1. Introduction

Surface plasmon resonance (SPR) has emerged as a highly powerful surface-sensitive analytical technique for chemical and biochemical sensing [1,2]. Most of the SPR sensors are based on a prism configuration. Although recent progress in portable SPR instrument is encouraging [3], the application of most SPR instruments for remote sensing is often limited by complex configuration and expensive instrument [4–6]. To address the limitations of traditional SPR devices, fiber-based SPR sensors are promising in many new and diversified applications because of their distinct advantages in portability, compactness, and low cost [7].

Conventional solid core fiber SPR sensors are sensitive to the refractive index (RI) of surrounding medium, which need to be lower than that of fiber core [8,9]. Therefore, it is difficult to detect a high RI medium. In order to demonstrate high refractive index measurement based on SPR, theoretical models of fiber SPR sensors have been proposed by holding a liquid medium inside the central holes of the photonics crystal fiber (PCF) or micro-structured optical fiber (MOF). One drawback of this configuration is the difficulty of evenly depositing the thin metal film on the internal face of the hollow cores [10–12]. Gratings and complicated PCF or MOF structures necessary for this configuration of SPR make the fabrication even more difficult [13,14]. Owing to its simple structure and low loss properties in the visible and infrared regions, hollow fiber has been widely studied and is now commonly used as a component in many applications [15–17]. Different from most fiber SPR sensors reported previously, the hollow fiber of several hundred microns diameter can be easily coated with a metal film on the inner wall. Then the hollow fiber SPR sensor holds the liquid medium inside the hollow core and the detection light is transmitted in the medium. To satisfy the condition of total reflection, the liquid medium should have a higher RI than the cladding of the fiber. Therefore, the application of hollow fiber provides a solution for detecting high RI liquids using SPR sensing. Additionally, the air-hole structure also broadens the potential applications of hollow fiber by introducing additional materials into the hollow fiber. Among these materials, LCs are ideal candidates with high RI. LCs are fluid materials with orientational order [18]. LC is of particular interest because its refractive index can be tuned by temperature or by electric field [19,20]. Hence, some LC-based photonic composite structures have been demonstrated by tuning the temperature [21,22], and large RI variations can be achieved by precisely controlling the thermotropic transition of LC [23–25]. In this paper, we present a novel SPR temperature sensor based on hollow fiber structure internally-coated with silver. Due to the strong change in RI of LC, a hollow fiber SPR sensor filled with LC is highly sensitive to temperature and offers a wide dynamic RI detection range. It also holds the advantage of an abrupt wavelength shift caused by the phase-transition of LC. When the temperature changed from 20 °C to 50 °C, we observed that the variation of LC RI is approximately 0.0345. As the temperature increased, the SPR wavelength of the SPR-based temperature sensor showed redshift and the total wavelength shift was up to 160 nm. From the linear fit, the temperature sensitivity was 4.72 nm/°C when the temperature was between 20 °C and 34.5 °C; while for temperature exceeding 34.5°C, the temperature sensitivity dropped to 0.55 nm/°C. This sensor is a good candidate for applications requiring temperature monitoring and alarming due to the abrupt wavelength shift and high temperature sensitivity of this LC SPR sensor.

2. Sensor preparation

In Fig. 1, we present the principle of this liquid crystal filled SPR thermometer. Figure 1(a) shows the configuration of the SPR-based temperature sensor. Here, the silver-coated hollow fiber filled with LC was connected to multi-mode fibers. Figures 1(b) and 1(c) depict the silver-coated hollow fiber and the cross sections of the sensor, respectively. The light beam from a halogen lamp was launched into the silver-coated hollow fiber via the multi-mode fiber. Then, the light beam propagated in the LC core and illuminated the mirror silver layer where surface plasmons wave were excited. The light beam underwent total internal reflection on the inner surface of the hollow fiber while passing through the sensor. Finally, the SPR wavelength was measured with a spectrophotometer (HORIBA, iHR550).

 

Fig. 1 Design of the SPR thermometer. (a) Schematic view of the hollow fiber filled with a liquid crystal medium, and temperature controlled by an external heat source. White light was launched into the hollow fiber and a spectrophotometer collected the response for different temperature values. (b) The structure of the silver coated hollow fiber. (c) The cross section of the hollow fiber filling with bulk homogeneous dielectric or a liquid crystal medium.

Download Full Size | PDF

To evaluate the effect of RI of LC with temperature, the sensor was subjected to external temperature perturbation and the SPR wavelength was tracked at different temperatures. Experimentally, in order to realize the temperature-induced RI change, the temperature was controlled by a heat bath, and a thermocouple was used for temperature calibration. Additionally, surface plasmon waves were excited on the interface between the silver layer and the supporting tube when appropriate light transmits in the liquid core of the hollow fiber. The RI of the liquid core had to be higher than that of the supporting tube (the RI for borosilicate glass is 1.4714) to satisfy the condition of total reflection [26]. Therefore, the RI of the bulk homogeneous dielectric and the LC medium were higher than 1.4714.

The structure of the silver-coated hollow fiber is shown in Fig. 1(b). A hollow fiber with 500 μm inner diameter was selected to implement the SPR-based temperature sensor. A segment of 5 cm length was cleaved and coated with a silver layer on its inner surface by using an improved liquid phase deposition method [27]. Specifically, the silver mirror reaction was an optimum selection of a fast chemical liquid phase deposition method to coat the tube inner surface, in which the thickness and smoothness of silver layer can be precisely controlled by tuning the deposition time, reaction temperature and flow rate of solution to meet the SPR sensing requirement. Before deposition, the inner wall was sensitized by SnCl₂ solution (0.01 g/ml SnCl₂, 5% HCl) for about 20 s, which was beneficial to obtain stronger and smoother silver layer. It also improved the adhesion between the glass surface and the silver particles, which were reduced in the silver mirror reaction. Furthermore, the Sn²⁺ ions that remained on the glass surface shortened the plating time. Generally, silver nitrate and glucose solution are used as the plating and reducing agent, respectively, and in order to increase the reaction speed, high temperature as 80 °C is mandatory in silver mirror reaction. However, in our experiment, the thoroughly mixed silver nitrate and glucose (silver nitrate was 0.02 g/ml and glucose was 0.01 g/ml, then mixed with the volume proportion of 1:1) flowed through the glass capillary in an alkaline solution (0.03 g/ml NaOH). Hence, with the temperature set at 20 °C, the deposition speed also fast, 12 seconds deposition time and 0.05ml/s flow rate were satisfactory for obtaining thickness of silver necessary for the SPR sensing. A shorter plating time avoided large silver particles, which were growing as the reaction time increased. Therefore, a smoother silver surface was obtained in this way.

The LC was 4-cyano-4^'-pentylbiphenyl (5CB), which is a nematic thermotropic LC [28]. The mesophases of 5CB depend on the temperature, and Fig. 2 shows the schematic diagram of different liquid crystal phases. The phase of 5CB tuned with the increasing temperature, and the endothermic transition from nematic to isotropic phase proceeded at nearly 35.5°C. In order to demonstrate the thermally-induced phase transition, we used a thermal platform polarizing microscope (DM4500P, Leica Inc.) to observe the whole phase transition process in Fig. 2(b). We fabricated a drop of LC between two parallel glass slides, and laid it on the thermal platform. These observations showed that 5CB maintained a liquid crystal phase under a broad temperature range of about 22.5 °C to 35.5 °C.

 

Fig. 2 Schematic diagram of the liquid crystal phases changing with temperature. (a) Schematic view of different liquid crystal phases. (b) Liquid crystal phase transition process observed under the polarizing microscope.

Download Full Size | PDF

The bulk homogeneous dielectric media were mixtures of polymethylphenyl siloxane fluid and kerosene with different volume ratios, and the RIs of the mixed liquids were varied from 1.5251 to 1.5734 and verified with an Abbe refractometer. The volume ratios for different RIs are listed in Table 1.

Table 1. Volume Ratios between Polymethylphenyl Siloxane Fluid and Kerosene for Different RIs

View Table

3. Results and discussion

The resonance wavelength of the SPR-based sensor strongly depends on the RI of 5CB liquid core, which is strongly sensitive to temperature in Fig. 3. As temperature increased, the resonant wavelength red shifted by 71.98 nm for a temperature increase from 20 °C to 34.5 °C. This temperature range corresponds to the nematic molecular order. Therefore, the temperature sensitivity was measured at 4.72 nm/°C [Fig. 3(a)]. For temperatures from 36 °C to 50 °C, exceeding the transition temperature, the resonant wavelength moved toward longer wavelength at lower sensitivity as temperature increased. The resonance wavelength maintained a linear relationship with temperature, with temperature sensitivity of 0.55 nm/°C. The different temperature response for the two temperature ranges was due to the thermally-induced phase transition of LC. In addition, when temperature was close to the nematic to isotropic phase transition temperature (35.5 °C), an abrupt wavelength shift of about 46 nm was observed for a temperature change of only 1.5 °C. This high temperature sensitivity coincided with thermally-induced phase transition. We repeated the measurement process for three times by using the same sensor, and good repeatability was shown in Fig. 3(b). This good repeatability of the experiment demonstrated the feasibility of using this SPR-based temperature sensor.

 

Fig. 3 (a) Normalized intensity transmission spectra with liquid crystal core for different temperature values. (b) Correlation of the resonance wavelength for different temperatures. The linear relationship is shown by the two red lines for the temperature ranges of 20~34.5 °C and 36~50°C, respectively. The blue dashed line represents the transition temperature between nematic and isotropic phases.

Download Full Size | PDF

Different bulk homogeneous dielectric media were analysed to calibrate the RI sensitivity of this SPR sensor. The RI of 5CB was then estimated at different temperatures. The bulk homogeneous dielectric media were mixtures of polymethylphenyl siloxane fluid and kerosene, and the RI was varied from 1.5251 to 1.6100 by modifying the components. The ray transmission model was calculated to compare the performance of this sensor with experimental data [Fig. 4(a)]. The theoretical results revealed that the SPR resonance dip shows a blue shift when the RI of the filled medium increased. In these calculations, it was assumed that the inner diameter of hollow fiber was 500 μm, the fiber length was 5 cm, the RI of the hollow fiber was 1.4714, and the thickness of the silver layer was 75 nm.

 

Fig. 4 Normalized intensity transmission spectra of hollow fiber SPR sensor with different RIs of the bulk homogeneous dielectric medium. (a) Theoretical. (b) Experimental. (c) Linear relationship between the resonance wavelength of the hollow fiber SPR sensor (λ) and the refractive index of the bulk solutions (n). (d) The RI for the liquid crystal medium calculated from the linear relationship shown in Fig. 4(c). Experimental data correlate to the RI calculated from the calibration in Fig. 3(b).

Download Full Size | PDF

The spectral response to different RIs was also investigated experimentally in Fig. 4(b). It illustrates that, as the surrounding RI increases, the dip wavelength showed blue shift, which is in agreement with the theoretical model. Usually, the resonance wavelength shifts towards longer wavelength as the RI increases for a solid core fiber SPR sensor [29]; however, we observed the precise opposite in our hollow fiber SPR sensor, the resonance wavelength shifts towards shorter wavelength when the RI increases.

Figure 4(c) presents the dip wavelength shift versus the RI. The experimental dip wavelengths decreased linearly as the RI increased from 1.5251 to 1.5734, which is consistent with its theoretical calculation. The maximum RI of the bulk homogeneous dielectric media we used for calibration was only 1.5734. Therefore, we had to extrapolate the theoretical dip wavelengths of different RIs from 1.5800 to 1.6100 to predict the RI of the LC. We observed that the experimental resonance wavelength results agreed well with the theoretical calculation in the RI range of 1.5251 to 1.5734. From the experimental linear fit, the RI sensitivity is –4.68 × 10³ nm/RIU for the SPR sensor. This was again in good agreement with the expected RI sensitivity from the theoretical model at –4.38 × 10³ nm/RIU. Through the comparative experiments and spectroscopic analysis, we calculated the corresponding RIs of 5CB under different temperature as shown in Fig. 4(d). Ultimately, when the temperature changed from 20 °C to 34.5 °C, the RI of 5CB changing from n_20°C = 1.6098 to n_34.5°C = 1.5945; while when the temperature was close to the transition temperature, the RI of 5CB varied about 0.01 for only 1.5 °C temperature change, and when temperature exceeded the transition temperature, the RI shows slow variation from n_36°C = 1.5846 to n_50°C = 1.5827.

The morphology of the interface between silver and the hollow fiber has been investigated using a scanning electron microscope (SEM), and the surface structure of the silver layer is shown in Fig. 5. The whole cross section is shown in Fig. 5(a), and the measured silver layer thickness from the SEM images is 70 nm. It deviated marginally from the calculated thickness in Fig. 4(c), which is mainly arising from the fluctuation in the silver layer thickness. Figure 5(b) shows the partial enlarged view of the silver layer, we can clearly observe the interface between glass and silver. The nicks should be fragments of the fiber in the cutting process.

 

Fig. 5 SEM pictures of the cross section of the silver layer. (a) Surface structure of the silver and the hollow fiber. (b) Enlarged image of the silver layer.

Download Full Size | PDF

4. Conclusions

In conclusion, we demonstrated a novel hollow fiber SPR temperature sensor based on silver coated and liquid crystal filled structure. The RI of LC decreasing with temperature increasing and the variation can be calibrated by resonance wavelength shift of this SPR-based temperature sensor. The temperature sensitivities were measured at 4.72 nm/°C and 0.55 nm/°C for the temperature ranges of 20 to 34.5 °C and 36 to 50 °C, respectively. An abrupt wavelength shift of more than 46 nm was caused by the transition between nematic and isotropic phase. These results demonstrate the suitability of the design for temperature monitoring. Moreover, the demonstrated sensor with simple structure is low cost (about 1 dollar for each silver-coated hollow fiber) and easily prepared. By filling liquid crystal inside the hollow core, the damageable and easily oxidizable silver layer can be protected which is beneficial to improve the service life. We keep the silver layer between the supporting tube and LC and close the two ends, which can protect the silver layer to isolate the air. This sensor can be further developed for many other physical parameter monitoring including thermal diffusion, electric field and magnetic field etc.