Branched Flow as a Local Signature of a Random Potential

Bo Liu, Eric J. HellerarXiv 2013 · arXiv

Abstract

In classically chaotic systems, small differences in initial conditions are exponentially magnified over time. However, experimental observations of electron flux from a quantum point contact (QPC) in two-dimensional electron gases (2DEGs) reveal that the resulting "branched flow" patterns remain remarkably invariant to changes in initial conditions. While previously conjectured to be a purely quantum effect, this paper demonstrates that this stability arises from the adiabatic nature of the QPC and can be reproduced both classically and quantum mechanically. We show that branches are local signatures of the random potential, with the QPC merely determining their relative illumination.

Introduction

Electrons traveling through a 2DEG do not move in straight lines; instead, they form intricate, tree-like patterns known as branched flow. This phenomenon occurs as electrons navigate a random background potential, which classically should be chaotic. According to chaos theory, a minor shift in the starting position—the "Butterfly Effect"—should lead to a completely different trajectory. Surprisingly, experiments show that shifting the electron injector (the QPC) by a small amount leaves the branch locations virtually unchanged. This paper investigates whether this robustness is a unique quantum signature or a more fundamental property of the system's geometry and source.

Methodology: Bridging Quantum and Classical Perspectives

To analyze the stability of branched flow, the authors utilize a dual-track approach:

  1. Wavepacket Dynamics: The evolution of electron wavepackets is simulated over time. A similarity metric is defined to calculate the overlap between two different wavepackets as they propagate through the random potential.
  2. Wigner Quasiprobability Distribution: This mathematical framework maps quantum states onto phase space (position and momentum). This allows for a direct comparison between quantum results and classical trajectory simulations.
  3. Numerical Simulations: The study employs synthetic numerical simulations of 2DEGs, modeling a random potential background and a harmonic confinement potential for the QPC, tuned to match experimental parameters.
Figure 3: Phase space distributions (a, b) and the resulting branched flow patterns (c, d). The similarity in branch locations despite variations in the source highlights the robustness of the flow.

The Role of the Quantum Point Contact

A key insight of this work is the function of the QPC as an adiabatic guide. Even if two electron wavepackets start with minimal spatial overlap, the process of being squeezed through the QPC's narrow channel ensures they emerge with highly similar momentum distributions. By the time the electrons encounter the random potential, their quantum overlap can increase from less than 5% to over 85%.

Furthermore, the branches themselves are identified as caustics—regions in space where multiple classical paths naturally converge. These caustics are a "local signature" of the underlying random potential landscape. The QPC (the source) determines which branches are "lit up" (their relative intensity), but the landscape itself dictates where those branches are located.

Results and Discussion

The simulations reveal a high degree of correlation between shifted sources. Shifting the QPC by one Fermi wavelength ($\lambda_F$) resulted in a quantum overlap of 85% and a classical phase space correlation of 79% after a short travel distance.

Figure 1: Comparison of branched flow patterns (a, b) and a schematic of the potential landscape (c). The branch structure remains consistent even when the source is perturbed.

Visual evidence from the simulations confirms that while the relative strength of individual branches may fluctuate when the injector is moved, the physical locations of the strongest branches remain identical, even at distances of 70 correlation lengths from the source. This consistency across both classical and quantum models debunks the notion that branched flow stability is a purely non-classical phenomenon.

Conclusion

The observed stability of branched flow in 2DEGs is not a mysterious quantum effect but a consequence of the QPC's role in preparing the electron flux and the geometric nature of caustics in random potentials. This understanding opens doors for "branch management"—the potential to steer electron flow in nanodevices—and suggests that branch patterns could serve as a diagnostic tool to map invisible potential landscapes within semiconductors.

Summary updated

Abstract

In classically chaotic systems, small differences in initial conditions are exponentially magnified over time. However, it was observed experimentally that the (necessarily quantum) pattern of electron flux from a quantum point contact (QPC) traveling over a random background potential in two-dimensional electron gases(2DEGs) remains substantially invariant to large changes in initial conditions. Since such a potential is classically chaotic and unstable to changes in initial conditions, it was conjectured that the origin of the observed stability is purely quantum mechanical, with no classical analog. In this paper, we show that the observed stability is a result of the physics of the QPC and the nature of the experiment. We show that the same stability can indeed be reproduced classically, or quantum mechanically. In addition, we explore the stability of the branched flow with regards to changes in the eigenmodes of quantum point contact.
BibTeX
@misc{liu2013branched,
  author = {Bo Liu and Eric J. Heller},
  title = {{Branched Flow as a Local Signature of a Random Potential}},
  year = {2013},
  eprint = {1309.1814},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/1309.1814},
}

Loading related content...

Figure