1. Introduction

Tapered optical fibers have been intensively used to attain a variety of complex photonic functions that range from supercontinuum generation [1], harmonic generation [2], wavelength-selective devices [3], dispersion compensation [4], other photonic devices [5], and sensors [6]. Compared with various kinds of fiber-optic sensors, e.g. long-period grating fibers [7], fiber Bragg gratings [8], side-polished fibers (D-type fibers) [9], and cladding-removed fibers [10], tapered fibers are much simpler devices. Studies show tapered fibers have been used for the measurement of refractive index of the external liquid environment [11], temperature [12], gases [13], chemical agents [14], biomolecules [15], and pathogens [16].

In a light-coupling tapered fiber, the reduction of the core and cladding diameters makes the evanescent fields spread out into the cladding region and eventually beyond the outer boundary [17, 18]. Single mode fiber tapers have been demonstrated to have higher fractional power surrounding the fiber waist than the exposed core of the cladding-removed fibers [19]. The fractional power of evanescent wave is sensitive to environmental changes, and thus the single mode fiber tapers are shown to have better sensitivity. The use of evanescent sensing on tapered fibers has been previously investigated with sensing principles, including measuring changes in output power due to refractive index changes [20], evanescent field absorption [19], evanescent wave fluorescence [21], and surface plasmon resonance (SPR) [22]. However, most of the studies are relied on the configuration of either with bare fiber waist or thin film depositions on the waist to excite SPRs. Controlling and optimizing the film thickness of the whole fiber waist region is comparatively difficult and depends on the experience of operators. Besides, the V-number matching is a unique problem in fluorescence tapered fiber sensors [23].

Utilizing the combination of tapered single mode optical fiber and noble metal nanoparticles as a localized surface resonance (LSPR) sensor for refractive index measurement and biochemical sensing is rare and less published. The exploitation of LSPR of noble metal nanoparticles or nanostructures in the development of plasmonic sensors has attracted considerable research interest for many years. LSPR is the collective oscillation of conduction electrons confined to metal nanoparticles, whose resonance frequency has been shown to be strongly dependent on the particle's size, shape, composition, and the dielectric properties of surrounding medium [24, 25]. If metal nanoparticles are in-resonance with the excitation wavelength, the nanoparticles can then both absorb and scatter light outside their geometric cross-sections. The biomolecular interactions on the nanoparticle surface can change the evanescent field distribution. Therefore, measuring the absorbance variation or spectral shift of LSPR through transmission or reflection methodologies is an unsophisticated means to interrogate the molecular binding events [26, 27]. A recent study has utilized the star-shaped gold nanoparticle modified tapered fiber as a refractive index sensor which relies on the measurement of transmission spectrum change by a monochromator [28]. The theory for the intensity modulation based on the evanescent absorption for the cladding-removed plastic clad silica multimode fiber modified with nanoparticles has been reported [29].

In this paper, a single mode tapered fiber which was chemically modified with gold nanoparticles and further functionalized with molecular recognition probes on the waist is proposed for the measurement of surrounding refractive index changes and biomolecular interactions. Corresponding model experiments are systematically demonstrated. Apart from the spectral investigation, the sensing strategy of the tapered fiber LSPR sensor mainly relies on the interrogation of transmission intensity difference at the endface of fiber by a photodiode. This configuration is unsophisticated. In addition, with the aid of LSPR of gold nanoparticles on the waist region, the refractive index resolution of the tapered fiber LSPR sensor is 3.2$\times$10⁻⁵ RIU that is comparable to other optical sensors [30–32].

2. Experimental

2.1 Preparation of the tapered fiber

The abrupt tapered fiber is manufactured by tapering a standard telecommunication single-mode fiber (SMF-28) through a hydrogen-oxygen flame-brushing technique. To taper the fiber, both sides of the fiber are pulled simultaneously while heating the fiber, as illustrated in Fig. 1(a) . The function of hydrogen-oxygen flame is to heat the fiber to its softening temperature. The pulling stages and the flame are controlled by the computer program. Figure 1(b) exhibits a microscopic image of a tapered optical fiber. Region A, B, and C are the tapered sensing region, transition-stretching region, and the general fiber region, respectively. The average waist diameter and waist length of our tapered fibers is 48 μm and 1.25 mm, respectively.

 

Fig. 1 (a) Schematic of fiber tapering. (b) The microscopic image of a tapered fiber. Region A: the tapered sensing region; Region B: the transition-stretching region; Region C: the general fiber region.

Download Full Size | PDF

2.2 Preparation of the tapered fiber LSPR sensor

All aqueous solutions used in experiments were prepared with ultrapure water with a specific resistance of 18 MΩ$\cdot$cm. Gold nanoparticles employed in our experiments were prepared according to the literature [33]. The tapered fibers were cleaned by a three-step process of ultrasonic bath in acetone for 20 min, soaking in a mixture consisting of 3 volumes of 30% H₂O₂ and 7 volumes of concentrated H₂SO₄ (Piranha solution) for 30 min, followed by a thorough flushing with ultrapure water, and drying in an oven at the temperature of 70 °C. Then, the clean fibers were filled with 1% solution of 3-mercaptopropyltrimethoxysilane (MPTMS, Sigma) in ethanol and allowed to react for 12 hr, leading to the formation of a thiol-terminated self-assembled monolayer (SAM) of MPTMS on the tapered fiber surface. The thiol-functionalized tapered fibers were subsequently rinsed by ethanol to remove unbound monomers from the surface and blow-dried with nitrogen gas. Afterwards, tapered fibers were submerged in the solution of prepared Au nanoparticles for 48 hr to form a metal nanoparticle submonolayer on the tapered fiber surface. Finally, the nanoparticle-immobilized tapered fibers were rinsed with ethanol and dried by nitrogen gas. Figure 2(a) shows the schematic diagram of Au nanoparticle-modified tapered fiber. After the treatment of Au nanoparticles, the surface of the tapered fiber is ruby-colored. Figure 2(b) exhibits a scanning electron microscope (SEM) image of immobilized Au nanoparticles on the fabricated tapered fiber surface. From the histogram analysis of SEM images, the mean diameter of Au nanoparticles is estimated to be 24$\pm$3 nm. The normalized absorption spectrum of Au nanoparticles measured from the tapered fiber LSPR sensor in water through the halogen lamp (Newport, Hg(Xe) Arc Lamp, 200 W) illumination is displayed in Fig. 2(d). The main absorption peak is at 537 nm. This spectral characterization is beneficial to select the wavelength of laser source to excite LSPR of Au nanoparticles for subsequent sensing experiments. Poly(methyl methacrylate) (PMMA) was used to manufacture the flow cell. The tapered fiber LSPR sensor was encapsulated in the ñow cell which has the inlet and outlet facilities for the liquid samples to be kept in contact with the sensing region. The free volume of the microfluidic channel is about 1 μl when the channel is occupied by a sensing fiber.

 

Fig. 2 (a) Schematic of self-assembled submonolayer of Au nanoparticles on the tapered fiber surface by using the chemical modification of MPTMS. (b) The SEM image of immobilized Au nanoparticles on Region A of a tapered fiber. (c) The histogram analysis of SEM image shows the mean diameter of immobilized Au nanoparticles is 24$\pm$3 nm. (d) The absorption spectrum of immobilized Au nanoparticles on the tapered fiber.

Download Full Size | PDF

3. Results and discussion

The ability of tapered fiber LSPR sensor to transduce the change of surrounding refractive index (RI) into the observable spectrum is first examined. For measuring the spectral evolution, the calibrated-track halogen lamp and spectrometer (NEWPORT, IS Series Minispectrometer) were employed, as the configuration shown in Fig. 3(a) . Sucrose solutions with various RIs ranging from 1.333 to 1.403 were successively pumped into the flow cell. The recorded transmission spectra are exhibited in Fig. 3(b). It is seen that the transmittance at the peak wavelength decreases and the peak red-shifts as the surrounding RI alters from low to high. The spectral shift of tapered fiber LSPR sensor in response to RI change is analyzed and the linear calibration result is given in the left-side plot of Fig. 3(c) with a correlation coefficient (r) of 0.9995. The result shows that the refractive index resolution based on interrogating the spectral wavelength shift is 51 nm/RIU. The peak wavelength of 537 nm was chosen for analysis to achieve the maximum sensitivity in the plots of relative intensity I_S/I₀ against RI, where I_S and I₀ respectively represent the intensity of optical signal from the tapered fiber LSPR sensor in contact with a sample and water. Sensor resolution is defined as the smallest change in the bulk refractive index that produces a detectable change in the sensor output. The magnitude of sensor output change that can be detected depends on the level of uncertainty of the sensor outputs-the output noise. The resolution of a sensor, $r_{RI}$, is typically expressed in terms of the standard deviation of noise of the sensor output, ${\sigma }_{SO}$, translated to the refractive index of bulk medium, $r_{RI}={\sigma }_{SO}/S_{RI}$, where $S_{RI}$ is the bulk refractive index sensitivity [31]. Here, the stability of the spectral measurement system is taken as the standard deviation of I_S/I₀ for five repetitive spectral collections on a single tapered fiber LSPR sensor, for which the spectral integration time is 200 ms and the number of scans averaged is 10. Thus, the calculated stability is 1.42$\times$10⁻⁴. The resultant linear fit of the relative intensity against RI plot (the right-side plot in Fig. 3(c)) yields the slope of linear regression of 3.8 RIU⁻¹. According to the above definition, the sensor resolution in Fig. 3(c) is obtained by dividing the stability over the slope of curve and, thus, is estimated to be 3.7$\times$10⁻⁵ RIU. The calculated correlation coefficient of is 0.9989. Since the transmission intensity at a particular wavelength has a linear relationship to the surrounding RI, the RI measurement using the tapered fiber LSPR sensor based on the transmission intensity interrogation can be achieved.

 

Fig. 3 (a) The halogen lamp illumination scheme with a spectrometer. (b) Transmission spectra obtained from a tapered fiber LSPR sensor in the presence of different solutions with RI ranging from 1.333 to 1.403. (c) Calibration curves of the spectral shift (left-side) and the relative intensity change (right-side) in response to different RI solutions.

Download Full Size | PDF

Figure 4(a) exhibits the system layout of the tapered fiber LSPR sensor based on the interrogation of transmission intensity change at the endface of fiber. The excitation light source of a 532-nm single-longitudinal-mode frequency-doubled diode-pumped solid-state laser (Lions DPL532-50-0.7) mounted on a temperature-stabilized copper base with a temperature controller (Newport 350B) was coupled into the fiber to induce the LSPR of immobilized Au nanoparticles. The laser source was modulated by the optical chopper with a chopper frequency of 500 Hz and was demodulated by the lock-in amplifier (SRS530, Stanford Research Systems, Inc.) to improve the signal-to-noise ratio. The transmitted optical signals were monitored in real-time by a photodiode, amplified by the lock-in amplifier, and recorded by a computer via GPIB interface. Figure 4(b) shows the sensorgram of a tapered fiber LSPR sensor based on the interrogation of transmission intensity change in response to the successive sucrose solutions with RI ranging from 1.333 to 1.403. It shows the step-down decreases of the relative intensity with rising RI. The temperature uncertainty of the excitation laser can be kept at 10⁻³ °C. The laser power stability of the entire sensing system based on transmission intensity interrogation was measured and recorded it for more than 4 hr with 1-second sampling time and, thus, was estimated to be 1.35$\times$10⁻⁴. The average signal intensity at each RI step in Fig. 4(b) as a function of the corresponding RI is given in Fig. 4(c). The relative intensity is linearly proportional to the RI of the injected solution (r = 0.9989). Error bars are obtained from five repetitive measurements using five tapered fiber LSPR sensors. The coefficient of variation (CV) is less than 3%, indicating that the sensor fabrication process has good reproducibility. The slope of the calibration curve is calculated to be 4.16 RIU⁻¹. According to the above definition of sensor resolution, the refractive index resolution of the tapered fiber LSPR sensor based on the transmission intensity interrogation scheme is 3.2$\times$10⁻⁵ RIU. This value is very close to the value of 3.7$\times$10⁻⁵ RIU obtained from Fig. 3(c), and is comparable to that of the prism-coupled SPR sensor [32]. The results indicate that the presented tapered fiber LSPR sensor can be applied as a sensitive refractive index sensor. For the consideration of a sensor applied in dynamic measurements during a period of time, measuring the intensity change is a superior method for simplifying the system setup and making it more compact, since the spectrometer is not needed and the excitation light source can be replaced by a LED.

 

Fig. 4 (a) The laser illumination scheme with a photodiode (PD) receiver. (b) The transmitted light intensity interrogation of a tapered fiber LSPR sensor in response to different RI solutions. The RI of each introduced solution is shown in the figure. Calibration curve is given in (c). Error bars are obtained from five repetitive measurements using five tapered fiber LSPR sensors.

Download Full Size | PDF

As a proof-of-concept and to further investigate the feasibility of a tapered fiber LSPR sensor relying on the interrogation of transmission intensity change for analyte sensing applications, we carried out model experiments in buffer solution to kinetically monitor the molecular recognition events. N-(2,4-dinitrophenyl)-6-aminohexanoic acid (DNP, MW = 297.27 Da, Sigma) was chemically functionalized on the tapered sensing region as the molecular recognition probe. To functionalize DNP onto the Au nanoparticle surface, tapered fiber LSPR sensors were immersed in a 0.02 M aqueous solution of cystamine dihydrochloride (Sigma), and allowed to react for 12 hr in order to form a monolayer of cystamine which acts as a linkage between DNP molecules and Au nanoparticles. The cystamine-modified Au nanoparticles were further functionalized with DNP by submerging the fibers in a mixed methanol/water solution (v/v $=$ 1:1) consisting of 0.01 M 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES, pH = 7.4, Sigma), 0.1 M 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide hydrochloride (EDC, Fluka), 0.025 M N-hydroxy-succinimide (NHS, Fluka), and 0.34 mM DNP, and allowed to react for 4 hr. Then, the DNP-functionalized tapered fiber LSPR sensors were rinsed with ultrapure water and drying by nitrogen gas. The transmission spectra of probe-functionalized tapered fiber LSPR sensor were measured and results showed that the transmittance peak decreased after the modification of cystamine onto the surface of Au nanoparticles, and then further decreased after the functionalization of probes. These spectral changes are consistent with the fact that the extinction spectra of metal nanoparticles are sensitive to the RI variation of surrounding environment at which the molecular interaction happens, and further indicate that the probe functionalization process is successful.

Prior to analyte detections, the nonspecific adsorption test was performed to verify the results. Before the injection of sample solution, the DNP-functionalized tapered fiber LSPR sensor was exposed to phosphate buffered saline (PBS, pH = 7.4, Sigma) to check the system stability. Afterward, a solution of 10⁻⁵ g/ml bovine serum albumin (BSA, Sigma) or 5$\times$10⁻⁹ g/ml monoclonal mouse IgE anti-DNP antibody (MW = 220 kDa, Sigma) [34] was injected into a sensor chip. Figure 5(a) shows the results. The sensor response of tapered fiber LSPR sensor in the presence of anti-DNP antibody shows a signal difference from the background signal, while in the presence of BSA it was indistinguishable from the background signal, indicating that there was negligible nonspecific adsorption from high concentration of BSA. Refractive indices of PBS, BSA, and anti-DNP antibody solutions were measured by a refractometer (Kyoto Electronics, RA-620) and all were 1.33395. There is no significant RI difference between the blank and the analyte solutions. Therefore, we can make sure that the measured signal change was caused by the specific binding event between analyte and corresponding recognition molecules on the Au nanoparticle surface.

 

Fig. 5 (a) Nonspecific adsorption test of DNP-functionalized tapered fiber LSPR sensor. (b) The relative transmission intensity change of DNP-functionalized tapered fiber LSPR sensor in response to anti-DNP antibody solutions with different concentration ranging from 5$\times$10⁻⁹ to 1$\times$10⁻⁶ g/ml. Calibration curve is exhibited in (c). The value of each data point is based on five repetitive measurements using five DNP-functionalized tapered fiber LSPR sensors.

Download Full Size | PDF

The sensorgram of tapered fiber LSPR sensor in response to the target molecule, anti-DNP antibody, spiked in buffer is displayed in Fig. 5(b) and 5(c). For each injection, the introduced fluid was kept at a static mode till the equilibrium was reached. Before the serial injection of sample solutions, the sensor was exposed to PBS to check the system stability. Hereafter, the anti-DNP antibody solutions with different concentration ranging from 5$\times$10⁻⁹ to 1$\times$10⁻⁶ g/ml and PBS were successively injected into the flow cell. As shown in Fig. 5(b), the transmitted optical intensity in the steady-state decreases with the increasing concentration of anti-DNP antibody solution. The corresponding calibration curve is exhibited in Fig. 5(c). The calculated correlation coefficient is 0.985. Error bars are obtained from five repetitive measurements using five DNP-functionalized tapered fiber LSPR sensors. The calculated CV is less than 3%. Here, the limit of detection (LOD) is defined as the sensor response (I_S/I₀) that yields a signal-to-noise ratio of 3 [31], where the noise is taken as the standard deviation of I_S/I₀ for 100-second measurement duration in buffer. From the calibration graph as shown in Fig. 5(c), a LOD of 1.06$\times$10⁻⁹ g/ml (4.8 pM) for anti-DNP antibody is estimated. The results reveal that the probe-functionalized tapered fiber LSPR sensor can be used to transduce the molecular binding events at the surface of Au nanoparticles into a detectable optical transmission intensity change with high sensitivity that is useful for biosensing applications.

4. Conclusion

We have systematically characterized the optical properties of transmission spectrum of the tapered fiber LSPR sensor in response to the RI change of surrounding medium. Both the resonance wavelength shift and the transmission optical intensity variation are linearly proportional to the surrounding RI change. The feasibility of probe-functionalized tapered fiber LSPR sensor for real-time analyte detection based on the transmission intensity interrogation scheme has been successfully demonstrated. Results suggest that the probe-functionalized tapered fiber LSPR sensor has potential to be used in biomolecular sensing applications.