Dynamics of Energy Transfer in Large. Plasmonic Aluminum Nanoparticles

Transcription

1 Supporting Information Dynamics of Energy Transfer in Large Plasmonic Aluminum Nanoparticles Kenneth J. Smith,#, Yan Cheng,#, Ebuka S. Arinze,#, Nicole E. Kim, Arthur E. Bragg, Susanna M. Thon Department of Chemistry, Johns Hopkins University, 3400 N Charles St, Baltimore 21218, United States Department of Electrical and Computer Engineering, Johns Hopkins University, 3400 N Charles St, Baltimore 21218, United States S1

2 Figure S1. TEM images of aluminum NPs in (A), (B) and Figure 1A, were used to obtain the histogram (C) of the synthesized aluminum nanoparticle size. This analysis yields a size distribution with an average particle diameter of 98±12 nm. (D) FDTD-calculated spatial electromagnetic field profile for a 93 nm-diameter aluminum NP at the quadrupole resonance wavelength of 269 nm (color scale in a.u.). Figure S2. TEM images of aluminum nanoparticles before (A) and after (B) prolonged exposure to a 400 nm excitation source show no observable changes in nanoparticle morphology. S2

3 Figure S3: Fluence dependent transient absorption data conducted at 101 uj cm -2 (Top), 64 uj cm -2 (Middle), and 34 uj cm -2 (Bottom). Similar spectral responses were observed at all fluences with intensity having linear dependence on pump fluence Shown in Figure S4B. S3

4 A) B) Figure S4. (A) Time-dependent traces ( nm) at different pump fluences ranging from 134 uj cm -2 (red) to 34 uj cm -2 (yellow); fitting models described in the text are overlaid as dashed black lines. (B) Fluence dependence of max positive and negative extinction obtained in TAS measurements; fit reveals closely linear relationship between signal and fluence (slope = 1.29, see main text). S4

5 Figure S5: FDTD simulations of (top) absorption and (bottom) scattering cross sections of Al nanoparticle in IPA at different refractive indices (and corresponding dielectric constants). These comparisons reveal that the scattering contribution dominates absorption by an order of magnitude in most regions. Furthermore, local solvent heating (which generally induces reduction in refractive index and dielectric constants) can be expected to result in a net negative change in spectral scattering profile. S5

6 τ 2 τ 1 τ R φ a 1 a 2 a 3 γ c ps 0.91 ps 33 ps nm ps 0.3 ps 22 ps nm ps 3.5 ps 22 ps nm Table S1: Parameters for fits shown in Figure 3A in the text. S6

7 τ 2 τ 1 τ R φ a 1 a 2 a 3 γ c 134 uj 272 ps 0.91 ps 33 ps cm uj 191 ps 0.7 ps 25 ps cm uj 216 ps 0.8 ps 27 ps cm uj 183 ps 0.7 ps 25 ps cm -2 Table S2: Parameters for fits shown in Figure S4A. 134 uj cm uj cm uj cm uj cm -2 τ R 14.0 ps 13.8 ps 13.7 ps 13.3 ps Table S3: Oscillation period calculated from the truncated data sets ps shown in Figure 3B in the text. S7

8 Figure S6: Crude isolation of interband bleach contribution. (Top) Linear trend lines were calculated for NIR spectra obtained at each delay using points at 1000 nm and 1100 nm. The value of the trend line at the wavelength where the bleach maximum occurred (869 nm) was subtracted from measured value at this wavelength. The difference is plotted in the bottom panel. Although this method is a crude approximation for the shape of the scattering contributions in this wavelength range, this analysis shows that the interband transition recovers roughly on the timescale of thermal energy transport. S8

9 Two Interface Model Our model uses the following equations, adapted from the model in Reference 1, 1 for the temperature dynamics of the aluminum NP core (p), the oxide shell (o), and the solvent medium (m): T p (t) = 2 + π k 1(Rg 1 ) 2 u 2 e κ 1u 2 t/r 2 T 0 (u 2 (1 + Rg 1 ) k 1 Rg 1 ) 2 + (u 3 k 1 Rg 1 u) 2 du 2R T o (t) = T 0 k πr 1 (Rg 1 ) 2 o + 0 T m (t) = T 0 2r o πr k 2(r o g 2 ) 2 T p 0 u 2 e κ 1u 2 t/r 2 u(1 + T o 0 u 2 e κ 2u 2 t/r o 2 u(1 u 2 k1rg1 ) cos(u(r o R) )+(1 u2 (1+Rg1) R k1rg1 ) sin(u(r o R) ) R du (u 2 (1+Rg 1 ) k 1 Rg 1 ) 2 +(u 3 k 1 Rg 1 u) 2 u 2 k2rog2 ) cos(u(r r o ) )+(1 u2 (1+rog2) ro k2rog2 ) sin(u(r r o ) ) ro du (u 2 (1+r o g 2 ) k 2 r o g 2 ) 2 +(u 3 k 2 r o g 2 u) 2 where T o is the initial temperature increase of the nanostructure after optical pump excitation, c is the heat capacity per unit volume, G is the interface thermal conductance, Λ is the thermal conductivity, κ 1 = Λ o /c o, κ 2 = Λ m /c m, k 1 = 3c o /c p, k 2 = 3c m /c o, g 1 = G 1 /Λ o, and g 2 = G 2 /Λ m. The numerical subscript refers to the interface investigated (p - nanoparticle, o - oxide and m - medium), and interface 1 refers to the aluminum/aluminum oxide interface while interface 2 refers to the aluminum oxide/isopropanol. S9

10 A B Figure S7. Normalized calculated temperature evolutions at the surface of varying sizes of aluminum nanoparticles (10 nm to 100 nm in diameter) and 3 nm into the surrounding medium: (A) without the presence of an oxide layer, and (B) with the presence of an oxide layer. S10

11 A B C D Figure S8: Temperature evolution time constant (1/e) 3 nm into the solvent medium as a function of oxide thickness in the case of: (a) constant total nanoparticle size (the metal core diameter decreases with increasing oxide thickness) and (b) constant metal core size (the total size of the nanoparticle increases with increasing oxide thickness). Associated spectra for (a) and (b) are plotted in (c) and (d), respectively. Solid lines are for temperature evolution in the particles, and dashed lines are for temperature evolution in the solvent medium. The blue, orange, yellow, purple, green, cyan, and red spectra correspond to 0, 0.5 nm, 3.5 nm, 6.5 nm, 9.5 nm, 12.5 nm, and 15.5 nm oxide thicknesses respectively. These results indicate that changing the size of the aluminum metal core has a much greater influence on thermal energy transfer to the medium than changing the oxide thickness itself beyond a drastic decrease in the time constant after introduction of an oxide layer. S11

12 Although the thermal conductivities are different (205 Wm -1 K -1 for Al vs. 30 Wm -1 K -1 for AlOx), the heat capacities of aluminum metal and aluminum oxide are fairly similar (2.43e10 6 Jm -1 K -1 for Al vs. 3.48e10 6 Jm -1 K -1 for AlOx). The temperature evolution in the two-interface model depends more strongly on the interface properties than the size variation of the associated layers. In this case specifically, the oxide layer s primary role is to act as a heat sink in facilitating the heat transfer from the core material. As a result, the addition of an intermediary oxide layer drastically changes the temperature time constant in the solvent, and subsequent increases in the oxide layer thickness have a minimal effect when compared to the core aluminum thickness due to the much lower thermal conductivity of the aluminum oxide material (almost ten-fold lower than in the aluminum metal). S12

13 Figure S9. Solvent coherences during the pump-probe overlap were used to determine the instrument response function. The inset displays the calculated IRF Gaussian obtained by fitting the solvent coherences at time zero. The full width half max was calculated to be 262 fs. S13

14 References (1) Stoll, T.; Maioli, P.; Crut, A.; Rodal-Cedeira, S.; Pastoriza-Santos, I.; Vallée, F.; Del Fatti, N. Time-Resolved Investigations of the Cooling Dynamics of Metal Nanoparticles: Impact of Environment. J. Phys. Chem. C 2015, 119 (22), S14