Another limitation of the methods described here results from the fact that the calculations have been performed for monodis-perse and perfectly spherical particles, whereas the experimental data are always collected over a distribution of particle sizes and shapes. As a result of that, the experimentally determined width and position of the surface plasmon resonance may be broadened and/or shifted as compared to the calculations. For example, the position of Aspr for oblate particles is shifted to larger wavelength, and hence the corresponding size calculated form eq 10 will be
too large, whereas the size calculated from eq 11 will be roughly consistent with the real size. For a broad distribution of particle sizes on the other hand, the size value calculated from eq 10 will be hardly influenced, whereas the size calculated from eq 11 will be too small. These considerations demonstrate that differences between the results of eqs 10 and 11 point to oblate particles and/ or to a broad size distribution, whereas consistency of the sizes calculated from eqs 10 and 11 can only be expected for fairly
spherical and monodisperse particles. However, the particles usedืhere were reasonably monodisperse and spherical exhibiting a standard deviation of particle sizes in the range from 10% to 20% of d and an average ratio of the longest to the shortest axis in the range of 1.1 to 1.2. These values are typical for the synthetic procedures described here, and the agreement between experi-ment and theory was quite good over the whole range of particle diameters (4-100 nm); hence, the particle sizes calculated from
eqs 10 and 11 were fairly consistent