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

Because HF treatment has been shown to remove or passivate surfacetraps on InP [9], the normal PL spectra for the HF-etched InP QDs show onlya small degree of deep-trap emission (manifested in Fig. 10 by a red tail in thePL spectra). For unetched and unpassivated CdSe QDs, the normal redshiftedemission from traps is more pronounced and is manifested as a peak at725 nm in Fig. 11. In Figs. 10 and 11, it is apparent that PL upconversion isalso occurring for trap emission in the red tail of the PL spectra. Withincreased trap density, the intensity of the UCPL from traps and from the QDband edge is greatly increased relative to the band-edge emission. It is notedthat the maximum degree of UCPL occurs with QDs that were aged (sitting inambient conditions for several weeks to months).There are several important features in the UCPL spectra of Figs. 10and 11. First, the UCPL cannot be detected at energies above the maximumenergy exhibited in the normal PL emission. Second, the intensity of theUCPL generally follows the intensity distribution of the global PL emissionacross the whole PL spectrum, except that its peak intensity is red-shiftedfrom the global PL peak intensity. Third, the UCPL spectra were obtained atlow light intensity (a Xe lamp in the fluorimeter was used as the excitationsource). A fourth feature of the data is that the intensity of the UCPL followsa linear dependence on excitation intensity at low excitation intensity and thenbegins to saturate; this is shown, for example, for the CdSe QDs as an insetin Fig. 11. The fifth feature is the critical role of surface states in the PLupconversion.The first two features of the UCPL spectra indicate that only band-edgeemission can be detected from the QDs. This means the upconversionmechanism cannot involve carrier ejection to a barrier surrounding theQDs, followed by radiative recombination in the barrier, as was proposedfor upconversion in semiconductor heterojunctions [52–55]; PL from a barrierwould exhibit higher energies than the QD bandgap and the UCPL would notbe confined to the range of normal PL emission from the QDs. Thus, Augerprocesses cannot be involved here. The fact that PL upconversion is restrictedonly to the QD bandgap energies means that subgap states are involved as anintermediate state.The third and fourth features indicate that nonlinear two-photonabsorption (TPA) cannot be the cause of upconversion because the lampexcitation source is too weak for generating a TPA process; TPA requires veryhigh light intensities usually generated by laser excitation, is generallyinefficient, and is nonlinear with intensity.The fourth and fifth features indicate that surface states or traps areplaying a critical role in the UCPL process. The relatively intensity of UCPL isdirectly correlated with the surface state density. The linear dependence ofUCPL intensity on excitation intensity is also consistent with photoexcitationto traps.Copyright 2004 by Marcel Dekker, Inc. All Rights Reserved.