Abstract
The possibility to use ferroelastic materials as components of neuromorphic devices is discussed. They can be used as local memristors with the advantage that ionic transport is constraint to twin boundaries where ionic diffusion is much faster than in the bulk and does not leak into adjacent domains. It is shown that nano-scale ferroelastic memristors can contain a multitude of domain walls. These domain walls interact by strain fields where the interactions near surfaces are fundamentally different from bulk materials. We show that surface relaxations (∼image forces) are curtailed to short range dipolar interactions which decay as 1/d2 where d is the distance between domain walls. In bigger samples such interactions are long ranging with 1/d. The cross-over regime is typically in the range of some 200–1500 nm using a simple spring interaction model.
Introduction
In contrast to ferroelectric domain walls, ferroelastic domain walls (mostly twin walls) have the advantage that their ‘emerging properties’ are localized within domain walls while the adjacent domains play no role. This remarkable property (Salje, 2010) stems from the local symmetry breaking inside the domain wall but not in the domains. Typical emerging properties are the development of polarity in non-polar materials (Van Aert et al., 2012), high conductivity (Aird and Salje, 1998), photovoltaics (Bhatnagar et al., 2013) and atomic-scale transistors (Chai et al., 2020). Several of these properties exist also in ferroelectric domain walls, both show enhanced conductivity compared to that of the domains (Seidel et al., 2009; Catalan et al., 2012; Meier and Selbach, 2022), and their density and topological complexity can be modulated by the choice of the substrate and the system dimensions (Venkatesan et al., 2007; Nesterov et al., 2013; Feigl et al., 2014), e.g., the film thickness. The non-trivial electronic and transport properties have been demonstrated to be suitable for new applications of domain wall nanoelectronics (Catalan et al., 2012; Meier and Selbach, 2022). Both ferroelectric and ferroelastic domain walls have been proposed as memristor materials (Maksymovych et al., 2011; Kim et al., 2012; Ma et al., 2020; Salje, 2021a; Liu et al., 2023), and as key elements of neuromorphic devices (Indiveri et al., 2013; Christensen et al., 2022). The topology of these wall-related networks was described in (Cipollini et al., 2024) for ferroelectric domain boundaries, although the ferroelasticity of domain walls in materials like BiFeO3 (Kubel and Schmid, 1990; Balke et al., 2009) was ignored in (Cipollini et al., 2024). It was speculated in (Bose et al., 2017) that wiring together memristor devices enables the realization of cross-bar arrays for vector-matrix multiplications. Self-assembled memristor networks of nano-objects, such as nanoparticle self-assembled networks (Bose et al., 2017; Mambretti et al., 2022) or nanowire networks (Cultrera et al., 2021; Hochstetter et al., 2021; Milano et al., 2022), were identified as potential candidates for neuromorphic devices (Lu and Salje, 2024). Ferroelastic twin walls relate to the interface between two domains and not necessarily to any an epitaxial strain imposed by the substrate. The movement of ferroelastic twin walls can be extremely swift and can be initiated by electric fields (Casals et al., 2019). The rearrangement of twin walls often occurs during avalanche processes (Harrison and Salje, 2010). The typical time scale for avalanche movements is some femtoseconds and even an extended kink inside a twin walls can assume a speed faster than the highest speed of sound in the material under consideration (Salje et al., 2017). This very high speed of the wall movement is very difficult to measure experimentally. Examples are avalanche process where events which last below a microsecond are commonly observed (Ding et al., 2012; Chen et al., 2022; Chen et al., 2023). The scale invariance of the process implies that the true local process is orders of magnitude faster but, alas, we are limited in their observation by our limited electronic and piezo-electric detector capabilities. Alternatively, local “swinging”’ of ferroelastic domain walls can be observed in resonance ultrasonic spectroscopy (RUS) and Resonant Piezoelectric Spectroscopy (RPS) (Carpenter, 2015) where the upper frequency is again some 10 MHz. It appears that a big technological advance would be needed to extend the frequency range of avalanches detection and RUS to the GHz regime and above. Such facilities would be most useful for the investigation of the memory capacity and the memristor functionality is a wide range of materials. The driving force for devices is often the external strain exerted from the counter electrode in a memristor (Salje, 2012). Under appropriate boundary conditions, twin boundaries form complex, self-assembled neuromorphic networks where the coupling between twin walls is a key element in the neuromorphic performance (Pastor-Satorras and Vespignani, 2001; Strogatz, 2001; Moreno et al., 2004; Ghavasieh and De Domenico, 2023). In this paper we describe how sample sizes, surfaces, and defects interplay to determine how twin walls interact and how they form the common ferroelastic tangles (Viehland and Salje, 2014; Salje et al., 2016a). Typical ferroelastic materials which may be explored further for neuromorphic computation include WO3 (Aird and Salje, 2000; Kim et al., 2010), CaTiO3 (Lee et al., 2006) and Pb3(PO4)2 (Salje et al., 1993; Wruck et al., 1994).
Theoretical model
Ferroelastic domains, domain walls and the topological atomic kinks inside the walls are described by a Landau-type double-well potential on the interatomic interactions, as schematically shown in Figure 1. Our simulations are based on a two-dimensional toy model with two base atoms carrying negative charges (red atoms) and positive charges (yellow atoms). The total interactions can be divided into two main parts, i. e., short range pairwise interactions and long range Columbic interactions. The potential forms for the short range interactions are summarized in Table 1.
FIGURE 1
TABLE 1
| Uanion-anion | First NN (black springs, spring1) | |
| Second NN (black stick, spring2) | ||
| Third NN (spring 3) | ||
| Fourth NN (spring 4) | ||
| Ucation-cation | First NN (green springs, spring 5) | |
| Second NN (spring 6) | ||
| Uanion-cation | First NN (red springs, spring 7) |
List of interaction potentials for the simulation of ferroelectric twin structures.
Where r is the distance between atoms. The first- and third-nearest interactions are related to the elastic interactions and constitute the elastic background in ferroelastic materials. The second-nearest interaction has two minima which are symmetric with respect to of the ‘cubic’ lattice parameter and hence pushes or pulls the atoms into the diagonal direction of the unit cell. This leads to a shear of the equilibrium structure. We choose the model parameter such that the shear angle is 2°. The additional fourth-nearest Landau-type interactions help to obtain a reasonable domain wall thickness and stability (Chrosch and Salje, 1999; Catalan et al., 2012). The model parameters were inspired by the well-known second-order phase transition of SrTiO3 with a typical ferroelastic shear angle of 2° (Hayward and Salje, 1999). The simulated microstructures are fairly robust with respect to the parameters in Table 1. The ratios of the prefectors of the various potential were optimized to reproduce a weakly first-order transition where the temperature evolution of the order parameter is smooth in the ferroelastic phase and shows a step at the transition temperature. The atomic interactions between cations, and cations and anions are purely harmonic to exclude any additional polar instability in the bulk. The polarity condenses only inside the twin walls where the inversion symmetry is broken by the change of the shear angle across the walls and the flexoelectric relaxation. Similar potential forms have been developed to successfully investigate internal frictions accompanying the dynamic motions of ferroelastic domains (Zhao et al., 2013; He et al., 2022), the interactions of fine microstructures inside the domain walls (Lu et al., 2022; Lu et al., 2023), piezoelectricity (Lu et al., 2019a; Lu et al., 2020a), ferroelectricity (Lu et al., 2019b), and magnetism (Lu et al., 2020b) emerging from the static and dynamic polar ferroelastic domain walls (often simple twin walls) (Salje et al., 2016b; Lu et al., 2019c). All simulations are performed using the LAMMPS code (Plimpton, 1995). Visualizations are performed using the OVITO software (Stukowski, 2010). A symmetry analysis of the model was published by Lu et al. (2024). The model is based on a “cubic” to “tetragonal” transition (in two dimensions). Extensions to other systems like “cubic” to “rhombohedral” have been envisaged but only very few papers have been published on this scenario (Yang et al.). Nevertheless, the phenomena which are described in this paper are indeed expected in rhombohedral systems like Pb3(PO4)2 (Bismayer and Salje, 1981; Salje and Wruck, 1983; Salje et al., 1993) and in LaAlO3. In LaAlO3 kinks and wall-wall interactions are predominant and even the extreme tweed configuration was experimentally observed (Salje et al., 2016a). No model simulations were performed so far in these systems, although no significant changes are needed for, e.g., LaAlO3, to describe the same effects as in this paper. The effect of magnetic boundaries is different because the typical wall thicknesses are much bigger and local links do not play any role.
Results
Wall-wall interactions
In most high-density domain patterns, the interaction between domain boundaries stems from junctions between intersecting boundaries (Salje and Ishibashi, 1996; Salje, 2012; Lu et al., 2019a). The resultant domain patterns are highly complex with the intersection of horizontal and vertical domain walls. Many studies focus on the question how this structural complexity influences the switching process (Casals et al., 2021). Here we discuss the case of much milder interactions when such junctions are absent. The relevant inter-wall interaction is first described in a defect free system when the total energy depends on the system size because the main interaction mechanism is the volume strain and not the local shear strain (Lu et al., 2019c). In the case of noninteracting walls the energy is a simple superposition of the wall and the bulk energy. Interactions can still exist in case of electronic interactions. However, such interactions are restricted to lengths scales of the wall thickness and are not relevant in this context (Lajzerowicz and Levanyuk, 1994; Nataf et al., 2020). In simulations using our potentials, the averaged bulk potential energy is −27.41 meV per atom. The averaged excess energy of two polar walls with respect to the bulk is 1.86 meV per atom, which corresponds to a wall energy of 1.19 × 10−9 mJ/m in our two-dimensional structure. Expanding into three dimensions, this value is close to wall energies estimated in three-dimensional structures, which is approximately 20 mJ/m2 (Salje et al., 2005; Barone et al., 2014; Viehland and Salje, 2014). The equivalent potential energy for the bulk and nonpolar walls (when the charges in then model are set to zero) is 10.95 meV and the excess energy of two nonpolar walls is 1.87 meV per atom. The effect of the two walls in this rather dense wall configuration is hence some 10%–20% of the total energy and is large enough to trigger pattern formation of the domain walls. In the nonpolar case, the resulting potential energy is independent of the interwall distance beside for very small distances when the local strain fields overlap (Figure 2A). This is different from the polar system where we find an energy minimum near 25 lattice units which corresponds to approximately 125 nm in a typical perovskite structure. For larger distances we find a weak attractive dipole-dipole interaction (Figure 2B).
FIGURE 2
The field dependence of the pair interaction is shown in Figure 3. These results show that purely elastic interactions lead to repulsion over very small distances while the additional dipole moments in domain walls lead to much longer-range interactions and a shallow energy minimum which is the origin of an “intrinsically” stable interwall distance. The dipoles can be swayed by electric fields perpendicular to the walls so that the wall-wall interaction can be modified by electric fields.
FIGURE 3
The wall-wall interactions are best seen in the strain fields of needle domains where the thickness of the needle is small. The strain fields are very complex near the needle tips, each bulging of the strain field in Figure 4 is generated by the kink in the needle domain. These additional strain fields give us the inspiration for an even stronger wall-wall interaction, namely, that generated by kinks in domain walls.
FIGURE 4
The combinations of elastic and columbic forces of polar domain walls were theoretically impact the formations of ferroelastic topological patterns, such as needle domains and comb domain structures when the interwall distances are sufficiently small. We now explore the origin of the equidistant configuration by keeping two outer domain walls at a constant distance while a third inner domain wall is allowed to move sideways. We choose d = 30 lattice units (Figure 5) for the two outer walls. We find an energy minimum for the third wall exactly in the middle. The energy difference between the middle position and the slightly shifted position (position 2 and 3) is very small. We find energy shifts in the order of magnitude of 0.001 meV/atom. This small energy explains why sideways movements are hard to observe experimentally when the walls are widely spaced: The energy gain is very small compared with common pinning energies of domain walls.
FIGURE 5
The role of topological kinks inside twin walls
It was first proposed in 2017 that wall bending may induce kinks in domain walls (Salje et al., 2017). These kinks residing inside domain walls and are not necessarily static but accelerate beyond the speed of sound under even modest external shear stress. The same kinks were subsequently found to be at the core of domain switching in metals (Yang et al., 2020; Yang et al., 2021) and during interactions with wall junctions which determine much of the mechanical properties of ferroelastic materials (He et al., 2019). Very high kink concentrations exist also in domain walls in uniaxial ferroelectrics like LiNbO3 (Gonnissen et al., 2016). The twin wall has often mirror symmetry (the so-called w walls, (Salje, 2012)). The mirror symmetry is broken by kinks and local stress fields and leads to significant strains emanating from the kinks. It is the purpose of this paper to characterize the strain fields and to show that strain-mediated interactions between kinks is a likely source for the interaction between parallel domain walls (Everhardt et al., 2019; Selke, 1988; Roitburd, 1976; Luk’yanchuk et al., 2009). A step forwards was achieved in 2000 when Pertsev et al. (Pertsev et al., 2000) calculated equilibrium shapes of curved ferroelastic domain walls in crystals. Smooth kinks in walls were investigated using their dislocation-disclination model on the basis that the domain walls are infinitely narrow. They found that elastic monopoles (the interaction energy decay as 1/d, where d is the kink-kink distance) exist in the bend regions in large samples.
Figure 6 shows the typical atomic kink structures residing inside the horizontal ferroelastic domain walls.
FIGURE 6
The kinks have two effects on the crystal structure. The first is that they generate large strain fields near the kink position. The second is that they curve the sample if this bending is not impeded by external forces, like a substrate underneath the sample. Both features give rise to interactions when two kinks are in either the same twin wall or in two parallel twin walls (Figure 7).
FIGURE 7
In our thin, freestanding sample (Figure 8) the interaction energy follows a typical dipole-dipole interaction for distances smaller than the equilibrium distance (Figure 9). The scaling exponent is −2 for this interaction energy E ∼ d-2. The size effect (surface strains) is very large in small systems, and strongly affects the kink-kink interaction energy while it is irrelevant for large system sizes. This scenario applies to small memristive elements where the fast diffusion is constrained to the twin walls. The crossover is at sizes equivalent to around 200 nm which shows that nanoparticles are prone to size effects while large single crystals are not. Similar interactions occur in parallel walls containing kink-antikink pairs.
FIGURE 8
FIGURE 9
All simulations were performed under open boundary conditions where the sample could change its shape. This approach makes the simulations much more realistic for most ferroelastic materials (Salje, 2012) and particularly for the large group of disordered materials (Salje and Dahmen, 2014) where porosity often plays a major role in the assessment of their elastic deformation (Casals and Salje, 2021). Most ferroelastic minerals, for example, display porosities between 10% and 60% (Salje, 2021b; Salje, 2022) which allows mineral grains to relax their shape when their microstructure changes as a function of temperature or pressure. The simulations highlight that shapes changes occur for small enough samples when kinks are generated in a twin wall. Depending on the kink or antikink configuration, the sample bends in one direction or the opposite. The bents are restricted to a small area near the (anti-)kink position while the rest of the sample simply tilts in two opposite directions. The bent region is ∼10 lattice units wide; the sample tilts continue to the sample surface. In typical crystal structures like perovskites with lattice units of ∼0.4 nm or feldspars with 1.3 nm in the monoclinic b direction these bent regions are some 4–13 nm wide and are hence observable under the transmission electron microscope. The tilt angles are in the order of 1.2° for single kinks and are hence observable.
In contrast to small, freestanding samples, thick samples were shown to have much stronger wall-wall interactions (Pertsev and Salje, 2000). These interactions are long ranging with decay scaling 1/d where d is that distance between kinks. They are very similar to dislocations so that many of the dislocation patterns are transferable to kinked ferroelastic twin walls–notwithstanding that the walls are essentially flat while the dislocations can loop. The size dependence of the wall-wall interaction was investigated by Lu et al. (2023). These authors showed that the fundamental features of the interaction is the same (namely, the lattice bending and the local distortion) but that the weak dipolar interaction in this samples continuously emerges into the much stronger mono-polar interaction (Figure 10).
FIGURE 10
The thin samples are hence dominated by surface relaxations, similar to the relaxations in dislocations, while such relaxations do not exist in thick samples. Kink-kink interactions in bulk samples interact as “monopoles” with a d-1 dependence when they are separated by the distance d. As the sample size decreases, the interaction for thin samples decays following a characteristic d-2 trend similar to that of dipoles. This behaviour of any singularity (dielectric, dislocations, interstitials, etc.) is commonly described analytically by the concept of “image forces”. The construction is based on the calculation of the surface relaxation as having the same energy as if a fictitious image force was placed outside the sample. Such image forces have also been used to describe the dynamics of dislocation movements (Gurrutxaga-Lerma et al., 2015). Our results clarify the role of a wide crossover regime near d = 1,000 l. u. The exact value of the crossover region depends on the model parameters but the order of magnitude is estimated to be between 200 nm and a few microns. This places the crossover behaviour in the typical region for nano-crystals and small grain devices in neuromorphic computer elements. The results show that the expected self-generated nano-structure differs greatly as a function of size: while larger grains show organized domain walls with the classic patterns of needle domains and junctions, this is not true for small grains where more complex structure with higher wall densities and curved walls are expected.
Discussion
Neuromorphic computing uses physical artificial neurons for computation (Burr et al., 2017; Marković et al., 2020). These neuromorphic elements often operate analogue rather than digital with memristors (Kumar et al., 2022), spintronic memories, threshold switches, transistors (Fuller et al., 2019; He et al., 2021), as attractive physical realisations. Domain boundary-based systems exist in all these options where the relevant elements (like ionic diffusion over interatomic distances, spin flips, charge induces switches and transistor junctions) can be constraint to domain walls. Following this route, we have then to consider how the domain walls form complex networks from ferroelastic stripe pattern to glassy structure (Salje and Carpenter, 2015), Skyrmions (Pantel et al., 2012) and tweed structures (Salje and Parlinski, 1991; Parlinski et al., 1993; Salje et al., 2016a). The fundamental idea is that neuromorphic engineering explores individual neurons, circuits, applications, and overall architectures that are desirable for computation but also the adaptation to fast change (plasticity). This plasticity may partly come from the artificial neuron but also from inter-neuron interactions. In addition, each neuron may have an internal adaptive structure generated by the domain boundaries (Viehland and Salje, 2014). In the approach of Seidel and Sharma (Sharma and Seidel, 2023), only two crossed domain walls are considered. If the domain wall density is much higher in tweed and glassy structures, we expect typical repetition units of the nano-structure of some few unit cells so that the total size of the artificial neuron remains below some 20 nm and still contains a fully adaptive and complex domain boundary array.
We have shown in this paper that one opportunity to ‘tailor’ responsive, plastic systems is to use adaptive domain boundary-based arrays which contain enough complexity to relax an-harmonically when stimulated. Such responses have been observed in test cases like SrTiO3 (Salje et al., 2013; Pesquera et al., 2018; Kustov et al., 2020), where the time constants of the glass relaxation stretch over several decades. The same effect is expected in WO3 where glassy behaviour was observed (Hutchins et al., 2007) and related to the existence of small polarons and low temperature relaxations (Schirmer and Salje, 1980). The smallness of the neuron makes it likely to see domain patterns related to our case of open boundary conditions, which is expected to contain much greater wall densities than under clamped boundary conditions. The exact solution will depend greatly on the interface between the neuron and the substrate which may reduce the relaxation and the image forces. As several different highly conducting interfaces are in use, this issue may need experimental clarification while the possible options for the wall configurations are described in this paper.
Statements
Data availability statement
The original contributions presented in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding authors.
Author contributions
GL: Conceptualization, Data curation, Formal Analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Software, Supervision, Validation, Visualization, Writing–original draft, Writing–review and editing. ES: Conceptualization, Data curation, Formal Analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Software, Supervision, Validation, Visualization, Writing–original draft, Writing–review and editing.
Funding
The author(s) declare that financial support was received for the research, authorship, and/or publication of this article. National Natural Science Foundation of China (Grant No. 12304130) and the Doctoral Starting Fund of Yantai University (Grant No. 1115- 2222006). EPSRC (EP/P024904/1) and the EU’s Horizon 2020 programme under the Marie Sklodowska-Curie Grant (861153).
Acknowledgments
GL is grateful for the financial support by the National Natural Science Foundation of China (Grant No. 12304130) and the Doctoral Starting Fund of Yantai University (Grant No. 1115-2222006). EKHS is grateful to EPSRC (EP/P024904/1) and the EU’s Horizon 2020 programme under the Marie Sklodowska-Curie Grant (861153).
Conflict of interest
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
Publisher’s note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
References
1
AirdA.SaljeE. K. H. (1998). Sheet superconductivity in twin walls: experimental evidence of WO3-x. J. Phys. Condens. Matter10, L377–L380. 10.1088/0953-8984/10/22/003
2
AirdA.SaljeE. K. H. (2000). Enhanced reactivity of domain walls in with sodium. Eur. Phys. J. B - Condens. Matter Complex Syst.15, 205–210. 10.1007/pl00011037
3
BalkeN.ChoudhuryS.JesseS.HuijbenM.ChuY. H.BaddorfA. P.et al (2009). Deterministic control of ferroelastic switching in multiferroic materials. Nat. Nanotechnol.4, 868–875. 10.1038/nnano.2009.293
4
BaroneP.Di SanteD.PicozziS. (2014). Improper origin of polar displacements at CaTiO3 and CaMnO3 twin walls. Phys. Rev. B89, 144104. 10.1103/physrevb.89.144104
5
BhatnagarA.Roy ChaudhuriA.Heon KimY.HesseD.AlexeM. (2013). Role of domain walls in the abnormal photovoltaic effect in BiFeO3. Nat. Commun.4, 2835. 10.1038/ncomms3835
6
BismayerU.SaljeE. (1981). Ferroelastic phases in Pb3(PO4)2-Pb3(AsO4)2; X-ray and optical experiments. Acta Crystallogr. Sect. A37, 145–153. 10.1107/s0567739481000417
7
BoseS. K.MallinsonJ. B.GazoniR. M.BrownS. A. (2017). Stable self-assembled atomic-switch networks for neuromorphic applications. IEEE Trans. Electron Devices64, 5194–5201. 10.1109/ted.2017.2766063
8
BurrG. W.ShelbyR. M.SebastianA.KimS.KimS.SidlerS.et al (2017). Neuromorphic computing using non-volatile memory. Adv. Phys. X2, 89–124. 10.1080/23746149.2016.1259585
9
CarpenterM. A. (2015). Static and dynamic strain coupling behaviour of ferroic and multiferroic perovskites from resonant ultrasound spectroscopy. J. Phys. Condens. Matter27, 263201. 10.1088/0953-8984/27/26/263201
10
CasalsB.NatafG. F.SaljeE. K. H. (2021). Avalanche criticality during ferroelectric/ferroelastic switching. Nat. Commun.12, 345. 10.1038/s41467-020-20477-6
11
CasalsB.SaljeE. K. H. (2021). Energy exponents of avalanches and Hausdorff dimensions of collapse patterns. Phys. Rev. E104, 054138. 10.1103/physreve.104.054138
12
CasalsB.van DijkenS.HerranzG.SaljeE. K. H. (2019). Electric-field-induced avalanches and glassiness of mobile ferroelastic twin domains in cryogenic SrTiO3. Phys. Rev. Res.1, 032025. 10.1103/physrevresearch.1.032025
13
CatalanG.SeidelJ.RameshR.ScottJ. F. (2012). Domain wall nanoelectronics. Rev. Mod. Phys.84, 119–156. 10.1103/revmodphys.84.119
14
ChaiX.JiangJ.ZhangQ.HouX.MengF.WangJ.et al (2020). Nonvolatile ferroelectric field-effect transistors. Nat. Commun.11, 2811. 10.1038/s41467-020-16623-9
15
ChenY.GouB.XuX.DingX.SunJ.SaljeE. K. H. (2023). Multibranches of acoustic emission as identifier for deformation mechanisms in additively manufactured 316L stainless steel. Addit. Manuf.78, 103819. 10.1016/j.addma.2023.103819
16
ChenY.GouB.YuanB.DingX.SunJ.SaljeE. K. H. (2022). Multiple avalanche processes in acoustic emission spectroscopy: multibranching of the Energy−Amplitude scaling. Phys. status solidi (b)259, 2100465. 10.1002/pssb.202270008
17
ChristensenD. V.DittmannR.Linares-BarrancoB.SebastianA.Le GalloM.RedaelliA.et al (2022). 2022 roadmap on neuromorphic computing and engineering. Neuromorphic Comput. Eng.2, 022501. 10.1088/2634-4386/ac4a83
18
ChroschJ.SaljeE. K. H. (1999). Temperature dependence of the domain wall width in LaAlO3. J. Appl. Phys.85, 722–727. 10.1063/1.369152
19
CipolliniD.SwierstraA.SchomakerL. (2024). Modeling a domain wall network in BiFeO3 with stochastic geometry and entropy-based similarity measure. Front. Mater.11, 1323153. 10.3389/fmats.2024.1323153
20
CultreraA.MilanoG.De LeoN.RicciardiC.BoarinoL.CallegaroL. (2021). Recommended implementation of electrical resistance tomography for conductivity mapping of metallic nanowire networks using voltage excitation. Sci. Rep.11, 13167. 10.1038/s41598-021-92208-w
21
DingX.ZhaoZ.LookmanT.SaxenaA.SaljeE. K. H. (2012). High junction and twin boundary densities in driven dynamical systems. Adv. Mater.24, 5385–5389. 10.1002/adma.201200986
22
EverhardtA. S.DamerioS.ZornJ. A.ZhouS.DomingoN.CatalanG.et al (2019). Periodicity-doubling cascades: direct observation in ferroelastic materials. Phys. Rev. Lett.123, 087603. 10.1103/physrevlett.123.087603
23
FeiglL.YudinP.StolichnovI.SlukaT.ShapovalovK.MtebwaM.et al (2014). Controlled stripes of ultrafine ferroelectric domains. Nat. Commun.5, 4677. 10.1038/ncomms5677
24
FullerE. J.LiY.BennetC.KeeneS. T.MelianasA.AgarwalS.et al (2019). Redox transistors for neuromorphic computing. IBM J. Res. Dev.63 (9), 1–9. 10.1147/jrd.2019.2942285
25
GhavasiehA.De DomenicoM. (2023). Generalized network density matrices for analysis of multiscale functional diversity. Phys. Rev. E107, 044304. 10.1103/physreve.107.044304
26
GonnissenJ.BatukD.NatafG. F.JonesL.AbakumovA. M.Van AertS.et al (2016). Direct observation of ferroelectric domain walls in LiNbO3: wall-meanders, kinks, and local electric charges. Adv. Funct. Mater.26, 7599–7604. 10.1002/adfm.201603489
27
Gurrutxaga-LermaB.BalintD. S.DiniD.SuttonA. P. (2015). Elastodynamic image forces on dislocations. Proc. R. Soc. A Math. Phys. Eng. Sci.471, 20150433. 10.1098/rspa.2015.0433
28
HarrisonR. J.SaljeE. K. H. (2010). The noise of the needle: avalanches of a single progressing needle domain in LaAlO3. Appl. Phys. Lett.97, 021907. 10.1063/1.3460170
29
HaywardS. A.SaljeE. K. H. (1999). Cubic-tetragonal phase transition in SrTiO3 revisited: Landau theory and transition mechanism. Phase Transitions68, 501–522. 10.1080/01411599908224530
30
HeX.LiS.DingX.SunJ.KustovS.SaljeE. K. H. (2022). Internal friction in complex ferroelastic twin patterns. Acta Mater.228, 117787. 10.1016/j.actamat.2022.117787
31
HeX.LiS.DingX.SunJ.SelbachS. M.SaljeE. K. H. (2019). The interaction between vacancies and twin walls, junctions, and kinks, and their mechanical properties in ferroelastic materials. Acta Mater.178, 26–35. 10.1016/j.actamat.2019.07.051
32
HeY.ZhuL.ZhuY.ChenC.JiangS.LiuR.et al (2021). Recent progress on emerging transistor-based neuromorphic devices. Adv. Intell. Syst.3, 2000210. 10.1002/aisy.202000210
33
HochstetterJ.ZhuR.LoefflerA.Diaz-AlvarezA.NakayamaT.KuncicZ. (2021). Avalanches and edge-of-chaos learning in neuromorphic nanowire networks. Nat. Commun.12, 4008. 10.1038/s41467-021-24260-z
34
HutchinsM. G.Abu-AlkhairO.El-NahassM. M.Abdel-HadyK. (2007). Electrical conductivity and dielectric relaxation in non-crystalline films of tungsten trioxide. J. Non-Crystalline Solids353, 4137–4142. 10.1016/j.jnoncrysol.2007.06.042
35
IndiveriG.Linares-BarrancoB.LegensteinR.DeligeorgisG.ProdromakisT. (2013). Integration of nanoscale memristor synapses in neuromorphic computing architectures. Nanotechnology24, 384010. 10.1088/0957-4484/24/38/384010
36
KimD. J.LuH.RyuS.BarkC. W.EomC. B.TsymbalE. Y.et al (2012). Ferroelectric tunnel memristor. Nano Lett.12, 5697–5702. 10.1021/nl302912t
37
KimY.AlexeM.SaljeE. K. H. (2010). Nanoscale properties of thin twin walls and surface layers in piezoelectric WO3-x. Appl. Phys. Lett.96, 032904. 10.1063/1.3292587
38
KubelF.SchmidH. (1990). Structure of a ferroelectric and ferroelastic monodomain crystal of the perovskite BiFeO3. Acta Crystallogr. Sect. B46, 698–702. 10.1107/s0108768190006887
39
KumarS.WangX.StrachanJ. P.YangY.LuW. D. (2022). Dynamical memristors for higher-complexity neuromorphic computing. Nat. Rev. Mater.7, 575–591. 10.1038/s41578-022-00434-z
40
KustovS.LiubimovaI.SaljeE. K. H. (2020). Domain dynamics in quantum-paraelectric SrTiO3. Phys. Rev. Lett.124, 016801. 10.1103/physrevlett.124.016801
41
LajzerowiczJ.LevanyukA. P. (1994). Fluctuation-induced interaction of domain walls: influence on the commensurate-incommensurate transition. Phys. Rev. B49, 15475–15484. 10.1103/physrevb.49.15475
42
LeeW. T.SaljeE. K. H.Goncalves-FerreiraL.DaraktchievM.BismayerU. (2006). Intrinsic activation energy for twin-wall motion in the ferroelastic perovskite CaTiO3. Phys. Rev. B73, 214110. 10.1103/physrevb.73.214110
43
LiuZ.WangH.LiM.TaoL.PaudelT. R.YuH.et al (2023). In-plane charged domain walls with memristive behaviour in a ferroelectric film. Nature613, 656–661. 10.1038/s41586-022-05503-5
44
LuG.CorderoF.HideoK.DingX.XuZ.ChuR.et al (2024). Elastic precursor softening in proper ferroelastic materials: a molecular dynamics study. Phys. Rev. Res.6, 013232. 10.1103/physrevresearch.6.013232
45
LuG.DingX.SunJ.SaljeE. K. H. (2022). Wall-wall and kink-kink interactions in ferroelastic materials. Phys. Rev. B106, 144105. 10.1103/physrevb.106.144105
46
LuG.HideoK.DingX.XuZ.ChuR.NatafG. F.et al (2023). Influence of kinks on the interaction energy between ferroelastic domain walls in membranes and thin films. Microstructures3, 2023033. 10.20517/microstructures.2023.28
47
LuG.LiS.DingX.SaljeE. K. H. (2019a). Piezoelectricity and electrostriction in ferroelastic materials with polar twin boundaries and domain junctions. Appl. Phys. Lett.114, 202901. 10.1063/1.5092523
48
LuG.LiS.DingX.SunJ.SaljeE. K. H. (2019b). Ferroelectric switching in ferroelastic materials with rough surfaces. Sci. Rep.9, 15834. 10.1038/s41598-019-52240-3
49
LuG.LiS.DingX.SunJ.SaljeE. K. H. (2019c). Electrically driven ferroelastic domain walls, domain wall interactions, and moving needle domains. Phys. Rev. Mater.3, 114405. 10.1103/physrevmaterials.3.114405
50
LuG.LiS.DingX.SunJ.SaljeE. K. H. (2020a). Enhanced piezoelectricity in twinned ferroelastics with nanocavities. Phys. Rev. Mater.4, 074410. 10.1103/physrevmaterials.4.074410
51
LuG.LiS.DingX.SunJ.SaljeE. K. H. (2020b). Current vortices and magnetic fields driven by moving polar twin boundaries in ferroelastic materials. npj Comput. Mater.6, 145. 10.1038/s41524-020-00412-5
52
LuG.SaljeE. K. H. (2024). Multiferroic neuromorphic computation devices. Apl. Mater.12, 061101. 10.1063/5.0216849
53
Luk’yanchukI. A.SchillingA.GreggJ. M.CatalanG.ScottJ. F. (2009). Origin of ferroelastic domains in free-standing single-crystal ferroelectric films. Phys. Rev. B79, 144111. 10.1103/physrevb.79.144111
54
MaC.LuoZ.HuangW.ZhaoL.ChenQ.LinY.et al (2020). Sub-nanosecond memristor based on ferroelectric tunnel junction. Nat. Commun.11, 1439. 10.1038/s41467-020-15249-1
55
MaksymovychP.SeidelJ.ChuY. H.WuP.BaddorfA. P.ChenL.-Q.et al (2011). Dynamic conductivity of ferroelectric domain walls in BiFeO3. Nano Lett.11, 1906–1912. 10.1021/nl104363x
56
MambrettiF.MiriglianoM.TentoriE.PedraniN.MartiniG.MilaniP.et al (2022). Dynamical stochastic simulation of complex electrical behavior in neuromorphic networks of metallic nanojunctions. Sci. Rep.12, 12234. 10.1038/s41598-022-15996-9
57
MarkovićD.MizrahiA.QuerliozD.GrollierJ. (2020). Physics for neuromorphic computing. Nat. Rev. Phys.2, 499–510. 10.1038/s42254-020-0208-2
58
MeierD.SelbachS. M. (2022). Ferroelectric domain walls for nanotechnology. Nat. Rev. Mater.7, 157–173. 10.1038/s41578-021-00375-z
59
MilanoG.PedrettiG.MontanoK.RicciS.HashemkhaniS.BoarinoL.et al (2022). In materia reservoir computing with a fully memristive architecture based on self-organizing nanowire networks. Nat. Mater.21, 195–202. 10.1038/s41563-021-01099-9
60
MorenoY.NekoveeM.PachecoA. F. (2004). Dynamics of rumor spreading in complex networks. Phys. Rev. E69, 066130. 10.1103/physreve.69.066130
61
NatafG. F.GuennouM.GreggJ. M.MeierD.HlinkaJ.SaljeE. K. H.et al (2020). Domain-wall engineering and topological defects in ferroelectric and ferroelastic materials. Nat. Rev. Phys.2, 634–648. 10.1038/s42254-020-0235-z
62
NesterovO.MatzenS.MagenC.VlooswijkA. H. G.CatalanG.NohedaB. (2013). Thickness scaling of ferroelastic domains in PbTiO3 films on DyScO3. Appl. Phys. Lett.103, 142901. 10.1063/1.4823536
63
PantelD.GoetzeS.HesseD.AlexeM. (2012). Reversible electrical switching of spin polarization in multiferroic tunnel junctions. Nat. Mater.11, 289–293. 10.1038/nmat3254
64
ParlinskiK.HeineV.SaljeE. K. H. (1993). Origin of tweed texture in the simulation of a cuprate superconductor. J. Phys. Condens. Matter5, 497–518. 10.1088/0953-8984/5/4/018
65
Pastor-SatorrasR.VespignaniA. (2001). Epidemic spreading in scale-free networks. Phys. Rev. Lett.86, 3200–3203. 10.1103/physrevlett.86.3200
66
PertsevN. A.NovakJ.SaljeE. K. H. (2000). Long-range elastic interactions and equilibrium shapes of curved ferroelastic domain walls in crystals. Philos. Mag. A80, 2201–2213. 10.1080/01418610008212157
67
PertsevN. A.SaljeE. K. H. (2000). Thermodynamics of pseudoproper and improper ferroelastic inclusions and polycrystals: effect of elastic clamping on phase transitions. Phys. Rev. B61, 902–908. 10.1103/physrevb.61.902
68
PesqueraD.CarpenterM. A.SaljeE. K. H. (2018). Glasslike dynamics of polar domain walls in cryogenic SrTiO3. Phys. Rev. Lett.121, 235701. 10.1103/physrevlett.121.235701
69
PlimptonS. (1995). Fast parallel algorithms for short-range molecular dynamics. J. Comput. Phys.117, 1–19. 10.1006/jcph.1995.1039
70
RoitburdA. L. (1976). Equilibrium structure of epitaxial layers. Phys. status solidi (a)37, 329–339. 10.1002/pssa.2210370141
71
SaljeE.ParlinskiK. (1991). Microstructures in high Tc superconductors. Supercond. Sci. Technol.4, 93–97. 10.1088/0953-2048/4/3/002
72
SaljeE.WruckB. (1983). Specific-heat measurements and critical exponents of the ferroelastic phase transition in Pb3(PO4)2 and Pb3(P1-xAsxO4)2. Phys. Rev. B28, 6510. 10.1103/PhysRevB.28.6510
73
SaljeE. K. H. (2010). Multiferroic domain boundaries as active memory devices: trajectories towards domain boundary engineering. ChemPhysChem11, 940–950. 10.1002/cphc.200900943
74
SaljeE. K. H. (2012). Ferroelastic materials. Annu. Rev. Mater. Res.42, 265–283. 10.1146/annurev-matsci-070511-155022
75
SaljeE. K. H. (2021a). Mild and wild ferroelectrics and their potential role in neuromorphic computation. Apl. Mater.9, 010903. 10.1063/5.0035250
76
SaljeE. K. H. (2021b). Ferroelastic twinning in minerals: a source of trace elements, conductivity, and unexpected piezoelectricity. Minerals11, 478. 10.3390/min11050478
77
SaljeE. K. H. (2022). Porosity in minerals. AIMS Mater. Sci.9, 1–8. 10.3934/matersci.2022001
78
SaljeE. K. H.AktasO.CarpenterM. A.LagutaV. V.ScottJ. F. (2013). Domains within domains and walls within walls: evidence for polar domains in cryogenic SrTiO3. Phys. Rev. Lett.111, 247603. 10.1103/physrevlett.111.247603
79
SaljeE. K. H.AlexeM.KustovS.WeberM. C.SchiemerJ.NatafG. F.et al (2016a). Direct observation of polar tweed in LaAlO3. Sci. Rep.6, 27193. 10.1038/srep27193
80
SaljeE. K. H.CarpenterM. A. (2015). Domain glasses: twin planes, Bloch lines, and Bloch points. Phys. status solidi (b)252, 2639–2648. 10.1002/pssb.201552430
81
SaljeE. K. H.DahmenK. A. (2014). Crackling noise in disordered materials. Annu. Rev. Condens. Matter Phys.5, 233–254. 10.1146/annurev-conmatphys-031113-133838
82
SaljeE. K. H.Graeme-BarberA.CarpenterM. A.BismayerU. (1993). Lattice parameters, spontaneous strain and phase transitions in Pb3(PO4)2. Acta Crystallogr. Sect. B49, 387–392. 10.1107/s0108768192008127
83
SaljeE. K. H.HaywardS. A.LeeW. T. (2005). Ferroelastic phase transitions: structure and microstructure. Acta Crystallogr. Sect. A61, 3–18. 10.1107/s0108767304020318
84
SaljeE. K. H.IshibashiY. (1996). Mesoscopic structures in ferroelastic crystals: needle twins and right-angled domains. J. Phys. Condens. Matter8, 8477–8495. 10.1088/0953-8984/8/44/004
85
SaljeE. K. H.LiS.StengelM.GumbschP.DingX. (2016b). Flexoelectricity and the polarity of complex ferroelastic twin patterns. Phys. Rev. B94, 024114. 10.1103/physrevb.94.024114
86
SaljeE. K. H.WangX.DingX.ScottJ. F. (2017). Ultrafast switching in avalanche-driven ferroelectrics by supersonic kink movements. Adv. Funct. Mater.27, 1700367. 10.1002/adfm.201700367
87
SchirmerO. F.SaljeE. (1980). Conduction bipolarons in low-temperature crystalline WO3-x. J. Phys. C Solid State Phys.13, L1067–L1072. 10.1088/0022-3719/13/36/005
88
SeidelJ.MartinL. W.HeQ.ZhanQ.ChuY. H.RotherA.et al (2009). Conduction at domain walls in oxide multiferroics. Nat. Mater.8, 229–234. 10.1038/nmat2373
89
SelkeW. (1988). The ANNNI model — theoretical analysis and experimental application. Phys. Rep.170, 213–264. 10.1016/0370-1573(88)90140-8
90
SharmaP.SeidelJ. (2023). Neuromorphic functionality of ferroelectric domain walls. Neuromorphic Comput. Eng.3, 022001. 10.1088/2634-4386/accfbb
91
StrogatzS. H. (2001). Exploring complex networks. Nature410, 268–276. 10.1038/35065725
92
StukowskiA. (2010). Visualization and analysis of atomistic simulation data with OVITO–the Open Visualization Tool. Model. Simul. Mater. Sci. Eng.18, 015012. 10.1088/0965-0393/18/1/015012
93
Van AertS.TurnerS.DelvilleR.SchryversD.Van TendelooG.SaljeE. K. H. (2012). Direct observation of ferrielectricity at ferroelastic domain boundaries in CaTiO3 by electron microscopy. Adv. Mater.24, 523–527. 10.1002/adma.201103717
94
VenkatesanS.KooiB. J.De HossonJ. T. M.VlooswijkA. H. G.NohedaB. (2007). Substrate influence on the shape of domains in epitaxial PbTiO3 thin films. J. Appl. Phys.102, 104105. 10.1063/1.2815657
95
ViehlandD. D.SaljeE. K. H. (2014). Domain boundary-dominated systems: adaptive structures and functional twin boundaries. Adv. Phys.63, 267–326. 10.1080/00018732.2014.974304
96
WruckB.SaljeE. K. H.ZhangM.AbrahamT.BismayerU. (1994). On the thickness of ferroelastic twin walls in lead phosphate Pb3(PO4)2 an X-ray diffraction study. Phase Transitions48, 135–148. 10.1080/01411599408200357
97
YangY.LiS.DingX.SunJ.WeissJ.SaljeE. K. H. (2020). Twisting of pre-twinned α-Fe nanowires: from mild to wild avalanche dynamics. Acta Mater.195, 50–58. 10.1016/j.actamat.2020.04.023
98
YangY.LinY.DingX.SunJ.ButterfieldN. J.AufortJ.et alThe twin boundary in calcite, its structure and physical properties. Submitted to PRB.
99
YangY.ZhangL.LiS.DingX.SunJ.WeissJ.et al (2021). Mild fluctuations in ferroelastic domain switching. Phys. Rev. B104, 214103. 10.1103/physrevb.104.214103
100
ZhaoZ.DingX.LookmanT.SunJ.SaljeE. K. H. (2013). Mechanical loss in multiferroic materials at high frequencies: friction and the evolution of ferroelastic microstructures. Adv. Mater.25, 3244–3248. 10.1002/adma.201300655
Summary
Keywords
ferroelastic materials, twin walls, memristors, wall-wall interactions, surface relaxations, domain walls
Citation
Lu G and Salje EKH (2024) Ferroelastic twin walls for neuromorphic device applications. Front. Mater. 11:1406853. doi: 10.3389/fmats.2024.1406853
Received
25 March 2024
Accepted
01 August 2024
Published
14 August 2024
Volume
11 - 2024
Edited by
Steven Koester, University of Minnesota Twin Cities, United States
Reviewed by
Tamalika Banerjee, University of Groningen, Netherlands
Pavan Nukala, Indian Institute of Science (IISc), India
Updates
Copyright
© 2024 Lu and Salje.
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
*Correspondence: Guangming Lu, luguangming1990@ytu.edu.cn; Ekhard K. H. Salje, ekhard@esc.cam.ac.uk
Disclaimer
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.