Optomechanics of Single Aluminum Nanodisks - Nano Letters (ACS

Letter pubs.acs.org/NanoLett

Optomechanics of Single Aluminum Nanodisks Man-Nung Su,† Pratiksha D. Dongare,‡ Debadi Chakraborty,⊥ Yue Zhang,‡,§ Chongyue Yi,† Fangfang Wen,† Wei-Shun Chang,† Peter Nordlander,§,∥ John E. Sader,⊥ Naomi J. Halas,†,§,∥ and Stephan Link*,†,∥ †

Department of Chemistry, ‡Applied Physics Graduate Program, §Department of Physics, and ∥Department of Electrical and Computer Engineering, Laboratory for Nanophotonics, Rice University, Houston, Texas 77005, United States ⊥ School of Mathematics and Statistics, University of Melbourne, Melbourne, Victoria 3010, Australia S Supporting Information *

ABSTRACT: Aluminum nanostructures support tunable surface plasmon resonances and have become an alternative to gold nanoparticles. Whereas gold is the most-studied plasmonic material, aluminum has the advantage of high earth abundance and hence low cost. In addition to understanding the size and shape tunability of the plasmon resonance, the fundamental relaxation processes in aluminum nanostructures after photoexcitation must be understood to take full advantage of applications such as photocatalysis and photodetection. In this work, we investigate the relaxation following ultrafast pulsed excitation and the launching of acoustic vibrations in individual aluminum nanodisks, using single-particle transient extinction spectroscopy. We find that the transient extinction signal can be assigned to a thermal relaxation of the photoexcited electrons and phonons. The ultrafast heating-induced launching of in-plane acoustic vibrations reveals moderate binding to the glass substrate and is affected by the native aluminum oxide layer. Finally, we compare the behavior of aluminum nanodisks to that of similarly prepared and sized gold nanodisks. KEYWORDS: Acoustic vibrations, aluminum nanostructures, nanomechanics, ultrafast spectroscopy, single-particle spectroscopy

P

nanoparticle cools to its initial temperature via phonon− phonon coupling with the surrounding environment.19 Because of the ultrafast heating, acoustic vibrations of the nanoparticle lattice can be launched impulsively and result in an oscillatory modulation of the transient extinction decay due to volume expansion and contraction that can shift and broaden the surface plasmon resonance.20−24 The measured acoustic vibration frequencies can be simulated with bulk elastic properties using a classical elastic model,25 and the damping times of the acoustic modes provide insight into the energy dissipation to the local environment.26 Therefore, transient extinction spectroscopy can be utilized as a noninvasive method to also study the mechanical properties of nanoscale objects. Single-particle measurements are advantageous over ensemble studies because they are capable of measuring intrinsic damping times, avoiding sample size and shape heterogeneities that strongly affect acoustic vibration frequencies.27,28 Significant research efforts have been devoted to the ultrafast dynamics and mechanical properties of noble metal nanostructures using both ensemble 23,24,28−45 and single-particle22,26,27,39,46−52 transient extinction measurements. Thus far, the vast majority of these studies has focused on gold nanostructures. However, the spectral tunability of plasmon resonances in gold is limited because of interband transitions.

lasmonic nanostructures have attracted significant attention because of their unique optical properties1,2 making them ideal platforms for a variety of applications such as sensors,3,4 medical therapy agents,5 catalysts,6−8 solar cells,9,10 photodetectors, 11 surface-enhanced spectroscopy substrates,12,13 ultrafast optical switches,14,15 and opto-mechanical devices.16,17 Their optical properties arise from highly wavelength tunable surface plasmon resonances, the collective oscillation of conduction band electrons. The resonance condition of this collective motion is determined by the nanostructure material, size, shape, and local environment.2 While plasmon line shape engineering is important for achieving efficient light-matter interactions, understanding the relaxation processes following photoexcitation is important for any application that is based on light absorption rather than scattering. Transient extinction spectroscopy is an ideal tool for studying ultrafast electron dynamics in metal nanoparticles. An ultrafast pump pulse excites a plasmon that decays into electron−hole pairs or excites directly intrinsic inter- and intraband transitions, while a time-delayed probe pulse measures the temporal changes in the optical extinction that are caused by carrier relaxation. Typically in metal nanoparticles, the change in electron distribution is described by an effective temperature corresponding to a thermalized Fermi− Dirac distribution following initial electron−electron scattering.18 Electron−phonon coupling equilibrates electron and lattice temperatures within a few picoseconds and the © 2017 American Chemical Society

Received: January 24, 2017 Revised: March 14, 2017 Published: March 16, 2017 2575

DOI: 10.1021/acs.nanolett.7b00333 Nano Lett. 2017, 17, 2575−2583

Letter

Nano Letters Recently, aluminum nanostructures have emerged as an attractive alternative that support plasmon resonances from the UV to the IR spectral ranges.53−58 Furthermore, aluminum is the third most abundant element and hence presents a sustainable alternative to noble metals. Narrow interband transitions, high electron densities, and native oxide layers all affect the optical response of aluminum nanostructures.59 However, very little is known about the ultrafast relaxation dynamics and this lack of knowledge hinders the further development of aluminum plasmonics. Energy relaxation in aluminum thin films has been investigated and has revealed faster electron−electron scattering and electron−phonon relaxation compared to gold.60−65 While lifetimes of 500 fs and ∼1 ps were reported for gold films for these processes,66,67 shorter times of 200 and 500 fs, respectively, were found for aluminum films.60,61 However, considering the electron−phonon coupling constants of gold and aluminum, which are 2.1 × 1016 and 4.9 × 1017 W m−3 K−1, respectively, the reported electron−phonon relaxation time may be an overestimation caused by high excitation powers. Regarding time-resolved measurements of aluminum nanostructures, we are only aware of the single probe wavelength study by Pomfret et al. who reported relaxation dynamics of aluminum nanowires fabricated by electrochemical deposition into a porous template.68 To the best of our knowledge, no studies exist on the ultrafast dynamics of zero-dimensional aluminum nanostructures. In this work, we studied the acoustic vibrations of aluminum nanodisks of varying diameters using single-particle transient extinction spectroscopy with tunable probe wavelengths. We found that the transient extinction was modulated by acoustic vibrations that indicate a thermal relaxation mechanism. From the size dependence of the in-plane breathing mode and simulations, we were furthermore able to characterize the binding to the substrate and the role of the native oxide layer. These results are contrasted to the well-studied relaxation dynamics in gold nanostructures. The aluminum nanodisks investigated here were fabricated on glass substrates by electron-beam lithography and were first characterized using a combination of dark-field spectroscopy, scanning electron microscopy (SEM), and finite difference time domain (FDTD) simulations as illustrated in Figure 1a−c. The disk diameters ranged from 170 to 350 nm with a disk thickness of 35 nm. SEM yielded the exact bottom diameters which were ∼5 nm larger (Table S1) and confirmed a small size variation on the order of 1% (Figure S1) as well as a tilt angle of 75° for the side walls caused by the aluminum evaporation process (Figure S2). Finally, Figure S3 reports the disk thickness as measured with an atomic force microscope (AFM). Single-particle scattering spectra show that the scattering response redshifts and broadens with increasing disk diameter (Figure 1a). The dipolar plasmon resonance depends sensitively on the aspect ratio, that is, the ratio of disk diameter and height.69 The 600 nm resonance for the 170 nm diameter nanodisk shifts to 700 nm for the 210 nm nanodisk and broadens due to increased radiation damping. Further increase in the disk diameter also leads to the appearance of higher order modes at shorter wavelength, producing for the largest nanodisk a rather featureless scattering spectrum except for a dip around 850 nm. These higher order modes originated from the inhomogeneous charge distribution due to phase retardation induced by oblique angle excitation in a dark-field microscope.70 This intensity dip, which is seen in all nanodisks

Figure 1. Optical characterization of aluminum and gold nanodisks. (a) Experimental single-particle scattering spectra, (b) SEM images (scale bar: 100 nm), and (c) FDTD modeling of aluminum nanodisks with designed diameters of 170, 210, 250, 290, and 350 nm (see SI for details regarding the experimental setup and simulations). A 210 nm diameter gold nanodisk is included for comparison (bottom). In (c), the solid lines correspond to scattering, the dotted lines to absorption, and the dashed lines to extinction. Simulations covering a wider spectral range from 200−2000 nm are given in Figure S4 and show the emergence of additional higher energy modes. Simulations for s- and p-polarized excitations confirm the isotropic geometry of the aluminum nanodisks (Figure S5).

except for the smallest one, is caused by aluminum interband transitions, consistent with previous work.71 These relatively large nanodisk sizes were chosen so that the optical response occurred in the visible spectral range that can be accessed by our transient extinction microscope. FDTD simulations were able to reproduce all features of the scattering spectra when taking into account all experimental conditions including the nanodisk geometries, a 3 nm thick native surface oxide layer,53 and the light excitation and collection geometries (solid lines, Figure 1c). We used a reflective objective (74× magnification and numerical aperture of 0.65) instead of a refractive objective in order to avoid spectral distortion over the 500 nm wavelength range (see the Supporting Information (SI) for details). From the FDTD simulations, we obtained the nanodisk absorption spectra (dotted lines, Figure 1c), which show the size-independent aluminum interband transitions around 850 nm. The absorption spectra are less intense than the scattering response, which dominates the overall extinction spectrum (dashed lines, Figure 1c), as expected from the relatively large nanodisk sizes.57 Knowing the quantitative absorption cross sections is important because it determines the maximum temperature 2576


Letter

Nano Letters change induced by the ultrafast excitation pulse (see below and the SI). In order to be able to compare the response of the aluminum disks with gold nanostructures whose optical properties and relaxation dynamics have been extensively studied,26,28,30,32,35 we also fabricated and characterized gold nanodisks with a diameter of 210 nm and a thickness of 35 nm (bottom, Figure 1) with no adhesion layer. The measured and simulated spectra are again in excellent agreement and reveal dipole and quadrupole modes at 880 and 630 nm, respectively, while at wavelengths shorter than 600 nm interband transitions in gold contribute to the absorption. Compared to the aluminum nanodisk with the same diameter (210 nm), the surface plasmon resonances of the gold nanodisk occur at lower energies due to the lower free electron density of gold.59 The relaxation dynamics of the nanodisks were measured using a home-built single-particle transient extinction microscope (see SI and Figure S6 for details). Part of the fundamental output of a Ti:sapphire oscillator at 810 nm was frequency-doubled in a beta barium borate crystal to generate 405 nm pulses. 810 or 405 nm pulses were used to pump the interband transitions of aluminum and gold nanodisks, respectively. Interband excitation was chosen as an efficient way for ultrafast heating of the nanodisks independent of nanodisk size and plasmonic metal. Another part of the fundamental beam was used to generate a white light continuum in a photonic crystal fiber so that tunable probe wavelengths between 550 to 700 nm could be subsequently obtained by transmission through various bandpass filters. Low pump (9.27 × 104 W/cm2) and probe (2.26 × 104 W/cm2) powers ensured that the nanodisks did not melt, as independently verified by recording single-particle scattering spectra before and after the transient extinction measurements (Figure S7). The high repetition rate of the laser (86 MHz) and pump beam modulation of 720 kHz allowed us to measure transmission changes of 10−4. Wavelength-dependent probe measurements revealed that the transient extinction signal of the 170 nm diameter aluminum nanodisk is caused by bleaching (i.e., increased transmission T, ΔT/T) when probed near the surface plasmon resonance maximum (Figure 2). The smallest nanodisk was chosen for this investigation because it has the narrowest, most clearly defined surface plasmon resonance line shape. The ultrafast excitation pulse creates a hot electron distribution following interband transitions and initial electron thermalization through electron−electron scattering. This hot electron gas couples to the nanoparticle lattice by exciting phonon modes, leading to the heating of the nanostructure and its expansion. Because the latter occurs on a time scale faster than an acoustic breathing period, a nanoparticle volume expansion and contraction can be driven by ultrafast optical excitation. These processes change the optical response that is probed in the differential transmission experiments, because a difference in the occupation of electron energy levels causes a change in the dielectric function. Volume oscillations further modify the dielectric function as well as the electron density, and hence the transient optical signal. This interpretation has been widely applied to gold nanostructures,72−74 which typically display surface plasmon broadening as a result of increased electron temperatures, resulting in both increased transmission (i.e., bleaching) observed at the plasmon resonance as well as decreased transmission (i.e., induced extinction) at the edges of the plasmon band. The bleaching of the plasmon resonance for

Figure 2. Transient transmission of an aluminum nanodisk with a diameter of 170 nm and a thickness of 35 nm probed at 550, 600, 650, and 700 nm. The pump wavelength was 810 nm. The pump and probe powers were 9.27 × 104 and 2.26 × 104 W/cm2, respectively. The extracted oscillation frequencies are 27.0, 28.3, 26.0, and 27.6 GHz, respectively. The solid black lines are the experimental data with the fits based on eq 1 overlaid as dotted red lines. Offsets are added for better comparisons.

the aluminum nanodisk in Figure 2 agrees with this interpretation, considering its broad resonance and that the strongest bleaching is observed at resonance. The transient extinction of the 170 nm diameter nanodisk in Figure 2 also shows a GHz modulation that we assign to impulsively launched vibrations of the nanodisk lattice. Following a thermal relaxation model, this oscillatory feature can be fit with the following equation20 ⎛ t ⎞ ΔTHO (t ) = AHO exp⎜ − ⎟cos(2πtν − ϕ) T ⎝ τHO ⎠

(1)

where AHO, τHO, ν, and ϕ are the amplitude, damping time, frequency, and phase of a damped harmonic oscillation corresponding the acoustic breathing mode. Fitting the experimental data with eq 1 yielded an acoustic vibration frequency of 26.4 ± 3.8 GHz and a damping time of 40.0 ± 9.1 ps for the 170 nm diameter nanodisk (Table S2), independent of probe wavelength, as expected for a thermal relaxation process. Probing the lattice oscillations as a function of nanodisk diameter confirms these assignments. Figure 3a shows the transient transmission of aluminum nanodisks with diameters of 170, 210, 250, 290, and 350 nm exciting interband transitions at 810 nm. Considering the wavelength independent response of the 170 nm diameter nanodisk, we selected as probe a wavelength of 650 nm as it overlapped with the plasmon resonances of all nanodisks. For these larger nanodisks, we observe very similar transient transmission due to plasmon bleaching. The transient signal and the oscillation amplitude become smaller as the nanodisk diameter increases, despite similar induced temperature changes with the same excitation powers due to the cancellation of the size-dependent absorption cross section and lattice heat capacity (see SI and 2577


Letter

Nano Letters

Figure 3. Acoustic vibrations of aluminum nanodisks. (a) Transient transmission of aluminum nanodisks with diameters of 170 (black), 210 (blue), 250 (red), 290 (cyan), and 350 (magenta) nm pumped at 810 nm and probed at 650 nm. Solid lines are the experimental data, while fits to eq 1 for the first 250 ps until the oscillation have been completely damped are shown as dashed gray lines. The data is offset for better comparisons. The pump and probe powers were 9.27 × 104 and 2.26 × 104 W/cm2, respectively. (b) Acoustic vibration frequencies of aluminum nanodisks as a function of inverse diameter. Black circles: experimental data with error bars based on the standard deviation calculated from more than 4 nanodisks for each diameter. Blue squares and red diamonds: FEM simulations for the limiting cases of the nanodisks being free and fixed to the surface, respectively. Magenta and green lines: analytical model for the free surface case considering only a side oxide layer versus a top oxide layer, respectively. (c) Schematic illustrations of an aluminum nanodisk with both top and side oxide layers sitting on a glass substrate as considered in the FEM calculations. (d,e) A top oxide layer (d) and an oxide layer surrounding the side walls (e) of the aluminum nanodisk were considered in the analytical model.

these two extreme cases, and we conclude that the aluminum nanodisks intrinsically bind to the substrate even without an additional adhesion layer. However, the binding is not sufficiently strong to rigidly fix the bottom surface to the glass. These results are similar to what we observed for gold nanodisks with an intermediate titanium adhesion layer thickness of 2 nm using the same approach.78 Analytical modeling of the size-dependent acoustic breathing frequencies reveals that the top oxide layer has a larger impact than the oxide on the nanodisk sides. The in-plane breathing frequencies were obtained analytically by solving Navier’s equation (see SI) considering the two cases highlighted in Figure 3d,e. In the first scenario, the aluminum nanodisk has only a top oxide layer (Figure 3d), while in the second case, an oxide layer surrounding only the sides of the nanodisk was included (Figure 3e). A substrate was not considered, and the calculated frequencies as functions of nanodisk diameter are given in Figure 3b as green and magenta lines. The results with only the top oxide layer match well with the trend predicted by the FEM simulations of the free surface case (blue squares and green line in Figure 3b), while the calculated frequencies including only a side oxide layer are consistently lower. This comparison lets us conclude that, while the native oxide layer stiffens the nanodisks, the mechanical stiffening of the in-plane breathing mode is due primarily to the top oxide layer. This observation can be rationalized by considering that the acoustic vibration frequencies are proportional to the square root of the Young’s modulus E and the inverse metal density ρ, ν ∝ E/ρ .25 While the densities of aluminum and aluminum oxide are similar (2.7 and 3.95 g/cm3, respectively), the Young’s modulus of aluminum oxide is significantly larger

Table S3). However, it is difficult to compare the magnitudes of the transient extinction signals and oscillation amplitudes for nanodisks of different sizes using a single probe wavelength, because the plasmon line widths are strongly size dependent. Nevertheless, the size dependence of the acoustic oscillations follows a clear trend of decreasing frequency with increasing nanodisk diameter. In fact, the frequencies scale with the inverse diameter (black circles in Figure 3b), a clear signature of an in-plane breathing mode, confirming the assignment of the signal oscillations to lattice vibrations. This result also supports our assertion that the origin of the transient extinction signal is photothermal relaxation. The overall decay of the transient bleaching (Figures 2 and 3) is on the order of hundreds of picoseconds, consistent with phonon−phonon coupling to the surrounding medium.75−77 A quantitative analysis of the size-dependent in-plane breathing mode of the aluminum nanodisks reveals a moderate binding strength to the substrate (Figure 3b). Following the approach used to determine the binding strength of gold nanodisks to a glass substrate,78 we compare in Figure 3b the experimental frequencies (Table S2) obtained from the data by fitting with eq 1 to finite element method (FEM) simulations considering two extreme cases: the free surface case (blue squares, Figure 3b) and the fixed surface case (red diamonds, Figure 3b). In the former case, the bottom of the aluminum nanodisk can expand freely on the substrate. In the latter case, the bottom of the aluminum nanodisk is rigidly fixed to the substrate and no displacement close to the surface can occur, causing an expansion gradient toward the top surface. A 3 nmthick aluminum oxide shell surrounding the aluminum nanodisks on the top and the sides was included in these simulations (Figure 3c). The measured frequencies fall between 2578


Letter

Nano Letters (215−413 GPa)79 than aluminum (70 GPa).80 The difference in elastic modulus has a larger impact on the acoustic frequency, and the aluminum oxide shell mechanically stiffens the aluminum nanodisk. The aluminum oxide on the side of the nanodisk causes zero radial stress to the in-plane vibration, while the aluminum oxide on top of the nanodisk contributes large nonzero radial stress and therefore has a larger impact on the breathing mode frequency. For comparison, we also calculated the acoustic breathing frequencies of pure aluminum oxide nanodisks (Table S4). The significantly higher frequencies of pure aluminum oxide nanodisks compared to the measured values support these conclusions, and further substantiate that the composition of the nanodisk core is primarily aluminum metal. Considering that the overall trend of the experimental data in Figure 3b falls between all models, we must conclude that the breathing frequencies were affected by a combination of binding strength to the substrate and interfacial area of the side and top oxide layers as the nanodisk size was varied. Considering the multitude of studies performed on gold nanostructures,22−24 it is interesting to directly compare the ultrafast dynamics of single gold and aluminum nanodisks of the same diameter. Although both the launching of acoustic modes and phonon−phonon relaxation are similar, a quantitative comparison of 210 nm aluminum and gold nanodisks reveals several differences in the transient transmission (Figure 4): (1) For the gold nanodisk probed at the plasmon resonance, an initial fast relaxation on the fewpicosecond time scale is observed, corresponding to electron− phonon coupling. (2) The transient extinction signal is an order of magnitude larger for the gold nanodisk. (3) The acoustic frequency and the damping time are different for the aluminum and gold nanodisks, despite having the same physical dimensions. It is interesting to consider the origin of these differences in greater detail. (1) The initial fast relaxation in gold nanostructures has been assigned to electron−phonon coupling, which establishes a thermal equilibrium between the hot electron gas and the lattice. The hot electron distribution created by ultrafast excitation leads to a broadening of the plasmon resonance and a transient bleaching feature when probed near the resonance maximum, as done here.20 The sign and amplitude of the early part of the transient transmission therefore depend on the probe wavelength.81 However, the selection of the probe wavelength for the aluminum nanodisks cannot explain the absence of the picosecond relaxation component as demonstrated by the probe wavelength independence in Figure 2. A reduced temporal resolution of the 650 nm probe beam, obtained by continuum generation and spectral filtering, compared to using the 810 nm fundamental beam also cannot explain this observed difference in relaxation dynamics. We measured the transient transmission of the 210 nm diameter gold nanodisk with different combinations of pump/probe wavelengths (405/810, 405/650, and 810/650 nm, Figure S8). Despite differences in absolute intensities and oscillation amplitudes caused by the differences in absorption cross sections and the relative position of the probe wavelength with respect to the plasmon resonance maximum, we find that the overall features remain the same and that the time resolution is not affected (Figure S8b). Our time resolution is limited to ∼1 ps for all probe wavelengths because of our use of a diffractive high numerical aperture objective. Nevertheless, we can resolve the electron−phonon relaxation of the gold nanodisk (Figure

Figure 4. Comparison of the ultrafast dynamics for aluminum and gold nanodisks. (a) Transient transmission of a gold nanodisk with a diameter of 210 nm (green) pumped at 405 nm and probed at 810 nm compared to an aluminum nanodisk with the same diameter of 210 nm (blue) pumped at 810 nm and probed at 650 nm. The different pump wavelengths were chosen so that in each case interband transitions were excited. Solid lines are the experimental data, while fits to eq 1 are shown as dashed black lines. The data is scaled and offset for better visualization. (b) Same data but plotted for shorter time delays only. No adhesion layer was used for the gold nanodisk. For the gold nanodisk, the pump and probe power densities were 7.92 × 104 and 9.27 × 103 W/cm2, respectively. For the aluminum nanodisks, the pump and probe powers densities were 9.27 × 104 and 2.26 × 104 W/ cm2, respectively.

4b), as the time constant is pump power dependent.82 However, changing the pump power has no effect on the transient transmission of the 210 nm diameter aluminum nanodisk (Figures S11). We therefore assign the absence of the fast relaxation component for the aluminum nanodisks to the much stronger electron−phonon coupling in aluminum, making it impossible to resolve with our setup. This conclusion is supported by the observation of a faster electron−phonon relaxation time for aluminum films61 and ensemble transient extinction measurements by Owrutsky and co-workers on template grown nanowires of various metals.38,68,83 Even with a time resolution of 120 fs fast electron−phonon relaxation dynamics could not be observed in aluminum. Because the electron−phonon coupling constant for gold (2.1 × 1016 W m−3 K−1) is an order of magnitude smaller than for aluminum (4.9 × 1017 W m−3 K−1)62,63 one would expect the relaxation in aluminum to be an order of magnitude faster than for gold (∼1 ps).67 A smaller difference between the time scales for electron 2579


Letter

Nano Letters

respectively (blue and green bars in Figure 5a). Their relative ratio is 2.0, which can almost entirely be explained by the effect

thermalization and electron−phonon coupling could also explain the slower rise of the transient signal for the aluminum nanodisk in Figure 4b. (2) The smaller transient extinction signal for the aluminum nanodisks could result from a smaller change in the aluminum dielectric function after ultrafast excitation, causing in turn a smaller perturbation on its optical properties. The broad spectra might also lead to an intrinsically smaller response of the surface plasmon modes to elevated electron temperatures and volume changes. Calculations using a modified dielectric function for aluminum, due to variations in the electron and hole occupation and volume changes, would be necessary to explore these explanations; but is beyond the scope of the current study. We are, however, able to estimate upper limits for the electron and lattice temperatures of the aluminum and gold nanodisks using the calculated absorption cross sections and the two-temperature model (see SI and Table S5).84 Although the calculated electron temperature for gold is only twice as large as the one for the aluminum nanodisk, the initial few picoseconds relaxation is missing in the aluminum nanodisk transient. For the thermal lattice relaxation, the transient signal is about 10 times larger in the gold than in the aluminum nanodisk, although the lattice temperature is only 1.25 times larger for the gold nanodisk under similar excitation and probe conditions (i.e., interband excitations and probe wavelengths that were about equally blueshifted from the resonance maxima). It is possible that the two-temperature model fails here for the aluminum nanodisks and significantly overestimates the actual electron and lattice temperatures because the assumption of a clear separation between electron and lattice heating may no longer be valid for stronger electron− phonon coupling. Furthermore, trapping of charge carriers at aluminum oxide defects inside the nanostructures and the outside oxide shell could possibly lead to significantly reduced temperatures. Interestingly, the ratios between transient signal and oscillation amplitude for the gold and aluminum nanodisks are similar, which might be due to the larger thermal expansion coefficient of aluminum if indeed heating is less efficient for the aluminum nanodisk.85 (3) Aluminum nanodisks have higher acoustic vibration frequencies than gold nanodisks even when comparing nanostructures with the same geometry. The acoustic vibration frequencies of the 210 nm diameter aluminum and gold nanodisks were determined as 23.9 ± 2.3 and 8.1 ± 0.3 GHz, respectively. Considering ν ∝ E/ρ ,25 the elastic constants E and densities ρ of bulk aluminum and gold explain the difference. The values for the Young’s modulus of aluminum and gold are comparable (70 and 79 GPa, respectively), but the densities are very different (2.7 and 19.3 g/cm3, respectively). Small shifts of the acoustic frequencies from this predicted behavior are likely due to the aluminum oxide layer and potentially different binding strengths to the glass substrate.78 Smaller quality factors of aluminum nanodisks seem to suggest a faster intrinsic damping compared to gold nanodisks. The quality factor Q is defined as the product of damping time and acoustic vibration frequency, πντHO, which equivalently represents the ratio of the maximum energy stored in the nanoparticle and the energy dissipated per oscillation period. The observed quality factor Qexp is composed of contributions from intrinsic and external damping according to 1/Qexp = 1/ Qint + 1/Qext.37 The average measured quality factors of aluminum and gold nanodisks are 3.2 ± 1.1 and 6.5 ± 2.9,

Figure 5. Quality factors of aluminum and gold nanodisks. (a) Histograms of quality factor for aluminum nanodisks (blue), gold nanodisks without an adhesion layer (green), and gold nanodisks with a 2 nm titanium layer (red). Inset: Same distributions normalized to their mean. Damping times plotted against measured vibration periods for individual (b) aluminum and (c) gold nanodisks grouped by color according to their designed diameters (see legend). The pump/probe wavelength combinations for the aluminum and gold nanodisks were 810/650 and 405/810 nm, respectively. The data for the bimetallic gold/titanium nanodisks was taken from ref 78.

of the metal density on the vibration frequency and hence Q, that is ρAu /ρAl = 19.3/2.7 ≈ 2.67. The aluminum nanodisks simply have a lower quality factor because aluminum is lighter than gold. This result furthermore implies that Q is dominated by intrinsic properties such as lattice defects86 rather than by the environment (i.e., through binding to the substrate) for lithographically prepared nanostructures. To further illustrate that substrate binding did not affect damping, we also added a comparison of the quality factors for gold nanodisks with a 2 nm titanium adhesion layer to Figure 5a (red bars). According to our previous results, this adhesion layer thickness results in moderate binding of gold to the substrate,78 similar to the aluminum nanodisks (Figure 3b). 2580


Letter

Nano Letters

aluminum oxide sites, might also play a role and lead to overall reduced electron and lattice temperatures. Furthermore, we find that the native oxide layer, and in particular on top of the nanodisk, increases the breathing frequency of the aluminum nanodisks, as concluded from the size-dependent acoustic vibration frequency and simulations. Importantly, we establish that the damping of acoustic vibrations in these lithographically prepared nanostructures is dominated by intrinsic processes; the surrounding medium, including the presence of a substrate and adhesion to it, do not matter. Our study provides important information about the thermal relaxation dynamics of aluminum nanostructures and their mechanical properties, which are relevant for future applications as nonlinear photoacoustic imaging agents,88 nanomechanical resonators,89 and microbalances.90

However, despite the increase in substrate binding, the presence of the titanium layer does not significantly alter the mean quality factor (7.4 ± 4.0) when compared to the gold nanodisks without an adhesion layer, ruling out that the substrate presents a major damping channel. Simulations considering only the bulk material parameters and the binding strength to the substrate further confirm these conclusions (Figure S12). For similar moderate binding (1 GPa nm−1) gold nanodisks have larger quality factors than aluminum nanodisks. The large drop in simulated Q values when transitioning from an unbound to a tightly fixed nanodisk, in strong contradiction to the experiment for gold nanodisks (Figure 5a), indicates that substrate damping is not important in the experiment. However, the absolute values of the simulated quality factors are more than 5 times larger, suggesting that the polycrystalline nature of the lithographically prepared nanostructures and their rough surfaces (Figure S3) contribute to additional intrinsic damping, consistent with larger quality factors measured for chemically prepared nanoparticles.22,37 For the latter, the substrate and even the capping material have been shown to influence Q.26,87 For the three different samples considered (Al, Au, Au/Ti), the quality factors vary significantly among the same type of nanodisk, while the overall relative standard deviation of Q is comparable (inset in Figure 5a). The distribution in quality factors for the same diameter aluminum and gold nanodisks arises from distributions of acoustic frequencies and mostly damping times (Figure 5b,c), consistent with our previous study of lithographically prepared gold nanodisks.78 When the damping times are plotted as a function of vibration periods for each measured aluminum nanodisk (Figure 5b), it becomes obvious that the small size variation of 1% (Figure S1 and Table S1) for these electron-beam fabricated nanostructures cannot completely explain the observed variation in vibration periods (3−15%). However, the damping times varied significantly more (21−52%). For the gold nanodisks in Figure 5c, a similar trend is observed, and the corresponding values for the variation in vibration periods and damping times are 3−10% and 28−52%, respectively. We attribute the large variation in damping times to the polycrystalline nature of nanostructures fabricated by metal evaporation, creating nanodisks that have almost the same size but potentially very different grain sizes and hence a substantial density of internal lattice defects.86 Note that the heterogeneous distributions of acoustic vibration frequencies and damping times demonstrate the necessity for single-particle measurements. In summary, we investigated the acoustic vibrations of single aluminum nanodisks as a function of disk diameter and compared the results to gold nanodisks that were similarly prepared by electron-beam lithography. While we can assign the origin of the transient extinction signal of the aluminum nanodisks and its time dependence to a thermal relaxation process in which photoexcited electrons couple to phonons of the metal and then to the surrounding medium, similar to the description used for gold nanostructures, we find distinct differences between the aluminum and gold nanostructures. Electron−phonon coupling is stronger in aluminum and we therefore did not observe an initial fast relaxation on the fewpicosecond time scale. Although detailed electronic structure calculations are still needed to quantitatively relate the optical response to changes in the dielectric function caused by the photoexcited electrons, the smaller signal intensities suggest that potentially other relaxation processes, such as trapping at

■

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.nanolett.7b00333. Single-particle transient extinction spectroscopy setup; sample preparation; single-particle scattering spectroscopy; FDTD simulations; experimental scattering spectra before and after transient extinction spectroscopy; instrument response with different pump and probe wavelength combinations; two-temperature model estimating maximum electron and lattice temperatures; pump power dependence of the transient extinction signal for an aluminum nanodisk; FEM simulations; analytical model for the acoustic vibration frequencies of core−shell particles; simulated quality factors of aluminum and gold; and nanodisk size characterization (PDF)

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. ORCID

Man-Nung Su: 0000-0003-2570-4285 Wei-Shun Chang: 0000-0002-0251-4449 Peter Nordlander: 0000-0002-1633-2937 Naomi J. Halas: 0000-0002-8461-8494 Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS P.N., N.J.H., and S.L. thank the Robert A. Welch Foundation (Grants C-1220 to N.J.H., C-1222 to P.N., and C-1664 to S.L.), the Army (MURI W911NF-12-1-0407), and the Air Force (MURI FA9550-15-1-0022) for financial support. S.L. acknowledges support from the National Science Foundation (ECCS1608917). D.C. and J.E.S. acknowledge support from the Australian Research Council grants scheme and the ARC Centre of Excellence in Exciton Science. We thank Professor Greg Hartland for stimulating discussions.

■

REFERENCES

(1) Perez-Juste, J.; Pastoriza-Santos, I.; Liz-Marzan, L. M.; Mulvaney, P. Coord. Chem. Rev. 2005, 249, 1870−1901.

2581


Letter

Nano Letters

(38) Sando, G. M.; Berry, A. D.; Owrutsky, J. C. J. Chem. Phys. 2007, 127, 074705. (39) Staleva, H.; Hartland, G. V. J. Phys. Chem. C 2008, 112, 7535− 7539. (40) Tang, Y.; Ouyang, M. Nat. Mater. 2007, 6, 754−759. (41) Wheeler, D. A.; Green, T. D.; Wang, H.; Fernández-López, C.; Liz-Marzán, L.; Zou, S.; Knappenberger, K. L.; Zhang, J. Z. Chem. Phys. Lett. 2012, 543, 127−132. (42) Juvé, V.; Crut, A. l.; Maioli, P.; Pellarin, M.; Broyer, M.; Del Fatti, N.; Vallée, F. Nano Lett. 2010, 10, 1853−1858. (43) Perner, M.; Gresillon, S.; März, J.; von Plessen, G.; Feldmann, J.; Porstendorfer, J.; Berg, K. J.; Berg, G. Phys. Rev. Lett. 2000, 85, 792− 795. (44) Wang, L.; Nishijima, Y.; Ueno, K.; Misawa, H.; Tamai, N. Appl. Phys. Lett. 2009, 95, 053116. (45) Wang, L.; Nishijima, Y.; Ueno, K.; Misawa, H.; Tamai, N. J. Phys. Chem. C 2012, 116, 17838−17846. (46) Fedou, J.; Viarbitskaya, S.; Marty, R.; Sharma, J.; Paillard, V.; Dujardin, E.; Arbouet, A. Phys. Chem. Chem. Phys. 2013, 15, 4205− 4213. (47) Lo, S. S.; Devadas, M. S.; Major, T. A.; Hartland, G. V. Analyst 2013, 138, 25−31. (48) Major, T. A.; Crut, A.; Gao, B.; Lo, S. S.; Fatti, N. D.; Vallee, F.; Hartland, G. V. Phys. Chem. Chem. Phys. 2013, 15, 4169−4176. (49) Marty, R.; Arbouet, A.; Girard, C.; Mlayah, A.; Paillard, V.; Lin, V. K.; Teo, S. L.; Tripathy, S. Nano Lett. 2011, 11, 3301−3306. (50) Schumacher, T.; Kratzer, K.; Molnar, D.; Hentschel, M.; Giessen, H.; Lippitz, M. Nat. Commun. 2011, 2, 333. (51) van Dijk, M. A.; Lippitz, M.; Orrit, M. Phys. Rev. Lett. 2005, 95, 267406. (52) Yu, K.; Zijlstra, P.; Sader, J. E.; Xu, Q.-H.; Orrit, M. Nano Lett. 2013, 13, 2710−2716. (53) Knight, M. W.; King, N. S.; Liu, L.; Everitt, H. O.; Nordlander, P.; Halas, N. J. ACS Nano 2014, 8, 834−840. (54) Olson, J.; Manjavacas, A.; Liu, L.; Chang, W.-S.; Foerster, B.; King, N. S.; Knight, M. W.; Nordlander, P.; Halas, N. J.; Link, S. Proc. Natl. Acad. Sci. U. S. A. 2014, 111, 14348−14353. (55) McMahon, J. M.; Schatz, G. C.; Gray, S. K. Phys. Chem. Chem. Phys. 2013, 15, 5415−5423. (56) Bohren, C. F.; Huffman, D. R. Absorption and scattering of light by small particles; Wiley, 1983. (57) Langhammer, C.; Schwind, M.; Kasemo, B.; Zorić, I. Nano Lett. 2008, 8, 1461−1471. (58) Brown, A. M.; Sundararaman, R.; Narang, P.; Goddard, W. A.; Atwater, H. A. ACS Nano 2016, 10, 957−966. (59) Zorić, I.; Zäch, M.; Kasemo, B.; Langhammer, C. ACS Nano 2011, 5, 2535−2546. (60) Rethfeld, B.; Kaiser, A.; Vicanek, M.; Simon, G. Phys. Rev. B: Condens. Matter Mater. Phys. 2002, 65, 214303. (61) Nie, S.; Wang, X.; Park, H.; Clinite, R.; Cao, J. Phys. Rev. Lett. 2006, 96, 025901. (62) Hohlfeld, J.; Wellershoff, S. S.; Güdde, J.; Conrad, U.; Jähnke, V.; Matthias, E. Chem. Phys. 2000, 251, 237−258. (63) Tas, G.; Maris, H. J. Phys. Rev. B: Condens. Matter Mater. Phys. 1994, 49, 15046−15054. (64) Arbouet, A.; Voisin, C.; Christofilos, D.; Langot, P.; Fatti, N. D.; Vallée, F.; Lermé, J.; Celep, G.; Cottancin, E.; Gaudry, M.; Pellarin, M.; Broyer, M.; Maillard, M.; Pileni, M. P.; Treguer, M. Phys. Rev. Lett. 2003, 90, 177401. (65) Park, S.; Pelton, M.; Liu, M.; Guyot-Sionnest, P.; Scherer, N. F. J. Phys. Chem. C 2007, 111, 116−123. (66) Del Fatti, N.; Bouffanais, R.; Vallée, F.; Flytzanis, C. Phys. Rev. Lett. 1998, 81, 922−925. (67) Sun, C. K.; Vallée, F.; Acioli, L. H.; Ippen, E. P.; Fujimoto, J. G. Phys. Rev. B: Condens. Matter Mater. Phys. 1994, 50, 15337−15348. (68) Pomfret, M. B.; Brown, D. J.; Epshteyn, A.; Purdy, A. P.; Owrutsky, J. C. Chem. Mater. 2008, 20, 5945−5947. (69) Gans, R. Ann. Phys. 1915, 352, 270−284.

(2) Olson, J.; Dominguez-Medina, S.; Hoggard, A.; Wang, L.-Y.; Chang, W.-S.; Link, S. Chem. Soc. Rev. 2015, 44, 40−57. (3) Mayer, K. M.; Hafner, J. H. Chem. Rev. 2011, 111, 3828−3857. (4) Willets, K. A.; Van Duyne, R. P. Annu. Rev. Phys. Chem. 2007, 58, 267−297. (5) Hu, M.; Chen, J.; Li, Z.-Y.; Au, L.; Hartland, G. V.; Li, X.; Marquez, M.; Xia, Y. Chem. Soc. Rev. 2006, 35, 1084−1094. (6) Kamat, P. V. J. Phys. Chem. B 2002, 106, 7729−7744. (7) Eustis, S.; El-Sayed, M. A. Chem. Soc. Rev. 2006, 35, 209−217. (8) Wu, K.; Chen, J.; McBride, J. R.; Lian, T. Science 2015, 349, 632− 635. (9) Atwater, H. A.; Polman, A. Nat. Mater. 2010, 9, 205−213. (10) Thomann, I.; Pinaud, B. A.; Chen, Z.; Clemens, B. M.; Jaramillo, T. F.; Brongersma, M. L. Nano Lett. 2011, 11, 3440−3446. (11) Zheng, B. Y.; Zhao, H.; Manjavacas, A.; McClain, M.; Nordlander, P.; Halas, N. J. Nat. Commun. 2015, 6, 7797. (12) Félidj, N.; Truong, S. L.; Aubard, J.; Lévi, G.; Krenn, J. R.; Hohenau, A.; Leitner, A.; Aussenegg, F. R. J. Chem. Phys. 2004, 120, 7141−7146. (13) Brown, L. V.; Yang, X.; Zhao, K.; Zheng, B. Y.; Nordlander, P.; Halas, N. J. Nano Lett. 2015, 15, 1272−1280. (14) Wurtz, G. A.; Pollard, R.; Hendren, W.; Wiederrecht, G. P.; Gosztola, D. J.; Podolskiy, V. A.; Zayats, A. V. Nat. Nanotechnol. 2011, 6, 107−111. (15) Abb, M.; Albella, P.; Aizpurua, J.; Muskens, O. L. Nano Lett. 2011, 11, 2457−2463. (16) Favero, I.; Karrai, K. Nat. Photonics 2009, 3, 201−205. (17) Kippenberg, T. J.; Vahala, K. J. Science 2008, 321, 1172−1176. (18) Voisin, C.; Christofilos, D.; Loukakos, P. A.; Del Fatti, N.; Vallée, F.; Lermé, J.; Gaudry, M.; Cottancin, E.; Pellarin, M.; Broyer, M. Phys. Rev. B: Condens. Matter Mater. Phys. 2004, 69, 195416. (19) Link, S.; El-Sayed, M. A. Annu. Rev. Phys. Chem. 2003, 54, 331− 366. (20) Hartland, G. V. Chem. Rev. 2011, 111, 3858−3887. (21) Wang, X.; Guillet, Y.; Selvakannan, P. R.; Remita, H.; Palpant, B. J. Phys. Chem. C 2015, 119, 7416−7427. (22) Zijlstra, P.; Tchebotareva, A. L.; Chon, J. W. M.; Gu, M.; Orrit, M. Nano Lett. 2008, 8, 3493−3497. (23) Soavi, G.; Tempra, I.; Pantano, M. F.; Cattoni, A.; Collin, S.; Biagioni, P.; Pugno, N. M.; Cerullo, G. ACS Nano 2016, 10, 2251− 2258. (24) Hartland, G. V. J. Chem. Phys. 2002, 116, 8048−8055. (25) Hodak, J. H.; Henglein, A.; Hartland, G. V. J. Chem. Phys. 1999, 111, 8613−8621. (26) Ruijgrok, P. V.; Zijlstra, P.; Tchebotareva, A. L.; Orrit, M. Nano Lett. 2012, 12, 1063−1069. (27) Staleva, H.; Hartland, G. V. Adv. Funct. Mater. 2008, 18, 3809− 3817. (28) Burgin, J.; Langot, P.; Del Fatti, N.; Vallée, F.; Huang, W.; ElSayed, M. A. J. Phys. Chem. C 2008, 112, 11231−11235. (29) Mahmoud, M. A.; O’Neil, D.; El-Sayed, M. A. Nano Lett. 2014, 14, 743−748. (30) Dowgiallo, A.-M.; Schwartzberg, A. M.; Knappenberger, K. L. Nano Lett. 2011, 11, 3258−3262. (31) Guillet, Y.; Audoin, B.; Ferrié, M.; Ravaine, S. Phys. Rev. B: Condens. Matter Mater. Phys. 2012, 86, 035456. (32) Hu, M.; Wang, X.; Hartland, G. V.; Mulvaney, P.; Juste, J. P.; Sader, J. J. Am. Chem. Soc. 2003, 125, 14925−14933. (33) Huang, W.; Qian, W.; El-Sayed, M. A. Nano Lett. 2004, 4, 1741−1747. (34) Jais, P. M.; Murray, D. B.; Merlin, R.; Bragas, A. V. Nano Lett. 2011, 11, 3685−3689. (35) Kelf, T. A.; Tanaka, Y.; Matsuda, O.; Larsson, E. M.; Sutherland, D. S.; Wright, O. B. Nano Lett. 2011, 11, 3893−3898. (36) Newhouse, R. J.; Wang, H.; Hensel, J. K.; Wheeler, D. A.; Zou, S.; Zhang, J. Z. J. Phys. Chem. Lett. 2011, 2, 228−235. (37) Pelton, M.; Sader, J. E.; Burgin, J.; Liu, M.; Guyot-Sionnest, P.; Gosztola, D. Nat. Nanotechnol. 2009, 4, 492−495. 2582


Letter

Nano Letters (70) Hao, F.; Larsson, E. M.; Ali, T. A.; Sutherland, D. S.; Nordlander, P. Chem. Phys. Lett. 2008, 458, 262−266. (71) Lecarme, O.; Sun, Q.; Ueno, K.; Misawa, H. ACS Photonics 2014, 1, 538−546. (72) Ahmadi, T. S.; Logunov, S. L.; El-Sayed, M. A. J. Phys. Chem. 1996, 100, 8053−8056. (73) Hartland, G. V.; Hodak, J. H.; Martini, I. Phys. Rev. Lett. 1999, 82, 3188−3188. (74) Del Fatti, N.; Vallée, F. Appl. Phys. B: Lasers Opt. 2001, 73, 383−390. (75) Dowgiallo, A.-M.; Knappenberger, K. L. J. Am. Chem. Soc. 2012, 134, 19393−19400. (76) Link, S.; Hathcock, D. J.; Nikoobakht, B.; El-Sayed, M. A. Adv. Mater. 2003, 15, 393−396. (77) Hu, M.; Hartland, G. V. J. Phys. Chem. B 2002, 106, 7029−7033. (78) Chang, W.-S.; Wen, F.; Chakraborty, D.; Su, M.-N.; Zhang, Y.; Shuang, B.; Nordlander, P.; Sader, J. E.; Halas, N. J.; Link, S. Nat. Commun. 2015, 6, 7022. (79) Auerkari, P. Mechanical and physical properties of engineering alumina ceramics; VTT Offsetpaino: Espoo, Finland, 1996. (80) Chinmulgund, M.; Inturi, R. B.; Barnard, J. A. Thin Solid Films 1995, 270, 260−263. (81) Hartland, G. V. Chem. Sci. 2010, 1, 303−309. (82) Hodak, J.; Martini, I.; Hartland, G. V. Chem. Phys. Lett. 1998, 284, 135−141. (83) Owrutsky, J. C.; Pomfret, M. B.; Brown, D. J. J. Phys. Chem. C 2009, 113, 10947−10955. (84) SINGH, N. Int. J. Mod. Phys. B 2010, 24, 1141−1158. (85) Nix, F. C.; MacNair, D. Phys. Rev. 1941, 60, 597−605. (86) Goubet, N.; Yan, C.; Polli, D.; Portalès, H.; Arfaoui, I.; Cerullo, G.; Pileni, M.-P. Nano Lett. 2013, 13, 504−508. (87) Yu, K.; Major, T. A.; Chakraborty, D.; Devadas, M. S.; Sader, J. E.; Hartland, G. V. Nano Lett. 2015, 15, 3964. (88) Lin, K.-H.; Lai, C.-M.; Pan, C.-C.; Chyi, J.-I.; Shi, J.-W.; Sun, S.Z.; Chang, C.-F.; Sun, C.-K. Nat. Nanotechnol. 2007, 2, 704−708. (89) Feng, X. L.; White, C. J.; Hajimiri, A.; Roukes, M. L. Nat. Nanotechnol. 2008, 3, 342−346. (90) Burg, T. P.; Godin, M.; Knudsen, S. M.; Shen, W.; Carlson, G.; Foster, J. S.; Babcock, K.; Manalis, S. R. Nature 2007, 446, 1066−1069.

2583


Optomechanics of Single Aluminum Nanodisks - Nano Letters (ACS

Recommend Documents