Orbital and spin Kondo effects in a double quantum dot

– Motivated by recent experiments, in which the Kondo effect has been observed for the first time in a double quantum-dot structure, we study electron transport through a system consisting of two ultrasmall, capacitively-coupled dots with large level spacing and charging energy. Due to strong interdot Coulomb correlations, the Kondo effect has two possible sources, the spin and orbital degeneracies, and it is maximized when both occur simultaneously. The large number of tunable parameters allows a range of manipulations of the Kondo physics – in particular, the Kondo effect in each dot is sensitive to changes in the state of the other dot. For a thorough account of the system dynamics, the linear and nonlinear conductance is calculated in perturbative and non-perturbative approaches. In addition, the temperature dependence of the resonant peak heights is evaluated in the framework of a renormalization group analysis. Introduction. Recently there has been substantial interest in many-body and correlation effects in ultrasmall semiconductor quantum dots. The dots may have strong electron-electron interactions, characterized by the charging energy EC, and large level spacing δ (typically EC > δ) [1]. At low temperature and strong coupling Γ between the dot and the leads, kBT ≪ Γ, quantum fluctuations of the charge and spin degrees of freedom strongly affect the transport through the dot [2–6]. The spin fluctuations lead to the Kondo effect, which has been verified in experiments on single quantum dots [7]. The Kondo effect has recently been discovered also in a double quantum-dot structure [8]. Motivated by these experiments, we consider in the present work a system of two capacitivelycoupled quantum dots, depicted in Fig. 1a). The level spacings in each dot are large, and effectively just one level per dot is coupled to the two reservoirs. For strong tunnel coupling and spin-degenerate levels, each dot would separately display the usual spin-Kondo effect in its I−V characteristics. In the present case, the interdot interaction is assumed strong rendering the charge states n1 and n2 of the two dots strongly correlated. Consequently, the Kondo effect of each dot is sensitive to and can be manipulated by voltages applied to the other dot. The strong interdot interaction has also another, more dramatic effect: the Kondo effect can arise from the orbital degeneracy (the two dot levels tuned to resonance) even in absence of