🔬 mesoscale physics

Orbital magnetization and magnetic susceptibility of interacting electrons

This paper rigorously derives formulas for the orbital magnetization and magnetic susceptibility of interacting electrons within the Hartree-Fock approximation, revealing that while magnetization follows the non-interacting form with Hartree-Fock replacements, susceptibility includes a distinct interaction-induced contribution, a finding validated through tests on an interacting Rashba model.

Original authors: Jian Kang, Minxuan Wang, Oskar Vafek

Published 2026-02-16

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Jian Kang, Minxuan Wang, Oskar Vafek

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: The "Crowded Dance Floor" Problem

Imagine a crowded dance floor where everyone is dancing to music. In a simple scenario (non-interacting electrons), you can predict how the dancers move just by looking at the music and the floor layout. If you spin the room (apply a magnetic field), you can easily calculate how the whole crowd starts to swirl or "magnetize."

However, in the real world, electrons are like a chaotic, energetic crowd. They don't just dance to the music; they constantly bump into, push, and pull on each other. This is the electron-electron interaction.

When scientists try to calculate how this chaotic crowd reacts to a magnetic field, it gets incredibly messy. Usually, to do this math, you have to simulate the crowd while the room is spinning. But if the room is spinning even a tiny bit, the math becomes a nightmare because the "bumping" rules change in complex ways. It's like trying to calculate the traffic flow of a city while simultaneously simulating a massive earthquake happening in the middle of the rush hour.

The Breakthrough: A Shortcut Through the Chaos

The authors of this paper (Jian Kang, Minxuan Wang, and Oskar Vafek) found a brilliant shortcut. They asked: "Do we really need to simulate the spinning room to know how the crowd will swirl?"

Their answer is no.

They developed a new mathematical recipe that allows you to calculate the Orbital Magnetization (how much the crowd wants to spin) and the Magnetic Susceptibility (how easily they can be made to spin) by only looking at the crowd when the room is perfectly still (zero magnetic field).

The Two Main Results

The paper delivers two distinct formulas, which behave differently:

1. The Magnetization Formula (The "Easy" Part)

The Analogy: Imagine you want to know how much a crowd wants to spin. The authors found that if you just update your map of the dance floor to account for how the dancers push each other while standing still, you can use the exact same simple formula you would use for a non-interacting crowd.
The Takeaway: The "interaction" (the pushing) is already baked into the map. Once you have the "Hartree-Fock" map (the self-consistent solution), you don't need to do anything special. The formula looks exactly the same as the simple version, just with a more complex map.

2. The Susceptibility Formula (The "Surprise" Part)

The Analogy: Now, imagine you want to know how sensitive the crowd is to a spin. If you just use the updated map, you get it wrong. The crowd has a hidden "groupthink" reaction. When you try to spin them, they don't just react individually; they react as a group in a way that creates a new force.
The Takeaway: The formula for susceptibility has an extra term. This is a "bonus" contribution that only exists because the electrons are interacting. You cannot get this by just swapping the map; you have to add this specific "interaction penalty" to the calculation. This is the first time such a rigorous formula has been derived for interacting systems.

Why This Matters: The "Twisted Graphene" Connection

Why do we care about this? Because of a new class of materials called Van der Waals heterostructures (like twisted layers of graphene or MoTe2).

The Problem: In these materials, electrons interact very strongly. They create "orbital magnets" (materials that generate their own magnetic fields without needing iron). To study them, scientists usually have to apply a magnetic field to see what happens.
The Difficulty: In these materials, even a tiny magnetic field (like 1 Tesla) is actually a huge disturbance to the quantum mechanics. It's like trying to hear a whisper in a hurricane. Simulating the math with the field on is computationally impossible for many systems.
The Solution: This new method lets scientists run their simulations without the magnetic field (which is easy and fast) and then use the new formulas to predict exactly what would happen if they did apply the field.

The "Rashba" Test Drive

To prove their theory works, the authors tested it on a specific model called the "interacting Rashba model."

They calculated the properties using their new "zero-field" shortcut.
They then did the hard, messy calculation with the magnetic field turned on (which is very difficult).
The Result: The two matched perfectly. The shortcut worked.

Summary in a Nutshell

Think of this paper as inventing a crystal ball for magnetic materials.

Previously, to predict how a material with interacting electrons would react to a magnet, you had to build a complex, expensive, and slow simulation of the material inside a magnet.

Now, thanks to this paper, you can just look at the material in a quiet room (zero magnetic field), take a snapshot of how the electrons are arranging themselves, and plug that snapshot into a new set of equations. These equations will tell you exactly how the material will behave in a magnetic field, saving massive amounts of computer power and opening the door to designing new types of magnetic switches and quantum computers.

The Golden Rule: For the "swirl" (Magnetization), just update the map. For the "sensitivity" (Susceptibility), update the map and add a special "crowd reaction" bonus.

1. Problem Statement

Recent experiments on van der Waals heterostructures (e.g., twisted bilayer graphene, MoTe $_2$ ) have revealed that electron-electron (e-e) interactions can induce spontaneous orbital magnetism, leading to phenomena like the quantum anomalous Hall effect and hysteretic valley switching. These states possess non-zero magnetic moments, making them sensitive to external magnetic fields ( $B$ ).

However, calculating the orbital magnetization ( $M$ ) and orbital magnetic susceptibility ( $\chi$ ) for interacting systems in the limit of vanishing magnetic field ( $B \to 0$ ) is theoretically and computationally challenging:

Computational Complexity: Standard approaches require solving the self-consistent Hartree-Fock (HF) equations at a non-zero magnetic field ( $B \neq 0$ ). In this regime, interactions occur within magnetic subbands (Landau levels), and the number of subbands grows as $B$ decreases, making calculations prohibitively expensive.
Theoretical Gap: While formulas for $M$ and $\chi$ exist for non-interacting systems, generalizing these to interacting systems has been difficult. Previous attempts often relied on perturbative treatments of spatially varying $B$ fields or gauge-invariant Green's functions, which are complex to implement for self-consistent interacting systems.

The authors aim to develop a rigorous, computationally efficient method to derive $M$ and $\chi$ for interacting electrons using only the self-consistent solution at zero magnetic field ( $B=0$ ).

2. Methodology

The authors employ the self-consistent Hartree-Fock (HF) approximation at zero temperature. Their derivation proceeds through the following key steps:

Hamiltonian Formulation: They start with a general interacting Hamiltonian including a single-particle term (with minimal coupling to vector potential $\mathbf{A}$ ) and a two-body interaction term $V$ .
Green's Function Decomposition: To handle the magnetic field dependence, they decompose the Green's function $G(\mathbf{r}, \mathbf{r}'; z)$ into a phase factor $e^{i\Phi(\mathbf{r}, \mathbf{r}')}$ (associated with the line integral of $\mathbf{A}$ ) and a gauge-invariant part $\tilde{G}$ . This allows the total energy to be expressed in terms of the density matrix $\tilde{P}$ , which carries the $B$ -dependence.
Variational Derivation: They calculate the derivative of the grand potential $\Omega$ with respect to $B$ at $B=0$ . This involves differentiating the HF Hamiltonian and the density matrix.
Operator Identities: By differentiating the defining equation of the Green's function with respect to $B$ and setting $B=0$ , they derive an operator identity (Eq. 14) relating the derivative of the Green's function to the commutators of the HF Hamiltonian with the position operator (velocity operator) and the derivative of the HF potential.
Cancellation of Interaction Terms: A crucial step involves showing that the term containing the derivative of the HF potential ( $d\tilde{V}^{HF}/dB$ ) vanishes when integrated over the contour enclosing the Fermi sea for the magnetization calculation. This is due to the coincidence of poles in the Green's functions and the factor $(E_n - \mu)$ .
Bloch Representation: For periodic systems, they express the results in terms of Bloch wavefunctions $|n, \mathbf{k}\rangle$ , utilizing the periodic part $u_{n,\mathbf{k}}$ and the crystal momentum $\mathbf{k}$ .

3. Key Contributions and Results

A. Orbital Magnetization ( $M$ )

The authors derive a formula for the orbital magnetization of interacting electrons (Eq. 20).

Main Result: The formula for $M$ $M$ in the interacting HF case has the exact same form as the non-interacting case, provided that:
1. The Hamiltonian is replaced by the self-consistent HF Hamiltonian at $B=0$ .
2. The wavefunctions are the eigenstates of this self-consistent HF Hamiltonian.
Implication: One does not need to solve the system at $B \neq 0$ . The magnetization can be computed entirely from the $B=0$ self-consistent solution.
Convergence Note: The authors emphasize that while the formula looks like a band sum, in moiré systems, one must include "remote bands" in the $B=0$ calculation to ensure the self-consistent potential converges, even if the final $M$ is dominated by "active bands" near the Fermi energy.

B. Orbital Magnetic Susceptibility ( $\chi$ )

The derivation for susceptibility (Eq. 21) reveals a more complex structure:
$\chi = \chi_{\text{band}} + \chi_{\text{int}}$

$\chi_{\text{band}}$ : This term follows the non-interacting formula but uses self-consistent HF quantities (Green's function, velocity operator, inverse effective mass).
$\chi_{\text{int}}$ : This is a new, interaction-induced contribution (Eq. 23) that has no analogue in non-interacting theory. It arises from the explicit dependence of the HF potential on the magnetic field (via the density matrix) and cannot be obtained by simply replacing non-interacting parameters with HF ones.

C. Numerical Validation

To test their theory, the authors applied the formulas to an interacting Rashba-like continuum model with contact interactions.

Method: They solved the model exactly at finite $B$ (using Landau levels) and compared the results with the predictions from their $B=0$ formulas.
Outcome: The grand potential $\Omega(B)$ calculated directly at finite $B$ matched the linear ( $M$ ) and quadratic ( $\chi$ ) expansions derived from the $B=0$ self-consistent solution in the small $B$ limit. This validates the rigorous derivation.

4. Significance

Computational Efficiency: The proposed method eliminates the need to solve self-consistent equations at non-zero magnetic fields. Since the number of magnetic subbands explodes as $B \to 0$ , this approach is significantly more efficient and stable for studying orbital magnets in the low-field regime.
Theoretical Rigor: It provides a rigorous proof that the orbital magnetization formula retains its non-interacting form under HF approximation, clarifying a long-standing ambiguity.
New Physics in Susceptibility: The identification of the interaction-induced term $\chi_{\text{int}}$ highlights that orbital susceptibility is fundamentally different from magnetization in interacting systems; it contains contributions that are purely interaction-driven and cannot be captured by single-particle band theory alone.
Applicability: The framework is applicable to periodic systems (moiré heterostructures) and can be extended to aperiodic systems with disorder or open boundaries, as the core operator identities do not rely on periodicity.

In summary, this work provides a powerful, computationally tractable framework for predicting and understanding interaction-induced orbital magnetism in modern 2D materials, bridging the gap between complex many-body interactions and observable magnetic properties.