Proximity Ferroelectricity Driven by Mobile High-Miller-Index Domain Walls
Changming Ke, Shi Liu
Abstract
Wurtzite ferroelectrics such as scandium-doped aluminum nitride (AlScN) are promising for next-generation memory because of their compatibility with semiconductor processes and strong spontaneous polarization. Ferroelectric switching in these materials is typically attributed to doping-induced softening of the bulk switching barrier. However, recent reports of proximity ferroelectricity, in which undoped AlN layers up to 500 nm thick fully switch in AlN/AlScN multilayers, challenge this view. Here, we reveal an alternative switching mechanism mediated by high-Miller-index domain walls, long overlooked due to their complex geometry and presumed instability. Using first-principles calculations and machine-learning molecular dynamics simulations, we show that these walls, once nucleated, migrate with exceptionally low barriers. The Sc dopants play a dual role: they stabilize high-index walls and thereby promote nucleation, while also introducing pinning that hinders wall motion. In multilayers, our simulations demonstrate that mobile domain walls nucleated in AlScN can propagate deep into adjacent AlN, where they move easily without dopant pinning, enabling low-field switching across thick undoped layers. This microscopic divide-and-conquer mechanism resolves the puzzle of proximity ferroelectricity and highlights high-index interfaces as an underexplored lever for controlling ferroelectric switching.
AI Impact Assessments
(3 models)Scientific Impact Assessment
Core Contribution
This paper addresses a fundamental puzzle in wurtzite ferroelectrics: how undoped AlN layers up to 500 nm thick can undergo complete polarization switching when adjacent to thin AlScN ferroelectric layers — the phenomenon of "proximity ferroelectricity" first reported experimentally in Nature (2025). The authors propose a microscopic "divide-and-conquer" mechanism centered on high-Miller-index domain walls (DWs), such as {11̄21} and {33̄62}. The key insight is that these weakly charged, high-index DWs possess an unusual combination of moderate formation energies and exceptionally low migration barriers, making them highly mobile once nucleated. Sc doping plays a dual, previously unrecognized role: it lowers high-index DW formation energies (facilitating nucleation) while simultaneously introducing pinning centers that hinder DW motion. In multilayers, the doped layer nucleates DWs that are then injected into adjacent undoped AlN, where they propagate freely without dopant pinning.
This represents a conceptual shift from the conventional view that Sc doping acts primarily by softening the bulk double-well switching barrier, instead reframing the problem in terms of DW nucleation and kinetics.
Methodological Rigor
The study employs a multi-scale computational approach combining:
1. DFT calculations for DW formation energies and migration barriers (NEB method) across multiple wall types and compositions.
2. Finite-field ab initio molecular dynamics (AIMD) to quantify DW mobility under applied electric fields, providing first-principles switching field estimates.
3. Machine-learning deep potential molecular dynamics (DPMD) for large-scale simulations (up to 256,000 atoms) that capture nucleation and domain growth in realistic AlN/AlScN heterostructures.
The multi-scale approach is well-justified: AIMD provides benchmarks for DW mobility but is limited to small cells and short timescales, while DPMD enables statistical sampling of nucleation events and heterostructure-level phenomena. The deep potential model is validated against DFT for DW energies and barriers, lending credibility to the large-scale simulations.
A notable methodological limitation is the definition of switching fields (EDW_s and ENLS_s) based on short simulation timescales (5 ps for AIMD, 100 ps for DPMD), which are orders of magnitude shorter than experimental switching times (~microseconds). The authors acknowledge this, but the quantitative gap between simulated and experimental fields remains large. The two-step field protocol for the heterostructure simulation (16 MV/cm followed by 4 MV/cm) is somewhat contrived, though it effectively demonstrates the mechanistic principle.
Potential Impact
Immediate field impact: This work directly addresses a high-profile experimental observation (proximity ferroelectricity, Nature 2025) with an atomistic mechanism that the existing continuum thermodynamic models could not provide. The identification of high-index DWs as the kinetically dominant switching pathway fundamentally reframes how the community should think about polarization reversal in wurtzite ferroelectrics.
Practical implications: The divide-and-conquer mechanism suggests concrete design principles for ferroelectric heterostructures — optimizing the doped layer as a "DW injector" rather than simply a polarization source. This could guide the engineering of lower coercive fields in memory devices based on AlScN, a material of intense commercial interest due to its CMOS compatibility.
Broader applicability: The prediction of mobile high-index DWs in GaN (~0.8 MV/cm) and ZnO (~1.1 MV/cm) extends the framework beyond AlScN, potentially opening ferroelectric functionality in widely used semiconductor materials.
Conceptual contribution: The inverse correlation between 180° DW stability and mobility — counter to the intuition from perovskite ferroelectrics — is a significant theoretical insight that could influence DW engineering across ferroelectric materials classes.
The explanation for anomalous Avrami exponents (n ≈ 11) through the convergence of nucleation and growth timescales at high Sc concentrations is an elegant secondary contribution.
Timeliness & Relevance
This paper is exceptionally timely. Proximity ferroelectricity was reported in Nature in early 2025, and the thermodynamic framework by Eliseev et al. appeared in Physical Review X in 2025. This work fills a critical gap by providing the missing atomistic mechanism. The broader context — wurtzite ferroelectrics for memory applications — is one of the most active areas in materials science, with substantial industrial interest from semiconductor manufacturers seeking alternatives to HfO₂-based ferroelectrics.
Strengths
Limitations
Overall Assessment
This is a high-impact computational study that provides a compelling and novel atomistic mechanism for proximity ferroelectricity in wurtzite heterostructures. The identification of high-index DWs as mobile switching agents and the dual role of Sc doping represent significant conceptual advances. While experimental validation remains needed, the mechanistic framework is physically sound and offers actionable design principles for ferroelectric device engineering.
Generated Apr 29, 2026
Comparison History (35)
Paper 2 offers a broadly impactful mechanism for ferroelectric switching in AlN/AlScN multilayers, resolving a prominent experimental puzzle (proximity ferroelectricity) with a new domain-wall-mediated picture. Its combination of first-principles and ML molecular dynamics suggests strong methodological rigor and provides actionable design levers (high-index walls, dopant roles) for scalable semiconductor-compatible memories and piezo/ferroelectric devices. The implications extend across materials science, device engineering, and computational modeling. Paper 1 is timely for quantum networking, but the practical leap to telecom-band interfaces may still face significant engineering constraints and narrower near-term adoption.
Paper 1 introduces a highly novel, generalizable AI framework for open hypothesis discovery in autonomous experiments, shifting the paradigm from closed-loop optimization to automated physical law generation. This methodological breakthrough has broader applicability across various scientific disciplines compared to Paper 2, which, while highly relevant for semiconductor memory applications, focuses on a specific material mechanism.
Paper 2 addresses a critical bottleneck in next-generation memory devices using semiconductor-compatible wurtzite ferroelectrics. By resolving the puzzle of proximity ferroelectricity through a novel domain-wall switching mechanism, it challenges existing paradigms and offers immediate, high-value implications for microelectronics. Paper 1 presents elegant fundamental materials science regarding strain gradients in perovskite membranes, but Paper 2's direct applicability to scalable semiconductor technologies and its use of advanced machine-learning molecular dynamics give it a broader and more immediate scientific and industrial impact.
Paper 2 introduces a novel microscopic mechanism (mobile high-Miller-index domain walls) that resolves an outstanding experimental puzzle (proximity ferroelectricity in AlN/AlScN multilayers) and provides a clear design lever for low-field switching in CMOS-compatible ferroelectrics, with strong implications for memory devices and ferroelectric materials physics. It combines first-principles and ML molecular dynamics for mechanistic rigor and has broad relevance across ferroelectrics, interfaces, and device engineering. Paper 1 is valuable as a benchmark, but its impact is more incremental and primarily methodological within ML-for-materials.
Paper 2 addresses a major puzzle in ferroelectric switching (proximity ferroelectricity in AlN/AlScN multilayers) with broad implications for next-generation memory technology. It introduces a novel mechanism—mobile high-Miller-index domain walls—that resolves an unexplained experimental observation and opens new design principles for ferroelectric devices. The combination of first-principles calculations with machine-learning molecular dynamics adds methodological sophistication. Its direct relevance to semiconductor industry applications (CMOS-compatible ferroelectrics) gives it significantly broader real-world impact compared to Paper 1, which provides a detailed but more specialized electronic structure study of a single material.
Paper 2 likely has higher impact: it introduces a broadly applicable first-principles framework for ultrafast light-induced phase transitions, explicitly incorporating quartic anharmonicity and nonlinear electron–phonon coupling, and validates it against experiments in TiSe2. This methodological advance is timely for nonequilibrium quantum materials and can transfer across many CDW, ferroelectric, and correlated systems. Paper 1 is novel and relevant for Al(Sc)N ferroelectric devices, but its immediate impact is more specialized to wurtzite nitride ferroelectrics and multilayer switching mechanisms.
Paper 1 resolves a fundamental anomaly in ferroelectric switching (proximity ferroelectricity) by discovering a novel mechanism involving high-Miller-index domain walls. Its insights directly impact the development of next-generation semiconductor memory, combining high novelty with broad technological application. Paper 2 provides a valuable but more incremental modeling framework for metamaterials, making Paper 1's breakthrough more scientifically impactful.
Paper 2 uncovers a fundamentally new microscopic mechanism to explain proximity ferroelectricity in materials critical for next-generation memory. Its use of advanced machine-learning molecular dynamics to reveal the role of high-Miller-index domain walls offers broad implications for semiconductor design. While Paper 1 provides a valuable diagnostic technique for diamond, Paper 2's direct relevance to advanced memory technologies and its paradigm-shifting theoretical insights suggest a broader and more transformative scientific impact.
Paper 2 reports the first direct optical visualization of altermagnetic domains in MnTe, a breakthrough observation in the rapidly emerging field of altermagnetism. This provides fundamental experimental confirmation of predicted altermagnetic domain properties using a simple laboratory-scale technique, opening the door for widespread experimental studies. While Paper 1 elegantly resolves the proximity ferroelectricity puzzle through computational methods with clear technological relevance to memory devices, Paper 2's pioneering experimental observation in a hot new field (altermagnetism) is likely to have broader and more immediate impact across condensed matter physics and spintronics.
Paper 2 offers immense potential for real-world applications in next-generation semiconductor memory. By utilizing modern machine-learning molecular dynamics to resolve the puzzle of proximity ferroelectricity, it demonstrates high methodological rigor and timeliness. While Paper 1 provides a highly novel fundamental physics discovery regarding the phonon thermal Hall effect, Paper 2 bridges fundamental materials science with immediate industrial relevance, giving it a broader and more substantial technological and scientific impact.