1. Introduction

Microstructured optical fibers (MOFs) enable efficient interaction between guided light modes and gases or liquids filling the airholes [1]. Thus MOFs provide the potential for developing compact devices with long interaction lengths that are suitable for high sensitivity optical sensing and spectroscopy applications [1–7]. Most practical implementations have relied on filling MOFs from one or both fiber ends. However, given that free diffusion of gases along the fiber typically occurs at a rate of several hours per meter [1–4], and vacuumor pressure-assisted filling is impractical if regular replacement of the filling substance is required, this approach is restrictive. Most importantly, difficulties arise when attempting to ensure stable optical coupling and simultaneous fluidic access, in a compact fashion.

Fabrication of micro-fluidic side-channels at several locations along the fiber length - leading from the fiber’s outer surface to the cladding airholes in index-guiding photonic crystal fiber (PCF), or to the core in hollow-core photonic bandgap fiber (HC-PBGF) - would significantly improve filling rates [3]. Micro-channels would also yield the freedom to alter the composition, concentration and pressure of the gas or liquid sample filling the airholes, without compromising optical coupling via the two fiber ends. To our knowledge, previous development of such channels has been limited to: heat- and pressure-assisted expansion and subsequent breach of the core in HC-PBGFs [8], ArF laser ablation in PCF [9], and focused ion beam (FIB) processing of MOFs [10]. A single channel in [8] reportedly caused a transmission loss of 6 dB, while values for induced fiber transmission loss were not provided in [9] and [10]. The need still exists for manufacturing micro-channels in MOF with limited damage to the microstructured region - the ultimate objective being to incur zero additional loss in fiber transmission.

Femtosecond lasers have emerged in recent years as a powerful tool for materials processing, with micro-machining being one of the foremost applications. In particular, micro-machining has been demonstrated in materials that are optically transparent at the processing laser wavelength [11–13]. Ultrashort pulses deliver energy to a sample over time-scales significantly shorter than that required for energy transfer to the lattice [12,13]. A combination of non-linear effects (such as multi-photon absorption and avalanche ionization) produces a high-density plasma at the laser focus, which deposits electron energy in the sample lattice at the end of each pulse. This causes permanent, highly localized structural changes without inducing collateral damage due to thermal diffusion.

Micro-machining by femtosecond laser has been exploited in a broad range of materials to manufacture structures specifically for micro-fluidic applications [11]. In silica, fluidic channels have been realized in bulk glass [14,15] as well as in conventional single-mode fiber (SMF) [16], and a clear distinction can be made between two different techniques [11]. The first method relies on selective etching with a hydrofluoric acid solution to remove the lasermodified section of the sample, while the second utilizes femtosecond pulses for actual material removal (with liquid-assisted expulsion of debris in some cases). Laser-based fabrication of micro-channels in MOF represents a significant challenge due to the fragility of the microstructured region, upon which optical guidance ultimately relies. Here we report, for the first time to our knowledge, on the successful femtosecond laser machining of low-loss micro-channels in MOFs. Such micro-channels were fabricated using a single-step process relying only on exposure to ultrashort pulses, without any subsequent chemical etching.

2. Experimental work

Both types of MOF used in the experiments were fabricated in-house. HC-PBGFs (Fig. 1) were manufactured by means of the stack-and-draw technique to have an outer diameter of 205 μm, a core diameter of 14.4 μm, a cladding pitch (Λ) of 4.5 μm, and an air filling factor of 94 %. White light measurements revealed that the fiber transmission loss is 90 dB.km^-1 at 1530 nm. The suspended-core holey fiber (SC-HF: refer to Fig. 2) was fabricated from an ultrasonic-drilled preform to an outer diameter of 180 μm [1]. A transmission loss of 0.15 to 0.20 dB.m^-1 was measured at 1530 nm. This fiber contains a 1 μm-diameter core supported by three thin struts, each 6 μm long and approximately 150 nm wide at the waist. Due to its small core size, the SC-HF possesses a comparatively large evanescent field in the hollow cladding. Finite-element modeling (based on scanning electron microscope (SEM) images) predicts that at least 20 % of the guided mode’s optical power overlaps with the cladding at 1530 nm [1].

 

Fig. 1. SEM image of HC-PBGF microstructure.

Download Full Size | PDF

 

Fig. 2. SEM images of the SC-HF structure.

Download Full Size | PDF

Micro-channels were fabricated in 2 m-long, acetylene-filled HC-PBGF and SC-HF cells (gas-filled HC-PBGF cells were first reported in [5]). HC-PBGFs were spliced to conventional SMF using an electric arc fusion splicer and a method developed independently from, but similar to that described in [6]. A solid-core highly-nonlinear fiber (HNLF) was used in conjunction with conventional SMF to minimize the splice loss in SC-HF cells. The input splice was performed first for both cell types, thus providing stable optical coupling to the MOF. Each open fiber end was placed in a gas-tight, windowed interface for filling with acetylene at atmospheric pressure. Acetylene diffusion into these single-ended MOFs was monitored by measuring the linestrengths of P10 (1530.98 nm) and P11 (1531.59 nm) absorption lines. Such measurements were conducted with an erbium amplified spontaneous emission (ASE) source and optical spectrum analyzer (OSA) at 0.01 nm nominal resolution. Strong and stable lines were obtained after approximately 2 hours, at which point the output end of each cell was sealed by splicing.

An amplified, mode-locked Ti:sapphire laser providing pulses of 110 fs duration, at a wavelength of 800 nm and a repetition rate of 1 kHz, was used for micro-channel fabrication. The 2 mm-diameter laser output was passed through a 200 μm-diameter aperture to produce a smooth transverse intensity profile. The beam at the aperture was subsequently imaged onto the target fiber through a lens pair consisting of a plano-convex lens and a 50x microscope objective (0.55 numerical aperture and 10.1 mm working distance) - yielding a 2 μm spot size at the focus. The output pulse energy was measured to be 13 μJ. MOF samples were mounted in V-grooves on a 3-D nano-translation stage, thereby allowing precision control of the fiber position relative to the laser beam. Visual confirmation of focusing was provided by means of a CCD camera imaging through the microscope objective.

The optical transmission of gas-filled MOF cells was continuously monitored during the micro-machining process at wavelengths around 1530 nm to detect any changes in fiber transmission (again utilizing an erbium ASE source). This configuration also provided a means of identifying the successful completion of each micro-channel: the onset of acetylene out-diffusion being indicated clearly by decreased absorption linestrengths.

3. Results and discussion

3.1 Micro-channels in HC-PBGF

HC-PBGF cells were produced with an average input splice loss of 1.7 ± 0.2 dB. Insertion loss for the sealed, acetylene-filled cells averaged 5.9 ± 0.9 dB: a value comparable to that reported in [6] for empty cells. Micro-machining was performed in several HC-PBGF cells: either one or two micro-channels were fabricated in the same cell - the latter with a view to the future development of sensor devices. Channels were manufactured at sites approximately 0.3 m away from the splices. Laser irradiation times of typically 20 minutes were required for the micro-channel tips to breach the cores of HC-PBGFs. No discernible change in optical transmission level was detected during any of these exposures. In principle, therefore, more than two micro-channels could be machined in any one cell with no significant loss penalty.

Micro-machining parameters were optimized specifically to reduce the amount of lase-rinduced modifications to the photonic crystal cladding. Due to the vast difference in damage threshold characteristic for the fiber jacket (solid fused silica), and for the mostly hollow cladding, it was expected that material damage at the interface between these two regions could be limited by maintaining the laser focus at a fixed point close to the fiber surface. Ideal conditions were selected whereby micro-machining occurred rapidly at the outset (close to the focus), and then decreased gradually as the channel tip moved further away from the focus. This led to a narrowing of the channel beyond the laser focus, with tapering continuing all the way to the hollow core. Although sustained irradiation did not cause significant transmission loss, micro-machining was usually discontinued soon after the core had been breached to avoid damaging the microstructured cladding beyond the core.

 

Fig. 3. SEM image of a micro-channel fabricated in HC-PBGF: arrows indicate damage caused by laser ‘scoring’ (insert shows channel and ‘scoring’ lines on uncoated fiber surface, prior to cleaving).

Download Full Size | PDF

 

Fig. 4. Optical microscope images of micro-channel fabricated in coated HC-PBGF: a) channel cross-section (showing guidance via higher-order bandgap); b) at polymer surface; c) at silica surface.

Download Full Size | PDF

Figure 3 displays a micro-channel in cross-section. The polymer coating of this HC-PBGF sample had been removed prior to laser irradiation. Once micro-machining had been completed, two ‘scoring’ lines were inscribed with the laser at the fiber’s outer surface: one on either side of the micro-channel, as shown in the insert of Fig. 3. These lines allowed a precise fiber cleave to be performed in the plane of the channel for more effective visualization, but also caused small regions of damage - indicated by the arrows in Fig. 3. The channel diameter tapers from 20 μm at the fiber surface to a few microns at the core. Figure 3 50 μm also reveals that laser-induced damage to the microstructure is most severe at the interface between the bulk silica and the microstructured region (where debris is also visible). However, the channel contains significantly less debris in the remainder of the microstructured cladding as it tapers gradually to the core. Figure 4 shows cross-section and top views of a micro-channel produced in acrylate polymer-coated HC-PBGF. The rate of ablation for the polymer far exceeds that of fused silica and as a result the channel entrance appears considerably rougher at the polymer surface. Nevertheless, the quality of this microchannel matches that of channels in uncoated fiber samples with respect to penetration depth and laser-induced damage. Coated MOFs naturally possess superior mechanical strength when compared to bare fibers. Thus the ability to fabricate successful micro-channels in coated HCPBGF represents considerable progress towards device implementation.

We observed accelerated out-diffusion from an acetylene-filled HC-PBGF cell containing two micro-channels. A clear decrease in absorption was detected as soon as the second microchannel breached the core. Figure 5 shows acetylene absorption spectra at specific points in time during the fabrication of two micro-channels. It is evident from Fig. 5 that the overall transmission level remained intact both during and after micro-machining. Figure 6 illustrates the difference between out-diffusion characteristics of cells containing one and two micro-channels, respectively. Diffusion curves were plotted using P10 linestrengths from absorption spectra such as those selected in Fig. 5 (minor saturation was observed for the P11 line). Initially, the two curves are almost identical, but diverge quite significantly at the point corresponding to the second micro-channel breaching the core (just before C in Fig. 6). Subsequently, the rate of diffusion in the cell containing two micro-channels clearly exceeds that of the cell containing a single channel for the remainder of the out-diffusion process. Curve fitting has been performed in Fig. 6 by using a summation of exponential diffusion curves. These preliminary curve fits indicate a more complex process than is characteristic for diffusion via the fiber ends [3,4], with a strong dependence on micro-channel location along the fiber. We are currently developing more accurate numerical models (incorporating channel dimensions) to obtain a quantitative description of the results.

 

Fig. 5. Acetylene absorption spectra recorded at several times during the fabrication of two micro-channels in a single HC-PBGF cell (curves correspond to times labeled in Fig. 6).

Download Full Size | PDF

 

Fig. 6. Acetylene out-diffusion curves obtained by plotting diminishing linestrengths of acetylene’s P10 absorption line during fabrication of one and two micro-channels in HC-PBGF cells.

Download Full Size | PDF

3.2 Micro-channels in SC-HF

Micro-machining was also investigated in acetylene-filled SC-HF cells. The input HNLF-to-SC-HF splice loss was approximately 7 dB in such cells, with a total SMF-to-SMF insertion loss of 15 dB. Our objective was to fabricate a micro-channel that would penetrate only one of the three airholes in the SC-HF microstructure, thus accessing a third of the evanescent field.

SC-HF samples were secured in V-grooves on a pair of fiber rotators. The CCD camera was used to verify that incident laser pulses were directed towards a single cladding airhole and not towards one of the silica struts. Compared to HC-PBGFs, these fibers required a small reduction in average pulse energy and slightly longer irradiation times for the channel tip to reach the microstructure (typically 30 minutes: the bulk silica region is larger in SC-HFs).

Under optimum fabrication conditions, we achieved a transmission loss of only 0.5 dB per micro-channel. In Fig. 7, it is evident that the struts and core remained undamaged during the micro-machining process, so this loss may be attributed to scattering resulting from small amounts of debris deposited onto the core. Acetylene’s P11 absorption line was monitored so that out-diffusion could be detected and micro-machining discontinued. Figure 8 displays P11 absorption before and after out-diffusion from a single cladding airhole - channel formation having resulted in approximately a one third decrease in linestrength.

 

Fig. 7. Optical microscope image showing cross-section of micro-channel fabricated in SC-HF.

Download Full Size | PDF

 

Fig. 8. Acetylene P11 absorption spectra recorded for a single micro-channel in SC-HF.

Download Full Size | PDF

4. Conclusion

We have demonstrated successful fabrication of functional micro-fluidic channels in both hollow-core PBGF and solid-core, index-guiding SC-HF by means of femtosecond laser micro-machining. The engineered technique proved to be highly selective and has the merit of causing no discernible transmission loss in HC-PBGF, and only a minor loss increase in SC-HF. This procedure should therefore be entirely scalable - allowing for the fabrication of numerous micro-channels along the length of a spliced MOF cell, without unduly compromising optical transmission. Acetylene gas diffusion was demonstrated through such micro-channels, and increased out-diffusion rates were recorded when two channels were fabricated in HC-PBGF.

This leads to an entirely new concept of MOF-based gas cells, in which the conditions of the substance used to fill the fiber’s microstructured region may be altered at will. MOF cells containing variable pressures and concentrations can readily be envisaged, with sample volumes that may be filled much more rapidly - extremely promising attributes for fiberbased sensing and spectroscopy. Prospects for further device development also appear favorable with regard to mechanical integrity of the fiber, as micro-channels have already been manufactured in polymer-coated HC-PBGFs. Current work is focused on investigating the long-term stability of such micro-channels, with a view to developing practical sensors.