Fabrication and Characterization of Homogeneous Surface-Enhanced

Nov 2, 2010 - The fabrication of SERS-active substrates, which offer high enhancement factors as well as spatially homogeneous distribution of the enh...
0 downloads 0 Views 3MB Size
Download PDF
pubs.acs.org/Langmuir © 2010 American Chemical Society

Fabrication and Characterization of Homogeneous Surface-Enhanced Raman Scattering Substrates by Single Pulse UV-Laser Treatment of Gold and Silver Films Konstantin Christou,*,† Inga Knorr,‡ J€urgen Ihlemann,§ Hainer Wackerbarth,† and Volker Beushausen† ‡

† Department of Photonic Sensor Technologies, Laser-Laboratorium G€ ottingen e.V., G€ ottingen, Institut f€ ur Materialphysik, University of G€ ottingen, G€ ottingen, and §Department Nanostructures, Laser-Laboratorium G€ ottingen e.V., G€ ottingen

Received July 29, 2010. Revised Manuscript Received October 1, 2010 The fabrication of SERS-active substrates, which offer high enhancement factors as well as spatially homogeneous distribution of the enhancement, plays an important role in the expansion of surface-enhanced Raman scattering (SERS) spectroscopy to a powerful, quantitative, and noninvasive measurement technique for analytical applications. In this paper, a novel method for the fabrication of SERS-active substrates by laser treatment of 20, 40, and 60 nm thick gold and of 40 nm thick silver films supported on quartz glass is presented. Single 308 nm UV-laser pulses were applied to melt the thin gold and silver films. During the cooling process of the noble metal, particles were formed. The particle size and density were imaged by atomic force microscopy. By varying the fluence, the size of the particles can be controlled. The enhancement factors of the nanostructures were determined by recording self-assembled monolayers of benzenethiol. The intensity of the SERS signal from benzenethiol is correlated to the mean particle size and thus to the fluence. Enhancement factors up to 106 with a high reproducibility were reached. Finally we have analyzed the temperature dependence of the SERS effect by recording the intensity of benzenethiol vibrations from 300 to 120 K. The temperature dependence of the SERS effect is discussed with regard to the metal properties.

1. Introduction Surface-enhanced Raman scattering (SERS) spectroscopy has become a versatile tool in chemical and biochemical research.1-4 This technique provides huge Raman signal enhancement for molecules in close vicinity to nanostructured metal surfaces, and therefore promises to overcome the intrinsic “weakness” of the Raman effect. In general, two major mechanisms are responsible for SERS: First, the electromagnetic enhancement arising from the increase of local electromagnetic field strength induced by plasmon resonances in the vicinity of metallic nansotructures.5 Second, the chemical enhancement due to the adsorption of molecules at the surface resulting in an increased scattering cross-section.6 While enhancement factors up to 12 orders of magnitude were theoretically predicted for the electromagnetic enhancement, the chemical enhancement factor is merely in the order of 102.7 Therefore, the electromagnetic enhancement is considered as the major contributing mechanism of the SERS effect. In contrast to other powerful analytical techniques (e.g., fluorescence) SERS has distinct advantages: there are no limitations for adsorbates and it works label-free.8 Many types of *To whom correspondence should be addressed. E-mail: konstantin. [email protected]. Tel.: þ49 (0)551 503527. Fax: þ49 (0)551 503599.

(1) Banholzer, M. J.; Millstone, J. E.; Qin, L.; Mirkin, C. A. Chem. Soc. Rev. 2008, 37, 885–897. (2) Wackerbarth, H.; Gundrum, L.; Salb, C.; Christou, K.; Vi€ol, W. Appl. Opt. 2010, 49, 4367–4371. (3) Wackerbarth, H.; Salb, C.; Gundrum, L.; Niederkr€uger, M.; Christou, K.; Beushausen, V.; Vi€ol, W. Appl. Opt. 2010, 49, 4362–4366. (4) Wackerbarth, H.; Hildebrandt, P. Chem. Phys. Chem. 2003, 4, 714–724. (5) Ru, E.; Meyer, M.; Blackie, E.; Etchegoin, P. P. G. J. Raman Spectrosc. 2008, 39, 1127–1134. (6) Adrian, F. J. Chem. Phys. 1982, 77, 5302–5314. (7) Kneipp, K.; Kneipp, H.; Itzkan, I.; Dasari, R. R.; Feld, M. S. J. Phys.: Condens. Matter 2003, 14, R597–R624. (8) Moskovits, M. J. Raman Spectrosc. 2005, 36, 485–496.

18564 DOI: 10.1021/la103021g

SERS-active substrates have been described for analytical applications, including chemically roughened electrodes, metal colloids, metal-sol gels and metal-coated substrates.9-12 To show Raman enhancement, these substrates need to fulfill several requirements: Basically, they have to consist of noble metals (preferentially gold or silver) because of their plasmon resonance in the visible spectral range and biocompatibility in the case of biodiagnostics. The highest enhancement factors were observed on nanostructures showing sharp features as well as a strong curvature which provide extremely high electromagnetic field enhancement due to the “lightning-rod” effect and the so-called “hot-spots”. Even single molecule detection was shown.13 Kneipp et al. reported a single molecule detection of a cyanine dye in silver colloidal solution using near-infrared surface-enhanced Raman scattering and has determined an enhancement factor of 1013.14 These enhancement factors are related to a clustering of the silver colloids;an ensemble that is “hot” according to Moskovits.8 So far “SERS-active” colloids were used seldom for quantitative SERS measurements because of the randomly distributed and hardly controllable number of “hot-spots” generated by the agglomeration of the colloids. Therefore periodic nanostructured substrates come into focus, when talking about SERS as a tool for analytical applications. Although most of the mentioned SERSactive substrates show spatially ensemble-averaged enhancement (9) Fleischmann, M.; Hendra, P.; McQuillan, A. Chem. Phys. Lett. 1974, 26, 163–166. (10) Angel, S.; Myrick, M.; Milanovich, F. J. Appl. Spectrosc. 1990, 44, 335–336. (11) Murphy, T.; Schmidt, H.; Kronfeldt, H. Appl. Phys. B: Laser Opt. 1999, 69, 147–150. (12) Szeghalmi, A.; Kaminskyj, S.; R€osch, P.; Popp, J.; Gough, K J. Phys. Chem. B 2007, 111, 12916–12924. (13) Nie, S.; Emory, S. R. Science 1997, 275, 1102–1106. (14) Kneipp, K.; Kneipp, H.; Deinum, G.; Itzkan, I.; Dasari, R. R.; Feld, M. S. J. Appl. Spectrosc. 1998, 52, 175–178.

Published on Web 11/02/2010

Langmuir 2010, 26(23), 18564–18569

Christou et al.

Article

Figure 1. Optical setup for the nonimaging beam homogenizer.

factors of up to six orders of magnitude, they suffer in many cases from poor signal stability and reproducibility. Furthermore, these substrates are associated very often with extensive fabrication steps especially for the fabrication of periodic nanostructures resulting in high manufacturing costs. Additionally they are not suitable as disposables in analytics, which is generally of huge interest. For this reason, the aim of this study is to develop a fabrication technique that overcomes these problems. In this paper a fabrication technique that generates SERSactive substrates by the laser treatment of thin gold and silver films supported on quartz glass substrates will be introduced. With the focus on the development of a novel and economical fabrication method which fulfills the current need for stable, reproducible, and disposable SERS-active substrates in view of qualitative and quantitative analytical applications, an extensive study of the parameter laser wavelength, fluence, and noble metal film layer thickness was done. For this purpose excimer laser single and double pulses with laser wavelengths of 193, 248, 308, and 351 nm, film thicknesses of 10, 20, 40, 60, and 80 nm, and fluences ranging from 100 to 700 mJ/cm2 were investigated. In this study we focus on the most suitable parameter. Besides the characterization of the generated nanostructured surfaces by atomic force microscopy and VIS-NIR reflectance spectroscopy, a calculation of the signal reproducibility as well as of the enhancement factor using SERS measurements of benzenethiol monolayers adsorbed on the generated substrates are reported. The SERS enhancement from 300 to 120 K of the fabricated substrates is determined to get a deeper insight of the parameters of the enhancement mechanism.

2. Experimental Section Fabrication. As a first step to the fabrication of SERS-active substrates, gold and silver films with a thickness of 20, 40, and 60 nm were vaporized (Auto 306, Edwards) on 1 mm thick quartz glass microscope slides (Suprasil 1, Aachener Quarz-Glas Technologie Heinrich GmbH). Afterward, each of these gold films was illuminated by one single laser pulse of an excimer laser (Compex 150, Coherent) emitting light at a wavelength of 308 nm with a pulse length of 20 ns. A nonimaging homogenizer, consisting of a cylindrical lens array and a spherical lens working as a Fourier optic, was used to generate a spatially homogeneous beam profile in the focal plane, a so-called flat-top (Figure 1). The gold films were treated in the focal plane with a square cross-section of approximately 4  4 mm2 generating the SERS-active area. Regarding the illuminated area, fluences between 100 and 350 mJ/cm2 could technically be applied. Atomic Force Microscopy (AFM). The most feasible way to scan large samples topographically is noncontact AFM imaging. For this purpose we used the cross-functional XE-150 AFM from Park Systems with a motorized sample stage. AFM images were recorded with a high resolution AFM tip (high aspect ratio of 7:1) and with scan sizes of 500  500 nm2 and 1 μm2. Dispersive Raman Microscope. SERS measurements for the determination of signal reproducibility and enhancement were acquired by using a Raman microscope built in the LaserLaboratorium G€ ottingen e.V., equipped with a near-infrared (785 nm) high power laser diode (800 mW) and a back illuminated Langmuir 2010, 26(23), 18564–18569

deep depletion DU420-BRDD CCD camera (Andor) coupled to a Spectra-Pro 300i-P spectrometer (Acton Research), especially designed for the near-infrared spectral region. Light from a pigtailed laser diode connected with a collimator (Sch€after and Kirchhoff) was focused onto the sample by guiding it through a custom built optical arrangement consisting of several long pass and bandpass filters (Semrock), mirrors and lenses (Linos) as well as a 40 (N.A. 0.45) objective (Zeiss). Samples were positioned by a motorized xyz-translation stage (Zaber Technologies) with a repeatability of 1 μm and a step size of 0.1 μm. To realize an efficient coupling of light into the spectrometer a cross-sectional area converting optical fiber bundle with randomly oriented fibers was used. On the one side of the fiber bundle the fibers were arranged circularly, whereas on the other side they were arranged linearly, which replaced the entrance slit of the spectrometer directly. With this configuration a spectral resolution of 8 cm-1 and an increase of light throughput by a factor of 12 compared to a conventional slit arrangement could be achieved. SERS measurements were acquired with 50 mW laser power and an acquisition time of 1 s. Transmissive Raman System. For the temperature dependent SERS measurements, a standard system (Kaiser Optical Systems Inc., Ann Arbor, MI) was used; the excitation wavelength of 785 nm (line width 0.06 nm) was provided by a GaAlAs diode-laser (Invictus, Kaiser Optical Systems, Inc.). The excitation and scattered light were guided by a multimode optical fiber equipped with a probe head onto the sample. The power of the laser emission was 100 mW at the probe. The scattered light was coupled by an aperture in the optical fiber and transmitted to the spectrograph (PhAT System, Kaiser Optical Systems, Inc.), which uses holographic transmissive gratings to perform filtering and dispersion functions. The diffracted light was recorded with a CCD camera (iDus, Andor Technology plc.). The spectral resolution was 5 cm-1. All SERS spectra were acquired with 2 s acquisition time. Visible-Near Infrared Reflectance Spectroscopy. Extinction spectra were acquired using the home-built dispersive Raman microscope extended with an additional optical setup guiding white light from a halogen lamp (HL-2000, Avantes) collinearly through the microscope objective onto the substrate. For the measurements the laser light was blocked and reflectance spectra with a wavelength range of 500-1100 nm were recorded. All reflectance spectra were collected against a gold mirror as the reference and converted into extinction. Material. Self-assembled monolayers of benezenethiol 99.9% (Sigma-Aldrich) were generated by soaking the substrates in a 1 mM benzenethiol/ethanol solution for 24 h. This procedure forms a self-assembled monolayer, as benzenethiol forms a stable covalent bond via the sulfur of the thiol group to gold and silver surfaces. After this treatment, the samples were rinsed with ethanol (p.a., Sigma-Aldrich) and left to dry in air for 10 min before the SERS measurement.

3. Results and Discussion Topographic Characterization. Besides great progress in understanding the nature of the SERS effect, novel engineered substrates have also allowed the applications of scientists from multiple disciplines to flourish. Although these applications and substrates are promising advances, a much greater understanding of how to control surface architecture in order to stabilize and maximize the SERS signal is needed.1 Imperfections in substrates and their nanostructures have a significant effect on the SERS signal and thus impair quantifications. Here we focus on the reproducible generation of gold and silver particles by laser pulse treatment of thin gold and silver films. The laser treatment results in a randomly and densely packed arrangement of spherical particles of different sizes. Due to laser light absorption, the thin metal films are melted and partly ablated. The ablation process DOI: 10.1021/la103021g

18565

Article

Christou et al.

Figure 2. AFM image of a 20 nm gold film before (left) and after treatment with a single 308 nm laser pulse and a fluence of 227 mJ/cm2 (right).

can be observed due to the deposition of the used metals on the substrate mount of the experimental setup. During the cooling phase of this thermal annealing process the surface tension leads to the formation of spherical metal particles with sizes in the nanoscale range and interparticular gaps that result in high SERS activity.7,15 In Figure 2 AFM images of an untreated and a nanostructured gold surface fabricated with a fluence of 227 mJ/ cm2 are presented. Before the treatment the AFM image shows gold islands on top of the 40 nm thick gold film. These islands with a maximum height of 4 nm are typical for the deposition of gold on substrates. According to the AFM image after the laser treatment the melting and cooling process leads to the formation of spherical particles much larger than the islands. To get a better understanding of the fabrication process both the fluence and the gold film layer thickness were varied to verify the effect on the generated particle sizes. The AFM provides a true three-dimensional surface profile of the generated substrates, so the particle height can be quantified. Figure 3 shows the dependence of the mean particle height measured across an area of 1 μm2 for different fluences and film thicknesses. Especially for the 40 nm gold layer the mean particle height strongly depends on the laser fluence. By increasing the fluence, the particle height also increases and reaches a maximum at 60 nm. In addition, Figure 3 shows also that an increase in thickness of the gold film tends to result in particles with increasing sizes because the mean particle height of the tested gold layer with 20 nm film thickness is slightly smaller compared to the generated one with films of 40 nm thickness for the same fluence. For fluences smaller than 200 mJ/cm2, the AFM images of the 20 nm gold film are distorted by a few particles showing sizes much bigger than 100 nm, so that the smaller ones could not be resolved by the used AFM tip. These results show clearly that the generated particle sizes can be controlled by the used gold film thickness and fluence. SERS Measurements. Self-assembled monolayers of pure and functionalized alkylthiols on metals, especially on gold, have been particularly extensively studied.16-18 These systems have facilitated studies at the molecular level and added comprehensive fundamental insight to this field. Here we use this approach to generate a monolayer of benzenethiol to determine the SERS activity of the generated substrates (Figure 4a)). In Figure 4b a comparison of a SERS spectrum recorded for benzenethiol on a nanostructured gold surface and a Raman spectrum of liquid benzenethiol is presented. The SERS spectrum differs from the Raman spectrum of the liquid benzenethiol in terms of the band position and relative intensities that is related to the surface (15) Xu, H.; Aizpuua, J.; K€all, M.; Apell, P. Phys. Rev. E 2000, 62, 4318–4324. (16) Schreiber, F. Prog. Surf. Sci. 2000, 65, 151–256. (17) Wackerbarth, H.; Tofteng, A. P. B.; Jensen, K. J.; Chorkendroff, I.; Ulstrup, J. Langmuir 2006, 22, 6661–6667. (18) Wackerbarth, H.; Grubb, M.; Hansen, A. G.; Zhang, J.; Ulstrup, J. Langmuir 2004, 20, 1647–1655.

18566 DOI: 10.1021/la103021g

Figure 3. Relationship between mean particle height and laser fluence for 20 and 40 nm thick gold films.

enhancement mechanism and molecular orientation. The ring stretching vibration at 1580 cm-1 and the vibrations of benzenethiol at 1025 and 1074 cm-1 are significantly enhanced in the SERS spectrum according to the fact that normal modes of the adsorbed molecules, which are involved in changes in molecular polarizibility with a component perpendicular to the surface, show the greatest enhancement.19 The typical SH-bands, the deformation vibration at 918 cm-1 and the stretching vibration at 2575 cm-1 (not shown) of benzenethiol disappear upon binding of the sulfur to the surface, indicating the formation of a monolayer at the surface. SERS measurements of benzenethiol adsorbed on the untreated gold film (Figure 2) indicate that the surface roughness, caused by the surface evaporation process, can be neclected as a reason for the SERS effect because no SERS signals could be measured under the same conditions as described in the experimental part of this study. To find out if there is any relation between the SERS activity of the fabricated nanostructures and the parameter layer thickness and fluence, SERS measurements of benezenethiol from five different equally spaced places across the whole nanostructured area were performed. The averaged SERS intensity of the strongest Raman band at 1074 cm-1 is plotted in Figure 5 against the used fluence for the gold film thicknesses of 20, 40, and 60 nm. The intensity of the SERS signal varies significantly with the fluence and layer thickness. The error bars represent the standard deviation of the measurements. Standard deviations of 5% were found in the Raman band intensity at 1074 cm-1 for each fluence, promising that the fabricated SERS active substrates are suitable for quantitative measurements. As expected, the SERS intensity also varies with the film thickness and inevitably with the generated particle sizes, indicating that the precise morphology of the nanostructured gold film is an important factor. Comparing the mean particle height in Figure 3 with the SERS intensity shown in Figure 5, a trend between the SERS intensity and the mean particle height can be observed. By increasing the fluence the particle height as well as the SERS intensity increases nonlinearly, whereas a film thickness of 40 nm shows the highest signal intensity. According to Figure 5 the laser treatment of a 40 nm gold film results in the most suitable particle height distributions in view of signal reproducibility and enhancement factor. This trend is not fully understood so that further investigations are necessary to point out the complex relation between the parameter layer thickness and fluence related to the SERS activity of the fabricated nanostructures. (19) Xue, G.; Ma, M.; Zhang, J.; Lu, Y. Y.; Carron, K. T. J. Colloid Interface Sci. 1992, 150, 1–6.

Langmuir 2010, 26(23), 18564–18569

Christou et al.

Article

Figure 4. (a) Illustration of a self-assembled monolayer of benzenethiol on a nanostructured surface. (b) (Top) Raman spectrum of liquid benzenethiol measured in a cuvette and (bottom) SERS spectrum of benzenethiol adsorbed on a nanostructured gold surface.

nethiol Raman band at 1074 cm-1 to the corresponding band measured from liquid benzenethiol.21 The normal Raman spectrum was measured in a quartz glass cuvette (Starna) with a path length of 100 μm filled with benzenethiol under the same collection conditions, a laser power of 50 mW, and an acquisition time of 1 s. The SERS enhancement factor Ω is then given by Ω ¼

Figure 5. Relation between SERS intensity of the Raman band at 1074 cm-1 and the fluence for the nanostructured 20, 40, and 60 nm thick gold films.

For this reason we used extinction spectroscopy measured in reflectance geometry in the vis-NIR spectral region. The extinction maximum corresponds to the maximum of the surfaceplasmon resonance. Figure 6 shows extinction spectra of the nanostructured 40 nm gold film for the extremal values of the signal intensity (fluences of 107 and 350 mJ/cm2). The extinction spectra show that the extinction maximum is red-shifted from 683 to 695 nm toward the excitation wavelength of 785 nm for the 350 mJ/cm2 fluence and thus for the largest intensity of the benzenethiol Raman band. The red-shift indicates that the particle sizes increase with an increasing fluence according to the Mie theory. According to Haynes et al. the maximum of the SERS signal is reached when the maximum of the surface-plasmon resonance wavelength is a value between the excitation wavelength and the vibrational band of interest.20 This agrees well with our observation. This result shows that the SERS enhancement can be tuned by the presented fabrication method. SERS-Enhancement Factor. The SERS-enhancement factors were determined by using a method proposed by McFarland et al. by comparing the SERS intensity of the strongest benze(20) Haynes, C. L.; Yonzon, C. R.; Zhang, X.; van Duyne, R. P. J. Raman Spectrosc. 2005, 36, 471–484. (21) McFarland, A.; Young, M.; Dieringer, J.; van Duyne, R. P. J. Phys. Chem. 2005, 109, 11279–11285.

Langmuir 2010, 26(23), 18564–18569

NF IS 1 NS IF G

ð1Þ

where NF and NS are the number of molecules probed in the liquid sample and on the SERS active substrates, respectively, and IF and IS are the corresponding normal Raman and SERS intensities. The factor G describes the increase in the surface due to the treatment in comparison to a flat surface. In a first approximation, a densely packed layer of halfspheres with the radius R was assumed for the surface of the generated substrates, so that G = 2πR2/(2R)2 = π/2. The laser spot diameter of 220 μm was determined on the basis of the knife-edge method.22 Assuming a number of benzenethiol molecules on a flat surface of 0.709 nmol/cm2 and a mean particle height of 60 nm in the case of the nanostructured 40 nm thick gold film for a fluence of 350 mJ/cm2, a number of molecules of NS = 1.6  1011 is probed on the substrate within the laser spot.23 The number of molecules NF = 2.1  1016 probed for the normal Raman experiment results from a probe volume of 3.6 nL approximating the probe volume as a cylinder with a diameter of 220 μm and a height of 100 μm corresponding to the path length of the cuvette. Substituting the measured values of IS and IF into eq 1, an enhancement factor of Ω ≈ 4  106 is obtained. Signal Reproducibility. To determine the spatial and batchto-batch signal reproducibility of the fabricated substrates, SERS mappings were done. Therefore, the substrates were scanned by the dispersive Raman microscope in a region of interest of 5  5 mm2 centered in the middle of the nanostructured area with a pitch of 200 μm. Figure 7 shows the SERS mapping of a substrate which was fabricated with a gold film layer of 40 nm and a fluence of 227 mJ/cm2. The color of the nanostructured surface is violet. This can be very well explained by the Mie theory after that the particle size correlates with the color of the scattered light. For the determination of signal reproducibility a standard deviation of 5% from the mean of the Raman band at 1074 cm-1 for the (22) Khosrofian, J.; Garetz, B. Appl. Opt. 1983, 22, 3406–3410. (23) Wan, L.-J.; Terashima, M.; Noda, H.; Osawa, M. J. Phys. Chem. B 2000, 104, 3563–3569.

DOI: 10.1021/la103021g

18567

Article

Christou et al.

Figure 6. (Left) Extinction curves of 40 nm gold films treated with fluences of 107 and 350 mJ/cm2. (Right) Polynomial fit of the extinction maxima.

Figure 7. (Left) Image of a nanostructured gold film next to an untreated gold film. (Right) SERS mapping for the standardized intensity of the strongest Raman band at 1074 cm-1.

highlighted area was calculated. Additionally it could be observed that some data points differ from the mean significantly, in some cases over 20% from point-to-point. Two more substrates from different batches were scanned the same way; signal reproducibility of 5% standard deviation of the mean were calculated (batch-to-batch variation) showing that the presented fabrication method has great potential for generating low-cost SERS active substrates which can be used for quantitative measurements. Temperature Dependence. Temperature dependence of SERS intensity has not been studied often so far, if mainly to identify the specific features of the surface contributing to SERS. Pang et al. studied the temperature dependence of the SERS intensity of 1-propanethiol adsorbed on silver islands in ultrahigh-vacuum.24 Owing to the washing procedure and the covalent sulfur-silver or sulfur-gold bond, the formation of a temperature independent monolayer can be assumed in our studies. We have not observed a decay of the SERS intensity by increasing acquisition time. The excitation wavelength (785 nm) and power thus are obviously not sufficient to cause a significant change of the surface structure. As we have performed the experiments in vacuum (0.1 Pa), changes caused by contaminants can be neglected. We have recorded SERS spectra of benzenethiol attached to gold and silver particle surfaces between 120 and 300 K. We have nanostructured a 40 nm silver film on quartz glass with a 308 nm single laser pulse, a fluence of 350 mJ/cm2, and additionally an (24) Pang, Y. S.; Hwang, H. J.; Kim, M. S. J. Phys. Chem. 1998, 102, 7203–7209.

18568 DOI: 10.1021/la103021g

Figure 8. AFM images of a 40 nm silver film before (left) and after the treatment with a single 308 nm laser pulse and a fluence of 350 mJ/cm2 (right).

AFM image before and after the laser treatment was acquired (Figure 8). The untreated silver film shows nearly the same island structures with nearly the same heights compared to the islands of the untreated gold film. For the untreated silver film no SERS signal of the benzenethiol could be measured under the same conditions as mentioned in the experimental part. The laser treatment of the silver film resulted in particles with a similar shape as the gold particles. Compared to the mean gold particle height of 60 nm fabricated with a fluence of 350 mJ/cm2 and a film thickness of 40 nm the silver particles show a smaller mean particle height of 45 nm for the same experimental conditions. The observed difference in particle size can be explained by the difference in the melting point of the used metals and by the interfacial metal-substrate interaction during the cooling phase Langmuir 2010, 26(23), 18564–18569

Christou et al.

Article

Figure 9. (Left) Temperature dependence of a nanostructured gold film. (Right) Temperature dependence of a nanostructured silver film. The solid line represents the Raman band at 1074 cm-1 and the dotted line represents the Raman band at 1580 cm-1. The inset shows a time and temperature depending SERS measurement of benzenethiol.

according to Henley et al.25 The intensities of the vibration modes at 1074 and 1580 cm-1 depending on the temperature are shown in Figure 9. The intensities of the benzenethiol modes at the silver particle surface show an increase of 2.5 by decreasing the temperature from 300 to 120 K. This is in line with studies from Pang et al., who reported an increase of the intensity by a factor of 3 of 1-propanethiol adsorbed on silver islands between 300 and 15 K.24 The inset in Figure 9 shows the intensity of the SERS signal versus the time of the cooling cycle. By going back to 300 K, the original SERS intensity of the benzenethiol is reached, indicating that the temperature dependence of the SERS effect is reversible. To discuss the temperature dependence of the SERS signals we consider the theoretical notion of the electromagnetic enhancement mechanism. In the theoretical approach of Kerker et al. the enhancement Ω is calculated for molecules adsorbed on spherical particles according to eq 2:26 ΩAbs ðω, ωS Þ ¼

jεM ðωÞ - εD j2 jεM ðωS Þ - εD j2 jεM ðωÞ þ 2εD j2 jεM ðωS Þ þ 2εD j2



r rþd

12 ð2Þ

In eq 2 the excitation frequency is given by ω, the Stokes-shifted frequency by ωS, the particle radius by r, the distance of the target molecule to the particle surface by d, the dielectric function of the surrounding medium by εD, and the dielectric function of the noble metal by εM. According to Chiang et al. the main dependence of the signal enhancement Ω on the substrate temperature T enters through the dielectric function εM of the noble metal. The optical properties of the metal can be characterized by a temperature-dependent Drude model (eq 3):27 εM ðTÞ ¼ 1 -

ωp 2 ðTÞ ωðω þ iωc ðTÞÞ

ð3Þ

In contrast, the benzenethiol modes of the gold particle surface do not show an increase of the intensity by decreasing the temperature. The particle shape and size (rAu = 60 nm; rAg = 45 nm) are pretty similar in our study, so that we assume that the differences are related to the temperature dependence of the dielectric properties of the noble metals. Further studies at this fundamental topic should provide a better understanding of the surface-enhanced Raman effect.

4. Conclusion A novel method for the fabrication of SERS-active substrates based on the treatment of thin gold and silver films supported on quartz glass with single UV-laser pulses is presented. The substrates show high enhancement factors in the order of 106 as well as high signal reproducibility of 5% standard deviation measured by probing self-assembled monolayers of benzenethiol. The fabricated nanostructures can be controlled by varying the noble metal layer thickness and the laser fluence offering a way to customize SERS active substrates. In addition to the SERS measurements a topographic characterization of the fabricated gold and silver particles was performed by atomic force microscopy. Furthermore vis-NIR reflectance spectra of the gold particles were acquired showing a red-shift of the plasmon resonance with increasing particle size resulting in an increase of the SERS intensity. The temperature dependent SERS measurements on gold and silver particles of benzenethiol show significant differences in the temperatue dependence. While the signal intensity is independent of the gold particles, the intensity increases upon cooling the silver particles. Because of the fast fabrication method and the high signal reproducibility, these substrates have a great potential as disposables for quantitative analytical applications in surface-enhanced Raman spectroscopy.

Here, the bulk plasmon frequency is given by ωp, the collision frequency by ωc, and the excitation frequency by ω. Thus the temperature dependence of the SERS signals can be explained by the electromagnetic mechanism.

Acknowledgment. Funding for this research was provided by the German Federal Ministry of Education and Research (Grant No. 01RI0643A).

(25) Henley, S. J.; Carey, J. D.; Silva, S. R. P. Phys. Rev. B 2005, 72, 195408-1– 195408-10. (26) Kerker, M.; Siiman, O.; Wang, D.-S. J. Phys. Chem. 1984, 88, 3168–3170. (27) Chiang, H.-P.; Leung, P. T.; Tse, W. S. J. Phys. Chem. 2000, 104, 2348– 2350.

Supporting Information Available: AFM image of 20 nm gold films after treatment with a single 308 nm laser pulse and a fluence of 227, 277, and 308 mJ/cm2. This material is available free of charge via the Internet at http://pubs.acs.org.

Langmuir 2010, 26(23), 18564–18569

DOI: 10.1021/la103021g

18569