1. Introduction

Fiber-delivered high-power ultraviolet laser light has numerous applications, extending beyond its use in the semiconductor industry to e.g. experiments on coherent manipulation of trapped ions for precision spectroscopy and optical clocks [1,2], quantum information processing [3], and trapped ion simulators [4], in which excellent beam quality and pointing stability are desirable. Synthetic fused silica has been widely used as an optical fiber material for its high transparency in the visible and near-IR spectral range as well as its mechanical robustness, high chemical resistance and good biomedical compatibility. Silica-based optical fibers are the mainstay of optical telecommunications and many other applications. However, in some demanding spectral regions, such as the mid-infrared (mid-IR) and the ultraviolet (UV), silica fibers exhibit high attenuation. The development of soft glass materials with reduced phonon absorption has extended the spectral range of optical fibers at long wavelengths [5]. At ultraviolet wavelengths no alternative glasses have been identified and silica fibers continue to exhibit high attenuation as well as fluorescence and photodarkening [6–8].

Multiphoton absorption [9,10] and optically-induced color center formation are primary causes of the poor performance of silica-based fibers in the UV, along with increasing material absorption and Rayleigh scattering. The formation of color centers is associated with photodarkening and this is referred to as solarization [8]. Solarization of silica fibers can be mitigated by loading hydrogen in the silica network to reduce the formation of color centers [11,12]. This requires loading the fiber at high-pressure with hydrogen gas and subsequent curing via exposure to UV radiation [13,14]. A large core area is advantageous as it reduces peak intensities within the fiber. Recently multi-trench silica fiber [11] and endlessly single-mode photonic crystal fiber silica fibers (PCFs) [13–16] have been proposed and demonstrated for UV light transmission. In PCF, the silica cladding comprises a periodic air-hole array that provide channels near the core for the loading of high-pressure hydrogen gas and make the post-processing of fibers more convenient and efficient [13,14]. Hydrogen passivation reduces the solarization effect and enables transmission of continuous wave incident UV laser power up to the milliWatt power level. Solarization-free delivery of 200 mW at 313 nm wavelength and 10 mW at 235 nm were recently reported; however, to maintain long-term fiber transparency for such laser intensities is still challenging at deeper UV wavelengths [10]. Karlitschek et al. reported that hydrogen loading could enable stable transmission in hydrogen-loaded multimode silica fiber for about 100 μJ, 3 ns incident pulses at 266 nm [11,12], where the laser repetition rate was 10 Hz and stable transmission was observed only for about 5000 pulses. It is noted that hydrogen passivation reduces the formation of a range of color centers but cannot change the intrinsic material absorption in the UV. The intrinsic loss of commercial solarization-resistant silica fiber typically increases from 30 dB/km to approaching 1 dB/m from 400 nm to 200 nm wavelength [17].

In hollow-core fiber (HCF) light is confined in an air/vacuum core rather than in solid silica. Anti-resonant hollow-core fiber (AR-HCF) is a type of HCF with a simple cladding design and spectrally limited transmission windows of low-loss [18,19]. In AR-HCF, the thin core wall can be considered a Fabry-Perot resonator at grazing incidence and the spectral dependence of light guidance is broadly explained by an ARROW model [20]. Because of the greatly reduced overlap between the guided light and the solid material in the cladding, AR-HCFs exhibit much lower material absorption, optical nonlinearity, and a higher optical damage threshold than solid-core fibers [18].

Low attenuation in silica AR-HCFs was first demonstrated in the spectral range of 3 - 4 μm where bulk silica has absorption rising from around 50 dB/m to nearly 1000 dB/m [21]. A few tens of dB/km or lower attenuation has now been reported from visible to beyond 3 μm wavelengths [18,19,22,23]. However, demonstrating low attenuation at even shorter wavelengths has been proven harder. ‘Kagome’ fiber was tested for UV laser transmission at 355 nm in 2009 [24]. 2 dB/m attenuation was measured at 355 nm and a rich content of higher-order modes were observed at the fiber output [24]. In 2014 Hartung and colleagues designed an AR-HCF for UV and demonstrated single-mode guidance in three transmission bands that partly covered from UVA to UVC, with minimum attenuation rising from about 1 dB/m to over 10 dB/m [25]. Similar loss performance was also reported in [26]. In 2015, another design of AR-HCF improved the performance to 3 dB/m average attenuation over the wavelength range from 216 nm to 310 nm [27]. Meanwhile, over 100 hours UV laser transmission in a ‘Kagome’ type AR-HCF at 280 nm demonstrated solarization-free transmission in HCF with 15 mW CW input power [28]. However, the simulations presented suggested that the transmission bands of this ‘Kagome’ fiber in the UV were limited to a few nm bandwidth. Fiber attenuation of 0.8 dB/m at 280 nm was measured by cutbacks of a few meters.

In this paper, we report two AR-HCFs with useful properties in the UV. Attenuations were measured to be about 0.1 dB/m at 218 nm and 0.26 dB/m at 355 nm in one fiber, by a cutback from 33.6 m to 8.4 m. In a second AR-HCF designed for 266 nm, 2 meters of AR-HCF was tested for delivery of high repetition-rate nanosecond laser pulses and compared with commercial high-OH and solarization-resistant silica multimode fibers for UV. By delivering over 0.46 µJ pulses at 30 kHz repetition rate for over an hour, we demonstrate that stable solarization-free laser transmission at 266 nm in AR-HCF does not require Hydrogen passivation of the fiber material. Comparing with experiments performed using commercial fibers we show that the single-mode AR-HCF clearly outperforms commercial multimode silica fibers at this wavelength.

2. Low-loss AR-HCFs for UV

We report two different AR-HCFs. Figure 1(a) shows an SEM picture of the first AR-HCF. The fiber was fabricated using the stack-and-draw technique as described in [21] with low-OH F300 fused silica as starting material. The fiber outer diameter is 109 μm and core diameter is about 17 μm defined by the diameter of inscribed circle fitting to the core region. The average core wall thickness and inner diameter of cladding capillaries are 132 nm (min. 100 nm, max. 165 nm) and 7.3 μm (min. 6.5 μm, max. 8.4 μm) respectively. SEM images of each capillary at higher magnifications can be found at [29].

 

Fig. 1 (a) SEM picture of the first AR-HCF. The core diameter is about 17 μm and the average thickness of core wall is 132 nm. (b) and (c) are the near-field-pattern images at the output of AR-HCF after 33.6 m and 8.4 m respectively in the cutback measurement, recorded using a 10 nm bandpass filter centered at 355 nm.

Download Full Size | PDF

We measured fiber attenuation using two techniques – traditional cutback measurement and a non-destructive ‘insertion loss’ method. In the non-destructive attenuation measurement, we calculated the attenuation of the AR-HCF by directly and repeatedly comparing multiple transmission spectra through AR-HCFs of different lengths under independent optimized coupling conditions and demonstrated the results consistent with the cutback measurement. The related data can be found at [29].

In both measurements, AR-HCF was rewound in loops of 1 m diameter and spread on an optical bench to minimize the risk of bend loss. A laser-driven plasma light source (LDLS Eq. (-99) was used with an emission spectrum from deep UV to near-IR. An ANDO optical spectral analyzer (OSA) (AQ6315E) and an Ocean Optics spectrometer (USB4000) were both used to record the transmitted spectrum, covering spectral ranges from 350 nm to 1750 nm and from 180 nm to 850 nm respectively. However, our system is not able to characterize optical spectra accurately at wavelengths below 200 nm.

In both fiber attenuation measurements, we used fiber butt coupling for excitation. At the input end of the AR-HCF, a 15 cm multimode silica fiber of 10 μm core diameter (Thorlabs FG010LDA, 0.1 NA) was used to couple light into the AR-HCF. Such excitation of AR-HCF was found to be more efficient than lens coupling for our broadband light source. This method also helped reduce the excitation of cladding modes of AR-HCF and avoid stray light interfering with the attenuation measurement.

When measuring the fiber transmission spectra with the OSA, we used a common bare fiber adaptor and inserted the output of AR-HCF into the instrument. We observed a stable and consistent light collection efficiency into the spectrum analyzer. When using the Ocean Optics spectrometer, which has a much shorter internal optical path, a significant insertion instability was found when using a conventional fiber adaptor, which we attribute to the small numerical aperture of AR-HCF. Therefore a 5 cm long high-OH multimode silica fiber (Thorlabs FG050UGA, 50 μm core diameter) that has a 0.22 NA (as preferred for use with the Ocean Optics spectrometer) was used to couple the transmitted light of AR-HCF into the Ocean Optics spectrometer. The AR-HCF was butt-coupled with the end of a short multimode fiber. The gap between the two fiber ends was minimized while observing using a lens. The butt-coupling efficiency was adjusted to maximum by monitoring the overall intensity measured by the Ocean Optics spectrometer. In this way, we were able to obtain a stable and consistent coupling efficiency between AR-HCF and Ocean Optics spectrometer.

In Fig. 2(a), the transmission spectra measured by the OSA and Ocean Optics spectrometers in the cutback measurement are normalized by their 33.6 m transmission intensities at 350 nm. The attenuation curves shown in Fig. 2(b) are calculated independently from the two sets of un-normalized data. The attenuation curves measured by the two instruments overlap well between 350 nm and 370 nm. In Fig. 2(b), the attenuation between 412 nm and 526 nm stays flat around 0.12 dB/m. At wavelengths shorter than 412 nm, the attenuation increases gradually to 0.4 dB/m at 310 nm. The plateau of attenuation curve between 280 nm and 310 is caused by background noise (stray light) as shown in Fig. 2(a). In the fundamental transmission band, the attenuation at 355 nm is 0.26 dB/m; and in the higher-order transmission band, the minimum attenuation is below 0.1 dB/m around 218 nm. These attenuation values were confirmed by repeated non-destructive attenuation measurement using 8.4 m and 25.2 m AR-HCFs [29].

 

Fig. 2 (a) Transmission spectra recorded by the OSA and Ocean Optics spectrometer for 33.6 m and 8.4 m lengths. Two groups of measured transmission spectra are normalized by their 33.6 m fiber transmitted intensities at 350 nm respectively. (b) Calculated attenuations based on raw data. The attenuation at 355 nm is 0.26 dB/m; and the averaged minimum attenuation is measured as 0.08 dB/m around 218 nm.

Download Full Size | PDF

The output near-field-pattern after 33.6 m fiber transmission was observed to be fundamental-mode like at 355 nm as shown in Fig. 1(b); higher-order modes started appearing [Fig. 1(c)] as the fiber was cut to 8.4 m. The appearance of higher-order modes in the near-field pattern correlates with transverse offset in the butt-coupling conditions. To optimize the single-mode performance, the dimension of the cladding tubes needs to be expanded so that the filtering of higher-order modes in the core can be achieved by phase matching with cladding modes [30,31].

3. Pulsed laser delivery in AR-HCF at 266 nm

3.1 AR-HCF for 266 nm

Figure 3(a) shows the SEM picture of a second AR-HCF fabricated for laser delivery of pulsed laser light at 266 nm. This fiber has the same outer diameter of 109 μm as in Fig. 2(a), and is drawn from the same group of preforms. Due to different fabrication conditions, the cladding of this fiber is less expanded but more uniform in size. With bigger gaps between the cladding capillaries, the core diameter expands as does the thickness of the core wall. As a result of the thicker core walls, the transmission bands shifts towards longer wavelengths as explained by the ARROW model [18,20,21], which was confirmed by two cutback measurements. Figure 3(d) shows the measured fiber attenuations, where the data are from two cutbacks from 25.1 m to 5.3 m (for the longer wavelength transmission band) and 19.8 m to 5.7 m (for the shorter wavelength transmission band) respectively.

 

Fig. 3 (a) SEM picture of the second AR-HCF designed for 266 nm laser delivery; (b) and (c) are the near-field-pattern images at the output of AR-HCF after 19.8 m and 5.7 m respectively, with no filter in use; (d) calculated attenuation from cutback measurements. 0.7 dB/m and 0.83 dB/m are measured minimum attenuations of two bands at 263.7 nm and 380 nm.

Download Full Size | PDF

In Fig. 3(d), the measured minimum attenuations are 0.7 dB/m and 0.83 dB/m at 263.7 nm and 380 nm respectively. At 266 nm and 355 nm, the measured values were 0.87 dB/m and 0.97 dB/m. We attribute the higher loss of the second AR-HCF to the gap between capillaries in the cladding being too large [31]. For AR-HCF of 7 capillary cladding design, when the ratio$r_{clad}/r_{core}$decreases below 0.6 ($r_{clad}$is the inner radius of cladding capillary and$r_{core}$the radius of fiber core), the leakage loss increases [31]. It is noted that both near field patterns at the output ends of short and long AR-HCFs are fundamental-mode like. We assume that any higher-order modes excited quickly dissipate along the fiber as a consequence of differential modal attenuation [32]. In the laser delivery experiment in Section 3.2 and 3.3, the near-field pattern was still found to be fundamental-mode like after 2 m length when substantial power was coupled into the fiber using different coupling conditions, and no higher-order mode could be observed. However, it is noted that the output mode pattern distorts when the fiber is compressed near the output end, and recovers when the pressure is removed.

3.2 Laser delivery experiment

We used Advanced Optowave AONano 266-5-30-V as a laser source at 266 nm. 266 nm laser pulses are generated by frequency-quadrupling a high-power Nd:YVO₄ laser, and the maximum average power output at 266 nm exceeds 5 W. The repetition rate is 30 kHz and the pulse length is 17 ns. No effort was made to characterize the polarization of laser beam or control the state of polarization of the delivered light either in the fiber design or in the experiments.

According to the laser manufacturer, the beam quality M² is less than 1.3 with roundness above 80%. In Fig. 4 inset (a), the recorded time-averaged far-field pattern at the laser output shows a two-lobe structure, and the beam width was measured as 3.5 mm by 2.8 mm. For a more efficient coupling with AR-HCF, a 50 μm pinhole was used for spatial filtering of the incident laser beam in the fiber coupling setup as shown in Fig. 4. All the lenses in the coupling setup were made of fused silica and were anti-reflection coated for 266 nm. It is noted that all focal lengths of lenses are those provided by the manufactures for visible wavelengths.

 

Fig. 4 Schematic of laser coupling setup. All lenses were made of fused silica glass, anti-reflection coated for 266 nm. All the focal lengths given are provided by the manufacturer for visible wavelengths. The pinhole is Newport PH50 made of Molybdenum designed for high- energy laser application. Four aluminum coated UV reflection enhanced silver mirrors were also used in the coupling setup and not displayed. Inset (a): far-field pattern directly recorded by a camera beam profiler at the output of laser; inset (b) near-field pattern of laser beam at the output of AR-HCF at 266 nm recorded after a lens.

Download Full Size | PDF

After the lens f₁, the beam waist at the focus was measured to be roughly 70 μm in diameter. However the measured transmission of light through the 50 μm pinhole was only up to 25% for low-power laser outputs below 200 mW, where the measured power fluctuation was about 4% (consistent with the 3% pulse-to-pulse stability specified by the manufacturer). As the laser output power increased, this transmitted power through the pinhole quickly dropped and the pinhole was typically burned and damaged at approximately 1 W incident power. We attribute the reduced transmission through the pinhole at higher incident powers to the pointing instability of the laser beam which degrades at higher power.

When the fiber coupling was optimized at low power using the pinhole, the laser power transmitted through the fiber remained the same when the pinhole was removed, and could not be further improved by fine adjustment. When the incident power was above 1W without use of pinhole, the fiber showed no sign of optical damage; however, because of deterioration of laser beam pointing stability, the delivered laser power in AR-HCF became unstable. Therefore, for a reliable measurement, all laser delivery results reported in this section are at low laser output powers below 200 mW. The temporal pointing instability of laser beam spot at the focal plane of lens f₁ was recorded by CCD camera when the laser output power was measured as 202 mW and is shown in Visualization 1.

Two commercial silica multimode fibers were also tested in the laser delivery experiments for comparison. They were high-OH silica multimode fiber (OH-MMF) (Thorlabs FG050UGA, 50 μm core diameter, 0.22 NA) and solarization-resistant silica multimode optical fiber (SR-MMF) (Thorlabs FG105ACA, 105 μm core diameter, 0.22 NA). Both fiber attenuations were reported to be around 0.24 dB/m at 266 nm due to the background material absorption and scattering loss (excluding solarization) [17].

3.3 Experiment results

Figure 5 summarizes the experimental results of pulsed laser delivery at 266 nm through AR-HCF and the two MMFs. With the pinhole, the maximum coupling efficiency into the AR-HCF was up to 60.2% and the transmitted power through the AR-HCF remained constant after removing the pinhole. The same setup was also used for the two types of MMFs.

 

Fig. 5 (a) Temporal evolution of transmissions in AR-HCF and two types of MMF. The incident powers are measured as 34.1 mW, 35.4 mW and 34 mW for AR-HCF, SR-MMF and OH-MMF respectively. (b) Temporal evolution of transmissions in AR-HCF with and without spatial filtering. The slight decline of transmission without pinhole is due to the reduced incident power from 120 mW to 107 mW in one hour. (c) Typical fluorescence of OH-MMF for about 35 mW incident power. The peak around 530 nm is residual light from the second harmonic generation stage of the laser.

Download Full Size | PDF

Figure 5(a) shows the temporal evolution of transmission in AR-HCF and two types of MMFs for different fiber lengths. The incident average powers were 34.1 mW, 35.4 mW and 34 mW for AR-HCF, SR-MMF and OH-MMF respectively. For the AR-HCF, about 13.75 mW was measured using a thermal meter (Gentec PS-310, power range 2 mW - 3 W) at the output end of 2 m AR-HCF giving an overall transmission efficiency of 40%. The delivered laser power through the fiber was measured every 5 minutes and was stable for one hour. For the MMFs, in order to avoid solarization during setup a low incident power (less than 5 mW) was used for the coupling alignment. For 2 m SR-MMF, up to 60% transmission efficiency (including the coupling efficiency and fiber loss) was observed for 35 mW incident power but the transmission dropped to 21.5% in the first 5 minutes. After 1 hour, the transmission eventually reduced to about 7% and became more stable. For OH-MMF, no power could be detected at the output of 2 meter fiber length for 35 mW incident power. For a greatly reduced 10 cm length, the transmission of OH-MMF quickly dropped from less than 33% at the beginning to around 3% due to solarization in one hour. For both MMFs, bright red fluorescence was visible along the whole fiber length. Figure 5(c) shows a measured typical fluorescence spectrum at the output of 10 cm OH-MMF. The peak at 530 nm is from the laser source. The fluorescence in SR-MMF under 35 mW incident power exhibits a similar spectrum as shown Fig. 5(c) but weaker. In AR-HCF no fluorescence is either visible to the eye or detected.

Figure 5(b) compares the transmission of AR-HCF with and without pinhole for one hour, where AR-HCF exhibits no sign of degradation over 120 mW incident power (peak power is over 235.3 W). For 120 mW incident power, about 10.4 mW output power was measured after 2 m fiber length. The slight reduction of transmission with time is due to the decline of incident power from 120 mW to 107 mW. The near-field-pattern at the output of AR-HCF was found to be fundamental-mode like in the laser delivery test [similar as in Fig. 4 inset (b) and (c)].

In many real-world applications, far longer exposure times would be required, possibly thousands of hours. In considering how those longer timescales might affect our results, we note that the fact that the light is guided in an air core affects the sensitivity to solarization in two ways. First, the peak intensity within the glass is greatly reduced. Consequently, the rate of solarization of the glass is reduced at least in proportion, greatly extending the time before the relevant glass (the core wall) suffers from degradation in transparency. Secondly, the overlap of the guided light with the core wall is tiny –1 part in 10,000 or possibly less [18,33]. Therefore, even when the glass is solarized, the effect of that on the performance of the fiber will be very small, possibly even too small to easily measure. It may be that the limiting factor is degradation of the physical properties of the fiber (e.g. increased brittleness of the fiber core wall) rather than reduction in the optical transmission. In order to analysis the limitations of the fiber we would therefore need a substantial study testing different properties of the fiber, possibly using an accelerated methodology and a higher power laser. Such a study is beyond the scope of this work.

4. Conclusions

In summary, we demonstrate AR-HCF with low-loss transmission bands covering part of UVC and the whole UVA and visible spectral regions in one single AR-HCF. About 0.1 dB/m attenuation at 218 nm and 0.26 dB/m at 355 nm were achieved. A lower loss performance of AR-HCF is expected by optimizing$r_{clad}/r_{core}$ [27], but more effort will be needed to overcome the challenges in fiber fabrication to obtain AR-HCFs with ultra-thin core walls and better longitudinal and transverse uniformities.

Solarization-free 266 nm nanosecond pulsed laser delivery has been successfully demonstrated in an optical fiber for the first time. When delivering 0.46 μJ pulses at 30 kHz repetition rate, AR-HCF shows stable single-mode transmission free of solarization and silica fluorescence for one hour, exhibiting an mediaoverwhelming advantage over commercial solarization-resistant large-core-area silica fibers in this spectral region. All data underlying the results presented in this paper can be found at [29].