Review of the tight-binding method applicable to the properties of moiré superlattices

This review provides a comprehensive theoretical and practical guide to atomistic tight-binding methods and numerical techniques for modeling the electronic, transport, and optical properties of various moiré superlattices, while also clarifying their connection to effective low-energy continuum models.

The Gist

Imagine you have two sheets of transparent, honeycomb-patterned plastic (like graphene). If you stack them perfectly on top of each other, they look like a single sheet. But if you rotate one sheet slightly, or stretch it a tiny bit, the patterns don't line up anymore. Instead, they create a giant, swirling interference pattern called a Moiré superlattice.

Think of it like holding two window screens up to the light and twisting one. You see a giant, slow-moving wave pattern appear that is much larger than the individual holes in the screens. In the world of atoms, these "giant waves" are where some of the most magical and strange physics happens, like electricity flowing without resistance (superconductivity) or materials becoming magnetic.

However, studying these giant atomic waves is a nightmare for computers. Because the pattern is so large, a single "unit" of this pattern contains thousands of atoms. Trying to calculate the behavior of every single atom in this giant crowd is like trying to predict the movement of every person in a stadium by asking each one individually—it takes too long and requires too much memory.

This paper is a guidebook for a specific shortcut called the Tight-Binding (TB) method. Here is how the paper explains it, using simple analogies:

1. The Problem: Too Many Atoms

The paper notes that while we have powerful tools to study small groups of atoms (like Density Functional Theory, or DFT), they are too slow for these giant Moiré patterns. On the other hand, simple mathematical models (Continuum models) are fast but miss the tiny details, like how the atoms physically shift and relax their positions.

2. The Solution: The Tight-Binding "Neighborhood" Map

The Tight-Binding method is like a neighborhood map. Instead of calculating the physics of the entire stadium at once, it only looks at how one atom interacts with its immediate neighbors (the people sitting right next to you).

• How it works: It assumes an atom's behavior is mostly determined by who its neighbors are and how far away they are.
• Why it's great: It keeps the detail of the individual atoms (so it can see if the atoms are squished or stretched) but is fast enough to handle thousands of them. It's the "Goldilocks" zone: not too simple, not too slow.

3. The Toolkit: Different Maps for Different Materials

The paper reviews how to build these "neighborhood maps" for three main types of materials:

• Graphene (The Carbon Honeycomb): The map is relatively simple, focusing on how electrons hop between carbon atoms. The paper shows that by tweaking the "distance" between atoms in the map, scientists can predict exactly when the material becomes a "magic angle" superconductor.
• TMDs (Transition Metal Dichalcogenides): These are like complex sandwiches with metal and other elements. The map here needs to be much more detailed (using 11 different types of "orbitals" or electron paths) to get the physics right.
• hBN (Hexagonal Boron Nitride): This is often used as a smooth bed for the other materials. The paper explains how to map the interaction between the carbon atoms of graphene and the boron/nitrogen atoms of this bed.

4. Handling the Math: The "Random Walk" Trick

When the Moiré pattern gets huge (containing millions of atoms), even the neighborhood map is too big to solve directly. The paper introduces a clever trick called Linear-Scaling Methods (like the Kernel Polynomial Method).

• The Analogy: Imagine you want to know the average height of everyone in a stadium. You don't need to measure everyone. Instead, you pick a few random people, measure them, and use a statistical formula to guess the average for the whole crowd.
• The Result: This allows scientists to simulate materials with millions of atoms on a standard computer, calculating things like how light interacts with the material or how electricity flows.

5. The "Magic" of Relaxation

One of the paper's key points is that atoms aren't static statues; they wiggle and settle into comfortable positions (relaxation).

• The Analogy: Imagine a crowd of people standing in a grid. If you twist the grid, the people in the middle might huddle closer together to save space, while those on the edges spread out.
• The Finding: The Tight-Binding method is special because it can account for this "huddling." The paper shows that if you ignore this relaxation, you get the wrong physics. If you include it, you can accurately predict the "flat bands" (energy levels where electrons get stuck and start interacting strongly), which leads to the exotic phenomena like superconductivity.

6. Real-World Examples in the Paper

The authors demonstrate this method with two specific stories:

• The 12-Sided Crystal: They studied a twisted graphene structure that forms a 12-sided (dodecagonal) pattern. Because this pattern doesn't repeat in a simple way, standard math fails. The Tight-Binding method, using the "random walk" trick, successfully predicted how light and electricity behave in this unique shape.
• The Trapped Exciton: They looked at a system where a layer of WSe2 sits on twisted graphene. They showed how the "huddling" of atoms in the graphene creates tiny traps that catch and hold "Rydberg excitons" (a type of excited particle), explaining a specific signal seen in experiments.

Summary

In short, this paper is a manual for building and using a specific type of computer model to understand giant, twisted atomic patterns. It argues that the Tight-Binding method is the best tool for the job because it strikes the perfect balance: it's detailed enough to see individual atoms moving and relaxing, but fast enough to handle the massive size of these Moiré superlattices. It bridges the gap between simple, fast theories and slow, ultra-precise simulations.

Technical Summary

Review of the Tight-Binding Method Applicable to the Properties of Moiré Superlattices

Problem Statement
Moiré superlattices, formed by stacking two-dimensional (2D) materials with relative rotation or lattice mismatch, have emerged as a versatile platform for exploring exotic quantum phenomena, including unconventional superconductivity, correlated insulating states, and topological effects. However, theoretical modeling of these systems faces significant challenges. Unlike angstrom-scale materials, moiré systems contain a large number of atoms (often thousands to millions) due to the long-wavelength interference patterns. Furthermore, their electronic structures are critically dependent on lattice relaxation and atomic reconstruction. While Density Functional Theory (DFT) offers high accuracy, its computational cost renders it inefficient for simulating large-scale moiré superlattices. Conversely, continuum models provide effective low-energy descriptions but lack the atomic resolution necessary to capture local disorder, strain, and specific chemical effects. There is a need for a theoretical framework that bridges the gap between the accuracy of atomistic simulations and the efficiency of continuum models.

Methodology
This review focuses on the Tight-Binding (TB) method as a central tool for modeling moiré materials. The paper systematically categorizes the construction of atomistic TB Hamiltonians for three primary material families: graphene-based, transition-metal dichalcogenide (TMD)-based, and hexagonal boron nitride (hBN)-based moiré superlattices.

1. Hamiltonian Construction:

   • Graphene: The standard approach utilizes a single-particle $p_z$ orbital model. Hopping parameters ($t_{ij}$) are determined using Slater-Koster (SK) relations, which depend on interatomic distances and decay factors. The model incorporates lattice relaxation by modifying hopping terms based on relaxed atomic positions. Electronic interactions are addressed via mean-field approximations, including the Hartree potential (for long-range Coulomb interactions), the Hartree-Fock approximation (including exchange), and the Hubbard-$U$ term (for local correlations). A rescaling strategy is introduced to reduce the computational cost of self-consistent calculations for large supercells.
   • TMDs: Due to the complexity of TMDs, these models typically employ an 11-orbital basis (5 $d$-orbitals from the metal and 3 $p$-orbitals from the chalcogen) per monolayer. Spin-orbit coupling (SOC) is incorporated via on-site terms. Interlayer interactions are modeled by adding hopping terms between specific orbitals (e.g., $p$-$p$ in homobilayers, $p$-$d_{z^2}$ in heterobilayers), with parameters often derived from Wannier transformations of DFT data.
   • hBN: Models for twisted bilayer hBN and graphene/hBN heterostructures utilize $p_z$ orbitals on Boron and Nitrogen atoms. The models account for the onsite energy difference between B and N atoms and employ various SK parametrizations for interlayer hopping, fitted to DFT results.

2. Numerical Techniques:
   To handle the large-scale Hamiltonian matrices (often $N > 10^5$ atoms), the review discusses methods beyond standard diagonalization:

   • Linear-Scaling Methods: The Kernel Polynomial Method (KPM) and the Tight-Binding Propagation Method (TBPM) are highlighted. These stochastic approaches scale as $O(N)$, allowing for the calculation of Density of States (DOS), optical conductivity, and transport properties in systems with millions of atoms.
   • Machine Learning: Recent developments using deep learning (e.g., DeepH, HamGNN, DeepTB) are reviewed for constructing ab initio-quality TB Hamiltonians from training data, accelerating the parameterization process.

3. Connection to Continuum Models:
   The paper details the derivation of low-energy effective continuum models from atomistic TB Hamiltonians. It explains how the moiré potential arises from the Fourier components of the interlayer tunneling. The review emphasizes that while local (momentum-independent) potentials capture the emergence of flat bands, non-local (momentum-dependent) tunneling terms are necessary to reproduce particle-hole asymmetry and other features observed in relaxed TB models.

Key Contributions and Results

• Comprehensive Overview: The paper provides a unified reference for the specific TB Hamiltonians, SK parameters, and relaxation potentials (e.g., LAMMPS potentials) used for graphene, hBN, and TMD moiré systems.
• Validation of TB Models: The review demonstrates that TB models, when properly parameterized and including relaxation, reproduce key experimental and DFT results. For instance, TB calculations of twisted bilayer graphene (TBG) successfully predict the "magic angle" ($\theta \approx 1.05^\circ - 1.1^\circ$) where the Fermi velocity vanishes and flat bands emerge.
• Role of Lattice Relaxation: The authors emphasize that atomic relaxation is not a minor correction but a critical factor. In TBG, relaxation leads to the formation of AA and AB domains, significantly altering the band structure and inducing particle-hole asymmetry. In TMDs and hBN, relaxation similarly dictates the flatness of bands and the magnitude of band gaps.
• Handling Interactions: The review illustrates how mean-field TB approaches (Hartree, Hartree-Fock, Hubbard) can be used to study correlated phases, such as Mott insulating states and ferromagnetism, in flat-band regimes.
• Case Studies: Two specific examples are provided to demonstrate the utility of TB methods:
  1. Dodecagonal Graphene Quasicrystal: Using TBPM on a system with 10 million atoms, the study reveals distinct peaks in the DOS and optical conductivity attributed to interlayer interactions, and predicts new Landau levels in magnetic fields.
  2. Rydberg Moiré Excitons in WSe$_2$/TBG: By combining TB with molecular dynamics, the authors explain the spatial confinement of Rydberg excitons in WSe$_2$ by the charge accumulation in the AA regions of the underlying relaxed TBG, matching experimental reflectance and photoluminescence spectra.

Significance and Scope
The paper positions the Tight-Binding method as a central, practical tool for the theoretical study of moiré superlattices. Its significance lies in its unique balance: it retains the atomic-level resolution required to model lattice relaxation, strain, and chemical specificity, while remaining computationally efficient enough to simulate realistic supercells containing thousands to millions of atoms.

The authors state that this review serves as a "theoretical and practical guide" for researchers applying TB methods to moiré systems. It bridges the gap between ab initio DFT calculations and effective continuum models, offering a systematic way to include many-body interactions and external fields. The paper concludes by noting that while TB is powerful, future work is needed to extend these methods to non-hexagonal lattices, improve linear-scaling algorithms for non-orthogonal bases, and utilize machine learning for the automated discovery of new moiré materials. The review does not claim to solve all correlation problems but asserts that TB provides the necessary accurate starting point for understanding the origin of flat-band-related phenomena.