Multi-Parameter Sensors Based on Optical Reflective Coupler Probe

We propose and demonst rate multi-parameter sensors based on optical reflective coupler probe (ORCP), for detection of liquid level, temperature, and refractive index (RI). Single-mode fiber ORCP (S-ORCP), few-mode fiber ORCP (F-ORCP) and multi-mode fiber ORCP (M-ORCP) are fabricated by adiabatically tapering and weak fusion technology. The highest liquid level sensitivity is 363.64 nm/mm, achieved by the M-ORCP, with detection limit of as low as 0.06 μm. In addition, multi-parameter measurement of liquid level, temperature and RI can be achieved by the M-ORCP, with temperature and RI sensitivities of −0.327 nm/°C and 7859.27 nm/RIU, respectively. With the merits of high sensitivity, stability, and compact half-coupler probe structure, the proposed ORCP sensors of ∼ cm length can measure the liquid level range of ∼3 mm, which have broad application prospects in the fields of fuel storage and chemical liquid level measurement, especially for chemical reagents, biological tissue liquids and flammable and explosive liquids.


I. INTRODUCTION
L IQUID level sensing is of particular significance and widely used in industrial and applications such as fuel storage, chemical synthesis, and biological analysis. Different liquid level sensors have been reported based on mechanical, electrical, and optical methods [1], [2]. Due to the inherent advantages of miniature size, electromagnetic immunity, resistance to corrosion, and remote sensing capability, the optical fiber liquid level sensors (OFLLSs) are extensively employed.
Among the physical variables which affect the OFLLSs, temperature is the most relevant in this application. Dong et al. proposed an OFLLS based on coreless-D-shape-coreless fiber structure with a sensitivity of 213.8 pm/mm and a temperature crosstalk of −0.128 mm/°C [11]. Kang et al. proposed a novel in-fiber modal interferometer for liquid level sensing based on the cascaded quasi-microfiber and double-cladding fiber with a response of −0.972 dB/mm and a temperature sensitivity of 0.164 nm/°C [6]. The composite schemes with embedded FBGs are utilized to eliminate the inaccuracy of the measured liquid level caused by temperature crosstalk [10], [14]. Camilo et al. proposed an OFLLS based on a FBG embedded into an epoxy resin diaphragm with a liquid level sensitivity of 2.8 pm/mm and a lower temperature crosstalk of 1.04 mm/°C [10]. In addition, the temperature-insensitive characteristics of photonic crystal fiber are utilized to solve the temperature crosstalk with a liquid level sensitivity of 0.585 nm/mm [12]. For most of OFLLSs, especially for application in dangerous chemical liquid with high risk of explosion or combustion, the sensitivity is still insufficient for liquid level measurement within a small measurement range of 3 mm. It is necessary to pursue higher sensitivity and lower temperature crosstalk [15].
The above reports have important applications in detecting relevant physical parameters, but few reports execute the multiparameter detection of liquid level, temperature, and RI in liquid. Recently, high sensitivity detection for temperature, RI, other biomolecular and virus molecular is achieved by interferometers based on optical microfiber couplers (OMCs) [16], [17]. Theoretical analysis in [18], [19] demonstrated that the temperature and RI sensitivity of the OMCs could be infinity near the dispersion turning point. However, as far as we know, there is no report of using OMCs for multi-parameter detection of liquid level and other parameters, especially with high sensitivity. In this paper, we propose and demonstrate multi-parameter sensors based on optical reflective coupler probe (ORCP) for detection of liquid level, temperature, and RI. The OFLLSs based on three kinds of ORCPs are fabricated by adiabatically tapering and weak fusion technology, including single-mode fiber ORCP (S-ORCP), the few-mode fiber ORCP (F-ORCP) and the multi-mode fiber ORCP (M-ORCP). The liquid level sensitivities of S-ORCP and F-ORCP are 43.71 nm/mm and 65.14 nm/mm. The highest liquid level sensitivity is achieved by the M-ORCP with 363.64 nm/mm and the detection limit is as low as 0.06 μm. In addition, the temperature and RI can also be detected by the M-ORCP, with sensitivities of −0.327 nm/°C and 7859.27 nm/RIU. The proposed ORCP sensors of ∼ cm length can measure the liquid level range of ∼3 mm due to the merits of high sensitivity, stability, and compact half-coupler probe structure, which has broad application prospects in the fields of fuel storage and chemical liquid level measurement, especially for chemical reagents, biological tissue liquids and flammable and explosive liquids.

II. THEORETICAL ANALYSIS
The proposed S-ORCP, F-ORCP and M-ORCP were fabricated with two single-mode fibers (SMF, core/cladding diameter = 8.2/125 μm, NA = 0.14), or two few-mode fibers (FMF, core/cladding diameter = 19/125 μm, NA = 0.15), or two multimode fibers (MMF, core/cladding diameter = 50/125 μm, NA = 0.2) to form OMC by weak fusion technology [20]. Then the waist region of the fabricated OMC is cut vertically to form a Fresnel reflection end face. Fig. 1(a) is the schematic diagram of the proposed ORCP. L air and L liquid are the coupling length in air and liquid. d is the waist diameter of the ORCP. According to the supermodes theory, the two parallel tapered fibers are regard as a new waveguide. When light injects into port 1 and reaches the coupling region, the even-and odd-supermodes are excited and interfere with each other. Due to the effective RI difference between the two supermodes, optical interference effect is generated with an optical phase difference (OPD) decided by liquid level. The light is then reflected at the end face of the ORCP owing to the Fresnel reflection. Most of the light is reflected into port 2 and a small part of light is reflected into port 1. Thus, the coupling length and the OPD are doubled. The reflection spectrum of the ORCP can be measured at port 2. Fig. 1(b)-(e), (f)-(i) and (j)-(m) are the simulation results of amplitude distribution along the S-ORCP, F-ORCP and M-ORCP by using beam propagation method (BPM), the yellow lines represent different liquid levels (level 1 to level 4). The surrounding RI and operation wavelength are 1.0 and 1550 nm. d is ∼2.5 μm. The coupling length (L air + L liquid ) is ∼2.29 mm. The interference intensity distribution of the light field in the coupling region takes on a typical 'snake' shape both in air and liquid. As the liquid level increased from 0 mm to 2.29 mm, the immersed part has the enhanced evanescent wave. The spectra of the ORCPs will change accordingly, due to mode interference is varied by different OPD. As for FMF and MMF, higher-order modes have stronger field in the cladding region [21]. From the simulation and comparison, it can be seen that as compared with the S-ORCP, more modal power in the cladding of F-ORCP and M-ORCP are converted into evanescent field. Thus, the F-ORCP and M-ORCP have larger evanescent field leakage, indicating they can achieve higher sensitivity.
According to the supermode theory, the output power R 2 at port 2 are expressed by the following equations [22]: where θ is the input light and the angle between the polarized plane and the horizontal axis. A = (n 1 −n 0 ) 2 /(n 1 +n 0 ) 2 . n 1 and n 0 are the refractive index of silica and air, respectively. The total OPD (ϕ i[i = x,y] ) in the sensor can be defined as: where λ is the operating wavelength, Δn air and Δn liquid are the effective RI differences (n even ef f − n odd ef f ) between the two interference supermodes in air and liquid, respectively. Δn air = (n even ef f − n odd ef f ) air , Δn liquid = (n even ef f − n odd ef f ) liquid , L is the total length of the ORCP (L = L air + L liquid ). When the ϕ is satisfied ϕ = (2N+1)π, N = 0,1,2,3 …, N is an integer, the destructive interference condition is satisfied. The dip wavelength λ N can be calculated as: The free spectral range (FSR) of the probe is expressed as: The response sensitivity of liquid level at resonant dip λ N can be given as: where G liquid and G air are the group effective RI difference between the two interference supermodes in liquid and air, respectively. Fig. 2(a) is the calculated Δn air and Δn liquid in the ambient RI of 1.0000 and 1.3333 as a function of the S-ORCP's diameter. It can be seen the Δn air is larger than the Δn liquid , while the |Δn| (Δn = Δn liquid -Δn air ) is gradually reduced. From (6) and (7), the group effective RI differences of the S-ORCP with different waist diameters in the liquid and air are shown in Fig. 2(b). As for G liquid , there is a turning point from positive to negative while the G air is always negative and is smaller than the G liquid . From (5) and Fig. 2, the liquid level sensitivity is affected by d and L. The change in liquid level can change L liquid and L air to influence the sensitivity. Fig. 2(c) and (d) are the relationship between effective RI difference or group effective RI difference and wavelength. As wavelength is increased from 1250 nm to 1650 nm, the calculated effective RI difference and group effective RI difference in air and liquid are different, which indicates the modal intensity distribution of supermodes will change with wavelength. The optical power in core and cladding of S-ORCP, F-ORCP and M-ORCP at different wavelength are also calculated. The optical power distribution is various at  Fig. 3(a) is the simulated reflection spectra of the S-ORCP near wavelength of 1550 nm with different L liquid (L liquid = 0 μm, 10 μm, 20 μm, 30 μm, 40 μm). d is 2.5 μm and L is fixed at 3 mm. The RI of the liquid is 1.3333. The reflection wavelength red-shifts as the L liquid is increased from 0 μm to 40 μm. Fig. 3(b) shows the relationship between the liquid level sensitivities and the liquid level with different ds (three line-types: 1.5 μm, 2 μm, 2.5 μm), and Ls (three color areas: 1.6 mm, 2 mm, 3 mm). For the fixed liquid levels, d decreases and the sensitivity becomes larger. For the same d, L decreases and the sensitivity becomes larger. The smaller the L and d, the larger the liquid level sensitivity. However, the smaller L limits the measurement range. There is a turnoff between the sensitivity and measurement range. For the S-ORCP with L = 3 mm, the simulated sensitivities with d = 1.5 μm, 2 μm and 2.5 μm are 332.77 nm/mm, 139.57 nm/mm, and 61.30 nm/mm, respectively.

III. SENSOR FABRICATION AND EXPERIMENTAL SETUP
In order to fabricate the ORCPs, it is necessary to obtain the OMCs firstly. Two fibers (SMFs, FMFs or MMFs) are aligned with each other before they are fused together using the flame modification method. During the fabrication process, the hydrogen gas flow, stretching speed and length, which determine the performance of the OMCs are controlled and optimized. Then, the proposed ORCPs are fabricated with a gem knife, by snapping the waist region of the OMCs to form the Fresnel reflection end faces. A broadband source (BBS, with wavelength ranging from 1250 nm to 1650 nm) is connected to the port 1 of the ORCP. The reflection spectra of the ORCP are recorded by an optical spectrum analyzer (OSA, AQ6370D, with bandwidth resolution of 0.02 nm, from 900 nm-1700 nm) real time through port 2. Fig. 4(a) and (b) show the microscopic images of the S-ORCP, L is ∼3 mm. Fig. 4(c) shows the cross-section image of the S-ORCP, d is ∼2.5 μm. Fig. 4(d) is comparison of the measured reflection spectra of the S-ORCP in air and liquid, respectively. When the S-ORCP is immersed in liquid from air gradually, the loss increases from 16.168 dB to 32.347 dB and the spectra shifts to longer wavelength, which is consistent with the results in Fig. 3.    5 shows the schematic diagram of the proposed OFLLS based on the ORCPs. The system consists of a BBS, an ORCP, a beaker, and an OSA. The packaged ORCP is dipping vertically into the beaker and contacting the lower end of the sensor region with the liquid level. For the ORCPs with appropriate d (≥2.5 μm), the buoyancy variation can hardly affect the spectra due to the high Young's modulus of silica fibers and the surface tension between liquid and fiber is small, which can be neglected [15]. The liquid level variation was characterized by repeatedly adding or remove liquid with a pipette of 20 μL that is equivalent to a calculated minimum level change of 3.88 μm (The diameter of the beaker is 80.97 mm). Fig. 6 shows the liquid level response of S-ORCP, F-ORCP and M-ORCP. Fig. 6(a) is the measured reflection spectra of S-ORCP with different liquid levels. It is clear that the reflection wavelengths shifted as the liquid level is increased from 0 mm to 87.3 μm with a step of 9.7 μm. The inset is the wavelength shift measured near 1550 nm. As the liquid level increases, the spectral drift is more obvious at longer wavelength. However, the longer the wavelength, the greater the absorption loss of water. There should be a compromise between the sensitivity and the loss. Fig. 6(b) shows the repeatability and the fitting curves of the wavelength drift versus the liquid level at the wavelength of 1350 nm, 1550 nm, and 1600 nm. When the liquid level is increased, the wavelength red-shifts with sensitivities of 38.81 nm/mm, 43.71 nm/mm, and 45.66 nm/mm, respectively. The linearity (R 2 ) is measured to be about 0.990. When the liquid level decreases, the wavelength blue-shifts accordingly and with sensitivities of 43.95 nm/mm, 43.94 nm/mm, and 44.87 nm/mm. Repeatedly testing the drift of the S-ORCPs at different liquid levels to research the repeatability of the proposed OFLLS. In Fig. 6(b), the inset is the detail measurement error around 1600 nm. The maximum non-repeatability of the wavelength drift is ∼±0.25 nm, due to the evaporation of water molecules and a small amount of water molecules carried by the S-ORCP. Hence, the proposed S-ORCP possesses good reversibility and repeatability in measuring liquid level.

IV. EXPERIMENT AND RESULTS
In order to further improve the liquid level sensitivity of the proposed OFLLS, FMF and MMF are used to fabricate F-ORCP and M-ORCP (d = 2.5 μm and L = 3 mm). Fig. 6(c) and (d) are the liquid level response of the OFLLS based on F-ORCP. When the liquid level is increased, the wavelength red-shifts with sensitivities of 42.6 nm/mm, 65.14 nm/mm, and 68.42 nm/mm, respectively. As the liquid level decreased, the wavelength blue-shifts accordingly and with sensitivities of 52.87 nm/mm, 64.87 nm/mm, and 74.07 nm/mm. The R 2 is ∼0.990 and the maximum non-repeatability of the wavelength drift is ∼±0.21 nm. For the OFLLS based on M-ORCP, the maximum measured sensitivity is 363.64 nm/mm around 1550 nm in Fig. 6(e). The detection limit for the M-ORCP is as low as 0.06 μm. In Fig. 6(f), the measured maximum non-repeatability of the wavelength drift is ∼±0.45 nm. The sensitivities for F-ORCP and M-ORCP are higher than the S-ORCP, which is consistent with the simulation results in Fig. 1.
With further increase of liquid level, the spectral wavelength shift tends to be stable. Fig. 7 shows the wavelength shift of M-ORCP around 1400 nm at each liquid level. The curve of liquid level response exhibits nonlinear in the whole measurement range from 0 to 4.5 mm, including linear (I), transition (II) and saturation (III) zones. Zone 1 has a maximum sensitivity of 363.64 nm/mm. As the liquid level increases to ∼3 mm (II), the wavelength shift decreases as well as sensitivity. In addition, when the liquid level is ∼3 mm, the wavelength shift tends to be stable.
The temperature characteristic of the proposed M-ORCP is also demonstrated. The M-ORCP is completely encapsulated in a capillary filled with water (RI = 1.3333) and is placed in a temperature controller with a resolution of 0.1°C. Fig. 8(a) and (b) are the temperature response of the M-ORCP in liquid. When the ambient temperature was raised from 25°C to 75°C with a temperature interval of 5°C, the wavelength dip blue-shifts and the sensitivities are −0.204 nm/°C, −0.327 nm/°C and −0.346 nm/°C around 1350 nm, 1550 nm, and 1600 nm, respectively. The reflection spectral responses of the liquid level and temperature in liquid are different, indicating the proposed sensor can be used for measurement of liquid and temperature [23]. The RI of liquid is inversely proportional to temperature [24]. The higher the temperature is, the more active the molecules are, the smaller the molecular density is, and the smaller the RI is, causing a blue shift.
The response of the M-ORCP in different RI solutions were tested by tracing the shifting of peaks in reflection spectra. The prepared solutions are five kinds of NaCl solutions with different concentrations. The RI of the solution is measured for 3 times by abbe refractometer and averaged, ranging from 1.3333 to 1.3588. As demonstrated in Fig. 9(a), the peaks red-shift dramatically with increase of the RI. Fig. 9(b) shows the measured wavelength as a function of the different RIs of the NaCl solutions. The RI sensitivity is 7859.27 nm/RIU, with R 2 of 0.996.
The interference dips formed between even-and odd-supermodes exhibits different characteristics. Due to the proposed ORCPs have different spectral response of liquid level and temperature, the sensor can be used for detection of liquid level and temperature. Selecting two interference dips as the observation spots and the multi-parameter measurement of liquid level and temperature can be realized by the sensitive matrix. The relationship between the wavelength drift and liquid level and temperature is expressed as [12]: where Δλ 1 and Δλ 2 represent wavelength drifts around 1550 nm (dip 1) and 1600 nm (dip 2), respectively. K L 1 and K T 1 are the liquid level and temperature sensitivity of dip 1. K L 2 and K T 2 are the liquid level and temperature sensitivity of dip 2. ΔL and ΔT represent the variations of liquid level and temperature, respectively. According to the change principle of the inverse matrix, the matrix (8) can be written as: The relationship among the liquid level and temperature and wavelength drifts can be obtained by substituting the measured experimental sensitivities into the matrix formula: Eq. (10) can be used for measuring the liquid level and temperature. Besides, the proposed sensor can realize multi-parameter detection of liquid level, temperature, and RI. Fig. 10 shows the stability of the proposed sensor based on M-ORCP. Fig. 10(a) is the reflection spectra measured for 1 h. The insets are the details of the wavelength dips around 1532 nm and 1557 nm and Fig. 10(b) shows the measured wavelength fluctuations. The wavelength shifts are all < ±0.01 nm, indicating the proposed sensor has high stability.

V. CONCLUSION
In conclusion, we demonstrate multi-parameter sensors based on ORCP, for detection of liquid level, temperature, and RI. The liquid level sensitivities of S-ORCP, F-ORCP and M-ORCP are 43.71 nm/mm, 65.14 nm/mm, and 363.64 nm/mm around 1550 nm, respectively. The highest liquid level sensitivity is achieved by the M-ORCP and the detection limit is as low as 0.06 μm. The temperature and RI can also be detected by the M-ORCP, with sensitivities of −0.327 nm/°C and 7859.27 nm/RIU, respectively. The proposed ORCP sensors of ∼ cm length can measure the liquid level range of ∼3 mm due to the merits of high sensitivity, stability, and compact half-coupler probe structure, which have broad application prospects in the fields of fuel storage and chemical liquid level measurement, especially for chemical reagents, biological tissue liquids and flammable and explosive liquids.