To investigate the influence of the particle size on the optical response, measurements of NPs deposited on ZnO are investigated in more detail, since this is a typical con- figuration for solar cell deposition. Fig. 4.9 shows the optical response of Ag NP ag-
Figure 4.9.:UV/Vis measurements of Ag NPs on ZnO, the ZnO absorption is subtracted in these measurements. Different NP distributions with mean particle sizes between 10 and 42 nm are applied. In addition the calculated normalized extinction for each NP size distribution is shown below. The tendency and roughly the position of the calculated resonance fits to the observed peaks in the absorption spectra.
glomerates deposited on LPCVD ZnO substrates with a rough surface morphology. This configuration is used, since it allows to measure the optical response without thin film interferences emerging at flat surfaces. The deposited silver layer thicknesses are chosen to be 3, 5, 7 and 10 nm, equal to those used for the formation of the NP films in Fig. 4.6. The measured absorption spectra shown here are already corrected for the absorption within the underlying ZnO layer, i.e. the ZnO absorption is subtracted. Hence, only the response of the Ag NPs is illustrated here. The nanoparticle island films show a strong absorption. For all particle sizes a significant peak, according to the plasmonic resonances of the particles, is observed. An increase of the mean particle
size results in a shift of the peak towards longer wavelengths. That this is an effect of the LSP resonances is checked by calculations shown on the bottom. The normalized extinction spectra are calculated by applying the measured Ag NP size distributions to a dielectric environment of n = 2, according to ZnO. The procedure of utilizing the size distribution histograms for the calculations was explained in more detail in section 2.4.2. Since the measurement of particle size on the rough ZnO surface is not possible in an adequate manner, the distributions measured for the planar surface in Fig. 4.6 are employed. The calculated spectra overlap with the observed absorption spectra. Their maximum positions agree quite well for the small particles and deviate for larger particles. The measured absorption peaks are broader than the calculated extinction spectra. This could be related to interactions of neighboring particles [188] due to their close separation that is not taken into account in the calculations. In addition the calculations expect a size distribution of spherical particles that is indeed not the case (as discussed before). However the overall trend is well reproduced with the conducted calculations. Obviously the deviation of the particle size in the vertical dimension is of minor importance, since the polarization vertical to the direction of incident light is more important. Furthermore the peaks in the optical measurements exhibit an increased amplitude with increasing particle size. This is related to the larger volume of silver, evident for the larger particles that results in a stronger absorption [189]. For the ZnO environment the particle resonances are centered around 500 nm according to its refractive index of n ≈ 2.
For the application of nanoparticles to solar cells the particles have to be integrated into the silicon absorber. This results in a strong variation of the plasmon resonance as discussed earlier. Incorporating NPs in a silicon environment (n ≈ 4) a distinct peak is observed, as shown in Fig. 4.10. This peak is shifted to 800-850 nm. Here very thin layers of a-Si:H are applied which facilitate a strong absorption below 600 nm, while they show a weak absorption above 700 nm. Therefore the LSP resonance is explicitly revealed. Again Mie simulations for the applied Ag NP size distribution are conducted and shown below. The shape of the calculated spectrum is a bit more irregular. This is related to the larger refractive index, which makes the simulation more sensitive to contributions of different particles sizes and results in a substructure of the spectra compared to the case for lower index material. However, the maximum position is in good agreement with the measurement. The absorption peak is again significantly broader, maybe related to interactions of the nanoparticles, that are not taken into account in the calculations.
4.3. SUMMARY
Figure 4.10.:UV/Vis measurements of Ag NPs in an a-Si:H environment. Here silver NPs of (23 ± 9) nm are incorporated in between two 30 nm thin a-Si:H layers. The calculated normalized extinction according the the particle size distribution in an environment of n = 4 is plotted below.
4.3. Summary
In this chapter the preparation of thin silver films and nanoparticle agglomerates, de- posited by thermal evaporation and sputtering, is investigated. The findings concerning the structural and optical properties of deposited layers and silver island films can be summarized as the following:
• Thermal evaporation of layers with a thickness below 10 nm results in the forma- tion of silver islands. This is most likely related to a Volmer-Weber growth mode of silver. A heat transfer from the evaporation source to the substrate may also contribute but is not the driving force.
• Using sputter deposition also island formation is observed. Under certain sputter conditions it is however possible to deposit dense, closed films.
• The optical response of a closed film considerably differs from island agglomerates. The island films exhibit a spectrally narrow, increased absorption, related to the
LSP resonance of the metal islands. The reflection is decreased over a broad range.
• With the setup for sputtering on large area substrates a deposition with nanome- ter precision is limited, compared to the small area deposition. However silver layers with a thickness below 10 nm are reproducibly deposited. The large area deposition allows for a better process integration into large area deposition of previous or following layers in device manufacturing.
• The silver films irrespective of deposition conditions and method are crystalline with a dominant orientation of the fcc (111) planes parallel to the substrate. • Annealing of the silver films forms nanoparticles or results in a better separation
of already formed islands. The particle size is dominantly influenced by the deposited silver layer thickness. With increasing layer thickness the formed NPs increase in average size. In addition the NP distribution width increases with growing particle size. AFM scans indicate that the particles are oblate shaped, their height is smaller than their lateral dimension.
• Optical measurements of annealed samples show an absorption peak, related to the LSP resonances of the particles. The position of this resonance is influenced by the size of the NPs as well as the used dielectric environment. An increase of the mean particle size red-shifts the absorption peak. By changing the NP environment from ZnO with a refractive index of n = 2 to a-Si:H with an index of n = 4, the resonance is shifted from about 500 nm to 850 nm. These effects are in agreement with Mie calculations for the used particle distributions in the according environments of ZnO and a-Si:H. Calculated spectra overlap with the measured absorption curves.
The observed optical properties of the silver NPs reveal their resonant absorption be- havior due to the LSP resonances. The controlled deposition of known particle sizes in combination with the ability to control the dielectric environment of the NPs gives a tool for manipulating the spectral resonance position. In the following chapter, this is utilized to influence the absorption of a-Si:H solar cells in its low absorption regime. Especially the strong fields in the direct NP environment are of importance in this context.