Demand for high-power laser delivery using flexible optical fibers has been growing out of the rapid development of laser machining, laser surgery, and many emerging laser applications [1–3]. Conventional solid-core silica optical fibers have already demonstrated a 10-kilowatt continuous laser power delivery at 1 µm [4]. Due to the laser damage of the host material [5], the conventional fiber for high-power laser delivery must require a large core design which inevitably brings in multimode transmission, bend sensitivity, and degraded output beam quality. To circumvent the hurdle of these limitations, the newly emerging anti-resonant hollow-core fibers (AR-HCFs) provide a new solution.

Different from traditional optical fibers, AR-HCFs [6–8] confine and transmit light in the vacuum/air core, where the laser light propagates as in a nearly free-space environment. By introducing proper designs [9], the AR-HCF is still able to maintain a single-mode guidance for a core diameter 30 times the length of the transmission wavelength. In this case, the contribution of host material to the modal properties of the AR-HCF could be reduced by nearly five orders of magnitude [10], resulting in low material absorption [11], low fiber dispersion [1], low optical nonlinearity [12], and higher laser damage threshold [13,14]. All the merits of the AR-HCF enable laser delivery of higher average and higher peak laser powers beyond the performance of conventional silica fibers and other microstructure hollow-core fibers [15].

In 2016, Michieletto and co-workers showed the potential of the AR-HCF to deliver picosecond pulsed laser light with an average power of approximately 70 W at 1-µm wavelength [16]. Later, an average power of pulsed laser light of over 200 W was delivered in an AR-HCF with the peak power reaching gigawatts [17]. In our previous work [18], light from a 300-W continuous wave (CW) laser at 1 µm was successfully delivered by an AR-HCF without cooling. Recently, the transmission of 1-kW CW laser light over a 1-km-long AR-HCF was reported with a near-diffraction-limited output [19].

In this Letter, we study the damage mechanism of the AR-HCF in the delivery of over 1-kilowatt continuous laser power at 1-µm wavelength and experimentally identify two types of laser-induced damage occurring in the polymer coating and jacket glass. The microstructure in the cladding of the AR-HCF is found to be free of damage for incident powers from a few hundred watts up to nearly 1.5 kilowatts where the deliberate deviation of the input laser beam only causes the deterioration of the coupling efficiency. The possible region of the laser damage threshold for the internal microstructure of the AR-HCF is discussed.

The AR-HCF used in our experiment is made by the stack-and-draw technique using Heraeus F300 fused silica tubes. The cladding of the AR-HCF consists of seven capillaries with an average inner diameter of 17 µm and thickness of approximately 335 nm (inset of Fig. 1), and the core diameter is approximately 35 µm. The 7-capillary cladding design of the AR-HCF is preferred to balance the single-mode guidance [20] and the large modal area for the high-power laser delivery purpose. The attenuation of the AR-HCF is measured by the cut-back method as shown in Fig. 1 and 0.13 dB/m is obtained at 1080-nm wavelength.

 

Fig. 1. (a) Measured attenuation of the AR-HCF. Inset: SEM picture of the AR-HCF. The core diameter is approximately 35 µm, and the average core wall thickness is approximately 335 nm. (b) Schematic of the laser delivery setup: L1, L2, L3, L4 are coated aspherical lenses with f1 = 100 mm, f2 = 150 mm, f3 = 120 mm, f4 = 100 mm; M1, M2 are dielectric mirrors, which have a reflectivity >99.5%; M3 is a back polished mirror used as a sampler (transmittance <<1%); a sampler (99:1) is used to monitor the output of laser source to calibrate the incident power at the input end of the AR-HCF in real time. PM1, PM2 are power meters which monitor the fluctuation of input power for calibration, and the transmitted laser power out of the AR-HCF, respectively. The AR-HCF is mounted on a silica V groove loaded in a homemade gas cell. Inside the shell of the gas cell are carved water channels, and water circulation is applied for thermal management in the experiment.

Download Full Size | PDF

The experimental setup is shown in Fig. 2. A commercial CW fiber laser (Maxphotonics MFSC-1500W) is used with a maximum output power up to 1500 W at 1080 nm. According to the specification, the M² of the laser output is less than 1.2. Similar to in our previous work [18], a 4-f lens system is used to scale the laser beam to match with the fundamental-like mode of the AR-HCF to optimize the coupling efficiency. A sampler is used to monitor the incident power and calibrate the coupling efficiency in real time. At the AR-HCF output end, the delivered laser power and beam quality are characterized after the collimating lens (L5).

 

Fig. 2. (a) Measured output power and coupling efficiency of the AR-HCF with low-refractive-index coating; (b) temporal measurement for 1150-W incident laser power for 30 minutes without any sign of AR-HCF damage.

Download Full Size | PDF

At the laser incidence, we strip off approximately 15 cm of coating from the fiber end which was demonstrated effective to prevent the coating from burning due to the leaked laser power at the incident end. However, laser-induced coating burning was still found at random locations along the fiber length when the incident laser power reached 230 watts and beyond. For a lower incident power, coating damage barely occurred for a laser delivery of hours. In most cases, the burning of the coating would cause catastrophic damage to the AR-HCF.

To resolve the coating burning issue, a low-refractive-index coating was applied in the fiber fabrication. The inversion of the refractive index contrast results in the total internal reflection at the interface of the jacket glass and coating which would effectively keep the stray light away from the coating. Figure 2 shows the measured transmitted laser power through the lower-index polymer-coated AR-HCF of 5-m length. The coupling efficiency was kept at approximately 80% by manual adjustment. As previously reported in [18], laser-induced thermal effects, including the reduction of the focal length of lens, and thermal drifting of the XYZ stage on which the fiber is mounted would both contribute to a noticeable degradation of the coupling efficiency in the short term. To keep the optimized coupling efficiency with an increasing laser power, the position of the incident end of the AR-HCF was tuned in real time, particularly along the optic axis in the fiber length direction due to the varied laser spot size. For the incident power of 1150 watts, stable laser delivery of 785 watt was tested for over half an hour [Fig. 2(b)]. However, in the test of a higher incident power, the damage of the AR-HCF began to appear occasionally and it was noted that all burned points were found in the silica V-groove inside the gas cell where the end of the AR-HCF was mounted and protected.

In the gas cell, the bare fiber sticking out from the V groove was shorter than 1 cm to avoid the wobble of fiber end causing the instable coupling. The gas cell seals the AR-HCF inside from the dust and possible pollution from the environment. More importantly, it could create a vacuum in the AR-HCF when necessary to further suppress the gas optical nonlinearity. As shown in Fig. 3, the bare AR-HCF was clamped by two identical V grooves in the gas cell. A metal V groove was found the be hazardous given that the laser induced air plasma was easily excited on the metal surface even for a low incident power of approximately 100 watts. As a result, the V groove made of fused silica was preferred instead.

 

Fig. 3. Schematic of assembly of the AR-HCF in the silica V-grooves. Inset: the air gap in between could be thin enough to function as a slot waveguide.

Download Full Size | PDF

 

Fig. 4. (a) Picture of the melted V groove and AR-HCF when delivering a 1150-W laser beam; (b) zoom-in of the damaged V groove under a microscope; (c) AR-HCF end remains intact.

Download Full Size | PDF

When the incident power was more than 1150 W, a weak light flash accompanied by cracking sounds was occasionally found to exit the gas cell and then the laser delivery of the AR-HCF ceased. As shown in Fig. 4, part of the AR-HCF on the V groove melting with the silica plate caused the failure of laser delivery rather than any damage to the fiber end or the part outside the V groove. It is noted that before the assembly of the gas cell, the V-groove was supersonically cleaned in a 99.9% pure ethanol bath and dried, and the AR-HCF was wiped with ethanol multiple times to remove the pollution of dust. The cause of melting damage is still an open question, but we tend to attribute it to the ionization of the thin air layer between the bare AR-HCF and V groove, where both laser speckle in the fiber jacket glass and the slot waveguide effect [21] could contribute to the local enhancement of the electric field in the air gap.

The air ionization threshold when pumped by a CW laser is approximately of the order of 100 W/µm² [22]. Therefore, an air gap of micrometer width must require a minimum laser intensity beyond 100 W for air plasma excitation. For 1.5-kW incidence, an 80% coupling efficiency allows 300 W of laser power to leak from the core into the jacket glass of the AR-HCF. The complex speckle pattern in the jacket may give rise to a high local field near the air gap out of constructive interference. In the air gap, the electric field could then be enhanced by ${{n_1^2} / {n_0^2}}$ times because of the slot waveguide effect. The ionized air could quickly heat up and even melt the surrounding silica glass, where the deformation would result in even stronger scattering and eventually trigger the damage of the AR-HCF.

To verify our hypothesis, we used the refractive index matching liquid to fill in the gap between the V groove and fiber to get rid of the possible slot waveguide effect. By using this method, we successfully demonstrated 1-kW laser delivery through a 5-meter AR-HCF for incident power of 1430 watts for 1 minute. After 1 minute, the laser delivery failed again because of the melting in the V groove. The index-matching liquid was found to have completely evaporated already in this case. The loose mechanical contact between the inner wall of water-cooled gas cell and the silica V groove failed to efficiently carry the heat away.

Unlike coating and jacket glass, the microstructure of the AR-HCF cladding seemed to remain intact regardless of incident power where the average of the silica structure thickness is only approximately 335 nm. In the AR-HCF, the modal field overlap with the core wall can be reduced to 0.002% [11]. As the CW laser damage threshold of silica glass reaches 4.75 kW/µm² [23], the trivial modal power density distribution in the microstructure of the cladding predicts an unrealistic damage threshold up to 1 GW. Therefore, we argue that the damage of the AR-HCF microstructure must result from the uncoupled laser power at the incidence leaking through the thin core wall and breaking down the glass structure.

A toy model is set up to illustrate the propagation of leaked laser power in the AR-HCF and predict the damage threshold of the core wall. We simply assume the uncoupled laser beam propagates and diffracts inside the core of the AR-HCF as in the free space where the reflection of the cladding is ignored. For a fundamental Gaussian incident beam, the intensity distributed at the core wall is approximated as a function of z as

For $\eta$ coupling efficiency, the total incident power is simply as

To find the possible ${I_{th}}$, we transversely scanned the focused laser beam across the end of the AR-HCF in the experiment when the incident power was 1150 W. Unfortunately, the offset of the incident beam could not bring in any AR-HCF damage but only deterioration of the coupling efficiency from 80% to nearly 0%. It is noted that the degraded coupling efficiency could be recovered to 80% again by manual adjustment after the scanning.

In our experiment, the focused beam waist at the incident end of the AR-HCF and the maximum intensity were calculated as 12.6 µm and $6.01 \times {10^8}$W/cm². We substitute this intensity into Eq. (3) and obtain the corresponding incident power as a function of the coupling efficiency as the lower limitation of the damage threshold. In Fig. 5, the damage threshold of the AR-HCF for 80%coupling efficiency is predicted to be above 25.52 kW and for a coupling efficiency higher than 95%, the AR-HCF possibly remains intact for over 97-kW incident power that could already ionize the air in the core [22]. In this case, a vacuum in the AR-HCF would be necessary to further increase the delivered laser power.

 

Fig. 5. Calculated threshold of the total incident power that possibly damages the microstructure of the AR-HCF as a function of the coupling efficiency.

Download Full Size | PDF

In summary, two types of laser-induced fiber damage were experimentally observed in our demonstration of laser delivery exceeding 1 kilowatt through a 5-meter AR-HCF at 1 µm. By scanning the focused high-power laser beam across the fiber end, we deduce that 25-kW incident power would not cause any damage to the microstructure of the AR-HCF with an 80% coupling efficiency. In the future, more effort would be necessary to explore the laser-induced air ionization phenomenon in an AR-HCF which plays a key role limiting the potential of the AR-HCF in the high-power laser delivery.