🔬 mesoscale physics

Relaxation-driven flat bands and topology in moiré transition metal dichalcogenide heterobilayers

This paper demonstrates that intrinsic lattice relaxation in moiré transition metal dichalcogenide heterobilayers generates a pseudomagnetic field that induces topologically non-trivial flat bands with non-zero Chern numbers, overturning previous conclusions of topological triviality and establishing these systems as a new class of relaxation-driven topological materials.

Original authors: Mitchell Luskin, Max Geier, Liang Fu, Ziyan Zhu

Published 2026-08-11

📖 8 min read🧠 Deep dive

CC BY 4.0

Original authors: Mitchell Luskin, Max Geier, Liang Fu, Ziyan Zhu

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine a world where you can build new materials by stacking thin sheets of atoms like pancakes. This is the playground of "moiré materials," a hot corner of physics where scientists twist two layers of a crystal slightly out of alignment. When you do this, the atoms don't just sit there; they create a giant, colorful pattern called a "moiré superlattice," kind of like the rippling interference pattern you see when you hold two window screens over each other. Usually, scientists thought these patterns were just static backdrops for electrons to dance on. But in reality, the atoms seek to relax. They shift and stretch to find the most comfortable spot, changing the landscape entirely. This "lattice relaxation" is the secret sauce that turns a boring, flat electronic stage into a dynamic, topological playground. Why does this matter? Because these materials might hold the keys to the next generation of super-fast, super-efficient electronics and even quantum computers that don't lose information.

The paper you're about to read dives deep into a specific type of these materials: a sandwich made of two different transition metal dichalcogenide (TMD) sheets, specifically WSe2 and WS2. For a long time, scientists used a simplified model that treated these sheets as rigid, unyielding boards. They thought that if you twisted them, you'd get flat energy bands (where electrons move very slowly) but nothing special in terms of "topology" (a fancy word for the shape of the electron's path that makes it robust against errors). However, the authors of this study, Mitchell Luskin, Max Geier, Liang Fu, and Ziyan Zhu, argue that this conclusion was a mistake caused by ignoring the atoms' natural urge to relax. They built a new, more realistic model that accounts for how the atoms shift, stretch, and squish. Their simulations show that this relaxation isn't just a small tweak; it's the main event. It generates invisible "pseudo-magnetic fields" (magnetic fields that act like real ones but are created by the stretching of the material) and "pseudo-electric fields." These fields reshape the energy bands, creating "flat bands" where electrons get stuck and interact strongly, and—most excitingly—turning previously boring bands into topological ones with a specific "Chern number" (a mathematical fingerprint of their shape). The authors suggest that this mechanism could lead to the discovery of "fractional Chern insulators," exotic states of matter that could revolutionize quantum computing, all driven by the simple act of atoms relaxing into their most comfortable positions.

The Story of the Stretchy Sandwich

Imagine you have two sheets of a special, stretchy fabric. One sheet is slightly bigger than the other. If you lay them on top of each other and twist them a tiny bit, they don't just sit flat. Because they are stretchy, the atoms in the fabric want to move to the spots where they fit together best. They pull and push, creating a new, relaxed shape. In the world of physics, this is called lattice relaxation.

For years, scientists studying these "moiré" materials (named after the pattern they make) used a "rigid" model. They pretended the fabric was made of steel—unbending and unchanging. They thought that if you twisted these materials, you'd get some interesting effects, but nothing truly magical like "topology" (which is like a knot in a string that can't be untied without cutting the string). They thought the top layers of these materials were just boring, flat roads for electrons.

But the authors of this paper said, "Wait a minute! Atoms aren't steel; they're more like rubber bands." They decided to build a new model that accounts for how the atoms actually move and stretch when they relax. They focused on a specific pair of materials: WSe2 (Tungsten Diselenide) and WS2 (Tungsten Disulfide).

The Three Magic Channels

The authors discovered that when these atoms relax, they don't just change the shape of the material; they create three distinct "channels" of magic that change how electrons behave:

The Sharpened Landscape (Modified Moiré Potential): As the atoms move to their comfortable spots, the "hills and valleys" of the energy landscape get sharper. It's like taking a smooth, rolling hill and turning it into a deep, narrow canyon. This traps the electrons in smaller spaces, making the "bandwidth" (how fast they can move) much narrower. Narrower bands mean electrons are more likely to stop and talk to each other, which is where the fun physics happens.
The Invisible Electric Field (Pseudoelectric Potential): The stretching of the atoms creates a scalar potential, which acts like an invisible electric field. This helps to widen the gaps between different energy levels, keeping the electrons in their specific lanes.
The Ghost Magnet (Pseudomagnetic Field): This is the big one. The stretching and twisting of the lattice creates a vector potential that acts exactly like a magnetic field, even though there are no magnets involved. The authors call this a "pseudomagnetic field." In their simulations, this field alone was enough to open a "topological gap" between the third and fourth energy bands.

The Big Discovery: Topology from Relaxation

Here is the punchline: In the old, rigid models, the third and fourth energy bands were topologically "trivial" (boring, like a plain circle). But in the new, relaxed model, the pseudomagnetic field twists these bands into a knot. The authors found that the third band gets a "Chern number" of +1, and the fourth gets -1.

Think of the Chern number as a label that tells you how many times the electron's path winds around a hole in a donut. A Chern number of 0 means no hole (boring). A Chern number of +1 or -1 means there is a hole, and the path is knotted. This knot makes the electron flow robust; it can't be easily stopped by impurities or defects.

The authors ran simulations using a method called "neural-network variational Monte Carlo." This is a fancy way of using artificial intelligence to guess the best possible state for a huge group of interacting electrons. They found that even when you add in the messy, real-world interactions between electrons, the "charge gap" (the energy needed to move an electron) stays large and healthy in the relaxed model. In fact, for certain twist angles (around 2.6 degrees), the relaxed model showed a much larger gap than the rigid model, suggesting that these topological states are real and stable.

The "Ideal" Playground for Quantum States

Why do we care about these knotted bands? Because they are the perfect playground for something called a Fractional Chern Insulator (FCI).

Imagine a highway where cars (electrons) usually drive in lanes. In an FCI, the cars get so crowded and interact so strongly that they start behaving like a single, giant fluid. This fluid can carry electricity in a very special way that is perfect for quantum computers. But for this to happen, the "road" (the energy band) needs to be very flat, and the "traffic rules" (the quantum geometry) need to be very uniform.

The authors found that lattice relaxation does a fantastic job of smoothing out the "traffic rules." In the rigid model, the "Berry curvature" (a measure of how the electron's path twists) was spiky and uneven, concentrated in tiny corners. But in the relaxed model, this curvature spreads out smoothly across the whole landscape. This makes the band much closer to the "ideal" state needed for FCI.

They calculated a few numbers to prove this. They looked at a "trace-condition violation" (a measure of how far the band is from being perfect) and a "Berry curvature fluctuation" (how bumpy the landscape is). In the relaxed model, these numbers dropped significantly, especially at small twist angles and small lattice mismatches. For example, at a twist angle of 0.5 degrees and a mismatch of 0.01, the relaxed model was an order of magnitude better than the rigid one.

What This Means for the Future

The authors are careful to say that these are results from simulations and models, not a physical experiment they performed in a lab. However, they provide a clear roadmap for what to look for.

They predict that if you take a WSe2/WS2 heterobilayer and twist it just right, you should see a Quantum Anomalous Hall Effect at a specific filling level (when there are 6 electrons per unit cell). This would mean the material conducts electricity on its edges without any magnetic field, a hallmark of topological materials. They also suggest that at fractional fillings (like 1/3 or 2/3 of a band), you might see those elusive Fractional Chern Insulator states.

The paper also rules out the old idea that these heterobilayers are just topologically trivial. It argues that the "trivial" results seen in previous studies were an artifact of ignoring the atoms' relaxation. By including the full set of strain-induced effects, the topology emerges naturally.

In short, this paper suggests that the secret to unlocking the most exotic quantum states in these materials isn't just twisting them; it's letting them relax. The atoms' natural desire to find a comfortable spot creates the very magnetic fields and smooth landscapes needed to build the quantum computers of the future. It turns a simple act of stretching into a powerful tool for engineering the quantum world.

Technical Summary: Relaxation-driven Flat Bands and Topology in Moiré TMD Heterobilayers

Problem Statement
Moiré transition metal dichalcogenide (TMD) heterobilayers, such as WSe $_2$ /WS $_2$ , are frequently modeled using continuum theories that predict topologically trivial bands. This stands in contrast to their homobilayer counterparts (e.g., twisted bilayer MoTe $_2$ ), which host topological bands and fractional Chern insulator (FCI) states. Previous continuum models for heterobilayers have treated lattice relaxation incompletely: while they account for the modification of the moiré potential due to relaxed stacking vectors, they neglect the strain-induced terms generated by the same displacement field. Specifically, these models omit the scalar deformation potential (pseudoelectric field) and the vector potential (pseudomagnetic field) that arise from spatially varying strain. Consequently, the role of intrinsic lattice relaxation in generating nontrivial topology in heterobilayers has remained an open question.

Methodology
The authors develop a comprehensive continuum model for TMD heterobilayers that fully incorporates lattice relaxation through three distinct channels:

Modified Moiré Potential: The relaxation-induced displacement field modifies the local stacking, requiring the inclusion of higher-order Fourier harmonics in the moiré potential.
Pseudoelectric Potential: A scalar deformation potential ( $\Phi$ ) arising from the trace of the strain tensor ( $\epsilon_0$ ).
Pseudomagnetic Potential: A vector gauge field ( $\mathbf{A}$ ) arising from the deviatoric strain components, which generates a pseudomagnetic field ( $\mathbf{B} = \nabla \times \mathbf{A}$ ).

The model is derived from a tight-binding Hamiltonian projected onto the valence band maximum of the higher-lying layer (WSe $_2$ ). The strain Hamiltonian is expanded to linear order in momentum and strain, yielding a pseudo-gauge form. The authors use WSe $_2$ /WS $_2$ as a prototype system, parameterizing the moiré potential and strain coefficients using density functional theory (DFT) calculations.

To bridge single-particle band structure with many-body physics, the authors employ a neural-network variational Monte Carlo (NNVMC) method. They compute the many-body charge gap ( $\Delta_c$ ) for a system of 18 particles in a $3 \times 3$ moiré supercell, comparing results from the rigid model (no relaxation) against the fully relaxed model.

Key Contributions and Results

Decomposition of Relaxation Effects: By isolating the three relaxation channels, the authors demonstrate that:
- The pseudoelectric potential and higher-order moiré harmonics primarily flatten the top two valence bands and increase the gap between them ( $\Delta_{12}$ ), but do not generate topology.
- The pseudomagnetic field is the sole driver of nontrivial topology. It opens a topological gap ( $\Delta_{34}$ ) between the third and fourth valence bands, endowing them with Chern numbers $C_3 = +1$ and $C_4 = -1$ .
Phase Diagram and Topology: Across a broad range of twist angles ( $\theta$ ) and lattice mismatches ( $\delta$ ), the relaxed model predicts a topological phase where the third and fourth valence bands are separated by a gap with nontrivial Chern numbers. This topology persists at the natural lattice mismatch of WSe $_2$ /WS $_2$ ( $\delta \approx 0.04$ ). In contrast, the top two bands remain topologically trivial, consistent with experimental observations of trivial correlated insulators at filling $\nu=1$ .
Many-Body Stability: The NNVMC calculations show that the relaxation-enhanced charge gap at filling $\nu=2$ survives many-body interactions. For twist angles $\theta \gtrsim 2.6^\circ$ , the relaxed model exhibits a systematically larger charge gap than the rigid model, confirming that relaxation effects are robust against Coulomb interactions.
Quantum Geometry and FCI Conditions: The authors analyze the quantum geometry of the bands using the integrated trace-condition violation ( $T/2\pi$ ) and Berry curvature fluctuation ( $F$ ). They find that relaxation smoothens the Berry curvature and quantum metric relative to the rigid model, moving the band closer to the ideal Chern limit. This smoothing is most effective at small twist angles and modest lattice mismatches, identifying small-mismatch TMD heterobilayers as promising platforms for FCI states.

Significance
The paper establishes a new framework connecting first-principles calculations to many-body observables via a continuum model that treats relaxation effects with full gauge-field consistency. The authors claim that this work reclassifies moiré heterobilayers as a new class of topological materials where topology is driven entirely by intrinsic lattice relaxation, rather than external strain or rigid-layer topology.

The study predicts several experimentally testable phenomena:

The emergence of a quantum anomalous Hall state at integer filling $\nu=6$ (filling the third valence band), characterized by a quantized Hall conductance $\sigma_{xy} = e^2/h$ .
The possibility of fractional Chern insulator states at fractional fillings of the third valence band, facilitated by the combination of flat bandwidth, nontrivial Chern number, and favorable quantum geometry.
A quantitative enhancement of the charge gap at $\nu=2$ due to relaxation, which can be probed via transport and compressibility measurements.

The framework is presented as generalizable to arbitrary heterobilayer TMD systems, requiring only the substitution of DFT-derived moiré potentials and tight-binding parameters for different material combinations.