Room temperature magnetoelectric coupling in a molecular ferroelectric ytterbium(III) complex

See allHide authors and affiliations

Science  07 Feb 2020:
Vol. 367, Issue 6478, pp. 671-676
DOI: 10.1126/science.aaz2795

Major-league magnetostriction

Magnetoelectric materials polarize in response to either electric or magnetic fields, making them attractive for data-storage applications. Long et al. discovered a ytterbium-based molecular magnetoelectric material with high magnetoelectric coupling (see the Perspective by Zhou and Han). An applied magnetic field strains the material, which changes its electrical properties. The required field is much lower than other magnetoelectric materials, and this work highlights the potential for using molecular materials in devices.

Science, this issue p. 671; see also p. 627


Magnetoelectric (ME) materials combine magnetic and electric polarizabilities in the same phase, offering a basis for developing high-density data storage and spintronic or low-consumption devices owing to the possibility of triggering one property with the other. Such applications require strong interaction between the constitutive properties, a criterion that is rarely met in classical inorganic ME materials at room temperature. We provide evidence of a strong ME coupling in a paramagnetic ferroelectric lanthanide coordination complex with magnetostrictive phenomenon. The properties of this molecular material suggest that it may be competitive with inorganic magnetoelectrics.

A ferroelectric material exhibits a permanent electrical polarization that can be switched by an electric field (1, 2). Such electroactive materials have a wide range of applications, including temperature sensing, data storage, piezoelectric devices, and electro-optics. The association of electrical and magnetic polarizabilities gives rise to multifunctional systems called magnetoelectrics. The term describes the influence of a magnetic (electric) field on the polarization (magnetization), allowing the constitutive properties to be simultaneously triggered (3, 4). For computing, modifying the polarization or magnetization using a low-magnitude magnetic or electric field may reduce the energy needed and speed up the processing rate of nonvolatile memory devices (3, 5, 6). Conventional approaches to designing single-phase magnetoelectrics are widely based on inorganic materials, such as oxides or fluorides (68). However, if the origins of ferroelectricity and magnetism are associated with different carriers, we should expect only a moderate magnetoelectric (ME) coupling (9). For instance, only a few examples exist of ME coupling at room temperature (6, 10, 11). In contrast, large ME coupling has been found in strain-mediated multiphase materials, such as composites or multilayers, by combining piezoelectricity with the magnetostrictive effect (3).

Molecular materials exhibit numerous advantages over inorganic ones, such as structural diversity, soft chemistry routes, environmentally friendly processing and shaping, optical transparency, and light density (12). For these reasons, molecular ferroelectrics (13, 14) are often considered an alternative to traditional metal oxides (15, 16). Although most molecular materials exhibit a magnetic ordering temperature below room temperature, important ME coupling may be expected from the association of paramagnetism and ferroelectricity (fig. S1) (7), as both these properties could involve the same chemical element. Although ME coupling in nonferroelectric molecular materials has been reported (17, 18), the interaction between magnetism and ferroelectricity has not been extensively studied (1926). For the few examples that exist, the ferroelectric ordering temperatures are typically lower than room temperature. Moreover, the modification of the polarization or magnetization by applying magnetic or electric fields, respectively—which also remains challenging in pure inorganic materials—requires a large magnitude operating field (27).

We designed a chiral lanthanide complex exhibiting an above–room temperature ferroelectricity that, in association with a strong magnetostriction, gives rise to a distinctive ME coupling. This allowed tuning of the ferroelectric domains at the nanometric scale by applying a relatively low magnetic field at room temperature. Our molecular approach to designing ME materials relied on the association of paramagnetic lanthanide ions, such as Yb3+, with a chiral diamagnetic zinc complex in order to favor the crystallization in 1 of the 10 polar point groups compatible with ferroelectricity. We chose the Yb3+ ion because it has a large total magnetic moment, which, being oriented along a magnetic field, can provide an anisotropic magnetostriction underlying the coupling between magnetic and structural subsystems.

The stoichiometric reaction of R,R-H2L [6,6'-((1E,1'E)-(((1R,2R)-1,2-diphenylethane-1,2-diyl)bis(azaneylylidene))bis(methaneylylidene))bis(2-methoxyphenol)] or S,S-H2L [6,6'-((1E,1'E)-(((1S,2S)-1,2-diphenylethane-1,2-diyl)bis(azaneylylidene))bis(methaneylylidene))bis(2-methoxyphenol)], Zn(OAc)2·2H2O, and Yb(NO3)3·5H2O in methanol yielded to a yellow solution, which, upon slow diffusion of diethylether, resulted in the crystallization of R,R-[Zn(OAc)(L)Yb(NO3)2] (R,R-1) or S,S-[Zn(OAc)(L)Yb(NO3)2] (S,S-2). Using single crystal x-ray diffraction, we found that R,R-1 and S,S-2 are isostructural to the dysprosium analog (28). The two enantiomers crystallized in the polar space group P21 with two independent homochiral dinuclear complexes within the asymmetric unit (Fig. 1 and figs. S2 and S3). These two complexes exhibit differences in the crystallographic distances and geometries (tables S1 and S2) that give rise to uncompensated dipolar moments along the b axis. We confirmed this by analyzing the location of the ions in the unit cell (table S3) (2931).

Fig. 1 Crystal structures of R,R-1 and S,S-2.

(A) Molecular structure of the dinuclear Zn2+-Yb3+ complexes R,R-1 and S,S-2 and their enantiomeric relationship. Orange, Yb3+; light blue, Zn2+; blue, N; red, O; gray, C. Hydrogen atoms have been omitted for clarity. (B) View of the crystal packing arrangement of R,R-1 along the a axis, emphasizing the two homochiral complexes. (C) Single-crystal facets assignment and view of the slice in the crystallographic (01¯1¯) plane.

We corroborated the enantiomeric nature of R,R-1 and S,S-2 using solid-state circular dichroism (CD) (fig. S4), with the mirror-symmetrical CD spectra showing Cotton effects of opposite signs at maximum wavelengths (λmax) of 261, 303, and 386 nm. The presence of the zinc complex also acted as a sensitizer toward lanthanide ions. Hence, the complexes R,R-1 and S,S-2 showed the typical Yb3+ luminescence in the 925- to 1075-nm near-infrared region at both room and low temperatures (figs. S5 to S11) (32).

We verified the absence of solvent molecules with thermogravimetric analysis, which also indicated that both enantiomers remain stable up to 550 K (fig. S12). We used single-crystal x-ray diffraction at 400 K and differential scanning calorimetry to confirm the absence of a phase transition up to the decomposition temperature (fig. S13). This demonstrated that the material remains crystallized in the P21 polar space group. We investigated the magnetic properties of R,R-1 and S,S-2 using SQUID (superconducting quantum interference device) magnetometry. We confirmed the paramagnetic behavior expected for a single Yb3+ ion with a 2F7/2 ground state (total angular moment J = 7/2; orbital moment L = 3; spin moment S = 1/2) and a room-temperature magnetic moment close to 4.32 bohr magnetons (fig. S14). We also observed a typical slow relaxation of the magnetization at low temperature (figs. S15 to S20 and tables S4 to S6) (32).

Both enantiomers crystallized in the acentric space group P21, giving a 2 point group. We confirmed ferroelectric behavior of R,R-1 (figs. S21 and S22) (32), which also applies to S,S-2 because the ferroelectric properties of pairs of enantiomers are known to be identical (33, 34). The absence of phase transition up to the decomposition of the material (550 K) indicates that the ferroelectric Curie point (Tc) will be located at a higher temperature, as observed in ferroelectrics such as [(CH3)4N]HgCl3 and [C(NH2)3][Al(H2O)6](SO4)2 (14).

We carried out piezoresponse force microscopy (PFM) measurements to investigate the polar behavior and ME coupling (32, 35). We measured the vertical and lateral responses, both of which consisted of amplitude and phase signals directly related to the magnitude of the polarization and to its orientation, respectively. This method provided sufficient spatial resolution for the detailed study of in-plane (IP; lateral PFM amplitude × phase) and out-of-plane (OOP; vertical PFM amplitude × phase) piezoelectric behavior at the nanoscale (35, 36). Because additional contributions having ionic transport or electrostatic nature could affect the PFM signals, we carried out the measurements taking into account the protocol described by Vasudevan et al. (37) and Balke et al. (38) (figs. S23 and S24) (32). Hence, the PFM responses we observed reflected the polarization state of the material. We obtained a clear piezoresponse coming from several planes of R,R-1 single crystal (fig. S25), but we detected the strongest simultaneous OOP and IP responses for the largest crystal facet accounting for the (01¯1¯) plane (fig. S26). The atomic force microscopy (AFM) for the (01¯1¯) plane reveals a stripe-like morphology (Fig. 2A). We found a peculiar organization in the OOP and IP PFM responses that demonstrated the presence of spontaneously polarized domains of opposite polarization states (shown in red and blue in Fig. 2B). Such striped organization was previously observed in other molecular or metal oxide ferroelectrics, such as BiFeO3 (35, 3941). The crystal symmetry predicts that the ferroelectric domain structure visualized by the IP and OOP components of the piezoresponse must be consistent. However, the resulting PFM response depends on the azimuthal angle between the scanning direction and the polar axis and contains different tensile and shear piezoelectric strain components (as described in the “PFM Measurements” subsection of the supplementary materials). This situation is reminiscent of that observed in the benchmark molecular ferroelectric diisopropylammonium bromide (40). According to the initial work of Aizu (42), which was recently applied to molecular ferroelectrics (13, 43, 44), the number of polarization directions in the ferroelectric phase depends on the symmetry of the paraelectric phase. Although the structure of the paraelectric phase cannot be examined (Tc is higher than the decomposition temperature)—thus precluding the determination of the uniaxial or multiaxial character of the polarization—a clear correlation between the IP and OOP responses can be observed in some other PFM experiments (fig. S27).

Fig. 2

Scanning probe microscopy (SPM) measurements of the (01¯1¯)-oriented R,R-1 single crystal. (A) AFM topography of the (01¯1¯) plane. (B) OOP (left) and IP (right) PFM responses without electrical bias. The color mapping reveals a distinctive organization demonstrating the presence of spontaneously polarized domains of opposite polarization states (shown in red and blue). (C) Magnified (×3) part of the scanned area [marked with a dashed square in (A) and (B)] with the OOP (left) and IP (right) PFM responses at 0 V. (D) OOP (left) and IP (right) PFM responses measured for the same area after the direct current (dc) bias voltage (±30 V) poling. (E) Schematic of the PFM experiment. DFL, deflection; E, electric field.

We investigated the polarization switchability using direct current (dc) bias voltage of ±30 V applied to different areas (dashed rectangles in Fig. 2, C and D). In these areas, we achieved uniform (monodomain-like) polarization states, visible on both IP and OOP images (Fig. 2D). Taking advantage of the absence of a phase transition, we performed PFM measurements at 450 K (fig. S24) (32). We found a notable decrease in the electromechanical response, directly related to the polarization, which reflected the temperature-dependent behavior for a ferroelectric material. The switchable character of the electric polarization was further demonstrated by applying opposite dc bias to generate box-in-box patterns that showed clear 180° phase contrast and domain walls (fig. S28). Moreover, the piezoresponse hysteresis loops obtained by switching spectroscopy (SS)–PFM were found to exhibit a centered square-like shape with a 180° switching for the phase component (Fig. 3A). We also measured the typical butterfly loops of the displacement signal, which confirmed the ferroelectric character (37). Previously switched areas (±30 V) revealed a shift of the phase and amplitude component toward either positive or negative voltage, owing to the formation of coherent remnant polarization states that caused a strong depolarization field (Fig. 3A). We determined a maximum local longitudinal piezoelectric coefficient dOOP=73 pm V1 at room temperature, which corresponded to a polarization of the order of magnitude of 10 μC cm−2. The theoretical spontaneous polarization that we estimated from the point charge model (2931) along the b axis by considering only the metal ions and some atoms of the ligand (Zn2+, Yb3+, O, and N+) gives 3.32 μC cm−2 along the [01¯1¯] direction (table S3) (32). We can rationalize this value, which was weaker than the experimental one, by pointing out the complexity of the structure. Also, we did not take into account the effect of additional molecular dipoles constituting the coordination complexes as well as covalency. Thus, the experimental value of the polarization could be positively compared with that of Rochelle salt (0.25 μC cm−2) and was found in the same order of magnitude as the better-performing molecular ferroelectrics (14, 15, 40, 41).

Fig. 3 SS-PFM hysteresis loops, multilevel states, and magnetostriction.

(A) SS-PFM hysteresis loops obtained for the virgin (0 V) area and the areas polished by dc bias voltage (± 30 V): phase and displacement components as a function of the voltage. (B) SS-PFM hysteresis loops obtained at zero and under applied magnetic field of ±1 kOe. (C) Normalized piezoresponse as a function of the magnetic field enlightening the six remanent polarization states that could be actuated by applying magnetic and/or electric fields. (D) Magnetostriction measured on a single crystal of R,R-1 at ±1 kOe (averaged on 5 loops). The lines are guides for the eye.

The ferroelectric character of R,R-1 crystals suggests the possibility of a ME coupling at room temperature. We performed PFM measurements on the same single crystal in the presence of a dc magnetic field of ±1 kOe applied along the (01¯1¯) plane. We observed the stripe-like morphology obtained by AFM (Fig. 4A). We found ferroelectric polarization redistribution with a low-magnitude magnetic field of ±1 kOe, similar to that produced by an electrical bias voltage (Fig. 4, B and C). The change of the polarization states and enhancement of the response make the effect easy to see. The ferroelectric domain’s modification appears only in some parts of the region, as has been systematically observed in the rare examples of materials investigated by PFM (27, 4547). The typical magnitude of the magnetoelectric tensor component α31=ΔudoopΔH D, where D is the thickness of the (01¯1¯)-oriented crystal and Δu is the change in vertical surface displacement induced by the change in lateral magnetic field ΔH, attains the order of 100 mV Oe−1·cm−1. Although this value reflects the polarization change for a given magnetic field, it greatly exceeds (by at least one order of magnitude) those observed in the bulk BiFeO3 multiferroics (ranging from 0.6 to 7 mV Oe−1·cm−1) (48) and is comparable to those of ferroelectric and ferro(ferri)magnetic composites with strain-mediated ME coupling (49). Additional confirmation of the change in responses comes from our in situ PFM measurements performed on other crystal facets and with various magnetic fields (figs. S29 to S32) (32).

Fig. 4 Room-temperature PFM studies under magnetic field evidencing the ME coupling.

(A) AFM topography of the (01¯1¯) plane for R,R-1. (B) OOP and IP PFM responses at H = 0 Oe. (C) OOP and IP PFM responses at H = 1 kOe evidencing the ME coupling as redistribution in the ferroelectric domains (change in colors) and increase in the electromechanical response. (D) Sketch illustrating the possible deformation of the individual complex under an applied magnetic field.

We found additional evidence of the ME interaction in the SS-PFM local hysteresis loop measurements that we performed under dc magnetic field. These loops were strongly affected by a magnetic field, specifically in their asymmetry and height, as well as in the coercive fields (Fig. 3B). Taking advantage of this ME coupling, we actuated variable polarization states via the dc electric and magnetic fields (Fig. 3C). This feature may be suitable for multilevel polarization state devices designed for high-density data systems (50). Notably, this interaction occurs in the paramagnetic state at room temperature using a moderate magnetic field (1 kOe). The low magnetic field we used clearly contrasts with other molecule-based materials that required fields of several tesla to induce a change in the pyroelectric currents (23). Such room-temperature low-field switching is also quite rare in metal oxides (11, 26, 27).

The strong ME effect we observed is because the same chemical element, Yb3+, is implicated in the two functionalities. Lanthanide ions present a larger spin-orbit coupling with respect to transition metal ions in classical inorganic magnetic ferroelectrics. Applying a magnetic field to a material containing anisotropic Yb3+ (nonzero orbital angular momentum, L = 3, spin-orbit coupling) should affect the crystal lattice producing magnetostriction. To confirm this, we investigated the magnetostriction, λ, at the microscopic and macroscopic levels. We measured the local surface displacement using contact AFM mode (no electric field applied) and evaluated it as a function of the applied magnetic field. We found a large room-temperature field-induced mechanical deformation (parastriction) (Fig. 3D). The displacement increased with magnetic field and reached ~10−4 (1 kOe). The magnetostriction value we found was consistent with the macroscopic data we obtained from a single-crystal x-ray diffraction experiment we conducted under a magnetic field of ~0.8 kOe applied with a deviation angle of 10 ± 5° along the [011¯] direction and yielding λ value of up to 1 × 10−3 (table S7) (32). We confirmed this effect on a different single crystal and by collecting three datasets to provide a statistical analysis (table S8). Reversing the direction of the magnetic field did not change the sign of the magnetostriction (table S7), which was fully consistent with the expected quadratic behavior. This large magnetostriction was comparable to those observed in ytterbium-based paramagnets. This phenomenon primarily had a single-ion character, and the related deformation can be equivalent in magnitude with that characteristic of magnetically long-range ordered systems (51). Notably, the association of magnetostriction and ferroelectricity engenders a pronounced ME coupling, as observed in inorganic multiphase ME materials (3).

Thus, we propose that the ME interaction in our Yb3+-based ferroelectric complex originates from magnetoelastic coupling that corresponds to the magnetic field–induced deformation of the crystal lattice in the paramagnetic phase. Because R,R-1 exhibited magnetostriction and ferroelectricity simultaneously, applying a magnetic field induced a mechanical strain via spin-lattice coupling that in turn influenced the polarization by affecting the overall dipole configuration. To support this, we made a comparative analysis of the crystal structures obtained under two directions of the magnetic field with respect to the zero-field one. Differences in the metal-ligand distances (tables S9 and S10) for both Yb3+ and Zn2+ ions could be discerned, which in turn affect the dipole order. Hence, the polarization values we calculated were found to be up to 10% greater or smaller (tables S3 and S11), depending on the orientation of the magnetic field, with respect to the zero-magnetic field value. Such results are in line with those obtained by PFM.

We have demonstrated room-temperature magnetoelectric control of ferroelectric domains in a molecule-based material. Thus, in the Yb3+-based chiral compound R,R-1, the combination of ferroelectric behavior with a magnetostrictive effect generates a strong ME coupling we observed at room temperature and with a relatively low magnetic field. These properties are useful for practical device application, including nonvolatile memory where information would be stored as electrically detectable and controllable by Yb3+ paramagnetism. More generally, such features appear particularly distinctive in single-phase materials and confirm that the genuine chemical design of multifunctional molecular materials may provide an alternative strategy to usual solid-state compounds for engineering ME devices.

Supplementary Materials

Materials and Methods

Figs. S1 to S32

Tables S1 to S11

References (5278)

Crystallographic Information Files

References and Notes

  1. Materials and methods are available as supplementary materials.
Acknowledgments: We thank J. Haines for fruitful discussions on crystallography. F. Salles is acknowledged for discussion on the point charge model. Funding: Financial support was provided by the University of Montpellier (UM), Centre National de la Recherche Scientifique (CNRS), Plateforme d’Analyze et de Caractérisation (PAC) et Institut Charles Gerhardt de Montpellier (ICGM), Fundação para a Ciência e a Tecnologia (FCT) Portugal (R&D project PDTC/QUI-QUI/098098/2008-FCOMP-01-0124-FEDER-010785), and NoE FAME. J.Lo. acknowledges the EMERGENCE@INC2019 funding. M.S.I. is grateful to FCT for financial support through the project MATIS–Materiais e Tecnologias Industriais Sustentáveis (CENTRO-01-0145-FEDER-000014). V.A.K. is grateful to FCT for financial support through the FCT Investigator Programme (project IF/00819/2014/CP1223/CT0011). This work was partly supported by funds from FEDER (Programa Operacional Factores de Competitividade COMPETE) and from FCT under the projects CICECO–Aveiro Institute of Materials (UIDB/50011/2020 and UIDP/50011/2020) and UID/FIS/04564/2019. Access to TAIL-UC facility funded under QREN-Mais Centro project ICT_2009_02_012_1890 is gratefully acknowledged as well. Author contributions: All authors contributed equally to this work. M.S.I. supervised, conceived of, and planned the SPM experiments. V.A.K. conducted the PFM and ME discussions. M.S.C.H. and J.A.P. performed the single-crystal XRD characterization. L.D.C. and R.A.S.F. measured the photoluminescence. B.D. resolved the crystal structures. E.M. and J.-M.T. performed the synthesis of the materials and standard characterizations. J.R. and M.B. measured the macroscopic dielectric and ferroelectric properties. D.G. and J.R. performed the single-crystal XRD characterization under magnetic field. J.La. and Y.G. provided critical feedback and helped with the analysis and writing process of the manuscript. J.Lo. conceived of the study, performed the magnetic characterization, and wrote the manuscript with input from all authors. Competing interests: All authors declare that they have no competing interests. Data and materials availability: All data are available in the main text or the supplementary materials.

Stay Connected to Science

Navigate This Article