1. INTRODUCTION

The efficient removal of high-power excess cladding light without inducing significant core loss and beam quality degradation is of significant interest for the production and operation of robust, high-power fiber lasers and amplifiers.

Light propagating in the cladding of a fiber laser can originate from a range of sources. It can be due to unabsorbed pump radiation that is transmitted through an active fiber, amplified spontaneously emitted light (ASE) leaving the core, copropagating signal radiation leaked from the core due to lossy splices, bend losses, or backward propagating signal radiation due to reflections external to the fiber laser, as illustrated in Fig. 1. For these reasons the cladding stripping method must account for light of varying power levels, propagating angles, direction and wavelength, without being thermally compromised.

 

Fig. 1. Mechanisms for unwanted cladding light: (a) unabsorbed pump light and ASE in an active fiber, (b) splice loss coupling light (signal and ASE) from the core into the cladding, (c) reflected light entering the cladding during materials processing.

Download Full Size | PDF

In materials processing applications, the laser is commonly exposed to substantial backreflections from, for example, highly reflective metal surfaces [1,2]. In such cases, isolation is necessary to prevent instability in the fiber laser. This isolation is typically achieved by applying a defocus to the beam, launching the beam in an off-center direction or by beam quality degradation. In these cases, there may be a substantial amount of backward propagating radiation coupled into the cladding of the fiber. It now becomes necessary to be able to remove this radiation from the cladding so that it does not propagate to and damage the various fiber laser components or even the pump laser diodes. The power level of the radiation in the cladding can be in the many kilowatts of average power.

At such high power levels, polymer-based stripping methods [3–5] are much more likely to fail. Furthermore, standard polymers are unsuitable for use in thulium, holmium, or erbium sources where the 1.5–2 μm radiation is subject to strong attenuation in the polymer, leading to localized heat zones, and rapid degradation of the polymer.

A number of groups have demonstrated all-glass cladding light strippers (CLSs) fabricated using mechanical polishing or ${\mathrm{CO}}_2$ laser processing [6,7]. However, there is little information provided about the exact process, the losses incurred, and the impact on the radiation propagating in the core. One group demonstrated the ablation and redeposition of glass to create a scattering surface and achieved a device with a cladding loss of 13 dB over 8 cm [8]. Another group used soft metals as an absorption stripping method suitable for high-power operation and demonstrated up to 8.4 dB cladding loss for device lengths of 7 cm [9]. An alternative technique has been to etch the fiber and create a rough surface [10–12]. While yielding excellent results, this process involves the use of hydrofluoric acid (HF) and a long process time to achieve reproducible results. A chemical-free approach is preferred for a production environment.

The different techniques previously demonstrated provide varying magnitudes of loss per length of device, as shown in Table 1.

Table 1. CLS Loss for Different Fabrication Techniques

View Table | View all tables in this article

In comparison, ${\mathrm{CO}}_2$ laser processing of optical fibers enables a contamination-free, rapid, noncontact manufacturing process of fiber-based devices [13]. The parameter space of possible cladding surface modifications achievable with this technique is extensive. For these reasons, we focus on the fabrication and comprehensive characterization of a specific subset of devices that are easily accessible in terms of general tolerances and fabrication.

The measurement of the CLS loss for different radiation profiles enables the fabrication of devices tailored for specific applications. A necessary step when fabricating CLS devices is to characterize the losses induced in the core. We propose a relatively simple method to characterize the devices that are fabricated. The induced core loss of a fiber with a 20 μm diameter, 0.09 NA core was measured to be $<0.008\pm 0.006\mathrm{dB}/\mathrm{cm}$ from 1950–2100 nm. The cladding loss is described in more detail for various processes.

Finally, we present a device that provides a cladding stripping loss of $>20\mathrm{dB}$ in a 400 μm diameter fiber. The output from a 300 W laser diode is launched into the cladding of the fiber at the input to the cladding stripper, and 2.7 W of remaining radiation is measured at the output. The thermal images of the packaged device are presented.

2. EXPERIMENTAL ARRANGEMENT

A. Device Fabrication

The CLS devices were fabricated using a 55 W, pulse-width-modulated ${\mathrm{CO}}_2$ laser (Coherent Scientific), operating at 10.6 μm. The laser was utilized to ablate channels of various depths, ranging from 5 to 75 μm on one side of an optical fiber. No rotational manipulation of the fiber was required in this process.

The optical fiber was mounted on a vertical stepper motor controlled stage using V-groove clamps, as shown in Fig. 2(a). A strain gauge was used to monitor the optical fiber tension. The tension was set to a value of 2 N and monitored throughout each CLS fabrication process. A drop in tension could indicate overheating of the fiber and subsequent core loss.

 

Fig. 2. (a) Experimental setup for the fabrication of a CLS. The laser beam was focused onto a vertically mounted fiber using a 50 mm focal length ZnSe lens which was scanned across the static ${\mathrm{CO}}_2$ laser beam. The fiber was mounted in v-groove clamps, and the tension is monitored with a strain gauge. A 2X telecentric lens orientated at 90 deg to the ${\mathrm{CO}}_2$ laser beam monitored the process and was used to measure the maximum cut depth. (b) Ablation of an optical fiber by scanning the focused ${\mathrm{CO}}_2$ laser beam across a 400 μm fiber as it was translated vertically, (c) viewed at 90° to beam after fabrication.

Download Full Size | PDF

A 25.4 mm diameter, 50 mm focal length ZnSe lens was used to focus the beam onto the optical fiber with a measured beam diameter of $\sim 50\mathrm{\mu m}$. The lens was mounted on a motorized stage enabling the lens to be swept across the beam over a distance of $\sim 1\mathrm{mm}$. This optical layout was chosen for repeatable device fabrication as it ensured that the ablation depth was not critically dependent on the alignment of the beam to the center of the optical fiber and the long working distance relative to the fiber diameter ensured that there was no critical dependence on the distance between the fiber and lens.

A copropagating HeNe laser was used to assist in the alignment of the swept ${\mathrm{CO}}_2$ beam to the fiber. The fabrication process was monitored visually using a CCD and a 2X telecentric lens with an 800 nm high-pass optical filter. The fiber was backlit by a collimated 800 nm LED. A movie of the fabrication process viewed at 45° to the incident ${\mathrm{CO}}_2$ laser beam can be accessed in Visualization 1.

Before fabrication, a portion of each fiber was mechanically stripped and then cleaned to remove the polymer coating. Following this, there was no other handling of the exposed glass in the process. The fiber was held by clamping over the original coating.

Control over the ablation depth of the surface features was achieved by varying the ${\mathrm{CO}}_2$ laser pulse duration from 7.8 μs at an average power of 170 mW, up to 15.68 μs at 920 mW. The other process parameters were held fixed; pulse interval of 1.31 ms, ZnSe lens sweep rate of $2\mathrm{mm}{\mathrm{s}}^{-1}$, and vertical stage velocity of $10\mathrm{mm}{\min }^{-1}$. These parameters resulted in the ablation depths shown in Fig. 3 for fibers of 250 and 400 μm outer diameter (OD).

 

Fig. 3. Maximum ablation depth achieved on the surface of 250 and 400 μm OD optical fibers as a function of ${\mathrm{CO}}_2$ laser pulse duration. The laser output ranges from 7.8 μs at an average power of 170 mW, to 15.68 μs at 920 mW. The pulse interval was constant at 1.31 ms. The feature depth was measured using a telecentric lens at 90° to the incident ${\mathrm{CO}}_2$ laser beam.

Download Full Size | PDF

The maximum ablation depth is greater for the 250 μm OD fiber compared to the 400 μm OD fiber as the evaporation rate of silica is temperature dependent [14] and the smaller volume reaches a higher temperature for a fixed set of ${\mathrm{CO}}_2$ laser ablation parameters.

In order to simplify the analysis we investigate only devices fabricated with varying pulse duration, although it should be noted that there are numerous combinations of these parameters that will yield the same ablation depth.

B. Method for Device Characterization

1. Cladding Loss Measurement

The optical loss of the glass cladding due to the CLS devices under investigation is a function of multiple parameters. These include ablation depth, length of the CLS, propagating angle of the radiation in the cladding fiber, and pitch of the ablation channels. In this paper, we detail measurement of the cladding loss as a function of incident angle of light and device length for CLS devices with various ablation depths at a fixed pitch.

To eliminate the noise from stray radiation that may be guided by the core, a coreless fiber with a 400 μm diameter, 0.46 NA cladding was utilized. This is representative of the cladding parameters for common fibers used to construct high-power fiber lasers.

The loss was measured during the fabrication process by launching light into the optical fiber at a range of angles. To facilitate this, one fiber end was mounted on a computer-controlled rotation stage, which was located 1.5 m from a collimated LED. The fiber was mounted such that the cleaved end-facet was located on the axis of rotation of the rotation stage. Irises were placed in the path of the LED to ensure that the radiation incident on the fiber was well collimated. Care was taken to ensure that the fiber was loosely spooled so as to minimize any mode-scrambling that could degrade the purity of the launched signal. By positioning the fiber near the CCD sensor without any lenses, we verified that the output radiation was propagating with the angle that corresponded to the launch conditions. The other end of the fiber was cleaved and a 4.5 mm aspheric lens with an acceptance NA of 0.48 was used to image the fiber end-face onto a CCD camera (Spiricon), as shown in Fig. 4. The large NA of the lens was essential to ensure that any radiation that can be guided by the 0.46 NA of the low-index polymer/silica cladding interface is captured by the camera. The cladding stripper was fabricated in between these two ends. A laser line filter at 532 nm with 1 nm FWHM was used to reject stray light due to the room lights and the blackbody radiation associated with the heating of the fiber during fabrication.

 

Fig. 4. Experimental setup for in situ measurement of CLS loss as a function of incidence.

Download Full Size | PDF

A single measurement involved scanning the rotary stage through 90° at a rate of $4.5°{\mathrm{s}}^{-1}$. A measurement was taken every 10 mm during the fabrication of the test devices.

In most applications, the cladding radiation that is required to be removed will typically be distributed over a large continuum of propagating angles.

2. Core Loss Measurement

A diagram illustrating the core loss measurement is shown in Fig. 5. A 2 μm ASE source and an optical spectrum analyzer were used to measure the core loss as a function of wavelength. For the purposes of this measurement we used a fiber with a 20 μm diameter, 0.09 NA core and a 400 μm diameter, 0.46 NA cladding. The cutoff wavelength for this core is 1950 nm. This fiber is representative of one that would be used in a high-power thulium-doped fiber laser [15], and has similar core and cladding diameters to fibers commonly used in building high-power ytterbium-doped fiber lasers [16].

 

Fig. 5. Diagram of the experimental setup for the measurement of core loss during ${\mathrm{CO}}_2$ laser fabrication of a CLS.

Download Full Size | PDF

CLS devices were placed before and after the device under fabrication to ensure that the loss measurement only examined the radiation propagating in the core. This ensured that we did not measure any excess loss as a result of radiation already present in the cladding, nor that we accepted radiation that had been coupled out from the core into the cladding as a result of the fabrication process.

The broadband measurement was also important to check for any spectrally dependent losses that may have occurred as a result of the periodic heating and ablation in the fabrication process.

3. EXPERIMENTAL RESULTS

A. Device Characterization

1. Cladding Loss Versus Launched Incidence Angle

The induced cladding loss for three different ablation depths is shown in Fig. 6. The graphs show that for increasing ablation depths, the losses increase with increasing incidence angle.

 

Fig. 6. Measured losses for radiation launched at different angles into the optical fiber for devices with ablation depths of (a) 15 μm, (b) 20 μm, and (c) 35 μm as a function of device length for a 620 μm pitch.

Download Full Size | PDF

2. Broadband Core Loss Characterization

The single-mode, broadband loss for devices with ablation depths of 15, 20, 35, and 57 μm are shown in Fig. 7. No wavelength-dependent losses were measurable because the transmission was within the noise of our 2 μm ASE source over the 1950–2100 nm region, giving $<0.008\pm 0.006\mathrm{dB}/\mathrm{cm}$ loss over the whole range. We did not have a more stable source and a longer device length could not be fabricated due to limits on our mounting arrangement. The loss measurement also showed that these particular processes do not induce any resonant core losses due to the periodic nature of the ablation. High-power measurements with a single-mode thulium-doped fiber laser at 1950 nm indicated the losses ranged from 0.03 to 0.05 dB for a 6 cm device (0.005–0.008 dB/cm). This is, however, also limited by the power stability and reproducibility of the laser.

 

Fig. 7. Core loss measurements for devices fabricated by different processes. The loss measurement was limited by the stability of our laser sources used for the characterization.

Download Full Size | PDF

3. High-Power Cladding Stripping Demonstration

We tested a device using a 300 W laser diode at 790 nm. The output from a fully filled 400 μm diameter, 0.22 NA fiber was launched into the cladding of a 400 μm diameter, 0.46 NA fiber. Radiation of 2.7 W was transmitted, resulting in a $\sim 20\mathrm{dB}$ loss, which is consistent with losses expected from Fig. 6(c). This fiber also had a 20 μm diameter, 0.1 NA core which would have accepted and guided some of the radiation thereby leading to the recording of a lower stripping efficiency than is actually the case. A 6 cm CLS with a 35 μm ablation depth and 620 μm pitch was used. The fiber was glued at each end of a 120 mm long aluminum heat sink with the CLS section itself suspended in air. This packaged CLS device was then conduction cooled on a separate water-cooled heat sink, as shown in Fig. 8.

 

Fig. 8. (a) Thermal camera image of a water-cooled packaged CLS device stripping $\sim 300\mathrm{W}$ of pump light. The maximum recorded temperature was 80°C. (b) Same device viewed with a normal camera.

Download Full Size | PDF

4. DISCUSSION

A. Device Performance Summary

The characteristics of the fabricated devices are summarized in Table 2. The cladding stripping efficiency can be varied over a large parameter space depending on the properties of the radiation to be removed from the cladding.

Table 2. Loss of CLS Devices of 620 μm Pitch for Various Ablation Depths and Launch Angles for Incident Radiation

View Table | View all tables in this article

We have demonstrated that comprehensive characterization of CLS devices requires angular-dependent loss measurements. Similar measurement techniques have been demonstrated as an important tool for the characterization of optical fiber sensors [17].

The core loss measurement did not reveal any significant wavelength-dependent losses. An average of $<0.008\pm 0.0006\mathrm{dB}/\mathrm{cm}$ of induced loss was measured for all of the devices.

B. Device Fabrication and Alternative Techniques

We have presented one approach to the ${\mathrm{CO}}_2$ laser fabrication of CLS devices with the main focus on producing short device lengths with high cladding loss and no perturbation to the core. For some devices, such as multimode core fibers, perturbations to the core introduced from bulk heating of the fiber may not adversely affect core light propagation as they would do for a single-mode fiber. In these cases, fabrication of the devices with longer pulse durations and larger ${\mathrm{CO}}_2$ beam sizes may be beneficial in achieving smoother ablation profiles while maintaining high tensile strength. An example of this is shown in Fig. 9. Following this process, we were able to fabricate devices that could be held under tensions higher than that required to cleave them.

 

Fig. 9. CLS fabricated through the rotation of a 400 μm optical fiber while it is translated longitudinally through a ${\mathrm{CO}}_2$ laser beam of $\sim 100\mathrm{\mu m}$ diameter.

Download Full Size | PDF

Rotating a fiber while simultaneously translating it through a focused ${\mathrm{CO}}_2$ laser beam can produce complex three-dimensional ablation profiles, as shown in Fig. 9. A movie of the process can be viewed in Visualization 2. While 360° ablation facilitates greater loss over shorter optical fiber lengths than single-sided processing, this fabrication technique relies on high-precision rotational control. Minimal precession is required to avoid large variations in ablation depths as the fiber moves in and out of the beam path. Such behavior may also result in unwanted deformation attributed to the torsion present in the fiber.

For these reasons, we chose to investigate the fabrication technique that utilizes a scanning beam to ablate one side of an optical fiber. This process facilitates highly reproducible, precise ablation profiles without stringent motion control requirements.

C. Scaling to Higher-Power Devices

${\mathrm{CO}}_2$ laser processing allows the fabricator to define the precise location and magnitude of the cladding light loss through the arbitrary control of the ablation depth. For these reasons, there is a plethora of different processes that could be envisioned to suit a particular application and power level. When considering packaging and high-power operation it may also be advantageous to direct the stripped light to a single side of a package. In addition, the period of the ablation can be chirped to enable a uniform power loss over the device length or heat sink.

5. CONCLUSION

We have presented a simple and robust CLS fabrication technique based on ${\mathrm{CO}}_2$ laser processing of an optical fiber surface. The process presented allows for the production of compact CLS devices of lengths less than 10 cm that are able to remove $>20\mathrm{dB}$ of radiation from a fully filled 400 μm diameter, 0.46 NA fiber. The induced core loss of these devices was measured to be $<0.008\pm 0.006\mathrm{dB}/\mathrm{cm}$ from 1950–2100 nm for the particular fiber that was used.

A simple technique was proposed to allow in situ characterization of the cladding losses for radiation launched at various incidence angles into the fiber. The methods described will assist in the design and fabrication of devices designed specifically for a given distribution of propagating radiation. These methods also enable the comparison of the efficiency of devices as new processes are developed.

Finally, we demonstrated the high-power operation of a cladding stripper by removing 300 W of radiation from the cladding of an optical fiber, which provides adequate power removal for kilowatt-level fiber laser fabrication. Minimal heating and no subsequent degradation of the polymer coating of the optical fiber was observed. Analysis of the thermal images indicated that the device was operating well below the damage threshold of the polymer fiber coating indicating that reliable long-term operation can be expected. We also suggest strategies for further power scaling, such as utilizing chirped profiles to enable more uniform power removal over the length of a device.