Anal. Chem. 2007, 79, 8584-8589
Linker-Molecule-Free Gold Nanorod Layer-by-Layer
Films for Surface-Enhanced Raman Scattering
Sukang Yun, Yong-Kyun Park, Seong Kyu Kim,* and Sungho Park*
Department of Chemistry, BK21 School of Chemical Materials Science and SKKU Advanced Institute of Nanotechnology,
Sungkyunkwan University, Suwon 440-746, South Korea
This paper reports a methodology for synthesizing and
ordering gold nanorods into two-dimensional arrays at a
water/hexane interface. This preparation method allows
the systematic control of the nanoparticle film thickness.
An investigation into the thickness-dependent surfaceenhanced Raman scattering (SERS) of the adsorbed
molecules revealed the nanorod (NR) films to have 1
order of magnitude stronger SERS enhancement than the
nanosphere (NS) under similar experimental conditions.
The exposed surface areas of the prepared NR and NS
films were analyzed using electrochemical methods, and
it was found that they had comparable exposed surface
areas. Therefore, the order of magnitude difference in the
enhancement factor was not due to the differences in
surface area but to their intrinsic difference in the optical
coupling of each film. The difference was attributed to the
high density of junction points with the NR films in
comparison with the corresponding NS films. Scanning
emission microscopy showed that the NR films have line
contacts with each other but the NS films have point
contacts, which can explain the difference in SERS
intensity between the NR and NS films.
The ability to acquire vibrational information for molecules
present in a monolayer (or even a single molecule) at surfaces or
within thin films has long been recognized as being essential to
the development of interfacial chemistry, sensors, and other
related applications. Among the plethora of vibrational spectroscopic techniques, surface-enhanced Raman scattering (SERS) has
attracted considerable attention since its introduction in 1977 due
to its unique advantages over other surface characterization
approaches.1,2 The SERS technique allows adsorbate vibrational
spectra to be obtained over a wide frequency range, and the more
liberal surface selection rules enable the detection of vertical, tilted,
and even “flat” oriented vibrational modes.3,4 Furthermore, the
apparent Raman cross sections of the adsorbates are enhanced
enormously (∼106-fold) by local electromagnetic fields when
adsorbates are on properly roughened coinage metals, Cu, Ag,
and Au.3,4 Theoretically, these enhancements have been attributed
* Corresponding authors. E-mail: [email protected] (S.K.K.); spark72@skku.
edu (S.P.). Fax: 82-31-290-7075.
(1) Jeanmaire, D. J.; Van Duyne, R. P. J. Electroanal. Chem. 1977, 84, 1.
(2) Albrecht, M. G.; Creighton, J. A. J. Am. Chem. Soc. 1977, 99, 5215.
(3) Chang, R. K.; Furtak, T. E. Surface Enhanced Raman Scattering; Plenum
Press: New York, 1982.
(4) Zou, S.; Weaver, M. J. Anal. Chem. 1998, 70, 2387.
8584 Analytical Chemistry, Vol. 79, No. 22, November 15, 2007
to two models, “electromagnetic (EM) mechanism” and “chemical
mechanism (CM)”.5 CM enhancement is due partly to electronic
resonance charge transfer between the adsorbate and metal
surface, where electrons are moved from the highest occupied
molecular orbital (HOMO) of the adsorbates to the metal (or vice
versa, from the metal to the lowest unoccupied molecular orbital
(LUMO)).5 This leads to an increase in the polarizability of the
adsorbate and an increase in the Raman scattering cross section.
In contrast to CM enhancement, there has been considerable
development with the EM enhancement theory. The EM enhancement theory states that the surface plasmons of the metal
nanostructures, particularly in an ensemble of such structures,
greatly enhance the local EM field intensities at the surface of
the particle and contribute to SERS enhancement.6 Recent calculations and experimental data have shown that the EM field can be
greatly enhanced at the gaps or the junctions between nanostructures, which are often called “hot spots”.6 As the interparticle
distance between two nanoparticles decreases, the degree of their
surface plasmon coupling increases, and the enhanced fields of
each particle begin to interfere coherently at the junctions.
Therefore, it is generally understood that in a number of systems,
aggregates of nanoparticles are better substrates for SERS
applications than individual nanoparticles due to the presence of
junctions.
Given that these considerations are important for large Raman
enhancements, there has been considerable research effort in
SERS focusing on the controlled and reproducible fabrication of
metallic nanostructures that generate such hot junctions. Although
an elaborate lithographic technique is a useful approach for
generating complex nanostructures, the fabrication of periodic
nanostructures with an interparticle distance of a few nanometers
is not practically feasible. Therefore, most efforts have focused
on the controlled assembly of nanoparticles synthesized using a
wet-chemical method.7-17 Traditionally, spherical nanoparticles
(5) Campion, A.; Kambhampati, P. Chem. Soc. Rev. 1998, 27, 241.
(6) Jiang, J.; Bosnick, K.; Maillard, M.; Brus, L. J. Phys. Chem. B 2003, 107,
9964.
(7) Murphy, C. J.; Gole, A. M.; Hunyadi, S. E.; Orendorff, C. J. Inorg. Chem.
2006, 45, 7544.
(8) Aroca, R. F.; Goulet, P. J. G.; dos Santos, D. S., Jr.; Alvarez-Puebla, R. A.;
Oliveira, O. N., Jr. Anal. Chem. 2005, 77, 378.
(9) Suzuki, M.; Niidome, Y.; Kuwahara, Y.; Terasaki, N.; Inoue, K.; Yamada, S.
J. Phys. Chem. B 2004, 108, 11660.
(10) Nikoobkht, B.; El-Sayed, M. A. J. Phys. Chem. A 2003, 107, 3372.
(11) Lee, S. J.; Morrill, A. R.; Moskovits, M. J. Am. Chem. Soc. 2006, 128, 2200.
(12) Hu, X.; Cheng, W.; Wang, T.; Wang, Y.; Wang, E.; Dong, S. J. Phys. Chem.
B 2005, 109, 19385.
10.1021/ac071440c CCC: $37.00
© 2007 American Chemical Society
Published on Web 10/16/2007
have been the main focus of such research.18,19 Currently, synthetic
routes for anisotropic nanoparticles such as rods, prisms, and
cubes are well-established, and their corresponding application
as SERS substrates is an interesting research topic.20 In line with
this research direction, nanorod aggregates were reported to show
stronger SERS enhancement than spherical nanoparticles under
similar experimental conditions.21
Recently, it was reported that nanoparticles synthesized in an
aqueous medium can be assembled into two-dimensional arrays
at the liquid-liquid interfaces, which is induced by the destabilization of nanoparticles.22-24 Destabilization was achieved by adding
alcohol to an aqueous nanoparticle suspension. This step led to a
contact angle variation of nanoparticles at the interfaces, which
approaches 90° at the liquid-liquid interfaces. Herein, we expand
this strategy to nanorod systems and show how a controlled
aggregate of gold nanorods with a relatively clean surface and
controlled nanorod density on a solid substrate can be achieved.
The comparison of the relative enhancement factors as SERS
substrates, which were obtained from the resulting nanorod (NR)
and nanosphere (NS) arrays, was made as a function of nanoparticle film thickness. This paper reports a method for obtaining
two- or three-dimensional NR arrays with clean surfaces in addition
to the issues related to the greater SERS enhancement from the
NR aggregates compared with NS ones.
EXPERIMENTAL SECTION
All chemicals used in this study were obtained from Aldrich
and used as received. The gold NS sol (d ) 13 ( 2 nm) was
prepared using the following conventional synthetic route. Briefly,
100 mL of a 1.0 mM aqueous HAuCl4‚3H2O solution was added
to 100 mL of triply deionized water (Millipore) and then boiled.
A 10 mL solution of 38.8 mM sodium citrate was added to the
solution which was followed by boiling for 20 min. The average
particle diameters were determined by transmission electron
microscopy (TEM) analysis. A gold NR sol was prepared using
the synthetic method reported elsewhere.25 A 200 mL aqueous
solution of 0.5 mM HAuCl4‚3H2O and 0.1 M cetyl-trimethylammonium bromide (CTAB) was mixed with 4 mL of an aqueous
5.89 mM AgNO3 solution. A volume of 1.2 mL of 0.1 M ascorbic
acid was then added. Finally, a reducing agent, 80 µL of a
(13) Schwartzberg, A. M.; Grant, C. D.; Wolcott, A.; Talley, C. E.; Huser, T. R.;
Bogomolni, R.; Zhang, J. Z. J. Phys. Chem. B 2004, 108, 19191.
(14) Yang, Y.; Xiong, L.; Shi, J.; Nogami, M. Nanotechnology 2006, 17, 2670.
(15) Orendorff, C. J.; Gearheart, L.; Jana, N. R.; Murphy, C. J. Phys. Chem. Chem.
Phys. 2006, 8, 165.
(16) Jeong, D. H.; Zhang, Y. X.; Moskovits, M. J. Phys. Chem. B 2004, 108,
12724.
(17) Tao, A.; Kim, F.; Hess, C.; Goldberger, J.; He, R.; Sun, Y.; Xia, Y.; Yang, P.
Nano Lett. 2003, 3, 1229.
(18) Park, S.; Yang, P.; Corredor, P.; Weaver, M. J. J. Am. Chem. Soc. 2002,
124, 2428.
(19) Grabar, K. C.; Smith, P. C.; Musick, M. D.; Davis, J. A.; Walter, D. G.;
Jackson, M. A.; Guthrie, A. P.; Natan, M. J. J. Am. Chem. Soc. 1996, 118,
1148.
(20) Orendorff, C. J.; Gole, A.; Sau, T. K.; Murphy, C. J. Anal. Chem. 2005, 77,
3261.
(21) Nikoobakht, B.; Wang, J.; El-Sayed, M. A. Chem. Phys. Lett. 2002, 366, 17.
(22) Duan, H.; Wang, D.; Kurth, D. G.; Mohwald, H. Angew. Chem., Int. Ed. 2004,
43, 5639.
(23) Reincke, F.; Hickey, S. G.; Kegel, W. K.; Vanmaekelbergh, D. Angew. Chem.,
Int. Ed. 2004, 43, 458.
(24) Li, Y.-J.; Huang, W.-J.; Sun, S.-G. Angew. Chem., Int. Ed. 2006, 45, 2537.
(25) Zijlstra, P.; Bullen, C.; Chon, J. W. M.; Gu, M. J. Phys. Chem. B 2006, 110,
19315.
1.6 mM NaBH4 solution, was added to the mixture at room
temperature and without agitation. The reaction was completed
after 3 h, which was signaled by a color change from colorless to
a weak brown color. The length and width were determined from
field emission scanning electron microscopy (FESEM) to be
52 ( 5 nm and 14 ( 2 nm, respectively.
Field emission scanning electron microscopy and TEM images
were obtained using a JEOL 7000F and JEOL JEM-3011, respectively. The electrode potentials were measured and are reported
versus a Ag/AgCl (saturated KCl) reference electrode (from
Autolab AUT12, and Bioanalytical Systems, respectively).
A micro-Raman spectrometer (Renishaw, InVia) was used to
record the SERS. A He-Ne laser at 632.8 nm was used as the
excitation laser. The laser power was reduced significantly by
lowering the tube current, which was further attenuated to
<0.5 mW by pinholes, optics, and filters before reaching the
sample. A 50× objective lens with a numerical aperture value of
0.75 focused the excitation beam onto an area of approximately
1 µm2. The scattered light was collected using the same objective
lens and passed though a double supernotch filter (Rayleigh
rejection bandwidth ,50 cm-1) before being introduced into a
polychromator with an 1800 grooves/mm grating and a Peltiercooled CCD detector. Each Raman spectrum was obtained with
a 30 s exposure time of the detector.
Figure 1 shows a schematic diagram of the steps involved in
preparing the close-packed two-dimensional nanoparticle arrays
with the corresponding experimental optical images. A 25 mL
sample of the gold nanoparticle aqueous solution was transferred
to a Teflon cell (inner dimension, 7.5 × 4.0 × 1.5 cm3), and
10 mL of hexane was added to the top of a colloid solution surface
in order to form an immiscible water/hexane interface. Then,
6 mL of ethanol was added dropwise to the surface of the water/
hexane layers (0.6 mL/min, using a mechanical syringe pump,
KDS101 from kdScienctific Inc.), leading to nanoparticle trapping
at the interfaces.
It was possible to tune the resulting nanoparticle twodimensional area depending on the amount of ethanol added. After
forming the desired area of nanoparticle array at the interfaces,
the hexane layer was evaporated. In this process, the nanoparticles
naturally assembled into close-packed islands, leading to the
appearance of a mirror-like metallic sheen due to optical coupling
of the nanoparticles. The resulting nanoparticle film was packed
into one large film as a monolayer using two Teflon bars until
the mirror reflection was observed. These films could then be
transferred onto a solid substrate by horizontal lifting for in-depth
analysis of the film profile by FESEM and TEM. Glass and silicon
wafer slides were used for the UV-vis absorption spectra and
SERS measurements, respectively. The substrates were modified
with 3-mercaptopropyltrimethoxysilane (MPTMS) before nanoparticle transfer, yielding thiol-terminated surface functionalities.18
RESULTS AND DISCUSSION
When the nanoparticles are hydrophilic they have a contact
angle of <90° at a water/oil interface and are suspended in the
water phase. When their contact angle approaches 90°, the
particles tended to adsorb to the water/oil interface. A highdielectric medium can effectively separate the charged particles
from each other. When the dielectric constant of medium is
decreased by the addition of a miscible solvent with a lower
Analytical Chemistry, Vol. 79, No. 22, November 15, 2007
8585
Figure 1. Schematic representation of the gold NR film formation at the water/hexane interface. The corresponding optical images are
represented in the second row.
Figure 2. ζ-Potential data as a function of added ethanol volume
rations for gold NSs (filled circles) and NRs before (open triangles)
and after (open squares) rinsing with a centrifugation.
dielectric constant, the surface charge of the charged particles
gradually decreases as a function of the amount of solvent added.
When ethanol is added, the surface charge density of the
nanoparticles decreases, which leads to the adsorption of nanoparticles to the interface. The decrease in interfacial energy at
the water/oil interface by the adsorption of nanoparticles is the
driving force for the entrapment of nanoparticles upon the
reduction of the dielectric constant of water.
The prepared NSs and NRs were stabilized by citrate anions
and CTAB cations, respectively. Therefore, the NSs and NRs have
negative and the positive surface charges due to the adsorbed
stabilizing agents, respectively. This feature was confirmed by
ζ-potential measurements, as shown in Figure 2. Before adding
ethanol, the NS solution had a ζ-potential of < -40 mV. Upon
the further addition of ethanol, this value increased to approximately -30 mV. With the NS system, the easy formation of
entrapped NSs at the interface can be understood by the facile
reduction of the surface charge upon the addition of ethanol. In
contrast to this system, the NRs showed a different profile with
the ζ-potential measurements. The initial value before adding the
ethanol was greater than +50 mV. Upon the addition of ethanol,
8586
Analytical Chemistry, Vol. 79, No. 22, November 15, 2007
this value decreased with a small slope. This means that the
destabilization of NRs is not as facile as the NS system due to the
firm capping of CTAB on the surface of the NRs. The destabilization originates from the high concentration (approximately
0.1 M) of CTAB in the NR solution. When the CTAB concentration
was decreased to 0.1 mM by rinsing with centrifugation, the
surface charge decreased as readily as the NS system upon the
addition of ethanol. This is consistent with the experimental
observations for the formation of an NR film at the interface
according to the synthetic route shown in Figure 1. Before
reducing the concentration of CTAB to ca. 0.1 mM, it was difficult
to form NR films at the interface. It should be noted that it is
important to reduce the concentration of the surfactant for the
easy formation of nanoparticle films at the interface given that
the many synthetic routes for anisotropic nanoparticles use a high
concentration of surfactant to induce anisotropic nanoparticle
growth. It is clear that the decrease in surface charge plays a key
role in inducing nanoparticle film formation.
FESEM was used to obtain more in-depth information regarding the film profile of the transferred nanoparticle films (Figure
3). The FESEM image of one monolayer of NR film shows a
homogeneous morphology over a large area, containing a closepacked nanoparticle architecture with small voids. This shows a
monolayer without significant aggregation. The analogous NS film
shown in Figure 3B shows comparable nanoparticle packing with
some voids. The appearance of voids is due to the competitive
electrostatic repulsion against the long-range van der Waals
interactions. The typical distance between the nanoparticles was
ca. 0 to ∼3 nm range without considering the voids. Figure 3,
parts C and E, shows FESEM images of the multilayer NR films
with two and four monolayers, respectively. The images show the
films to have a porous and networked architecture, retaining the
individual NR shape without significant aggregation. The multilayer NS films also showed a similar morphology to the NRs, as
shown in Figure 3, parts D and F. No large aggregate of particles
was observed in any of the nanoparticle samples. These results
suggest that the transferred nanoparticles retain a trace amount
of adsorbate molecules.
Figure 3. FESEM images for gold NR films with (A) one, (C) two,
and (E) four monolayers, and the corresponding images for NS films
with (B) one, (D) two, and (F) four monolayers on silicon wafer
substrates.
Figure 4. (A) UV-vis absorption spectra of gold nanospheres (NSs)
and nanorods (NRs) in solution phase. The diameter of the NS
particles is 13 ( 2 nm. The length and diameter of the NRs are 52 (
5 nm and 14 ( 2 nm, respectively. (B) UV-vis absorption spectra
(on glass slides in air) for different numbers of gold NR monolayer
films (from bottom to top, one, two, three, and four layers). (C) The
corresponding spectra for NS films.
The spectrum of NRs in the aqueous medium show two distinct
absorption bands centered at 510 and 806 nm, as shown in Figure
4A. The bands at the shorter and longer wavelengths were
assigned to the transverse surface plasmon mode and longitudinal
surface plasmon modes of NRs, respectively. The absorption bands
were fairly narrow, representing the monodispersity of NRs, which
is consistent with the observation from the FESEM images. Figure
4B shows the UV-vis spectra of the NRs on a glass substrate as
a function of the number of monolayers. In contrast to the
spectrum obtained in the aqueous phase, the NR films showed
different absorption profiles. The peak positions of the dried NR
films showed a red-shift compared with the NRs in an aqueous
solution. The transverse and longitudinal surface plasmon bands
appeared at 746 and 1214 nm, respectively. According to the
reported experimental and theoretical considerations for gold NR
films,26 this was attributed to the strong surface plasmon dipole
coupling among closely neighboring NRs. In comparison with
previous reports on NR films on solid substrates, the NR films
assembled in this study showed several noticeably different
features. Most studies used a dip-coating method, in which a solid
substrate covered with negatively charged materials, such as poly
electrolytes, was immersed into an NR suspension for spontaneous
immobilization.27,28 In those systems, the resulting monolayer NR
films were formed by an electrostatic interaction and were well
separated from each other. The typical surface coverage after
immersion for 14 h was <50%, which was attributed to steric
mismatch between the empty spaces between the preadsorbed
NRs and additional rods, in addition to the electrostatic repulsion
from the preadsorbed NRs.28 The NR films typically showed a
surface coverage of >50%, and the preparation step required a
relatively short period of time (typically, a few minutes). The high
surface NR coverage of the monolayer led to a significant redshift that originated from the heavy optical coupling between NRs.
This was induced by a short inter-NR distance. These NR films
showed a broader bandwidth, which also originates from the high
NR surface coverage. As evident in the FESEM image (Figure
3A), there were large numbers of contact points between the NRs,
which are tip-to-tip, side-to-side, and tip-to-side. Band broadening
occurred with increasing degree of contact points. The spectra
of the multilayered NR films showed a gradual increase in
absorbance without noticeable peak shifts. The linear increase (a
correlation coefficient for linearity, R2 ) 0.9923) in absorbance
as a function of the number of transferred NR monolayers clearly
shows that each transfer step was successful and an equal amount
of NRs had been transferred. The invariable peak positions for
the transverse and longitudinal modes highlight the low degree
of optical coupling between the NRs in the direction perpendicular
to the substrate. Because the UV-vis beam pathway is also
perpendicular to the substrate, optical coupling between the first
and subsequently transferred NR films should not be represented
in these spectra. This feature was obvious in the UV-vis spectra
for the multilayered NR films. In other studies, the buildup of
multilayer structures was achieved through the alternate deposition of oppositely charged anionic and cationic polyelectrolytes.12,27,28 In these studies, the measured UV-vis spectra of the
resulting NR films showed complex features depending on the
number of transferred layers.12,28 A comparison of these results
with the reported ones showed that the complex features resulted
from the low coverage after each immobilization step. In the
present system, the high surface coverage of NRs per each layer
prevents the additional filling up of NRs in the voids of the
predeposited NR films due to the insufficient void areas. Each
(26) Jain, P. K.; Eustis, S.; El-Sayed, M. J. Phys. Chem. B 2006, 110, 18243.
(27) Gole, A.; Murphy, C. J. Chem. Mater. 2005, 17, 1325.
(28) Vial, S.; Pastoriza-Santos, I.; Perez-Juste, J.; Liz-Marzan, L. M. Langmuir
2007, 23, 4606.
Analytical Chemistry, Vol. 79, No. 22, November 15, 2007
8587
Figure 5. (A) SERS spectra of adsorbed mercaptobenzene on a
different numbers of gold NR and NS monolayers (ML). (B) A graph
for SERS intensity of a band at 1583 cm-1 vs different numbers of
NR and NS monolayer films.
layer simply contributes to the increase in the NR concentration,
and a simple increase in absorbance was expected. A similar
tendency was observed with the NS system, as shown in Figure
4C. The peak position of the NS films showed a red-shift to 694
nm compared with NSs in aqueous solution. The resulting NS
multilayers also showed a systematic increase in absorbance with
increasing thickness of the NS films without a shift in the peak
position.
Of essential interest here is the comparison of the SERS
spectra obtained as a function of the number of NR and NS
monolayers. A systematic comparison of SERS enhancement
depending on the nanoparticle shape (NRs and NSs) and the
corresponding film thickness was performed using adsorbed
benzenethiol as a model system. Figure 5A shows representative
SERS spectra of benzenethiol adsorbed on a single monolayer
and eight monolayers of the NR film. A comprehensive peak
assignment of the spectral features has been reported and will
not be described in this report29,30 because the aim of this study
was to compare the relative SERS enhancement on various
nanoparticle films. It was demonstrated that aggregates of nanoparticles induce strong SERS enhancement due to the large EM
fields at the junctions of the nanoparticles.6,9,11-13,16,21 In such
ensembles of nanoparticles in the aggregate state, where the
plasmon modes interact closely, it was theoretically shown that
the highly localized plasmon modes could be generated at the
gaps between the nanoparticles. These sites are often referred to
as “hot spots”. Because these NR films have high coverage and
a close inter-NR distance (0 to ∼3 nm), these metallic layers are
expected to show large local EM field enhancement, mainly at
the gaps between the NRs. The short internanoparticle distance
plays an important role in the enhancement, particularly when
the separation length (L) between two metallic objects is less than
the radius of the object.31 It was also shown that the average
enhancement grew from a factor of 103 for isolated cylinders to a
maximum of 106 in the case L ) 0 where the cylinders were
touching.31 Therefore, these NR films are expected to show the
maximum enhancement compared with other analogous films
prepared using different preparation methods. The SERS intensity
of the adsorbate on the NR multilayer films showed a linear
(29) Joo, T. H.; Kim, M. S.; Kim, K. J. Raman Spectrosc. 1987, 18, 57.
(30) Taylor, C. E.; Pemberton, J. E.; Goodman, G. G.; Schoenfisch, M. H. Appl.
Spectrosc. 1999, 53, 1212.
(31) Garcia-Vidal, F. J.; Pendry, J. B. Phys. Rev. Lett. 1996, 77, 1163.
8588 Analytical Chemistry, Vol. 79, No. 22, November 15, 2007
Figure 6. (A) Cyclic voltammograms of one monolayer of NR (solid
traces) and NS (dashed traces) in 0.1 M H2SO4 solution with a scan
rate of 50 mV/s. (B) Raman spectra of an NR film prepared at the
water/hexane interface (bottom), one prepared by drop-casting of NR
solution (middle), and solid CTAB (top). The symbol (/) indicates a
band from a silicon wafer substrate.
increase with increasing number of layers up to eight monolayers
(Figure 5B). For this figure, three samples for each plot point
were utilized. Also, the film homogeneity for SERS scattering was
tested with repeated measurements on different spots with an
individual sample. As shown in the Figure 5B, the standard
deviation for each measurement is within 10%, showing the good
reproducibility. The good reproducibility resulted from the
relatively homogeneous morphology of the films compared to
those from other conventional preparation methods, such as
electrochemical oxidation and reduction cyclings. The SERS
intensity leveled off or decreased when the number of film layers
exceeded eight. This phenomenon is complicated by the diffusion
of light and penetration of the analyte through the nanopores
structures. However, the key feature is the linear increase in SERS
enhancement up to a certain point. From this tendency, the
relationship, y ) Ax, can be easily obtained, where y is the
enhancement, A is the enhancement factor per nanoparticle
monolayer, and x is the number of nanoparticle layers. For NR
films, the A value obtained from the linear fit was ca. 104/layer.
This value is less than the calculated value (106) but comparable
to the experimental enhancement factor from NRs reported by
other workers.21,31 This value is dependent on experimental
conditions, such as the laser power, excitation wavelength,
acquisition time, etc. In addition, the analyte coverage in a given
laser spot size should be included in order to calculate the
conventional enhancement factors by comparing the normal
Raman signal intensities. However, a qualitative comparison of
the enhancement factors with other shaped nanoparticles under
identical experimental conditions can be made. The NS film
showed an A value of approximately 103/layer obtained from
Figure 5B under the same experimental conditions, which is an
order of magnitude lower than that observed from the NR films.
Electrochemical analysis was used to check if the exposed NR
and NS films surface areas to the laser spot were comparable, as
shown in Figure 6A. The cyclic voltammograms were obtained
in a 0.1 M H2SO4 solution versus a Ag/AgCl reference electrode
and Pt as a counter electrode. Typical gold oxidation and reduction
peaks were obtained in the anodic and cathodic scan directions,
respectively. The integrated charge densities calculated from the
oxide reduction peak were 3.7 × 10-7 C/cm2 and 3.6 × 10-7 C/cm2
for the NS and NR monolayer films, respectively. The oxide
reduction peak area increased with increasing exposed surface
area. It is obvious that there was no difference in the exposed
surface area for a given geometrical area. Therefore, the 1 order
of magnitude difference in the enhancement factor A was not due
to differences in surface area but to their intrinsic difference in
the optical coupling of each film. As shown in Figure 5A, the NR
films showed stronger SERS signals than the NS films at
comparable film thicknesses. This difference might be due to the
high density of junction points with the NR films compared with
the corresponding NS films. As clearly shown in the FESEM
images, the NR films have line contacts with each other but the
NS films have point contacts. This observation can partly explain
the difference in SERS intensity between the NR and NS films.
Another property of these NR films is the low degree of
interference from the stabilizing agents, such as CTAB. The dropcast NR films showed a great deal of background signals from
the adsorbed CTAB molecules. Although those features decrease
when strongly adsorbing analytes are introduced, it can be
problematic when the analytes have weak adsorption energy.
Furthermore, in order to induce maximum CM enhancement, a
relatively clean nanoparticle surface is important for effective
electron transfer between the analytes and metal surface. Figure
6B shows that the NR films prepared at the interface exhibit clean
background features compared with the drop-cast NR films.
preparation methods, such as the layer-by-layer approach, in which
the buildup of multilayer structures was achieved through the
alternate deposition of oppositely charged anionic and cationic
polyelectrolytes, the assembly method at the water/oil interface
allowed nanoparticle films to be fabricated easily with a shorter
time period and a cleaner surface nature. These features, which
resulted from the absence of linker molecules, are important for
the applications of such structures as SERS substrates. A short
internanoparticle distance is essential for maximizing the EM field,
and the clean surface is important for facile electron transfer
between the adsorbed molecules and metal surface. The latter
feature will play an important role in the process where the CM
is dominant. In addition, the well-controlled two-dimensional
assembly enables a systematic comparison of the SERS enhancement factors of nanoparticle aggregates. The enhancements were
dependent on the shape and size of such building blocks on the
nanometer scale. The suggested enhancement factor per layer
(A) depending on the particle shape could be obtained from a
plot of the peak intensity and the number of layers. It is expected
that a comparison of the SERS enhancement factors in aggregated
forms with other shaped nanostructures, such as prisms, cubes,
and wires, will provide a new methodology for highly effective
SERS measurements.
CONCLUSIONS
It should be noted that since SERS is a collective effect
emerging in surfaces composed of closely interacting nanostructures, there should be a general synthetic route for forming
controlled aggregates. In addition, methods for systematically
comparing the SERS enhancement on such aggregates are needed
in order to fabricate an effective SERS substrate with various
nanoparticle shapes. The current example shows how a controlled
aggregate of isotropic and anisotropic nanoparticles can be
prepared as two- or three-dimensional films with a high packing
density and relatively clean surface. In comparison with other
ACKNOWLEDGMENT
This work was supported by the Korea Research Foundation
Grant funded by the Korean Government (MOEHRD, KRF-2005005-J11902, and KRF-C00050) and the Korea Science and Engineering Foundation (R01-2006-000-10426-0-2006). S. Park thanks
SKKU for start-up funds. S. K. Kim thanks the KOSEF-SRC
program (Center for Nanotubes and Nanostructured Composites).
Received for review July 6, 2007. Accepted September 8,
2007.
AC071440C
Analytical Chemistry, Vol. 79, No. 22, November 15, 2007
8589