1. Introduction

Fiber Bragg gratings (FBGs) are well known for their use in sensing physical parameters such as temperature, strain, and pressure [1, 2 ]. An attractive feature is that FBGs with different pitches can be written at different points along the length of the optical fiber, with each FBG producing a discrete narrowband reflection that can be wavelength division multiplexed. Depending on the particular design and application, FBG sensors also offer high accuracy and precision, high speed interrogation, and reflection mode operation.

Fiber Bragg gratings fabricated with the traditional technique of ultra-violet (UV) inscription in photosensitive silica-based fibers have a maximum operating temperature limit of approximately 500°C due to thermal annealing out of the refractive index modifications above this temperature [2, 3 ]. Developing grating-based sensors that can measure high temperatures, such as 1000°C and beyond, is an area of active research. Sapphire crystal fibers with femto-second laser written Bragg gratings are perhaps the leading technology with sensing up to 1900°C demonstrated [4, 5 ]. However, such fibres are highly multimode with broad Bragg reflections. It is also not possible to splice sapphire crystal fibers to conventional silica fiber. For applications up to 1300°C a number of fabrication techniques have been developed to inscribe gratings in silica-based fibers, such as chiral gratings [6], regenerated fiber Bragg gratings [7, 8 ], and femto-second laser written gratings [9, 10 ].

Chiral gratings are fabricated by rotating and translating an elliptical core fiber through a short heat zone [6]. These fibers have been demonstrated as temperature sensors up to 1000°C and are commercially available. Alternatively, regenerated fiber Bragg gratings are fabricated by thermally annealing a “seed” grating, such as a standard type I UV inscribed grating. These gratings have been shown to survive to temperatures up to 1100°C [8]. In one particular example a regenerated FBG was shown to survive up to 1295°C, however the fiber itself was shown to become extremely brittle after reaching such temperatures [7]. It should also be noted that the fabrication of such gratings requires treating the fiber with high pressure hydrogen for time periods on the order of 24 hours and gratings prepared in this manner have poorer reflectivity compared to the seed gratings [8].

A promising approach to fabricate optical fiber temperature sensors up to temperatures the order of 1300°C is the use of undoped single-material (e.g. pure fused silica) optical fibers, such as microstructured optical fibers, which avoid the potential for dopant diffusion at high temperatures [9]. To create an FBG in an undoped fiber requires techniques that do not rely on glass photosensitivity. To create a high temperature stable grating a technique known as femtosecond laser ablation can be used where by using sufficiently short duration pulses, multiphoton absorption can lead to ionization of electrons and the physical removal (ablation) of material [11, 12 ]. However, a particular challenge of microstructured optical fibers is that the complex cross section distorts and scatters the inscription beam [11].

In this paper we demonstrate the inscription of fiber Bragg gratings on the core of pure-silica suspended-core microstructured optical fibers (SCFs) using femtosecond laser ablation. The simple geometry of the SCF (three large holes) allows the direct writing of FBGs as the focused femtosecond laser beam is not significantly distorted when passing through the fiber cladding. We demonstrate that these can survive up to 1300°C and can be wavelength division multiplexed. Here we present three sensing elements operating in a single device.

2. Sensor fabrication

The optical fiber used was silica (Suprasil F300, Heraeus) suspended-core microstructured optical fiber [13], with an outer diameter of 160 µm [Fig. 1(a) ] and a core diameter of approximately 10 µm [Fig. 1(b)]. These fibers were then spliced to standard FC/APC connectorized single-mode fiber (SMF28) using an arc splicer (Fujikura FSM-100P) and the same technique previously determined for exposed-core microstructured optical fiber (MOF) [14]. That is, standard SMF-SMF splicer settings but with reduced arc current (12.5 mA compared to 16.5 mA), increased arc duration (3.0 s compared to 2.0 s), and manual alignment.

 

Fig. 1 Scanning electron images (SEMs) of the suspended-core fiber with a femtosecond laser ablation grating. (a) The fiber cross section. (b, c) magnified images of (a).

Download Full Size | PDF

Bragg gratings were written into the SCFs by adapting a technique previously developed for exposed-core MOFs [14, 15 ]. That is, the FBGs were written by focusing 800 nm femtosecond laser (Hurricane Ti:sapphire) pulses using a 50X long working distance microscope objective onto the core of the fiber. A repetition rate of 100 Hz was used and the SCF was translated to yield second order Bragg reflections at the desired wavelength (e.g. 1550 nm). In comparison to the exposed-core MOF, for the SCF the fs laser pulses must traverse across the fiber cladding and thus some distortion of the pulses was expected. This can be seen in Figs. 1(b) and 1(c), where the ablation points overlap. However, there was sufficient periodicity in the gratings that were formed in order to measure a defined Bragg reflection, as will be seen in the proceeding sections. In addition, there were several technical considerations that were introduced by using the enclosed structure. Firstly, the pulse energy required was greater: 400 nJ pulses were used compared to 200-250 nJ for open structures. It was also difficult to determine the location of the optical fiber core and thus a visible laser was coupled to the spliced single-mode fiber so that scattering from the ablation spots could be observed as a confirmation of the femto-second laser focusing position.

3. High temperature sensing

To test the operating temperature limit of the sensors a fiber as prepared in Sec. 2 was first packaged within a silica tube (approx. 2 mm outer diameter (OD)) for physical protection. The packaged sensor was then inserted into the graphite resistance furnace of a silica optical fiber draw tower. The length of the microstructured optical fiber was sufficiently long such that the spliced single-mode fiber was located well outside of the heat zone of the furnace. The particular grating used had a pitch of 1080 nm and a length of 20 mm. The resulting reflected spectrum at room temperature is shown in Fig. 2(a) , measured using an Optical Sensor Interrogator (OSI, National Instruments PXIe-4844). The reflected spectrum shows a peak reflectivity at 1556 nm, a full-width at half maximum of 48 pm, and a peak reflectivity greater than 30 dB above the background. The reflected spectrum shows peaks from higher order modes, as expected, however the fundamental mode reflectivity was approximately 8.5 dB above that of the higher order modes (at room temperature) and can further be improved by using SMF with a better mode field match to the fundamental mode of the SCF.

 

Fig. 2 Reflected spectra measured from the SCF Bragg grating at increasing temperatures within a silica fiber draw tower. The spectra shown are at temperatures (a) 20°C, (b) 500°C, (c) 1000°C, (d) 1200°C, (e) 1300°C, and (f) 1350°C. For temperatures of 400°C and above the temperature was recorded with a co-located B-type thermocouple.

Download Full Size | PDF

To measure the temperature experienced at the grating a B-type thermocouple was located at the top point of the 20 mm long grating, noting that the spliced connection (i.e. the light path) was directed through the top of the furnace. The furnace set-point was initially set at 850°C and at this temperature the location of the grating was adjusted so that the top point of the grating was at the hottest point within the furnace. The furnace was then increased in temperature in steps of 50°C and then held for five minutes. The spectra recorded at several points are shown in Figs. 2(b)-2(f). The results in Fig. 2 show the expected shift to longer wavelengths at higher temperatures. In addition, the reflected spectrum was found to be stable for the duration of the hold time (5 minutes) up to 1300°C. Once the temperature was increased beyond this temperature (1350°C) the grating was found to decrease in reflectivity rapidly. The peak reflectivity reduced by 2.0 dB in the first 5 minutes and 8.6 dB after one hour, in addition to a change in the spectral shape [Fig. 2(f)]. The grating did not recover after returning to room temperature, indicating permanent damage to the sensor. The temperature at which the sensor failed (between 1300°C and 1350°C) lies significantly above the annealing point of fused silica (1100°C [16]), but also well below the softening point (1600°C [16]). At a temperature of 1300°C fused silica (F300) glass has viscosity of 10^9.87 Pa.s (calculated using Eq. (2) in [13]), which is close to the dilatometric softening temperature (10¹⁰ Pa.s) [17]. This is the temperature at which a sample (e.g. a glass rod) with an external load will no longer expand in length for increasing temperature due to sample dilation and deformation. This implies that deformation happens on the micron-scale and thus micron-scaled holes such as those created by femto-second ablation can (at least partially) collapse, destroying the grating.

This experiment has successfully demonstrated the ability of femtosecond laser ablation gratings in suspended-core microstructured optical fibers to measure temperature up to 1300°C. A limitation of this experiment is that the furnace used has a short hot zone that varies by approximately 100°C over the length of the grating (20 mm) at 1300°C [13]. This leads to the broadening of the reflected spectrum that can be seen in Fig. 2 and a reduction in the maximum reflectivity. Experiments using either a furnace with a more even temperature distribution or by using gratings of shorter length will allow the performance limits of these devices to be explored more fully.

4. Calibration and multiplexed sensing

The results in Fig. 2 are insufficient to form a calibration curve due to the uneven temperature distribution of the silica draw tower furnace used for these tests. In order to create a calibration curve a second sensor was inserted centrally into a tube furnace (ModuTemp) and a K-type thermocouple was located at the center of the Bragg grating. The temperature of the tube furnace was raised to 1000°C and held there for several hours until both the thermocouple and the FBG showed stable readings, at which point the temperature profile of the furnace was deemed stable. The temperature was then reduced in steps of 100°C and held for at least one hour to ensure the furnace temperature was stable before a spectrum was recorded. The shift in the position of the fundamental mode reflection relative to room temperature is shown in Fig. 3 . A quadratic function was used to form the calibration curve due to the nonlinear thermo-optic response of the sensor. This is likely due to a combination of the silica thermo-optic coefficient and the mode confinement within the suspended-core [11]. Note that the calibration curve is valid for a particular set of parameters such as fiber geometry, material, and grating structure. In the following experiment we have assumed the impact of grating pitch, and thus the Bragg wavelength, to be negligible for this proof-of-concept demonstration. This is an approximation as material and waveguide dispersion will cause a slight change to the calibration curve at different wavelengths.

 

Fig. 3 Calibration curve measured using a SCF Bragg grating in a tube furnace. The shift in temperature (ΔT) and shift in wavelength (Δλ) are relative to values at room temperature. The quadratic fit was set to zero at the origin.

Download Full Size | PDF

A wavelength division multiplexed sensor was fabricated by writing three physically separated gratings into a suspended-core fiber using the same technique described in Sec. 2. In this case the gratings were made 10 mm long and three different pitches were used (Λ = 1055.7 nm, 1069.6 nm, 1083.5 nm). The gap between the gratings was 25 mm and 17 mm, respectively. The fiber was then spliced to conventional single mode fiber and packaged within a 2.0 mm OD silica tube. The packaged sensor was inserted into a soft glass fiber draw tower (graphite induction furnace), with the spliced connection passed through the bottom of the furnace. The reflected spectrum recorded at room temperature is shown in Fig. 4(a) . Note that the lower grating had the strongest reflection, indicating optical loss across the gratings.

 

Fig. 4 (a) Reflected spectrum from the multiplexed sensor at room temperature. *Indicates the peaks that were tracked. (b) Change in temperature recorded by the FBGs, corresponding to the peaks indicated in (a) and calibrated using the results in Fig. 3.

Download Full Size | PDF

Prior to the experiment we determined which reflections corresponded to each grating by locally heating the gratings separately and observing which reflection peaks shift in wavelength, which is indicated by the colored bands in Fig. 4(a). For the temperature sensing experiment only the fundamental mode reflections were tracked, which are indicated with asterisks. The furnace set-point was then raised in steps of 100°C up to 800°C and, using the calibration shown in Fig. 3, the results are shown in Fig. 4(b). Note that the temperatures measured by the gratings (up to 525°C) were less than 800°C because the furnace’s thermocouple is located in a different physical location (within the susceptor) compared to the FBG sensor (center line of the susceptor). The measured difference of approximately 275°C between the susceptor’s center line (FBG sensor) and the furnace’s thermocouple agrees with measurements we have previously made using a thermocouple within a silica tube at the center line of the susceptor.

While the maximum temperature tested in this particular experiment was only 525°C, the results of Fig. 4 demonstrate that a suspended-core microstructured optical fiber femtosecond laser ablation temperature sensor can be successfully multiplexed with up to three elements.

5. Discussion and conclusions

We have demonstrated temperature sensing up to 1300°C using femtosecond laser ablation gratings within silica suspended-core microstructured optical fiber. This relatively simple fiber geometry has been used as the femtosecond laser beam is not significantly distorted when passing through the cladding structure. The use of silica glass allows for direct splicing to conventional single-mode fiber, thus easy integration with commercial interrogators, and provides the possibility of long length sensors due to low propagation loss. The maximum operating temperature of the sensor was found to lie between 1300°C and 1350°C. Further research needs to be conducted to determine the exact maximum operating temperature limit of the sensor. The long term stability of the sensor at these temperatures also needs to be investigated further as the current study was limited to five minute intervals for each temperature measured.

Wavelength division multiplexed (WDM) sensing has been demonstrated, with three temperature sensing elements shown here. One limitation in the number of WDM sensors is the multi-mode nature of the suspended-core fiber used for this work, leading to several peak reflections per grating. The wavelength separation of the gratings must be large enough to accommodate the modes (approximately 12 nm) plus the shift over the desired wavelength range (14 nm for 1000°C). For an interrogator bandwidth of 80 nm this limits the number of sensors to three. The multiplexing capability can be improved through further designing the suspended-core MOF to reduce the number of propagating modes. The most obvious way of doing this is by reducing the core diameter. Improvements in the grating inscription process, such as the use of optimized laser pulse energy and filling the MOF air hole with near index-matching liquid to allow the writing beam to focus better in the core region, should also help reduce the optical loss observed in our multiplexed sensing experiment.

An additional parameter that requires further optimization is the grating length. Profiling of short-length furnaces, such as in fiber drawing towers, would benefit from better spatial resolution than can be achieved using the 10-20 mm gratings used here. To reduce grating length whilst maintaining reflectivity a stronger index modulation is required. One way to achieve this could be to, again, reduce the core diameter in order to increase the overlap of the propagating modes with the ablation points. However, careful design optimization is required as smaller core diameters may lead to increased FBG inscription and splicing difficulty.