1. Introduction
Research on the preparation of localized surface plasmon resonance (LSPR) optical fiber sensors utilizing the association of noble metal nanoparticle (NP) LSPR properties and optical fibers has become a popular subject. The localized electromagnetic field around metal surfaces is very sensitive to environmental refractive indexes [
1,
2]. Environmental changes at the interfaces between media and metals can be traced by the changes of metal LSPR characteristics. The coupling of LSPR sensors and optical fibers bring several advantages. First, multiplying the total reflection of light transmission in the optical fiber excites a sample many times, so the detection sensitivity of analytes is expected to be improved. Secondly, the penetration depth of evanescent field on the sidewall of unclad optical fiber is only 100–200 nm [
3]. Therefore, the sensor is only sensitive to the change of the environment nearby the optical fiber surface. In this way, the background noise can be greatly suppressed. Thirdly, the optical fiber sensor prepared based on the metal NP LSPR effect has the benefits of real-time detection, fast response and easy modification[
4].
At present, the preparation of metal NP sensing films for optical fiber sensors mainly makes use of silane as a linking agent to assemble metal NPs [
5–
11]. It was required to make optical fibers hydroxylated first. Then the optical fibers were immersed in the (3-mercaptopropyl) trimethoxysilane (MPTMS) toluene solution for 8 h and a gold colloid for 5 h [
5]. Alternatively, the optical fibers were immersed in the (3-aminopropyl) trimethoxysilane (APTMS) ethanol solution with little acetic acid for 10 min, and then left at 120 °C to complete silylation, and finally assembled metal NPs [
7]. The complex operation and lengthy manipulation limit the application of optical fiber LSPR sensors.
In this paper, we utilize the method of polyelectrolyte (PE) alternate assembly of metal NPs to prepare the optical fiber LSPR sensor [
12]. Our goal was to simplify the operating procedures for the preparation of optical fiber LSPR sensors and to quickly prepare a high-capacity sensing layer. The approach of the PE alternate assembly of metal NPs can achieve this goal. We monitored both the etching process of optical fibers and the assembly process of metal NPs. The sensitivity of the optical fiber LSPR sensor prepared by this method was evaluated and compared with other sensors. In addition, the optical fiber LSPR sensor was utilized for the immunoassay of goat anti-rabbit IgG.
2. Results and Discussion
Scheme 1 shows the setup of the optical fiber LSPR sensor. The optical fiber we used is quartz clad silica fiber (P600-4 UV-VIS, Ocean Optics Co.). The incident light was focused on the fiber core and limited in the core. A 2-cm long the cladding layer of the optical fiber was removed through the etching treatment by HF solution.
Figure 1a presents the serial transmission (I/I
0) curves collected at different etching time. The transmission at 559.0 nm was plotted with the etching time (
Figure 1b). The transmittance dropped gradually. The etching process can be divided into four stages (1–4 in
Figure 1b,c). Stage 1 presents the start of light leakage during the initial etching process. Stage 2 shows the moment where the transmittance had a sudden drop, which indicates that the etching reached the fiber core. The etching around the core continued in Stage 3 and the penetrated light largely increased. The cladding layer was completely removed in Stage 4, and more and more light leaked. We usually stopped the etching process prior to Stage 4. At this moment, the cladding layer was completely removed and most of the core was kept, thus ensuring the optical fiber has the greatest light capacity and mechanical strength. The corresponding transmittance is 0.65.
For immobilizing a LSPR sensing layer, a trilayer PE structure was chosen as a linker for self-assembly of Au NPs. The gold NPs were immobilized by the electrostatic interaction. Since the fiber was functionalized by negative charged hydroxy groups, the first PE layer we used was PDDA, a strong cationic PE. After the PDDA assembly, the fiber surface was positive charged. Then the ionic PE (PSS) and the comparatively weak cationic PE (PAH) were sequentially assembled. The citrate stabilized Au NPs were finally immobilized onto the outmost PAH layer.
The whole processing takes less than four hours. It includes 100 min for the etching of the optical fiber (the etched optical fiber can be repeatedly used), 40 min for the hydroxylation of optical fiber by H
2SO
4/H
2O
2 solution, and 75 min for the assembly of PDDA/PSS/PAH/Au (15 min for PDDA, 10 min for PSS, 10 min for PAH and 40 min for Au NPs). This method is very time-saving compared with the assembly of Au NP
via the silane coupling agent (about 13 h) [
5].
It was reported that the responses of sensitivities of LSPR sensors varied when different sized metal NPs were employed [
13]. In present study, we explored two sizes of Au NPs (
d = 48 and 23 nm). A series of LSPR spectra of Au NPs in their assembly process were recorded (
Figure 2a,d). A red shift of its LSPR band and an increase of peak intensities were observed. This means that the quantity of Au NPs immobilized on the optical fiber sidewall increased and the interparticle spacing between Au NPs reduced [
14].
Figure 2b and e show the trends of the LSPR peak intensities as a function of the Au NPs assembly time. At the beginning, the adsorbing process was fast, because there were a great number of active spots on the PE trilayer for the adsorption of Au NPs. Then the peak intensities increasd slowly. There might be two reasons to explain this phenomenon: one is the reducing of active spots on the trilayer PE matrix, and the other is that the interparticle electrostatic repulsion arose from more and more Au NPs. For the 48-nm Au colloid, when the assembly time is about 150 min, its peak intensity is 0.655 and the peak position is 577.7 nm. For the 23-nm Au colloid, the peak intensity is 0.149 and the peak position is 528.8 nm when the assembly time is about 39 min. It indicates that the size of Au NPs could influence the assembly time. The peak positions of the Au NP assembled films had 50.7 nm (for the 48-nm Au NPs) and 9.8 nm (for the 23-nm Au NPs) red shifts relative to their original Au colloids, respectively (
Figure 2 c,f).
SEM was used to characterize the Au NP films deposited on the trilayer PE substrates.
Figure 3a and c show the SEM images of the Au NP films prepared by different sized Au NPs. The assembled Au NPs are spherical in morphology, with the size of 48 ± 6 nm and 23 ± 2 nm. The distribution of the Au NPs on the assembled film is uniform. No aggregates of Au NPs can be seen. The coverage fraction of colloid A and B in the film is 6.934% ± 1.625 and 5.686% ± 0.946, for 48-nm and 23-nm Au, respectively.
It is well known that the aggregation of Au NPs will broaden and decrease the LSPR band (at ∼520 nm) and generate multipolar resonance bands (within 600–700 nm) in the long wave region [
15]. The aggregation will strongly affect the LSPR sensing performance and experimental reproducibility. Both the LSPR spectra of Au NP sensing films and SEM results indicate that the Au NPs prepared by the present method were well dispersed. Obviously, the multilayered PEs play an important role in the Au NPs assembly. The three consecutively alternate PE layers are composed of a strong cationic PE as the first layer, an ionic PE as the second layer and a comparatively weak cationic PE as the third layer [
12]. This trilayer structure would limit the PE layer interpenetration to a certain degree [
16], which would lead to the relative uniform surface charge distribution on the PE multilayer for Au NP assembly. In present study, both well-dispersed Au NPs on SEM images and single plasmon band in LSPR spectra agree on that the trilayer PE structure is favorable for the preparation of LSPR sensing films.
The response of these Au NP optical fiber LSPR sensors to the bulk phase was inspected.
Figure 4a,c present a series of transmission spectra of the LSPR sensors incubated with different concentrations of sucrose solutions. It can be seen that the red shift of the LSPR band positions took place and the transmission peak intensities reduced gradually with the concentrations of sucrose solutions increasing. The transmittance at 659.0 and 546.3 nm was chosen to plot the LSPR response to the refractive index of sucrose solutions (
Figure 4b and d). The sensitivity and the linear correlation coefficient (R) of the LSPR sensor are 9.46 TU/RIU(R = 0.9860) for 48-nm Au and 5.94 TU/RIU (R = 0.9939) for 23-nm Au. According to the formula of A = −logT (A is the absorbance and T is the transmittance), the sensitivity expressed by the absorbance and R are 13.09 AU/RIU (R = 0.9678) for 48-nm Au and 5.85 AU/RIU (R = 0.9666) for 23-nm Au. Compared with the literatures, the sensitivity was 4.21 AU/RIU for the optical fiber LSPR sensor with the silane coupling agent assembling Au NPs [
4], and 0.46 AU/RIU for the LSPR sensing film made by the silane coupling agent assembling Au NPs on glass substrates [
17], 0.21 AU/RIU for the sensing film prepared by the trilayer PE assembling Au NPs on glass substrates [
12]. Obviously, the sensitivity of present optical fiber LSPR sensor prepared by trilayer PE assembling Au NPs is much higher. This may be caused by the larger Au NP adsorption capacity due to a higher LSPR intensity relative to the silane modified LSPR sensors. The benefit of repeated reflecting light in an optical fiber may be another reason.
The reproducibility of the present optical fiber LSPR sensor was evaluated by the repetitious preparation of Au NP sensing films.
Figure 5 shows the five absorption spectra of a 23-nm Au optical fiber LSPR sensor in water. Hardly any difference of these absorption curves can be seen. The standard deviation of LSPR peak intensity is only 0.02%. This illustrates that the sensor prepared by the assembly method is controllable and reproducible.
Here, the optical fiber LSPR sensor modified by rabbit IgG was employed to detect goat anti-rabbit IgG (insert in
Figure 6a). Different concentrations of goat anti-rabbit IgG (in PBS solution) were pumped into a liquid cell. The 23-nm Au assembled optical fiber LSPR sensor was used here due to its shorter preparing time (
Figure 2e).
Figure 6a presents the dynamic process of goat anti-rabbit IgG adsorption. Both rabbit IgG and goat anti-rabbit IgG cause the peak intensities to remarkably increase. Bovine serum albumin (BSA) was used to block the unoccupied active spots. After pumping different concentrations of goat anti-rabbit IgG, the peak intensities elevated gradually with the goat anti-rabbit IgG concentration increasing.
Figure 6b shows the relation between the peak intensities
versus goat anti-rabbit IgG concentration. A working curve based on an exponential equation was obtained (insert in
Figure 6b).
Figure 6c is the LSPR spectra before (A) and after (B) a 200 μL of goat anti-rabbit IgG (100 ng/mL) were pumped into the liquid cell (the volume of 1.8 mL) and mixed adequately. The final concentration of goat anti-rabbit IgG equals to 11.1 ng/mL. The absorbance of curve B increased as 0.0003 unit. For the concentrations lower than 11.1 ng/mL, the present sensor had no visible response. So the lowest detection concentration of goat anti-rabbit IgG is supposed to be 11.1 ng/mL.
The anti-interference ability of the optical fiber LSPR sensor was inspected here. Goat anti-human IgG was pumped into the liquid cell instead of goat anti-rabbit IgG. The response of 100 μg/mL of mismatching goat anti-human IgG was similar to that of 0.1 μg/mL of matching goat anti-rabbit IgG (only differ from 0.0004 in intensity). Therefore, the fake positive for immunoassay by using the optical fiber LSPR sensor can be ignored except when the interfering concentration is 1,000-times of the analytes. This optical fiber sensor prepared by this method has the potential application as a portable immuno-sensor.