1. Introduction

Hollow-core photonic bandgap fibers (HC-PBGFs) have attracted great interest and opened up new opportunities to fiber optics in several scientific and technological areas [1, 2]. Unlike conventional fibers that guide light by total internal reflection, in a HC-PBGF light is confined and guided in the hollow core by a photonic bandgap that prohibits the propagation of light in the cladding region under certain conditions. The cladding of these fibers is a periodic array of air holes in silica glass. HC-PBGFs offer unique and exotic properties making them highly desirable for a range of applications, such as high power [3] and femtosecond pulse [4, 5] delivery, and gas-phase nonlinear optics [6].

Much effort in designing HC-PBGFs has been focused towards reducing the fiber attenuation, and impressive experimental progress has been reported [8, 9]. However, several emerging application areas such as power delivery [3], ultrashort pulse delivery [4] and pulse compression [7] require improvements in other characteristics, most notably increased bandwidth, reduced higher-order dispersion, and control of the nonlinear response. Optimising the attenuation has relied on engineering the core wall - the thin ring of glass immediately surrounding the hollow core - so as to minimise the overlap of the guided mode with the glass surfaces thus reducing surface scattering [10]. However, in doing so, surface-guided modes which are confined to the core wall have been introduced into the bandgap [11]. These have the effect of reducing the effective usable bandwidth, and simultaneously increasing the dispersion and dispersion slope. These surface modes can be seen as a consequence of having imperfect termination of the periodic cladding structure at the core/cladding interface. In order to remove the impact of surface modes on the guidance, one needs to design and fabricate a fibre in which the cladding structure terminates as naturally as possible at the core/cladding interface. One way to do this is to simply form the core by terminating the cladding at the natural edge of a unit cell.

We have fabricated a HC-PBGF with a core formed by the omission of seven unit cells which presents no sign of surface modes interactions within the bandgap. As a result of the absence of surface modes crossings, our fiber has wider bandwidth and approximately halved dispersion and dispersion slope compared to previous fibers. Our fiber design is based on numerical computations which show that if the core walls have just half the thickness of the thinnest features of the cladding, then surface mode interference can be suppressed [12, 13]. Fabricating such a structure represents a challenge which we have overcome using a modified fabrication procedure that allows for the production of low-loss HC-PBGFs from scratch within a single day.

2. Fiber fabrication

The fibers reported here were fabricated entirely from high-purity synthetic silica glass (F300 from Heraeus Quarzglas) using the stack-and-draw technique. Our previous hollow-core fibers [9] were formed using a two-dimensional array of capillaries stacked around a thin-walled core tube. Additional solid rods were inserted into the interstitial holes in the stack to create the required array of strands of glass, joined and supported by thin silica webs in the final fiber. The relative scale of the whole structure was roughly preserved in drawing the preform down to a fiber. In contrast, the fibers being reported here were drawn from a stack with neither core tube nor the addition of extra glass in the interstitials. Circular capillaries with thicker walls (relative to their diameters) than required in the final fiber were used to form a stack in which the core was created by simply omitting 7 central capillaries without using an extra tube. Short capillaries were used at both ends of the stack to support the core defect. Omitting the core tube resulted in a core wall thickness just half that of the struts in the cladding, as required to suppress surface modes [12, 13]. The stack was drawn to fiber in two drawing stages, with pressure applied at the top of the preform being used to inflate the structure to a high air fraction during the final draw. During drawing, inflation leaves bigger strands at the interstitial sites joined by thinner webs. This process allows for easy and very rapid fabrication of fibers with core wall thickness close to the thickness required to eliminate surface modes, and having both broadband operation and state-of-the-art attenuation. The simplicity of the process makes it possible to draw a hollow-core fiber from scratch, within a single day. A typical SEM image of one of our canes before the final draw to fiber is shown in Fig. 1.

 

Fig. 1. Typical SEM image of a cane used to draw our fibers. The outer diameter shown is 3.3 millimeters, and an addition silica jacket has yet to be added during the final draw to fiber.

Download Full Size | PDF

3. HC-PBGF free from surface modes

A scanning electron microscopy (SEM) image of the fabricated thin core wall fiber is shown in Fig. 2(a). Our thin core wall fiber features a 7-cell core surrounded by 6 rings of cladding holes plus one incomplete ring to facilitate the stacking. Detailed structural analysis was performed by SEM which allowed us to verify that the thickness of the core wall was within the regime for eliminating surface mode resonances within the bandgap. The core geometry of the fiber is slightly different to the core designs previously studied [12, 14], because during the fabrication process we have further experimentally optimized the fiber’s core and the ring of holes immediately surrounding it in order to reduce loss and maximize the transmission bandwidth.

 

Fig. 2. SEM micrographs of 7-cell HC-PBGF designed to operate at 1550 nm, to the same scale. (a) Our fiber incorporating a thin core wall, and (b) conventional HC-PBGF similar to that available commercially. Scale bars correspond to 10 µm.

Download Full Size | PDF

For comparison, Fig. 2(b) shows an SEM image of a fiber similar to those commercially available and designed for 1550 nm transmission - in what follows it will be referred to as “old fiber”. Note that although the size of the microstructured region is almost the same for the two fibers, our fiber has two fewer rings of cladding holes than the old fiber.

The photonic crystal cladding of our thin core wall fiber has a pitch Λ=6.7 µm and an air filling fraction of ~96% that give rise to a photonic bandgap centered approximately at 1550 nm. The high air-filling fraction in our cladding enables the formation of a wide photonic bandgap crossing the airline, which covers around 22% of the central bandgap wavelength. The cladding pitch of previously-reported HC-PBGFs designed for transmission at 1550 nm has typically been between 4 µm to 4.8 µm [9, 8]. Due to its large pitch, our fiber features a larger core than previous designs (see Fig. 2), which in general can be expected to result in reduced interaction of the air-guided mode with the core-cladding interface, reduced effect of waveguide dispersion and lower nonlinear response.

3.1. Attenuation spectrum

The optical attenuation of our thin core wall 7-cell HC-PBGF was measured using the cut-back technique. A sample of 270 m was cut back to 50 m and a fiber-based supercontinuum was used as the light source. The measured attenuation spectrum is shown in Fig. 3 (red curve). The loss remains low over a wide bandwidth and there are no sharp peaks indicating the presence of surface modes anticrossings. The sharp increases in loss below 1450 nm and at around 1750 nm indicate the edges of the photonic bandgap and are not due to surface mode coupling. Consequently, low loss is achieved over a broad spectral range; from 1450 nm to 1750 nm, covering the full spectral width of the photonic bandgap formed in the cladding. The minimum attenuation is 15 dB/km and remains below 50 dB/km over approximately 300 nm. This is the first time to our knowledge that light can be transmitted in a HC-PBGF with low loss over such a wide spectral window.

 

Fig. 3. Measured attenuation spectrum of (red) our thin core wall fiber, and (blue) old fiber. (top) Near field images of the “fundamental” air-guided mode after 50 m of our fiber at different wavelengths within the bandgap.

Download Full Size | PDF

The near field images shown at the top of Fig. 3 were recorded after 50m of the thin core wall fiber by using 10 nm bandpass filters in the wavelength range from 1440 nm to 1700 nm. These mode images confirm that light propagates in a single mode which does not couple to surface modes. In addition, the mode field patterns do not vary significantly across the transmission window.

In contrast, the attenuation spectrum of the old fiber (Fig. 3 blue curve) shows that while both our thin core wall fiber and the old fiber have a photonic bandgap centered at almost the same wavelength, the attenuation spectrum of the old fiber is not a smooth curve but presents distinct attenuation peaks. These high loss regions are due to surface mode anticrossings occurring near the short wavelength edge of the bandgap, at around 1450 nm and at around 1480 nm. At these wavelengths surface modes couple with the core-guided mode increasing the loss and thus decreasing the effective bandwidth of the fiber [11]. Therefore, by eliminating surface modes we have been able to increase the effective bandwidth by approximately 70 nm.

3.2. Group velocity dispersion

The variation of group index with wavelength was measured by low-coherence interferometery on 25 cm of fiber. Light from a supercontinuum source was launched into the fiber while ensuring that only the fundamental core mode was exited. The group velocity dispersion (GVD) for one polarization of the thin core wall fiber, together with the transmission spectrum of 50 m of the fiber, are given in Fig. 4(a). For the second polarization similar dispersion values were found, indicating the very good structural homogeneity of our fiber.

Again as before, neither the transmission spectrum nor the GVD of the thin core wall fiber present signs of surface modes anticrossings, resulting in smooth curves right across the bandgap Fig. 4(a). The chromatic dispersion goes from normal to anomalous with 20 ps/nm/km at the central bandgap wavelength. The dispersion slope is found to be approximately 0.3 ps/nm²/km over a broad 200 nm spectral range from 1490 nm to 1690 nm. This is the lowest dispersion slope yet reported for a HC-PBGF and represent a factor of almost two reduction compared to the previous state-of-the-art. For the old fiber, dispersion at the central wavelength is equal to 52 ps/nm/km with a steeper GVD-slope of 0.54 ps/nm²/km over less than 130 nm, as can be seen from Fig. 4(b). It is worth noting that the third order dispersion of previously available HC-PBGF has been a profound limitation on their performance for at least one important application: soliton compression of high-power ultrashort pulses [5, 7].

 

Fig. 4. Group velocity dispersion measured on 25 cm of fiber and normalized transmission through 50 m of fiber. (a) Thin core wall fiber, and (b) Old fiber.

Download Full Size | PDF

4. Low attenuation (9.5 dB/km) in thin-core-wall 7-cell HC-PBGF

Figure. 5(a) shows an SEM image of a different thin-core-wall fiber, designed for transmission at wavelengths around 1650 nm. The measured attenuation spectrum in Fig. 5 shows that the low loss transmission window starts at around 1560 nm. However, the long wavelength edge of the bandgap region could not be determined due to the limited spectral range of the optical spectrum analyzer used for the measurements. The minimum loss is 9.5 dB/km, and is almost constant over more than 100 nm. This is the lowest loss value ever reported for a 7 cell hollow core fiber (although still somewhat above the 1.2 dB/km reported in 19-cell fibers [9]). The apparent increase in loss after 1720 nm is due to the limitations of our measurement equipment and not due to surface mode anticrossings nor to the bandgap edge, and we expect that the low loss transmission window should extend to around 1900 nm. The reduced attenuation compared to previous 7-cell fibers is at least partly due to the use of a longer wavelength and the larger core size, but the thinner core wall is expected to reduce the scattering loss [12]. The low attenuation is strong evidence that our revised fabrication process can produce structures with sufficient regularity and integrity to compare with the previous state-of-the-art.

 

Fig. 5. (a) SEM image of a low loss 7-cell HC-PBGF designed for transmission at around 1650 nm, (bar, 10 µm). (b) Attenuation spectrum obtained from a cut-back measurement using a sample of 300 m cut back to 100 m.

Download Full Size | PDF

5. Conclusions

In this paper we have presented the fabrication and characterization of 7 cell HC-PBGF with a novel core geometry incorporating a thin core/cladding interface. Experimental studies indicated that the fibers are able to guide light over a broader spectral window and with comparable or lower attenuation than previous state-of-the-art HC-PBGF. This is a consequence of the elimination of surface mode resonances within the bandgap. Our fiber also presents the lowest dispersion slope yet reported for a HC-PBGF - a factor of almost two reduction compared to the prior state-of-the-art. In addition, our fabrication process enables the production of low-loss HC-PBGFs in less than a day and offers flexibility for further optimization of the cladding structure for the formation of broader bandgaps. We expect that a low loss transmission window of approximately 350 to 400 nm for a bandgap centered at 1550 nm can be obtained [13].

Although all the fibers presented here were designed to operate around the 1550 nm telecommunications window, due the ease of comparison with commercially available fibers, we have also fabricated several fibers with operational wavelengths from 1064 nm to 1750nm which, like the fibers reported here, also do not exhibit surface modes crossings. This technique could also be applied to the fabrication of 19-cell HC-PBGFs, where it would greatly extend the useable bandwidth of the fibers. The transmission spectra of previously-reported 19-cell fibers have been completely dominated by surface mode crossings, and so we anticipate that the larger core size and greatly reduced dispersion slope in the new fibers would enable a new regime of ultrashort-pulse solitons and high-power beam delivery.