View Online / Journal Homepage / Table of Contents for this issue

PAPER www.rsc.org/pccp | Physical Chemistry Chemical Physics

Tuning plasmons on nano-structured substrates for NIR-SERS

Sumeet Mahajan,a Mamdouh Abdelsalam,a Yoshiro Suguwara,b Suzanne Cintra,a
Andrea Russell,a Jeremy Baumbergb and Philip Bartlett*a
Received 16th August 2006, Accepted 9th November 2006
First published as an Advance Article on the web 23rd November 2006
DOI: 10.1039/b611803h

Surface-Enhanced Raman Spectroscopy (SERS) is a very sensitive and selective technique for
detecting surface species. Colloidal crystal-templated ‘inverse opal’ nanostructured gold films have
Published on 23 November 2006 on http://pubs.rsc.org | doi:10.1039/B611803H

been demonstrated to be excellent SERS substrates by various researchers around the globe.
However, visible excitation laser sources commonly used in SERS experiments can cause
photochemical reactions on the surface as well as fluorescence from the adsorbed molecules. A
way to circumvent this possibility is the use of Near Infra-Red (NIR) laser sources. This demands
appropriate design of substrates for NIR-SERS in order to obtain maximum enhancement of
signals from analytes. In the current paper, we use systematic variation of sphere size and
electrochemical control over film height to tune plasmons on such nanovoid substrates. We use
plasmon maps as a tool for predicting NIR-SERS enhancements recorded with a 1064 nm laser
source for benzenethiol as the probe molecule. Direct correlation is observed between Raman
enhancements and plasmonic resonances with ingoing and outcoming radiation. Our study
demonstrates the feasibility of plasmon engineering and the predictive power of their mapping on
Downloaded on 05 July 2012

our substrates. It also demonstrates the ability to design reproducible NIR-SERS substrates and
its empirical fruition.

Introduction and resonance of ingoing laser excitation and outcoming

scattered radiation bearing the signature of the molecule with
Plasmon-active materials are being intensely investigated ow- plasmons on these nanovoid substrates has been probed.8,13
ing to their wide ranging applications.1 One of the most The inter-dependence of optical properties, plasmon modes
promising is their application in Surface-Enhanced Raman and SERS implies that it would be imperative to design
Scattering (SERS).2 SERS is an extremely sensitive and selec- substrates for use with a particular excitation laser and target
tive technique for detecting surface species. Most of the signal molecule for optimum SERS.
enhancement in SERS compared to conventional Raman is A typical SERS experiment is carried out using a visible
attributed to the large increase in the local electric field laser. This can cause photochemical reactions, interference due
generated.3 Since plasmons lead to such field enhancements, to fluorescence, and degeneration, especially of large biologi-
surfaces that support various plasmon modes could serve as cal molecules.14 Hence, the need for benign sources of excita-
SERS active substrates. However, the ability to produce such tion for SERS of biological molecules necessitates tailored
structures reproducibly is critical to their exploitation. substrates for Near Infra-Red (NIR) wavelengths and the
In recent years, we have demonstrated the ability to make particular analyte under investigation. In this report, we
reproducible nanostructured substrates using colloidal crystal- demonstrate the predictive power of plasmon mapping and
templated electrodeposition.4,5 Colloidal crystal-templated its use as a design tool for engineering substrates for NIR-
‘inverse opal’ nanostructured gold films have been demon- SERS. This represents the first report demonstrating the
strated to be excellent SERS substrates.6–8 Further, though the tunability and experimental realization of colloidal-templated
optical and enhancement properties of SERS substrates have gold ‘inverse opals’ for NIR-SERS using a 1064 nm laser
been found to be structure dependent,9,10 no reports exist source.
wherein such nanovoid substrates are targeted and tailored
for use with a particular excitation wavelength, especially in
the NIR region.
Experimental details
We have shown recently that plasmon modes on such
nanovoid substrates can be tuned as a function of sphere size Preparation of nanostructured substrates
and film height.11,12 Further, the relationship between SERS
Substrates were prepared following the previously published
method by Bartlett et al.4,10 A typical synthesis of a substrate
a
School of Chemistry, University of Southampton, Southampton, UK involves preparation of conductive surfaces by thermal vapour
SO17 1BJ. E-mail: pnb@soton.ac.uk; Fax: +44-23-8059-3781; Tel: deposition of a 10 nm layer of chromium, followed by 200 nm
+44-23-8059-3333 of gold on to glass slides. These gold surfaces were used, after
b
School of Physics and Astronomy, University of Southampton,
Southampton, UK SO17 1BJ. Fax: +44-23-8059-2093; Tel: +44- treatment with 10 mM cysteamine solution in ethanol for
23-8059-2094 48 h, for electrodeposition of gold through a template of

104 | Phys. Chem. Chem. Phys., 2007, 9, 104–109 This journal is c the Owner Societies 2007
 View Online

polystyrene latex spheres assembled on the surface. The images were obtained using the same magnification and it can
spheres were obtained as 1 wt % aqueous solutions from be clearly seen that the electrodeposited substrate is much
Duke Scientific Corp. The assembly was carried out in a smoother in comparison to the evaporated film. Samples
manner such that a monolayer template of hexagonally prepared in this way were subjected to reflectance measure-
packed, well-ordered spheres was obtained. Gold was electro- ments and spectra were obtained. Thereafter, FT-Raman
chemically grown from a commercial cyanide-free gold plating spectra were recorded on the sample from approximately the
solution (Tech. Gold 25, Technic Inc.). After the electrodepo- same positions as the reflectance spectra.
sition, the polymer spheres were removed by dissolving in
DMF. This resulted in substrates with an array of intercon- Reflectance and SERS measurements
nected voids of varying depth depending on the charge passed
during the electrodeposition. The reflectivity of samples prepared as described above was
In the current study, substrates were prepared with a range studied using a BX51TRF Olympus microscope illuminated
of sphere diameters from 700 nm to 1100 nm. Substrates were with an incoherent white light source coupled to an Ocean
Published on 23 November 2006 on http://pubs.rsc.org | doi:10.1039/B611803H

prepared in which the film height was graded by retracting the Optics NIR 512 (850–1800 nm) spectrometer. The spectra
template from the plating solution in incremental steps using a were recorded in the normal incidence mode. A 20X IR
micrometer stage. In the current report the film heights quoted objective with a numerical aperture of 0.40 was used to record
are as estimated from their pore-mouth diameters, as de- the spectra. All spectra were normalized with respect to that
scribed by Bartlett et al.10 recorded on flat gold deposited on glass slides by vapour
deposition. Gold was chosen for normalization as its absor-
bance is fairly constant over the wavelength range studied and
Structure determination
it is chemically stable. A CCD camera (Olympus DP2)
The structure and morphology of the templated films were mounted on the microscope enabled the simultaneous record-
investigated using Philips XL30 Environmental Scanning ing of optical images of the area whose spectrum was being
Downloaded on 05 July 2012

Electron Microscope (ESEM). Representative SEM images monitored. NIR-SERS measurements were made using a
presented in Fig. 1 for substrates prepared with 900 nm Perkin Elmer System 2000 FTIR spectrometer with an
Diameter (D) sphere templates show that uniform gold films FT-Raman attachment equipped with a 1064 nm Nd:YAG
were obtained with almost negligible defects. laser. Raman spectra were recorded in the back-scattered
High resolution images were obtained with a Jeol JSM 6500 geometry between 200 cm 1 to 3500 cm 1 wavenumber shift
F thermal Field Emission Scanning Electron Microscope using an indium–gallium–arsenide (InGaAs) detector. A
(FESEM). FESEM images obtained of both the evaporated spectral resolution of 4 cm 1 was used with the incident laser
gold, upon which the substrates are grown, and an electro- power at the sample measured to be 250 mW. For all spectra,
chemically deposited substrate are presented in Fig. 2. The 50 scans were averaged.

Fig. 1 SEM images of a graded gold film fabricated using 900 nm D sphere template recorded at various film heights. Images corresponding to
film thicknesses of: (a) 0.25 D; (b) 0.45 D; (c) 0.7 D; and (d) 0.9 D are shown. All images were obtained under 10 000 magnification. The scale bar
for the images is 2 mm.

This journal is c the Owner Societies 2007 Phys. Chem. Chem. Phys., 2007, 9, 104–109 | 105
 View Online

Fig. 2 FESEM images showing the smoothness of (b) electrodeposited structured film compared to (a) flat evaporated gold used as substrate.
Published on 23 November 2006 on http://pubs.rsc.org | doi:10.1039/B611803H

Evaporated gold also visible in the middle at the bottom of the voids is clearly rougher than electrodeposited structured gold film. The voids shown
in (b) correspond to a film height of 0.9 D templated with 900 nm spheres. The scale bar for both images is 200 nm.

Results and discussion same data is presented as an extinction map in Fig. 3b.
Extinction, being the sum of absorbance and scattering, gives
Reflectance spectra were recorded, in the normal incidence an upper bound on the extent of plasmonic interactions on the
and back reflection geometry with the apparatus described substrate. However, we have previously shown that absorp-
above, on gold films templated with different sphere sizes. tion dominates on these substrates.11 Thus, Fig. 3 gives the
Cintra et al. have shown that the various plasmon modes relationship between reflectance spectra obtained at different
Downloaded on 05 July 2012

generated on nanostructured gold substrates gradually red thicknesses and the corresponding wavelength–position map
shift as the sphere diameters are increased from 350 nm to showing the extent of extinction for an 800 nm sphere-tem-
800 nm.8 Therefore, by increasing the sphere sizes further we plated graded gold film.
expected to generate plasmons at NIR wavelengths. Hence, in Extinction maps in the NIR range of 850 to 1800 nm were
the present study, sphere diameters of 700, 800, 900, 1000 and obtained for graded gold nanovoid films templated with
1100 nm were used. The reflectance recorded from graded gold 700 nm, 800 nm, 900 nm, 1000 nm and 1100 nm diameter
films was normalized with respect to evaporated flat gold, as in spheres. The maps for 700, 900 and 1100 nm sphere-templated
the NIR region gold has an almost negligible and wavelength- gold films are shown in Fig. 4. Localized plasmons have been
independent absorbance. The spectra for the graded 800 nm shown to exist at B0.6 D for our nanostructured substrates.15
sphere-templated substrate are presented on a logarithmic This appears as a weak band for the 700 nm nanovoid film, as
scale (analogous to absorbance) with the normalized reflec- evident in Fig. 4a. This band undergoes a red shift on
tance being plotted against wavelength in Fig. 3a. For visua- increasing the sphere size to 900 and 1100 nm (Fig. 4b and
lizing the plasmon resonances that occur on the substrates, the c). The extent of absorption is largest for 900 nm sphere-

Fig. 3 Data obtained from reflectance measurements at different film thicknesses on a graded gold film is converted to an extinction intensity
map. In (a), reflectance spectra recorded at different film thicknesses on a graded gold film templated with 800 nm spheres are offset for clarity. The
film heights in the figure, shown under each spectrum, increase from top to bottom. R and R0 are reflection intensities obtained from the
nanostructured film and flat gold, respectively. The corresponding extinction intensity contour map is shown in (b). The dips correspond to the
bright areas. The extinction intensity is normalized with respect to flat gold. Red indicates a maximum of 1 and violet stands for the least (complete
reflection) extinction of 0 with respect to flat gold. The extinction intensity maps relate directly (with an element of over-estimation, as extinction
could occur due to scattering as well) to possible plasmonic resonances on the substrate as a function of film height. The film thickness is
normalized with respect to sphere diameter.

106 | Phys. Chem. Chem. Phys., 2007, 9, 104–109 This journal is c the Owner Societies 2007
 View Online
Published on 23 November 2006 on http://pubs.rsc.org | doi:10.1039/B611803H

Fig. 4 Extinction contour maps for (a) 700 nm, (b) 900 nm and (c) 1100 nm sphere-templated gold nanostructured films. All maps are colour-
coded using the same spectrum scale of 0 (violet) to 1 (red), as in Fig. 3. The film heights are normalized with respect to sphere diameter.

templated films compared to the 700 nm and 1100 nm sphere-templated film showed a SERS enhancement profile
substrates. Extinction maps with 800 and 1000 nm sphere- similar to that of the 700 nm sphere-templated film, with the
templated films display the same trend and, hence, have not maximum signal being obtained at B0.75 D, albeit with much
been included in this figure. One can also see Bragg plas- higher intensities. The enhancement factor for the 800 nm
mons11,13 emerging in the NIR region, centered around nanovoid film at B0.75 D thickness was only slightly lower
900 nm and 1100 nm in Fig. 4b and c, respectively, corre- than that obtained for the 900 nm film at B0.6 D. The
sponding to the grating-like behaviour of shallow dishes. 1000 nm nanovoid film behaved very similarly to the
Downloaded on 05 July 2012

Raman spectra of adsorbed benzenethiol were obtained as a 1100 nm sphere-templated film and showed very little SERS
function of position (film height) using the same substrates enhancement.
that were studied by reflectance. The spectrum for benzene- To understand the origin of such structural dependence of
thiol obtained at B0.75 D using the 800 nm sphere-templated Raman enhancements, we carried out plasmon mapping and
sample is shown in Fig. 5a. The peak assignments indicated in SERS spectral profiling on all substrates from 700 to 1100 nm
the figure are based on those reported by Han et al.16 The sphere-templated structured films and studied them in tandem.
absence of a peak associated with the n(S–H) stretching As expected, the SERS enhancement was higher at those film
vibration confirms the formation of no more than a monolayer thicknesses where existence of a plasmon mode was indicated
of adsorbed benzenethiol and its linkage with gold; the S–H on the extinction map compared to those where a weak or no
bond is cleaved when the Au–S bond forms. All samples were plasmonic interaction was observed. However, the effect is
treated with benzenethiol solutions for 24 h and dried under more subtle than this. On overlaying the SERS intensities
argon before recording the SERS spectra. No significant corresponding to 420 cm 1, 997 cm 1 and 1570 cm 1 Raman
increase in SERS intensities were observed on increasing the bands of benzene thiol at the appropriate outcoming wave-
exposure time to 48 h. The SERS enhancement factors were lengths on the extinction maps along with the ingoing excita-
determined by comparison to the spectrum of neat benze- tion laser, we can see an obvious correlation (Fig. 6). Wherever
nethiol and assuming the formation of a compact monolayer, both the ingoing and outcoming wavelengths matched with
as described previously.8,17 The enhancement factor of the strong extinction on the substrate, we can see a large enhance-
B0.75 D 800 nm substrate corresponding to the spectrum ment in SERS.
shown in Fig. 5a was calculated to be 2.6  106. The SERS The matching of wavelengths correspond to the desired
from the substrates were stable and repeatable: thus, the resonance coupling between the laser excitation as well as
standard deviation was found to be less than 6.3 % for spectra scattered radiation with the plasmons generated on the struc-
(n = 14) recorded from approximately the same position tured nanovoid surface and, hence, a large increase in SERS.
(within 0.5 mm) on different occasions over several days. Existence of plasmon modes on the structured substrates not
The SERS spectral profiles with respect to film height for only helps in the coupling of energy between ingoing radiation
700, 900 and 1100 nm sphere-templated substrates are shown with the molecule adsorbed on the surface, but also assists in
in Fig. 5b–d. The variation of SERS signals with sphere the coupling of scattered radiation back from the surface.
diameter of the template and film height is clearly evident. Hence, resonance between ingoing excitation and outcoming
The intensity of peaks in the SERS spectra exhibit a strong scattered radiation with plasmon modes on the surface is
dependence on the thickness of the film as well as the sphere critical for optimization of SERS. For the 700 nm sphere-
diameter used for templating the structure. The maximum templated film, Fig. 6a, the plasmon resonances are not strong
enhancements occur between 0.4 and 0.8 D depending on the enough in the NIR; consequently, they do not contribute to
templating sphere diameter. Among all the samples, the great- large NIR-SERS enhancement. Nevertheless, the maximum
est SERS enhancements were obtained with the 900 nm enhancements do occur when the ingoing and outcoming
sphere-templated nanovoid film at around B0.6 D. The radiation resonates with the plasmons on the substrate. The
enhancement factor18 was calculated to be 3  106, which effect of matching the incident and exiting radiation with
was the highest among the substrates in this study. The 800 nm plasmon modes is most pronounced in Fig. 6b and c for 800

This journal is c the Owner Societies 2007 Phys. Chem. Chem. Phys., 2007, 9, 104–109 | 107
 View Online
Published on 23 November 2006 on http://pubs.rsc.org | doi:10.1039/B611803H
Downloaded on 05 July 2012

Fig. 5 Spectrum of benzenethiol adsorbed on (a) 800 nm sphere-templated gold film at 0.75 D film height, showing the absence of an –SH peak.
The SERS spectra recorded at different film heights for: (b) 700 nm; (c) 900 nm; and (d) 1100 nm sphere-templated gold films. The peak intensities
pass through a maximum with respect to film height. The enhancements are most pronounced in (c), in which the intensity scale is 0–15, while in (a)
and (d) it is 0–4 AU. The spectra were recorded with an FT-Raman spectrometer equipped with a 1064 nm Nd:YAG laser. 50 scans were averaged
under a laser incident power of 250 mW at the sample. Spectral resolution was 4 cm 1.

and 900 nm sphere-templated structured films. Large signal sphere size and film height. With respect to the latter control,
enhancement occurs on resonance between strong plasmonic electrodeposition appears much more powerful than other
absorption on the surface with the ingoing and outcoming techniques of metal deposition such as nanoparticle-assisted6,7
radiation. The importance of plasmon resonances with both or vapour deposition.19,20 In addition, our high resolution
incoming and outgoing radiation can be gauged from Fig. 6d FESEM images show that electrodeposited films are much
and e, where either the ingoing or outcoming radiation, or smoother than the vapour deposited gold. The flat evaporated
both, are not able to couple with plasmonic resonances on the gold films yield no SERS, hence, we conclude that it is the
surface and therefore there is reduced SERS enhancement. macro-structure of the templated films that is responsible for
In this study we have shown the scalability of the plasmon SERS rather than generation of nanometer scale surface
engineering approach into the NIR region. Although it is a roughness or ‘hot spots’. This option of tuning plasmons on
matter of further study on our part as to what part of our colloidal-templated nanostructured films offers an unprece-
nano-structured film contributes to the plasmon modes in the dented tool for designing and tailoring substrates for obtain-
NIR energy range, it is clear that we can control the generation ing SERS with a particular excitation source and a target
of suitable plasmon modes by varying parameters like the analyte molecule.

108 | Phys. Chem. Chem. Phys., 2007, 9, 104–109 This journal is c the Owner Societies 2007
 View Online
Published on 23 November 2006 on http://pubs.rsc.org | doi:10.1039/B611803H

Fig. 6 Extinction intensity contour maps superimposed with Raman scattering peak intensities for gold substrates templated with: (a) 700 nm; (b)
800 nm; (c) 900 nm; (d) 1000 nm; and (e) 1100 nm spheres. ( ) denotes the ingoing laser excitation at 1064 nm in the NIR-SERS experiment. ( ),
( ) and ( ) represent the outcoming radiation at 1114 nm, 1191 nm and 1277 nm corresponding to 420 cm 1, 1000 cm 1 and 1570 cm 1 Raman
bands of benzenethiol, respectively. The lengths of the arrows indicate the intensity of the scattered radiation (on a scale of 0 to 8.5 AU).

Conclusions 8 S. Cintra, M. Abdelsalem, P. N. Bartlett, J. J. Baumberg, T. Kelf,

Y. Sugawara and A. E. Russell, Faraday Discuss., 2005, 132,
In the present study, we have used variation in sphere size of 1–9.
colloidal templates and electrochemical control over film 9 R. C. Schroden, M. Al-Daous, C. F. Blanford and A. Stein, Chem.
Mater., 2002, 14, 3305–3315.
height to tune substrates for NIR-SERS in conjunction with
Downloaded on 05 July 2012

10 P. N. Bartlett, J. J. Baumberg, S. Coyle and M. E. Abdelsalam,

a 1064 nm laser source for a particular probe molecule, in this Faraday Discuss., 2004, 125, 117–132.
case benzenethiol. Proper combination of the two parameters 11 T. A. Kelf, Y. Sugawara, J. J. Baumberg, M. Abdelsalam and P. N.
Bartlett, Phys. Rev. Lett., 2005, 95, 116802.
of sphere size and film height resulted in significant enhance- 12 S. Coyle, M. C. Netti, J. J. Baumberg, M. A. Ghanem, P. R.
ment of signals from the adsorbed analyte. We have demon- Birkin, P. N. Bartlett and D. M. Whittaker, Phys. Rev. Lett., 2001,
strated the effect of matching incident and scattered radiation 8717.
with plasmonic resonances on the enhancement of Raman 13 J. J. Baumberg, T. Kelf, Y. Sugawara, S. Cintra, M. Abdelsalam,
P. N. Bartlett and A. E. Russell, Nano Lett., 2005, 5, 2262–2267.
bands of benzenethiol adsorbed on our nanostructured gold 14 P. Hendra, C. H. Jones and G. M. Warnes, Fourier Transform
substrates. SERS intensities could be significantly improved by Raman Spectroscopy—Instrumentation and Chemical Applications,
designing substrates that can sustain plasmonic resonances to Ellis Horwood Limited, Chichester, 1991.
15 T. A. Kelf, Y. Sugawara, R. M. Cole, J. J. Baumberg, M. E.
couple with the ingoing and outcoming radiation. Our study
Abdelsalam, S. Cintra, S. Mahajan, A. E. Russell and P. N.
proves the feasibility of the design, and the experimental Bartlett, Phys. Rev. B, 2006, accepted.
realization, of reproducible NIR-SERS substrates. 16 S. W. Han, H. J. Lee and K. Kim, Langmuir, 2001, 17, 6981–6987.
17 J. C. Love, L. A. Estroff, K. J. K. R. G. Nuzzo and G. M.
Whitesides, Chem. Rev., 2005, 105, 1103–1169.
References 18 Enhancement Factors (EF) were calculated by comparing the
intensity of the SERS spectra to those for neat benzene thiol.
1 W. I. Barnes, A. Dereux and T. W. Ebbesen, Nature, 2003, 424, The expression for the enhancement factor is EF = IsurfAmVlrNA/
824–830. IbulkAlRMbt, where Isurf is the intensity of the benzenethiol peak for
2 Z. Q. Tian, J. Raman Spectrosc., 2005, 36, 466–470. the substrate, Am is the area occupied by an adsorbed thiol
3 K. Kneipp, H. Kneipp, I. Itzkan, R. R. Dasari and M. S. Feld, J. molecule (from ref. 17), Vl is the laser volume for the bulk
Phys.: Condens. Matter, 2002, 14, R597–R624. experiment, NA is Avogadro’s number, r is the density of neat
4 P. N. Bartlett, P. R. Birkin and M. A. Ghanem, Chem. Commun., benzenethiol, Ibulk is the intensity of neat benzenethiol peak, Al is
2000, 1671–1672. laser spot area, R is the geometrical roughness factor for the
5 M. E. Abdelsalam, P. N. Bartlett, J. J. Baumberg, S. Cintra, T. A. surface due to the templated structure (1 r R r 4.63) and Mbt
Kelf and A. E. Russell, Electrochem. Commun., 2005, 7, 740–744. is the molecular weight of benzenethiol.
6 P. M. Tessier, O. D. Velev, A. T. Kalambur, J. F. Rabolt, A. M. 19 J. C. Hulteen, D. A. Treichel, M. T. Smith, M. L. Duval, T. R.
Lenhoff and E. W. Kaler, J. Am. Chem. Soc., 2000, 122, Jensen and R. P. Van Duyne, J. Phys. Chem. B, 1999, 103,
9554–9555. 3854–3863.
7 P. M. Tessier, O. D. Velev, A. T. Kalambur, A. M. Lenhoff, J. F. 20 T. R. Jensen, G. C. Schatz and R. P. Van Duyne, J. Phys. Chem. B,
Rabolt and E. W. Kaler, Adv. Mater., 2001, 13, 396–400. 1999, 103, 2394–2401.

This journal is c the Owner Societies 2007 Phys. Chem. Chem. Phys., 2007, 9, 104–109 | 109