1. Introduction

Fiber Bragg grating (FBG) based optical sensors fabricated by methods including phase mask, interferometric and direct-writing technologies, and by using ultraviolet (UV) or near-infrared (IR) femtosecond laser as the source have been studied extensively [1–6]. Among them, the UV laser exposure with phase-mask technique has gained the most interest and been more widely used owing to the inherent photosensitivity of the doped silica glass to UV light [7]. Compared with the interferometric or direct-writing technique for making FBGs, the phase-mask technique is more robust and reproducible which is highly advantageous for mass production. In recent years, various grating structures with multi-resonance excitation features provide an opportunity to enhance the environmental medium monitoring (referring to the refractive index change), and compared to fiber interferometric sensors [8–10] the FBG-based sensing system shows many advantages such as repeatable fabrication, spectral stability and multi-parameter measurement. For example, fiber tapers or discontinuities are combined with standard FBG in order for light to be coupled to cladding modes along the mismatched transitions modes [11–13]. These hybrid FBG structures have been demonstrated to measure surrounding refractive index (SRI) [11,14,15], bending [16,17] and other parameters [18] that a standard FBG usually fails to do. The other kind of asymmetric grating structures such as tilted [18,19], off-axis [20–22] and highly-localized [23–25] FBG have also been widely exploited, and the azimuthal symmetry breaking properties of such gratings will be used to achieve similar measurements with a single FBG. Specifically, tilted FBG (TFBG) couples light from the core to the cladding at a multitude of wavelengths (depending on tilt angle and grating period) and the coupling strength depends on the overlap between the modes involved as well as the tilted grating planes, and because of that, it always requires nontrivial angular modifications of standard FBG writing tools. Off-axis FBG with UV radiation can be fabricated with the conventional FBG writing setup and results in coupling to cladding modes due to the departure from pure cylindrical symmetry, but it confines relatively weak and lower-order resonances near the core mode that usually not very sensitive to refractive index [26]. In this respect, the improved off-axis TFBG written by femtosecond laser has been recently reported that extends the spectral coverage of the cladding modes [27]. Highly-localized FBG written by ultrafast laser also exhibits wide cladding mode resonance range that is suitable for lower SRI measurement, while as, much more expensive lasers and precise 3D alignment and positioning of the focal point of the laser pattern are required in IR femtosecond FBG fabrication.

Taking into account that higher order modes are generally suppressed in UV inscribed FBG [28], here we propose a simple method to fabricate a conventional FBG accompanied by a set of strong cladding modes excitation with a normal process of ultraviolet radiation through a phase mask by employing an ultrathin SMF with a 50 µm diameter cladding. Several thin [29–31] or thinned-processed fibers [32,33] have been reported recently that can improve the sensing sensitivities of the grating-based device such as SRI, temperature and strain, et.al. Without applying sophisticated cladding thinning methods such as grinding and chemical etching which may decrease the fiber strength [34], the ultrathin SMF is much more flexible and highly bend-resistant that can be embedded into structures/laminates or biological organization for robustness and noninvasiveness. In this paper, we theoretically and experimentally verify the occurrence of strong and widely separated cladding resonances in a spectral region where refractive index sensing is very important, and that are closer to cut-off and hence more sensitive than the corresponding resonances in a normal cladding fiber. We investigate the improved refractive index sensitivity of the sensor as well as the strain and temperature characters by tracing different cladding and core resonances. The experiment results indicate that this sensor therefore cannot only measure the basic parameters such as temperature and strain just as the conventional FBG, but also the wide cladding mode excitation enables it to sense the change of environmental parameters such as SRI. The proposed approach differs strongly from previous approaches, even that of Ref. 31 where a FBG in a similar thin fiber with a 25 µm diameter was used, because the current fiber allows for strong coupling to much lower order cladding modes providing high sensitivity in media with refractive indices down to 1.3. This is the simplest possible fiber Bragg grating device for enhanced SRI measurement without the need for complex symmetry-breaking fabrication methods.

2. Grating fabrication and characterizations

Figure 1(a) shows the schematic diagram of the designed ultrathin FBG. The employed thin cladding fiber (Fibercore SM1500) has a core/cladding diameter of approximately 3.8/50 µm with a high numerical aperture of 0.3. This ultrathin SMF was hydrogen-loaded at 10 MPa for 2 weeks in order to increase the photo-sensitivity (the laser intensity threshold for FBG formation is significantly reduced as a consequence) [35]. The standard optical setup for UV laser writing based on the scanning phase mask technique was employed for grating fabrication. A frequency doubled continuous wave Ar⁺ laser of 244 nm wavelength was used as the light source, and an uniform phase mask with a period of 1072 nm was selected to generate inference patterns in the whole fiber core. Under these conditions a uniform FBG are formed involving a core mode and a comb of cladding modes that induced by coupling between the forward propagating core mode and backward-propagating modes along the cladding (shown by the colorized arrows in Fig. 1(a)). Following grating fabrication, the fiber was annealed at 80 °C for 48 hours to release the residual hydrogen and stabilized the grating structure. As shown in Fig. 1(b), the grating fringes in the fiber core were inspected with a high high-resolution optical microscope to show an axial grating period of ∼536 nm. Visible red light was then launched into the ultrathin FBG giving rise to a clear side diffraction pattern with a length of 16 mm, as shown in Fig. 1(c). The ultrathin SMF was spliced with two standard SMFs (SM-28), thereby facilitating the connection with other fiber components (the jacketed ultrathin fiber is actually structurally robust away from the stripped region where the grating is fabricated, so the whole system is reliable as long as the splices are protected with the usual methods). Figure 1(d) shows the image of the splicing point between the thin cladding fiber and the SMF, obtained with a commercial fusion splicer (Fujikura FSM-60S). The estimated splicing loss was less than 0.15 dB by adjusting the arc discharge power of “standard -40 bit”, time of 800 ms and discharge location of “R-30 µm” (offset to SMF). As shown in Fig. 2, a super-continuum source (OYSL, SC-5) was launched into the grating, and an optical spectrum analyzer (OSA) (Yokogawa, AQ6370) with a resolution of 0.01 nm was used to collect the spectrum.

 

Fig. 1. (a) Schematic configuration of the designed ultrathin FBG. (b) Photomicrograph of the fabricated ultrathin FBG. (c) Side diffraction patterns with a visible light insertion. (d) Microscopic image of the splicing point.

Download Full Size | PDF

 

Fig. 2. Schematic of experimental setup.

Download Full Size | PDF

The corresponding reflection and transmission spectra are shown in Fig. 3. We can observe a strong Bragg reflection as well as a series of resonances over a broad spectral bandwidth from 1420 nm to 1557 nm, indicating the efficient coupling to cladding modes. And most particularly, the attenuation of the cladding modes was larger than -15 dB near 1545 nm. The vector modes with azimuthal orders of 1 were considered in the simulation below due to their maximum coupling coefficients for FBG, and the electric field distributions in the fiber cross section of some excited cladding modes at different resonating wavelengths are shown in the lower panel of Fig. 3. For comparison, cladding modes in a standard FBG similarly fabricated were generally less than -1 dB in attenuation and mainly concentrated near the Bragg resonance [25]. Furthermore, the spectral spacing in wavelength between different cladding modes of the designed FBG is increased to over 2 nm. In addition to a natural larger spacing between modes of a thinner fiber, the coupling in a uniform FBG does not break the cylindrical symmetry and only involves cladding modes of the same azimuthal order as that of the core (i.e. order 1). The increased spacing increases the range of unambiguous readings for the positions of individual resonance relative to its reference position. The closest comparable FBG reported previously was written in a 25 µm cladding diameter fiber but the smaller diameter resulted in only six measurable cladding mode resonances with attenuations lower than -5 dB [31]. The fiber used here is widely available commercially, easier to handle and to splice to regular fiber, and provides resonances over a much larger wavelength (and mode effective index) range.

 

Fig. 3. Measured transmission and reflection spectra of the fabricated ultrathin FBG by UV exposure. Mode couplings appear as numerous loss peaks in transmission, but only the Bragg (core mode) reflection appears in the reflected spectrum. Low level broadband noise in the reflection spectrum is due to the broadband supercontinuum light source used. The electric field distributions of the four vector cladding modes that have separated locations are simulated with a three-layer optical fiber model (inset).

Download Full Size | PDF

Simulations were carried out based on the full-vector complex coupled mode theory to verify the measured spectra of the inscribed ultrathin FBG [36], as shown in Fig. 4. Here the refractive indices of fiber core and cladding were considered as 1.4749 and 1.444, respectively, which leads to the numerical aperture equal to about 0.3. As is known the strongest interaction occurs when the wave vectors satisfy the phase match condition (PMC):${K_{in}} = {K_{out}} + {K_G}$ between a forward propagating core mode and one backward propagating core mode plus several back-propagating cladding modes. The resulting relation between the phase matching wavelength and the grating period is ${\lambda _{\textrm{co}}} = 2n_{\textrm{eff}}^{\textrm{co}}\Lambda $ for the core mode (“Bragg” resonance) and for cladding modes it becomes ${\lambda _{\textrm{clad,} m}} = (n_{\textrm{eff}}^{\textrm{co}} + n_{\textrm{eff}}^m)\Lambda $, where $n_{\textrm{eff}}^{\textrm{co}}$ and $n_{\textrm{eff}}^m$ are the effective refractive indices of the core mode and m^th HE/EH_1,m (m = 1,2,3…) cladding mode, respectively. It is clear in Fig. 4(a) that the period line intersects with the core mode and numerous HE/EH_1,m cladding modes at the considered wavelength range, which indicates the mode coupling and hence the resonances in the transmission spectrum. Some specific cases are illustrated in Figs. 4(b) and 4(c), from which it is important to note that higher order modes (larger m value) excited at shorter wavelength are more widely separated than those of lower order modes at longer wavelength. This makes it possible to attain dual resonances consisting of pairs of HE and EH mode resonances at shorter wavelengths while the HE and EH resonances tend to overlap at longer wavelengths in the spectrum of the ultrathin FBG.

 

Fig. 4. (a) Phase-matching condition for mode coupling. (b) and (c) Specific cases at shorter and longer wavelengths, respectively. Some modes are marked with the mode number for illustration.

Download Full Size | PDF

The comparison between simulated and experimental transmission spectra is shown in Fig. 5. Here the core index modulation of the ultrathin FBG was set to be 3.5×10⁻⁴ to obtain the best theoretical fit. It is clear that a high agreement in terms of both locations and strengths of the resonance bands between simulation and experiment results is obtained. As shown in the insets, dual resonances at shorter wavelength and single resonances at longer wavelength are observed in both simulated and experimental spectra, which is remarkably consistent with the PMC in Fig. 4. Compared with the experimental spectrum in Fig. 3 we can also see that the marked four cladding resonances (that will be used for comparison in the following experiments) clearly correspond to the resonances of the 5^th, 10^th, 16^th, and 18^th HE/EH cladding mode. The remaining small difference in the resonance wavelength is caused by the use of a constant refractive index for the fiber core and cladding in the simulations, thus neglecting dispersion over the wavelengths considered and UV-induced average index changes resulting from the fabrication of the FBG, whereas, the real ultrathin optical fiber appears a wavelength-dependent dispersion determined by composite materials of fiber core and cladding [37].

 

Fig. 5. Simulated and experimental spectra. The insets show spectra at shorter and longer wavelengths, respectively.

Download Full Size | PDF

3. Experiments and discussion

The spectral response of the ultrathin FBG with SRI variation was first investigated. A series of index oils (from Cargille Labs) with refractive indices range from 1.300 to 1.456 (values at 589 nm) were applied to the grating. The two sides of the FBG were fixed very straight on two translation stages, and methanol was used between each index measurement to clean the residual refractive index liquid on the fiber surface. The experiment results are shown in Fig. 6. It can be seen that when immersing the fiber in liquid oil a well identified cut-off mode, i.e. a sudden decrease in the amplitudes of the individual cladding mode resonance (marked with an asterisk) is observed. This can be used to provide an estimate of the liquid refractive index since the cut-off condition corresponds to a mode with an effective index equal to the liquid index. With increasing SRI, the wavelength of the cut-off mode red-shifts until all the cladding resonances disappear when the index oil reaches the index value of silica. The inset in Fig. 6 displays a linear fitting curve for the cut-off mode wavelength corresponding to each SRI, yielding a sensitivity of ∼490.2 nm/RIU, in good agreement with TFBG refractometry using normal cladding fiber [38].

 

Fig. 6. Spectral evolution of the ultrathin FBG immersed in air and liquids with increasing refractive indices. The asterisks indicate the signal of the cut-off mode. The inset shows extracted SRI-induced cut-off wavelength shifts.

Download Full Size | PDF

Four individual cladding resonances located at 1497.86 nm, 1509.53 nm, 1537.19 nm and 1551.39 nm (previously identified in Fig. 3) were then selected for comparative measurements of SRI sensitivity based on their wavelength shifts. As mentioned above, these dips have m orders corresponding to 5, 10, 16 and 18, respectively, and thus will show different responses to external medium. Figure 7(a) shows the transmission spectra of the typical 10^th HE/EH cladding mode located ∼28 nm away from the Bragg resonance (λ_B) in external media from 1.300 to 1.420. It can be seen that the cladding wavelength shifts more and more rapidly as the RI increases until totally disappears, and the total wavelength drift is achieved up to 1.05 nm. This value is two times larger than that of the cladding resonance that has the same separation from λ_B in a TFBG written in conventional fiber (and therefore the same m order) [39]. Figure 7(b) further shows the fitting results of the wavelength shifts of the marked 5^th, 10^th, 16^th, and 18^th cladding mode resonances relative to λ_B when the ultrathin FBG immersed in different liquids. It confirms that higher order modes have relatively higher SRI sensitivities. For all these resonances near cut-off (evaluating the first derivative of the fitting curve there) the sensitivity is larger than 10 nm/RIU. Especially for the 18^th cladding mode, the maximum sensitivity obtained is 28.3 nm/RIU at 1.35, which is more than twice that of the 10° TFBG (∼11.2 nm/RIU at 1.31 in Ref. [39]). The RI sensing performance is thus seen to equal or even exceed that of TFBGs and other non-conventional grating sensor devices while benefiting from the ease of fabrication and low cost of standard FBG technology, with some additional advantages of wider spacing between resonances, higher bending tolerance and minimal disruption in embedded applications.

 

Fig. 7. (a) Transmission spectra of individual resonance near 1537 nm (for the 10^th cladding mode) and (b) wavelength shifts of the selected 5^th, 10^th, 16^th, 18^th cladding modes relative to core mode λ_B in various external media.

Download Full Size | PDF

Secondly, the strain characteristic of the sensor was studied by increasing the distance between the FBG fixed points with one of the translation stages. Similarly to all fiber gratings, positive linear strain along the fiber axis leads to a redshift of the spectrum due to the change of the axial grating period. This red shift is strongest for the Bragg resonance and increasingly smaller for resonances at larger spectral distance from the Bragg [15]. Figure 8 shows the experiment results for the wavelength shifts of the ultrathin FBG under axial strain from 0 to 1600 µɛ and the corresponding strain sensitivities of the Bragg resonance and of four cladding resonances. As seen in Fig. 8, the calculated strain sensitivities are 1.11 pm/µɛ for the core mode and 1.11 pm/µɛ, 1.07 pm/µɛ, 1.00 pm/µɛ and 0.95 pm/µɛ for 5^th,10^th, 16^th,18^th cladding modes, respectively, which confirms that the red-shift is gradually decreasing for the higher order modes as in normal fibers. It is also clear that the magnitude of the strain sensitivity in general is almost the same as that of FBGs and TFBGs in normal SMF. This is because the dominant source of the shift is the change in period due to the strain (dΛ/dɛ) instead of the size of the fiber cross-section. It must be pointed out however that the axial force sensitivity is different (a thinner fiber strains more than a thicker one, for the same force applied) [31]. For the same reason, a thinner fiber will also be able to support much larger strains before breaking than a thicker one.

 

Fig. 8. Wavelength shifts of the selected modes in ultrathin FBG versus the applied axial strain.

Download Full Size | PDF

Finally, a similar set of experiments was carried out to quantify the cross sensitivity of the measurements to temperature change by placing the grating in a temperature-controlled oven that was set to increase the temperature from room temperature to 90 °C. Figure 9 shows that for this case, the wavelength shifts of 5^th, 10^th, 16^th and 18^th cladding mode follow similar linear trends with average slopes from 10.41 pm/°C to 10.67 pm/°C, and for core mode the temperature sensitivity is 12.03 pm/°C, which is relatively higher than others. Further tracking the relative shift between the core and 18^th mode resonance it yields a temperature sensitivity of the differential wavelength shift equal to 1.63 pm/°C, which is more than triple the response of TFBG that is 0.35 pm/°C in Ref. [40]. These differences in temperature sensitivity are likely due to the fact that this parameter is dominated by the thermo-optic coefficient of the fiber core, which in turn depends on the Germanium doping level of the core, and that the fiber used here had a larger Germanium concentration to increase the numerical aperture. It must be mentioned that the known temperature sensitivity of the Bragg and cladding mode resonances can be used to remove temperature effects from RI measurements since the Bragg resonance is totally insensitive to RI changes.

 

Fig. 9. Temperature responses of the core and the cladding modes in ultrathin FBG spectrum.

Download Full Size | PDF

4. Conclusion

In summary, FBGs inscribed in ultrathin fiber with the conventional UV laser and phase mask technology have been produced that exhibit efficient high order cladding modes coupling. The transmitted spectrum of the ultrathin FBG has been characterized and the refractive index sensing characteristics has been investigated experimentally. Unlike standard FBGs, and interestingly FBGs written in even thinner cladding fibers, a fiber with cladding diameter of 50 µm results in the excitation of cladding modes with a wide spectral span over 140 nm so that the grating is capable of measuring much lower values of refractive index with high sensitivity. Individual cladding mode resonances showed highly improved refractive index sensitivities up to 28.3 nm/RIU (near the cut-off). We also have shown that the strain sensitivities of different cladding resonances for the proposed grating are approximately 1 pm/µɛ, mainly determined by the change in grating period due to axial strain and are impervious to the smaller size of the fiber. In addition, the temperature sensitivity of the differential wavelength shift between the Bragg and cladding resonance has been increased by 3 times from 0.35 to 1.63 pm/°C. These results indicate that the FBGs UV-inscribed in 50 µm diameter SMF with slightly increased core doping provide a different combination of multi-parameter sensitivities that may prove important in certain technologies, especially those where fibers are embedded in materials, due to their smaller diameter, higher strain tolerance, and improved SRI and temperature sensitivities.