---
title: Triangular Triple Quantum Dot
url: https://www.emergentmind.com/topics/triangular-triple-quantum-dot-ttqd
type: topic
---

# Triangular Triple Quantum Dot

A triangular triple quantum dot (TTQD) is a three-dot artificial molecule in which the dots occupy the vertices of a closed triangle and are coupled either by coherent tunneling, by mutual capacitance, or by both. Relative to linear triple dots, the closed loop adds multiple interfering paths between source and drain, geometric frustration, chiral correlations, and orbital sectors associated with \(C_3\) symmetry; these features make TTQDs a canonical platform for studying correlated transport, Kondo screening, charge frustration, superconducting proximity, and electrically controlled qubits [2603.11488] [1605.05803] [1402.6451].

## 1. Geometry, realizations, and charge-state structure

In the strict molecular sense, a TTQD is a tunnel-coupled three-site ring. A representative gate-defined realization uses a GaAs/AlGaAs two-dimensional electron gas about \(100\) nm below the surface, three positively biased plunger gates \(P1,P2,P3\) to create the dots, negatively biased surface gates \(D1\)–\(D6\) to define the outer boundaries, inter-dot tuning gates \(T1,T2,T3\), and a negatively biased top gate separated by a \(50\) nm insulator to deplete only the small center region of the triangle. In that device the intended inter-dot separation is about \(250\) nm, and an Aharonov–Bohm oscillation period of about \(1.5\) T implies a central depleted area consistent with an equilateral triangle of side length about \(80\) nm. The same work established that the occupancies and the three tunnel couplings are independently tunable by gate voltages, with \(T1\), \(T2\), and \(T3\) acting predominantly on the \(QD1\)–\(QD2\), \(QD2\)–\(QD3\), and \(QD3\)–\(QD1\) links, respectively [1705.10053].

A second class of TTQD realization emphasizes electrostatics rather than coherent molecular orbitals. In a symmetric capacitive triangle, charge frustration appears when the interdot couplings satisfy \(X_{ij}=X\) for all \(i\neq j\). The frustrated manifold then consists of the six degenerate charge states
\[
(1,0,0),\ (0,1,0),\ (0,0,1),\ (1,1,0),\ (1,0,1),\ (0,1,1),
\]
which is the highest possible non-spin degeneracy in that model. Experimentally, these six-fold points produce omnidirectional charge transport between any pair of the three reservoirs and are accompanied by nearby nontrivial triple points in the stability diagram. In the reported GaAs/AlGaAs device, the fitted charging parameters are roughly \(U_i\sim 0.27\) meV and \(X\sim 0.06\) meV, while the mismatch in the three voltage coordinates of the frustrated points is less than \(0.7\) mV, corresponding to about \(8.2\,\mu\text{eV}\), comparable to \(2k_B T\) at \(52\) mK [1402.6451].

The term TTQD is also used in an asymmetric capacitive sense. In a GaAs/AlGaAs triple-dot structure with separate source and drain paths for each dot, the coupling hierarchy
\[
E_{12}\approx E_{13} > E_{23}
\]
stabilizes a two-fold ground-state degeneracy between
\[
(1,0,0)\quad \text{and}\quad (0,1,1),
\]
called the electron-pairing degeneracy. Along a one-parameter cut through that point, the effective charges satisfy
\[
Q_1=-Q_2=-Q_3=\pm \frac{e}{2},
\]
so the third dot mediates an induced negative-\(U\)-like attraction between the other two dots despite the underlying Coulomb interaction remaining repulsive. The reported coupling energies are approximately \(E_{12}\sim 180~\mu\text{eV}\), \(E_{13}\sim 160~\mu\text{eV}\), and \(E_{23}\sim 70~\mu\text{eV}\) [1807.08641].

## 2. Microscopic Hamiltonians and effective degrees of freedom

A standard interacting TTQD model is a three-impurity Anderson or Hubbard Hamiltonian. In one widely used form,
\[
H = H_{\mathrm{dots}} + H_{\mathrm{leads}} + H_{\mathrm{hyb}},
\]
with
\[
H_{\mathrm{dots}} = \sum_{i\mu} \varepsilon_i \hat d_{i\mu}^\dagger \hat d_{i\mu}
+ U \sum_i \hat n_{i\uparrow}\hat n_{i\downarrow}
+ \sum_{i\neq j,\mu} t_{ij}\hat d_{i\mu}^\dagger \hat d_{j\mu},
\]
noninteracting reservoirs
\[
H_{\mathrm{leads}}=\sum_{\alpha k\mu}\varepsilon_{\alpha k} c_{\alpha k\mu}^\dagger c_{\alpha k\mu},
\]
and hybridization
\[
H_{\mathrm{hyb}}=\sum_{\alpha k i\mu} V_{\alpha k i\mu}\, d_{i\mu}^\dagger c_{\alpha k\mu}+{\rm H.c.}
\]
A common choice is electron-hole symmetry,
\[
\varepsilon_d=-\frac{U}{2},
\]
with dot \(1\) attached to the left lead, dot \(3\) attached to the right lead, and dot \(2\) remaining internal to the loop. In nonequilibrium HEOM treatments the reservoirs are often encoded by a Lorentzian hybridization function
\[
\Delta_{\mu\nu}(\omega)=\delta_{\mu\nu}\,\Delta\,\frac{W^2}{\omega^2+W^2},
\]
where \(\Delta\) is the dot-lead coupling strength and \(W\) the bandwidth [2603.11488].

In the strong-coupling limit \(t\ll U\), the loop geometry generates an effective spin Hamiltonian that already contains the essential TTQD structure:
\[
H_{\mathrm{eff}} = -t(1-n)\sum_{jk,\mu} (\hat d_{j\mu}^\dagger \hat d_{k\mu}+{\rm H.c.})
+ J\sum_{j<k}\left(\hat{\mathbf S}_j\cdot \hat{\mathbf S}_k-\frac14 \hat n_j\hat n_k\right)
+ \chi\,\hat{\mathbf S}_1\cdot(\hat{\mathbf S}_2\times \hat{\mathbf S}_3).
\]
Here
\[
J = \frac{4t^2}{U},\qquad \chi = \frac{24 t^3 \sin\varphi}{U^2},\qquad \varphi = \frac{2\pi\phi}{\phi_0}.
\]
The exchange term encodes frustrated antiferromagnetism on the triangle, while the scalar-spin-chirality term is specific to the closed loop and underlies later discussions of chiral current, bias-induced time-reversal-symmetry breaking, and flux sensitivity [2603.11488].

For equilateral TTQDs, the natural one-particle basis is the \(C_3\)-adapted molecular-orbital basis. One convenient representation is
\[
d_{A \sigma}^\dagger = \frac{d_{a \sigma}^\dagger + d_{b \sigma}^\dagger + d_{c \sigma}^\dagger}{\sqrt{3}},
\]
\[
d_{E_+, \sigma}^\dagger = \frac{2 d_{a \sigma}^\dagger - d_{b \sigma}^\dagger - d_{c \sigma}^\dagger}{\sqrt{6}},
\qquad
d_{E_-, \sigma}^\dagger = \frac{d_{b \sigma}^\dagger - d_{c \sigma}^\dagger}{\sqrt{2}}.
\]
In the vertically contacted single-lead geometry, the \(A\) orbital is treated as localized with \(\Gamma_A=0\), while the two degenerate \(E_\pm\) channels control the orbital part of the Kondo problem. The equality
\[
\Gamma_{E_+}=\Gamma_{E_-}
\]
defines the SU(4)-symmetric point, whereas \(|\Gamma_{E_+}-\Gamma_{E_-}|\neq 0\) drives a crossover toward SU(2) behavior [1605.05803].

The same strong-coupling expansion shows that local charge is not independent of spin order. For example,
\[
\hat n_a = 1 + 8\left(\frac{t}{U}\right)^3
\left[\mathbf S_a\cdot(\mathbf S_b+\mathbf S_c)-2\mathbf S_b\cdot\mathbf S_c\right],
\]
with cyclic permutations for the other sites. This directly motivates the electric-polarization observable
\[
\delta n \equiv \langle n_a\rangle - \langle n_{\rm dot}\rangle
= \frac13\left(2\langle n_a\rangle - \langle n_b\rangle - \langle n_c\rangle\right),
\]
which vanishes in the symmetric high-temperature state but becomes finite once the spin sector is symmetry-selected by Kondo screening, orbital asymmetry, or external perturbations [1606.00522].

## 3. Correlation physics, Kondo regimes, and Fermi-liquid structure

At half filling and \(t/U\ll 1\), the isolated equilateral TTQD is a frustrated three-spin Heisenberg cluster with a fourfold-degenerate ground state consisting of two spin-\(\tfrac12\) doublets. Once one apex site is coupled to a metallic lead, Kondo screening lifts this degeneracy in a highly structured way: the spin on the contacted site forms a Kondo singlet with the lead, the remaining two spins form a local singlet on the opposite bond, and the charge operator above then implies \(\langle n_a\rangle>1\). The maximal strong-coupling polarization is
\[
\delta n_{\max}=12\left(\frac{t}{U}\right)^3,
\]
and NRG shows that the saturated low-temperature polarization obeys approximately
\[
\delta n^* = 12\left[c_3\left(\frac{t}{U}\right)^3 - c_5\left(\frac{t}{U}\right)^5\right],
\]
with \(c_3\to 1\) as \(\Gamma/U\to 0\). Smaller hybridization produces larger polarization, while a magnetic field suppresses the Kondo effect and therefore suppresses the polarization mechanism itself [1606.00522].

When the lead geometry preserves access to the degenerate \(E_+\) and \(E_-\) orbitals, the low-energy fixed point can instead be SU(4). In the vertical single-lead configuration, the condition \(\Gamma_{E_+}=\Gamma_{E_-}\) yields an SU(4)-symmetric Kondo state in which both spin and orbital pseudospin are screened. At that point
\[
\delta n(T)=0 \quad \text{for all } T,
\]
because orbital symmetry forbids any charge imbalance. Departing from orbital symmetry produces a continuous SU(4)-to-SU(2) crossover controlled by \(|\Gamma_{E_+}-\Gamma_{E_-}|\), and the induced electric polarization changes sign when the dominant parity of the screened orbital changes. The fixed-point spectra interpolate between an SU(4) pattern and an SU(2) “SC + F” structure composed of one strongly coupled Kondo channel and one free channel [1605.05803].

With two normal leads, lower symmetries, and gate-voltage tuning over a broad filling range, the TTQD supports additional many-body sectors. NRG analyses have identified a robust \(S=1\) Nagaoka state near four-electron filling and an SU(4)-related regime near three-electron filling. In the inversion-symmetric two-lead problem the zero-temperature transport is parametrized by even and odd phase shifts \(\delta_e\) and \(\delta_o\), with
\[
g_s=\frac{2e^2}{h}\sin^2(\delta_e-\delta_o),\qquad
g_p=\frac{2e^2}{h}\left(\sin^2\delta_e+\sin^2\delta_o\right),
\]
and
\[
N_{\mathrm{tot}}=\frac{2}{\pi}(\delta_e+\delta_o).
\]
The \(S=1\) Kondo regime is especially robust, whereas the SU(4) regime is much more sensitive to off-diagonal and diagonal distortions because it relies on orbital degeneracy. The Kondo scale \(T^*\) is correspondingly sensitive to where the local moment resides inside the triangle; moments concentrated on the apex site are screened more weakly than those residing on orbitals with larger direct lead overlap [1008.1821].

A complementary classification emphasizes the distinction between regular and singular Fermi liquids. For a TTQD symmetrically coupled to two leads, the zero-temperature conductance satisfies
\[
G=G_0 \sin^2\delta(\epsilon_F),\qquad G_0=\frac{2e^2}{h},
\]
with
\[
\delta(\epsilon_F)=\frac{\pi}{2}n_{\rm imp}+I_L.
\]
In the regular Fermi-liquid phase, \(I_L=0\), whereas in the singular Fermi-liquid phase, associated with underscreened Kondo behavior, \(|I_L|=\pi/2\). This yields
\[
G/G_0=\sin^2\!\left(\frac{\pi n_{\rm tot}}{2}\right)
\]
for the regular phase and
\[
G/G_0=\cos^2\!\left(\frac{\pi n_{\rm tot}}{2}\right)
\]
for the singular phase. In the same framework, dark states appear naturally when the one-electron molecular state avoids the lead-connected site and becomes effectively decoupled from transport [1511.01235].

## 4. Nonequilibrium transport, interference, chirality, and thermoelectricity

A central nonequilibrium result is that Coulomb repulsion can enhance, rather than suppress, transport in a TTQD. Using numerically exact HEOM, one study of a symmetric triangle attached to left and right reservoirs through dots \(1\) and \(3\) found that the stationary current is non-monotonic in \(U\): it increases at moderate interaction, reaches a maximum near \(U\approx 3.5t\), and decreases again at larger \(U\). The mechanism is spectral rather than merely electrostatic. As \(U\) grows, a right-hand antibonding or chiral resonance moves toward \(\omega=0\); at \(U\approx 3.5t\) it is optimally aligned with the bias window, whereas at larger \(U\) it moves past the Fermi level and loses spectral weight. The same work showed that the effect persists across the tested lead couplings \(\Gamma\), and that a three-dimensional current map in \((t,U)\) space is organized mainly by the ratio \(U/t\), with the strongest enhancement in an intermediate regime where \(U\) and \(t\) are comparable [2603.11488].

The same triangular topology also supports genuinely chiral nonequilibrium phenomena. In a local-moment TTQD ring connected to two reservoirs through dots \(2\) and \(3\), a finite bias alone can generate a chiral current without magnetic field. The corresponding current operator is
\[
\hat I_c = -\frac{24e}{\hbar}\frac{t^3}{U^2}\,
\hat{\bm S}_1\cdot(\hat{\bm S}_2\times \hat{\bm S}_3),
\]
and the chiral current oscillates with bias inside the Coulomb-blockade regime \(V<U/2\). Spectral analysis further identifies a topological blockade in which circulating current remains large while left-to-right transport is strongly suppressed because the relevant chiral states are localized. In the reported low-bias blockade region \(0\le V\le 0.15\) mV, the blockade measure
\[
\gamma \equiv \frac{\int_{V_1}^{V_2} |I_c|\, dV}{\int_{V_1}^{V_2} I_t\, dV}
\]
reaches \(\gamma=25\). The same work found a differential magnetoelectric susceptibility
\[
\alpha=\mu_0 \frac{\partial I_c}{\partial V}
\]
of order \(1\,{\rm ns\cdot m^{-1}}\) at low temperature [1911.12174].

Short-time fluctuation statistics reveal still finer structure. In Born–Markov treatments of triangular ring geometries, the waiting time distribution
\[
w(\tau)= \frac{(\mathbf I,\mathbf J_D^F e^{(\mathbf L-\mathbf J_D^F)\tau}\mathbf J_D^F\bar{\mathbf P})}
{(\mathbf I,\mathbf J_D^F\bar{\mathbf P})}
\]
shows coherent oscillations that track the time-dependent occupations of the drain-connected dot. In multi-electron regimes, successive waiting times become correlated, and the correlations are tunable by Aharonov–Bohm flux. A notable result is that distinct TTQD configurations can have essentially identical full counting statistics while being distinguishable by their joint waiting-time correlations, so long-time current cumulants do not exhaust the transport information content of the triangular geometry [2012.09039].

Thermoelectric response adds another interference-controlled sector. In a linear-response HEOM treatment, geometric blockade suppresses transport near the \(C_{3v}\)-symmetric triangle, while lifting that blockade causes the heat current to increase faster than the electric current at low temperature. For the parameter set \(t_{12}=t_{23}=0.4~\mathrm{meV}\), \(U=1~\mathrm{meV}\), \(\varepsilon=-U/2\), and \(k_B T=0.04~\mathrm{meV}\), the thermoelectric figure of merit reaches about \(ZT\approx 4.46\) near \(t_{13}\approx 0.55~\mathrm{meV}\). The reported interpretation is spectral: blockade relief splits and shifts the peaks in the local spectral function so that heat transport, which samples a broader energy range, responds more strongly than charge transport [2602.00625].

## 5. Superconducting and hybrid-lead TTQDs

When two vertices of the triangle are coupled to superconducting leads, the TTQD becomes a Josephson nanocluster in which orbital frustration competes with induced pairing. In the large-gap limit
\[
\Delta_{\mathrm{SC}} \gg \max(\Gamma_S,\Gamma_N,U,|\epsilon_d|),
\]
the superconducting quasiparticle continuum can be integrated out, yielding
\[
H_\mathrm{eff} = H_\mathrm{QD} +\sum_{\alpha=L,R}
\left( \Delta_{d,\alpha}\, d^\dagger_{\alpha\uparrow} d^\dagger_{\alpha\downarrow}
+\Delta_{d,\alpha}^*\, d_{\alpha\downarrow} d_{\alpha\uparrow} \right),
\qquad
\Delta_{d,\alpha}=\Gamma_S e^{i\theta_\alpha}.
\]
The ground state is then organized by overlaps with a small set of fixed-point wavefunctions: the singlet-bond state, the resonating-valence-bond state, the local-Cooper-pair state, and two mixed singlets. At small \(\Gamma_S\), the singlet-bond state dominates for \(\phi=0\) and the RVB state dominates for \(\phi=\pi\); for \(U=2\pi t\), the crossover near half filling occurs around \(\phi\simeq 2\pi/3\). Around \(\Gamma_S/t\simeq 3\), mixed singlets become important, while for \(\Gamma_S/t\gtrsim 4\) the local-Cooper-pair state dominates. In the low-filling \(\pi\)-junction regime, a Bogoliubov zero mode appears along
\[
\Gamma_S^2\,\epsilon_d + (\epsilon_d-2t)(\epsilon_d+t)^2 = 0,
\]
and this zero mode determines the extension of the \(S=\tfrac12\) phase boundary [1412.6721].

Adding one normal lead to the remaining vertex introduces a three-way competition among Kondo screening, RVB physics, and superconducting proximity. In the equilateral \(L\)-\(R\)-\(N\) triangle with one normal lead attached to dot \(N\) and two superconducting leads attached to \(L\) and \(R\), the isolated half-filled cluster has two low-lying spin-\(\tfrac12\) states: a singlet-bond state
\[
\left| \Phi_{\text{SB};\sigma} \right\rangle
=
\frac{1}{\sqrt{2}}
d^\dagger_{N,\sigma}
\left( d^\dagger_{L,\uparrow} d^\dagger_{R,\downarrow}
- d^\dagger_{L,\downarrow} d^\dagger_{R,\uparrow} \right)|0\rangle
\]
and an RVB state with the unpaired spin located away from the normal contact. The normal lead fully screens the singlet-bond state, but leaves the RVB moment largely unscreened; for four electrons, the Nagaoka \(S=1\) state is only partially screened because there is only one conduction channel. NRG phase diagrams show that superconducting proximity shrinks the \(S=\tfrac12\) region and induces a reentrant transition between the three- and four-electron fillings [2401.11444].

## 6. Qubit, optical, and definitional extensions

The TTQD is also a quantum-information platform. For three singly occupied dots, the low-energy Hilbert space decomposes into a quadruplet with total spin \(S=3/2\) and two doublets with total spin \(S=1/2\). In one exchange-only qubit proposal, the logical states are
\[
|0\rangle\equiv |D_1\rangle,\qquad |1\rangle\equiv |D_2\rangle,
\]
and projection of the Heisenberg model onto the doublet subspace gives
\[
\hat{H_\sigma} =
-\frac{1}{2}(3J+g \mu_B B_z)\mathbf{1}
+\frac{\delta}{2}\sigma_z
+\frac{\gamma}{2}\sigma_x,
\]
with
\[
J = \frac{1}{3}(J_{12}+J_{23}+J_{31}),\quad
\delta = \frac{1}{2}(J_{12}+J_{31}-2J_{23}),\quad
\gamma = \frac{\sqrt{3}}{2}(J_{12}-J_{31}).
\]
Gate-controlled symmetry breaking selects the initial state: \(\theta=4\pi/3\) prepares \(|D_1\rangle\), while \(\theta=\pi/3\) prepares \(|D_2\rangle\). Setting \(\delta=0\) yields \(X\)-rotations, setting \(\gamma=0\) yields \(Z\)-rotations, and transport readout is provided by a doublet blockade caused by asymmetric tunneling matrix elements. Under ac driving \(\gamma(t)=\gamma_0 e^{-i\omega t}\), the current is resonantly enhanced at \(\omega=\delta\) [1407.5400].

A distinct qubit encoding uses chirality rather than the doublet basis. In a half-filled triangular molecule, the degenerate chiral states \(q_+\) and \(q_-\) define a coded qubit, and the effective Hamiltonian in that basis contains only off-diagonal terms, so the effective pseudofield lies in the \(x\)-\(y\) plane. Adiabatically encircling the degeneracy point in parameter space implements a Herzberg circuit and produces a Berry phase
\[
\phi_{\pm}(C)=\mp \frac{1}{2}\Omega(C).
\]
For a loop that encloses the origin once, the accumulated phase is \(\pi\). A crucial geometric conclusion is that this construction works for the triangular molecule but not for the linear triple dot, because the linear geometry cannot generate a circuit enclosing the degeneracy in the effective parameter space [1110.4983].

Optical proposals exploit the same loop topology. Three identical quantum dots at the vertices of an equilateral triangle form a chiral artificial molecule once a perpendicular magnetic field threads the loop. In the chiral basis \(|\lambda\rangle\), \(\lambda=0,\pm\), the flux-dependent single-particle energies are
\[
E_{\lambda}=\epsilon-2t\cos\!\left[\frac{2\pi(\nu-\lambda)}{3}\right],
\qquad
\nu=\frac{\Phi}{\Phi_0}.
\]
The corresponding circular components of the dielectric response,
\[
\varepsilon_{\pm}(\omega)=\varepsilon(\omega)\pm \varepsilon_H(\omega),
\]
yield different refractive indices for right- and left-handed circular polarizations. The model predicts a threshold frequency \(\Omega\) above which birefringence is strong, as well as Kerr rotation, circularly polarized lasing, and polarization conversion. Weak deviations from the equilateral geometry or weak stray electric fields mix the ideal chiral states but do not remove the basic circular-polarization selectivity [1406.6432].

Not every triple-dot system discussed alongside TTQDs is a TTQD in the strict topological sense. A star-like device composed of one interacting dot coupled to two noninteracting dots and then to leads is explicitly not a triangular TTQD, even though it displays split Kondo resonances, Fano–Kondo interference, and a pseudogap Kosterlitz–Thouless transition that are relevant to broader three-dot interference physics [1608.05120]. Likewise, a serial InAs nanowire triple quantum dot with an integrated charge sensor reproduces triple points, quadruple points, all eight charge states, and quantum cellular automata behavior in a triple-dot setting, but it is not a closed triangle [2303.01114]. A three-terminal linear chain can also exhibit cooperative lifting of spin blockade through three-electron doublet and quadruplet states, yet it remains distinct from a TTQD because only adjacent dots are tunnel-coupled [1311.6582]. This distinction matters because the defining TTQD phenomena—loop interference, chirality, and geometrical frustration—depend on the closed triangular connectivity itself.

Source: https://www.emergentmind.com/topics/triangular-triple-quantum-dot-ttqd