1. Introduction

Photonic bandgap (PBG) materials have properties that make them useful for next generation photonic device applications. Wavelengths that lie within the bandgap are forbidden from propagating through the material, lending PBG structures naturally to the applications in filtering and confinement [1]. For example, in 2-D photonic crystals (PCs), PBG materials typically comprised of a slab dielectric with a regular pattern of holes etched in it, a line of defects in the pattern can form a planar waveguide. Photonic crystal fibers (PCFs) are optical fibers with a regular pattern of holes that run parallel to the fiber axis such that the cross-section comprises a PBG structure which can radially confine light [2]. The fiber cross-section can be designed to achieve different guiding mechanisms and to exploit various physical phenomena such as single mode operation over an extended wavelength range [3], hollow-core guidance [4] and non-linear effects such as super-continuum generation [5].

Conventionally, light propagates parallel to the holes in these fibers and is guided either by the interface between a high index core and low effective index microstructure, or by the stop bands of the PBG structure. However, one may also use these fibers in a transverse geometry, where the light propagates normal to the holes, similar to 2-D PC slabs. We recently demonstrated [6] that in this geometry we can directly measure the stop bands associated with microstructure of an index-guiding PCF. In the experiments, we found a fundamental partial gap at ~3.0 µm and a higher order gap at ~1.5 µm. This work has potentially important applications in microphotonics using existing PCF manufacturing technology, or in improving that technology by providing feedback during the fiber draw process.

In this paper we extend this novel application of PCF by tapering the fiber to reduce the pitch (Λ) of the PC microstructure, thus shifting the fundamental partial bandgap to shorter wavelengths. We have achieved reductions in the diameter of PCF by a factor of five whilst still retaining the PC structure. Using this post-processing technique, we fabricated PC structures with Λ<0.3 µm, which to our knowledge have not been demonstrated in a PCF before; this corresponds to a shift in the fundamental gap down to visible wavelengths. We experimentally probe the PCF taper in the transverse geometry and correlate our measurements with numerical simulations.

2. Basic principle

A schematic of this transverse probing is shown in fig. 1. An SEM image of the cross-section of the tapered fiber is shown in the inset, with the crystallographic symmetry axes as indicated. The untapered PCF microstructure is comprised of 10 rings of air holes, 0.9 µm in diameter, forming a triagonal lattice with a 1.28 µm pitch, and a single missing air hole forming a central defect; the fiber was supplied by Crystal Fibre A/S. Figure 2 shows the calculated band diagram [7] associated with the untapered PCF microstructure. Here we define the TE direction to be parallel to the fiber axis. When light is launched across the fiber as in Fig. 1, we expect that wavelengths which lie outside the bandgaps will propagate to the other side and wavelengths within the gaps will not; there will also be some scattering associated with edges of the microstructure. In our experiments, described in detail below, light is launched from a broadband source and the transmitted signal is measured on an optical spectrum analyzer (OSA), where the bandgaps appear as a loss notch in the transmission spectrum. There is no complete bandgap in this system, only partial gaps along the symmetry axes of the structure, so it is necessary to probe the fiber along one of these axes; in this work we only consider the partial gaps in the Γ-M direction (see Fig. 1). If we presume that the fiber cross-section is preserved by the tapering process, then the band structure shown should apply for all points along the taper. Experimentally, we expect that the wavelength of the notch in the transmission spectrum measurements should shift as we probe different points of the taper.

 

Fig. 1. Schematic of a tapered PCF probed in a transverse direction to the axis of the taper. Inset shows SEM image of cleaved tapered PCF, with local outer diameter of 37.5µm and pitch <0.5 µm

Download Full Size | PDF

 

Fig. 2. Band diagram for triagonal lattice of air holes in silica with Λ=1.28 µm and hole diameter=0. 9 mm. Crosshatched regions indicate partial bandgaps along symmetry axes.

Download Full Size | PDF

3. PCF taper

Chandalia et al. [8] and Huntington et al. [9] have previously reported tapering holey fibers by factors of 18:1 and 8:1, respectively, whilst retaining the cross-sectional geometry within the taper region; of course, PCFs are manufactured by drawing preforms down by factors of 100:1–1000:1. Chandalia et al. considered a fiber with a microstructure comprising only six very large (~20µm) air holes; Huntington et al. achieved Λ of 1.1µm, though they considered a structure with a relatively large initial air hole diameter (3.3 µm) and Λ (9 µm). This work presents the first demonstration to our knowledge of a tapered holey fiber with Λ<300 nm. The PCF tapers were produced on a taper rig using the flame brushing technique [10, 11]. The PCF was mounted on either end in a fiber chuck, each of which in turn was mounted on a motorized stage. The fiber was heated in the center by a narrow butane flame, also mounted on a motorized stage, which moved back and forth along the fiber axis as the taper was pulled by the fiber chuck stages. The motion of all three stages was controlled by an integrated computer interface. The cross-sectional dimensions of the tapers were found to be sensitive to the taper pull parameters such as the flame travel distance and fiber elongation rate. We found that the holes would collapse if the elongation rate was too slow or if the final fiber diameter was too small. The onset of collapse could be delayed by maintaining sufficient gas pressure within the holes, which we achieved by sealing the ends of the PCF.

 

Fig. 3. (a) SEM micrograph of untapered fiber (OD=108 µm, Λ=1.28 µm) and (b) tapered fiber: OD=47.1±0.6 µm, Λ ~0.56 µm

Download Full Size | PDF

Figure 3 shows SEM micrographs of the untapered fiber and of a taper with a local OD of 47.1 µm, respectively. One notes that the air/glass fraction and the quality of the hole geometry are essentially unchanged (the apparent shearing in Fig. 3(b) is due to the cleave). From the band structure in fig. 1, one expects that the tapered structure in Fig. 3(b) should have a fundamental Γ-M gap centered at ~1200 nm, demonstrating that one can draw tapers to have fundamental stop bands past the 1550 nm and 1300 nm communications bands without significant distortions of the microstructure.

We produced even smaller tapers to shift the fundamental partial gap to visible wavelengths. In general, we found that for OD <~40 µm, the holes begin to collapse relative to the pitch. This is shown in Figs. 4(a)–(c), where the ratio of hole diameter to pitch is very obviously reduced. Even so, these tapers still exhibit stop bands. Figure 5 is a color image obtained from an optical microscope in which the PCF taper has been illuminated by white light from the front. Bragg diffracted visible light can be clearly seen reflecting from the PCF taper with the reflected color varying monotonically along its length. The taper dimensions at either end of the visible spectrum roughly correspond to the dimensions for Figs. 4(a) and 4(b); the diameter of taper corresponding to the yellow/green area is ~24µm, corresponding to Λ~250 nm.

 

Fig. 4. SEM micrographs of cleaved tapered fiber ends (a) OD=26.4±0.6µm, Λ~290 nm, (b) 21.3±0.3µm, Λ~217nm, and (c) OD=17.3±0.3µm, Λ~170nm

Download Full Size | PDF

 

Fig. 5. Optical microscope image of tapered PCF illuminated by white light from the front; taper OD at the center is ~24 µm. The color of the first order Bragg reflected light varies monotonically to shorter wavelengths as the taper dimensions are reduced.

Download Full Size | PDF

4. Transverse probing experiment

Figure 6 shows a schematic of the transverse probing experiment. Light is launched from the unpolarized white light source to a cleaved end of single mode fiber (SMF-28), where it then propagates transversely through the PCF taper to another cleaved end of SMF-28 and is guided to the OSA. The PCF taper is held in a rotation chuck in order to adjust the orientation of the microstructure. The taper is cleaved at the waist and the end is imaged through a 10X objective lens to a CCD camera so that the orientation can be verified; an image from the camera is shown in Fig. 7(a). For the sample used in this experiment, the taper was pulled at a rate of 5 mm/min down to a minimum waist diameter of ~43 µm, where the waist was 5 mm in length. The local taper OD is a slowly varying function of the fiber axis (<25 µm/mm), ensuring that the local OD and hence the local pitch are effectively constant over the width of the probe beam. For local taper OD <59 µm, the separation between the two pieces of SMF-28 is 63 µm; for larger OD, the separation is increased to 69 µm. All of the fibers are mounted on 3 axis microblock stages to control the alignment. Optical spectra are taken at different points along the taper, where the taper is translated towards the camera to change the center wavelength of the stop band; a representative spectrum and the corresponding gaps calculated by assuming linear scaling of Λ with OD are shown in Fig. 7(b).

 

Fig. 6. Schematic diagram of transverse probing experiment

Download Full Size | PDF

 

Fig. 7. (a) Image of tapered fiber in experimental setup with PC microstructure oriented along Γ-M axis relative to optical axis. (b) Representative measured transmission spectrum for local taper OD=46.7±0.5 µm, with predicted TE and TM gaps superimposed.

Download Full Size | PDF

Figure 8 shows the measured transmission spectra where taper OD varies from 63.9 to 43.2 µm (all experimental measures of local OD given are ±0.5 µm). This corresponds to a change in Λ of ~770–530 nm, and manifests in a shift of the fundamental gap from 1670 nm to 1100 nm. The transmission is normalized against the transmitted spectrum with the input and output SMF butted up against each other. The background insertion loss is quite low, varying from ~2.5–5.0 dB. Relative to the background loss, the notch is ~21 dB deep, though it weakens with decreasing OD to 16 dB for OD < 45 µm. We expect this occurs because at these dimensions the probe beam is comparable in size to the PC microstructure; this can be verified with finite difference-time domain (FDTD) calculations and will be the subject of future work. The large size of the probe beam relative to the microstructure is also the reason why there is no apparent narrow band resonance in the transmission spectra due to the central core defect. If one considers this structure to be analogous to a 1-D fiber grating, a single missing hole would correspond to a phase shift of ~0.3π, which would manifest as a narrow transmission spike on the short wavelength side of the stop band. However, in our measurements, the probe beam has a width of ~10 µm, whereas the hole diameter varies from 533 nm down to 360 nm over the length of the taper, so one expects the effect of a single missing hole to be negligible. This was also borne out by spectra obtained from 2-D FDTD calculations [12] for a pulsed excitation, where the local taper OD was taken to be 61.5 µm. We found that the spectrum corresponding to the structure with a defect actually had a slightly smoothed amplitude ripple compared to no-defect case, particularly on the short wavelength side of the transmission notch. However, if one makes the defect larger by removing holes on either side of the core (creating a line defect transverse to the propagation direction), one does see a phase shift in the output spectrum. Also, it is interesting to note that the transmission of wavelengths in the bandgap is measurably suppressed even where the microstructure ~10 µm wide. This is shown in the animations in Fig. 9, calculated using 2-D FDTD using the experimental parameters for OD=43.2 µm, though we assume uniform Λ and hole size, unlike the actual structure shown in the SEMs above. In both animations, the probe light is a monochromatic CW input; Fig. 9(a) corresponds to λ=1100 nm, in the stop band, and Fig 9(b) corresponds to λ=1300 nm, which transmits. Returning to Fig. 8, note that for the larger OD spectra, we also see the secondary gap predicted in Fig. 2, though the detector response of the OSA is insufficient at short wavelengths to give an accurate measurement of this gap.

 

Fig. 8. Measured transmission spectra in tapered PCF transverse probing experiment.

Download Full Size | PDF

 

Fig. 9. (572 KB each) 2-D FDTD simulations corresponding to experiment with OD=43.2 µm and CW excitation for (a) 1100 nm (inside stop band) and (b) 1300 nm (outside stop band). [Media 1] [Media 2]

Download Full Size | PDF

 

Fig. 10. Fundamental partial gap along the Γ-M axis plotted as a function of local taper OD. Square points indicate experimentally measured wavelength at the minimum of the transmission notch and bars indicate FWHM of the notch.

Download Full Size | PDF

Figure 10 plots the center wavelength of the transmission notch as function of OD; the bars indicate the FWHM of the notch. The data superimposed on the fundamental gap, and we see very good correlation between a) the center of the gap and the measured center wavelength and b) the gap edges and the measured FWHM, verifying the validity of the 2-D band structure picture to explain the physics of this geometry. In particular, the experiments verify our observation in the SEMs that Λ/OD is conserved and that there is little distortion of the PC microstructure due to the tapering process.

5. Conclusion

We have demonstrated the feasibility of reducing the pitch of a PCF to submicron dimensions. We have achieved up to a five times reduction in the taper diameter as compared with original fiber diameter whilst retaining the PC structure. First order Bragg reflection was observed in the visible spectrum from the PCF taper. We have also experimentally probed this taper transversely as a function of OD and showed that wavelengths in the fundamental gap are suppressed in transmission by >20 dB, and that both the wavelengths and FWHM of measured gaps coincide with theory.