AmericanChemicalSociety.com

The objective of our research is to study the interrelationship between impurity formation, surface state characteristics, and photogenerated charge carrier diffusion in ZnO nanowires.

During the current project period (September 1, 2009 to August 31, 2010), we have made important progress on revealing surface effects on photogenerated charge carrier diffusion, as well as the surface state characteristics as functions of the nanowire diameter, as described below.

Using a scanning photocurrent microscopy (SPCM) technique implemented on a near-field scanning optical microscope (NSOM), with schematics shown in Figure 1 (a), we have obtained the photogenerated hole diffusion length in individual n-type ZnO nanowires. Particularly, the near-field photocurrent maps obtained by the SPCM technique on a single nanowire Schottky diode, shown in Figure 1(b), exhibit a uniform (highly localized) photocurrent spatial profile under the forward (reverse)- biased condition. This is consistent with the rectifying electrical characteristics of the Schottky diodes. Under the reverse-biased condition, where the electric field is negligible in the nanowire except in the space-charge region associated with the Schottky diode, the photocurrent is induced by the collection of photogenerated holes (which diffuse from the optical generation region under the NSOM probe) by the Schottky electrode. Only when the NSOM probe is within a short distance (comparable or smaller than the hole diffusion length, L_D) from the Schottky electrode will such a photocurrent be detected. Therefore, the decay of the photocurrent, I_P, away from the Schottky electrode, can be used to extract L_D via a single exponential relation, I_P ~ exp (-x/L_D).

In our case, as L_D is comparable to the optical aperture (~ 200 nm) of the NSOM probe, we approximated the optical field of the NSOM probe with a Gaussian function and used an exponential-Gaussian convoluted function to fit the spatial profile of I_P, as shown in Figure 2. This procedure allowed us to extract L_D significantly smaller than the NSOM probe size. Figure 3(a) shows measured L_D (solid circles) as a function of nanowire diameter. For nanowires with diameters larger (smaller) than ~ 40 nm, L_D exhibits a negligible (significant) diameter dependence. Particularly, L_D approaches ~ 250 nm as the diameter increases beyond 40 nm. This value is between 440 nm and 130 nm, which were reported for bulk n-type ZnO with the doping levels of 10¹⁴ cm^-3 and 10¹⁸ cm^-3, respectively. As L_D generally increases with the decreasing electron concentration, we believe that L_D obtained for large-diameter nanowires (~ 250 nm) corresponds approximately to the bulk value, considering that the electron concentration of ZnO nanowires is ~ 10¹⁷ cm^-3. This absence of surface effects is consistent with previous findings on bulk ZnO that surface states due to atomic dangling bonds are resonant with conduction and valence bands, thus these surface states do not contribute to carrier capture/recombination processes.

For carrier transport in nanowires with diameter smaller than 40 nm, surface effects play an important role. The main mechanism of surface effects is the carrier recombination via mid-bandgap surface states. To elucidate this, we used a simple model based on carrier recombination mediated by mid-bandgap surface states. In particular, with L_D given by (Dτ_eff)^1/2, where D and τ_eff are the diffusion constant (=k_BTµ_p/q, with µ_p as the hole mobility) and the effective lifetime of holes, respectively, the surface-mediated carrier recombination leads to a decrease in τ_eff and thus L_D. This process can be described by a simple model based on the continuity equation. With this, we modeled L_D as a function of nanowire diameter [dotted blue line in Figure 3(a)].

A diameter-dependent surface recombination velocity (S) is required for a good fitting of the data using this model. The value of S ranges from ~ 1.5 × 10⁴ cm/s to ~ 0.1 cm/s as the diameter changes from 25 nm to 60 nm. At room temperature, using a carrier capture cross section (σ) on the order of 10^-15 cm², we obtain a surface state density (N) on the order of 10¹¹ cm^-2 and 10⁶ cm^-2 for 25 nm and 60 nm diameter ZnO nanowires, respectively, as shown in Figure 3(b).

To the best of our knowledge, this is the first experimental observation of a diameter-dependent surface electronic structure in semiconductor nanowires, and this might be a universal phenomenon for many semiconductor nanostructures. The increasing importance of surface effects, as the material dimension shrinks, is commonly understood as a consequence of the increasing surface-to-volume ratio. Our results suggest that, besides the surface-to-volume ratio, possible changes in the surface electronic structures need to be considered in understanding surface effects. Specifically for ZnO nanowires in solar cells, where the carrier transport is central to energy conversion efficiency, our results indicate that the surface passivation is only necessary for nanowires with diameters smaller than 40 nm.