Pesquisar no resumo do documento
DOI: 10.1021/la9020614 13685Langmuir 2009, 25(23), 13685–13693 Published on Web 10/15/2009
© 2009 American Chemical Society
Self-Assembled Plasmonic Nanohole Arrays
Si Hoon Lee,†,^ Kyle C. Bantz,‡,^ Nathan C. Lindquist,§ Sang-Hyun Oh,*,†,§ and Christy L. Haynes*,‡
†Department of Biomedical Engineering, 312 Church Street SE and ‡Department of Chemistry, 207 Pleasant Street SE and §Department of Electrical and Computer Engineering, 200 Union Street SE, University of Minnesota, Twin Cities, Minneapolis, Minnesota 55455. ^These authors contributed equally to this work.
Received June 9, 2009. Revised Manuscript Received September 4, 2009
We present a simple and massively parallel nanofabrication technique to produce self-assembled periodic nanohole arrays over a millimeter-sized area of metallic film, with a tunable hole shape, diameter, and periodicity. Using this method, 30 30 μm2 defect-free areas of 300 nm diameter or smaller holes were obtained in silver; this area threshold is critical because it is larger than the visible wavelength propagation length of surface plasmon waves (∼27 μm) in the silver film. Measured optical transmission spectra show highly homogeneous characteristics across the millimeter-size patterned area, and they are in good agreement with FDTD simulations. The simulations also reveal intense electric fields concentrated near the air/silver interface, which was used for surface-enhanced Raman spectroscopy (SERS). Enhancement factors (EFs) measured with different hole shape and excitation wavelengths on the self-assembled nanohole arrayswere 104-106.With an additionalAg electroless plating step, theEFwas further increasedup to 3 106. The periodic nanohole arrays produced using this tunable self-assembly method show great promise as inexpensive SERS substrates as well as surface plasmon resonance biosensing platforms.
Since Ebbesen and co-workers discovered extraordinary opti- cal transmission (EOT) through subwavelength noble metal nanohole arrays1,2 there has been significant effort to fabricate nanohole arrays with well-controlled electromagnetic properties. Surface plasmon (SP) waves;collective oscillations of conduction electrons;contribute directly to the EOT effect,3,4
while also facilitating surface-enhanced spectroscopies. For the most part, nanohole arrays have been fabricated using expensive and time-intensive high-resolution serial techniques such as electron beam lithography (EBL)5 and focused ion beam (FIB) milling.1,6 In some cases, advanced soft lithographymethods such as PEEL (a combination of phase-shift lithography, etching, electron-beam deposition, and lift-off) or soft nanoimprint tech- niques have been successfully employed to fabricate nanohole arrays.7,8 In a recent example, Chen et al. fabricated square lattice gold nanohole arrays with sub-250 nm diameter using UV nanoimprint lithography combined with reactive ion etching and a Cr/Au lift-off process.9 With this method, the authors varied the nanohole diameter and periodicity with well-ordered nanohole arrays up to 1 cm2 and examined systematic shifts in the
transmission spectra of structural variations, baths with varied refractive index, and thiol chemisorption. While these methods have facilitated important fundamental studies, the field would greatly benefit from a simpler massively parallel fabrication method that can pattern nanohole arrays with a deep ultraviolet (DUV) patterning resolution, i.e., around 200 nm, without using an exposure tool, photomask, or imprint mold. The work presented herein capitalizes on the nanosphere lithography (NSL) technique conceived of (as “natural lithography”) by Deckman et al.10 and popularized by Van Duyne and co-work- ers.11 Instead of employing an as-assembled 2D colloidal array as a shadowmask for nanostructure deposition, this work employs a reactive ion etching (RIE) step to shrink the nanospheres before metal deposition, facilitating the formation of nanohole arrays after removal of the nanospheres. By controlling the original nanosphere size, etching time, metal deposition thickness, and metal deposition angle, it is possible to tune the nanohole spacing, size and aspect ratio, and, accordingly, the plasmonic properties.
While EBL and FIB have been employed in the vast majority of nanohole array studies, there have been a few advances in addition to the aforementioned PEEL and nanoimprint techni- ques toward implementing high throughput methods. As early as 1995,Masuda et al. used ananodic alumina template andmultiple polymer/metal deposition steps to achieve a nanohole array structure; however, this work did not focus on topographic tuna- bility, optical characterization, or use of the substrate as a surface- enhanced spectroscopy platform.12 A decade later, Jiang and McFarland employed a specialized spin-coating technique to create 2D non-close-packed nanospheres for use as a deposition mask, again focusing only on structural fabrication and charac- terization.13 More recently, Peng and Kamiya formed randomly
*To whom correspondence should be addressed. E-mail: firstname.lastname@example.org (S.-H.O); email@example.com (C.L.H.). (1) Ebbesen, T. W.; Lezec, H. J.; Ghaemi, H. F.; Thio, T.; Wolff, P. A. Nature
1998, 391 (6668), 667. (2) Genet, C.; Ebbesen, T. Nature 2007, 445 (7123), 39-46. (3) Gao, H.; Henzie, J.; Odom, T. W. Nano Lett. 2006, 6 (9), 2104-2108. (4) Liu, H.; Lalanne, P. Nature 2008, 452 (7188), 728-731. (5) Altewischer, E.; Genet, C.; van Exter, M. P.; Woerdman, J. P.; Alkemade,
P. F. A.; van Zuuk, A.; van der Drift, E. W. J. M. Opt. Lett. 2005, 30 (1), 90-92. (6) Brian, B.; Sepulveda, B.; Alaverdyan, Y.; Lechuga, L. M.; Kaell, M. Opt.
Express 2009, 17 (3), 2015-2023. (7) Kwak, E.-S.; Henzie, J.; Chang, S.-H.; Gray, S. K.; Schatz, G. C.; Odom,
T. W. Nano Lett. 2005, 5 (10), 1963-1967. (8) Stewart, M. E.; Mack, N. H.; Malyarchuk, V.; Soares, J.; Lee, T. W.; Gray,
S. K.; Nuzzo, R. G.; Rogers, J. A. Proc. Natl. Acad. Sci. U.S.A. 2006, 103 (46), 17143-17148. (9) Chen, J.; Shi, J.; Decanini, D.; Cambril, E.; Chen, Y.; Haghiri-Gosnet, A.
Microelectron. Eng. 2009, 86 (4-6), 632-635.
(10) Deckman, H. W.; Dunsmuir, J. H. Appl. Phys. Lett. 1982, 41 (4), 377-9. (11) Haynes, C. L.; Van Duyne, R. P. J. Phys. Chem. B 2001, 105 (24),
5599-5611. (12) Masuda, H.; Fukuda, K. Science 1995, 268 (5216), 1466-68. (13) Jiang, P.; McFarland, M. J. J. Am. Chem. Soc. 2005, 127 (11), 3710-3711.
13686 DOI: 10.1021/la9020614 Langmuir 2009, 25(23), 13685–13693
Article Lee et al.
ordered nanohole arrays by spin-coating a solution of polystyrene beads onto the surface of a silicon wafer and controlling relative bead density by adjusting the ionic strength of the nanoparticle solution, followed by Au film deposition.14 Nanoholes were then formed by scanning probe microscopy tip machining, where the tip is used to expose the gold-coated polystyrene bead, and then wet chemistry is used to remove the remaining polystyrene. These nanoholes were characterized using atomic force microscopy and used to create nanoscale transistors. With a similar focus on only fabrication methodology, close-packedNSL-assembled nano- spheres were recently used byWu et al. as nanolenses for forming a nanohole array.15Earlier this year, Lee and co-workers reported the combination of NSL and vertical angle ion milling to create nanocrescent holes; however, the holes are disordered, with variable spacing throughout the sample. Surface plasmon reso- nance (SPR) characterization of the nanocrescent holes was completed, but the nanostructures were not assessed as surface- enhanced spectroscopy substrates.16
While NSL is most often employed as a direct templating method, where metal is deposited into the void space between as- assembled nanospheres, there is precedent for creating more complex nanostructures by combining NSL with plasma or reactive ion etching (RIE) as is done in this work. Van Duyne and co-workers combined NSL and RIE to fabricate embedded triangular nanoparticles that are less prone to solvent remodeling than their nonembedded counterpart.17 After assembling the nanosphere mask, they use CF4 to etch the glass between nano- spheres and thendepositAg through the nanospheres and into the wells. Variation of CF4 etch time facilitated plasmon tuning, and SERS enhancement factors as high as 108 were estimated for chemisorbed benzenethiol. Huang et al. used colloidal templating to create a double-layer colloidal crystal mask with two different nanosphere diameters and performed an oxygen etch, a Au deposition, and removal of the top colloidal crystal layer to create gold nanohole substrates.18 While plasmonic characterization was not performed, the specular reflection was shown to be attenuated compared to glass substrates. Live and Masson pre- pared microhole arrays by drop-coating a polymer microsphere suspension onto glass slides followed by plasma treatment to reduce the size of the spheres and then metal evaporation.19
Controlled evaporation of the microsphere solution yielded large defect-free areas. Both the localized and propagating plasmons were characterized for these substrates with varied hole diameters and refractive index environments, suggesting utility for micro- hole arrays in plasmonic biosensing; however, nanoscale features were not attempted in this work. Ctistis et al. recently reported the use of the same method to fabricate nanohole arrays but was focused on the size-dependent magnetic properties of nanoholes in a cobalt thin film.20 Earlier this year, Lou et al. published a short report where NSL and RIE were combined to create nanoholes in Au.21 This work characterized plasmonic transmis- sion with a fixed hole diameter and varied spacing and used the nanohole array as a substrate for cell growth. Unfortunately, the
typical defect-free domain sizes were small, only a few micro- meters, impeding many potential sensor applications. Clearly, there is a great interest and need for massively parallel nanohole fabrication with well-characterized plasmonic and surface-en- hanced spectroscopic properties; this is the goal of the work presented herein.
Significant progress has been made toward understanding and modeling the plasmonic properties of nanohole arrays fabricated by traditional EBL or FIBmethods. A recent systematic study by Lee et al. compared the SPR spectra and refractive index sensitivity of EBL-fabricated nanohole and nanoslit arrays.22
K€all and co-workers demonstrated the influence of the dielectric substrate’s refractive index on the SPR shift per refractive index unit (i.e., the sensitivity) that can be achieved using FIB-fabri- cated nanohole arrays.6 Parsons et al. examined the effect of nanohole periodicity on the resultant SPR both experimentally using FIB fabrication and in finite-difference time-domain (FDTD) modeling.23 Schatz and co-workers performed FDTD modeling of both isolated and square arrays of nanoholes inAu to predict the electromagnetic fields due to the interaction of the propagating and localized surface plasmons as well as the optical transmission spectra.24 Their modeling predicted concentrated fields at both the top and bottom surface of the nanoholes and transmission spectrawithmultiple features fromnanohole arrays. Optical properties have also been characterized in more exotic versions of the nanohole array. For example, Kim et al. char- acterized optical transmission through arrays of equilateral triangle nanoholes made by ion milling,25 and Brolo and co- workers demonstrated enhancements in spontaneous emission by coupling quantum dots to plasmonic FIB-fabricated nanohole arrays.26 These fundamental studies of nanohole optical proper- ties, and correlations with FDTD modeling, are critical for the application of these substrates.
Several groups have also used nanohole arrays for SPR biosensing and imaging.8,27-31 In recently published work, Oh and co-workers achievedmultiplex SPR biosensing and measure- ment of binding kinetics using FIB-fabricated nanohole arrays in amicroarray format.32-34 Brolo and co-workers also integrated a FIB-manufactured nanohole array into a fluidic device where chemical species could flow through the etched nanoholes, rather than just over the top, in an effort tomaximize molecular binding events at the hole edges where the electromagnetic fields should be most intense.35 In an alternate approach to achieve maxi- mal refractive index sensitivity, Brolo and co-workers also demonstrated SPR sensing of binding events after coating the
(14) Peng, X.; Kamiya, I. Nanotechnology 2008, 19 (31). (15) Wu, W.; Dey, D.; Memis, O. G.; Katsnelson, A.; Mohseni, H. Proc. SPIE
2008, 7039, 70390P/1-70390P/8. (16) Wu, L. Y.; Ross, B. M.; Lee, L. P. Nano Lett. 2009, 9 (5), 1956-1961. (17) Hicks, E.; Lyandres, O.; Hall, W.; Zou, S.; Glucksberg, M.; Van Duyne, R.
J. Phys. Chem. C 2007, 111 (11), 4116-4124. (18) Huang, W.-H.; Sun, C.-H.; Min, W.-L.; Jiang, P.; Jiang, B. J. Phys. Chem. C
2008, 112 (45), 17586-17591. (19) Live, L.; Masson, J. J. Phys. Chem. C 2009, 113 (23), 10052-10060. (20) Ctistis, G.; Papaioannou, E.; Patoka, P.; Gutek, J.; Fumagalli, P.; Giersig,
M. Nano Lett. 2009, 9 (1), 1-6. (21) Lou, Y.; Westcott, N.; Mcglade, J.; Muth, J.; Yousaf, M. Mater. Res. Soc.
Symp. Proc. 2009, 1133.
(22) Lee, K.-L.; Wang, W.-S.; Wei, P.-K. Plasmonics 2008, 3 (4), 119-125. (23) Parsons, J.; Hendry, E.; Burrows, C. P.; Auguie, B.; Sambles, J. R.; Barnes,
W. L. Phys. Rev. B: Condens. Matter Mater. Phys. 2009, 79 (7), 073412/1-073412/4. (24) Chang, S.; Gray, S.; Schatz, G. Opt. Express 2005, 13 (8), 3150-3165. (25) Kim, J.H.;Moyer, P. J. Appl. Phys. Lett. 2006, 89 (12), 121106/1-121106/3. (26) Brolo, A. G.; Kwok, S. C.; Cooper, M. D.; Moffitt, M. G.; Wang, C. W.;
Gordon, R.; Riordon, J.; Kavanagh, K. L. J. Phys. Chem. B 2006, 110 (16), 8307-8313.
(27) Brolo, A. G.; Arctander, E.; Gordon, R.; Leathem, B.; Kavanagh, K. L. Nano Lett. 2004, 4 (10), 2015-2018.
(28) Stark, P. R. H.; Halleck, A. E.; Larson, D. N. Methods 2005, 37 (1), 37-47. (29) Tetz, K. A.; Pang, L.; Fainman, Y. Opt. Lett. 2006, 31 (10), 1528-1530. (30) Lesuffleur, A.; Im, H.; Lindquist, N. C.; Oh, S. H. Appl. Phys. Lett. 2007,
90, 243110. (31) Yang, J.-C.; Ji, J.; Hogle, J. M.; Larson, D. N. Nano Lett. 2008, 8 (9),
2718-2724. (32) Lesuffleur, A.; Im, H.; Lindquist, N. C.; Lim, K. S.; Oh, S. H. Opt. Express
2008, 16 (1), 219-224. (33) Im,H.; Lesuffleur, A.; Lindquist, N. C.; Oh, S.-H. Anal. Chem. 2009, 81 (8),
2854-2859. (34) Lindquist, N. C.; Lesuffleur, A.; Im, H.; Oh, S. H. Lab Chip 2009, 9 (3),
382-387. (35) Eftekhari, F.; Escobedo, C.; Ferreira, J.; Duan, X.; Girotto, E. M.; Brolo,
A. G.; Gordon, R.; Sinton, D. Anal. Chem. 2009, 81 (11), 4308-4311.
DOI: 10.1021/la9020614 13687Langmuir 2009, 25(23), 13685–13693
Lee et al. Article
nanohole array with a dielectric layer such that the only revealed Au surface was within the interior area of the nanoholes.36 While each of these examples demonstrates the capability of nanohole arrays to act as sensor signal transduction elements, efforts to date have largely been limited to model analyte systems (e.g., biotin/ streptavidin) partly due to the low throughput and expensive nature of nanohole array fabrication strategies.
Because nanohole arrays show great promise for SPR sensing, they are also inherently promising as surface-enhanced spectros- copy substrates. Surface-enhanced Raman spectroscopy (SERS) is the most commonly employed of the surface-enhanced phe- nomena. Since its discovery in 1977 by Van Duyne and co- workers,37 at least 9000 papers have been published on the topic of SERS with a drastic uptick in interest in the past decade as nanofabrication methods and electromagnetic modeling tools are refined and as SERS becomes an accepted analytical technique. SERS occurs when a Raman-active molecule is localized within the large electromagnetic fields that are generated upon excitation of the SPR and provides many orders of magnitude enhancement over normal Raman scattering. To date, there are only a few examples of SERS measurements using nanohole arrays as the enhancing substrate. Brolo and co-workers measured SERS spectra from both copper and gold nanohole arrays fabricated using FIB but did not quantify the enhancement factor based on complications in their experimental system.27,38 Wallace and co-workers also measured SERS spectra from EBL-fabricated Au nanohole arrays, achieving enhancement factors as high as 4.2 105.39 Rowlen and co-workers used EBL-fabricated nano- hole arrays to demonstrate that SERS enhancement decreased as hole lattice spacing increased.40While these results are promising, further demonstration of SERS on nanohole arrays is warranted, especially if this can be accomplished using arrays produced using a massively parallel fabrication strategy.
Herein we report a simple method to produce nanohole arrays in a massively parallel fashion where the shape of the nanohole profile can be varied based on the tilt angle during metal deposition. The surface plasmon-enhanced transmission is char- acterized and correlates well with FDTD simulations. The simu- lations reveal that the electromagnetic fields are focused at the hole edges, as expected, and that the two measured transmission peaks can be attributed to separate plasmons at the Ag/air and Ag/glass interfaces. Support of two separate plasmons will facil- itate optimized SERS enhancement at two different laser excita- tion wavelengths. High S/N SERS spectra are measured from theses nanohole array substrates, and the resulting enhancement factors are an order ofmagnitude better than a silver filmwithout the nanoholes. Further, we demonstrate that the achieved en- hancements can be improved by another order of magnitude by using electroless plating to superimpose a textured Ag film over the nanohole array.
Materials. All the chemicals were of reagent grade or better and used as purchased. All water samples were ultrapure (18.2 MΩ cm) from a Millipore water purification system
(Long Beach, CA). Commercially available 400 and 600 nm diameter carboxylated polystyrene (PS) nanospheres (Invitro- gen, 4 wt %) were used to form a 2D self-assembled single layer. For nanosphere dilution and surface tension control, 1:1 (v/v) Milli-Q water:ethanol (Sigma-Aldrich) were used. Benzenethiol (99%) was purchased from Sigma-Aldrich. 99.999% purity Ag pellets for vapor deposition were purchased from Kamis Inc.
Preparation of Substrates. To obtain the hydrophilic sub- strate necessary for nanosphere self-assembly, standard glass microscope slides were rinsed with acetone, methanol, isopropyl alcohol (IPA), and deionized water and then sonicated in a cleaning solution of NH4OH/30% H2O2/H2O (volume ratio 1:1:10) for 1 h. Finally, they were cleaned in a piranha solution (70% H2SO4/30% H2O2) for 30 min. After rinsing thoroughly, they were stored in deionizedwater and dried with high-purityN2 immediately before nanosphere deposition. To facilitate large areas of perfect nanosphere packing, the nanosphere solution spreading area was controlled by the presence of lithographically defined polydimethylsiloxane (PDMS, Sylgard) wells. Standard soft lithography methods were employed to create the PDMS wells.41 A negative photoresist, SU-8 50, was spin-coated onto a siliconwafer andpatterned to create themastermold.A 10:1 ratio of PDMS and curing agent was degassed in vacuum, cast onto the master mold, and wiped with a plastic film to remove excess uncured PDMS, creating a 50-200 μm thick layer and creating holes several millimeters in diameter. After curing the PDMS at 70 C overnight, the PDMS sheet was carefully peeled off and aligned onto the glass slide substrate, defining the PDMSwells for nanosphere assembly. The PDMS wells were reused many times.
Colloidal Self-Assembly of Hexagonal Close-Packed Nanohole Arrays. Figure 1a shows a schematic of the colloidal self-assembly using PDMS-defined assembly areas. After drop- ping the nanosphere suspensiononto the glass slides, a single layer of nanospheres is formed from the center and grows toward the outside as the solution evaporates. To pattern large areas, the nanospheres were diluted in 1:1 (v/v) Milli-Q water:ethanol to control the concentration and the surface tension at the li- quid-vapor interface. A 0.5 μL droplet of the colloidal suspen- sion was placed onto the glass slides within the 3 mm diameter
Figure 1. (a) Schematic representation of the colloidal assembly using PDMS-defined areas and (b) process steps for fabricating nanohole arrays using NSL.
(36) Ferreira, J.; Santos, M. J. L.; Rahman, M. M.; Brolo, A. G.; Gordon, R.; Sinton, D.; Girotto, E. M. J. Am. Chem. Soc. 2009, 131 (2), 436-437. (37) Jeanmaire, D. L.; Van Duyne, R. P. J. Electroanal. Chem. Interfacial
Electrochem. 1977, 84 (1), 1. (38) Anema, J. R.; Brolo, A. G.;Marthandam, P.; Gordon, R. J. Phys. Chem. C
2008, 112 (44), 17051-17055. (39) Yu, Q.; Guan, P.; Qin, D.; Golden, G.; Wallace, P. M. Nano Lett. 2008, 8
(7), 1923-1928. (40) Reilly, T. H.; Chang, S. H.; Corbman, J. D.; Schatz, G. C.; Rowlen, K. L.
J. Phys. Chem. C 2007, 111 (4), 1689-1694. (41) Xia, Y. N.; Whitesides, G. M. Annu. Rev. Mater. Sci. 1998, 28, 153-184.
13688 DOI: 10.1021/la9020614 Langmuir 2009, 25(23), 13685–13693
Article Lee et al.
PDMS-defined wells and dried at room temperature. The close packed polystyrene nanospheres were then etched with RIE (Reactive Ion Etch, STS320), causing them to shrink while retaining their original crystalline lattice location. A mixture of oxygen and argon gas was used in the etch process with the experimental parameters of O2(35 sccm)/Ar(10 sccm), 60 mTorr, 60 W. After etching the PS nanospheres, an e-beam evaporator (Temescal) was used to deposit a 50 nm thick silver film onto the templated substrate. The substrate holder was rotating during normal incidence deposition, whereas it was not in the case of angleddeposition.The evaporation chamber pressure anddeposi- tion rate were 5 10-6 Torr and 1 Å/s, respectively, and the deposition thickness was monitored by a quartz crystal micro- balance. Finally, the PS nanospheres were removed with 3M Scotch tape, leaving behind the nanohole array. Residue and any remaining nanospheres were dissolved using dichloromethane and ultrasound sonication at 70 W for 20 min. The processing steps for nanohole array fabrication using NSL are illustrated in Figure 1b.
Electroless Plating. Electroless plating of silver onto the nanohole array was accomplished using a protocol described by Halas and co-workers.42 A nanohole array substrate was sub- merged in a solution of 1.2 mL of Acacia (500 mg/L) with 0.2 mL of buffer solution (1.5M citric acid/0.5M sodium citrate, pH=3) and 0.3 mL of silver lactate (37 mM in water). Then, 0.3 mL of hydroquinone (0.52 M in water) was added while stirring the mixture vigorously. The Acacia stabilizes the silver ions and allows for slower reduction of the silver, which causes the forma- tion of “spiky” silver deposition. The substrates were removed from the solution after 4, 6, or 8 min.
SEM Characterization. Scanning electron microscope (SEM) images were taken with a JEOL 6500 field emission gun scanning electron microscope with an accelerating voltage of 5.0 kV. The samples were sputtered with 50 Å of Pt prior to imaging to reduce charging effects.
Optical Characterization. Transmission spectra were mea- sured using a tungsten-halogen lamp and amicroscope objective (50, NA=0.55). The transmitted light was collected using an optical fiber (200 μmdiameter core), and the zero-order transmis- sion spectrumwas analyzedwith anOceanOpticsUSB4000 fiber- optic spectrometer in the 400-850 nm spectral range. More detailed data collection and spectral normalization methods can be seen in previous work.30
SERSMeasurements. BothAg nanohole arrays andAg film over nanosphere (AgFON) substrates, prepared as previously described,43were dosedwith 1mMbenzenethiol (BZT) in ethanol unless otherwise specified. Each substrate was placed in 2 mL of the ethanolic BZT solution for 16 h at ambient conditions to facilitatemonolayer formation.The substrateswere then removed from the ethanolic solution and rinsed with ethanol before they weremounted in themicroscaleRaman sample holder. SERSwas excited with a Millenia Vs 532 nm excitation laser with p-polari- zation (Spectra-Physics, Mountain View, CA) or a 12 mW HeNe 632.8 nm excitation laser with random polarization (Newport Corp., Irvine, CA). Neutral density filters were added to the optical path to achieve desired incident powers. In both cases, the laser beam was first passed through the appropriate interference filter (Melles-Griot, Rochester, NY) and was then guided into a 20 microscope objective (Nikon, Melville, NY) and onto the sample. The scattered light was collected back through same 20 objective, and Rayleigh scattered light was removed using the appropriate notch filter (Semrock, Rochester, NY) before collec- tion.Detectionwas accomplished using a 0.5mSpectra-Pro 2500i single monochromator and a Spec 400B liquid nitrogen-cooled CCDchip (both fromPrinceton Instruments/Acton, Trenton,NJ). In all cases, SERS spectra were measured from three ran- domly chosen areas on a given substrate. Spectra recorded with
near-infrared excitation (752 nm) were excited using an Innova90 krypton laser (Coherent Inc., SantaClara,CA) and collectedwith aWITec alpha300R scanning confocalRamanmicroscopewith a UHTS300 spectrometer and DV401 CCD detector (all from WITec, Savoy, IL) with a 100 objective (Olympus, Center Valley, PA). In this case, spectra were measured from a 10 μm 10 μm area. EF Calculation. EFs were all calculated with the intense
benzene breathing stretch at approximately 1000 cm-1 shift for both the liquid and surface-adsorbed BZT using the equation:
EF ¼ NvolIsurf Nsurf Ivol
where Nvol is the number of BZT molecules contributing to the normal Raman signal, Nsurf is the number of BZT molecules contributing to the SERS signal, and Isurf and Ivol are the inten- sities of the scattering band of interest in the SERS and normal Raman spectra, respectively. The packing density for BZT onAg, as reported in the literature, is 6.8 1014 molecules/cm2.44,45 Using the surface area obtained from AFM or SEM measure- ments and the laser spot size (Table 1), it is then possible to calculate Nsurf and the resulting enhancement factor.
Results and Discussion
NSL Process for Self-Assembly of Periodic Nanohole Arrays. While the simple “drop and dry” method commonly used for nanosphere lithography can produce a relatively large area 2D packed structure (∼1 mm2), it is difficult to obtain a single crystalline domain over a large area. When a droplet of colloidal suspension is placed onto the substrate, it spreads out and forms concentric circles before drying. The convex, dome- shaped structure of the droplet hinders the collision and redis- tribution of the nanospheres at the drying edge, degrading the quality of 2D packing. In addition, since it is difficult to control the exact area over which the droplet spreads out, the local concentration of the nanospheres is not precisely controlled.
Since Nagayama and co-workers observed capillary force- induced 2D self-assembly using a Teflon ring,46 many researchers have been working to improve the quality of the 2D self- assembled structures by using a concave (instead of convex) meniscus relative to the surface of a substrate.47-49 Using the concavemeniscus facilitates the convective flow and enhances the rearrangement of nanospheres just before leading edge solvent evaporation. In this work, a PDMSwell encourages formation of a concave meniscus on the substrate and controls the local
Table 1. SEM-Determined Substrate Surface Areas for Enhancement Factor Calculation
surface area (cm2)
sample tilt angle (deg) λex = 532 nm λex = 633 nm λex = 752 nm
0 9.15 10-9 1.30 10-8 5.66 10-9 30 8.34 10-9 1.18 10-8 5.16 10-9 45 8.34 10-9 1.18 10-8 5.16 10-9 50 8.34 10-9 1.18 10-8 5.16 10-9 55 7.93 10-9 1.12 10-8 4.91 10-9 60 7.53 10-9 1.07 10-8 4.65 10-9
(42) Jackson, J. B.; Halas, N. J. J. Phys. Chem. B 2001, 105 (14), 2743-2746. (43) Bantz, K. C.; Haynes, C. L. Langmuir 2008, 24, 5862-5867.
(44) Whelan, C. M.; Smyth, M. R.; Barnes, C. J. Langmuir 1999, 15 (1), 116-126.
(45) Wan, L. J.; Terashima, M.; Noda, H.; Osawa, M. J. Phys. Chem. B 2000, 104 (15), 3563-3569.
(46) Denkov, N. D.; Velev, O. D.; Kralchevsky, P. A.; Ivanov, I. B.; Yoshimura, H.; Nagayama, K. Nature 1993, 361 (6407), 26-26.
(47) Dimitrov, A. S.; Nagayama, K. Langmuir 1996, 12 (5), 1303-1311. (48) Fulda, K. U.; Tieke, B. Adv. Mater. 1994, 6 (4), 288-290. (49) Wang, D. Y.; Mohwald, H. Adv. Mater. 2004, 16 (3), 244.
DOI: 10.1021/la9020614 13689Langmuir 2009, 25(23), 13685–13693
Lee et al. Article
concentration of nanospheres, facilitating the formation of a packed monolayer over ∼1 mm diameter region with a typical single crystalline, defect-free domain size of ∼30 30 μm2. A larger PDMSwell can increase the size of a single layer although it becomes more difficult to control the drying pattern and uni- formity. We used a 3 mm diameter well (hole through the PDMS sheet) and could easily produce polycrystalline single nanosphere layers ∼1 mm across. The PDMS wells also allow parallel fabrication of multiple arrays in a reproducible way.
The self-assembly of nanospheres was followed by an RIE process in oxygen plasma to reduce the nanosphere size in a controlled manner. Importantly, the RIE process left the lattice position of each nanosphere intact (Figure 2b). Subsequently, a 50 nm thick Ag filmwas evaporated onto the nanosphere template (Figure 2c), using each size-reduced PS nanosphere as a shadow mask. Then the nanospheres were removed with tape and solvent cleaning. Figure 2d shows the nearly defect-free, crystalline nanohole arrays produced by these steps. The average diameter of the hole after bead removal was 196( 11 and 298( 7 nmwhen using 400 and 600 nm diameter nanospheres, respectively.
Using the NSL process, the size of each nanohole can be easily tuned by changing the RIE processing time. Furthermore, ellip- tical hole shapes can be created by simply tilting the substrate during the metal deposition, adding another degree of freedom to tailor the plasmon resonance characteristics of the nanohole arrays, shown in Figure 3a. Since Haynes and co-workers demonstrated an angle-resolved deposition process using NSL,11 many researchers have utilized tilted metal deposition techniques for making nanostructures.50-52 The ellipticity of the holes depends on the size of nanospheres and the tilt angle of the substrate; smaller nanospheres with a steeper substrate tilt angle produce ellipseswith higher aspect ratios. Parts b and c ofFigure 3 show SEM images of 45 and 60, respectively, tilted metal depo- sition onto etched nanosphere templates before removing the nano- spheres. In Figure 3d, the aspect ratio measured after removing the nanospheres was 1.6 when the substrate was tilted to 60.
Figure 2. Scanning electron micrographs of (a) single layer for- mation with PS nanospheres (600 nm diameter), (b) nanosphere template after size reduction using reactive ion etching, (c) nano- sphere template after 50 nm thick Ag deposition, and (d) the resulting large area (∼30 30 μm2) single crystalline hexagonal nanohole array.
Figure 3. (a) Schematic illustration of the angle-resolved metal deposition process. Scanning electron micrographs of (b) an ellip- tical nanohole array (45 tilt angle) before nanosphere removal, (c) an elliptical nanohole array (60 tilt angle) before nanosphere removal, and (d) after nanosphere removal.
(50) Kosiorek, A.; Kandulski, W.; Chudzinski, P.; Kempa, K.; Giersig, M. Nano Lett. 2004, 4 (7), 1359-1363.
(51) Lu, Y.; Liu, G. L.; Kim, J.; Mejia, Y. X.; Lee, L. P. Nano Lett. 2005, 5 (1), 119-124.
(52) Choi, D. G.; Jang, S. G.; Kim, S.; Lee, E.; Han, C. S.; Yang, S. M. Adv. Funct. Mater. 2006, 16 (1), 33-40.
13690 DOI: 10.1021/la9020614 Langmuir 2009, 25(23), 13685–13693
Article Lee et al.
Transmission Spectra andFDTDSimulation.Based on the critical optical properties for SPR, EOT, or SERS biosensing, transmission extinctionmeasurementswere completedon circular nanohole arrays. Figure 4a presents the experimental results obtained from substrates with circular holes made using 400 nm diameter nanospheres. Finite-difference time-domain (FDTD) calculations (Fullwave, Rsoft Design Group) were performed on the analogous structure and are represented as a dashed line. The spectral position of the maximum transmission peaks can be approximated with the following equation:53
λmax ¼ a0 4 3 ði2 þ ijþ j2Þ
-1=2 ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiεdεm εd þ εm
where a0 is the periodicity of the array, the integers (i, j) represent the Bragg resonance orders, and εm and εd are the dielectric functions of the metal and the dielectric, respectively. The calculations and the experimental measurements are in good agreement. The (1,0) transmission peaks at the silver/glass inter- face calculated by FDTD and measured by the spectrometer appear at 672 and 695 nm, respectively. The (1,1) transmission peak at the silver/glass interface and the (1,0) transmissionpeak at the silver/air interface overlap at∼500 nm, resulting in a broader resonance. Maintenance of two separate surface plasmon reso- nances on one sample will be convenient because it will facilitate a broader range of possible excitation wavelengths that yield good SERS enhancement factors. Although the probe area in the optical measurements was larger than the size of the perfectly crystalline areas, both the transmitted intensity and the peak wavelengths
measured at nine different locations on the nanohole array sample, as shown in Figure 4b, are very similar, centered at λ=695( 5 nm.
FDTD modeling facilitates visualization of the electromag- netic field distribution in plasmonic structures. Figure 5a shows a time-averaged intensity map of the z-component of the electric field. Periodic boundary conditions are used to simulate an infinite hexagonal array with a periodicity of 400 nm. Since the domain size of the fabricated nanohole arrays are larger than the propagation length of the surface plasmon at visible frequencies, ∼10 μm on a smooth Ag film, using such periodic boundary conditions is justified. Furthermore, the propagation length of a surface plasmon within an array is reduced due to scattering introduced by the nanoholes. A nonuniform mesh was used for computational efficiency, with a nominal grid size of 5 nm and an edge/interface grid size of 2 nm. The dispersion of the silver film was modeled with a Drude/Lorentzian model fit to experimental optical constants. Both electric fieldmapswere extracted from the FDTD calculation at the (1,0) resonance direction (x-axis) at different wavelengths. Light is incident from the top (þz) and excites surface plasmons on the nanohole array at each interface. The optical energy is confined strongly at the air/silver inter- face with λmax = 422 nm and to the glass/silver interface with λmax=672 nm. These results confirm the wavelength of the trans- mission peak at each interface in Figure 4a. Furthermore, the air/ silver interface shows high electric field intensity at both wave- lengths, which contributes to the SERS enhancement. Figure 5b shows a top-down view of the electric field intensity at the air/ silver interface with λmax=422 nm. As has been seen in previous nanohole modeling, the electromagnetic fields are largely con- centrated to the edges of the nanhole features. The surface
Figure 4. (a) Experimental and computational (FDTD) transmis- sion spectra for a hexagonal nanohole array with a periodicity of 400 nm and a circular hole shape. (inset) The direction of (1,0) and (1,1) resonances. (b) Experimental spectra measured at nine dif- ferent locations on a nanohole array substrate. Figure 5. (a) Time-averaged intensity map of the plasmonic field
(z-component of the electric field) at λ=422 nm and λ=672 nm. The FDTD result confirms the wavelength of the transmission peak at each interface. (b) A top-down view of the electric field intensity at the air/silver interface at λ=422 nm.
(53) Thio, T.; Ghaemi, H. F.; Lezec, H. J.; Wolff, P. A.; Ebbesen, T. W. J. Opt. Soc. Am. B 1999, 16 (10), 1743-1748.
DOI: 10.1021/la9020614 13691Langmuir 2009, 25(23), 13685–13693
Lee et al. Article
plasmon resonance was measured and modeled for the elliptical nanoholes as well, but continued work with these and higher aspect ratio nanoholes will be the subject of future work. SERS Measurements. On the basis of the predicted and
measured optical properties, these large area nanohole arrays are good candidate SERS substrates. For initial SERS characteriza- tion, all substrates were dosed with benzenethiol, an efficient Raman scatterer with well-characterized surface coverage. SERS measurements were performed on the circular nanohole arrays as well as the elliptical nanohole arrays and compared to both flat Ag substrates and the commonly used AgFON substrate. As can be seen in Figure 6, the circular nanohole arrays yield high signal- to-noise SERS spectra, with no spectral evidence for polystyrene residue on the substrate. At 532 nm excitation wavelength, the calculated enhancement factor, 1.70 104, is comparable to that measured from a AgFON, 2.62 104, and significantly higher than that measured from a flat 50 nm thick Ag film, 1.07 103. On the basis of the understanding that SERS enhances both exciting and scattered photons, the ideal nanohole substrate would have a SPR centered between the excitation wavelength and the Stokes scattering wavelength of interest. When consider- ing the 1000 cm-1 shift band of benzenethiol, the ideal SPR wavelengths for 532, 632.8, and 752 nm excitations are 547, 654, and 782 nm, respectively. As can be seen in the measured transmission spectra in Figure 4a, the 532 nm excitation is off-
resonance compared to the other twowavelengths and is expected to yield lower enhancement factors even though Raman scatter- ing is more efficient at shorter excitation wavelengths.
Table 2 summarizes the enhancement factors measured at all three excitation wavelengths for the circular nanohole arrays, with values up to 8 105. These enhancement factors are compa- rable to or higher than those reported in the literature for nanohole arrays fabricated using traditional methods39,40 and, as expected, are significantly larger with the on-resonance 632.8 and 752 nm excitation wavelengths than with the off-resonance 532 nm excitation. Comparison to a flat Ag film of the same thickness demonstrates that the nanohole array gives an order of magnitude better average enhancement. In addition, SERS spec- tra were also measured from the elliptical nanohole array sub- strates (Table 2). In general, the elliptical profile did not yield a significantly different enhancement factor than the circular profile. Based on the predicted concentration of electromagnetic fields in higher aspect ratio nanostructures, future work will focus on polarized SERSmeasurements from the elliptical nanohole arrays.
While the periodic structure of the nanohole array clearly yields improved SERS performance, further improvement of the en- hancement factor would facilitate lower detection limits in future sensor applications. The FDTD simulations reported herein
Figure 6. SERS spectra of (A) 1mMBZTonaAgFON, (B) 1mM BZT on a circular hole nanoarray, and (C) 1 mM BZT on 50 nm flat silver. P= 0.32 mW, t= 180 s, and λex = 532 nm.
Table 2. Average EF Values over Three Different Wavelengths
sample tilt angle (deg) λex = 532 nm λex = 633 nm λex = 752 nm
0 1.70 104 8.13 105 1.24 105 30 1.85 104 1.88 106 4.18 104 45 1.63 104 7.51 105 2.05 104 50 1.35 104 7.15 105 8.98 104 55 2.23 104 5.42 105 7.03 104 60 2.97 104 1.11 106 4.48 104 flat Ag 1.07 103 5.55 104 N/A AgFON 2.62 104 9.04 104 N/A Figure 7. Scanning electronmicrographof (a) 4minAgelectroless
plating onto the nanohole array and (b) 6 min Ag electroless plating onto the nanohole array.
13692 DOI: 10.1021/la9020614 Langmuir 2009, 25(23), 13685–13693
Article Lee et al.
suggest that the majority of the electromagnetic enhancement originates at the edges of the nanoholes. In an effort to exploit the remainder of the substrate, the nanohole arrays were immersed in a Ag plating bath known for producing high surface area “spiky” features.42 One literature precedent applying electrolessly plated Ag for SERS was performed by Green et al., where noble metal nanoscale pillars were fabricated using island lithography.54 In this case, fabrication beganwith deposition ofCsCl islands onto a silica film followed by Cr deposition on top of the CsCl islands. When the islands are removed, the SiO2 layer was etched, leaving pillarswithCr caps. TheCr is then removed andAg is electrolessly plated into the gaps between the SiO2 pillars, and then the SiO2 is etched away to reveal the final nanostructure. The conditions of the electroless plating bath were altered to give different Ag structures, and high signal-to-noise SERS spectra were measured from adsorbed pyridine though characterization of the LSPR is
not done. Herein, for control purposes, the previously described nanohole substrate and a flat Ag film were immersed in the plating bath simultaneously. Figure 7a shows the resulting structures after 4 min in the plating solution; 6 min plating bath exposure was also performed on identical substrates (Figure 7b). Figure 8 includes representative benzenethiol SERS spectra recorded from both the normal and enhanced substrates with 532 nm excitation while Table 3 delineates themeasured enhance- ment factors using both 532 and 632.8 nm excitation for both flat and nanohole array substrates before plating and after either 4 or 6 min in the plating bath. These data demonstrate that a 4 min immersion in the plating bath gives an additional order of magnitude SERS enhancement on either a flat or a nanohole array substrate at either excitation wavelength, yielding enhance- ment factors as large as 3 106. Because of the roughness of the electrolessly plated silver, it was not possible to obtain AFM data for the enhanced substrates. EF estimates are based on the origi- nal surface areas found for the nanohole substrate. Transmission measurements of the SPRwere attempted on these substrates, but the recorded spectra were very broad, indicating extremely heterogeneous topographical features. While the substrates that were immersed in the plating bath for 6 min do give a slightly improved enhancement factor over the unplated substrates, these substrateswere not as efficient as the substrates immersed for only 4 min. Electron microscopy analysis of the substrates immersed for 6 min (Figure 7b) revealed that the nanohole structure was nearly entirely obscured, largely negating the enhancement in- troduced by the nanohole grating structure. Effort was made to avoid obscuring the nanohole substrate by performing the plating directly on the RIE-etched template before sphere removal (without a vapor deposited metal layer). Plating bath immersion times of 4, 6, and 8 min all produced a substrate where the nanospheremask could not be removed by normalmeans; thus, it was not possible to achieve the desired nanohole array structure.
This work has demonstrated a simple and massively parallel nanofabrication technique to produce large defect-free areas (>30 30 μm2) of noblemetal nanohole arrays, with demonstrated hole sizes of 300 nmdiameter, across amillimeter-sized patterning area. The nanohole profiles can be varied simply by controlling the deposition angle onto the etched nanosphere template. The optical transmission properties of the nanohole arrays are extremely homogeneous across a sample, and FDTD simulations predict the optical characteristics accurately, revealing two resonance peaks due to surface plasmons at the two different dielectric interfaces. The nanohole arrays give good signal-to-noise ratio SERS spectra at three demonstrated excitation wavelengths with the largest enhancement factors, 8 105, occurring when the plasmon resonance is nearly centered between the excitation wavelength and the Raman scattering wavelength. The nanohole structure contributes an order of magnitude to this enhancement factor regardless of the excitationwavelength employed.To exploit the substrate region between the nanoholes, Ag electroless plating was implemented on the nanohole arrays to introduce additional surface roughness, yielding enhancement factors as large as 3 106. These substrates hold great promise for further application as SERS substrates as well as platforms for extraordinary optical transmission and surface plasmon resonance biosensing.
Acknowledgment. S.H.L. was supported by a Samsung Fel- lowship, and K.C.B. was supported primarily by the MRSEC Program of the National Science Foundation under Award
Figure 8. SERS spectra of (A) 4 min Ag plated nanohole array, (B) 4 min Ag plated flat silver film, (C) a standard nanohole array, and (D) 5 a flat silver film. All substrates were exposed to 1 mM BZT prior to SERS measurements. P=0.32 mW, t=180 s, and λex=532 nm.
Table 3. Average EFValues for TwoDifferent PlatingConditions and Wavelengths
sample λex = 532 nm λex = 633 nm
nanohole array 1.70 104 8.13 105 nanohole array þ 4 min plating 1.58 105 3.26 106 nanohole array þ 6 min plating 7.51 104 2.08 105 flat Ag film 1.07 103 5.55 104 flat Ag film þ 4 min plating 2.39 104 1.18 106 flat Ag film þ 6 min plating 2.09 104 6.11 104
(54) Green, M.; Liu, F. J. Phys. Chem. B 2003, 107 (47), 13015-13021.
DOI: 10.1021/la9020614 13693Langmuir 2009, 25(23), 13685–13693
Lee et al. Article
DMR-0819885.N.C.L. acknowledgesNIHBiotechnologyTrain- ingGrant fromNIGMS, and S.H.O. acknowledges support from 3M Nontenured Faculty Award, the ACS Petroleum Research Fund (Doctoral New Investigator Award), and the Minnesota Partnership for Biotechnology and Medical Genomics. C.L.H.
acknowledges support from the ACS Petroleum Research Fund and U. of Minnesota start-up funds. Device fabrication was performed at the University of Minnesota NanoFabrication Center (NFC), which receives support from NSF National Nanotechnology Infrastructure Network (NNIN).