November 1, 2014 / Vol. 39, No. 21 / OPTICS LETTERS
6221
Miniature fiber-optic sensor for simultaneous measurement of pressure and refractive index Simon Pevec and Denis Donlagic* University of Maribor, Faculty of EE & Computer Science, Smetanova 17, SI-2000 Maribor, Slovenia *Corresponding author: ddonlagic@uni‑mb.si Received September 2, 2014; revised October 3, 2014; accepted October 5, 2014; posted October 6, 2014 (Doc. ID 222361); published October 23, 2014 This Letter presents a fiber-optic sensor created at the tip of an optical fiber for simultaneous measurements of pressure and refractive index. The sensor diameter does not exceed the standard fiber diameter and is shorter than 300 μm. Measurement resolutions of 0.2 mbar and 2 × 10−5 RIU were demonstrated experimentally by using spectral interrogation and Fourier-transform-based measurement algorithms (interrogation system bandwidth corresponded to 1 Hz). A micromachining process based on selective chemical etching of specially designed phosphorus-doped fibers, and a sequence of splice and cleave steps were used to fabricate the sensor. © 2014 Optical Society of America OCIS codes: (060.2370) Fiber optics sensors; (230.4000) Microstructure fabrication; (120.2230) Fabry-Perot; (120.5710) Refraction; (120.5475) Pressure measurement; (070.0070) Fourier optics and signal processing. http://dx.doi.org/10.1364/OL.39.006221
Growing man-made system complexities and rising levels of automation are driving the demand for increasing numbers of physical and/or chemical parameters that need to be measured or monitored within various systems. Reductions in sensor sizes and minimizations of sensor connection lines (i.e., number of optical fibers in the cases of fiber-optic sensors) are vital for microfluidic, industrial, biomedical, and other similar applications. Multiparameter [1–14] fiber-optic sensors created at the tip or along a single fiber have thus been intensively investigated during the recent past. The choices of parameters that can be simultaneously measured are, however, limited by available sensor designs and manufacturing technologies. Most of the reported multiparameter fiber-optic sensors are dual-parameter sensors, which are capable of measuring temperatures in combination with another physical parameter; for example, refractive index [3,4], pressure [5–7], strain [8–11], displacement [12,13], or other [14]. While the temperature is an important and frequently demanded measurement parameter, the extending of dual-parameter miniature fiber-optic sensor designs into combination of measurement parameters involving other parameters than just temperature are proving to be a challenge. The temperature measurement capability is usually added straightforwardly to an existing sensor design by incorporating additional grating [13] or an all-fiber Fabry–Perot [5] cavity at the end of the lead-in fiber. Sensors that can measure multiple parameters, excluding the temperature, require more complex sensor designs, which are often difficult to manufacture, especially in a miniaturized form. This Letter presents a highly sensitive dual-parameter sensor for simultaneous measurements of refractive index (RI) and pressure within a dimension, which do not exceed the standard fiber diameter. Pressure and RI are typical fluidic parameters that are often measured in various industrial systems. The proposed sensor is produced by the selective chemical etching (SCE) micromachining process [15], which involves simple fiber cleave-splice-etch steps, which has good production potential. 0146-9592/14/216221-04$15.00/0
The proposed sensor design is shown in Fig. 1. The sensor consists of a standard lead-in single-mode fiber (SMF), the first disk-shaped spacer, two parallel silica supporters, the second disk-shaped spacer, a cylindrical spacer, and a thin silica diaphragm. The end face of the first disk-shaped spacer, two parallel silica supporters, and the back surface of the second disk-shaped spacer form an open-path Fabry–Perot cavity/microcell (the front surface of the second disk-shaped spacer was made intentionally uneven and nonflat to minimize backreflection). The structure is further coated by a thin layer of TiO2 (about 50 nm thick, n 2.45), which provides considerable enhancement of RI contrast between optical surfaces and most liquids of practical interest. TiO2 is also straightforward for deposition and assures good chemical inertness and biocompatibility. The surrounding fluid can thus freely enter the microcell to provide the fluid’s RI through measurement of the microcell’s optical path length. The back surface of the second diskshaped spacer, the cylindrical spacer, and an inner surface of the thin flexible silica diaphragm form another Fabry–Perot cavity, in which length is pressure sensitive. The pressure and RI sensing FP cavities have initial lengths of 155 and 90 μm, respectively. These lengths are made differently to allow for effective sensor interrogation, as described further below. The entire structure is made of silica glass and is fusion spliced to form a chemically and temperature-robust assembly.
Fig. 1. Proposed RI pressure sensor design. © 2014 Optical Society of America
6222
OPTICS LETTERS / Vol. 39, No. 21 / November 1, 2014
Sensor fabrication utilizes a micromachining process based on the SCE of phosphorus-doped optical fibers [15]. The sensor fabrication includes the production of two subassemblies (designated by 1 and 2 in Fig. 2). The first subassembly fabrication is started by splicing a coreless fiber to a flat cleaved lead-in SMF and cleaving it 20 μm away from the splice to form the first diskshaped spacer [Fig. 2(a)]. In the next step, a specially designed phosphorus-doped microcell-forming fiber (MFF), with the cross section as shown in Fig. 3(a), is spliced to the first disk-shaped spacer and cleaved 150 μm away from the splice [Fig. 2(b)] (a similar process was used in [16] to produce microcells). This step is followed by splicing a standard 62.5 μm telecom multimode fiber (MMF) to the free end of MFF and cleaving it 55 μm away from the last splice [Fig. 2(c)]. This step concludes the preparation of the first subassembly. A flat cleaving of the coreless fiber starts the second subassembly preparation. This coreless fiber is then spliced to a specially designed cylindrical-spacer-forming fiber (CSFF), which is then cleaved 90 μm away from this splice [Fig. 2(d)]. This structure is further etched for about 3 min in 40% hydrofluoric acid (HF) at 25°C to selectively remove the entire phosphorus-doped section [Fig. 2(e)] (the phosphorus-doped silica glass etches about 40 times faster than pure silica in the particular case of CSFF, which contained about 9.6 mol. % of P2 O5 ). This technique allows for the formation of a cylindrical spacer with a very thin wall that later allows for the formation of a large diaphragm pressure sensor. This concludes the preparation of the second subassembly. The first and the second subassemblies are then spliced together [Fig. 2(f)]. Finally, the coreless fiber is cleaved about 60 μm away from the cavity [Fig. 2(g)] and polished down to a thickness below 5 μm [Fig. 2(h)]. The final setting of the pressure sensor diaphragm thickness is achieved by controlled etching, as described in [17]; thus, the sensor
Fig. 2. Sensor production sequence.
assembly is connected to the signal interrogation system and exposed to HF within a closed vessel, where we cyclically vary the pressure [Fig. 2(i)]. During etching, the backreflected spectra is continuously acquired and processed in order to obtain cavity-length readings. This allows for determining the sensor’s pressure sensitivity (and, consequently, the diaphragm thickness). When the target pressure sensitivity has been achieved (400 nm/bar), the sensor is removed from the HF and neutralized. This allows for calibrations of the pressure sensors during the production process while achieving high-pressure sensitivity. An RI sensing microcell is also formed during the above-described diaphragm formation etching process. Upon exposure of MFF to HF, HF first removes the pure silica layers of MFF. When HF comes into contact with the elliptical phosphorus-doped region, it removes this region at a considerably higher rate than the pure silica regions to form the RI measurement FP cavity. Thus, during this final etching step, the diaphragm and the microcell structures are etched simultaneously. This simultaneous etching approach eliminates the need for any masking. However, it requires careful tunings of structure-forming fiber dimensions and diaphragm thicknesses before etching. The diaphragm thickness is thus checked by a pressure test before etching (the sensor is connected to the interrogation system and pressurized). Further, the MFF elliptical P2 O5 -doped core is designed in a way that the doped region approaches the outer boundary of the fiber as far as the minimal possible distance (to 1–2 μm). This distance still provides easy fiber drawing and fiber splicing (highly P2 O5 -doped region has low viscosity that affects both processes), while it minimizes the etching time required for the microcell formation (P2 O5 -doped core is removed at about a 55 times higher rate than pure silica in the case of the used MFF). This allows for entire microcell formation within a time that is less than the time required for pure-silica diaphragm thickness adjustment. Further, CSFF dimensions are selected to yield a cylindrical spacer with sidewalls of about 14 μm, which is about the minimum thickness that guarantees survival of the cylinder spacer over the time required for diaphragm/microcell etching. While the selection of a thicker cylinder spacer wall would be feasible, a thicker wall is not desirable since it leads to smaller diaphragm size and, consequently, to reduced sensor pressure sensitivity. The exact geometry of both sensor-forming fibers used for the sensors’ experimental production is given in Fig. 3. Finally, the chosen role and designs of both diskshaped spacers deserve further explanation: the first disk-shaped spacer protects the SMF front surface from HF during etching. As explained above, a microcell is
Fig. 3. Structure-forming fibers under optical microscope: (a) MFF; (b) CSFF.
November 1, 2014 / Vol. 39, No. 21 / OPTICS LETTERS
formed over a shorter period of time than in the average time required to fine-tune the diaphragm thickness. This leaves microcell side surfaces exposed to HF for an extra amount of time, and, during this time, the germaniumdoped SMF core would preferentially etch (germaniumdoped silica is etched faster than pure silica in HF). This would cause a fiber’s end-face surface degradation with poor optical reflectivity. The inserted coreless spacer etches uniformly and retains reasonable surface flatness that can thus act as a semi-reflective mirror even after prolonged HF exposure. The second disk-shaped spacer separates RI and pressure measurement cavities by serving as a reference optical surface. In order to avoid the formation of two spatially separated semi-reflective surfaces (one on each side of the disk, which would yield additional components within the FT of the spectrum that could further lead to cross-talk generation), MMF is used during the production of the second disk. As the first surface of the second disk is also exposed to the HF for an extra amount of time, as described above, this surface becomes curved due to the SCE of a graded multimode core, which minimizes the disk’s first surface reflectance to a negligible level. An example of the produced sensor is shown in Fig. 4. Figure 4(a) shows the scanning electron microscope (SEM) image of the sensor; Fig. 4(b) shows the same sensor under an optical microscope; Fig. 4(c) shows the second disc’s first (exposed) surface. Figure 5(a) shows the backreflected optical spectrum as acquired by the signal interrogator (sensor was immersed in water). Interrogation of the sensor was performed by a National Instrument NI PXIe-4844 spectral interrogation unit, which provides optical spectrum sampling within a 10 Hz rate. The acquired spectra were further processed online by a custom-developed LabVIEW code, which further averaged 10 acquired samples before displaying/storing the result (which yielded the final sampling rate of 1 Hz for both RI and pressure measurements). Figure 5(b) depicts the amplitude FT of the same
Fig. 4. Fabricated sensor: (a) SEM image; (b) optical image; (c) front surface of the second disk-shaped spacer.
6223
Fig. 5. (a) Sensor’s backreflected optical spectrum. (b) FT of spectrum multiplied by c/2.
spectrum (after applying Gaussian window to the original spectrum). The x axis is multiplied by c/2 to display cavity lengths. The first peak in the FT of the optical spectrum represents the pressure measurement cavity (which has an optical length of 90 μm), and the second peak corresponds to the RI measurement cavity (optical length in water 290 μm). Figure 6 shows the phase change of the component in the FT of the backreflected spectrum that corresponds to the RI measurement cavity length when we varied the RI of the surrounding fluid over the 1.320–1.325 RIU range. The response is linear with a sensitivity of 7.1 × 104 degrees/RIU (equivalent of 830 nm/RIU). A well-defined backreflected spectrum and high spectral sensitivity also provide opportunities for achieving high measurement resolution. This is demonstrated in Fig. 6(b), where we immersed the sensor in a water-glycerol solution (9∶1 ratio) followed by multiple additions of small amounts of water and glycerol to raise and lower the RI of the fluid for the same value. A resolution better than 2 × 10−5 RIU can be clearly observed in the presented figure. Peaks/ notches appearing in Fig. 6(b) are RI transients caused by local injections of small amounts of glycerin/water into the test solution, i.e., injected liquid needs several seconds to mix with the test solution in order to reach an equilibrium concentration across the entire test volume. Similar tests were also performed on the pressure sensor. Figure 7(a) shows the phase change of the component within the FT of the backreflected spectrum that corresponded to the pressure measurement cavity when we varied the pressure of the surrounding fluid over 0– 1.7 bar. This test was also repeated regarding four fluids with different RIs and demonstrated the absence of any significant pressure-RI measurement cross sensitivity (all four pressure readings coincided, regardless of RI value). Figure 7(b) represents a pressure-sensor resolution demonstration (10 samples averaged at 10 Hz optical
Fig. 6. (a) Sensor response to RI change. (b) RI resolution demonstration (at 1 Hz interrogation system bandwidth).
6224
OPTICS LETTERS / Vol. 39, No. 21 / November 1, 2014
Fig. 8. Cross talk: (a) pressure reading error as a function of RI change; (b) RI reading error as a function of applied pressure.
Fig. 7. (a) Sensor response to pressure change in fluids with different RI. (b) Pressure resolution demonstration (at 1 Hz interrogation system bandwidth).
spectrum interrogation). The sensor was raised and dropped by a few millimeters in a test tube filled with water to induce small changes in hydrostatic pressure. Resolution within the range of better than 0.1 mbar can be observed from the presented experiment. Higher unambiguous pressure ranges at lower sensitivities also can be achieved by increasing diaphragm thickness (by limiting etching time, for example). Cross sensitivity between both parameters was also directly evaluated/measured. We varied RI, at constant pressure, over the entire unambiguous measurement range (i.e., for 5 × 10−3 RIU), while observing the change in pressure reading [Fig. 8(a)]. The total change in pressure reading corresponded to 0.44 mbar. Similarly, changing pressure over the 1.7 bar range (full unambiguous range) while keeping the RI constant caused change in the RI reading of about 4.2 × 10−5 RIU [Fig. 8(b)]. Both cross sensitivities were low, i.e., near the system resolution range. It should be stressed that this cross talk mainly arises from the overlapping of individual time components in the FT of the backreflected optical spectrum and could be likely further reduced by making both cavity lengths more dissimilar or by increasing the interrogator wavelength sweep range. This Letter presented a miniature, all-silica dualparameter pressure RI sensor. This sensor was created at the tip of an optical fiber with a diameter that is equal to the standard fiber diameter and length that does not exceed 300 μm. High measurement resolutions (better than 0.1 mbar and 2 × 10−5 RIU at 1 Hz) were demonstrated experimentally by using spectral interrogation and an FT-based measurement algorithm. The sensor has low intrinsic temperature sensitivity (provided by an allsilica design). While the pressure measuring range can be set by tuning the diaphragm’s thickness and is independent of the RI measurement part; the RI measurement range (which in our case corresponded to 5 × 10−3 RIU) depends on the RI measuring cavity’s free spectral range.
Shorter RI measurement cavities can provide a broader unambiguous measurement range, but at the cost of lower resolution (due to the broader spectral fringes). RI and pressure measurement cavities should, however, be sufficiently different in their lengths to allow for their interrogation with low cross talk. This particular sensor was fabricated by the application of a micromachining process based on SCE of specially designed phosphorusdoped fibers and a sequence of spliced cleave steps. This work was supported by the Slovenian Research Agency (grant nos. P2-0368 and L2-5494) and CORE@UM program sponsored by European Commission under the ERDF. Optical fibers were manufactured by Optacore d.o.o. from Slovenia. References 1. X. K. Zeng and Y. J. Rao, Chin. Phys. Lett. 18, 1617 (2001). 2. S. M. Lee, S. S. Saini, and M. Y. Jeong, IEEE Photon. Technol. Lett. 22, 1431 (2010). 3. D. W. Kim, F. Shen, X. P. Chen, and A. B. Wang, Opt. Lett. 30, 3000 (2005). 4. S. Pevec and D. Donlagic, Opt. Express 22, 16241 (2014). 5. S. Pevec and D. Donlagic, Appl. Opt. 51, 4536 (2012). 6. K. Bremer, E. Lewis, G. Leen, B. Moss, S. Lochmann, and I. A. R. Mueller, IEEE Sens. J. 12, 133 (2012). 7. H. D. Bae, D. Yun, H. J. Liu, D. A. Olson, and M. Yu, J. Lightwave Technol. 32, 1585 (2014). 8. Y. Zhan, L. Li, F. Yang, K. Gu, H. Wu, and M. Yu, OptoElectron. Rev. 21, 283 (2013). 9. W. L. Liu, W. Z. Li, and J. P. Yao, IEEE Photon. Technol. Lett. 23, 1340 (2011). 10. A. Zhou, B. Qin, Z. Zhu, Y. Zhang, Z. Liu, J. Yang, and L. Yuan, Opt. Lett. 39, 5267 (2014). 11. O. Frazao, L. M. Marques, S. Santos, J. M. Baptista, and J. L. Santos, IEEE Photon. Technol. Lett. 18, 2407 (2006). 12. Y. L. Yu, H. Y. Tam, W. H. Chung, and M. S. Demokan, Opt. Lett. 25, 1141 (2000). 13. L. A. Ferreira, A. B. L. Ribeiro, J. L. Santos, and F. Farahi, IEEE Photon. Technol. Lett. 8, 1519 (1996). 14. D. A. Pereira, O. Frazao, and J. L. Santos, Opt. Eng. 43, 299 (2004). 15. S. Pevec, E. Cibula, B. Lenardic, and D. Donlagic, IEEE Photon. J. 3, 627 (2011). 16. D. Donlagic, Opt. Lett. 36, 3148 (2011). 17. D. Donlagic and E. Cibula, Opt. Lett. 30, 2071 (2005).
6221
Miniature fiber-optic sensor for simultaneous measurement of pressure and refractive index Simon Pevec and Denis Donlagic* University of Maribor, Faculty of EE & Computer Science, Smetanova 17, SI-2000 Maribor, Slovenia *Corresponding author: ddonlagic@uni‑mb.si Received September 2, 2014; revised October 3, 2014; accepted October 5, 2014; posted October 6, 2014 (Doc. ID 222361); published October 23, 2014 This Letter presents a fiber-optic sensor created at the tip of an optical fiber for simultaneous measurements of pressure and refractive index. The sensor diameter does not exceed the standard fiber diameter and is shorter than 300 μm. Measurement resolutions of 0.2 mbar and 2 × 10−5 RIU were demonstrated experimentally by using spectral interrogation and Fourier-transform-based measurement algorithms (interrogation system bandwidth corresponded to 1 Hz). A micromachining process based on selective chemical etching of specially designed phosphorus-doped fibers, and a sequence of splice and cleave steps were used to fabricate the sensor. © 2014 Optical Society of America OCIS codes: (060.2370) Fiber optics sensors; (230.4000) Microstructure fabrication; (120.2230) Fabry-Perot; (120.5710) Refraction; (120.5475) Pressure measurement; (070.0070) Fourier optics and signal processing. http://dx.doi.org/10.1364/OL.39.006221
Growing man-made system complexities and rising levels of automation are driving the demand for increasing numbers of physical and/or chemical parameters that need to be measured or monitored within various systems. Reductions in sensor sizes and minimizations of sensor connection lines (i.e., number of optical fibers in the cases of fiber-optic sensors) are vital for microfluidic, industrial, biomedical, and other similar applications. Multiparameter [1–14] fiber-optic sensors created at the tip or along a single fiber have thus been intensively investigated during the recent past. The choices of parameters that can be simultaneously measured are, however, limited by available sensor designs and manufacturing technologies. Most of the reported multiparameter fiber-optic sensors are dual-parameter sensors, which are capable of measuring temperatures in combination with another physical parameter; for example, refractive index [3,4], pressure [5–7], strain [8–11], displacement [12,13], or other [14]. While the temperature is an important and frequently demanded measurement parameter, the extending of dual-parameter miniature fiber-optic sensor designs into combination of measurement parameters involving other parameters than just temperature are proving to be a challenge. The temperature measurement capability is usually added straightforwardly to an existing sensor design by incorporating additional grating [13] or an all-fiber Fabry–Perot [5] cavity at the end of the lead-in fiber. Sensors that can measure multiple parameters, excluding the temperature, require more complex sensor designs, which are often difficult to manufacture, especially in a miniaturized form. This Letter presents a highly sensitive dual-parameter sensor for simultaneous measurements of refractive index (RI) and pressure within a dimension, which do not exceed the standard fiber diameter. Pressure and RI are typical fluidic parameters that are often measured in various industrial systems. The proposed sensor is produced by the selective chemical etching (SCE) micromachining process [15], which involves simple fiber cleave-splice-etch steps, which has good production potential. 0146-9592/14/216221-04$15.00/0
The proposed sensor design is shown in Fig. 1. The sensor consists of a standard lead-in single-mode fiber (SMF), the first disk-shaped spacer, two parallel silica supporters, the second disk-shaped spacer, a cylindrical spacer, and a thin silica diaphragm. The end face of the first disk-shaped spacer, two parallel silica supporters, and the back surface of the second disk-shaped spacer form an open-path Fabry–Perot cavity/microcell (the front surface of the second disk-shaped spacer was made intentionally uneven and nonflat to minimize backreflection). The structure is further coated by a thin layer of TiO2 (about 50 nm thick, n 2.45), which provides considerable enhancement of RI contrast between optical surfaces and most liquids of practical interest. TiO2 is also straightforward for deposition and assures good chemical inertness and biocompatibility. The surrounding fluid can thus freely enter the microcell to provide the fluid’s RI through measurement of the microcell’s optical path length. The back surface of the second diskshaped spacer, the cylindrical spacer, and an inner surface of the thin flexible silica diaphragm form another Fabry–Perot cavity, in which length is pressure sensitive. The pressure and RI sensing FP cavities have initial lengths of 155 and 90 μm, respectively. These lengths are made differently to allow for effective sensor interrogation, as described further below. The entire structure is made of silica glass and is fusion spliced to form a chemically and temperature-robust assembly.
Fig. 1. Proposed RI pressure sensor design. © 2014 Optical Society of America
6222
OPTICS LETTERS / Vol. 39, No. 21 / November 1, 2014
Sensor fabrication utilizes a micromachining process based on the SCE of phosphorus-doped optical fibers [15]. The sensor fabrication includes the production of two subassemblies (designated by 1 and 2 in Fig. 2). The first subassembly fabrication is started by splicing a coreless fiber to a flat cleaved lead-in SMF and cleaving it 20 μm away from the splice to form the first diskshaped spacer [Fig. 2(a)]. In the next step, a specially designed phosphorus-doped microcell-forming fiber (MFF), with the cross section as shown in Fig. 3(a), is spliced to the first disk-shaped spacer and cleaved 150 μm away from the splice [Fig. 2(b)] (a similar process was used in [16] to produce microcells). This step is followed by splicing a standard 62.5 μm telecom multimode fiber (MMF) to the free end of MFF and cleaving it 55 μm away from the last splice [Fig. 2(c)]. This step concludes the preparation of the first subassembly. A flat cleaving of the coreless fiber starts the second subassembly preparation. This coreless fiber is then spliced to a specially designed cylindrical-spacer-forming fiber (CSFF), which is then cleaved 90 μm away from this splice [Fig. 2(d)]. This structure is further etched for about 3 min in 40% hydrofluoric acid (HF) at 25°C to selectively remove the entire phosphorus-doped section [Fig. 2(e)] (the phosphorus-doped silica glass etches about 40 times faster than pure silica in the particular case of CSFF, which contained about 9.6 mol. % of P2 O5 ). This technique allows for the formation of a cylindrical spacer with a very thin wall that later allows for the formation of a large diaphragm pressure sensor. This concludes the preparation of the second subassembly. The first and the second subassemblies are then spliced together [Fig. 2(f)]. Finally, the coreless fiber is cleaved about 60 μm away from the cavity [Fig. 2(g)] and polished down to a thickness below 5 μm [Fig. 2(h)]. The final setting of the pressure sensor diaphragm thickness is achieved by controlled etching, as described in [17]; thus, the sensor
Fig. 2. Sensor production sequence.
assembly is connected to the signal interrogation system and exposed to HF within a closed vessel, where we cyclically vary the pressure [Fig. 2(i)]. During etching, the backreflected spectra is continuously acquired and processed in order to obtain cavity-length readings. This allows for determining the sensor’s pressure sensitivity (and, consequently, the diaphragm thickness). When the target pressure sensitivity has been achieved (400 nm/bar), the sensor is removed from the HF and neutralized. This allows for calibrations of the pressure sensors during the production process while achieving high-pressure sensitivity. An RI sensing microcell is also formed during the above-described diaphragm formation etching process. Upon exposure of MFF to HF, HF first removes the pure silica layers of MFF. When HF comes into contact with the elliptical phosphorus-doped region, it removes this region at a considerably higher rate than the pure silica regions to form the RI measurement FP cavity. Thus, during this final etching step, the diaphragm and the microcell structures are etched simultaneously. This simultaneous etching approach eliminates the need for any masking. However, it requires careful tunings of structure-forming fiber dimensions and diaphragm thicknesses before etching. The diaphragm thickness is thus checked by a pressure test before etching (the sensor is connected to the interrogation system and pressurized). Further, the MFF elliptical P2 O5 -doped core is designed in a way that the doped region approaches the outer boundary of the fiber as far as the minimal possible distance (to 1–2 μm). This distance still provides easy fiber drawing and fiber splicing (highly P2 O5 -doped region has low viscosity that affects both processes), while it minimizes the etching time required for the microcell formation (P2 O5 -doped core is removed at about a 55 times higher rate than pure silica in the case of the used MFF). This allows for entire microcell formation within a time that is less than the time required for pure-silica diaphragm thickness adjustment. Further, CSFF dimensions are selected to yield a cylindrical spacer with sidewalls of about 14 μm, which is about the minimum thickness that guarantees survival of the cylinder spacer over the time required for diaphragm/microcell etching. While the selection of a thicker cylinder spacer wall would be feasible, a thicker wall is not desirable since it leads to smaller diaphragm size and, consequently, to reduced sensor pressure sensitivity. The exact geometry of both sensor-forming fibers used for the sensors’ experimental production is given in Fig. 3. Finally, the chosen role and designs of both diskshaped spacers deserve further explanation: the first disk-shaped spacer protects the SMF front surface from HF during etching. As explained above, a microcell is
Fig. 3. Structure-forming fibers under optical microscope: (a) MFF; (b) CSFF.
November 1, 2014 / Vol. 39, No. 21 / OPTICS LETTERS
formed over a shorter period of time than in the average time required to fine-tune the diaphragm thickness. This leaves microcell side surfaces exposed to HF for an extra amount of time, and, during this time, the germaniumdoped SMF core would preferentially etch (germaniumdoped silica is etched faster than pure silica in HF). This would cause a fiber’s end-face surface degradation with poor optical reflectivity. The inserted coreless spacer etches uniformly and retains reasonable surface flatness that can thus act as a semi-reflective mirror even after prolonged HF exposure. The second disk-shaped spacer separates RI and pressure measurement cavities by serving as a reference optical surface. In order to avoid the formation of two spatially separated semi-reflective surfaces (one on each side of the disk, which would yield additional components within the FT of the spectrum that could further lead to cross-talk generation), MMF is used during the production of the second disk. As the first surface of the second disk is also exposed to the HF for an extra amount of time, as described above, this surface becomes curved due to the SCE of a graded multimode core, which minimizes the disk’s first surface reflectance to a negligible level. An example of the produced sensor is shown in Fig. 4. Figure 4(a) shows the scanning electron microscope (SEM) image of the sensor; Fig. 4(b) shows the same sensor under an optical microscope; Fig. 4(c) shows the second disc’s first (exposed) surface. Figure 5(a) shows the backreflected optical spectrum as acquired by the signal interrogator (sensor was immersed in water). Interrogation of the sensor was performed by a National Instrument NI PXIe-4844 spectral interrogation unit, which provides optical spectrum sampling within a 10 Hz rate. The acquired spectra were further processed online by a custom-developed LabVIEW code, which further averaged 10 acquired samples before displaying/storing the result (which yielded the final sampling rate of 1 Hz for both RI and pressure measurements). Figure 5(b) depicts the amplitude FT of the same
Fig. 4. Fabricated sensor: (a) SEM image; (b) optical image; (c) front surface of the second disk-shaped spacer.
6223
Fig. 5. (a) Sensor’s backreflected optical spectrum. (b) FT of spectrum multiplied by c/2.
spectrum (after applying Gaussian window to the original spectrum). The x axis is multiplied by c/2 to display cavity lengths. The first peak in the FT of the optical spectrum represents the pressure measurement cavity (which has an optical length of 90 μm), and the second peak corresponds to the RI measurement cavity (optical length in water 290 μm). Figure 6 shows the phase change of the component in the FT of the backreflected spectrum that corresponds to the RI measurement cavity length when we varied the RI of the surrounding fluid over the 1.320–1.325 RIU range. The response is linear with a sensitivity of 7.1 × 104 degrees/RIU (equivalent of 830 nm/RIU). A well-defined backreflected spectrum and high spectral sensitivity also provide opportunities for achieving high measurement resolution. This is demonstrated in Fig. 6(b), where we immersed the sensor in a water-glycerol solution (9∶1 ratio) followed by multiple additions of small amounts of water and glycerol to raise and lower the RI of the fluid for the same value. A resolution better than 2 × 10−5 RIU can be clearly observed in the presented figure. Peaks/ notches appearing in Fig. 6(b) are RI transients caused by local injections of small amounts of glycerin/water into the test solution, i.e., injected liquid needs several seconds to mix with the test solution in order to reach an equilibrium concentration across the entire test volume. Similar tests were also performed on the pressure sensor. Figure 7(a) shows the phase change of the component within the FT of the backreflected spectrum that corresponded to the pressure measurement cavity when we varied the pressure of the surrounding fluid over 0– 1.7 bar. This test was also repeated regarding four fluids with different RIs and demonstrated the absence of any significant pressure-RI measurement cross sensitivity (all four pressure readings coincided, regardless of RI value). Figure 7(b) represents a pressure-sensor resolution demonstration (10 samples averaged at 10 Hz optical
Fig. 6. (a) Sensor response to RI change. (b) RI resolution demonstration (at 1 Hz interrogation system bandwidth).
6224
OPTICS LETTERS / Vol. 39, No. 21 / November 1, 2014
Fig. 8. Cross talk: (a) pressure reading error as a function of RI change; (b) RI reading error as a function of applied pressure.
Fig. 7. (a) Sensor response to pressure change in fluids with different RI. (b) Pressure resolution demonstration (at 1 Hz interrogation system bandwidth).
spectrum interrogation). The sensor was raised and dropped by a few millimeters in a test tube filled with water to induce small changes in hydrostatic pressure. Resolution within the range of better than 0.1 mbar can be observed from the presented experiment. Higher unambiguous pressure ranges at lower sensitivities also can be achieved by increasing diaphragm thickness (by limiting etching time, for example). Cross sensitivity between both parameters was also directly evaluated/measured. We varied RI, at constant pressure, over the entire unambiguous measurement range (i.e., for 5 × 10−3 RIU), while observing the change in pressure reading [Fig. 8(a)]. The total change in pressure reading corresponded to 0.44 mbar. Similarly, changing pressure over the 1.7 bar range (full unambiguous range) while keeping the RI constant caused change in the RI reading of about 4.2 × 10−5 RIU [Fig. 8(b)]. Both cross sensitivities were low, i.e., near the system resolution range. It should be stressed that this cross talk mainly arises from the overlapping of individual time components in the FT of the backreflected optical spectrum and could be likely further reduced by making both cavity lengths more dissimilar or by increasing the interrogator wavelength sweep range. This Letter presented a miniature, all-silica dualparameter pressure RI sensor. This sensor was created at the tip of an optical fiber with a diameter that is equal to the standard fiber diameter and length that does not exceed 300 μm. High measurement resolutions (better than 0.1 mbar and 2 × 10−5 RIU at 1 Hz) were demonstrated experimentally by using spectral interrogation and an FT-based measurement algorithm. The sensor has low intrinsic temperature sensitivity (provided by an allsilica design). While the pressure measuring range can be set by tuning the diaphragm’s thickness and is independent of the RI measurement part; the RI measurement range (which in our case corresponded to 5 × 10−3 RIU) depends on the RI measuring cavity’s free spectral range.
Shorter RI measurement cavities can provide a broader unambiguous measurement range, but at the cost of lower resolution (due to the broader spectral fringes). RI and pressure measurement cavities should, however, be sufficiently different in their lengths to allow for their interrogation with low cross talk. This particular sensor was fabricated by the application of a micromachining process based on SCE of specially designed phosphorusdoped fibers and a sequence of spliced cleave steps. This work was supported by the Slovenian Research Agency (grant nos. P2-0368 and L2-5494) and CORE@UM program sponsored by European Commission under the ERDF. Optical fibers were manufactured by Optacore d.o.o. from Slovenia. References 1. X. K. Zeng and Y. J. Rao, Chin. Phys. Lett. 18, 1617 (2001). 2. S. M. Lee, S. S. Saini, and M. Y. Jeong, IEEE Photon. Technol. Lett. 22, 1431 (2010). 3. D. W. Kim, F. Shen, X. P. Chen, and A. B. Wang, Opt. Lett. 30, 3000 (2005). 4. S. Pevec and D. Donlagic, Opt. Express 22, 16241 (2014). 5. S. Pevec and D. Donlagic, Appl. Opt. 51, 4536 (2012). 6. K. Bremer, E. Lewis, G. Leen, B. Moss, S. Lochmann, and I. A. R. Mueller, IEEE Sens. J. 12, 133 (2012). 7. H. D. Bae, D. Yun, H. J. Liu, D. A. Olson, and M. Yu, J. Lightwave Technol. 32, 1585 (2014). 8. Y. Zhan, L. Li, F. Yang, K. Gu, H. Wu, and M. Yu, OptoElectron. Rev. 21, 283 (2013). 9. W. L. Liu, W. Z. Li, and J. P. Yao, IEEE Photon. Technol. Lett. 23, 1340 (2011). 10. A. Zhou, B. Qin, Z. Zhu, Y. Zhang, Z. Liu, J. Yang, and L. Yuan, Opt. Lett. 39, 5267 (2014). 11. O. Frazao, L. M. Marques, S. Santos, J. M. Baptista, and J. L. Santos, IEEE Photon. Technol. Lett. 18, 2407 (2006). 12. Y. L. Yu, H. Y. Tam, W. H. Chung, and M. S. Demokan, Opt. Lett. 25, 1141 (2000). 13. L. A. Ferreira, A. B. L. Ribeiro, J. L. Santos, and F. Farahi, IEEE Photon. Technol. Lett. 8, 1519 (1996). 14. D. A. Pereira, O. Frazao, and J. L. Santos, Opt. Eng. 43, 299 (2004). 15. S. Pevec, E. Cibula, B. Lenardic, and D. Donlagic, IEEE Photon. J. 3, 627 (2011). 16. D. Donlagic, Opt. Lett. 36, 3148 (2011). 17. D. Donlagic and E. Cibula, Opt. Lett. 30, 2071 (2005).
