Photoionization of Individual CdSe/CdS Core/Shell Nanocrystals on Silicon with 2-nm Oxide Depends on Surface Band Bending

CdSe/CdS nanocrystals are studied by electrostatic force microscopy (EFM) on nand p-type silicon with 2-nm surface oxide and graphite for comparison. In the absence of above-band-gap excitation, the nanocrystals are not in electrical equilibrium with the substrates. Upon continuous photoexcitation, the particles behave differently on each substrate. On n-type silicon, the particles exhibit predominantly positive charge. However, on p-type silicon, fewer charged particles overall are seen, blinking is more common, and the ratio of negative to positive particles is higher. The low ionization yield on p-Si is apparently due to the fast return of photoexcited electrons that are confined near the silicon/oxide interface by band bending. Nanocrystal photoionization dynamics are sensitive to the state of the silicon across the 2-nm oxide. On graphite, the particles show a greater propensity to oscillate between positive, negative, and neutral states than on the other substrates. CdSe nanocrystals are a model system for zero-dimensional semiconductor nanostructures and have attracted a lot of attention in the past decade. Their size-tunable electronic properties make them a prime candidate for electrooptical applications. The nanocrystal charge state strongly influences electrical transport properties in photovoltaic and electroluminescent devices, optical gain in nanocrystal lasers,1,2 and the brightness of nanocrystals in biological luminescence imaging. It has been postulated that single nanocrystal fluorescence “blinking” (i.e., intermittency) and spectral diffusion result from charge redistribution and/or photoionization.3-9 The kinetics of fluorescence intermittency is complex and implies that a range of quenching states are weakly coupled to the photoexcited internal state, even in single passivated core/shell particles made by the best current synthetic methods.4,10,11 In this paper, we report observations of single nanocrystal photoionization using electrostatic force microscopy (EFM). EFM allows the simultaneous mapping of surface topography and electrostatic field gradients12-15 and provides direct measurements of sample charge and dielectric properties with appropriate modeling. Previously, we have shown that single CdSe particles on HOPG graphite slowly (ca. 10-6 quantum yield) photoionize when excited above their band gap.16-18 In this study, we investigate the effect of the substrate on the photoionization of CdSe /CdS core/shell nanocrystals. The EFM experiment is quantitatively improved by the use of a recently developed tip-surface capacitive calibration method19 and by systematically measuring the force dependence upon the height above the surface. We observe substantially different charging behavior on p-type silicon, n-type silicon, and HOPG substrates. The particles have a strong tendency to lose one electron upon photoexcitation; however, negative charge, as well as multiple positive charges per nanocrystal, are observed whereas the equilibrium concentration of positive and negative charges is determined by the substrate type. CdSe/CdS core/shell passivated nanocrystals have greater stability and a higher fluorescence quantum yield than TOPO-capped particles. It is thought that the outer shell of these particles protects them from photoionization and photooxidation, which would quench luminescence.20,21 In CdS, overcoated CdSe particles the conduction band offset is small, resulting in a delocalization of the electron wave function over the entire particle while the hole is confined to the core.9,22 This property should make it easier for electrons to tunnel in and out of the system, compared with the tunneling in ZnS-coated particles in which both the * To whom all correspondence should be addressed. E-mail: brus@ chem.columbia.edu. † Department of Chemistry, Columbia University. ‡ IBM Research Laboratories. § Department of Applied Physics, Columbia University. NANO LETTERS 2003 Vol. 3, No. 4 497-501 10.1021/nl0340529 CCC: $25.00 © 2003 American Chemical Society Published on Web 03/08/2003 electron and hole are confined to the core, while still removing surface charge traps, increasing fluorescence yield, and protecting the particle from photooxidation. Thus, it is interesting to investigate the photoionization of these particles. Core/shell CdSe/CdS nanocrystals (∼4.5 nm ) ∼2-nm CdSe + 2-nm monolayers of CdS + TOPO) used in these experiments were made by organometallic synthesis.22 They showed an exciton absorption peak in solution at 544 nm with a fwhm of 30.9 nm and a resolved higher exciton at 437 nm. Before deposition onto the substrates, the particles were stored in a solution of hexane and TOPO in a dry atmosphere. The particles were spin-coated onto degenerately doped p-type (B-doped, 0.005-0.01 Ω cm) and n-type (Sbdoped, 0.008-0.03 Ω cm) silicon substrates with a 2-nm thermal oxide layer (obtained from IBM Research, Yorktown Heights, NY). The substrates were cleaned with ethanol and hexane prior to particle deposition. Highly ordered pyrolitic graphite (HOPG) was spin-coated with a thin layer of polyvinylbutyral (PVB) to prevent aggregation because single nanocrystals are highly mobile on graphite.23,24 The exposure of spun samples to air was minimized typically to no more than 10 min to prevent possible oxidation.16,18,22 EFM images of photoexcited nanocrystals (grazing angle, HeCd laser (Laconix 200 series), λ ) 442 nm, ∼0.2 W/cm2) and those exposed only to ambient light were obtained at room temperature in an argon purge box using a Digital Instruments (Santa Barbara, CA) Nanoscope IIIa Multimode AFM with an extender module. Pt-Ir-coated EFM tips (Nanosensors EFM-20) from Molecular Imaging (Phoenix, AZ) were used in all experiments. Their resonance frequency was around 65 kHz, and spring constants were measured to be around 1.35 N/m. Figure 1 shows the setup for an EFM experiment. A conductive AFM probe is electrically connected to a conductive substrate, forming a capacitor. A tapping-mode topographic scan of the sample is recorded on the first pass of a given line with no bias applied between the surface and the probe. On the second pass, the probe is lifted by a set amount, zlift (Figure 1), and scanned at a constant height above the substrate while a voltage, V ) VDC + VAC sin(ωt), is applied between them, and the probe is dithered mechanically at its natural frequency. The electrostatic force acting on the probe has the form where Vtot ) æ + VDC + VAC sin(ωt) and æ is the contact potential difference between the surface and the probe. The gradient of this force effectively shifts the cantilever’s mechanical resonance frequency,25 ν, by The components of ∆ν that oscillate at ω and 2ω are recorded using lock-in detection. During imaging, VDC was set to zero out the contact potential between the substrate and the probe. Thus, there is no DC electric field between the tip and substrate. Such static fields, if present, can polarize substrates and have been used to image Si doping profiles.26 By using previously reported models, the tipsurface interaction can be interpreted quantitatively.19 The details of the charge calculation methodology will be reported elsewhere. Figure 2a shows the images of surface topography, polarizability (∂F2ω/∂z), and charge (∂Fω/∂z) for a sample of CdS-capped CdSe nanocrystals on p-type silicon with a 2-nm layer of thermally grown oxide. The particles in this Figure are exposed only to room light and the below-band-gap 670nm diode laser that tracks the AFM cantilever motion. Most of the particles are uncharged; only 2 particles out of about 100 appear to carry chargesone positive (bright) and one negative (dark). Figure 2b shows the same sample area as in Figure 2a within ∼15 min of the 442-nm laser being turned on. In this first image, more charged particles appear. Some of the charged particles are circled to show the correspondence between the three images. Some of the particles appear to be blinking on and off on a time scale of one line scan of the image (∼1 s), which can be seen by an increase in the positive signal over nanocrystals that fluctuates from line to line whereas others maintain their state over longer time periods. The number of charged particles does not vary much with laser exposure time after Figure 1. EFM experimental setup. FEFM ) 1 2 dC dz Vtot 2 + EzQtip