The spectrum of the lamp used in the short-circuit current measurement, the nanorod absorption and emission spectra (of the homogeneous sample) and the silicon photodetector QE. The lamp is not well suited for the rods as it peaks in the region where the rod absorption is small. The lamp spectrum is cut off below 400nm by a filter. A realistic solar spectrum, such as the AM1.5 spectrum, would be much blue richer compared to the lamp and would therefore be absorbed better by the rods. 

Context in source publication

Context 1

... nanorods [5] (see Fig. 3) were grown at the National Nanotechnology Laboratory of CNR-INFM. They consist of a CdSe core and a CdS shell. Several properties of the nanorods depend on the aspect ratio (AR). The rods used had a diameter of about 5nm and a length of about 20nm, and thus an aspect ratio of 4. The corresponding luminescence quantum efficiency (QE) in solution was about 70% [5]. Moreover, the nanorods are expected to exhibit anisotropic emission, with a maximal emission in the plane perpendicular to the long axis. This effect, however, is only visible when the rods collectively align along a preferred orientation. In our samples we assume random and uncorrelated orientations. The absorption (see Fig. 4) and photoluminescence (PL) spectra of the nanorods in the concentrators were measured and found to be comparable to the spectra in solution. The absorption spectrum features a peak at longer wavelengths that arises from excitons in the core and overlaps strongly with the emission peak (see Fig. 5). However, the overall absorption is dominated by excitons in the shell, such that self-absorptions are relatively suppressed. A short-circuit current measurement technique was applied [6] (see Fig. 6), in which the top surface of the sample was illuminated using a lamp with a known spectrum (see Fig. 7). A silicon photodetector was used to map out the light intensity reaching the top surface. The same cell was used to scan one edge of the concentrator and detect the luminescent output. The arising short-circuit current was registered and the associated current density (J SC ) deduced. Based on the incident intensity on the front surface, the experimental PL leaving the edge and the known spectral and angular response of the detector, the photon number and concentration ratio (CR) were computed. The CR is defined as the photon flux leaving the detection edge divided by the incident flux. The results of the characterization are shown in Table 1. The photon concentration ratios (CRs) are low because these LSCs are test devices and not large enough to achieve concentration. Furthermore, the nanorods will perform better under a real solar spectrum with high intensities in the short wavelength regime, where the nanorod absorption is high. We have developed a versatile raytrace model, which can be applied to a variety of LSC configurations including thin films. The model is based on geometrical optics and uses a Monte Carlo approach to generate the outcome of events such as the absorption and emission of light by luminescent centers and reflection or transmission at surfaces. It can simulate PV cells and compute output spectra, short-circuit currents and other properties. The model has been verified through comparison with other, previously established models [7]. The short-circuit current measurement was simulated with the raytrace model. Beside the LSC characteristics, the lamp spectrum and intensity distribution over the surface were input to the model. The silicon photodetector was modeled as well, using measured spectral and angular responses. A further parameter in the model was the fundamental PL spectrum (at infinite dilution) of the nanorods, which was not known. In order to deduce it, a generic PL spectrum was adjusted iteratively until the model output matched the PL spectrum measured at the edge of the sample (see Fig. 8). The fundamental PL is not an entirely intrinsic property of the nanorod, but also depends on the host material. Unlike the homogeneous sample, the two thin film samples comprised the same host material and were therefore expected to have the same fundamental PL, despite different measured PLs (with peaks at 601nm and 605nm). The raytrace model proved consistent as it extracted the same fundamental PL (with a peak at 600nm) for both. Fig. 8. An example of fitting the fundamental photoluminescence (PL) spectrum so that the output from the raytrace model matches the experimentally measured PL. The output PL depends not only on the fundamental PL, but also on the quantum efficiency (QE) of the luminescent centers. Therefore the QE needed to be fitted simultaneously with the optimization of the PL position. This fit for the homogeneous sample yielded a value of (67±4)%, which is in good agreement with the ~70% quoted for the rods with this aspect ratio in solution [5]. Using this QE in the simulation of the thin film samples led to J SC and photon ratio predictions that were much higher than the measured values. This meant that the thin film LSCs were performing significantly worse than the homogeneous one. Their effective QEs were found to be 30% and 40% respectively. However, previous experiments have shown that homogeneous and thin film samples have similar performance for comparable absorptivities. The low output from the thin film samples could be a consequence of nanorod agglomerations in the highly concentrated film or of some visible macroscopic defects in the film. In order to find out whether the nanorods under examination showed less self-absorption compared to quantum dots, we used the absorption and emission spectra of a commercially available green QD and made a comparison with the raytrace model. All LSC specifications were identical for both the QD concentrator and the nanorod concentrator. Practical dimensions of 1m x 1m x 3mm and AM1.5 irradiance were chosen. The nanorod QE of 67% was applied to the QD and the absorption was adjusted such that equal amounts of incident light were absorbed by the concentrators (see Fig. 9). The relative difference in the absorption was less than 0.5%. The modeling showed that the QD LSC had (7±1) % more re-absorptions than the nanorod LSC. This small difference, however, affected the emission out of the edges strongly. The nanorod concentrator emitted (1.12±0.05) % of all incident photons (in the range from 300 to 700nm) out of the edges, whereas the QD concentrator emitted only (0.65±0.05) %. This demonstrated that the nanorods have less re-absorption and associated losses. Nanorod LSCs were successfully fabricated at the Fraunhofer IAP using core-shell nanorods. The absorption and emission spectra of the nanorods show a small overlap, which motivates their use in the LSC. The nanorod spectra in the LSC matrix were comparable to those in solution. A homogeneous and two thin film concentrators were characterized by means of a short- circuit current measurement technique. This experiment was simulated with the raytrace model. As part of the modeling, the fundamental PL of the nanorods and their luminescence QE were fitted. The QE value obtained by this method for the nanorods in the homogeneous LSC (67 ± 3)% was in good agreement with the value of 70% for nanorods with this aspect ratio in solution [5]. This result confirmed the validity of the raytrace model and showed that nanorods can be incorporated into LSCs without a significant reduction in luminescence quantum efficiency. The thin film LSCs showed significant losses compared to the prediction. These could be explained by agglomeration of nanorods or macroscopic defects in the thin film. Finally, the raytrace model was applied to demonstrate that nanorods have less self-absorption compared to quantum dots and consequently less re- absorption losses. This makes nanorods very attractive for incorporation in luminescent solar concentrators. The authors would like to acknowledge EPSRC for a studentship and the European Commission Framework VI Integrated Project FULLSPECTRUM (Ref. N: SES6-CT- 2003-502620) for ...