Quantum oscillation in the Hofstadter Model

Wenhu Xu, Li‐Ping Yang, Mingpu Qin, Tao XiangarXiv 2008 · arXiv

Abstract

We have developed a novel quantum transfer matrix method to study thermodynamic properties of the Hofstadter model on square lattices. This method overcomes the finite size effect and allows the magnetization and other thermodynamic quantities to be accurately and efficiently calculated. The interplay between the lattice potential and the applied magnetic field leads to substantial corrections to the conventional de Haas-van Alphen (dHvA) oscillation of magnetization. It yields many paramagnetic peaks or diamagnetic deeps above the background of dHvA oscillation. This correction bears the fractal feature of energy spectra and can be used to probe the hierarchical structure of Hofstadter butterfly by experimental measurements.

Introduction

The Hofstadter butterfly is an iconic fractal pattern representing the energy levels of electrons in a 2D lattice subjected to a magnetic field. While mathematically elegant, it presents significant theoretical and experimental challenges. The system's behavior depends critically on whether the magnetic flux per lattice cell ($\phi$) is a rational or irrational number. Traditional numerical methods, such as Exact Diagonalization or the Bethe Ansatz, struggle as the denominator of the rational flux ($q$) increases, typically failing beyond $q > 1000$. Furthermore, previous studies often focused on the zero-temperature ground state, where the fractal nature renders physical quantities like magnetic susceptibility non-analytic. This research introduces a method to calculate these properties in the thermodynamic limit at finite temperatures, revealing how the fractal structure manifests in measurable physical quantities.

Methodology: The Quantum Transfer Matrix Method

To address the limitations of existing methods, a novel quantum transfer matrix (QTM) approach was developed. Instead of direct diagonalization to find individual energy levels, the method focuses on the partition function. The process involves:

  1. Dimensional Reduction: Decoupling the 2D Hofstadter model into 1D Hamiltonians using the Landau gauge.
  2. Trotter-Suzuki Decomposition: Approximating the partition function by breaking the Hamiltonian exponential into manageable components.
  3. Fermionic Mapping: Expressing local transfer operators via fermion operators.
  4. Matrix Product Simplification: Reducing the system's partition function calculation to the product of $2 \times 2$ matrices.

This approach is highly efficient, with computation time scaling linearly with lattice size ($N_y$). This allows for simulations of massive systems (up to $10^8$ sites), ensuring convergence in the thermodynamic limit.

Results and Discussion

Lattice Corrections to Quantum Oscillations

In a standard electron gas, magnetization oscillates smoothly with the magnetic field—a phenomenon known as the de Haas-van Alphen (dHvA) effect. This study reveals that the underlying lattice introduces "lattice corrections"—subtle extra peaks and dips—onto this smooth background.

Magnetization as a function of $1/\phi$ for different temperatures. The inset highlights the fine structure emerging at lower temperatures.

As shown in Figure 1, temperature acts as a filter for the fractal details. At higher temperatures (e.g., $T=0.1$), thermal fluctuations smear out fine structures, leaving only standard oscillations. As the temperature decreases to $T=0.02$, the "resolution" increases, and the hierarchical sub-cells of the Hofstadter butterfly emerge as distinct magnetic features.

Fractal Fingerprinting in Susceptibility

The nature of the magnetic response is deeply tied to the topology of the energy bands. When the flux $\phi = p/q$ has an odd denominator $q$, the system encounters a van Hove singularity, resulting in a paramagnetic peak. Conversely, an even $q$ corresponds to a Dirac point where the density of states vanishes, leading to a diamagnetic dip.

Magnetic susceptibility $\chi$ vs. flux $\phi$ at $T=0.1$ and $T=0.02$. Labeled peaks and dips correspond to specific rational flux values.

Figure 2 illustrates these "fractal fingerprints" in the magnetic susceptibility ($\chi$). At low temperatures, $\chi$ oscillates wildly, with local maxima and minima aligning perfectly with rational flux values such as $4/13$ (peak) or $3/10$ (dip).

Conclusion

The study demonstrates that quantum corrections to magnetization are large enough to be measured in modern experimental setups, such as semiconductor superlattices or ultra-cold atoms in optical lattices. This provides a direct experimental pathway to "see" the density of states of the Hofstadter butterfly. While the $T \to 0$ limit remains mathematically complex due to infinite oscillations, the QTM method provides a robust framework for exploring these fractal properties at finite temperatures. Future work may extend this method to hexagonal lattices or incorporate many-body effects.

Summary updated

Abstract

We have developed a novel quantum transfer matrix method to study thermodynamic properties of the Hofstadter model on square lattices. This method overcomes the finite size effect and allows the magnetization and other thermodynamic quantities to be accurately and efficiently calculated. The interplay between the lattice potential and the applied magnetic field leads to substantial corrections to the conventional de Haas-van Alphen (dHvA) oscillation of magnetization. It yields many paramagnetic peaks or diamagnetic deeps above the background of dHvA oscillation. This correction bears the fractal feature of energy spectra and can be used to probe the hierarchical structure of Hofstadter butterfly by experimental measurements.
BibTeX
@misc{xu2008quantum,
  author = {Wenhu Xu and Li‐Ping Yang and Mingpu Qin and Tao Xiang},
  title = {{Quantum oscillation in the Hofstadter Model}},
  year = {2008},
  eprint = {0808.0099},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/0808.0099},
}
Figure