We report the design and performance of a high power femtosecond laser source near 1 micron wavelength which is generated from an octave-spanning supercontinuum (SC) pumped by an Er-doped mode-locked laser. The laser system delivers >5W average power at 35MHz repetition rate and 135 fs pulse duration.
©2010 Optical Society of America
Mode-locked lasers have found many important applications such as frequency metrology , precision machining [2,3] and multiphoton imaging [4,5]. Mode-locked fiber lasers especially attract attention from many researchers owing to their low cost, compactness and virtually maintenance-free operation. The most popular mode-locked fiber lasers are based on Er-doped and Yb-doped materials which operate near 1550 nm and 1000 nm, respectively. In general, it is easier and cheaper to build a compact oscillator based on Er-doped fiber due to the availability of fiber components developed for telecommunication. In addition, a large selection of fibers with different signs of dispersion for wavelength around 1550 nm is commercially available which facilitates the construction of mode-locked laser oscillators at this wavelength. On the other hand, fiber sources at around 1 micron wavelength are technologically important due to existing highly efficient Yb-doped gain fibers. This wavelength region is also desirable for many applications including multiphoton microscopy where linear absorption of most biological samples is small. However, it is more difficult to build compact and low cost femtosecond fiber oscillators with Yb-doped fibers. Standard optical fibers have normal dispersion at Yb gain wavelength region. For dispersion compensation at these wavelengths one therefore needs to use more exotic and expensive fibers such as PCF, chirped fiber Bragg grating, higher order mode fibers (HOM) or bulk gratings [6–9]. Dispersion compensation is not required in the all-normal dispersion cavity design  and relatively compact all-fiber oscillators have been demonstrated recently [11–13].
We show here the generation of high power femtosecond pulses near 1 micron wavelength by amplification of a slice of an octave-spanning SC. The starting point of the whole system is a compact Er-doped mode-locked laser with carbon nanotube (CNT) saturable absorber (SA) . This could be an attractive alternative solution to Yb-based oscillators, which gets the benefits of established and low cost telecom components. The pulse duration of the system is comparable to all-fiber Yb-doped systems reported to date [11–13]. Furthermore, we also demonstrate that power scaling using this approach is feasible as well. As the result, multi-Watt level output power, 135 fs pulse duration and > 150 nJ pulses were generated with a relatively compact system.
Recently, ~5 mW mode-locked picosecond pulse train at 1110 nm was demonstrated . This system was also based on spectral filtering of an octave-spanning supercontinuum (SC) source pumped at 1550 nm followed by amplification in Yb gain fiber. However, the pulse duration in that work is in the picosecond domain and the output power is much lower than what we report here. In addition, our Er-doped mode-locked fiber laser oscillator is based on CNT SA which has real advantages (such as fabrication simplicity, low cost and broadband operation) over current established mode-locking techniques using SESAMs or nonlinear polarization evolution (NPE).
2. Experimental setup and results
The detailed description of the Er-doped laser system (Fig. 1 ) has been reported in . In short, the oscillator was an integrated passively mode-locked fiber laser with carbon nanotube saturable absorber operating in soliton mode-locking regime. The saturable absorber was based on fiber taper embedded in carbon nanotube/polymer composite (FTCNT SA). The design of the SA was similar to the one reported in Ref. 14. The dispersion of the laser cavity was optimized so that it generated ~300 fs transform limited pulses at ~35 MHz repetition rate. The average output power of the oscillator in single-pulsing mode was about 1.5 mW.
The laser output was amplified in a single stage Er-doped amplifier to about 40mW. The gain fiber was ~1.5 m in length (which was about 2 times shorter than in ). The short gain fiber made the portion of the unabsorbed pump laser larger which could then be used to pump a segment of Yb-doped fiber downstream. This technique removed the need for a separate pump laser for the Yb-doped preamplifier. The un-absorbed 976nm pump was about 250mW at max pump power of 450mW at 976nm after the Er-doped fiber. The Er-doped gain fiber had normal dispersion at the laser wavelength so significant spectral broadening without wave-breaking during the amplification process occurred. At the output of the amplifier the pulses were highly chirped with much larger spectral bandwidth. We used a single mode fiber (SMF28) which has anomalous dispersion at this wavelength to compress the pulses before launching them into a short piece (4.5cm) of highly nonlinear fiber (HNLF) having ~10.6 μm2 effective mode area. By carefully optimizing the splicing parameters we could achieve low loss (< 1 dB) splice between the HNLF and SMF28 despite the large mismatch in core diameters. This HNLF was kindly provided to us by Sumitomo Electric. The dispersion of the HNLF at 1550nm was ~-5 ps2/km. An extremely stable octave-spanning supercontinuum was generated after the HNLF extending down to 1 μm. The longer wavelength side was measured only up to 1750 nm due to the limited range of the spectrum analyzer that we used. The SC actually extended to beyond 2100 nm revealed by a separate experiment where the spectral component around 2 μm was frequency doubled in a PPLN crystal and beat with the spectral component around 1 μm for detection of the carrier-envelope offset frequency (fceo). A stable fceo beat note was observed in that experiment, indicating that the generated SC was coherent. In another experiment, we have also compressed a portion of a similar SC to 14 fs pulse duration which contained only 3.5 optical cycles .
The spectral distribution of the energy of the SC could be controlled by adjusting the pump power of the Er-doped amplifier and the length of the SMF28 compression fiber. The length of the compression fiber was optimized (~50cm) to shift a large portion of the energy of the SC toward the 1 micron wavelength region. We adjusted these parameters to have some level of enhancement of the energy around 1 micron for further amplification purpose. The spectrum of the SC right after the HNLF is shown in Fig. 2(a) . It is obvious that the portion of the wavelength near 1040nm was enhanced. The output power contained in the spectral region from 1000nm to 1100nm was estimated at ~2 mW. This power level was enough to saturate the singlemode Yb-doped preamplifier.
As visible from the graph shown in Fig. 2(a) a significant portion of the 976 nm pump light was not absorbed by the 1.5 m segment of the Er-doped fiber. This radiation was then used to pump 60 cm of singlemode Yb-doped fiber (having 6 μm core diameter) after going through ~2 m of HI1060 passive fiber stretcher. The stretcher was needed in the system to avoid nonlinear effect in the singlemode Yb-doped preamplifier which tends to develop sharp features in the output spectrum at around 1 micron and degrades the pulse quality. The residual pump was fully absorbed in the short segment of Yb-doped fiber and the light around 1040 nm was amplified efficiently as shown in Fig. 2(b). The output power of the 1 micron light was measured to be around 90mW after filtering off the longer wavelength region. The output pulse energy of ~2.5nJ was calculated.
The output pulses were highly chirped but could be de-chirped down to ~110fs with a grating compressor. The efficiency of our current grating compressor was only ~50% making the compressed pulsed energy ~1.25 nJ. The spectrum and autocorrelation of the compressed output pulses are shown in Fig. 3 . Zero phase Fourier-transform of the output spectrum gave a pulse duration of ~45 fs, so the compressed pulses were 2.4 times the transform limit.
3. Power scaling
The output power of the laser system at this point is comparable to the performance of all-fiber normal dispersion lasers [11–13]. In the next step we show that the output power of the system could be scaled to multi-Watts level using a high power cladding pump Yb-doped fiber amplifier. The design of the high power amplifier was reported in detail in . In short, a pump/signal combiner was fabricated in house to couple the pump light from a multimode diode laser into the cladding of a large-mode-area Yb-doped gain fiber (which had 20 micron core size and 125 micron pump cladding size). The length of the gain fiber was ~1m. The power amplifier was spliced to the output of the singlemode Yb-doped amplifier described above through an isolator designed to work at around 1 micron to avoid feedback. The complete system up to this point used fiber-coupled components and thus no alignment was needed. To mitigate distortion to the pulses due to nonlinear effect in the power amplifier we increased the singlemode fiber stretcher to ~20 m of HI1060 fiber. The HI1060 fiber had estimated group velocity dispersion of 22 fs2 /mm and third-order dispersion (TOD) of 74 fs3 /mm. The pulse duration of the pulses before the power amplifier was about 30 ps. It turned out that the output power from the singlemode preamplifier in the previous stage (90 mW) was enough to saturate the DC power amplifier. We generated ~11.5 W of 1 micron output power at 22.5 W pumping at ~980 nm. The efficiency of the power amplifier was around 50%. The output power is currently limited by the available pump power.
The spectrum of the pulses before the DC power amplifier is shown in Fig. 4(a) (red curve). Some level of gain narrowing occurred after the amplifier as expected (black curve). As the result we measured a pulse duration of about 135 fs after compression with a grating compressor. The time-bandwidth-product (TBWP) was ~0.65. The output power after the compressor was 5.7 W which was limited mainly by the poor diffraction efficiency of the gratings. The pulse energy of compressed pulses was about 160 nJ. In principle, the overall efficiency in the compression stage could be improved by using highly efficient transmission gratings. The pulse quality was also slightly degraded witnessed by the appearance of a small pedestal in the autocorrelation trace. We attribute this degradation to the uncompensated TOD of the fiber stretcher. Using fiber with an anomalous TOD for pulse stretching may help to improve the pulse quality .
The stability of the 1 μm output pulse train was investigated using a RF spectrum analyzer and a fast silicon photodetector (detection range: 400-1100 nm). The results of such measurement are shown in Figs. 5(a) and 5(b). No visible sidebands and high signal contrast (> 80 dB) are observed indicating low timing jitters between pulses and low intensity fluctuation from pulse to pulse.
We report the design and performance of a high power femtosecond laser system operating near 1 micron by amplification of a portion of a SC generated by a compact Er-doped mode-locked laser. This alternative approach is scalable and gets the benefits of low cost, reliable components developed for telecommunication without sacrificing the performance.
We would like to thank Masaaki Hirano of Sumitomo Electric for providing the HNLF. Support is appreciated from National Science Foundation (NSF) ERC Center for Integrated Access Networks, CIAN.
References and links
2. X. Liu, D. Du, and G. Mourou, “Laser ablation and micromachining with ultrashort laser pulses,” IEEE J. Quantum Electron. 33(10), 1706–1716 (1997). [CrossRef]
3. C. B. Schaffer, A. Brodeur, J. F. García, and E. Mazur, “Micromachining bulk glass by use of femtosecond laser pulses with nanojoule energy,” Opt. Lett. 26(2), 93–95 (2001). [CrossRef]
6. H. Lim, F. ¨. O. Ilday, and F. Wise, “Femtosecond ytterbium fiber laser with photonic crystal fiber for dispersion control,” Opt. Express 10(25), 1497–1502 (2002), http://www.opticsinfobase.org/abstract.cfm?URI=oe-10-25-1497. [PubMed]
7. I. Hartl, G. Imeshev, L. Dong, G. C. Cho, and M. E. Fermann, “Ultra-compact dispersion compensated femtosecond fiber oscillators and amplifiers,” Conference on Lasers and Electro-Optics 2005, Baltimore, MD, paper CThG1.
9. M. Schultz, O. Prochnow, A. Ruehl, D. Wandt, D. Kracht, S. Ramachandran, and S. Ghalmi, “Sub-60-fs ytterbium-doped fiber laser with a fiber-based dispersion compensation,” Opt. Lett. 32(16), 2372–2374 (2007). [CrossRef] [PubMed]
10. A. Chong, J. Buckley, W. Renninger, and F. Wise, “All-normal-dispersion femtosecond fiber laser,” Opt. Express 14(21), 10095–10100 (2006), http://www.opticsinfobase.org/abstract.cfm?URI=oe-14-21-10095. [CrossRef] [PubMed]
12. M. Schultz, H. Karow, O. Prochnow, D. Wandt, U. Morgner, and D. Kracht, “All-fiber ytterbium femtosecond laser without dispersion compensation,” Opt. Express 16(24), 19562–19567 (2008), http://www.opticsinfobase.org/abstract.cfm?URI=oe-16-24-19562. [CrossRef] [PubMed]
15. D. Kielpinski, M. G. Pullen, J. Canning, M. Stevenson, P. S. Westbrook, and K. S. Feder, “Mode-locked picosecond pulse generation from an octave-spanning supercontinuum,” Opt. Express 17(23), 20833–20839 (2009), http://www.opticsinfobase.org/oe/abstract.cfm?URI=oe-17-23-20833. [CrossRef] [PubMed]
16. K. Kieu, J. Jones and N. Peyghambarian, “Generation of few-cycle pulses from an amplified carbon nanotube mode-locked fiber laser system,” early posting PTL, 10.1109/LPT.2010.2063423.
18. I. Hartl, T. R. Schibli, A. Marcinkevicius, D. C. Yost, D. D. Hudson, M. E. Fermann, and J. Ye, “Cavity-enhanced similariton Yb-fiber laser frequency comb: 3×1014 W/cm2 peak intensity at 136 MHz,” Opt. Lett. 32(19), 2870–2872 (2007). [CrossRef] [PubMed]