WorksPhysics and Astronomy
Strain-driven stabilization of a room-temperature chiral multiferroic with coupled ferroaxial and ferroelectric order
Abstract
Noncollinear ferroic materials are sought after as testbeds to explore the intimate connections between topology and symmetry, which result in electronic, optical and magnetic functionalities not observed in collinear ferroic materials. For example, ferroaxial materials have ordered rotational structural distortions that break mirror symmetry and induce chirality. When ferroaxial order is coupled with ferroelectricity arising from a broken inversion symmetry, it offers the prospect of electric-field-control of the ferroaxial distortions and opens up new tunable functionalities. However, chiral multiferroics, especially ones stable at room temperature, are rare. We report the discovery of a strain-stabilized, room-temperature chiral multiferroic phase in single crystals of BaTiS, a quasi-one-dimensional (1D) hexagonal chalcogenide. Using first-principles calculations, we predict the stabilization of this multiferroic phase having space group for biaxial tensile strains exceeding 1.5% applied on the basal ab-plane of the room temperature phase of BaTiS. The chiral multiferroic phase is characterized by rotational distortions of select TiS octahedra around the long -axis and polar displacement of Ti atoms along the -axis. We used an innovative approach using focused ion beam milling to make appropriately strained samples of BaTiS. The ferroaxial and ferroelectric distortions, and their domains in -BaTiS were directly resolved using atomic resolution scanning transmission electron microscopy. Landau-based phenomenological modeling predicts a strong coupling between the ferroelectric and the ferroaxial order making -BaTiS an attractive test bed for achieving electric-field control of chirality-related phenomena such as circular photo-galvanic current and the Rashba effect.
Notice: This manuscript was authored by UT-Battelle, LLC, under Contract No. DE-AC05000OR22725 with the U.S. Department of Energy. The United States Government retains and the publisher, by accepting the article for publication, acknowledges that the United States Government retains a nonexclusive, paid-up, irrevocable, worldwide license to publish or reproduce the published form of this manuscript or allow others to do so, for the United States Government purposes. The Department of Energy will provide public access to these results of federally sponsored research in accordance with the DOE Public Access Plan (http://energy.gov/downloads/doe-public-access-plan).
Main Text
Ferroic materials are characterized by spontaneous symmetry breaking with the order parameter being responsive to external fields. They form important components of modern-day electronics[1-2]. Conventional ferroic materials include ferroelectrics – that undergo inversion symmetry breaking leading to electric polarization, ferromagnets – that undergo time-reversal symmetry breaking leading to spin-polarization or magnetization, and ferroelastics – that undergo symmetry lowering leading to spontaneous strain. Recently, there is a growing interest in ferroic materials having non-collinear order due to their fascinating connections between topology and symmetry, which can enable new electronic, magnetic and photonic functionalities[3-6]. For example, the stabilization of magnetic skyrmions have opened up pathways for low-power information storage[6]. Non-collinear ferroics include ferrotoroidal and ferroaxial materials[3, 7-9]. Ferrotoroids have a toroidal arrangement of magnetic moments and break both space-inversion and time-reversal symmetries[9-10]. Ferroaxial (or ferrorotational) materials have rotational structural distortions that preserve both space-inversion and time-reversal symmetries, but break mirror symmetry[7-9, 11-12]. Ferroaxial materials can potentially demonstrate bulk or planar chirality and lead to such physical phenomena as circular dichroism, optical rotation[8, 13-15], and topological structures such as dipolar skyrmions[18] [16, 17]. Dynamic control over these phenomena can enable energy efficient devices for information storage and processing. However, purely ferroaxial materials cannot couple with external fields[19], and hence, external control of chirality is challenging, although few solutions do exist[18, 20, 21].
Ferroaxial order, if coupled with either ferroelectric or ferroelastic order, can overcome the above limitation and enable control over chirality using an external field. It can also result in rich physical phenomena such as ferroelectric Rashba effect[22], and tunable circular photogalvanic response effect[23] and Kramer-Weyl fermions[4]. A group theoretical analysis shows that there are 124 unique phase transitions from a high-symmetry parent phase that result in a lower symmetry ferroaxial phase; 64 of those also result in ferroelectricity[19]. Despite the large probability of finding chiral multiferroics with simultaneous ferroelectric and ferroaxial order, there are only a handful of reports of such materials. Examples include hexagonal below 74 K[22], below 24 K[12], below 4 K[23], below 90 K[11], and certain organic crystals including Rochelle salt below 297 K and triglycine sulfate below 321 K[24]. If one ignores the organic crystals due to their incompatibility with the processing conditions of modern electronics, all the reported inorganic materials are stabilized in a chiral ferroic phase much below room temperature.
In this Article, we report the strain-induced stabilization of a chiral multiferroic phase having a strong coupling between the ferroelectric and ferroaxial order parameters in single crystals of a quasi-one-dimensional (1D) chalcogenide . At room temperature, exists in the achiral phase, which has antipolar displacements of Ti atoms along the -axis[25, 26]. Using group theory and first-principles calculations, we predict that a biaxial strain can stabilize in the space group — that involves a rotation of octahedra about the -axis and polar displacements of Ti atoms along the -axis. Using phenomenological Landau theory-based modeling, we predict a strong coupling between the two order parameters. We devised an innovative focused ion beam (FIB)-based strategy to biaxially strain single crystals by over 3% — that is outside the range of strains achievable using epitaxy — and stabilize the chiral multiferroic phase at room temperature. We used scanning transmission electron microscopy (STEM)-based imaging and diffraction methods to directly map the ferroelectric and ferroaxial order parameters with atomic resolution, thus confirming the theoretical predictions. We also show presence of nanoscale chiral domains with sharp domain boundaries. We expect that the room-temperature chiral multiferroic phase in can enable strong electric-field switching of chirality.
Theoretical prediction of strain-induced stabilization of a chiral multiferroic phase in
is a quasi-1D semiconductor having a small band gap (0.3 eV)[25, 26, 27]. It shows giant optical anisotropy[26, 27], and glass-like thermal properties[28]. It also shows multiple electronic and structural phase transitions upon cooling[25]. At room temperature, the global refinement of synchrotron X-ray diffraction of single crystals leads to an average structural symmetry corresponding to the space group[25, 29]. This structure consists of face-shared octahedral chains along the -axis. The chains are arranged in a hexagonal manner and are stacked between chains of Ba atoms, as illustrated in Figure 1(a). is a noncentrosymmetric space group lacking space-inversion symmetry along the -axis, which is caused by the off-centering or polar displacements of Ti atom from the centroid of the octahedron. Within a chain, the Ti polar displacements along the -axis are aligned parallel to each other. Between neighboring chains, these displacements are aligned in an antiparallel manner, resulting in an anti-polar ordering. There is a net polarization along the -axis due to unequal upward () and downward () Ti polar displacements in the unit cell, as illustrated in Figure 1(a). The Ti atoms are also off-centered along the -plane. At room temperature, these -plane displacements are globally disordered, but are ordered locally over a few unit cells[26]. Below 240 K, the -plane Ti displacements order to form a three-dimensional toroidal dipolar structure with antiparallel dipoles along the -axis and an array of vortex-vortex-antivortex dipolar pattern along the -plane[29]. Given its susceptibility to form non-collinear dipolar textures at cryogenic temperatures, we selected to explore whether non-collinear distortions can also be stabilized at room temperature with external perturbations such as strain.
Figure 1. Atomic models showing chiral multiferroic order in BaTiS. a. Achiral -BaTiS lacks space-inversion symmetry along the -axis due to off-center displacements of Ti atoms within the TiS octahedra, as shown in the bottom panel. These Ti displacements have an overall ferrielectric order due to a mixture of parallel and antiparallel alignment of the dipoles, and lead to two net polar states ( and ) depending on the direction of the displacements. For simplicity, only the positive polar state () is shown. The dash lines in the top panel denote the mirror planes including a 6-fold rotation axis. b, c. Simultaneous ferroaxial order and ferroelectric order in -BaTiS. Two (out of the four) domain states, and , are shown. In the top-panels, a clockwise rotation () of TiS octahedra is shown in b and an anticlockwise rotation () in c. Blue and red arrows represent the direction of the rotational distortions. The rotated TiS octahedra and the unrotated ones, with respect to -BaTiS, are highlighted in red and blue, respectively, in b,c. The bottom panels in b,c show a parallel alignment of the Ti off-center displacements along the -axis, with the net polarization point up in b, and down in c. Barium, titanium and sulfur atoms are shown in orange, blue and grey, respectively.
We used group-theoretical tools to search for symmetry-related sub-groups of the room-temperature phase that belong to one of the 65 Sohncke groups and are expected to show chirality[30]. We find that the chiral sub-group can be accessed with minimal distortions, and it also lacks inversion symmetry. The mirror-symmetry breaking from - can be introduced through a distortion mode, which results in either clockwise or anticlockwise rotations of the octahedral chains lying at the corner of a unit cell, with the rotations being in-phase along a chain. These rotational distortions introduce a ferroaxial order, or parallel to the 6-fold rotation axis (-axis), as shown schematically in Figure 1(b–Figure 1c). Meanwhile, a distortion mode drives the Ti polar displacements along the -axis to align in a parallel manner between neighboring chains in the phase, as opposed to their antipolar ordering in the phase. Additionally, the distortion involves outward displacements of neighboring Ba atoms, causing an expansion of some of the Ba hexagons and leading to the dimerization of Ba atoms, as illustrated in Fig. S1(d–e). Further details of the distortion mode analysis are provided in Supplementary Section 1. Together, these two distortion modes lead to simultaneous ferroelectric and ferroaxial transitions from achiral - to chiral -. The different combinations of chirality ( or ) and polarization ( or ) in - can be expected to result in a multi-domain structure with four degenerate chiral multiferroic states (, , , and ), which are related to each other by glide and mirror symmetries.
Next, we used first-principles density-functional theory (DFT) calculations to find the conditions that can activate the and distortion modes in achiral - and stabilize the chiral multiferroic
-BaTiS. A geometry optimization was performed with the and distortion modes imposed in achiral -BaTiS. The optimized crystal structure was found to be in a local minimum and retained both the ferroelectric and ferroaxial order, as shown in Figure 1(b-Figure 1c). Moreover, the optimized -BaTiS lattice was found to have larger lattice constants along the plane with Å as compared to Å in -BaTiS, which suggests that the chiral multiferroic phase in BaTiS can be potentially stabilized using biaxial tensile strain along the -plane. The calculation details can be found in Methods and Supplementary Section 2. We use the optimized lattice parameters of -BaTiS as a reference (i.e., zero strain). At zero strain, -BaTiS is energetically unfavorable. However, its energy is lowered with respect to -BaTiS with increasing biaxial tensile strain along the -plane. For biaxial strains (tantamount to uniaxial-compressive strain along the -axis of ), we find that the -BaTiS phase becomes lower in energy than -BaTiS [Fig. S2(a)], thus providing a pathway to realize the chiral multiferroic phase experimentally. In addition to the rotational distortions of TiS octahedra, the optimized -BaTiS lattice also displays concomitant -plane displacements of Ba atoms that lead to three distinct local BaTiS structural motifs [Supplementary Section 1 Fig. S1(d-f)].
Direct observation of ferroelectric and ferroaxial order in BaTiS
To verify the theoretical predictions, we synthesized BaTiS single crystals using an iodine-assisted vapor transport method, which has been reported earlier[31]. The typical way to impose biaxial strains is through the growth of thin films that are epitaxially strained to a substrate of appropriate dimensions. The growth of epitaxially strained thin films of ternary chalcogenides is still in a nascent state[32, 33, 34, 35]. A recent report demonstrated the growth of BaTiS thin films that were found to be polycrystalline in nature[34]. To overcome these challenges, we used an innovative strategy involving focused ion beam (FIB) milling of a confined single crystal to introduce biaxial strain. Briefly, this process involves attaching an appropriately oriented single crystal to a TEM post such that the two opposite lateral sides of the single crystal are fixed to provide the confinement, instead of the conventional approach of attaching the specimen to the TEM post with only one side fixed. This was followed by plasma FIB milling to make a thin, electron-transparent lamella (more details in Supplementary Section 3 Fig. S3). The FIB milling also introduces stresses due to ion impact and surface amorphization[36, 37]. For the TEM lamellae with two sides fixed, we expect the intrinsic stress from the FIB milling will be confined to trigger the phase transition, whereas the TEM lamellae with only one side fixed to the post, the residual stresses can relax through the unattached end. We repeated this process three times and could reproducibly achieve a biaxially strained sample with strains ranging between 3.08–3.21%.
To directly resolve the crystal structure of the strained BaTiS sample and compare with the unstrained sample, we performed atomically resolved imaging using an aberration-corrected STEM. A high-angle annular dark field (HAADF) STEM image of a typical strain-relaxed lamella of -BaTiS viewed along [001] is shown in Figure 2(b). In this imaging mode, the intensity of the atomic columns is proportional to the square of their effective atomic number ()[38]. The heaviest Ba columns appear brightest and form a regular hexagonal network. The lighter Ti columns have lower intensity, and are present about the centers of the Ba hexagons. A quantitative analysis of the Ba and Ti atomic columns shows that the Ti atoms are off-centered with respect to the centroids of the surrounding Ba hexagons. These Ti displacements are ordered locally over 2–5 nm, but are disordered over the field of view of the image (see Fig. S4 in Supplementary Section 4). These observations are in good agreement with a recent study[26]. The lightest S columns around the Ti atoms show up as weak streaks.
Figure 2. Strain-induced stabilization and atomic-scale observation of ferroaxial order in . a. Phase stability as a function of biaxial strain on the -plane. is stable over for tensile strains exceeding 1.5%. A representative atomic-resolution STEM-HAADF image of - in b and biaxially strained - in c. Both the samples are viewed along the [001] axis. The structure models of and - phases are overlaid onto the STEM-HAADF images to provide clearer identification of the atomic columns. d-g. The top four principal components extracted from a -means clustering analysis of the HAADF image of -. The overlaid white arrows represent the displacement of Ba atomic columns relative to the undistorted hexagonal Ba sublattice in -. The rotational distortion of octahedron can be seen in f, and contrasted with unrotated octahedra in d, e, and g. Ba–Ba distance mapping for - in h and - in i. The overlaid lines connect neighboring Ba atoms. The magnitude of Ba–Ba distance is represented using the colorscale shown on the right. j. Histogram plot showing the distribution of Ba–Ba distances in - and -. - shows a unimodal distribution with an average Ba–Ba distance of 3.9 Å, while - has a bimodal distribution with an average value of 4.02 Å. It is equivalent to an average biaxial tensile strain of 3.08%. The histograms were fitted with Gaussian curves shown with solid lines. Scale bars are 2 nm in (b-c), 2 Å in (d-g), and 5 nm in (h-i).
A HAADF image of the -plane of a typical biaxially strained lamellae is shown in Figure 2(c). In contrast to the corresponding image of -BaTiS [Figure 2(b)], the strained structure shows a periodic distortion of the TiS octahedra involving a -degree rotation. The Ba atoms also undergo breathing displacements along the -plane to form irregular hexagons. Together, the TiS octahedral rotations and -plane displacements of Ba result in a mirror-symmetry breaking parallel to the 6-fold rotation axis, as we had discussed for the DFT-optimized -BaTiS chiral multiferroic phase [Figure 1(b-Figure 1c)]. The features of -BaTiS in the strained crystal were further confirmed by comparing them with the corresponding simulated HAADF image of a DFT-optimized structure, and its FFT analysis, as shown in Fig. S5(c-d, g-h). We find excellent agreement between the simulated and experimental images. To better resolve the rotated TiS octahedra in -BaTiS, we performed differential phase contrast (DPC) imaging by collecting four-dimensional (4D) datasets using a pixelated detector[39, 40, 41]. The reconstructed atomic potential images (see Fig. S6 in Supplementary Section 4) provide a clear view of the lighter sulfur atomic columns forming rotated and unrotated TiS octahedra in the strained BaTiS sample.
To gain further insights into the local structure and identify the different configurations of the Ba sublattice in the two samples, we used principal component analysis (PCA) (details can be found in Supplementary Section 5). The scree plot obtained after performing PCA on the HAADF image of - shows a gradual variation, suggesting similar local configurations with only small variations in the spacing between Ba columns. In contrast, we find four distinct components in the scree plot of the strained - [Fig. S7(c)] sample, suggesting four different local configurations. We used unsupervised -means clustering analysis to group Ba atoms based on the similarity of the local structural motifs centered around columns (6 Ba atoms per structural motif). The four local structural motifs in [001] - are shown in Figure 2(d-Figure 2g). In the strained sample, we observe nanoscale (~15 nm) domains consisting of three distorted Ba sublattice configurations, with the domains being separated by sharp interfaces consisting of a less distorted Ba sublattice [Fig. S8(d)]. We extracted the atomic displacements in the four motifs by comparing the column positions to that of the regular hexagons in -. As shown in Figure 2(b), we find the Ba atoms in - undergo different kinds of displacements in the -plane. For the rotated octahedron shown in Figure 2(f), its neighboring Ba atoms undergo outward displacements, which expand the Ba hexagon. These outward displacements subsequently lead to a dimerization of Ba atoms, as shown in Figure 2(d-Figure 2e). The structural motif in Figure 2(g) was extracted from the sharp domain boundaries, as shown in Fig. S8(d). These local symmetry features of -, as resolved by atomic-resolution STEM characterization, exhibit a consistent match with the DFT-optimized - lattice, as shown in Fig. S1(d-f).
We used the Ba–Ba distances from the HAADF images of the -plane of - and - to measure the biaxial strain in the latter. As opposed to the uniform Ba–Ba distances in - [Fig. 2(h)], there is substantial modulation of Ba–Ba distances in - [Fig. 2(i)]. The nanoscale domains (of around ~15 nm size) are found to be separated by sharp interfaces consisting of hexagons with shorter Ba–Ba distance [Fig. 2(i)]. A similar domain pattern was also revealed in our -means clustering analysis in Fig. S8(d). Notably, the average Ba–Ba distance in - is measured to be larger than - by 3.08%, which implies a biaxial tensile strain state in - [Fig. 2(j)], and is consistent with the DFT-predicted stability of the latter under tensile strains. The strain results were also obtained from Ti–Ti distance mapping for - and -, as shown in Fig. S9, and are consistent with the Ba–Ba distance results. More domain patterns can be seen from a larger field of view observation of - in Fig. S10(e-f).
Having confirmed the ferroaxial distortions along the -plane in the strained sample, we next performed STEM imaging of the strained and strain-relaxed samples oriented along [1\bar{1}0] to probe the predicted change from antipolar Ti displacements along the -axis in - to polar displacements in -. The antipolar displacements in - lead to a separation of the S and Ti columns along the -axis, as shown in Figure 3(a). As a result, the horizontal intensity profile between two adjacent Ba atomic columns exhibits only one peak in the middle corresponding to an S atomic column [Fig. 3(c, e)], with the Ti atomic column displaced either upwards or downwards. In the HAADF image of the strained - lattice shown in Figure 3(b), the polar displacements of Ti atoms along the -axis aligns them with the S atomic columns (the Ti and S columns are offset along the -axis). Consequently, two peaks, instead of one, show up in a horizontal line profile of the HAADF intensity across two Ba atoms [Fig. 3(d, e)]. The experimentally observed changes in intensity are in excellent agreement with the simulated STEM results for - and - oriented along [1\bar{1}0], as shown in Figure 3(f). Overall, using STEM, we are able to confirm the presence of both ferroaxial and ferroelectric order in the strained sample of -.
!Atomic-scale structures of BaTiS3 viewed along the [1\bar{1}0] zone axis and valence EELS showing changes in electronic structure with the phase transition bbox=0.20,0.09,0.80,0.59
Figure 3. Atomic-scale structures of BaTiS viewed along the [10] zone axis and valence EELS showing changes in electronic structure with the phase transition. a. Atomic arrangement in P6cm-BaTiS along the -axis, where S atoms and Ti atoms misalign with each other along the horizontal direction within the dashed box region. b. Atomic arrangement in P6-BaTiS having ferroelectric distortion of Ti atoms along the -axis, which aligns the S atoms with Ti along the horizontal direction shown with the dashed box. c-d. Experimental and simulated STEM-HAADF images of P6cm-BaTiS and P6-BaTiS, both oriented along the [10] zone axis showing the atomic arrangements between the Ba columns. e-f. Horizontal intensity profiles along the dashed lines in the experimental and simulated STEM images of P6cm-BaTiS and P6-BaTiS lattices in c and d, respectively. Compared to the single peak due to S atomic columns between Ba peaks in P6cm-BaTiS, an additional peak corresponding to Ti atomic columns indicated by the red arrow shows up in P6-BaTiS. g. DFT calculated absorption spectra of P6cm-BaTiS and P6-BaTiS for comparison. h. Monochromated valence EELS measurements for valence electron excitation showing the strong absorption peak in P6-BaTiS which is consistent with DFT calculations in g. Scale bars are 1 nm in (c-d).
While the STEM-HAADF images show the presence of multiple domains in the strained -BaTiS sample [see Figure 2(i)], it does not provide evidence that the different domains have different chirality. Chiral domains can be mapped by scanning a circularly polarized light over the sample, which will interact differently with domains of different handedness or chirality[18] [13, 14, 15]. However, the 15 – 20 nm wide domains in strained -BaTiS are much smaller than the spot size of a typical laser, thus making it challenging to map their chirality using polarized light. Instead, we used convergent beam electron diffraction (CBED) in a STEM to map domains having different chirality. STEM-CBED is a powerful tool to extract information about the local crystal structure and symmetry[42, 43, 44]. We used a nanometer-sized (~1 nm) electron probe and rastered it across the -BaTiS lamella (see experimental details in Methods). At each probe position (2 dimensions in the real space), we recorded the diffraction pattern (2 dimensions in the reciprocal space) using a pixelated direct-electron detector. The acquired 4D dataset, formed by zero-order Laue zone patterns, includes information about the rotational and mirror symmetries of the structure[43, 44], and was used to extract the distribution of chiral domains. We carried out unsupervised -means clustering analysis to automatically separate the domains based on the symmetry information recorded in each CBED pattern. As shown in Fig. S11(a-b), two distinctive regions possessing different rotational symmetries are identified in the scanned area. The corresponding CBED patterns from these two regions, as shown in Fig. S11(c-d), demonstrate distinct 3-fold rotational symmetry originating from screws in -BaTiS. Moreover, these two CBED patterns are related through a mirror reflection with the mirror plane being parallel to (110). This mirror operation is one of the symmetry elements which is present in the high-symmetry -BaTiS phase, but is lost in the -BaTiS phase due to the ferroaxial distortions involving clockwise or counterclockwise rotation of TiS octahedra. Thus, the STEM-CBED experiments provide confirmation that the domains identified using STEM-HAADF imaging indeed have different chirality. We also performed composition analysis of the two samples using atomically resolved electron energy-loss spectroscopy (EELS) analyses (see details in Supplementary Section 7 Fig. S12), and observed no elemental segregation in either sample, even at the domain boundaries in the chiral multiferroic phase. It confirms that the observed ferroelectric ferroaxial order in -BaTiS is a displacive-type transition, as opposed to the order-disorder type transition reported in NiTiO, a ferroaxial material[15].
Electronic structure of chiral multiferroic -BaTiS
Having successfully demonstrated the presence of ferroelectric ferroaxial distortions that lead to chiral multiferroic ordering in biaxially strained -BaTiS, we next discuss its electronic properties (see details in Supplementary Section 8). We observe a slight increase in the theoretical bandgap of -BaTiS (0.32 eV) with 3.5 % strain over unstrained -BaTiS (0.24 eV), calculated at the PBE level (see Fig. S13). The ferroaxial distortions in -BaTiS introduce localized S-3p states originating from rotated TiS octahedra [Fig. S13(d)], which results in flat valence bands close to Fermi energy [Fig. S13(c)]. To confirm the presence of the flat bands, we performed valence EELS measurements on the two samples using a monochromated STEM. The normalized EELS results for valence electron excitations are compared in Figure 3(h). We observe a prominent absorption peak at ~0.34 eV in -BaTiS, due to the flat bands, which are absent in -BaTiS. The corresponding EELS spectra exhibit excellent agreement with the absorption spectra of band edge excitations calculated using DFT in Figure 3(g). The deviations between the EEL and the DFT-calculated spectra at higher energies (above 0.45 eV) may be attributed to the coexistence of other energy excitations, such as collective excitations of valence electrons (plasmons) and Cerenkov radiation[45], and the instrumental broadening, all of which tend to smear the spectroscopic features. Additionally, the influence the chiral domain boundaries can also affect the absorption features in strained -BaTiS, which warrants further investigation.
Coupling between ferroelectric and ferroaxial order in -BaTiS
We now discuss the origin of the ferroelectric ferroaxial transition and the coupling between the two order parameters in strained -BaTiS. As mentioned before, each of the order parameters has two degenerate configurations ( and ), and together they lead to four domains ( to , or to ) that are degenerate in energy. To estimate the barriers related to the transition between the four configurations, we combined DFT calculations with a Landau model of phase transitions[46]. The free energy of strained BaTiS, , was expanded as a power series of the two primary order parameters, and (see details in supplementary section 9). We used the centrosymmetric and achiral structure as the reference phase for the transition between opposite ferroelectric ferroaxial order ( to , or to ). Symmetry analysis confirms that the transition from to is a valid ferroelectric ferroaxial transition[19]. We note that the actual switching pathway in the experiments may involve complex domain boundaries or other metastable phases[25, 29], and using the structure as the saddle point likely gives an upper bound of the switching barriers. The amplitudes of ferroelectric and ferroaxial modes, and , have been normalized to those in -BaTiS under 3.5% tensile strain, which is close to the experimentally measured strain. The free energy expression for the coupled transition in terms of and up to their eighth order coupling can then be written as follows:
where and are, respectively, the coefficients for individual ferroaxial and ferroelectric mode, while and are coefficients accounting for the coupling between the two modes. Figure 4a shows the DFT-calculated free energy as a function of the amplitude of the individual ferroelectric mode (), ferroaxial mode (), and the coupled mode in strained -BaTiS. We find that ferroelectric is characterized by a typical double-well potential, and by itself, leads to a relatively modest lowering of free energy over the reference strained -BaTiS phase. In contrast, the ferroaxial mode, , by itself, is unstable as it forms a single-well potential with a minimum at , and increasing energy for finite . However, when the two modes are coupled, we observe an appreciable lowering of energy for -BaTiS, which indicates a strong coupling between the two order parameters. Energy contributions for the -BaTiS transition were calculated to be 658.0 meV/f.u. for pure , 311.6 meV/f.u. for pure , and -1066.3 meV/f.u. for the coupling term, respectively (Fig. S17; The fitted Landau coefficients can be found in Table S1 in supplementary section 9). It also implies that the chirality is improper in nature, as it is stabilized only due to its strong coupling with the ferroelectric mode. The resulting energy landscape between the four degenerate chiral multiferroic states, which is obtained using the fitted energy model, is shown in Figure 4b. We find that the lowest energy pathway to switch the direction of the polarization is also expected to switch the handedness or the chirality , due to the strong coupling between the two modes. Thus, we predict that -BaTiS should allow for electric-field tunability of chirality.
Figure 4. Landau free energy landscape for the coupled ferroelectric ferroaxial transition. a. DFT-calculated potential energy of the primary modes ( and ), and the coupled distortion mode in strained -BaTiS. The normalized distortion amplitude () = 0 corresponds to the centrosymmetric phase, while 1 corresponds to the phase, with both the phases under 3.5% biaxial strain. b. Contour plot of the Landau free energy as a function of the primary order parameters (, ). The four degenerate chiral multiferroic states are labeled.
Conclusion and outlook
In summary, we have demonstrated the successful stabilization of a room-temperature chiral multiferroic phase in single-crystals of quasi-1D -BaTiS under biaxial tensile strain. This multiferroic state includes ferroelectric order along the long -axis, and ferroaxial order around the same axis, both of which were confirmed using atomic-resolution STEM imaging, diffraction and EELS. We observe the presence of nanoscale chiral domains that are separated by sharp domain boundaries. In pure ferroaxial materials, it is non-trivial to switch their chirality as the ferroaxial order is invariant under both time-reversal and space-inversion operations, and hence, does not couple to typical thermodynamic fields. However, in -BaTiS, we predict the presence of a strong coupling between the ferroelectric and ferroaxial order. In fact, the ferroaxial order, by itself, is found to be unstable, and is stabilized due to the coupling with the ferroelectric order in -BaTiS. Thus, we predict -BaTiS to be a strong candidate to show electric-field tuning of chirality.
To demonstrate electric-field control of chirality in -BaTiS, several open questions will need to be addressed. Our STEM studies show that the chiral domains in strained -BaTiS are extremely small, spanning about 15 – 20 nm in size. Such small domains pose a challenge for probing using conventional optical probes. It might be possible to probe their chiral properties using electron energy-loss magnetic chiral dichroism carried out within a STEM (STEM-EMCD)[47], or through electron vortex beams[48], which are sensitive to local orbital angular momentum. Annealing the material might lead to a reduction in the chiral domain walls and increase the size of the chiral domains making them suitable for probing using conventional optical techniques. Lastly, many ternary oxides and chalcogenides adapt the hexagonal perovskite structure. It might be possible to stabilize them in a chiral multiferroic phase using strain to obtain electric-field tunable chiral materials for different wavelengths and applications. In fact, the oxide analogue of , i.e., hexagonal already undergoes a ferroelectric ferroaxial transition below 74 K[22]. It might be possible to push the transition temperature of towards room temperature through epitaxial growth, which is well established for oxides.
Methods
Sample preparation
Single crystals of were synthesized using an iodine-assisted chemical vapor transport (CVT) method as reported earlier[25, 26, 27, 28, 31]. [001]-orientated TEM lamellae were prepared using a Thermo Scientific Helios G4 Plasma focused ion beam (PFIB) Xe Dual Beam system. A Pt layer was deposited over the region of interest prior to ion milling. Xe PFIB was used to make TEM sample with a final thickness of ~100 nm using 30 kV and 0.1 nA as the final voltage/current. TEM lamella with different strain states were prepared through a standard lift-out routine: one [001]- lamella was welded onto a TEM half grid with one side fixed such that strain relaxation was possible through the unconstrained side. For another lamella, we attached it to a TEM half-grid with two lateral sides fixed such that biaxial strain can be introduced inside the lamella upon thinning. All the TEM lamellae were further thinned down to an electron transparent thickness utilizing a Fischione’s Model 1040 NanoMill TEM specimen preparation system at an operating voltage of 900 V and at incident angels of 5° and -5°.
Electron microscopy
Atomic resolution -contrast STEM-HAADF imaging was carried out using a Nion UltraSTEMTM equipped with a spherical aberration corrector operating at 100 kV. The STEM-HAADF images were acquired with a convergence semi-angle of 30 mrad and collection semi-angles from 80 to 200 mrad. To correct the sample drift effect, we recorded each STEM dataset as a series of 20 successive frames with fast scanning (1 μs/pixel dwell time) for post-processing correction through image registration. STEM-HAADF simulations were performed using the multi-slice method as implemented in STEM[49]. and crystal structures optimized by DFT were used for the STEM simulations. To match the experimental conditions, we performed the simulations using an aberration-free probe with an accelerating voltage of 100 kV and a convergence semiangle of 30 mrad. Thermal scattering was included in our simulation through the phonon excitation model proposed by Forbes et al[50]. The sample thickness was set to 15 nm and the defocus value was set to 10 Å to obtain good agreement with the experimental results.
Lattice distortions were obtained from the atomic-scale STEM-HAADF images by mapping the atomic distances between neighboring Ti–Ti and Ba–Ba atoms. Atomic positions were determined by fitting 2D Gaussian peaks to Ti and Ba columns. The local crystallography for Ba and Ti sublattices was mapped out using data analytics tools available within the Pycroscopy suite[51]. All the image data analyses and visualizations were carried out using the home-developed python-based codes.
4D-STEM experiments were carried out on a Nion UltraSTEMTM at 100 kV using a Dectris ELA high-speed electron-counting detector. Differential phase contrast (STEM-DPC) datasets were acquired with pixel size using a 30 mrad convergent beam. At each pixel position, a pixel Ronchigram image was recorded with a dwell time of 0.5 ms. Convergent-beam electron diffraction (STEM-CBED) patterns were obtained with a probe semi-angle of 2.5 mrad to avoid the overlap of Bragg discs. Diffraction patterns were collected using a step size of 1 nm with scan positions. Dwell time for each pixel was set at 2 ms. Each diffraction pattern was recorded on Dectris ELA detector with frame size of pixels. -means clustering was applied for differentiating the local rotational and mirror symmetries of the projected structure.
Core-loss electron energy loss spectra (EELS) were acquired with pixel dwell times of 3 ms, energy dispersion of 1 eV per channel, and a collection semi-angle of 35 mrad on a Nion Iris spectrometer attached to the Nion UltraSTEM. Ba-, Ti-, and S- edges were used for elemental quantification. The spectrum images were denoised using principal component analysis (PCA). A power law was used to model the background signal prior to the core-loss signal. Core-loss EELS data were processed using Digital Micrograph.
Electronic structures of -BaTiS and -BaTiS were characterized by valence EELS acquired at an operating voltage of 100 kV using a Nion IRIS EEL spectrometer attached to the Nion aberration-corrected monochromated STEM-EELS (Ultra-HERMES) at Oak Ridge National Laboratory (ORNL). The Ultra-HERMES at ORNL is equipped with a Dectris ELA high-speed electron-counting detector. The probe convergence semi-angle of 32 mrad, and an entrance aperture collection angle of 25 mrad were used to record the spectra. An energy dispersion of 0.85 meV per channel was selected to achieve an energy resolution of 13.2 meV and 18.7 meV for unstrained and strained BaTiS samples, respectively (see Supplementary Fig. S15). The spectrum-images ( pixels) were recorded with dwell times of 10 ms per pixel. To process the valence EELS spectra, we first calibrated the recorded EEL spectra by aligning the center of the zero-loss peak (ZLP) at 0 meV using a correlation-based routine as implemented in the Nion Swift package. For a good approximation of scattering probability, all the EELS signals were normalized to the ZLP intensity. Background subtraction was performed by fitting a power-law function with a fitting window from 0.25 to 0.30 eV in low-loss EELS spectra.
Density-functional theory calculations
First-principles calculations based on density functional theory were performed using the Vienna Ab initio Simulation Package (VASP)[52]. We used projector augmented-wave (PAW) potentials to treat the electron-ion interactions. The generalized gradient approximation within the Perdew-Burke-Ernzerhof (GGA-PBE) functional was adopted for exchange-correlation interactions[53, 54, 55]. A plane-wave basis set with a cutoff energy of 650 eV was used according to convergence tests. The Brillouin zone was sampled using a -centered -point mesh with spacing of 0.025 Å. Convergence for electronic calculations was reached with energy change under eV. To investigate the effect of strain on the coupled ferroelectric ferroaxial order, we firstly initialized rotational distortion of TiS octahedra in strained -BaTiS lattice and then optimized the cell volume along the axis under biaxial strain conditions. The ionic positions were relaxed until the forces were smaller than eV/Å.
To compare the valence EELS results, the frequency-dependent dielectric function of -BaTiS and unstrained -BaTiS crystals were simulated using the independent particle approximation (IPA)[56]. The imaginary part is obtained by calculating the direct transitions between occupied and unoccupied states. The absorption coefficient is calculated as (cm, Gaussian unit), where is the refractive index and is the speed of light in vacuum. To match the energy resolution with valence EELS spectra, we set the CSHIFT tag in VASP to 0.01 and NEDOS = 10000.
The distortion modes of -BaTiS from the centrosymmetric -BaTiS structure were investigated using the ISODISTORT Software Suite[57, 58].
Acknowledgements
This work was supported, in part, by an ARO MURI program with award no. W911NF-21-1-0327, and the National Science Foundation (NSF) of the United States under grant numbers DMR-2122070 (G.Y.J.), DMR-2122071, and DMR-2145797 (R.M.). M.C., A. R. L., and J. A. H. are supported by the U.S. Department of Energy, Office of Science, Basic Energy Sciences, Materials Sciences and Engineering Division. The Microscopy work was conducted as part of a user project at the Center for Nanophase Materials Sciences (CNMS), which is a DOE Office of Science User Facility using instrumentation within ORNL’s Materials Characterization Core provided by UT-Battelle, LLC, under Contract No. DE-AC05-00OR22725 with the DOE and sponsored by the Laboratory Directed Research and Development Program of Oak Ridge National Laboratory, managed by UT-Battelle, LLC, for the U.S. Department of Energy. This work used computational resources through allocation DMR160007 from the Advanced Cyberinfrastructure Coordination Ecosystem: Services & Support (ACCESS) program, which is supported by NSF grants #2138259, #2138286, #2138307, #2137603, and #2138296.
Author Contributions
G.R. and R.M. conceived the idea and designed the experiments. H.C., B.Z. and J.R. synthesized the single crystals of BaTiS. H.C. performed PFIB sample preparations. G.R. and R.M. carried out all the STEM measurements and their analyses. R.K.V., J.A.H., A.R.L., and M.C. supervised STEM experiments and analyses. G.Y.J., G.R., and R.M. performed the first-principles DFT calculations and Landau free energy analyses. C.W. and D.X. contributed to the discussion of theoretical results. G.R. and R.M. drafted the manuscript with edits from all authors.
Competing interests
The authors declare no conflict of interest.
Data availability
Data presented in the main text of this paper are openly available in Zenodo at https://doi.org/10.5281/zenodo.12810613. Data presented in the Supplementary Information are available upon request from the corresponding authors.
References
- [18]P. Behera, M. A. May, F. Gómez-Ortiz, S. Susarla, S. Das, C. T. Nelson, L. Caretta, S.-L. Hsu, M. R. McCarter, B. H. Savitzky, Science advances 2022, 8, eabj8030.
- [19]J. Hlinka, J. Privratska, P. Ondrejkovic, V. Janovec, Physical review letters 2016, 116, 177602.
- [20]D. Di Sante, P. Barone, R. Bertacco, S. Picozzi, Advanced materials (Deerfield Beach, Fla.) 2012, 25, 509.
- [21]F. De Juan, A. G. Grushin, T. Morimoto, J. E. Moore, Nature communications 2017, 8, 15995.
- [22]Y. Akishige, G. Oomi, T. Yamaoto, E. Sawaguchi, Journal of the Physical Society of Japan 1989, 58, 930.
- [23]A. J. Hearmon, F. Fabrizi, L. C. Chapon, R. Johnson, D. Prabhakaran, S. V. Streltsov, P. Brown, P. G. Radaelli, Physical review letters 2012, 108, 237201.
- [24]G. L. Rikken, N. Arvarvari, Nature Communications 2022, 13, 3564.
- [25]H. Chen, B. Zhao, J. Mutch, G. Y. Jung, G. Ren, S. Shabani, E. Seewald, S. Niu, J. Wu, N. Wang, Advanced Materials 2023, 35, 2303283.
- [26]B. Zhao, G. Ren, H. Mei, V. C. Wu, S. Singh, G. Y. Jung, H. Chen, R. Giovine, S. Niu, A. S. Thind, Advanced Materials 2024, 2311559.
- [27]S. Niu, G. Joe, H. Zhao, Y. Zhou, T. Orvis, H. Huyan, J. Salman, K. Mahalingam, B. Urwin, J. Wu, Nature Photonics 2018, 12, 392.
- [28]B. Sun, S. Niu, R. P. Hermann, J. Moon, N. Shulumba, K. Page, B. Zhao, A. S. Thind, K. Mahalingam, J. Milam-Guerrero, Nature Communications 2020, 11, 6039.
- [29]B. Zhao, G. Y. Jung, H. Chen, S. Singh, Z. Du, C. Wu, G. Ren, Q. Zhao, N. S. Settineri, S. J. Teat, arXiv preprint arXiv:2406.09530 2024.arxiv.org/abs/2406.09530
- [30]G. H. Fecher, J. Kübler, C. Felser, Materials 2022, 15, 5812.
- [31]B. Zhao, M. S. B. Hoque, G. Y. Jung, H. Mei, S. Singh, G. Ren, M. Milich, Q. Zhao, N. Wang, H. Chen, Chemistry of Materials 2022, 34, 5680.
- [32]M. Surendran, S. Singh, H. Chen, C. Wu, A. Avishai, Y. T. Shao, J. Ravichandran, Advanced Materials 2024, 2312620.
- [33]I. Sadeghi, K. Ye, M. Xu, Y. Li, J. M. LeBeau, R. Jaramillo, Advanced Functional Materials 2021, 31, 2105563.
- [34]M. Surendran, B. Zhao, G. Ren, S. Singh, A. Avishai, H. Chen, J.-K. Han, M. Kawasaki, R. Mishra, J. Ravichandran, Journal of Materials Research 2022, 37, 3481.
- [35]M. Surendran, H. Chen, B. Zhao, A. S. Thind, S. Singh, T. Orvis, H. Zhao, J.-K. Han, H. Htoon, M. Kawasaki, Chemistry of Materials 2021, 33, 7457.
- [36]W. van Mierlo, D. Geiger, A. Robins, M. Stumpf, M. L. Ray, P. Fischione, U. Kaiser, Ultramicroscopy 2014, 147, 149.
- [37]L. Pastewka, R. Salzer, A. Graff, F. Altmann, M. Moseler, Nuclear Instruments and Methods in Physics Research Section B: Beam Interactions with Materials and Atoms 2009, 267, 3072.
- [38]S. Pennycook, D. Jesson, Ultramicroscopy 1991, 37, 14.
- [39]J. A. Hachtel, J. C. Idrobo, M. Chi, Advanced structural and chemical imaging 2018, 4, 1.
- [40]J. Hwang, J. Y. Zhang, J. Son, S. Stemmer, Applied Physics Letters 2012, 100.
- [41]N. Shibata, S. D. Findlay, Y. Kohno, H. Sawada, Y. Kondo, Y. Ikuhara, Nature Physics 2012, 8, 611.
- [42]J. M. LeBeau, S. D. Findlay, L. J. Allen, S. Stemmer, Ultramicroscopy 2010, 110, 118.
- [43]H. Kim, M. Goyal, S. Salmani-Rezaie, T. Schumann, T. N. Pardue, J.-M. Zuo, S. Stemmer, Physical Review Materials 2019, 3, 084202.
- [44]B. Buxton, J. A. Eades, J. W. Steeds, G. Rackham, Philosophical Transactions of the Royal Society of London. Series A, Mathematical and Physical Sciences 1976, 281, 171.
- [45]R. F. Egerton, Electron energy-loss spectroscopy in the electron microscope, Springer Science & Business Media, 2011.
- [46]P. Chandra, P. B. Littlewood, in Physics of Ferroelectrics: A Modern Perspective, Springer 2007, p. 69.
- [47]P. Schattschneider, S. Rubino, C. Hébert, J. Rusz, J. Kuneš, P. Novák, E. Carlino, M. Fabrizioli, G. Panaccione, G. Rossi, Nature 2006, 441, 486.
- [48]J. Verbeeck, H. Tian, P. Schattschneider, Nature 2010, 467, 301.
- [49]L. J. Allen, S. Findlay, Ultramicroscopy 2015, 151, 11.
- [50]B. Forbes, A. Martin, S. D. Findlay, A. J. D’Alfonso, L. J. Allen, Physical Review B—Condensed Matter and Materials Physics 2010, 82, 104103.
- [51]S. Somnath, C. R. Smith, N. Laanait, R. K. Vasudevan, S. Jesse, Microscopy and Microanalysis 2019, 25, 220.
- [52]G. Kresse, J. Furthmüller, Physical review B 1996, 54, 11169.
- [53]J. P. Perdew, K. Burke, M. Ernzerhof, Physical review letters 1996, 77, 3865.
- [54]P. E. Blöchl, Physical review B 1994, 50, 17953.
- [55]G. Kresse, D. Joubert, Physical review b 1999, 59, 1758.
- [56]M. Gajdoš, K. Hummer, G. Kresse, J. Furthmüller, F. Bechstedt, Physical Review B—Condensed Matter and Materials Physics 2006, 73, 045112.
- [57]H. T. Stokes, S. v. Orden, B. J. Campbell, Journal of Applied Crystallography 2016, 49, 1849.
- [58]B. J. Campbell, H. T. Stokes, D. E. Tanner, D. M. Hatch, Journal of Applied Crystallography 2006, 39, 607.