Copyright 2004 by Marcel Dekker, Inc. All Rights Reserved.

through the discrete QD levels is onset in this junction. Hence, the apparentlevel spacing in the tunneling spectra is larger than the real level spacing by afactor of V B /V 1 = (1+C 1 /C 2 ). Therefore, upon reducing the tip–QDdistance, C 1 increases and so does the measured gap. An additional differenceis that in the dashed curve, a doublet is observed at the onset of tunneling intothe CB, in contrast to a corresponding single peak seen in the solid curve. Thesecond peak in the dashed curve cannot be associated with the p state becausethe apparent s–p separation must be larger here as compared to the solid curvedue to the effect of voltage division, and the observed spacing is smaller. Thepeak spacing within this doublet is 170 meV, comparable to E C valuesmeasured for InAs QDs of similar size. Hence, we attribute this doublet tosingle-electron charging of the 1S e level. As the tip approaches the QD, G 1increases towards the value of G 2 and thus the process of resonant tunnelingbecomes accompanied by QD charging.Figure 16 represents the size dependence on the tunneling spectra forQDs on HOPG in the absence of charging effects. The measured VB–CB gaps(1.57, 1.37, and 1.2 eV) and s–p level separations (0.49, 0.43, and 0.32 eV) forthe nanocrystals of radii 1.8, 2.5, and 3.4 nm, respectively, are in relativelygood agreement with the values obtained in the QD/linker/Au case (SectionIII.B) and exhibit the expected quantum-size effect. The absence of chargingmultiplets allowed us to observe (for the larger QDs) a third peak at positivebias that may be related to the next CB state, presumably 1D e [82]. This peakFigure 16 Size evolution of tunneling spectra of InAs QDs on HOPG. The QDradii are denoted in the figure. For clarity, the spectra are offset vertically and shiftedalong the bias axis to center the measured gaps at zero bias.Copyright 2004 by Marcel Dekker, Inc. All Rights Reserved.