1. Introduction
The ocean covers more than 70% of the Earth’s surface, influencing the global environment and climate while supporting human survival and development. Therefore, as a prerequisite for understanding ocean ecology, studying the ocean environment, and exploiting ocean resources, ocean exploration is of great significance to human society [
1,
2]. Modern ocean exploration technology primarily utilizes sound, light, electricity, and magnetic detection platforms to sense and analyze the physical, chemical, and biological parameters of the ocean [
3]. These technologies include space remote sensing, ship-borne observation, ocean buoys, submersibles, and other advanced techniques. Recently, the ocean detection network technology composed of multi-functional small sensors has become the focus of development because of its unique advantages of wide detection range, low cost, and real-time monitoring [
4,
5].
Optical fiber sensors are emerging as an innovative sensing technology, gradually replacing traditional electronic sensors, and are recognized as a crucial component of ocean networking detection technology [
6]. These sensors offer several advantages, such as multifunctionality, miniaturization, resistance to electromagnetic interference, easy waterproofing, and integration into optical fiber communication networks. In ocean research, numerous parameters require detection, among which salinity, temperature, and pressure are indispensable parameters for the study of ocean physics, as they enable researchers to calculate key factors such as ocean density, depth, and dynamics to achieve real-time analysis of ocean currents, tides, and stratification. Consequently, optical fiber sensors predominantly find application in the measurement of seawater salinity, temperature, and pressure for ocean exploration purposes [
7,
8]. Numerous institutions and scholars have conducted research on optical fiber sensors for the measurement of salinity, temperature, and pressure [
9,
10,
11,
12,
13,
14,
15,
16]. For instance, L. Ji et al. [
17] introduced a π phase-shifted FBG sensor enclosed in a metal thin-walled cylinder, enabling the extensive range and high-resolution measurement of seawater pressure. D. Xue et al. [
18] combined microelectromechanical systems (MEMSs) with FPI interferometers to propose a highly sensitive fiber optic sensor that can simultaneously measure seawater salinity and temperature. Similarly, Y. Liu et al. [
19] utilized tapered polarization-maintaining fiber and FBG to design a fiber optic sensor based on the Sagnac loop for the simultaneous measurement of seawater salinity and temperature. Numerous research studies have explored optical fiber sensors for measuring ocean parameters. However, only a limited number of sensors have been able to simultaneously measure salinity, temperature, and pressure because incorporating additional measurement channels significantly complicates the sensor design, fabrication, and signal demodulation processes. Y. Zhao et al. [
20] proposed a three-channel optical fiber sensor based on the SPR effect, which can measure seawater salinity, pressure, and temperature concurrently by employing different sensitive films. Nonetheless, the process of coating multiple sensitizing materials at the end of the fiber is extremely challenging. Moreover, accommodating three SPR dips within a limited spectral range increases the risk of spectral overlap. J. Liu et al. [
21] developed a micro-nano fiber-based three-channel sensor using the MZI technique to simultaneously measure the temperature, salinity, and pressure of seawater. However, the mechanical strength of the micro-nano fiber is notably low, posing a high risk of fracture during actual measurements. Furthermore, utilizing a single measurement mechanism to obtain three-parameter measurements introduces the challenge of sensitivity crosstalk, complicating signal demodulation. A variety of sensing mechanisms have been successfully applied to fiber optic sensors for seawater measurements, the most commonly used mechanisms being SPR, FPI, and FGB. SPR-sensing structures are particularly useful for seawater salinity measurements due to their high refractive index (RI) sensitivity, simple construction, and low cost. Fiber-ended FP sensing structures are frequently utilized for pressure measurements because of their high sensitivity and the absence of additional mechanical sensitizing structures. FBG is an excellent choice for temperature measurement or compensation of sensors. Compared to using a single sensing mechanism for multi-parameter measurements, utilizing different measurement mechanisms can effectively prevent channel crosstalk and spectral overlap and simplify signal demodulation. However, integrating multiple sensing mechanisms on a single optical fiber is challenging and requires innovative structural design and complex preparation processes. To address these challenges, G. An et al. [
22] designed a fiber optic sensor capable of measuring all three parameters by integrating three different sensing mechanisms—namely SPR, FBG, and FPI. A tilted fiber Bragg grating (TFBG) was used to achieve the excitation of the SPR effect. The SPR transmission spectrum was then reflected to the input fiber by a chirped fiber Bragg grating (CFBG), effectively solving the problem of incompatibility between the transmitted SPR-sensing structure and the reflected fiber end FP sensing structure. However, TFBG-SPR typically has a low sensitivity and requires a complex polarization control optical path.
Complex structural design and difficult signal demodulation have been the main problems hindering the development and application of multi-parameter fiber optic sensors, while multi-core fiber optic sensors provide an effective idea to solve these problems. In this article, we propose and demonstrate a probe-type optical fiber sensor that allows for multi-parameter measurement in seawater. The sensor is composed of an MCF and COFs and integrates three measurement channels. The CHSPR based on the side-polished MCF is utilized to measure salinity. The CHFPI relies on an air cavity to measure pressure and temperature. The CHFBG inscribed in the central core of the MCF serves as a temperature compensation component. To enhance the performance of the sensor, we incorporate two distinct microstructures. The first is the SIDC structure, which streamlines the process of filling the PDMS and enhances the repeatability of sensor preparation. The second is the SCR structure, which effectively directs the transmission spectrum of SPR into the symmetric side core of the MCF, solves the compatibility problem between the SPR-sensing structure and fiber end FP sensing structure, and realizes the probe-type structure design. The sensor employs three distinct sensing principles in its three channels, which, combined with SDM technology, enables independent demodulation of the measurement results for each channel. This approach effectively prevents crosstalk between the channels and significantly reduces the complexity of demodulating multi-parameter measurement results. The sensor has demonstrated promising performance in experimental tests, with sensitivity values of 0.36 nm/‰ for salinity (range: 0 to 60‰), −10.62 nm/MPa for pressure (range: 0.1 to 0.5 Mpa), and −0.19 nm/°C for temperature (range: 25 to 85 °C). As a highly integrated and easily demodulated probe-type optical fiber sensor, it has exciting application prospects in the multi-parameter measurement of shallow seawater, tidal estuaries, and saltwater lakes.
2. Operating Principle and Fabrication
A schematic diagram of the probe-type optical fiber sensor for the measurement of seawater salinity, pressure, and temperature is shown in
Figure 1a; the sensor consists of an MCF, a COF with an inner diameter of 50 μm (COF
50), and a COF with an inner diameter of 20 μm (COF
20). The COF
50 and COF
20 have lengths of approximately 50 and 30 μm, respectively. The interior of the COF
50 comprises air, while the interior of the COF
20 is filled with PDMS. These components, along with the MCF central core, create an air cavity to generate FPI. Changes in temperature and pressure cause the PDMS to deform, thus affecting the size of the air cavity. Consequently, the CH
FPI based on the air cavity can be utilized to measure the temperature and pressure of seawater. The MCF is side-polished until one side core is exposed, and the polished surface is then plated with a 50 nm thick gold film to create a CH
SPR to measure seawater salinity. The fiber end is shaped into a 45° cone frustum using micro-grinding technology and coated with a gold reflective film on the grinding surface, resulting in an SCR structure. The SPR transmission spectrum, which carries information about the sample salinity, is reflected twice by the structure, entering the symmetric side core, and subsequently received by the spectrometer. Furthermore, the FBG inscribed in the central core of the MCF is referred to as CH
FBG and is employed to compensate for the temperature crosstalk of the CH
SPR and the CH
FPI. The MCF used by the sensor is a seven-core fiber with a core diameter of 9 μm and a cladding diameter of 125 μm. The side core is located 37.5 μm from the central core. As depicted in
Figure 1b, only the three cores enclosed within the dotted line frame are utilized in this study.
The change in seawater salinity causes a simultaneous change in its RI. Therefore, the CH
SPR, which is highly sensitive to the change in the RI, can be employed to measure seawater salinity. The theoretical model of the side-polished MCF SPR is based on the Kretschmann configuration [
23], which has been extensively discussed in our previous work [
24]. The transmission spectrum of the CH
SPR can be calculated by the following formula:
where
neff represents the effective RI of the side-polished fiber,
λ denotes the wavelength of light transmitted in the core, and
L corresponds to the SPR-sensing region, which measures approximately 10 mm in this study. The light transmitted in the fiber core is a broad-spectrum light, and the SPR effect induces a resonance dip in the transmission spectrum, with the lowest point of the dip referred to as the resonance wavelength. Changes in seawater salinity (Δ
S) cause shifts in the resonant wavelength, so the salinity measurement sensitivity of the CH
SPR can be expressed as follows:
Pressure and temperature measurements are performed using the CH
FPI. According to Fresnel’s law, the sensor consists of three reflector surfaces: M
1, located at the interface between the central core and air; M
2, situated at the interface between air and PDMS; and M
3, positioned at the interface between PDMS and seawater, as depicted in
Figure 1. The reflectance values of M
1 and M
2 remain unaffected by seawater salinity, measuring 3.60% and 2.85%, respectively. However, the reflectance of M
3 is only 0.06% at a seawater salinity of 30‰. Given that M
1 and M
2 have similar reflectance characteristics and M
3 exhibits an extremely low reflectance, the FPI primarily occurs within the air cavity of the COF
50. Accordingly, the resonance wavelength of the FPI can be determined using the following formula:
where
n represents the RI of the medium inside the FP cavity,
L denotes the length of the cavity, and
m corresponds to the interference order.
Further, under the disturbance of seawater pressure (Δ
P) and temperature (Δ
T), the shift of the
m-order resonance wavelength can be expressed as follows:
Elastic polymers such as PDMS and ultraviolet glue are commonly used materials for FPI-based optical fiber pressure sensors. They change the
L or
n of the FP cavity through elastic deformation when subjected to pressure. There are two typical pressure-sensing structures: a cavity filled with polymer [
25,
26] or a cavity with a polymer film placed at the end [
27,
28]. The former structure is easier to prepare as the liquid polymer completely fills the cavity due to the capillary effect. However, it results in a poor contrast of the FPI spectrum due to the polymer and fiber interface having extremely low reflectivity. On the other hand, the FPI spectrum based on the latter structure offers suitable contrast but is challenging to prepare due to the precise operations required for filling the polymer at the picolitre level. To achieve a stable and controllable PDMS filling process and obtain a high spectral contrast, we propose a novel structure called the SIDC structure. This structure is created by fusing the COF
50 and COF
20.
Figure 2 illustrates the SIDC structure filled with PDMS. In this structure, the strength of the capillary effect is inversely proportional to the inside diameter of the capillary. As a result, PDMS only exists within the COF
20 region, while the COF
50 remains an air-filled cavity. The thickness of the filled PDMS film is determined by the length of the COF
20, while the length of the FP air cavity is determined by the length of the COF
50. This allows for precise control over the dimensions and properties of the PDMS and air cavity sections within the sensor structure.
Previous studies have established the viability of the symmetric core reflection structure [
29], but further investigation is necessary to determine the impact of the SIDC structure on reflection efficiency. Based on the actual preparation technique, we initially established the length of the COF
20 for the PDMS filling to be 30 μm. Within our laboratory, we have two types of COFs suitable for creating the FP air cavity, and their inner diameters are 50 and 30 μm (COF
30). Therefore, we first discuss the effect of the capillary inner diameter on the reflection efficiency of the SIDC structure.
Figure 3a,b depict the simulation results of the reflectance of the SIDC structure based on the COF
50 and COF
30, respectively, at an incident wavelength of 635 nm. The two simulated structures differ solely in the inner diameter of the COF, while the other structural parameters and filling media remain constant. It is evident that the presence of PDMS in the COF
20 does not significantly affect the beam propagation, and the COF
50 only slightly impacts the propagation of the divergent beam. The reflection efficiency of the proposed structure is determined by the ratio of the output power of core 1 to the input power of core 2. The reflection efficiencies of the structures depicted in
Figure 3a,b were 61.02% and 61.25%, respectively, with no significant difference between them. In the subsequent PDMS filling test, the COF
30 is filled with PDMS, as illustrated in
Figure 3c. This occurred because the capillary effect of the COF
30 is similar to that of the COF
20. As a result, our final approach is a combination of the COF
50 and COF
20.
Considering the typical length range of the FP cavity, which is usually between 50 and 100 μm, we also investigated the effect of the COF50 length on the reflectance. Simulation results indicate that with COF50 lengths of 50, 75, and 100 μm, the corresponding reflectance values of the structure were 61.02%, 54.76%, and 46.16%, respectively. Hence, it can be concluded that shorter structures exhibit higher reflectivity.
The fabrication process of the optical fiber sensor involves six steps: FBG inscribing, fiber side-polishing, COF fusion and cutting, fiber end micro-grinding, PDMS filling, and gold film deposition. In the first step of FBG inscribing, a phase mask method is employed utilizing a 248 nm KrF excimer laser. The FBG inscribing system includes a CCD for observing the core, which ensures that the FBG is accurately inscribed in the central core of the MCF. This step is crucial for the precise positioning of the FBG within the fiber core [
30,
31]. In the next step, the MCF is side-polished using the wheel side-polishing method, which includes a pair of fiber rotators and a pair of CCDs to adjust the core position and observe the remaining thickness of the fiber to ensure that the side core is just exposed. The microscopic image of the side-polished MCF is shown in
Figure 4a. In the third step, the MCF and COF
50 are fused together with cladding alignment using an optical fiber fusion splicer (NT-600s, Notian, Nanjing, China). After fusion, the COF
50 is cut to a fixed length with a cutting device. This cutting device allows for clear observation of the fusion point between the MCF and COF
50, helping to accurately adjust and control the cutting position to achieve the desired length of the remaining COF
50. Following this, the COF
20 is fused to the front end of the COF
50 using the same method.
Figure 4c illustrates the MCF-COF
50-COF
20 sample that was prepared, providing a visual representation of the completed structure. In the subsequent step, the 45° cone frustum structure is created using the fiber end micro-grinding method, as depicted in
Figure 4b. The prepared sample is secured in a fiber sleeve and brought into contact with the grinding paper. Two motors are utilized to drive both the fiber sleeve and the rotating platform, enabling the grinding process of the fiber end.
Figure 4d shows the MCF-COF
50-COF
20 sample after undergoing the grinding process. The diameter of the remaining fiber end, denoted as
d, is directly influenced by the duration of the grinding process. Furthermore, the cone angle, represented as
α, is determined by the angle formed between the fiber and the rotating platform. In the fifth step, the prepared sample is immersed in a PDMS droplet. Due to the capillary effect, the PDMS gradually fills the capillary structure, flowing along the channel created by the COF
20. The filling process continues until it reaches the interface between the COF
20 and COF
50, as depicted in
Figure 4e. At this point, the PDMS stops advancing further, resulting in a PDMS-filled capillary structure that is confined within the designated region. In the subsequent step, any excess PDMS remaining at the end of the fiber is carefully wiped off. The sample is then cured at a temperature of 75 °C.
Figure 4f displays a microscopic image of the cured sample. Then, a gold reflection film with a thickness of approximately 200 nm is deposited on the fiber end surface, and a gold film with a thickness of approximately 50 nm is deposited onto the side-polishing surface by an ion sputtering apparatus (ETD-900M, Elaborate, Beijing, China). The thickness of the gold films is measured by a three-dimensional surface morphology analyzer (S neox-90, Sensofar, Barcelona, Spain). The measured thicknesses are represented in
Figure 5a for the side-polishing surface and
Figure 5b for the fiber end surface. Finally, the gold film on the top surface of the 45° cone frustum structure is gently wiped with grinding paper to prevent the transmitted light from being reflected to the central core, thus improving the signal-to-noise ratio of the FPI and FBG reflection spectra.
To establish a connection between the proposed probe-type optical fiber sensor and the light source and detector, a fan-in-fan-out device (FIFO) is utilized. This FIFO device, based on SDM technology, facilitates the efficient coupling of light signals between the external components and any core of the MCF.
To measure the reflectance of the SIDC structure, a 635 nm semiconductor laser and a silicon-based optical power detector (PM121D, Thorlabs, Newton, NJ, USA) are used. The measured reflectance, found to be 35%, was lower than the simulation result. Several factors contributed to this discrepancy. Firstly, the symmetry of the structure could affect the reflectance, as any deviation from perfect symmetry may result in reduced reflectivity. Secondly, the roughness of the fiber end grinding surface can also impact the reflectance, as it may introduce the scattering and absorption of light. Lastly, the insertion loss of the FIFO can cause a reduction in the overall signal strength, thus affecting the measured reflectance.