1. Introduction

Optical damage currently limits the ability of conventional silica fiber optics to deliver the high beam quality, high peak power pulses required for micro-machining applications. High beam quality is particularly important for micro-machining, as it allows a small intense focused spot which influences the quality of the features being machined [1]. Laser systems are now commercially available which generate short (nanosecond) pulses at the high repetition rates (tens of kHz) and high peak powers suited to micro-machining. This is currently carried out using motorized stages to move the workpiece beneath the laser and hence describe the machining pattern. However, greater flexibility in the design of the machining systems would be possible if pulses could be delivered through a lightweight optical fiber system, in particular when considering the processing of more complex non-planar workpieces.

Single mode pulse delivery at 1064nm has been achieved previously [1] with peak powers of the order of 250W. However, the regime investigated in that report involved much longer pulses (0.125ms) than studied here, and they are not ideal for high precision micro-machining. Other work for fiber delivery of Nd:YAG laser light for machining purposes has concentrated on large core or multimode delivery [2–4]. This enables high energy pulses (10–30J) to be delivered through the silica fiber without exceeding the laser induced damage threshold (LIDT) of silica. In this situation, however, the beam quality is much lower (M²~20) than that required for high precision machining [2].

 

Fig. 1. Scanning electron micrograph image of hollow-core photonic bandgap fiber designed for use at 1064nm wavelength.

Download Full Size | PDF

Recently developed hollow-core photonic bandgap (PBG) fibers [5] have the potential to overcome the limitations discussed above as the majority of the power is contained within a hollow-core and is delivered in a single mode. These fibers are currently under intensive investigation for high power amplification applications [6] and also for transmission of femtosecond solitons with very high (5.5 MW) peak powers [7], the pulse energies, however, remaining low. In this report the focus is on delivering the nanosecond pulses frequently used for machining. We study the delivery of high power pulses with low M² which, when delivered at high repetition rates, can be used for precision machining. The damage limitations of the hollow-core fiber are also investigated. The PBG fiber used in this investigation (see Fig. 1) has a hollow core approximately 8.2µm in diameter, surrounded by a 7-layer photonic bandgap structure. A measured attenuation curve is given in Fig. 2. The low-loss PBG region spans 180nm centered around 1060nm wavelength. Within the bandgap, attenuation drops to 60dB/km, while on the band edges it oscillates around 1000dB/km. The fiber operates as if single mode at 1064nm with an NA of approximately 0.12 and a mode field diameter of approximately 6.5µm. (In principle, several modes are guided, but these are not observed in our experiments. [8])

 

Fig. 2. Attenuation in the hollow-core PBG fiber. The minimum attenuation is around 60dB/km, and the fiber guides over a band of roughly 180nm width. The inset shows the low-loss region in greater detail.

Download Full Size | PDF

2. Damage limitations of the PBG fiber

As the fibers used in this study have a hollow core, their pulse delivery capability will be limited by the LIDT of the silica/air PBG cladding structure surrounding the core. Therefore to determine this limit Q-switched pulses at 1064nm wavelength were deliberately focussed onto the PBG cladding structure (as opposed to guiding through the hollow core) to test the damage threshold of the PBG web structure (see Fig. 1). A pulse width of 8ns (M²~5–6) at repetition rate of 10Hz was used for the tests.

Two focussed spot sizes were used: 22µm (with a 10× microscope objective) and 14µm (with a 20× microscope objective). The fiber end was viewed using a CCD camera to ensure the energy was incident only on the PBG cladding region. Although significant power was still incident on air between the silica webs it was considered that this would give a good estimate of the LIDT of the fiber cladding material. The PBG structure was completely ablated at pulse energies of 450µJ for the 10× objective, and 200µJ for the 20× objective within a few 10’s of pulses but survived several thousand pulses (the duration of the test) at energies below these levels. Figure 3 shows a PBG fiber with the PBG cladding structure completely ablated. The surrounding unstructured silica cladding is unaffected although it is covered with the scattered debris from the PBG cladding. Using the calculated spot size the pulse energies 450µJ and 200µJ correspond to energy fluxes of 117 Jcm^-2 and 129 Jcm^-2 respectively. According to previous work, [9], the LIDT for bulk silica with an 8ns pulse at 1064nm is around 100 to 120 Jcm^-2 but in this case all the laser energy is incident on the silica. In our measurements with the PBG cladding region the spot size would result in significant power being incident on air between the silica webs. Nevertheless it was considered that this would still give a good practical estimate of the LIDT of the fiber cladding material. It should also be noted that in this initial investigation the LIDT statistics were not performed as a function of cleave quality. However, based on these measurements, the fact that the guided mode in hollow-core fibers is strongly peaked in air [8], and with fiber design optimization (see Section 3), we expect to be able to transmit nanosecond pulses up to the level of tens of milliJoules in a single mode.

 

Fig. 3. Optical micrograph showing damage to hollow-core fiber. The PBG cladding structure is completely ablated by 450µJ pulses, guidelines represent 50µm spacing.

Download Full Size | PDF

 

Fig. 4. Temporal pulse profile delivered from Q-switched Nd:YAG at 1064nm and repetition rate of 15kHz

Download Full Size | PDF

3. Pulse delivery

For the pulse delivery investigation, a diode-pumped Nd:YAG laser was used with a high-quality, low divergence beam. This operated at 1064nm in TEM₀₀ (M²~1.2), delivering approximately 65ns pulses (see figure 4) at a repetition rate of 15kHz. The laser output was expanded and launched into 1m of PBG fiber using a 10× microscope objective. The objective lens was not completely filled with resulting in an effective launch NA of 0.21 and a corresponding spot size of ~4.2µm. It should be pointed out that the spot size was deliberately chosen to be smaller than the core size of the PBG fiber to avoid significant energy being incident on the PBG cladding. Naturally, this resulted in the NA being overfilled and hence a loss of efficiency on launch, the best efficiency achieved being approximately 30%. We would expect this to be improved through the use of more careful beam conditioning optics. Optimum launch conditions were achieved by a combination of viewing the launch end of the fiber via CCD camera to ensure the spot was centred on the air core and, optimising output with a power meter. Figure 5 shows the observed near-field pattern recorded using a low-power cw source through a long length of fiber (20m) and a far-field pattern observed using the ns pulse train and one meter of fiber. In both cases, the output shows a high-quality guided mode.

 

Fig. 5. Near-field (left, recorded with a low-power cw source) and false-color far field image (recorded using ns pulses) at 1064nm after transmission through 20m (left) and 1m (right) of PBG fiber. Excellent beam quality is observed in both cases.

Download Full Size | PDF

The maximum pulse energy delivered was measured as 380µJ, corresponding to an average power of 5.6W and a peak power of around 6kW. However, this did not represent the power at which the fiber was damaged but rather the limit of the power available from the Q-switched laser due to the beam conditioning optics and the efficiency of the launch (roughly 30%).

Previous work [10] with conventional step-index single-mode fiber using a Nd:YAG Q-switched laser at 532nm with an 8ns pulse found that fiber damage occurred at much lower levels with a maximum pulse energy delivered of around 5µJ (corresponding to a peak power of around 600W). The repetition rate of the laser used in this instance was 10Hz and test duration was around 5 minutes. Using the commonly accepted approximate factor of ${t_p}^{1/2}$ [9], where t_p is the pulse width, to scale energies from an 8ns to a 65ns pulse, the maximum pulse energy would increase to around 15µJ. We predict, therefore, that use of the focusing conditions in this investigation to focus the pulse train onto the surface of a standard single-mode fiber would result in immediate and permanent damage to the fiber end-face whereas the PBG fiber can deliver 380µJ without catastrophic damage. The increase of around 25 times in pulse energy deliverable by PBG, compared to standard single-mode fiber (which is an underestimation of the improvement as the 380µJ pulse did not cause catastrophic failure in the PBG fiber), clearly demonstrates the benefits of the hollow-core PBG fiber. A further two orders of magnitude increase is predicted by optimization of the fiber design including a larger relative core size using a 19 cell defect in the fiber, and optimization of the core surround to decrease the light-in-silica fraction and reduce the intensity at the core wall.

The pulse energy delivered by the fiber remained constant (within a few percent) for the duration of the test (roughly 2 mins or 1,800,000 pulses) suggesting that no significant damage was occurring at the launch or delivery end. Although 2 minutes does not represent a practical time scale for micro machining, due to the high laser repetition rate (15kHz), it was regarded that the number of pulses was sufficient for this initial investigation. As the PBG fiber did not fail catastrophically during this period longer test times combined with actual material processing are planned in the future. On inspection however, there was a noticeable difference between the fiber before and after the tests (Fig. 6.) We believe that non-catastrophic damage may be occurring to the PBG fiber cladding. Damage can also be seen on the unstructured silica cladding surrounding the PBG structure. However, these effects did not measurably degrade the ability of the fiber to deliver light. It is also worth observing that our experiments were performed in an atmospheric environment, and the results might well be different in an inert gas.

The heat dissipation in a microstructured fiber has previously been modelled [11], and although those workers did not consider an air-core PBG fiber the principles appear to apply. It is suggested that the properties of the PBG fiber cladding have a large effect on the heat dissipation and temperatures in the fiber. In particular thinner silica webs might lead to higher thermal loads and stresses within the PBG fiber cladding because the conductivity is decreased (in the case where conduction is the main heat transfer mechanism). In the fiber tested in this investigation the PBG fiber cladding is relatively large and contains very thin silica webs (of the order of a few hundred nanometers). This structure might not be ideally suited to removal of heat from the central region although the heat absorbed is unlikely to cause significant damage to the fiber. A further investigation of this phenomenon will require both experimental and modeling work.

 

Fig. 6. Optical micrographs of fiber endfaces before (left) and after (right) pulse delivery. No noticeable change in the pulse delivery was observed.

Download Full Size | PDF

The PBG fiber was also observed using an Environmental Scanning Electron Microscope (ESEM), in which non-conducting samples do not have to be coated as static charge is removed using a gas flow. This allows direct imaging of the silica PBG fiber cladding to inspect for damage on a sub-micron scale. In Fig. 7 we see a PBG fiber as cleaved and after pulsed delivery. Although not conclusive (further statistical damage studies with subsequent ESEM analysis are planned in the near future) it would appear that after pulse delivery small-scale damage may be occurring. The end faces of the silica capillaries used in the PBG fiber cladding structure appear to become detached and thin layers or rings are splaying off the surface. Although similar effects can be seen in freshly cleaved samples (Fig. 1) the frequency of these sites appears to increase after laser pulse delivery.

 

Fig. 7. Environmental scanning electron micrographs of fiber end-faces freshly-cleaved (left) and after being used for pulse delivery (right).

Download Full Size | PDF

4. Conclusion

Although the laser induced damage threshold of the PBG fiber cladding structure is not particularly high (around 120 Jcm^-2 using 8ns pulses and low beam quality), the low overlap of the guided mode with the glass enables high-power nanosecond pulse transmission in a single mode through a hollow core. The high energies and high beam quality delivered through the PBG fibers are of the level required for precision machining applications, in particular when combined with the high repetition rates used here. They are more than an order of magnitude higher than has been reported with conventional step-index single-mode fibers. The damage limit of the PBG fibers has not been reached and it is expected that higher peak powers/energies will be delivered in the near future. With further optimization of the fiber design for this specific application, we expect a further order of magnitude increase in the delivered pulse energy. This promises a new generation of single-mode fibers for high-precision laser machining applications.