Silica-Void-Gold Nanoparticles: Temporally Stable
Surface-Enhanced Raman Scattering Substrates
Maryuri Roca and Amanda J. Haes*
Department of Chemistry, UniVersity of Iowa, Iowa City, Iowa 52242
Received July 28, 2008; E-mail: firstname.lastname@example.org
Abstract: Reproducible detection of a target molecule is demonstrated using temporally stable solution-
phase silica-void-gold nanoparticles and surface-enhanced Raman scattering (SERS). These composite
nanostructures are homogeneous (diameter ) 45 ( 4 nm) and entrap single 13 nm gold nanoparticle
cores inside porous silica membranes which prevent electromagnetic coupling and aggregation between
adjacent nanoparticles. The optical properties of the gold nanoparticle cores and structural changes of the
composite nanostructures are characterized using extinction spectroscopy and transmission electron
microscopy, respectively, and both techniques are used to monitor the formation of the silica membrane.
The resulting nanostructures exhibit temporally stable optical properties in the presence of salt and
2-naphthalenethiol. Similar SERS spectral features are observed when 2-naphthalenethiol is incubated
with both bare and membrane-encapsulated gold nanoparticles. Disappearance of the S-H Raman
vibrational band centered at 2566 cm-1with the composite nanoparticles indicates that the target molecule
is binding directly to the metal surface. Furthermore, these nanostructures exhibit reproducible SERS signals
for at least a 2 h period. This first demonstration of utilizing solution-phase silica-void-gold nanoparticles
as reproducible SERS substrates will allow for future fundamental studies in understanding the mechanisms
of SERS using solution-phase nanostructures as well as for applications that involve the direct and
reproducible detection of biological and environmental molecules.
Since its discovery over 30 years ago, surface-enhanced
Raman scattering (SERS) has become a valuable but limited
spectroscopic technique for the sensitive detection of molecules.1,2
Similar to normal Raman spectroscopy, SERS provides both
chemical identification and structural information of a molecule
based on its unique vibrational fingerprint. Although Raman
spectroscopy provides highly specific information regarding a
molecule, its use is limited because of inherently low signal
intensities and small molecular cross sections versus other
spectroscopic techniques.3In comparison to normal Raman
scattering, SERS has been shown to increase the magnitude of
the Raman effect for a given analyte by up to 106to 1014orders
of magnitude1,2,4and has contributions from both chemical and
electromagnetic enhancement effects.5-7
Control over the composition, shape, size, and local environ-
ment surrounding the metallic substrate is vital for achieving
consistent SERS enhancements. In the past 15 years, much
progress has been made in controlling these properties via both
fabrication2,8and synthetic9-12techniques. The simplicity of
nanoparticles synthesized by metal salt reduction techniques has
warranted their widespread use in SERS detection; however,
the high surface energy and resulting instability of these
materials in solution is often translated into irreproducible SERS
Inconsistencies in these solution-phase measurements have
been attributed to changes in the electromagnetic or localized
surface plasmon resonance (LSPR) properties of the nanopar-
ticles.14Slight variations in either the shape or size of a
nanostructure will greatly influence the LSPR of the nanopar-
ticles and, as a consequence, their SERS enhancements.
Furthermore, when nanoparticles aggregate, the LSPR of the
structures will couple resulting in a new, lower energy extinction
band which will impact the intensity of a SERS signal.
To both maintain consistent electromagnetic properties and
prevent aggregation, metal nanoparticles have been protected
with stabilizing molecules. For instance, nanoparticles can be
modified with capping molecules which increase electrostatic
(1) Jeanmaire, D. L.; Van Duyne, R. P. J. Electroanal. Chem. Interfacial
Electrochem. 1977, 84 (1), 1–20.
(2) Haes, A. J.; Haynes, C. L.; McFarland, A. D.; Schatz, G. C.; Van
Duyne, R. P.; Zou, S. MRS Bull. 2005, 30 (5), 368–375.
(3) Le Ru, E. C.; Blackie, E.; Meyer, M.; Etchegoin, P. G. J. Phys. Chem.
C 2007, 111 (37), 13794–13803.
(4) Nie, S.; Emory, S. R. Science 1997, 275 (5303), 1102–1106.
(5) Campion, A.; Kambhampati, P. Chem. Soc. ReV. 1998, 27 (4), 241–
(6) Otto, A. J. Raman Spectrosc. 1991, 22 (12), 743–752.
(7) Schatz, G. C.; Young, M. A.; Van Duyne, R. P. Top. Appl. Phys.
2006, 103, 19–46.
(8) Haynes, C. L.; Haes, A. J.; McFarland, A. D.; Van Duyne, R. P. Top.
Fluoresc. Spectrosc. 2005, 8, 47–99.
(9) Nehl, C. L.; Liao, H.; Hafner, J. H. Nano Lett. 2006, 6 (4), 683–688.
(10) Doering, W. E.; Piotti, M. E.; Natan, M. J.; Freeman, R. G. AdV. Mater.
2007, 19 (20), 3100–3108.
(11) Lal, S.; Grady, N. K.; Kundu, J.; Levin, C. S.; Lassiter, J. B.; Halas,
N. J. Chem. Soc. ReV. 2008, 37 (5), 898–911.
(12) Le, F.; Brandl, D. W.; Urzhumov, Y. A.; Wang, H.; Kundu, J.; Halas,
N. J.; Aizpurua, J.; Nordlander, P. ACS Nano 2008, 2 (4), 707–718.
(13) Rodger, C.; Rutherford, V.; White, P. C.; Smith, W. E. J. Raman
Spectrosc. 1998, 29 (7), 601–606.
(14) Haes, A. J.; Stuart, D. A.; Nie, S.; Van Duyne, R. P. J. Fluoresc.
2004, 14 (4), 355–367.
Published on Web 10/03/2008
10.1021/ja8059039 CCC: $40.75 2008 American Chemical Society
J. AM. CHEM. SOC. 2008, 130, 14273–14279 9 14273
repulsive forces between nanoparticles and prevent aggrega-
tion.15An additional advantage of using surface ligands to
improve the stability of nanoparticles is that this surface
chemistry can be selectively displaced by the target molecule
which improves its detectability with SERS; however, these
capping molecules can experience chemical degradation, reac-
tion with the metal, or be affected by environmental changes
(pH, temperature, ionic strength, etc.).16Furthermore, when the
surface ligand is displaced or degraded, the nanoparticles often
aggregate and cause large changes in the LSPR of the nano-
particles which can either improve or degrade the SERS signal.
The robustness of nanoparticles to chemical and environ-
mental changes has been greatly improved by encapsulating the
metal nanoparticle cores in microporous silica shells.17Alter-
natively, composite nanoparticle structures with polymer
shell-void-noble metal core architectures have been synthe-
sized18and recently used as catalysts.19In particular, the use
of silica shells offers many advantages including (1) reduction
of electromagnetic coupling between metal nanoparticle cores,
(2) optical transparency, (3) tunability in the optical properties
of metal nanoparticles,20(4) versatility in the design of diverse
surface morphologies and functionalizations,21and (5) improved
biocompatibility via surface modification.22
Despite these advantages, the use of silica-coated nanopar-
ticles in SERS applications has been limited to the employment
of reporter or tag molecules that have been entrapped between
the metal core and the silica shell during synthesis.23-25
Alternatively, silica has been used to eliminate SERS enhance-
ments by purposefully blocking the metal substrate to prevent
direct interactions between the molecule and metal.26Because
silica is composed of a disordered, microporous structure, the
diffusion of molecules toward the metal core is limited.
Consequently, only slow, inward diffusion of small ions has
been reported with these core shell nanostructures.27
In this work, we demonstrate the encapsulation of gold
nanoparticle cores in porous silica membranes for reproducible
and temporally consistent SERS detection. First, silica shells
that entrap gold nanoparticle cores will be formed. Next, the
less dense and cross-linked internal silica matrix will be
selectively etched versus the highly cross-linked external silica
membrane. These internally etched silica membranes which
contain gold nanoparticle cores (IE Au@SiO2) will be demon-
strated to be viable SERS substrates for the reproducible
detection of 2-naphthalenethiol. This work presents the first
report of reproducible SERS detection using solution-phase
silica-void-gold nanoparticles. These composite nanostructures
should allow for fundamental studies in understanding the
mechanisms of SERS using solution-phase nanostructures as
well as for applications that involve the direct and reproducible
detection of biological and environmental molecules.
Reagents and Chemicals. Amberlite MB-150 mixed bed
exchange resin, citric acid, gold(III) chloride trihydrate, (3-
aminopropyl)trimethoxysilane (APS), sodium chloride (NaCl),
sodium citrate dehydrate, sodium trisilicate (27%), 2-naphthalene-
thiol (2-NT), and tetraethyl orthosilicate (TEOS) were purchased
from Sigma (St. Louis, MO). Ammonium hydroxide, ethanol,
hydrochloric acid (HCl), methanol, and nitric acid, were purchased
from Fisher Scientific (Pittsburgh, PA). Ultrapure water (18.2 MΩ
cm-1) was obtained using a Nanopure System from Barnstead
(Dubuque, IA). For all experiments, glassware was cleaned with
aqua regia (3:1 HCl/HNO3), rinsed thoroughly with ultrapure water,
and oven-dried overnight before use.
Gold Nanoparticle Synthesis. Gold nanoparticles were synthe-
sized via citrate reduction.28Briefly, 50 mL of 1 mM gold(III)
chloride trihydrate solution was refluxed and vigorously stirred for
10 min. Once a rolling boil was achieved, 5 mL of 39 mM citrate
solution was quickly added, and the solution changed from pale
yellow to deep red within 1 min. The solution was refluxed for an
additional 10 min and allowed to equilibrate to room temperature
while stirring. The resulting gold nanoparticles have a diameter of
d ) 13.3 ( 0.6 nm, as determined from transmission electron
microscopy, and an extinction maximum of λmax) 520 nm. The
nanoparticle concentration was calculated to be 13 nM using the
approached reported by Haiss et al.29
particles (Au@SiO2). Au@SiO2nanoparticles were synthesized
by combining two methods reported by Liz-Marzan and co-
workers.20,30Briefly, 40 mL of the gold nanoparticle stock solution
was diluted with 40 mL of water and mixed with Amberlite resin
to reduce the ionic strength of the solution. The resin was then
removed, and the pH of the nanoparticle solution was adjusted to
5.0 using 1 M HCl. Next, the nanoparticles were made vitreophilic
via the addition of 1 mM APS (0.52 mL). After 15 min, 27% silicate
(60 µL) was added to the solution, and the mixture was stirred for
24 h. The addition of 352 mL of ethanol (to a final ratio of 1:4.4
water/ethanol) induced the precipitation of unreacted silicate. After
6 h, the silica shells were thickened by adding concentrated
ammonium hydroxide (0.2 mL), TEOS (20 µL), and 1 mM APS
(20 µL). After 16 h, the reaction was stopped and the ethanol
removed from the mixture by centrifugation (30 min, 10 000 rpm).
Synthesis of Internally
Nanoparticles (IE Au@SiO2). IE Au@SiO2nanoparticles were
synthesized by increasing the pH of the Au@SiO2solution. First,
the pH of a Au@SiO2 solution (stirring) was increased via the
addition of ammonium hydroxide (1.5 M final concentration). The
etching process was quenched with concentrated HCl until the
overall solution pH approached ∼2.
Nanoparticle Purification. In all composite nanoparticle solu-
tions, careful steps were taken to remove uncoated or partially silica-
coated nanoparticles from the Au@SiO2or IE Au@SiO2nanopar-
(15) Hunter, R. J. Introduction to Modern Colloid Science; Oxford
University Press: Oxford, England, 1993.
(16) Neouze, M. A.; Schubert, U. Monatsh. Chem. 2008, 139 (3), 183–
(17) LizMarzan, L. M.; Giersig, M.; Mulvaney, P. Chem. Commun. 1996,
(18) Kim, M.; Sohn, K.; Na, H. B.; Hyeon, T. Nano Lett. 2002, 2 (12),
(19) Liu, G. Y.; Ji, H. F.; Yang, X. L.; Wang, Y. M. Langmuir 2008, 24,
(20) LizMarzan, L. M.; Giersig, M.; Mulvaney, P. Langmuir 1996, 12 (18),
(21) Brinker, C. J.; Scherer, G. W. Sol-Gel Science: The Physics and
Chemistry of Sol-Gel Processing; Harcourt Brace Jovanovich: Boston,
(22) Gupta, R.; Chaudhury, N. K. Biosens. Bioelectron. 2007, 22 (11),
(23) Mulvaney, S. P.; Musick, M. D.; Keating, C. D.; Natan, M. J. Langmuir
2003, 19 (11), 4784–4790.
(24) Gong, J. L.; Jiang, J. H.; Yang, H. F.; Shen, G. L.; Yu, R. Q.; Ozaki,
Y. Anal. Chim. Acta 2006, 564 (2), 151–157.
(25) Doering, W. E.; Nie, S. Anal. Chem. 2003, 75 (22), 6171–6176.
(26) Olson, L. G.; Lo, Y. S.; Beebe, T. P.; Harris, J. M. Anal. Chem. 2001,
73 (17), 4268–4276.
(27) Ung, T.; Liz-Marzan, L. M.; Mulvaney, P. Langmuir 1998, 14 (14),
(28) Grabar, K. C.; Freeman, R. G.; Hommer, M. B.; Natan, M. J. Anal.
Chem. 1995, 67 (4), 735–743.
(29) Haiss, W.; Thanh, N. T. K.; Aveyard, J.; Fernig, D. G. Anal. Chem.
2007, 79 (11), 4215–4221.
(30) Grzelczak, M.; Correa-Duarte, M. A.; Liz-Marzan, L. M. Small 2006,
2 (10), 1174–1177.
14274 J. AM. CHEM. SOC. 9 VOL. 130, NO. 43, 2008
Roca and Haes
These nanoparticles were purified using a Sephadex-25 column
where the bare or partially coated gold nanoparticles were retained
while the Au@SiO2or IE Au@SiO2nanoparticles were eluted and
redispersed in water. The concentration of the Au@SiO2and IE
Au@SiO2 nanoparticle solutions were estimated via extinction
spectroscopy and assumed to have an extinction coefficient of 13
nm bare gold nanoparticles (ε520nm) 2 × 108M-1·cm-1).
Transmission Electron Microscopy (TEM). TEM was per-
formed using a JEOL JEM-1230 microscope equipped with a Gatan
CCD camera. Samples were prepared on 400 mesh copper grids
coated with a thin film of Formvar and carbon. The nanoparticle
solution was diluted in a 50% water-ethanol mixture. Ap-
proximately 50 µL of this solution was pipetted onto a grid and
promptly drained using filter paper. The grids were allowed to
thoroughly dry before imaging.
Extinction Spectroscopy. Extinction spectra were collected
using a UV-vis spectrometer (Ocean Optics HR4000) configured
in a transmission geometry. Disposable methacrylate cuvettes (path
length ) 1 cm) were used. All measurements were taken using the
following parameters: 30 ms integration time, average of 30 scans,
and boxcar of 10. For continuous spectral acquisition, spectra were
recorded every 2 s.
SERS. All SERS spectra were collected using an Advantage200A
Raman system (DeltaNu). All measurements were taken using the
following parameters: excitation wavelength, λex ) 633 nm,
integration time ) 50 s, power ) 2 mW. All spectra were
background-corrected using Excel. Spectral intensities were normal-
ized to units of ADU·mW-1·s-1. Enhancement factors (EF) were
calculated versus a normal Raman spectrum of 8 mM 2-NT in
Results and Discussion
Nanoparticles. The stepwise synthesis of IE Au@SiO2nano-
particles is depicted in Scheme 1. First, gold nanoparticles are
coated in a composite silica shell. Solid silica shells are prepared
via a modified Sto ¨ber method31in a two-step process whereby
a vitreophilic layer which consists of APS and silicate is initially
formed on the gold surface. The resulting thin silica layer is
required to minimize nanoparticle aggregation and to protect
the nanoparticle cores from degradation in subsequent steps.
Next, the initial silica layer is thickened using an ethanolic
mixture of APS, TEOS, and ammonium hydroxide. Finally,
conversion of the silica shell into a silica membrane (i.e.,
silica-void-gold structure) occurs in an aqueous solution of
ammonium hydroxide. The porosity of the resulting IE
Au@SiO2 nanoparticles is dictated by the concentration of
ammonium hydroxide and etching time. At short times (0-10
min), the resulting nanoparticles are gold nanoparticle cores
entrapped in porous silica membranes (2). By extending the
etching time, Au@SiO2nanoparticles with freely moving gold
cores (3) and membrane-etched (4) or completely dissolved
silica membranes are formed.
In Figure 1, representative TEM images that depict the
formation of bare gold (Figure 1A), Au@SiO2(Figure 1B), as
well as IE Au@SiO2nanoparticles after 10 (Figure 1C) and 15
(Figure 1D) min etching periods are included. As shown in
Figure 1B, silica nanoparticles with and without gold cores are
formed in this process. Next, the internal silica matrix is slowly
etched under basic conditions. Two important observations are
noted during this process. First, the silica etching process does
not depend on the presence of the gold core. As visible in Figure
1C, all nanoparticles have been internally etched including those
that likely contained no original gold core. Second, the diameter
of the composite nanoparticle structures remains constant (∼45
nm) before and after etching. This indicates that the etching
process is occurring primarily at the inner silica matrix. As
etching time increases, the internal silica matrix is sufficiently
etched allowing the gold core to freely move within the silica
membrane. Subsequently, the silica membrane becomes thinner
and eventually exposes the gold core.
The formation and subsequent dissolution of the silica
membrane on Au@SiO2nanoparticles has also been monitored
using extinction spectroscopy (Figure 2). It is well-established
that noble metal nanoparticles exhibit a strong extinction
(absorption + scattering) band that can be tuned throughout
visible to near-infrared wavelengths.2This extinction band
results when the incident photon frequency is in resonance with
the collective oscillation of the conduction band electrons and
is known as the LSPR.2
In these studies, LSPR spectroscopy has been used to
understand the formation and subsequent dissolution of the silica
membrane on gold nanoparticles. The optical properties of these
composite structures reveal both dielectric changes near the gold
nanoparticle cores and electromagnetic coupling between nano-
particles when the silica membrane becomes sufficiently thin.
It should be noted that when the electromagnetic fields from
two different nanoparticles interact, a complex, lower energy
LSPR is produced.32As shown in Figure 2A, Au@SiO2
nanoparticles exhibit a single LSPR band centered at 525 nm.
Initial formation of the silica membrane via dissolution of the
silica shell is apparent from the initial drop in the extinction
intensity and background of this LSPR band (<30 min). As
(31) Sto ¨ber, W.; Fink, A.; Bohn, E. J. Colloid Interface Sci. 1968, 26 (1),
(32) Ghosh, S.; Pal, T. Chem. ReV. 2007, 107 (11), 4797–4862.
Scheme 1. Encapsulation of Gold Nanoparticle Cores in
Microporous Silica Membranes
representative TEM images: (A) Au nanoparticles (d ) 13.3 ( 0.6 nm),
(B) Au@SiO2nanoparticles (d ) 45 ( 5 nm), (C) optimized IE Au@SiO2
nanoparticles (10 min etching time) (d ) 45 ( 4 nm), and (D) IE Au@SiO2
(15 min etching time). The scale bar represents 50 nm in all micrographs.
Synthesis of IE Au@SiO2 nanoparticles as revealed in
J. AM. CHEM. SOC. 9 VOL. 130, NO. 43, 2008
etching continues, a new extinction band grows in at longer
wavelengths (∼620 nm). This new LSPR band is characteristic
of uncontrollably aggregated or electromagnetically coupled
nanoparticles. Comparison of these extinction data to structural
characterization with TEM suggests that the silica shell has been
sufficiently thinned or fully etched thereby facilitating electro-
magnetic coupling between multiple nanoparticle cores.
Upon analyzing these data more precisely, three distinct
phases in the etching process are clearly observed (Figure 2B).
In the first etching phase (phase I), a blue-shift in the extinction
maxima of the IE Au@SiO2 nanoparticles is observed. This
decrease in wavelength indicates that the local dielectric
environment surrounding the nanoparticles is decreasing and is
consistent with silica being replaced by a lower dielectric
constant material (water).2,33Representative TEM images of
these nanoparticles after 10 min of etching (Figure 1C) support
this observation and clearly show both a complete outer silica
membrane and a less dense internal silica matrix. The progres-
sive blue-shift in the LSPR of the gold nanoparticle cores with
increasing etching time indicates the continuous etching of the
inner silica matrix. Throughout this phase, both optical and
structural characterization support that the external silica shell
remains intact. However, as initially revealed in Figure 1D, at
the end of phase I (∼15 min), the gold nanoparticle cores do
not remain in the center of the silica membranes indicating that
the inner silica matrix closest to the core is fully etched.
Interestingly, at the end of phase I/beginning of phase II, the
extinction maxima of the IE Au@SiO2nanoparticles stabilize.
During the second phase of the etching process, silica is
hypothesized to undergo continual etching at the internal surface
of the silica membrane; however, this process occurs beyond
the distance dependence of the nanoparticle to dielectric constant
variations, and no change in the extinction spectra of the
nanoparticles is observed.
After an ∼50 min etching period, the optical properties of
the nanoparticles begin shifting to longer wavelengths. This final
phase of the etching process indicates that the nanoparticles are
electromagnetically coupling and that the silica membrane has
been breached. Ultimately, dissolution of the silica membrane
results in gold nanoparticles with optical properties consistent
with uncontrolled aggregation.
Understanding Silica Membrane Formation. The dissolution
of the inner silica matrix is clearly observed in Figures 1 and
2. Furthermore, TEM studies reveal that the external silica
surface is not etched during this process. In previous studies,
similar silica membrane formation was observed for silica-coated
carbon nanotubes CNT@SiO2.30This dissolution began at the
surface of the CNT and extended radially inward forming an
internal void volume with a constant external shell diameter.
In contrast, the present IE Au@SiO2dissolution studies reveal
no radially forming internal void volume but a decrease in the
density of the inner silica matrix in smaller domains followed
by the dissolution of the outer silica membrane.
These IE Au@SiO2nanoparticle data suggest that two distinct
silica matrixes (inner and outer) are formed on the surface of
the gold nanoparticle cores. As discussed in the synthesis of
the (unetched) Au@SiO2nanoparticles, both APS and TEOS
were used during the formation of the silica shell. As a result,
two hydrolysis and condensation reactions will occur to form
the composite Au@SiO2nanoparticle structure. The following
represent the hydrolysis reactions for APS and TEOS, respec-
tively. Silicate ions formed from TEOS (reaction 2) are less
acidic than those produced from APS (reaction 1); however, in
both cases, subsequent condensation/precipitation will occur as
where R ) CH3CH2-O-, CH3-O-, or NH2-(CH2)3-. As a
result, the acidity and relative concentrations of APS and TEOS
are hypothesized to drive the formation of the two distinct silica
matrixes. Hydrolyzed APS is more acidic and lower in
concentration than the hydrolyzed TEOS in the silica matrix.
Consequently, APS should be consumed before TEOS in
reaction 3. Because silicate bonds that have been formed from
APS are less cross-linked than silicate bonds formed from
TEOS, the APS-rich inner matrix will be more porous and have
a lower density than the external membrane.34
(33) Kreibig, U.; Vollmer, M. Cluster Materials; Springer-Verlag: Heidel-
berg, Germany, 1995.
Figure 2. Monitoring the etching process for IE Au@SiO2nanoparticles
using extinction spectroscopy. (A) Extinction spectra are collected over a
2 h period. (B) The formation of the silica membrane occurs in three distinct
stages. Phase I: etching the internal silica matrix. Phase II: dissolution of
the remote areas of the membrane vs the metal core. Phase III: complete
membrane dissolution. A dashed line has been added to guide the eye.
14276 J. AM. CHEM. SOC. 9 VOL. 130, NO. 43, 2008
Roca and Haes
When the silica shell is etched (or silica membrane is formed),
reaction 3 will reverse as follows:
and as a result, dissolution of silica will occur. The more porous
and less dense APS-containing matrix will react faster than the
denser and highly cross-linked external membrane. This dis-
solution mechanism is consistent with the observations of silica
membrane formation that results in silica-void-gold nano-
structures as observed in Figure 1. More complete investigations
of the silica shell dissolution mechanism are currently underway.
Assessing the Optical Stability of IE Au@SiO2 Nano-
particles. As mentioned previously, the optical properties of
noble metal nanoparticles exhibit unique extinction spectra
which arise from their LSPR, a phenomenon not observed in
bulk materials.7,35-37The local electromagnetic fields ac-
companying excitation of the LSPR is a key factor in the intense
signals observed in all surface-enhanced spectroscopies, includ-
ing SERS.2,38One limitation of using standard solution-phase
noble metal nanoparticles for surface-enhanced spectroscopic
experiments is the instability of these nanostructures in the
presence of target molecules. As a result, nanoparticle optical
properties can vary significantly which influences their elec-
tromagnetic properties and prevents quantitative detection of
The optical stability of citrate-reduced gold and IE Au@SiO2
nanoparticles in the presence of salt are compared in Figure 3.
As expected, the optical stability of bare gold nanoparticles is
disrupted by the addition of NaCl (Figure 3A). When salt is
added to the solution, the electrostatic repulsive forces induced
by the citrate ions at the surface of the nanoparticles are
diminished in the presence of salt ions, and consequently, the
nanoparticles aggregate. When this occurs, the LSPR of the
nanoparticles couple and a new, lower energy extinction band
centered at ∼650 nm is detected.
In contrast, IE Au@SiO2nanoparticles are optically stable
and have an extinction band centered at ∼520 nm which is
conserved in the presence of salt (Figure 3B). The silica
membrane physically impedes electromagnetic coupling between
the metal cores. Furthermore, the LSPR of IE Au@SiO2
nanoparticles is similar to that of bare gold nanoparticles (at t
) 0) and Au@SiO2nanoparticles, that is, they agree within a
few nanometers of each other (Au@SiO2 data not shown).
Clearly, silica membranes conserve the optical properties of the
gold nanoparticle cores in harsh environments without signifi-
cantly altering their optical properties versus bare gold
SERS Activity of Bare and Composite Nanoparticles. To
assess how the optical properties of nanoparticles influence
resulting SERS signals, gold, Au@SiO2, and IE Au@SiO2
nanoparticles have been incubated with 2-NT. This molecule
was chosen for several reasons. First, 2-NT has a high affinity
for the gold core. Second, the molecular size of 2-NT permits
its diffusion through the microporous silica membrane which
has an estimated pore size of ∼15 Å.34Finally, the molecule
has a large Raman cross section and is moderately soluble in
In preliminary experiments, the concentrations of both
nanoparticles and 2-NT were optimized to minimize bare gold
nanoparticle aggregation. Under these optimized conditions, bare
gold nanoparticles exhibit large SERS signals (Figure 4A-1);
however, these signals vary drastically with time and from run
to run. After a silica shell is grown on these structures (Figure
4A-3), no SERS signal is observed regardless of 2-NT concen-
tration and incubation time. Au@SiO2nanoparticles exhibit no
SERS activity because the molecule is not able to interact with
the metal core surface required for satisfying the chemical
enhancement and/or the silica shell extends beyond the strong
electromagnetic fields near the gold surface, an interaction
required for electromagnetic enhancement.
As shown in Figure 4A-2, IE Au@SiO2nanoparticles exhibit
SERS enhancement for 2-NT. To understand this spectroscopic
response, we must understand the role of both the thickness
and porosity of the silica membrane on the IE Au@SiO2
nanoparticles. We hypothesize that the observed SERS response
is from molecules that have diffused inside the silica membrane.
This is based on three experimental observations. First, the SERS
signal is not detected immediately after the molecule is added
to the nanoparticle solution. Instead, more than 1 min is required
before a SERS signal is detected, thereby indicating that the
molecule needs time to diffuse through the silica membrane
and bind to the gold nanoparticle core. Second, the silica
membrane has an estimated thickness of ∼3-4 nm (from TEM).
J. AM. CHEM. SOC. 9 VOL. 130, NO. 43, 2008
(34) Van Blaaderen, A.; Vrij, A. J. Colloid Interface Sci. 1993, 156 (1),
(35) Haynes, C. L.; Van Duyne, R. P. J. Phys. Chem. B 2001, 105 (24),
(36) Link, S.; El-Sayed, M. A. J. Phys. Chem. B 1999, 103 (40), 8410–
(37) Kreibig, U.; Gartz, M.; Hilger, A.; Hovel, H. Optical Investigations
of Surfaces and Interfaces of Metal Clusters. In AdVances in Metal
and Semiconductor Clusters; Duncan, M. A., Ed.; JAI Press Inc.:
Stamford, CT, 1998; Vol. 4, pp 345-393.
(38) Schatz, G. C.; Van Duyne, R. P. Electromagnetic Mechanism of
Surface-Enhanced Spectroscopy; Wiley: New York, 2002; Vol. 1, pp
nanoparticles in the absence and presence of 0.1 M NaCl. (A) When bare
gold nanoparticles are exposed to NaCl, the LSPR centered at 520 nm
decreases in intensity while a new band forms at longer wavelengths. (B)
IE Au@SiO2 nanoparticles are stable in the presence of NaCl. Digital
photographs of the nanoparticle solutions are included above the respective
extinction spectra. Photographs at t ) 0 (left) and t ) 10 s (right) for
uncoated gold nanoparticles and t ) 0 (left) and t ) 4 min (right) are
included for IE Au@SiO2nanoparticles.
Assessing the optical stability of bare and IE Au@SiO2
Approximately 1 order of magnitude in SERS intensity is lost
per nanometer that the target molecule is from the enhancing
substrate.39-41If the SERS signal observed with IE Au@SiO2
nanoparticles results from molecules outside the silica mem-
brane, the SERS signals for both the Au@SiO2and IE Au@SiO2
should be identical. Clearly, this is not the case. Instead, these
results suggest that 2-NT is able to diffuse through the silica
membrane of IE Au@SiO2nanoparticles and subsequently bind
to the surface of the gold nanoparticle cores.
Finally, the direct adsorption of 2-NT to the gold core in IE
Au@SiO2 nanoparticles is evident by comparing spectral
differences between Raman and SERS spectra for 2-NT (Figure
4B). Vibrational band assignments were made according to the
literature.42In all spectra for 2-NT, vibrational bands at 1381
(strong, ring stretching) and 786 (weak, ring deformation) cm-1
are present; however, several spectral differences are also noted.
First, a C-H bending mode centered at 1087 cm-1in the normal
Raman spectrum shifts to 1069 cm-1in both SERS spectra.
Previously, it was reported that 2-NT binds perpendicular to
the metal surface thereby preferentially enhancing and shifting
the vibration mode at 1069 cm-1.42This is observed in the
spectral data for both bare gold and IE Au@SiO2nanoparticles
but not in the Raman spectrum. Additionally, the ring stretching
mode centered at 1628 cm-1in the Raman spectrum shifts to
1622 cm-1in the SERS spectra. Finally, the S-H stretch at
2566 cm-1is only present in the Raman spectra suggesting that
this bond is no longer present in the IE Au@SiO2nanoparticle
studies and that a chemical bond has formed between the gold
surface and sulfur group. These spectral changes further support
the hypothesis that 2-NT is binding directly to the gold surface
in the IE Au@SiO2nanoparticle structures.
Time-Dependent SERS Studies. When engineering nanopar-
ticles for reproducible SERS detection, it is important to
compare the optical properties of the nanoparticles to the
observed SERS enhancement at a fixed vibrational frequency
over time. Following the addition of 2-NT, the solution was
briefly vortexed and spectra were collected every 50 s. The
extinction intensity at 633 nm and SERS intensity for the ring
stretching mode centered at 1381 cm-1is shown in Figure 5,
parts A and B, respectively. As can be seen in Figure 5A, the
extinction intensity of bare gold nanoparticles at 633 nm doubles
within the first minute of exposure to 2-NT; but after 2 min,
this extinction intensity steadily decreases. This response is
consistent with nanoparticles that are aggregating and settling
out of solution.43In contrast, the extinction intensity at 633 nm
for IE Au@SiO2remains constant after the addition of 2-NT
and electromagnetic coupling between the gold cores is pro-
hibited by the silica membrane.
(39) Kennedy, B. J.; Spaeth, S.; Dickey, M.; Carron, K. T. J. Phys. Chem.
B 1999, 103, 3640–3646.
(40) Ye, Q.; Fang, J. X.; Sun, L. J. Phys. Chem. B 1997, 101 (41), 8221–
(41) Tsen, M.; Sun, L. Anal. Chim. Acta 1995, 307 (2-3), 333–340.
(42) Alvarez-Puebla, R. A.; Dos Santos, D. S.; Aroca, R. F. Analyst 2004,
129 (12), 1251–1256.
(43) Tantra, R.; Brown, R. J. C.; Milton, M. J. T. J. Raman Spectrosc.
2007, 38, 1469–1479.
Figure 4. Raman and SERS data for 2-NT. (A) Solution-phase SERS
spectra of 4 mM 2-NT with 4 nM (1) bare gold nanoparticles (irreproduc-
ible), (2) IE Au@SiO2, and (3) Au@SiO2nanoparticles. Spectral averages
of 1, 10, and 6 were used in 1, 2, and 3, respectively. (B) Normalized spectra
of 2-NT with (1) IE Au@SiO2(SERS), (2) bare gold nanoparticles (SERS),
and (3) a supersaturated suspension of 2-NT in water (Raman). The most
significant vibrational bands are labeled. Experimental conditions are as
follows: λex) 633 nm, power ) 2.0 mW, SERS integration time ) 50 s,
Raman integration time ) 10 s. Spectral averages of 10, 1, and 10 were
used in 1, 2, and 3, respectively.
Figure 5. Monitoring the extinction and SERS intensity for nanoparticle
samples upon 2-NT addition as a function of time. (A) Real-time changes
in the extinction intensity at the Raman excitation wavelength (λex) 633
nm) are shown for bare gold and IE Au@SiO2nanoparticles. Spectra were
collected every 1 min. (B) Real-time behavior of the SERS intensity at
1381 cm-1reveals distinct responses to the nanoparticle samples. λex )
633 nm, power ) 2.0 mW, integration time ) 50 s. In both panels, solid
squares ) bare gold nanoparticles and open circles ) IE Au@SiO2
nanoparticles. Identical nanoparticle and 2-NT concentrations are used in
14278 J. AM. CHEM. SOC. 9 VOL. 130, NO. 43, 2008
Roca and Haes
It is clear that from these data that bare gold and IE Au@SiO2
nanoparticles exhibit different optical responses when 2-NT is
added to the solution as observed in Figure 5A. Next, to compare
and quantify SERS signals for the different nanoparticle
structures, Raman enhancement factors (EF) were calculated
using the following equation:3
where CRaman is the concentration of analyte that produces a
Raman signal IRaman, and CSERSis the concentration of analyte
that produces a SERS signal ISERS.
We should note that the EFs for all SERS measurements have
a contribution from both electromagnetic and chemical com-
ponents.44The electromagnetic component relies on the optical
properties of the nanoparticles and can contribute up to 1012
orders of magnitude in SERS enhancement.45,46The chemical
enhancement mechanism, on the other hand, is related to the
charge transfer as well as a physical interaction between the
nanoparticle and target molecule and contributes up to 102orders
of magnitude in SERS experiments.5
The observed changes in the optical properties of the
nanoparticles have large impacts on SERS enhancement of
2-NT. In Figure 5B, two important trends are observed. First,
the SERS enhancement for both bare and IE Au@SiO2
nanoparticles follows the time-dependent extinction data ob-
served in Figure 5A. Because the bare gold nanoparticles have
constantly changing optical properties, the observed SERS
signals also fluctuate. IE Au@SiO2nanoparticles have stable
optical properties, and as a result, SERS intensities are also
stable and quantitative. Second, the magnitude of the SERS
intensity for the bare gold nanoparticles is much larger than
the SERS intensity for the IE Au@SiO2 nanoparticles. This
response is expected as the extinction intensity of the IE
Au@SiO2nanoparticles is much smaller than that for the bare
gold nanoparticles at the selected SERS excitation wavelength
because no electromagnetic coupling between nanoparticle cores
(aggregation) has occurred.
It is important to note that bare nanoparticles exhibit
irreproducible SERS enhancements which have been attributed
to the aggregation of the nanoparticles upon 2-NT addition. A
maximum EF of 4 × 104is observed for bare gold nanoparticles,
whereas an average EF of 3 × 102is obtained for IE Au@SiO2
nanoparticles. It should be noted that the average signal strength
for bare gold nanoparticles varies by 3000% over a 2 h period.
In comparison, both the detector response for normal Raman
scattering and SERS signal strength from IE Au@SiO2nano-
particles vary by ∼10%. This suggests that signal variation from
IE Au@SiO2 nanoparticles arises from fluctuations in the
detector response (or laser power)snot from changes in the
electromagnetic properties or stability of the nanoparticles!
Finally, the extinction intensity for IE Au@SiO2nanoparticles
is weak at the excitation wavelength. As a result, the magnitude
of the SERS enhancement observed for these nanoparticles is
small and consistent in magnitude with a dominant chemical
enhancement and/or slight electromagnetic enhancement. Further
refinement and improvement in the optical properties of the
nanoparticle core and structure of the silica membrane will aid
in improving the magnitude of this enhancement for IE
Au@SiO2nanoparticles and understanding molecular diffusion
of analytes through the silica membrane, respectively.
This work presents the first report of temporally consistent
SERS detection using solution-phase silica-void-gold nano-
particles. Specifically, the principal discovery reported here is
that quantitative and temporally stable SERS signals are
achieved using solution-phase gold nanoparticle cores that have
been entrapped in porous silica membranes. These composite
silica-void-gold nanostructures were synthesized in a stepwise
process. First, gold nanoparticles were encapsulated in a thin
silica matrix formed from APS and TEOS. Next, a denser silica
shell composed of TEOS encapsulated the initial silica matrix.
Finally, the less cross-linked (APS-containing) silica domains
were etched before the dissolution of the outer silica membrane
and resulted in microporous silica membranes that both encap-
sulated and prevented electromagnetic coupling/aggregation
between gold nanoparticle cores.
Detection of target species using SERS occurred after the
molecules diffused through the silica membrane. The magnitude
and reproducibility of the SERS enhancements for 2-NT that
was mixed with gold and internally etched Au@SiO2nanopar-
ticles were compared. In both cases, the intensity of the SERS
signal correlated with changes in the extinction intensity of the
nanoparticles at the SERS excitation wavelength. Furthermore,
after a brief equilibration period for the diffusion of the
molecules into the nanostructure core, the optical properties and
SERS enhancements of the target molecule using the internally
etched Au@SiO2nanoparticles were demonstrated to be constant
for at least 2 h. Further refinement of the electromagnetic
properties of the gold nanoparticle core should lead to larger
SERS enhancements for target molecules. The demonstration
of these nanostructures as stable SERS substrates for the
quantitative and reproducible detection of a target molecule
suggests that these composite structures could be used in a
variety of biomedical and environmental applications where
sensitive detection is required. Finally, expansion of the
presented composite nanostructure synthesis to other core
materials has the potential to increase the long-term stability of
any nanomaterial where instability and high surface energy
Acknowledgment. The authors gratefully acknowledge financial
support the Department of the Navy, Office of Naval Research (YIP
Award No. N00014-07-1-0827), the Society of Analytical Chemists
of Pittsburgh, and the University of Iowa. The authors also
acknowledge the Central Microscopy Research Facility (CMRF)
for use of the TEM.
(44) Moskovits, M. J. Raman Spectrosc. 2005, 36 (6-7), 485–496.
(45) Kneipp, K.; Kneipp, H.; Kneipp, J. Acc. Chem. Res. 2006, 39 (7),
(46) McFarland, A. D.; Young, M. A.; Dieringer, J. A.; Van Duyne, R. P.
J. Phys. Chem. B 2005, 109 (22), 11279–11285.
J. AM. CHEM. SOC. 9 VOL. 130, NO. 43, 2008