🔬 mesoscale physics

Atomistic substrate relaxation effects in the band gaps of graphene on hexagonal boron nitride

This study demonstrates that atomistic substrate relaxation significantly modulates the primary and secondary band gaps of graphene on hexagonal boron nitride, reducing the zero-twist primary gap from ~30 meV to ~3 meV in rigid structures while revealing a maximum near 0.6° and a persistent ~1 meV gap across all twist angles that switches sign between 30° and 60°.

Original authors: Jiaqi An, Nicolas Leconte, Srivani Javvaji, Youngju Park, Jeil Jung

Published 2026-02-24

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Jiaqi An, Nicolas Leconte, Srivani Javvaji, Youngju Park, Jeil Jung

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 you have two very special, ultra-thin sheets of material. One is Graphene (a single layer of carbon atoms arranged in a honeycomb pattern, like a chicken wire fence), and the other is Hexagonal Boron Nitride (h-BN, a similar honeycomb pattern but made of boron and nitrogen atoms).

When you stack these two sheets on top of each other, they don't just sit there perfectly flat. Because the atoms in the carbon fence are slightly smaller than the atoms in the boron-nitrogen fence, they don't line up perfectly. This mismatch creates a giant, repeating pattern of ripples and waves across the surface, known as a Moiré pattern. Think of it like holding two slightly different window screens over each other; you see a new, larger pattern emerge where the holes overlap.

This paper is about what happens when you twist these two sheets relative to each other, like turning a dial, and how the "softness" or "stiffness" of the bottom sheet changes the electronic properties of the top sheet.

Here is the breakdown of their discovery using everyday analogies:

1. The "Electronic Gap" (The Traffic Light)

In these materials, electrons usually zoom around freely like cars on a highway. However, because of the Moiré pattern, a "traffic light" appears. Sometimes, the light turns red, stopping the electrons. This stoppage creates a Band Gap.

Primary Gap: The main traffic light right at the center of the highway.
Secondary Gap: A smaller traffic light further down the road.

The size of this gap determines how well the material can act as a switch (like in a computer chip). If the gap is too small, the switch is leaky. If it's just right, it's a perfect switch.

2. The "Rigid vs. Soft" Substrate (The Concrete vs. The Mattress)

The researchers wanted to know: Does it matter if the bottom sheet (h-BN) is hard as a rock or soft like a mattress?

The "Rigid" Scenario: Imagine the bottom sheet is glued to a concrete floor. It cannot move. When you twist the top sheet, the bottom sheet stays perfectly flat.
The "Relaxed" Scenario: Imagine the bottom sheet is a soft mattress. When you twist the top sheet, the mattress squishes and deforms to try to fit the top sheet better.

The Finding:
When the bottom sheet is rigid (concrete), the "traffic light" (the gap) is very dim—only about 3 meV (milli-electron volts). It's barely a stop sign.
But when the bottom sheet is soft and allowed to relax (the mattress), the gap gets much bigger—up to 30 meV! The mattress squishes in a way that creates a much stronger "stop" for the electrons.

3. The "Goldilocks" Twist Angle (The 0.6° Sweet Spot)

The researchers twisted the top sheet at different angles, from 0° (perfectly aligned) to 30°.

The Surprise: They found a "sweet spot" at a tiny twist of about 0.6 degrees.
The Analogy: Imagine trying to fit a square peg in a round hole. Usually, it's a bad fit. But at this specific 0.6° angle, the giant Moiré pattern (the ripples) lines up perfectly with the atoms in the graphene, like a key fitting into a lock.
The Result: At this specific angle, the system becomes extra stable (energetically happy), and the "traffic light" (the gap) gets a little boost. It's like the mattress found the perfect way to hug the top sheet.

4. The "Disappearing" Secondary Gap

As they twisted the angle further (beyond 1°), the Secondary Gap (the second traffic light) simply vanished. It closed up completely. This means that at larger twist angles, the material loses one of its special electronic features, but the Primary Gap (the main one) stays open and robust, even up to 30°.

5. Why This Matters

This study is crucial because scientists have been arguing about how big these gaps actually are. Some computer models said they were big, others said they were small.

The Lesson: The size of the gap depends entirely on whether you treat the bottom layer as a rigid rock or a squishy mattress.
The Takeaway: If you want to build future electronics using these materials, you can't just assume the bottom layer is hard. You have to account for how it "relaxes" and deforms. If you ignore this, your calculations will be off by a factor of 10!

Summary in a Nutshell

Think of Graphene on h-BN as a dance floor.

If the floor is concrete (rigid), the dancers (electrons) can't move much, but the pattern is weak.
If the floor is a trampoline (relaxed), the dancers can bounce and interact more strongly, creating a much stronger pattern (a bigger gap).
There is a specific dance move (0.6° twist) where the trampoline and the dancers sync up perfectly, making the system extra stable.

This paper tells us that to understand how these materials work, we must stop treating the bottom layer as a static stage and start treating it as a dynamic, squishy partner that actively helps shape the electronic properties.

1. Problem Statement

Graphene on hexagonal boron nitride (G/h-BN) heterostructures exhibit moiré-induced electronic band gaps (primary and secondary) near the Dirac point and higher energies. While experimental and theoretical studies have confirmed these gaps, there is a significant discrepancy in reported values:

Continuum models often predict smaller gaps (e.g., ~8 meV).
Atomistic simulations with full relaxation predict larger gaps (20–35 meV).
Experiments report intermediate values (~14 meV).

The authors identify that previous atomistic studies often neglected the specific relaxation constraints of the substrate layers (specifically the remote h-BN layers beneath the interface). The central problem is to quantitatively assess how different substrate relaxation schemes (rigid vs. relaxed) and twist angles affect the magnitude and evolution of these band gaps, resolving the discrepancies between models and experiments.

2. Methodology

The study employs a multi-scale approach combining classical molecular dynamics (MD) for structural relaxation and a refined Tight-Binding (TB) model for electronic structure calculations.

Structural Relaxation (MD):
- Simulations were performed using the LAMMPS package.
- Force Fields: The interlayer interaction uses the DRIP potential (reparametrized with EXX-RPA DFT data), while intralayer interactions use ExTeP (for B-N) and REBO2 (for C-C).
- Relaxation Schemes: Four distinct configurations were analyzed to isolate substrate effects:
  1. Suspended: Both graphene and h-BN layers are fully relaxed.
  2. Rigid Interface: The top h-BN layer (interface) is rigid; graphene relaxes.
  3. Rigid Remote: The interface h-BN relaxes, but a remote h-BN layer (rotated 180°) is kept rigid to mimic a bulk substrate.
  4. Completely Rigid: No relaxation allowed.
- Commensurate Supercells: Constructed using four integer indices ( $p, q, p', q'$ ) to generate twist angles up to 60° with minimal lattice mismatch noise.
Electronic Structure (Tight-Binding):
- A Hybrid Tight-Binding (HTC) model was developed, combining a refined two-center (TC) interlayer model with short-range F2G2 intralayer terms derived from DFT Wannierization.
- Truncated Atomic Plane Wave (TAPW) Method: Used to project the full atomistic Hamiltonian onto a reduced subspace. This allows for efficient calculation of the average mass term ( $\bar{m}$ ), which directly determines the primary band gap at the Dirac point.
- Strain Corrections: Global and local strain corrections were applied to hopping parameters to account for lattice reconstruction and interlayer distance variations.

3. Key Contributions

Quantification of Substrate Effects: The paper demonstrates that the treatment of the remote substrate layers is critical. Keeping remote layers rigid significantly reduces the predicted band gaps compared to fully suspended systems.
Discovery of a Local Energy Minimum: The authors identify a non-monotonic behavior where the total energy exhibits a shallow minimum at a twist angle of $\sim 0.6^\circ$ . This coincides with a local enhancement of the primary band gap.
Mechanism of Stabilization: The energy minimum at $0.6^\circ$ is attributed to a commensuration effect where the moiré lattice vectors align with the graphene lattice vectors, enhancing structural stability and lattice reconstruction (specifically the expansion of the energetically favorable BA-stacking region).
Robustness of the Primary Gap: Contrary to continuum models predicting gap closure at large angles, the study shows the primary gap remains finite ( $\sim 1$ meV) up to $30^\circ$ , though it becomes sensitive to the choice of rotation center (AA vs. AB stacking) at larger angles.

4. Key Results

Primary Band Gap Evolution:
- Zero Twist ( $0^\circ$ ): The primary gap varies drastically based on relaxation:
  - Fully relaxed (suspended): ~30 meV.
  - Rigid remote substrate: ~9 meV.
  - Completely rigid: ~3 meV.
- Small Angles ( $\sim 0.6^\circ$ ): In the presence of a substrate, the primary gap shows a weak local maximum (enhancement) near $0.6^\circ$ , driven by increased lattice reconstruction.
- Large Angles: The gap decreases gradually but does not close, saturating around 1 meV up to $30^\circ$ .
Secondary Band Gap:
- The secondary (valence band) gap is less sensitive to relaxation schemes.
- It starts at ~12 meV at zero twist but drops to zero for twist angles beyond $\sim 1^\circ$ .
Lattice Reconstruction:
- Relaxation leads to an expansion of the BA-stacking region (most stable) and a reduction of AA and AB regions.
- At small finite angles ( $\sim 0.6^\circ$ ), the system exhibits enhanced local rotation angles and larger atomic displacements compared to the perfectly aligned case, correlating with the observed energy minimum and gap enhancement.
Mass Term and Sign Switching:
- The average mass term accounts for >70% of the primary gap magnitude.
- A finite primary gap of ~1 meV persists across all angles ( $0^\circ$ to $30^\circ$ ) for all sliding configurations.
- The sign of the mass term (favoring occupation of Carbon atop Boron) remains constant up to $30^\circ$ but switches for angles between $30^\circ$ and $60^\circ$ .

5. Significance

Resolving Discrepancies: The study explains why different theoretical models yield different gap values: the degree of substrate relaxation assumed in the model is the primary variable.
Design Guidelines: It suggests that for experimental setups where the substrate is rigid (e.g., thick h-BN on SiO2), the expected band gaps will be significantly smaller (factor of 2–4 reduction) than in suspended or weakly coupled systems.
Stability of Non-Zero Angles: The identification of an energetic stabilization at $\sim 0.6^\circ$ suggests that non-zero twist angles might be naturally stabilized in G/h-BN flakes due to the alignment of moiré and graphene lattice vectors, a phenomenon relevant for the fabrication of moiré superlattices.
Methodological Advancement: The combination of the HTC model and TAPW method provides a computationally efficient yet accurate framework for studying large-scale moiré systems with atomistic detail.