Inter-dot Spin-Orbit Coupling in Quantum Dots

• Inter-dot spin-orbit coupling is defined as the spin-nonconserving tunneling component between spatially separated quantum dots that induces spin flips.
• It modifies tunneling rates and opens avoided crossings, thereby enabling electric-dipole spin resonance and anisotropic exchange in various semiconductor systems.
• Experimental studies quantify its impact (e.g., ~4% spin-flip events) and demonstrate tunability across GaAs, InSb, and Ge platforms for enhanced qubit performance.

Inter-dot spin-orbit coupling is the spin-nonconserving component of electron or hole tunneling between spatially separated quantum dots. In its minimal form, it supplements the ordinary spin-conserving tunnel matrix element by an additional off-diagonal amplitude that couples localized states in different dots with opposite spin, thereby linking orbital motion across the interdot barrier to spin dynamics. In double quantum dots and related few-dot structures, this mechanism modifies tunneling rates, opens avoided crossings, enables electric-dipole spin resonance, produces anisotropic exchange interactions, and can either limit or enhance qubit control depending on operating regime and material platform (Maisi et al., 2015).

1. Conceptual definition and physical scope

In a localized left/right-dot basis, inter-dot spin-orbit coupling is the part of the tunnel Hamiltonian that transfers a carrier from one dot to the other while changing its spin or pseudospin. It is therefore distinct from ordinary tunnel coupling, which preserves spin, and also from intradot spin-orbit mixing, which hybridizes orbital states within the same dot. A concise single-electron representation is

$H = H_0 + H_t + H_{\rm SO},$

with $H_0$ the isolated-dot Hamiltonian, $H_t$ the spin-conserving tunnel term, and $H_{\rm SO}$ the spin-orbit-mediated spin-nonconserving tunneling term. In the GaAs double-dot formulation,

$$H_t = t_c \sum_{\sigma=\uparrow,\downarrow} \left(c^\dagger_{L\sigma}c_{R\sigma}+c^\dagger_{R\sigma}c_{L\sigma}\right),$$

while

$$H_{\rm SO}=\sum_{\sigma,\sigma'} t_{\sigma\sigma'}\,c^\dagger_{L\sigma}c_{R\sigma'}+\mathrm{h.c.},$$

with off-diagonal elements $$t_{\uparrow\downarrow}=-t_{\downarrow\uparrow}\equiv t_{\rm SO}$$ and diagonal parts much smaller than $t_c$ (Maisi et al., 2015).

The same structure reappears in one-electron nanowire double dots, where spin-momentum locking in the localized dot states generates a spin-flipped tunneling amplitude after projection into a left/right-dot basis, and in two-electron singlet-triplet descriptions, where spin-orbit coupling links $(1,1)$ triplets to the $(0,2)$ singlet through phenomenological amplitudes $H_0$0 or $H_0$1 (Liu et al., 2018, Rolon et al., 2016). In hole systems the same notion is used for spin-flipping tunnel couplings between triplets and singlets in Pauli-spin-blockaded transport, with the characteristic scale often re-expressed as a spin-orbit length $H_0$2 (Liu et al., 2022).

A common misconception is that inter-dot spin-orbit coupling is simply “strong spin-orbit coupling in a double dot.” The literature distinguishes more sharply. Inter-dot spin-orbit coupling refers specifically to the off-diagonal tunneling structure between dots; intradot spin-orbit coupling refers to orbital-spin mixing within one dot; and anisotropic exchange is a many-body consequence that emerges only after combining tunneling, Coulomb blockade, and often magnetic-field-dependent mixing [(Liu et al., 2018); (Baruffa et al., 2010)].

2. Microscopic Hamiltonians and matrix-element structure

In two-dimensional GaAs, the microscopic origin is the Rashba and Dresselhaus interaction,

$H_0$3

Projecting $H_0$4 onto the left/right localized orbitals yields an effective spin-flip matrix element

$H_0$5

with $H_0$6 the interdot distance and $H_0$7 the spin-orbit length. In the single-electron GaAs analysis, this gives

$H_0$8

for $H_0$9 and $H_t$0 (Maisi et al., 2015).

A related one-dimensional formulation uses

$H_t$1

with

$H_t$2

In the localized basis $H_t$3, the interdot spin-orbit matrix elements are

$H_t$4

so the same microscopic term generates both spin-conserving and spin-flip tunnel amplitudes once expressed in the dot basis (Khomitsky et al., 2010).

In nanowire double quantum dots, the effective Hamiltonian is written as

$H_t$5

After constructing dot-localized Kramers-pair states and projecting into the four-state basis, the tunnel block contains both $H_t$6 and $H_t$7. The spin-flipped amplitude is written as

$H_t$8

and in schematic form

$H_t$9

so it depends on interdot separation, confinement, SOC strength, and the angle between the magnetic-field axis and the SOC axis (Liu et al., 2018).

The strong-SOC nanowire treatment makes an additional point that is not generic but is important in that limit: in the symmetric-well, large-$H_{\rm SO}$0 approximation,

$H_{\rm SO}$1

hence

$H_{\rm SO}$2

independent of $H_{\rm SO}$3. This means that SOC redistributes tunneling weight between spin-conserving and spin-flip channels without changing the total tunneling magnitude in that model (Li et al., 2013).

3. Direct single-electron observation in GaAs

A direct experimental identification of inter-dot spin-orbit coupling at the single-electron level was obtained in an electrically isolated AlGaAs/GaAs double quantum dot by real-time charge sensing. The device was tuned to the $H_{\rm SO}$4 charge-degeneracy line, a nearby quantum point contact monitored charge motion, and the current $H_{\rm SO}$5 was sampled in real time at $H_{\rm SO}$6 to produce alternating $H_{\rm SO}$7 and $H_{\rm SO}$8 plateaus (Maisi et al., 2015).

The central observable was the waiting-time distribution in the $H_{\rm SO}$9 configuration. Its statistics showed two exponentials. Fast events with rate $$H_t = t_c \sum_{\sigma=\uparrow,\downarrow} \left(c^\dagger_{L\sigma}c_{R\sigma}+c^\dagger_{R\sigma}c_{L\sigma}\right),$$0 were identified with spin-conserving tunneling, while slow events with rate $$H_t = t_c \sum_{\sigma=\uparrow,\downarrow} \left(c^\dagger_{L\sigma}c_{R\sigma}+c^\dagger_{R\sigma}c_{L\sigma}\right),$$1 corresponded to tunneling possible only after a spin flip, either during tunneling or by intradot relaxation. On resonance, the rates were of the order of hundreds of hertz, and by varying the barrier gate the experiment swept $$H_t = t_c \sum_{\sigma=\uparrow,\downarrow} \left(c^\dagger_{L\sigma}c_{R\sigma}+c^\dagger_{R\sigma}c_{L\sigma}\right),$$2 from $$H_t = t_c \sum_{\sigma=\uparrow,\downarrow} \left(c^\dagger_{L\sigma}c_{R\sigma}+c^\dagger_{R\sigma}c_{L\sigma}\right),$$3 up to $$H_t = t_c \sum_{\sigma=\uparrow,\downarrow} \left(c^\dagger_{L\sigma}c_{R\sigma}+c^\dagger_{R\sigma}c_{L\sigma}\right),$$4 (Maisi et al., 2015).

The measured dependence was

$$H_t = t_c \sum_{\sigma=\uparrow,\downarrow} \left(c^\dagger_{L\sigma}c_{R\sigma}+c^\dagger_{R\sigma}c_{L\sigma}\right),$$5

at $$H_t = t_c \sum_{\sigma=\uparrow,\downarrow} \left(c^\dagger_{L\sigma}c_{R\sigma}+c^\dagger_{R\sigma}c_{L\sigma}\right),$$6, so about $$H_t = t_c \sum_{\sigma=\uparrow,\downarrow} \left(c^\dagger_{L\sigma}c_{R\sigma}+c^\dagger_{R\sigma}c_{L\sigma}\right),$$7 of all interdot tunneling events involved a spin flip. The linear scaling of $$H_t = t_c \sum_{\sigma=\uparrow,\downarrow} \left(c^\dagger_{L\sigma}c_{R\sigma}+c^\dagger_{R\sigma}c_{L\sigma}\right),$$8 with $$H_t = t_c \sum_{\sigma=\uparrow,\downarrow} \left(c^\dagger_{L\sigma}c_{R\sigma}+c^\dagger_{R\sigma}c_{L\sigma}\right),$$9 over nearly two orders of magnitude matched the expectation for a spin-orbit-mediated mechanism rather than an independent relaxation bottleneck. A weak residual offset at low $$H_{\rm SO}=\sum_{\sigma,\sigma'} t_{\sigma\sigma'}\,c^\dagger_{L\sigma}c_{R\sigma'}+\mathrm{h.c.},$$0 and zero field was attributed to intradot spin relaxation once the external field was too small to define a common quantization axis (Maisi et al., 2015).

The same experiment also established a practical event-classification rule. For

$$H_{\rm SO}=\sum_{\sigma,\sigma'} t_{\sigma\sigma'}\,c^\dagger_{L\sigma}c_{R\sigma'}+\mathrm{h.c.},$$1

one chooses a threshold

$$H_{\rm SO}=\sum_{\sigma,\sigma'} t_{\sigma\sigma'}\,c^\dagger_{L\sigma}c_{R\sigma'}+\mathrm{h.c.},$$2

and classifies any waiting time $$H_{\rm SO}=\sum_{\sigma,\sigma'} t_{\sigma\sigma'}\,c^\dagger_{L\sigma}c_{R\sigma'}+\mathrm{h.c.},$$3 as a spin-flip event. The stated misclassification probability is below $$H_{\rm SO}=\sum_{\sigma,\sigma'} t_{\sigma\sigma'}\,c^\dagger_{L\sigma}c_{R\sigma'}+\mathrm{h.c.},$$4, corresponding to $$H_{\rm SO}=\sum_{\sigma,\sigma'} t_{\sigma\sigma'}\,c^\dagger_{L\sigma}c_{R\sigma'}+\mathrm{h.c.},$$5 readout fidelity. This is significant because it turned inter-dot spin-orbit coupling from an inferred admixture in spectroscopy into a directly countable single-electron process (Maisi et al., 2015).

4. Driven dynamics and electric-dipole spin resonance

Under time-dependent electric driving, inter-dot spin-orbit coupling converts charge motion into spin motion. In the one-dimensional pulse-pumped double dot, a single half-period pulse $$H_{\rm SO}=\sum_{\sigma,\sigma'} t_{\sigma\sigma'}\,c^\dagger_{L\sigma}c_{R\sigma'}+\mathrm{h.c.},$$6 induces dynamics in which the charge density and spin density become irregular, and the charge density distribution becomes strongly spin-dependent. The analysis identifies three regimes in terms of

$$H_{\rm SO}=\sum_{\sigma,\sigma'} t_{\sigma\sigma'}\,c^\dagger_{L\sigma}c_{R\sigma'}+\mathrm{h.c.},$$7

For $$H_{\rm SO}=\sum_{\sigma,\sigma'} t_{\sigma\sigma'}\,c^\dagger_{L\sigma}c_{R\sigma'}+\mathrm{h.c.},$$8, tunneling remains dominated by $$H_{\rm SO}=\sum_{\sigma,\sigma'} t_{\sigma\sigma'}\,c^\dagger_{L\sigma}c_{R\sigma'}+\mathrm{h.c.},$$9 and is spin-independent to leading order; for $$t_{\uparrow\downarrow}=-t_{\downarrow\uparrow}\equiv t_{\rm SO}$$0, tunneling and spin-orbit coupling combine coherently and produce highly irregular, multi-frequency oscillations; for $$t_{\uparrow\downarrow}=-t_{\downarrow\uparrow}\equiv t_{\rm SO}$$1, over-barrier motion sets in and high-energy states participate, with strong spin dependence in the charge distribution (Khomitsky et al., 2010).

In nanowire double dots, electrically driven spin resonance can arise by two mechanisms: SOC-induced intradot pseudospin mixing and interdot spin-flipped tunneling. With an AC electric perturbation

$$t_{\uparrow\downarrow}=-t_{\downarrow\uparrow}\equiv t_{\rm SO}$$2

the on-resonance Rabi rate between states $$t_{\uparrow\downarrow}=-t_{\downarrow\uparrow}\equiv t_{\rm SO}$$3 and $$t_{\uparrow\downarrow}=-t_{\downarrow\uparrow}\equiv t_{\rm SO}$$4 is

$$t_{\uparrow\downarrow}=-t_{\downarrow\uparrow}\equiv t_{\rm SO}$$5

Near the spin-flip anticrossing at $$t_{\uparrow\downarrow}=-t_{\downarrow\uparrow}\equiv t_{\rm SO}$$6, where interdot SOC dominates, the pure spin-flip EDSR rate scales as

$$t_{\uparrow\downarrow}=-t_{\downarrow\uparrow}\equiv t_{\rm SO}$$7

Because $$t_{\uparrow\downarrow}=-t_{\downarrow\uparrow}\equiv t_{\rm SO}$$8, the EDSR strength depends strongly on interdot distance and magnetic-field orientation (Liu et al., 2018).

The InSb nanowire estimates in that work give $$t_{\uparrow\downarrow}=-t_{\downarrow\uparrow}\equiv t_{\rm SO}$$9, $t_c$0 when $t_c$1, and $t_c$2 for $t_c$3 and $t_c$4; the zero-field anticrossing is quoted as $t_c$5. The same analysis argues that phonon-emission-induced spin-flip relaxations can be effectively suppressed by the phonon bottleneck effect even at relatively low magnetic fields in strong-SOC materials such as InSb (Liu et al., 2018).

A plausible implication is that inter-dot spin-orbit coupling is best viewed not merely as a leakage channel, but as a charge-to-spin transducer whose utility depends on whether the device is operated for transport, spectroscopy, or coherent control.

5. Two-electron manifolds, singlet-triplet dynamics, and anisotropic exchange

For two electrons in GaAs double dots, inter-dot spin-orbit coupling is commonly expressed in the basis

$t_c$6

through the decomposition

$t_c$7

Here

$t_c$8

while

$t_c$9

with $(1,1)$0 and $(1,1)$1. After Bloch-Feshbach projection into the $(1,1)$2 subspace, the effective couplings

$(1,1)$3

show explicitly that the $(1,1)$4-$(1,1)$5 and $(1,1)$6-$(1,1)$7 dynamics are controlled by the interplay of spin-orbit and hyperfine terms (Rolon et al., 2016).

This interplay produces experimentally accessible signatures in Landau-Zener sweeps across the $(1,1)$8-$(1,1)$9 resonance. The singlet return probability $(0,2)$0 exhibits oscillations at frequencies associated with $(0,2)$1-$(0,2)$2 and $(0,2)$3-$(0,2)$4 evolution, and the spin-orbit contribution appears as both a frequency shift and a modulation of the normalized Fourier amplitudes. The detuning dependence of $(0,2)$5 and the broadening and shifting of the Fourier lineshapes provide a route to extracting the interdot spin-orbit scale $(0,2)$6 (Rolon et al., 2016).

At lower energies in the singly occupied $(0,2)$7 sector, projecting out doubly occupied charge states gives an effective anisotropic exchange Hamiltonian. In one Hubbard-like derivation,

$(0,2)$8

with

$(0,2)$9

and symmetric anisotropy tensor elements built from quadratic combinations of $H_0$00 (Nowak et al., 2010). In the complementary GaAs two-electron effective-Hamiltonian treatment, the standard spin form

$H_0$01

is quantitatively reliable across couplings, with the remaining small discrepancy between analytical and numerical results traced mostly to the cubic Dresselhaus term (Baruffa et al., 2010).

It is therefore inaccurate to identify inter-dot spin-orbit coupling with anisotropic exchange in a one-to-one way. In the GaAs two-electron treatment, $H_0$02 to second order at zero magnetic field, so $H_0$03 exactly in that limit. By contrast, in strong-SOC nanowire pseudospin qubits, competition between spin-conserving and spin-flip hopping naturally yields XXZ-plus-Dzyaloshinskii-Moriya structure. The relation between inter-dot SOC and exchange anisotropy is thus model- and regime-dependent rather than universal [(Baruffa et al., 2010); (Li et al., 2013)].

6. Material realizations, tunability, and qubit implications

Different material platforms realize very different inter-dot spin-orbit scales and control strategies.

Platform	Representative inter-dot SOC result	Citation
GaAs isolated DQD	$H_0$04; 99% spin-flip-event fidelity	(Maisi et al., 2015)
GaAs weakly coupled DQD	$H_0$05–$H_0$06, $H_0$07–$H_0$08; $H_0$09–$H_0$10	(Raith et al., 2013)
InSb nanowire DQD	$H_0$11, $H_0$12, $H_0$13	(Liu et al., 2018)
Ge hut-wire hole DQD	$H_0$14–$H_0$15 by gate tuning	(Liu et al., 2022)
GaAs flopping-mode TQD	$H_0$16; $H_0$17	(Matsumoto et al., 29 Aug 2025)

In GaAs weakly coupled double dots, phonon-assisted relaxation reveals “spin hot spots” when the detuning-controlled orbital splitting matches the Zeeman energy, $H_0$18. The resulting anticrossing gap is

$H_0$19

and the relaxation rate is strongly anisotropic with magnetic-field orientation. The analysis identifies an easy-passage direction in which the anticrossing gap vanishes and the hot spots are suppressed, and a perpendicular direction in which they are strongest. Fits to experiment gave Rashba and Dresselhaus spin-orbit lengths of about $H_0$20–$H_0$21 (Raith et al., 2013).

In Ge hut-wire hole double dots, the inter-dot spin-flip matrix element is parameterized by

$H_0$22

and transport in the Pauli spin blockade regime returns a gate-tunable $H_0$23–$H_0$24. At representative central-gate voltages, the extracted values include $H_0$25 with $H_0$26 and $H_0$27, as well as $H_0$28 with $H_0$29 and $H_0$30. The interpretation given there is that electrical tuning of interdot coupling and local electric field can act as a spin-orbit “switch,” balancing fast electrical control against reduced spin-orbit-mediated decoherence (Liu et al., 2022).

A later GaAs triple-dot implementation exploits the same logic in flopping-mode operation. With

$H_0$31

the key matrix element

$H_0$32

is maximal near $H_0$33, where charge-noise sensitivity is also suppressed. The reported experimental performance includes $H_0$34 rising nearly linearly to $H_0$35 before saturation, $H_0$36 with the third-dot orbital tuned, $H_0$37 without feedback and $H_0$38 with FNN feedback, and $H_0$39-gate fidelities improving from $H_0$40 to $H_0$41 with feedback and pulse optimization (Matsumoto et al., 29 Aug 2025).

Related quantum-dot-molecule work on holes extends the idea beyond simple ground-state-to-ground-state tunneling. There, spin mixing in the $H_0$42 shell of one dot is transferred through interdot $H_0$43-$H_0$44 tunneling into an effective spin-orbit term in the neighboring dot’s $H_0$45 shell, leading to enhanced spin relaxation around the $H_0$46-$H_0$47 resonance when the dots are misaligned and the magnetic field is tilted from the sample plane. This suggests that “inter-dot spin-orbit coupling” can also denote SOC transfer through excited-state channels rather than only direct spin-flipped hopping between lowest orbitals (Kawa et al., 2019).

Taken together, these results establish inter-dot spin-orbit coupling as a unifying mechanism across semiconductor dot architectures. In weak-coupling transport it appears as a small but directly measurable spin-flip probability per tunnel event; in spectroscopy it opens avoided crossings and relaxation hot spots; in two-electron manifolds it seeds singlet-triplet mixing and anisotropic exchange; and in electrically driven devices it becomes a controllable resource for fast spin manipulation. The practical challenge is not whether inter-dot spin-orbit coupling is present, but how to place the device in the regime where its spin-charge hybridization is either minimized for idling or maximized for control.