We report fundamental mode operation in ytterbium-doped leakage channel fibers with effective areas up to 3160µm2. These fibers are bend-resistant. Lasers with slope efficiency ~60% and M2~1.3 has been demonstrated. Bend loss and amplifier performance have also been characterized.
© 2006 Optical Society of America
In the last few years, there has been significant progress in commercialization of fiber lasers. Fiber lasers have many advantages, which include the ease of manufacturing, turn-key operation, robust and stable performance, low maintenance, and better thermal load dissipation. These benefits contributed to the commercial success of fiber lasers in the recent years. Major limitations in further power scaling of fiber lasers are nonlinear effects in the forms of Raman, Brillouin and self-phase modulations. These nonlinear effects come from tight confinement of optical modes in an optical fiber. Although these nonlinear effects can be mitigated by individually tailored solutions, an increase of effective mode area can significantly ease all of the nonlinear limits.
There have been significant efforts in the last several years to extend effective mode areas of ytterbium-doped optical fibers. There are two major approaches. The first approach uses conventional step index (SI) design. The basic concept of this approach is to propagate fundamental mode in a multi-mode fiber . Ytterbium-doped fibers with effective areas of 500–600µm2 and reasonable bending performance have been commercially available in the last several years. It, however, becomes prohibitively difficult to launch and maintain a fundamental mode if the core size is further increased. A key limitation is inter-modal coupling due to a much reduced mode spacing in a multi-mode optical fiber, when the number of propagating modes increases significantly at larger core size. There is a recent demonstration of an ytterbium-doped fiber with an effective area of 2000µm2 . This fiber, however, supports well over 100 modes. It is consequently very difficult to maintain correct launch condition and to avoid inter-modal coupling. It is potentially possible to ease this limit by reducing the number of supported modes by lowering numerical aperture (NA). This is, however, currently limited by the controllability of the fabrication process used to make SI fibers. In addition, bending loss of fundamental mode will significantly increase as NA is further reduced. Recently, efforts have been made to flatten the fundamental mode profile and to utilize a doping profile that amplifies preferentially the fundamental mode . Both of these efforts can only make marginal improvements. The second major approach is based on a design using single mode photonic crystal fibers (PCF). In a PCF, it is possible to achieve very small effective NA to implement designs close to single mode operations. A key advantage of this approach is its ability to control effective NA of a waveguide precisely by controlling hole size in the cladding. This is equivalent to making a large-core conventional fiber with a very small NA so as to reduce the number of propagating modes. The extremely low equivalent NA of the fibers makes them very bend sensitive. For effective areas above 1000µm2, fibers have to be kept straight. This is accomplished by making the fiber into a rod that has a typical diameter of ~1.5mm . Recently, an ytterbium-doped rod-fiber with ~2000µm2 effective area and un-doped fiber with 4500µm2 have been demonstrated . These rods are, however, limited to a short length of ~0.5m and do not dissipate heat as well as optical fibers do.
We recently demonstrated a new approach in fiber design for large-core fibers . This new approach uses a small number of large holes to define a waveguide. The large holes provide much improved bending loss performance, while the leakage channels between holes can be engineered to implement significant higher order mode loss at very large effective mode areas. Consequently, in a first demonstration, robust fundamental mode operation in a passive fiber with an effective area of ~1400µm2 was achieved. It was also shown there is very little fundamental mode loss at bending radius as small as ~8cm. In this paper, we further demonstrate fundamental mode operation in two ytterbium-doped double clad fibers with effective areas of ~1500µm2 and ~3160µm2 respectively. Laser operation and bending characteristics have been characterized.
2. Experimental details
The cross section of the first fiber is shown in Fig. 1(a). We will refer to this fiber as 1st fiber. Six holes with a diameter d of ~41µm form the core. The center-to-center average separation of holes Λ is ~50µm. This gives an average normalized hole diameter d/Λ of 0.82 and an average core diameter 2Λ-d of 59µm. Due to the distortion from the fabrication process, the core is slightly elliptical. The longest edge-to-edge hole separation is ~65µm, and two shorter ones are ~49µm. A circular portion of ~30µm in diameter in the center of the core is doped with ytterbium. The refractive index of the ytterbium-doped region is closely matched to that of the rest of the core so that there is negligible impact on the guiding property of the LCF due to the doped part of the core. The fiber, with an outer diameter of 260µm, is coated with a low refractive index polymer coating that has a refractive index of 1.37, corresponding to a pump NA of 0.46. This low refractive index polymer coating provides the multimode pump guide for the double clad LCF The measured near field mode intensity distribution is shown in Fig. 1(b). This near field was taken from the output of an amplifier arrangement. The distribution is scaled so that the outlines of the measured intensity distribution is consistent with edge-toedge hole separations. An effective area of ~1500µm2, i.e. a MFD of 43.7µm, is computed from the measured near field intensity distribution by performing the standard effective mode area integration A simulation was also performed with the known hole diameter and average hole separation using a mode solver based on multi-pole algorithm without considering the deformations from the fabrication process. The modeled mode intensity distribution of the fundamental mode is shown in Fig. 1(c). An effective area of ~1550µm2, i.e., a MFD of 44.4µm, is computed from the simulated intensity distribution. The pump absorption, measured with a 980 nm multimode diode and cut-back method is 2.6dB/m. The multimode pump diode power was launched into the multimode pump guide and its power level was carefully chosen so that the fiber output was not dominated by ytterbium ASE and there was minimum ytterbium inversion. Although the 1st fiber is slightly elliptical, negligible birefringence is expected. At large core regime as in this fiber, there is very little interaction between the guided fundamental mode and core boundary. The negligible birefringence of the fiber is further confirmed by our simulations. Details of theoretical design analysis in term of differential mode loss and wavelength dependence of the LCFs are addressed in our previous work . A more comprehensive review of LCF designs and simulation of bending loss are addressed in a paper submitted to Journal of the Optical Society of America B in the special fiber laser issue to be published in 2007.
Laser performance was also tested. A 3m long fiber was used. At 1st end, the holes in the fiber was collapsed with a fusion splicer and then cleaved. The second end was prepared by tapering the fiber down to ~100µm at the waist on a fusion splicer. Effort was made to ensure that the holes at the waist were totally collapsed. The taper was then cleaved in the middle of the waist. We have observed that the taper is not essential for the single mode propagation. The LCF has sufficient mode discriminations for robust single mode propagation. The taper does make launch easier and more efficient due to an improved angular tolerance and reduced mode-size mismatch between the LCF and launch fiber. A laser was constructed using the 1st end as the output coupler using the reflection off the cleaved end face. Light from the 2nd end was collimated and then reflected off a high reflector mirror. Figure 2 gives the laser performance. The threshold is ~2W of launched power. The slope efficiency was extracted from the slope of a linear fit to the measured data to be ~60%. The high slope efficiency indicates that there is a strong interaction between the multimode pump power and the ytterbium-doped core. Pump power was launched into the multimode pump guide. Since the 6 holes are designed to let higher order modes to pass freely, we expect higher order pump modes to pass through the gaps between holes with ease. The scattering off the large hole boundaries is also expected to help coupling among pump modes. This will further help pump absorption. It is evident from the high slope efficiency that there is a strong overlap between the multimode pump power and the ytterbium-doped core. Expected pump absorption calculated directly from the absorption level in the ytterbium-doped region and the area ratio of the doped region to multimode pump guide, i.e. total multimode pump guide minus 6 holes, is ~2.6B/m, a good match to the measured pump absorption. Pump absorption can be further improved by optimization of the holes for more efficient pump pass-through and using standard techniques such as non-circular pump guide to scatter skew rays into core-intersecting rays. It needs to be further noted that the multimode pump guide of the 1st fiber is slightly non-circular and this non-circularity can be further improved if required. M2 was measured to be ~1.26.
The fiber was also tested in an amplifier configuration. An ytterbium ASE source with 20mW power was launched into the 2nd end of the 3m fiber. M2 was measured to be ~1.3 in this configuration. Bend loss was measured in this amplifier configuration with a single coil of varying radius placed in the middle of the fiber. The rest of the fiber was placed so that it bent only slightly. Effort was made to ensure that the amplifier was far from saturation. The outputs were measured for various bend radii. This was then translated into dB/m and plotted in Fig. 3. Since the amplifier output was dominated by the power at peak gain wavelength of ~1020nm, this bending loss is mostly for this peak gain wavelength of ~1020nm. There was no effort made to control the plane of the bend. There is a small but insignificant bending loss dependence on bending planes as noted in our previous work . It can be seen from the bend loss measurement that the fiber can be coiled down to 8.5cm radius with negligible power penalty. Output degradation of laser and amplifier performance is expected for coils with less than 8.5cm of bend radius.
The 2nd fiber was drawn from the same preform as the 1st fiber. It will be refereed to as the 80µm fiber. The cross section is shown in Fig. 4(a). It was drawn into an outer diameter of 350µm. It is also coated with the low refractive index polymer coating to give a pump NA of 0.46. The diameter of the holes is 55µm. The average center-to-center spacing of holes is 67µm. The doped part of the core has a diameter of 40µm. The core diameter is 88µm in its longest dimension and 66µm in its two shorter dimensions. The average core diameter is 79µm. The average normalized hole diameter is 0.82. The pump absorption, measured with a 980nm multimode diode and a cut-back method is 3.6dB/m. This is higher than the expected 2.6dB/m. This higher pump absorption in the 2nd fiber may mainly come from easier excitation of lower order pump modes in this much larger fiber. The measured near field intensity distribution from an amplifier arrangement of the 2nd fiber output is shown in Fig. 4 (b). The near field intensity distribution is again scaled so that the outline of the intensity distribution is consistent with the fiber cross sectional geometry. Effective mode area is computed from this measured mode profile to be ~3160µm2, i.e. a MFD of 63.4µm. The fiber is simulated with the known hole diameter and average hole separation without considering the deformations from fiber fabrication. The simulated mode intensity distribution is shown in Fig. 4(c). Effective mode area calculated from the simulation is 2990µm2, i.e. a MFD of 61.7µm.
Laser performance was also tested in a similar configuration as for the 50µm core fiber. The fiber ends were also prepared similarly as for the 50µm core fiber. The 2nd end of the 80µm core fiber was tapered down to a diameter of 104µm. The taper is again not essential for single mode propagation, but added to ease launch alignment tolerance and to improve launch efficiency. Since pump guide is increased from 260µm to 350µm, this leads to a reduction of available pump intensity in the fiber. Fiber length has to be increased to provide sufficient gain for the laser test. Figure 5 gives the laser performance with a laser cavity consisting of ~5m of the 80µm core fiber. The fiber was loosely laid on an optical table with a coil diameter of ~40cm. The threshold is increased to ~9W of launched power. The slope efficiency was measured to be ~60%. M2 of the output beam was measured to be ~1.3.
The 80µm core fiber was also tested in amplifier configurations with the ytterbium ASE source as input. M2 measured from the amplifier output is ~1.3. Spectral gain in 5m of 80µm core fiber was measured at various launched powers. This was done by measuring output spectra at various pump power levels as well as when pump power was turned off. Fiber loss spectrum was measured independently with a cut-back method and is included in the gain calculation. The measured gain is shown in Fig. 6. ASE spectrum is shown in the inset. Gain peaks at ~1020nm. A maximum gain of 38dB was measured with a launched pump power of 12W. Amplifier performance at various bending radii was also characterized. The ytterbium ASE source and 5m long fiber were again used. A single coil was introduced in the middle of the fiber while the rest of fiber is kept to a much larger bend radius. Effort was made to ensure that the amplifier operates far from gain saturation. The bend loss was calculated from the measured amplifier output. This is given in Fig. 7. No effort was made to control the bending plane as for the 1st fiber. It can be seen that a minimum bend radius of 15cm needs to be maintained for negligible bend loss penalty.
To summarize, we demonstrate bend resistant ytterbium-doped fiber with an effective areas up to ~3160µm2. We have demonstrated efficient laser performance and reasonable M2. Bend loss measurement indicates that, with negligible loss penalty, the fiber with 1500µm2 can be coiled down to 8.5cm radius and the fiber with 3160µm2 to 15cm radius.
The authors would like to thanks William S. Wong for his assistance in the fabrication of the ytterbium-doped core glass.
References and links
1. M. E. Fermann, “Single-mode excitation of multimode fibers with ultrashort pulses,” Opt. Lett. 23, 52–54(1998). [CrossRef]
2. K. H. Liao, K. C. Hou, G. Q. Change, V. Smirnov, L. Glebov, R. Changkakoti, P. Mamidipudi, and A. Galvanauskas, “Diffraction-limited 65µm core Yb-doped LMA fiber based high energy fiber CPA systems in Proceedings of Conference of Lasers and Electro-Optics, (Long Beach, California., 2006), paper CPDB4.
3. W. E. Torruellas, Y. Chen, B. McIntosh, J. Faroni, K. Tankala, S. Webster, D. J. Hagan, and M. J. Soileau, “High peak power ytterbium-doped fiber amplifiers,” presented in Photonic West, San Jose, California, USA, 21–26 January 2006.
4. J. Limpert, N. Deguil-Robin, I. Manek-Hönninger, F. Salin, F. Röser, A. Liem, T. Schreiber, S. Nolte, H. Zellmer, A. Tünnermann, J. Broeng, A. Petersson, and C. Jakobsen, “High-power rod-type photonic crystal fiber laser,” Opt. Express 13, 1055–1058 (2005). [CrossRef]
5. J. Limpert, O. Schmidt, J. Rothhardt, F. Röser, T. Schreiber, A. Tünnermann, S. Ermeneux, P. Yvernault, and F. Salin, “Extended single-mode photonic crystal fiber laser,” Opt. Express 14, 2715–2720 (2006). [CrossRef]
6. W. S. Wong, X. Peng, J. M. McLaughlin, and L. Dong, “Breaking the limit of maximum effective area for robust single-mode propagation in optical fibers,” Opt. Lett. 30, 2855–2857(2005). [CrossRef]