expected for a low dimensional nonfrustrated AF such as CuWO 4 (the paramagnetic susceptibility of CuWO4 has a broad susceptibility maximum at ≈90 K15,16). It is also evident that all samples order at temperatures TN much lower than |θ|. Overall, the paramagnetic susceptibilities show features that are consistent with the presence of dominant frustrated AF interactions with a frustration index that decreases with the doping. G

DOI: 10.1021/acs.inorgchem.5b01416 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry

We also measured the dielectric responses under applied magnetic fields up to 9 T; see Figure 7. For our sample of MnWO4, a broad peak is observed at TFE2, consistent with published single-crystal data obtained for 9 T magnetic fields applied along different crystal directions.5 As in zero magnetic field, this peak at TFE2 shifts to higher temperature for x = 0.05 and 0.10. Both the changes in slope at the Néel temperatures and the capacitance minimum in the x = 0.20 sample are still observed for μ0H = 9 T. As in zero magnetic field, these two latter features are frequency-independent and show no thermal hysteresis. The position of capacitance minimum for x = 0.20 decreases with increasing magnetic field, indicating the existence of a new magnetoelectric effect for this composition.



CONCLUDING REMARKS Our structural and magnetic data demonstrated that our polycrystalline samples of Mn1−xCuxWO4 with low Cu-content (x ≤ 0.25) belong to a solid solution and are not phase segregated MnWO4−CuWO4, regardless of the synthesis method. Both diffraction and Raman data suggested that these compounds form in the monoclinic P2/c space group and indicated smooth changes in the average structure as Cu is progressively incorporated into MnWO4. The introduction of Cu2+ ions leads to an increase in the Néel temperature. Despite the propensity of Cu2+ to display JT distortions, the multiferroic transition survives the Cu/Mn subtitution for small Cu concentrations x < 0.15 and the multiferroic transition temperature increases with the Cu concentration. New trends in some of the magnetic properties and in dielectric properties appear for higher compositions x = 0.2 and 0.25. Finally, we examine how results obtained with polycrystalline samples in this work and in ref 19 compare with very recent data obtained on crystals that were grown using the floating method.18 The nominal compositions xn = 0.05 and 0.10 of the polycrystalline feed rods were employed to designate the compositions of the crystals. As in polycrystalline pellets, the lowest ferroelectric transition at TFE1 is strongly affected by the Cu doping. The substitution of Cu for Mn in these crystals has, however, less effect on TN, TM2, and TFE2 than it has in polycrystalline samples.

Figure 7. Temperature dependence of the capacitance, plotted in the form of C(T) − C(25K), of Mn1−xCuxWO4 polycrystalline pellets (x ≤ 0.25). Data were taken at 385 kHz during the warming run in an external magnetic field of μ0H = 9 T. The values of the capacitance at 25 K are in the range of 3.0 × 10−12 to 5.4 × 10−12 Faraday.

and 0.2, the data at zero magnetic field (Figure S11) are consistent with those recently obtained for x = 0.05, 0.1, and 0.19 by Kumar et al.19 For x = 0.05 and 0.1, the peaks at TFE2 shift to higher temperature with increasing Cu-doping for all frequencies. These shifts are frequency-independent, indicating the ferroelectric nature of these capacitance peaks. The xdependences of TFE2 is given in Figure S9. TFE2 coincide with the magnetic phase transitions TM2, suggesting that a magnetoelectric multiferroic state survives a small doping x < 0.15. For x = 0.05 and x = 0.10, very weak peaks are visible in zero magnetic field below TFE2 at ≈8.5 K and ≈9.5 K, respectively (also shown in Figure S9). According to recent neutron diffraction data,19 these peaks are not associated with a collinear-to-noncollinear AF1-to-AF2 transition, as in pure MnWO 4 . It should be recalled that low-temperature capacitance peaks have already been detected in Mn1−x(Mg,Zn)xWO4 and MnW1−xMoxO4 solid solutions.13,49 A study of pyroelectric properties of a ceramic sample of Mn0.85Mg0.15WO4 showed that this Mg-doped composition remains ferroelectric below the low-temperature capacitance anomaly.50 For our Cu-doped sample x = 0.15, the zero-field capacitance is rather featureless. On the other hand, the zerofield capacitance of the x = 0.20 samples displays frequencyindependent marked changes in slope at TN (0.20) ≈ 17.5 K with no thermal hysteresis. Also, a step-like feature with a local maximum at ≈7 K and a minimum at ≈10.5 K is observed; see Figure S11. For a given sample, both the shapes and the positions of these two extrema are frequency-independent and do not depend on whether the data are taken while warming or cooling the sample. Interestingly, the local minimum occurs at the magnetic phase transition temperature TM2(0.2) that was observed in the same pellet. Because the measured capacitance is dominated both by the electronic and ionic polarization mechanisms, the two anomalies at TN and TM2 in the capacitance of the x = 0.20 sample may be associated with either magnetoelastic effects19 or subtle changes in the electronic structure occurring at the antiferromagnetic orderings. In fact, a change in slope at TN is seen in all samples and is therefore a general feature of capacitance in zero magnetic field (see Figure S11). To our knowledge, no marked change in slope at TN has been observed in other substituted MnWO4.



ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.inorgchem.5b01416. The crystal structure of CuWO4, thermogravimetry curves of Mn1−xCuxWO4 (x = 0.15; 0.25), XRD pattern for Mn0.75Cu0.25WO4, lattice parameters and cell volume of Mn1−xCuxWO4 (x ≤ 0.25), combined Rietveld refinement plots of the XRD data and NPD data for Mn0.85Cu0.15WO4, variations of the metal-to-oxygen distances in Mn1−xCuxWO4, variation of the distortion parameter in Mn1−xCuxWO4, variation of selected O−O distances within or between WO 4 chains for Mn1−xCuxWO4 (x ≤ 0.25), phase transition temperatures in Mn1−xCu x WO 4 (x ≤ 0.25), inverse magnetic susceptibility for Mn1−xCuxWO4 (x ≤ 0.25), temperature dependence of the capacitance in zero magnetic field for Mn1−xCuxWO4 (x ≤ 0.25) (PDF) H

DOI: 10.1021/acs.inorgchem.5b01416 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry



(27) Stephens, P. W. J. Appl. Crystallogr. 1999, 32, 281−289. (28) Obermayer, H. A.; Dachs, H.; Schröcke, H. Solid State Commun. 1973, 12, 779−784. (29) Schofield, P. F.; Redfern, S. a. T. J. Phys.: Condens. Matter 1992, 4, 375. (30) Ruiz-Fuertes, J.; Bernert, T.; He, M.; Winkler, B.; Vinograd, V. L.; Milman, V. Appl. Phys. Lett. 2014, 105, 071911. (31) Ruiz-Fuertes, J.; Friedrich, A.; Pellicer-Porres, J.; Errandonea, D.; Segura, A.; Morgenroth, W.; Haussühl, E.; Tu, C.-Y.; Polian, A. Chem. Mater. 2011, 23, 4220−4226. (32) Iliev, M. N.; Gospodinov, M. M.; Litvinchuk, A. P. Phys. Rev. B: Condens. Matter Mater. Phys. 2009, 80, 212302. (33) Hoang, L. H.; Hien, N. T. M.; Choi, W. S.; Lee, Y. S.; Taniguchi, K.; Arima, T.; Yoon, S.; Chen, X. B.; Yang, I.-S. J. Raman Spectrosc. 2010, 41, 1005−1010. (34) Dai, R. C.; Ding, X.; Wang, Z. P.; Zhang, Z. M. Chem. Phys. Lett. 2013, 586, 76−80. (35) Gupta, H. C.; Jindal, R.; Sinha, M. M. Acta Phys. Pol., A 2012, 122, 142−145. (36) Schofield, P. F.; Knight, K. S.; Redfern, S. A. T.; Cressey, G. Acta Crystallogr., Sect. B: Struct. Sci. 1997, 53, 102−112. (37) Schofield, P. F. Mineral. Mag. 1994, 58, 185−199. (38) Šroubek, Z.; Ž d̆ánský, K. J. Chem. Phys. 1966, 44, 3078−3083. (39) Ehrenberg, H.; Weitzel, H.; Fuess, H.; Hennion, B. J. Phys.: Condens. Matter 1999, 11, 2649. (40) Tian, C.; Lee, C.; Xiang, H.; Zhang, Y.; Payen, C.; Jobic, S.; Whangbo, M.-H. Phys. Rev. B: Condens. Matter Mater. Phys. 2009, 80, 104426. (41) Ye, F.; Fishman, R. S.; Fernandez-Baca, J. A.; Podlesnyak, A. A.; Ehlers, G.; Mook, H. A.; Wang, Y.; Lorenz, B.; Chu, C. W. Phys. Rev. B: Condens. Matter Mater. Phys. 2011, 83, 140401. (42) Doumerc, J. P.; Dance, J. M.; Chaminade, J. P.; Pouchard, M.; Hagenmuller, P.; Krussanova, M. Mater. Res. Bull. 1981, 16, 985−990. (43) Melot, B. C.; Drewes, J. E.; Seshadri, R.; Stoudenmire, E. M.; Ramirez, A. P. J. Phys.: Condens. Matter 2009, 21, 216007. (44) Matityahu, S.; Aharony, A.; Entin-Wohlman, O. Phys. Rev. B: Condens. Matter Mater. Phys. 2012, 85, 174408. (45) Liu, J.; Koo, H.-J.; Xiang, H.; Kremer, R. K.; Whangbo, M.-H. J. Chem. Phys. 2014, 141, 124113. (46) Forsyth, J. B.; Wilkinson, C.; Zvyagin, A. I. J. Phys.: Condens. Matter 1991, 3, 8433. (47) Hollmann, N.; Hu, Z.; Willers, T.; Bohatý, L.; Becker, P.; Tanaka, A.; Hsieh, H. H.; Lin, H.-J.; Chen, C. T.; Tjeng, L. H. Phys. Rev. B: Condens. Matter Mater. Phys. 2010, 82, 184429. (48) Solovyev, I. V. Phys. Rev. B: Condens. Matter Mater. Phys. 2013, 87, 144403. (49) Meddar, L.; Josse, M.; Maglione, M.; Guiet, A.; La, C.; Deniard, P.; Decourt, R.; Lee, C.; Tian, C.; Jobic, S.; Whangbo, M.-H.; Payen, C. Chem. Mater. 2012, 24, 353−360. (50) Josse, M.; Meddar, L.; Deniard, P.; Jobic, S.; Payen, C.; Decourt, R.; Maglione, M. Ferroelectrics 2012, 428, 94−100.

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS We thank Lynda Meddar, Vincent Hardy, Bernard Humbert, and Myung-Hwan Whangbo for helpful discussions. We thank Rodolphe Decourt and Eric Gautron for their expert assistance during the PPMS and TEM experiments. The authors acknowledge the exploratory work carried out by Lynda Meddar. Vincent Hardy is thanked for his initial magnetization measurements.



REFERENCES

(1) Tokura, Y.; Seki, S. Adv. Mater. 2010, 22, 1554−1565. (2) Khomskii, D. Physics 2009, 2, 20. (3) Katsura, H.; Nagaosa, N.; Balatsky, A. V. Phys. Rev. Lett. 2005, 95, 057205. (4) Mostovoy, M. Phys. Rev. Lett. 2006, 96, 067601. (5) Arkenbout, A. H.; Palstra, T. T. M.; Siegrist, T.; Kimura, T. Phys. Rev. B: Condens. Matter Mater. Phys. 2006, 74, 184431. (6) Taniguchi, K.; Abe, N.; Takenobu, T.; Iwasa, Y.; Arima, T. Phys. Rev. Lett. 2006, 97, 097203. (7) Heyer, O.; Hollmann, N.; Klassen, I.; Jodlauk, S.; Bohatý, L.; Becker, P.; Mydosh, J. A.; Lorenz, T.; Khomskii, D. J. Phys.: Condens. Matter 2006, 18, L471. (8) Macavei, J.; Schulz, H. Z. Kristallogr. - Cryst. Mater. 1993, 207, 193−208. (9) Lautenschläger, G.; Weitzel, H.; Vogt, T.; Hock, R.; Böhm, A.; Bonnet, M.; Fuess, H. Phys. Rev. B: Condens. Matter Mater. Phys. 1993, 48, 6087−6098. (10) Chaudhury, R. P.; Lorenz, B.; Wang, Y. Q.; Sun, Y. Y.; Chu, C. W. Phys. Rev. B: Condens. Matter Mater. Phys. 2008, 77, 104406. (11) Song, Y.-S.; Chung, J.-H.; Park, J. M. S.; Choi, Y.-N. Phys. Rev. B: Condens. Matter Mater. Phys. 2009, 79, 224415. (12) Song, Y.-S.; Chung, J.-H.; Shin, K. W.; Kim, K. H.; Oh, I. H. Appl. Phys. Lett. 2014, 104, 252904. (13) Meddar, L.; Josse, M.; Deniard, P.; La, C.; André, G.; Damay, F.; Petricek, V.; Jobic, S.; Whangbo, M.-H.; Maglione, M.; Payen, C. Chem. Mater. 2009, 21, 5203−5214. (14) Kihlborg, L.; Gebert, E. Acta Crystallogr., Sect. B: Struct. Crystallogr. Cryst. Chem. 1970, 26, 1020−1026. (15) Van Uitert, L. G.; Sherwood, R. C.; Williams, H. J.; Rubin, J. J.; Bonner, W. A. J. Phys. Chem. Solids 1964, 25, 1447−1451. (16) Anders, A. G.; Zvyagin, A. I.; Kobets, M. I.; Pelikh, L. N.; Khats’ko, E. N.; Yurko, V. G. Zh. Eksp. Teor. Fiz. 1972, 62, 1798−1802. (17) Lake, B.; Tennant, D. A.; Cowley, R. A.; Axe, J. D.; Chen, C. K. J. Phys.: Condens. Matter 1996, 8, 8613. (18) Liang, K.-C.; Wang, Y. Q.; Sun, Y. Y.; Lorenz, B.; Chu, C. W. Ferroelectrics 2014, 470, 43−51. (19) Kumar, C. M. N.; Xiao, Y.; Lunkenheimer, P.; Loidl, A.; Ohl, M. Phys. Rev. B: Condens. Matter Mater. Phys. 2015, 91, 235149. (20) Yourey, J. E.; Kurtz, J. B.; Bartlett, B. M. Inorg. Chem. 2012, 51, 10394−10401. (21) Petricek, V.; Dusek, M.; Palatinus, L. Z. Kristallogr. - Cryst. Mater. 2014, 229, 345−352. (22) Cheary, R. W.; Coelho, A. A. J. Appl. Crystallogr. 1998, 31, 851− 861. (23) Cheary, R. W.; Coelho, A. A. J. Appl. Crystallogr. 1998, 31, 862− 868. (24) Bain, G. A.; Berry, J. F. J. Chem. Educ. 2008, 85, 532. (25) Tomaszewicz, E.; Typek, J.; Kaczmarek, S. M. J. Therm. Anal. Calorim. 2009, 98, 409−421. (26) Becker, P.; Bohatý, L.; Eichler, H. j.; Rhee, H.; Kaminskii, A. a. Laser Phys. Lett. 2007, 4, 884−889. I

DOI: 10.1021/acs.inorgchem.5b01416 Inorg. Chem. XXXX, XXX, XXX−XXX