Home
Search
Collections
Journals
About
Contact us
My IOPscience
Polarization-dependent SERS effects of laser-generated sub-100 nm antenna structures
This content has been downloaded from IOPscience. Please scroll down to see the full text. 2014 Nanotechnology 25 265302 (http://iopscience.iop.org/0957-4484/25/26/265302) View the table of contents for this issue, or go to the journal homepage for more
Download details: IP Address: 155.33.16.124 This content was downloaded on 17/06/2014 at 13:35
Please note that terms and conditions apply.
Nanotechnology Nanotechnology 25 (2014) 265302 (8pp)
doi:10.1088/0957-4484/25/26/265302
Polarization-dependent SERS effects of laser-generated sub-100 nm antenna structures Limei Chen1, Tianrui Zhai2, Xinping Zhang2, Claudia Unger3, Jürgen Koch3, Boris N Chichkov3 and Peter J Klar1 1
I. Institute of Physics, Justus-Liebig University of Giessen, Heinrich-Buff-Ring 16, Giessen, D-35392, Germany 2 College of Applied Sciences, Beijing University of Technology, 100124, People’s Republic of China 3 Laser Zentrum Hannover e.V., Hollerithallee 8, Hannover, D-30419, Germany E-mail: [email protected] and [email protected] Received 20 November 2013, revised 4 May 2014 Accepted for publication 12 May 2014 Published 11 June 2014 Abstract
Sub-100 nm antenna arrays consisting of a star-like ridge or dome-like structures with needles in their centers are prepared in thin gold films on glass substrates using femtosecond laser pulses. The needles can be bent mechanically to be horizontally aligned to the substrate surface. Controlled variation of the pulse energy allows one to obtain nanostructures of different defined morphologies. These arrays of nanostructures are covered with a thin homogeneous layer of rhodamine molecules. Raman spectra using linearly polarized laser light of 632.8 nm are taken with the laser spot centered on individual nanostructures and at positions on the unstructured film. The average Raman enhancement within the laser spot focused onto a nanostructure is two orders of magnitude higher than on the unstructured film. The nanostructures with bent needles exhibit a polarization dependence of the SERS effect, i.e., typically the enhancement is larger by about a factor of two for excitation light polarized parallel to the needle direction than for the perpendicular case. The enhancement factor of the star-like ridge structures with needles is analyzed by the finite-element method, which agrees with the experiment. We show that the variation of the SERS activity of almost similar structures arises from the inherent randomness of the hot spots created in the fabrication process. Nevertheless, these antenna structures may be useful as elements in novel SERS devices as they can be accurately positioned on a device using a cheap fabrication process compatible with microfabrication technology. Keywords: SERS effects, antenna arrays, polarization dependence (Some figures may appear in colour only in the online journal) enhancement of the electric field amplitude of the light impinging on the sample, placing small samples, e.g. single molecules, into such hot spots causes an enhancement of their Raman signals by factors of up to 1015 under resonant excitation conditions [6]. For this reason, Raman spectroscopy can be performed on single molecules [7, 8]. The antenna effects are based on the interaction of the light field with the free carrier plasma in the metallic nanostructures. This relationship is further confirmed by the finding that the Raman enhancement is largest in resonance with the plasmon resonance of metal nanoparticles [9–11].
1. Introduction Surface-enhanced Raman spectroscopy (SERS) and tipenhanced Raman spectroscopy (TERS) make use of local antenna effects that occur in the vicinity of edges and tips of metallic nanostructures or in narrow gaps between nanostructures [1–5]. The light-field impinging on the sample is focused in so called sub-wavelength hot-spots where the intensity of the light field is locally enhanced by several orders of magnitude. As the Raman effect is to a first approximation proportional to the fourth power of the local 0957-4484/14/265302+08$33.00
1
© 2014 IOP Publishing Ltd Printed in the UK
L Chen et al
Nanotechnology 25 (2014) 265302
Samples exhibiting SERS can be divided into two major groups. One comprises films of random arrangements of metal nanostructures that may be realized by various means, e.g. annealing of substrates coated with nanostructures or deposition of very thin metal films etc [12–15]. The other group comprises regular or designed arrangements of metal nanostructures that can be prepared by approaches either based on electron beam lithographic methods or on selfassembly of nanostructures [16–21]. In the case of regular arrangements, the hot spots can be adjusted in a controlled way in the fabrication process, but the fabrication of the structures is usually expensive and time consuming. A prominent example of such an adjustable hot spot is the gap between the two triangles forming a metal bow-tie structure [22]. Typically, the identification and optimization of hot spots in random structures is not as straight forward as for regular arrangements of metallic nanostructures. In general, a compromise between the simplicity and costs of the fabrication process, on the one hand, and the degree of spatial definition and magnitude of the signal enhancement of the hot spots, on the other hand, needs to be made. Several theoretical models have been developed to explain SERS. The local electric field distribution can be calculated by solving Maxwell’s equations for the light field impinging on the nanostructures. Often the static approach of solving Laplace’s equation for a constant external electric field in the presence of the metallic nanostructure arrangement yields useful estimates of the local field enhancement and, thus, the Raman enhancement [23]. Finite-element methods are preferably used to solve the differential equations, either for the static or the time-dependent case [24, 25]. However, there are also efforts to combine electrodynamics and quantum mechanics to account for the response of the analyte [26, 27]. Here, we present a fast approach for fabricating regular arrays of metal nanostructures that exhibit polarizationdependent SERS effects at spatially defined hot spots. The structures are tested by studying the local enhancement of the Raman signals of a thin homogeneous rhodamine film covering the nanostructure array. The Raman enhancement is modeled by a finite-element approach where the morphology of the structures is determined by a scanning electron microscopy (SEM) analysis. The spot size and polarization of the excitation light are also taken into account. We show by numerical simulations based on the structural parameters extracted from the SEM images of the arrays of metal nanostructures that the average polarization-dependent Raman enhancement factor of a metal nanostructure of an array of similar metal nanostructures can be well reproduced by theory. The modelling also explains the differences of the Raman enhancement factors of individual metal structures of the same array. To a large extent, these differences reflect the sensitivity of the Raman enhancement to minor structural variations, such as a variation of gap width between edges, of curvature or of sharpness of edges, etc.
Figure 1. SEM images of parts of the antenna arrays prepared by
laser processing. (a) Arrays prepared with pulse energies of about 200 nJ. (b) Arrays prepared with the same parameters as (a), but with additional mechanical bending of the needles. (c) Arrays prepared with pulse energies of about 100 nJ. (d) Arrays prepared with the same parameters as (c), but the needles are bent mechanically. Scale bar, 5 μm.
2. Fabrication of sub-100 nm antenna arrays Regular arrays of SERS-active nanostructures of defined morphology are prepared by laser processing of thin gold films on glass substrates. First, a 100 nm thick gold film is coated onto a glass substrate by thermal evaporation. Nanostructures at specific positions on the gold film can be formed by irradiating these positions with single femtosecond laser pulses centered at a wavelength of 800 nm with pulse energies ranging from 50 to 200 nJ [28–30]. In the laser focus, the gold film is locally melted and forms a liquid metal jet which by solidification yields a vertical gold needle on a metal dome. The process has been studied in time-resolved pump-probe experiments [28]. The morphology of the needle and the dome can be controlled by varying the energy of the single pulse and the focusing conditions. As examples, figures 1(a) and (c) depict SEM images of regular arrays of needle-like nanostructures obtained by using pulse energies of 200 nJ and 100 nJ, respectively. At the lower pulse energy, a semisphere-like dome with the upright needle in its center is formed in the solidification process. At higher pulse energies, this dome collapses in the formation process, yielding on solidification a star-like radial structure of gold edges facing the needle in its center. The needles arising are about 1 μm in height with the tip shape depending on the process parameters, e.g., a sharp tip with a tip diameter of less than 100 nm in figure 1(a) and a ball-like tip in figure 1(c). Furthermore, it can be seen in the images that the gold nanostructures formed for defined process parameters are very similar and can be regularly arranged. The SEM images in figures 1(b) and (d) show needle arrays where the needles are 2
L Chen et al
Nanotechnology 25 (2014) 265302
Figure 2. Raman images of the antenna array in figure 1(b) using a 50× objective with the laser excitation polarized (a) parallel or (b)
perpendicular to the bent needles. (c) Raman image of a single nanostructured antenna of the type depicted in figure 1(a) obtained using a 160× objective. The images are taken with the band pass set to detect the scattered light of the prominent Raman mode of rhodamine at 1350 cm−1.
bent mechanically to be horizontally aligned to the substrate surface, which breaks the symmetry of the structure to introduce polarization dependence. The mechanical bending is achieved by sliding a glass plate across the nanostructured surface containing the prepared needle structures. The upright needles are then bent preferentially along the sliding direction. The SEM images demonstrate that the morphology of the needle structures can be tuned by the processing parameters. Furthermore, the needle structures within an array prepared by using the same process parameters are very similar in appearance, demonstrating the reproducibility of the fabrication process. However, the minor differences, which remain between individual needle structures, such as variations in gap widths between edges, tilting angle or length of the needle still affect the enhancement factor observed in the Raman experiment, as will be shown below. For the Raman experiments, a homogeneous thin film of a 0.05 mM rhodamine 110 solution in isopropanol was spincoated onto the nanostructure arrays. It caused the adhesion of a few monolayers of rhodamine molecules on the sample surface after evaporation of the solvent. The Raman experiments were performed in back-scattering geometry using a Renishaw In Via Raman microscope system equipped with polarization optics. A He-Ne laser with 632.8 nm emission wavelength was used as excitation source. Raman signals were detected with the system operating either in imaging mode or in spectral mode. In the imaging mode, a single pixel of the CCD detector corresponds to about 80 × 80 nm2 sample area when a 160× objective is used. Raman spectra of the nanostructure arrays with needles tilted parallel to the sample surface were acquired with the linear polarization of the excitation laser parallel as well as perpendicular to the needle direction.
imaging unit was set to detect the most prominent Raman band of rhodamine at 1350 cm−1. All three images are not corrected for the Gaussian profile of the intensity distribution within the laser spot. Using a 50× objective, the spot illuminated by the Gaussian intensity profile of the laser is about 15 μm in diameter and covers several nano needle sites. Corresponding Raman images obtained of an array with needle structures with bent needles of the type depicted in figure 1(b) with the linear polarization of the laser parallel and perpendicular to the needle direction are shown in figures 2(a) and (b), respectively. The regular arrangement of the nanostructures is clearly visible in the Raman image, indicating that each nanostructure exhibits an enhanced Raman intensity compared to the sample surroundings. It should also be noted that the intensity enhancement is about a factor of two larger in case of the parallel polarization geometry than for the perpendicular one. This is a first indication of the antenna action of the bent needle. Figure 2(c) shows a Raman image recorded of a single antenna structure of the type depicted in figure 1(a). A 160× objective was used, yielding a width of the Gaussian intensity profile of about 3 μm. The enhancement of the Raman intensity at the position of the nanostructure is clearly visible in the image and confined to an area of a few hundred nm. This demonstrates that hot spots are formed within the individual nanostructures even in the case of upright needles. Of course, individual hot spots on the nanostructures cannot be identified due to the diffraction limit. To quantify the differences caused by Raman enhancement of the nanostructured areas, we obtained spectra with the laser focused onto a nanostructured needle and focused onto the plain sample between nanostructures, respectively. Typical spectra (red) obtained of the needle structures of the different types depicted in figures 1(a), (b), and (d), respectively, are shown in figure 3, together with the corresponding spectra (black) taken between nanostructures on the plain areas of the samples. Taking into account that the black spectra were acquired with a 10 times higher laser power, it can be estimated that for all three structures the enhancement factors are of the same order of magnitude, about 200, i.e., 210, 215 and 230, respectively, in figures 3(a) to (c). The enhancement of the Raman signal is more than two orders of magnitude
3. SERS effects based on antenna arrays In figure 2, three Raman images are shown. The laser spot is defocussed in such a way that a larger circular spot is illuminated with a Gaussian intensity profile. The spatial distribution of the Raman signal in a specified wavenumber band is imaged onto the CCD. In all cases, the spectral filter of the 3
L Chen et al
Nanotechnology 25 (2014) 265302
Figure 3. Comparison of Raman spectra of rhodamine taken on the unstructured sample surface (black) and different nanostructures (red). (a) Star-like nanostructures with upright needles as shown in figure 1(a), (b) Star-like nanostructures with bent needles as shown in figure 1(b), (c) Nanostructures with bent ball-like needles as shown in figure 1(d). The intensity of the exciting laser was 10 times higher for spectra taken between needle structures compared to spectra taken on needle structures.
Figure 4. Comparison of the Raman spectra of rhodamine measured on bent needles for two polarization geometries, either polarization
parallel to the needle direction (red) or perpendicular to the needle direction (black). (a) Star-like nanostructures with bent needles as shown in figure 1(b), and (b) nanostructures with bent ball-like needles as shown in figure 1(d).
the two polarizations (I|| − I⊥)/(I|| + I⊥) is 30% ± 20%. The comparatively large standard deviation has two main causes. First, despite the good control and reproducibility of the fabrication process of the needle structures, a residual randomness of the nanostructure shape remains and has an impact on the hot-spot formation. Second, several hot spots within the nanostructure consisting of a star-like ridge with bent needles (figure 1(b)) or a deformed dome with bent needles (figure 1(d)) are formed, which contributes to the observed average enhancement within the laser spot. Due to the randomness of the positions and strengths of the hot spots (which are formed at the ridges or within the deformed dome), what may occur is excitation by light polarized perpendicular to the needle direction. This causes a stronger enhancement, despite the antenna effects due to the bent needle. We will come back to this point in section 4. On the other hand, thermal drift in the experiment also has an impact on the experimental error bars, as it difficult to ensure that the measurements for the two polarization directions are indeed performed on exactly the same spot on the sample. We find no significant polarization dependence for the upright needles of the type depicted in figure 1(a), i.e.
compared to the unstructured regions on the sample. It should be noted that the polarization direction with respect to the needle direction was always perpendicular to the needle direction, also in case of the bent needles. It is worth noting that for needle structures of the type depicted in figure 1(c) hardly any enhancement of the Raman signal was observed. In case of the bent needles depicted in figures 1(b) and (d), we have also performed polarization-dependent measurements with the incoming laser light polarized parallel and perpendicular to the needle direction. The corresponding sets of spectra for the two types of needle structures are shown in figures 4(a) and (b), respectively. The Raman signals for the two polarization geometries differ by almost a factor of two. The Raman enhancement is larger if the incoming light is polarized along the needle direction, indicating that the needles act as antennas. The behavior is reproducible and in agreement with the results of the Raman imaging in figures 2(a) and (b). Furthermore, the same results are obtained when turning the polarization direction in the beam path of the excitation light and when rotating the sample by 90 degrees, i.e., the effect is no artifact of the measurement system. The majority of the bent needles (eight out of ten structures) show a stronger enhancement for the incoming light being linearly polarized along the needle direction. On average, the relative difference of the Raman intensities for
(I
∥
− I
) (I
∥
)
− I is 0% ± 20%. The origin of the standard
deviation is the same and has been described above. 4
L Chen et al
Nanotechnology 25 (2014) 265302
its intensity profile in the xy-plane was Gaussian, with a width of λ = 632.8 nm. To model the average Raman enhancement factor of the metal nanostructure within the laser spot, finite element calculations were performed using the software package COMSOL Multiphysics 3.5a. In the calculations, the electric field distributions are derived using the 3D/RF Module/ Electromagnetic waves/Harmonic propagation. The solver is chosen as ‘Parametric’. All boundary settings are chosen as ‘scattering boundary condition’. The maximum element size is set to 2 nm in ‘Subdomain mesh parameters’ to guarantee the accuracy of calculation. The refractive index of gold is chosen as 0.181 + 3.068·i in the vicinity of λ = 632.8 nm [31]. For a given model structure, the following four steps of the calculation were carried out to derive its average Raman enhancement factor and to identify the hot spots of the nanostructure: (1) Calculation of the electric field distribution E0(x, y) of the incident laser beam with a wavelength of 632.8 nm in air. The corresponding values are calculated on a square mesh of 2 μm × 2 μm size with a pitch of 1 nm and the laser beam centered at (x, y) = (1 μm, 1 μm). The data (E0(x, y)4) are stored as a data array which is independent of z. (2) Calculation of the local electric field distribution E (x, y, z) when the laser impinges on a star-like ridge structure with a needle in its center. The laser spot and the nanostructure illuminated are both centered at ( x, y ) = ( 1 μm, 1 μm). The corresponding values are again calculated on a mesh of 2 μm × 2 μm size with a 1 nm pitch. The data (E (x, y, z)4) are stored as a data array which depends on z. (3) Derivation of the local Raman enhancement factors as g ( x, y, z ) = E ( x, y, z )4 E0 ( x, y )4 . Typical maximal values g are in the range of 104–109. (4) Derivation of the average Raman enhancement factor by averaging across the beam profile in the xy-plane and along the z-direction of the nanostructure:
Figure 5. Computing models for star-like ridge structures with an
upright needle (a) or a horizontally tilted needle (c) based on the SEM images of the corresponding nanostructures (b) and (d). The radially arranged ridges and the needle base are characterized by the structural parameters a, b, c, d, e and N the number of ridges. The needle itself is characterized by the needle length L and the tilting angle α.
4. Modeling of the SERS effect of star-like needle structures To understand the origin of the Raman enhancement and the effect of the nanostructure morphology on its magnitude, we modelled exemplarily the field enhancement of the star-like ridge structures with needles. For this purpose, the morphology of the star-like needle structure is approximated by an upright (or tilted) needle surrounded by a set of radially arranged ridges. The corresponding simplified models are shown in figures 5(a) and (c), and are based on the SEM images in figures 5(b) and (d). The radially arranged ridges and the needle base are characterized by the structural parameters a, b, c, d, e and N, the number of ridges. The needle itself is characterized by the needle length L and the tilting angle α defining the gap between needle tip and ridges. The crucial parameters determining the magnitude of the Raman enhancement are the gap b between the needle base and ridge edge, the number of ridges N, the needle length L and the tilting angle α. For simplicity, the other structural parameters were kept constant at values corresponding to the SEM images: ridge length c = 250 nm, radius of the ridge on the outside e = 40 nm, inner radius of the ridge d = 10 nm, and needle diameter at the base a = 130 nm. The scattering geometry of the Raman experiment can be explained by the following. The probe light impinging on the metal nanostructure travelled along the z-direction (perpendicular to the sample plane). The light was linearly polarized either along x or y (the xy-plane being the sample plane), and
g =
∑z
∑ g ( x , y , z ) E 0( x , y ) 4 d A ∑ E 0( x , y ) 4 d A
(1)
where the sums in numerator and denominator are in the xyplane, dA = 1 nm2. Typical values derived for are in the range of 10–104. As an illustration of typical results, figure 6 depicts the normalized local intensity distributions E (x, y, z)4 of the incoming light field for cross sections in the xy-plane and the yz-plane for both types of needle structures, i.e. with the needle pointing up and with the needle being tilted horizontally, respectively. In both cases the incoming light is linearly polarized along the y-direction, as denoted by the red arrows in figure 6. It can be seen from the cross sections in the xyplane that hot spots occur at gaps between needle base and ridges, when the polarization direction is parallel to the corresponding radial orientation of the ridge, i.e., for ridges pointing in y-direction or the ridges closest to the y-direction and not screened by the needle when tilted. Thus, in case of a 5
L Chen et al
Nanotechnology 25 (2014) 265302
of the gaps formed and the number of the gaps is part of the explanation of the significant variation of the SERS properties observed in section 3. Marginal differences in the morphology of needle structures, such as a variation of the gap by a nanometer, may occur, although these structures are prepared using the same process parameters. Furthermore, we have performed calculations for different geometries in terms of needle orientation and polarization direction of the excitation light with respect to the needle orientation for a constant number of ridges N = 8. If a needle oriented horizontally is excited with the linear polarization parallel to the needle direction, the average Raman enhancement factor is larger than in case of excitation with polarization perpendicular to the needle direction. This confirms the antenna action of a horizontal needle when excited with light polarized along the needle direction. Both other geometries where the electric field of the light is perpendicular to the needle direction, i.e., an upright needle pointing in z-direction with the electric field in the xy-plane or a horizontal needle pointing in x-direction with the electric field along y, show basically the same average Raman enhancement factor. This indicates that the needle has hardly any additional impact in these cases apart from its base forming the gaps with the ridges. The difference between the Raman enhancement factors for incoming light polarized perpendicular and parallel to needle direction decreases when the tilting angle α of the needle increases, i.e., the curve for parallel polarization (purple curve with triangles) would approach the curve for perpendicular polarization (dashed green line with crosses). However, this difference between parallel and perpendicular polarization not only depends on the gap formed between needle tip and ridges, but even more severely on needle length, as can be seen in figure 7(b). Figure 7(b) shows the dependence of the average Raman enhancement factor on needle length L for a nanostructure with a needle tilted horizontally. Both the number of ridges and the gap width were kept constant, N = 8 and b = 15 nm. Calculations were performed for two excitation geometries, exciting with light polarized parallel (blue curve, full circles) and perpendicular (red curve, open circles) to the needle direction. An oscillatory dependence of on needle length with a decaying envelope function is found in the parallel case. The maxima probably correspond to antenna resonances of different order, i.e. multipolar plasmon excitations [32]. Of course, this strong dependence on needle length also contributes to the variation of the Raman enhancement observed for needle structures prepared using the same process parameters. However, if the needle length can be well controlled in the fabrication process, the average Raman enhancement factor can be maximized for this particular excitation geometry. This can be accomplished by optimizing the antenna action of the needle, i.e. tuning its length in resonance with the incoming light wavelength. This finding is further corroborated by two insets of figure 7(b) depicting two cross sections of the yz-plane for needle lengths, where is minimal and maximal, respectively. In contrast, an almost flat curve is observed in the perpendicular case, which is always lower than in the parallel case confirming our experimental
Figure 6. Modeling of the hot spots for needle structures with (a)
upright needles (α = 90°) and (b) horizontally tilted needles (α = 0°). N = 12, L = 850 nm, b = 10 nm. Plots of |E (x, y, z)|4 in the xy-plane and in the yz-plane are for z0 = 1 μm and x0 = 1 μm, respectively. The red arrows denote the polarization orientation of probe light.
horizontally tilted needle, the average Raman enhancement factor depends on the orientation of the polarization with respect to the needle direction in agreement with the experimental findings. The cross section in the yz-plane for the structure with the upright needle reveals that the needle has no additional impact on the average Raman enhancement factor apart from forming gaps to the ridges. Figure 7(a) shows the dependence of the average Raman enhancement factor on the gap width b between the base of the needle and the ridges. The length of the needle was kept constant, L = 250 nm. Three curves are calculated for different numbers of ridges surrounding an upright needle radially, i.e. N = 4, 8, 12. As expected, the Raman enhancement increases with the number of ridges N. It can also be seen that the average Raman enhancement factor decreases rapidly with increasing gap size b between needle base and ridges for all N. For example, the relative change of the Raman enhancement factor with b in case for N = 12 ridges in the range between b = 5 and 10 nm is about 20% per nm. This sensitivity of the average enhancement on the width 6
L Chen et al
Nanotechnology 25 (2014) 265302
Figure 7. (a) Dependence of the average Raman enhancement factor on the gap width b between the base of the needle and the ridges. Length of the needle L = 250 nm. The number of ridges, the orientation of the needle, as well as the excitation geometry are given in the legend. (b) Dependence of the average Raman enhancement factor on needle length L for a nanostructure with needle tilted horizontally and excited with light polarized parallel (blue line)/perpendicular (red line) to the needle direction. The number of ridges is N = 8, and the gap width is b = 15 nm.
findings. However, it should be noted here again that the behavior of two curves for parallel and perpendicular polarization isvery sensitive to the tilting angle of the needle. The tilting angle α changes the widths of the gaps between the needle and ridges where hot spots occur and, thus, the corresponding contributions to the Raman enhancement. If the tilting angle is increasing by a few degrees away from α = 0, i.e., the gap between tip and ridge becomes wider, a reversal of the order of the curves in figure 7(b) may occur for long needles. The reason is that the effect of shading by the needle becomes more important than the additional Raman enhancement at the gap between tip and ridge. Thus, the Raman enhancement and the polarization effects are also very sensitive to fluctuations of the needle orientation, which explains the statistical variations in the polarization effects observed for different individual needle structures prepared using the same process parameters. Nevertheless, if a high degree of control of tilting angle and length of the needle can be achieved in the fabrication process, a much better performance in terms of Raman enhancement will be expected for bent-needle structures than for upright-needle structures. Finally, we would like to briefly discuss the Raman enhancement of the structures with upright needles on intact semispherical domes (figure 1(c)) and the corresponding structures with needles tilted horizontally (figure 1(d)). The former do not show a significant Raman enhancement, as no ridges and gaps are present and the antenna effect of an upright needle is negligible. In case of the corresponding structures with needles tilted horizontally (figure 1(d)) the situation is different, as the dome is deformed and edges occur when the needles are mechanically bent. The folds and bends which are present in this type of needle structure are not aligned in the regular star-like fashion as are the structures depicted in (figures 1(a) and (b)). However, hot spots are formed in the gaps and edges caused by the deformation of the dome in the bending process of the needle. We have not modelled this type of structure in more detail. However, the
results are expected to be essentially the same as for the needle structures with the star-like ridges.
5. Conclusions We have analyzed the SERS effect of nanostructures composed of star-like ridge structures with needles in the center fabricated by femtosecond laser pulses. The average Raman enhancement factors can be explained quantitatively in terms of their behavior on nanostructure morphology and excitation geometry. In case of the horizontally tilted needles the average Raman enhancement factor will depend on the orientation of the polarization with respect to the needle direction. For polarization of the incoming light parallel to the needle direction, the average Raman enhancement factor can be maximized when the antenna action of the needle is optimized by tuning its length in resonance with the incoming light wavelength. In contrast, upright needles have no additional impact on the average Raman enhancement factor apart from forming gaps to the ridges. A quantitative agreement between experimentally determined average Raman enhancement factors and their theoretical counterparts can be obtained by a simplified theoretical model. The model also indicates that the SERS effects of such needle structures are very sensitive to minor fluctuations in the morphology, which explains the experimentally observed variations of the SERS activity for almost identical nanostructures. Nevertheless, the SERS enhancement averaged over the area of the gold nanostructure is reproducibly two orders of magnitude larger than on the plain gold film. Therefore, the fast and low-cost laser processing of gold films presented here offers an interesting alternative to accurately define SERS-active areas on the micrometer scale as part of integrated optical devices. 7
L Chen et al
Nanotechnology 25 (2014) 265302
Acknowledgements
[15] Albrecht M G and Creighton J A 1977 J. Am. Chem. Soc. 99 5215 [16] Lindquist N C, Nagpal P, McPeak K M, Norris D J and Oh S-H 2012 Rep. Prog. Phys. 75 036501 [17] Beermann J, Novikov S M and Leosson K Bozhevolnyi S I 2009 Optics Express 17 12698 [18] Hou Y, Xu J, Wang P and Yu D 2010 Appl. Phys. Lett. 96 203107 [19] Bahns J T, Imre A, Vlasko-Vlasov V K, Pearson J, Hiller J M, Chen L H and Welp U 2007 Appl. Phys. Lett. 91 081104 [20] Chen G, Wang Y, Yang M, Xu J, Goh S J, Pang M and Chen H 2010 J. Am. Chem. Soc. 132 3644 [21] Schuck P J, Fromm D P, Sundaramurthy A, Kino G S and Moerner W E 2009 Phys. Rev. Lett. 94 017402 [22] Kravets V G, Zoriniants G, Burrows C P, Schredin F, Casiraghi C, Klar P, Geim A K, Barnes W L and Grigorenko A N 2010 Phys. Rev. Lett. 105 246806 [23] Behr N and Raschke M B 2008 J. Phys. Chem. C 112 3766 [24] Krug J T 2nd, Sanchez E J and Xie X S 2002 J. Chem. Phys. 116 10895 [25] Crozier K B, Sundaramurthy A, Kino G S and Quate C F 2003 J. Appl. Phys. 94 4632 [26] Geshev P I, Fischer U and Fuchs H 2010 Phys. Rev. B 81 125441 [27] Mullin J, Valley N, Blaber M G and Schatz G C 2012 Phys. Chem. A 116 95774 [28] Unger C, Koch J, Overmeyer L and Chichkov B N 2012 Opt. Express 20 24864 [29] Koch J, Korte F, Bauer T, Fallnich C, Ostendorf A and Chichkov B N 2005 Appl. Phys. A 81 325 [30] Kuznetsov A I, Koch J and Chichkov B N 2009 Appl. Phys. A 94 221 [31] Johnson P B and Christy R W 1972 Phys. Rev. B 6 4370 [32] Krenn J R, Schider G, Rechberger W, Lamprecht B, Leitner A, Aussenegg F R and Weeber J C 2000 Appl. Phys. Lett. 77 3379
The authors from Giessen and Beijing thank the German Exchange Service (DAAD) and the Chinese Science Council for supporting this work in the course of their bilateral exchange programme (PPP). We thank Arseniy Kuznetsov for technical assistance.
References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14]
Moskovits M 1985 Rev. Mod. Phys. 57 783 Hartschuh A 2008 Ang. Chem. Int. Ed. 47 8178 Pettinger B 2010 Molecular Phys. 108 2039 Bailo E and Deckert V 2008 Chem. Soc. Rev. 37 921 Stiles P L, Dieringer J A, Shah N C and Van Duyne R P 2008 Ann. Rev. Anal. Chem. 1 601 Dieringer J A, Wustholz K L, Masiello D J, Camden J P, Kleinman S L, Schatz G C and Van Duyne R P 2009 J. Am. Chem. Soc. 131 849 Nie S M and Emory S R 1977 Science 275 1102 Kneipp K, Wang Y, Kneipp H, Perelman L T, Itzkan I, Dasari R and Feld M S 1997 Phys. Rev. Lett. 78 1667 McFarland A D, Young M A, Dieringer J A and Van Duyne R P 2005 J. Phys. Chem. B 109 11279 Neacsu C C, Berweger S and Raschke M B 2007 Nanobiotechnol. 3 172 Haynes C L and Van Duyne R P 2003 J. Phys. Chem. B 107 7426 Liu Y, Yu C and Sheu S 2006 J. Mater. Chem. 16 3546 Zhang X, Liu H and Feng S 2009 Nanotechnology 20 425303 Jeanmarie D L and Van Duyne R P 1977 J. Electroanal. Chem. 84 1
8
Search
Collections
Journals
About
Contact us
My IOPscience
Polarization-dependent SERS effects of laser-generated sub-100 nm antenna structures
This content has been downloaded from IOPscience. Please scroll down to see the full text. 2014 Nanotechnology 25 265302 (http://iopscience.iop.org/0957-4484/25/26/265302) View the table of contents for this issue, or go to the journal homepage for more
Download details: IP Address: 155.33.16.124 This content was downloaded on 17/06/2014 at 13:35
Please note that terms and conditions apply.
Nanotechnology Nanotechnology 25 (2014) 265302 (8pp)
doi:10.1088/0957-4484/25/26/265302
Polarization-dependent SERS effects of laser-generated sub-100 nm antenna structures Limei Chen1, Tianrui Zhai2, Xinping Zhang2, Claudia Unger3, Jürgen Koch3, Boris N Chichkov3 and Peter J Klar1 1
I. Institute of Physics, Justus-Liebig University of Giessen, Heinrich-Buff-Ring 16, Giessen, D-35392, Germany 2 College of Applied Sciences, Beijing University of Technology, 100124, People’s Republic of China 3 Laser Zentrum Hannover e.V., Hollerithallee 8, Hannover, D-30419, Germany E-mail: [email protected] and [email protected] Received 20 November 2013, revised 4 May 2014 Accepted for publication 12 May 2014 Published 11 June 2014 Abstract
Sub-100 nm antenna arrays consisting of a star-like ridge or dome-like structures with needles in their centers are prepared in thin gold films on glass substrates using femtosecond laser pulses. The needles can be bent mechanically to be horizontally aligned to the substrate surface. Controlled variation of the pulse energy allows one to obtain nanostructures of different defined morphologies. These arrays of nanostructures are covered with a thin homogeneous layer of rhodamine molecules. Raman spectra using linearly polarized laser light of 632.8 nm are taken with the laser spot centered on individual nanostructures and at positions on the unstructured film. The average Raman enhancement within the laser spot focused onto a nanostructure is two orders of magnitude higher than on the unstructured film. The nanostructures with bent needles exhibit a polarization dependence of the SERS effect, i.e., typically the enhancement is larger by about a factor of two for excitation light polarized parallel to the needle direction than for the perpendicular case. The enhancement factor of the star-like ridge structures with needles is analyzed by the finite-element method, which agrees with the experiment. We show that the variation of the SERS activity of almost similar structures arises from the inherent randomness of the hot spots created in the fabrication process. Nevertheless, these antenna structures may be useful as elements in novel SERS devices as they can be accurately positioned on a device using a cheap fabrication process compatible with microfabrication technology. Keywords: SERS effects, antenna arrays, polarization dependence (Some figures may appear in colour only in the online journal) enhancement of the electric field amplitude of the light impinging on the sample, placing small samples, e.g. single molecules, into such hot spots causes an enhancement of their Raman signals by factors of up to 1015 under resonant excitation conditions [6]. For this reason, Raman spectroscopy can be performed on single molecules [7, 8]. The antenna effects are based on the interaction of the light field with the free carrier plasma in the metallic nanostructures. This relationship is further confirmed by the finding that the Raman enhancement is largest in resonance with the plasmon resonance of metal nanoparticles [9–11].
1. Introduction Surface-enhanced Raman spectroscopy (SERS) and tipenhanced Raman spectroscopy (TERS) make use of local antenna effects that occur in the vicinity of edges and tips of metallic nanostructures or in narrow gaps between nanostructures [1–5]. The light-field impinging on the sample is focused in so called sub-wavelength hot-spots where the intensity of the light field is locally enhanced by several orders of magnitude. As the Raman effect is to a first approximation proportional to the fourth power of the local 0957-4484/14/265302+08$33.00
1
© 2014 IOP Publishing Ltd Printed in the UK
L Chen et al
Nanotechnology 25 (2014) 265302
Samples exhibiting SERS can be divided into two major groups. One comprises films of random arrangements of metal nanostructures that may be realized by various means, e.g. annealing of substrates coated with nanostructures or deposition of very thin metal films etc [12–15]. The other group comprises regular or designed arrangements of metal nanostructures that can be prepared by approaches either based on electron beam lithographic methods or on selfassembly of nanostructures [16–21]. In the case of regular arrangements, the hot spots can be adjusted in a controlled way in the fabrication process, but the fabrication of the structures is usually expensive and time consuming. A prominent example of such an adjustable hot spot is the gap between the two triangles forming a metal bow-tie structure [22]. Typically, the identification and optimization of hot spots in random structures is not as straight forward as for regular arrangements of metallic nanostructures. In general, a compromise between the simplicity and costs of the fabrication process, on the one hand, and the degree of spatial definition and magnitude of the signal enhancement of the hot spots, on the other hand, needs to be made. Several theoretical models have been developed to explain SERS. The local electric field distribution can be calculated by solving Maxwell’s equations for the light field impinging on the nanostructures. Often the static approach of solving Laplace’s equation for a constant external electric field in the presence of the metallic nanostructure arrangement yields useful estimates of the local field enhancement and, thus, the Raman enhancement [23]. Finite-element methods are preferably used to solve the differential equations, either for the static or the time-dependent case [24, 25]. However, there are also efforts to combine electrodynamics and quantum mechanics to account for the response of the analyte [26, 27]. Here, we present a fast approach for fabricating regular arrays of metal nanostructures that exhibit polarizationdependent SERS effects at spatially defined hot spots. The structures are tested by studying the local enhancement of the Raman signals of a thin homogeneous rhodamine film covering the nanostructure array. The Raman enhancement is modeled by a finite-element approach where the morphology of the structures is determined by a scanning electron microscopy (SEM) analysis. The spot size and polarization of the excitation light are also taken into account. We show by numerical simulations based on the structural parameters extracted from the SEM images of the arrays of metal nanostructures that the average polarization-dependent Raman enhancement factor of a metal nanostructure of an array of similar metal nanostructures can be well reproduced by theory. The modelling also explains the differences of the Raman enhancement factors of individual metal structures of the same array. To a large extent, these differences reflect the sensitivity of the Raman enhancement to minor structural variations, such as a variation of gap width between edges, of curvature or of sharpness of edges, etc.
Figure 1. SEM images of parts of the antenna arrays prepared by
laser processing. (a) Arrays prepared with pulse energies of about 200 nJ. (b) Arrays prepared with the same parameters as (a), but with additional mechanical bending of the needles. (c) Arrays prepared with pulse energies of about 100 nJ. (d) Arrays prepared with the same parameters as (c), but the needles are bent mechanically. Scale bar, 5 μm.
2. Fabrication of sub-100 nm antenna arrays Regular arrays of SERS-active nanostructures of defined morphology are prepared by laser processing of thin gold films on glass substrates. First, a 100 nm thick gold film is coated onto a glass substrate by thermal evaporation. Nanostructures at specific positions on the gold film can be formed by irradiating these positions with single femtosecond laser pulses centered at a wavelength of 800 nm with pulse energies ranging from 50 to 200 nJ [28–30]. In the laser focus, the gold film is locally melted and forms a liquid metal jet which by solidification yields a vertical gold needle on a metal dome. The process has been studied in time-resolved pump-probe experiments [28]. The morphology of the needle and the dome can be controlled by varying the energy of the single pulse and the focusing conditions. As examples, figures 1(a) and (c) depict SEM images of regular arrays of needle-like nanostructures obtained by using pulse energies of 200 nJ and 100 nJ, respectively. At the lower pulse energy, a semisphere-like dome with the upright needle in its center is formed in the solidification process. At higher pulse energies, this dome collapses in the formation process, yielding on solidification a star-like radial structure of gold edges facing the needle in its center. The needles arising are about 1 μm in height with the tip shape depending on the process parameters, e.g., a sharp tip with a tip diameter of less than 100 nm in figure 1(a) and a ball-like tip in figure 1(c). Furthermore, it can be seen in the images that the gold nanostructures formed for defined process parameters are very similar and can be regularly arranged. The SEM images in figures 1(b) and (d) show needle arrays where the needles are 2
L Chen et al
Nanotechnology 25 (2014) 265302
Figure 2. Raman images of the antenna array in figure 1(b) using a 50× objective with the laser excitation polarized (a) parallel or (b)
perpendicular to the bent needles. (c) Raman image of a single nanostructured antenna of the type depicted in figure 1(a) obtained using a 160× objective. The images are taken with the band pass set to detect the scattered light of the prominent Raman mode of rhodamine at 1350 cm−1.
bent mechanically to be horizontally aligned to the substrate surface, which breaks the symmetry of the structure to introduce polarization dependence. The mechanical bending is achieved by sliding a glass plate across the nanostructured surface containing the prepared needle structures. The upright needles are then bent preferentially along the sliding direction. The SEM images demonstrate that the morphology of the needle structures can be tuned by the processing parameters. Furthermore, the needle structures within an array prepared by using the same process parameters are very similar in appearance, demonstrating the reproducibility of the fabrication process. However, the minor differences, which remain between individual needle structures, such as variations in gap widths between edges, tilting angle or length of the needle still affect the enhancement factor observed in the Raman experiment, as will be shown below. For the Raman experiments, a homogeneous thin film of a 0.05 mM rhodamine 110 solution in isopropanol was spincoated onto the nanostructure arrays. It caused the adhesion of a few monolayers of rhodamine molecules on the sample surface after evaporation of the solvent. The Raman experiments were performed in back-scattering geometry using a Renishaw In Via Raman microscope system equipped with polarization optics. A He-Ne laser with 632.8 nm emission wavelength was used as excitation source. Raman signals were detected with the system operating either in imaging mode or in spectral mode. In the imaging mode, a single pixel of the CCD detector corresponds to about 80 × 80 nm2 sample area when a 160× objective is used. Raman spectra of the nanostructure arrays with needles tilted parallel to the sample surface were acquired with the linear polarization of the excitation laser parallel as well as perpendicular to the needle direction.
imaging unit was set to detect the most prominent Raman band of rhodamine at 1350 cm−1. All three images are not corrected for the Gaussian profile of the intensity distribution within the laser spot. Using a 50× objective, the spot illuminated by the Gaussian intensity profile of the laser is about 15 μm in diameter and covers several nano needle sites. Corresponding Raman images obtained of an array with needle structures with bent needles of the type depicted in figure 1(b) with the linear polarization of the laser parallel and perpendicular to the needle direction are shown in figures 2(a) and (b), respectively. The regular arrangement of the nanostructures is clearly visible in the Raman image, indicating that each nanostructure exhibits an enhanced Raman intensity compared to the sample surroundings. It should also be noted that the intensity enhancement is about a factor of two larger in case of the parallel polarization geometry than for the perpendicular one. This is a first indication of the antenna action of the bent needle. Figure 2(c) shows a Raman image recorded of a single antenna structure of the type depicted in figure 1(a). A 160× objective was used, yielding a width of the Gaussian intensity profile of about 3 μm. The enhancement of the Raman intensity at the position of the nanostructure is clearly visible in the image and confined to an area of a few hundred nm. This demonstrates that hot spots are formed within the individual nanostructures even in the case of upright needles. Of course, individual hot spots on the nanostructures cannot be identified due to the diffraction limit. To quantify the differences caused by Raman enhancement of the nanostructured areas, we obtained spectra with the laser focused onto a nanostructured needle and focused onto the plain sample between nanostructures, respectively. Typical spectra (red) obtained of the needle structures of the different types depicted in figures 1(a), (b), and (d), respectively, are shown in figure 3, together with the corresponding spectra (black) taken between nanostructures on the plain areas of the samples. Taking into account that the black spectra were acquired with a 10 times higher laser power, it can be estimated that for all three structures the enhancement factors are of the same order of magnitude, about 200, i.e., 210, 215 and 230, respectively, in figures 3(a) to (c). The enhancement of the Raman signal is more than two orders of magnitude
3. SERS effects based on antenna arrays In figure 2, three Raman images are shown. The laser spot is defocussed in such a way that a larger circular spot is illuminated with a Gaussian intensity profile. The spatial distribution of the Raman signal in a specified wavenumber band is imaged onto the CCD. In all cases, the spectral filter of the 3
L Chen et al
Nanotechnology 25 (2014) 265302
Figure 3. Comparison of Raman spectra of rhodamine taken on the unstructured sample surface (black) and different nanostructures (red). (a) Star-like nanostructures with upright needles as shown in figure 1(a), (b) Star-like nanostructures with bent needles as shown in figure 1(b), (c) Nanostructures with bent ball-like needles as shown in figure 1(d). The intensity of the exciting laser was 10 times higher for spectra taken between needle structures compared to spectra taken on needle structures.
Figure 4. Comparison of the Raman spectra of rhodamine measured on bent needles for two polarization geometries, either polarization
parallel to the needle direction (red) or perpendicular to the needle direction (black). (a) Star-like nanostructures with bent needles as shown in figure 1(b), and (b) nanostructures with bent ball-like needles as shown in figure 1(d).
the two polarizations (I|| − I⊥)/(I|| + I⊥) is 30% ± 20%. The comparatively large standard deviation has two main causes. First, despite the good control and reproducibility of the fabrication process of the needle structures, a residual randomness of the nanostructure shape remains and has an impact on the hot-spot formation. Second, several hot spots within the nanostructure consisting of a star-like ridge with bent needles (figure 1(b)) or a deformed dome with bent needles (figure 1(d)) are formed, which contributes to the observed average enhancement within the laser spot. Due to the randomness of the positions and strengths of the hot spots (which are formed at the ridges or within the deformed dome), what may occur is excitation by light polarized perpendicular to the needle direction. This causes a stronger enhancement, despite the antenna effects due to the bent needle. We will come back to this point in section 4. On the other hand, thermal drift in the experiment also has an impact on the experimental error bars, as it difficult to ensure that the measurements for the two polarization directions are indeed performed on exactly the same spot on the sample. We find no significant polarization dependence for the upright needles of the type depicted in figure 1(a), i.e.
compared to the unstructured regions on the sample. It should be noted that the polarization direction with respect to the needle direction was always perpendicular to the needle direction, also in case of the bent needles. It is worth noting that for needle structures of the type depicted in figure 1(c) hardly any enhancement of the Raman signal was observed. In case of the bent needles depicted in figures 1(b) and (d), we have also performed polarization-dependent measurements with the incoming laser light polarized parallel and perpendicular to the needle direction. The corresponding sets of spectra for the two types of needle structures are shown in figures 4(a) and (b), respectively. The Raman signals for the two polarization geometries differ by almost a factor of two. The Raman enhancement is larger if the incoming light is polarized along the needle direction, indicating that the needles act as antennas. The behavior is reproducible and in agreement with the results of the Raman imaging in figures 2(a) and (b). Furthermore, the same results are obtained when turning the polarization direction in the beam path of the excitation light and when rotating the sample by 90 degrees, i.e., the effect is no artifact of the measurement system. The majority of the bent needles (eight out of ten structures) show a stronger enhancement for the incoming light being linearly polarized along the needle direction. On average, the relative difference of the Raman intensities for
(I
∥
− I
) (I
∥
)
− I is 0% ± 20%. The origin of the standard
deviation is the same and has been described above. 4
L Chen et al
Nanotechnology 25 (2014) 265302
its intensity profile in the xy-plane was Gaussian, with a width of λ = 632.8 nm. To model the average Raman enhancement factor of the metal nanostructure within the laser spot, finite element calculations were performed using the software package COMSOL Multiphysics 3.5a. In the calculations, the electric field distributions are derived using the 3D/RF Module/ Electromagnetic waves/Harmonic propagation. The solver is chosen as ‘Parametric’. All boundary settings are chosen as ‘scattering boundary condition’. The maximum element size is set to 2 nm in ‘Subdomain mesh parameters’ to guarantee the accuracy of calculation. The refractive index of gold is chosen as 0.181 + 3.068·i in the vicinity of λ = 632.8 nm [31]. For a given model structure, the following four steps of the calculation were carried out to derive its average Raman enhancement factor and to identify the hot spots of the nanostructure: (1) Calculation of the electric field distribution E0(x, y) of the incident laser beam with a wavelength of 632.8 nm in air. The corresponding values are calculated on a square mesh of 2 μm × 2 μm size with a pitch of 1 nm and the laser beam centered at (x, y) = (1 μm, 1 μm). The data (E0(x, y)4) are stored as a data array which is independent of z. (2) Calculation of the local electric field distribution E (x, y, z) when the laser impinges on a star-like ridge structure with a needle in its center. The laser spot and the nanostructure illuminated are both centered at ( x, y ) = ( 1 μm, 1 μm). The corresponding values are again calculated on a mesh of 2 μm × 2 μm size with a 1 nm pitch. The data (E (x, y, z)4) are stored as a data array which depends on z. (3) Derivation of the local Raman enhancement factors as g ( x, y, z ) = E ( x, y, z )4 E0 ( x, y )4 . Typical maximal values g are in the range of 104–109. (4) Derivation of the average Raman enhancement factor by averaging across the beam profile in the xy-plane and along the z-direction of the nanostructure:
Figure 5. Computing models for star-like ridge structures with an
upright needle (a) or a horizontally tilted needle (c) based on the SEM images of the corresponding nanostructures (b) and (d). The radially arranged ridges and the needle base are characterized by the structural parameters a, b, c, d, e and N the number of ridges. The needle itself is characterized by the needle length L and the tilting angle α.
4. Modeling of the SERS effect of star-like needle structures To understand the origin of the Raman enhancement and the effect of the nanostructure morphology on its magnitude, we modelled exemplarily the field enhancement of the star-like ridge structures with needles. For this purpose, the morphology of the star-like needle structure is approximated by an upright (or tilted) needle surrounded by a set of radially arranged ridges. The corresponding simplified models are shown in figures 5(a) and (c), and are based on the SEM images in figures 5(b) and (d). The radially arranged ridges and the needle base are characterized by the structural parameters a, b, c, d, e and N, the number of ridges. The needle itself is characterized by the needle length L and the tilting angle α defining the gap between needle tip and ridges. The crucial parameters determining the magnitude of the Raman enhancement are the gap b between the needle base and ridge edge, the number of ridges N, the needle length L and the tilting angle α. For simplicity, the other structural parameters were kept constant at values corresponding to the SEM images: ridge length c = 250 nm, radius of the ridge on the outside e = 40 nm, inner radius of the ridge d = 10 nm, and needle diameter at the base a = 130 nm. The scattering geometry of the Raman experiment can be explained by the following. The probe light impinging on the metal nanostructure travelled along the z-direction (perpendicular to the sample plane). The light was linearly polarized either along x or y (the xy-plane being the sample plane), and
g =
∑z
∑ g ( x , y , z ) E 0( x , y ) 4 d A ∑ E 0( x , y ) 4 d A
(1)
where the sums in numerator and denominator are in the xyplane, dA = 1 nm2. Typical values derived for are in the range of 10–104. As an illustration of typical results, figure 6 depicts the normalized local intensity distributions E (x, y, z)4 of the incoming light field for cross sections in the xy-plane and the yz-plane for both types of needle structures, i.e. with the needle pointing up and with the needle being tilted horizontally, respectively. In both cases the incoming light is linearly polarized along the y-direction, as denoted by the red arrows in figure 6. It can be seen from the cross sections in the xyplane that hot spots occur at gaps between needle base and ridges, when the polarization direction is parallel to the corresponding radial orientation of the ridge, i.e., for ridges pointing in y-direction or the ridges closest to the y-direction and not screened by the needle when tilted. Thus, in case of a 5
L Chen et al
Nanotechnology 25 (2014) 265302
of the gaps formed and the number of the gaps is part of the explanation of the significant variation of the SERS properties observed in section 3. Marginal differences in the morphology of needle structures, such as a variation of the gap by a nanometer, may occur, although these structures are prepared using the same process parameters. Furthermore, we have performed calculations for different geometries in terms of needle orientation and polarization direction of the excitation light with respect to the needle orientation for a constant number of ridges N = 8. If a needle oriented horizontally is excited with the linear polarization parallel to the needle direction, the average Raman enhancement factor is larger than in case of excitation with polarization perpendicular to the needle direction. This confirms the antenna action of a horizontal needle when excited with light polarized along the needle direction. Both other geometries where the electric field of the light is perpendicular to the needle direction, i.e., an upright needle pointing in z-direction with the electric field in the xy-plane or a horizontal needle pointing in x-direction with the electric field along y, show basically the same average Raman enhancement factor. This indicates that the needle has hardly any additional impact in these cases apart from its base forming the gaps with the ridges. The difference between the Raman enhancement factors for incoming light polarized perpendicular and parallel to needle direction decreases when the tilting angle α of the needle increases, i.e., the curve for parallel polarization (purple curve with triangles) would approach the curve for perpendicular polarization (dashed green line with crosses). However, this difference between parallel and perpendicular polarization not only depends on the gap formed between needle tip and ridges, but even more severely on needle length, as can be seen in figure 7(b). Figure 7(b) shows the dependence of the average Raman enhancement factor on needle length L for a nanostructure with a needle tilted horizontally. Both the number of ridges and the gap width were kept constant, N = 8 and b = 15 nm. Calculations were performed for two excitation geometries, exciting with light polarized parallel (blue curve, full circles) and perpendicular (red curve, open circles) to the needle direction. An oscillatory dependence of on needle length with a decaying envelope function is found in the parallel case. The maxima probably correspond to antenna resonances of different order, i.e. multipolar plasmon excitations [32]. Of course, this strong dependence on needle length also contributes to the variation of the Raman enhancement observed for needle structures prepared using the same process parameters. However, if the needle length can be well controlled in the fabrication process, the average Raman enhancement factor can be maximized for this particular excitation geometry. This can be accomplished by optimizing the antenna action of the needle, i.e. tuning its length in resonance with the incoming light wavelength. This finding is further corroborated by two insets of figure 7(b) depicting two cross sections of the yz-plane for needle lengths, where is minimal and maximal, respectively. In contrast, an almost flat curve is observed in the perpendicular case, which is always lower than in the parallel case confirming our experimental
Figure 6. Modeling of the hot spots for needle structures with (a)
upright needles (α = 90°) and (b) horizontally tilted needles (α = 0°). N = 12, L = 850 nm, b = 10 nm. Plots of |E (x, y, z)|4 in the xy-plane and in the yz-plane are for z0 = 1 μm and x0 = 1 μm, respectively. The red arrows denote the polarization orientation of probe light.
horizontally tilted needle, the average Raman enhancement factor depends on the orientation of the polarization with respect to the needle direction in agreement with the experimental findings. The cross section in the yz-plane for the structure with the upright needle reveals that the needle has no additional impact on the average Raman enhancement factor apart from forming gaps to the ridges. Figure 7(a) shows the dependence of the average Raman enhancement factor on the gap width b between the base of the needle and the ridges. The length of the needle was kept constant, L = 250 nm. Three curves are calculated for different numbers of ridges surrounding an upright needle radially, i.e. N = 4, 8, 12. As expected, the Raman enhancement increases with the number of ridges N. It can also be seen that the average Raman enhancement factor decreases rapidly with increasing gap size b between needle base and ridges for all N. For example, the relative change of the Raman enhancement factor with b in case for N = 12 ridges in the range between b = 5 and 10 nm is about 20% per nm. This sensitivity of the average enhancement on the width 6
L Chen et al
Nanotechnology 25 (2014) 265302
Figure 7. (a) Dependence of the average Raman enhancement factor on the gap width b between the base of the needle and the ridges. Length of the needle L = 250 nm. The number of ridges, the orientation of the needle, as well as the excitation geometry are given in the legend. (b) Dependence of the average Raman enhancement factor on needle length L for a nanostructure with needle tilted horizontally and excited with light polarized parallel (blue line)/perpendicular (red line) to the needle direction. The number of ridges is N = 8, and the gap width is b = 15 nm.
findings. However, it should be noted here again that the behavior of two curves for parallel and perpendicular polarization isvery sensitive to the tilting angle of the needle. The tilting angle α changes the widths of the gaps between the needle and ridges where hot spots occur and, thus, the corresponding contributions to the Raman enhancement. If the tilting angle is increasing by a few degrees away from α = 0, i.e., the gap between tip and ridge becomes wider, a reversal of the order of the curves in figure 7(b) may occur for long needles. The reason is that the effect of shading by the needle becomes more important than the additional Raman enhancement at the gap between tip and ridge. Thus, the Raman enhancement and the polarization effects are also very sensitive to fluctuations of the needle orientation, which explains the statistical variations in the polarization effects observed for different individual needle structures prepared using the same process parameters. Nevertheless, if a high degree of control of tilting angle and length of the needle can be achieved in the fabrication process, a much better performance in terms of Raman enhancement will be expected for bent-needle structures than for upright-needle structures. Finally, we would like to briefly discuss the Raman enhancement of the structures with upright needles on intact semispherical domes (figure 1(c)) and the corresponding structures with needles tilted horizontally (figure 1(d)). The former do not show a significant Raman enhancement, as no ridges and gaps are present and the antenna effect of an upright needle is negligible. In case of the corresponding structures with needles tilted horizontally (figure 1(d)) the situation is different, as the dome is deformed and edges occur when the needles are mechanically bent. The folds and bends which are present in this type of needle structure are not aligned in the regular star-like fashion as are the structures depicted in (figures 1(a) and (b)). However, hot spots are formed in the gaps and edges caused by the deformation of the dome in the bending process of the needle. We have not modelled this type of structure in more detail. However, the
results are expected to be essentially the same as for the needle structures with the star-like ridges.
5. Conclusions We have analyzed the SERS effect of nanostructures composed of star-like ridge structures with needles in the center fabricated by femtosecond laser pulses. The average Raman enhancement factors can be explained quantitatively in terms of their behavior on nanostructure morphology and excitation geometry. In case of the horizontally tilted needles the average Raman enhancement factor will depend on the orientation of the polarization with respect to the needle direction. For polarization of the incoming light parallel to the needle direction, the average Raman enhancement factor can be maximized when the antenna action of the needle is optimized by tuning its length in resonance with the incoming light wavelength. In contrast, upright needles have no additional impact on the average Raman enhancement factor apart from forming gaps to the ridges. A quantitative agreement between experimentally determined average Raman enhancement factors and their theoretical counterparts can be obtained by a simplified theoretical model. The model also indicates that the SERS effects of such needle structures are very sensitive to minor fluctuations in the morphology, which explains the experimentally observed variations of the SERS activity for almost identical nanostructures. Nevertheless, the SERS enhancement averaged over the area of the gold nanostructure is reproducibly two orders of magnitude larger than on the plain gold film. Therefore, the fast and low-cost laser processing of gold films presented here offers an interesting alternative to accurately define SERS-active areas on the micrometer scale as part of integrated optical devices. 7
L Chen et al
Nanotechnology 25 (2014) 265302
Acknowledgements
[15] Albrecht M G and Creighton J A 1977 J. Am. Chem. Soc. 99 5215 [16] Lindquist N C, Nagpal P, McPeak K M, Norris D J and Oh S-H 2012 Rep. Prog. Phys. 75 036501 [17] Beermann J, Novikov S M and Leosson K Bozhevolnyi S I 2009 Optics Express 17 12698 [18] Hou Y, Xu J, Wang P and Yu D 2010 Appl. Phys. Lett. 96 203107 [19] Bahns J T, Imre A, Vlasko-Vlasov V K, Pearson J, Hiller J M, Chen L H and Welp U 2007 Appl. Phys. Lett. 91 081104 [20] Chen G, Wang Y, Yang M, Xu J, Goh S J, Pang M and Chen H 2010 J. Am. Chem. Soc. 132 3644 [21] Schuck P J, Fromm D P, Sundaramurthy A, Kino G S and Moerner W E 2009 Phys. Rev. Lett. 94 017402 [22] Kravets V G, Zoriniants G, Burrows C P, Schredin F, Casiraghi C, Klar P, Geim A K, Barnes W L and Grigorenko A N 2010 Phys. Rev. Lett. 105 246806 [23] Behr N and Raschke M B 2008 J. Phys. Chem. C 112 3766 [24] Krug J T 2nd, Sanchez E J and Xie X S 2002 J. Chem. Phys. 116 10895 [25] Crozier K B, Sundaramurthy A, Kino G S and Quate C F 2003 J. Appl. Phys. 94 4632 [26] Geshev P I, Fischer U and Fuchs H 2010 Phys. Rev. B 81 125441 [27] Mullin J, Valley N, Blaber M G and Schatz G C 2012 Phys. Chem. A 116 95774 [28] Unger C, Koch J, Overmeyer L and Chichkov B N 2012 Opt. Express 20 24864 [29] Koch J, Korte F, Bauer T, Fallnich C, Ostendorf A and Chichkov B N 2005 Appl. Phys. A 81 325 [30] Kuznetsov A I, Koch J and Chichkov B N 2009 Appl. Phys. A 94 221 [31] Johnson P B and Christy R W 1972 Phys. Rev. B 6 4370 [32] Krenn J R, Schider G, Rechberger W, Lamprecht B, Leitner A, Aussenegg F R and Weeber J C 2000 Appl. Phys. Lett. 77 3379
The authors from Giessen and Beijing thank the German Exchange Service (DAAD) and the Chinese Science Council for supporting this work in the course of their bilateral exchange programme (PPP). We thank Arseniy Kuznetsov for technical assistance.
References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14]
Moskovits M 1985 Rev. Mod. Phys. 57 783 Hartschuh A 2008 Ang. Chem. Int. Ed. 47 8178 Pettinger B 2010 Molecular Phys. 108 2039 Bailo E and Deckert V 2008 Chem. Soc. Rev. 37 921 Stiles P L, Dieringer J A, Shah N C and Van Duyne R P 2008 Ann. Rev. Anal. Chem. 1 601 Dieringer J A, Wustholz K L, Masiello D J, Camden J P, Kleinman S L, Schatz G C and Van Duyne R P 2009 J. Am. Chem. Soc. 131 849 Nie S M and Emory S R 1977 Science 275 1102 Kneipp K, Wang Y, Kneipp H, Perelman L T, Itzkan I, Dasari R and Feld M S 1997 Phys. Rev. Lett. 78 1667 McFarland A D, Young M A, Dieringer J A and Van Duyne R P 2005 J. Phys. Chem. B 109 11279 Neacsu C C, Berweger S and Raschke M B 2007 Nanobiotechnol. 3 172 Haynes C L and Van Duyne R P 2003 J. Phys. Chem. B 107 7426 Liu Y, Yu C and Sheu S 2006 J. Mater. Chem. 16 3546 Zhang X, Liu H and Feng S 2009 Nanotechnology 20 425303 Jeanmarie D L and Van Duyne R P 1977 J. Electroanal. Chem. 84 1
8
