---
title: 'Proximity Ferroelectricity: Mechanisms & Applications'
url: https://www.emergentmind.com/topics/proximity-ferroelectricity
type: topic
---

# Proximity Ferroelectricity: Mechanisms & Applications

Proximity ferroelectricity denotes a family of phenomena in which ferroelectric order, ferroelectric switchability, or ferroelectric control emerges because a material is placed near a polar instability or near a ferroelectric, multiferroic, metallic, or electrochemically active environment. In the current literature, the term does not have a single usage. It can describe an incipient ferroelectric crystal poised near a soft polar mode, a non-ferroelectric electronic system whose band structure is controlled by an adjacent ferroelectric, an otherwise unswitchable polar material rendered switchable by a thin ferroelectric layer, or a nanoscale film whose polar phase is stabilized by electrodes, surfaces, or screening charges. Across these uses, the unifying theme is nonlocal electrostatic, lattice-dynamical, or interfacial coupling rather than an isolated bulk instability of the proximitized subsystem alone [1001.5458] [2501.05219] [2403.17326].

## 1. Terminology and scope

Across the cited literature, “proximity ferroelectricity” is used in several distinct but related senses. In bulk lattice-dynamical work, it refers to a crystal that is not ferroelectric at equilibrium but lies close to a ferroelectric instability. In heterostructure work, it often refers to ferroelectric control of an adjacent material without chemical modification. In wurtzite multilayers, it denotes collective switching of a polar but otherwise unswitchable layer when it is directly contacted by a thin switchable ferroelectric. In ultrathin capacitors and surface-controlled oxides, it denotes stabilization or transformation of polar order by boundary conditions, screening, or electrochemical state [1001.5458] [1909.02844] [2501.05219] [2402.14176].

| Usage in the literature | Representative systems | Operative mechanism |
|---|---|---|
| Incipient or near-soft-mode ferroelectricity | TlCl, TlBr, TlI; litharge-type layered semiconductors | Soft polar TO mode, large Born effective charges, giant lattice dielectric response |
| Interfacial electronic proximity | graphene/BiFeO\(_3\); \(\alpha\)-Bi/SnSe; graphene/SnTe; moiré vdW devices | Bound-charge electrostatics, symmetry breaking, hybridization, Rashba SOC, exchange coupling |
| Collective switching in multilayers | AlScN/AlN, ZnMgO/ZnO, graded nitrides and oxides | Self-consistent depolarization fields, field splitting, domain nucleation and propagation |
| Boundary-condition-driven polar stabilization | Pt/PbZrO\(_3\)/Pt, HfO\(_2\)-based films, granular metal/FE substrates | Electrode screening, finite surface/interface DOS, image-charge control of electrostatics |

A recurrent misconception is that proximity ferroelectricity always means that the proximitized material itself acquires a bulk ferroelectric ground state. The literature explicitly distinguishes several other cases. In graphene/SnTe, what is transferred is spin–orbit coupling controlled by the ferroelectric polarization of SnTe, not spontaneous ferroelectric polarization in graphene. In graphene/BiFeO\(_3\), the key observables are polarization-controlled doping, sublattice asymmetry, exchange splitting, and Rashba SOC. In contrast, ultrathin PbZrO\(_3\) and expanded Tl halides are discussed in terms of genuinely stabilized polar states [2402.09045] [1909.02844] [2402.14176].

## 2. Incipient ferroelectricity: soft modes, anomalous charges, and strain

In the incipient sense, proximity ferroelectricity describes a material that sits on the verge of a ferroelectric transition because a polar transverse-optical mode is strongly softened but not yet unstable. The central lattice-dynamical quantity is the Born effective charge tensor,
$$
Z^{*}_{\kappa,\alpha\beta} \;=\; \Omega \;\left.\frac{\partial P_{\alpha}}{\partial u_{\kappa\beta}}\right|_{\mathcal{E}=0},
$$
which measures dynamic charge transfer during ionic displacement. Large \(Z^*\) amplifies long-range Coulomb forces and produces large longitudinal–transverse optical phonon splitting. The static dielectric response is then linked to the optical phonons through the Lyddane–Sachs–Teller relation,
$$
\frac{\varepsilon_{0}}{\varepsilon_{\infty}} \;=\; \prod_{j}\left(\frac{\omega_{\mathrm{LO},j}}{\omega_{\mathrm{TO},j}}\right)^2.
$$
This framework is explicit in both thallium halides and the litharge-type layered semiconductors [1001.5458] [2412.11577].

The thallium halides TlCl, TlBr, and TlI are the canonical example of an “apparently ionic” system that nonetheless lies close to ferroelectricity. In the simple cubic CsCl structure \(Pm\bar{3}m\), with Tl at \((0,0,0)\) and the halogen at \((1/2,1/2,1/2)\), the electronegativity differences are large, yet density-functional perturbation theory gives anomalous Born charges well above the nominal \(\pm 1\): \(Z^*(\mathrm{Tl})=+2.02\), \(+2.10\), and \(+2.21\) for TlCl, TlBr, and TlI, with equal and opposite values on the halogens. The microscopic driver is cross-band-gap hybridization between Tl \(6p\) and halogen \(p\) states. These enhanced dynamic charges generate giant LO–TO splitting, exceptionally large lattice dielectric response, and a near-soft polar mode at \(\Gamma\). For TlBr at 290 K, \(\varepsilon_0 \approx 30.6\) and \(\varepsilon_\infty \approx 5.34\); for TlCl at 293 K, \(\varepsilon_0 \approx 32.7\) and \(\varepsilon_\infty \approx 4.76\). A modest lattice expansion drives the system across the instability: at the PBE level, more than about \(2\%\) expansion produces a lower-energy tetragonal state with a [001] polar displacement, while in LDA the onset occurs at \(\sim 1\%\) expansion. The published optic-phonon table likely has LO/TO labels swapped, but the physically important point is the very large LO–TO gap and soft TO mode [1001.5458].

The litharge-type \([\mathrm{Pb}_2\mathrm{F}_2]\)- and \([\mathrm{Bi}_2\mathrm{O}_2]\)-based layered semiconductors extend the same incipient-ferroelectric logic to a broad mixed-anion family. Here the universal microscopic origin is strong cross-bandgap hybridization of cation \(p\) orbitals enabled by \(6s^2\) lone-pair electrons and the litharge-type structural unit. The resulting polarizability is unusually large for semiconductors: the in-plane static dielectric constants exceed 50 and reach 292.7 in Bi\(_5\)O\(_4\)S\(_3\)Cl, while BiOI and BiOCl have \(\varepsilon_0=119.2\) and \(59.8\), respectively. In BiOCl, the lowest polar TO mode has \(\omega_{\mathrm{TO}}\approx 10\) meV and mode effective charge \(Z_m\approx 0.85\), contributing more than \(88\%\) of the in-plane lattice dielectric response. The in-plane Born effective charges are anomalous, for example Bi \(+5.34\), O \(-3.23\), and Cl \(-2.11\) in BiOCl; Pb and Bi reach \(+3.97\) in PbFBr and \(+5.92\) in BiOI. A few percent tensile strain then condenses the soft mode: the critical biaxial strain lies in the \(0.5\)–\(3.5\%\) range across the family, while archetypal \(\alpha\)-PbO requires \(\approx 13\%\) biaxial strain. Beyond the critical strain, the systems develop classical double wells and spontaneous polarizations of \(15\)–\(35\ \mu\mathrm{C/cm}^2\) at \(2\%\) strain beyond threshold [2412.11577].

These cases establish proximity ferroelectricity, in the incipient sense, as a problem of anomalous lattice polarization rather than only of chemical ionicity or structural complexity. A plausible implication is that the decisive descriptor is not whether a material is nominally ionic or covalent, but whether its occupied and low-lying empty states permit large dynamical charge transfer during polar displacements.

## 3. Interfacial proximity in electronic, spintronic, and moiré systems

In heterostructures, proximity ferroelectricity often means that a ferroelectric layer acts on a neighboring conductor or semimetal through electrostatics, symmetry breaking, and hybridization, while the proximitized layer itself remains non-ferroelectric. The channel is then a detector of polarization rather than the seat of the order parameter. This is explicit in graphene/BiFeO\(_3\), where ferroelectric polarization in the multiferroic substrate controls graphene’s doping, sublattice mass term, exchange splitting, and Rashba SOC. For Fe-terminated BFO(111), the head and tail interfaces carry markedly different bound charges, \(\sigma_{\mathrm{head}}\approx 151\ \mu\mathrm{C/cm}^2\) and \(\sigma_{\mathrm{tail}}\approx 25\ \mu\mathrm{C/cm}^2\), leading to adsorption distances of \(2.35\) and \(2.7\) Å. Both graphene layers are \(n\)-doped, but the Dirac-point shifts differ strongly: \(E_D=-0.85\) eV for GrP.head and \(-0.47\) eV for GrP.tail. The spin-resolved gaps at \(K\) are \(55\) and \(26\) meV at the head interface and \(6\) and \(1.5\) meV at the tail interface; the exchange spin splittings are \(104\) and \(75\) meV for electrons and holes at the head interface, and \(-35\) and \(-40\) meV at the tail interface. A KWANT transport analysis then defines proximity electro-resistance, proximity magneto-resistance, and proximity multiferroic electro-magneto-resistance, with room-temperature values spanning \(-44\%\) to \(+33\%\), \(-22\%\) to \(+48\%\), and \(+7\%\) to \(+13\%\), respectively [1909.02844].

A related but more symmetry-driven realization occurs in \(\alpha\)-Bi/SnSe, where the in-plane ferroelectric polarization of monolayer SnSe couples to the spin–orbit-entangled \(p_z\) states of \(\alpha\)-Bi. The proximity terms are written as
$$
H_{\mathrm{prox}} = \Delta\,\sigma_z s_0 + \lambda_R\,\sigma_y s_y,
$$
with \(\Delta\) tracking the ferroelectric state. First-principles calculations and ARPES show that the induced symmetry breaking drives an insulator-to-semimetal transition and yields two-dimensional pseudo-Weyl fermions along \(\Gamma\)–Y. The heterostructure has spontaneous polarization \(P_s=34.48\ \mu\mathrm{C/cm}^2\) and switching barrier \(E_B=28.9\) meV per cell. Reversing \(P\) flips the spin polarization at fixed \(k\), changes the sign of the Berry-curvature dipole \(D_y\), and reverses both the nonlinear anomalous Hall response and the sign of \(\sigma_{xy}^z\). Wannierized calculations place \(D_y\) at approximately \(1\)–\(4\) nm near the Fermi level, while the spin–charge interconversion efficiencies are of order \(10^3\times(h/e^2)\) [2311.18026].

In twisted graphene/SnTe, the same broad idea appears in a more restricted sense: the ferroelectric polarization of SnTe transfers SOC to graphene, rather than transferring ferroelectric order itself. The \(3^\circ\)-twisted graphene/SnTe bilayer has no common point symmetry, so the effective graphene Hamiltonian requires symmetry-free linear-in-\(k\) SOC terms. DFT and model fitting show spin splittings greater than \(20\) meV near the Dirac point, with strong momentum asymmetry due to hybridization of the right-moving branch with a SnTe band. The dominant Rashba contribution is in-plane, with generalized norms \(|\beta_1|=47.61\) meV, \(|\beta_2|=368.73\) meV, and \(|\lambda|=155.60\) meV, and the effective in-plane Rashba field is inferred to lie at an angle of approximately \(7.35^\circ\), nearly aligned with the in-plane ferroelectric polarization of SnTe. Reversal of \(P\) is therefore expected to reverse the effective Rashba field and the spin texture [2402.09045].

A separate line of work uses “proximity ferroelectricity” to describe conductive detector layers coupled to an interfacial sliding ferroelectric in moiré van der Waals devices. In BN/WSe\(_2\)/monolayer-graphene/WSe\(_2\)/BN heterostructures, the large \(\sim 30\%\) lattice mismatch suppresses graphene moiré bands, yet transport hysteresis survives. Hysteresis appears only under top-gate sweeps, not under bottom-gate sweeps, and Landau-level spectroscopy at \(B=9\) T shows nearly vertical charge-neutrality ridges, indicating anomalous screening by a conductive moiré ferroelectric with pinned domain walls. The switching thresholds inferred from Hall sweeps lie around \(V_{b,\max}\approx 4.5\)–\(5\) V. The same behavior is reproduced by multilayer twisted MoS\(_2\), while a reference monolayer graphene on the same BN flake without MoS\(_2\) shows no hysteresis. In this interpretation, the proximitized graphene is again not the ferroelectric medium; it is capacitively gated by a conductive sliding ferroelectric whose polarization is set by domain-wall motion in BN or twisted MoS\(_2\) [2403.17326].

Taken together, these interfacial systems show that proximity ferroelectricity can mean ferroelectric control of electronic structure, exchange, Berry curvature, or spin relaxation without requiring a ferroelectric ground state in the channel itself. This suggests that the term is often most precise when read as a statement about where the order parameter resides and where its measurable consequences appear.

## 4. Boundary conditions, screening, and electrochemical stabilization

Another major usage concerns systems in which proximity to electrodes, surfaces, or a ferroelectric substrate changes the thermodynamic landscape itself. In Pt/PbZrO\(_3\)/Pt nanocapacitors, first-principles calculations show that ultrathin PbZrO\(_3\), though antiferroelectric in bulk, becomes polar in the extreme thickness limit because well-compensated Pt interfaces stabilize R3c-derived polar phases relative to the Pbam-derived antipolar phase. The famous bulk triple-well potential is effectively inverted. In 3-, 5-, and 7-layer capacitors, polar states such as Pmc2\(_1\) and Pc lie below the antipolar competitors. For example, in the 3-layer nanocapacitor the lowest state is polar Pmc2\(_1\) at \(-133.8\) meV per Pb, while the Pbam-derived state lies at \(+15.9\) meV per Pb; in the 7-layer case, Pmc2\(_1\) reaches \(-242.9\) meV per Pb whereas Pbam is at \(-160.8\) meV per Pb. The residual depolarizing field falls from \(4.08\pm0.17\) to \(0.19\pm0.03\) GV/m in the 5-layer system and from \(2.81\pm0.08\) to \(0.38\pm0.05\) GV/m in the 7-layer system when Pt screening and interfacial relaxation are included. The layer-resolved LDOS remains insulating, ruling out pathological metallic spill-out [2402.14176].

In hafnia-based films, the controlling boundary condition is not only metallic screening but also finite surface and interface densities of electronic and ionic states. The proposed antiferroionic model couples the bulk competition between ferroelectric and structural antipolar instabilities to potential-dependent screening charges with distinct kinetics. Environmental BE-PFM on 20 nm Hf\(_{0.5}\)Zr\(_{0.5}\)O\(_2\) shows that ambient conditions give pinched, AFE-like loops; argon yields narrower and nearly paraelectric-like off-field behavior; UHV at \(2\times10^{-10}\) Torr produces wider FE-like loops and clear remanent domains at 300 K. The same framework explains temperature-driven changes: at 108 K the loops are again AFE-like, at 360 K pinching disappears, and at 420 K pinching reappears. The phase diagram is thus organized by oxygen partial pressure, temperature, and the kinetics of surface compensation, rather than by bulk order alone [2207.12525].

Granular metals on ferroelectric substrates provide a more electrostatic realization. A quasi-2D nanogranular metallic film above a ferroelectric substrate experiences image-charge screening that changes the Coulomb blockade energy of each grain. For the layered geometry considered,
$$
E_c = E_c^0 \frac{1}{\varepsilon}\left[1 + \frac{a}{2h}\frac{\varepsilon-\varepsilon_{\mathrm{FE}}}{\varepsilon+\varepsilon_{\mathrm{FE}}}\right],
$$
so the field-dependent ferroelectric permittivity renormalizes the charging energy. In the sequential-tunneling regime, the conductivity becomes
$$
\sigma(T,\varepsilon_{\mathrm{FE}})=\sigma_0 \exp[-E_c(\varepsilon_{\mathrm{FE}})/T].
$$
Because the transport depends exponentially on \(E_c\), hysteretic changes in \(\varepsilon_{\mathrm{FE}}\) produce giant electro-resistance. For representative parameters, conductivity ratios approaching \(\sim 10\) are feasible at room temperature, and the effect can be used for nonvolatile memory, electric-field sensing, and temperature sensing near \(T_{C,\mathrm{FE}}\) [1407.7719].

These examples show that proximity ferroelectricity can be created or destroyed by compensation quality. In one case, good metallic screening stabilizes a polar phase that is not the bulk ground state; in another, surface electrochemistry moves the system between FE-like and AFE-like responses; in a third, ferroelectric screening acts on a nearby electronic system without current passing through the ferroelectric.

## 5. Collective switching in wurtzite multilayers and graded structures

The most explicit modern use of the term concerns polar materials such as AlN and ZnO that are strongly polar but unswitchable below dielectric breakdown in bulk or as isolated films. A general thermodynamic theory treats multilayers with polarization normal to the interfaces and constant electric displacement,
$$
D = \varepsilon_0 \varepsilon_i E_i + P_i,
$$
so that internal depolarization fields are set self-consistently by the polarizations, dielectric constants, and thicknesses of all layers. Within this Landau–Ginzburg–Devonshire framework, the internal field renormalizes the local double-well potential, lowers the effective barrier in the hard polar layer, and produces two regimes: “proximity switching,” in which the multilayers collectively switch, and “proximity suppression,” in which they collectively do not switch. The same framework predicts that dielectric or paraelectric layers can in principle acquire induced ferroelectricity when the renormalized quadratic coefficient becomes negative [2501.05219].

Applied to nitride and oxide bilayers, the theory explains why collective switching can occur at fields well below the intrinsic coercive field of the hard polar constituent. For Al\(_{0.83}\)Sc\(_{0.27}\)N/AlN, the single-domain intrinsic coercive fields are approximately \(9\) MV/cm for the AlScN layer and \(\sim 26\) MV/cm for AlN, the latter exceeding the breakdown range of \(\sim 6\)–\(8\) MV/cm; the bilayer single-domain stack coercive field is \(\sim 15\) MV/cm. For Zn\(_{0.66}\)Mg\(_{0.34}\)O/ZnO, the corresponding bulk values are \(\sim 7.7\) and \(\sim 25\) MV/cm, while the bilayer stack spans roughly \(8\) to \(23\) MV/cm depending on thickness ratio. Finite-element calculations further show that randomly distributed monopolar defects act as nucleation centers and reduce the practical coercive field substantially [2501.05219].

A microscopic resolution of the experimental switching puzzle in AlN/AlScN multilayers is provided by domain-wall kinetics. Undoped AlN layers up to about 500 nm thick were observed to switch when coupled to a much thinner Sc-doped trigger layer occupying only \(\approx 3\%\) of the total stack volume. First-principles calculations and machine-learning molecular dynamics identify high-Miller-index \(180^\circ\) domain walls, such as \(\{11\bar{2}1\}\) and \(\{33\bar{6}2\}\), as the decisive kinetic object. In AlN, these walls have formation energy \(\gamma \approx 0.15\ \mathrm{eV/\AA^2}\) and migration barriers \(E_m<0.1\) eV, and finite-field AIMD shows motion below \(2.5\) MV/cm, as low as \(\approx 1.6\) MV/cm depending on wall and configuration. By contrast, neutral vertical walls \(\{1\bar{2}10\}/\{10\bar{1}0\}\) have \(E_m>1\) eV and remain immobile up to 15 MV/cm. In the proposed divide-and-conquer mechanism, Sc lowers the nucleation cost of high-index walls in AlScN but also pins them; once injected across the interface into undoped AlN, the same walls propagate easily because the pinning drops sharply. In a 256,000-atom AlN/Al\(_{0.6}\)Sc\(_{0.4}\)N heterostructure, a 16 MV/cm pulse nucleates and injects the walls, and a subsequent 4 MV/cm pulse completes switching in pristine AlN [2604.25343].

Local probe geometries make the same physics visible in nanodomain nucleation. A Landau–Ginzburg–Devonshire plus finite-element treatment of Al\(_{1-x}\)Sc\(_x\)N/AlN bilayers and trilayers under a shielded biased tip again finds only the two collective regimes, switching and suppression, but the loop shape and threshold depend strongly on layer order. When AlN is under the tip in an AlN/Al\(_{0.73}\)Sc\(_{0.27}\)N bilayer, the simulated coercive biases are \(U_c\approx -23\) V and \(U_t\approx 21\) V, with loop width \(\sim 44\) V. Reversing the order to Al\(_{0.73}\)Sc\(_{0.27}\)N/AlN raises these to \(U_c\approx -45\) V and \(U_t\approx 32\) V, with width \(\sim 79\) V. Effective coercive fields in tip geometry lie in the \(5\)–\(7.5\) MV/cm range, far below the parallel-plate thermodynamic values of \(\sim 26\) MV/cm for AlN and \(\sim 9.5\) MV/cm for AlScN. The physical mechanism is again an internal depolarizing field that opposes the polarization in the layer with larger spontaneous polarization and lowers its barrier [2506.09196].

Compositionally graded structures generalize this to a continuous profile of Landau coefficients and permittivity. In 40 nm AlN–Al\(_{1-x}\)Sc\(_x\)N and ZnO–Zn\(_{1-x}\)Mg\(_x\)O stacks with gradients over diffusion lengths \(h_d=3\)–10 nm, the depolarization field created by the composition gradient lowers the switching barrier in the otherwise unswitchable regions. In parallel-plate geometry, graded AlN–Al\(_{0.73}\)Sc\(_{0.27}\)N and ZnO–Zn\(_{0.63}\)Mg\(_{0.37}\)O switch collectively with coercive biases of 43 and 40 V, corresponding to \(10.6\) and \(10.0\) MV/cm, respectively, lower than the corresponding uniform bilayers. Under a shielded probe, the coercive voltages fall further to approximately 15 and 20 V, or \(3.8\) and \(5.0\) MV/cm. In graded MgO–Zn\(_{1-x}\)Mg\(_x\)O, the same internal field can even induce a shallow double well in the MgO-like region, with vertical stripe domains in capacitor geometry and coercive field \(\approx 0.97\) MV/cm under the tip [2509.13443].

The wurtzite literature is also where the distinction between mechanism and phenomenology is sharpest. The dominant microscopic controversy is whether switching should be attributed mainly to homogeneous bulk-barrier softening or to nucleation and transport of the right domain walls. The high-index-wall results argue decisively for the latter in AlN/AlScN multilayers [2604.25343].

## 6. Experimental signatures, device implications, and unresolved issues

The experimental signatures depend on which sense of proximity ferroelectricity is under discussion. For incipient systems, the defining observables are giant LO–TO splitting, large \(\varepsilon_0/\varepsilon_\infty\), soft polar modes, and strain- or pressure-induced condensation of a \(\Gamma\)-point instability. For interfacial electronic systems, the key observables are polarization-controlled band shifts, gap openings, exchange splittings, ARPES-visible band topology, Hall and nonlinear Hall signals, and anomalous screening in transport maps. For multilayers and nanoscale capacitors, the decisive signatures are collective hysteresis, sequence-dependent coercive fields, local PFM butterfly loops, injected domain walls, and screening-controlled changes in the free-energy landscape [1001.5458] [2311.18026] [2403.17326] [2506.09196].

The technological implications are correspondingly diverse. In TlBr, the large static dielectric constant associated with proximity to ferroelectricity is proposed to screen charged defects and impurities, reducing ionized-impurity scattering and trap capture. TlBr radiation detectors show \(\mu\tau \approx 10^{-3}\ \mathrm{cm^2/V}\) for electrons and \(\approx 10^{-4}\ \mathrm{cm^2/V}\) for holes, resistivity \(\sim 10^{10}\ \Omega\cdot\mathrm{cm}\), and a band gap of 2.68 eV. In graphene/BFO, the multiferroic proximity effect supports a six-resistance device concept with room-temperature PER, PMR, and PMER. In \(\alpha\)-Bi/SnSe, ferroelectric switching reconfigures pseudo-Weyl states, Berry-curvature dipoles, nonlinear anomalous Hall response, and spin Hall signals. In conductive moiré ferroelectrics, anomalous screening and pinned domain walls support multilevel memory and synaptic responses. In granular films, ferroelectric screening gives giant electro-resistance without current through the ferroelectric. In nitrides and oxides, proximity switching opens routes to memory, actuation, sensing, nonlinear optics, and nanoscale domain control in materials that would otherwise remain unswitchable [1001.5458] [1909.02844] [2311.18026] [1407.7719] [2501.05219].

Several controversies and open problems remain. One is terminological: the same phrase covers at least four distinct physical situations, from incipient bulk ferroelectricity to purely interfacial band-structure control. Another is mechanistic. In unconventional moiré ferroelectrics, the debate between correlation-driven charge transfer and interfacial sliding ferroelectricity motivated the WSe\(_2\)/graphene/BN control experiments and the twisted-MoS\(_2\) replication. In wurtzite multilayers, continuum electrostatic models captured the thermodynamic feasibility of switching thick AlN but overestimated coercive fields until domain-wall injection was included explicitly. In hafnia, the unresolved issues center on microscopic identification of the surface and interface compensating species, their kinetics, and how buried interfaces in real MIM devices couple electrochemical and electrostatic control [2403.17326] [2604.25343] [2207.12525].

A plausible synthesis is that proximity ferroelectricity is best understood as a relational property. A material is “proximate” not simply because it is polar, but because its polarization, switchability, or electronic structure is redefined by nearby degrees of freedom—soft phonons, anomalous dynamic charges, electrode screening, mobile domain walls, interface electrostatics, finite surface DOS, or a neighboring ferroelectric order parameter. The breadth of current usage reflects that shared structure rather than a single microscopic mechanism.

Source: https://www.emergentmind.com/topics/proximity-ferroelectricity