1. Introduction

In the era of rapid development of Internet of Things and big data, sensor technology has reached an unprecedented level of importance and been widely used. Among various sensor technologies, optical fiber sensors have attracted great attentions in the applications of biological, chemical and environmental industries during the past few decades, due to their inherent advantages of compactness, real-time response and natural immunity to external electromagnetic interference [1–3]. Optical fiber based temperature sensors have attracted significant research attention since temperature monitoring is essential in environmental monitoring. In many optical fiber based temperature sensors applications, it is highly desirable to have high sensitivity and broad range to improve the device performance. Earlier efforts in achieving the temperature sensor with super-sensitivity and wide range include plasmonics-based optical fiber sensors [2,4–6], photonic crystal fiber sensor [6–14], fiber Bragg gratings (FBG) [15,16], interferometers [17–30] and so on. Although tremendous effort has been dedicated towards developing fiber sensor with super-sensitivity, broad working temperature ranges, excellent linearity and high stability, challenges still remain in achieving all of the four parameters due to certain technical bottlenecks [6,9,10,27]. The fiber sensors based on MZIs have attracted significant research attention with advantages of super-sensitivity, low cost, simplicity and compactness [8,18–20,22,26,28,30]. In order to solve the above-mentioned thorny problems, many researchers have prepared fiber temperature sensors based on MZI by combining sensitive materials with optical fibers, such as solid [8,22,31], liquid [10,18–20] and liquid crystal [30]. However, the temperature sensitivities of previous reports about MZI or LC refilled into the PCF sensor are less than 40.1 nm/°C [14].

LC is a great candidate for sensors due to its optical anisotropy and super-sensitivity for environmental change, which is stimulus-responsive material under stimuli such as electric field [32,33], magnetic field [34], optical irradiation [35] or environmental temperature [36–40]. Several photonic devices combining LCs and fiber technologies have been proposed [41–48]. The most commonly used method is utilizing LCs refilled into hole of the microstructured fiber and dynamic tuning the refractive index with changing temperature [6–7,17,34]. Another approach is to use the LC coating on the surface of the side-polished fiber [49]. However, previous reports have not been well addressed due to randomly distributed LC molecules, where the lights were scattered and unstable in the sensors without LC orientation treatment [50,51].

In this paper, we demonstrate a temperature sensor with super-sensitivity, broad working temperature range, excellent linearity and high stability by combining LCs, single mode fiber (SMF) and exposed-core microstructure fiber (ECMF). The fiber sensor is fabricated via a MZI based on the structure of SMF₁-ECMF-SMF₂ and a microtube, where the microtube is filled with LC after side polishment and orientation treatment. The LC fiber sensor has super temperature sensitivity of more than −21.6 nm/°C with a good linearity of 0.976 from 22°C to 31°C, −558.5 nm/°C from 31°C to 32°C, −37.3 nm/°C with a good linearity of 0.999 from 32°C to 34°C and −6.7 nm/°C with a good linearity of 0.999 from 34°C to 110°C, respectively. The sensitivity of the LC fiber sensor is increased by up to 155 times compared with the previously reported fiber sensor filled with LC based on the MZI [30], and up to 13.9 times compared to the temperature sensors based on LC-filled photonic crystal fiber without orientation treatment [14]. The fiber sensor based on LCs with super-sensitivity, wide temperature range, excellent linearity and high stability show great potential applications such as environment monitoring, food detection, medicine and chemical industry.

2. Experimental

A setup for sensor consisting of a super-continuum broadband light source (the center wavelength is 1550 nm, Anyang Laser Co. Ltd), an MZI, and an optical spectrum analyzer (OSA, Yokogawa AQ6370C) is used to collect the spectra at the fiber-end, as shown in Fig. 1(a). The MZI is composed of an ECMF (Shenzhen Engineering Laboratory for Optical Fiber Sensors and Networks) sandwiched between two single mode fibers (SMFs, core size of 8.3 µm, Yangtze Soton Laser Co. Ltd). In order to split the beam into two beams to form MZI, the core of the ECMF and SMF1 are spliced with lateral offset of 7 µm along Y direction by fusion splicer (Fujikura FSM-100P+, Japan). A light beam passes through the SMF₁ and then separating at spliced point. One beam is transmitted along the core of the ECMF, and another beam is passed through LCs (5CB, cleaning point is 34 °C, HCCH). Since the refractive index of the LC and the ECMF are different, there is an optical path difference (OPD) when the light passes through the cladding mode (LCs area) and the core mode (ECMF), respectively. The core mode and cladding mode are recouped into the lead-out fiber (SMF₂), resulting in the intermodal interference fringes from the spectral output terminal. Figure 1(b) shows photograph of sensing area observed under the optical microscope (Leica CTR6000). It can be clearly seen that a 292 µm ECMF is spliced between two SMFs by fusion splicer. The cross-sectional shape of the ECMF is shown in Fig. 1(c). The core size of the ECMF is about 8 µm. There are four holes surrounding the fiber core, with the big one exposed to the external environment.

 

Fig. 1. (a) The setup of the temperature sensor. (b) The photograph of sensing area observed under the optical microscope. (c) The cross-sectional shape of the ECMF.

Download Full Size | PDF

In our experiment, the ECMF is placed in a side-polished microtube after orientation treatment and LC filling. Figure 2 depicts the fabrication process of the high-performance fiber sensor based on ECMF and LCs. (i) Firstly, a 300 µm-diameter microtube is side polished by sandpaper with 150 µm polish depth, and then fixed on a glass substrate by epoxy resin. (ii) The microtube is cleaned by ethanol and spin-coated with polyimide (PI). The PI film on the inside wall is about 190 nm thickness. (iii) The microtube is rubbed by the flannel, where the rubbing direction is parallel to the radial direction of the microtube, and then it is filled with LC (5CB, HCCH) by capillary action. The LC molecules are oriented aligned along the axial direction of the microtube. For the MZI fabrication, (iv) the ECMF is spliced with a SMF₁ by fusion splicer with a 7 µm position offset along Y direction relative to SMF₁ after the automatic alignment. (v) The ECMF is cleaved to leave a 292 µm length. (vi) The SMF₂ is spliced manually by tuning the splice position to obtain optimal interference spectrum. (vii) Finally, the cleaned structure of SMF₁-ECMF-SMF₂ is immersed in the LC-filled microtube, and the ends of the microtube are sealed with epoxy resin to prevent the LCs from leaking. The OPD will occur between the cladding mode and core mode due to the different refractive indexes of LCs and fiber core, leading to intermodal interference when the two beams are combined at the SMF₂.

 

Fig. 2. The fabrication process of the high performance sensor. (i). The microtube is polished by sandpaper. (ii). A PI film is coated on the inside wall of microtube. (iii). The microtube is rubbed by the flannel, and it is filled with LC. (iv). The ECMF is spliced with a SMF₁ by fusion splicer. (v). The ECMF is cleaved by fiber cutter to a certain length. (vi). The SMF₁-ECMF was spliced with SMF₂ by fusion splicer. (vii). Assembly of the SMF₁-ECMF-SMF₂ with the microtube.

Download Full Size | PDF

3. Results and discussion

In order to verify the feasibility of the SMF₁-ECMF-SMF₂ structural design, we use Rsoft software to simulate the beam propagation. Figure 3 shows the simulated photon distribution inside the SMF₁-ECMF-SMF₂. The fiber consisted of lead-in SMF₁, ECMF and lead-out SMF₂, with lengths of 200 µm, 292 µm and 200 µm, respectively. The core refractive index is 1.4502 for all the three parts. The position offset of the ECMF was 7 µm relative to the position of the SMF₁ and SMF₂. Figures 3(a) and (b) show the simulated results with LC refractive index of 1.5 and 1.68 at exposed-core area of the ECMF respectively. The optical power density is close to 1 when the light is transmitted from bottom to top through the SMF₁, and then the optical power density drops sharply at the splicing point of SMF₁ and ECMF as shown in Fig. 3. The light beam spreads out in the ECMF and LCs. Because the refractive index of the cladding mode (LCs, n = 1.5∼1.68, 1550 nm) is larger than the refractive index of the core mode (ECMF, n = 1.4502, 1550 nm, 20°C), a large amount of photons are transmitted in the cladding mode, resulting in a greatly reduced optical power density. However, at the splicing point of ECMF and SMF₂, the two beams are recouped into the lead-out fiber SMF₂ to cause an increase in optical power density compared to the ECMF area, rising from 0.04 to 0.2 and 0.09 to 0.2 corresponding to the LC refractive index of 1.5 and 1.68 respectively. According to the simulation results, the optical field can be propagating in the structure of SMF₁-ECMF-SMF₂ with either 1.5 or 1.68 refractive index of LC around ECMF. Therefore, the MZI can be realized based on the structure of SMF₁-ECMF-SMF₂ with uniformly-aligned LC filling.

 

Fig. 3. Simulated electric field propagating in the SMF₁-ECMF-SMF₂ with different refractive indexes of LC around the ECMF. The refractive index of LC is 1.5 (a) and 1.68 (b).

Download Full Size | PDF

It is well known that LC molecules are randomly distributed without external field modulation (such as electric field, magnetic field) or surface anchoring orientation. The light propagation in the randomly distributed LC will be strongly scattered due to its optical anisotropy [51], which is a great disadvantage for applications of MZI. Here, to illustrate the adverse effects of scattering on the MZI, we prepared four samples for comparison. Figures 4(a)-(f) show the optical characteristics of 150 µm-thick LC devices with random orientation in nematic phase, uniform orientation in nematic phase and in isotropic phase, respectively. In Fig. 4(a), the LC (5CB) is filled into an empty cell formed by two indium-tin-oxide (ITO) glass substrates without surface treatment, named as A1. The LC molecules are randomly distributed in the cell, resulting in an uneven distribution of the refractive index within the cell. Light will be scattered as it passes through A1. Figure 4(b) shows the LC device with uniform surface treatment of anti-parallel rubbing using PI coated on the ITO glass, named as A2. The orientation of the LC molecules is along the rubbing direction, and light passing through the A2 is hardly scattered. When the LC goes from anisotropic to isotropic phase above the cleaning temperature, the light could pass through the device without scattering, as shown in Fig. 4(c). The Figs. 4(d)-(f) show photos of the two samples taken under the natural light. The size of samples was 3 cm×3 cm. The transmittance (T) of the A1 under visible light was measured to be 52.6% at 25 °C, and a blurred image of Shenzhen University logo can be seen through A1. However, the transmittance of A2 reaches 86% with the LC molecules uniformly oriented by PI layer at 25 °C, and the logo can be clearly seen through A2 as shown in Fig. 4(e). The transmittance of the A2 increases to 99.8% when the temperature rises to 45 °C above the cleaning temperature, and a very clear logo can be seen at this time as shown in Fig. 4(f). According to the experimental results, it has been demonstrated that the light could pass through the LC device without scattering in the case of the uniformly oriented LC molecules in nematic phase or in isotropic state, while the light is strongly scattered when the LC molecules are disordered.

 

Fig. 4. The schematic diagrams of LC molecules orientations in the cell (a) without orientation treatment (A1) and (b) with uniform orientation treatment (A2) at 25 °C. (c) The LCs is heated to 45 °C in isotropic phase with uniform orientation treatment. (d-f) Photos of A1 without orientation treatment at 25 °C, A2 with uniform orientation treatment at 25 °C, and A2 with orientation treatment at 45 °C, respectively. The transmission spectra of the sensors immersed in a microtube filled with LC (g) without orientation treatment at 25 °C, (h) with uniform orientation treatment at 25 °C, and (i) with uniform orientation treatment at 45 °C, respectively. The insets show the corresponding optical images of the sensors observed under POM.

Download Full Size | PDF

Herein, the sensor based on the structure of SMF₁-ECMF-SMF₂ is immersed in a microtube filled with LCs. The orientation of LC molecules plays a vital role. Figure 4(g) shows the transmission spectrum of the sensor when it is immersed in a microtube without orientation treatment. At 25 °C, the experimental results show that no obvious interference signals are collected at the lead-out SMF₂ due to the randomly distributed LC molecules. The insets of Fig. 4(g) show the device images observed under a polarizing optical microscope (POM, Nikon Co.), where the directions of polarizer and analyzer are represented by P and A, respectively. The microtubes are rotated to 0° or 45° with respect of the optical axis of polarizer. It can be seen from the insets that the device is in a bright state regardless of the angles, indicating that the LC molecules are distributed randomly within the microtube. However, when the sensor is immersed in a microtube filled with uniformly oriented LC confined by PI layer at 25°C, a series of interference fringes are collected at lead-out SMF₂, as shown in Fig. 4(h). We can clearly see that the device shows dark to bright transition when the angle changes from 0° to 45° (inset of Fig. 4(h)). The dark and bright transition indicates that the LC molecules are uniformly oriented in the microtube. When the LC is in isotropic state as the temperature rising to 45 °C, no birefringence is observed in the device under POM, where the device is in a dark state regardless of the angles as shown in the inset of Fig. 4(i). A clear interference fringe could be collected at lead-out SMF₂ (Fig. 4(i)). The contrast of interference fringe in isotropic phase is improved compared to that of Fig. 4(h) due to even less scattering. Therefore, the experimental results show that the LC molecule orientation is a vital role for the sensor based on the structure of SMF₁-ECMF-SMF₂.

According to the principle of the MZI, the sensors have a phase difference (Δφ) between the core mode and the cladding mode of ECMF, Δφ=2πΔn_effL/λ, where Δn_eff is the effective refractive index difference of the two modes, L is the length of the ECMF, and λ is the operating wavelength. When the phase difference satisfies the condition Δφ= (2m + 1)π, where m is the order of the MZI, the attenuation peak wavelength (λ_m) can be expressed as λ_m=2Δn_effL/(2m + 1). According to the formulas of the λ_m, the order of the MZI (m) is (λ_m/Δλ_m) + 1/2, where the Δλ_m is the spacing between the adjacent attenuation peak wavelengths Δλm=λ_m-1-λ_m. The sensor is placed on a hot-stage (INSTEC, MK2000) and an OSA is used to measure the transmission spectra of the sensor at different temperatures. Figure 5(a) plots the measured transmission spectra of the sensor with m = 39, while the corresponding two adjacent attenuation peak wavelengths were λ₃₉=1303.9 nm, λ₃₈=1337.7 nm at 31 °C, respectively. The attenuation peak wavelength blue-shifts from 1495.8 nm to 1303.9 nm (red dashed line) as the temperature increases from 22 °C to 31 °C, and the sensitivity is −21.3 nm/°C. In Fig. 5(b), the attenuation peak wavelength blue-shifts from 1436.2 nm to 1361.7 nm with m = 20 (red dashed line) when the temperature increases from 32 °C to 34 °C, and the corresponding two adjacent attenuation peak wavelengths are λ₂₀=1436.2 nm and λ₁₉=1506.6 nm at 32 °C respectively. Here, a sharply shift occurs between 31 °C and 32 °C, since the refractive index (ne) of LC sharply decreases when the temperature is at the clearing point of 5CB. It can be seen from experiments that the 1303.9 nm attenuation peak corresponds to m = 39 at 31 °C, and 1436.2 nm peak corresponds to m = 20 at 32 °C, respectively. Based on the formula of λ_m=2Δn_effL/(2m + 1), where the Δn_effL value does not change when the temperature hold on 32 °C, λ₃₉=745.4 nm at 32 °C can be roughly calculated. Therefore, the attenuation peak blue-shifts from 1303.9 nm to 745.4 nm while the temperature increases from 31 °C to 32 °C with the sensitivity of 558.5 nm/°C. At 34 °C, the LC is in isotropy state, λ₂₀=1361.7 nm, λ₁₉=1429.8 nm from bottom curve of the Fig. 5(b). According to the MZI interference law, it can be seen that the attenuation peak order (m) at the far right of the bottom curve in the Fig. 5(b) is 16 with λ₁₆=1674.3 nm. When the LC is in isotropic state, the attenuation peak wavelength blue-shifts 507.9 nm when the temperature changes from 34 °C to 110 °C and the sensitivity is calculated to be −6.7 nm/°C for m = 16, as shown in the Fig. 5(c). The summarized relationship of attenuation peak wavelength versus temperature is plotted in Fig. 5(d). In the temperature range of 22 °C to 31 °C where the LC is in nematic phase, the attenuation peak wavelength blue-shifts about 191.9 nm within 9 °C temperature change and the mean sensitivity is calculated to be −21.3 nm/°C with a good linearity of 0.976, shown as the blue squares in Fig. 5(d). When the temperature changes from 31 °C to 32 °C, the temperature sensitivity reaches the highest value of −558.5 nm/°C, which is at least 155 times larger than the reported value of LC fiber sensor based on MZI in Ref. [30], and 13 times larger than the results of LC filled-PCF sensor in Ref. [14] as well. In the temperature range of 32 °C to 34 °C, the mean sensitivity is −37.3 nm/°C with a good linearity of 0.999 when the temperature is just above the clearing point of 5CB, shown as the green spheres in Fig. 5(d). When the temperature is in the range of 34 °C to 110 °C, the LC is in the isotropic phase. In this range, the attenuation peak wavelength linearly blue-shifts about 507.9 nm, and the mean sensitivity is calculated to be −6.7 nm/°C with a good linearity of 0.999, shown as the blue triangles in Fig. 5(d). According to the experimental data and fitting using Origin software, the empirical equations of temperature sensitivity in different temperature regions are obtained: S = −37.3-16/[1 + e^(T-30.99)/0.04] when the 22 °C ≤ T ≤ 34°C; S = −6.7-30.6/[1 + e^(T-33.99)/0.04] when the 34 °C ≤ T ≤ 110 °C, where S is the temperature sensitivity and T is the temperature. The device has been measured several times, where the wavelength shift was always the same and the error bar was less than 1 nm as shown in Fig. 5(d). Since the power of the light source is very weak outside the wavelength range of 1100 nm to 1700nm, the data can be hardly collected above 110 °C or below 22 °C. But theoretically this device can detect broader temperatures range. Here, four different sensitivities have been detected in this sensor, −21.6 nm/°C at nematic state, −558.5 nm/°C and −37.3 nm/°C at the phase transition states, and −6.8 nm/°C at isotropic state, respectively. To understand the reason, we measured the refractive index of 5CB at different temperatures using the methods reported in Ref. [37,38,40]. Since the orientation direction of the LC is parallel to the long direction of the microtube, the propagating light is under the influence of refractive index ne in the sensor. In Fig. 5(e), the blue spheres represent the measured refractive index values of ne as a function of temperature at 1550 nm. The value of n_e changes from 1.6233 to 1.6057 at the rate of 0.0019 /°C when the temperature increases from 22 °C to 31 °C, varies from 1.6057 to 1.5508 at the rate of 0.0549 /°C when the temperature increases from 31 °C to 32 °C, decreases from 1.5508 to 1.5444 at the rate of 0.0032 /°C when the temperature changes from 32 °C to 34 °C, and varies from 1.5444 to 1.5178 at the rate of 0.0004 /°C when the temperature increases from 34 °C to 110 °C. Based on the refractive index measurement, the experimental results of the four sensitivities of Fig. 5(d) can be well explained. The better sensitivity and linearity are mainly due to the large gradient variation of the refractive index with temperature changing, resulting from good oriented LC in the MZI, and the very narrow temperature range of the 5CB at nematic phase state.

 

Fig. 5. The transmission spectra of the sensor measured at different temperatures, with m = 39, m = 20 and m = 16 corresponding to the temperature from 22 °C to 31 °C (a), from 32 °C to 34 °C (b), and from 34 °C to 110 °C (c). (d) The summarized relationship of attenuation peak wavelength versus temperature. (e) The measured refractive index values of n_e as a function of temperature at the measurement wavelength of 1550 nm.

Download Full Size | PDF

Table 1 shows the performance comparison (sensitivity, measurement range and linearity) of the fiber temperature sensors based on MZI or LCs of different configurations. It can be seen that our sensor shows better performance. This sensor via MZI based on ECMF combined with uniformly aligned LCs is characterized by high temperature sensitivity and a broad working temperature range, offering great potential for applications in temperature monitoring and alarming.

Table 1. Comparisons of optical fiber temperature sensors based on MZI or LC by different configurations and their performances

View Table

4. Summary

In summary, the LC fiber temperature sensor with super-sensitivity, wide temperature range, excellent linearity and high stability is fabricated via MZI based on the structure of SMF₁-ECMF-SMF₂ and a LC-filled microtube, where the LC in the microtube is uniformly orientated. The fiber sensor combined with LC has super temperature sensitivity of more than −21.6 nm/°C with a good linearity of 0.976 from 22°C to 31°C, −558.5 nm/°C from 31°C to 32°C, −37.3 nm/°C with a good linearity of 0.999 from 32°C to 34°C and −6.7 nm/°C with a good linearity of 0.999 from 34°C to 110°C, respectively. The sensitivity of the LC fiber sensor is increased by up to 155 times, compared with the previously reported fiber sensor based on the MZI which were filled with LC without orientation treatment. The sensor sensitivity is increased by up to 13.9 times compared to the temperature sensors based on photonic crystal fibers filled with LC. The LC fiber sensor with super-sensitivity, wide temperature range, high stability and excellent linearity provides great potential applications in such as environment monitoring, food detection, medicine, and chemical industry.