Surface-enhanced Raman scattering from oxazine 720 adsorbed on scratched gold films

Transcription

1 JOURNL OF RMN SPECTROSCOPY J. Raman Spectrosc. 2005; 36: Published online in Wiley InterScience ( DOI: /jrs.1337 Surface-enhanced Raman scattering from oxazine 720 adsorbed on scratched gold films lexandre G. rolo and Christopher J. ddison Department of Chemistry, University of Victoria, P.O. ox 3065, Victoria, ritish Columbia, V8W 3V6, Canada Received 2 October 2004; ccepted 3 January 2005 Surface-enhanced Raman scattering (SERS) and surface-enhanced resonance Raman scattering (SERRS) from oxazine 720 (oxa) dye adsorbed on scratched gold films are reported. The SERS-active surface was prepared by performing a series of scratches in a 100 nm thick gold film deposited in glass. tomic force microscopic imaging revealed a sub-structure within the scratches containing a set of parallel gold wires of different sizes and shapes. The 1-D order imposed by the parallelism between these wires is responsible for an interesting polarization effect observed in forward scattering experiments. It is shown that the maximum enhanced signal is observed when the polarization of the incident field is perpendicular to the direction of the scratches. This polarization discrimination may be useful in the design of SERS applications in chemical sensing and optical switching. Moreover, we also show that these scratched gold surfaces can be used as ordinary SERS substrates for experiments in backscattering using a common Raman microscope in non-resonance conditions with the excitation energy. This was accomplished by obtaining the electrochemical SERS of oxa in situ (under electrochemical control). The potential dependence of the SERS from oxa adsorbed on scratched u is compared with previous results obtained with g electrodes. Copyright 2005 John Wiley & Sons, Ltd. KEYWORDS: surface-enhanced Raman scattering; surface plasmons; nanowires; oxazine 720; gold nanostructures INTRODUCTION Metallic nanostructures support direct excitation of localized surface plasmon (LSP) modes. 1 The LSP modes are accessible in the visible range for gold nanoparticles. The exceptional optical properties of these structures and their potential applications in biochemical sensing justify the large amount of work that is being developed in this field. 2 5 particularly interesting aspect is the fact that the excitation of LSP modes translates into a large increase in the local electromagnetic (EM) field. Consequently, an increase in the spectroscopic response is expected from molecules adsorbed in these structures since they are influenced by the enhanced EM field. 6 Surface-enhanced Raman scattering (SERS) is the most important of the surface plasmons (SP) mediated enhanced spectroscopic methods. SERS (in certain conditions) may Ł Correspondence to: lexandre G. rolo, Department of Chemistry, University of Victoria, P.O. ox 3065, Victoria, ritish Columbia, V8W 3V6, Canada. agbrolo@uvic.ca Contract/grant sponsor: University of Victoria. Contract/grant sponsor: Natural Sciences and Engineering Research Council of Canada (NSERC). Contract/grant sponsor: Canada Foundation for Innovation. Contract/grant sponsor: ritish Columbia Knowledge and Development Fund (CKDF). rival the most sensitive of the spectroscopic probes, such as fluorescence spectroscopy, with detection limits in the single-molecule realm. 7 In contrast to fluorescence, however, the bandwidth of the features in SERS is much narrower because they originate from vibrational transitions rather than electronic in fluorescence. Despite this important advantage, SERS is still not a widely used technique in sensing and biomedical analysis. The main challenges that need to be overcome to turn SERS into a routine analytical tool are related to the reproducibility of the substrate. SERS has been observed from a variety of substrates from random structures, such as colloids 8 and rough electrode surfaces, 9 to organized architectures, such as gratings, 10,11 which are created using modern fabrication tools. The organized substrates offer the advantage of better control over the location of the enhanced field, reproducibility and polarization effects. The random substrates present a fractal symmetry which optimizes the field localization 12 leading to a much larger enhancement factor, but show lower reproducibility. The development of substrates for SERS is then a very active field, and it is the subject of this work. Here we report a SERS substrate which is fabricated simply by creating a parallel scratch on gold films. 13 substructure in the nanometric range is observed Copyright 2005 John Wiley & Sons, Ltd.

2 630. G. rolo and C. J. ddison within the scratch, and surface-enhanced (resonance) Raman scattering [SE(R)S] from oxazine 720 (oxa) dye adsorbed on these substructures are easily observed. Moreover, the parallelism of the scratches introduced a 1-D order in this system; therefore, this substrate is an intermediate between a completely random system (gold colloids dispersed in glass) and the 2-D organized systems (gratings on gold, for instance). This 1-D order is responsible for a striking polarization dependence on the SERRS signal obtained in forward scattering geometry. The magnitude of the SERRS intensity is related to the direction of the scratch relative to the polarization of the incident beam. We will also show the generality of this substrate by obtaining SERS from oxa using backscattering geometry in an electrochemical environment. One of the main promises for application of SERS in analytical chemistry is in DN analysis The polarization dependence reported here should be suitable to introduce another level of control on the analysis of mixtures of genetic materials from different organisms. Therefore, this effect can be explored for the creation of biochips for multiplex analysis. Moreover, the fabrication of the scratched substrate studied here does not require lithography or electron/ion milling. Hence they are significantly cheaper than organized nanopatterned gratings that could produce similar polarization effects. The simple fabrication procedure of the scratched substrates could also allow their production just before the measurements, and therefore they may be significantly less prone to surface contaminations. set of parallel lines can clearly be seen within the ca 10 µm scratch in Fig. 1(). The characteristic length and distribution of the features can be better visualized by the line scan presented in Fig. 1(). Figure 1 indicates that a random set of parallel lines and grooves were formed as consequence of the scratching procedure. The thickness of the gold layer varies within the scratch and it is smaller in some regions than the 100 nm of the original film. The illumination of this sub-structure from the glass side would allow an evanescent field of incident light to tunnel through the gold film and excite the LSP modes supported by the nanostructures. The scratched surface was then modified by a droplet evaporation procedure. 13,17 In this case, a drop of 10 µm solution of oxa (from Lambdachrome) in HPLC-grade methanol (ldrich) was added to the surface. Oxa is a common laser dye, with the main absorption band in the visible region around 620 nm. The solvent was allowed to evaporate, after which the surface was rinsed with a copious amount of ultra-pure water (18.2M cm, from a arnstead NNOpure Diamond water purification system). This procedure ensures that at most one monolayer of the adsorbate was present at the surface during the Raman measurements. Figure 2 shows the Raman instrumentation used in the forward scattering measurements. This experimental geometry guarantees that the amount of excitation photons at the gold air interface will be directly related to the evanescent EXPERIMENTL The substrate (from EMF) was 100 nm thick gold films depositedona5nmchromiumlayerona1ð 1inchglass slides. The gold film was mechanically scratched according to the following procedure. 10 µm metallic tip was gently pressed against the gold surface and dragged parallel to the sides. The complete removal of the gold from the surface was avoided, since the main idea was to create a set of randomly distributed parallel gold lines. n atomic force microscopic (FM) picture of the scratched substrate is shown in Fig. 1. The FM images were obtained using a Thermomicroscope Explorer SPM. ll scans were run in noncontact mode at 200 µms 1. The scan range and resolution were typically 100 µm and 200 lines per scan, respectively. Figure 1. tomic force microscope image of a scratched gold surface. () 83.5 ð 83.5 µm overall image, showing one scratch at the left and another one at the far right; () line scan over the scratch at the left of (). The sub-structure within scratch contains parallel lines and grooves of different sizes. abinet-soleil compensator Glass Notch filter Laser light at nm Objective lens scratches u Objective lens focusing lens Raman scattering Spectrograph and CCD Figure 2. Experimental setup used in the forward scattering measurements.

3 SERS from oxazine 720 adsorbed on scratched gold films 631 field and the surface plasmon excitation. 35 mw He Ne laser (from Melles Griot) was used as excitation source; the wavelength of excitation was nm. The laser was directed through a polarizer to ensure the polarization purity. Then the polarized beam was passed through an adjustable abinet Soleil compensator. The compensator is basically a quartz crystal of variable thickness that works as a variable wave plate. 18 The polarization state of the output beam can then be linear, elliptically or circularly polarized depending on the settings of the compensator. The laser light reached the sample from the glass side of the substrate and it was focused through the scratches containing the nanostructures using a 10ð Olympus microscope objective (numerical aperture D 0.25). The transmitted light (containing the Raman information) was collected using a similar microscope lens. The fundamental laser light was rejected using a Kaiser super-notch filter and the remaining radiation was directed through a Kaiser Holospec f /1.4 spectrograph coupled with an ndor DV-401-V CCD detector. The substrates and the optical components were mounted in mechanical stages that allow fine positioning control, and a wide-view microscope was used to aid the alignment of the incoming laser light with the scratches. The spectroelectrochemical measurements were performed by adapting the scratched slide in a custom-made Teflon cell containing a Pt ring and a Pt wire (from lfa esar), which were used as auxiliary and pseudo-reference electrodes, respectively. The electrolyte solution was 0.1 M KCl (from ldrich). The potential was controlled using a PR 173 potentiostat/galvanostat system and a Hokuto Denko H-111 function generator. The in situ (under electrochemical control) SERS measurements was obtained in backscattering mode using a Renishaw invia microscope Raman spectrometer system equipped with a Spectra-Physics argon ion laser to produce excitation at 782 nm. RESULTS ND DISCUSSION Polarization-dependent SERRS from scratched u ultra-thin films strong polarization dependence has been reported for the transmission of light through several sub-wavelength metallic nanostructures, such as slits, 19 gratings of nanowires 20 and gratings of elliptical nanoholes in thin films. 21 In all these cases, the polarization dependence is related to the differences in the conditions for electron confinement between longitudinal and the transverse directions in elongated nanostructures. n equivalent effect should be expected in the Raman scattering from oxa adsorbed on the nanostructures presented in Fig 1. In fact, Fig. 3 shows two SERRS spectra of oxa adsorbed on a scratch obtained at orthogonal polarization directions (using the experimental setup described in Fig. 2). The position of the scratch and the relative orientation of the polarization state of the incident beam are depicted in the inset scratches Wavenumber /cm -1 u Figure 3. SERRS spectra of oxa adsorbed on a scratched gold surface at two orthogonal incident polarizations, using the setup presented in Fig. 2. () The polarization of the incident light (represented by an arrow) is perpendicular to the direction of the scratches (see inset); () the polarization of the incident light is parallel to the scratches (see inset). The wavenumbers of the main vibrations observed in the SERRS spectra are indicated in Fig. 3, and they can all be assigned to the movements of the phenoxazine ring of the dye. 17,22 The spectra shown are similar to those observed from the oxa adsorbed on roughened silver surfaces under electrochemical conditions. 17 Since oxa has an electronic absorption band around 620 nm, the Raman spectra in Fig. 3 contain contributions from charge-transfer processes in a resonance Raman-like effect (SERRS effect). 23 The dependence of the SERRS intensities on the polarization state of the incident beam is also evident in Fig. 3. The SERRS spectrum shown in Fig. 3() was obtained for incident polarization perpendicular to the scratch (see inset) and the spectrum shown in Fig. 3() is for polarization parallel to the scratch. It is clear that more efficient scattering was observed in Fig. 3(). This result indicates that the random set of parallel structures within the scratch (as shown in Fig. 1) is acting as a series of nanowires. The perpendicularly polarized light can efficiently excite the LSP modes in the transverse direction of the wires, maximizing the SERRS efficiency. 13,20 In order to confirm this polarization effect, three scratches in different directions were created in the same gold film, as depicted in Fig. 4() (scratches a, b and c). The incident laser light was initially polarized along the y-axis [as also indicated in Fig. 4()] before interacting with the abinet Soleil compensator. The compensator was then adjusted for different values of phase retardation and the SERRS signal was measured. The state of the polarization of the incident light is not changed when the compensator is adjusted at 0 and 360 of retardation. The compensator is a quarter-wave plate (output is circularly polarized) at 90 and 270 of retardation. t 180 of retardation the output is linear polarized, but rotated 90 from the original polarization

4 632. G. rolo and C. J. ddison C z x a b c u y Retardation / degrees D Retardation / degrees Retardation / degrees Figure 4. () Representation of three scratches (a, b and c) at different directions in an u thin film. The incident light is originally polarized along the y-direction before going through the compensator. () SERRS intensity of the ca 585 cm 1 mode of oxa 720 adsorbed on scratch (a) against the retardation phase of the compensator. The triangles represent the experimental data points and the line is the best fit. (C) SERRS intensity of the ca 585 cm 1 mode of oxa 720 adsorbed on scratch (b) against the retardation phase of the compensator. The diamonds represent the experimental data points and the line is the best fit. (D) SERRS intensity of the ca 585 cm 1 -mode of oxa 720 adsorbed on scratch (c) against the retardation phase of the compensator. The circles represent the experimental data points and the line is the best fit. (half-wave plate). t all the other intermediate retardations the output of the compensator is elliptically polarized. Notice that in these experiments (with three scratches of different directions in a gold film) we ensured that all optical conditions were the same for all scratches; therefore, the polarization effects could not be related to optical artifacts. Plots of the SERRS intensity of the 585 cm 1 band against the retardation is shown for all three scratch directions [a, b and c in Fig. 4()] in Fig. 4() (D). The pattern of the plots in Fig. 4() (D) is different for different scratches direction. ccording to Fig. 3, it is expected that the sub-structured scratch should behave as a linear polarizer, producing a maximum SERRS signal when the polarization of the incident light is perpendicular to direction of the scratch and a minimum signal when they are parallel. Therefore, the expected trend of the plots in Fig. 4() (D) should then be equivalent to the output intensity of an optical system formed by a polarized input, a abinet Soleil compensator and a linear polarizer in the output adjusted at three different angles relative to the incident polarization. The output intensity of this simple optical equivalent is given by 18 I / [1 sin υ ] 1 where υ is the retardation phase and is the angle between the incident polarization [before the variable wave-plate and aligned along the y-axis, as shown in Fig. 4()] and the direction of the scratches. In fact, the solid line in Fig. 4() (D) corresponds to the calculated best fit of the function given by Eqn (1). It is shown in Fig. 4() (D) that this model agrees very well with the experimental data for all scratch orientations. The fit was performed using as a free variable. small discrepancy was observed between the expected values of (90, 0 and 45 for scratches a, b and c, respectively) and the values obtained from the best fit [108, 18 and 54, from Fig. 4(), (C) and (D), respectively]. These differences can in principle be attributed to small tilts in the position of the optical elements, including the direction of the scratches relative to the edge of the slides. 13 In any event, the experiments shown in Fig. 4 confirm that the interesting polarization behaviour reported here is indeed related to the direction of the nanostructures and, consequently, to the preferential excitation of transverse SP modes. Spectroelectrochemical measurements The scratched gold surface was also incorporated in a spectroelectrochemical cell for experiments under electrochemi-

5 SERS from oxazine 720 adsorbed on scratched gold films 633 cal control. The main objective here was to show that the scratched u surface is also SERS active in a backscattering arrangement and it can be used as an ordinary SERS substrate. The in situ SERS spectrum of oxa adsorbed on u recorded at several different applied potentials is shown in Fig. 5. These experiments were performed using 782 nm excitation, and therefore away from the internal resonances of oxa (absorption band around 620 nm). ttempts to obtain SERS from oxa adsorbed on the smooth surface were unsuccessful. However, intense Raman signals (as presented in Fig. 5) were easily obtained by repositioning the incident laser beam to allow direct illumination of the scratch. This further confirms that the SERS reported in this work indeed originates from the nanostructures in the scratch and cannot be explained by a simple resonance Raman effect. Reversible electrochemical behaviour has been reported for oxa and similar molecules containing the phenoxazine ring adsorbed on electrode surfaces. 24 These molecules can also be used as mediators in redox processes involving biological molecules and in electrocatalysis. 25 We have recently reported an investigation on the electrochemistry of oxa adsorbed on g electrodes. 17 combination of cyclic voltammetric measurements, surface-enhanced Raman spectroscopy at different excitation energies and density functional theory (DFT) calculations was used to elucidate the electrochemical behaviour and the charge-transfer contributions to the SER(R)S mechanism. 17 That work demonstrated that oxa adsorbs on g in an upright orientation and that the molecule remains adsorbed in the SERS-active sites even after the reduction of the ring (which occurs at 500 mv vs gjgcljcl sat). 17,22 The spectral features shown in Fig. 5 are highly dependent on the applied potential and show several similarities to our previous report on g, although the presence of a potential-dependent shoulder at 568 cm 1, indicated with an arrow in Fig. 5, is the most notable exception. New spectral features (not observed previously for g) are also presented in Fig. 5, trace f (at 900 mv). These bands, at 655 and 791 cm 1, may be attributed to the reduced oxa species Wavenumber / cm -1 Figure 5. SERRS spectra of oxa at different applied potentials: (a) C500; (b) C200; (c) 100; (d) 400; (e) 700; (f) 900 mv. a b c d e f 2 2 SERS Intensity SERS Intensity Wavenumber / cm E / mv Figure 6. () Curve fitted SERS spectra of oxa adsorbed on a scratched gold electrode. () Potential profile (SERS intensity vs applied potential plot) obtained using the strongest band in () (around 590 cm 1 ). The circles are experimental data points and the line is to guide the eye. curve fit program was used to estimate the intensities of the oxa bands and these values were used to calculate potential profiles (SERS intensity versus applied potential plot) as presented in Fig. 6. Similarly to the SERS behaviour of oxa adsorbed on g, 17 a sharp decrease in the SERS intensity is observed in Fig. 6() for potentials more negative than 500 mv. This intensity drop should be related to the reduction of the phenoxazine ring. t potentials more positive than 300 mv; however, an increase in the SERS signal is observed, and the SERS intensities maximize again at C200 mv, which could be related to charge-transfer effects. 9 This positive potential region was not accessible in the experiments with g, owing to the anodic instability of that metal. CONCLUSIONS We demonstrated that scratched gold films can be used as substrates for enhanced-raman scattering. The scratching procedure creates a set of parallel nanometric substructures that support polarized Raman effects in forward scattering measurements. These polarization effects may be useful for the design of SERS applications. For instance, these substrates should provide another degree of freedom (polarization

6 634. G. rolo and C. J. ddison state) in analytical schemes for biomedical applications or they can also be used as optical switches in nanophotonics (or plasmonics) devices. The scratched gold can also be used as common SERS substrates in backscattering experiments using standard Raman microscopes. This was demonstrated by performing in situ spectroelectrochemical measurements for the adsorbed dyes at different applied potentials. The results obtaining with the scratched gold substrate were in accord with previous work using roughened silver electrodes. On the other hand, the gold substrate allows accessibility to a wider potential range. Experiments at potentials more positive than C100 mv showed intensities changes not observed for silver. These changes could be related to charge-transfer effects, but this can only be confirmed by a detailed study of the spectroelectrochemical behaviour of oxa on SER(R)S-active gold substrates. cknowledgements This work was supported by operating grants from the University of Victoria (Start Up grant) and the Natural Sciences and Engineering Research Council of Canada (NSERC) and by equipment grants from the Canada Foundation for Innovation (CFI New Opportunities) and the ritish Columbia Knowledge and Development Fund (CKDF). We thank Kelly kers for providing access to the Renishaw invia system. REFERENCES 1. Kelly KL, Coronado E, Zhao LL, Schatz GC. J. Phys. Chem. 2003; 107: Taton T, Mucic RC, Mirkin C, Letsinger RL. J. m. Chem. Soc. 2000; 122: Nam JM, Stoeva SI, Mirkin C. J. m. Chem. Soc. 2004; 126: Taton T, Lu G, Mirkin C. J. m. Chem. Soc. 2001; 123: Nath N, Chilkoti. nal. Chem. 2002; 74: Moskovits M. Rev. Mod. Phys. 1985; 57: Kneipp K, Kneipp H, Itzkan I, Dasari RR, Feld MS. Chem. Rev. 1999; 99: osnick K, Jiang J, rus LE. J. Phys. Chem. 2002; 106: rolo G, Irish DE, Smith D. J. Mol. Spectrosc. 1997; 405: Kahl M, Voges E. Phys. Rev. 2000; 61: rolo G, ctander E, Gordon R, Leathem, Kavanagh KL. NanoLetters 2004; 4: Markel V, Shalaev VM, Zhang P, Huynh W, Tay L, Haslett TL, Moskovits M. Phys. Rev. 1999; 59: rolo G, rctander E, ddison CJ. J. Phys. Chem. 2005; 109: Gearheart L, Ploehn HJ, Murphy CJ. J. Phys. Chem. 2001; 105: Kneipp K, Kneipp H, Kartha V, Manoharan R, Deinum G, Itzkan I, Dasari RR, Feld MS. Phys. Rev. E 1998; 57: R Taton T, Mirkin C, Letsinger RL. Science 2000; 289: rolo G, Sanderson C. Can. J. Chem. 2004; 82: Jenkins F, White HE. Fundamentals of Physical Optics. McGraw- Hill: New York, Porto J, Vidal FJG, Pendry J. Phys. Rev. Lett. 1999; 83: Schider G, Krenn JR, Gotschy W, Lamprecht, Ditlbacher H, Leitner, ussenegg FR. J. ppl. Phys. 2001; 90: Gordon R, rolo G, McKinnon, Rajora, Leathem, Kavanagh KL. Phys.Rev.Lett.2004; 92: rolo G, Sanderson C, Smith P. Phys. Rev. 2004; 69: Schneider S, Grau H, Halbig P, Freunscht P, Nikel U. J. Raman Spectrosc. 1996; 27: Santos D, Gorton L, Kubota LT. Electrochim. cta 2002; 47: Malinauskas, Ruzgas T, Gorton L. J. Electroanal. Chem. 2000; 484: 55.