Harnessing electrostatics through temperature modulations to control ferroelectrics

Problem Statement
Dynamically controlling the state of ferroelectric materials is a challenge of fundamental and technological importance. While direct coupling to external electric fields and strain offer established methods for manipulation, temperature variations provide an alternative strategy. Specifically, this paper investigates spatial temperature modulations (T-modulations) rather than homogeneous temperature changes. The core physical problem addressed is the electrostatic consequence of a temperature gradient across a ferroelectric sample: such a gradient induces a polarization gradient, which generates bound charges ($\rho_b = -\nabla \cdot P$) and a resulting depolarizing field. This field penalizes polarization oriented parallel to the gradient. The paper explores how harnessing this depolarizing field can lead to non-electrical poling, dipole rotation, and the formation of persistent polar textures.

Methodology
The authors employ a dual-method approach combining scale-independent theoretical modeling and predictive atomistic simulations:

1. Landau Potential Modeling: The study utilizes a sixth-order Landau potential to model a perovskite oxide (specifically PbTiO$_3$) divided into "hot" and "cold" regions. The model couples these regions via the continuity of the electric displacement component parallel to the temperature modulation ($D_\parallel$). An electrostatic energy term is included to model the bound charges and the associated depolarizing field. The system assumes periodic boundary conditions along the modulation direction, equivalent to perfect screening.
2. Atomistic Simulations: To validate the Landau model, the authors perform Langevin molecular dynamics (MD) simulations based on second-principles potentials derived from atomistic simulations. These simulations use supercells (10$\times$10$\times$20 repetitions of the elemental cell) with a sinusoidal temperature profile assigned to individual atoms. The simulations explore two scenarios:
   • Dipole Rotation: Starting from a polarized state, the temperature difference ($\Delta T$) is increased to observe the reorientation of dipoles.
   • Domain Formation: The system is cooled through the Curie temperature ($T_C$) under a constant $\Delta T$ to observe the selection of polar axes.
   • Strain Constraints: To simulate uniaxial materials or strained thin films (e.g., PbTiO$_3$ on SrTiO$_3$), in-plane lattice vectors are fixed to prevent dipole rotation, forcing the system to explore multidomain configurations.

Key Results

• Dipole Rotation: Both the Landau model and MD simulations demonstrate that a sufficiently strong temperature modulation forces the polarization vector to rotate from being parallel to the gradient ($P_\parallel$) to being perpendicular ($P_\perp$). This rotation eliminates the bound charges and the associated depolarizing penalty.
• Phase Diagram: A phase diagram constructed as a function of $T_{cold}$ and $\Delta T$ reveals that the $P_\perp$ state is the global free energy minimum for $T_{cold} < T_C$. The $P_\parallel$ state exists as a local minimum only for small $\Delta T$. Notably, the electrostatic coupling allows the "hot" region to remain polarized even when $T_{hot} > T_C$, provided it is coupled to a colder, polarized region.
• Agreement of Scales: The atomistic simulations show near-perfect agreement with the scale-independent Landau model, confirming that these electrostatic effects are scale-free. While the gradients in the simulations are large ($\sim$100 K/nm), the authors argue the physics applies to experimentally accessible modulations.
• Persistent Textures: When dipole rotation is suppressed (e.g., by uniaxial constraints or compressive strain), the system does not simply rotate but breaks into multidomain states. Specifically, the combination of compressive strain and T-modulation induces the formation of electric stripe domains that persist even after the temperature modulation is removed.
• Cooling Protocols: Simulations of cooling through $T_C$ under a constant $\Delta T$ show a high probability (up to 97% for $\Delta T = 100$ K) of the polarization aligning perpendicular to the temperature modulation, significantly exceeding random chance.

Significance and Claims
The paper claims to provide a numerical proof-of-concept for controlling ferroelectric states through temperature modulation alone. The primary significance lies in demonstrating that:

1. Temperature gradients can be used to reorient electric dipoles and select specific polar axes without the application of external electric fields.
2. This control mechanism relies on long-range electrostatic interactions (depolarizing fields) and is largely independent of sample size or the specific details of the temperature modulation profile.
3. It offers a dynamical route to creating persistent polar textures, such as multidomain states, in materials where dipole rotation is constrained.

The authors position this work as an alternative to electrical poling and strain manipulation, suggesting that these effects should be viable in the nanoscale regime due to the relatively low thermal conductivity of ferroelectric oxides, which allows temperature differences to persist long enough for dipole reconfiguration (estimated at 20–50 ps). The paper concludes by proposing temperature modulation as a viable method for the dynamical control of ferroelectricity, inviting experimental verification.