The propagation loss of EHC PBGF is measured by

The propagation loss of EHC-PBGF is measured by the cutback method. The transmission spectra of EHC-PBGF with lengths of 0.6 and 1.2m are obtained firstly using the same splicing conditions, as shown in Fig. 2. Then the propagation loss versus wavelength is obtained by subtraction and normalization as shown in Fig. 2 in blue color. There is a cut-off region of photonic bandgap located from 1360 to 1416nm. The propagation loss at 1550nm is obtained to be about 4.3dB/m and the splicing loss between EHC-PBGF and SMF-28 is about 7.0dB at 1550nm. The high average insert loss we considered results by two reasons. Firstly, the homogeneous distribution of air holes in the cladding is not good, which leads to the weak photonic bandgap effect of EHC-PBGF and secondly the core is mismatched between EHC-PBGF and SMFs. The insertion loss is much higher in the short wavelength region because of the poorer bandgap effect. So in the sensing experiments, we only record the transmission spectra of EHC-PBGF in the long wavelength region. The photographs in Fig. 3(a) and (b) show near field mode distribution of EHC-PBGF at 1550 nm captured by infrared microscope (Leica DM6000M, Leica Inc.) with a tunable laser. It indicates the existing LP01 and LP11 in the air core and cladding supermode in the solid silica rods. A full-vector finite E-4031 cost method with the commercial software COMSOL Multiphysics was applied to simulate the modal characteristic of EHC-PBGF. The photographs shown in Fig. 4(a) and (b) show the two core modes of LP01 and LP11 at the wavelength of 1550nm and the supermodes distributed in the solid silica rods are clearly co-existing. Comparing Figs. 3 and 4, we can see the experimental results and simulated results are well consistent. For a modal interference, the accumulated phase difference of different modes with the propagational length of L can be expressed as , here λ is the operating wavelength, is the effective refractive index difference of the two modes that propagated in the core or cladding. When the phase difference satisfies the condition , a resonant dip appears at:where is an integer. When the surrounding physical parameters are applied on the modal interference, the phase difference between the two modes will be changed, which indicates the wavelength shift of the interference dip.
Results and discussion The transmission spectrum of the modal interferometer with a length of 651.8μm and without twist at room temperature is shown in Fig. 5. The maximum fringe visibility of the interference resonance dips is 19.1dB. Moreover, some minor dips can also be observed, which indicates that multiple modes contribute to the interference [17] and it is an interference superposition of some modes. In Fig. 5, there are two dips marked by Dip 1 at wavelength of 1446nm and Dip 2 at wavelength of 1545nm, respectively, which are recorded in the sensing experiments rather than other dips due to these two dips are cleanly and easily distinguished with high contrast and not overlapped by other dips. The response of the modal interferometer with a length of 651.8μm to twist was investigated experimentally. The EHC-PBGF under test is positioned at the middle of a section of fiber, and then one end of the fiber is fixed on a fiber holder, the other end is stuck in a rotated disk and the total twist length d is set to10 cm, as shown in Fig. 1(a). Fig. 6(a) shows the transmission spectra of the modal interferometer at two anti-clockwise twist rates of and 2.618rad/m. Two resonance dips marked by Dip 1 and Dip 2 are clear. It can be seen that the center wavelength of Dip 1 remains almost still, while the resonant wavelength of Dip 2 is redshift when the twist rate applies from to 2.618rad/m. Fig. 6(b) shows the dependence of wavelength shift upon the twist rate. Firstly, the fiber is twisted in anti-clockwise in a total angle of 25 degree by step of 5 degree and then turned back to the original point for clockwise twist experiment. As shown in Fig. 6(b), when the fiber is twisted anti-clockwise, wavelengths of resonance Dip 1 and Dip 2 are both redshift, and while twisted clockwise, the wavelengths are both blueshift, and it is completely reversible and repeatable in our experiments. Obviously, the wavelength shift of Dip 1 at 1446nm is dependent weakly on the twist; on the other hand, Dip 2 at 1545nm is stronger dependent on the twist. By using linear fit, the twist sensitivities of Dip 1 and Dip 2 are obtained to be −31.95 and −585.8pm/(rad/m), respectively. It indicates the sensitivity of Dip 2 is much higher than that of Dip 1. We consider that Dip 1 is resulted from an interference of some supermodes in bar-type silica rods of the cladding. The high birefringence of the supermodes leads to the polarization maintaining of guided light. So the wavelength shift of Dip 1 is dependent weakly on the twist. While Dip 2 is resulted from the resonant interference of two guided air core modes with a low birefringence, which leads to that the wavelength shift is strongly dependent on the twist. In many papers published before, various high-birefringence fibers (HBFs) and low-birefringence fibers (LBFs) were analyzed the twist effects [23,24]. Twist effect on the fiber is one of the major causes of the circular birefringence in the fibers. For the case of HBF, the linear birefringence is greater than circular birefringence induced by the twist effect. Thus the twisted fiber acts as a rotator. The circular birefringence caused by the twist effects will be greatly swamped by the large linear birefringence of HBF itself [25]; While the circular birefringence caused by the twist effects and the linear birefringence are co-working in LBF, hence torsion sensitivity of the twist sensor based on LBF can be achieved to be more higher than the one based on HBF. By fixing one end of the fiber and twisting on the other end, the retardance (θ) between the two guided modes is expressed as a function of twist angle θ which is given as [25]: