🔬 mesoscale physics

Plasmons in twisted bilayer graphene across dispersive and flat bands

This study employs time-dependent density-functional theory to characterize the plasmonic response of twisted bilayer graphene, revealing that large-angle structures preserve Dirac plasmons with geometric scaling while small-angle magic-angle configurations exhibit distinct interband plasmons arising from transitions within quasi-flat bands.

Original authors: Antonio Palamara, Michele Pisarra, Antonello Sindona

Published 2026-09-18

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Antonio Palamara, Michele Pisarra, Antonello Sindona

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 built from sheets of carbon atoms, so thin they are essentially two-dimensional. When scientists stack two of these sheets, called graphene, and twist them slightly relative to one another, they create a new material with surprising properties. This twisting generates a giant, repeating pattern known as a moiré pattern, which acts like a new landscape for electrons moving through the material. Depending on the angle of the twist, this landscape can either let electrons zip around freely or trap them in place, creating a "flat" energy state where they barely move. This ability to control how electrons behave has made these twisted sheets a hot topic for researchers hoping to build better electronics or even understand how superconductivity works. A key question in this field is how these electrons respond when they are jostled by energy, specifically how they move together in waves known as plasmons. Understanding these waves is crucial because they might help explain how electrons pair up to conduct electricity without resistance.

In a recent study, researchers Antonio Palamara, Michele Pisarra, and Antonello Sindona used powerful computer simulations to map out these electron waves in twisted bilayer graphene. They focused on two very different scenarios: one where the layers are twisted at a large angle, allowing electrons to move with high energy, and another where the layers are twisted at a very small angle, creating the flat, trapped energy states. To make the complex calculations possible for the small-angle case, which would normally require a computer to handle an impossibly large number of atoms, the team used a clever workaround. They slightly squeezed the distance between the two layers of carbon. This adjustment allowed them to recreate the flat energy states of the famous "magic angle" without needing to simulate the massive number of atoms usually required for that specific angle. This approach let them compare the two regimes side by side using the same detailed atomic model.

The researchers found that in the large-angle, high-energy scenario, the electrons behave much like they do in a single sheet of graphene. They identified a strong, high-energy wave called the pi plasmon, which is a collective vibration of the electrons that form the chemical bonds in the carbon sheet. When they added extra electrons to the system, a new type of wave emerged, known as a two-dimensional Dirac plasmon. This wave follows a predictable pattern based on the angle of the twist and the amount of added charge, behaving very much like a wave on a flat pond. However, they also found that as they added more electrons, the behavior became more complicated and less uniform, suggesting that the simple rules governing the electrons start to break down when the energy levels get too high.

In the small-angle, flat-band scenario, the story changes significantly. Here, the electrons are crowded into a narrow energy range, and the researchers discovered that the expected low-energy wave did not appear as a clear, distinct signal. Instead, they observed a blurry, broad excitation that did not have the sharp characteristics of a well-defined wave. This suggests that in this flat state, the electrons are not moving in a coordinated, collective rhythm in the way scientists had hoped. However, they did find a different, clearer wave at a higher energy level. This wave, which they call an optical plasmon, arises from electrons jumping between different energy levels. By adjusting the energy scale of their simulation, they estimated that this wave would appear in the mid-infrared range of light, a frequency that could be measured in future experiments.

The study concludes that while the flat-band regime offers a unique environment for electrons, it does not automatically produce the clean, collective waves that some theories predicted. The behavior of the electrons is highly sensitive to the exact arrangement of the atoms and the specific energy levels they occupy. The researchers suggest that their findings provide a solid starting point for understanding how these materials might support superconductivity, but they also highlight that the path to a clear answer is complex. The ability to tune the distance between the layers, as they did in their simulation, might offer a new way to control these electronic properties, potentially making it easier to study these materials without needing to twist them to a perfect, microscopic angle. Ultimately, this work paints a more detailed and realistic picture of how electrons dance in these twisted carbon sheets, showing that the dance is far more intricate than previously imagined.

Technical Summary: Plasmons in Twisted Bilayer Graphene Across Dispersive and Flat Bands

Problem Statement
The dynamical dielectric response of twisted bilayer graphene (TBG) is critical for understanding charge dynamics, collective excitations, and potential pairing mechanisms in magic-angle systems. While theoretical models (continuum or tight-binding) have predicted various plasmonic modes, fully atomistic first-principles calculations of the momentum-dependent dielectric response for large-angle and near-magic-angle TBG have been computationally prohibitive. The number of atoms in commensurate supercells scales rapidly with decreasing twist angle (reaching $\sim 10^4$ atoms near the first magic angle $\theta \approx 1.1^\circ$ ), making standard plane-wave (PW) density-functional theory (DFT) and time-dependent DFT (TDDFT) calculations extremely demanding. This study addresses the gap in atomistic descriptions of plasmon spectra as the electronic bands transition from dispersive (large-angle) to quasi-flat (small-angle) regimes.

Methodology
The authors employ time-dependent density-functional theory (TDDFT) within the random-phase approximation (RPA) to calculate the macroscopic dielectric function and the electron energy-loss function (ELOSS).

Structural Models: Three commensurate TBG configurations are analyzed:
1. Large-angle dispersive regimes: $\theta_1 = 21.7868^\circ$ and $\theta_2 = 13.1736^\circ$ at the reference interlayer distance $d_0 = 3.349$ Å.
2. Flat-band regime: A smaller angle $\theta_7 = 4.4085^\circ$ treated at a reduced "magic distance" $\tilde{d}_7 = 2.587$ Å. This "magic-distance configuration" is a constrained model designed to reproduce the quasi-flat band manifold of magic-angle TBG without requiring the prohibitively large supercell of the actual $\theta \approx 1.1^\circ$ structure.
Computational Framework: Calculations utilize the Quantum ESPRESSO suite with PBE functionals and norm-conserving pseudopotentials. The dielectric response is computed using a custom in-house framework. To handle the large supercell of the $\theta_7$ configuration (676 atoms), the authors employ a slab-truncated Coulomb kernel to isolate the 2D response and truncate the kinetic-energy cutoff for wavefunctions during the response calculation to manage memory constraints.
Analysis: The study focuses on the loss function $E_{LOSS}(q, \omega) = -\text{Im}[\epsilon_M(q, \omega)^{-1}]$ . Plasmonic character is assessed by examining the real and imaginary parts of the macroscopic dielectric function at loss resonances, specifically looking for zero-crossings in $\text{Re}[\epsilon_M]$ accompanied by small $\text{Im}[\epsilon_M]$ .

Key Results

Dispersive-Band Regime (Large Angles):
- Intrinsic Response: At charge neutrality, the $\theta_1$ and $\theta_2$ structures retain the semimetallic Dirac character of graphene. The loss spectrum is dominated by a strongly dispersive $\pi$ plasmon ( $\sim 4.4$ eV at long wavelengths), redshifted relative to graphite, arising from $\pi \to \pi^*$ transitions. Weaker features correspond to van Hove singularities (VHS) induced by moiré folding.
- Doped Response: Electron doping activates a two-dimensional Dirac plasmon (2DP) with $\sqrt{q}$ dispersion. The study establishes an approximate geometric scaling relation between the $\theta_1$ and $\theta_2$ configurations: $\omega_{2DP}^{\theta_1}(q_1) \approx s_{1/2} \omega_{2DP}^{\theta_2}(q_2)$ , where $s_{1/2} \approx \sqrt{19/7}$ is the ratio of moiré lattice constants. This scaling holds well within the linear Dirac cone region but breaks down at higher doping levels due to nonlinear and anisotropic band effects.
- Acoustic-like Branches: Acoustic-like excitation (AE) branches are observed embedded within the single-particle excitation continuum. These are heavily Landau-damped and highly sensitive to momentum direction, preventing their unambiguous identification as well-defined layer-antisymmetric acoustic plasmons.
Flat-Band Regime (Magic-Distance Configuration):
- Electronic Structure: Reducing the interlayer distance to $\tilde{d}_7$ creates a narrow, quasi-flat manifold of four bands around the Fermi level, resulting in a pronounced density of states (DOS) peak.
- Low-Energy Excitations: At intrinsic filling, a low-energy feature (Flat-Band Excitation, FBE) is observed. However, dielectric analysis reveals no clear zero-crossing in $\text{Re}[\epsilon_M]$ at the loss maximum, suggesting this is not a well-defined plasmon but rather a collective-like excitation driven by intra- and interband transitions involving the quasi-flat manifold and high-DOS regions.
- Interband Plasmon: A distinct, finite-energy peak (Optical Plasmon, OP) is identified at $\omega_{OP} \approx 0.40$ eV. This mode exhibits a clear dielectric zero-crossing and is attributed to interband transitions between the quasi-flat manifold and neighboring high-DOS regions.
- Filling Dependence: The FBE vanishes when the quasi-flat bands are fully filled, confirming its dependence on partial occupation. The OP persists across all filling conditions (intrinsic, electron-doped, hole-doped) because the relevant interband transition channels remain available, though its energy and spectral weight shift.

Significance and Claims
The paper claims to provide the first fully atomistic TDDFT-RPA characterization of the dielectric response in TBG across both dispersive and flat-band regimes.

Methodological Contribution: The study demonstrates that the "magic-distance" construction offers a computationally tractable route to estimate plasmonic energy scales of magic-angle TBG without simulating the massive $\sim 10^4$ atom supercells required for the actual magic angle at the reference distance.
Physical Insights: The results highlight a hierarchy in charge dynamics: robust graphene-derived responses (like the $\pi$ plasmon) persist in large-angle TBG, while the flat-band regime is governed by the specific topology of the quasi-flat manifold and its coupling to high-DOS regions.
Implications for Superconductivity: The authors suggest that the identified collective modes, particularly the interband plasmon in the flat-band regime, provide a starting point for assessing plasmon-mediated contributions to Cooper pairing in magic-angle TBG. They propose that incorporating these frequency-dependent screened interactions into superconducting DFT could clarify the interplay between plasmonic and phonon-mediated pairing mechanisms.

The paper remains modest regarding experimental feasibility, noting that the structural compression required for the magic-distance model must be experimentally verified and that structural relaxation effects are not included in the current constrained geometry.

More like this