Time-Resolved Photodynamics of Triangular-Shaped Silver

Mahmoud A. Mahmoud , Paul Szymanski , and Mostafa A. El-Sayed ..... J. Fedou , S. Viarbitskaya , R. Marty , J. Sharma , V. Paillard , E. Dujardin , A...
0 downloads 0 Views 213KB Size
NANO LETTERS

Time-Resolved Photodynamics of Triangular-Shaped Silver Nanoplates

2006 Vol. 6, No. 1 7-10

Luigi Bonacina, Andrea Callegari, Camilla Bonati, Frank van Mourik, and Majed Chergui* Laboratoire de Spectroscopie Ultrarapide, Ecole Polytechnique Fe´ de´ rale de Lausanne (EPFL), CH-1015 Lausanne, Switzerland Received October 28, 2005; Revised Manuscript Received November 16, 2005

ABSTRACT We measured the ultrafast response of triangular silver nanoparticles upon femtosecond excitation of their plasmon resonance. After a fast electron relaxation, the signature of a bimodal mechanical vibration of the particle is apparent as a spectral modulation. We identify the two lowest frequency, totally symmetric vibrations of the particle as responsible for this modulation, through their influence on the plasmon peak position and width, in full agreement with the results of a variational elastodynamic model that is also presented. From the analysis of the phase we conclude that thermal expansion and electron pressure, respectively, are responsible for the excitation of the two vibrations.

The short-time optical response of metals involves the excitation-relaxation dynamics of electrons, plasmons, and phonons. In nanoparticles (NPs), these dynamics are further modified by confinement and by the interplay between bulk and surface effects. Characterizing the optical response of metal nanoparticles, therefore, provides fundamental insight into the physics of these systems and essential information for nanoparticle-based technological applications. Ultrafast pump-probe spectroscopy has been applied to probe the response of spherical metal nanoparticles as a function of particle size (from a few to a few hundred nanometers), composition (mainly noble metals), and environment (liquid solvents, glassy matrixes).1-9 The influence of shape has received much less attention, mostly because of the difficulties associated with preparing uniform samples of nonspherical nanoparticles and with interpreting the results. Measurements of the dynamics of such particles remain, at present, limited to a handful of cases: Ag nanoellipsoids,10 Au nanorods11,24 and, very recently, Ag nanoprisms,12,13 and arrays of Au nanoprisms.14 The chain of processes triggered by impulsive laser excitation with photon energies below the intraband transition (≈4 eV, for silver) has been characterized for spherical metal NPs.6,8,15 Conduction band electrons driven out of equilibrium rapidly evolve into a hot thermal gas which cools in a few picoseconds, warming up the lattice and causing it to expand. Both the pressure from the hot electrons and the sudden thermal expansion can excite mechanical vibrations of the particle. Eventually, all of this mechanical and thermal energy is lost to the surrounding environment and the particle returns to its initial state. Each of these processes is accompanied 10.1021/nl052131+ CCC: $33.50 Published on Web 12/06/2005

© 2006 American Chemical Society

by a change of the optical spectrum as the surface plasmon resonance (SPR) shifts and/or broadens. The lower symmetry of nonspherical particles results in a richer optical response but, at the same time, complicates its interpretation with practical and conceptual issues. Specifically, the triangular nanoparticles investigated here differ from spherical ones in two important ways. First, the dominant mechanism of vibrational excitation for spherical particles is thermal expansion when the particle is large (r > 6 nm) and electron pressure when it is small.16 Nonspherical particles have at least two characteristic length scales, and it is not a priori obvious which is the relevant one in determining the dominant mechanism. Second, the SPR frequency of spherical NPs is rather insensitive to their size (because of scale invariance of the underlying equations17), up to sizes of several tens of nanometers where retardation effects become nonnegligible. Therefore, the finite dispersion of the sample size does not affect its optical spectrum. Conversely, nonspherical particles do not have properties of scale invariance, and at least one of their dimensions is usually large enough to make retardation effects nonnegligible; hence, the sample size dispersion does affect the optical spectrum. The system under study is a dilute aqueous suspension of NPs in the shape of equilateral triangles, synthesized in our laboratory with a photochemical method that allows control of their size and shape and,18 hence, tuning of the SPR peak position. The sample studied here (Figure 1) consists of NPs of ∼8 nm thickness, ∼70 nm edge length, and SPR peaking, in water, at 730 nm. We measure its time- and wavelengthdependent optical response via transient absorption, in a

Figure 1. (a-c) Transmission electron microscope images of a typical sample of silver nanoparticles deposited on a carbon film. (c) Stacks of particles, easily found on the film, allow an estimation of their thickness (∼8 nm). (d) Histogram: lateral size distribution of the triangles in the sample; Line: fit of a log-normal function to the histogram, average size: 68.9 nm. (e) UV-vis extinction spectrum of the sample. The dotted line indicates the position of the pump-pulse excitation at 800 nm, and the hatched area indicates the spectrum of the subset of particles excited by the laser, as determined by the model described in the text.

pump-probe scheme. We use a pump pulse from an amplified Ti:sapphire laser (∼70 fs, 800 nm, 1 kHz) and a broadband (white-light) probe pulse generated in a CaF2 window. The white-light beam is split into a probe and a reference to correct for pulse-to-pulse intensity fluctuations.19 The nanoparticle solution is allowed to constantly flow in a closed cycle through a quartz flow cell, to refresh the sample at each shot and minimize photodegradation. Figure 2 shows a full two-dimensional dataset of wavelength-resolved transient signal and representative time slices at various wavelengths. Given the wealth of information contained therein, we first focus on and analyze a single slice, shown in Figure 3. The most prominent features, at different time scales, are a strong and short-lived initial bleach (