Quantitative surface enhanced Raman scattering detection based on the sandwich structure substrate

Article abstract of DOI:10.1016/j.saa.2011.03.045

A sandwich structured substrate was designed for quantitative molecular detection using surface enhanced Raman scattering (SERS), in which the probe molecule was sandwiched between silver nanoparticles (SNPs) and silver nanoarrays. The SNPs was prepared using Lee-Meisel method, and the silver nanoarrays was fabricated on porous anodic aluminum oxide (AAO) using electrodepositing method. The SERS studies show that the sandwich structured substrate exhibits good stability and reproducibility, and the detection sensitivity of Rhodamine 6G (R6G) and Melamine can respectively reach up to 10^-19 M and 10^-9 M, which is improved greatly as compared to other SERS substrates. The improved SERS sensitivity is closely associated with the stronger electromagnetic field enhancement, which stems from localized surface plasmon (LSP) coupling between the two silver nanostructures. Furthermore, the SERS intensity increased almost linearly as the mother concentration increased, which indicates that such a sandwich structure may be used as a good SERS substrate for quantitative analysis.

Full text of DOI:10.1016/j.saa.2011.03.045

Spectrochimica Acta Part A 79 (2011) 625–630
Contents lists available at ScienceDirect
Spectrochimica Acta Part A: Molecular and
Biomolecular Spectroscopy
journal homepage: www.elsevier.com/locate/saa
Quantitative surface enhanced Raman scattering detection based on the
“sandwich” structure substrate
Junmeng Zhanga, Shengchun Qua,∗, Lisheng Zhanga,b, Aiwei Tanga, Zhanguo Wanga
a Key Laboratory of Semiconductor Materials Science, Institute of Semiconductors, Chinese Academy of Sciences, Beijing 100083, China
b Beijing Key Lab of Nano-Photonics and Nano-Structure, Capital Normal University, Beijing 100037, China
a r t i c l e i n f o
a b s t r a c t
Article history:
A sandwich structured substrate was designed for quantitative molecular detection using surface
enhanced Raman scattering (SERS), in which the probe molecule was sandwiched between silver
nanoparticles (SNPs) and silver nanoarrays. The SNPs was prepared using Lee–Meisel method, and the sil-
ver nanoarrays was fabricated on porous anodic aluminum oxide (AAO) using electrodepositing method.
The SERS studies show that the sandwich structured substrate exhibits good stability and reproducibility,
Received 16 September 2010
Received in revised form 24 February 2011
Accepted 16 March 2011
Keywords:
SERS
Silver nanoparticles
Silver nanoarrays
AAO template
and the detection sensitivity of Rhodamine 6G (R6G) and Melamine can respectively reach up to 10−19
M
and 10−9 M, which is improved greatly as compared to other SERS substrates. The improved SERS sen-
sitivity is closely associated with the stronger electromagnetic field enhancement, which stems from
localized surface plasmon (LSP) coupling between the two silver nanostructures. Furthermore, the SERS
intensity increased almost linearly as the mother concentration increased, which indicates that such a
sandwich structure may be used as a good SERS substrate for quantitative analysis.
© 2011 Elsevier B.V. All rights reserved.
1. Introduction
providing the required roughness features for SERS enhancement,
such as nanospheres [19], nanowires [20], nanofilms [21], nanoar-
SERS as a powerful analytical tool has gained much attention in
biological and chemical field due to its high sensitivity and selec-
tivity [1–5]. Large enhancement of Raman signals on the surfaces of
metal nanostructures makes SERS possible to achieve both abun-
dant structural and quantitative information in a nondestructive
manner. Additionally, the SERS are often used as a sensitive analyt-
ical tool for single molecules detection under optimum conditions.
For the detection of single molecules, a very high enhancement
(up to 1014–1015) is accordingly required. Theoretically, the strong
enhancement is attributed to the so-called hot spots, which may be
junctions between nanostructures [6–9].
It is well known that the SERS enhancement strongly depends
on the detailed morphology of the nanostructures. The critical issue
in achieving the ultimate SERS sensitivity anchors on obtaining the
“ideal” substrates, which can yield not only strong SERS effects but
also quantitative, reproducible and stable signals. Various methods
have been developed to fabricate the “ideal” SERS substrates includ-
ing the dry methods (such as vacuum deposition, sputtering and
lithography) [10–14] and wet methods (such as sol–gel method and
electrochemical deposition) [15–18]. The aforementioned meth-
ods produce different-shaped nanostructures which are capable of
rays [22,23] and nanorods [24]. In fact, however, these SERS-active
substrates mentioned above provide either limited sensitivity or
irreproducible Raman signals. Therefore, it is necessary to fabricate
an ideal substrate which can construct highly regular and repro-
ducible hot spots.
Up to date, several groups have developed SERS-active sand-
wich substrates that can exploit LSP coupling between either two
nanoparticle layers [25] or one thin film and one nanoparticle layer
[26,27], in which the probe molecules are embedded. As reported
in previous, the electromagnetic field between two closely spaced
SNPs was substantially enhanced by an order of 1011 [28]. The SERS
intensity of the double sandwich structured substrates based on sil-
ver colloid layer is four times stronger than a single substrate [29].
Due to such a strong additional enhancement, the sandwich struc-
tured substrates have potential applications in achieving highly
improved SERS sensitivity when probe molecules are embedded
between the double nanostructures.
In this paper, a sandwich structured substrate was fabricated
based on the probe molecules embedded between SNPs and silver
nanoarrays, in which SNPs were obtained using Lee–Meisel method
[30], and silver nanoarrays were fabricated by electrodepositing
silver in the pores of AAO templates. This sandwich structured sub-
strate makes probe molecules surrounded by silver nanostructures
in multiple directions, which should help to achieve high SERS sen-
sitivity. The results demonstrate that the detection sensitivity of
∗
Corresponding author. Tel.: +86 10 82304568; fax: +86 10 82304240.
E-mail address: qsc@semi.ac.cn (S. Qu).
1386-1425/$ – see front matter © 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.saa.2011.03.045

626
J. Zhang et al. / Spectrochimica Acta Part A 79 (2011) 625–630
such a sandwich structured substrate is extremely high, and a good
linear relationship between the SERS intensity and the concentra-
tions of probe molecular solutions is obtained.
2. Experimental
2.1. Materials
High purity aluminum (99.999%, 0.25 mm in thickness) was
obtained from Beijing Cuibolin Non-Ferrous Technology Develop-
ing Co. Ltd. Melamine (99.5%) was purchased from Beijing Chemical
Reagents Co, Ltd., and Rhodamine 6G (R6G 95%) was obtained from
Sigma–Aldrich Co. Ltd. Other reagents (99.8% silver nitrate, 99.5%
boric acid, 99.5% oxalic acid, 85% phosphoric acid, 99.99% chromium
trioxide, and 99% trisodium citrate) were purchased from Beijing
Chemical works. All chemicals were of analytical grade purity and
used as received without further purification.
Fig. 1. Schematic diagram of the SNPs/nanoarrays sandwich substrate for SERS.
2.2. Preparation of SNPs
at 70 ◦C, SERS spectra were recorded using a RENISHAW H13325
Raman spectrophotometer with an excitation laser wavelength of
532 nm and excitation laser power of 5 mW. The laser beam was
focused to a spot about 3 ␮m in diameter by a 50× objective lens.
The Raman band of a silicon wafer at 520 cm−1 was used to calibrate
the spectrometer. In order to see the importance of the presence of
silver nanoarrays, SERS spectra of R6G/SNPs solution without silver
nanoarrays were also recorded.
Another probe molecule, Melamine was diluted to solutions of
different concentrations (10−3 M to 10−9 M) with deionized water.
Then they were respectively sandwiched between SNPs and silver
nanoarrays for SERS measurements using the same method above-
mentioned.
SNPs were prepared according to the Lee–Meisel method. Typi-
cally, 45 mg silver nitrate was dissolved in 250 ml deionized water,
and the solution was heated to the boiling point. Then 10 ml of a
0.5% trisodium citrate aqueous solution was added into the boiling
silver nitrate solution drop by drop, accompanied by vigorous stir-
ring. Afterwards, the mixture was kept boiling for another 10 min
and finally it became green-gray. The SNPs prepared in this way
was stable for at least several days or weeks.
2.3. Preparation of silver nanoarrays
A highly ordered porous AAO template was fabricated using a
classical two-step anodization technique [10,31,32]. Briefly, an alu-
minum plate (0.25 mm × 50 mm × 75 mm) was anodized in a 0.4 M
oxalic acid solution at 8 ◦C at a constant voltage of 45 V for 1.5 h
after annealing and electro-polishing. Then the aluminum oxide
film formed by the first anodization was removed and the second
anodization was performed for 30 min under the same conditions
as the first one. Where after, the anodization voltage was reduced
step by step to 12 V to thin the bottom barrier layers at a velocity
of 2 V/min.
Silver nanoarrays was deposited in the pores of AAO template
by applying AC voltage of 17.8 V with a frequency of 50 Hz for 120 s
in the electrolyte containing 0.05 M silver nitrate and 30 g/L boric
acid at 8 ◦C. After electrodeposition, the aluminum substrate on the
bottom of AAO template was etched by an aqueous mixture of per-
chloric acid and copper dichloride, and then the barrier layer was
removed in phosphoric acid solution (5 wt%) at ambient tempera-
ture. Finally, a transparent-brown thin film was obtained and was
kept in absolute ethanol in order to prevent oxidation.
The surface morphology of AAO template, silver nanoarrays and
sandwich structure was analyzed using a JSM-7401F scanning elec-
tron microscope. UV–visible absorption spectra were recorded on a
SHIMADZU UV-3101 spectrometer. The resolution was set to 5 nm
and the scan region was 300–800 nm.
3. Results and discussion
Fig. 2 displays the UV–visible absorption spectra of SNPs solu-
tion and silver nanoarrays. As shown in Fig. 2(a), a single absorption
maximum at 436 nm was observed in the UV–visible absorption
spectrum of SNPs solution due to the surface plasmon resonance
a
b
1.6
1.2
0.8
0.4
0.0
2.4. Measurements
The R6G was used as probe molecule to study the SERS enhance-
ment of the sandwich substrate. R6G was dissolved in deionized
water and prepared for various concentrations (2 × 10−8 M to
2 × 10−20 M). Starting with aqueous R6G solution of 2 × 10−8 M,
SNPs solution was added to the R6G solutions. 1 ml R6G solution
and 1 ml SNPs solution were mixed and, therefore, 2 ml sample with
a concentration of 10−8 M of R6G molecules was obtained. Solutions
of various concentrations from 10−8 M to 10−20 M, respectively,
were prepared using the similar method. The SERS measurements
were performed by dropping the R6G/SNPs solution onto the sur-
face of silver nanoarrays, and the schematic illustration of the
sandwich structured substrates is given in Fig. 1. The diameter of
the spreading area is about 3 mm. After evaporation in drying oven
300
400
500
600
700
800
Wavelength (nm)
Fig. 2. UV–visible absorption spectra of (a) SNPs solutions and (b) silver nanoarrays
fabricated on an AAO template.

J. Zhang et al. / Spectrochimica Acta Part A 79 (2011) 625–630
627
Fig. 3. SEM images of AAO templates (a) in the top view and (b) the cross sectional view, silver nanoarrays (c) in the top view and (d) the cross sectional view, and (e) SNPs
adsorbed on silver nanoarrays. (f) TEM image of SNPs.
(SPR) of SNPs. In Fig. 2(b), the characteristic peak of silver nanoar-
rays deposited on AAO template is located at about 504 nm due to
the resonant excitation of plasma oscillations in the confined elec-
tron gas of silver nanorods. It should be mentioned that the AAO
template is a good transparent matrix for studying the optical prop-
erties of silver nanoarrays because the absorption of AAO template
is weak at the wavelength region of interest in the present research.
The characteristic peak of the spectrum is very close to the 532 nm
laser line of Raman spectrophotometer, therefore, the excitation of
LSP by the resonant laser line will be very efficient [33,34].
The SEM images of AAO template, silver nanoarrays, and SNPs
adsorbed on silver nanoarrays are shown in Fig. 3(a)–(e), respec-
tively. In the top view of AAO template shown in Fig. 3(a), the
nanopores of a uniform size are in the form of a perfect two-
dimensional array with a hexagonal pattern. The distribution of the
pores is very uniform, which indicates that the as-fabricated AAO
template is of high quality. In the cross sectional view (Fig. 3(b)),
the pores are almost perfectly aligned vertically with respect to
the surface of the templates and the height of the AAO templates
is estimated to be about 22 ␮m. Fig. 3(c) and (d) gives the SEM
images of silver nanoarrays deposited onto AAO template. It can
be observed from the top view (Fig. 3(c)) that the silver nanoar-
rays is deposited in nearly all the pores of AAO template. They are
fixed in the pores with their tips exposed exclusively at the top.
As shown in the cross sectional view (Fig. 3(d)), the nanorods is
parallel to each other and perpendicular to the Al substrate at the
bottom. The average diameter of the nanorods is approximately
50 nm, which corresponds to that of the pores of AAO templates.
However, it also can be found that some parts of nanorods in pores
are not visible, which may be slipped out from the pores during
sampling. The SEM image of the SNPs adsorbed on silver nanoar-
rays is shown in Fig. 3(e). It can be seen that the distribution of
SNPs on the nanoarrays surface is very uniform. Upon dropping the
mixed solution onto the silver nanoarrays substrate, the SNPs and
probe molecules in the solution will interact and bind to the nanoar-
rays. Statistically, some of the molecules will be sandwiched in the
randomly formed junction between the SNPs and the nanoarrays,
where the electromagnetic field would be further enhanced leading
to stronger SERS signals. In order to further demonstrate the mor-
phology of the SNPs, the TEM image of SNPs is given in Fig. 3(f). It can
be seen that the nanoparticles are irregularly in shape, with many
edges on themselves. This property may lead to an increased SERS
activity, because the nanoparticles with clear edges can have more
enhanced electromagnetic field compared with the nanospheres
[35].
For a quantitative analysis, The SERS spectra of the sample solu-
tions with different concentrations of R6G ranging from 10−8 to
10−20 M were measured. Fig. 4(A) shows the SERS spectra of R6G
adsorbed on the sandwich substrates, and the intensity profile of
the peak at 1649 cm−1 as a function of the concentration of sam-
ple solutions is also shown in Fig. 4(B). As shown in Fig. 4(A), it
can be seen that the SERS spectrum is very similar to the reported
SERS spectrum of R6G. According to previous reports [36], there
are three characteristic peaks for identification of R6G molecule:
612 cm−1, 1362 cm−1 and 1649 cm−1, in which the 612 cm−1 and
1362 cm−1 bands are attributed to the characteristic ring in-plane
bending mode and the C–C stretching mode, respectively, and the
last band is arising from the in-plane mode of ring stretching vibra-
tion. For the sake of comparison, we have chosen the strongest band
at highest peak at 1649 cm−1 as the diagnostic peak, and the inte-
grated intensity of diagnostic bands of each spectrum was used
to represent their intensity. It can be seen that the intensity of

628
J. Zhang et al. / Spectrochimica Acta Part A 79 (2011) 625–630
1649
1509
A
612
3x105
2x105
1362
1312
1573
n
1186
773
m
b
l
k
j
i
h
g
4x104
3x104
2x104
f
e
a
d
c
×
8
8
b
×
×
1x104
0
a
8
600
800 1000 1200 1400 1600 1800
Raman shift (cm-1)
300
600
900
1200
1500
1800
Raman shift (cm-1)
B
Fig. 5. SERS spectra of R6G in the R6G/SNPs solutions with concentrations of 10−16
M
and 10−15 M, respectively.
106
105
wich substrate. Therefore, the importance of the presence of silver
nanoarrays was proved.
As we know, R6G is electronically resonant molecule for SERS
due to its electronic energy levels. When the laser line is focused on
532 nm, the contribution to high Raman EF from the electronic res-
onance of R6G with laser wavelength is also very large. Therefore, to
be precise, the Raman signal obtained should be a surface enhanced
resonance Raman scattering (SERRS), not pure SERS effect. Then,
we chose one non-electronically resonant molecule, Melamine,
as another probe molecule to measure SERS performance of the
above substrate. Melamine is one of nitrogen-containing organic
compounds, which adulterated intentionally in milk products can
invite dangers in public-health. Fig. 6(A) indicates the inten-
sity of Melamine’s characteristic peaks decreases with decreasing
Melamine concentration. Significantly, it is observed that the detec-
tion limit of the sandwich substrate for Melamine could reach as
low as 10−9 M. Moreover, Fig. 6(B) reveals a linear relation between
the SERS intensity and Melamine concentration similar to the case
for R6G detection.
Based on the above observations, the detection sensitivity of the
sandwich substrate is higher than other SERS substrates no matter
the probe molecular is resonant or non-resonant. The electromag-
netic mechanism is considered to play an important role in the
additional Raman enhancement. In recent studies [29,37], a coop-
erative coupling effect of double SERS-active nanostructures was
proposed to account for the additional SERS intensity. It has been
suggested that the interaction between LSP of SNPs and LSP of sil-
ver nanoarrays are responsible for such enhancement. When SNPs
is irradiated with light, the oscillating electrons in the SNPs induce
electron oscillations in the silver nanoarrays, thus producing LSP
with the frequency equal to that of SNPs. The LSP in SNPs is cou-
pled to the LSP in the silver nanoarrays, resulting in a new set of
coupled LSP modes. In our study, the optical response range of silver
nanoarrays is close to the laser line, and also overlaps well with the
absorption band of SNPs, thus LSP resonance of the sandwich sub-
strate can be easily obtained. As a result, it is possible for the strong
SERS enhancement due to the hot spots in the junction between
the two silver substrates resulting from the LSP resonance cou-
pling. Statistically, some of the probe molecules can be sandwich
in the junctions where the hot spots exist, which makes the SERS
signals become stronger. Therefore, it is proposed that the addi-
tional Raman enhancement of this substrate is attributed to the
existence of strong electromagnetic field generated by the cooper-
ative coupling of LSP of SNPs and LSP of silver nanoarrays in the
junctions between the two silver nanostructures.
logI=0.1498*logC+2.6337
10-17 10-15 10-13 10-11 10-9
Concentration of analyte (M)
10-7
Fig. 4. (A) SERSspectra ofR6Gembedded in thesandwich structure with theconcen-
trations of (b) 10−20 M, (c) 10−19 M, (d) 10−18 M, (e) 10−17 M, (f) 10−16 M, (g) 10−15 M,
(h) 10−14 M, (i) 10−13 M, (j) 10−12 M, (k) 10−11 M, (l) 10−10 M, (m) 10−9 M, (n) 10−8 M,
and (a) Raman spectrum of 10−2 M R6G dropped on a glass slide. (B) The inten-
sity profile of the peak at 1649 cm−1 as a function of the concentration of the R6G
solutions.
the diagnostic peak of R6G is decreased with the R6G concentra-
tion decreasing. As depicted in Fig. 4(B), a good linear relation was
observed between the logarithmic concentrations of R6G and the
intensities of the diagnostic peak, which provides a calibration for
quantitative detection of R6G.
For the SERS spectrum of 10−19 M in Fig. 4(A), the characteristic
peaks of R6G molecules are distinguishable from the background.
Therefore, the 10−19 M concentration can be taken as the limit of
detection for R6G, which is much lower than that derived from
other SERS-active substrates. The SERS enhancement factor (EF)
is an important parameter for demonstrating the SERS perfor-
mance. Herein, we estimate the SERS EF from the standard equation
defined as: EF = ISERSCRS/IRSCSERS, wherein ISERS and IRS correspond
to the Raman intensity of diagnostic band of the probe molecules
adsorbed on the SERS substrate and the non-SERS substrate, respec-
tively; CSERS and CRS are, respectively, the concentration of probe
molecules under SERS conditions and that under non-SERS condi-
tions, taking into account that the experimental conditions, such as
the laser wavelength, laser power, microscope objective lens, and
measuring conditions on the substrate, are identical in all cases. As
a result, a high Raman EF of 3 × 1016 can be obtained when using
CSERS of 10−19 M and CRS of 10−2 M.
In order to evaluate the SERS contribution from silver nanoar-
rays in the sandwich substrate, single SNPs as substrate without
silver nanoarrays were tested for SERS sensitivity and were com-
pared with the sandwich substrate. The experimental conditions
are identical to that for sandwich substrate. Using this single SNPs
substrate, the lowest detectable concentration of R6G was around
10−16 M (Fig. 5), which was lower than the 10−19 M of sand-

J. Zhang et al. / Spectrochimica Acta Part A 79 (2011) 625–630
629
the reproducibility quantitatively, the integrated intensities of each
spectrum at 1649 cm−1 is calculated, and a 14.1% relative standard
deviation is obtained. This indicates that the sandwich-based SERS
of our substrate has good reproducibility. The reasons for the good
reproducibility of SERS can be summarized as follows: our silver
nanoarrays is very uniform in diameter and length and the SNPs are
well-distributed on the surface of nanoarrays. Therefore, the num-
ber of the junctions irradiated by the focused laser beam is similar
in each measurement. On the other hand, the R6G molecules and
SNPs are mixed homogeneously before dropping on the surface,
which makes the surface concentration of R6G become relatively
uniform on the entire surface. It should be mentioned that the SERS
activity of the sandwich substrate preserved almost the same after
keeping in absolute alcohol for one week. It indicates that this sand-
wich substrate has good stability, which is important for practical
applications in the future.
683
A
1617
985
h
g
573
f
e
d
c
b
a
400
600
800 1000 1200 1400 1600 1800
Raman Shift (cm-1)
B
4. Conclusions
In conclusion, a simple and reproducible SERS-active sand-
wich substrate based on probe molecular embedded between SNPs
and silver nanoarrays has been presented. This sandwich struc-
tured substrate exhibits a good linear relationship between the
SERS intensity and the concentration of probe molecular solutions.
The detection limit of this substrate can reach extremely low for
either resonant or non-resonant probe molecular, which is signifi-
cantly superior to other SERS-active substrates. This improvement
of SERS sensitivity is attributed to the enhanced local electromag-
netic field associated with LSP produced by plasmon cooperative
coupling between the SNPs and the silver nanoarrays. In addition,
the sandwich substrate shows excellent reproducibility and good
stability. Therefore, a SERS sensor based on this SNPs/nanoarrays
sandwich structured substrate used for rapid, sensitive, and quan-
titative detection of molecules in chemistry or biology fields will
be realized in the coming future.
10 5
logI=0.22545*logC+3.88076
10 4
10 -10 10 -9 10 -8 10 -7 10 -6 10 -5 10 -4 10 -3 10 -2
Concentration of analyte (M)
Fig. 6. (A) SERS spectra of Melamine embedded in the sandwich structure with
the concentrations of (b) 10−9 M, (c) 10−8 M, (d) 10−7 M, (e) 10−6 M, (f) 10−5 M, (g)
10−4 M, (h) 10−3 M, and (a) Raman spectrum of solid Melamine. (B) The intensity
profile of the peak at 683 cm−1 as a function of the concentration of the Melamine
solutions.
Acknowledgments
Reproducibility is another important parameter in SERS per-
formance. Fig. 7 shows the reproducibility test of SERS spectra of
10−12 M R6G obtained from six randomly selected points on the
surface of sandwich substrate. It can be observed in the SERS spec-
tra that remarkable reproducibility is achieved in the characteristic
peaks of R6G at 612, 1362, and 1649 cm−1. In order to analyze
This work was financially supported by the National Basic
Research Program of China (973 Program) under Grant No.
2010CB933800 and National Natural Science Foundation of China
under Grant Nos. 61076009, 60736034 and 50990064.
References
[1] G.V. Pavan Kumar, B.A. Ashok Reddy, M. Arif, T.K. Kundu, C. Narayana, J. Phys.
Chem. B 110 (2006) 16787–16792.
[2] Z.P. Zhou, D.Y. Wan, X.Y. Dou, L. Song, X.Q. Yan, D.F. Liu, H.J. Yuan, Y. Gao, J.X.
Wang, L.F. Liu, W.Y. Zhou, S.S. Xie, Physica E 28 (2005) 360–364.
[3] Y.P. Sun, W. Song, X. Zhu, R. Zhang, Q.Y. Pang, Z.R. Zhang, H.F. Yang, J. Raman
Spectrosc. 40 (2009) 1306–1311.
1649
1509
1573
1362
[4] D.A. Stuart, C.R. Yonzon, X.Y. Zhang, O. Lyandres, N.C. Shah, M.R. Glucksberg,
J.T. Walsh, R.P. Van Duyne, Anal. Chem. 77 (2005) 4013–4019.
[5] C.S. Levin, S.W. Bishnoi, N.K. Grady, N.J. Halas, Anal. Chem. 78 (2006)
3277–3281.
[6] E.C. Le Ru, P.G. Etchegoin, M. Meyer, J. Chem. Phys. 125 (2006) 204701.
[7] J.P. Camden, J.A. Dieringer, Y.M. Wang, D.J. Masiello, L.D. Marks, G.C. Schatz, R.P.
Van Duyne, J. Am. Chem. Soc. 130 (2008) 12616–12617.
1312
1186
612
600
773
800
[8] S.J. Lee, A.R. Morrill, M. Moskovits, J. Am. Chem. Soc. 128 (2006) 2200–2201.
[9] K. Imura, H. Okamoto, M.K. Hossain, M. Kitajima, Nano Lett.
2173–2176.
6 (2006)
[10] L.S. Zhang, Y. Fang, P.X. Zhang, Spectrochim. Acta A 69 (2008) 91–95.
[11] V.L. Schlegel, T.M. Cotton, Anal. Chem. 63 (1991) 241–247.
[12] L.S. Zhang, P.X. Zhang, Y. Fang, Anal. Chim. Acta 591 (2007) 214–218.
[13] L.S. Zhang, Y. Fang, P.X. Zhang, Chem. Phys. Lett. 451 (2008) 102–105.
[14] S. Chattopadhyay, H.C. Lo, C.H. Hsu, L.C. Chen, K.H. Chen, Chem. Mater. 17 (2005)
553–559.
[15] W.H. Ke, D.F. Zhou, J.Z. Wu, K. Ji, Appl. Spectrosc. 59 (2005) 418–423.
[16] C. Schmuck, P. Wich, B. Kustner, W. Kiefer, S. Schlucker, Angew. Chem. Int. Ed.
46 (2007) 4786–4789.
1000
1200
1400
1600
Raman shift (cm-1)
Fig. 7. SERS spectra of 10−12 M R6G obtained from six randomly selected points on
the surface of sandwich substrate.
[17] N. Leopold, B. Lendl, J. Phys. Chem. B 107 (2003) 5723–5727.

630
J. Zhang et al. / Spectrochimica Acta Part A 79 (2011) 625–630
[18] L. Rivas, S. Sanchez-Cortes, J.V. Garcia-Ramos, G. Morcillo, Langmuir 17 (2001)
574–577.
[19] C.H. Wang, D.C. Sun, X.H. Xia, Nanotechnology 17 (2006) 651–657.
[20] I. Yoon, T. Kang, W. Choi, J. Kim, Y. Yoo, S.W. Joo, Q.H. Park, H. Ihee, B. Kim, J.
Am. Chem. Soc. 131 (2009) 758–762.
[21] L.A. Dick, A.D. McFarland, C.L. Haynes, R.P. Van Duyne, J. Phys. Chem. B 106
(2001) 853–860.
[22] Q. Zhou, Y. Yang, J. Ni, Z.C. Li, Z.J. Zhang, Physica E 42 (2010) 1717–1720.
[23] L.S. Zhang, P.X. Zhang, Y. Fang, J. Colloid Interface Sci. 311 (2007) 502–506.
[24] R.L. Zong, J. Zhou, Q. Li, B. Du, B. Li, M. Fu, X.W. Qi, L.T. Li, J. Phys. Chem. B 108
(2004) 16713–16716.
[27] C.J. Orendorff, A. Gole, T.K. Sau, C.J. Murphy, Anal. Chem. 77 (2005) 3261–3266.
[28] H.X. Xu, J. Aizpurua, M. Kall, P. Apell, Phys. Rev. E 62 (2000) 4318–4324.
[29] C.D. Keating, K.K. Kovaleski, M.J. Natan, J. Phys. Chem. B 102 (1998) 9414–9425.
[30] P.C. Lee, D. Meisel, J. Phys. Chem.-US 86 (1982) 3391–3395.
[31] S.B. Chaney, S. Shanmukh, R.A. Dluhy, Y.P. Zhao, Appl. Phys. Lett. 87 (2005)
031908.
[32] N. Ji, W.D. Ruan, C.X. Wang, Z.C. Lu, B. Zhao, Langmuir 25 (2009) 11869–11873.
[33] G.H. Gu, J. Kim, L. Kim, J.S. Suh, J. Phys. Chem. C 111 (2007) 7906–7909.
[34] G.H. Gu, J.S. Suh, J. Raman Spectrosc. 41 (2010) 624–627.
[35] A. Gutes, C. Carraro, R. Maboudian, ACS Appl. Mater. Interf. 1 (2009) 2551–2555.
[36] X.T. Wang, W.S. Shi, G.W. She, L.X. Mu, S.T. Lee, Appl. Phys. Lett. 96 (2010)
053104.
[25] C. Shi, H. Yan, C. Gu, D. Ghosh, L. Seballos, S.W. Chen, J.Z. Zhang, B. Chen, Appl.
Phys. Lett. 92 (2008) 103107.
[37] Y.A. Yang, A.M. Bittner, K. Kern, J. Solid State Electrochem. 11 (2007) 150–154.
[26] J.K. Daniels, G. Chumanov, J. Phys. Chem. B 109 (2005) 17936–17942.