Hot-Spot Engineering in Polygonal Nanofinger Assemblies for Surface

LETTER pubs.acs.org/NanoLett

Hot-Spot Engineering in Polygonal Nanofinger Assemblies for Surface Enhanced Raman Spectroscopy Fung Suong Ou,† Min Hu,† Ivan Naumov,† Ansoon Kim,† Wei Wu,‡ Alexander M. Bratkovsky,† Xuema Li,† R. Stanley Williams,‡ and Zhiyong Li*,† †

Intelligent Infrastructure Lab and ‡NanoElectronics Research Group, Hewlett-Packard Laboratories, 1501 Page Mill Road, Palo Alto, California 94304, United States ABSTRACT: Multiparticle assemblies of nanoscale structures are the fundamental building blocks for powerful plasmonic devices. Here we show the controlled formation of polygonal metal nanostructure assemblies, including digon, trigon, tetragon, pentagon, and hexagon arrays, which were formed on top of predefined flexible polymer pillars that undergo self-coalescence, analogous to finger closing, with the aid of microcapillary forces. This hybrid approach of combining top-down fabrication with self-assembly enables the formation of complex nanoplasmonic structures with sub-nanometer gaps between gold nanoparticles. On comparison of the polygon-shaped assemblies, the symmetry dependence of the nanoplasmonic structures was determined for application to surface enhanced Raman spectroscopy (SERS), with the pentagonal assembly having the largest Raman enhancement for the tested molecules. Electromagnetic simulations of the polygonal structures were performed to visualize the field enhancements of the hot spots so as to guide the rational design of optimal SERS structures. KEYWORDS: Plasmonics, SERS, nanofingers, nanoimprint lithography, gold nanocluster

T

he assembly of subwavelength, or nanoscale, metallic structures into particular geometries can have profound impact because such metal aggregates can support localized surface plasmons.1,2 Plasmons in the metal assemblies provide very intense local fields that enable applications such as plasmonic waveguides,3 biochemical sensors,4,5and surface enhanced Raman spectroscopy (SERS)6 because of extremely large dielectric contrast with an environment. Precise control of both the shape and geometrical symmetry of the assemblies has been an active focus of many research efforts. For example, by using either topdown or bottom-up approaches, nanoparticle dimers7
9 and trimers10 have been studied extensively. However, each of the two approaches has certain limitations. For example, the topdown approach allows one to introduce symmetry and gap size control at certain scale, but sub-5-nm gap sizes have been elusive. On the other hand, the bottom-up approach can achieve small gap sizes, but it is difficult to achieve highly uniform structures of arbitrary symmetries across a large area, such as tetramers, pentamers, or heptamers, for systematic study. Yet, the complex structures such as these have extremely interesting photonic properties; for example, it has been recently shown that when nanoparticles or nanopillars are assembled in a heptamer, one can observe a Fano-like resonance.11
14 It is therefore extremely interesting to study the plasmonic properties of these more complicated assemblies and also examine their performance for various applications, such as SERS sensing. Despite a significant amount of progress in the exploration of dimers,7
9 few extensive studies have been performed on the higher order assemblies because of the difficulty in fabrication. In this Letter, r 2011 American Chemical Society

we present a systematic study of the plasmonic properties of regular polygons, including 2-mer (digon), 3-mer (trigon), 4-mer (tetragon), 5-mer (pentagon), and 7-mer (hexagon arrangement with the seventh particle in the center), and the SERS signal that arises from them. We provide a controlled method to generate these structures, which is based on the nanoimprinting technique described previously.15 We also present simulations of the interactions of the structures with light in order to understand and correlate them with the experimental observations, which ultimately can guide the rational design of the optimal SERS structures. Figure 1 shows schematic illustrations of the nanostructure assemblies we studied. Each circle as shown in Figure 1 can represent a nanoscale metal sphere, disk, or cylinder, with a small separation between neighboring circles. For simplicity of simulation, we chose a nanosphere to model each nanoparticle. The small separation ensures a strong coupling between the neighboring metal nanospheres so that the electromagnetic near-field is greatly enhanced.16,17 The symmetries of the metal nanostructures lead to interesting plasmonic modes, in analogy with molecular-orbital hybridization in conjugated systems.8 Since the formation of such nanostructure assemblies poses significant challenges for the fabrication process, a systematic experimental study for all the symmetries shown in Figure 1 is not readily available in the literature. Here, well-controlled polygon-shaped Received: April 12, 2011 Revised: May 13, 2011 Published: May 23, 2011 2538

dx.doi.org/10.1021/nl201212n | Nano Lett. 2011, 11, 2538–2542

Nano Letters

LETTER

Figure 1. Schematic drawings of the polygonal assemblies using nanospheres for the digon (2-mer), trigon (3-mer), tetragon (4-mer), pentagon (5-mer), and hexagon (7-mer) shapes.

gold nanostructures with various symmetries were formed on top of predefined flexible polymer fingers, which then self-assembled with the aid of microcapillary forces. The experimental procedure for the fabrication of the goldcoated nanofingers is described schematically in Figure 2. First, a silicon mold with pillar arrays of the various symmetries in Figure 1 was formed by a combination of e-beam lithography and dry Si etching as described previously.15 Then, as shown in Figure 2b, UV-curable nanoimprint lithography (NIL) was utilized to transfer the Si pillar pattern to a polymeric reversetone mold using a custom-designed nanoimprint machine.18 The polymeric reverse-tone mold was used in a subsequent NIL step (Figure 2c) to form the final polymer nanofingers shown in Figure 2d, with the procedure similar to that previously reported.19,20 Finally, Au with nominal thickness of 50
80 nm was deposited on the sample by e-beam evaporation at normal incidence to form the Au nanoparticles on the tips of the polymer fingers, as shown in Figure 2e. After the arrays were exposed to solvent and air-dried, the fingers closed together in the designated symmetry, as shown in Figure 2f. The capillary force drives the coalescence of the nanofingers; a similar phenomenon was observed in high aspect ratio microscale structures.21
23 Since the patterns in our case can be precisely defined by the initial e-beam lithography and reproduced faithfully by NIL, an array with any arbitrary unit cell can be fabricated uniformly with this method over a large area. Figure 3 shows the top and side view scanning electron microscopy (SEM) images of different polygonal structures after closing. The fingers for all geometries were ca. 520 nm tall before Au deposition. The average diameter of the metal particles on the fingertips was about 136 nm. The periods of the 2-mer, 3-mer, and 4-mer arrays were 500 nm, while the periods of the 5-mer and the 7-mer arrays were 700 nm. Since the nanoparticles self-assembled during finger closing, this process can be used to trap molecules between adjacent

Figure 2. Fabrication procedure for the nanofingers: (a) The fabrication of the nanofinger silicon mold using e-beam lithography. (b) The making of the daughter mold using nanoimprinting, (c, d) The fabrication of the polymer nanofingers from the polymer daughter mold using nanoimprinting. (e) e-beam deposition of 80 nm Au onto the nanofingers. (f) Soaking and drying of the solvent to induce the closure of the nanofingers.

finger tips with sub-nanometer spacing dictated by the size of the trapped molecules.24 These sites are also potential “hot spots” to enable ultrahigh Raman scattering enhancement from the molecules residing there,15 which is extremely important for SERS sensing applications. The 2-mer has just one touching point, and thus the ratio of potential hot spots to nanofingers is 1:2. For the 3-mer, square 4-mer, and 5-mer, the ratio of touching points to nanofingers is 1:1, and thus these structures might be expected be more efficient for SERS than 2-mers and roughly equal to one another. Finally, the ratio of touching points to nanofingers is 12:7 for the 7-mer, which might (naively) be expected to have the most intense SERS signal. In order to study the symmetry dependence of the nanofinger assemblies for SERS, we chose a model molecule, trans-1,2-bis (4-pyridyl)-ethylene (BPE), for our experiments. Either a standard upright confocal Raman microscope (Jobin-Yvon T64000) or a custom-built microRaman system (Bayspec) were employed for the Raman measurements. All spectra were collected with the micro-Raman setup either through a 100 (N.A. 0.9) objective with the illumination of 633 and 785 nm light or through a 50 IR corrected objective (N.A. 0.65) with 1064 nm illumination. The optical power at the sample was about 30 μW for both the 633 and 785 nm lasers and was about 20 mW for the 1064 nm 2539

dx.doi.org/10.1021/nl201212n |Nano Lett. 2011, 11, 2538–2542

Nano Letters

LETTER

Figure 4. (a) Normalized Raman spectra of BPE excited at 785 nm for the digon, trigon, tetragon, pentagon, and hexagon nanofingers. (b) Comparison of the normalized intensity of the Raman signal at 1600 cm
1measured for 633, 785, and 1064 nm incident radiation.

Figure 3. SEM image showing the top view and the side view (45° tilt from surface normal) of the closed nanofinger assemblies for (a, b) digon, (c, d) trigon, (e, f) tetragon, (g, h) pentagon, and (i, j) hexagonshaped 7-mer. Scale bars in the SEM images are 200 nm.

laser. The integration time was 1 s for the entire spectrum. To test the performance of the SERS substrate, we soaked the test substrates in 1 mM BPE in ethanol for 10 min. Then the substrates were washed with copious amounts of ethanol to remove the physisorbed BPE molecules and subsequently dried in a nitrogen stream. Figure 4a shows the Raman spectra of BPE excited at 785 nm when using 2-mer, 3-mer, 4-mer, 5-mer, 7-mer, and uniform arrays as the control for the SERS substrates. The nanofingers on the control sample formed random assemblies after the closing process.15 All the substrates generated qualitatively the same sets

of Raman peaks with similar peak intensity ratios from the test molecule, BPE, suggesting that there is no significant difference in terms of the enhancement mechanism from the various symmetries finger assemblies. However, on comparison of the intensity of the Raman peaks from the different symmetries, it is clear that 5-mers outperformed the other symmetry designs. Note the intensity of the Raman signal was normalized against the number of pillars per unit area; hence the figure shows the intensity/finger in the given designs. Figure 4b shows the comparison of the normalized intensity of the Raman signal at 1600 cm
1 measured at three different laser wavelengths. As in the case of using the laser at 785 nm, the 5-mers outperformed the rest of the symmetries when excited using laser wavelengths of 633 and 1064 nm as well. In our previous paper,15 we calculated the enhancement factor (EF) for the BPE molecules trapped between the nanofinger tips in the random nanofinger assemblies (dot array sample shown in Figure 4b) to be 2  1010 at 785 nm illumination. In this study, we can estimate the enhancement factor to be ∼1011 at 785 nm illumination for the same BPE molecules trapped in the 5-mers, a further improvement of about a factor of 6 when compared to the random nanofinger assemblies (dot array sample in Figure 4b). The dominant SERS intensity of the 5-mer was initially surprising, since one would intuitively expect that the 7-mer, with nearly two touching points per nanofinger, would be the most efficient structure. However, our measurements revealed that the 2540

dx.doi.org/10.1021/nl201212n |Nano Lett. 2011, 11, 2538–2542

Nano Letters SERS intensity from 7-mers was even lower than that for 2-mers, and there were considerable and consistent differences among the 3-mers, 4-mers, and 5-mers. Thus, the local symmetry is a critical factor in the plasmonic behavior of the nanoparticle polygons. In order to analyze the underlying physics of the observed enhancements, simulations were performed using the discretedipole approximation (DDA),25,26 which is applicable to any arbitrary particle shape and configuration of particles. In this approach, the object of interest was represented as a set of polarizable spheres on a cubic grid with a lattice period a = 1 nm, and we assumed that all the Au nanospheres were identical to each other and had the same diameter of 136 nm to match the physical measurement of the structures in the SEM images. The separation between adjacent spheres in all clusters was 1.5 nm, which was chosen more to ensure a reliable simulation rather than to mimic the actual gap size, which was too small to measure reliably. Each elemental sphere felt the field of the incident beam Einc and the fields generated by all other spheres, as described by the system of linear equations: pj = RjEj, where the field is the sum of an incident field and the field produced by all the dipoles in the system with the exception of the jth dipole.25,26 The polarizability is given by the standard Lorentz
Lorenz expres
1 = r
3
2ik3/3 for a sphere with sion R
1 i i (εi þ 2)(εi
1) radius ri, with the last complex term giving the radiative correction.27 The dielectric constants were chosen using empirical data for Au for different wavelengths.28 The distributions of the electric field intensity |E(r)|2 for the various symmetries corresponding to 785 nm illumination laser wavelength at normal incidence along with the representative SEM images of individual polygons are shown in Figure 5. The “hot spots” or maxima in |E(r)|2 are located precisely in the gaps between adjacent spheres and characterized by the peak value of the nearfield enhancement factors |E|/|E0| shown in the figure. The field intensity maps from the DDA calculation shown in Figure 5 clearly indicated that the symmetry of the polygonal nanosphere aggregates determined the location of the hot spots; many of the touching points between adjacent spheres were “dark”. Among all the symmetries studied, the 2-mer had the highest field enhancement (|E|/|E0| ∼ 180) in the gap when the incident field was parallel to the axis of the dimer; however, under the orthogonal incident field, the same gap showed much smaller field enhancement (|E|/|E0| ∼ 10). Thus, when considering both the facts that there was only one touching point for the 2-mer and it only lit up strongly for parallel polarization, this structure had a relatively weak SERS enhancement. But how do we understand the results for the other symmetries? Nordlander and co-workers have analyzed the group theoretic properties of the plasmons for the 3-mer through the 7-mer structures.10,29,30 We can qualitatively understand the experimental results and the DDA field-intensity calculations from the symmetry-adapted basis functions for the plasmons, based on two in-plane dipoles for each nanoparticle in the structure, and their irreducible representations. Although there are twice as many in-plane basis functions as nanoparticles for each polygon, only a subset of those for each structure, belonging to a particular irreducible representation for the space group of the polygon, have the correct symmetry (2Eu) to couple to the in-plane polarization vector (electric field) of the incident radiation. Within the set of basis functions that are dipole active, some of those may have nodes that pass between adjacent pairs of nanoparticles in the structure, and thus those potential hot spots are symmetry forbidden to “light up”. Both Figure 3 of ref 30 and

LETTER

Figure 5. (a
e) Distributions of electric field intensity |E|2 at 785 nm in the central plane of all the polygon shaped assemblies shown in Figure 1 for the two different in-plane electric field polarizations illustrated on the top panel. Scale bar in SEM is 200 nm.

Figure 5e of this paper show that for the heptamer all of the potential hot spots that exist between the central particle and the six surrounding particles are dark and thus do not contribute to Raman enhancement of trapped molecules. Furthermore, because of the high symmetry of the heptamer and the requirement that the basis functions that couple to electric dipolar excitation must have dipole character, the extent of the in-phase overlap of the dipoles between adjacent particles on the ring of the heptamer is limited. This gives rise to the relatively weak field enhancements calculated in Figure 5e and corresponding SERS intensities in Figure 4; the 7-mer with its close-packed structure is too symmetric to couple efficiently to normally incident radiation. 2541

dx.doi.org/10.1021/nl201212n |Nano Lett. 2011, 11, 2538–2542

Nano Letters In contrast, for the lower symmetry trigon and pentagon structures in panels b and d of Figure 5b and 5d, the basis functions of the dipole-allowed plasmon irreducible representations have significant overlap between nearly every pair of adjacent nanoparticles in the structures for both orientations of the polarization vector. This leads to the stronger plasmon excitations for these structures, both of which lack inversion symmetry, compared to the other polygons. Of course, group theory alone cannot be used to predict quantitative field enhancement differences between the various polygons. That requires a detailed understanding of the material properties and structural details of the polygons and the wavelength of the incident light. For the 7-mer, there is the additional issue that structures with a central atom inside a polygon may exhibit a Fano-like resonance,11,13,30 which adds a level of complexity to the spectral response of the structure. At certain wavelengths, there can be a strong antiresonance between modes described by different basis functions of the same symmetry, with local dipole modes pointing in mostly opposite directions with an apparently reduced local field for the 7-mer and thus make it even less desirable for SERS. In this Letter, we demonstrated a simple and scalable method to produce Au nanoparticles on flexible nanofingers that can come into nanometer proximity at a predefined symmetry. This high precision for assembling nanoparticles opens a new path for the design and fabrication of arbitrary geometries of nanostructures. We also showed that pentagon nanofinger arrays outperformed other polygon symmetries studied in this work for SERS applications. By examining DDA simulations of the gold nanosphere assemblies, we found that the polygon symmetry determined the number and the strength of the potential plasmonic hot spots between adjacent Au nanoparticles that actually light up under incident optical excitation and thus contribute to SERS. The 5-mer, which had no inversion symmetry, provided a set of plasmon basis functions that both coupled to dipolar excitation and provided significant in-phase field overlap between adjacent nanoparticles. This suggests a means for further engineering multiparticle plasmon resonances by exploring various broken symmetry structures.

’ AUTHOR INFORMATION Corresponding Author

*E-mail: [email protected].

LETTER

(6) Moskovits, M. J. Raman Spectrosc. 2005, 36, 485–496. (7) Brown, L.; Sobhani, H.; Lassiter, J.; Nordlander, P.; Halas, N. ACS Nano 2010, 4, 819–832. (8) Sonnichsen, C.; Reinhard, B.; Liphardt, J.; Alivisatos, A. Nat. Biotechnol. 2005, 23, 741–745. (9) Nordlander, P.; Oubre, C.; Prodan, E.; Li, K.; Stockman, M. I. Nano Lett. 2004, 4, 899–903. (10) Brandl, D. W.; Mirin, N. A.; Norlander, P. J. Phys. Chem. B 2006, 110, 12302–12310. (11) Fan, J. A.; Wu, C. H.; Bao, K.; Bao, J. M.; Bardhan, R.; Halas, N. J.; Manoharan, V. N.; Nordlander, P.; Shvets, G.; Capasso, F. Science 2010, 328, 1135–1138. (12) Verellen, N.; Sonnefraud, Y.; Sobhani, H.; Hao, F.; Vandenbosch, G. A. E.; Moshchalkov, V. V.; Van Dorpe, P.; Nordlander, P.; Maier, S. ACS Nano 2010, 4, 1664–1670. (13) Hentschel, M.; Saliba.; Vogelgesang, R.; Giessen, H.; Alivisatos, A. P.; Liu, N. Nano Lett. 2010, 10, 2721–2726. (14) Fan, J.; Bao, K.; Wu, C.; Bao, J.; Bardhan, R.; Halas, N.; Manoharan, V.; Shvets, G.; Nordlander, P.; Capasso, F. Nano Lett. 2010, 10, 4680–4685. (15) Hu, M.; Ou, F. S.; Wu, W.; Naumov, I.; Li, X.; Bratkovsky, A.; Williams, R. S.; Li, Z. J. Am. Chem. Soc. 2010, 132, 12820–12822. (16) Jiang, J.; Bosnick, K.; Maillard, M.; Brus, L. J. Phys. Chem. B 2003, 107, 9964–9972. (17) Camden, J. P.; Dieringer, J. A.; Wang, Y.; Masiello, D. J.; Marks, L. D.; Schatz, G. C.; Van Duyne, R. P. J. Am. Chem. Soc. 2008, 130, 12616–12617. (18) Wu, W.; Jung, G. Y.; Olynick, D. L.; Straznicky, J.; Li, Z.; Li, X.; Ohlberg, D. A. A.; Chen, Y.; Wang, S. Y.; Liddle, J. A.; Tong, W. M.; Williams, R. S. Appl. Phys. A: Mater. Sci. Process. 2005, 80, 1173–1178. (19) Ge, H. X.; Wu, W.; Li, Z.; Jung, G. Y.; Olynick, D.; Chen, Y. F.; Liddle, J. A.; Wang, S. Y.; Williams, R. S. Nano Lett. 2005, 5, 179–182. (20) Wu, W.; Hu, M.; Ou, F. S.; Li, Z.; Williams, R. S. Nanotechnology 2010, 21, 25502. (21) Chandra, D.; Yang, S. Langmuir 2009, 25, 10430–10434. (22) Kotera, M.; Ochiai, N. Microelectron. Eng. 2005, 78
79, 515–520. (23) Chini, S. F.; Amirfazli, A. Langmuir 2010, 26, 13707–13714. (24) Kim, A.; Ou, F. S.; Hu, M.; Ohlberg, D. A. A.; Williams, R. S.; Li, Z. J. Am. Chem. Soc. 2011, 133, 8234–8239. (25) Drain, B.; Flatau, P. J. Opt. Soc. Am. A 1994, 11, 1491–1499. (26) Purcell, E.; Pennypacker, C. Astrophys. J. 1973, 186, 705–714. (27) Wokaun, A.; Gordon, J.; Liao, P. Phys. Rev. Lett. 1982, 48, 957–960. (28) Palik, E. D. Handbook of Optical Constants of Solids; Academic Press: New York, 1985. (29) Bao, K; Mirin, N. A.; Nordlander, P. Appl. Phys. A: Mater. Sci. Process. 2010, 100, 333–339. (30) Mirin, N. A.; Bao, K.; Nordlander, P. J. Phys. Chem. A 2009, 113, 4028–4034.

’ ACKNOWLEDGMENT We thank Michael Stuke for helpful discussions. This work was partially supported by DARPA. The views expressed are those of the author and do not reflect the official policy or position of the Department of Defense or the U.S. Government. ’ REFERENCES (1) Barnes, W.; Dereux, A.; Ebbesen, T. W. Nature 2003, 424, 824–830. (2) Maier, S. A. Plasmonics: Fundamentals and Applications; Springer: Berlin, 2007. (3) Brongersma, M. L.; Hartman, J. W.; Atwater, H. A. Phys. Rev. B 2000, 62, R16356–R16359. (4) Lal, S.; Link, S.; Halas, N. J. Nat. Photonics 2007, 1, 641–648. (5) Anker, J.; Hall, W. P.; Lyandres, O.; Shah, N. C.; Zhao, J.; Van Duyne, R. P. Nat. Mater. 2008, 7, 442–453. 2542

dx.doi.org/10.1021/nl201212n |Nano Lett. 2011, 11, 2538–2542