J. Phys. Chem. B 2001, 105, 9913-9917
9913
ARTICLES Adsorbate-Induced Quenching of Hot Electrons in Gold Core-Shell Nanoparticles S. L. Westcott Department of Electrical and Computer Engineering, Rice UniVersity, Houston, Texas 77005
R. D. Averitt MST-10, Condensed Matter and Thermal Physics, Los Alamos National Laboratory, Los Alamos, New Mexico 87545
J. A. Wolfgang and P. Nordlander Department of Physics, Rice UniVersity, Houston, Texas 77005
N. J. Halas* Department of Electrical and Computer Engineering and Department of Chemistry, Rice UniVersity, Houston, Texas 77005 ReceiVed: March 29, 2001; In Final Form: June 26, 2001
We have investigated the effect of molecular adsorbates on the ultrafast electron dynamics in nanoshell particles. These nanoparticles consist of a thin gold shell surrounding a gold sulfide core. Pump-probe transmission measurements at 1.5 eV on unmodified nanoshells in aqueous solution yield an electron relaxation lifetime of ∼2.7 ps. The lifetime decreased with the adsorption of p-aminobenzoic acid (∼1.7 ps) or aniline (∼1.9 ps) on the nanoshells. With adsorbed p-mercaptobenzoic acid or n-propylamine, electron thermalization occurred in ∼2.4 or ∼2.8 ps, respectively. Surface-enhanced Raman signals were detected from the aromatic molecules, confirming their adsorption on the nanoshells. Density functional theory calculations indicate that the molecules providing the strongest modification of the dynamics possess the largest induced dipole moments near a metal surface. This suggests that the adsorbate-induced perturbation of the nanoshell electron dynamics is primarily electronic in nature.
1. Introduction The dynamics of electron relaxation in nanostructures and systems of reduced dimensionality is a subject of considerable scientific and technological interest. As the size of a metallic nanoparticle is decreased, the electronic structure and the screening properties of the free electrons are modified, changing the relative importance of the various relaxation mechanisms that are present in bulk systems.1-3 The ultrafast dynamics of electrons in metals can be investigated by studying their response to femtosecond optical pulses. In bulk metals, electron relaxation is primarily determined by electron-phonon coupling.4,5 For metal nanoparticles, the electron dynamics are modified by increased electron-surface collisions,1 reduced electron-lattice phonon coupling,6 and modified screening of the electronelectron interaction.2,3,4,7 As predicted for electron-surface phonon coupling,6 the hot electron decay lifetime for tin, gallium,8 and silver9 nanoparticles decreased with decreasing particle size. For gold nanoparticles, the electron decay lifetime was independent of particle size10 and shape.11 The electron * Author to whom correspondence should be addressed. Fax: (713) 3485686. E-mail:
[email protected].
relaxation lifetime in silver nanoparticles was longer in silicate glass than in alumina,9 but was unaffected by changing the pH or iodide ion concentration in aqueous solution.12 Studies of hot electron relaxation in semiconductor nanoparticles and at semiconductor surfaces have clearly demonstrated the importance of adsorbates on the relaxation dynamics.13,14 The introduction or destruction of impurity-induced trap states can modify relaxation times by 5 orders of magnitude.14 A similar effect of adsorbate-induced electron relaxation in solid metallic nanoparticles has not yet been observed. However, diverse electron relaxation lifetimes have been measured for nanoparticles made by different methods. Metal nanoshells are uniquely well suited for studying the effect of adsorbed impurities on electron relaxation in nanoparticles. These nanoshells are novel nanoparticles consisting of a dielectric core coated by a thin metal shell. For gold sulfide core-gold shell nanoparticles, the shell is typically 3-5 nm thick. The electron-surface scattering rate in the shell is an order of magnitude larger than the electron-phonon scattering rate of bulk gold, suggesting that the surface and chemisorbed impurities could have a significant influence on the nanoshell properties. Like other metal nanoparticles, the optical properties
10.1021/jp011213t CCC: $20.00 © 2001 American Chemical Society Published on Web 09/19/2001
9914 J. Phys. Chem. B, Vol. 105, No. 41, 2001 of the nanoshells are dominated by the collective electron oscillation known as the plasmon resonance.15 In the nanoshell geometry, the surface plasmon resonance frequency is determined by the relative size of the core radius and the shell thickness, as well as the dielectric functions of the core, shell, and embedding media.16,17 Thus, the strong optical absorption and scattering associated with the surface plasmon resonance can be placed at wavelengths from 600 nm to 2500 nm17-19 and, in particular, at wavelengths where competing relaxation mechanisms such as interband transitions are less important.20 In the present work, we demonstrate for the first time that molecules adsorbed onto the surface of gold nanoshells can significantly modify the electron thermalization rate in these nanoparticles. As we show, the influence of an adsorbate molecule on the relaxation of hot electrons in a nanoscale metallic substrate appears to be directly related to the dipole moment of the adsorbate-metal complex.
Westcott et al.
Figure 1. The solid lines are the experimentally measured transient bleaching which occurs in nanoshells in aqueous solutions (A) and in poly(vinyl alcohol) films (B). The dotted lines were obtained by fitting the data to eq 1, with relaxation lifetimes of 3.3 ps and 1.6 ps, respectively.
2. Experimental Section Nanoshells with a gold sulfide core and a gold shell were prepared by mixing aqueous solutions of sodium sulfide and gold chloride.17,22 The reaction also produces solid gold nanoparticles. These nanoshell solutions were concentrated by centrifuging and diluted in the desired solvent to a volume of 200 µL. A 1 mm path length cell was used for both pumpprobe and UV-visible absorbance measurements. The optical density was typically 0.3 to 0.8 at the nanoshell resonance wavelength (near 840 nm). Several batches of nanoshells were used in this work with 20% size distributions and total average diameters ranging from 30 to 50 nm. Comparing measured UVvisible absorbance spectra to Mie scattering theory, we determined shell thicknesses of 3 to 5 nm.17 The ultrafast dynamics of the metal nanoshells were investigated using optical transmission pump-probe measurements. A cavity-dumped Ti:sapphire laser operating at a nominal wavelength of 840 nm with a repetition rate of 400 kHz and pulse width of 70 to 80 fs was used. The beam was split and attenuated to provide a pump beam fluence ranging from 500 to 550 µJ/cm2 and a probe beam with 5% the fluence of the pump beam. For this fluence range, the calculated peak electron temperature is 2500 to 3000 K, using Te ) (TL + 2U/γ)1/2, where Te is the peak electron temperature, TL is the lattice temperature, U is the absorbed energy divided by the total volume of gold in the shells and γTe is the electronic specific heat. 3. Results and Discussion Initial measurements of the induced change in transmission of gold-gold sulfide nanoshells were taken with the nanoshells embedded in a poly(vinyl alcohol) film. Calculations of how a hot electron distribution would change the dielectric function of the metal shell agreed with the observed magnitude of transient absorption and transient bleaching in gold-gold sulfide nanoshells.21,23 In those films, the electron relaxation lifetime was 1.65 ( 0.10 ps. This is similar to the lifetime measured for solid gold nanoparticles in water excited to 2200 K.11 Contrary to expectations, when the pump power was halved (and the peak electron temperature decreased), the lifetime did not change for nanoshells in PVA films. Furthermore for nanoshells in aqueous solution, the electron relaxation was notably longer (2.5 to 3.5 ps), as can be seen in Figure 1. This clearly demonstrates that the nanoshell-embedding medium interface plays an important role in determining the electronic relaxation. However, the electron relaxation was not changed by suspending nanoshells in either water or D2O.
Figure 2. Solid lines are transient bleaching data (average of 3 scans) while dotted lines correspond to eq 1. Curves are displaced for clarity. In (a), A is from nanoshells in water with τd ) 2.6 ( 0.1 ps, B is in 7.2 µM pABA (τd ) 2.5 ( 0.1 ps), and C is in 14.4 µM pABA (τd ) 1.7 ( 0.1 ps). In (b), A is in water (τd ) 2.7 ( 0.1 ps) and B is in 7.2 µM aniline (τd ) 1.9 ( 0.1 ps). In (c), A is in water (τd ) 2.8 ( 0.2 ps), B is in 20 µM n-propylamine (τd ) 2.8 ( 0.3 ps), and C is in 300 µM n-propylamine (τd ) 3.0 ( 0.2 ps). In (d), A is in water (τd ) 2.6 ( 0.1ps), B is in 7.2 µM pMBA (τd ) 2.4 ( 0.4 ps), and C is in 21.6 µM pMBA (τd ) 2.6 ( 0.3 ps).
Therefore changes in solvent vibrations did not affect the relaxation. For multiple samples diluted from the same concentrated batch of nanoshells, the variation in lifetime was (0.5 ps. Because the plasmon resonance of solid gold nanoparticles occurs near 520 nm, no transient changes in transmission at 850 nm were measured from solutions or films23 containing only solid gold nanoparticles made by other preparation techniques. For one nanoshell batch, the relative concentration of solid gold nanoparticles was reduced by a factor of 2 after three additional cycles of centrifuging and redispersing the solution. This did not affect the relaxation lifetime of either the film or the solution samples. In either solutions or films, the relaxation was also independent of nanoshell size and sample absorbance. The effect of surface impurities was investigated by adding molecules with amine (NH2) or thiol (SH) functional groups which bind to the gold nanoshell surface. Pump-probe data for nanoshells in water and in various concentrations of p-aminobenzoic acid (pABA), aniline, n-propylamine, and p-mercaptobenzoic acid (pMBA) is shown in Figure 2. To quantify the electron relaxation, we use an equation that accounts for the initially excited nonthermalized electrons, the hot Fermi distribution of electrons, and the phonons as coupled subsystems.24 The expression for the induced change in transmission ∆T/T with time t is
Hot Electrons in Gold Core-Shell Nanoparticles
[
( )] ( )
∆T -t ) 1 - exp T τr
exp
-t + Υoff τd
J. Phys. Chem. B, Vol. 105, No. 41, 2001 9915
(1)
in which τr is the non-Fermi electron thermalization rise time, τd is the hot electron Fermi distribution decay time, and Υoff is an offset due to effects that persist after the electrons and the lattice have come to the same (elevated) temperature. Other researchers have observed that this feature decays over a time scale of 50 to 200 ps as heat diffuses from the excitation region11,12 but that decay cannot be resolved in our 5 ps scans. Equation 1 was convolved with a 78 fs Gaussian pulse to account for the laser pulse width. All 3 variables (τr, τd, and Υoff) were determined by optimally fitting the data using the Nelder-Mead simplex method. For all of the samples, Υoff was 0.30 ( 0.05. τr varied from 75 to 105 fs due to a combination of our data resolution (50 fs) and laser pulse width. The values of τd are given in the caption for Figure 2 demonstrating how electron relaxation was affected by molecular adsorbates. In Figure 2a, the electron relaxation was similar in water or in 7.2 µM pABA, but was much more rapid in 14.4 µM pABA. For aniline, faster electron relaxation occurred at a 7.2 µM concentration (Figure 2b). These molecular concentrations were on the order of monolayer coverage of the nanoshell surfaces. Increasing the pABA or aniline concentrations to 300 µM did not further decrease the lifetime. As shown in Figure 2c,d, the electron relaxation was insignificantly affected by the presence of n-propylamine or pMBA, even at a concentration of 300 µM. Since aggregation of the nanoshells might also decrease the electron relaxation lifetime,25 dynamic light scattering and transmission electron microscopy were used to confirm that none of the added molecules caused aggregation. For each molecule, these measurements were repeated on 3-5 samples made from several batches of nanoshells. The average lifetimes are given in Table 1. To confirm that pABA, aniline, and pMBA actually bind to the nanoshells, we used Raman spectroscopy. Raman scattering by molecules is generally quite weak and no signal would normally be detected from 300 µM concentrations. However Raman scattering can be enhanced by the local electric fields at a roughened metal or metal colloid surface.26 The local electric field enhancement due to the nanoshell plasmon resonance allows nanoshells to be used as substrates for surface-enhanced Raman spectroscopy (SERS) in the near-infrared.27 Using a Nicolet FT-Raman spectrometer with a 1064 nm Nd:YAG laser source, the spectra shown in Figure 3 were obtained from molecules adsorbed to the surface of the nanoshells in solution. Even at higher, near-saturation, concentrations, the Raman signal from these molecules in aqueous solutions was below the detection threshold. Detecting a Raman signal from the solutions of nanoshells and molecules verified that the molecules were indeed bound to the nanoshell surfaces. The Raman spectra of the pure molecules and of the molecules adsorbed on nanoshell surfaces are not identical. When the molecules bind to a surface, the molecular configuration changes and a degree of charge transfer may occur, leading to peak shifts and broadening in the Raman spectra. The selection rules that determine peak strength of an enhanced mode depend on the molecular orientation at the surface and result in different relative peak heights in comparison to the spectra of unbound molecules.26 Pure n-propylamine has a Raman signal 10 times weaker than the signal from the other molecules and no Raman signal was detected from bound n-propylamine. The binding of the npropylamine was instead confirmed by demonstrating that
Figure 3. (a) Raman spectra of 300 µM pABA adsorbed to nanoshells (solid line) and solid pABA (dotted line, ×0.05). (b) Raman spectra of 300 µM aniline adsorbed to nanoshells (solid line) and neat aniline (dotted line, ×0.05). (c) Raman spectra of 21.6 µM pMBA adsorbed to nanoshells (solid line) and solid pMBA (dotted line, ×0.05).
TABLE 1: Average Electron Decay Lifetimes for Nanoshells with Several Adsorbates adsorbate
τd (ps)
none pABA aniline pMBA propylamine
2.8 ( 0.4 1.7 ( 0.2 1.9 ( 0.1 2.4 ( 0.3 2.8 ( 0.2
nanoshells with n-propylamine did not aggregate when a bifunctional molecule such as p-aminobenzenethiol(pABT) was added. Without n-propylamine, the pABT caused nanoshells to aggregate with an approximate doubling of their hydrodynamic radius (as measured by dynamic light scattering) and a broadening of their UV-visible absorbance. 4. Quantum Chemical Calculations All four molecules bonded to the nanoshell surface, but only pABA and aniline strongly modified the nanoshell electron dynamics. To understand why, the interaction between the molecule and the surface was modeled using quantum chemical techniques. A molecular mechanics method with empirical interaction parameters was used for an initial determination of the configuration and position of each adsorbate molecule on a 31 atom Cu(111) surface.28 Because the core electrons of gold atoms travel at relativistic velocities,29 it is not computationally feasible to use gold atoms in the Density Functional formalism. For chemisorption effects mediated by the response of the free electrons of the substrate, such as induced dipole moments and adsorbate potentials, we expect only minor differences between a copper and a gold surface. The results of the molecular mechanics simulations were used as starting points for ab initio calculations. The Density Functional formalism30,31 was used to optimize the minimum energy configuration of each molecule at several heights above the metal surface. The calculation used the exchange correlation potential by Barth and Hedin32 and numerical atomic basis functions comprising all occupied ground state orbitals plus all orbitals for the doubly ionized atomic states together with d-type polarization functions. For both pABA and pMBA, the molecular mechanics simulation determined that the plane of the benzene ring was parallel to the metal surface if the molecules were in a vacuum. However, including several water molecules as well as the pABA (or pMBA) molecule above the metal surface resulted in a minimum energy orientation in which the benzene ring was
9916 J. Phys. Chem. B, Vol. 105, No. 41, 2001
Westcott et al.
Figure 4. The dipole moment as a function of height above a copper surface for pABA (thick solid line), aniline in the perpendicular orientation (dashed), aniline in the parallel orientation (dotted), propylamine (dash-dot), and pMBA (thin solid line). The circles on each line indicate the height and dipole moment at the minimum energy position.
calculations). Note that for aniline the calculated dipole moment for the parallel orientation agrees with the observed trend, also suggesting that the orientation of aniline on nanoshells is parallel to the surface. Physically, the electronic perturbation induced by the adsorbate, as reflected by the magnitude of the induced dipole moment, introduces an additional channel by which hot electrons can decay. In this picture, the interaction between the molecules and the metal perturbs the electronic potential resulting in a local change in the screening of the Coulomb interaction between electrons, modifying the relaxation. Among other possible decay mechanisms, there could be a direct energy transfer to the molecules.35 or an adsorbate-induced enhanced electron-phonon scattering mechanism. Clearly, a more detailed analysis is needed to determine the microscopic origin of the present effect. In summary, the electron relaxation rate in gold-gold sulfide nanoshells was increased when p-aminobenzoic acid or aniline was adsorbed to the surface. Other molecules, n-propylamine and p-mercaptobenzoic acid, had a negligible effect on the electron relaxation lifetime. The change in the electron relaxation rate appears to be electronic and the molecules which cause a change in electron relaxation lifetime have stronger electronic interactions with the metal, according to calculations of the induced dipole moment of the molecule-metal complex. We expect that a similar increase in electronic relaxation rate would occur for other molecules which interact strongly with the metal surface. While these measurements were performed on gold nanoshells, we expect that these results may apply more generally to other confined metallic systems.
Figure 5. Comparison of calculated molecule-metal dipole moment and electron relaxation in nanoshells with adsorbed pABA (circle), aniline (square for a parallel adsorption orientation, triangle for a perpendicular adsorption orientation), pMBA (diamond), or propylamine (star).
Acknowledgment. This work was supported by the Robert A. Welch Foundation, the National Science and Engineering Research Council of Canada (S.L.W.), and the Multi-University Research Initiative of the Army Research Office. References and Notes
perpendicular to the metal surface such that the carboxylic acid group was solvated. Thus the perpendicular orientation of pABA, pMBA, and aniline were examined in the ab initio calculations. Aniline does not have a carboxylic acid group to solvate and previous experimental studies have determined that its benzene ring was oriented parallel to the metal surface.33,34 Therefore, ab initio calculations were also done with aniline oriented parallel to the surface. Using the electronic structure determined by the ab initio calculations, the dipole moment of the adsorbate/metal complex and the total energy were calculated. The dipole moment as a function of height above the metal surface is shown in Figure 4 for pABA, aniline, pMBA, and propylamine. For each adsorbate, the height (and dipole moment) corresponding to minimum energy is marked with a circle. The dipole moments of the molecules which have a larger effect on the electron relaxation, pABA (5.8 Debye) and aniline (4.6 Debye when parallel to the surface), are considerably stronger than the dipole moments of pMBA (2.4 Debye) and n-propylamine (1.5 Debye). The dipole moment was stronger for aniline in the parallel orientation (4.6 Debye) than in the perpendicular orientation (2.2 Debye). In the parallel orientation the benzene ring π electrons can couple with the metal surface. The connection between the induced dipole moment and the average electron relaxation lifetime inside the metal can be seen in Figure 5. The range of dipole moments is the dipole moment calculated at the minimum energy height and at the minimum energy height (0.026 nm (the height increment for the ab initio
(1) Del Fatti, N.; Valle´e, F.; Flytzanis, C.; Hamanaka, Y.; Nakamura, A. Chem. Phys. 2000, 251, 215. (2) Voisin, C.; Christofilos, D.; Del Fatti, N.; Valle´e, F.; Pre´vel, B.; Cottancin, E.; Lerme´, J.; Pellarin, M.; Broyer, M. Phys. ReV. Lett. 2000, 85, 2200. (3) Shahbazyan, T. V.; Perakis, I. E.; Bigot, J.-Y. Phys. ReV. Lett. 1998, 81, 3120. (4) Halte´, V.; Guille, J.; Merle, J.-C.; Perakis, I.; Bigot, J.-Y. Phys. ReV. B 1999, 60, 11738. (5) Kaganov, M. I.; Lifshitz, I. M.; Tanatarov, L. V. Zh. Eksp. Teor. Fiz. 1956, 31, 232; SoV. Phys. JETP 1957, 4, 173. (6) Belotskii, E. D.; Luk’yanets, S. N.; Tomchuk, P. M. Zh. Eksp. Teor. Fiz. 1992, 101, 163; SoV. Phys. JETP 1992, 74, 88. (7) Shahbazyan, T. V.; Perakis, I. E. Chem. Phys. 2000, 251, 37. (8) Nisoli, M.; Stagira, S.; De Silvestri, S.; Stella, A.; Tognini, P.; Cheyssac, P.; Kofman, R. Phys. ReV. Lett. 1997, 78, 3575. (9) Halte´, V.; Bigot, J.-Y.; Palpant, B.; Broyer, M.; Pre´vel, B.; Pe´rez, A. Appl. Phys. Lett. 1999, 75, 3799. (10) Hodak, J. H.; Henglein, A.; Hartland, G. V. J. Chem. Phys. 2000, 112, 5942. (11) Link, S.; El-Sayed, M. A. J. Phys. Chem. B 1999, 103, 8410. (12) Roberti, T. W.; Smith, B. A.; Zhang, J. Z. J. Chem. Phys. 1995, 102, 3860. (13) Halas, N. J.; Bokor, J. Phys. ReV. Lett. 1989, 62, 1679. (14) Burda, C.; Green, T. C.; Link, S.; El-Sayed, M. A. J. Phys. Chem. B 1999, 103, 1783. (15) Kreibig, U.; Vollmer, M. Optical Properties of Metal Clusters; Springer: Berlin, 1995. (16) Neeves, A. E.; Birnboim, M. H. J. Opt. Soc. Am. B 1989, 6, 787. (17) Averitt, R. D.; Sarkar, D.; Halas, N. J. Phys. ReV. Lett. 1997, 78, 4217. (18) Averitt, R. D.; Westcott, S. L.; Halas, N. J. J. Opt. Soc. Am. B 1999, 16, 1824. (19) Oldenburg, S. J.; Jackson, J. B.; Westcott, S. L.; Halas, N. J. Appl. Phys. Lett. 1999, 75, 2897.
Hot Electrons in Gold Core-Shell Nanoparticles (20) Although the relaxation is not affected by direct interband relaxation mechanisms at these wavelengths (∼840 nm), dynamic screening of the d-band electrons by the conduction electrons is still important (see ref 21). (21) Averitt, R. D.; Westcott, S. L.; Halas, N. J. Phys. ReV. B 1998, 58, 10203. (22) Zhou, H. S.; Honma, I.; Komiyama, H.; Haus, J. W. Phys. ReV. B 1994, 50, 12052. (23) Averitt, R. D.; Westcott, S. L.; Halas, N. J. J. Opt. Soc. Am. B 1999, 16, 1814. (24) Sun, C.-K.; Valle´e, F.; Acioli, L. H.; Ippen, E. P.; Fujimoto, J. G. Phys. ReV. B 1994, 50, 15337. (25) Feldstein, M. J.; Keating, C. D.; Liau, Y.-H.; Natan, M. J.; Scherer, N. F. J. Am. Chem. Soc. 1997, 119, 6638. (26) Moskovits, M. ReV. Mod. Phys. 1985, 57, 783.
J. Phys. Chem. B, Vol. 105, No. 41, 2001 9917 (27) Oldenburg, S. J.; Westcott, S. L.; Averitt, R. D.; Halas, N. J. J. Chem. Phys. 1999, 111, 4729. (28) HyperChem Computational Chemistry Part 2: Theory and Methods; Autodesk: New York, 1992. (29) Balasubramanian, K. J. Mol. Struct. 1989, 202, 291. (30) Kohn, W.; Sham, L. J. Phys. ReV. A 1965, 140, 1133. (31) Lou, L.; Guo, T.; Nordlander, P.; Smalley, R. E. J. Chem. Phys. 1993, 99, 5301. (32) von Barth, U.; Hedin, L. J. Phys. C 1972, 5, 1629. (33) Shindo, H. J. Chem. Soc., Faraday Trans. 1 1986, 82, 45. (34) Fic¸ iciogˇlu, F.; Kuliyev, S.; Kadirgan, F. J. Electroanal. Chem. 1996, 408, 231. (35) Germer, T. A.; Stephenson, J. C.; Heilweil, E. J.; Cavanagh, R. R. Phys. ReV. Lett. 1993, 71, 3327.