A monolithic fiber chirped pulse amplification system that generates sub-500 fs pulses with 913 µJ pulse energy and 4.4 W average power at 1.55 µm wavelength has recently been demonstrated. The estimated peak power for the system output approached 1.9 GW. The pulses were near diffraction-limited and near transform-limited, benefiting from the straight and short length of the booster amplifier as well as adaptive phase shaping for the overall mitigation of the nonlinear phase accumulation. The booster amplifier employs an Er3+-doped large mode area high efficiency media fiber just 28 cm in length with a fundamental mode (LP01) diameter of 54 µm and a corresponding effective mode area of 2290 µm2.
© 2014 Optical Society of America
Ultrafast lasers are compelling due to the unique interaction between matter and their extremely short pulses of light . Traditional material modification applications rely upon very high average powers with continuous wave (CW) lasers to induce catastrophic thermal effects or very high pulse energies with pulsed lasers to induce dielectric breakdown or other peak power driven events. By contrast, ultrafast lasers—where the pulse duration is less than a picosecond—readily invoke highly desirable modifications to objects of interest with a relatively low average power and pulse energy. This is an inherent benefit of the ultrafast laser pulse peak power and the extreme brevity of its pulse duration. Several advanced applications have been identified for which the ultrafast laser would be a unique solution but would require an order of magnitude increase in available fluence and time-averaged irradiance. Such requirements would then demand energies in excess of 1 millijoule in each compressed pulse. Ultrafast fiber laser systems are limited in output pulse energy by the temporal pulse distortion caused by self-phase modulation (SPM) as the pulses propagate through the system amplifiers. The magnitude of the SPM is quantified by the B-integral [2,3], which is proportional to both the beam irradiance inside the amplifier and the propagation length.
Increasing the amplifier’s mode area will reduce the beam irradiance and thus the B-integral. Most efforts to develop large-mode area (LMA) waveguides have focused on approaches such as low NA photonic crystal fiber , leakage channel fiber , chirally-coupled core fiber , large-pitch fiber , and distributed mode filtering rod fiber . These established approaches all operate in the fundamental fiber mode (LP0,1) which suffers from bend-induced reductions in the mode size and increases in modal instability. This may be avoided by making the fiber into a rigid rod with typical diameter of 1.5 mm or more: Tino at el. have demonstrated record sub-500 fs pulses with 3.8 GW at a 1.03 µm wavelength . These rods are limited in length to ~1 m however, due to practical packaging constraints. Such laser assemblies are further complicated as they necessitate free-space coupling due to the challenges of direct fiber splicing to these rods without impacting their beam guiding properties. It has been demonstrated that these traditional LMA fiber limitations may be overcome by using a higher-order mode (HOM) fiber amplifier. Such a device is scalable in mode size, resists bend-induced mode size reductions, and possesses superior stability characteristics . However, the gain fiber is still required to be quite long, a couple of meters, and fabricating these amplifiers with an integral long period grating (LPG) for mode conversion remains complicated.
Optical fibers as waveguides use their small diameters to retain single mode characteristics. As such, they pose an inherent disadvantage with respect to reducing irradiance as compared to laser rod or thin disk geometries. Whereas most other energy scaling strategies attempt to reduce the B-integral by implementing LMA fiber amplifiers with ever larger effective areas (Aeff), this approach results in substantial beam quality degradation and practical packaging challenges. In contrast, we have proposed using a high efficiency media (HEM) amplifier to achieve the B-integral reduction by shortening the required gain fiber length while maintaining reasonable gain and efficiency . In order to dramatically reduce the fiber length, a heavily erbium doped (1.6 × 1026 ions/m3 and greater) phosphate glass fiber was developed that prevented the parasitic energy loss associated with ion clustering in more traditional glass fibers, e.g. silica. With the combination of the HEM fiber amplifier and Raydiance’s advanced adaptive pulse shaping techniques, the first monolithic HEM based fiber-optic CPA system for high energy femtosecond pulse generation was demonstrated. The HEM Er-doped phosphate gain fiber was just 28 cm in length with an Aeff = 2290 µm2. The resultant 1.55 µm wavelength monolithic fiber chirped pulse amplification laser architecture produced a record 913 µJ pulse energy with just 485 fs pulse durations at a 4.8 kHz repetition rate (4.4 W average power and ~1.9 GW peak power).
2. LMA Er-doped HEM fiber amplifier for high energy CPA systems
The LMA HEM Er-doped phosphate fiber was fabricated using a rod-in-tube technique  with an erbium concentration in the core glass of 1.6 × 1026 ions/m3. This corresponds to 104 dB/m absorption at 1480 nm. The core and cladding diameters were 75 µm and 250 µm, respectively, as shown in Fig. 1(b), and did not have a polymer coating. The fiber core NA was 0.13 and the background loss, measured at 1310 nm by a standard cut-back technique, was 3 dB/m. Assuming a step-index profile, the Aeff for this fiber design is 2290 µm2 (simulated by MODE Solutions, product of Lumerical Solutions, Inc.).
To build an amplifier using the LMA HEM fiber, the input splice between the silica mode field adapter fiber and the HEM fiber was optimized by minimizing both the splice loss and the observed back-reflection. The splice was produced with an angle of 8°, as shown in Fig. 1(a). This design minimizes the back-reflection that may result from the spliced interfaces with significant refractive index contrasts (1.578 for phosphate glass and 1.444 for silica glass). This precaution minimizes the risk of parasitic lasing by avoiding the formation of a resonant cavity in the HEM fiber. Coreless, un-doped fiber with a diameter of 1000 µm was produced with a refractive index closely matching that of the amplifier fiber core for the purpose of end-capping the high energy HEM fiber. The addition of the end cap allowed the beam to expand in diameter, via normal Gaussian beam diffraction, prior to encountering a glass-to-air interface. This was a critical element in the design of the amplifier to prevent dielectric damage at the interface that may result from the extremely high peak powers of the emitted chirped pulses. Following splicing to the LMA HEM amplifier fiber, the end-cap was cleaved at a length of 8.5 mm and angle polished at 8°, again in order to minimize back reflection and avoid beam clipping at the glass-to-air interface.
A 10/90 splitter was used to monitor the signal input to the Er-doped HEM fiber amplifier as well as to monitor the back reflected light. The 1480 nm Raman fiber pump laser and the signal were combined by a single mode high power 1480/1553 nm wavelength division multiplexer (WDM). The WDM output was fusion spliced to the Er-doped HEM phosphate glass fiber amplifier. Finally, the assembled LMA Er-doped HEM fiber amplifier was mounted on a constant temperature water chilled aluminum plate. This new HEM fiber amplifier generates pulse energies more than 10 times higher than recently reported monolithic fiber femtosecond laser demonstrations . This new amplifier however retains the previously demonstrated low M2 and a fiber format that allows for compact packaging and industrial applications.
3. Experimental setup and results
In the present laser systems, the above described LMA Er-doped HEM fiber amplifier was pumped by a high power, all-fiber Raman laser with its output shifted to 1480 nm from ytterbium fiber lasers . Figure 2 shows a high level schematic of our fiber-optic industrial-grade femtosecond lasers which rely upon a standard chirped pulse amplification (CPA) architecture in a monolithic, fiber-optic format. This architecture includes discrete, active feedback and control loops based on laser average power and pulse quality.
Pulses from a mode-locked laser (MLL) at 40 MHz are spectrally broadened from 7 nm to 14 nm at full width, half maximum (FWHM) by a spectral broadening Er-doped single mode polarization maintaining (PM) fiber amplifier. A pair of chirped fiber Bragg gratings (CFBG) stretch the pulse duration to >4 ns (FWHM). The pulse rate is reduced from 40 MHz to 4.8 kHz in two steps using >35 dB extinction PM fiber acousto-optic modulators (AOM) labeled Picker 1 and Picker 2: first from 40 MHz to 200 kHz, and then from 200 kHz to 4.8 kHz. Two single mode PM Er-doped fiber amplifiers, Pre-Amp 1 and Pre-Amp 2, are used here to compensate for signal losses. To further manage the pulse quality, an active fiber coupled liquid crystal pulse shaper  is used between pre-amplifiers to precompensate for nonlinear phase accumulated in the amplifier chain. A single mode Er-doped non-PM fiber amplifier is spliced between the pulse shaper and the LMA Er-doped HEM fiber amplifier, Booster-Amp, to compensate for signal loss within the pulse shaper. Finally, a free-space Treacy grating compressor with groove density of 1200 lines per millimeter is used to compress the pulses from the LMA Er-doped HEM amplifier to less than 500 fs, as characterized by intensity autocorrelation. The sequence of light generation and amplification components are entirely fiber optic and fusion spliced together to form a rugged and stable optical path. Final output control through pulse compression is then performed in an industrialized free-space opto-mechanical assembly.
Figure 3(a) shows the optical spectra of the signal at the pulse shaper output when the pulse shaper is enabled and disabled, respectively. With the pulse shaper disabled, the optical spectrum of the signal has a spectral bandwidth of 11 nm (FWHM) centered at 1552.5 nm. In addition, there is a significant amplified spontaneous emission (ASE) peak around 1532 nm. A programmed flat-top pulse shaping function was then applied to the pulse shaper aimed at increasing the spectral bandwidth as well as reducing the ASE. As shown by the black curve of Fig. 3(a), the spectrum bandwidth at the pulse shaper output was increased to 20 nm (FWHM) with a quasi-flat-top spectral shape centered at 1552.5 nm. In addition, the ASE around 1532 nm has been filtered out.
With 63 W of pump power at 1480 nm and a 22.5 mW input signal coupled into the HEM amplifier, the output signal reaches 8.5 W (13.5% optical-to-optical efficiency) and 1.77 mJ per pulse at 4.8 kHz. The calculated gain is 0.92 dB/cm. The amplifier efficiency can be further improved by increasing the input signal power. However, care should be taken to avoid optical damage and strong nonlinearity generated in the single mode pre-amplifiers and fiber components. The pulse duration at the HEM booster amplifier is 4.35 ns as measured by a fast photo diode connected to a fast sampling scope (200 MHz, 2GS/s). The measured beam propagation parameter, or M2, at the maximum pulse energy of 1.77 mJ from the HEM fiber amplifier output was just 1.11 in both the horizontal and vertical axes, demonstrating a nearly diffraction limited spatial beam quality. Figure 3(b) shows the optical spectra of the signal at the HEM booster amplifier input (red curve) and at the compressor output (black curve). The calculated saturation energy of the LMA HEM fiber amplifier with a 54 µm effective mode diameter is 0.4 mJ. As the output signal energy exceeds this value, the resulting output signal exhibits significant spectral tilt and distortion to the leading edge of the up-chirped amplified optical pulse. Due to the chirped nature of the spectrum, the more red wavelengths experience greater amplification than the blue as it enters the amplifier first, scavenging more of the available energy. This behavior may increase the nonlinearities in a fiber CPA system as the optical spectral bandwidth is reduced. Such spectral tilting is clearly shown in Fig. 3(b) as the LMA HEM fiber output reaches 1.77 mJ. The B-integral generated from the HEM fiber amplifier here is estimated to be 2 radians.
Based on our evaluations, at low signal repetition rates a measureable portion of the continuous wave ASE exists in the laser output signal. The out-of-band ASE is filtered by the free space compressor gratings and does not contribute to the laser output signal. However, there will be a small amount of in-band ASE remaining within the bandwidth of the pulse spectrum which is spread over time. In order to distinguish “pure pulses” from the background, the laser output is sent through the AOM and the optical power in the first order diffracted beam is measured under two different driver configurations. A first measurement is made over a narrow window that includes the signal pulse along with the small contribution from the ASE. A second measurement is then made over a much longer time span (by comparison) that includes the signal pulse in addition to the ASE energy present in between adjacent pulses. With knowledge of the two time windows and the measured energies, the total contribution of continuous wave ASE energy can be calculated and subtracted to identify the “pure pulse” energy. In these measurements, the first time window was set to be 0.7 µs, much wider than the pulse duration, but much narrower than the second time span: 288.3 µs (which is the time between pulses with 4.8 kHz repetition rate). The estimated ASE energy portion was 12%, and the resultant pure pulse energy of the individual compressed pulses with the maximum pulse energy was 913 µJ.
Figure 4 shows the background-free, second harmonic generation (SHG) intensity autocorrelation of the 913 µJ pulses output from the system pulse compressor (black line). The FWHM of the measured autocorrelation trace is 746 fs. From this, a sech2 pulse autocorrelation with assumed 0.65 deconvolution factor was used to estimate the duration of the pulse to be 485 fs. The estimated peak power of these pulses is ~1.9 GW. It is possible the actual pulse duration may be slightly longer than 485 fs considering the difference between the actual pulse shape and the assumed sech2 shape. Similarly, the actual peak power may be slightly lower than the estimated peak power due to the non sech2 shape of the pulse and that portion of the energy that lies in the pedestal. The theoretical sech2-shaped pulse autocorrelation is included in Fig. 4 for reference (red dashed line). The near field beam profile at the compressor output is also shown in the inset. The slight asymmetry of the beam may be attributed to the 8° angled splice at the input to the HEM fiber amplifier. The measured beam propagation parameter, or M2, at the maximum compressor output pulse energy of 913 µJ was 1.21 (horizontal) and 1.21 (vertical), demonstrating nearly diffraction limited spatial beam quality.
With a very short LMA Er-doped HEM fiber with an Aeff of 2290 µm2, we have demonstrated a fiber amplifier producing 913 µJ energy pulses with less than 500 fs duration and 1.9 GW peak power from a monolithic CPA fiber-optic amplification architecture. This represents better than 9 × increase in the pulse energy output capability of a state-of-the-art all-fiber-optic CPA ultra-short pulse laser systems .
We have experimentally demonstrated a 1.55 µm monolithic fiber-optic CPA system generating 913 µJ per pulse with less than 500 fs pulse duration at a repetition rate of 4.8 kHz (4.4 W average power and ~1.9 GW peak power). This system output performance was achieved using a LMA Er-doped HEM fiber amplifier with a B-integral of approximately 2 radians, and advanced adaptive pulse shaping and spectral broadening techniques. It is anticipated that further optimization of the pulse shaping technique could further minimize spectral tilting and SPM in the HEM fiber amplification. Additional system improvements such as further increases in the Aeff, increasing Er-doping concentration, or longer pulse stretch factors may well result in shorter fiber lengths and permit pulse energy well in excess of this 1 mJ pulse energy milestone.
Elements of this work were sponsored by Navy Contract N00164-11-C-BT07. NAVAIR Public Release Distribution Statement A-“Approved for Public release; distribution is unlimited”.
References and links
1. P. S. Banks, B. C. Stuart, A. M. Komashko, M. D. Feit, A. M. Rubenchik, and M. D. Perry, “Femtosecond laser materials processing,” Proc. SPIE 3934, 14–21 (2000). [CrossRef]
2. A. E. Siegman, Lasers (University Science Books, 1986), pp. 385–386.
3. X. Peng, K. Kim, M. Mielke, S. Jennings, G. Masor, D. Stohl, A. Chavez-Pirson, D. T. Nguyen, D. Rhonehouse, J. Zong, D. Churin, and N. Peyghambarian, “High efficiency, monolithic fiber chirped pulse amplification system for high energy femtosecond pulse generation,” Opt. Express 21(21), 25440–25451 (2013). [CrossRef] [PubMed]
5. L. Dong, X. Peng, and J. Li, “Leakage channel optical fibers with large effective area,” J. Opt. Soc. Am. B 24(8), 1689–1697 (2007). [CrossRef]
6. J. Želudevičius, R. Danilevičius, K. Viskontas, N. Rusteika, and K. Regelskis, “Femtosecond fiber CPA system based on picosecond master oscillator and power amplifier with CCC fiber,” Opt. Express 21(5), 5338–5345 (2013). [CrossRef] [PubMed]
7. F. Stutzki, F. Jansen, T. Eidam, A. Steinmetz, C. Jauregui, J. Limpert, and A. Tünnermann, “High average power large-pitch fiber amplifier with robust single-mode operation,” Opt. Lett. 36(5), 689–691 (2011). [CrossRef] [PubMed]
8. M. Laurila, M. M. Jørgensen, K. R. Hansen, T. T. Alkeskjold, J. Broeng, and J. Lægsgaard, “Distributed mode filtering rod fiber amplifier delivering 292W with improved mode stability,” Opt. Express 20(5), 5742–5753 (2012). [CrossRef] [PubMed]
9. T. Eidam, J. Rothhardt, F. Stutzki, F. Jansen, S. Hädrich, H. Carstens, C. Jauregui, J. Limpert, and A. Tünnermann, “Fiber chirped-pulse amplification system emitting 3.8 GW peak power,” Opt. Express 19(1), 255–260 (2011). [CrossRef] [PubMed]
10. J. W. Nicholson, J. M. Fini, X. Liu, A. DeSantolo, P. Westbrook, R. Windeler, E. Monberg, F. DiMarcello, C. Headley, and D. DiGiovanni, “Single-frequency pulse amplification in a higher-order mode fiber amplifier with fundamental-mode output,” in CLEO: 2013, OSA Technical Digest (online) (Optical Society of America, 2013), paper CW3M.3.
11. X. Wang, Q. Nie, T. Xu, and L. Liu, “A review of the fabrication of optic fiber,” Proc. SPIE 6034, 60341D (2006). [CrossRef]
12. M. Mielke, X. Peng, K. Kim, T. Booth, W. Lee, G. Masor, X. Gu, R. Lu, M. Hamamoto, R. Cline, J. Nicholson, J. Fini, X. Liu, A. DeSantolo, P. Westbrook, R. Windeler, E. Monberg, F. DiMarcello, C. Headley, and D. DiGiovanni, “High energy, monolithic fiber femtosecond lasers,” in Conference on Lasers and Electro-Optics Europe/Quantum Electronics and Laser Science and Photonic Applications Systems Technologies, Technical Digest (CD) (Optical Society of America, 2013), paper PD-A.3.