Effect of Picosecond Postirradiation on Colloidal Suspensions of

Nov 11, 2010 - Maurizio Muniz-Miranda , Cristina Gellini , and Emilia Giorgetti ... Emilia Giorgetti , Paolo Marsili , Francesco Giammanco , Silvana T...
2 downloads 0 Views 3MB Size
ARTICLE

Effect of Picosecond Postirradiation on Colloidal Suspensions of Differently Capped AuNPs Emilia Giorgetti,*,‡ Francesco Giammanco,§ Paolo Marsili,§ and Anna Giusti§,|| ‡

INSTM and Istituto dei Sistemi Complessi, Consiglio Nazionale delle Ricerche, Via Madonna del Piano 10, 50019 Sesto Fiorentino (Firenze), Italy § Department of Physics “E. Fermi”, University of Pisa, Largo Bruno Pontecorvo 3, Pisa, Italy

bS Supporting Information ABSTRACT: Fifth generation ethylendiamine-core poly(amidoamine) (PAMAM-G5) and poly(ethyleneimine) capped gold nanoparticles were prepared by picosecond laser ablation in water, by using the fundamental wavelength of a picosecond Nd:YAG laser. The role of PAMAM-G5 inner cavities in the processes of photofragmentation of the nanoparticles by postirradiation with the second or third harmonic from the same laser is assessed through comparison of the experimental results obtained with the two stabilizers and the shot-by-shot theoretical modeling of the photobleaching process. We also show how thorough bleaching of the suspensions by postirradiation with 532 nm causes a strong increase of their fluorescence emission. Such fluorescence, which cannot be attributed to new fluorescent species, such as gold quantum dots, is due to a strong enhancement of the intrinsic emission bands of both stabilizers and is related to formation of gold subnanometer fragments and their subsequent interactions with the molecular compounds.

1. INTRODUCTION In the past years, metal nanoparticles have been widely investigated in view of their applications in the fields of photonics, sensing, and nanomedicine.1 Chemical reduction methods permit a high versatility in terms of shape of the nanostructures, allowing preparation of exotic products such as nanocages, nanostars, nanorods, and nanoprisms.2 However, pulsed laser ablation of a metallic target in a liquid environment, although more expensive and less versatile in terms of particle shape, is extremely simple, is fully compatible with different solvents, and, above all, guarantees the highest degree of purity of the final products.3 In particular, gold nanoparticles (AuNPs) produced in the form of colloidal suspensions by pulsed laser ablation are of great interest, especially for the fast developing fields of human biology and medicine, where nanoparticles seem suitable for drug delivery systems and in vivo and in situ diagnostics.4 In this framework, laser-induced fragmentation of preformed capped metal nanoparticles is of particular interest either as a means to control their size and shape or as a method to obtain nanocomposite materials with novel or enhanced spectroscopic properties, originating from the interaction of the nanostructured metal with the capping molecular compound.5 In refs 6 and 7 we reported on an analysis of the fragmentation caused by 532 or 355 nm picosecond pulses in a previously prepared colloidal suspension of gold nanoparticles (AuNPs) capped by fifth generation ethylendiamine-core poly(amidoamine) (PAMAMG5). Such a process is due to the simultaneous onset of different physical mechanisms, i.e., evaporation and electron emission via thermoionic effect or multiphoton absorption.8 Irradiation with r 2010 American Chemical Society

these wavelengths permitted thorough bleaching of the plasmon band with no evidence of gold precipitation. This suggests that, in both cases, the existing nanoparticles are gradually reduced to smaller dimensions and finally broken into subnanometer fragments (AuSNF). Such fragments, consisting of less than ∼37 atoms, are also stabilized by the dendrimer and cannot recombine to generate new nanoparticles. Their dimensions are too small for allowing the onset of the plasmonic absorption. If large enough, they can only contribute to the blue-UV region of the spectrum, through interband transitions of metallic gold. However, when their size becomes comparable with the Fermi wavelength of an electron in gold (i.e., ∼0.7 nm) or lower, they lose their “metallic” nature and exhibit discrete electronic transitions.9 Such clusters, also called quantum dots (QDs), represent a further intermediate state of the matter, between the atomic and the bulk one.10 The products of fragmentation can be either neutral or positively charged. Their formation and subsequent stabilization cause strong changes in the extinction spectra of the suspensions. The fingerprint of the photofragmentation effect in PAMAMG5-stabilized suspensions, along with the bleaching of the plasmon band, is the growth of a new band in the UV, which we attributed, at least to a certain extent, to a ligand to metal charge transfer (LMCT) among gold cations trapped within the Special Issue: Laser Ablation and Nanoparticle Generation in Liquids Received: August 25, 2010 Revised: October 28, 2010 Published: November 11, 2010 5011

dx.doi.org/10.1021/jp108042m | J. Phys. Chem. C 2011, 115, 5011–5020

The Journal of Physical Chemistry C PAMAM-G5 cavities and the PAMAM-G5 molecule itself.6,8 At the same time, the conversion of AuNPs into AuSNFs during 532 and 355 nm postirradiation is demonstrated by an initial percentage growth of the absorption in the blue-violet region of the spectrum at the expense of the plasmon band, followed by thorough bleaching of the absorption above 400 nm for sufficiently long exposures. During this initial phase, Mie fitting of the spectra permits direct evaluation of the conversion process of AuNPs into AuSNFs.11 Such efficient production of AuSNFs and, by further bleaching, of AuQDs has a strong potential interest, being associated with a considerable enhancement of the fluorescence emission of the samples,11 which some authors attributed to the intrinsic fluorescence properties of AuQDs.12,13 For example, PAMAM-capped Au8 is expected to emit around 460 nm.13 In light of the previous results, we investigated the role of PAMAM-G5 and of its inner cavities in the photofragmentation process of AuNPs, by performing a comparison with another stabilizing agent, poly(ethyleneimine) (PEI), which is a cavityfree, linear polymer containing the same functional groups as PAMAM-G5, except amides. In this paper, we describe the preparation of PAMAM-G5 or PEI -capped AuNPs via picosecond laser ablation at 1064 nm (section 3) and their photofragmentation with both 532 and 355 nm pulses, with particular attention to the production of AuSNFs (section 44). Finally, section 5 presents a comparison between the fluorescence properties of suspensions of PAMAM-G5 and PEI -capped AuNPs after thorough bleaching with 532 nm pulses. The key results of the investigation are discussed in section 6.

2. EXPERIMENTAL METHODS We prepared AuNPs by laser ablation of a gold target in a liquid environment. All the suspensions used for photofragmentation and fluorescence tests were prepared with the fundamental frequency (1064 nm) of a mode-locked Nd:YAG laser (EKSPLA PL2143A: rep. rate 10 Hz, pulse width 25 ps). The pulse energy was 15 mJ and the ablation time was varied in order to achieve different initial concentrations of NPs. The infrared wavelength was chosen because it does not cause any damage either on NPs or on stabilizing molecules, as noticed in ref 6. The focusing conditions of the laser beam were maintained constant through all the experiments, and the diameter of the laser spot on the target was fixed at 1.4 mm. The target was placed in a 1 cm  1 cm quartz cuvette and was kept 2 cm in front of the focal plane of the laser beam. In each ablation we used 2 mL of stabilizing solution. We postirradiated the obtained suspensions by using the third harmonic (355 nm, pulse width 15 ps) or the second harmonic (532 nm, pulse width 20 ps) from the same laser. The focusing conditions of the laser beam were maintained constant, and the diameter of the laser spot on the bottom of the 1 cm  1 cm quartz cuvette was 1.4 mm. The volume of the suspension used for postirradiation tests was 1 or 2 mL. The ablation process or postirradiation of the NPs was monitored by measuring in situ the visible spectra with an Ocean Optics fiber spectrophotometer and a deuterium-tungsten lamp. The sampling beam was perpendicular to the picosecond laser beam and crossed the quartz cuvette 0.5 cm above the bottom of the cell. We prepared AuNPs suspensions in solutions of PAMAM-G5 or PEI in water. In the case of PAMAM-G5, the solutions were

ARTICLE

3.8 mM referred to superficial amino groups and were obtained by dilution of 6.67% aqueous solution of amino-terminated PAMAM-G5 from Dendritech with ultrapure water (18.2 MΩ 3 cm at 25 C). According to tabulated values provided by the producer, PAMAM-G5 is a monodisperse compound, having a molecular diameter of 5.4 nm. In the case of PEI, we diluted 1:100 in ultrapure water the Fluka 50% water solution purchased from Sigma-Aldrich. The gold target was purchased from Goodfellow. We recorded UV-vis spectra some days after the preparation of the suspensions with a double beam spectrophotometer (Perkin-Elmer model Lambda19) and fluorescence spectra with a Jasco FP-750 spectrofluorimeter. Transmission electron micorscopy (TEM) samples were obtained by dipping carbon-coated copper grids in the suspensions, and the images were recorded with a JEOL2010, 200 kV high-resolution TEM (HRTEM). Particle mean diameter and dispersivity were determined by fitting the measured statistical distributions by an asymmetric Gaussian, i.e., a Gaussian with different standard deviations below (σ-) and above (σþ) the average particle diameter 2Ro. Although a log-normal function is typically adopted to describe the statistical distributions of AuNPs obtained by the diffusion growth dynamic,10 nevertheless, in our case, such modeling would enhance the contribution of particles of size much larger than the average radius, which seems in contrast with our experimental data. Even if both distributions give quite similar results for the purposes of this paper, e.g., Mie fitting of the absorbance and modeling of the fragmentation, we adopted the asymmetric Gaussian, since its degree of confidence is larger than that of the log-normal. This finding, which might indicate a relatively reduced influence of diffusion growth in our experimental conditions compared with other schemes of NP production, would deserve a dedicated investigation.

3. PARTICLE PREPARATION We compared the behavior of PAMAM-G5 with that of PEI. PEI, at least to our knowledge, had never been used before for production of AuNPs by laser ablation methods. Although with a very similar chemical composition, the two stabilizing agents exhibit some differences, as shown by Figure 1, which reports their chemical structure. The first difference between PAMAMG5 and PEI is in the shape. PAMAM-G5 is a globular dendritic molecule containing inner cavities, while PEI is a linear polymer. Moreover, while both molecules contain primary and tertiary amines, PAMAM-G5 contains amides and PEI contains secondary amines. Both compounds act as efficient stabilizing agents for AuNPs prepared by laser ablation. Figure 2 shows absorption spectra and TEM images of typical samples of PAMAM-G5 or PEI-capped AuNPs obtained by 1064 nm ablation with 15 mJ energy per pulse. In both cases, the plasmon peak is well-resolved and centered at 525 nm (Figure 2a). The TEM pictures show spherical, disaggregated NPs. In the case of PAMAM-G5 (Figure 2b), the average diameter 2R0 is 3.4 nm, with asymmetric statistical distribution characterized by σþ = 2.5 nm and σ- = 1.7 nm. In the case of PEI (Figure 2c), we obtained 2R0 = 3.7 nm, σþ = 2.5 nm and σ- = 1.3 nm. In addition to laser ablation procedures, either PAMAM-G5 or PEI can be used for the preparation of AuNPs by chemical reduction. Moreover, both molecules can cause spontaneous 5012

dx.doi.org/10.1021/jp108042m |J. Phys. Chem. C 2011, 115, 5011–5020

The Journal of Physical Chemistry C

Figure 1. Chemical structure of PAMAM-G3 (a) and of a possible unit of PEI polymer (b). Here, we considered PAMAM-G3, instead of G5 for reasons of graphical clarity.

reduction of HAuCl4 in water, in the absence of any specific reductant. Indeed, we observed that 1 mM solutions of HAuCl4 in water containing the same concentration of PAMAM-G5 or PEI used in the ablation experiments turn spontaneously wine red in several days, at ambient temperature and in the dark, as reported by the spectra of Figure 3. Although the plasmon band forms in both cases, indicating that either the dendrimer and the polymer act as reductants themselves, nevertheless such a reduction process is slower with PEI. An important difference between the two spectra is observed in the UV region. The spectrum of PAMAM-G5-capped AuNPs (red curve in Figure 3) is characterized by a band at 287 nm. The formation of this band, which is peculiar for the dendrimer and disappears upon addition of proper reductant, such as NaBH4,6 is related to interaction with charged gold species. As already reported in refs 6 and 7 an analogous UV band also appears in PAMAM-G5 solutions under picosecond irradiation with 532 and 355 nm pulses and grows more intense when the same irradiation dose is combined with ablation of a gold target or takes place in AuNPs-containing suspensions. If we consider the electronic spectra of PAMAM-G5 and PEI in water recorded prior to irradiation (reported in Figure 4a with continuous red and blue lines, respectively), we observe that the spectrum of PAMAM-G5 exhibits a weak broad absorption around 280 nm, extending up to 400 nm with a tail, which can be reasonably assigned to n f π* transitions of the carbonyl group.14 Highenergy green light irradiation of PAMAM-G5 causes the growth of a band centered around 282 nm (dashed red line in Figure 4a).

ARTICLE

Figure 2. Electronic spectra (a), statistical distributions and typical TEM images (b, c) of PAMAM-G5 (red line and hystogram) and PEI (blue line and hystogram) capped suspensions of AuNPs used throughout this paper. Optical path length (OPL) = 1 cm.

Figure 3. Electronic spectra of suspensions of PAMAM-G5 (red line) and PEI (blue line) capped AuNPs obtained by spontaneous chemical reduction of HAuCl4 in water. OPL = 1 mm.

In contrast, it has almost no effect on PEI (dashed blue line in Figure 4a). When the same irradiation dose is delivered during 532 nm ablation of a gold target, the differences between the two capping agents are further enhanced, as reported in the inset of Figure 4a. In particular, in PEI solutions (blue line in the inset), AuNP production with 15 mJ pulses at 532 nm does not generate any appreciable features in the UV spectral region. In contrast, as already reported in refs 6-8 the same ablation conditions in the presence of PAMAM-G5 lead to a dramatic increase of the UV absorption, represented by an intense band centered around 290 nm (red line in the inset). Also note the different scales between Figure 4a and its inset. 5013

dx.doi.org/10.1021/jp108042m |J. Phys. Chem. C 2011, 115, 5011–5020

The Journal of Physical Chemistry C

ARTICLE

Figure 5. Statistical distribution and typical TEM image of a PAMAMG5-stabilized suspension after postirradiation with 355 nm, 8 mJ, and 2500 shots.

Figure 4. (a) Electronic spectra of PAMAM-G5 (red lines) and PEI (blue lines) solutions before (continuous lines) and after (dashed lines) irradiation with 532 nm, 15 mJ, and 10000 shots. The inset shows the result of the same irradiation dose, when delivered during ablation of a gold target for PAMAM-G5 (red line) and PEI (blue line), respectively. (b) Electronic spectra of PEI solutions after irradiation with 355 nm, 15 mJ, and 5000 shots without (solid blue line) and with (dashed blue line) a gold target. OPL = 1 cm.

The 355 nm irradiation of PAMAM-G5 was already reported in refs 7 and 15. In particular, we showed that, differently from the 532 nm case, exposure to 355 nm light has a strong effect on the UV absorption of PAMAM-G5, even in the absence of the gold target and even in the continuous wave regime.15 In the case of PEI, the effect of high-energy 355 nm irradiation is shown in Figure 4b. In the absence of the gold target, it causes the growth of a new broad weak band around 350 nm (solid blue line), which might indicate some damage of the molecule. However, the changes with respect to the initial spectrum shown in Figure 4a are quite small, particularly if compared to the case of PAMAMG5. In the presence of a gold target (dashed blue line), 355 nm irradiation leads to formation of the plasmon band, with no significant additional features in the UV region of the spectrum.

4. PARTICLE PHOTOFRAGMENTATION WITH 532 AND 355 NM In ref 8 we extensively analyzed experimentally and theoretically the fragmentation of PAMAM-G5-capped AuNPs, in order to disentangle the main phenomena responsible for the bleaching of the colloids. We found that, in all experiments, the presence of a stabilizing agent played a key role in preventing fragmented AuNPs from reaggregation, which would have made the interpretation of results difficult. Therefore, since PAMAM-G5 and PEI are efficient stabilizers for AuNPs and exhibit a similar composition, we performed new experiments aimed at elucidating the differences or similarities between the two molecules. To better follow the photofragmentation and to highlight the differences between the two stabilizing compounds, we irradiated

Figure 6. Statistical distribution and typical TEM image of a PEI-stabilized suspension after postirradiation with 355 nm, 8 mJ, and 2500 shots.

until thorough bleaching the 2-fold diluted PAMAM-G5 and PEI-capped AuNPs suspensions of Figure 2 with both 532 nm and 355 pulses. We recorded shot by shot either the in situ UV-vis full spectra and the absorption in the plasmon maximum and evaluated the decay rate of postirradiated colloids versus the laser energy. We recall that, in ref 8 we defined the decay rate as the inverse of the number of shots required to reduce the plasmon peak absorption by 1/e. In this section, we first analyze the morphological evolution of AuNPs during the bleaching with both stabilizers. Then we compare the experimental decay rates with the theoretical curves calculated by the model of ref 8, which was improved in order to include the time evolution of the statistical distribution of the AuNPs. Finally, by using Mie fitting of the absorption spectra recorded in situ after a fixed number of shots, we evaluate the density of AuNPs and AuSNFs contained in the colloids and their temporal evolution during the fragmentation process. Examples of the effect of 355 nm postirradiation on the statistical distribution of the AuNPs are reported in Figures 5 and 6. Figure 5 shows a typical TEM image and corresponding statistical distribution for the PAMAM-G5 case, after irradiation of 1 cm3 of suspension with 355 nm, 8 mJ per pulse, and 2500 shots. The particle average diameter reduces by ≈10%, and the statistical distribution becomes narrower than that in Figure 2b (σþ = 2.2 nm and σ- = 0.8 nm), showing a reduced contribution of large particles. Figure 6 reports the results regarding PEIcapped AuNPs, which were postirradiated in the same conditions as Figure 5. Here, the average particle diameter decreased by about 25% with respect to Figure 2c. The green curves in the statistical plots of Figures 5 and 6 represent the expected particle distribution, as calculated by an upgraded version of the model described in ref 8. The new model includes the effective shot-by-shot evolution of the distribution function which, in ref 8, had been kept constant. Basically, the 5014

dx.doi.org/10.1021/jp108042m |J. Phys. Chem. C 2011, 115, 5011–5020

The Journal of Physical Chemistry C

Figure 7. Theoretical (solid lines) and experimental (squares) decay rate versus laser pulse energy for photofragmentation of PAMAM-G5 (red) and PEI (blue) suspensions of AuNPs with 355 nm pulses.

upgrade takes into account that, during the interaction with laser, larger radius AuNPs can (i) fragment into AuSNFs and smaller radius AuNPs when evaporation prevails or (ii) be directly converted into AuSNFs, due to Coulomb explosion caused by ionization. As a consequence, the initial AuNP distribution is expected to evolve to a bimodal distribution at some stage of the bleaching, as the statistical plots of Figures 5 and 6 confirm. Our initial average radius being relatively small as compared with, for instance, that of ref 16, the differences between a bimodal and a rescaled asymmetric Gaussian distribution are barely distinguishable. However, it appears that the green curves of Figures 5 and 6 fit the histograms better than an asymmetric Gaussian one. Figures 7 and 8 show the decay rate versus the laser pulse energy for Au colloids in PAMAM-G5 and PEI for bleaching with 355 and 532 nm. The continuous lines were calculated ab initio by using the upgraded version of the model of ref 8. Such a procedure consists, first, in calculating the fragmentation of a single AuNP, which can take place either for evaporation or ionization, including thermoionic effect and photon assisted transitions of different orders. Then, the laser focusing cone is divided into elementary volumes, where the laser intensity can be assumed constant, and the conditions for thorough bleaching of the particles contained within the cone are evaluated. Such conditions are verified when all AuNPs of initial radius Ro are fragmented to a radius e0.5 nm; that is, they no longer contribute to the plasmon resonance. In this condition, we restart simulation with a new particle density reduced by an amount given by the ratio between the elementary volume and the total one. In ref 8, we assumed self-similar evolution of the initial AuNP distribution. Indeed, we verified that this assumption holds true until the plasmon peak reduces to 1/e. Nevertheless, in principle, large radius AuNPs can gradually loose fragments and transform into smaller AuNPs, which can still contribute to plasmon resonances. Therefore, we implemented the model to account for this effect and hence to calculate the effective time evolution of the AuNP statistical distribution in the laser cone. Indeed, as stressed in ref 8, the laser-induced fragmentation process, as a whole, includes many quantities, which are radius sensitive. For example, the first step of the model deals with laser heating of a AuNP, which then dissipates by transferring heat to electrons and hence to lattice through electron-phonon collisions. The heat diffusion coefficient depends quadratically on the radius of the NP.17 Besides, also the coupling term between electrons and laser pulse contains Fermi corrections to the bulk values of the dielectric constant, which are more significant in the lower radius

ARTICLE

Figure 8. Theoretical (solid lines) and experimental (squares) decay rate versus laser pulse energy for photofragmentation of PAMAM-G5 (red) and PEI (blue) suspensions of AuNPs with 532 nm pulses.

portion of the distribution. Since we recognized that the fragmentation, in our range of fluence, can occur by either evaporation or ionization, the time evolution of lattice temperature during a single laser pulse is crucial to evaluate the predominance of one effect on another. In fact, according to ref 8, ionization prevails if the fissility condition18 occurs before the lattice temperature reaches the evaporation limit. This value, which is assumed ≈0.3 eV, is also radius sensitive although, in our conditions, variations of some tenths of percent do not affect the overall results. Finally, the work function of a AuNP undergoes an increase as the ionization goes on, due to the positive charge left on the surface. This effect turns out to be proportional to the inverse of the radius.8 The theoretical simulations of Figures 7 and 8 confirm that, in our conditions, with both PAMAM and PEI, the fragmentation occurs mainly through ionization. The main source of ionization is due to photon-assisted transitions, which start to be effective when electron heating populates the electronic states which are below the ionization threshold, corresponding to the energy of the laser photon. As a comment to Figures 7 and 8, we note that the curves in PAMAM-G5 and PEI are very similar to each other, particularly for the UV case. However, at 532 nm, the bleaching in PEI proceeds at a slower rate than in PAMAM-G5. In the theoretical model, the role played by the stabilizer is simply that of preventing fragmented AuNPs from reaggregation, regardless of its composition and/or shape. Hence, the differences in the theoretical results can only be attributed to a slightly different initial distribution of AuNP dimensions, which is confirmed by Figure 2. In this sense, the fraction of larger size particles, which undergo a multiple steps fragmentation and which are more abundant in the PEI case, seems to play a key role. In fact, when evaporation takes place, larger size AuNPs can split into AuSNFs and smaller size AuNPs, leading to a bimodal distribution. Owing to the quadratic dependence on radius of the dissipation factor,8,17 the temperature of larger size AuNPs can grow above the evaporation threshold before the onset of a Coulomb explosion. Therefore, this part of the distribution contributes to the onset of a bimodal distribution at some stage of bleaching. Smaller size AuNPs are then fragmented by ionization. It explains why, in our conditions, ionization is the main cause of fragmentation. This effect leads, in general, to an increase of the number of shots required to thoroughly bleach the colloid. Actually, if compared with those of ref 8, where only the PAMAM-G5 case was considered, the theoretical curves of Figures 7 and 8 increase 5015

dx.doi.org/10.1021/jp108042m |J. Phys. Chem. C 2011, 115, 5011–5020

The Journal of Physical Chemistry C

Figure 9. (a) Absorbance in the plasmon peak and (b) UV-vis spectrum vs laser shots for a PAMAM-G5 (red lines) and a PEIstabilized (blue lines) suspension of AuNPs bleached with 10 mJ pulses at 532 nm. OPL = 1 cm. The black circle marks the presence of an isosbestic point. (c) Calculated density of atoms belonging to AuNPs (circles) and AuSNFs (triangles) vs laser shots, during the bleaching process. Solid lines represent the total equivalent atoms in both species.

at a lower rate, of the order of 20-30%, for energies above 10 and 20 mJ for 532 and 355 nm, respectively. However, owing to its larger bleaching efficiency, this effect is barely detectable when using 355 nm, as shown in Figure 7. Figures 9 and 10 report the detailed bleaching behavior observed with PAMAM-G5 (red lines) or PEI (blue lines)capped AuNPs, when postirradiated with both 532 nm (Figure 9) and 355 nm (Figure 10) pulses and with a value of the energy of the single pulse, which is intermediate with respect to the available energy range of our laser source, i.e., 10 mJ. Note that the shape and intensity of the in situ spectra below 300 nm is to be considered just an indication, due to both saturation of the signal and low response of the spectrometer in this wavelength range. In the case of 532 nm irradiation (Figure 9a), the decay of the absorption in the plasmon maximum is quite similar with both stabilizing agents up to around 10000 shots. Beyond this value, while in the case of PEI-stabilized AuNPs the bleaching of the plasmon proceeds smoothly and regularly toward 0, in the case of PAMAM-G5-stabilized particles it nearly stops up to 45000 shots, with some oscillations and, finally, restarts. Figure 9b shows the full UV-vis spectra taken in situ at fixed numbers of shots, i.e., 2400, 7200, 16000, and 28000. As expected on the basis of previous considerations, the main differences are in the

ARTICLE

Figure 10. (a) Absorbance in the plasmon peak and (b) UV-vis spectrum vs laser shots for a PAMAM-G5 (red lines) and a PEIstabilized (blue lines) suspension of AuNPs bleached with 10 mJ pulses at 355 nm. OPL = 1 cm. The black circle marks the presence of an isosbestic point. (c) Calculated density of atoms belonging to AuNPs (circles) and AuSNFs (triangles) vs laser shots, during the bleaching process. Solid lines represent the total equivalent atoms in both species.

blue-UV region. A first important difference between the two samples is the growth of the UV band around 290 nm in the case of PAMAM-G5-capped AuNPs, with an isosbestic point around 330 nm (black circle in Figure 9b). Moreover, in the case of PEIstabilized NPs, Figure 9b shows the progressive and parallel reduction of both plasmon and interband absorption of gold, which are thoroughly canceled after 28000 shots. In contrast, in the case of PAMAM-G5-stabilized NPs, even after thorough bleaching of the plasmon band, a considerable pedestal due to interband absorption remains after 28000 shots, which is not completely canceled even after further irradiation up to 69800 shots (black line). This is an indication that metallic gold is still present, but in the form of AuSNFs. Mie fitting of the spectra of Figure 9b permits shot by shot calculation of the density of atoms belonging to “metallic” gold, that is to particles with size larger than ∼0.7 nm, which can be present in the sample in the form of either AuNPs or AuSNFs. We performed the fitting by following the method illustrated in ref 11. As observed in ref 11, the absorption spectrum of our suspensions is the sum of two contributions: the one originating from sufficiently large NPs, whose conduction electrons undergo plasmonic oscillations, and the one originating from subnanometer gold clusters, whose size is small enough that plasmonic 5016

dx.doi.org/10.1021/jp108042m |J. Phys. Chem. C 2011, 115, 5011–5020

The Journal of Physical Chemistry C oscillations are strongly damped by quantum effects but large enough that they contribute to interband transitions. The latter contribution mainly affects the blue-violet region of the spectrum. By using the initial NP distributions obtained from TEM analysis as a reference, we first fitted the long-wavelength side of the spectrum (from plasmon resonance up to the near-infrared). Then we adjusted the AuSNF contribution to fit the blue side. To compare on a same scale both SNFs and NPs contributions, we converted the NPs density resulting from the fit in an equivalent number of atoms by using the formula Z NNP ¥ 3 R f ðRÞ dR ð1Þ Nat ¼ 3 Rat 0 where f(R) is the distribution function obtained from TEM analysis. As the fragmentation proceeds, the initial distribution maintains almost the same shape until the absorbance is reduced by 1/e and then begins to change (see above). We adjusted the AuNP distribution when the fit departs in some portions of the spectrum by more than 10%, by using TEM analysis as a reference. Nevertheless, we observed that the equivalent number of atoms given by eq 1 is quite insensitive even to significant changes in the distribution function. The results are reported in Figure 9c for the first 10000 shots. Beyond this value, the plasmon peak fades out and hence it becomes very difficult to perform a reliable fit. However, the trend is clear enough and also the differences between PAMAM and PEI. It appears that, while the density of atoms contained inside the NPs (blue and red circles) follows the same monotonic behavior with both capping compounds, the density of Au atoms belonging to SNFs (blue and red triangles) behaves very differently. In the case of PEI, after an increase during the first period of exposure, it starts decaying. At the same time, the total number of atoms belonging to “metallic” structures also decreases (blue solid line). In contrast, in the case of PAMAM-G5, the total number of gold atoms (red solid line) does not exhibit any substantial change, at least within a 10% fitting error, and the number of atoms pertaining to AuSNFs grows up to a stable value. This suggests that fragmentation converts NPs into SNFs, which do not undergo any further fragmentation into atomic gold. In contrast, they are trapped and stabilized by PAMAM-G5 molecules and poorly affected by subsequent photoirradiation. It also explains why, in the PAMAM-G5 case, the absorption in the plasmon peak reported in Figure 9a keeps roughly constant between 10000 and 50000 shots, being almost completely assigned to the interband background of AuSNFs and not to plasma oscillations of the conduction electrons of NPs. In the case of 355 nm bleaching, the decay of the absorption in the plasmon maximum for PAMAM-G5-stabilized NPs is more regular than that obtained with 532 nm (Figure 10a). Nevertheless, also now, after an initial phase of about 1500-2000 laser shots during which the two curves proceed at the same rate, the bleaching of the plasmon band of PAMAM-G5-stabilized AuNPs decreases its rate. This results in the need for a longer time to achieve thorough bleaching of the suspension. The improved regularity of the bleaching process for PAMAM-G5 suspensions of AuNPs is confirmed by the full spectra (Figure 10b), where a considerable, although again not complete, bleaching of the interband absorption is observed. Note that no difference is found between the spectra of PAMAM-G5-stabilized NPs taken after 19400 or 29000 shots (black line). Also in this case, it is useful to check the decay of the concentration of Au atoms belonging to AuNPs or AuSNFs in

ARTICLE

the two samples. The results of Mie fitting are reported in Figure 10c. The decay of the concentration of atoms belonging to AuNPs is again monotonic and similar with both stabilizers (blue and red circles). Concerning AuSNFs (blue and red triangles), we observe that, with both capping agents, their concentration increases up to a maximum (around 3000 laser shots for PEI and 6000 for PAMAM G5) and afterward it starts decaying. As a whole, AuSNFs production and persistence are favored with PAMAM-G5, whose total number of “metallic” atoms (solid red line), although exhibiting a more pronounced decay behavior with respect to 532 nm postirradiation, always keeps larger than that pertaining to PEI suspensions. From these data, we can infer that, although 355 nm light is more efficient than 532 nm light in AuSNFs fragmentation, due to the strong absorption of AuSNFs in the blue-UV region of the spectrum, nevertheless, it appears that PAMAM-G5-stabilized samples exhibit a strong resistance to AuSNFs pulverization.

5. FLUORESCENCE SPECTROSCOPY OF BLEACHED SUSPENSIONS OF PAMAM-G5 AND PEI-CAPPED AUNPS Photofragmentation of PAMAM-G5 or PEI-capped AuNPs produces AuSNFs and AuQDs until complete dissolution of gold. Among the properties of AuQDs, fluorescence seems a particularly interesting one. Indeed, there are reports in the literature describing the fluorescence properties of AuQDs and claiming that these systems exhibit a high quantum yield and good resistance to bleaching.12,13 In particular, PAMAM-G5capped AuQDs obtained by chemical reduction and formed by 5-13 atoms would be strongly fluorescent in the blue-green region of the spectrum.19 Such observations have been questioned by other authors, who claim that the observed fluorescence comes from the dendrimer and not from the metal.20,21 Our photobleaching procedure seems very efficient in dissolving AuNPs down to structures which no longer exhibit any metallic behavior. Therefore, we decided to measure the fluorescence of our bleached suspensions, in order to check its presence, if any, and the emission properties of AuQDs contained therein. For this purpose, we thoroughly bleached PAMAM-G5 and PEIcapped AuNPs suspensions with 532 nm pulses and, after subtraction of the contribution of pure water, we compared their fluorescence spectra with those of PAMAM-G5 or PEI aqueous solutions before and after irradiation in identical conditions. In both cases, the AuNPs had been previously obtained by 1064 nm ablation with 15 mJ and the initial absorbance in the plasmon peak was 1.1. Figure 11a shows the excitation spectra for emission at 420 nm and the fluorescence spectra under excitation at 350 nm of a PAMAM-G5 solution (blue lines), of 1 cm3 of the same solution irradiated with 532 nm pulses, 8 mJ, 40000 shots without (green lines) and with (red lines) thoroughly fragmented AuNPs. Note that the blue and green lines in Figure 11a are multiplied by a factor of 5. In all cases, the excitation spectrum is characterized by a band around 350 nm, while the emission exhibits two bands centered at 417 and 465 nm. In Figure 11b we compared the fluorescence properties of two samples obtained from an identical initial suspension, but bleached in different conditions, namely, 8 mJ and 40000 shots (thin lines) and 16 mJ and 10000 shots (thick lines). The emission spectra were recorded for excitation with 300 nm (red lines), 350 nm (green lines), and 400 nm (blue lines). 5017

dx.doi.org/10.1021/jp108042m |J. Phys. Chem. C 2011, 115, 5011–5020

The Journal of Physical Chemistry C

ARTICLE

Figure 12. Excitation spectra for emission at 454 nm (black lines) and emission spectra for excitation at 300 nm (red lines), 350 nm (green lines), and 400 nm (blue lines) of PEI-stabilized suspensions of AuNPs after thorough bleaching with 532 nm, 8 mJ, 40000 shots (thin lines) compared with those of a concentrated PEI solution in water (thick lines).

Figure 11. (a) Excitation spectra for emission at 420 nm an fluorescence spectra for excitation at 350 nm of a PAMAM-G5 solutions before (blue lines) and after irradiation with 532 nm, 8 mJ, and 40000 shots without (green lines) and with (red lines) AuNPs. Blue and green spectra are multiplied by a factor of 5. (b) Excitation spectra for emission at 454 nm (black lines) and emission spectra for excitation at 300 nm (red lines), 350 nm (green lines), and 400 nm (blue lines) of PAMAM-G5stabilized suspensions of AuNPs after thorough bleaching with 532 nm, 8 mJ, 40000 shots (thin lines) or 532 nm, 16 mJ, 10000 shots (thick lines).

The excitation spectra recorded for emission at 454 nm are represented by black lines. The spectra of Figure 11 were deconvoluted. In all cases, deconvolution did not show any evidence of new bands, which could be attributed to AuQDs, but only a redistribution of the relative weight of bands already present in Au-free PAMAM-G5 solutions. As far as spectral features and intensity are concerned, the behavior of PAMAM-G5 solutions is not significantly modified by 532 nm irradiation. In contrast, the presence of Au causes a strong enhancement of the overall fluorescence emission for excitation with 350 and 400 nm and no significant changes for excitation with 300 nm. Moreover, according to Figure 11b, the different irradiation regime does not modify the spectral features, the only difference being the intensity of the emission, which is roughly doubled in the sample obtained with 8 mJ pulses. This suggests that, being the total initial amount of gold identical and the photobleaching complete, an important parameter ruling the fluorescence enhancement is the total energy released into the sample during the photobleaching process, i.e., 320 or 160 J for the 8 and 16 mJ cases, respectively. Indeed, analogous experiments were performed with an initial suspension of PAMAM-G5-stabilized AuNPs obtained by 1064 nm ablation and exhibiting an initial absorbance in the plasmon peak of 0.8. It was bleached with 532 nm pulses and the same amount of total energy (48 J), but in different conditions (5 m J and 9600 shots, or 10 m J and 4800 shots). The two samples gave identical fluorescence emission, both as far as spectral features and intensity are concerned. The data are reported as Supporting Information in Figure S1. Figure 12 shows the case of PEI. Thin lines describe the fluorescence behavior of 1 cm3 of the suspension of PEI-capped

AuNPs after bleaching with 532 nm, 8 mJ, and 40000 shots. Differently from PAMAM-G5, PEI solutions used for ablation tests do not exhibit any significant fluorescence, neither before nor after 532 nm irradiation. Therefore, in order to put in evidence of the contribution of Au in the spectra of Figure 12, we performed fluorescence spectroscopy of PEI by using more concentrated solutions; i.e., we increased the concentration by a factor of 5. In this case, we found the spectra reported in Figure 12 as thick lines. The strong scattering component of this concentrated solution masks the fluorescence in the low-wavelength region, so that some of the bands, which are clearly resolved in the Au-containing sample (i.e., at 506 nm for excitation at 400 nm; at 418 nm for excitation at 350 nm and at 403 nm for excitation at 300 nm), appear as broad shoulders in the Au-free one. However, the spectral position of the shoulders exhibits a good correspondence with that of the emission bands observed in the Au-containing sample. As a consequence, also in this case, it seems more reasonable to assign the emission to an enhancement of the intrinsic fluorescence of the capping agent, rather than to any new fluorescent species.

6. DISCUSSION The growth of the UV absorption under 355 or 532 nm photoirradiation of PAMAM-G5 water solutions can be explained in terms of formation of carbonyls, due to photodegradation of the dendrimer,22 as the result of the oxidative photochemical reaction, which may lead from CH2 to CO or COOH.23 As expected, UV light is more efficient than visible light, while 1064 nm light has no effect at all.6,15 The enhancement in the presence of photofragmentation processes of the AuNPs can be related to the production of Au(III) or, more generally, of positively charged Au clusters, as fragmentation byproduct, which is very efficient with 532 and 355 nm picosecond pulses, while it is completely absent with 1064 nm pulses.6 It was reported that Au(III) acts as a catalyst for many photochemical reactions, for example, for that leading from -CH2OH to -COOH.24 Moreover, on the basis of a FT-IR analysis, it was suggested that Au(III) is likely to complex within the cavities of PAMAM-G5, and to connect amides and tertiary ammines located on different sides of the same cavity, the formation of such complex requiring at least two different functional groups.25 The role of Au(III) in our samples is demonstrated by the selfreduction tests shown in Figure 3, where a new UV band forms also in the dark. Au(III) can complex with PAMAM-G5, and even in the absence of light, a redox process can take place whose 5018

dx.doi.org/10.1021/jp108042m |J. Phys. Chem. C 2011, 115, 5011–5020

The Journal of Physical Chemistry C fingerprint is the formation of a band around 290 nm. In general, once trapped within a dendrimer cavity, Au(III) can act both as the catalyst of photodegradation reactions of the dendrimer and as the origin of a LMCT with PAMAM-G5. In ref 25, Au(III) complexation is proved by the presence of an isosbestic point in the UV-vis spectra, which we also observe, as pointed out by the black circle in Figures 9b and 10b. In our case, the isosbestic point stems from the crossing of the absorption band at 290 nm and the interband transitions curve, indicating an equilibrium condition between metallic particles and non-metallic charged particles interacting through LMCT with the dendrimer. Due to the lack of proper complexation sites, the behavior of PEI results is completely different. Indeed, in this case, a growth of the UV absorption is never observed, confirming the role of PAMAM-G5 cavities and amides therein in the trapping of Au-charged clusters. Beyond Au cations, photofragmentation of AuNPs also produces AuSNFs. The observed stability of photofragmentationgenerated AuSNFs is different with the two capping agents. In the case of PAMAM-G5, they exhibit a considerable resistance to further 532 nm fragmentation (Figure 9c), while they are less stable when irradiated with 355 nm pulses (Figure 10c). This fact is related to a more efficient absorption of UV wavelengths by interband transitions. AuSNFs generation is observed with PEI, as well, but with a lower efficiency and persistence. In this case, their contribution rapidly disappears from the spectra of PEI-stabilized suspensions of AuNPs exposed to photobleaching (Figures 9c and 10c). This again suggests a major role of dendrimer cavities in the process. Fluorescence spectroscopy performed on thoroughly bleached PAMAM-G5 or PEI-stabilized suspensions of AuNPs showed that the presence of a photofragmentation byproduct causes a strong increase of the fluorescence of both stabilizing compounds, particularly for excitation at 350 nm. In the case of PAMAM-G5, such an increase, which can be as high as 2 orders of magnitude for some emission lines,11 is roughly proportional to the overall energy released into the sample during the photofragmentation. Moreover, the comparison among the spectra of PAMAM-G5 before and after irradiation and of PAMAM-G5-stabilized suspensions of AuNPs after thorough bleaching did not show any evidence of new bands which could be assigned to new fluorescent species, such as AuQDs, at least in the spectral range, which we could span with our equipment. A similar behavior was observed also with PEI, although in this case the poor intrinsic fluorescence of the organic compound and the high scattering level of its solutions impaired a direct comparison between Au-containing and Au-free samples. Reference 26 describes the fluorescence properties of different types of PAMAM dendrimers and of poly(propylene imine) (PPI), a compound which is very similar to PEI. The authors conclude that, although the nature of the color center is unclear, the fluorescence of PAMAM comes from its cavities, that is, from amides or tertiary amines, and is strongly enhanced by different effects, such as pH, aging, or oxidation. Comparison with PPI suggests that fluorescence is likely to originate from tertiary amines in both compounds. In our case, photochemical processes, such as those related to the formation of the UV band in PAMAM-G5, as assumed in ref 11 or others could be at the origin of the observed Au-induced fluorescence enhancements. Indeed, either charged or neutral Au fragments would behave as efficient catalyzers, thus improving the overall emission. Although this point deserves a deeper investigation, nevertheless, the strong enhancement of the fluorescence observed with both PAMAM-

ARTICLE

G5 and PEI in the presence of Au subnanometer particulate and the reducing ability of both capping agents demonstrated by Figure 3 seem to exclude a major role for amides and suggest that the fluorescence emission is related to Au-catalyzed photooxidation of tertiary amines.

7. CONCLUSIONS This paper reports on a comparison between the properties of a dendritic molecule, PAMAM-G5, and a linear polymer, PEI, as stabilizers for AuNPs obtained by picosecond laser ablation in water. While PAMAM-G5 had already been extensively used as a particle stabilizer in such experiments, to our knowledge, this is the first time that PEI is used in laser ablation experiments. The strong similarity in chemical composition between the two capping agents, associated with a completely different morphology, was exploited to clarify the role of PAMAM-G5 inner cavities and its interactions with AuNPs, SNFs, or QDs. In particular, since both compounds are efficient stabilizers either for AuNPs and for photofragmentation byproduct, in both cases, we could perform an accurate shot-by-shot monitoring of the photofragmentation processes of AuNPs, which occurs during postirradiation with 532 or 355 nm picosecond pulses. In the PAMAM-G5 case, the interaction of the molecule with charged Au fragments originates an UV band in the electronic spectrum, which represents the fingerprint of the fragmentation process. It never appears with PEI and is reasonably associated with the presence of amides in the dendrimer inner cavities. Moreover, we observed that, although the process of progressive decay of AuNPs concentration does not exhibit any substantial differences with the two stabilizers, nevertheless PAMAM-G5 is characterized by a strong affinity with AuSNFs, which causes resistance to their further pulverization into AuQDs and subsequent thorough dissolution. This effect is particularly pronounced with 532 nm postirradiation, while it is weaker with 355 nm postirradiation, due to the strong interband absorption of AuSNFs at this wavelength. PAMAM-G5 and PEI are weak fluorofores. However, their complexation with fragmentation byproduct of AuNPs permits a strong enhancement of their initial fluorescence. In the case of thoroughly bleached suspensions of PAMAM-G5-capped AuNPs, a more than 10-fold enhancement of the overall fluorescence can be observed for excitation at 350 or 400 nm, and for excitation at 400 mn, the emission concentrates into the band at 470 nm, which can be intensified up to 2 orders of magnitude.11 Although we observed a redistribution of the relative weight of the emission bands, depending on the excitation wavelength, we could not find any evidence of new bands, which could be attributed to AuQDs. In contrast, we believe that, with our samples and our fabrication procedure, we observe a metalenhanced fluorescence (MEF) effect, which is related to the onset of Au-catalyzed photochemical reactions. The reducing nature of our compounds is consistent with oxidative processes, which can be strongly favored during high-intensity UV-visible light irradiation. Throughout the paper, the experimental data were simulated with an upgraded version of the theoretical model already proposed in ref 8, which not only describes the multiple physical process, which are active during AuNPs fragmentation, but also permits following the evolution of the morphology of AuNPs contained in the suspension under fragmentation and to calculate, shot-by-shot, the percentage of atoms contained either in 5019

dx.doi.org/10.1021/jp108042m |J. Phys. Chem. C 2011, 115, 5011–5020

The Journal of Physical Chemistry C AuNPs or in AuSNFs. Further experiments are currently in progress to check the validity of the model with nanoparticles of different metals, dimensions, and shapes, such as branched gold nanostructures.

’ ASSOCIATED CONTENT

bS

Supporting Information. Figure showing excitation and fluorescence spectra of two PAMAM-G5 suspensions of AuNPs after 532 nm bleaching with the same total amount of energy (48 J) but different energies of the single picosecond pulses. This information is available free of charge via the Internet at http:// pubs.acs.org.

’ AUTHOR INFORMATION Corresponding Author

* E-mail: emilia.giorgetti@fi.isc.cnr.it. )

Present Addresses

Universite’ Paris-Sud 11, Institut de Chimie Moleculaire et de Materiaux d’Orsay, UMR CNRS 8613, F-91405, Orsay, France.

ARTICLE

(13) Tran, M. L.; Zvyagin, A. V.; Plakhotnik, T. Chem.Commun. 2006, 2400–2401. (14) Gu, T.; Ye, T.; Simon, J. D.; Fox, M. A. J. Phys. Chem. B 2003, 107, 1765–1771. (15) Giorgetti, E.; Giusti, A.; Giammanco, F.; Laza, S.; DelRosso, T.; Dellepiane, G. Appl. Surf. Sci. 2007, 254, 1140–1144. (16) Inasawa, S.; Sugiyama, M.; Yamaguchi, Y. J. Phys. Chem. B 2005, 109, 9404–9410. (17) Hu, M.; Hartland, G. V. J. Phys. Chem. B 2002, 106, 7029–7033. (18) Saunders, W. A. Phys. Rev. A 1992, 46, 7028–7041. (19) Zheng, J.; Zhang, C.; Dickson, R. M. Phys. Rev. Lett. 2004, 93 (7), 77402–77405. (20) Wang, D.; Imae, T. Chem. Lett. 2005, 34, 640–642. (21) Wang, D.; Imae, T. J. Am. Chem. Soc. 2004, 126, 13204–13205. (22) Esumi, K.; Suzuki, A.; Aihara, N.; Usui, K.; Torigoe, K. Langmuir 1998, 14, 3157–3159. (23) Pal, K. Talanta 1998, 46, 583–587. (24) Hedden, R. C.; Bauer, B. J.; Smith, A. P.; Gr€ohn, F.; Amis, E. Polymer 2002, 43, 5473–5481. (25) Torigoe, K.; Suzuki, A.; Esumi, K. J. Colloid Interface Sci. 2001, 241, 346–356. (26) Wang, D.; Imae, T.; Miki, M. J. Colloid Interface Sci. 2007, 306, 222–227.

’ ACKNOWLEDGMENT Funding from the Italian Project PRIN2007 “Metal-organic plasmonic nanostructures for sensor applications” is acknowledged. ’ REFERENCES (1) Jain, P. K.; Lee, K. S.; El-Sayed, M. A. J. Phys. Chem. B 2006, 110, 7238–7248. (a) Jain, P. K.; Yeo, B. S.; Syadler, J.; Schmid, T.; Zenobi, R.; Zhang, W. Chem. Phys. Lett. 2009, 472, 1–13. Kumar, A.; Kumar Vemula, P.; Ajayan, P. M.; John, J. Nat. Mater. 2008, 7, 236–241. (2) Chen, S.; Wang, J.; Foulger, S. H.; Carroll, D. L. J. Am. Chem. Soc. 2003, 125, 16186–16187. Alloisio, M.; Demartini, A.; Cuniberti, C.; Dellepiane, G.; Jadhav, S.; Thea, S.; Giorgetti, E.; Gellini, C.; MunizMiranda, M. J. Phys. Chem. C 2009, 113, 19475–19581. Seo, D.; Park, J. C.; Song, H. J. Am. Chem. Soc. 2006, 128, 14863–14870. Loo, C.; Lowery, A.; Halas, N.; West, J.; Drezek, R. Nano Lett. 2005, 5, 709–711. Chen, J.; Saeki, F.; Wiley, B. J.; Cang, H.; Cobb, M. J.; Li, Z.; Au, L.; Zhang, H.; Kimmey, M. B.; Xia, Y. Nano Lett. 2005, 5, 473–477. (3) Mafune, F.; Khono, J.; Takeda, Y.; Kondow, T.; Sawabe, H. J. Phys. Chem. B 2001, 105, 5114–5120. Werner, D.; Hashimoto, S.; Tomita, T.; Matsuo, S.; Makita, Y. J. Phys. Chem. C 2008, 112, 16801– 16808. Silvestre, J. P.; Poulin, S.; Kabashin, A. V.; Sacher, E.; Meunier, M.; Luong, J. H. T. J. Phys. Chem. B 2004, 108, 16864–16869. (4) Jain, P. K.; El-Sayed, I. H.; El-Sayed, M. A. Nano Today 2007, 2, 18–29. (5) Yamada, K.; Tokumoto, Y.; Nagata, T.; Mafune’, F. J. Phys. Chem. B 2006, 110, 11751–11756. Crespo, P.; Litran, L.; Rojas, T. C.; Multigner, M.; Sanchez-Lopez, J. C.; Garca, M. A.; Hernando, A.; Penades, S.; Fernandez, A. Phys. Rev. Lett. 2004, 93, 87204–87208. (6) Giusti, A.; Giorgetti, E.; Laza, S.; Marsili, P.; Giammanco, F. J. Phys. Chem. C 2007, 111, 14984–14991. (7) Giorgetti, E.; Giusti, A.; Giammanco, F.; Marsili, P. Opt. Spectrosc. 2009, 107, 505–509. (8) Giammanco, F.; Giorgetti, E.; Marsili, P.; Giusti, A. J. Phys. Chem. C 2010, 114, 3354–3363. (9) Idrobo, J. C.; Walkosz, W.; Yip, S. F.; Og€ut, S.; Wang, J.; Jellinek, J. Phys. Rev. B 2007, 76, 205422–205433. (10) Kreibig, U. Vollmer, M. Optical Properties of Metal Clusters; Springer-Verlag: Berlin, 1995. (11) Giorgetti, E.; Giusti, A.; Giammanco, F.; Marsili, P.; Laza, S. Molecules 2009, 14, 3731–3753. (12) Zheng, J.; Petty, J. T.; Dickson, R. M. J. Am. Chem. Soc. 2003, 125, 7780–7781. 5020

dx.doi.org/10.1021/jp108042m |J. Phys. Chem. C 2011, 115, 5011–5020