A novel design for a microstructure fiber (MSF) laser consisting of a large core and a single annulus of 5 air holes is described. The fiber design incorporates a silica core that was doped in the liquid phase with 1300 ppm Nd2O3. The light guiding losses in the structurally very simple MSF are ≈0.7 dB/m. Single transverse mode emission is demonstrated with a mode field area larger than 200 μm2. The laser simultaneously emits at two groups of wavelengths centered at 1060 nm and 1090 nm. Pumped by a cw Ti:sapphire laser, the fiber laser yields a maximum output power of 280 mW (pump power limited) at a slope efficiency of 52%. Our results indicate how the advanced possibilities of MSF’s can be used for optimized fiber laser designs.
©2005 Optical Society of America
Compared to conventionally designed fibers, microstructure fibers (MSF’s) offer new options for designing doped fibers and for building fiber lasers. With the very tight confinement possible with MSFs, refractive index engineering and tailoring of nonlinear properties is possible to an extent that allows for a substantial manipulation of fiber properties. For example, manipulation of the chromatic dispersion properties via microstructuring is now in wide use in fiber optics [1–3]. For building fiber lasers, however, a reduction of nonlinear optical effects, such as Brillouin scattering, is of paramount importance . This can be accomplished by scaling up the effective mode area of the fiber. While in regular fibers, such scaling is paid for by increased bending losses, microstructure fibers offer an alternative that is less susceptible to such problems. Therefore, large-mode-area MSFs offer a wide range of application opportunities [5,6] and are very promising for the realization of high-power fiber lasers with superior beam quality.
A common technological problem is the incorporation of lasing dopants (Nd3+, Yb3+) into the core material, for which silica found nearly exclusive use. The modified chemical vapor deposition (MCVD, ) technology conventionally used for doping silica is very time-consuming. It results in a preform surrounded by undoped material, which has to be removed by an additional etching step. Collapsing the preform, a doping dip inevitably occurs in the center of the core with conventional methods. We avoid these drawbacks of the MCVD doping method by doping silica in its liquid state instead . This method neither requires additional etching steps nor does it exhibit doping anomalies or result in a degradation of light guiding properties of the fibers.
In MSF’s, the core is often suspended in a complicated structure, even though guiding only relies on total internal reflection. In contrast, we use a much simpler design, consisting of only a single annulus of 5 air holes surrounding the active core area. Because of its fivefold symmetry, we refer to this fiber as the penta fiber in the following. We have confirmed, both, by simulations of the light guiding losses and also experimentally, that our minimalist approach fully suffices for low-loss guiding of light inside a fiber laser. Still, our approach preserves all advantages of MSF’s, such as low bending losses even at large mode diameters and the enhanced scaling potential of the fiber geometry compared to conventional fiber technology. Below we will describe operation of a Nd-doped penta fiber laser relying on our improved doping technology.
2. The penta fiber
The core of the penta fiber was manufactured from a silica rod, which was doped with 1300 ppm Nd2O3 in the liquid phase. Because of volume doping, this rod. The rod was inserted into the central bore of a preform with five more surrounding bores. Subsequently, the preform was pulled out to a fiber with 125 microns outer diameter and 15 microns core diameter, which is nearly 20 times larger than the lasing wavelength (λ=1.064 μm) in this medium. The doped region extends well into the bridges between the air holes, which results in a completely doped core. The resulting fiber cross-section exhibits about equal core size and pitch (Λ=19 μm) as shown in the scanning-electron micrograph in Fig.1(a). The hole diameter is d=18 μm.
Figure 1(b) shows the mode field obtained by a numerical simulation . From the structural parameters of the fiber, a numerical aperture of about 0.156 was calculated . Similar to conventional fibers, a V-parameter can be defined for PCFs , which we calculate as VPCF=17.8. This value is well above the higher-mode cutoff value of V=π, which indicates that our pentafiber supports multimode operation at the lasing wavelength. We will show below, however, that an effective single-transverse-mode operation of this fiber is possible without further measures. Our numerical calculations  yield extremely low radiation losses of below 1 dB/km for the lowest 4 modes. Experimentally, we find substantially higher values using the cut-back-method. The measured losses amount to approximately 1dB/m for the fundamental mode. This may be taken as an indication for manufacturing imperfections, but is still absolutely sufficient for operation inside a laser cavity, given the high gain of the fiber. The general handling properties of the doped penta fibers are excellent. We observed the best quality of the end face with manual cleaving. In particular, we did not observe noticeable deviations of the structure geometry over a 10-m length, inspecting the fiber end faces after multiple cuts into 10 pieces. Nevertheless, we observed small differences of the lasing characteristics between the different fiber segments.
3. Emission properties of the microstructure fiber laser
A cw Ti:sapphire laser tuned to λ = 809 nm (Tsunami) was used as the pump source for core pumping. This pumping wavelength ensures optimum absorption efficiency of the active ions in the fiber core. The experimental set-up used for the optical excitation of the ions in the microstructured fiber is shown in Fig. 2.
Using coupling lenses with different focal length f, the launch efficiency for the pump light was experimentally optimized to 75%. Best laser performance was observed with an f = 15.35 mm molded aspheric lens. Using this lens, the pump spot diameter is about 30 μm in diameter, which is about twice as large as the core itself. In the experiments, we lost less than 10% of the pump light at the output coupler.
The resonator mirror at the coupling port was dichroic with a reflectivity of <1% at 809 nm and ~99% at the lasing wavelength. For optimum laser performance the required fiber length is 1 m or slightly less. This is confirmed by comparing the small signal gain produced by the pump to the laser radiation losses. The output power was optimized using output coupling mirrors with different reflectivities. For a fiber length of 1 m and using the 4% Fresnel reflectivity of the fiber end face as the output coupling mirror yields the maximum wavelength-integrated output power.
3.1 Near-field characteristics
Figures 3(a) and (b) show linear cuts through the near field intensity distributions at the threshold and about 4 times above the threshold. These images were measured imaging the fiber end face on to a CCD camera and are indicative of a single-transverse mode operation of the laser. The full width at half maximum at threshold is measured as w = 16.6 μm. For operation well above threshold, a slightly narrower mode pattern (w = 10.3 μm) is observed, other than one would expect with excited higher transverse modes. From the calculated near field of the fundamental HE11-mode we expect a width of ~ 8 μm, which we take as an indication that our laser essentially operates in the fundamental transverse mode if only the pump power is high enough. This behavior is in contrast to the operation of rod lasers, which tend to escape from single transverse mode operation at elevated pump powers. The observed contraction of the mode size with higher powers in our laser can potentially be explained by a dependence of the refractive index on intensity, most likely induced by thermal lensing inside the doped core.
3.2 Output characteristics
Figure 4 displays the measured laser output characteristics of our penta fiber laser as a function of launched power. We observe a laser threshold at 140 mW and a slope efficiency of 37% for a straight fiber configuration. Coiling the fiber to a 3 cm diameter does not reveal any increase of losses. We can conclude that bending losses in our fiber are negligible which is extremely important for practical applications. Using circular polarization of the pump and a very well prepared fiber, we observed doubling of the output power up to 280 mW, see the following section. This result was only limited by available pump power. The laser did not show signs of saturation or other degradation when operated at the highest pump powers available. Maximum efficiency pumping is achieved with λ pump=808 nm. Efficient direct diode pumping appears possible in a 5-nm range around this central wavelength.
3.3 Polarization properties
The penta structure of the fiber does neither provide a clear reference for polarization measurements nor does it define clear polarization axes as many other structures do. Therefore output polarization depends only on the pump efficiency of the different polarization modes in the fiber. In the experiments described so far, we did not use any additional polarization-sensitive elements inside the laser cavity. Still, our fiber displays birefringence, which we experimentally determined as B ≈ 6×10-5, with a slight variation from sample to sample. Due to this intrinsic birefringence polarization eigenmodes exist and can be selectively excited. Using a polarizer-wave plate combination, we selectively excited either one of these polarization eigenmodes. Employing a second rotatable polarizer at the output as an analyzer we monitor laser emission parallel and orthogonal to the input state. As can be seen in Fig. 5, we always detect light in both polarization eigenmodes, regardless of what polarization mode we excite. The degree of polarization for the linearly polarized input amounts to ≈30 %, both, for the straight and for the coiled fiber. Each of the 360° plots indicates output power vs. input polarizer setting at a fixed pump power of 300 mW. Intentionally, we did not use a normalized representation to not obscure the influence of pump radiation polarization on emitted power.
With linear input polarization, the output is elliptically polarized, with a preferred orientation perpendicular to the input. For a circularly polarized input the output appears circularly polarized, too. Note, however, that we observe a doubling of the output power for circular pump polarization.
3.4 Dual-wavelength operation
It is well known that Ti:sapphire lasers or Er3+ doped fiber lasers can simultaneously oscillate on several distinct wavelengths, both in cw and pulsed operation [12,13]. Even though slightly less common, dual wavelength operation has also been observed with standard single mode Nd–doped silica fibers, which has also been exploited for short pulse generation . We observe dual-wavelength operation for a microstructure fiber laser. Emission appears on two distinct transitions within the 4F3/2→4I11/2 manifold of the Nd – ion at 1060 and 1090 nm, as shown in Figs. 6(a) and (b) for two different input powers. At higher pump powers, laser operation appears in two bands in the range from 1055 to 1095 nm, i.e. a total 10 THz frequency span, with a possible tuning range of about 10 nm in either band.
Emission at the rather uncommon 1090 nm wavelength sets in at a second laser threshold of about 200 mW launched pump power, which is about 5 times higher than the threshold for 1060-nm operation, see Fig. 7. As the splitting between the two wavelength bands is suspiciously close to the Raman shift in silica, we estimated the critical power P cr necessary for the onset of Raman oscillation . We calculate P cr = 32 kW for our case, where we assumed an effective mode area of 200 μm2, an effective fiber length of 1 m and a Raman gain coefficient of g R = 10-13 m/W. Even if self-Q-switching should occur this value of P cr is well above the intrafiber power levels (<1 W cw) in our laser. To further investigate the dual-band operation of the laser, we separated both components by a grating and monitored their temporal evolution with a fast photodiode and a digital sampling oscilloscope (see Fig. 8). If the long-wavelength band were only a Raman shifted copy of the 1060-nm emission in the fiber, one would expect a strong power correlation of the emission in both bands, e.g. by saturation effects. Another mechanism that would cause a strong anticorrelation is gain competition in a homogeneously broadened medium . If the observed dual wavelength operation is due to two independent laser transitions, however, output power fluctuations should also be widely independent, and no correlation would be visible. From the data in Fig. 8 we extract a correlation coefficient of –0.1, which means that, in fact, emission at both wavelengths is uncorrelated. Another indication against a Raman type gain mechanism is the absence of further Stokes sidebands in the vicinity of 1120 nm and above for even higher excitation levels. Therefore we believe that the two spectral channels have to be attributed to the Stark - splitted upper level of the ionic transition 4F3/2 → 4I11/2 in Nd.
This result is encouraging for the generation of short pulses in such a structure, as it appears possible to bridge the gap between the two spectral bands for generating ultrashort pulses with a properly designed resonator. If the mode-locking process could extend over both wavelength bands, a pulse duration of < 100fs directly from a Nd-doped fiber laser appears possible.
Using a novel design approach, we manufactured and characterized a highly efficient Nd-doped fiber laser. Volume doping of silica widely avoids doping inhomogeneities and opens up a perspective for much higher doping levels than accessible with competing technologies. We use a penta fiber MSF design of extraordinary simplicity, which yet allows for low transmission losses. Our approach greatly simplifies manufacturing, as it does not require a delicate photonic lattice section, which requires much greater care during fiber pulling. As we experimentally demonstrated, our fiber laser design allows for laser efficiencies within two thirds of the theoretical limit. For the first time to our knowledge, co-emission at 1060 and at 1090 nm was observed in a Nd-doped fiber laser. We believe that the appearance of the uncommon 1090-nm emission is yet another indication for the very high quality of the doped laser-active core. Our approach greatly simplifies MSF design in two important ways. First manufacturing of doped core regions can be brought to higher efficiencies and higher doping levels. Second, manufacturing can be greatly simplified, owing to the much simpler structure of the penta design. Both these improvements together open up a perspective for a much wider use of microstructuring for fiber lasers.
References and Links
1. D. Mogilevtsev, T. A. Birks, and P. St. J. Russell, “Group-velocity dispersion in photonic crystal fibers,“ Opt. Lett. 23, 1662–4 (1998). [CrossRef]
2. Lin-Ping Shen, Wei-Ping Huang, and Shui-Sheng Jian, “Design of Photonic Crystal Fibers for Dispersion- Related Applications,“ J. Lightwave Technol. 21, 1644–51 (2003). [CrossRef]
3. W. H. Reeves, D. V. Skryabin, F. Biancalana, J. C. Knight, P. St. J. Russell, F. G. Omenetto, A. Efimov, and A. J. Taylor, “Transformation and control of ultrashort pulses in dispersion-engineered photonic crystal fibres,“ Nature 424, 511–5 (2003). [CrossRef] [PubMed]
5. W. J. Wadsworth et al.: “High power air clad photonic crystal fibre laser,” Opt. Express 11, 48 (2003), http://www.opticsexpress.org/abstract.cfm?URI=OPEX-11-1-48. [CrossRef] [PubMed]
6. J. C. Knight et al.: “Large mode area photonic crystal fiber,” Electron. Lett. 34, 1347 (1998). [CrossRef]
7. S. R. Nagel, J. B. Macchesney, and K. L. Walker, “An overview of the modified chemical vapor-deposition (MCVD) process and performance,“ IEEE J. Quantum Electron. 18, 459–476 (1982). [CrossRef]
8. W. L. Barnes et al., “Detailed characterization of Nd3+ doped SiO2-GeO2 glass fibre lasers,” Opt. Commun. 82, 282 (1991). [CrossRef]
9. B.T. Kuhlmey, T.P. White, G. Renversez, D. Maystre, L. C. Botten, C.M. de Sterke, and R.C. McPhedran, “Multipole method for microstructured optical fibers. II. Implementation and results,“ J. Opt. Soc. Am. B 19, 2331–40 (2002), http://www.physics.usyd.edu.au/cudos/mofsoftware/. [CrossRef]
10. N. A. Mortensen, J.R. Folkenberg, M.D. Nielsen, and K.P. Hansen,” Modal cutoff and the V parameter in photonic crystal fibers,” Opt. Lett. 28, 1879 (2002). [CrossRef]
11. N. A. Mortensen, “Effective area of photonic crystal fibers,” Opt. Express 10, 341 (2002), http://www.opticsexpress.org/abstract.cfm?URI=OPEX-10-7-341. [PubMed]
12. J. M. Evans, D. E. Spence, D. Burns, and W. Sibett, “Dual-wavelength self-mode-locked Ti:sapphire laser,“ Opt. Lett. 29, 409–11 (1993).
13. D. Pudo, L. R. Chen, D. Giannone, Lin Zhang, and I. Bennion, “Actively mode-locked tunable dual-wavelength erbium-doped fiber laser,” IEEE Photonics Technol. Lett. 17, 1 (2002).
15. G. P. Agrawal, Nonlinear Fiber Optics, 2nd ed., (Academic Press, San Diego, CA, 1998)