- Email: [email protected]
“Bellows spring-shaped” ultrasensitive fiber-optic Fabry-Perot interferometric strain sensor
Accepted Manuscript Title: “Bellows spring-shaped” ultrasensitive fiber-optic Fabry-Perot interferometric strain sensor Authors: Pinggang Jia, Guocheng Fang, Zhe Li, Hao Liang, Yingping Hong, Ting Liang, Jijun Xiong PII: DOI: Reference:
S0924-4247(18)30211-5 https://doi.org/10.1016/j.sna.2018.04.041 SNA 10754
To appear in:
Sensors and Actuators A
Received date: Revised date: Accepted date:
2-2-2018 13-4-2018 26-4-2018
Please cite this article as: Jia P, Fang G, Li Z, Liang H, Hong Y, Liang T, Xiong J, “Bellows spring-shaped” ultrasensitive fiber-optic FabryPerot interferometric strain sensor, Sensors and Actuators: A. Physical (2010), https://doi.org/10.1016/j.sna.2018.04.041 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
“Bellows spring-shaped” ultrasensitive fiber-optic Fabry-Perot interferometric strain sensor Pinggang Jia
a,c
, Guocheng Fang
a,b,c
, Zhe Li a, Hao Liang a, Yingping Hong a,
Ting Liang a, Jijun Xiong
Key Laboratory of Instrumentation Science & Dynamic Measurement, Ministry of Education, North
SC RI PT
a
a*
University of China, Taiyuan 030051, China b
Institute for Biomedical Materials and Devices, Faculty of Science, University of Technology Sydney, New South Wales 2007, Australia
M
A
N
U
HIGHLIGHTS 1. A novel fiber-optic strain sensor based on bellows spring-shaped structure is firstly proposed, which significantly enhances the sensitivity and reduces the temperature-induced error of the Fabry-Perot strain sensor. 2. The cascaded method of the bellows spring (microbubble) unit can multiply the sensitivity effectively. 3. The sensitivity of the developed sensor is reinforced by 5-to-12 folds compared with the highest level ever reported.
Abstract
D
All-silica fiber-optic Fabry-Perot strain sensor is rapidly gaining widespread adoption in many fields due to its compact structure, low cost and immunity to electromagnetic interference. However, the conventional configuration
TE
suffers from low sensitivity due to the limitation of its inherent structure feature. In this paper, we demonstrate an ultrasensitive fiber-optic Fabry-Perot interferometric strain sensor based on silica bellows spring structure. Utilizing
EP
the cascaded microbubbles, the sensitivity of the single-microbubble sensor is enhanced up to 203.8 pm/με and that of the cascaded two-microbubble sensor reaches 518.8 pm/με. The sensitivity is reinforced by 5-to-12 folds compared
CC
with the highest level ever reported, which could also be multiplied by increasing the microbubble number. The temperature-induced strain error is much low, less than 0.01 με/°C, showing great thermal stability and hightemperature application potential. Moreover, the strain sensor also provides high-quality spectrum, controllable free-
A
spectral range and low cost. Keywords: fiber-optic sensor, Fabry–Perot interferometer, strain, ultrasensitive 1. Introduction Fiber-optic strain sensors have been proved to be useful in fields such as structural health monitoring, aerospace, and nanotechnology for their simple configuration, rapid response, low cost and immunity to electromagnetic 1
interference[1–3]. However, the conventional device greatly suffers from low sensitivity, leading to the large temperature-induced strain error and the application limit in high-precision strain measurement. The widely-used fiber-optic strain sensors are mainly based on the following three optical devices: Fiber Bragg gratings (FBG), Mach-Zehnder interferometer and Fabry-Perot interferometer. The common FBG strain sensors have the typical low sensitivity of 0.89 pm/με with a large temperature-induced strain error of 10 με/°C[4]. The FBG is
SC RI PT
achieved by periodically modulating the refractive index of fiber-optic core through ultraviolet, thus the * Corresponding author: [email protected] c
Pinggang Jia and Guocheng Fang contributed equally to this work.
intrinsic solid cylinder structure of the optical fiber hinders the improvement of the strain sensitivity. Furthermore, due to the dramatical thermal expansion effect on the nanoscale grating, the temperature drift is inevitably large, leading to a magnified temperature-induced strain error. Although some techniques, such as etching FBG on tapered optical fiber, coating graphene oxide on FBG and using etched polymer fiber, have been investigated to enhance the
U
sensitivity, the current largest sensitivity can only reach up to 5.5 pm/με with a temperature-induced strain error of 6
N
με/°C[5,6]. Similarly, the conventional core-mismatched Mach-Zehnder strain sensors typically has the sensitivity of approximately 1 pm/με within the temperature-induced strain error of 11.4 με/°C[7]. The optimized sandwich
A
microcavity-based Mach-Zehnder strain sensor changes the situation in nature and enables a 6.8 pm/με sensitivity[8].
M
A long Mach-Zehnder interference arm can benefit the optimization of strain sensitivity, for instance, a 1.8m-lengthmultimode-fiber-imbedded Mach-Zehnder strain sensor reaches the sensitivity of 18.6 pm/με[9]. However, such
D
large size violates the compact developing trend of fiber-optic sensors. Traditional fiber-optic Fabry-Perot strain sensor, that is fabricated by fusing a section of silica capillary between
TE
the optic fibers, has a low sensitivity of about 1 pm/με owing to the strain force applied on the axial direction of the silica capillary[10,11]. One method to enhance the sensitivity of the device is to shrink the length of the middle silica
EP
capillary (namely the interference length)[12]. Greice et al achieved 9.5 pm/με sensitivity by using a short capillary of 25μm[13]. However, the extremely short length of the capillary challenges the incision and fusion techniques using
CC
optical fiber cleaver and optical fiber fusion splicer. A 3.5μm-long interference length is obtained by plugging a cantilever tapered optical fiber into a silica microcapillary[14], which skillfully reduces the interference length but makes the whole device fragile.
A
Another method of optimizing the sensitivity is to make the sensors more deformable. As a result, the fiber-optic
Fabry-Perot strain sensors based on microcavity have emerged as the new generation for their potential in high sensitivity and small temperature-induced strain error. Many methods and materials have been investigated to create the microcavity, mainly comprising of the chemical etching[15–17], femtosecond laser[18], electronic arc discharge[19,20] and special-structural fiber[21] (photonic crystal fiber[22], etc.). The strain sensors based on etched microcavity have different sensitivities varying from 0.65 pm/με to 10.3 pm/με owing to their various shape and 2
size[17,18,20]. Because the chemical etching method is relatively uncontrollable, the acquired microcavity seems to be ununiform and grotesque. The femtosecond laser, which exactly makes up the insufficiency, can produce inline microcavity with desirable size, shape and surface, allowing the sensitivity reach up to 22.5 pm/με[18]. The natural microcavity in photonic crystal fiber also promises a sensitivity of 15.4 pm/με, which has high-quality interference spectrum at the same time. In addition, the microbubbles fabricated by heating the etched microcavity or tapering the
SC RI PT
liquid-induced microcavity can effectively enable ultrahigh sensitivity up to 30.66[23] or 43.0 pm/με[24]. Unfortunately, considering state of the art, the microbubble-based inline Fabry-Perot strain sensor has the following drawbacks: a) Size of the microcavity couldn’t be fabricated large enough, leading to the overmuch work on reduction of the microbubble wall (even to submicro scale), which in turn makes the microbubble fragile. b) Fabry-Perot interference length couldn’t be independent of the microbubble size and further controllably reduced. Because the microbubble wall forms the two reflective Fabry-Perot facets, the axis diameter of the microbubble is almost the Fabry-Perot interference length.
U
In this letter, we firstly demonstrated an ultrasensitive fiber-optic Fabry-Perot interferometric strain sensor based
N
on silica bellows spring structure with low-temperature coefficient. The sensor was fabricated by inserting two sections of single-mode fiber (SMF), from opposite ends, into the microbubbles which were fabricated by fusing a
A
gas-pressurized hollow silica tube (HST). This design can enable the sensors more deformable and the interference
M
length easily adjustable simultaneously. Interestingly, the strain sensitivity can be greatly improved with the increase number of the cascaded microbubbles. Experimental results show that the sensitivity of the single-microbubble sensor
D
is enhanced up to 203.8 pm/με and that of the cascaded two-microbubble sensor reaches 518.8 pm/με. The temperature-induced strain error is much low, less than 0.01 με/°C, showing great thermal stability and high-
EP
spectral range and low cost.
TE
temperature application potential. Moreover, the strain sensor also provides high-quality spectrum, controllable free-
A
CC
2. Structure and theory analysis
3
Fig. 1. Schematic configuration of the proposed fiber-optic Fabry-Perot strain sensor: (a) sensor based on single microbubble, (b) sensor based on cascaded two microbubbles. (c) Schematic of the bellows spring and the real object of the bellows spring.
Figure 1 illustrates the schematic configuration of the proposed fiber-optic Fabry-Perot interferometric strain sensor, which consists of two sections of SMF and the cascaded microbubbles. The number of the microbubbles can be single or more, for instance, the single-microbubble and two cascaded-microbubbles modalities are shown in
SC RI PT
Fig.1(a) & (b) respectively. The schematic of the bellows spring and the corresponding real object image are shown in Fig.1(c). We could see that the proposed microbubbles play the roles of the bellows disc, which could increase the deformability of the strain sensor in theory as the same function with the bellows spring. The microbubbles are fabricated by fusing a gas-pressurized HST which is then fused together with the SMF. According to Fig. 1, when light propagates along the left SMF, part of the light is reflected from the reflecting end of the left SMF, another part arrives at the end of the right SMF and reflects. Multi-beam interference occurs in the Fabry-Perot cavity. Due to the relative low reflectivity of the silica, higher-order reflection that has low energy could be ignored, thus the
4 nL
0 )
N
I I1 I2 2 I1 I 2 cos(
U
corresponding interference spectrum can be simplified as two-beam interference and determined as follows[19]:
(1)
A
where I1 and I2 are the intensities of the light reflected from the left SMF and the right SMF, n is the refractive index
M
of the air, L is the distance between the ends of the two SMFs, λ is the wavelength of the light, φ0 is the initial phase of the light. The interference valley or peak is usually used in the peak-tracing demodulation. According to Eq. (1)
D
the interference spectrum is minimized when the phase value becomes an odd multiple of the m-order spectrum
TE
valley. That is[25]
4 nL
m
0 2m 1 ,
(2)
A
CC
EP
where m is an integer, m is the initial central wavelength of the m-order spectrum valley. When the SMFs are pulled by the force towards opposite directions, the deformation of the microbubble will lead to a small change ΔL on the Fabry-Perot cavity, producing a wavelength shift of the interference spectrum. The L becomes L L and the m becomes m m . The n could be considered as constant when the sensor is used to measure the strain. The phase value of the cosine term remains identical to that of the m-order spectrum valley. That is 4 nL
m
0
4 n( L L) 0 2m 1 , m m
(3)
which could be simplified as m
4
L m . L
(4)
Thereby the sensitivity of the strain sensor depends on the ΔL/L when a certain force is loaded, which illustrates that an easy-deformed shape and a small distance between the ends of the two SMFs could promote the strain sensitivity. Due to the similar superposed bellows spring shape, according to the corresponding theory, the singlemicrobubble change induced by force F can be estimated from[26,27] 2 E Lt 3 1 2 H
D L h L 1 2t t 4t ， t
SC RI PT
F
(5)
where E is Young's modulus, μ is Poisson's ratio, t is the average thickness of the microbubble, H and D are the longitudinal diameter and axial length, as shown in Fig. 1. α can be obtained from C 1 / C ， C 1 / C 1 (2 / lnC) 2
(6)
where C=H/h , h is the outer diameter of the SMF, as shown in Fig. 1.
U
According to Eq. (5) and Eq. (6), Fig 2 indicates the numerical simulation for the force and the length change ΔL
N
of the cascaded two microbubbles (S1), the single microbubble (S2), and the conventional Fabry-Perot strain sensor
A
based on in-line silica tube (S3), respectively. Inset is the simulation parameters and the schematic structure of the S3. As we can see, compared with the conventional Fabry-Perot strain sensor based on in-line silica tube, the bellows
M
spring shape and the large size of the microbubble ensure a larger ΔL when a certain force F is loaded. The cascade configuration can improve the ΔL greatly. It is all generally known that the low sensitivity is partially caused by large
D
stiffness coefficients. The axial loading force hardly cause the length change of the capillary owing to the large
TE
stiffness factor. The microbubble in fact makes the sensors more deformable. More specifically, the cascaded two microbubbles can double the sensitivity referring to the single-microbubble sensitivity. It is also conceivable the sensitivity can be multiplied based on the cascaded number of the microbubble. Furthermore, the distance L between
EP
the reflecting facets of SMFs can be easily adjusted according to the following fabrication process, indicating the
A
CC
flexible control of the free spectrum range of the interferometric spectrum.
5
SC RI PT
Fig. 2. Numerical simulation for the force and the length change ΔL of the cascaded microbubbles (S1), the single microbubble (S2), and the conventional Fabry-Perot strain sensor based on in-line silica tube (S3), respectively. Inset is the simulation parameters
A
N
U
and the schematic structure of the S3.
3. Fabrication
M
Figure 3 illustrates the fabrication process of the strain sensor based on cascaded two microtubules, which mainly involves 6 steps. The HST (YN126200, Yongnian Ruipu Chromatogram Equipment Co., Ltd. China) has the inner
D
and outer diameters of 128 μm and 200 μm, respectively. First of all, the well-cut SMF (G652D, Yangtze Optical
TE
Fiber and Cable Co., Ltd. China) was inserted into the HST using a commercial fusion splicer (FITEL, S183 version 2, Japan), and then fused with the HST, as shown in Fig. 3(a). The fusion intensity and fusion duration are set as 100 uints and 700 ms, respectively. The electronic arc discharge is induced by 1 ~ 2 times. In step 2, the HST was filled
EP
with air using a pneumatic pump (ConST, 162, China) and the absolute pressure was maintained of 110 ~120 kPa, which was followed by moving the left motor until an appropriate distance between the electrodes and the right end
CC
of the fusion area, as shown in Fig. 3(b). In step 3, the HST was heated by implementing the electronic arc discharge with 100 units fusion intensity, 800 ms fusion duration and 4 ~ 5 times electronic arc discharge, as shown in Fig. 3(c).
A
Due to the slightly higher pressure in HST, the heating area was thermally expanded into a microbubble, and the thickness of the microbubble was simultaneously reduced. The parameters, place and times of the electronic arc discharge can be adjusted according to the observation on the fusion splicer screen. Then the electrodes were moved in front of the microbubble with an appropriate distance. The distance can be estimated according to the size of the first microbubble. The valley gap should also be considered which is approximately half of the D, according to our experience. In step 4, another microbubble was formed by repeating the step 3, as shown in Fig. 3(d). In step 5, the right end of the HST was cut off using a fiber cleaver under a microscope, as shown in Fig. 3(e). In the final step, 6
another well-cut SMF was inserted into the microbubble and fused together with the HST following the same parameters in step 1, as shown in Fig. 3(f). In this step, during the insertion process, the SMF can connect with an optical spectrum analyzer to adjust the free spectrum range by controlling the interference distance in real time. Considering the whole process, the reflecting ends of the SMFs were not damaged or contaminated, which ensured
N
U
SC RI PT
a high-quality interference spectrum.
A
Fig. 3. Fabrication process of the proposed fiber-optic Fabry-Perot strain sensor based on two cascaded microbubbles.
The microscopic image of the typical strain sensor based on single microbubble is shown in Fig. 4(a), with the
M
diameter H approximately 370 μm. Fig. 4(b) shows the microscopic image of the typical strain sensor based on cascaded two microbubbles. We could see that the microbubble has fine uniform and good sphericity. The cascaded
D
two microbubbles form a visible ravine between the microbubbles, enhancing the stability of the structure to some
TE
extent. The slight asymmetric wall thickness may be caused by the fact that the electrode position is not in the exact center of the microbubble, which could be optimized by rotating the HST during the fabrication or relying on MEMS technology to enhance the uniformity. With the protection of the microbubble, the reflecting facets of the SMF aren’t
EP
damaged, ensuring the high quality of the spectrum. The Fig. 4(c) illustrates the cross-sectional image of the HST. The typical interference obtained by an optical spectrum analyzer (Micron Optics Inc., SM125, USA) is shown in
CC
Fig. 4(d). The spectrum has the optical intensity of approximately -20 dB and the contrast ration of approximately 10 dB. Compared with the fabrication process of the other fiber-optic strain sensors, we need only a commercial fiber-
A
optic fusion splicer and a small pneumatic pump. The proposed strain sensor also has advantages of high-quality spectrum, controllable free-spectral range and low cost.
7
SC RI PT
Fig. 4. (a) Microscopic image of the proposed fiber-optic strain sensor based on single microbubble. (b) Microscopic image of the proposed fiber-optic strain sensor based on cascaded two microbubbles. (c) Cross-sectional image of the HST. (d) Typical interference spectrum of the strain sensor.
As for the fabrication uniformity of the proposed sensor, the corresponding validation confirms that the proposed
U
electronic arc discharge can guarantee the mechanical and spectrum uniformity in an accepted degree. We could
N
imagine that the same parameters of silica capillary, electronic arc discharge and filled pressure could ensure the basically same expansion of the microbubble. The used silica capillary has a good uniformity and homogeneity due
A
to its traction processing. Meanwhile, the parameters (intensity, duration etc.) of the fusion splicer has neglect
M
variation at each time and the pneumatic pump has a adjustive accuracy as small as 10Pa. Thus the microbubbles attained in this method has good uniformity under the unchanged processing condition. Following the
D
abovementioned parameters, the microbubbles have a vertical diameter H of around 370 μm, an axial radius D of around 190 μm and a insertion loss of around -20 dB. So the proposed electronic arc discharge fabricating method
TE
can satisfy the requirement of the uniformity. Here, the sensor attained by the electronic arc discharge only used for the proof-of-concept validation. In the future batch fabrication, the MEMS technology will enhance the uniformity
EP
and reduce the cost simultaneously, significantly enlarging the practical application.
A
CC
4. Experimental results and discussion
Fig. 5. Experimental setup for the strain measurement.
8
The experimental setup for the strain measurement is shown in Fig.5, which consisted of an optical spectrum analyzer with a scanning laser inside, a fixed platform, a moving platform and a personal computer. The strain sensor can be fixed on the platforms using the strain glue (KYOWA, #2129, Japan), as shown in Fig. 5. If the distance between the strain glue points is Lg and the moving platform runs △Lg, the average strain (in terms of uε) along the fiber is equal to (Lg / Lg ) 106 . So we could preciously control
SC RI PT
the strain loaded by adjusting the Vernier gauge. Two samples were used to characterize the strain: sensor A is based on single microbubble which has the H of 376 μm and the D of 164 μm, sensor B is based on two cascaded microbubbles which have the H1 of 372 μm, D1 of 184 μm, H2 of 376 μm, D2 of 177 μm, respectively. The smallest thickness t of the two samples was observed and estimated under a light microscope (OLS54100 LEXT, Japan), indicating the outmost thickness being approximately 5 μm. The strain was increased and decreased with a step of 25 με when sensor A was tested. Due to the large wavelength
U
shift, the strain was increased and decreased with a step of approximately 12.5 με when sensor B was tested. The spectrum was recorded at each step. To clearly illustrate the wavelength shift, the spectra of sensor B under 0, 12.5
N
,25 and 37.5 με are shown in Fig. 6, which shows that the wavelength shifted towards the long wavelength when the
A
strain was applied owing to the increased distance between the SMFs ends. The valley around 1540 nm was traced.
M
Because the swept wavelength of the laser integrated in the optical spectrum analyzer covers 1510~1590 nm, the spectrum cannot be obtained once the wavelength shift exceeds 1590 nm. Fortunately, according to Eq. (2), the wavelength shift beyond the wavelength scope can be calculated by the wavelength within the scope. To be more
TE
D
specific, when a step of strain is applied, the invisible wavelength shift can be obtained by[24] 2
2 1 1
(7)
EP
where λ1 is the valley wavelength within 1510 ~ 1590 nm, Δλ1 is the wavelength shift we could observe obviously, λ2 is the valley wavelength which now is within the observable scope but will overstep the scope
A
CC
after the strain applied. Thus, the wavelength can be calculated by (7) step by step.
9
SC RI PT
D
M
A
N
U
Fig. 6. Different spectra of sensor B when the loaded strain is 0, 12.5, 25 and 37.5 με, respectively.
TE
Fig.7. Wavelength shift with respect to strain.
Experimental results showed that the sensor A and sensor B showed linear response to strain with sensitivities of
EP
approximately 203.8 and 518.8 pm/με respectively over the strain range of approximately 800 and 1100 με, as shown in Fig. 7. The correlation coefficient squares reached 99.98% and 99.95%, respectively. Such sensitivity of 518.8
CC
pm/με is much high, approximately 583 times higher than that of the FBG strain sensor[4], 12 times larger than the highest that of the microcavity based Fabry- Perot strain sensor ever reported[23,24]. The up and down measured points agreed well with each other, illustrating the well repeatability of the sensor. Compared with the sensitivity of
A
the two samples, we can conclude that the sensitivity of the sensor can be multiplied by increasing the number of the cascaded microbubbles. To investigate the temperature coefficient of the proposed strain sensor, the two sample sensors were placed in a muffle furnace (Nabertherm, sn209012, Germany). The temperature was increased with a step of 40 °C, and was kept for 5 mins at each step. The corresponding spectrum was recorded. Due to the outward thermal expansion of the microbubbles, the spectrum of the sensor A shifted toward the long wavelength when the temperature increased. We 10
assume that the coefficient of long-wavelength shift is positive. Experimental results showed that the sensor A had linear response to temperature with a sensitivity of approximately 2.3 pm/°C within the temperature range 20 - 300 °C as shown in Fig. 8. The spectrum of the sensor B shifted toward the short wavelength when the temperature increased, which may be caused by the inward expansion of the SMFs exceeding the outward expansion of the microbubble. The phenomena results from that the thermal expansion coefficient of SMF is larger than that of the
SC RI PT
silica owing to the Ge-doped fiber core[25]. Another contribution to the mitigation of the axial expansion may be the spherical structure of the microbubble. Thus, compared with the simple capillary, the microbubble has a smaller axial thermal-expansion elongation. According to the definition, the sensor B had linear response to temperature with a sensitivity of approximately -2.5 pm/°C. The proposed sensor had a temperature-induced strain error of approximately 0.01 με/°C, which can be negligible to some extent. Such temperature-induced strain error is only one thousand order of magnitude lower than that of the FBG strain sensor, and a quarter of the smallest ever reported of the microcavity-based Fabry-Perot strain sensor. The positive and negative temperature coefficients indicate that an
U
approximately zero temperature coefficient may be obtained by preciously controlling the fusion position, which
Fig. 8. Wavelength shift with respect to temperature.
CC
EP
TE
D
M
A
N
makes great sense for the high-temperature application and will be our future work.
To characterize the stability of the proposed sensor, we implented the following experiments with another sensor
A
(two cascaded microbubble type). Firstly, the stability was investigated under the condition of free strain loaded at the of temperature 300°C. Over the course of experiment, the spectra were recorded with the time step of 10 min over the range of 2 hours. To avoid the complexity, Fig.9 (a) only shows the spectra with the time step of 20 min. The top and right axes demonstrate the wavelength of the monitored valley (@ around 1545nm) with respect to time. We could see that the whole spectra almost keep stable except some fluctuation on the peak, showing the good stability of the sensor at high temperature under the condition of free strain. Then to research the thermal stability under the condition of strain loaded, the sensor was loaded with the strain of 800 με. As the previous setup, the spectra were 11
recorded with the time step of 10 min over the range of 2 hours, as shown in Fig.9 (b). The top and right axes also show the wavelength of the monitored valley (@ around 1554nm) with respect to time. According to the spectra, we could see that even there is strain and high temperature loaded, the whole spectra seem to be unchanged, indicating the good stability under the strain loaded. Because the stability results at room temperature are the same with that at high temperature, so the corresponding results are not discussed here. The whole stability experiments show the great
A
N
U
SC RI PT
practical application of the proposed sensor.
Fig. 9. (a) Spectra of the proposed sensor with free strain over the time range of 120 min at 300°C and the corresponding
M
wavelength shift with respect to time; (b) spectra of the proposed sensor with 800 με strain over the time range of 120 min at 300°C
D
and the corresponding wavelength shift with respect to time.
6. Conclusion
TE
An ultrasensitive fiber-optic Fabry-Perot interferometric strain sensor based on bellows spring-shaped structure was experimentally demonstrated. The special shape and the cascaded structure can improve the strain sensitivity
EP
efficiently. Experimental results showed that the sensitivity of the single microbubble and the two cascaded microbubbles can be enhanced to 203.8 pm/με and 518.8 pm/με, respectively. Moreover, such a strain sensor ensures
CC
low temperature-induced strain error, high-quality spectrum and easily-adjustable free spectrum range, indicating the great potential of the future application. The sensor also indicates a large possibility of high-temperature application and hydraulic pressure application. Furthermore, the MEMS technology could potentially optimize the uniformity,
A
batch fabrication and low cost. Acknowledgment This work was supported by the National Science Fund for Distinguished Young Scholars of China (51425505); National Natural Science Foundation of China (51405454); Fund for Shanxi “1331 Project” Key Subject Construction; China Scholarship Council CSC scholarships (Guocheng Fang 201708140082).
12
A
CC
EP
TE
D
M
A
N
U
SC RI PT
References [1] Y. Wang, Review of long period fiber gratings written by CO2 laser, J. Appl. Phys. 108 (2010). doi:10.1063/1.3493111. [2] Y. Zhang, P. Yuan, H.Y. Choi, K.S. Park, S.J. Park, U.C. Paek, B.H. Lee, E.S. Choi, C. Wu, Z. Liu, A.P. Zhang, B.O. Guan, H.Y. Tam, Temperature-insensitive fiber optic Fabry-Perot interferometer based on special air cavity for transverse load and strain measurements, Opt. Express. 25 (2017) 9443–9448. doi:10.1364/OE.25.009443. [3] M. Rajibul Islam, M. Mahmood Ali, M.H. Lai, K.S. Lim, H. Ahmad, Chronology of fabry-perot interferometer fiber-optic sensors and their applications: A review, Sensors (Switzerland). 14 (2014) 7451–7488. doi:10.3390/s140407451. [4] L. Lv, S. Wang, L. Jiang, F. Zhang, Z. Cao, P. Wang, Y. Jiang, Y. Lu, Simultaneous measurement of strain and temperature by two peanut tapers with embedded fiber Bragg grating, Appl. Opt. 54 (2015) 10678. doi:10.1364/AO.54.010678. [5] S. S, K.S. Vasu, A. S., S. A. K., Enhanced strain and temperature sensing by reduced graphene oxide coated etched fiber Bragg gratings, Opt. Lett. 41 (2016) 2604–2607. doi:10.1364/OL.41.002604. [6] S. Sengupta, S.K. Ghorai, P. Biswas, Design of superstructure fiber bragg grating with efficient mode coupling for simultaneous strain and temperature measurement with low cross-sensitivity, IEEE Sens. J. 16 (2016) 7941–7949. doi:10.1109/JSEN.2016.2611002. [7] Z. Cao, X. Ji, R. Wang, Z. Zhang, T. Shui, F. Xu, B. Yu, Compact fiber sensor with high spatial resolution for simultaneous strain and temperature measurement, IEEE Sens. J. 13 (2013) 1447– 1451. doi:10.1109/JSEN.2012.2237092. [8] C.R. Liao, D.N. Wang, Y. Wang, Microfiber in-line Mach-Zehnder interferometer for strain sensing., Opt. Lett. 38 (2013) 757–9. doi:10.1364/OL.38.000757. [9] Y. Liu, L. Wei, Low-cost high-sensitivity strain and temperature sensing using graded-index multimode fibers, Appl. Opt. 46 (2007) 2516–2519. doi:10.1364/AO.46.002516. [10] H. Liu, H.Z. Yang, X.G. Qiao, M.L. Hu, Z.Y. Feng, R. Wang, Q. Rong, D.S. Gunawardena, K.S. Lim, H. Ahmad, Strain measurement at high temperature environment based on Fabry-Perot interferometer cascaded fiber regeneration grating, Sensors Actuators, A Phys. 248 (2016) 199– 205. doi:10.1016/j.sna.2016.07.028. [11] Q. Shi, Z. Wang, L. Jin, Y. Li, H. Zhang, F. Lu, G. Kai, X. Dong, A hollow-core photonic crystal fiber cavity based multiplexed Fabry-Pérot interferometric strain sensor system, IEEE Photonics Technol. Lett. 20 (2008) 1329–1331. doi:10.1109/LPT.2008.926948. [12] Y. Zhao, M. qing Chen, R. qing Lv, F. Xia, In-fiber rectangular air fabry-perot strain sensor based on high-precision fiber cutting platform, Opt. Commun. 384 (2017) 107–110. doi:10.1016/j.optcom.2016.10.005. [13] G.K.B. Costa, P.M.P. Gouvêa, L.M.B. Soares, J.M.B. Pereira, F. Favero, A.M.B. Braga, P. PalffyMuhoray, A.C. Bruno, I.C.S. Carvalho, In-fiber Fabry-Perot interferometer for strain and magnetic field sensing, Opt. Express. 24 (2016) 14690. doi:10.1364/OE.24.014690. [14] Y. Liu, C. Lang, X. Wei, S. Qu, Strain force sensor with ultra-high sensitivity based on fiber inline Fabry-Perot micro-cavity plugged by cantilever taper, Opt. Express. 25 (2017) 2691–2695. doi:10.1364/OE.25.007797. [15] S. Pevec, D. Donlagic, All-fiber, long-active-length Fabry-Perot strain sensor, 19 (2011) 2621– 2625. [16] P.A.R. Tafulo, P.A.S. Jorge, J.L. Santos, O. Frazão, Fabry–Pérot cavities based on chemical etching for high temperature and strain measurement, Opt. Commun. 285 (2012) 1159–1162. doi:10.1016/j.optcom.2011.11.097. 13
A
CC
EP
TE
D
M
A
N
U
SC RI PT
[17] Yuan Gong, Yun-Jiang Rao, Yu Guo, Zeng-Ling Ran, Yu Wu, Temperature-Insensitive Micro Fabry-Perot Strain Sensor Fabricated by Chemically Etching Er-Doped Fiber, IEEE Photonics Technol. Lett. 21 (2009) 1725–1727. doi:10.1109/LPT.2009.2032662. [18] Y. Liu, S. Qu, W. Qu, R. Que, A Fabry-Perot cuboid cavity across the fibre for high-sensitivity strain force sensing, J. Opt. (United Kingdom). 16 (2014). doi:10.1088/2040-8978/16/10/105401. [19] S. Liu, Y. Wang, C. Liao, G. Wang, Z. Li, Q. Wang, J. Zhou, High-sensitivity strain sensor based on in-fiber improved Fabry – Perot interferometer, Opt. Lett. 39 (2014) 2121–2124. [20] a Zhou, B.Y. Qin, Z. Zhu, Y.X. Zhang, Z.H. Liu, J. Yang, L.B. Yuan, Hybrid structured fiber-optic Fabry-Perot interferometer for simultaneous measurement of strain and temperature, Opt. Lett. 39 (2014) 5267–5270. doi:Doi 10.1364/Ol.39.005267. [21] M.S. Ferreira, P. Roriz, J. Bierlich, J. Kobelke, K. Wondraczek, C. Aichele, K. Schuster, J.L. Santos, O. Frazão, Fabry-Perot cavity based on silica tube for strain sensing at high temperatures, Opt. Express. 23 (2015) 16063. doi:10.1364/OE.23.016063. [22] M.S. Ferreira, J. Bierlich, J. Kobelke, K. Schuster, J.L. Santos, O. Frazão, Towards the control of highly sensitive Fabry-Pérot strain sensor based on hollow-core ring photonic crystal fiber, Opt. Express. 20 (2012) 21946. doi:10.1364/OE.20.021946. [23] C. Yin, Z. Cao, Z. Zhang, T. Shui, R. Wang, J. Wang, L. Lu, S. Zhen, B. Yu, TemperatureIndependent Ultrasensitive Fabry-Perot All-Fiber Strain Sensor Based on a Bubble-Expanded Microcavity, IEEE Photonics J. 6 (2014). doi:10.1109/JPHOT.2014.2345883. [24] S. Liu, K. Yang, Y. Wang, J. Qu, C. Liao, J. He, Z. Li, G. Yin, B. Sun, J. Zhou, G. Wang, J. Tang, J. Zhao, High-sensitivity strain sensor based on in-fiber rectangular air bubble, Sci. Rep. 5 (2015) 1–7. doi:10.1038/srep07624. [25] P. Jia, G. Fang, T. Liang, Y. Hong, Q. Tan, X. Chen, W. Liu, C. Xue, J. Liu, W. Zhang, J. Xiong, Temperature-compensated fiber-optic Fabry–Perot interferometric gas refractive-index sensor based on hollow silica tube for high-temperature application, Sensors Actuators, B Chem. 244 (2017) 226–232. doi:10.1016/j.snb.2016.12.123. [26] H.K. Dubey, V. Bhope, Stress and Deflection Analysis of Belleville Spring, IOSR J. Mech. Civ. Eng. 2 (2012) 2278–1684. www.iosrjournals.org. [27] J.R. Jang, ANFIS : Adap tive-Ne twork-Based Fuzzy Inference System, 23 (1993).
14
Pinggang Jia received the Ph.D degree in instrument science and technology from Chongqing University, Chongqing, China, in 2013. He is currently a teacher with North University of China, where he is also a staff of the Micro and Nano technology research center, North University of China. His current research interests include fiber optic sensors, sensor interrogation systems, and laser-based measurement technology, high temperature sensors.
SC RI PT
Guocheng Fang received his master degree from North University of China in 2017. He is currently completing his Ph.D degree in Institute for Biomedical Materials and Devices, Faculty of Science, University of Technology Sydney. His current research interests include optical sensors, MESM technology and point-of-care devices. Zhe Li received his bachelor degree from North University of China in 2015. He is currently completing his master degree in North University of China. His current research interests include fiber-optic sensors, high-temperature sensors and microfabrication technology.
N
U
Hao Liang received his bachelor degree from North University of China in 2015. He is currently completing his master degree in North University of China. His current research interests include fiber-optic sensors, fiber bragg grating sensors and microfabrication technology.
M
A
Yingping Hong receive the Ph.D degree in measurement technology and instrument from the North University of China. Now he is a Lecturer with North University of China. He has been engaged in the research of mechanical parameter in extreme environment, involving pressure testing for high temperature, data recording and transmission.
EP
TE
D
Ting Liang received the Ph.D degree in microelectronics and solid-state electronics from Beijing University of Technology, Beijing, China, in 2007. Now he is an associate professor with North University of China. His current research interests include optical gas sensors, high temperature sensing technology, MEMS technology
A
CC
Jijun Xiong received the Ph.D degree from Tsinghua University, Beijing, China, in 2003. From 2003 to 2005, he was a Post-Doctoral Researcher with Tsinghua University, Beijing, China. He is now a distinguished professor with North university of China. His research interests include the fundamental theory and instrumentation research on dynamic mechanical parameters measurement, mainly focus on the scale effect in silicon microstructures, dynamic measurement, MESM sensing technology, high temperature sensing technology.
15
S0924-4247(18)30211-5 https://doi.org/10.1016/j.sna.2018.04.041 SNA 10754
To appear in:
Sensors and Actuators A
Received date: Revised date: Accepted date:
2-2-2018 13-4-2018 26-4-2018
Please cite this article as: Jia P, Fang G, Li Z, Liang H, Hong Y, Liang T, Xiong J, “Bellows spring-shaped” ultrasensitive fiber-optic FabryPerot interferometric strain sensor, Sensors and Actuators: A. Physical (2010), https://doi.org/10.1016/j.sna.2018.04.041 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
“Bellows spring-shaped” ultrasensitive fiber-optic Fabry-Perot interferometric strain sensor Pinggang Jia
a,c
, Guocheng Fang
a,b,c
, Zhe Li a, Hao Liang a, Yingping Hong a,
Ting Liang a, Jijun Xiong
Key Laboratory of Instrumentation Science & Dynamic Measurement, Ministry of Education, North
SC RI PT
a
a*
University of China, Taiyuan 030051, China b
Institute for Biomedical Materials and Devices, Faculty of Science, University of Technology Sydney, New South Wales 2007, Australia
M
A
N
U
HIGHLIGHTS 1. A novel fiber-optic strain sensor based on bellows spring-shaped structure is firstly proposed, which significantly enhances the sensitivity and reduces the temperature-induced error of the Fabry-Perot strain sensor. 2. The cascaded method of the bellows spring (microbubble) unit can multiply the sensitivity effectively. 3. The sensitivity of the developed sensor is reinforced by 5-to-12 folds compared with the highest level ever reported.
Abstract
D
All-silica fiber-optic Fabry-Perot strain sensor is rapidly gaining widespread adoption in many fields due to its compact structure, low cost and immunity to electromagnetic interference. However, the conventional configuration
TE
suffers from low sensitivity due to the limitation of its inherent structure feature. In this paper, we demonstrate an ultrasensitive fiber-optic Fabry-Perot interferometric strain sensor based on silica bellows spring structure. Utilizing
EP
the cascaded microbubbles, the sensitivity of the single-microbubble sensor is enhanced up to 203.8 pm/με and that of the cascaded two-microbubble sensor reaches 518.8 pm/με. The sensitivity is reinforced by 5-to-12 folds compared
CC
with the highest level ever reported, which could also be multiplied by increasing the microbubble number. The temperature-induced strain error is much low, less than 0.01 με/°C, showing great thermal stability and hightemperature application potential. Moreover, the strain sensor also provides high-quality spectrum, controllable free-
A
spectral range and low cost. Keywords: fiber-optic sensor, Fabry–Perot interferometer, strain, ultrasensitive 1. Introduction Fiber-optic strain sensors have been proved to be useful in fields such as structural health monitoring, aerospace, and nanotechnology for their simple configuration, rapid response, low cost and immunity to electromagnetic 1
interference[1–3]. However, the conventional device greatly suffers from low sensitivity, leading to the large temperature-induced strain error and the application limit in high-precision strain measurement. The widely-used fiber-optic strain sensors are mainly based on the following three optical devices: Fiber Bragg gratings (FBG), Mach-Zehnder interferometer and Fabry-Perot interferometer. The common FBG strain sensors have the typical low sensitivity of 0.89 pm/με with a large temperature-induced strain error of 10 με/°C[4]. The FBG is
SC RI PT
achieved by periodically modulating the refractive index of fiber-optic core through ultraviolet, thus the * Corresponding author: [email protected] c
Pinggang Jia and Guocheng Fang contributed equally to this work.
intrinsic solid cylinder structure of the optical fiber hinders the improvement of the strain sensitivity. Furthermore, due to the dramatical thermal expansion effect on the nanoscale grating, the temperature drift is inevitably large, leading to a magnified temperature-induced strain error. Although some techniques, such as etching FBG on tapered optical fiber, coating graphene oxide on FBG and using etched polymer fiber, have been investigated to enhance the
U
sensitivity, the current largest sensitivity can only reach up to 5.5 pm/με with a temperature-induced strain error of 6
N
με/°C[5,6]. Similarly, the conventional core-mismatched Mach-Zehnder strain sensors typically has the sensitivity of approximately 1 pm/με within the temperature-induced strain error of 11.4 με/°C[7]. The optimized sandwich
A
microcavity-based Mach-Zehnder strain sensor changes the situation in nature and enables a 6.8 pm/με sensitivity[8].
M
A long Mach-Zehnder interference arm can benefit the optimization of strain sensitivity, for instance, a 1.8m-lengthmultimode-fiber-imbedded Mach-Zehnder strain sensor reaches the sensitivity of 18.6 pm/με[9]. However, such
D
large size violates the compact developing trend of fiber-optic sensors. Traditional fiber-optic Fabry-Perot strain sensor, that is fabricated by fusing a section of silica capillary between
TE
the optic fibers, has a low sensitivity of about 1 pm/με owing to the strain force applied on the axial direction of the silica capillary[10,11]. One method to enhance the sensitivity of the device is to shrink the length of the middle silica
EP
capillary (namely the interference length)[12]. Greice et al achieved 9.5 pm/με sensitivity by using a short capillary of 25μm[13]. However, the extremely short length of the capillary challenges the incision and fusion techniques using
CC
optical fiber cleaver and optical fiber fusion splicer. A 3.5μm-long interference length is obtained by plugging a cantilever tapered optical fiber into a silica microcapillary[14], which skillfully reduces the interference length but makes the whole device fragile.
A
Another method of optimizing the sensitivity is to make the sensors more deformable. As a result, the fiber-optic
Fabry-Perot strain sensors based on microcavity have emerged as the new generation for their potential in high sensitivity and small temperature-induced strain error. Many methods and materials have been investigated to create the microcavity, mainly comprising of the chemical etching[15–17], femtosecond laser[18], electronic arc discharge[19,20] and special-structural fiber[21] (photonic crystal fiber[22], etc.). The strain sensors based on etched microcavity have different sensitivities varying from 0.65 pm/με to 10.3 pm/με owing to their various shape and 2
size[17,18,20]. Because the chemical etching method is relatively uncontrollable, the acquired microcavity seems to be ununiform and grotesque. The femtosecond laser, which exactly makes up the insufficiency, can produce inline microcavity with desirable size, shape and surface, allowing the sensitivity reach up to 22.5 pm/με[18]. The natural microcavity in photonic crystal fiber also promises a sensitivity of 15.4 pm/με, which has high-quality interference spectrum at the same time. In addition, the microbubbles fabricated by heating the etched microcavity or tapering the
SC RI PT
liquid-induced microcavity can effectively enable ultrahigh sensitivity up to 30.66[23] or 43.0 pm/με[24]. Unfortunately, considering state of the art, the microbubble-based inline Fabry-Perot strain sensor has the following drawbacks: a) Size of the microcavity couldn’t be fabricated large enough, leading to the overmuch work on reduction of the microbubble wall (even to submicro scale), which in turn makes the microbubble fragile. b) Fabry-Perot interference length couldn’t be independent of the microbubble size and further controllably reduced. Because the microbubble wall forms the two reflective Fabry-Perot facets, the axis diameter of the microbubble is almost the Fabry-Perot interference length.
U
In this letter, we firstly demonstrated an ultrasensitive fiber-optic Fabry-Perot interferometric strain sensor based
N
on silica bellows spring structure with low-temperature coefficient. The sensor was fabricated by inserting two sections of single-mode fiber (SMF), from opposite ends, into the microbubbles which were fabricated by fusing a
A
gas-pressurized hollow silica tube (HST). This design can enable the sensors more deformable and the interference
M
length easily adjustable simultaneously. Interestingly, the strain sensitivity can be greatly improved with the increase number of the cascaded microbubbles. Experimental results show that the sensitivity of the single-microbubble sensor
D
is enhanced up to 203.8 pm/με and that of the cascaded two-microbubble sensor reaches 518.8 pm/με. The temperature-induced strain error is much low, less than 0.01 με/°C, showing great thermal stability and high-
EP
spectral range and low cost.
TE
temperature application potential. Moreover, the strain sensor also provides high-quality spectrum, controllable free-
A
CC
2. Structure and theory analysis
3
Fig. 1. Schematic configuration of the proposed fiber-optic Fabry-Perot strain sensor: (a) sensor based on single microbubble, (b) sensor based on cascaded two microbubbles. (c) Schematic of the bellows spring and the real object of the bellows spring.
Figure 1 illustrates the schematic configuration of the proposed fiber-optic Fabry-Perot interferometric strain sensor, which consists of two sections of SMF and the cascaded microbubbles. The number of the microbubbles can be single or more, for instance, the single-microbubble and two cascaded-microbubbles modalities are shown in
SC RI PT
Fig.1(a) & (b) respectively. The schematic of the bellows spring and the corresponding real object image are shown in Fig.1(c). We could see that the proposed microbubbles play the roles of the bellows disc, which could increase the deformability of the strain sensor in theory as the same function with the bellows spring. The microbubbles are fabricated by fusing a gas-pressurized HST which is then fused together with the SMF. According to Fig. 1, when light propagates along the left SMF, part of the light is reflected from the reflecting end of the left SMF, another part arrives at the end of the right SMF and reflects. Multi-beam interference occurs in the Fabry-Perot cavity. Due to the relative low reflectivity of the silica, higher-order reflection that has low energy could be ignored, thus the
4 nL
0 )
N
I I1 I2 2 I1 I 2 cos(
U
corresponding interference spectrum can be simplified as two-beam interference and determined as follows[19]:
(1)
A
where I1 and I2 are the intensities of the light reflected from the left SMF and the right SMF, n is the refractive index
M
of the air, L is the distance between the ends of the two SMFs, λ is the wavelength of the light, φ0 is the initial phase of the light. The interference valley or peak is usually used in the peak-tracing demodulation. According to Eq. (1)
D
the interference spectrum is minimized when the phase value becomes an odd multiple of the m-order spectrum
TE
valley. That is[25]
4 nL
m
0 2m 1 ,
(2)
A
CC
EP
where m is an integer, m is the initial central wavelength of the m-order spectrum valley. When the SMFs are pulled by the force towards opposite directions, the deformation of the microbubble will lead to a small change ΔL on the Fabry-Perot cavity, producing a wavelength shift of the interference spectrum. The L becomes L L and the m becomes m m . The n could be considered as constant when the sensor is used to measure the strain. The phase value of the cosine term remains identical to that of the m-order spectrum valley. That is 4 nL
m
0
4 n( L L) 0 2m 1 , m m
(3)
which could be simplified as m
4
L m . L
(4)
Thereby the sensitivity of the strain sensor depends on the ΔL/L when a certain force is loaded, which illustrates that an easy-deformed shape and a small distance between the ends of the two SMFs could promote the strain sensitivity. Due to the similar superposed bellows spring shape, according to the corresponding theory, the singlemicrobubble change induced by force F can be estimated from[26,27] 2 E Lt 3 1 2 H
D L h L 1 2t t 4t ， t
SC RI PT
F
(5)
where E is Young's modulus, μ is Poisson's ratio, t is the average thickness of the microbubble, H and D are the longitudinal diameter and axial length, as shown in Fig. 1. α can be obtained from C 1 / C ， C 1 / C 1 (2 / lnC) 2
(6)
where C=H/h , h is the outer diameter of the SMF, as shown in Fig. 1.
U
According to Eq. (5) and Eq. (6), Fig 2 indicates the numerical simulation for the force and the length change ΔL
N
of the cascaded two microbubbles (S1), the single microbubble (S2), and the conventional Fabry-Perot strain sensor
A
based on in-line silica tube (S3), respectively. Inset is the simulation parameters and the schematic structure of the S3. As we can see, compared with the conventional Fabry-Perot strain sensor based on in-line silica tube, the bellows
M
spring shape and the large size of the microbubble ensure a larger ΔL when a certain force F is loaded. The cascade configuration can improve the ΔL greatly. It is all generally known that the low sensitivity is partially caused by large
D
stiffness coefficients. The axial loading force hardly cause the length change of the capillary owing to the large
TE
stiffness factor. The microbubble in fact makes the sensors more deformable. More specifically, the cascaded two microbubbles can double the sensitivity referring to the single-microbubble sensitivity. It is also conceivable the sensitivity can be multiplied based on the cascaded number of the microbubble. Furthermore, the distance L between
EP
the reflecting facets of SMFs can be easily adjusted according to the following fabrication process, indicating the
A
CC
flexible control of the free spectrum range of the interferometric spectrum.
5
SC RI PT
Fig. 2. Numerical simulation for the force and the length change ΔL of the cascaded microbubbles (S1), the single microbubble (S2), and the conventional Fabry-Perot strain sensor based on in-line silica tube (S3), respectively. Inset is the simulation parameters
A
N
U
and the schematic structure of the S3.
3. Fabrication
M
Figure 3 illustrates the fabrication process of the strain sensor based on cascaded two microtubules, which mainly involves 6 steps. The HST (YN126200, Yongnian Ruipu Chromatogram Equipment Co., Ltd. China) has the inner
D
and outer diameters of 128 μm and 200 μm, respectively. First of all, the well-cut SMF (G652D, Yangtze Optical
TE
Fiber and Cable Co., Ltd. China) was inserted into the HST using a commercial fusion splicer (FITEL, S183 version 2, Japan), and then fused with the HST, as shown in Fig. 3(a). The fusion intensity and fusion duration are set as 100 uints and 700 ms, respectively. The electronic arc discharge is induced by 1 ~ 2 times. In step 2, the HST was filled
EP
with air using a pneumatic pump (ConST, 162, China) and the absolute pressure was maintained of 110 ~120 kPa, which was followed by moving the left motor until an appropriate distance between the electrodes and the right end
CC
of the fusion area, as shown in Fig. 3(b). In step 3, the HST was heated by implementing the electronic arc discharge with 100 units fusion intensity, 800 ms fusion duration and 4 ~ 5 times electronic arc discharge, as shown in Fig. 3(c).
A
Due to the slightly higher pressure in HST, the heating area was thermally expanded into a microbubble, and the thickness of the microbubble was simultaneously reduced. The parameters, place and times of the electronic arc discharge can be adjusted according to the observation on the fusion splicer screen. Then the electrodes were moved in front of the microbubble with an appropriate distance. The distance can be estimated according to the size of the first microbubble. The valley gap should also be considered which is approximately half of the D, according to our experience. In step 4, another microbubble was formed by repeating the step 3, as shown in Fig. 3(d). In step 5, the right end of the HST was cut off using a fiber cleaver under a microscope, as shown in Fig. 3(e). In the final step, 6
another well-cut SMF was inserted into the microbubble and fused together with the HST following the same parameters in step 1, as shown in Fig. 3(f). In this step, during the insertion process, the SMF can connect with an optical spectrum analyzer to adjust the free spectrum range by controlling the interference distance in real time. Considering the whole process, the reflecting ends of the SMFs were not damaged or contaminated, which ensured
N
U
SC RI PT
a high-quality interference spectrum.
A
Fig. 3. Fabrication process of the proposed fiber-optic Fabry-Perot strain sensor based on two cascaded microbubbles.
The microscopic image of the typical strain sensor based on single microbubble is shown in Fig. 4(a), with the
M
diameter H approximately 370 μm. Fig. 4(b) shows the microscopic image of the typical strain sensor based on cascaded two microbubbles. We could see that the microbubble has fine uniform and good sphericity. The cascaded
D
two microbubbles form a visible ravine between the microbubbles, enhancing the stability of the structure to some
TE
extent. The slight asymmetric wall thickness may be caused by the fact that the electrode position is not in the exact center of the microbubble, which could be optimized by rotating the HST during the fabrication or relying on MEMS technology to enhance the uniformity. With the protection of the microbubble, the reflecting facets of the SMF aren’t
EP
damaged, ensuring the high quality of the spectrum. The Fig. 4(c) illustrates the cross-sectional image of the HST. The typical interference obtained by an optical spectrum analyzer (Micron Optics Inc., SM125, USA) is shown in
CC
Fig. 4(d). The spectrum has the optical intensity of approximately -20 dB and the contrast ration of approximately 10 dB. Compared with the fabrication process of the other fiber-optic strain sensors, we need only a commercial fiber-
A
optic fusion splicer and a small pneumatic pump. The proposed strain sensor also has advantages of high-quality spectrum, controllable free-spectral range and low cost.
7
SC RI PT
Fig. 4. (a) Microscopic image of the proposed fiber-optic strain sensor based on single microbubble. (b) Microscopic image of the proposed fiber-optic strain sensor based on cascaded two microbubbles. (c) Cross-sectional image of the HST. (d) Typical interference spectrum of the strain sensor.
As for the fabrication uniformity of the proposed sensor, the corresponding validation confirms that the proposed
U
electronic arc discharge can guarantee the mechanical and spectrum uniformity in an accepted degree. We could
N
imagine that the same parameters of silica capillary, electronic arc discharge and filled pressure could ensure the basically same expansion of the microbubble. The used silica capillary has a good uniformity and homogeneity due
A
to its traction processing. Meanwhile, the parameters (intensity, duration etc.) of the fusion splicer has neglect
M
variation at each time and the pneumatic pump has a adjustive accuracy as small as 10Pa. Thus the microbubbles attained in this method has good uniformity under the unchanged processing condition. Following the
D
abovementioned parameters, the microbubbles have a vertical diameter H of around 370 μm, an axial radius D of around 190 μm and a insertion loss of around -20 dB. So the proposed electronic arc discharge fabricating method
TE
can satisfy the requirement of the uniformity. Here, the sensor attained by the electronic arc discharge only used for the proof-of-concept validation. In the future batch fabrication, the MEMS technology will enhance the uniformity
EP
and reduce the cost simultaneously, significantly enlarging the practical application.
A
CC
4. Experimental results and discussion
Fig. 5. Experimental setup for the strain measurement.
8
The experimental setup for the strain measurement is shown in Fig.5, which consisted of an optical spectrum analyzer with a scanning laser inside, a fixed platform, a moving platform and a personal computer. The strain sensor can be fixed on the platforms using the strain glue (KYOWA, #2129, Japan), as shown in Fig. 5. If the distance between the strain glue points is Lg and the moving platform runs △Lg, the average strain (in terms of uε) along the fiber is equal to (Lg / Lg ) 106 . So we could preciously control
SC RI PT
the strain loaded by adjusting the Vernier gauge. Two samples were used to characterize the strain: sensor A is based on single microbubble which has the H of 376 μm and the D of 164 μm, sensor B is based on two cascaded microbubbles which have the H1 of 372 μm, D1 of 184 μm, H2 of 376 μm, D2 of 177 μm, respectively. The smallest thickness t of the two samples was observed and estimated under a light microscope (OLS54100 LEXT, Japan), indicating the outmost thickness being approximately 5 μm. The strain was increased and decreased with a step of 25 με when sensor A was tested. Due to the large wavelength
U
shift, the strain was increased and decreased with a step of approximately 12.5 με when sensor B was tested. The spectrum was recorded at each step. To clearly illustrate the wavelength shift, the spectra of sensor B under 0, 12.5
N
,25 and 37.5 με are shown in Fig. 6, which shows that the wavelength shifted towards the long wavelength when the
A
strain was applied owing to the increased distance between the SMFs ends. The valley around 1540 nm was traced.
M
Because the swept wavelength of the laser integrated in the optical spectrum analyzer covers 1510~1590 nm, the spectrum cannot be obtained once the wavelength shift exceeds 1590 nm. Fortunately, according to Eq. (2), the wavelength shift beyond the wavelength scope can be calculated by the wavelength within the scope. To be more
TE
D
specific, when a step of strain is applied, the invisible wavelength shift can be obtained by[24] 2
2 1 1
(7)
EP
where λ1 is the valley wavelength within 1510 ~ 1590 nm, Δλ1 is the wavelength shift we could observe obviously, λ2 is the valley wavelength which now is within the observable scope but will overstep the scope
A
CC
after the strain applied. Thus, the wavelength can be calculated by (7) step by step.
9
SC RI PT
D
M
A
N
U
Fig. 6. Different spectra of sensor B when the loaded strain is 0, 12.5, 25 and 37.5 με, respectively.
TE
Fig.7. Wavelength shift with respect to strain.
Experimental results showed that the sensor A and sensor B showed linear response to strain with sensitivities of
EP
approximately 203.8 and 518.8 pm/με respectively over the strain range of approximately 800 and 1100 με, as shown in Fig. 7. The correlation coefficient squares reached 99.98% and 99.95%, respectively. Such sensitivity of 518.8
CC
pm/με is much high, approximately 583 times higher than that of the FBG strain sensor[4], 12 times larger than the highest that of the microcavity based Fabry- Perot strain sensor ever reported[23,24]. The up and down measured points agreed well with each other, illustrating the well repeatability of the sensor. Compared with the sensitivity of
A
the two samples, we can conclude that the sensitivity of the sensor can be multiplied by increasing the number of the cascaded microbubbles. To investigate the temperature coefficient of the proposed strain sensor, the two sample sensors were placed in a muffle furnace (Nabertherm, sn209012, Germany). The temperature was increased with a step of 40 °C, and was kept for 5 mins at each step. The corresponding spectrum was recorded. Due to the outward thermal expansion of the microbubbles, the spectrum of the sensor A shifted toward the long wavelength when the temperature increased. We 10
assume that the coefficient of long-wavelength shift is positive. Experimental results showed that the sensor A had linear response to temperature with a sensitivity of approximately 2.3 pm/°C within the temperature range 20 - 300 °C as shown in Fig. 8. The spectrum of the sensor B shifted toward the short wavelength when the temperature increased, which may be caused by the inward expansion of the SMFs exceeding the outward expansion of the microbubble. The phenomena results from that the thermal expansion coefficient of SMF is larger than that of the
SC RI PT
silica owing to the Ge-doped fiber core[25]. Another contribution to the mitigation of the axial expansion may be the spherical structure of the microbubble. Thus, compared with the simple capillary, the microbubble has a smaller axial thermal-expansion elongation. According to the definition, the sensor B had linear response to temperature with a sensitivity of approximately -2.5 pm/°C. The proposed sensor had a temperature-induced strain error of approximately 0.01 με/°C, which can be negligible to some extent. Such temperature-induced strain error is only one thousand order of magnitude lower than that of the FBG strain sensor, and a quarter of the smallest ever reported of the microcavity-based Fabry-Perot strain sensor. The positive and negative temperature coefficients indicate that an
U
approximately zero temperature coefficient may be obtained by preciously controlling the fusion position, which
Fig. 8. Wavelength shift with respect to temperature.
CC
EP
TE
D
M
A
N
makes great sense for the high-temperature application and will be our future work.
To characterize the stability of the proposed sensor, we implented the following experiments with another sensor
A
(two cascaded microbubble type). Firstly, the stability was investigated under the condition of free strain loaded at the of temperature 300°C. Over the course of experiment, the spectra were recorded with the time step of 10 min over the range of 2 hours. To avoid the complexity, Fig.9 (a) only shows the spectra with the time step of 20 min. The top and right axes demonstrate the wavelength of the monitored valley (@ around 1545nm) with respect to time. We could see that the whole spectra almost keep stable except some fluctuation on the peak, showing the good stability of the sensor at high temperature under the condition of free strain. Then to research the thermal stability under the condition of strain loaded, the sensor was loaded with the strain of 800 με. As the previous setup, the spectra were 11
recorded with the time step of 10 min over the range of 2 hours, as shown in Fig.9 (b). The top and right axes also show the wavelength of the monitored valley (@ around 1554nm) with respect to time. According to the spectra, we could see that even there is strain and high temperature loaded, the whole spectra seem to be unchanged, indicating the good stability under the strain loaded. Because the stability results at room temperature are the same with that at high temperature, so the corresponding results are not discussed here. The whole stability experiments show the great
A
N
U
SC RI PT
practical application of the proposed sensor.
Fig. 9. (a) Spectra of the proposed sensor with free strain over the time range of 120 min at 300°C and the corresponding
M
wavelength shift with respect to time; (b) spectra of the proposed sensor with 800 με strain over the time range of 120 min at 300°C
D
and the corresponding wavelength shift with respect to time.
6. Conclusion
TE
An ultrasensitive fiber-optic Fabry-Perot interferometric strain sensor based on bellows spring-shaped structure was experimentally demonstrated. The special shape and the cascaded structure can improve the strain sensitivity
EP
efficiently. Experimental results showed that the sensitivity of the single microbubble and the two cascaded microbubbles can be enhanced to 203.8 pm/με and 518.8 pm/με, respectively. Moreover, such a strain sensor ensures
CC
low temperature-induced strain error, high-quality spectrum and easily-adjustable free spectrum range, indicating the great potential of the future application. The sensor also indicates a large possibility of high-temperature application and hydraulic pressure application. Furthermore, the MEMS technology could potentially optimize the uniformity,
A
batch fabrication and low cost. Acknowledgment This work was supported by the National Science Fund for Distinguished Young Scholars of China (51425505); National Natural Science Foundation of China (51405454); Fund for Shanxi “1331 Project” Key Subject Construction; China Scholarship Council CSC scholarships (Guocheng Fang 201708140082).
12
A
CC
EP
TE
D
M
A
N
U
SC RI PT
References [1] Y. Wang, Review of long period fiber gratings written by CO2 laser, J. Appl. Phys. 108 (2010). doi:10.1063/1.3493111. [2] Y. Zhang, P. Yuan, H.Y. Choi, K.S. Park, S.J. Park, U.C. Paek, B.H. Lee, E.S. Choi, C. Wu, Z. Liu, A.P. Zhang, B.O. Guan, H.Y. Tam, Temperature-insensitive fiber optic Fabry-Perot interferometer based on special air cavity for transverse load and strain measurements, Opt. Express. 25 (2017) 9443–9448. doi:10.1364/OE.25.009443. [3] M. Rajibul Islam, M. Mahmood Ali, M.H. Lai, K.S. Lim, H. Ahmad, Chronology of fabry-perot interferometer fiber-optic sensors and their applications: A review, Sensors (Switzerland). 14 (2014) 7451–7488. doi:10.3390/s140407451. [4] L. Lv, S. Wang, L. Jiang, F. Zhang, Z. Cao, P. Wang, Y. Jiang, Y. Lu, Simultaneous measurement of strain and temperature by two peanut tapers with embedded fiber Bragg grating, Appl. Opt. 54 (2015) 10678. doi:10.1364/AO.54.010678. [5] S. S, K.S. Vasu, A. S., S. A. K., Enhanced strain and temperature sensing by reduced graphene oxide coated etched fiber Bragg gratings, Opt. Lett. 41 (2016) 2604–2607. doi:10.1364/OL.41.002604. [6] S. Sengupta, S.K. Ghorai, P. Biswas, Design of superstructure fiber bragg grating with efficient mode coupling for simultaneous strain and temperature measurement with low cross-sensitivity, IEEE Sens. J. 16 (2016) 7941–7949. doi:10.1109/JSEN.2016.2611002. [7] Z. Cao, X. Ji, R. Wang, Z. Zhang, T. Shui, F. Xu, B. Yu, Compact fiber sensor with high spatial resolution for simultaneous strain and temperature measurement, IEEE Sens. J. 13 (2013) 1447– 1451. doi:10.1109/JSEN.2012.2237092. [8] C.R. Liao, D.N. Wang, Y. Wang, Microfiber in-line Mach-Zehnder interferometer for strain sensing., Opt. Lett. 38 (2013) 757–9. doi:10.1364/OL.38.000757. [9] Y. Liu, L. Wei, Low-cost high-sensitivity strain and temperature sensing using graded-index multimode fibers, Appl. Opt. 46 (2007) 2516–2519. doi:10.1364/AO.46.002516. [10] H. Liu, H.Z. Yang, X.G. Qiao, M.L. Hu, Z.Y. Feng, R. Wang, Q. Rong, D.S. Gunawardena, K.S. Lim, H. Ahmad, Strain measurement at high temperature environment based on Fabry-Perot interferometer cascaded fiber regeneration grating, Sensors Actuators, A Phys. 248 (2016) 199– 205. doi:10.1016/j.sna.2016.07.028. [11] Q. Shi, Z. Wang, L. Jin, Y. Li, H. Zhang, F. Lu, G. Kai, X. Dong, A hollow-core photonic crystal fiber cavity based multiplexed Fabry-Pérot interferometric strain sensor system, IEEE Photonics Technol. Lett. 20 (2008) 1329–1331. doi:10.1109/LPT.2008.926948. [12] Y. Zhao, M. qing Chen, R. qing Lv, F. Xia, In-fiber rectangular air fabry-perot strain sensor based on high-precision fiber cutting platform, Opt. Commun. 384 (2017) 107–110. doi:10.1016/j.optcom.2016.10.005. [13] G.K.B. Costa, P.M.P. Gouvêa, L.M.B. Soares, J.M.B. Pereira, F. Favero, A.M.B. Braga, P. PalffyMuhoray, A.C. Bruno, I.C.S. Carvalho, In-fiber Fabry-Perot interferometer for strain and magnetic field sensing, Opt. Express. 24 (2016) 14690. doi:10.1364/OE.24.014690. [14] Y. Liu, C. Lang, X. Wei, S. Qu, Strain force sensor with ultra-high sensitivity based on fiber inline Fabry-Perot micro-cavity plugged by cantilever taper, Opt. Express. 25 (2017) 2691–2695. doi:10.1364/OE.25.007797. [15] S. Pevec, D. Donlagic, All-fiber, long-active-length Fabry-Perot strain sensor, 19 (2011) 2621– 2625. [16] P.A.R. Tafulo, P.A.S. Jorge, J.L. Santos, O. Frazão, Fabry–Pérot cavities based on chemical etching for high temperature and strain measurement, Opt. Commun. 285 (2012) 1159–1162. doi:10.1016/j.optcom.2011.11.097. 13
A
CC
EP
TE
D
M
A
N
U
SC RI PT
[17] Yuan Gong, Yun-Jiang Rao, Yu Guo, Zeng-Ling Ran, Yu Wu, Temperature-Insensitive Micro Fabry-Perot Strain Sensor Fabricated by Chemically Etching Er-Doped Fiber, IEEE Photonics Technol. Lett. 21 (2009) 1725–1727. doi:10.1109/LPT.2009.2032662. [18] Y. Liu, S. Qu, W. Qu, R. Que, A Fabry-Perot cuboid cavity across the fibre for high-sensitivity strain force sensing, J. Opt. (United Kingdom). 16 (2014). doi:10.1088/2040-8978/16/10/105401. [19] S. Liu, Y. Wang, C. Liao, G. Wang, Z. Li, Q. Wang, J. Zhou, High-sensitivity strain sensor based on in-fiber improved Fabry – Perot interferometer, Opt. Lett. 39 (2014) 2121–2124. [20] a Zhou, B.Y. Qin, Z. Zhu, Y.X. Zhang, Z.H. Liu, J. Yang, L.B. Yuan, Hybrid structured fiber-optic Fabry-Perot interferometer for simultaneous measurement of strain and temperature, Opt. Lett. 39 (2014) 5267–5270. doi:Doi 10.1364/Ol.39.005267. [21] M.S. Ferreira, P. Roriz, J. Bierlich, J. Kobelke, K. Wondraczek, C. Aichele, K. Schuster, J.L. Santos, O. Frazão, Fabry-Perot cavity based on silica tube for strain sensing at high temperatures, Opt. Express. 23 (2015) 16063. doi:10.1364/OE.23.016063. [22] M.S. Ferreira, J. Bierlich, J. Kobelke, K. Schuster, J.L. Santos, O. Frazão, Towards the control of highly sensitive Fabry-Pérot strain sensor based on hollow-core ring photonic crystal fiber, Opt. Express. 20 (2012) 21946. doi:10.1364/OE.20.021946. [23] C. Yin, Z. Cao, Z. Zhang, T. Shui, R. Wang, J. Wang, L. Lu, S. Zhen, B. Yu, TemperatureIndependent Ultrasensitive Fabry-Perot All-Fiber Strain Sensor Based on a Bubble-Expanded Microcavity, IEEE Photonics J. 6 (2014). doi:10.1109/JPHOT.2014.2345883. [24] S. Liu, K. Yang, Y. Wang, J. Qu, C. Liao, J. He, Z. Li, G. Yin, B. Sun, J. Zhou, G. Wang, J. Tang, J. Zhao, High-sensitivity strain sensor based on in-fiber rectangular air bubble, Sci. Rep. 5 (2015) 1–7. doi:10.1038/srep07624. [25] P. Jia, G. Fang, T. Liang, Y. Hong, Q. Tan, X. Chen, W. Liu, C. Xue, J. Liu, W. Zhang, J. Xiong, Temperature-compensated fiber-optic Fabry–Perot interferometric gas refractive-index sensor based on hollow silica tube for high-temperature application, Sensors Actuators, B Chem. 244 (2017) 226–232. doi:10.1016/j.snb.2016.12.123. [26] H.K. Dubey, V. Bhope, Stress and Deflection Analysis of Belleville Spring, IOSR J. Mech. Civ. Eng. 2 (2012) 2278–1684. www.iosrjournals.org. [27] J.R. Jang, ANFIS : Adap tive-Ne twork-Based Fuzzy Inference System, 23 (1993).
14
Pinggang Jia received the Ph.D degree in instrument science and technology from Chongqing University, Chongqing, China, in 2013. He is currently a teacher with North University of China, where he is also a staff of the Micro and Nano technology research center, North University of China. His current research interests include fiber optic sensors, sensor interrogation systems, and laser-based measurement technology, high temperature sensors.
SC RI PT
Guocheng Fang received his master degree from North University of China in 2017. He is currently completing his Ph.D degree in Institute for Biomedical Materials and Devices, Faculty of Science, University of Technology Sydney. His current research interests include optical sensors, MESM technology and point-of-care devices. Zhe Li received his bachelor degree from North University of China in 2015. He is currently completing his master degree in North University of China. His current research interests include fiber-optic sensors, high-temperature sensors and microfabrication technology.
N
U
Hao Liang received his bachelor degree from North University of China in 2015. He is currently completing his master degree in North University of China. His current research interests include fiber-optic sensors, fiber bragg grating sensors and microfabrication technology.
M
A
Yingping Hong receive the Ph.D degree in measurement technology and instrument from the North University of China. Now he is a Lecturer with North University of China. He has been engaged in the research of mechanical parameter in extreme environment, involving pressure testing for high temperature, data recording and transmission.
EP
TE
D
Ting Liang received the Ph.D degree in microelectronics and solid-state electronics from Beijing University of Technology, Beijing, China, in 2007. Now he is an associate professor with North University of China. His current research interests include optical gas sensors, high temperature sensing technology, MEMS technology
A
CC
Jijun Xiong received the Ph.D degree from Tsinghua University, Beijing, China, in 2003. From 2003 to 2005, he was a Post-Doctoral Researcher with Tsinghua University, Beijing, China. He is now a distinguished professor with North university of China. His research interests include the fundamental theory and instrumentation research on dynamic mechanical parameters measurement, mainly focus on the scale effect in silicon microstructures, dynamic measurement, MESM sensing technology, high temperature sensing technology.
15