1 of 21

cubic Dresselhaus

hot spots

ORBITRONICS · BERRY CURVATURE · APCTP

Giant Orbital Hall Effect from

Cubic Dresselhaus Orbital Coupling

KIST, Center for Semiconductor Technology, 26.08.31

Kyoung-Min Kim

Quantum Matter Theory Group · Asia Pacific Center for Theoretical Physics (APCTP)

with Gwen Sevilen (ITB)

Seminar · arXiv:2608.19648

2 of 21

AGENDA

Where we're going

01

Motivation

Orbitronics, the orbital Hall effect, and the linear-coupling ceiling

02

Model

A two-band Hamiltonian with cubic Dresselhaus orbital coupling

03

Mechanism

Momentum-space hot spots and why they amplify the Berry curvature

04

Results

A giant, tunable orbital Hall conductivity

05

Outlook

Experimental signatures, candidate materials, open directions

2

3 of 21

01 · MOTIVATION

Orbitronics: a second angular-momentum channel

Spintronics moves information using electron spin. Orbitronics uses the electron's orbital angular momentum instead — and it survives even where spin–orbit coupling is weak.

The orbital Hall effect (OHE): a charge current generates a transverse flow of orbital angular momentum.

No spin–orbit coupling required — orbital texture alone is enough.

Confirmed experimentally in light and 3d/4d metals (Ti, Cr, Zr, Cu, Al …), well beyond the heavy elements spintronics needs.

Orbital Hall effect — schematic

charge

current J

 

 

Bernevig, Hughes & Zhang, PRL 95, 066601 (2005); Go, Jo, Kim & Lee, PRL 121, 086602 (2018).

3

4 of 21

01 · MOTIVATION

Every enhancement strategy so far: linear-in-k coupling

A recent “universal” two-band formalism shows the orbital Berry curvature (OBC) is fixed entirely by the ordinary Berry curvature and the band splitting — but every microscopic realization studied to date builds that Berry curvature from a Rashba-type coupling that is linear in momentum.

The open question

In the analogous spin-orbit setting, higher-order (cubic) corrections to the linear Rashba/Dresselhaus coupling are a well-known refinement.

​

For orbital coupling, this next order had never been explored, and no Hamiltonian-level design principle existed for pushing the OHC past the linear paradigm.

This talk

We introduce cubic Dresselhaus orbital coupling as exactly such a design principle — and show it boosts the orbital Hall conductivity by more than an order of magnitude.

Universal two-band formalism: J. M. Lee, Phys. Rev. B 112, 054441 (2025).

4

5 of 21

01 · MOTIVATION

The idea, in one picture

Add a cubic-in-k term to the orbital coupling → it creates four extra momentum-space “hot spots” where the orbital Berry curvature spikes.

4×

larger Jacobian factor at each hot spot vs. the Γ point

16×

orbital Berry curvature enhancement per hot spot (Δ→0 limit)

32×

peak orbital Hall conductivity, summed over all four hot spots

Same mechanism should apply wherever an interband coupling of this functional form appears — orbital, sublattice, or spin.

5

6 of 21

02 · BACKGROUND

A short history of the orbital Hall effect

2005

Bernevig, Hughes & Zhang

First prediction: intrinsic orbital Hall current in p-doped silicon.

2008–09

Kontani et al.

Giant orbital (and spin) Hall effect predicted in 4d/5d transition metals.

2018

Go, Jo, Kim & Lee

Orbital texture identified as the microscopic origin of intrinsic spin & orbital Hall effects.

2023–25

Choi et al.; many others

Experimental observation of OHE in light metals (Ti, Cr, Zr, Cu, Al); orbital torques.

2025

J. M. Lee

Universal two-band formalism: OBC fixed by ordinary Berry curvature × band splitting.

All the way to 2025, the underlying ingredient stayed the same: a linear-in-momentum coupling.

6

7 of 21

02 · BACKGROUND

The universal two-band Berry-curvature formalism

For any two-band Hamiltonian written in terms of an interband coupling and a band splitting, the orbital Berry curvature (OBC) that drives the orbital Hall response takes one fixed functional form.

 

 

 

J. M. Lee, Phys. Rev. B 112, 054441 (2025).

7

8 of 21

03 · MODEL

A minimal two-orbital band model

 

What's new here

 

This is the object we redesign → (next slide)

8

9 of 21

03 · MODEL

The cubic Dresselhaus orbital coupling

β — linear coefficient

The familiar Rashba-type term. By itself (α = 0) it reproduces the standard, previously-studied case.

α — cubic coefficient

The new ingredient. Borrowed directly from cubic Dresselhaus spin-orbit coupling in zinc-blende crystals.

Physical origin — not phenomenology

 

Krich & Halperin, PRL 98, 226802 (2007) — cubic Dresselhaus spin-orbit coupling in 2D electron quantum dots.

9

10 of 21

03 · MODEL

Where α = β(d/π)² actually comes from

The bulk (3D) Dresselhaus spin-orbit Hamiltonian in a zinc-blende crystal is purely cubic in momentum — there is no separate linear term to begin with:

 

 

 

This splits cleanly into two pieces:

 

10

11 of 21

03 · MODEL

Plugging in: the orbital Berry curvature

 

compare: for the purely linear case, α = 0, so

is just a constant — no k-dependence at all

The key structural fact: J(k) now depends on momentum, and can therefore be locally larger somewhere away from k = 0 than it is at the zone center.

That possibility — realized concretely — is the whole mechanism. Next: where does J(k) peak?

11

12 of 21

03 · MODEL → MECHANISM

Where the OBC peaks: hot spots at Γ and beyond

 

Γ point:

Hot spots:

Four extra zeros, related to each other by the crystal's C₄ rotation symmetry — the four hot spots.

Using α = β(d/π)², the hot-spot location reduces to this remarkably simple relation — tunable via sample thickness.

OBC in momentum space — Fig. 1(a)–(b)

Linear (left) has a single peak at Γ. Cubic (right) adds four hot spots (red ×), related by C₄ symmetry.

hot-spot momentum, set by

sample thickness d

Fig. 1(a)–(b), Kim & Sevilen, arXiv:2608.19648.

12

13 of 21

03 · MECHANISM

Fourfold Jacobian → sixteenfold Berry curvature

At the Γ point

At each hot spot

The Jacobian is four times larger in magnitude at the hot spots, with a sign flip.

This 16× ratio holds exactly in the Δ → 0 limit; away from it, the ratio grows further — up to 57× at Δ = 0.4 eV, as shown by Fig. 1(c)–(d):

13

14 of 21

03 · MECHANISM

The intuitive picture: vortex vs. antivortex winding

 

At Γ

A simple vortex — winds once, at rate β.

At a hot spot

 

Fig. 2, Kim & Sevilen, arXiv:2608.19648 — vortex at Γ (square), antivortices at the four hot spots (triangles).

14

15 of 21

04 · RESULTS

From curvature to conductivity: the linear-case baseline

Linear case (α = 0): as Δ grows from zero, cancellation between the two partially-occupied bands weakens and the OHC rises — until the upper band fully depopulates at Δ = μ, where it peaks and then decays.

Peak condition: Δ = μ

Beyond this threshold the OHC is governed entirely by the (decaying) OBC of the lower band — exactly solvable in closed form at T = 0.

Fig. 3(a), Kim & Sevilen, arXiv:2608.19648.

15

 

16 of 21

04 · RESULTS

The cubic case: a giant orbital Hall conductivity

Peak values (β = 3 eVÅ, α = 800 eVų)

μ = 0.1 eV

31.0 e/2π

32× linear

μ = 0.2 eV

22.8 e/2π

49× linear

μ = 0.3 eV

10.3 e/2π

36× linear

Enhancement isn't purely the 16× hot-spot factor: multiplicity (×4 hot spots) offsets their narrower extent in k-space, compounding to 32–49×.

Fig. 3(a)–(b), Kim & Sevilen, arXiv:2608.19648.

16

17 of 21

04 · RESULTS

A richer peak structure: two population thresholds

Unlike the linear case, the cubic OHC shows a more complex peaking behavior — which threshold controls the peak depends on whether the hot spots themselves are already populated at Δ = 0.

Small μ (hot spots start empty)

OHC grows until the hot spots become populated — that's the peak.

Larger μ (hot spots already full)

The peak instead comes from full depopulation of the upper band at the hot spots.

Fig. 3(c)–(d), Kim & Sevilen, arXiv:2608.19648 — peak OHC vs. α, showing both threshold regimes.

17

18 of 21

04 · RESULTS

A universal 1/Δ divergence, 17× stronger

In the small-splitting regime, the OHC diverges as 1/Δ in both cases — but the divergence coefficient itself is enhanced.

Linear case

Cubic case

Where does the 17 come from?

16 = four hot spots, each contributing 4× the Γ-point weight. 1 = the Γ point itself, still present.

A clean, closed-form decomposition — the 16× Berry-curvature enhancement from Slide 12 propagates directly into the transport coefficient.

18

19 of 21

05 · OUTLOOK

Experimental signatures: a switch, not a dial

Because k₀ = π/d, the hot-spot location is set by the sample thickness — giving three independently tunable knobs.

Thickness d

sets k₀ = π/d directly

Chemical potential μ

sets which states are occupied

Temperature T

(de)populates the hot spots

A sharply peaked ridge in (d, μ) — not a broad enhancement — plus a non-monotonic, thickness-dependent thermal response.

Fig. 4, Kim & Sevilen, arXiv:2608.19648.

19

20 of 21

05 · OUTLOOK

Beyond this paper: candidates and generalizations

What a host material needs

A pair of orbitally (near-)degenerate bands to serve as the pseudospin (a, b).

Bulk inversion asymmetry of the zinc-blende symmetry class, which permits the cubic-in-momentum coupling.

Identifying specific candidate materials is left for future work.

Related platforms worth exploration: light-induced cubic Rashba coupling; even-order intersubband coupling.

Beyond the orbital Hall effect

The mechanism is generic: any two-band interband coupling of this cubic functional form produces the same 4× Jacobian / 16× curvature enhancement.

The ordinary Berry curvature and the quantum metric both show the same fourfold hot-spot enhancement (Supplementary Note 3).

Suggests a universal fourfold enhancement for transport phenomena built from these quantities, in the small-splitting regime.

20

21 of 21

SUMMARY

Take-home message

1

Cubic Dresselhaus orbital coupling — the textbook higher-order correction from zinc-blende crystals — creates four extra Berry-curvature hot spots beyond the Γ point.

2

Winding number doubles at each hot spot → Jacobian factor 4× → orbital Berry curvature 16× → orbital Hall conductivity boosted 32–49×, with a universal 17× divergence coefficient.

3

A concrete, falsifiable experimental signature: a sharply peaked ridge in thickness × doping space, plus a distinctive non-monotonic thermal response.

Thank you — questions welcome.

Kyoung-Min Kim · kyoungmin.kim@apctp.org · arXiv:2608.19648

21