Development of highly reproducible nanogap SERS substrates: Comparative performance analysis and its application for glucose sensing

Biosensors and Bioelectronics 26 (2011) 1987–1992

Contents lists available at ScienceDirect

Biosensors and Bioelectronics journal homepage: www.elsevier.com/locate/bios

Development of highly reproducible nanogap SERS substrates: Comparative performance analysis and its application for glucose sensing U.S. Dinish a , Fu Chit Yaw a , Ajay Agarwal b , Malini Olivo a,c,d,e,∗ a

Bio Optical Imaging Group, Singapore Bioimaging Consortium, Agency for Science Technology and Research, 11 Biopolis Way, Singapore 138667, Singapore Institute of Microelectronics, Agency for Science Technology and Research, Science Park 2, Singapore 117685, Singapore Division of Medical Science, National Cancer Centre Singapore, 11 Hospital Drive, Singapore 169610, Singapore d Department of Pharmacy, National University of Singapore, 18 Science Drive 4, Singapore 117543, Singapore e School of Physics, National University of Ireland, Galway, Ireland b c

a r t i c l e

i n f o

Article history: Received 21 June 2010 Received in revised form 18 August 2010 Accepted 20 August 2010 Available online 24 September 2010 Key words: Surface-enhanced Raman scattering Deep UV lithography Nanogap SERS substrates Reproducibility Glucose sensing

a b s t r a c t We report a new class of a SERS substrate with ordered nanostructures fabricated on silicon wafer using a deep UV (DUV) lithography technique followed by surface coating of silver and/or gold film. These substrates possess sharp edged nanogaps, which are responsible for the SERS enhancement. SERS performance of these substrates was analyzed by studying its reproducibility, repeatability and signal enhancement measured from 2-naphthalene thiol (NT) molecule covalently anchored on to the substrate. SERS performance of this substrate was also compared with a commercial substrate and metal film over nanosphere (MFON) substrate, which is one of the most promising reported substrates. It was found that MFON substrate showed a slightly higher SERS intensity among all three chosen substrates, but the relative standard deviation (RSD) of the intensity for the two prominent peaks of NT was about 7–14% while for our nanogap DUV substrate the RSD was less than 3% with comparable SERS signal intensities to MFON. For the commercial substrate, the relative standard deviation was about 7–9% but with a much lower SERS signal intensity. To our knowledge, this observed reproducibility along with good SERS enhancement with nanogap substrate is the best among the reported SERS substrates. These observed results with the nanogap substrate show great potential for the development of a sensitive SERS biosensing platform. Efficacy of the nanogap DUV substrate for biosensing was demonstrated for in vitro glucose sensing under physiologically relevant conditions. © 2010 Elsevier B.V. All rights reserved.

1. Introduction The discovery of the surface-enhanced Raman scattering (SERS) effect about 33 years ago helped to overcome the inherent sensitivity limitation of Raman spectroscopy (Jeanmaire and Van Duyne, 1977; Albrecht and Creighton, 1977). Since then SERS has been emerged as a powerful analytical tool for various biosensing applications. When developing a SERS sensor, the substrate on to which analyte molecules are physically or chemically adsorbed plays the important role. Several types of novel SERS-active substrates such as metal film over nanosphere (MFON) (Dick et al., 2002), biomimetic substrate (Garrett et al., 2009), porous film (Qian et al., 2007; Wijnhoven et al., 2000; Giorgis et al., 2008), nanoparticle substrate (Grabar et al., 1995; Addison and Brolo, 2006; Bright et al., 1996; Musick et al., 2000; Mulvaney et al., 2003; Fan and Brolo,

∗ Corresponding author at: Bio Optical Imaging Group, Singapore Bioimaging Consortium, Agency for Science Technology and Research, 11 Biopolis Way, Singapore 138667, Singapore. Tel.: +65 64368317; fax: +65 63720161. E-mail address: [email protected] (M. Olivo). 0956-5663/$ – see front matter © 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.bios.2010.08.069

2008, 2009), nanowell and nanopore array (Liu and Lee, 2005; Choi et al., 2010) lithography based substrate (Vo Dinh et al., 1986; Li et al., 2008; De Jesus et al., 2005) etc. have been developed to improve the applicability of SERS for sensing applications. Among these SERS substrates, the most promising one is the MFON substrate due to the strong SERS enhancement and ease to fabricate. This substrate has shown great promise and has been successfully used for in vitro (Lyandres et al., 2005) and in vivo glucose sensing in animal models (Stuart et al., 2006). The major factor that determines the efficacy of a substrate in sensing is its reproducibility and repeatability. Reproducibility of SERS substrate refers to the ability to produce enhanced signal at various part of the substrate with minimum intensity variation. Repeatability refers to the minimum batch to batch variation of substrate. Reproducibility and repeatability of the substrates are highly critical since most of the SERS-based sensing is relying on the intensity of particular spectral peak/bands. Surface roughness and hence reproducibility of MFON substrates is determined by the close packing and arrangement of PS beads. It is very difficult to obtain well packed PS in a long range order, especially when smaller PS (less than 200 nm diameter) is used (Lin et al., 2009).

1988

U.S. Dinish et al. / Biosensors and Bioelectronics 26 (2011) 1987–1992

This causes some limitations in the reproducibility and stability of the substrate. Though, less than 20% variation in sample intensity between various points of the substrate is well acceptable in SERS measurements (Natan, 2006; Lin et al., 2009), there is always a high demand for better reproducible and more stable SERS substrates. In this context, we report a new class of substrate fabricated on silicon (Si) wafer with nanostructures in the form of nanogaps, which was patterned using deep UV (DUV) photolithography technique followed by e-beam deposition of metal. Compared to conventional lithography, this approach allows better control of the nanostructures resulting in uniform roughness at metal film and in turn provides better reproducibility and enhancement for the substrate. SERS performance of these nanogap DUV substrates is compared with MFON and commercial substrate by studying the reproducibility, repeatability and signal enhancement from 2naphthalene thiol (NT), which can covalently bind to the substrate. In this study, monometallic and bimetallic MFON substrates like AgFON, Ag–AuFON, nanogap DUV Ag and nanogap DUV Ag–Au substrates are analyzed. Bimetallic substrates were fabricated by coating Au over the Ag layer. Though Ag substrates have shown excellent SERS enhancement, it is associated with inherent problems such as (i) easily get oxidized, (ii) ineffective for testing biological samples in phosphate buffer saline (PBS) medium where substrate gets peeled-off (Scholes et al., 2008), (iii) fluctuating SERS activity resulting from laser-induced structural changes in the silver oxide layers (Buchel et al., 2001) and (iv) photodynamics of Ag under oxygen-containing atmosphere (Jacobson and Rowlen, 2006). Even though Au substrates provide weaker SERS enhancement compared to Ag, it is relatively chemically inert and hence we fabricated bimetallic substrates. Applicability of nanogap DUV substrate for sensitive glucose sensing under physiologically relevant condition is demonstrated in this paper using self assembled monolayer (SAM) surface functionalization (Lyandres et al., 2005). SERS-based direct glucose sensing is highly challenging because of the poor Raman crosssection of the glucose molecule. Unfortunately, surface roughness and related stability issues associated with substrates often add further challenge to the SERS-based glucose sensing platform. In this context, direct glucose sensing is demonstrated using highly reproducible nanogap substrates where change in SERS intensity of the prominent peaks in the glucose spectra can be monitored as a function of its concentration.

2. Materials and methods 2.1. Substrate fabrication 2.1.1. DUV lithography nanogap substrates The highly ordered arrays of nanostructures coated with noble metals were fabricated on 8 in. diameter single crystal p-type Si wafers. The whole fabrication process comprises of DUV photolithography to pattern nanostructures, silicon etching to realize nano-dimensional patterns followed by oxidation to control the gap between nanostructures. For lithography, positive photoresist (4100 A˚ thick) with baking conditions of 130 ◦ C/90 s and 60 s puddle developments were used. A single binary mask with circular patterns was utilized to achieve different inter-structure gaps by varying the exposure dosage (66, 70, 74 and 78 mJ cm−2 ). The larger exposure dosages lead to smaller structures with larger interparticle spacing in the photoresist patterns, which was translated to silicon after etching. Silicon etching was accomplished in inductively coupled deep reactive ion etching system using SF6 and C4 F8 chemistry. The depth of the etching was maintained at about 150 nm. Dry oxidation at 900 ◦ C for 2–6 h was used to control the nanostructure’s spacing. Subsequently, sharp tip nanogap struc-

tures were coated with metal using an e-beam evaporation system with Ag (30 nm) to form DUV Ag and Au (15 nm) over DUV Ag substrate to result in DUV Ag–Au substrate. 2.1.2. MFON substrates In-house MFON substrate with ordered nanostructures were fabricated by deposition of closely packed polystyrene (PS) nanospheres (diameter 384 nm, 2.5 wt.%, Kisker) on glass substrates followed by metal coating (Fu et al., 2009). Details of the substrate fabrication are given in supplementary material. 2.1.3. Commercial substrate Commercial SERS substrate (KlariteTM , D3 technologies) was purchased and used as received without further modification. This substrate consists of a 6 mm × 10 mm chip coated with Au with an active nanostructured area of 4 mm × 4 mm fabricated on Si. 2.2. Surface modification of substrates for glucose sensing Surface functionalization of the substrates was carried out using mixed monolayers (Lyandres et al., 2005) to form SAM. SERS substrates cleaned with ethanol and dried in Argon gas stream was first incubated in 1 mM decanethiol (DT, Fluka analytical) solution in ethanol for 45 min. After that, it was thoroughly washed with ethanol and transferred to 1 mM mercaptohexanol (MH, Fluka analytical) solution in ethanol and incubated for 12 h. Later, these substrates were washed thoroughly with ethanol followed by drying in Argon gas stream. All the functionalized substrates were stored in dry box. SAM surface functionalization is expected to effectively partition the glucose molecule near to the SERS-active substrate. 2.3. SERS instrumentation and measurement SERS experiments were carried out using Raman microscope (Renishaw InVia) system operating at 633 nm laser. System was connected to the microscope (Leica) and laser light was coupled through an objective lens (50×, 0.75 N.A.). The maximum laser power at the sample was 6.2 mW. Stokes shifted Raman spectra were collected with a spectral resolution of about 1 cm−1 . The instrument was calibrated with the Raman signal from a silicon standard centered at 520 cm−1 . Baseline correction of the measured spectra was performed to remove the broad background and fluorescence band. SERS performance of the substrates was evaluated by measuring the signal intensity from NT (Sigma–Aldrich) molecule. Monolayer of NT on the substrate was formed by incubating it in 10 ␮M solution in ethanol for 2 h and followed by washing with ethanol to remove unbound molecule. SERS measurement of NT in the 600–1800 cm−1 range was carried out with an integration time of 10 s. Throughout the analysis, stokes shifted Raman intensity at 1379 and 1066 cm−1 peak was used to compare the SERS performance among the various substrates. Measurements were done at ten random locations of substrates, which were about 10 ␮m apart and subsequently relative standard deviation (RSD) of the intensity was calculated. 2.4. Glucose sensing in SERS platform First of all, un-enhanced Raman spectrum from a concentrated 1 M, 20 ␮L glucose solution in PBS (pH 7.4) was measured. Spectral acquisition was carried out with a laser power of 6.2 mW and with an integration time of 500 s in the 450–1400 cm−1 range. In the SERS measurement, the spectrum of the control substrate (bare DT–MH treated DUV Ag–Au with 20 ␮L of PBS solution) was initially recorded. Later, the functionalized substrates were incubated with

U.S. Dinish et al. / Biosensors and Bioelectronics 26 (2011) 1987–1992

1989

Fig. 1. Characterization of bimetallic nanogap DUV substrate (A) SEM image and (B) high resolution AFM image showing nanogap is 30 nm.

glucose solutions at different concentrations such as 5, 10, 15, 20 and 25 mM for 10 min and SERS spectrum was acquired. All these experiments were repeated at three different locations of the substrate with 6.2 mW laser power and with an integration time of 2 min. Spectral intensity obtained from five different locations of the substrate was averaged. Subsequently, the intensity values of the spectra obtained from the control substrate were subtracted and variation of the intensity of the prominent peaks of glucose versus its concentrations can be monitored. 3. Results and discussion

nanogap to be about 30 nm over the active region of the substrate. Ag and Au coated nanogap DUV substrates were analyzed by cross-sectioning the active area using focused ion beam followed by transmission electron microscopy (TEM) technique. As shown in Fig. 2, energy dispersive X-ray (EDX) analysis of these TEM sample reveals that Au layer is deposited firmly on top of Ag due to the excellent solid solubility of the metals. The Ag and Au thicknesss ratio was found to be 2:1 and almost similar concentrations of the two metals were revealed at these three positions where analysis was carried out.

3.1. Surface characterization of substrates To understand the surface morphology and nanometric roughness that determines the SERS enhancement and reproducibility, three substrates were characterized by scanning electron microscopy (SEM) and atomic force microscopy (AFM, applied materials working in tapping mode with tip radius about 10 nm). It can be noted that the topography of all the substrates are different and possess unique nanopattern, which is required for SERS enhancement. As illustrated in Fig. 1A, SEM image of the Ag–Au DUV substrate reveals that the nanogap is uniform throughout the active area. These nanogaps are considered as the ‘hot spots’ for enhancing the Raman signal. By changing the lithography exposure dosage it is possible to control the distance of closest approach between adjacent nanostructures (Tan et al., 2007). In SERS, such sharp edged nanogaps are expected to generate intense hotspots. Typically, 22–24 nanogaps are covered by the laser spot of 1 ␮m2 . The edges of the star island structures may also contribute to the electromagnetic enhancement. Size of these nanogaps was measured to be in the range of 30 ± 5 nm. However, it is difficult to precisely control the nanogaps with dimension less than 30 nm. Fig. 1B shows the high resolution AFM image and the cross-section along the two adjacent nanostructures that confirms the size of

Fig. 2. TEM/EDX image of nanogap DUV substrate coated with 30 nm silver and 15 nm gold. Marked three circles denote the area where elemental analysis was carried out.

1990

U.S. Dinish et al. / Biosensors and Bioelectronics 26 (2011) 1987–1992

Fig. 3. Comparison of the SERS performance of MFON, nanogap DUV and commercial substrates at normalized experimental conditions. The error bars denoting their relative standard deviation. Sample = 10 ␮M NT, observed Raman peak = 1378 and 1066 cm−1 , laser = 633 nm, power = 60 ␮W and integration time = 10 s.

SEM and AFM images of the MFON substrate that was fabricated in our lab are given in supplementary material (Fig. S1). It shows that MFON substrates possess well-ordered hexagonal packing of PS beads (Fig. S1A). At the high magnification of SEM and AFM images (Fig. S1B), it can be noted that there is noticeable roughness on the PS beads due to metal coating, which contributes to the SERS enhancement. This observed roughness is also expected to be contributing to the SERS enhancement. SEM and AFM images of the commercial substrate are given in supplementary material (Fig. S2) and shows that SERS-active region is dominated by an array of inverted pyramidal depressions arranged into a square lattice (Fig. S2A). This substrate possesses high spatial uniformity with no noticeable defects over relatively large area. AFM image (Fig. S2B) reveals that these pyramidal nanostructures are separated by about 600 nm in both horizontal and vertical direction, which agree with the reported data (Alexander and Le, 2007). 3.2. SERS study of the substrates In order to compare the performance of various substrates, RSD of the SERS intensity at two prominent peaks of NT at 1066 and 1378 cm−1 , was calculated at normalized experimental settings and is given in Fig. 3. Among the three substrates, MFON shows the highest SERS intensity. Further, Ag substrates possesses higher signal enhancement than the bimetallic (Ag–Au) configuration, which is due to the better plasmonic properties of Ag. Our experiments revealed that despite the strong enhancement of Ag substrates, it was chemically unstable when tested with some biological samples in PBS and resulted in the peeling off the metal coating. Comparing the performance of Ag and Ag–Au substrates under same category, it can be noted that Ag–Au configuration always yields less signal intensity compared to Ag substrates. However, we noticed that the Au-film could adhere strongly to the Ag-film and has shown excellent stability when tested with biological samples in PBS. Such bimetallic configuration not only helps to tap the better enhancement from Ag but also provide better stability to overcome the limitations of Ag substrates. Here, plasmonic properties of this configuration can be considered as a ‘hybrid’ of Ag and Au. We observed that nanogap DUV substrates always showed better reproducibility compared to MFON and commercial substrates. When developing SERS substrates, one has to often compromise either on signal enhancement or on reproducibility. Due to this, it is always challenging to have a SERS substrate with adequate

Fig. 4. Representation of the 10 ␮M NT SERS spectra from Ag–AuFON, nanogap DUV Ag–Au and commercial substrates before background subtraction. Laser = 633 nm, power = 60 ␮W and integration time = 10 s.

signal enhancement and lower intensity variation. We noticed that AgFON substrates have provided almost double the signal intensity compared to the bimetallic configuration, but with a RSD of about 11–14%. Bimetallic configuration has shown better performance with a lower variation of only 6–9%. Again, though AgFON and nanogap DUV Ag substrates possesses similar enhancement, the intensity variation from our nanogap Ag substrate is only one-third. The RSD of the signal intensity from the commercial substrate was in the range of 7–9%, which is in good agreement with reported data (Perney et al., 2007). However, the signal enhancement from commercial substrate was weak and 1–2 orders lower than that of other two types of substrates. Nanogap DUV Ag substrates have shown good reproducibility with a variation of 3–5% while its bimetallic configuration showed excellent SERS performance with RSD as low as only 2–3% without much compromising on the signal enhancement. Experiments were repeated for substrates from different batches and all showed consistent results as mentioned above. SERS signal intensity variation of the substrates is summarized in supplementary material (Table S1). The observed smaller intensity variation for nanogap substrates is remarkable and probably the least among the previously reported substrates fabricated by various techniques, which possess reproducibility with a RSD in the range of 10–20% (Zhou et al., 2009; Brown and Martin, 2008; Oran et al., 2008; Sutherland and Winefordner, 1991; Mulvaney et al., 2003; Khan et al., 2009; De Jesus et al., 2005; Driskell et al., 2008). Fig. 4 shows the NT SERS spectra measured at 60 ␮W laser excitation from Ag–AuFON, DUV Ag–Au and commercial substrates before the background subtraction. We observed that MFON possess an inherently strong background and fluorescence compared to nanogap DUV substrate. We assume that it may be due the PS beads or the impurities associated with it. At lower laser power, signal to noise ratio from the nanogap substrate was always better than the other two types of substrates. Moreover, at higher laser power (600 ␮W), the background signal from MFON substrate also increases and saturates the detector. This prevents the further increase of the excitation power to yield better signal to noise ratio. Since bimetallic nanogap substrates have shown comparatively less background signal intensity, it was possible to work at higher laser power to yield better signal to noise ratio. Commercial substrate has also shown strong background spectra at relatively lower signal enhancement.

U.S. Dinish et al. / Biosensors and Bioelectronics 26 (2011) 1987–1992

1991

3.3. Estimation of SERS enhancement factor In order to analyze the SERS signal enhancement capability of SERS substrates, its absolute enhancement factor, G is studied. The absolute quantification of the enhancement factor G of nanoroughened SERS substrates is still a matter of debate and is based on many assumptions regarding the surface coverage of the adsorbed analytes (Cai et al., 1998; Smythe et al., 2009; Ru et al., 2007). G can be defined as G=

(ISERS /NSERS ) (IRAMAN /NBULK )

(1)

where ISERS is the SERS intensity of the particular peak of NT while IRAMAN represents the un-enhanced Raman intensity measured in liquid form. NSERS is the number of NT molecule on the substrate contributing to the SERS signal and NBULK is the number of molecules contributing to the un-enhanced Raman signal. Usage of covalently anchoring dye such as NT will ease the analysis because it can form a uniform monolayer on the substrate. By incorporating the surface coverage (), roughness factor (R), density () and apparent height (H) of the NT liquid layer contributing to the Raman signal, G can be rewritten as G=

 H   I  SERS R

IRAMAN

(2)

Details of the G calculation and results are given in supplementary material. MFON substrate possesses an enhancement factor in the range of 1011 , while nanogap substrates have higher enhancement factor in the order of 1013 . As given in Eq. (2), R of substrate is an intrinsic parameter that defines the G value. R can vary between different substrates based on the type of nanoroughening. In the case of MFON substrates, whole surface area of PS beads, which is exposed to the metal coating was considered to be contributing to the roughness and hence to the SERS enhancement. R of the MFON substrate was calculated to be 1.57 while it is 0.008 for nanogap substrates. The R value of nanogap substrate might have been slightly underestimated, which is due to the assumption that the dominant enhancement comes from the nanogap. The calculated G value is the maximum possible for the nanogap substrates when roughness contribution from nanogap alone is considered. However, in practical cases, there may be peripheral contribution from the edges of the star island structures to the electromagnetic enhancement and hence the roughness factor value will be higher than that used in the calculation. Taking into account of these, we estimate that the G value for nanogap substrates can be in the range of 1011 , which agree with the best possible reported data (Kneipp et al., 1996; Jackson and Halas, 2004; Ru et al., 2007). Due to the unique nanostructure possessed by nanogap DUV substrates, its R is lower than that of MFON and result in higher G value. We did not calculate the G value for commercial substrate because its signal enhancement is significantly low. G value is a quantitative parameter to gauge the performance of the substrate. But, for practical applications, the measured signal intensity and reproducibility are the important factors deciding the performance of a SERS substrate (Ru et al., 2007). The observed higher reproducibility and SERS enhancement from nanogap DUV substrates are likely due to the precise fabrication methodology, which allows the better control of process parameters to generate uniformly nanoroughened surface. However, the SERS enhancement from MFON substrates was contributed by the nanometric roughness provided by PS beads and also by the metal deposition. Practically, precise controlling of this roughness in nanometric scale is difficult. This affects the uniformity in the surface roughness leading to the comparatively larger variation of SERS intensity. From these results it can be concluded that nanogap bimetallic sub-

Fig. 5. Un-enhanced Raman spectra of 1 M glucose solution using 633 nm excitation, 6.2 mW power at 500 s integration time.

strates with maximum intensity variation of less than 3% can be used to develop a sensitive SERS biosensing platform, especially when dealing with low concentration samples. 3.4. Glucose sensing at physiological conditions A viable glucose sensor must be capable of detecting glucose in the 0–450 mg/dL (0–25 mM) range under physiological conditions with minimum variation. Though the reported SERS glucose sensing using MFON substrate was successful, they faced some problem with intensity variation of signal and this variation was attributed to the poor reproducibility of the substrates (Lyandres et al., 2005). Hence, in this context, direct glucose sensing was carried out using nanogap DUV substrates. Un-enhanced Raman spectrum from a concentrated 1 M glucose solution in PBS was measured and shown in Fig. 5. It indicates that the prominent peaks are 519, 1067, 1131 and 1365 cm−1 in the stokes shifted range of 450–1400 cm−1 . Later, SERS spectra of glucose solutions at various physiologically relevant concentrations were measured using the functionalized substrates. The intensity variation of these four prominent peaks of glucose solutions under SERS mode as a function of concentration is shown in Fig. 6.

Fig. 6. Response showing the intensity variation of prominent SERS peaks of glucose at physiological concentrations.

1992

U.S. Dinish et al. / Biosensors and Bioelectronics 26 (2011) 1987–1992

It can be noted that the intensity value of all these peaks follows a positive linear relation to the concentration of the glucose. Intensity of the 519 cm−1 peak shows better response with a clear linear variation compared to other peaks in these concentration ranges. The experimental conditions used in our case were comparable to the reported glucose sensing methodology adopted using MFON substrates (Lyandres et al., 2005; Stuart et al., 2006). In our experiments, we could consistently achieve stable SERS spectra and hence could provide a straight forward platform for glucose sensing where change in intensity of glucose Raman peaks can be directly monitored as a function of concentration. We achieved the glucose sensing at various concentrations with minimum intensity variation using different functionalized nanogap substrates. Sensitivity of a SERS biosensor is strongly dependent on the substrate performance and hence there is a great demand for highly reproducible substrates. This is especially true when detecting biological samples at low concentration where changes in the intensity of a particular SERS peak have to be monitored. Our experiment proved that SERS glucose sensing was achieved due to the superior SERS performance possessed by nanogap substrates. 4. Conclusions Novel nanogap-structured SERS substrates fabricated on Si wafer using the DUV lithography technique allows better control of fabrication parameters that improves the uniformity in surface roughness to achieve homogenous SERS enhancement. Moreover, this fabrication process offers a relatively cheap and viable option for mass production of SERS substrates. SERS performance of this substrate was compared with a commercial and MFON substrate and found that nanogap substrates possess highest reproducibility with strong SERS enhancement. To the best of our knowledge, the observed high reproducibility with intensity variation of less than 3% for nanogap substrate is the best among the other reported substrates. Efficacy of DUV substrates for sensitive biosensing was demonstrated by glucose sensing under conditions of physiologically relevant concentrations. The high reproducibility and signal enhancement of DUV substrates helped for glucose sensing by analyzing the intensity change of the prominent peaks in glucose SERS spectra as a function of concentration. These results show a significant advancement towards the development of an implantable, real-time continuous SERS-based biosensor. Currently we are investigating on the development of various biosensors based on these nanogap SERS substrates. Acknowledgements The authors would like to thank students, Ms. Low Vway Hau and Mr. Quan Lam Zhung, from Nanyang Technological University, Singapore, for their assistance in the measurement. They would also like to thank Praveen Thoniyot and Shashi Rautela from the Bio Optical Imaging Group, Singapore Bioimaging Consortium and Kho Kiang Wei from National Cancer Centre, Singapore for their help in this project.

Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.bios.2010.08.069. References Addison, C.J., Brolo, A.G., 2006. Langmuir 22 (21), 8696–8702. Albrecht, M.G., Creighton, J.A., 1977. J. Am. Chem. Soc. 99, 5215–5217. Alexander, T.A., Le, D.M., 2007. Appl. Opt. 46 (18), 3878–3890. Bright, R.M., Walter, D.G., Musick, M.D., Jackson, M.A., Allison, K.J., Natan, M.J., 1996. Langmuir 12 (3), 810–817. Brown, R.J.C., Martin, J.T.M., 2008. J. Raman Spectrosc. 39, 1313–1326. Buchel, D., Mihalcea, C., Fukaya, T., Atoda, N., Tominaga, J., Kikukawa, T., Fuji, H., 2001. Appl. Phys. Lett. 79 (5), 620–622. Cai, W.B., Ren, B., Li, X.Q., She, C.X., Liu, F.M., Cai, X.W., Tian, Z.Q., 1998. Surf. Sci. 406, 9–22. Choi, D., Choi, Y., Hong, S., Kang, T., Lee, L.P., 2010. Small 6 (16), 1741–1744. De Jesus, M.A., Giesfeldt, K.S., Oran, J.M., Abu-Hatab, N.A., Lavrik, N.V., Sepaniak, M.J., 2005. Appl. Spectrosc. 59 (12), 1501–1508. Dick, L.A., McFarland, A.D., Haynes, C.L., Van Duyne, R.P., 2002. J. Phys. Chem. B 106 (4), 853–860. Driskell, J.D., Shanmukh, S., Liu, Y., Chaney, S.B., Tang, X.J., Zhao, Y.P., Dluhy, R.A., 2008. J. Phys. Chem. C 112, 895–901. Fan, M., Brolo, A.G., 2008. ChemPhysChem 9, 1899–1907. Fan, M., Brolo, A.G., 2009. Phys. Chem. Chem. Phys. 11, 7381–7389. Fu, C.Y., Koh, Z.Y., Kho, K.W., Dinish, U.S., Praveen, T., Malini, O., 2009. Proc. of SPIE 7397, 739717-1–739717-10. Garrett, N.L., Vukusic, P., Ogrin, F., Sirotkin, E., Winlove, C.P., Moger, J., 2009. J. Biophotonics 2 (3), 157–166. Giorgis, F., Descrovi, E., Chiodoni, A., Froner, E., Scarpa, M., Venturello, A., Geobaldo, F., 2008. Appl. Surf. Sci. 254, 7494–7497. Grabar, K.C., Freeman, R.G., Hommer, M.B., Natan, M.J., 1995. Anal. Chem. 67, 735–743. Jackson, J.B., Halas, N.J., 2004. Proc. Natl. Acad. Sci. U.S.A. 101 (52), 17930–17935. Jacobson, M.L., Rowlen, K.L., 2006. J. Phys. Chem. B 110, 19491–19496. Jeanmaire, D.L., Van Duyne, R.P., 1977. J. Electroanal. Chem. 84, 1–20. Khan, M.A., Hogan, T.P., Shanker, B., 2009. J. Raman Spectrosc. 40, 1539–1545. Kneipp, K., Wang, Y., Kneipp, H., Itzkan, I., Dassari, R.R., Feld, M.S., 1996. Phys. Rev. Lett. 76, 2444–2447. Li, J., Iu, H., Luk, W.C., Wan, J.T.K., Ong, H.C., 2008. Appl. Phys. Lett. 92, 213106–213108. Lin, X.M., Cui, Y., Xu, Y.H., Ren, B., Tian, Z.Q., 2009. Anal. Bioanal. Chem. 394 (7), 1729–1745. Liu, G.L., Lee, L.P., 2005. Appl. Phys. Lett. 87, 074101-1–074101-3. Lyandres, O., Shah, N.C., Yonzon, C.R., Walsh Jr., J.T., Glucksberg, M.R., Van Duyne, R.P., 2005. Anal. Chem. 77, 6134–6139. Mulvaney, S.P., He, L., Natan, M.J., Keating, C.D., 2003. J. Raman Spectrosc. 34 (2), 163–171. Musick, M.D., Keating, C.D., Lyon, L.A., Botsko, S.L., Pena, D.J., Holliway, W.D., McEvoy, T.M., Richardson, J.N., Natan, M.J., 2000. Chem. Mater. 12 (10), 2869–2881. Natan, M.J., 2006. Faraday Discuss. 132, 321–328. Oran, J.M., Hinde, J.R., Nahla, AH., Scott,.T.R., Michael, J.S., 2008. J. Raman Spectrosc. 39, 1811–1820. Perney, N.M.B., Garcia de Abajo, F.J., Baumberg, J.J., Tang, A., Netti, M.C., Charlton, M.D.B., Zoorob, M.E., 2007. Phys. Rev. B 76, 035426-1–035426-5. Qian, L.H., Yan, X.Q., Fujita, T., Inoue, A., Chen, M.W., 2007. Appl. Phys. Lett. 90 (15), 153120–153122. Ru, E.C.L., Blackie, E., Meyer, M., Etchegoin, P.G., 2007. J. Phys. Chem. C 111, 13794–13803. Scholes, F.H., Bendavid, A., Glenn, F.L., Critchley, M., Davis, T.J., Sexton, B.A., 2008. J. Raman Spectrosc. 39, 673–678. Smythe, E.J., Dickey, M.D., Bao, J., Whitesides, G.M., Capasso, F., 2009. Nano Lett. 9 (3), 1132–1138. Stuart, D.A., Yuen, J.M., Shah, N.C., Lyandres, O., Yonzon, C.R., Glucksberg, M.R., Walsh, J.T., Van Duyne, R.P., 2006. Anal. Chem. 78, 7211–7215. Sutherland, W.S., Winefordner, J.D., 1991. J. Raman Spectrosc. 22, 541–549. Tan, R.Z., Agarwal, A., Balasubramanian, N., Kwong, D.L., Jiang, Y., Widjaja, E., Garland, M., 2007. Sens. Actuators A 139, 36–41. Vo Dinh, T., Meier, M., Wokaun, A., 1986. Anal. Chim. Acta 181, 139–148. Wijnhoven, J.E.G.J., Zevenhuizen, S.J.M., Hendriks, M.A., Vanmaekelbergh, D., Kelly, J.J., Vos, W.L., 2000. Adv. Mater. 12 (12), 888–890. Zhou, J., Xu, S., Xu, W., Zhao, B., Ozaki, Y., 2009. J. Raman Spectrosc. 40, 31–37.