magnetic-structure-stabilized polarization in an above-room-...
TRANSCRIPT
-
Polar MagnetDOI: 10.1002/anie.201406180
Magnetic-Structure-Stabilized Polarization in an Above-Room-Temperature Ferrimagnet**Man-Rong Li, Maria Retuerto, David Walker, Tapati Sarkar, Peter W. Stephens,Swarnakamal Mukherjee, Tanusri Saha Dasgupta, Jason P. Hodges, Mark Croft,Christoph P. Grams, Joachim Hemberger, Javier S�nchez-Ben�tez, Ashfia Huq, Felix O. Saouma,Joon I. Jang, and Martha Greenblatt*
Abstract: Above-room-temperature polar magnets are ofinterest due to their practical applications in spintronics. Herewe present a strategy to design high-temperature polarmagnetic oxides in the corundum-derived A2BB’O6 family,exemplified by the non-centrosymmetric (R3) Ni3TeO6-typeMn2+2Fe
3+Mo5+O6, which shows strong ferrimagnetic orderingwith TC = 337 K and demonstrates structural polarizationwithout any ions with (n�1)d10ns0, d0, or stereoactive lone-pair electrons. Density functional theory calculations confirmthe experimental results and suggest that the energy of themagnetically ordered structure, based on the Ni3TeO6 proto-type, is significantly lower than that of any related structure,and accounts for the spontaneous polarization (68 mC cm�2)and non-centrosymmetry confirmed directly by second har-monic generation. These results motivate new directions in thesearch for practical magnetoelectric/multiferroic materials.
The design of polar and magnetic materials with magneto-electric/multiferroic behavior is a key issue to developspintronic devices for nonvolatile memories, faster dataprocessing speeds with less power usage, larger storagedensities, and additional functionalities.[1,2] There are nowseveral demonstrated strategies to achieve electric polariza-tion of magnetic compounds, including the lone-pair-electron-active cation,[3] charge ordering,[4] and magnetic-structure(spin spiral)[5]-driven multiferroics, the strain-induced thinfilm technique,[6] and polarization from geometric structuredistortion,[7] including second-order Jahn–Teller (SOJT) dis-tortion[8] and the recently predicted hybrid improper ferro-electricity in corner-sharing BO6 network with “layered” A/A’ cations in perovskite,[9a,b] Ruddlesden–Popper,[9b,c] andDion–Jacobson[9d] phases. So far, most of the materials-by-design work has focused on ABO3 perovskites and relatedphases with either (n�1)d10ns0,[10] d0,[7] or stereoactive lone-pair electron configuration cations,[3] to favor hybridization
[*] Dr. M. R. Li, M. Retuerto, T. Sarkar, M. GreenblattDepartment of Chemistry and Chemical Biology, RutgersThe State University of New Jersey610 Taylor Road, Piscataway, NJ 08854 (USA)E-mail: [email protected]
Dr. D. WalkerLamont-Doherty Earth Observatory, Columbia University61 Route 9W, Palisades, NY 10964 (USA)
Dr. P. W. StephensDepartment of Physics & Astronomy, State University of New YorkStony Brook, NY 11794 (USA)
Dr. S. Mukherjee, T. S. DasguptaDepartment of Condensed Matter Physics and Materials SciencesS. N. Bose National Centre for Basic SciencesJD Block, Sector III, Salt Late, Kolkata 700098 (India)
Dr. J. P. Hodges, A. HuqSpallation Neutron Source, Oak Ridge National LaboratoryOak Ridge, TN 37831 (USA)
Dr. M. CroftDepartment of Physics and Astronomy, RutgersThe State University of New Jersey136 Frelinghuysen Road, Piscataway, NJ 08854 (USA)
Dr. C. P. Grams, J. HembergerII. Physikalisches Institut, Universit�t zu Kçln50937 Kçln (Germany)
Dr. J. S�nchez-Ben�tezDepartamento de Qu�mica F�sica I, Facultad de Ciencias Qu�micasUniversidad Complutense de Madrid28040 Madrid (Spain)
Dr. F. O. Saouma, J. I. JangDepartment of Physics, Applied Physics and AstronomyBinghamton UniversityP.O. Box 6000, Binghamton, NY 13902 (USA)
[**] This work was supported by the NSF-DMR-0966829 and ARO-434603 (DOD-VV911NF-12-1-0172) grants. The use of the NationalSynchrotron Light Source, Brookhaven National Laboratory, wassupported by the U.S. Department of Energy, Office of Science,Office of Basic Energy Sciences, under contract no. DE-AC02-98CH10886. A part of this research at ORNL’s High Flux IsotopeReactor and Spallation Neutron Source was sponsored by theScientific User Facilities Division, Office of Basic Energy Sciences,U.S. Department of Energy. We thank J. Hanley at LDEO (ColumbiaUniversity) for making the high-pressure assemblies, Dr. W. Zhangand P. S. Halasyamani (University of Houston) for fruitful dis-cussion about the SHG measurements, and Dr. F. Mompean for hishelp in the conductivity measurements. Dr. J. S�nchez-Ben�tez issupported by the Spanish project MAT2013-41099-R.
Supporting information for this article is available on the WWWunder http://dx.doi.org/10.1002/anie.201406180. Further details onthe crystal structure investigation may be obtained from theFachinformationszentrum Karlsruhe, 76344 Eggenstein-Leopold-shafen, Germany (Fax: (+ 49)7247-808-666; E-Mail: [email protected]), on quoting the depository numbers CSD-427790 and-427791.
.AngewandteCommunications
10774 � 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Angew. Chem. Int. Ed. 2014, 53, 10774 –10778
http://dx.doi.org/10.1002/anie.201406180
-
between the metal open-shell d and oxygen 2p states,[11a] orprovide ferroelectrically active electron pairs.[11b] In mostobserved cases the multiferroic order is well below roomtemperature (RT); only a few candidates, such as thehexaferrites and related phases,[12a] BiFeO3,
[3] Cr2O3,[12b] e-
Fe2O3’, and its doped analogues,[12c] show promising magneto-
electric phenomena above RT. Therefore, new practicalmagnetoelectric materials are needed.
Recently, the corundum-derived A2BB’O6 oxides withunusually small A-site cations have drawn much atten-tion,[13a,b] particularly Ni3TeO6, which was reported to shownonhysteretic colossal magnetoelectricity.[13b] The crystalstructure of A2BB’O6 (Figure S1 in the Supporting Informa-tion, SI) makes it possible to incorporate strong magnetictransition metal ions on both the A- and B-sites for magneticand potential magnetoelectric behavior. A good example isour discovery of Mn2+2Fe
3+M5+O6 (M = Nb5+ and Ta5+; d0
ions), with LiNbO3 (LN)-type polar structures, antiferromag-netic (TN� 90 K) ordering, spin localization, and structuraldistortions that predict many more interesting and usefulpolar magnetic phases.[13a] Therefore, it appeared promising todesign above-RT polar ferri- or ferromagnets by compositionmodulation. In this work, we introduced Mo5+ into the B’-siteof Mn2FeMoO6, and studied its crystal structure, oxidationstates, optical second harmonic generation (SHG) activity,and the magnetic and electrical properties. Theoreticalcalculations were also performed for further insight intostructure–property relationships. We have found a new mag-netic and polar phase without lone-pair-electron or d0
configuration cations, or incommensurate or helicoidal mag-netic structures. Mn2FeMoO6 stabilizes a magnetic structurethat brings on spontaneous polarization.
Mn2FeMoO6, prepared at 1623 K under 8 GPa, crystal-lizes in the non-centrosymmetric Ni3TeO6-type (R3) structureas determined from combined refinements of powder syn-chrotron X-ray (PSXD) and neutron diffraction (PND) data(Section 2 of SI). There are two crystallographically inde-pendent Mn ions (Mn1 and Mn2) at the A-sites; Fe and Moare ordered over the B-sites. No oxygen deficiency wasobserved on the two oxygen positions (O1 and O2). Thepresence of cationic antisite disorder was also examined, withthe best refinement indicating about 7% B-site Fe/Modisorder, yielding the nominal composition: Mn2-(Fe0.93(1)Mo0.07(1))(Mo0.93(1)Fe0.07(1))O6 (Table S1). No Mn/Fedisordering was observed during the refinements. The crystalstructure along the c-axis is displayed in Figure 1a. TheMn1O6/MoO6 and Mn2O6/FeO6 face-sharing octahedral pairsalong the c-axis form zigzag chains through edge-sharing withedge-shared Mn1O6/FeO6 and Mn2O6/MoO6 octahedral pairsin the ab-plane. The cations near the centers of the face-shared octahedral pairs are displaced by electrostatic repul-sions (see the very large displacements in Mn1O6/MoO6 andMn2O6/FeO6 octahedral pairs in Figure 1b and c) and causelarge octahedral distortions comparable to those in LiNbO3(Table S2), in which the extended structure displays a polarmoment along the c-axis. The ordering of Fe and Mo over theB-sites renders different atomic displacements (dMn2 and dMn1in Figure 1b and c) of their face-shared pairs and conclusivelyaccounts for the spontaneous polarization (PS). The polar
structure of Mn2FeMoO6 is directly confirmed by the activeSHG response as shown in Figure S5. The crystal structureanalysis together with the bond valence sums (BVS) calcu-lations (Table S2) indicate that the formal oxidation states areMn2+ (d5), Fe3+ (d5), and Mo5+ (d1), as further confirmed bythe X-ray absorption near-edge spectroscopic (XANES)studies (Section 4 of SI).
Figure 2a shows the evolution of magnetic susceptibilitycdc versus T for Mn2FeMoO6. When the temperature islowered from 400 K, the sample undergoes a paramagnetic toferromagnetic transition with a sharp rise of the magnet-ization below a magnetic transition temperature (TC) of ca.350 K. A more precise TC = 337 K can be estimated from theinflection point in the dcdc/dT versus T curve (top inset ofFigure 2a). The high-temperature 1/cdc versus T plot (bottominset of Figure 2a) shows a deviation from the Curie–Weisslaw but can be nicely fitted by the N�el model
Figure 1. a) Crystal structure of Mn2FeMoO6 viewed along the [110]direction. b) and c) Enlarged views of the face-sharing Mn1O6/MoO6and Mn2O6/FeO6 octahedral pairs along the c-axis to show thedistortions and atomic displacements from half-way between the c-axisoxygen planes marked by dashed lines. dMn1 = 0.506 �, dMo = 0.196 �,dMn2 = 0.486 �, dFe = 0.267 �.
AngewandteChemie
10775Angew. Chem. Int. Ed. 2014, 53, 10774 –10778 � 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.angewandte.org
http://www.angewandte.org
-
1c ¼
1c0þ TC �
s
T�q
� �for a two-sublattice ferrimagnet (black line
in top inset of Figure 2a),[14] giving c0 = 0.016(1) emumol�1 Oe, C = 14.4(2) emuK mol�1 Oe, s = 8329(4) molOeemu�1 K, and q = 239.9(1) K. From the value of C, we couldestimate an effective magnetic moment meff = 10.7mB, which is
close to the calculated spin-only moment per formula unit(mcalc. = 10.4mB). The ferrimagnetism is also confirmed by theisothermal magnetization curves recorded at 5 and 300 K(Figure 2b). The PND data show magnetic contributions insome low-angle reflections at both 10 and 300 K. Themagnetic intensities at 10 K have been refined in a modelthat considered the following ferrimagnetic arrangement (seeTableS1): in the face-sharing Mn1O6/MoO6 pairs the spins ofMn1 (4.0(3)mB) are along the c direction and antiparallel tothe Mo moments (�1.1(1) mB). Similarly, in the face-sharingMn2O6/FeO6 pairs, the spins of Mn2 (�3.8(2) mB) and Fe(3.4(5)mB) are also antiparallel along the c axis. In the abplanes, Mn1 and Fe spins order ferromagnetically in onelayer, whereas those of Mn2 and Mo in the next layer alsoorder ferromagnetically, but antiparallel to those of Mn1 andFe. The magnetic structure is shown in Figure 2 c. The netmagnetic moment from PND refinements (2.5(3) and0.81(8)mB for 10 and 300 K, respectively) is comparable tothe saturation magnetization in Figure 2b (� 1.98 and0.95mB mol
�1 at 5 and 300 K, respectively).The plot of resistivity 1 versus T for Mn2FeMoO6 at H =
0 T (Figure 3) is characteristic of semiconducting behaviorwith a resistivity value of around 20 Wcm at 300 K. Theresistivity becomes too high to measure below 80 K. In theentire temperature range between 80 and 300 K, the resis-tivity follows a Mott�s variable range hopping (VRH)[15]
conduction mechanism:
1 ¼ 10 exp T0T� �1
=4" #
as seen in the linear fit in the plot of ln1 versus 1/T1/4 (blackline in inset of Figure 3). The fitting allowed us to extract theparameters T0 and 10, 1.15 � 10
8 K and 2.6 � 10�10 Wcm,respectively, which are similar to the observed values inother transition metal oxides.[16] Thus, Mn2FeMoO6 is a ferri-magnetic VRH semiconductor.
Figure 3. 1 versus T plot at zero-field showing semiconductor behavior.The inset shows the linear fit to the plot of ln1 versus 1/T1/4, indicatingMott’s VRH conduction mechanism.
Figure 2. a) cdc versus T in the zero-field cooled as well as field cooledmodes from 5 to 400 K with H =0.1 T; the bottom (left) inset showsthe dcdc/dT versus T curve; the top (right) inset shows the 1/cdc versusT curve, which could be fitted to the N�el model for a two-sublatticeferrimagnet at higher temperature. b) M versus H curves at T = 5 and300 K; the inset shows the expanded region between �1.5 and 1.5 T,showing clear hysteresis loops at both temperatures. c) Magneticstructure at 10 K refined from PND data demonstrating the spinpolarization, the net magnetic moment M =m(Mn1)+ m(Fe)�m(Mn2)�m(Mo).
.AngewandteCommunications
10776 www.angewandte.org � 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Angew. Chem. Int. Ed. 2014, 53, 10774 –10778
http://www.angewandte.org
-
DFT calculations (Section 6 of SI) confirmed the high-spin nominal d5 state of both Fe3+ and Mn2+, and that of Mo5+
to be nominally d1. A somewhat large value of magneticmoment (� 0.10–0.12 mB) is found to reside at the O site,suggesting strong d–p hybridization between the transitionmetals and O. The ferrimagnetic spin structure (cited as up-up-down-down along the zigzag chain) in Figure 2c isconfirmed to be lower in energy than the ferromagnetic(spins of Mn, Fe, and Mo all aligned parallel) and ferrimag-netic (antiparallel alignment of Mn1 and Mn2, but parallelalignment of Fe and Mo, cited as up-up-down-up) solutions bya large energy difference of about 300 and 80 meV, respec-tively, which is consistent with PND data analysis. Thissuggests a very strong antiferromagnetic exchange betweenMn1 and Mn2 and a moderately strong antiferromagneticexchange between Fe and Mo. In Figure 4 the density of states(DOS) in the ground state (up-up-down-down magneticconfiguration) supports the above conclusions. The magneticas well as the conducting properties (the presence or absenceof band gap) seem to be highly influenced by the assumedconfiguration of the Fe–Mo arrangement whereas the nom-inal valences remain unaffected, a fact well established fordouble perovskites.[17] This presumably explains the VRHsemiconducting behavior dominated by disorder (� 7%) aswell as the experimental observation of a net moment of ca.
2.5 mB, which is highly suppressed compared to the theoreticalestimate of 4.0 mB based on a perfectly ordered state, and alsosuggests possible resistivity tuning by controlling the Fe/Modisordering degree.
The crystal structure of Mn2FeMoO6 breaks the polar-ization rule, because known polar perovskite-related andcorundum-based compounds contain ions with either(n�1)d10ns0, or d0 electronic configuration,[11a] or at leastone stereoactive lone pair.[3] To the best of our knowledge,polar e-Fe2O3 is the only exception with only d
n (0
-
.Keywords: density functional calculations · ferromagnets ·polar magnets · second harmonic generation
[1] N. A. Spaldin, M. Fiebig, Science 2005, 309, 391 – 392.[2] W. Eerenstein, M. D. Mathur, J. F. Scott, Nature 2006, 442, 759 –
765.[3] J. Wang, et al., Science 2003, 299, 1719 – 1722 (see the
Supporting Information).[4] N. Ikeda, et al., Nature 2005, 436, 1136 – 1138 (see the
Supporting Information).[5] T. Kimura, T. Goto, H. Shintani, K. Ishizaka, T. Arima, Y.
Tokura, Nature 2003, 426, 55 – 58.[6] a) R. Ramesh, N. A. Spaldin, Nat. Mater. 2007, 6, 21 – 29; b) J. H.
Lee, et al. Nature 2010, 466, 954 – 958 (see the SupportingInformation).
[7] T. Varga et al., Phys. Rev. Lett. 2009, 103, 047601 (see theSupporting Information).
[8] P. S. Halasyamani, K. R. Poeppelmeier, Chem. Mater. 1998, 10,2753 – 2769.
[9] a) J. M. Rondinelli, C. J. Fennie, Adv. Mater. 2012, 24, 1961 –1968; b) A. T. Mulder, N. A. Benedek, J. M. Rondinelli, C. J.Rennie, Adv. Funct. Mater. 2013, 23, 4810 – 4820; c) N. A.Benedek, C. J. Fennie, Phys. Rev. Lett. 2011, 106, 107204;d) N. A. Benedek, Inorg. Chem. 2014, 53, 3769 – 3777.
[10] A. A. Belik, T. Furubayashi, H. Yusa, E. Takayama-Muromachi,J. Am. Chem. Soc. 2011, 133, 9405 – 9412.
[11] a) R. E. Cohen, Nature 1992, 358, 136 – 138; b) N. A. Hill, J.Phys. Chem. B 2000, 104, 6694 – 6709.
[12] a) Y. Kitagawa, Y. Hiraoka, T. Honda, T. Ishikura, H. Naka-mura, T. Kimura, Nat. Mater. 2010, 9, 797 – 802; b) G. T. Rado,V. J. Folen, Phys. Rev. Lett. 1961, 7, 310 – 311; c) J. Tuček, R.Zbořil, A. Namai, S. i. Ohkoshi, Chem. Mater. 2010, 22, 6483 –6505.
[13] a) M. R. Li, et al. Angew. Chem. 2013, 125, 8564 – 8568; Angew.Chem. Int. Ed. 2013, 52, 8406 – 8410 (see the SupportingInformatioin); b) Y. S. Oh, S. Artyukhin, J. J. Yang, V. Zapf,J. W. Kim, D. Vanderblt, S. W. Cheong, Nat. Commun. 2014, 5,3201.
[14] L. N�el, Ann. Phys. 1948, 3, 62.[15] N. F. Mott, Conduction in non-crystalline materials, 2nd ed.,
Clarendon, Oxford, 1993.[16] J. Fontcuberta, B. Mart�nez, A. Seffar, S. PiÇol, J. L. Garc�a-
MuÇoz, X. Obradors, Phys. Rev. Lett. 1996, 76, 1122 – 1125.[17] T. Saha-Dasgupta, D. D. Sarma, Phys. Rev. B 2001, 64, 064408.[18] R. Shannon, Acta Crystallogr. Sect. A 1976, 32, 751 – 767.[19] S. C. Abrahams, S. K. Kurtz, P. B. Jamieson, Phys. Rev. 1968, 172,
551 – 553.[20] a) J. Y. Son, G. Lee, M. H. Jo, H. Kim, H. M. Jang, Y. H. Shin, J.
Am. Chem. Soc. 2009, 131, 8386 – 8387; b) Y. Inaguma, D.Sakurai, A. Aimi, M. Yoshida, T. Katsumata, D. Mori, J. Yeon,P. S. Halasyamani, J. Solid State Chem. 2012, 195, 115 – 119.
[21] W. Eerenstein, F. D. Morrison, J. Dho, M. G. Blamire, J. F. Scott,N. D. Mathur, Science 2005, 307, 1203 – 1203.
.AngewandteCommunications
10778 www.angewandte.org � 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Angew. Chem. Int. Ed. 2014, 53, 10774 –10778
http://dx.doi.org/10.1021/cm980140whttp://dx.doi.org/10.1021/cm980140whttp://www.angewandte.org
-
Supporting Information
� Wiley-VCH 2014
69451 Weinheim, Germany
Magnetic-Structure-Stabilized Polarization in an Above-Room-Temperature Ferrimagnet**Man-Rong Li, Maria Retuerto, David Walker, Tapati Sarkar, Peter W. Stephens,Swarnakamal Mukherjee, Tanusri Saha Dasgupta, Jason P. Hodges, Mark Croft,Christoph P. Grams, Joachim Hemberger, Javier S�nchez-Ben�tez, Ashfia Huq, Felix O. Saouma,Joon I. Jang, and Martha Greenblatt*
anie_201406180_sm_miscellaneous_information.pdf
-
1
SUPPORTING INFORMATIOIN
Full information of References [3], [4], [7] and [13a]:
[3] J. Wang, J. B. Neaton, H. Zheng, V. Nagarajan, S. B. Ogale, B. Liu, D. Viehland, V.
Vaithyanathan, D. G. Schlom, U. V. Waghmare, N. A. Spaldin, K. M. Rabe, M. Wuttig, R.
Ramesh, Science 2003, 299, 1719-1722
[4] N. Ikeda, Ohsumi, K. Ohwada, K. Ishii, T. Inami, K. Kakurai, Y. Murakami, K. Yoshii, S.
Mori, Y. Horibe, H. Kitô, Nature 2005, 436, 1136-1138.
[7] T. Varga, A. Kumar, E. Vlahos, S. Denev, M. Park, S. Hong,T. Sanehira, Y. Wang, C. J. Fennie, S. K.
Streiffer,X. Ke, P. Schiffer, V. Gopalan, & J. F. Mitchell, Phys. Rev. Lett. 2009, 103, 047601.
[13a] M. R. Li, D. Walker, M. Retuerto, T. Sarkar, J. Hadermann, P. W. Stephens, M. Croft, A. Ignatov,
C. P. Grams, J. Hemberger, I. Nowik, P. S. Halasyamani, T. T. Tran, S. Mukherjee, T. Saha-Dasgupta, M.
Greenblatt, Angew. Chem. 2013, 125, 8564-8568; Angew. Chem. Int. Ed. 2013, 52, 8406-8410.
-
2
Content
1. Cation Ordering in Corundum Derivatives
2. High Pressure Synthesis and Powder Synchrotron X-ray and Neutron Diffraction Studies
3. Second Harmonic Generation Measurements
4. X-ray Absorption Near-Edge Spectroscopy
5. Magnetic, Magnetotransport, and Electrical Conductivity Properties Measurements
6. Theoretical Calculations
7. Dielectric and Polarization Measurements
1. Cation Ordering in Corundum Derivatives
The crystal structure of corundum-type (A2O3) oxides consists of hexagonal close packing
of the oxygen atoms, where the metal atoms occupy two-thirds of the octahedral sites, the
remaining one-third of the octahedral sites are vacant (Fig. S1a).1-3
Each AO6 octahedron shares
edges with three other AO6 octahedra to form an intralayer edge-sharing (ES) infinite sheet in the
ab plane which is 2/3 occupied and 1/3 vacant. The intralayer ES octahedral sheets are bonded to
form a three-dimensional structure via interlayer face-sharing (FS) to form octahedral pairs along
the c direction, these pairs corner share with the edge-sharing slabs of alternative layers. The
metal atoms near the center of FS octahedral pairs tend to displace away in opposite directions
towards the vacant sites in order to reduce the electrostatic repulsions.
-
3
(a) Corundum (R-3c)
(b) Ilmenite (R-3) (c) LiNbO3 (R3c)
(d) Ordered ilmenite (R3) (e) Ni3TeO6 (R3)
Intralayer ES
Interlayer FS
AO6/BO6
BO6
AO6
A1O6/A2O6
BO6/B O6
A1A2
A1O6/BO6
A2O6/B O6
Fig. S1 Cation ordering tree in corundum-related phases with (a) A2O3-type corundum;
(b) ABO3-type ilmenite; (c) ABO3-type LiNbO3; (d) A2BB´O6-type ordered ilmenite, and
(e) A2BB´O6-type Ni3TeO6 crystal structures viewed along the [110] direction.
Octahedral color: carmine, AO6; purple, BO6; cyan, B´O6.
-
4
Given the same corundum-type rhombohedral stacking of octahedra, different crystal
structures can be derived when the octahedral metal sites are occupied by more than one distinct
cation (similar to the double perovskite related A2BB´O6). Fig. S1 shows the cation ordering tree
in corundum-derived systems, including ilmenite (IL), LiNbO3 (LN), ordered IL (OIL), and
Ni3TeO6 type structures,2,4,5
where the FS octahedral pairs along the c axis are formed by two
different cation octahedra. The crystal structures are highly strained: stretched in the ab plane
and compressed in the c direction due to the electrostatic repulsions between the cations across
the ES and FS octahedral pairs, respectively.
1.1 IL and LN type Structures
The IL and LN structures have two distinct cation-orderings in ABO3. IL adopts a
centrosymmetric R-3 space group, built by alternating ES AO6 and BO6 octahedral layers in the
ab plane stacked via FS between unlike A and B cations (Fig. S1b). While in the non-
centrosymmetric LN structure (R3c), the ES octahedral layers consist of both AO6 and BO6
octahedra to avoid intralayer ES between identical cations (Fig. S1c). The number of reports on
related phases increased in the past years, but because of the requirement of relatively high
pressures for their preparation, these systems are far less explored than normal ABO3 perovskites.
1.2 Ordered Ilmenite and Ni3TeO6-type Structures
The A2BB´O6 system is more complex (than IL and LN ABO3), as an additional ordering
parameter is introduced into the B-sites. The crystal structure transforms to either OIL or
Ni3TeO6 type. Generally, B´ has a higher oxidation state than B; A is typically a monovalent-to-
trivalent cation. So far, only a few compounds have been reported in this system: Li2GeTeO6 is
-
5
the only phase with OIL structure bearing some B-site cationic disorder, which results in
Li2[(Ge0.82Te0.18)B(Te0.82Ge0.18)B’]O6.2 Several Ni3TeO6-type compounds were reported with
A+
2B4+
B´6+
O6,6 A
2+2B
2+B´
6+O6,
4,5 A
2+2B
3+B´
5+O6,
7 and A
3+2B
2+B´
4+O6
8 formulae, all of which
contain an ion with electronic configuration (n-1)d10
ns0 such as Ge
4+,8 Sb
5+,7 and Te
6+,4-6
at the
B'-site. Compared with IL, the OIL (R3) has two crystallographically independent A cations
(designated as A1 and A2). The ES layers are formed by either A or B octahedra and the
intralayer ordering gives cystallographically unlike neighbors in ES A1O6/A2O6 layers and ES
BO6/B'O6 layers (Fig. S1d), respectively. The interlayer FS A1O6/B'O6 and A2O6/BO6
octahedral pairs form A1-B'-A2-B columns parallel to the c axis. As shown in Fig. S1d, the A
cations displace away from their FS partner cations (A1-B'; A2-B) toward opposite directions
along the c axis; this displacement is largely responsible for the two crystallographically
different A-site positions due to the size and charge difference between B and B’ cations. The
Ni3TeO6 structure is a close derivate of the LN structure; compared with OIL it has additional
intralayer octahedral ordering to avoid ES between like A or B cations, and thus forming
alternative stacking of A1O6/BO6 and A2O6/B'O6 octahedral layers in the ab-plane (Fig. S1e).
The cation ordering in corundum derivatives is complex and can be driven by many factors,
including the cationic size and charge difference, structural tolerance factor (t), electron
configuration, and synthesis condition. The larger the size or charge difference between the
cations, the greater is the chance of ordering. The corundum-derived ABO3 and A2BB'O6
systems can be regarded as highly distorted perovskite-related structures with very small
tolerance factor (t) values9, due to the unusually small size of A cations. The A-site 12-fold
coordination of the ideal perovskite structure transforms to an octahedral 6-fold coordination.
These structures are highly distorted and compact, most are metastable at ambient pressure and
-
6
often require very high pressure for their synthesis. 10-12
For example, some IL-type structures
transform into LN-polymorphs under higher pressure.13,14
Another important effect on the cation
ordering is the electronic configuration. For example, Mn2FeMO6 (M = Nb5+
and Ta5+
, d0
,)
adopt a polar LN structure,12
while Mn2Fe3+
Sb5+
O6 (Sb5+
, d10
) a non-polar IL structure. We
believe the latter behavior is due to the absence of a second-order Jahn-Teller d0 cation at the B’-
sites, which favors distortion of the octahedra.11
Generally, the cation ordering is affected by the competition and/or combination of several
factors for the lowest energy of the system, and the real outcome can be complicated and subtle
case by case. For example, both Bi2FeCrO6 and Bi2FeTiO6 were predicted to stabilize in the
Ni3TeO6 structure.15,16
However, Bi2FeCrO6 prepared under HP crystallizes in the LN structure
(R3c),11
while Bi2FeTiO6 has not been experimentally prepared yet.17
2. High Pressure Synthesis and Powder Synchrotron X-ray and Neutron Diffraction Studies
Polycrystalline Mn2FeMoO6 was prepared from stoichiometric mixtures of MnO (99.99%,
Alfa Aesar), Fe (99.999%, Alfa Aesar), Fe2O3 (99.999%, Sigma Aldrich) and MoO3 (99.998%
Alfa Aesar) under high pressure. The oxide mixture was placed in a Pt capsule, pressurized
typically over 8-12 hours, and reacted at 1623 K under 8 GPa for 30 minutes in a LaCrO3 heater
inside a MgO crucible in a multi-anvil press,18
and then quenched to RT by turning off the
voltage supply to the resistance furnace, which reduced the temperature to RT in a few seconds.
The pressure is maintained during the temperature quenching and then decompressed in 8-12
hours. Powder synchrotron x-ray diffraction (PSXD) data were recorded on beam line X-16C (λ
= 0.69991 Å) at the National Synchrotron Light Source (NSLS), Brookhaven National
Laboratory (US). Powder neutron diffraction (PND) data were collected at 10 and 300 K on
-
7
~220 mg sample at the POWGEN instrument in the Spallation Neutron Source in the Oak Ridge
National Laboratory (US). Diffraction data analysis and Rietveld refinements19
were performed
with the TOPAS software package20
and EXPGUI interface of GSAS programs21
.
Original powder synchrotron x-ray diffraction (PSXD) data can be indexed to either
centrosymmetric R-3 or noncentrosymmetric R3 cell, corresponding to IL (R-3), OIL (R3), or
Ni3TeO6 (R3) type structure. Although the two space groups have the same reflection conditions,
Rietveld refinements on PSXD data indicated unambiguously the non-centrosymmetric Ni3TeO6
structure based on the structural model in Ref. 4. More important than the relatively small
difference in the refinement statistics (Rwp = 7.72% (OIL), 7.29% (Ni3TeO6), and 8.62% (IL))
was the fact that the IL and OIL structural models yielded unrealistic atomic displacement
parameters. The Ni3TeO6-type structure was further reinforced by combined PSXD and PND
refinements attributed to the different counting of the electron and neutron scattering ability
(electron numbers for PSXD: ZMn = 25, ZFe = 26, ZMo = 42, ZO = 8; neutron scattering lengths for
PND: bc,Mn = -3.73 × 10-15
m; bc,Fe = 9.45 × 10-15
m, bc,Mo = 6.715 × 10-15
m; bc,O = 5.803 × 10-15
m), as totally unacceptable fitting of the 300 K data in IL (χ2 = 21.28, Rp/Rwp = 10.99/10.50%)
and OIL (χ2 = 26.17, Rp/Rwp = 14.75/11.75%) models obtained. The final refined plots from
Ni3TeO6 structure are presented in Fig. S2 (combined PSXD and PND refinements on 300 K
data (χ2 = 1.91, Rp/Rwp = 7.69/3.04%)) and Fig. S3 (PND data at 10 K (χ
2 = 2.33, Rp/Rwp =
3.73/1.99%), the fitting and crystallographic parameters are listed in Table S1. Selected
interatomic distances and bond angles are listed in Table S2. There are two crystallographically
independent Mn atoms (Mn1 and Mn2) occupying the A-sites (3a, (0 0 z)), and Fe (3a, (0 0 z))
and Mo (3a, (0 0 0)) are ordered over the B-sties. The oxygen atoms are driven to two positions
(9h (x ,y ,z)) due to the higher degree of cation ordering compared with that of the IL structure.
-
8
The anti-site disorder was also refined (constrained to the initial stoichiometry Fe : Mo = 1 : 1), by
assuming that some Fe can randomly replace Mo atoms, and vice versa, giving ~7% of Fe/Mo
disordering. The octahedral chains in the crystal structure of Mn2FeMoO6 are connected to form
infinite octahedral slabs (thickness ~3.4 Å, Fig. S4, left) via corner-sharing between Mn1O6-
MoO6 and Mn2O6-FeO6 neighboring chains (Fig. S4, right). The zigzag chains and slab packing
leave vacancies right above and below the face-shared octahedral pairs, which favor the
stabilization of the structure.
The octahedral distortions (Figures 1b and c) lead to three short and three long metal-
oxygen bonds for each metal site (Table S2), with the average and
distances of 2.195(3) Å at RT, which are comparable to those of the MnO6 octahedra in the
corundum derivatives, Mn22+
FeSbO6 (2.173 Å)10
and Mn22+
FeMO6 (M = Nb (2.16 Å) and Ta
(2.22 Å))12
. The distance (2.033(4) Å) is also comparable to octahedral in
similar phases as in Mn22+
Fe3+
SbO6 (2.006 Å)10
and Mn2Fe3+
MO6 (M = Nb (2.030 Å) and Ta
(2.012 Å));12
the average distance (2.008(2) Å) is larger than in other related ordered
double perovskites, e.g., Sr2FeMoO6 where it is 1.947 Å22
, which indicates a lower oxidation
state of Mo in Mn2FeMoO6 compared to the ordered double perovskite where Mo is considered
5+/6+ mixed valent23
. The crystal structure at 10 K is similar to that at 300 K, but with smaller
unit cell and atomic displacement parameters as expected.
-
9
2θ (degree, λ = 0.69991 Å)10 20 30 40
Inte
nsi
ty (
a.u
.)
2 4 6 8D-space (Å)
Inte
nsi
ty (
a.u
.)
1 2 3
Inte
nsi
ty (
a.u
.)
(a)
(b)
(c)
Fig. S2 Experimental, calculated, and difference of (a) PSXD and (b-c) PND patterns of
Mn2FeMoO6 at 300 K, corresponding to the combined refinements in Ni3TeO6-type
structure. In (b) and (c) PND data are from the low and high D-spacing banks,
respectively. The vertical bars (│) show the peak index of vanadium (sample can), and
nuclear and magnetic structure, respectively, from top to bottom.
-
10
Fig. S3 Experimental, calculated, and difference of PND patterns of Mn2FeMoO6 at 10
K, corresponding to the refinements in Ni3TeO6-type structure. (a) and (b): PND data
from low and high D-spacing banks, respectively. Inset of (b) shows the peak intensity
differences of data at 10 and 300 K due to the evolution of the magnetic structure. The
vertical bars (│) show the peak index of vanadium (sample can), and nuclear and
magnetic structure, respectively, from top to bottom.
-
11
Table S1 Atomic parameters and agreement factors after the Rietveld refinements of
PND (10 K) and combined PND and PSXD (300 K) data of Mn2FeMoO6 in Ni3TeO6-type
structure (rhombohedral, space group of R3 (No. 146), Z = 3).
Temperature/K 10 300
a/Å
c/Å
V/Å3
Mn1 (3a, 0 0 zp)
zp
Uiso (× 100)/Å2
Mag. mom./µB
Mn2 (3a, 0 0 zp)
zp
Uiso (× 100)/Å2
Mag. mom./µB
(Fe/Mo)1 (3a, 0 0 zp)
Occ.
zp
Uiso (× 100)/Å2
Mag. mom./µBiii
(Mo/Fe)2 (3a, 0 0 0)
Occ.
Uiso (× 100)/Å2
Mag. mom./µBiii
O1 (9b, xp yp zp)
5.2394(1)
13.8305(2)
328.8(1)
0.2258(6)
0.71(2) i
+4.0(3)
0.7144(7)
0.71(2) i
-3.8(2)
0.93(1)/0.07(1)
0.4995(8)
0.71(2) i
+3.4(5)
0.93(1)/0.07(1)
0.71(2) i
-1.1(1)
5.2505(1)
13.8355(1)
330.3(1)
0.2175(2)
0.99(2) i
+1.8(7)
0.7160(2)
0.99(2) i
-1.8(7)
0.93(1)/0.07(1)
0.4949(4)
0.99(2) i
+1.1(9)
0.93(1)/0.07(1)
0.99(2) i
-0.3(9)
-
12
xp
yp
zp
Uiso (× 100)/Å2
O2 (9b, xp yp zp)
xp
yp
zp
Uiso (× 100)/Å2
Rp/wRp/% (PND #1)
Rp/wRp/% (PND #2)
Rp/wRp/% (SXRD)
Rp/wRp/% (Total)
RMag
χ2
0.0338(12)
0.3120(7)
0.0996(5)
0.52(2)ii
0.0410(12)
0.7121(12)
0.6007(5)
0.52(2)ii
2.93/1.68
4.19/3.12
-
3.73/1.99
2.15
2.33
0.0376(1)
0.3198(1)
0.0976(2)
1.00(2)ii
0.0417(1)
0.7164(1)
0.5975(2)
0.98(2)ii
3.14/1.95
4.66/3.74
7.84/10.29
7.69/3.04
5.91
1.91
i Uiso for all cations are constrained to be the same value during refinements of 10 and
300 K data, respectively; ii Uiso for all oxygen atoms are constrained to be the same
value during refinements of 10 and 300 K data, respectively; iii The moments on Fe and
Mo are constrained to be 5.92 : 1.73 according to the expected full spin moments.
-
13
Table S2 Selected bond lengths (Å), octahedral distortion parameters (Δ), atomic bond
valence sums (BVS), and bond angles (º) of crystal structure of Mn2FeMoO6 at 10 and
300 K, respectively
10 K 300 K
Selected bond lengths (Å)
Mn1O6
Mn1-O1 × 3
-O2 × 3
ΔMn1 (× 10-3
)i
BVS
2.337(9)
2.055(7)
2.196(8)
4.12
2.14
2.298(3)
2.093(2)
2.195(3)
2.18
2.08
Mn2O6
Mn2-O1 × 3
-O2 × 3
ΔMn2 (× 10-3
) i
BVS
2.108(6)
2.262(9)
2.185(8)
1.24
2.11
2.093(1)
2.298(3)
2.195(3)
2.18
2.08
-
14
(Fe/Mo)1O6
(Fe/Mo)1-O1 × 3
-O2 × 3
Δ(Fe/Mo)1 (× 10-3
) i
BVS
1.961(7)
2.146(9)
2.054(6)
2.03
2.89
1.920(3)
2.146(4)
2.033(4)
3.09
3.10
(Mo/Fe)2O6
(Mo/Fe)2-O1 × 3
-O2 × 3
Δ(Mo/Fe)2 (× 10-3
) i
BVS
2.076(7)
1.897(5)
1.986(7)
2.03
4.85
2.085(2)
1.931(2)
2.008(2)
1.47
4.56
Selected bond angles (º)
O1-Mn1-O1
O2-Mn1-O2
O1-Mn1-O2
70.3(3)
112.5(3)
80.3(3)
73.6(1)
110.9(1)
79.9(1)
-
15
87.2(3)
147.5(3)
88.2(1)
151.1(2)
O1-Mn2-O1
O2-Mn2-O2
O1-Mn2-O2
109.0(3)
77.0(3)
78.9(3)
89.9(3)
154.6(4)
110.3(1)
74.7(1)
79.3(1)
89.1(1)
152.1(2)
O1-(Fe/Mo)1-O1
O2-(Fe/Mo)1-O2
O1-(Fe/Mo)1-O2
99.8(4)
82.1(3)
87.4(4)
89.4(4)
167.2(4)
100.4(2)
81.0(1)
87.8(1)
89.0(1)
166.0(2)
O1-(Mo/Fe)2-O1
O2-(Mo/Fe)2-O2
O1-(Mo/Fe)2-O2
80.8(2)
98.8(3)
88.7(3)
90.3(3)
82.6(1)
97.6(1)
88.5(1)
90.3(1)
-
16
167.1(3) 169.3(1)
Mn1-O1-Mn2
-(Fe/Mo)1
-(Mo/Fe)2
Mn1-O2-Mn2
-(Fe/Mo)1
-(Mo/Fe)2
122.6(3)
93.3(3)
89.9(3)
118.4(4)
96.6(3)
117.0(3)
120.9(1)
95.4(1)
86.6(1)
119.6(1)
95.1(2)
116.9(1)
Mn2-O1-(Fe/Mo)1
-(Mo/Fe)2
Mn2-O2-(Fe/Mo)1
-(Mo/Fe)2
115.4(4)
95.5(3)
84.7(3)
95.8(3)
116.1(1)
96.1(1)
86.9(2)
94.1(1)
(Fe/Mo)1-O1-(Mo/Fe)2 140.4(4) 140.2(1)
(Fe/Mo)1-O2-(Mo/Fe)2 140.6(4) 141.7(1)
i The octahedral distortion parameter Δ is defined as Δ = 1/6Σ{(d -dav)/dav}2, where d and
dav denote the individual interatomic distance and the average distance, respectively. 24
In the well-known polar LiNbO325, ΔLi = 1.8 × 10
-3, ΔNb = 4.0 × 10-3.
-
17
Mo
Mn1
Fe
Mn2
Fig. S4 Octahedral slab (thickness of ~3.4 Å) in the crystal structure of Mn2FeMoO6
viewed along the [-110] (left) and the [110] (right) directions, respectively, to show the
octahedral zigzag chains by edge-sharing connection of face-shared octahedral pairs
along the c-axis. Mn1O6, blue; Mn2O6, yellow; FeO6, carmine; MoO6, light green; O1,
red; O2, deep purple.
3. Second Harmonic Generation Measurements
Broadband SHG experiments were conducted at RT on a powder sample26
. The
fundamental beam covering a wide wavelength range ( = 1100 – 2000 nm) was produced from
an optical parametric oscillator at increments of 100 nm, which was synchronously pumped by
an Nd:YAG laser with a pulse width of 30 ps and a repetition rate of 50 Hz. The input irradiance
was about 0.7 GW/cm2 and the SHG signals (SHG = /2 = 550–1000 nm) were collected with a
-
18
reflection geometry by a fiber-optic bundle, which was coupled to a spectrometer equipped with
a charge-couple device camera. Surface-induced SHG as well as SHG signals from other optical
components were negligible. The relative SHG signals (Fig. S5) spectrally resolved in a broad
wavelength range were precisely calibrated with the known and measured efficiencies of all
optical components to produce the wavelength-dependent SHG response.
Fig. S5 Wavelength-dependent SHG spectra of Mn2FeMoO6
4. X-ray Absorption Near-Edge Spectroscopy
X-ray absorption near-edge spectra (XANES) were collected on the beam line X-19A at
Brookhaven NSLS, to confirm the formal oxidation state of the cations. Mn and Fe spectra were
collected in both transmission and fluorescence modes with simultaneous standards. The Mo
-
19
XANES was collected in fluorescence mode in a He-atmosphere-chamber with standards run in
temporal proximity. The main edge features at 3-d transition metal K edges are dominated by 1s
to 4p transitions. These features, and the step- continuum onset which lies underneath them,
manifest a chemical shift to higher energy with increasing valence. The 4p features can also be
split into multiple features by the local atomic coordination/bonding and by final state effects (i.e.
admixed 3d configurations). The chemical shift of the K edge has been widely used to chronicle
the evolution of the transition metal valence state in oxide-based materials.12,27-34
.
In Fig. S6a-c, the Mn- and Fe-K edges are compared to a series of standard compounds.27-
29 The nominal proximity of the main edge rise for various formal valence states is indicated by a
square in the figures. It is worth noting that the MnO and FeO standards (with edge sharing
Mn/Fe-O octahedra) manifest a robust two feature rise and the chemical shift is identified (here)
roughly with the inflection point between the two. The less structured steeply rising Mn-K edge
of the Mn2FeMoO6 exhibits a chemical shift consistent with MnO, which is the Mn2+
standard
and are clearly much lower in energy than the Mn3+
and Mn4+
standards. Thus, the formal
oxidation state at the Mn sites in this compounds is consistent with Mn2+
. Similarly, the chemical
shift for the Fe-K edge for Mn2FeMoO6 (Fig. S6b) falls in the formal Fe3+
valence range. The Fe
K-pre-edge region (7.11-7.12 KeV, Fig. S6c) is shown on an expanded scale. The features in the
pre-edge are quadrupole and dipole (through p-d hybridization) allowed transitions into final
states with 3d character. The onset energy and structure of the pre-edge features are typically
coupled to the transition metal valence. The Fe-K pre-edge features of Mn2FeMoO6 occurs at an
energy comparable to that of the Fe~3+
standards and intermediate between those of the Fe2+
and
Fe~4+
standards. This provides additional support for the Fe~3+
state in this material.
-
20
(a) (b)
(c) (d)
Fig. S6 XANES spectra of Mn2FeMoO6. The Mn-K edge spectra (a) along with a series
of octahedral O-coordinated Mn compounds with varying formal valences: Mn2+O,
Mn2+2FeNbO6 , LaMn3+O3, and CaMn
4+O3; the Fe-K edge spectra (b) and its pre-edge
region (c) along with a series of octahedral O-coordinated Fe compounds with varying
formal valences: Fe2+O, LiFe2+PO4, LaSrFe~3+O4, Mn2Fe
~3+NbO6 and SrFe~4+O3
; the
Mo-L3 edge (d) compared to reference edges for elemental Mo, the pyrochlore
Sm2Mo2O7 (Mo4+~4d2), the double perovskite SrMo0.5Fe0.5O3 (Mo
5+~4d1) and, the
(Mo6+~4d0) MoO3 and the quadruple perovskite Sr4Fe3MoO12 (Mo6+~4d0). Here the
formal valence is used ignoring more subtle hybridization/covalency effects.
Fig. S6d shows the Mo L3-edges for Mn2FeMoO6 along with a number of standard spectra.
The intense peak features at the L3-edges of 4d transition metals involve dipole allowed 2p-core
to 4d final-state transitions. These features can provide a probe of the empty 4d state energy
distribution, albeit modified by the transition matrix element, core–hole interaction and multiplet
-
21
effects.35-37
The transition metal d-states are split by the octahedral coordination ligand- field into
lower energy, t2g (sextet) and a higher energy eg (quartet) states. In low-d-occupancy transition
metal ions like Mo5+
(d1), the T L3-edges display a robust two peak structure with the lower
energy peak (A) involving transitions into empty t2g states; and the high-energy peak (B)
involving excitations into empty eg states. For d-hole counts greater than 4, the B-feature
intensity reflects the empty eg states and the intensity of the A-feature scales with the number of
t2g-holes. Thus the A-feature intensity, relative to that of the B-feature, provides an indicator of
the T 4d count/valence state. This trend can be seen in the Mo4+
/d2, Mo
5+/d
1, and Mo
6+/d
0
standard spectra sequence. The relative A/B feature intensity of the Mn2FeMoO6 spectrum places
it in the nominally Mo5+
/d1 configuration regime. Comparing to the conventionally synthesized
perovskite, SrMo0.5Fe0.5O3,35
one notes that the t2g-eg ligand field splitting in the Mn2FeMoO6
compound is substantially larger resulting in a larger/better resolved A-B feature splitting. Thus,
the formal oxidation states of Mn2+
2Fe3+
Mo5+
O6 were manifested. Here, the d1-electron
configuration of Mo5+
is ascribed to the formation of Ni3TeO6-type structure compared with the
Nb5+
and Ta5+
(d0-electron configuration) analogs, which adopt the polar LN-type structure.
12
5. Magnetic, Magnetotransport Properties, and Electrical Conductivity Measurements
Magnetization measurements were carried out with a commercial Quantum Design
superconducting quantum interference device (SQUID) magnetometer. The susceptibility was
measured in zero field cooled (ZFC) and field cooled (FC) conditions under a 0.1 T magnetic
field, for temperatures ranging from 5 to 400 K. Isothermal magnetization curves were obtained
at T = 5-300 K under an applied magnetic field varied from -5 to 5 T. The dc electrical
-
22
conductivity properties were measured on the pellet sample with the standard four probe
technique in a physical property measurement system (PPMS) from Quantum Design. The
magnetotransport properties were measured on the pellet sample with the standard four probe
technique in a physical property measurement system (PPMS) from Quantum Design. To avoid
the Joule heating effect, measurements were carried out with less than 0.5 μA current. The
magnetoresistance is defined as)0(
)0()(100)(
R
RHRHMR
, where R(H) is the resistivity at applied
magnetic field H and R(0) is the resistivity without magnetic field. The magnetotransport
properties were measured between -9 and 9 T at 100 and 300 K, respectively, showing a
maximum negative magnetoresistance about -2% and -2.5% at 9 T at 300 and 100 K,
respectively as shown in Fig. S7, thus enhance the multifunctional properties of this material.
Fig. S7 Magnetoresistance measurements results of Mn2FeMoO6 with magnetic field
between 0 and 9 T,showing maximum negative magnetoresistance of about -2.5% and -
2% at 9 T for 100 and 300 K, respectively.
-8 -6 -4 -2 0 2 4 6 8
0.0
0.5
1.0
1.5
2.0
2.5
3.0
-MR
(%
)
H (T)
100 K
300 K
-
23
6. Theoretical Calculations
The theoretical calculations have been performed within the generalized gradient
approximation (GGA)38
of the exchange-correlation functional with the choice of Perdew-Burke-
Ernzerhof (PBE) functional39
in a spin polarized scheme. To improve the description of
correlation effects in Mn and Fe-d electrons, a DFT+U40-42
method within the Dudarev et al.’s
approach was used. We have varied the U value between 4-8 eV at Fe and Mn sites and between
1-4 eV at Mo site, with a choice of Hund's coupling constant, JH = 0.8 eV. The variation did not
produce significant changes in the result. The calculations were performed in plane wave basis as
implemented in Vienna Ab-initio Simulation Package (VASP)43
. For these plane wave
calculations, we used projector augmented wave (PAW) potentials44
and the wave functions
were expanded in the plane wave basis with a kinetic energy cutoff of 600 eV. Reciprocal space
integrations were carried out with a k mesh of 5 x 5 x 5.
Calculations of electronic and magnetic structure have been carried out using the
generalized gradient approximation (GGA) of density functional theory (DFT), supplemented
with Hubbard U correction (GGA+U) to take care of strong electron-electron correlation at the
transition metal sites, as implemented in plane wave based Vienna Ab-initio Simulation code45
.
The calculated spin-polarized electronic structure, with chosen U values of 5 eV at the 3d
transition metal sites, Mn and Fe, and 2 eV at the 4d transition metal site of Mo, shows Fe and
Mn d-dominated states to be fully filled in the majority spin channels and empty in the minority
spin channel, while Mo d states to be partially filled in one of the spin channel, and empty in
other. The calculated electron configuration and dp hybridization are found to hold good for a
-
24
variation of U value between 4-8 eV at the 3d transition metal sites (Mn and Fe), 1-4 eV at the
4d transition metal (Mo) site, confirming the robustness of the obtained results.
The ferromagnetic solution with spins of Mn, Fe, Mo all aligned parallel are found to be
energetically unfavorable by a large energy difference of about 300 meV compared to the
ferrimagnetic solution where the spins of two nonequivalent Mn sites (Mn1 and Mn2) are
aligned antiparallel. The lowest energy magnetic structure, turned out to be the ferrimagnetic
structure with spins of Mn1 and Fe aligned parallel, which are antiparallel to the spins of Mn2
and Mo, giving rise to layers of ferromagnetically ordered Mn1 and Fe, coupled
antiferromagnetically to the next layer of ferromagnetically ordered Mn2 and Mo, in perfect
agreement with the magnetic structure obtained from PND data analysis. This magnetic
structure, cited as up-up-down-down (along the zig-zag chain) in the following, is found to be
energetically lower compared to the ferrimagnetic solution with antiparallel alignment of Mn1
and Mn2, and parallel alignment of Fe and Mo, (up-up-down-up) by an energy difference of
about 80 meV. This suggests a very strong antiferromagnetic exchange between Mn1 and Mn2
and a moderately strong antiferromagnetic exchange between Fe and Mo, mediated via the
corner sharing super-exchange paths. We also computed the polarization, for the insulating
solution assuming perfect ordering of Fe and Mn, in ground state magnetic configuration, using
Berry phase formalism46,47
.
Mn2FeMoO6 lacks any (n-1)d10
ns0 or d
0 ion, and adopts the polar Ni3TeO6 structure
instead of others in the corundum family shown in Fig. S1, breaking the polarization rules in
perovskite-related oxides. As far as we know, to date, all known polar perovskite-related and
corundum based compounds, including the above LN, ordered IL, and Ni3TeO6 structures,
-
25
contain at least an (n-1)d10
ns0 electronic configuration ion at the A-site (such as Zn
2+ in
ZnSnO325
, Sc3+
in ScFeO348
, and In3+
in (In1-xMx)MO3 (M = Fe0.5Mn0.5)49
), or B-site (such as Ge4+
in La2MgGeO68, Sb
5+ in Ni2InSbO6 and Ni2ScSbO6,
7 and Te
6+ in Ni3TeO6 and Li2ZrTeO6
4-6), or
a lone pair electron ion at the A-site (such as Bi3+
in BiFeO350
and Pb2+
in PbVO351
), or a SOJT
ion at the B-site (such as Nb5+
and Ta5+
in Mn2+
2Fe3+
M5+
O6 (M = Nb, Ta)12
and W6+
in
NaLaFeWO652
), no polarization was observed in oxides without any of the above ions, which
either favor the essential hybridization between the metal nd and oxygen 2p states53
, or provide
stereoactive lone pair electrons such as the 6s2 lone pair electron in Bi
3+ in BiFeO3
50. As far as
we knew, the polar ε-Fe2O3 is an exception but the crystal structure is different from
perovksite/corundum and contains mixed FeO6 and FeO4 coordination environment54-66
.
We calculated the total energy of different possible structural arrangements (Fig. S1) to
reveal the reason for the formation of the Ni3TeO6 structure. In order to check the influence of
disorder on the electronic structure, we considered three different configurations in a supercell,
which is three times the primitive unit cell, namely (i) perfect ordering (Fe-Mo-Fe-Mo-Fe-Mo),
(ii) segregation (Fe-Fe-Fe-Mo-Mo-Mo), and (iii) mixed configuration (Fe-Fe-Mo-Fe-Mo-Mo).
Our total energy calculations suggest that the ordered arrangement of Fe and Mo [case (i)] is
preferred over the segregated arrangement of Fe-Mo [case (ii)] by a large energy difference of
about 600 meV/f.u., and by an energy difference of about 150 meV/f.u. over the mixed
configuration [case (iii)] when the magnetism is turned on. This result is evidence for the
important role of magnetism in the finite d configuration of Fe and Mo in stabilizing the
observed Ni3TeO6 structure, which allows for two nonequivalent positions of B and B' ions. Note
that ilmenite (R-3) or LiNbO3 (R3c) structures do not allow for two nonequivalent positions of B
and B'. The Ni3TeO6 structure is also found to be stable over the ordered ilmenite (R3) structure
-
26
by about 30 meV/f.u., which arises due to the large size mismatch between the ionic sizes of
Mn2+
in octahedral environment (~0.83 Å) and that of Fe3+
or Mo5+
(0.65 and 0.61 Å,
respectively)67
. In the ordered ilmenite structure Mn1-Mn2 edge share within the layer, while in
Ni3TeO6, the unlike atoms, e.g. Mn1 and Fe, edge share within the layer.
7. Dielectric and Polarization Measurements
Dielectric measurements were carried out in two-point geometry. For these investigations
polycrystalline pellets were prepared as plate-like capacitors with typical area of 4 mm2, a
thickness of 0.5 mm and silver-paint electrodes. For the spectroscopic measurements in the
frequency range from 1 Hz < < 1 MHz a frequency response analyzer [Novocontrol] was
utilized; the polarization measurements at low temperatures were carried out in a Sawyer-Tower
circuit using an electrometer (Keithley 6517) and linear E-field ramping with a frequency of one
cycle per minute.
At high temperatures the dielectric properties are completely determined by the finite
conductivity', which effectively is connected to the measured imaginary part of the complex
permittivity ' = 20''. In addition to the experimental obstacles to perform polarization or
pyro-current measurements in this conductive regime, and to precisely determine ' in the
presence of a large dielectric loss '', one has to consider further intrinsic and non-intrinsic
contributions to the dielectric response. On the one hand the presence of VRH as demonstrated in
in Figure 3 will lead to a contribution to the real part of the permittivity ' 68,69
. On the other hand
heterogeneities in the sample, i.e. grain boundaries and electrodes (contacts) will lead to highly
-
27
capacitive depletion layers, which (like Schottky diodes) give rise to highly dispersive dielectric
response as they form effective RC-elements 68,69
. Measurements of the complex dielectric
permittivity are shown in Fig. S9. The high values of ' for high temperatures and small
frequencies are accompanied by much higher values in '' denoting the influence of sample
conductivity together with non-intrinsic resistivities and capacitances of contacts and grain
boundaries. Nevertheless, one can try to evaluate the resulting effective conductivity as it is
shown in Fig. S8 in the vicinity of the magnetic phase transition. The data is presented as
Arrhenius-plot ln'(1/T) between 295 and 385 K to demonstrate the thermally activated nature of
Fig. S8 Logarithm of the conductivity measured at 1 Hz (left scale) and magnetization
measured in a field of 0.1 T (right scale) plotted vs reciprocal temperature. The linear
regions above and below the magnetic transition temperature of 337 K are fitted
assuming thermal activation with an energy barrier .
0.0026 0.0028 0.0030 0.0032 0.0034
-3.0
-2.5
-2.0
-1.5
' ~ e-/(k
BT)
/kB= 1830 K
ln(
'(S
/m))
1/T (1/K)
/kB= 2135 K
385 K 333 K 295 K
0.0
0.1
0.2
0.3
M (
B/f
.u.)
-
28
Fig. S9 Real (upper frame) and imaginary (lower frame) part of the complex permittivity
* vs temperature as measured in zero magnetic field for frequencies between 1 Hz and
1 MHz. the inset shows a P(E)-curve measured with a frequency of one cycle per
minute.
the transport in this temperature regime. At the ferrimagnetic transition (which in Fig. S8 is
nicely denoted by the onset of spontaneous magnetization) the effective energy barrier changes:
102
103
'
0 100 200 300
100
103
106
1 Hz ... 1 MHz
''
T (K)
-200 -100 0 100 200-100
0
100
T = 2 KP
(C
/m2)
E (V/mm)
-
29
in the magnetically ordered regime the energy barrier is lowered. This meets the evaluation of
the magneto-resistive properties of the previous section, even though the details of this effect, e.g.
the role of inter-grain contacts, cannot be elucidated at this point.
For low temperatures and high frequencies the impact of the residual conductivity on the
dielectric properties is reduced. On cooling, the large values for ' drop via two steps (due to
electrode and inter-grain contacts) towards the intrinsic value of ε'which still is a relatively
high value compared to other transition metal oxides. In this low temperature regime the residual
conductivity ' vanishes and the dielectric loss '' becomes much smaller than ' Therefore it is
possible to conduct direct polarization measurements. The inset of Fig. S9 displays a P(E)-loop
measured at 2 K in fields up to 170 V/mm. No hint towards a ferroelectric component can be
found as denoted e.g. via switchable or at least non-linear polarization. However, these findings
do not exclude the presence of possible structurally induced pyroelectricity in the polycrystalline
sample.
References
1 Werfel, F. & Brümmer, O. Corundum structure oxides studied by XPS. Phys. Scripta 28, 92 (1983).
2 Woodward, P. M., Sleight, A. W., Du, L.-S. & Grey, C. P. Structural Studies and order–disorder phenomenon in a series of new quaternary tellurates of the type A2+M4+Te6+O6 and A1+2M
4+Te6+O6. J.Solid State Chem. 147, 99-116 (1999). 3 Hoel, C. A. et al. High-pressure synthesis and local structure of corundum-type In2−2xZnxSnxO3 (x
≤ 0.7). J. Am. Chem. Soc. 132, 16479-16487 (2010). 4 Becker, R. & Berger, H. Reinvestigation of Ni3TeO6. Acta Crystallogr. E 62, i222-i223 (2006). 5 Živković, I., Prša, K., Zaharko, O. & Berger, H. Ni3TeO6 —a collinear antiferromagnet with
ferromagnetic honeycomb planes. J. Phys. Condens. Matter 22, 056002 (2010). 6 Choisnet, J., Rulmont, A. & Tarte, P. Les tellurates mixtes Li2ZrTeO6 et Li2HfTeO6: un nouveau
phénomène d'ordre dans la famille corindon. J. Solid State Chem. 75, 124-135 (1988). 7 Ivanov, S. A. et al. Spin and dipole ordering in Ni2InSbO6 and Ni2ScSbO6 with corundum-related
Structure. Chem. Mater. 25, 935-945 (2013). 8 Swaffer, M. et al. La2MgGeO6: a novel Ge based perovskite synthesised under ambient pressure.
Chem. Commun. 1776-1777 (2002).
-
30
9 Michael W. Lufaso, B. S. Perovskite synthesis and analysisi using structure prediction diagnostic software. Dissertation (2002).
10 Mathieu, R. et al. Magnetic order near 270 K in mineral and synthetic Mn2FeSbO6 ilmenite. Appl. Phys. Lett. 98, 202505 (2011).
11 Suchomel, M. R. et al. High pressure bulk synthesis and characterization of the predicted multiferroic Bi(Fe1/2Cr1/2)O3. Appl. Phys. Lett. 90, 112909-112903 (2007).
12 Li, M.-R. et al. Polar and Magnetic Mn2FeMO6 (M = Nb, Ta) with LiNbO3-type structure: high-pressure synthesis. Angew. Chem. Int. Ed. 52, 8406-8410 (2013).
13 Aimi, A. et al. High-pressuressynthesis and correlation between structure, magnetic, and dielectric properties in LiNbO3-type MnMO3 (M = Ti, Sn). Inorg. Chem. 50, 6392-6398 (2011).
14 Varga, T. et al. Coexistence of weak ferromagnetism and ferroelectricity in the high pressure LiNbO3-type Phase of FeTiO3. Phys. Rev. Lett. 103, 047601 (2009).
15 Baettig, P., Ederer, C. & Spaldin, N. A. First principles study of the multiferroics BiFeO3, Bi2FeCrO6, and BiCrO3: structure, polarization, and magnetic ordering temperature. Phys. Rev.B 72, 214105 (2005).
16 Baettig, P. & Spaldin, N. A. Ab initio prediction of a multiferroic with large polarization and magnetization. Appl. Phys. Lett. 86, 012505 (2005).
17 Feng, H.-J. & Liu, F.-M. First-principles prediction of coexistence of magnetism and ferroelectricity in rhombohedral Bi2FeTiO6. Phys. Lett. A 372, 1904-1909 (2008).
18 Walker, D., Carpenter, M. A. & Hitch, C. M. Some simplifications to multianvil devices for high pressure experiments. Am. Mineral. 75, 1020-1028 (1990).
19 Rietveld, H. A profile refinement method for nuclear and magnetic structures. J. Appl. Crystallogr. 2, 65-71 (1969).
20 Coelho, A. TOPAS-Academic, General Profile and Structure Analysis Software for Powder Diffraction Data, version 4.1 (2007).
21 Toby, B. EXPGUI, a graphical user interface for GSAS. J. Appl. Crystallogr. 34, 210-213 (2001). 22 Retuerto, M., Alonso, J. A., Martinez-Lope, M. J., Martinez, J. L. & Garcia-Hernandez, M. Record
saturation magnetization, Curie temperature, and magnetoresistance in Sr2FeMoO6 double perovskite synthesized by wet-chemistry techniques. Appl. Phys. Lett. 85, 266-268 (2004).
23 Serrate, D., Teresa, J. M. D. & Ibarra, M. R. Double perovskites with ferromagnetism above room temperature. J.Phys.: Condens. Matter 19, 023201 (2007).
24 Brown, I. D. & Shannon, R. D. Empirical bond-strength-bond-length curves for oxides. Acta Crystallogr. A 29, 266-282 (1973).
25 Inaguma, Y., Yoshida, M. & Katsumata, T. A polar oxide ZnSnO3 with a LiNbO3-type structure. J. Am. Chem. Soc. 130, 6704-6705 (2008).
26 Kurtz, S. K. & Perry, T. T. A powder technique for the evaluation of nonlinear optical materials. J. Appl. Phys. 39, 3798-3813 (1968).
27 Croft, M. et al. Systematic Mn d-configuration change in the La1-xCaxMnO3 system: a Mn K-edge XAS study. Phys. Rev. B 55, 8726-8732 (1997).
28 Mandal, T. K. et al. La2MnVO6 double perovskite: a structural, magnetic and X-ray absorption investigation. J. Mater. Chem. 19, 4382-4390 (2009).
29 Bune, R. O. et al. Crystal structure and properties of Ru-stoichiometric LaSrMnRuO6. Chem. Mater. 18, 2611-2617 (2006).
30 Zeng, Z., Fawcett, I. D., Greenblatt, M. & Croft, M. Large magnetoresistance in double perovskite Sr2Cr1.2Mo0.8O6-δ. Mater. Res. Bull. 36, 705-715 (2001).
31 Veith, G. M., Greenblatt, M., Croft, M. & Goodenough, J. B. Synthesis and characterization of the oxynitride pyrochlore - Sm2Mo2O3.83N3.17. Mater. Res. Bull. 36, 1521-1530 (2001).
-
31
32 Ramanujachary, K. V. et al. Substitutional effects of 3d transition metals on the magnetic and structural properties of quasi-two-dimensional La5Mo4O16. J. Solid State Chem. 164, 60-70 (2002).
33 Veith, G. M. et al. Synthesis and characterization of the new Ln2FeMoO7 (Ln = Y, Dy, Ho) compounds. J. Mate. Chem. 14, 1623-1630 (2004).
34 Veith, G. M. et al. Synthesis and characterization of Sr3FeMoO6.88: an oxygen-deficient 2D analogue of the double perovskite Sr2FeMoO6. Chem. Mater. 17, 2562-2567 (2005).
35 Huang, Y.-H. et al. Double-Perovskite Anode Materials Sr2MMoO6 (M = Co, Ni) for Solid Oxide Fuel Cells. Chem. Mater. 21, 2319-2326 (2009).
36 Yang, T. et al. B-site ordered perovskite LaSrMnNbO6: Synthesis, structure and antiferromagnetism. J. Solid State Chem. 183, 2689-2694 (2010).
37 Retuerto, M. et al. Magnetic and structural studies of the multifunctional material SrFe0.75Mo0.25O3−δ. Inorg. Chem. 51, 12273-12280 (2012).
38 Perdew, J. P. et al. Atoms, molecules, solids, and surfaces: Applications of the generalized gradient approximation for exchange and correlation. Phys. Rev. B 46, 6671-6687 (1992).
39 Perdew, J. P., Burke, K. & Ernzerhof, M. Generalized gradient approximation made simple. Phys. Rev. Lett. 77, 3865-3868 (1996).
40 Anisimov, V. I., Zaanen, J. & Andersen, O. K. Band theory and Mott insulators: Hubbard U instead of Stoner I. Phys. Rev. B 44, 943-954 (1991).
41 Liechtenstein, A. I., Anisimov, V. I. & Zaanen, J. Density-functional theory and strong interactions: Orbital ordering in Mott-Hubbard insulators. Phys. Rev. B 52, R5467-R5470 (1995).
42 Dudarev, S. L., Botton, G. A., Savrasov, S. Y., Humphreys, C. J. & Sutton, A. P. Electron-energy-loss spectra and the structural stability of nickel oxide: an LSDA+U study. Phys. Rev. B 57, 1505-1509 (1998).
43 Kresse, G. & Hafner, J. Ab initio molecular-dynamics simulation of the liquid-metal–amorphous-semiconductor transition in germanium. Phys. Rev. B 49, 14251-14269 (1994).
44 Kresse, G. & Joubert, D. From ultrasoft pseudopotentials to the projector augmented-wave method. Phys. Rev. B 59, 1758-1775 (1999).
45 Kresse, G. & Hafner, J. Ab initio molecular dynamics for open-shell transition metals. Phys. Rev. B 48, 13115-13118 (1993).
46 King-Smith, R. D. & Vanderbilt, D. Theory of polarization of crystalline solids. Phys. Rev. B 47, 1651-1654 (1993).
47 Vanderbilt, D. & King-Smith, R. D. Electric polarization as a bulk quantity and its relation to surface charge. Phys. Rev. B 48, 4442-4455 (1993).
48 Li, M.-R. et al. A Polar corundum oxide displaying weak ferromagnetism at room temperature. J. Am. Chem. Soc. 134, 3737-3747 (2012).
49 Belik, A. A., Furubayashi, T., Yusa, H. & Takayama-Muromachi, E. Perovskite, LiNbO3, corundum, and hexagonal polymorphs of (In1-xMx)MO3. J. Am. Chem. Soc. 133, 9405-9412 (2011).
50 Wang, J. et al. Epitaxial BiFeO3 Multiferroic thin film heterostructures. Science 299, 1719-1722 (2003).
51 Singh, D. J. Electronic structure and bond competition in the polar magnet PbVO3. Phys. Rev. B 73, 094102 (2006).
52 Retuerto, M. et al. Polar and magnetic layered A-site and rock salt B-Site-ordered NaLnFeWO6 (Ln = La, Nd) perovskites. Inorg. Chem. 52, 12482-12491 (2013).
53 Cohen, R. E. Origin of ferroelectricity in perovskite oxides. Nature 358, 3 (1992).
-
32
54 Tuček, J., Zbořil, R., Namai, A. & Ohkoshi, S.-i. ε-Fe2O3: an advanced nanomaterial exhibiting giant coercive field, millimeter-wave ferromagnetic resonance, and magnetoelectric coupling. Chem. Mater. 22, 6483-6505 (2010).
55 Ohkoshi, S.-i. et al. A millimeter-wave absorber based on gallium-substituted ε-iron oxide nanomagnets. Angew. Chem. Int. Ed. 46, 8392-8395 (2007).
56 Namai, A. et al. Synthesis of an electromagnetic wave absorber for high-speed wireless communication. J. Am. Chem. Soc. 131, 1170-1173 (2008).
57 Kadlec, C. et al. Electromagnon in ferrimagnetic ɛ-Fe2O3 nanograin ceramics. Phys. Rev. B 88, 104301 (2013).
58 Tucek, J., Ohkoshi, S.-i. & Zboril, R. Room-temperature ground magnetic state of ɛ-Fe2O3: in-field Mössbauer spectroscopy evidence for collinear ferrimagnet. Appl. Phys. Lett. 99, 253108 (2011).
59 Lefevre, C. et al. Magnetic and polar properties’ optimization in the magnetoelectric Ga2–xFexO3 compounds. J. Phys. Chem. C 117, 14832-14839 (2013).
60 Namai, A. et al. Hard magnetic ferrite with a gigantic coercivity and high frequency millimetre wave rotation. Nature Commun. 3, 1035 (2012).
61 Rado, G. T. Observation and possible mechanisms of magnetoelectric effects in a ferromagnet. Phys. Rev. Lett. 13, 335-337 (1964).
62 Arima, T. et al. Structural and magnetoelectric properties of Ga2−xFexO3 single crystals grown by a floating-zone method. Phys. Rev. B 70 (2004).
63 Saha, R. et al. Multiferroic and magnetoelectric nature of GaFeO3, AlFeO3 and related oxides. Solid State Commun. 152, 1964-1968 (2012).
64 Dmitriev, A. I. et al. Magnetic phase transition in ɛ-InxFe2−xO3 nanowires. Phys. Solid State 55, 2252-2259 (2013).
65 Yamada, K. et al. The phase transition of ɛ-InxFe2−xO3 nanomagnets with a large thermal hysteresis loop (invited). J. Appl. Phys. 111, 07B506 (2012).
66 Sakurai, S., Kuroki, S., Tokoro, H., Hashimoto, K. & Ohkoshi, S. Synthesis, crystal structure, and magnetic properties of ϵ-InxFe2–xO3 nanorod-shaped magnets. Adv. Funct. Mater. 17, 2278-2282 (2007).
67 Shannon, R. Revised effective ionic radii and systematic studies of interatomic distances in halides and chalcogenides. Acta Crystallogr. A 32, 751-767 (1976).
68 Lunkenheimer, P. et al. Origin of apparent colossal dielectric constants. Phys. Rev. B 66, 052105 (2002).
69 Niermann, D., Waschkowski, F., de Groot, J., Angst, M. & Hemberger, J. Dielectric properties of charge-ordered LuFe2O4 revisited: the apparent influence of contacts. Phys. Rev. Lett. 109, 016405 (2012).
203-Mn2FeMoO6-anie_201406180203-SI-Mn2FeMoO6-anie_201406180