Sensing Characteristics of Fiber Fabry-Perot Sensors Based on Polymer Materials

,

There are different methods to fabricate optical fiber FP sensors [22].For example, the dual-beam FP interferometer has a relatively low cost, and better stability under severe conditions [23], and thus is the most common FP sensor structure.In 2012, Liao et al manufactured a FP RI sensor using The associate editor coordinating the review of this manuscript and approving it for publication was Sukhdev Roy.
high powerlaser etching and welding technique [24], which isn't cost effective.In 2016, Liu used a fusion splice to connect a chemically-etched multimode fiber and a singlemode fiber (SMF) to fabricated a high-sensitivity FP strain sensor [25].In the same year, Xu et al fabricated a high-pressure FP measurement sensor by connecting a short section hollow core fiber with an SMF, where a silicon dioxide diaphragm was coated at the end of HCF section to form an FP cavity [26].However, the sensitivity is relatively low due to the low thermo-optic effect of silica fiber.A FP interferometer using (ultraviolet) UV polymer droplets was prepared in 2015 by Sun et al. [27].In 2018, Oliveira et al used UV glue to fill the FP cavity to make FP sensor [28], and Zhao et al used polyimide film as FP cavity for humidity sensing [29].The use of polymer materials makes the fabrication process simple with additional advantage of low cost and high sensitivity to temperature and strain due to the high thermo-optic and elastic effect of polymer materials.In the case of UV sensing, narrow band gap semiconductors Si and GaAs are widely used to sense UV intensity.However, their UV sensitivity is relatively low [30], and the semiconductor material is normally sensitive to infrared and visible light and thus cause cross-effects [31]- [33].Moreover, the processing semiconductor material is usually complicated and the cost is high.
In this paper, a simple FP cavity manufacturing technique is proposed by using polymer material to fill the gap between the ends of two SMFs.The sensing responses of four different polymer materials (UV88, NOA68, Loctite3525 and BCB) for strain, temperature and RI were tested and analyzed.Furthermore, a new application for UV sensing with the FP sensor was proposed, and experimental results showed that it has a good sensitivity and linear response to UV intensity without cross-sensitivity to visible light.

II. THEORETICAL ANALYSIS
The schematic diagram of the proposed optical fiber FP sensor is shown in Fig. 1.There are two reflection interfaces in the FP structure: one is at the interface between SMF 1 and filled polymer material (R 1 ) and the other is at the interface between the polymer material and SMF 2 (R 2 ).Assume the reflectivity R 1 = R 2 = R and the RI of the F-P cavity (polymer material) is n, the reflected light intensity I R is [22]: where L, λ and I 0 are cavity length, wavelength and input light intensity, respectively.When R is relatively low (R ≤5%), the reflected light intensity I R can be expressed:

III. FABRICATION OF THE FIBER FP STRUCTURES
A section SMF (YOFC, China) was prepared for fabrication of the sensor.Four different polymer materials, UV88 (Relentless, China), Loctite3525 (HenKel, Germany), NOA68 (Norland, USA) and BCB (an optical waveguide resin material) were used to fill in the gap of the optical fiber FP cavity, to investigate the influence of these polymer materials on the performance of the FP sensor.UV88, Loctite3525 and NOA68 are composed of oligomer, active diluent, photosensitizer and auxiliary agent, and the ratio of various components is also different.BCB is composed of photosensitizer and BCB monomer.The physical properties of the four polymer materials are listed in Table 1.The manufacturing set-up of optic-fiber FP structure which consists of two continuous zoom telecentric lens and a three-dimensional translation stage, is shown in Fig. 2(a).UV curing light source (SPCM-0800, China), thermostat (Sunne, China) and electric iron (China) were used to cure the polymer materials.The fabrication process is divided into five steps as illustrated in Fig. 3(a): i. Cut two SMFs with flat end surface by using a traditional fiber cleaver (CT30, Japan), and rough one end of the output SMF to reduce background reflection with a sandpaper.ii.Fix one SMF onto a metal block and another SMF onto

IV. EXPERIMENTS AND ANALYSIS
Firstly, the strain responses of the FP sensors filled with above four polymer materials were tested and compared.The schematic diagram of the experimental setup for strain sensing is shown in Fig. 5.The sensor is fixed on a translation  stage and the axial strain f can be expressed as: In the experiments, the distance between the two fixed points is L 0 = 2 cm and the step change of the strain applied to the sensor is 50 µε.
Figures 6(a), (b) and (c) show the spectral responses vs. increasing and decreasing strain applied to the sensors filled with UV88, NOA68 and BCB.The spectra of the three materials was red-shifted because the cavity length of the sensor was changed after applying strain to the fiber FP sensor, the material RI also changes due to the stress-light effect.As a result, the spectrum experienced a drift.The strain sensitivity is related to Young's modulus and viscosity of the material.The sensor strain f (t) can be expressed as [34]: where f 0 is a time-independent stress component, E, v and t are Young's modulus, viscosity and time.Arnaldo G. Leal-Junior et al. concluded that the stress applied to the material has a negative effect on Young's modulus and a positive effect on the viscosity [34].This conclusion has been verified in our experiments.The positive and reverse travel strain sensing curves in Figs 6(d), (e) and (f) demonstrated good linearity between strain and wavelength shift for these three materials.BCB has the largest Young's modulus and the smallest viscosity, so it has the lowest sensitivity of 0.0032 nm/µε.NOA68 has the second-largest Young's modulus and the second-smallest viscosity, so it has a greater sensitivity of 0.1198 nm/ µε.The sensitivity value of UV88 is 0.2021 nm/ µε.By contrast, FP sensor with UV88 is more sensitive to strain changes than that of NOA68 and BCB, due to its smallest Young's modulus and largest viscosity.It is worth noting that strain sensitivity is not linearly related to Young's modulus and viscosity.The data of Loctite3525 is missing in Figs. 6, because the spectrum for Loctite3525 showed a red-shift when a stress was put on the sensor.When the translation stage rotates 1 µm, which spectrum suddenly changes greatly and continuously red shifts.However, this sensor could not be obtained for stress sensing characteristics due to an unstable spectrum.We think it may be caused by its high sensitivity because its Young's modulus is very small and its viscosity is very large.
In addition, the hysteresis error is used to measure the hysteresis phenomenon, that the positive and reverse travel  sensing curves do not coincide, can described as: where max and y max are the maximum hysteresis difference in the full range and the output full range.The hysteresis errors of strain sensing are calculated as 0.1102, 0.024 and 0.003 nm/µε for UV88, NOA68 and BCB, respectively.
Next, temperature responses of the FP sensors filled with the above four polymer materials were studied.In our experiments, the sensor was heated gradually in a temperature control box, and the spectral responses were recorded by an OSA (Data is recorded every 5 • C for UV88, NOA68 and Loc-tite3525 and every 10 • C for BCB).The spectral responses of the above four FP sensors vs. increasing and decreasing temperature are shown in Figs.7(a), (b), (c) and (d), respectively.As temperature increase, the wavelength shifts to longer wavelength for all the above four sensors.The reason is that the temperature variation changes the RI and dimension of the polymer material, which can be described as: where α and ξ are thermal expansion effect coefficient and thermo-optical effect coefficient, respectively.The maximum temperature measured by the FP sensors filled with UV88, Nor68 and Loctite3525 is about 90 • C. When the temperature is higher than 90 • C, those polymers will be soften and thus not suitable for temperature measurement higher than 90 • C. For BCB, the highest measurable temperature can reach as high as 250 • C. The positive and reverse travel temperature sensing curves in Figs 7(e)-(h) demonstrated a good linearity between temperature and dip wavelength shift for the FP sensor filled with above four different polymer materials.The temperature rising sensitivities of UV88, Nor68, Loctite3525 and BCB are 1.3355, 0.968, 0.938 and 0.2164 nm/ • C, respectively.UV88 has the highest temperature sensitivity.and the temperature cooling sensitivities of UV88, Nor68, Loctite3525 and BCB are 1.3085, 1.3597, 1.044 and 0.2185 nm / • C, respectively.It cannot be ignored that the sensitivity of heating and cooling is not the same, where the worst repeatability material is NOA68.The hysteresis errors of temperature sensing are 0.1536, 0.3885, 0.0599 and 0.0112 nm/ • C for UV88, NOA68, Loc-tite3525 and BCB, respectively.It may be that the temperature has changed the properties of the material.For BCB, although the temperature sensitivity is lower than the above three polymer materials, it can measure temperatures up to 250 • C and has good repeatability and hysteresis.In addition, the price of UV88 is far less than of the other three materials, which shows a high-cost performance.Therefore, according to the requirements of different temperature measurement range and sensitivity, different polymer materials can be selected to prepare optic-fiber FP temperature sensor.
Although the polymer materials have been cured, but the waterproof performance is not very good.Therefore, the refractive index solution infiltrates into the polymer FP cavity, resulting in the response of the FP sensor to surrounding RI.We put the sensor in pure water completely, after waiting for the spectrum to stabilize, using a dropper to drop Dimethyl sulfoxide(It can increase the refractive index of the solution), recording the spectrum when it becomes stable, and repeating the operation several times to obtain multiple sets of spectra.The RI responses of the FP structure filled with these polymer materials are tested in experiment.Figs.8(a), (b) and (c) demonstrated that the spectra with UV88, Nor68 and Loctite3525 have red-shift as increasing of surrounding RI.It is possible the polymer material hasn't completely filled the space between the fibers.The optical path difference increases as RI, which can be described as: where n, L are the refractive index and length of the FP cavity.
The corresponding RI sensitivities of UV88, Nor68 and Loc-tite3525 are 24.678,81.096 and 34.394 nm/RIU, respectively  The results show that, for all the four sensors, the wavelength undergoes red-shift and blue-shift as UV intensity increases and decreases respectively.The reason is the fact that increase of UV intensity will introduce the increase of RI of the polymer materials, resulting in the red-shift of the FP sensor.Good linear fits for their sensors were demonstrated in Figs.10(e), (f), (g) and (h), and the sensitivities are calculated as 0.0067, 0.00573, 0.00834 and 0.0023 nm/(mw/cm 2 ), respectively.Moreover, the good hysteresis for UV sensing was demonstrated by calculating the hysteresis errors [8.56 × 10 −4 , 2.37 × 10 −4 , 3.8 × 10 −4 and 2.39 × 10 −4 nm/(mw/cm 2 ) for UV88, NOA68, Loc-tite3525 and BCB].FP sensor filled with Loctite3525 has both highest sensitivity than that of other three materials.Therefore, it has become the first choice for UV intensity measurement.The resolution of UV light measured by the spectrometer could reach 0.733 mw/cm 2 , and it also was found to be insensitive to visible light, showing low cross sensitivity to visible light and thus is an idea sensor for UV intensity measurement.
Finally, through the integration and comparison of the above experimental results, the temperature, stress, RI and UV light intensity sensing characteristics of the optics-fiber FP structures filled with UV88, Loctite3525, NOA68 and BCB are completely summarized in Table 2.

V. CONCLUSION
A simple and low-cost optical fiber FP sensor structure was proposed by filling polymer materials (UV88, Locatite3525, NOA68 and BCB) between the ends of two SMFs.The sensing performance to strain, temperature, RI and UV light intensity were studied experimentally for the four different polymer materials.The results show that UV88 has the best sensitivity to both strain and temperature with sensitivity of 0.2021 nm/µε and 1.3355 nm/ • C, respectively.The FP sensor filled with BCB has wide temperature measurement range from room temperature to 250 • C, although its temperature sensitivity is as low as 0.2164 nm/ • C. In addition, the FP sensors filled with the four polymer materials can measure the change of UV light intensity, where the sensor filled with Loctite3525 has the highest sensitivity of 0.0087 nm/(mw/cm 2 ) to UV intensity compared to that of the other three polymer materials.The proposed sensors can be used in aerospace, electric power, astronomical observation technologies, fire alarm, missile launch and other scenarios that require temperature, strain or UV measurement [35], [36].

FIGURE 1 .
FIGURE 1.A schematic diagram of the optics-fiber FP structure.

a 3 -
D translation stage, which can precisely align the two SMFs.iii.FP cavity lengths can be controlled by adjusting the translation stage along the axial direction of the fiber.iv.Use a drop of polymer material to fill gap between two SMFs, which is effectively the FP cavity of the fiber sensor.v. Cure the polymers with different conditions: UV88, Locatite3525 and NOA68 are cured for 20 seconds by UV light source, followed 12 h curing at 70 • C; BCB were cured 20 seconds by UV light source firstly, and then heated at 300 • C for 10 seconds.Figures 2(b) and (c) show the microscope photos of fabricated optical fiber FP structures filled with NOA68 and BCB, respectively.Figure 3(b) shows a schematic diagram of the experimental setup, which consists of a broadband source (MAX-RAY PHOTONICS, China), an optical spectrum analyzer (OSA, YOKOGAWA AQ5370D, Japan) and a circulator.The reflective spectra of the FP sensors of four different polymer materials with cavity length of 40 µm (Polymer material will shrink after curing, which will also cause the deviation of the cavity length) were shown in Figs.4(a)-(d).

FIGURE 5 .
FIGURE 5. A schematic diagram of the strain measurement setup.

FIGURE 6 .
FIGURE 6. Reflection spectra response versus stress for the fiber FP structure with (a) UV88, (b) NOA68 and (c) BCB; (d)-(f) Dip wavelength shift as a function of stress.

FIGURE 9 .
FIGURE 9. (a) UV sensing device; (b) Calibration of UV intensity; (c) Device schematic.[shown in Figs.8(e), (f) and (g)].But for BCB which waterproof performance is excellent, we found that the reflection spectrum has no significant shift with increasing of surrounding RI [shown in Figs.8(d) and (h)], the FP structure with BCB has residual sensitivity to RI, which can be used to avoid the problem of cross-sensitivity between temperature and refractive index in temperature sensing applications.The proposed FP sensor is applied for UV sensing.Figure 9(a) shows an experimental photo for UV light intensity sensing.An UV radiometer (linshang-125, China) is used for calibration of UV light intensity [shown in Fig. 9(b)] and Fig. 9(c) shows a schematic diagram of the UV sensing setup.Figures 10(a), (b), (c) and (d) show the spectral responses and repeatability to UV intensity for the FP sensor filled with UV88, NOA68, Loctite3525 and BCB, respectively.The results show that, for all the four sensors, the wavelength undergoes red-shift and blue-shift as UV intensity increases and decreases respectively.The reason is the fact that increase of UV intensity will introduce the increase of RI of the polymer materials, resulting in the red-shift of the FP sensor.Good linear fits for their sensors were demonstrated in Figs.10(e), (f), (g) and (h), and the sensitivities are calculated as 0.0067, 0.00573, 0.00834 and 0.0023 nm/(mw/cm 2 ), respectively.Moreover, the good hysteresis for UV sensing was demonstrated by calculating the hysteresis errors [8.56 × 10 −4 , 2.37 × 10 −4 , 3.8 × 10 −4 and 2.39 × 10 −4 nm/(mw/cm 2 ) for UV88, NOA68, Loc-tite3525 and BCB].FP sensor filled with Loctite3525 has both highest sensitivity than that of other three materials.Therefore, it has become the first choice for UV intensity measurement.The resolution of UV light measured by the spectrometer could reach 0.733 mw/cm 2 , and it also was found to be insensitive to visible light, showing low cross sensitivity to visible light and thus is an idea sensor for UV intensity measurement.Finally, through the integration and comparison of the above experimental results, the temperature, stress, RI and UV light intensity sensing characteristics of the optics-fiber FP structures filled with UV88, Loctite3525, NOA68 and BCB are completely summarized in Table2.

Figure 9 (
a) shows an experimental photo for UV light intensity sensing.An UV radiometer (linshang-125, China) is used for calibration of UV light intensity [shown in Fig. 9(b)] and Fig. 9(c) shows a schematic diagram of the UV sensing setup.Figures 10(a), (b), (c) and (d) show the spectral responses and repeatability to UV intensity for the FP sensor filled with UV88, NOA68, Loctite3525 and BCB, respectively.

FIGURE 10 .
FIGURE 10.Reflection spectra response versus UV light intensity for the fiber FP structure with (a) UV88, (b)NOA68, (c) Loctite3525 and (d)BCB; (e)-(h) Dip wavelength shift as a function of UV light intensity.

TABLE 1 .
Properties values of four polymer materials.

TABLE 2 .
Sensing Summarized sensing characteristics values of the proposed sensor.