Anatomy of the modern theory of orbital magnetism from first-principles: term-by-term analysis in the gauge-covariant formalism

This paper presents a gauge-covariant, first-principles analysis of orbital magnetism across various material classes, demonstrating how the interplay between localized electronic character and band structure-induced Berry phases dictates magnetic responses and suggests new avenues for enhancing orbital magnetism in orbitronics.

The Gist

The Big Picture: What is this paper about?

Imagine you are trying to understand how a magnet works. For a long time, scientists thought the magnetism came mostly from the "spin" of electrons (like tiny tops spinning in place). But recently, a new field called orbitronics has emerged, which looks at the "orbit" of electrons (how they circle around the nucleus, like planets around the sun).

This paper is a "forensic investigation" into how we calculate this orbital magnetism. The authors are asking: Are our old ways of calculating this correct, or do we need a new, more sophisticated map?

They found that the answer depends entirely on where the electrons live.

• If the electrons are "homebodies" (stuck close to the atom), the old way works fine.
• If the electrons are "travelers" (zooming between atoms), the old way fails completely, and you need the new, high-tech map.

---

The Two Competing Maps

To understand the paper, you need to know about the two methods the authors compared:

1. The Old Map: The "Atom-Centered Approximation" (ACA)

The Analogy: Imagine a city where every house has a fence. The old method says: "To count the traffic, we only look at the cars driving inside the fences of individual houses. We ignore the cars driving on the streets between the houses."

• How it works: It calculates magnetism by looking only at the electrons spinning inside a tiny sphere around each atom.
• The Flaw: It assumes electrons stay put. If an electron is zooming from one atom to another (like a delivery driver), this method misses that movement entirely.

2. The New Map: The "Modern Theory" (Berry Phase)

The Analogy: This is a GPS that tracks the entire journey. It doesn't just look at the house; it tracks the car as it drives down the street, turns corners, and interacts with other cars. It accounts for the "twist" in the road (the Berry Phase) that happens when electrons move through the complex landscape of a material.

• How it works: It calculates magnetism by looking at the whole material at once, including the "interstitial" space (the empty space between atoms) and how electrons hop between them.
• The Benefit: It captures the full picture, including the "quantum weirdness" that happens when electrons interfere with each other.

---

The Investigation: Who is right?

The authors tested these two maps on three different types of "neighborhoods" (materials) to see which method gives the correct answer.

Neighborhood 1: The "Homebodies" (d-transition Metals like Iron, Cobalt, Nickel)

• The Vibe: In these metals, the electrons are like introverts. They stay very close to their home atoms. They don't wander far.
• The Result: The Old Map (ACA) works great here! It captures about 70–90% of the magnetism.
• The Takeaway: For these materials, the "fence" method is a good enough approximation. The electrons aren't doing much traveling, so ignoring the streets between houses isn't a big deal.

Neighborhood 2: The "Travelers" (sp metals like Aluminum, Bismuth)

• The Vibe: In these metals, the electrons are extroverts. They are "delocalized," meaning they zoom freely between atoms. They have high kinetic energy and love to travel.
• The Result: The Old Map (ACA) fails miserably. It only captures about 40% of the magnetism. It's like trying to count traffic in a city by only looking at driveways and ignoring the highways.
• The Takeaway: For these materials, you must use the New Map. The magnetism comes almost entirely from the electrons moving between atoms, which the old method completely misses.

Neighborhood 3: The "Twisty Roads" (Transition Metal Dichalcogenides like MoS2)

• The Vibe: These are 2D materials (like a single sheet of graphene). Here, the "roads" (energy bands) are very twisty and close together.
• The Result: The New Map reveals a massive amount of magnetism that the Old Map misses.
• The Secret Sauce: In these materials, the electrons can hop between different energy levels (valence and conduction bands) very easily. This creates a "coherent hybridization"—imagine two dancers perfectly synchronized. This synchronization creates a huge orbital magnetism that is valley-dependent (it changes depending on which "valley" or direction the electron is in).
• The Takeaway: The Old Map sees a flat, boring landscape. The New Map sees a rollercoaster with huge loops, and that's where the real magnetism is hiding.

---

The "J-Decomposition": Breaking it Down

The authors didn't just say "Old vs. New." They broke the New Map down into three ingredients to see exactly what was contributing to the magnetism:

1. The "Atomic" Part (M0): The electrons spinning in place. (This is what the Old Map sees).
2. The "Hybrid" Part (M1 & M2): The electrons mixing and hopping between atoms. (This is what the Old Map misses).

The Discovery:

• In Iron/Nickel, the "Atomic" part is the main ingredient. The "Hybrid" part is just a garnish.
• In Aluminum/Bismuth, the "Hybrid" part is the main course. The "Atomic" part is barely a crumb.
• In MoS2, the "Hybrid" part creates a massive spike in magnetism near the energy gaps, which is crucial for future electronics.

---

Why Does This Matter? (The "So What?")

This paper is a guidebook for the future of Orbitronics.

• If you are building a computer chip using Iron or Nickel: You can stick to the simpler, older methods. They are fast and accurate enough.
• If you are designing next-gen quantum devices using Aluminum, Bismuth, or 2D materials: You cannot use the old methods. You will get the wrong answer, and your device won't work. You must use the "Modern Theory" to harness the power of the Berry Phase.

The Final Metaphor:
Think of orbital magnetism as a symphony.

• The Old Method only listens to the soloists playing in their own rooms (the atoms).
• The New Method listens to the whole orchestra, including how the musicians talk to each other across the stage (the hopping electrons).
• In some songs (Iron), the soloists are enough. In others (Aluminum, MoS2), the magic is entirely in the conversation between the musicians. If you ignore the conversation, you miss the music.

Summary

This paper proves that to understand and control the magnetism of modern materials, we need to stop looking at atoms in isolation and start looking at how electrons dance together across the entire material. The "Modern Theory" is the only way to see the full dance.

Technical Summary

1. Problem Statement

The study addresses critical theoretical and practical challenges in calculating orbital magnetism in solids, specifically the discrepancy between the traditional Atom-Centered Approximation (ACA) and the Modern Theory of Orbital Magnetism.

• The ACA Limitation: The ACA calculates orbital angular momentum (OAM) by integrating within muffin-tin (MT) spheres around atomic nuclei. While effective for localized $d$-electrons, it fails for delocalized states (e.g., $sp$-metals, topological materials) because it ignores interstitial contributions and itinerant electron motion.
• The Modern Theory Challenge: The modern theory, based on Berry phases, provides a gauge-invariant total orbital magnetization ($M$) that includes both local and itinerant contributions. However, standard implementations often rely on naive perturbative expansions (e.g., $k \cdot p$ or simple tight-binding models) that insert a "complete" identity operator.
• The Core Issue: Naive implementations often violate gauge invariance because they neglect the $k$-derivative of the basis functions (Wannier functions). This leads to missing "anomalous position" terms and incorrect results, particularly for materials where band hybridization is strong. The paper aims to clarify which specific terms in the modern theory account for the difference between ACA and the full theory, and under what conditions the ACA is valid.

2. Methodology

The authors developed a rigorous gauge-covariant formalism based on the work of Lopez et al. (2012) and applied it to first-principles calculations.

• Gauge-Covariant Formulation:

  • They utilize a formulation where the orbital magnetization is expressed in terms of covariant derivatives of Bloch states. This ensures the result is independent of the specific choice of Wannier gauge (unitary transformation of occupied bands).
  • They introduce gauge-related matrices $A$ (Berry connection/position), $B$ (velocity), $C$ (orbital angular motion), and $J$ (gauge correction due to band mixing).
  • Crucially, they construct a quantum mechanical orbital moment operator ($\hat{M}$) based on occupation-weighted covariant derivatives, allowing for band-resolved analysis.

• J-Decomposition Scheme:

  • The total orbital magnetization ($M$) is decomposed into terms based on powers of the gauge correction matrix $J$ (which represents coherent band hybridization):
    • $M^{(0)}$: Terms proportional to $J^0$. Describes atomic-like contributions (intra-cell circulation) derived solely from Wannier basis matrix elements.
    • $M^{(1)}$: Terms proportional to $J^1$. Describes intermediate-level circulation (mixed intra/inter-atomic).
    • $M^{(2)}$: Terms proportional to $J^2$. Describes highly non-local contributions arising from coherent band hybridizations (inter-cell circulation).
  • This decomposition allows the authors to map the ACA (which captures mostly $M^{(0)}$) against the full modern theory.

• Computational Approach:

  • DFT: Calculations performed using the FLAPW method (Fleur code) with SOC and DFT+$U$ where necessary.
  • Wannierization: Bloch states transformed into Maximally Localized Wannier Functions (MLWFs) using Wannier90.
  • Materials Studied:
    • $d$-transition metals (Fe, Co, Ni, Pt, W, etc.).
    • $sp$-metals (Al, Bi).
    • Transition Metal Dichalcogenides (1H-MoS$_2$, Td-WTe$_2$).

3. Key Contributions

1. Rigorous Gauge-Covariant Operator: The authors derived a gauge-invariant orbital moment operator that correctly includes the $k$-derivative of the basis states, resolving the inconsistencies found in naive tight-binding or perturbative approaches.
2. Term-by-Term Decomposition ($J$-decomposition): They established a clear physical link between the ACA and the modern theory. They proved that the ACA corresponds essentially to the $M^{(0)}$ term (intra-cell self-rotation), while the deviation arises from $M^{(1)}$ and $M^{(2)}$ terms driven by band hybridization.
3. Validation of Gauge Invariance: They demonstrated that the total orbital magnetization is only gauge-invariant when all terms ($M^{(0)} + M^{(1)} + M^{(2)}$) are summed; individual terms are gauge-dependent.
4. Band-Resolved Analysis: The construction of the orbital moment operator enables the analysis of orbital magnetism in excited states and specific bands, which is crucial for understanding non-equilibrium phenomena.

4. Key Results

A. $d$-Transition Metals (Localized Electrons)

• 3d Metals (Fe, Co, Ni): The ACA captures >70% of the total orbital magnetization. The $M^{(0)}$ term dominates, consistent with the localized nature of $d$-electrons.
• 5d Metals (W): Significant deviation occurs. Tungsten (W) shows a large contribution from $M^{(1)}$ and $M^{(2)}$ due to the delocalized nature of 5$d$ electrons. The ACA fails to reproduce the sign and magnitude of the total magnetization in W.
• Conclusion: The validity of the ACA is directly correlated with the spatial localization of the electronic wavefunctions.

B. $sp$-Metals (Delocalized Electrons)

• Al and Bi: The ACA captures only a small fraction (e.g., ~42% in hexagonal Bi) of the total magnetization.
• Mechanism: The large kinetic energy and delocalization of $sp$ electrons lead to strong inter-site hopping. The $M^{(2)}$ term (inter-band hybridization) becomes dominant.
• Van Hove Singularity: In hexagonal Bi, a sharp peak in orbital magnetization near the Fermi level is driven by inter-band hybridization between nearly degenerate $p$-bands. The ACA, being an intraband approximation, completely misses this feature.

C. Transition Metal Dichalcogenides (TMDs)

• 1H-MoS$_2$:
  • The valley-dependent orbital moment at the $K$ and $K'$ points is four times larger in the modern theory than in the ACA.
  • The $M^{(2)}$ term dominates near the band gap due to coherent hybridization between valence and conduction bands ($p$-$d$ mixing).
  • Energy Dependence: Within the band gap, the modern theory orbital magnetization varies linearly with energy (linked to Berry curvature), whereas the ACA remains constant.
• Td-WTe$_2$:
  • Near avoided band crossings (topological features), the modern theory predicts orbital moments nearly two orders of magnitude larger than the ACA.
  • The strong Berry curvature induced by broken inversion symmetry drives these large non-local contributions.

5. Significance and Implications

• Beyond Atomic Orbitals: The paper demonstrates that controlling orbital magnetism in modern materials (TMDs, topological insulators) cannot rely solely on atomic orbital manipulation. The Berry phase and band hybridization are the primary drivers of enhanced orbital moments.
• Orbitronics: The results suggest that orbitronic devices (utilizing orbital currents) can achieve significantly higher efficiency in materials with strong band hybridization and non-trivial topology, where the modern theory effects dominate.
• Theoretical Framework: The proposed gauge-covariant formalism provides a robust, physically transparent tool for calculating orbital-related quantities (Orbital Hall Effect, Orbital Edelstein Effect) in real materials, correcting previous approximations that violated gauge invariance.
• Experimental Guidance: The distinction between $M^{SR}$ (Self-Rotation, linked to MCD) and $M^{CM}$ (Center-of-Mass, itinerant) offers a pathway to interpret experimental data, suggesting that $sp$-metals and TMDs may exhibit large ratios of itinerant-to-local orbital magnetization.

In summary, this work provides a comprehensive "anatomy" of orbital magnetism, proving that while the ACA is sufficient for localized $d$-electrons, the modern theory's non-local, Berry-phase-driven terms are essential for understanding and engineering orbital phenomena in delocalized and topological systems.