1. Nano-Scale Characterization: M. Pinar Mengüç RADIATIVE TRANSFER LABORATORY Mechanical Engineering Department UNIVERSITY OF KENTUCKY College of Engineering Experimental Methods: (a) Ellipsometry; (b) Surface Plasmon reflectance; (c) Surface Plasmon scattering.

2. Nano-Scale Characterization: M. Pinar Mengüç RADIATIVE TRANSFER LABORATORY Mechanical Engineering Department UNIVERSITY OF KENTUCKY College of Engineering PMT Scattered light optics Incident light optics Translation stage to control incident angle Rotational stage Light source Motor Control Unit

3. RADIATIVE TRANSFER LABORATORY Mechanical Engineering Department UNIVERSITY OF KENTUCKY College of Engineering Sensitivity Analysis A a sensitivity analysis is performed to determine the optimum conditions for characterization of gold nanoparticles and 2D-agglomerates via evanescent wave/surface plasmons. This analysis is of primary importance in order to determine the conditions for which a particular parameter can be estimated, and is a necessary step for the development of inversion techniques and optimal experiments. Sensitivities of M 11, M 12, M 33, and M 34 are calculated via the normalized sensitivity coefficients: The normalized sensitivity coefficients provide the variation of the output/measurement (normalized scattering matrix elements, M ij ) associated to a relative variation of one parameter of the system (ψ k ), when all other parameters (known η or to be estimated ψ l,l≠k ) are fixed. In a general way, the estimation of a parameter is considered to be conceivable when the normalized sensitivity coefficients are greater than 0.1, difficult but feasible between 0.01 and 0.1, and very difficult or even impossible below the threshold 0.01. It is assumed that the system is composed of single nanoparticles, as well as agglomerates (triangle, square, and linear chains). A given system is defined as a function of its composition in single spherical nanoparticles. 1- Sensitivity to composition For this case, all particles have a diameter d m of 40 nm, and are all located on a thin gold film (h = 0). Results for M 12 and M 33 are reported in Fig. 1. Figure 1: Averaged normalized sensitivity coefficient to the percentage of single nanoparticles % (d m = 40 nm, h = 0). (a) Sensitivity of M 12. (b) Sensitivity of M 33. (a)(b) (a) (b) Figure 2: Averaged normalized sensitivity coefficient to d m (h = 0). (a) Sensitivity of M12. (b) Sensitivity of M33. 2- Sensitivity to diameter It is assumed that there is a non-uniform distribution of diameters from 38 to 42 nm (h = 0); the averaged normalized sensitivity coefficients of M 12 and M 33 are presented in Fig. 2. Conclusions The averaged normalized sensitivity coefficients of M 11 (not shown) are always very low, which imply that the characterization should be done by using the polarization information. Moreover, in all cases performed in this work, the scattering matrix element M 33 in the range of observation angles  from 110 to 150º is found to have quite high sensitivity to all the pertinent parameters of interest. This window can therefore be used as a starting point for an experimental investigation and the development of an inversion technique. 020406080100120140160180 10 -3 10 -2 10 10 0 0 % 25 % 50 % 75 % 100 % X d m norm,avg [ M 12 ]  020406080100120140160180 10 -3 10 -2 10 10 0 10 % 15 % 25 % 50 % 75 % 100 % X % norm,avg [ M 12 ]  020406080100120140160180 10 -3 10 -2 10 10 0 10 % 15 % 25 % 50 % 75 % 100 % X % norm,avg [ M 33 ]  020406080100120140160180 10 -3 10 -2 10 10 0 0 % 25 % 50 % 75 % 100 % X d m norm,avg [ M 33 ]  Nano-Scale Characterization: M. Pinar Mengüç (w/ Mathieu Francoeur)