Article pubs.acs.org/Langmuir
Surface Plasmon Resonance Properties of Single Elongated Nanoobjects: Gold Nanobipyramids and Nanorods Anna Lombardi, Matthieu Loumaigne, Aurélien Crut, Paolo Maioli, Natalia Del Fatti, and Fabrice Vallée* FemtoNanoOptics group, LASIM, Université Lyon 1-CNRS, 43 Boulevard du 11 Novembre, 69622 Villeurbanne, France
Miguel Spuch-Calvar, Julien Burgin, Jérome Majimel, and Mona Tréguer-Delapierre CNRS-Université Bordeaux, ICMCB, 87 Avenue du Dr. A. Schweitzer, 33608 Pessac, France ABSTRACT: The spectral characteristics (wavelength and line width) and the optical extinction cross-section of the longitudinal localized surface plasmon resonance (LSPR) of individual gold nanobipyramids have been quantitatively measured using the spatial modulation spectroscopy technique. The morphology of the same individual nanoparticles has been determined by transmission electron microscopy (TEM). The experimental results are thus interpreted with a numerical model using the TEM measured sizes of the particles as an input, and either including the substrate or assuming a mean homogeneous environment. Results are compared to those obtained for individual nanorods and also show the importance of the local environment of the particle on the detailed description of its spectral position and extinction amplitude.
1. INTRODUCTION Plasmonic effects on the optical response of metal nanoparticles and nanostructures have been extensively experimentally and theoretically investigated during the past decade. This interest has been motivated by the large possibilities they offer for manipulating and controlling the optical properties of metalbased nanoparticles, playing with their shape, size, composition, and organization.1−4 This versatility mostly exploits the dependence of the localized surface plasmon resonance (LSPR) on the morphology and environment of metal nanoparticles. Its characteristics, e.g., spectral position, quality factor (or spectral width), and amplitude (or extinction crosssection), can thus, to a certain extent, be adapted to specific applications. This has fostered the development of new methods to synthesize nano-objects of different morphologies such as rods, bipyramids, decahedra, triangles, cubes, or stars.5−11 In this context, elongated shaped objects such as gold nanorods present a strong LSPR at a wavelength adjustable in the red and near-infrared part of the spectrum by modifying their aspect ratio.1−7 Together with the now reached excellent control of their synthesis, this makes them particularly suited for many applications, for instance, in photothermal therapy or in nanosensing.3,12,13 Most of the targeted applications require further enhancement of nanoparticle−electromagnetic field interaction to improve their absorption or their sensitivity to changes of their environment. This can be done modifying the particle shape, for instance, sharpening the rod tips,3,12−16 or using nanobipyramids instead of nanorods, but requires precise © 2012 American Chemical Society
correlation of the LSPR characteristics with the nanoparticle morphology. Because of the unavoidable residual size and shape dispersion of synthesized nanoparticles, such correlation can only be done at a single particle level, combining optical spectroscopy and morphology characterization tools.17−24 The spatial modulation spectroscopy (SMS) technique is here particularly relevant because, as compared to other individual nano-object spectroscopic tools such as dark field microscopy,8,17 it yields access not only to the shape of the optical spectrum of the object but also to the amplitude of its extinction crosssection.14,25,26 Its combination with morphology determination of the same nano-object such as transmission electron microscopy (TEM) permits full and detailed comparison of its LSPR properties with predictions of theoretical models, adding the requirement of reproducing the amplitude of the extinction spectra in addition to their shape.18,19,26,27 This detailed comparison makes precise characterization of the investigated object and of its environment necessary. Consequently, comparison of the experimental results to those of theoretical models is also a powerful tool to obtain information on the particle and environment characteristics. It also permits computation of other important but difficult to access LSPR features, such as the electromagnetic field Special Issue: Colloidal Nanoplasmonics Received: January 14, 2012 Revised: February 24, 2012 Published: February 27, 2012 9027
dx.doi.org/10.1021/la300210h | Langmuir 2012, 28, 9027−9033
Langmuir
Article
Figure 1. Nanoparticle shape model and environment geometry used in numerical modeling of the optical extinction of individual gold nanobipyramids (a) and nanorods (d) on a silica substrate. The optical extinction spectra of two bipyramids and two nanorods measured by SMS are shown in panels b and e, respectively, and their corresponding TEM images are in panels c and f. The deduced sizes are L = 67 nm, W = 24.5 nm, r = 3.5 nm, and L = 77 nm, W = 24 nm, r = 4.2 nm for the bipyramids BP1 and BP2, respectively, and L = 36.5 nm, W = 12.5 nm and L = 39 nm, W = 11.7 nm for the rods NR1 and NR2, respectively. Fits of the individual extinction spectra (full lines) were performed using the TEM sizes and the environment refractive index nm as a parameter; nm = 1.05, 1.15, 1.42, and 1.41 for BP1, BP2, NR1, and NR2, respectively. The theoretical width has been adjusted to the experimental one. The dashed lines in b and e are the normalized ensemble extinction spectra of the initial colloidal solutions.
solubility), and the crystallized CTAB was centrifuged at 700 rpm for 5−10 min. The solution was subsequently further centrifuged between 1000 and 2000 rpm (depending on the particle size) for 2 h. The centrifugation steps were repeated several times to reduce the CTAB concentration as much as possible, but with enough care to avoid particle aggregation (excessive purification can lower the CTAB molecule density on the particle surface to a level that destabilizes the colloidal solution). Gold nanorods stabilized by CTAB in aqueous solutions were synthesized using the procedures described in refs 6 and 14. All the particles were characterized by UV−vis spectroscopy and TEM. The mean width (i.e., equatorial diameter) W of the nanobipyramids is about 25 nm, and their mean length L is in the 60−80 nm range. The mean width and length of the nanorods are 10 and 40 nm, respectively. The optical spectra of the aqueous solutions are dominated by the strong longitudinal LSPR of the nano-object. It shows up as a resonance in the red part of their spectrum, as illustrated Figure 1.1,2,5−7,15 As for nanorods,14,28−30 and as observed using dark-field microscopy,8 the large sensitivity of the LSPR wavelength, λR, on the particle aspect ratio η = L/W leads to inhomogeneous broadening of the resonance observed in ensemble measurements. This makes difficult precise analysis of the intrinsic characteristics of the LSPR that can be determined using individual nano-object spectroscopy. The extinction cross-section of one single nano-object from the previous solutions has been quantitatively measured using the SMS technique.14,25,26 Briefly, the SMS technique is based
amplitude and distribution in and around the object. We have used this approach to investigate the characteristics (amplitude, wavelength, and width) of the LSPR of single bipyramids, and to compare them to those of other elongated objects such as nanorods. The experimental data are compared to the results of a numerical model, using the measured morphology as input and explicitly taking into account the substrate used in the experimental studies.
2. EXPERIMENTS AND METHODS Gold bipyramids were synthesized using a modified version of the seed mediated approach developed by Liu et al.8 Briefly, gold seeds of 3−4 nm were first prepared by mixing, using fast stirring, an aqueous solution containing 0.25 mM HAuCl4 and 0.25 mM sodium citrate, with 0.6 mL of 12 mM NaBH4. The seed solution was stored in the fridge for 4 h prior to its use. Anisotropic growth of gold seeds was performed using a growth solution of 0.45 mL of 11 mM HAuCl4, 0.038 mL of 26 mM AgNO3, 0.2 mL of HCl, and 0.076 mL of ascorbic acid dissolved in 10 mL of 0.1 M cetyltrimethylammonium bromide (CTAB). This solution was prepared in a thermal bath over 27 °C due to the solubility properties of the CTAB. For preparation of different sized bipyramids, the seed amount added to the growth solution was changed (i.e., 0.02, 0.2, and 0.4 mL), allowing the production of elongated nanoparticles with different aspect ratios. A purification of the Au bipyramids solution was systematically performed prior to structural and optical characterizations. It was done in two steps: the CTAB was first frozen in the fridge for 1 day (in order to reduce its 9028
dx.doi.org/10.1021/la300210h | Langmuir 2012, 28, 9027−9033
Langmuir
Article
presence of the substrate.9 Though this has been neglected in previous SMS spectra modeling, as in most single particle optical studies, it is an important parameter leading to an anisotropic dielectric environment. Details of the numerical simulations will be given elsewhere, and we will only describe here the model geometries for the present experiment. The bipyramids are usually described as two base-joined pentagonal pyramids with rounded tips.8,16 Their actual morphology frequently deviates from this model shape, exhibiting an irregular 6-fold twinning structure with highly stepped dominant facets as shown by TEM tomography.32 For the sake of simplicity they were modeled as bicones (i.e., truncated bicone capped with hemisphere of radius r, Figure 1), a geometry easier to handle in numerical treatments. The computed extinction cross-sections have been found to be very similar to those of the bipyramid model (i.e., within the precision of the experiments). The single bicones were assumed to lay on one of their sides on the silica substrate (Figure 1).16 Following previous studies,14 the nanorods were modeled as cylinders capped by hemispheres, which corresponds to the shapes measured by TEM (Figure 1). In all simulations in which the substrate was assumed to be infinite, inclusion of its finite thickness only introduced a weak correction to the computed σext amplitude (about 10%) for the relatively small objects investigated here. To take into account the presence of residual water and surfactant molecules on the silica surface, the objects were assumed to be embedded in a nonabsorbing homogeneous environment with refractive index nm filling the half-space above the silica substrate (Figure 1). Small particles (
