Colloidal Gold-Modified Optical Fiber for Chemical and Biochemical

Reagents and Materials. Multimode plastic-clad silica optical fiber (model F-MBC) was purchased from Newport (Irvine, CA) with core and cladding diame...
0 downloads 0 Views 96KB Size
Anal. Chem. 2003, 75, 16-21

Colloidal Gold-Modified Optical Fiber for Chemical and Biochemical Sensing Shu-Fang Cheng and Lai-Kwan Chau*

Department of Chemistry, National Chung Cheng University, Ming-Hsiung, Chia-Yi, Taiwan 621, R.O.C

A novel class of fiber-optic evanescent-wave sensor was constructed on the basis of modification of the unclad portion of an optical fiber with self-assembled gold colloids. The optical properties and, hence, the attenuated total reflection spectrum of self-assembled gold colloids on the optical fiber changes with different refractive index of the environment near the colloidal gold surface. With sucrose solutions of increasing refractive index, the sensor response decreases linearly. The colloidal gold surface was also functionalized with glycine, succinic acid, or biotin to enhance the selectivity of the sensor. Results show that the sensor response decreases linearly with increasing concentration of each analyte. When the colloidal gold surface was functionalized with biotin, the detection limit of the sensor for streptavidin was 9.8 × 10-11 M. Using this approach, we demonstrate proof-ofconcept of a class of refractive index sensor that is sensitive to the refractive index of the environment near the colloidal gold surface and, hence, is suitable for labelfree detection of molecular or biomolecular binding at the surface of gold colloids. Noble metal nanoparticles characteristically exhibit a strong absorption band that is not present in the spectrum of the bulk metal. This absorption band results when the incident photon frequency is resonant with the collective oscillation of the conduction electrons and is known as the localized surface plasmon resonance (LSPR). The resonance frequency of the LSPR is highly dependent upon the local environment of the nanoparticle.1-5 As such, the optical properties (e.g., absorbance and peak wavelength) of noble metal nanoparticles are sensitive to the refractive index of the surrounding solvent and, additionally, the binding events to those functionalized nanoparticles.5-8 * To whom correspondence should be addressed. Tel.: (886) 5-2720411, ext. 66411. Fax: (886) 5-2721040. E-mail: [email protected]. (1) Underwood, S.; Mulvaney, P. Langmuir 1994, 10, 3427-3430. (2) Jensen, T. R.; Duval, M. L.; Kelly, K. L.; Lazarides, A. A.; Schatz, G. C.; Van Duyne, R. P. J. Phys. Chem. B 1999, 103, 9846-9853. (3) Templeton, A. C.; Pietron, J. J.; Murray, R. W.; Mulvaney, P. J. Phys. Chem. B 2000, 104, 564-570. (4) Okamoto, T.; Yamaguchi, I.; Kobayashi, T. Opt. Lett. 2000, 25, 372-374. (5) Nath, N.; Chilkoti, A. Anal. Chem. 2002, 74, 504-509. (6) Englebienne, P. Analyst 1998, 123, 1599-1603. (7) Eck, D.; Helm, C. A.; Wagner, N. J.; Vaynberg, K. A. Langmuir 2001, 17, 957-960. (8) Malinsky, M. D.; Kelly, K. L.; Schatz, G. C.; Van Duyne, R. P. J. Am. Chem. Soc. 2001, 123, 1471-1482.

16 Analytical Chemistry, Vol. 75, No. 1, January 1, 2003

In the past decade, many optical systems have been developed utilizing the sensitivity of propagating surface plasmon resonance (PSPR) of a thin noble metal layer to the refractive indexes of bulk and thin film dielectrics and to the thickness of thin films.9 Conventional PSPR sensors using the Kretschmann configuration require bulky and expensive optical equipment and, hence, are difficult to be miniaturized. Fiber-optic probe-based PSPR sensors allow for a small sensing element and sample volume, simplified optical design, potential use for disposable fiber-optic sensors, and the capability for remote sensing.10 Nevertheless, Malinsky et al. suggest that LSPR sensors have at least three unique properties in comparison to the PSPR sensors, namely, faster response time, smaller pixel size, and capability of simultaneous LSPR sensing and surface-enhanced Raman scattering.8 Furthermore, fiber-optic sensing based on absorption of the evanescent field is analogous to attenuated total reflection (ATR) spectroscopy.11 As a result, the sensitivity of the method depends on the length of the sensing fiber.12 Thus, the sensitivity of the LSPR sensor will be improved with an optimized length of the optical fiber that has been modified with noble metal colloids. This paper presents a novel class of label-free fiber-optic LSPR sensor which retains many of the desirable features of the PSPR sensors, namely, the sensitivity to the refractive indices of bulk liquids and the ability to interrogate biomolecular interactions without a label. The sensor was constructed on the basis of modification of the unclad portion of an optical fiber with selfassembled Au colloids. The sensor is easy to fabricate and can be constructed by simple optical designs. Moreover, the sensor has the potential capability for on-site, in vivo, and remote sensing, can be easily multiplexed to enable high-throughout screening of biomolecular interactions, and has the potential use for disposable sensors. EXPERIMNETAL SECTION Reagents and Materials. Multimode plastic-clad silica optical fiber (model F-MBC) was purchased from Newport (Irvine, CA) with core and cladding diameters of 400 and 430 µm, respectively. The following chemicals, n-hexadecyltrimethylammonium chloride (CTAB, Fluka), sodium borohydride (Lancaster), 3-(mercaptopropyl)-trimethoxysilane (MPTMS, Acros), N-2-(mercaptopropionyl) glycine (MG, Fluka), mercaptosuccinic acid (MSA, Acros), (9) Homola, J.; Yee, S. S.; Gauglitz, G. Sens. Actuators B 1999, 54, 3-15. (10) Jorgenson, R. C.; Yee, S. S. Sens. Actuators B 1993, 12, 213-220. (11) Harrick, N. J. Internal Reflection Spectroscopy; Harrick Scientific Corp.: New York, 1979. (12) DeGrandpre, M. D.; Burgess, L. W. Anal. Chem. 1988, 60, 2582-2586. 10.1021/ac020310v CCC: $25.00

© 2003 American Chemical Society Published on Web 11/28/2002

cystamine dihydrochloride (Acros), 2-(4-(2-hydroxyethyl)-1-piperazinyl)-ethanesulfonic acid (HEPES, Fluka), 1-ethyl-3-(3-dimethylaminopropyl)-carbodiimine hydrochloride (EDC, Fluka), biotin (Aldrich), and streptavidin (Fluka), were used as received. All aqueous solutions were prepared with water that had been purified by using a YMDI-100 water purification (Yeameei Membrane) with a specific resistance of 18 MΩ cm. Preparation of Au Sols. Aqueous solution of hydrogen tetrachloroaurate (1.78 mL, 25.4 mM), 8.22 mL of chloroform, and 0.4 mL of a 0.02 M ethanol solution of CTAB were mixed and stirred for 10 min to form a 4.52 × 10-4 M hydrogen tetrachloroaurate solution. Freshly prepared NaBH4 ethanol solution (0.8 mL, 0.15 M) was added to the hydrogen tetrachloroaurate solution with vigorous stirring. After the solution was further stirred for 30 min, the ruby-colored organic phase was separated. Absorption spectra of the samples were obtained by using a HP 8453 spectrophotometer. Transmission electron microscopic (TEM) observations of the samples, which had been dispersed and allowed to dry on copper grids, were taken with a Joel TEM 1200 EX instrument. Histograms derived from TEM image analysis showed that the mean diameter of the Au colloids was 8.4 ( 2.8 nm. Preparation of Colloidal Au-Modified Optical Fibers. Unclad portions (∼6 cm) of the optical fibers was cleaned for 30 min in a bath consisting of 3 parts 30% H2O2 to 7 parts concentrated H2SO4. Caution: the above solution reacts violently with organic materials and must be handled with extreme care. The clean unclad portion of the optical fibers were then submerged into vials of 1% solution of MPTMS in toluene. After 8 h, the optical fibers were removed and rinsed with methanol to remove unbound monomers from the surface. After thorough rinsing, the unclad portions of the optical fibers were immersed in Au sol (with absorbance of about 1) for 5 h to form a self-assembled colloid monolayer of gold (CMAu) on the core surfaces. Subsequently, the modified optical fibers were rinsed sequentially with water, methanol, and chloroform. Functionalization of CMAu. CMAu was modified by the formation of a self-assembled monolayer (SAM) of MG, MSA, or cystamine by immersion of CMAu in a solution of MG (0.01 M), MSA (0.01 M), or cystamine dihydrochloride (0.02 M) in deionized water for 2 h at room temperature, respectively. The cystaminemodified CMAu was further functionalized with biotin by immersion of the cystamine-modified CMAu in a methanol/water (1:1) solution of 0.01 M HEPES, 0.01 M EDC, and 5 mM biotin for 30 min at room temperature, rinsed with water and methanol, and air-dried at room temperature. During the functionalization process, both a UV-vis spectrophotometer (model 8453, Hewlett Packard) and a quartz crystal microbalance (P-Sensor 2000, ANT Technology Co. Ltd.) were used to monitor the progress. Instrumentation and Measurements. As shown in Figure 1a, the fiber-optic sensing system consisted of a tungsten halogen light source (model LS-1, Ocean Optics, Inc.), a fiber coupler (Newport), a sensing fiber, a cell, and a fiber-optic spectrometer (model S2000, Ocean Optics, Inc.). Each ATR spectrum was referenced to the background spectrum of an optical fiber (∼6 cm unclad) in air. The temporal response of the sensor was obtained by a fiber-optic system as shown in Figure 1b. This setup consisted of a diode laser (673 nm, model SRT 9225, Micro Laser

Figure 1. Schematic representation of the experimental setup used to make measurements with the CMAuOF. (a) Apparatus used to obtain ATR spectra: A, tungsten halogen light source; B, collimating lens; C, fiber coupler; D, colloidal Au-modified optical fiber; E, liquid cell; F, fiber-optic spectrometer; G, computer. (b) Apparatus used to obtain temporal responses: A, diode laser; B, chopper; C, fiber coupler; D, colloidal Au-modified optical fiber; E, liquid cell; F, photodiode; G, lock-in amplifier; H, computer.

System), an optical chopper (model OC-4000, Photon Technology International), a lock-in amplifier (model 7220, EG&G Instruments), a fiber coupler (Newport), a sensing fiber, a cell, and a photodiode (model 2001, New Focus, Inc.). A peristaltic pump (model 7524-10, Cole-Parmer) was used to generate flowing streams through a flow cell. The flow cell had a volume of ∼1.5 cm3, and a flow rate of 30 mL/min was used throughout the study. For all photometric and spectrophotometric titrations, a constant ionic strength was maintained during the titration. Spectrophotometric titrations of dissolved MG and MSA were taken by a UV-vis spectrophotometer (model 3E, Varian). RESULTS AND DISCUSSION The Au colloids were synthesized by NaBH4 reduction of hydrogen tetrachloroaurate.13 This is a simple, one-step reduction, and yields a high surface coverage of isolated Au colloids.13 The average coverage of the CMAu’s was determined by atomic absorption spectroscopy (AAS) together with TEM. By this approach, Au colloids in solutions were independently measured by TEM prior to assembly. The average size was then used to estimate the number of Au atoms in an average Au colloid. After the formation of a self-assembled CMAu, the chemisorbed Au colloids were dissolved in a 1:3 HNO3/HCl mixture. Consequently, the number of chemisorbed Au colloids and, hence, the surface coverage (relative to a close-packed monolayer) was estimated by measuring the concentration of the dissolved Au using AAS, (13) Tseng, J.-Y.; Lin, M.-H.; Chau, L.-K. Colloids Surf., A 2001, 182, 239-245.

Analytical Chemistry, Vol. 75, No. 1, January 1, 2003

17

Figure 2. Absorbance spectrum of (a) colloidal Au in chloroform and that of (b) a monolayer of Au colloids on a MPTMS-functionalized glass.

assuming that the size distribution of the chemisorbed Au colloids was the same as that measured by TEM prior to self-assembly. The average surface coverage of our CMAu was determined to be 33% of the hypothetical full coverage. The self-assembled CMAu’s on glass were characterized by UV-vis spectrophotometry. When Au colloids get very close, coupling of plasmons of individual particles results in an increased absorbance at wavelengths >600 nm.14,15 As shown in Figure 2, the spectrum of CMAu on glass, when compared with the spectrum of Au colloids in solution, has an increased absorbance at wavelengths >600 nm. This spectral characteristic suggests that the Au colloids self-assembled on glass are close enough to effect the coupling of plasmons of individual particles. Such a conclusion is further supported by the high coverage of our CMAu as compared with those of other CMAu’s.4,14,15 The ability of the colloidal Au-modified optical fiber (CMAuOF) to transduce changes in the surrounding refractive index into ATR spectra was first examined. When the concentration and, hence, the refractive index (RI) of a sucrose solution increased in the range of 1.33-1.41,16 the ATR spectrum of the CMAuOF exhibited a decrease in the reflectance (I/I0, where I ) intensity of the ATR signal from a CMAuOF that is immersed in a sample and I0 ) intensity of the ATR signal from a CMAuOF that is immersed in air) at the peak wavelength and a red shift in the peak wavelength, as shown in Figure 3a. Figure 3b shows a linear fit (correlation coefficient, R ) 0.9993) to the plot of reflectance at 542 nm as a function of RI. This wavelength is chosen in the plot because a maximum sensitivity is found at that wavelength. In addition, since I0 is a constant at a particular wavelength, a plot of intensity (I) versus RI is also linear. On the other hand, a plot of absorabance (-log(I/I0)) at 542 nm as a function of RI is slightly less linear (R ) 0.9924) with a slope of 4.21 absorbance unit (AU)/refractive index unit (RIU). The F-ratio for lack of fit on the plot of absorbance versus (14) Grabar, K. C.; Freeman, R. G.; Hommer, M. B.; Natan, M. J. Anal. Chem. 1995, 67, 735-743. (15) Schmitt, J.; Machtle, P.; Eck, D.; Mohwald, H.; Helm, C. A. Langmuir 1999, 15, 3256-3266. (16) Hwang, T.-C. Experiments in Physical Chemistry; Gaulih Book Co.: Taipei, 1994.

18

Analytical Chemistry, Vol. 75, No. 1, January 1, 2003

RI is 19.6, which is significant at the 99.9% level of confidence and suggests that the model is probably not adequate. On the contrary, the F-ratio for lack of fit on the plot of reflectance versus RI is 1.68, which is not significant at the same level of confidence and indicates that the model is probably adequate at such a level of significance. At other wavelengths, as well, the plots of absorbance versus RI are generally less linear than the plot of reflectance versus RI, indicating the change of absorbance of CMAu with respect to RI is a nonlinear function. Because of the better linearity of the plot of reflectance or intensity versus RI, reflectance or intensity was used as the sensor response in this study. Previous work5 used both LSPR maximum and absorbance changes to measure the surrounding RI of CMAu. However, the shift in the LSPR maximum is only ∼13 nm over the RI range used in this study, and its resolution is limited by the broadness of the reflectance bands. Hence, reflectance changes rather than shifts in LSPR maximum were measured in this study. Furthermore, with a large number of reflections at the CMAuOF, the absorbance changes of the sensor should be significantly enhanced (4.21 AU/RIU) as compared to that of a single CMAu on a glass slide (0.46 AU/RIU).5 Figure 4 compares a spectrum of CMAu on the unclad portion of an optical fiber obtained by ATR with a spectrum of CMAu on glass obtained by the transmission mode. Since bound-mode attenuation in the CMAuOF increases with wavelength, a positive sloping baseline in the plot of absorbance versus wavelength is expected.17 As a result, distortion of the evanescent field spectrum can be observed. In addition, bound mode attenuation in the CMAuOF also increases with RI of the surrounding medium. Hence, a plot of absorbance versus RI should have a positive deviation from linearity.12 Previous study demonstrated that, within a limited RI range, the absorbance of CMAu is linearly dependent on the RI of the surrounding medium.5 Thus, the linear sensor response (either absorbance or reflectance) of our sensor over the range of RI as shown in Figure 3b indicates that the change of the CMAuOF sensor response due to the change of numerical aperture and V-number (normalized frequency) at the different RI surrounding media17 is negligible here. Sensor resolution is usually defined as the minimum change in the parameter (in this case, RI) to be determined that can be resolved by the sensor. Operationally, the sensor resolution in this case is the change in RI corresponding to the intensity or reflectance that yields a signal-to-noise (S/N) ratio of 3, where the noise is taken as the standard deviation of the intensity or reflectance for five repetition measurements of a sample. By this definition, the sensor resolution with the setup as shown in Figure 1a is 2.1 × 10-4 RIU. Such sensor resolution is comparable to those of the fiber-optic PSPR sensor.10 The sensor resolution may be further improved by employing a green laser (∼540 nm) in the setup, as shown in Figure 1b, to enhance the signal-to-noise ratio. Figure 5 shows the time course of the sensor response as a 1.31 M sucrose solution was injected into the sensing flow cell that contained water. It can be seen that the sensor response decreased with time and reached a steady state in less than 20 s. Such response time matches the time response of our flow cell. With the injection of water, the sucrose solution could be eluted (17) DeGrandpre, M. D.; Burgess, L. W. Appl. Spectrosc. 1990, 44, 273-279.

Figure 3. (a) ATR spectra of CMAuOF in aqueous solutions containing increasing concentration of sucrose (from top to bottom). (b) Plot of reflectance at 542 nm versus refractive index of the sucrose solution.

Figure 4. Spectra of (a) a monolayer of Au colloids on a MPTMSfunctionalized glass obtained by transmission mode and that of (b) a self-assembled monolayer of Au colloids on the unclad portion of an optical fiber obtained by ATR. Both spectra were obtained with the sensing layers immersed in water.

Figure 5. Temporal response of the CMAuOF with respect to injection of (a) water, (b) 1.31 M sucrose solution, and (c) water.

quickly and completely, and the sensor response was back to the original level. Such injection and elution processes were repeatable; that is, the response was reversible. The results as shown in Figures 3 and 5 indicate that the CMAuOF is sensitive to the RI of the surrounding bulk medium. Previous work suggests that the electromagnetic fields of the LSPR penetrate into the adjacent solution layer ∼50-60 nm and are

Figure 6. Calibration graph obtained with the MG-functionalized CMAuOF for Ni2+. The response of the sensor was obtained from the setup as shown in Figure 1b. Inset: a plot of 1/(I′ - I) versus reciprocal concentration of Ni2+ (1/CNi) with correlation coefficient of 0.9961.

potentially sensitive to analyte binding events at the colloid/ solution interface.8 Hence, SAMs were used to create functionalized surfaces on CMAu’s. First, a SAM of MG was formed on a CMAu to study whether the MG-functionalized CMAuOF is sensitive to the binding event of a metal ion, Ni2+, which is small in comparison with most biomolecules. MG and Ni2+ were chosen in this study because MG, Ni2+, and the Ni2+-MG complex are transparent at the detection wavelength (673 nm). As shown in Figure 6, the intensity of a MG-functionalized CMAuOF decreases linearly with the increase in Ni2+ concentration (R ) 0.9930). In a control experiment with unfunctionalized CMAuOF, an increase in Ni2+ concentration causes negligible change in the sensor response, indicating that the change in RI due to the increase in Ni2+ concentration can be neglected here. With the limit of detection (LOD) defined as the intensity at a Ni2+ concentration that yields a signal-to noise (S/N) ratio of 3, where the noise is taken as the standard deviation of the intensity for five repetition measurements of a blank, the LOD of the sensor is 2.2 × 10-4 M. To estimate the minimum detectable Ni2+ coverage on the CMAu, the binding capability for Ni2+ by the dissolved MG was studied by a spectrophotometric titration (data not shown). The general reaction for the equilibrium is

M + L ) ML Analytical Chemistry, Vol. 75, No. 1, January 1, 2003

19

where M is the metal ion and L is the ligand. By the Ketelaar approach,18

CM/(A - A°) ) 1/(C - M)KfCL + 1/(C - M)

Table 1. Parameters (LOD, MDC, and Kf) Obtained from Three CMAuOF’s that Were Functionalized with Different Modifiers and the Kf of Their Dissolved Forms modifier

where A is the total absorbance due to the metal ion and the complex concentration, A° is the the absorbance of the initial concentration of the metal ion, C is the molar absorptivity of the complex, M is the molar absorptivity of the metal ion, CM is the initial concentration of the metal ion, CL is the initial concentration of the ligand, and Kf is the conditional formation constant of the reaction. Hence, a linear plot of CM/(A - A°) versus 1/CL would allow the calculation of Kf by dividing the y intercept by the slope. On the basis of the above calculation, the conditional formation constant of the dissolved Ni2+-MG complex in an acetate buffer (0.2 M, pH 5.6) is 1.6 × 102 M-1. However, upon immobilization, a direct translation of reactivity from the solution form to the surface structure is rarely observed.19 Hence, it is necessary to estimate the binding capacity of the surface structure for Ni2+. The general reaction for the equilibrium between a solution phase metal ion or receptor protein, M, and an immobilized ligand, L, is

M + L ) ML where ML is the immobilized metal-ligand or receptor-ligand complex. Considering a Langmuir adsorption model and assuming that the deviation of the equilibrium solution concentration of M from its initial concentration is negligibly small, then

Kf ) x/[M]0(1 - x)

(1)

where x is the fractional coverage (in molar ratio) of M on a full monolayer of L and [M]0 is the initial concentration of the metal ion or receptor protein in solution. If we further assume that the change in reflectance or intensity from a CMAuOF is proportional to the change in x, then

I′ - I ) kx

(2)

where I′ is the intensity from the CMAuOF which is immersed in water, I is the intensity from the CMAuOF that is immersed in a sample solution, and k is the a proportionality constant. Combining eqs 1 and 2 results in the following expression:

1/(I′ - I) ) 1/k + 1/kKf[M]0

(3)

Hence, a linear plot of 1/(I′ - I) versus 1/[M]0 would allow the calculation of Kf by dividing the y intercept by the slope. Using the intercept and slope of the plot as shown in the inset of Figure 6, the conditional formation constant of the immobilized Ni2+-MG complex in an acetate buffer (0.2 M, pH 5.6) is calculated to be 3.6 × 102 M-1. This number has the same order (18) Rose, N. J.; Drago, R. S. J. Am. Chem. Soc. 1959, 81, 6138-6141. (19) Stole, S. M.; Jones, T. P.; Chau, L.-K.; Porter, M. D. In Chemical Sensors and Microinstrumentation; Murray, R. W., Dessy, R. E., Heineman, W. R., Janata, J., Seitz, W. R., Eds.; ACS Symposium Series 403; American Chemical Society: Wangshington, DC, 1989; Chapter 19.

20 Analytical Chemistry, Vol. 75, No. 1, January 1, 2003

MG MSA biotin a

LOD (M)

MDC (%)

Kf, immoblized form (M-1)

Kf, dissolved form (M-1)

2.2 × 10-4 1.1 × 10-4 9.8 × 10-11

7.9 0.65 0.22

3.6 × 102 5.9 × 101 2.2 × 107

1.6 × 102 1.9 × 101 1 × 1015 a

From ref 20.

of magnitude as that of the dissolved form, suggesting that the reactivity of dissolved MG does not change appreciably upon immobilization. Furthermore, the above results suggest that this sensing technique can be used to estimate the binding constant of an immobilized ligand with an adsorbate if both of them lack a spectroscopic signature in the UV-vis region. The fractional coverage of M on L can be estimated from eq 1. When x is small, the fractional coverage of M on L is equal to Kf[M]0. At the detection limit, we can assume x to be small; hence, the minimum detectable coverage (MDC) of Ni2+ on a MGfunctionalized CMAu is ∼7.9%. Such detection capability indicates that the sensor is sensitive to submonolayer coverage. To further investigate the size of the metal ion on the sensitivity of the CMAuOF, a SAM of MSA was formed on a CMAu to study the binding event of MSA with a larger metal ion, Pb2+. By the same approach as above, the data as summarized in Table 1 indicate that the functionalized CMAuOF sensors are sensitive to submonolayer coverage of metal ions and that a larger metal ion appears to increase the sensitivity of the measurement. Judging from the results as shown in Table 1, LSPR can be used to measure changes in RI induced by analyte binding events to functionalized CMAu. Since Ni2+ and Pb2+ are small ions, their contributions to the changes of effective RI should be small as compared to those of biomolecules. Hence, we also tested the sensor for use in a receptor-ligand binding study using the model streptavidin-biotin pair. In this well-characterized model system, streptavidin has a molecular mass of 60 000 and binds biotin at four binding sites with an extremely high binding constant (Ka ) 1015 M-1).20 The binding experiments were performed with a biotinfunctionalized CMAu. Upon introduction of streptavidin, decrease in sensor response occurred quickly. Previous study showed that the binding of streptavidin to surface-immobilized biotin is diffusion-limited.21 After the streptavidin had time to adsorb, buffer containing a large molar excess of biotin (5 mM biotin in HEPES) was injected into the flow cell to initiate competitive desorption. However, desorption of streptavidin during the time frame of measurement is not apparent. The lack of dissociation of the surface-immobilized streptavidin-biotin complex during competitive desorption with biotin was reported previously5,22,23 and is consistent with the extremely slow off-rate constant of the surfaceimmobilized complex.23 (20) Weber, P. C.; Ohlendorf, D. H.; Wendoloski, J. J.; Salemme, F. R. Science 1989, 243, 85-88. (21) Spinke, J.; Liley, M.; Schmitt, F.-J.; Guder, H.-J.; Angermaier, L.; Knoll, W. J. Chem. Phys. 1993, 99, 7012-7019. (22) Perez-Luna, V. H.; O’Brien, M. J.; Opperman, K. A.; Hampton, P. D.; Lopez, G. P.; Klumb, L. A.; Stayton, P. S. J. Am. Chem. Soc. 1999, 121, 64696478.

The concentration-dependent responses of a biotin-functionalized CMAuOF were also investigated in order to determine the sensitivity of this sensor for the study of streptavidin-biotin binding. Over the range of streptavidin concentration from 0.5 to 3.0 nm, the sensor response is linear (R ) 0.9978). The sensor response saturates when streptavidin concentration is higher than ∼30 nM. The LOD of the sensor was determined to be 9.8 × 10-11 M, which is much lower than the above two examples that involve binding of only small metal ions. In comparison to the biotin-functionalized CMAu sensor based on transmission mode,5 the LOD of our biotin-functionalized CMAuOF is lower by ∼2 orders of magnitude. Furthermore, based on the above metalligand binding models and the receptor-ligand binding model, the minimum detectable coverage or the sensitivity of the sensor appears to be related to the size of the adsorbate. A similar finding has also been reported previously3,24,25 and is attributed to the attendant change in particle volume and, hence, the particle polarizability.3 The apparent association constant of the biotin-functionalized CMAu with streptavidin was also estimated by the same approach as above and is equal to 2.2 × 107 M-1. As shown in Table 1, this Kf value is comparable to that of the immobilized form reported previously26 but is much lower than that of the dissolved form. Also shown in Table 1, the minimum detectable streptavidin coverage on the biotin-functionalized MCAu is only ∼0.22%, which is smaller than that of the metal ions. As such, these results indicate that a functionalized CMAuOF can be used to transduce (23) Jung, L. S.; Nelson, K. E.; Campell, C. T.; Stayton, P. S.; Yee, S. S.; PerezLuna, V.; Lopez, G. P. Sens. Actuators 1999, 54, 137-144. (24) Mulvaney, P. Langmuir 1996, 12, 788-800. (25) Ung, T.; Liz-Marzan, L. M.; Mulvaney, P. J. Phys. Chem. B 2001, 105, 34413452. (26) Zhao, S.; Walker, D. S.; Reichert, W. M. Langmuir 1993, 9, 3166-3173.

receptor-ligand binding at a surface into a reflectance or intensity change with a sensitivity that is useful for biosensor applications. CONCLUSION Studies presented in this paper illustrate the feasibility of making evanescent-wave sensors with colloidal Au modified optical fibers for chemical and biochemical sensing. We have demonstrated proof-of-concept of a class of RI sensor that exploits LSPR of self-assembled Au colloids on the unclad portion of an optical fiber for monitoring the RI of solution bulk and for label-free detection of molecular or biomolecular binding at the surface of Au colloids. The advantage of this type of sensor is its simplicity in construction and easy of use. The realization of the sensors is through the measurement of reflectance or intensity from the colloidal Au-modified optical fiber. The sensitivity of the sensors is useful for biosensor applications but has not yet been optimized. We believe a better sensitivity of the sensor can be achieved by optimization of some key parameters, such as the size and density of immobilized Au colloids, surface structure of the functionalized monolayer, core diameter of the optical fiber, length of the unclad fiber for immobilization of Au colloids, detection wavelength, and optical design. ACKNOWLEDGMENT Support of this research by the National Science Council (R.O.C.) and National Chung Cheng University through Grant no. NSC 89-2113-M-194-014 is acknowledged.

Received for review May 8, 2002. Accepted October 17, 2002. AC020310V

Analytical Chemistry, Vol. 75, No. 1, January 1, 2003

21