An all-fiber Mach–Zehnder interferometer (MZI) for two quasi-continuous points’ temperature sensing in seawater is proposed. Based on the beam propagation theory, transmission spectrum is designed to present two sets of clear and independent interferences. Following this design, MZI is fabricated and two points’ temperature sensing in seawater are demonstrated with sensitivities of 42.69pm/°C and 39.17pm/°C, respectively. By further optimization, sensitivity of 80.91pm/°C can be obtained, which is 3-10 times higher than fiber Bragg gratings and microfiber resonator, and higher than almost all similar MZI based temperature sensors. In addition, factors affecting sensitivities are also discussed and verified in experiment. The two points’ temperature sensing demonstrated here show advantages of simple and compact construction, robust structure, easy fabrication, high sensitivity, immunity to salinity and tunable distance of 1-20 centimeters between two points, which may provide references for macroscopic oceanic research and other sensing applications based on MZIs.
© 2018 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
Temperature of seawater is one of the most important parameters in oceanology and critical driving factors in Marine dynamics . Previous research in oceanology usually focused on the process happened within a large space scale of hundreds of kilometers or even global scale, thus few spatial resolution of oceanic parameters are required. However, with the in-depth investigation of ocean, it is found that many parameters changed within a very small scale (around several centimeters to meter) can reveal the change or process happened on the large scale oceanic phenomenon. Such as the distributions and changes of temperature within a tiny area will reveal the heat exchange and material circulation happened in the air-sea interaction and turbulent mixing [2,3]. In addition, taking the ocean internal waves for example, when internal wave phenomena happens, usually along with seawater temperature ﬂuctuation of about 1.5-3.5°C in the direction of perpendicular to the sea surface . Furthermore, considering the layering construction of internal wave , it is necessary to monitor the change of temperature around the interface, which indicates that one point temperature measurement is incompetent for internal wave research. In other words, measuring temperatures of two quasi-continuous points around the boundary layer with high enough sensitivity may help us to monitor and judge of the occurrence of internal wave phenomenon. All above examples show the demands on the spatial resolution of centimeter level and capability of two or multi-point synchronous measurement with high sensitivity.
Contrasting typical seawater temperature detection method such as Conductivity, Temperature, and Depth sensor system (CTD), optical fiber is more suitable to solve this problem due to its small footprint, low price and immunity to electromagnetic interference. However, the existing fiber-optic temperature sensors include fiber Bragg gratings (FBG), Fabry–Pérot interferometry (FPI), and microfiber resonator (MKR), may suffer from relative low sensitivity of about 7pm/°C to 29pm/°C, and special equipment, such as FIB facility, complex knot or UV-exposure process when fabricating interferometers, resonators or gratings [6–10]. Compared with the above methods, fiber sensors based on in-line all-fiber Mach–Zehnder interferometers (MZIs) show advantages of high sensitivity, simple structure, easy fabrication, and low cost, which have been widely used in strain [11–15], bending [16,17], liquid level , air or liquid refractive index (RI) [19–24] and temperature sensing [12–14,16,18,20,22,25,26]. Especially for MZI based temperature sensing, single-mode fiber is only tapered or simply combined with those of thin-core fiber, thinned fiber, multimode fiber, or multi-core fiber, show distinct advantages of high sensitivity and easy fabrication than those of FBGs, FPIs and MKRs. However, all above mentioned in-line MZIs for temperature sensing are designed for one point sensing. Though two points temperature sensing can be realized by considering the cascade of two same MZIs, however, the transmission spectrum will become to be confused and illegible due to the overlap of two sets of similar interference signals within the same band range , which may bring some difficulty in signal analysis and spectral demodulation, and prevent its development into three or more points sensing. In addition, the too large and nonadjustable distance between the measuring points (usually larger than 10cm) will take some trouble in internal wave monitoring and other investigations of fine structures.
In this paper, we design and fabricate an all-fiber MZI by simply splicing multiple single-mode and thin-core fibers successively for quasi-continuous two points’ temperature sensing in seawater. Combined with the beam propagation theory and abundant experiment, the structure of MZI (including the splicing method, type of thin-core fibers, lengths of every section of single-mode and thin-core fibers) is carefully designed, and the transmission spectrum is obtained with two sets of clear and independent interferences. To be specific, the transmitted spectrum is typically composed by three sections, the front part comes from the interference happened in thin-core fiber, the last part is mainly from the single-mode fiber, and the middle part is the transition between above two parts. In addition, to make the offset splicing joint between two fibers more robust, a practical and simple strengthening method is introduced. Due to the sealing of the splicing section, cross-sensitivity between salinity can also be avoided. Based on the fabricated MZI, quasi-continuous two points temperatures sensing are demonstrated, and sensitivities are estimated to be 42.69pm/°C and 39.17pm/°C, respectively. By further optimization, sensitivity of 80.91pm/°C can be obtained, which is 2-10 times higher than those of FBGs, FPIs, and MKRs [6–10] and is almost the highest sensitivity reached by current similar in-line MZI sensors. In addition, some factors that may affect sensitivities are also discussed. Results show that sensitivity mainly depends on the wavelength of sensing dips and total length immersed in seawater, being independent on the locations of measuring points and salinity of seawater. The insensitivity of locations indicates the tunable distance between two points tuning from 1cm to 20cm and we treat the distance of 1cm as the quasi-continuous two points. The two points temperature sensors based on MZIs demonstrated here show advantages of simple and compact construction, robust structure, easy fabrication, high sensitivity and tunable distance between two measuring points.
2. Design and fabrication
2.1 Design on the fiber type and splicing order
In principle, for almost all such kind of in-line MZIs, their operating principle is described as follows: when light is launched from a fiber with larger core diameter into a fiber with smaller core diameter, at the splicing point, due to the mode field mismatch, a part of the power is coupled into the cladding mode of the thin-core fiber. The interference between the core mode and cladding more leads to the interference spectrum we observed. It is evident that the more significant mismatch happened in the splicing point, the more easily cladding mode is excited.
Based on the above principle, the structure is designed and shown in Fig. 1(a) based on the following considerations. Firstly, considering the cost performance, we choose the standard single-mode fiber (SMF-28e) produced by Corning Co. and thin-core fibers (980-HP: $2.7/meter and 780-HP: $3.25/meter) produced by Nufern Co. to assemble this sensor. By contrasting the structures of 980-HP (with a core diameter of 3.6μm) and 780-HP (with a core diameter of 4.4μm), it can be found that the difference between the standard single-mode fiber (SMF1) and 980-HP fiber is larger than that of 780-HP fiber, thus it can be predicted that in the respect of exciting cladding mode, choosing the 980-HP fiber as TCF1 is more suitable. To verify it, by aligned splicing the single-mode fiber with different thin-core fibers, transmission spectra are collected and shown in Fig. 1(b). It can be seen that the interference can be observed in the short wave band when we choose the structure of SMF-980-HP and there is almost no interference observed when the structure of SMF-780-HP is used. In other words, at the splicing point between SMF and 980-HP fiber, cladding mode is excited more effectively than that of 780-HP fiber, as we predicted.
2 2 Design on the offset of splicing joint
It is noticing that though the cladding mode has been excited successfully in the case of 980-HP fiber used, however, the extinction ratio is too low for sensing application. To obtain a larger extinction ratio, we need to excite more power of cladding mode, thus offset splicing method is used to further aggravate the mode field mismatch. It can be analyzed that too small offset has little help in exciting the cladding mode in fiber and too large offset will introduce extra insertion loss. To choose an optimal offset, different offsets are selected, and their transmission spectra are shown in Fig. 1(c). It can be seen that with the increasing offset, interference becomes to be more and more obviously, however, along with more and more loss. In addition, due to the high dependence of transmission on offset, the offset may be slightly changed in the fusion process, as a result, its transmission will also be changed inevitably, which can be seen by contrasting the solid and dotted blue lines in Fig. 1(c). Considering the above factors, offset between 7 and 15μm is recommended, by which sufficient cladding mode is excited and the insert loss is also acceptable. In experiment, the offset can be estimated by means of the fusion splicer with a screen showing the X-cut and Y-cut views in the spliced area (Z means the direction along the fiber axis) and be changed by the motor-driven.
Based on the above design, a section of 0.85-cm-length 980-HP fiber with a core diameter of 3.6μm is firstly spliced as TCF1 between two single-mode fibers SMF1 and SMF2 with offset of about 11μm. Similar with other in-line MZIs, at the offset splicing joint light propagating in the lead-in SMF (SMF1) core is divided into two parts: a fraction of light propagates into the core of the TCF1 as the core mode and a fraction of light propagates into the cladding of TCF1 as the cladding mode. The cladding modes interfere with the core mode due to the phase difference, thus the TCF1 acts as the first inter-modal interference sensing unit (named IMI-A) and its transmission spectrum is shown in black line of Fig. 1(d), in which typical interference is observed, especially in the wavelength range of 1150nm to 1270nm, more that 10dB extinction ratio can be obtained. The reason why the interference happened in the short wave band is more obvious than that of long wave band mainly dues that the 980-HP fiber is originally designed for the single-mode operation with cutoff wavelength around 980nm, when the operating wavelength exceeds 980nm, the longer wavelength is, the more inclined to support the single mode operation. In other words, in the longer wave band, the energy of core mode LP01 is much larger than energy distributed in other LPmn modes (cladding mode). The larger difference between two energy, the smaller extinction ratio obtained. Thus the interference happened in the long wave band will show lower extinction ratio in spectrum. The relative power of LP0n mode and LP1n mode can be calculated by the beam propagation method as mentioned in reference  and refer to the commercial software “Optiwave” .
In addition, it is found that compared with the aligned splicing, the SMF1-TCF1 offset splicing point is often fragile and easy to be broken. To strengthen the splicing point, offset splicing area is immersed into a partial removed metal tube, which is filled with liquid Polydimethylsiloxane (PDMS). Then heat to 120°C and solidify for 15 minutes. To evaluate the influence of PDMS, transmission spectrum with and without PDMS are contrasted in Fig. 1(d). It can be seen that the spectrum can still keep clear interference after sealing. Based on the above process, a robust IMI-A for one point temperature sensing is constructed. By comparing the red line and blue line in Fig. 1(d), we can estimate that the insertion loss induced by the unit IMI-A is about 8-16 dB, which is mainly caused by the offset splicing method.
2.3 Design on the independent interference of two sensing units
Firstly, considering the demand for temperature measurement at two quasi-continuous points, two sensing units are essential, and it is better that the two sets of interference are independent with each other, which is convenient for signal analysis. Due that the interference of TCF1 mainly occupy the short wave band, it is expected that the second interference unit behaves its interference in the long wave band. It is noticing that due to the mode field mismatch at the TCF1-SMF2 splicing point, part of the cladding mode travelling in the cladding of the TCF1 will directly couple into the cladding mode of the SMF2 and redistribution of the fractional power of each cladding mode will be performed. In the same way, in section of SMF2, according to the single-mode condition and fiber parameter used in experiment (core diameter D = 8.2μm, NA≈0.14), λcutoff is calculated to be about 1500nm. When operating wavelength exceeds 1500nm, due that at the TCF1-SMF2 splicing point, part of the cladding mode travelling in the cladding of the TCF1 will directly couple into the cladding mode of the SMF2, thus except for the LP01 mode, there is still some cladding mode existing and can interfere with the LP01 mode, as a result obvious interference can be observed. However, in the short wave band, due to the support of multiple modes, the energy of LP01 mode may become to be very weak due to the mode competition and cannot form the interference with the cladding mode, thus the interference of SMF2 mainly takes over the longwave band of the spectrum and form another interference sensing unit (named IMI-B) for the second point sensing.
Considering the cladding mode in SMF2 can just propagate for a short distance, to keep the propagation of cladding mode in a short SMF and prevent the degeneration of the cladding mode in too long SMF, the SMF2 is cut off and spliced with a very short section of TCF2. Due that the role of TCF2 is just to make the light propagation in SMF2 to be a limited distance, thus no cladding mode is hoped to be excited at the SMF2-TCF2 splicing point. According to the spectrum shown in Fig. 1(b), it's not hard to conclude that using the 780-HP as the TCF2 and splicing it with the SMF2 in aligned splicing method can suppress the strong excitation of cladding mode at the SMF2-TCF2 splicing point, especially in the short wave band. In addition, the length of TCF2 should be as short as possible, because ultra-short length makes the very large free spectral rang (FSR), thus avoid the remarkable impact of its interference on the whole spectrum. Finally, SMF3 is connected with the TCF2 as the lead-out fiber.
Secondly, to further verify the above analysis, based on the beam propagation method, the power evolution process as the light propagates through the MZI with structure of SMF1-0.85cmTCF1-6.45cmSMF2-0.35cmTCF2-SMF3 has been simulated, whose structure is shown in Fig. 2(a). It can be seen from Fig. 2(b) and Visualization 1 that for typical wavelength, such as 1230nm, the interference happened in the section of TCF1 (IMI-A) is much stronger than that of section SMF2 (IMI-B). However, for a larger wavelength, such as 1600nm, the opposite is true, which indicates that the interference of IMI-A mainly takes over the shortwave band of the spectrum and the IMI-B is dominant in the longwave band as we predicted.
Then, based on the above design and analysis, corresponding sensor is fabricated experimentally and its transmission spectrum is shown in Fig. 3(a). By contrasting the red line and black line in Fig. 3(a), it can be found that due to the splicing of SMF2 and TCF2, although little insert loss is introduced, but there is a significant modulation on longwave band, which is consistent with the above theoretical simulations. And as we predicted, transmission spectrum show the following characters: interference of IMI-A dominates the short wavelength of the spectrum, and interference of IMI-B dominates the long wave band. Between them, a transition region of 1300nm to 1400nm exists.
In addition, the above spectrum makes the mode analysis of the spectrum more clear and easy. For example, by Fourier transform of the IMI-A and IMI-B dominant band spectrum, we can easily get the spatial frequency of the IMI-A and IMI-B, respectively, as shown in Fig. 3(b). It can be seen that for both IMI-A and IMI-B, there are only one dominantly excited cladding mode, along with other weakly excited ones. The dominant cladding mode interferes with the core mode to form the main interference pattern. In addition, as mentioned in reference , the relation between the spatial frequency and the interferometer length as well as the differential modal group index was given as:
2.4 Design on the fiber length of IMI-B
At last, to further verify the IMI-B dominant spectrum mainly comes from the interference in section of SMF2 and investigate the effect of fiber length on transmission spectrum, we fabricate and measure the spectrum of another MZI with different lengths of SMF2 based on the theory mentioned in references  and Fig. 4(a), with the increasing length of SMF2, FSR decreases gradually, which is consistent with the theory and verify that the SMF2 indeed acts as the second interference unit. Moreover, Fig. 4(a) clearly shows the degeneration of the cladding mode in SMF with the increasing length of SMF2. It can be estimated that when the length of SMF2 exceeds about 21cm, the interference in SMF2 almost vanishes due to the thorough degeneration of the cladding mode.
On the other hand, the extra short length of SMF2 is also not recommended due that with the decreasing length of SMF2, FSR increases gradually, when the FSRs of two interferences approach with each other gradually, the overlap of the two interferences will make the spectrum to be very confused. As shown in Fig. 4(b), for the transmission spectrum of the sensor with structure of SMF1-0.85cmTCF1-2.1cmSMF2-0.35cmTCF2-SMF3, it is difficult to tell that which peak comes from the interference in IMI-A, and which one comes from IMI-B. It can be predicted that similar difficulty will also exist when simply cascading two same MZIs. Thus the fiber length of the IMI-B is recommended within the range of 6-20 centimeters.
Finally, by summarizing the above section of 2.1 to 2.4, we can conclude that choosing the 980-HP as TCF1 and splicing it after the SMF1 with offset mainly due to the effective excitation of the cladding mode. Then by use of two fibers with different cutoff wavelengths, two independent sensing units are fabricated, one mainly takes over the shortwave band of the spectrum and the other is dominant in the long wave band. Finally, a short section of 780-HP fiber is spliced after the SMF2 to collect the signal.
3. Results of sensing experiments
Based on the fabricated MZI shown in Fig. 1(a), temperature sensing is then performed with the sensing system shown in Fig. 5(a), which consists of a broad-band light source (SuperKTM Compact), a fabricated in-line MZI sensor, an optical spectrum analyzer (OSA, Ando AQ6379C), two temperature-controlled heating platforms and thermocouple thermometers (TCT). In addition, the details of the sealed splicing point are also shown in the insets of Fig. 5(a). To demonstrate the temperature sensing in seawater, seawater used in this experiment is obtained from Yellow Sea in Qingdao with 34‰ salinity and then poured into two containers. Both containers are located on heating platforms to control the temperature of seawater, and a TCT is used for calibration. The section of IMI-A is immersed into the first container with length of 3.5cm for point A sensing and IMI-B is immersed into the second container with length of 3.5cm for point B sensing. The distance between them is about 10mm, which is considered to be quasi-continuous.
Considering that the IMI-A and IMI-B are cascaded, cross influence between two points on spectrum is inevitable. So we firstly change the temperature of point A from 1.7°C to 30.5°C and keep that of point B unchanged. By tracking two typical peaks in IMI-A dominant region (peak A, with wavelength around 1245nm) and IMI-B dominant region (peak B, with wavelength around 1547nm), respectively, we plot their peak shifts with the increasing temperature in Fig. 5(b) and 5(c). It can be seen that both peaks shift to the long wave band with the increasing temperature. However, the shift of peak A is much larger than that of peak B, which dues that the peak A lies in the IMI-A dominant region, the temperature change happened in point A is bound to attract more obvious shift of peak A than that of peak B. To evaluate them quantitatively, the linear fittings of wavelength of two peaks at different temperatures are plotted in the insets of Fig. 5(b) and 5(c). It shows that sensing dips shift approximately linearly with the increase of the temperature with slopes of 42.69pm/°C and 9.67pm/°C, respectively.
In addition, to evaluate the effect of the introduction of SMF2 and TCF2 on sensitivity of TCF1, a MZI with structure like SMF-TCF-SMF is fabricated and its sensitivity for one point temperature sensing is measured. The schematic diagram of the structure and its transmission spectrum are shown in Fig. 5(d). For contrast, the sensing peak is also selected around 1245nm. By tracking the shift of this peak, the sensitivity is estimated to be about 44.57pm/°C, which is almost the same with that of IMI-A in MZI shown in Fig. 1(a).
Similarly, change the temperature of point B from 2.1°C to 32.2°C and keep that of point A unchanged, we obtain the transmission spectra of peak A and peak B under different temperatures, which are plotted in Fig. 5(e) and 5(f). The corresponding values of two peaks at different temperatures are shown in the insets of Fig. 5(e) and 5(f), in which the slopes of the lines are about 8.37pm/°C and 39.17pm/°C, respectively.
The above experiment results show that though the cross influence between two points is almost negligible, strictly, a matrix can still be applied to compensate the cross-sensitivity between two points, and simultaneous measurement of temperature of two points can also be realized using this matrix:
Based on this matrix, simultaneous two points’ temperature measurements are then demonstrated as follows: we choose the seawater sample at 12.8°C and 14.4°C as the standard sample and the corresponding wavelengths of peak A and B are 1244.62nm and 1547.38 nm, respectively, which are regarded as the initial values (denoted by λA0 and λB0). Assuming that the being tested temperatures of point A and point B at any moment are denoted by TA and TB. By collecting the transmission spectrum and checking the wavelength of peak A λA and peak B λB, temperatures of two points at any time TA and TB can be obtained based on the relation between the wavelength and their relationship, which can be expressed as:
Table 1 shows the comparisons of two tests on measured temperatures of point A and point B by our MZI with results measured by TCT. It can been seen that the temperatures measured by the MZI and TCT almost consistent with each other, which indicates the good accuracy of the MZI.
In the above used MZI, the total length of IMI-B is about 6.45cm and the distance between two containers is about 10mm, however, in experiment the length of IMI-B immersed into seawater is only 3.5cm. To further investigate the dependence of sensitivity and improve it, it is necessary to consider several factors. In theory, the temperature-induced shift of the sensing dips can be expressed as [13,20]
4.1. Influence of wavelength of sensing dips
Firstly, we investigate the effect of wavelength of sensing dips on sensitivities. A MZI with similar structure is fabricated (SMF1-0.85cmTCF1-6.45cmSMF2-0.4cmTCF2-SMF3), whose transmission spectrum is shown in Fig. 6(a). To investigate the effect of wavelength of sensing dips on sensitivities, different sensing dips are selected as typical dips, which are peak 1 to peak 5 in Fig. 6(a). By tracking their respective shifts with the increasing temperature, every peaks’ sensitivities are estimated, as shown in Fig. 6(b). It can be seen that with the increasing wavelength, sensitivity increases gradually. When the wavelength of 1673.88nm is used, maximum sensitivity can reach to be about 80.91pm/°C, which is almost the same with the highest sensitivity reached by current in-line MZI sensors .
4.2 Influence of fiber length immersed into seawater
Secondly, we investigate the effect of length of IMI-B immersed into seawater on sensitivity of peak 2. The same MZI like the one mentioned in Fig. 6 is used. To investigate the effect of length of IMI-B immersed into seawater on sensitivity, different immersed lengths are selected as typical length, which are 1.5cm, 2cm, 3cm, 5cm, and 6.5cm, respectively. Correspondingly, we track this peak shifts with the increasing temperature and estimate their sensitivities. A solid line is plotted to fit the measured sensitivity in Fig. 7, it can be seen that with the increasing length, sensitivity almost increases linearly.
4.3 Influence of the position of point B
According to the above analysis, we can also conclude that the position of point B has no effect on the sensitivity. To verify it, we chose three different positions of point B (denoted by B1, B2 and B3) as typical position, as shown in Fig. 8(a). The length of IMI-B immersed into seawater is fixed to be 3.0cm. Correspondingly, the shifts of peak B1, B2 and B3 are plotted in Fig. 8(b)-8(d). Similarly, linear fitting of these peaks’ values show that the sensitivities are 39.41pm/°C, 38.29pm/°C and 37.10pm/°C, respectively, which are almost equal to each other, as we predicted.
4.4. Influence of the salinity of seawater
For such sensors operated in seawater, it is necessary to evaluate its cross-sensitivity between temperature and salinity of seawater. Firstly, we investigate the sensitivity of this sensor operated under other salinity, such as the pure water with special salinity of zero. We fabricate another sensor same like the one used in Fig. 5 and repeat the experiments. For contrast, the sensing peaks are also selected around 1245nm and 1547nm. As is shown in Fig. 9(a) and 9(b), by changing the temperature of two points successively, and tracking the shifts of two peaks, sensitivities of IMI-A and IMI-B are estimated to be 40.26pm/°C and 38.64pm/°C, respectively, which are almost the same with the sensor operated in 34‰ salinity seawater shown in Fig. 5. In other words, salinity of seawater around the sensor has no influence on the temperature sensing, which also indicates the applicability of this MZI in other liquids.
In addition, to fully evaluate the cross-sensitivity, using the same sensor, we also investigate the change of salinity of seawater on the shift of the sensing peaks. We immerse both section IMI-A and IMI-B into the sample, by adding salt into pure water repeatedly, the salinity of sample is artificially tuned from 0‰ to 40‰, and dependence of two peak shifts on salinity changes for IMI-A and IMI-B are plotted in Fig. 10(a) and 10(b), respectively. As is shown, with the increasing salinity, both sensing peaks almost keep unmoved. And the salinity sensitivity can be estimated to be 3.3pm/‰ and 2.7pm/‰, which indicates the negligible cross-sensitivity between temperature and salinity sensing. It mainly due to the sealing of the splicing point, which prevents the contact between the seawater and the fiber cladding and avoids the cross-sensitivity of salinity.
4.5 Reproducibility of the sensor
Finally, reproducibility of a senor is also investigated. Based on the above design described in section 2, it seem that there are too many strict restrictions for fabricating such a senor and it is difficult to duplicate two identical sensor. After several attempts, it is found that only if the types of fiber and cascaded order are correct, it is very likely to obtain two sets of very similar spectra. For example, if we try to duplicate the sensor with the spectrum shown in Fig. 3(a), all parameters are repeated to fabricate another sensor. It can be seen from Fig. 11(a) that they are very similar with each other and show the same spectral characteristic. For convenient comparison, two sets of the transmission spectra are plotted in the same figure and to be staggered. What's more, by tracking the sensitivities of the peaks around the same wavelength, such as peak B and peak B’, it can be estimated that the sensitivities are 39.17pm/°C and 38.29 pm/°C, respectively. In other words, though all details in spectrum cannot be duplicated absolutely, sensitivity is still not affected by different spectral details.
However, once the wrong fiber type is chosen, such as using the 980-HP fiber as TCF2, even the same length of fiber is used, the spectrum will also tend to be messy inevitably, especially in the short wave band due to the strong excitation of the cladding mode at the splicing point between SMF2 and 980-HP fiber, which can be clearly seen from Fig. 11(b).
In conclusion, we propose and fabricate an all-fiber MZI by splicing multiple single-mode and thin-core fibers successively. By carefully designing the structure of MZI, an optimized transmission spectrum with two sets of independent interferences is obtained. Based on the fabricated MZI, quasi-continuous two points temperatures sensing are demonstrated with sensitivities of 42.69pm/°C and 39.17pm/°C, respectively. By further optimization, sensitivity of 80.91pm/°C can be obtained, which is higher than most of similar in-line MZI sensors for temperature sensing. When the resolution of OSA used is 0.02 nm, the temperature resolutions are evaluated to be 0.24°C, which is low enough for monitoring and judging of the occurrence of internal wave. Furtherly, factors affecting sensitivities are fully investigated and discussed from several aspects, such as wavelength of sensing dips, fiber length immersed into seawater, position of measuring points and salinity of seawater. The two point temperature sensing demonstrated here show advantages of simple construction, robust structure, easy fabrication, high sensitivity and tunable distance between two measuring points, which can satisfy the demands on the present oceanographic researches and may provide useful references for other sensing applications based on all-fiber MZIs, such as strain, temperature in air, and refractive index in other liquid surroundings.
National Key R & D Program of China (2017YFC1405600); Shandong Provincial Natural Science Foundation, China (ZR2017MD030).
References and links
1. N. Bruneau, J. Zika, and R. Toumi, “Can the ocean’s heat engine control horizontal circulation? Insights from the caspian sea,” Geophys. Res. Lett. 44(19), 9893–9900 (2017). [CrossRef]
2. D. P. Alappattu, Q. Wang, R. Yamaguchi, R. J. Lind, M. Reynolds, and A. J. Christman, “Warm layer and cool skin corrections for bulk water temperature measurements for air-sea interaction studies,” J. Geophys. Res. Oceans 122(8), 6470–6481 (2017). [CrossRef]
3. A. Mashayek, R. Ferrari, S. Merrifield, J. R. Ledwell, L. St Laurent, and A. N. Garabato, “Topographic enhancement of vertical turbulent mixing in the Southern Ocean,” Nat. Commun. 8, 14197 (2017). [CrossRef] [PubMed]
4. X. H. Fang, F. M. Boland, and G. R. Cresswell, “Further observations of high-frequency current variations on the continental shelf near Sydney, New South Wales,” Mar. Freshw. Res. 35(6), 611–618 (1984). [CrossRef]
5. A. K. Liu, F. C. Su, M. K. Hsu, N. J. Kuo, and C. R. Ho, “Generation and evolution of mode-two internal waves in the South China Sea,” Cont. Shelf Res. 59(59), 18–27 (2013). [CrossRef]
6. X. Shu, D. Zhao, L. Zhang, and I. Bennion, “Use of dual-grating sensors formed by different types of fiber Bragg gratings for simultaneous temperature and strain measurements,” Appl. Opt. 43(10), 2006–2012 (2004). [CrossRef] [PubMed]
7. L. Q. Men, P. Lu, and Q. Y. Chen, “A multiplexed fiber Bragg grating sensor for simultaneous salinity and temperature measurement,” J. Appl. Phys. 103(5), 053107 (2008). [CrossRef]
8. L. V. Nguyen, M. Vasiliev, and K. Alameh, “Three-wave fiber Fabry–Pérot interferometer for simultaneous measurement of temperature and water salinity of seawater,” IEEE Photonics Technol. Lett. 23(7), 450–452 (2011). [CrossRef]
10. H. J. Yang, J. Wang, Y. P. Liao, S. S. Wang, and X. Wang, “Dual-point seawater temperature simultaneous sensing based on microfiber double knot resonators,” IEEE Sens. J. 17(8), 2398–2403 (2017). [CrossRef]
12. J. Shi, S. Xiao, M. Bi, L. Yi, and P. Yang, “Discrimination between strain and temperature by cascading single-mode thin-core diameter fibers,” Appl. Opt. 51(14), 2733–2738 (2012). [CrossRef] [PubMed]
13. J. Zhou, C. Liao, Y. Wang, G. Yin, X. Zhong, K. Yang, B. Sun, G. Wang, and Z. Li, “Simultaneous measurement of strain and temperature by employing fiber Mach-Zehnder interferometer,” Opt. Express 22(2), 1680–1686 (2014). [CrossRef] [PubMed]
14. C. Li, T. G. Ning, C. Zhang, J. Li, C. B. Zhang, X. D. Wen, H. Lin, and L. Pei, “All-fiber multipath Mach–Zehnder interferometer based on a four-core fiber for sensing applications,” Sens. Actuators A Phys. 248, 148–154 (2016). [CrossRef]
15. C. Li, T. G. Ning, J. Li, C. Zhang, C. B. Zhang, H. Lin, and L. Pei, “Fiber-optic laser sensor based on all-fiber multipath Mach–Zehnder interferometer,” IEEE Photonics Technol. Lett. 28(18), 1908–1911 (2016). [CrossRef]
17. S. Zhang, W. Zhang, S. Gao, P. Geng, and X. Xue, “Fiber-optic bending vector sensor based on Mach-Zehnder interferometer exploiting lateral-offset and up-taper,” Opt. Lett. 37(21), 4480–4482 (2012). [CrossRef] [PubMed]
19. J. Villatoro and D. Monzón-Hernández, “Low-cost optical fiber refractive-index sensor based on core diameter mismatch,” J. Lightwave Technol. 24(3), 1409–1413 (2006). [CrossRef]
20. P. Lu, L. Q. Men, K. Sooley, and Q. Y. Chen, “Tapered fiber Mach–Zehnder interferometer for simultaneous measurement of refractive index and temperature,” Appl. Phys. Lett. 94(13), 131110 (2009). [CrossRef]
21. D. W. Duan, Y. J. Rao, L. C. Xu, T. Zhu, D. Wu, and J. Yao, “In-fiber Mach–Zehnder interferometer formed by large lateral offset fusion splicing for gases refractive index measurement with high sensitivity,” Sens. Actuators B Chem. 160(1), 1198–1202 (2011). [CrossRef]
22. F. Ahmed, V. Ahsani, A. Saad, and M. B. G. Jun, “Bragg grating embedded in Mach–Zehnder interferometer for refractive index and temperature sensing,” IEEE Photonics Technol. Lett. 28(18), 1968–1971 (2016). [CrossRef]
23. Y. Zhao, X. Li, L. Cai, and Y. Yang, “Refractive index sensing based on photonic crystal fiber interferometer structure with up-tapered joints,” Sens. Actuators B Chem. 221, 406–410 (2015). [CrossRef]
24. Q. Wang, L. X. Kong, Y. L. Dang, F. Xia, Y. W. Zhang, Y. Zhao, H. F. Hu, and J. Li, “High sensitivity refractive index sensor based on splicing points tapered SMF-PCF-SMF structure Mach–Zehnder mode interferometer,” Sens. Actuators B Chem. 225, 213–220 (2016). [CrossRef]
25. L. V. Nguyen, D. Hwang, S. Moon, D. S. Moon, and Y. Chung, “High temperature fiber sensor with high sensitivity based on core diameter mismatch,” Opt. Express 16(15), 11369–11375 (2008). [CrossRef] [PubMed]
26. N. Zhao, Q. Lin, W. Jing, Z. Jiang, Z. Wu, K. Yao, B. Tian, Z. Zhang, and P. Shi, “High temperature high sensitivity Mach–Zehnder interferometer based on waist-enlarged fiber bitapers,” Sens. Actuators A Phys. 267, 491–495 (2017). [CrossRef]
27. Z. B. Tian and S. S.-H. Yam, “In-line single-mode optical fiber interferometric refractive index sensors,” J. Lightwave Technol. 27(13), 2296–2306 (2009). [CrossRef]
28. “Optiwave,” Opti BPM [Online]. Available: http://optiwave.com.