1 Supporting Information for Shape Transformation of Gold Nanoplates and their Surface Plasmon Characterization: Triangular to Hexagonal Nanoplates Soonchang Hong, Kevin L. Shuford *,, and Sungho Park *,, Department of Chemistry, Department of Energy Science & SKKU Advanced Institute of Nanotechnology Sungkyunkwan University Suwon 440-746, South Korea Fax #: 82-31-290-7075 Department of Chemistry Drexel University 3141 Chestnut Street, Philadelphia, PA 19104, USA *Corresponding Author s Email: Kevin L. Shuford (shuford@drexel.edu), Sungho Park (spark72@skku.edu)
2 Experimental Section Synthesis of Au triangular nanoplates: The Au triangular nanoplates solution (edge L = 136 ± 8 nm, thickness = 8 ± 1 nm) was prepared using a conventional methodology. S1 Briefly, 0.5 ml of 20 mm aqueous HAuCl 4 3H 2 O solution was added to 36.5 ml of deionized water (Millipore). And then one milliliter of a 10 mm aqueous solution of sodium citrate and 1 ml of 100 mm aqueous NaBH 4 (Ice-cold) solution were added with vigorous stirring. This solution contains Au spherical seed nanoparticles. In order to prepare triangular nanoplates, growth solutions were prepared as follows. A mixture of 108 ml of 0.05 M aqueous CTAB (from Fluka) solution and 54 μl of 0.1 M aqueous NaI solution was divided into three containers labeled with 1, 2, and 3. Container 1 and 2 hold 9 ml of the mixture and container 3 holds the rest solution of 90 ml. Then, a mixture of 125 μl of 20 mm aqueous HAuCl 4 3H 2 O solution, 50 μl of 100 mm NaOH, and 50 μl of 100 mm ascorbic acid were added to each container 1 and 2. A mixture of 1.25 ml of 20 mm HAuCl 4 3H 2 O, 0.5 ml of 100 mm NaOH, and 0.5 ml of 100 mm ascorbic acid were added to container 3. One ml of the seed solution was added to the container 1 with mild shaking. Then, one ml of container 1 solution was added container 2. After 5 second shaking, the whole solution of container 2 was added to container 3. After 30 minutes, the color of container 3 shows magenta-purple. Synthesis of Au hexagonal nanoplates: 75 ml of 0.5 M CTAB was mixed with 37.5 μl of 0.1 M NaI. Then, 3 ml of 20 mm HAuCl 4 3H 2 O and 5 ml of the prism solution were added to the solution. Then, 22.5 ml of 5.3 mm ascorbic acid was added dropwise to the solution over a period of 45min (0.5mL/min) with vigorous stirring. The color of solution changed from yellow to greenish brown and allowed to sit overnight. In order to control the thickness of Au hexagonal nanoplates, the similar experimental procedure was followed but different amount of reactants was used as described below.
3 Thickness control (See Figure S2) A: 75 ml of 0.5 M CTAB + 37.5 μl of 0.1 M NaI + 2 ml of 20 mm HAuCl 4 3H 2 O + prism seed 1 ml +5.3 mm AA 15 ml (0.5 ml/min) B: 75 ml of 0.5 M CTAB + 37.5 μl of 0.1 M NaI + 3 ml of 20 mm HAuCl 4 3H 2 O + prism seed 1mL +5.3 mm AA 22.5mL (0.5 ml/min) C: 75 ml of 0.5 M CTAB + 37.5 μl of 0.1 M NaI + 4 ml of 20 mm HAuCl 4 3H 2 O + prism seed 1 ml +5.3 mm AA 30 ml(0.5 ml/min) D: 75 ml of 0.5 M CTAB + 37.5 μl of 0.1 M NaI + 5 ml of 20 mm HAuCl 4 3H 2 O + prism seed 1 ml +6.6 mm AA 30 ml (0.5 ml/min) DDA Calculations The optical properties of Au nanoplates have been calculated using the Discrete Dipole Approximation (DDA). S2,S3 DDA represents the nanoparticles volume as a square array of point dipoles. Each dipole obtains an oscillating polarization from the local field at that lattice site, which is composed of the incident plane wave and the fields radiated from the other dipoles in the array. The dipole polarizability incorporates the optical constants of the metal and is assigned based upon a lattice dispersion relation. S4 Here we have utilized experimentally determined values for the refractive index of Au. S5 The set of coupled dipole equations compose a large, dense matrix equation that is solved iteratively for the induced polarizations, which are then used to calculate the nanoparticle extinction. The calculated extinction spectra presented in the manuscript have been averaged over several orientations (i.e. nanoparticle positions relative to an incident plane wave). This is done to account for the numerous excitations present in a solution, where
4 nanoparticles can freely translate and rotate. The coordinate system was chosen such that the shortest dimension of the scatterer morphology is along x, and the flat nanoplate cross section is parallel to the y-z plane. For the highly symmetric particles (disks and hexagonal prisms), a good approximation for the full orientational average is obtained by sampling over a 90 o rotation of the particle around z (we call this angle θ), while keeping the incident plane wave fixed, propagating along x and polarized along y. For large particles that support higher order modes, it is important to sample the intermediate angles, as particles with this symmetry have selection rules that suppress excitation of certain modes under normal excitation. S6 This will be demonstrated below. The spectrum for triangular prism nanoparticles was obtained by averaging three representative orientations: propagation vector k along the prism axis and polarization vector E in the triangular plane, both k and E in the triangular plane, and k in the plane and E out of the plane. Previous studies on the optical properties of triangular prisms have determined that the full orientational average is well approximated by sampling these orientations. S7,S8 The nanoparticle morphologies used for the calculations were based upon the SEM images and experimentally determined dimensions. In Fig. 3D, trace a was modeled as a perfect triangular prism with a 7 nm thickness and a 136 nm edge length. Trace b was modeled as a perfect disk with a 100 nm diameter and a 7 nm thickness. This diameter is ~20 nm larger than the experimentally determined value. During the morphological transition stage modeled here, the experimental samples are a mixture of disks and polygon prisms, which results in a broader peak that is red-shifted because of the polygon vertices. The disk diameter used in the simulation was adjusted for this effect to more accurately represent the experimental spectrum in this case. Traces c and d were
5 modeled as perfect hexagonal prisms with 97 nm and 210 nm edges respectively. These morphologies were shown experimentally to thicken, so the particle thickness for traces c and d were taken as 10 nm and 19 nm respectively. The peaks in the extinction spectra were assigned to a particular multipole order based upon the induced polarization of the nanoparticle at that wavelength. DDA discretizes the polarization using many dipoles. Vector plots of the dipoles collectively point from regions of negative induced surface charge to regions of positive charge. The number and arrangement of charges indicates the order of the electric multipole primarily excited at that frequency. Note that complex geometries support modes that deviate from those of spherical particles, and as a result, the multipole patterns can be quite different than the familiar properties of spherical harmonics. This is especially true for high-order multipoles of anisotropic particles, where assignment becomes arduous. Figures S3-S6 show vector plots of the induced polarization for a triangular prism, a disk, a small hexagonal prism, and a large hexagonal prism, respectively. The great majority of dipoles used in the calculation have been removed for viewing clarity. Moreover, charges have been added to the plots to denote the polarity of a given spatial region and aid in the assignment of the multipole polarization state. Each of the vector plots is for a single orientation, which was the primary contributor to the orientationally averaged spectrum presented in the manuscript (Fig. 3D). Figure S3, Panels A-C correspond to the L = 1, 2, and 3 (dipole, quadrupole, etc.) modes of a triangular prism nanoparticle. These polarization patterns are consistent with previous studies on this particle shape. S7 The second and third modes are very close in energy, and as a result, there is a slight mixing of character. This is most evident in the
6 quadrupole pattern (Panel B) on the far right edge, which bears some resemblance to the well defined charged regions present in the L = 3 mode in the same location. Figure S4 shows the polarization corresponding to the dipole mode of a disk shaped nanoparticle. This is a very clear example of a dipole excitation. Figure S5 displays the vector plots for a small hexagonal prism. The lower energy mode (Panel A) is the dipole mode and the higher energy mode (Panel B) corresponds to a quadrupole mode. The quadrupole mode is most efficiently excited when the incident polarization is rotated 45 o out of plane, which is the orientation for the vector plot shown in Panel B. As a result, the polarization pattern appears slightly skewed. Figure 5 in the main text shows the polarization of a large hexagonal prism. The inset displays the extinction calculated for normal excitation (θ = 0) and when the incident field is rotated 45 o from normal excitation (θ = 45). Note that the lower energy modes at 1037 and 734 nm are not excited when the incident polarization is parallel to the hexagonal cross section. This is reminiscent of the selection rule effect in nanorods under normal excitation. S6 Panels A and B correspond to the L = 2 (quadrupole) and L = 3 modes of a hexagonal prism. The pattern in Panel B is slightly unusual; however, the number of charges is consistent with what is expected for a L = 3 mode. The L = 4 mode is shown in Panel C and is a straightforward assignment. The dipole mode (L = 1) resides beyond 1400 nm and is not presented here.
7 Figure S1 FESEM images of Au hexagonal nanoplates when (A) they are stacked and (B) dispersed. Figure S2 FESEM images of Au hexagonal nanoplates with different physical dimensions, (A) edge L = 358 (±88) nm, thickness t = 35 (±7) nm, (B) edge L = 340 (±51) nm, thickness t = 51 (±9) nm, (C) edge L = 344 (±80) nm, thickness t = 79 (±4) nm, and (D) edge L = 387 (±61) nm, thickness t = 170 (±32) nm. Insets show their sideview. (E) UV-vis-NIR spectra of each sample (A)~(D) are plotted (a)~(d), respectively.
Figure S3 Polarization of a triangular prism nanoparticle with a 136 nm edge length and 7 nm height. The plots show a surface plane of the particle, where 80% of the dipoles have been removed for viewing clarity. Panels A, B, and C correspond to the first three multipoles (dipole, quadrupole, etc.). The particle orientation with respect to incident field and the excitation wavelength are displayed in each panel. 8
9 Figure S4 Polarization of a disk nanoparticle with a 100 nm diameter and a 7 nm height. The plot shows a surface plane of the particle, where 90% of the dipoles have been removed for viewing clarity. The particle orientation with respect to incident field and the excitation wavelength are displayed in the plot. Figure S5 Polarization of a hexagonal prism nanoparticle with a 97 nm edge length and a 10 nm height. The plots show a surface plane of the particle, where 80% of the dipoles have been removed for viewing clarity. Panels A and B correspond to the first two multipoles (dipole, quadrupole). The particle orientation with respect to incident field is shown in Panel A. The orientation for Panel B corresponds to 45 o rotation into the page about the axis perpendicular to both k and E depicted in Panel A.
10 References S1. J. E. Millstone, W. Wei, M. R. Jones, H. Yoo, C. A. Mirkin, Nano Lett. 2008, 8, 2526. S2. B. T. Draine, Astrophys. J. 1988, 333, 848. S3. B. T. Draine, P. J. Flatau, J. Opt. Soc. Am. A 1994, 11, 1491. S4. B. T. Draine, J. Goodman, Astrophys. J. 1993, 405, 685. S5. P. B. Johnson, R. W. Christy, Phys. Rev. B 1972, 6, 4370. S6. E. K. Payne, K. L. Shuford, S. Park, G. C. Schatz, C. A. Mirkin, J. Phys. Chem. B 2006, 110, 2150. S7 K. L. Shuford, M. A. Ratner, G. C. Schatz, J. Chem. Phys. 2005, 123, 114713. S8 R. Jin, Y. Cao, C. A. Mirkin, K. L. Kelly, G. C. Schatz, J. G. Zheng, Science 2001, 97, 3831.