Controlled Dephasing of a Quantum Dot in the Kondo Regime

Kondo correlation in a spin polarized quantum dot (QD) results from the dynamical formation of a spin singlet between the dot’s net spin and a Kondo cloud of electrons in the leads, leading to enhanced coherent transport through the QD. We demonstrate here significant dephasing of such transport by coupling the QD and its leads to potential fluctuations in a nearby “potential detector.” The qualitative dephasing is similar to that of a QD in the Coulomb blockade regime in spite of the fact that the mechanism of transport is quite different. A much stronger than expected suppression of coherent transport is measured, suggesting that dephasing is induced mostly in the “Kondo cloud” of electrons within the leads and not in the QD.

Figure 1

(a) Top view scanning electron microscope image of the structure. The gray areas are metallic gates deposited on the surface of a GaAs-AlGaAs heterojunction embedding the 2DEG. (b) Conductance of the QPC detector (top gray trace) and conductance of the QD (the lower black trace) as a function of the plunger gate voltage, $V_p$. The numbers 0–13 indicate the number of electrons in the QD. (c) The black trace presents a Kondo peak, namely, conductance $G_{\mathrm{QD}}$ as a function of $V_p$, while the gray trace is the detector differential signal, namely, the derivative of the current through the QPC with respect to the modulation of the QD plunger gate. Here $V_d=1.5\mathrm{meV}$ and $t_d=0.3$. (d) Detector sensitivity $\Delta t_d$ as a function of $t_d$ (Here $T_{k1}\sim 41\mu \mathrm{eV}$.) Measurements are made on both sides of the Kondo resonance: the circles between 12 and 13 electrons while the diamonds between 13 and 14 electrons.

Figure 2

Differential conductance of the QD as a function of plunger gate voltage and applied bias. The ZBA’s FWHM, measured at the narrowest neck of the conductance peak, yields $T_k$.

Figure 3

(a) Conductance $G_{\mathrm{QD}}$ as a function of $V_p$ for $V_d=0$ and 1.2 mV, when the detector is tuned to $t_d=1$. (b) The same data as in (a) but for $t_d=0.26$. The applied bias on the detector SD, $V_d$, serves as a gate, which shifts the conductance picture, and breaks the symmetry of the two QPCs forming the QD. This results in a uniform suppression of conductance at the peaks. However, the dip, which is developed in the Kondo valley for $t_d\ne 1$, is a consequence of the dephasing interaction with the detector. (c) $g_m$ (full diamonds) and $g_v$ (empty squares) as a function of $V_d$, for $t_d=0.26$.

Figure 4

The suppression strength ($\gamma$) as a function of $t_d$ for a few values of $V_d$ (squares, $V_d=1.5\mathrm{mV}$; triangles, $V_d=1.2\mathrm{mV}$; circles, $V_d=0.9\mathrm{mV}$). Two sets are presented: (a) $T_{k1}\sim 41\pm 3\mu \mathrm{eV}$ and (b) $T_{k2}\sim 12\pm 5\mu \mathrm{eV}$. (The error bars are determined by the least-square fit of a quadratic function of $V_d$ for fixed $T_k$ and $t_d$.) In the inset (c), an example of the measured and the predicted effect multiplied by 30 (diamonds) for $V_d=1.2\mathrm{mV}$ and $T_{k1}\sim 41\pm 3\mu \mathrm{eV}$ (using the measured values of $\Delta t_d$ and $t_d$).