Ss

Ma

b

c

a

ARRA

KSON

1

bhsin[miu

tea(e

1d

Spectrochimica Acta Part A 78 (2011) 706–711

Contents lists available at ScienceDirect

Spectrochimica Acta Part A: Molecular andBiomolecular Spectroscopy

journa l homepage: www.e lsev ier .com/ locate /saa

urface-enhanced Raman scattering of a Ag/oligo(phenyleneethynylene)/Agandwich

elissa Fletchera, D.M. Alexsonb, Sharka Prokesb, Orest Glembockib, Alberto Vivonic, Charles Hostena,∗

Department of Chemistry, Howard University, Washington, DC 20059, United StatesNaval Research Laboratory, Electronics Science and Technology, Washington, DC 20375, United StatesDepartment of Biology, Chemistry, and Environmental Sciences, Inter American University, Bayamon Campus, Bayamon, PR, United States

r t i c l e i n f o

rticle history:eceived 24 September 2010eceived in revised form 1 November 2010ccepted 30 November 2010

eywords:urface-enhanced Ramanligo(phenyleneethynylenes)anosandwich

a b s t r a c t

�,�-Dithiols are a useful class of compounds in molecular electronics because of theirability to easily adsorb to two metal surfaces, producing a molecular junction. We haveprepared Ag nanosphere/oligo(phenyleneethynylene)/Ag sol (AgNS/OPE/Ag sol) and Agnanowire/oligo(phenyleneethynylene)/Ag sol (AgNW/OPE/Ag sol) sandwiches to simulate the archi-tecture of a molecular electronic device. This was achieved by self-assembly of OPE on the silvernanosurface, deprotection of the terminal sulfur, and deposition of Ag sol atop the monolayer. Thesesandwiches were then characterized by surface-enhanced Raman scattering (SERS) spectroscopy. Theresulting spectra were compared to the bulk spectrum of the dimer and to the Ag nanosurface/OPE SERSspectra. The intensities of the SERS spectra in both systems exhibit a strong dependence on Ag deposition

time and the results are also suggestive of intense interparticle coupling of the electromagnetic fieldsin both the AgNW/OPE/Ag and the AgNS/OPE/Ag systems. Three previously unobserved bands (1219,1234, 2037 cm−1) arose in the SER spectra of the sandwiches and their presence is attributed to thestrong enhancement of the electromagnetic field which is predicted from the COSMOL computationalpackage. The 544 cm−1 disulfide bond which is observed in the spectrum of solid OPE but is absent inthe AgNS/OPE/Ag and AgNW/OPE/Ag spectra is indicative of chemisorption of OPE to the nanoparticles

ation

through oxidative dissoci

. Introduction

Surface-enhanced Raman scattering (SERS) spectroscopy hasecome a useful technique in many areas of research [1–7]. SERSas been applied to chemical analysis [8], electronics [9], and sen-ors [10]. While this method proves to be informative, it also facests own challenges. In order to get a significant enhancement of sig-al, on the order of 104–106, noble metal substrates must be used1,2]. In addition to type of metal used, the topography of the metal

ust be of atomic roughness [11,12]. Despite the advances maden technology, the mechanism of enhancement has yet to be fullynderstood.

The SERS phenomenon is largely due to the scattering cross sec-ions of molecules interacting with the magnetic field induced by

xciting surface plasmons of metals such as silver [13], gold [14],nd copper [15]. This is known as the electromagnetic mechanismEM) of enhancement [13–15]. The secondary portion of the SERSffect is due to charge-transfer complexes between the substrate

∗ Corresponding author. Tel.: +1 202 806 7505.E-mail address: chosten@howard.edu (C. Hosten).

386-1425/$ – see front matter © 2010 Elsevier B.V. All rights reserved.oi:10.1016/j.saa.2010.11.053

of the disulfide bond.© 2010 Elsevier B.V. All rights reserved.

and molecule namely, the chemical mechanism of enhancement(CHEM) [16,17].

More recently “Hot Spot” theory has been used to explain thelarge enhancement factors (EFs) of 108–1010 which have beenobserved for monolayers adsorbed on silver surfaces [4,5,18]. Thefirst attempt at explaining these huge EFs used the geometriesof the fractal aggregates formed by the silver substrate [19]. Itwas thought that the CHEM was the principal mode of enhance-ment that contributed to hot spots. Today, the theory has evolved,naming the EM the primary mode of enhancement. If two metalsurfaces’ local surface plasmons come into proximity (1–2 nm) ofeach other, their transition dipoles are anticipated to couple, pro-ducing coherent interference [19,20]. If molecules are trapped inthat 1–2 nm spacing, the collected Raman signal is expected to besignificantly larger. The intensified SERS spectra observed for 1,4-benzenedithiol, 4-aminobenzenethiol, and rhodamine 6G on Ag soland Ag coated nanospheres have been accredited to the hot spots

that are created by two close spheroids. Hot spot theory has led tothe detection of single molecules.

The Raman signal has been further enhanced by sandwich-ing molecules between metallic surfaces. When an azobenzenemonolayer was sandwiched between Au and Ag the Raman spec-

ica Ac

tasstoAecatsll

tsessnsnbon

fi[of[mhiw

mbrvipffmtO

mOmsrtaaottosrtas

M. Fletcher et al. / Spectrochim

rum was more intense than that of the monolayer on Au or Aglone [21]. Poly(vinylpyridine) and dipicolinic acid have also beentudied between smooth or roughened silver surfaces and silverol [22]. The spectrum of dipicolinic acid was enhanced enougho allow it to be used as a chemical marker for the detectionf Bacillus anthracis. Near-infrared SERS spectroscopy has probedu|4-aminobenezene thiol|Ag sandwich spectra which showednhancement factors of 105, where a factor of 102 was said to beaused by chemical enhancement [21]. These elevated EFs werettributed to the coupling of the electromagnetic fields from thewo metal layers. Several sandwiched monolayers have yieldedimilar results, and all were said to be resultant of the amplifiedocal field produced by plasmon coupling between the two metallicayers [22–24].

Trapping monolayers between metals have proven to increasehe Raman signal of the molecule; however, the morphology of theurfaces is crucial to the magnitude of the enhancement. Orendorfft al. fabricated several types of gold nanoparticles [1]. including:pheres, triangles, cubes, and dog bones. Their study showed thatandwiches using planar nanoparticles scattered more than theon-planar particles, implying usefulness for the analysis of poorcatterers. Hu and Wang [25] also used variously shaped Au and Aganoparticles to sandwich 4-aminothiophenol, showing that the2 vibrational bands were dramatically enhanced (CHEM). Thesebservances were ascribed to the tunneling from the Ag to Auanoparticles through 4-aminothiophenol.

Molecular electronic devices (MED) are based on a similar con-guration to the metal|molecule|metal configuration of hot spots25–29]. For a hot spot to exist, the critical factor is the proximityf the two metal particles. For a MED to be fabricated, the principalactor is the overall electronic character of the molecular bridge29,30]. Oligo(phenyleneethynylenes), in particular 4-[4′-(4′′-

ercapto-1-phenylethynyl)phenylethynyl]benzenethiol (OPE),ave been named the prototypical bridges for molecular electron-

cs [23,31,32–33]. The unsubstituted backbone is non-rectifying,hile substituted OPE are promising rectifiers [23,34].

Like alkanethiols, OPE has been found to form self-assembledonolayers (SAMs). SAMs continue to be vigorously studied

ecause of their applications in corrosion prevention, molecularecognition, and optical devices. Monolayers of OPE can pro-ide practical information on structure-function relationships andnterfacial dynamics. Single wave ellipsometry and contact anglesut films of OPE to be 21 ± 1 A thick [35]. OPE has been shown toorm dense SAMs that align in a herringbone fashion [34,35]. Therontier orbitals are both conducting channels [28]. From IR data,

olecular tilt and twist were deduced to be 31◦ and 33◦, respec-ively [29,23,35]. To the best of our knowledge, no sandwiches ofPE have been characterized in the literature.

In order to fabricate a rectifying device, the bridging moleculeust be uniformly oriented in the monolayer. The non-rectifyingPE backbone has been made into a rectifier by substitution on theiddle ring. Using the example of OPE–NO2, a monolayer would

pontaneously orient itself in an alternating fashion in order toeduce dipole interaction and maximize entropy. Care must beaken to ensure that this does not happen. The use of dithiolsffords a method to adsorb a monolayer to a metal substratend to deposit a second contact atop it. Pollack’s method [36]f using two protecting groups that have different chemistrieso distinguish the sulfur head groups was employed here. Inhis study, we have used SERS spectroscopy to investigate therientation of OPE sandwiched between varying combinations of

ilver surfaces. The trimethylsilylethyl TMSE, protecting group wasemoved by exposure to tetrabutylammonium fluoride (TBAF) inetrahydrofuran (THF). Sandwiching gave rise to two new bandst 1234 and 2036 cm−1 assigned to the CrIII–S stretch and the C(Cymmetric stretching, respectively.

ta Part A 78 (2011) 706–711 707

2. Experimental

OPE was synthesized as a dimer in order to take advantage ofthe ability of disulfide bonds to oxidatively cleave when exposedto a metallic surface. The synthetic details are published elsewhere[37]. Briefly, a series of Sonogashira coupling reactions were car-ried out to produce the acetylated monomers. Subsequently, theacetyl groups were removed and the disulfide formed. H1 NMR(400 MHz) spectra were recorded on a Bruker DRX400 spectrome-ter using CDCl3 solutions. H1 NMR (CDCl3, ppm) 7.52 (s, 8H), 7.46(d, 8H), 7.27 (d, 8H), 3.0 (m, 4H), 1.0 (m, 4H), and 0.1 (s, 18H).

For Raman experiments, silicon substrates were cut into 1-cm2

pieces and modified in one of the three ways. First, substrates werecleaned and 60 nm of silver was deposited to form a macroscop-ically smooth surface. Second, the Si substrate had 500 nm SiO2spheres spin-coated using an in-house designed and built coater.These surfaces were then covered with silver in the same fashionas the smooth surfaces. The last batch of Si surfaces was coveredwith GaO2 nanowires that were then coated in Ag.

SERS-active silver hydrosol was made by Lee and Meisel’s citratereduction of Ag+ [37]. Sodium citrate was added to boiling waterand allowed to dissolve. The solution was brought back to boiling.Then, silver nitrate was added to the citrate solution. That solutionwas allowed to reflux for 2 h. The solution became transparent yel-low in color. The UV–visible spectrum of the sol solution was takenand is presented in Fig. 2. The broad absorption band, centered atabout 430 nm, is characteristic of 50–60 nm silver particles. Thecolor was also consistent with this size range. The sol was storedunder argon at 4 ◦C [38].

The FT-Raman experiments were performed using a ThermoNicolet 6700 FT-IR equipped with a NXR FT-Raman Module. TheNd-YAG 1064 nm laser line was used at a power of 350 mW onthe sample. 96 accumulations at a 4 cm−1 resolution producedthe given spectrum. Spectra were collected with an InGaAs detec-tor. SERS spectroscopy was carried out with a Coherent Innova 90argon ion laser used at �ex = 514.5 nm and a power of 10 mW onthe sample. The laser was outfitted with a 514.5 nm band pass fil-ter. The scattered light was delivered to the monochromator slitvia a 514.5 nm notch filtered fiber optic bundle. SERS spectra weretaken using a Mitutoyo Ultraplan FS 110 microscope using the 100×objective with a 0.9 numerical aperture. An Ocean Optics HR4000high-resolution spectrometer was used to collect over the spectralrange set to 4 cm−1 resolution.

Electromagnetic field calculations were done using the COSMOLcomputational package.

3. Results and discussion

Fig. 1(a) shows the molecular structure of solid OPE’ withthe trimethysilylethyl (TMSE) protecting group and the disulfidelinkage. Chemisorption of OPE’ onto the silver surface results incleavage of the disulfide linkage (Fig. 1b). Creation of the nanosand-wich required the removal of the TMSE protecting group. This wasachieved by soaking the Ag/OPE, substrate in a solution of 0.1 Mtetrabutylammonium fluoride in THF. The deprotection of OPE’produced OPE (terminated as a thiolate) (Fig. 1c), (CH3)SiF, andH2C CH2. The thiolated OPE was then able to have a second contactdeposited atop the monolayer.

3.1. Deposition of Ag sol

The UV–visible absorption spectrum of the silver sol which wasused in creating the nanoassembly is shown in Fig. 2. The band at430 nm is characteristic of the plasma resonance for aggregates of60 nm silver spheres [38]. A single drop of the silver sol was added

708 M. Fletcher et al. / Spectrochimica Acta Part A 78 (2011) 706–711

surface, (c) OPE after deprotection and prior to the deposition of the Ag sol.

tT

psOapftwc

3

ottOstfis

FA

Fig. 1. (a) Molecular structure of solid OPE, (b) OPE adsorbed on a solid

o an OPE coated wafer and allowed to rest for 15, 30, and 60 min.he wafer was then washed with THF followed by EtOH.

No intercalation of Ag sol was assumed due to the denseacking of the monolayer and the extrapolated size of the silverol. However, if intercalation were to occur, the Raman spectra ofPE sandwiches should be altered due to changes in conformationnd tilt angles. The SERS spectra of the sandwiched monolayerresented in this paper strongly resemble the spectrum obtainedrom the Ag/OPE surface. Moreover, if the SERS-active Ag sol wereo have penetrated the monolayer, the b2 vibration at 1141 cm−1

ould have been enhanced by ∼102 because of the resultantharge transfer.

.2. Raman spectroscopy

To aid in the understanding of OPE as a monolayer, the spectrumf the free molecule was collected (Fig. 3). The full assignment ofhe OPE Raman bands is published elsewhere [39]. In order to provehat Figs. 4 and 5 show surface-enhanced spectra, a monolayer of

PE was adsorbed to a macroscopically smooth silver coated silicon

urface. The collected spectrum showed no identifiable peaks otherhan that of silicon (521.6 cm−1). When OPE was adsorbed to Aglm over nanospheres (AgFON) or nanowires (AgNW), and whenandwiched between an inactive surface and Ag sol, the resulting

ig. 2. UV–vis absorption spectrum of silver sol prepared by citrate reduction ofgNO3 (�max = 425 nm, 1 cm path length).

Fig. 3. FT-Raman spectrum (1064 nm excitation) of bis[4′′-trimethylsilylethyl-sulfanyl-4,4′-di(phenyleneethynylene)benzene] disulfide between 300 and2500 cm−1.

Fig. 4. SER spectra (514.5 nm excitation) of AgNS/OPE/Ag sol sandwiches at (a)15 min, (b) 30 min, and (c) 60 min of exposure to Ag sol solution. The spectra werecollected with an excitation power of 10 mW at the sample, 3-s exposure time, and4 cm−1 resolution.

M. Fletcher et al. / Spectrochimica Acta Part A 78 (2011) 706–711 709

Table 1The observed Raman lines and their vibrational assignments of the bulk and sandwiched OPE. str = stretching, oop = out-of-plane bending, bnz = benzene.

Vibrational assignment Raman shift (cm−1)

Bulk frequency Sandwich frequencya

Ag/OPE/sol AgNS/OPE/sol AgNW/OPE/sol

rI, rII, rIII, bend 402C(C oop, r–H oop 489r–H oop 526S–S str 544(C–C bend 603

725ip ring bend 739rI, rII, rIII, bend 748

838rI, rIII, bend 989rII bend 1016CrI–S str, bnz breathing 1086 1079 1076 1078CrI–S str 1131 1130 1134 1134Bnz breathing 1141 1139ring I brth, CrIII–S str 1180 1185 1183 1186CrIII–S str 1235 1234ring III str, CrIII–S str 1247CH3–Si sym def 1296(C–C str 1345Bnz ring str 1586 1590 1587 1587Bnz ring str 1599 1602Sym C(C str 2040 2037Asym C(C str 2207 2224 2214 2214S–H str 2519

sn

sscsssm

a

F1t

Sym CH3 str 2892Asym CH3 str 2950

3050

a 15 min. sol deposition, 3-s laser exposure.

pectra were indicative of OPE. We conclude the Ag sol and theanosurfaces are in fact SERS active.

The observed spectra for AgFON/OPE/Ag sol and AgNW/OPE/Agol sandwiches are shown in Figs. 4 and 5, respectively. The Ramanignal increases as deposition time of the silver sol increases untilomplete coverage is approached. Once this occurs, the transmis-ion of laser light through the Ag sol layer diminishes, and the inten-ity of the Raman signal begins to decrease. Similar trends were

een in both nanosphere/sol and nanowire/sol sandwiches. Theaximum signal we observed was after 15 min of sol deposition.After 15 min of deposition the SERS spectra were collected, and

re presented in Fig. 6. Fig. 6(c) shows the spectrum of the smooth

ig. 5. SER spectra (514.5 nm excitation) of AgNW/OPE/Ag sol sandwiches at (a)5 min, (b) 30 min, and (c) 60 min of exposure to Ag sol solution. Collected underhe same conditions as Fig. 3.

Ag/OPE/Ag sol sandwich. This is identical to the spectra of OPEadsorbed to a single SERS active contact. Fig. 6(a and b) showsthe SER spectra of the AgNW/OPE/Ag sol and AgFON/OPE/Ag solsandwiches, respectively. The AgFON surface has a single uniformlayer of spheres across the surface. Based on the curvature of thenanospheres, an aggregate could fit in the dips formed when thenanospheres come into contact. The arrangement of the spheresand the sol would allow the laser to probe the area between the sub-

strates, from where the greatest enhancement originates. While theAgFON surfaces have crevices that may increase the Raman scatter-ing signal, the possible hot spot density is greater in the nanowiresurfaces than the nanosphere surfaces. The nanowire sandwich

Fig. 6. SER spectra (514.5 nm excitation) of (a) AgNW/OPE/Ag sol, (b)AgNS/OPE/Agsol, and (c) Ag/OPE/Ag sol sandwiches with a 15-min Ag sol deposition time. Col-lected under the same conditions as Fig. 3.

710 M. Fletcher et al. / Spectrochimica Acta Part A 78 (2011) 706–711

F pheres

ptwifi

si1s11

ig. 7. Predicted enhancement of the electromagnetic field (a) around a 50 nm Ag shell and 50 nm sphere.

rovides the most intense SERS signal; this may be the result ofhe orientation of the wires on the surface. The random array ofires, best compared to dropped pick-up-sticks, creates pockets

n the topography of the surface where the wires’ electromagneticelds can couple and form the hypothesized hot spots.

While the sandwiched spectra resemble that of the AgFON/OPEpectrum, there are some differences in the spectra as shown

n Fig. 6. It is necessary to call attention to the region between000 cm−1 and 1300 cm−1 in Fig. 3. This portion of the spectrumhows four distinct peaks at 1085 cm−1, 1131 cm−1, 1141 cm−1, and180 cm−1, and a series of weak peaks, one of which is the peak at247 cm−1. In the sandwich spectra, the same regions show bands,

, (b) between a flat Ag surface and the 50 nm sphere, and (c) between a 500 nm Ag

at 1078 cm−1, 1134 cm−1, 1186 cm−1, and a new band 1234 cm−1.An examination of a wider spectral window indicates another bandat 2037 cm−1 which is absent in the AgFON/OPE spectrum.

SERS spectra obtained of analytes adsorbed on Ag sols haveshown anomalous bands the frequencies and intensities of whichhave been shown to be dependent on factors such as the reduc-ing agents used in sol preparation, the degree of aging of the sol,

and nature of the aggregating agents. Any analysis of SERS spectraobtained from sols must specifically identify and distinguish thebands which have their origin in the colloid and those which are dueto the analyte. Garcia-Ramos [40] and Sanchez-Cortez [41] identi-fied a number of SERS bands for citrate reduced colloids which they

ica Ac

aSch9fNp1tditaa[wD

amwabaoAiibpd1

3

tFii

AWfitlw

tcvhtoepac

4

Ads

[

[

[[[[[[

[[[[[[[[

[

[[

[[

[

[[[

[

[

[[

[

M. Fletcher et al. / Spectrochim

ssigned to citrate. In an earlier study of the SERS of NAD and NADHiiman et al. [42] reported anomalous bands which originated in theolloid. Yaffe and Blanch [43] in a SERS study of citrate-reduced andydroxylamine-reduced colloids identified bands at 805, 839, 925,47, 1025, 1088, 1298, 1400, 1643 and 1704 cm−1 as originatingrom the citrate-reduced colloid when the aggregating agents wereaNO3 and KNO3. Aggregating agents KNO3, K2SO4 and Na2SO4roduced bands at 614, 776, 930, 953, 1031, 1088, 1129, 1182, 1365,421, 1515, 1575, 1600, 1624 and 1652 cm−1. The frequencies ofhe new SERS bands which are observed in the sandwich assemblyo not match any of those which have been identified as result-

ng from citrate reduce colloids. These bands have their origin inhe OPE molecule which is sandwiched between the silver colloidnd silver substrate. Table 1 lists the Raman active bands of OPEnd their vibrational assignments. Assignments were based on Ref.37] and results from density functional theory calculations whichere performed on OPE using the DGAUSS B88-PW91, B88-LYP and-VWN functional of the CACHE 6.01 software package.

The previously unobserved band at 1234 cm−1 has beenssigned to the rIII-breathing mode and the CrIII–S stretchingode. The weak 1234 cm−1 band exhibits a 13 cm−1 red shifthich is reminiscent of the 1086 cm−1 peak shift, and has been

ttributed to changes in electron density when the second Ag–Sond is formed. A second previously unobserved peak appearst 2032 cm−1 and has been assigned to the asymmetric stretchf the C(C bonds. In the SERS spectra of AgNS/OPE/Ag sol andgNW/OPE/Ag sol (30-min deposition time) the broad band which

s observed in the 1230 cm−1 region of the spectra is resolvednto two bands at 1234 and 1219 cm−1. The resolution of the twoands is dependent on the Ag sol deposition time. Because of theroximity of the two bands and the fact that the DFT calculationso not predict an additional band is this spectral region the219 cm−1 band remains unassigned.

.3. Field enhancement calculations

The COSMOL modeling program was used to calculate the elec-romagnetic field around a Ag shell and between two Ag substrates.ig. 7(a–c) depicts the strengths of the fields. While there is littlencrease in the field around the sphere alone, there is a significantncrease when a metal or dielectric is introduced.

When the Ag sphere is in proximity to a macroscopically smoothg surface, the field increases by nearly five orders of magnitude.hen the sphere is put into proximity with a Ag-coated shell, the

eld increases by about four orders of magnitude. If molecules wereo get trapped between the two silver surfaces and excited by aaser, the vibrations that lie perpendicular to the surface planes

ould be severely enhanced.Contrary to the calculation results, the sphere/shell combina-

ion gave rise to a stronger SERS signal than that of the flat/shellombination. The SERS results can only be understood if the cur-ature of the substrates is taken into account. On a flat surface, theot spot that is generated is less likely to be sampled by the laserhan the hot spots in the AgFON/Ag sol sandwich. The roundnessf the nanosphere allows the sol to deposit at various angles,xposing the hot spots to the laser. While no calculations wereerformed for a nanowire/shell combination, the observed signalsre as expected—more greatly enhanced than the sphere/shellombination.

. Conclusions

It has been shown that OPE can be sandwiched between twog contacts, resembling the architecture of a molecular electronicevice. Differences between the intensities of the presented sets ofpectra can be observed. Using two silver contacts to sandwich the

[

[

[[

ta Part A 78 (2011) 706–711 711

monolayer provides two origins of enhancement. The constructiveinterference caused by the overlap of amplified electromagneticfields from the two SERS-active surfaces generates a considerableenhancement of the Raman signal. The given Raman spectra do notsuggest that any significant change in orientation occurs from thedeposition of the sol. The ring III breathing and CrIII–S stretchingmodes are significantly more enhanced when sandwiched as com-pared to the FT-Raman and the single surface enhanced spectrum.

Acknowledgements

The authors would like to thank the CREST Nanoscale Analyt-ical Sciences Education and Research Center Grant Number NSF0833127. M.F. would like to thank the NSF AGEP/Bridge to theDoctorate Program and the Department of Chemistry at HowardUniversity for financial support. C.M.H. would like to acknowledgethe ASEE/Summer Faculty Research Program at the Naval ResearchLaboratory.

References

[1] C. Orendorff, A. Gole, T. Sau, C. Murphy, Anal. Chem. 77 (2005) 3261–3266.[2] M. Muniz-Miranda, N. Neto, J. Colloid Surf. A 249 (2004) 79–84.[3] R.M. Metzger, C.A. Panetta, Synth. Met. 28 (1989) C807–C814.[4] L. Kvitek, R. Prucek, A. Panacek, J. Novotny, R. Hrbac, R. Zboril, J. Mater. Chem.

15 (2005) 1099–1105.[5] S. Emory, S. Nie, Anal. Chem. 69 (1997) 2631–2635.[6] Y. Wu, B. Zhao, W. Xu, G. Li, B. Li, Y. Ozaki, Langmuir 15 (1999) 1247–1251.[7] R.M. Metzger, C.A. Panetta, J. Mol. Electron. 5 (1989) 1–17.[8] G. Braun, S.J. Lee, S.M. Dante, T.-Q. Nguyen, M. Moskovits, N. Reich, J. Am. Chem.

Soc. 129 (2007) 6378–6379.[9] F. Svedberg, Y. Alaverdyan, P. Johansson, M. Kall, J. Phys. Chem. B 110 (2006)

25671–25677.10] C.R. Yonzon, C.L. Haynes, X. Zhang, J.T. Walsh, R.P. Van Duyne, Anal. Chem. 76

(2004) 78–85.11] W.-W. Zhang, X.M. Ren, H-F. Li, C.S. Lu, C.J. Hu, H.Z. Zhu, Q.J. Meng, Colloid

Interface Sci. 25 (2002) 150–157.12] L.C. Lu, Electrochem. Commun. 7 (2005) 1071–1076.13] K.L. Kim, S.J. Lee, K. Kim, J. Phys. Chem. B 108 (2004) 9216–9220.14] L. Baia, M. Baia, J. Popp, S. Astilean, J. Phys. Chem. B 110 (2006) 23982–23986.15] H. Luo, M.J. Weaver, Langmuir 15 (1999) 8743–8749.16] S. Sanchez-Cortes, J.V. Garcia-Ramos, Langmuir 16 (2000) 764–770.17] J.F. Arenas, D.J. Fernandez, J. Soto, I. Lopez-Tocon, J.C. Otero, J. Phys. Chem. B

107 (2003) 13143–13149.18] S.J. Lee, A.R. Morrill, M. Moskovits, J. Am. Chem. Soc. 128 (2006) 2200–2201.19] K.A. Bosnick, J. Jiang, L.E. Brus, J. Phys. Chem. B 106 (2002) 8096–8099.20] H. Wang, C.S. Levin, N.J. Halas, J. Am. Chem. Soc. 127 (2005) 14992–14993.21] H.-Y. Yu, J. Zhang, H.L. Zhang, Z.F. Liu, Langmuir 15 (1999) 16–19.22] J.K. Daniels, G. Chumanov, J. Phys. Chem. B. 109 (2005) 17936–17942.23] K. Kim, J.K. Yoon, J. Phys. Chem. B 109 (2005) 20731–20736.24] C.D. Keating, K.K. Kovaleski, M.J. Natan, J. Phys. Chem. B 102 (1998) 9414–9425.25] Q. Wang, S. Ward, A. Duda, J. Hu, P. stradins, R.S. Crandall, H.M. Branz, C. Perlov,

W. Jackson, P. Mei, C. Taussing, App. Phys. Lett. 85 (2004) 2122–2124.26] (a) S. Hou, J. Ning, Z. Shen, X. Zhao, Z. Xue, Chem. Phys. 327 (2006) 1–9;

(b) L. Kim, J. Kim, G.H. Gu, J.S. Suh, Chem. Phys. Lett. 426 (2006) 137–141.27] J. Semenario, Nat. Mater. 4 (2005) 111–113.28] C. Hacker, J. Batteas, J. Garno, M. Marquez, C. Richter, L. Richter, R. van Zee, C.

Zangmeister, Langmuir 20 (2004) 6195–6209.29] R.L. McCreery, Chem. Mater. 16 (2004) 4477–4496.30] K. Kneipp, H. Kneipp, R. Manoharan, I. Itzkan, R. Dasari, M. Feld, J. Raman Spec-

trosc. 29 (1998) 743–747.31] W.-W. Zhang, X.-M Ren, H.-F Li, C.-S Lu, C.-J. Hu, H.-Z. Zhu, Q.J. Meng, J. Colloid

Interface Sci. 255 (2002) 150–157.32] Y.C. Lu, Electrochem. Commun. 7 (2005) 1071–1076.33] R. Zehner, L. Sita, Langmuir 13 (1997) 2973–2979.34] J. Stapleton, P. Harder, T. Daniel, M. Reinard, Y. Yao, D. Price, J. Tour, D. Allara,

Langmuir 19 (2003) 8245–8255.35] F.-R. Fan, J. Yang, S. Dirk, D. Price, D. Kosynkin, J. Tour, A. Bard, J. Am. Chem. Soc.

123 (2001) 2454–2455.36] S. Pollack, J. Naciri, J. Mastrangelo, C. Patterson, J. Torres, M. Moore, R. Shashid-

har, J. Kushmerick, Langmuir 20 (2004) 1838–1842.37] P.C. Lee, D. Meisel, J. Phys. Chem. 86 (1982) 3391–3395.38] M. Fletcher, A. Vivoni, M. Moore, J. Lui, S. Prokes, O. Glembocki, C.M. Hosten,

Surf. Sci. 602 (2008) 1614–1621.39] M. Pickholz, S. Stafstrom, Chem. Phys. 270 (2001) 245–251.

40] R.F. Aroca, R.A. Alvarez-Puebbla, N. Pieczonka, S. Canchez-Cortez, J.V. Garcia-

Ramos, Adv. Colloidal Interface Sci. 116 (2005) 116, 45.41] M.V. Canamares, J.V. Garcia-Ramos, J.D. Gomez-Varga, C. Domingo, S. Sanchez-

Cotes, Langmuir 21 (2005) 8546.42] O. Siiman, R. Rivellini, R. Patel, Inorg. Chem. 27 (1988) 3940.43] N. Yaffe, E.W. Blanch, Vib. Spectrosc. 48 (2008) 196.