This paper presents a high-sensitivity, all-silica, all-fiber Fabry-Perot strain-sensor. The proposed sensor provides a long active length, arbitrary length of Fabry-Perot cavity, and low intrinsic temperature sensitivity. The sensor was micro-machined from purposely-developed sensor-forming fiber that is etched and directly spliced to the lead-in fiber. This manufacturing process has good potential for cost-effective, high-volume production. Its measurement range of over 3000 µε, and strain-resolution better than 1 µε were demonstrated by the application of a commercial, multimode fiber-based signal processor.
© 2011 OSA
Optical strain sensors have become one of the more successful and widely-used fiber optic sensing technologies. Electrical passivity, immunity to electromagnetic interferences, dielectric design, chemical inertness, a broad temperature range, small dimensions, and in-line cylindrical geometry are just a few of the properties that make optical-fibers unique for strain-measurement applications. Whilst most of today’s optical point and quasi-distributed strain sensing technologies depend on fiber Bragg gratings (FBGs), which have been at the forefront of strain-sensing technology over the past decade, several limitations such as the need for high-resolution spectral interrogation, high intrinsic temperature sensitivity, limited temperature range and the need for use of single-mode fiber systems, might restrict further FBGs penetration into the broader areas of application.
On the other hand, fiber Fabry–Perot (FP) sensors [1–17], especially short air-cavity FP sensors [3–13], provide various possibilities for cost-effective signal processing [18–25]. Economic signal processors, for example white-light interferometry-based interrogators [2,18,25], are readily-available commercially, and in widespread commercial use. Some of these interrogation techniques  utilize multimode fiber systems that importantly add to the robustness required in non-optic application fields. Furthermore, when spectral interogation is used, the short-cavity FP sensors require considerably lower spectral resolution than comparable FBG systems and are, therefore, less complex to produce. Air-cavity FP sensors can also provide low intrinsic temperature sensitivity [3,4], and a broad operational temperature range.
In spite of these advantages and the already wide-spread use of FP sensors in, for example, pressure measurements [2,26–28], air-cavity FP sensor have made insignificant progress into the strain-measurements application field. The reasons for this can be found in the fact that an FP air-cavity usually determines the active length of the sensor [3–7]. Since an air-cavity length is limited by losses and fringe contrast reduction to a few tens of micrometers, the strain-induced elongation of such a short cavity is insufficient for providing the desired strain-sensitivity during most practical applications.
Therefore, in order to produce sensitive short-cavity FP sensors, the sensor active length must be separated from the FP cavity length. “Fiber-in-capillary” [8–12], and similar approaches  have been reported in the past in order to achieve such designs. Such designs unfortunately compromise the sensor size, stability and tensile strength, and the simplicity of the sensor mounting onto the measurement object. These increased dimensions limit the applicabilities of such sensors in size-critical applications, and impose stringent requirements on the bonding process, in order to ensure high-fidelity (e.g. low-creep) strain transfer from the measured object to the sensor. These sensors are also difficult to produce due to the more complex alignment process required during manufacturing, and the problems encountered when bonding the capillary over the fiber. Direct fusion of the capillary over the fiber compromises the tensile strength of such a bond  (unlike direct fiber to fiber or fiber-to-capillary splicing) and alternative, non-fusion bonding processes [8,9] are, therefore, frequently applied in the production of “fiber-in-capillary” types of sensors. This leads to additional sensor stability and temperature range limitations. The current long active length short FP cavity sensors are, thus, more complex to produce, less robust, and more demanding for use than, for example, all-fiber sensors such as FBGs.
Consequently, despite the many advantages offered by FP sensing technology, FP sensors solutions are infrequently used in optical strain sensing applications. A long active length, short air-cavity FP sensor is required in order to take advantage of FP-sensing technology in strain measurements. Furthermore, such a sensor should have a good high-volume production potential, a small diameter of for example 125 µm, and an all-silica design. The latter is required to achieve broad temperature range, low intrinsic temperature sensitivity, good chemical inertness, and good long-term stability.
This paper presents a sensor concept that meets most of the requirements presented above. In addition, the presented sensor can utilize single-mode or multimode lead-in fiber that opens up additional possibilities for the use of robust and cost-effective signal interrogation.
2 Sensor design and manufacturing
The proposed sensor is shown in Fig. 1 . The sensor is composed of a lead-in fiber that also forms the first FP semi-reflective surface, an outer (semi-conical) wall, a second FP semi-reflective surface, a gutter that surrounds the second FP semi-reflective surface, and a tail section of the sensor (that can be of an arbitrary length).
When the structure in Fig. 1 is exposed to the uniform strain ε along its length, the change in the FP cavity length ΔLc can be expressed as:
A part of the sensor that includes the outer wall, the second semi-reflective surface, the gutter, and the tail section was micro-machined out of a single piece of optical fiber. An efficient way of micromachining such a structure can be obtained by the application of phosphorus pentoxide (P2O5) doping. By doping P2O5 into silica fiber, the etching rate of the silica in hydrofluoric acid (HF) or a similar etching agent, can be significantly increased , which can be further used for the formation of deep structures within the fiber cross-section. We have therefore designed a special strain-sensor forming fiber in order to micro-machine the proposed sensors.
The cross-sectional view of the proposed strain-sensor forming fiber is shown in Fig. 2a and consists of a central region that is doped with TiO2, a pure silica barrier ring, a (strongly) P2O5 doped gutter-forming ring, and an outer cladding. TiO2 doping has a relatively low impact on the etching rate of silica and thus allows for the well-controlled formation of a flat, retracted, centrally-positioned surface, upon exposure to the etching medium. In contrast, the P2O5 doped ring etches at a high-rate and thus forms a deep gutter that surrounds the retracted centrally-positioned (TiO2 doped) flat surface. The scanning electron microscope micrograph and optical image of the sensor-forming fiber after 175 min of etching in a 1:1 mixture of 40% HF and isopropyl alcohol at 10°C are shown in Figs. 2b and 2c (after completion of the etching process, the outer-dimension of the sensor forming fiber was reduced to 125 μm). The addition of isopropyl alcohol to HF reduces the etching medium surface tension, improves the wetting of the gutter region  and thus contributes to the creation of a deeper gutter (about 25% increase in gutter-depth was archived by the addition of isopropyl alcohol to HF). Similarly, reduction in the etching agent temperature improves the etching selectivity of the P2O5 doped region ).
The entire sensor production process thus consists of the few simple, sequential steps shown in Fig. 3 : flat-cleaving of sensor-forming fiber, etching of the sensor-forming fiber for predetermined time, and splicing of the etched-fiber to the lead-in fiber.
The lead-in fiber used during sensor production was standard 50/125 μm-graded index multimode fiber. The splicing was accomplished by a filament fusion splicer (Vytran FSS 2000) using properly-optimized fusion splicing parameters. In spite of thin sensor outer wall, splicing process proved to be stable and repeatable with no detectable influence on the final sensor performance. Splicing process, using FSS 2000 command-language, and splicing parameters used in this investigation are presented in Table 1 .
A typical example of the produced sensor is shown in Fig. 4 . The active length lo of the experimentally produced sensors was 360 µm, whilst the cavity length Lc corresponded to approximately 5 µm. Whilst the sensor cavity-length can be roughly tuned by controlling dopant concentration within sensor forming fiber’s central region, fine (nanometer range) tuning can be accomplished during fusion splicing by the active use of fusion splicer axial motors and feedback from the signal interrogator. For example, by using axial motors, and active feedback during fusion, it is possible to repeatable set the cavity length within a 50 nm range or better.
The experimental strain-sensor forming fiber preform was produced by the conventionally-modified chemical vapour deposition process (MCVD), followed drawing the preform into the fiber. The attributes of the experimentally-produced strain-sensor forming fiber are: a diameter of the central region of 57 µm, content of TiO2 within the central region of 3.4 mol%, thickness of the pure silica barrier ring of 11 µm, thickness of the P2O5 doped gutter forming region of 11 µm, content of P2O5 in the gutter forming region 7.2 mol%, and a fiber outer diameter of 147 µm (after etching this diameter is reduced to 125 µm). In this particular experimental case, we chose the dimensions of the strain-sensor forming fiber in such a way as to allow for the formation of a 125 µm outer-diameter sensor with 50 µm or a greater diameter of the second semi-reflective surface. The later allowed for the use of a 50 µm standard graded index multimode lead-in fiber or any other lead-in fiber with a core smaller or equal to 50 µm.
The purpose of the pure silica barrier ring within the sensor-forming fiber’s cross section is to protect the doped central region from the side (radial) etching. Without implementation of this barrier-ring, the higher etching rate of the central region would shorten the allowable etching-time, that would further result in a shallower gutter. The diameter of the central region should thus determine the diameter of the second semi-reflective FP surface.
It should also be stressed that the fiber attributes of the experimentally-produced fiber used in this investigation were not optimal due to restricted access to fiber manufacturing, that would otherwise be required to fully optimize the sensor-forming fiber (for example, the optimum central region size would be, in this particular case, 50 µm and not 57 µm, the gutter-forming region would probably be narrower, and higher concentration of P2O5 would likely be achieved within the same region).
When the sensor-forming fiber was etched in the controlled etching environment (e.g. at constant temperature, using constant etching time, constant chemical composition of the etching solution and continuous agitation) individual and sequentially etched structures proved to be very repeatable. This allowed for repeatable and consistent production of proposed sensors. For example, it was straightforward to maintain produced sensors active length within ±2 μm tolerance and cavity length within ±0.2 μm tolerance (without fine-tuning during splicing).
Figure 5 shows spectral characteristics of a typical produced sensor. This characteristic was recorded under broadband illumination using tungsten-filament bulb as a source, and multimode lead-in fiber. The recorded spectrum exhibited broad spectral fringes with increasing wavelength period, which is typical for a short FP cavity. The FP cavity had fringe contrast of about 20 dB and recorded spectral peak wavelengths indicated 5.4 µm long cavity.
3 Experimental results
Several experimentally produced sensors were bonded onto the steel and polymethil-meta-acrylate (PMMA) test bars. This bonding was accomplished by a purposely-developed epoxy adhesive for strain-gauge mounting applications (Vishay M-Bond AE-10). The sensors tails (e.g. sensor excess lengths) and the size of the area covered by the bonding material were at least 10 mm long (significantly more than sensor active length). Electrical strain-gauges were also mounted next to the tested fiber-optic sensors in order to provide reference strain measurements. These test bars were then longitudinally strained by calibrated weight loadings. The sensors were interrogated by a multimode fiber-based white-light FP signal interrogator from FISO Technologies Inc . The measured cavity length change versus induced-strain, is shown in Fig. 6 . The cavity length expansion was larger than predicted by Eq. (1), but almost identical for both test bar materials. This discrepancy can be simply explained by the differences in the Young modulus of the bonding material and the fiber that caused strain concentration in the weakened areas of fiber sensors (e.g. areas that consist of thin side-wall). Thus Eq. (1) needs to be adjusted and the correlation between the strain ε and cavity length change ΔLc expressed as:Fig. 6. Figure 6 shows static characteristics for 7 typical produced and mounted sensors (four sensors were mounted on steel and three on PMMA bar). Tested sensors exhibit liner and nearly identical static characteristics with variations of k below 10% from the average value, in the case of the steel bar mounting.
Steel bars were tested for strains up to 1000 µε, which was also the limit of our experimental setup. PMMA bars were tested within the range between 0 and 3000 µε. Cyclical loading and unloading of bars was applied in both cases, and none of the tested sensors failed within these test ranges.
Mounted sensors failed when static tensile strain exceeded about 3500 με −5000 με (these test were performed by using PMMA bars). These failures were, however, did not result in complete signal loss, but they rather reflected in considerable increases in sensor’s static characteristics hysteresis, which likely indicates cracking of sensors embedded in the adhesive. It should be noted that redesign of the sensor-forming fiber might be used to manipulate side wall thickens and thus tensile strength of the sensor.
Figure 7 demonstrates measurement system responses to small changes in the applied strain. The steel test bar, containing a strain-sensor, was cyclically-strained and relaxed for different predetermined strain values. The experimentally achieved resolution using produced strain-sensor and FISO’s signal processor corresponded to about 2 µε at 10 Hz sampling rate and 0.5 µε at a sampling rate corresponding to 1 Hz.
The intrinsic sensor temperature-sensitivity was tested by placing a non-mounted sensor into the temperature-controlled oven. Cavity length change during heating of the sensor up to 650 °C, is shown in Fig. 8 . No damage or optical degradation occurred during this temperature cycle. The cavity-length changed for approximately 0.04 nm/°C, which is equivalent to a temperature strain-sensitivity of 0.055 µε/°C for the gauge factor k=2. This measured temperature-sensitivity is (very) low in comparison to other available strain-sensing technologies. For example, typical (temperature non-compensated) FBG has intrinsic strain-temperature sensitivity of 10 µε/°C@1550nm. Thus in many applications, the intrinsic sensitivity of the proposed sensor can even be neglected, however when the sensor is mounted onto the measured object, the temperature-sensitivity will be mainly governed by the measured object’s thermal expansion, which can be considerable. The intrinsic temperature-sensitivity of the proposed sensor is likely to have originated from several sources: firstly, the thermal expansion of the pure silica outer-wall is slightly different from the region surrounded by the gutter made of doped silica, secondly, the air-cavity expands at a rate that corresponds to the pure silica outer-wall thermal expansion, and finally, a large, strongly-doped, lead-in multimode fiber core has a significantly higher coefficient of thermal expansion, and lower Young modules than pure silica cladding; therefore the core of the lead-in fiber can be deformed (pushed outwards) on the nanometer scale at elevated temperatures. When desired, further reduction of this already-low intrinsic temperature sensitivity could probably be achieved by adjusting the doping (concentration and dopant type) of the central region, and the selection of other lead-in fibers.
This paper presented an all-fiber, miniature, long-active length, short cavity FP strain-sensor. The presented concept could provide an effective solution in those applications requiring cost-efficient signal-processing, robustness of multimode lead-in fibers, broad temperature ranges and/or low intrinsic temperature sensitivity.
The proposed sensor was produced by micromachining process that includes production of a special sensor forming fiber. This sensor-forming fiber was cleaved, etched, and spliced to the lead-in fiber in order to form the final sensor. Since the single strain-sensor forming fiber production cycle usually yields kilometers to tens of kilometers of fiber, which is sufficient to produce 105-108 sensors, and since etching can be performed in large batches, the proposed production technology has good potential for low-cost high volume production.
The presented sensor prototypes were successfully applied to strain-measurements exceeding 3000 µε, which would accommodate most of those requirements encountered in practical industrial applications. A strain-resolution of 0.5 µε @ 1 Hz sampling rate was demonstrated by the application of cost-efficient commercially-available signal interrogator. The intrinsic temperature sensitivity proved to be below 0.04 nm/°C (approximately equivalent to 0.055µε/°C). The sensor was also exposed to temperatures exceeding 650 °C, without observable damage.
There are many possibilities for further optimizing the presented concept for specific applications in terms of active sensor length (sensitivity), sensor diameter, compatibility with available signal-processing schemes, and strain-measurement ranges. This optimization could be achieved through the design of a sensor-forming fiber, and the selection of desired lead-in fiber types. For example, the presented sensor-forming fiber was designed to provide compatibility with 50 µm core multimode lead-in fiber. The reduction of the second semi-reflective surface diameter, for example, to the diameter required to accommodate a single-mode lead-in fiber core, might directly allow for the formation of sensors with longer active lengths (higher strain-sensitivity) or sensors with thicker side walls that could be used in higher strain-range measurements applications. A similar effect could also be achieved by increasing both the sensor’s and lead-in fiber’s outer diameters.
We would like to thank to Borut Lenardic who made the MCVD recipe and supervised the strain-sensor forming fiber production. This work was supported by the Slovenian Public Research agency under grant no. P2-0368, OptaCore Slovenia, and FISO Technologies, Canada.
References and links
1. Y. J. Rao, “Recent progress in fiber-optic extrinsic Fabry–Perot interferometric sensors,” Opt. Fiber Technol. 12(3), 227–237 (2006). [CrossRef]
2. É. Pinet, “Fabry–Perot fiber-optics sensors for physical parameters measurements in challenging conditions,” J. Sensors 2009, Article ID 720980, 9 pages, (2009). [CrossRef]
3. J. Sirkis, T. A. Berkoff, R. T. Jones, H. Singh, A. D. Kersey, E. J. Friebele, and M. A. Putnam, “In-line fiber etalon (ILFE) fiber-optic strain sensors,” J. Lightwave Technol. 13(7), 1256–1263 (1995). [CrossRef]
7. Y. J. Rao, M. Deng, D. W. Duan, X. C. Yang, T. Zhu, and G. H. Cheng, “Micro Fabry–Perot interferometers in silica fibers machined by femtosecond laser,” Opt. Express 15(21), 14123–14128 (2007). [CrossRef] [PubMed]
9. T. A. Tran, W. V. Miller, K. A. Murphy, A. M. Vengsarkar, and R. O. Claus, “Stabilized Extrinsic Fiber-Optic Fizeau Sensor for Surface Acoustic Wave Detection,” J. Lightwave Technol. 10(10), 1499–1506 (1992). [CrossRef]
10. J. A. Etches and G. F. Fernando, “Evaluation of embedded optical fiber sensors in composites: EFPI sensor fabrication and quasi-static evaluation,” Polym. Composite. 30(9), 1265–1274 (2009). [CrossRef]
12. Y. Zhang, Y. Li, T. Wei, X. Lan, Y. Huang, G. Chen, and H. Xiao, “Fringe visibility enhanced extrinsic Fabry–Perot interferometer using a graded index fiber collimator,” IEEE Photon. J. 2(3), 469–481 (2010). [CrossRef]
13. A. Sakamoto and J. Nishii, “Fiber Fabry–Perot interferometer with precision glass-ceramic jacketing,” IEEE Photon. Technol. Lett. 17(7), 1462–1464 (2005). [CrossRef]
14. Q. Shi, F. Y. Lv, Z. Wang, L. Jin, J. J. Hu, Z. Liu, G. Kai, and X. Dong, “Environmentally stable Fabry–Perot-type strain sensor based on hollow-core photonic bandgap fiber,” IEEE Photon. Technol. Lett. 20(4), 237–239 (2008). [CrossRef]
16. Y. Gong, Y. J. Rao, Y. Guo, Z. L. Ran, and Y. Wu, “Temperature-insensitive micro Fabry–Perot strain sensor fabricated by chemically etching Er-doped fiber,” IEEE Photon. Technol. Lett. 21(22), 1725–1727 (2009). [CrossRef]
17. T. Zhao, Y. Gong, Y. Rao, Y. Wu, Z. Ran, and H. Wu, “Fiber-optic Fabry–Perot strain sensor based on graded-index multimode fiber,” Chin. Opt. Lett. 9(5), 050602–050605 (2011). [CrossRef]
19. Y. Jiang, “Fourier-transform phase comparator for the measurement of extrinsic Fabry–Perot interferometric sensors,” Microw. Opt. Technol. Lett. 50(10), 2621–2625 (2008). [CrossRef]
20. Y. L. Lo, J. S. Sirkis, and C. C. Chang, “Passive signal processing of in-line fiber etalon sensors for high strain-rate loading,” J. Lightwave Technol. 15(8), 1578–1586 (1997). [CrossRef]
21. A. J. Willshire, P. Niewczas, and J. R. McDonald, “An arrayed waveguide grating based multiplexer and interrogator for Fabry–Perot sensors,” IEEE Sens. J. 5(5), 964–969 (2005). [CrossRef]
22. B. Yu, A. B. Wang, and G. R. Pickrell, “Analysis of fiber Fabry–Perot interferometric sensors using low-coherence light sources,” J. Lightwave Technol. 24(4), 1758–1767 (2006). [CrossRef]
23. M. Schmidt, B. Werther, N. Fuerstenau, M. Matthias, and T. Melz, “Fiber-optic extrinsic Fabry–Perot interferometer strain sensor with <50 pm displacement resolution using three-wavelength digital phase demodulation,” Opt. Express 8(8), 475–480 (2001). [CrossRef] [PubMed]
24. F. B. Shen and A. B. Wang, “Frequency-estimation-based signal-processing algorithm for white-light optical fiber Fabry–Perot interferometers,” Appl. Opt. 44(25), 5206–5214 (2005). [CrossRef] [PubMed]
26. Y. Z. Zhu and A. B. Wang, “Miniature fiber-optic pressure sensor,” IEEE Photon. Technol. Lett. 17(2), 447–449 (2005). [CrossRef]
28. S. Pevec, E. Cibula, B. Lenardic, and D. Donlagic, “Micromachining of optical fibers using selective etching based on phosphorus pentoxide doping,” IEEE Photon. J. 3(4), 627–632 (2011). [CrossRef]
29. J. Kochan, T. Wintgens, R. Hochstrat, and T. Melin, “Impact of wetting agents on the filtration performance of polymeric ultrafiltration membranes,” Desalination 241(1-3), 34–42 (2009). [CrossRef]