🔬 materials science

Excitons in Large Disordered Boron-Nitride Layer using Linear-Scaling Bethe-Salpeter Simulations

This paper introduces a real-space, linear-scaling Bethe-Salpeter framework that enables efficient excitonic spectroscopy in large, disordered boron-nitride systems, revealing disorder-induced asymmetric broadening, a crossover in peak redshift, and Anderson localization of excitons in structures with up to $10^5$ orbitals.

Original authors: Thomas Galvani, Lorenzo Sponza, Hakim Amara, Sylvain Latil, Stephan Roche

Published 2026-06-30

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Thomas Galvani, Lorenzo Sponza, Hakim Amara, Sylvain Latil, Stephan Roche

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

The Big Picture: A New Way to See "Dancing Pairs" in a Messy World

Imagine a single layer of Boron Nitride (a material like graphene but made of Boron and Nitrogen atoms) as a perfectly organized dance floor. In this dance, an electron (a negatively charged dancer) and a "hole" (a missing dancer, acting like a positive charge) are attracted to each other. They hold hands and spin around one another, forming a pair called an exciton. These pairs are the stars of the show; they determine how the material absorbs light and what color it glows.

In a perfect crystal, everyone knows the steps. The dance floor is symmetrical, and the pairs move in perfect, predictable patterns. Scientists have known how to calculate this for a long time.

The Problem:
Real-world materials are rarely perfect. They have defects, impurities, or "disorder." Imagine the dance floor suddenly getting covered in random obstacles, or the music changing tempo unpredictably. The dancers bump into things, their paths get messy, and the perfect symmetry breaks.

Calculating how these "messy" dancers behave is incredibly hard. Traditional computer methods are like trying to map every single step of every dancer in a stadium of 100,000 people. It takes too much computing power, so scientists usually have to give up and only look at tiny, perfect sections of the floor.

The Solution: A "Linear-Scaling" Shortcut

This paper introduces a new mathematical trick (a framework) that allows scientists to simulate these messy, huge dance floors without needing a supercomputer the size of a city.

The Analogy: The "Sublattice" Map
Think of the Boron Nitride layer as a checkerboard with black squares (Boron) and white squares (Nitrogen).

The Old Way: To understand the dance, you had to calculate the position of every single atom on the board simultaneously.
The New Way: The authors realized they could treat the black squares and white squares as two separate, slightly connected groups. They used a "perturbative decoupling" (a fancy way of saying "a smart approximation") to simplify the math. They mapped the complex electron-hole pairs onto a much simpler, sparse model.

The "Kernel Polynomial Method" (KPM): The Flashlight
Usually, to find the energy of these excitons, you have to solve a massive equation that requires "diagonalizing" a giant matrix (finding the exact roots of a huge polynomial). This is like trying to find the exact height of every person in a crowd by measuring them one by one.

Instead, this method uses the Kernel Polynomial Method. Imagine shining a flashlight through the crowd. You don't need to measure everyone individually; you just need to see how the light scatters to understand the overall shape and density of the crowd. This allows them to calculate the "absorption spectrum" (how the material reacts to light) with a cost that grows linearly with the size of the system. They can now simulate systems with 100,000 atoms (10^5 orbitals), which was previously impossible.

What They Discovered: How Disorder Changes the Dance

Using this new tool, they simulated a Boron Nitride layer with "Anderson disorder" (random energy variations, like random bumps on the dance floor). Here is what they found:

1. The "Asymmetric Blur"
In a perfect crystal, the light absorption peak is a sharp, symmetrical spike. When disorder is added, the peak doesn't just get wider; it gets asymmetrically broadened.

The Metaphor: Imagine a sharp bell sound. If you add disorder, the bell doesn't just get quieter; the sound drags out more on one side than the other. The paper found that the "tail" of the light absorption stretches more toward the blue (higher energy) side because the disorder mixes the perfect dancers with higher-energy, messy ones, but there are no "lower-energy" dancers to mix with.

2. The Redshift (The Slow-Down)
As the disorder gets stronger, the main peak of light absorption moves to lower energies (a "redshift").

The Metaphor: Think of a runner on a track. In a perfect track, they run at a steady speed. If the track gets bumpy (disorder), the runner slows down.
The Twist: At first, as the bumps get slightly worse, the runner slows down slowly (quadratic relationship). But once the bumps get really bad, the runner slows down much faster (linear relationship). The paper found a "crossover" point where this behavior changes.

3. The "Center of Mass" Gets Stuck (Localization)
This is a crucial finding. The electron and hole are still holding hands tightly (the pair remains compact), but the pair itself stops moving freely across the material.

The Metaphor: Imagine a couple dancing in a large ballroom. In a perfect room, they can glide across the whole floor. In a messy room full of random obstacles, they are forced to stay in one small corner, spinning in place. They are still a tight couple, but they are localized. The disorder traps the "center of mass" of the pair in a specific spot.

Why This Matters (According to the Paper)

The authors state that this method opens the door to studying "excitonic physics" in materials that are too big, too messy, or too complex for old methods. Specifically, they mention:

Disordered monolayers: Materials with random defects.
Moiré superlattices: Patterns created when stacking layers at weird angles.
Quasicrystals: Materials with complex, non-repeating patterns.
Structurally complex quantum materials: Any material where the usual rules of symmetry don't apply.

They emphasize that this approach allows for statistical investigations (running the simulation hundreds of times to see the average behavior of disorder) and extends calculations to systems with millions of atoms, which was previously out of reach for standard "ab initio" (first-principles) methods.

In Summary:
The paper presents a new, efficient mathematical "flashlight" that lets scientists see how light interacts with massive, messy materials. They used it to show that when you mess up a perfect crystal, the light-absorbing "dancers" don't just get blurry; they change their rhythm, slow down in a specific way, and get trapped in small corners, all while staying tightly linked together.

Technical Summary: Excitons in Large Disordered Boron-Nitride Layer using Linear-Scaling Bethe-Salpeter Simulations

Problem Statement
While many-body perturbation theory (combining $GW$ corrections and the Bethe-Salpeter equation, BSE) has achieved high accuracy for crystalline materials like hexagonal boron nitride (hBN), it faces significant bottlenecks when applied to atomistically large, disordered, or structurally complex systems (e.g., moiré superlattices, quasicrystals, and functionalized nano-objects). Conventional $ab$ initio Green's function methods rely on reciprocal-space calculations and full diagonalization, which scale poorly with system size. This limits the ability to study optical spectra in systems where translational symmetry is broken and where the interplay between electron-hole attraction, local registry, and disorder is critical. Existing linear-scaling approaches, such as those using Wannier functions, often require explicit eigenstate constructions or pristine-cell projections that are not feasible for large disordered cells.

Methodology
The authors introduce a real-space, linear-scaling framework to compute excitonic spectroscopy in large disordered systems. The methodology proceeds through the following steps:

Sublattice-Resolved Perturbative Decoupling: Starting from a sparse tight-binding Hamiltonian separable into Boron ( $\Lambda_B$ ) and Nitrogen ( $\Lambda_N$ ) sublattices, the authors treat the inter-sublattice coupling via second-order perturbation theory. This constructs approximate effective single-particle Hamiltonians for electrons and holes, mapping the problem onto localized orbitals within their respective sublattices.
Sparse Real-Space BSE Mapping: The Bethe-Salpeter Hamiltonian (BSH) is mapped onto a sparse real-space Hamiltonian for localized electron-hole pairs. The basis consists of localized excitations $|p, n\rangle$ , where $p$ and $n$ denote hole and electron positions.
Hilbert Space Truncation: To ensure linear scaling, the authors introduce a cutoff radius $R_{cut}$ . Pairs with an electron-hole separation $|R| > R_{cut}$ are truncated from the Hilbert space. This reduces the dimension of the BSH to scale as $O(N_{orb} R_{cut}^d)$ , where $N_{orb}$ is the number of orbitals and $d$ is the dimension.
Kernel Polynomial Method (KPM): Instead of diagonalizing the large sparse BSH, the authors employ the Kernel Polynomial Method to compute the imaginary part of the dielectric function ( $\varepsilon_2$ ). This allows for the calculation of absorption spectra with a computational cost scaling linearly with the system size ( $O(N)$ ), avoiding the need for full eigenstate determination.
Disorder Implementation: The method is applied to Anderson-disordered monolayer hBN, where diagonal disorder is added to the onsite potentials. The disorder term in the BSH is derived as the difference in onsite energies between the electron and hole sites.

Key Contributions and Results
The authors applied this framework to Anderson-disordered monolayer hBN supercells containing up to $10^5$ orbitals (approximately $10^4$ atoms), averaging over hundreds of disorder realizations. The key findings include:

Disorder-Induced Asymmetric Broadening: The method reveals that disorder causes an asymmetric broadening of bright exciton peaks. The broadening is stronger toward the blue (higher energy) side. This is attributed to the hybridization of the lowest bright states (located at the bottom of the excitonic band) with higher-energy states, while no states are available below the band edge to hybridize with.
Redshift Crossover: The main absorption peak exhibits a redshift that transitions from a quadratic dependence on disorder strength ( $W_0^2$ ) at low disorder to a linear dependence ( $W_0$ ) at higher disorder values. This behavior is consistent with level repulsion from states above the band edge.
Activation of Dark States: In the pristine case, optical activity is restricted to discrete states in the $Q=0$ sector. Disorder breaks translational invariance, folding states from different $Q$ sectors into the supercell's $Q=0$ sector. This activates a continuum of optically active states, effectively "dark" in the pristine limit, contributing to the spectral weight.
Anderson Localization of the Center of Mass: Real-space analysis of the excitonic eigenfunctions shows that while the relative electron-hole separation (compactness) remains largely unchanged (preserving the tight binding of the pair), the center of mass of the exciton undergoes Anderson localization. The localization length scales inversely with the square of the disorder strength ( $\sim 1/W_0^2$ ) for $W_0 > 0.1$ eV.

Significance and Claims
The paper claims to provide a scalable, atomistic framework that extends excitonic calculations beyond the reach of conventional $GW$-BSE methods. By mapping the BSE onto a sparse real-space Hamiltonian and utilizing KPM, the approach enables optical spectroscopy for:

Large-scale disordered systems (up to millions of atoms in principle).
Moiré superlattices and quasicrystalline approximants where symmetry cannot be relied upon.
Structurally complex quantum materials.

The authors emphasize that this method opens a path for investigating exciton engineering in materials with intrinsic broken translational symmetry, such as transition-metal dichalcogenides (TMDs) and BN-based quasicrystals. They specifically note the potential to study exciton diffusion limited by static disorder, a regime previously difficult to access with standard diagonalization techniques. The work is presented as a route to solving the BSE with atomic resolution using light computational schemes, bridging the gap between accuracy and computational workload for large, disordered structures.

The Big Picture: A New Way to See "Dancing Pairs" in a Messy World

The Solution: A "Linear-Scaling" Shortcut

What They Discovered: How Disorder Changes the Dance

Why This Matters (According to the Paper)

Technical Summary: Excitons in Large Disordered Boron-Nitride Layer using Linear-Scaling Bethe-Salpeter Simulations

More like this