We demonstrate high-power high-efficiency cavity-enhanced second harmonic generation of an in-house built ultra-high spectral density (SBS-suppressed) 1178nm narrowband Raman fibre amplifier. Up to 14.5W 589nm CW emission is achieved with linewidth Δν 589<7MHz in a diffraction-limited beam, with peak external conversion efficiency of 86%. The inherently high spectral and spatial qualities of the 589nm source are particularly suited to both spectroscopic and Laser Guide Star applications, given the seed laser can be easily frequency-locked to the Na D2a emission line. Further, we expect the technology to be extendable, at similar or higher powers, to wavelengths limited only by the seed-pump-pair availability.
©2009 Optical Society of America
Signal distortions produced by atmospheric turbulence have long since been a challenge to high-resolution ground-based astronomy. Provided a reference source with sufficient brightness, advanced adaptive optics systems can now compensate for such distortions in real time. In some cases, natural guide stars can be used, but sky coverage can be significantly increased using laser guide stars (LGS). These are produced by the resonant backscatter of Sodium atoms in the Earth’s mesosphere (~90 km altitude), when illuminated by high power (typically 10W-class) lasers frequency-locked to the D2a emission line of atomic sodium (λ~589.159nm) . Unfortunately, no current solid-state medium has sufficient optical gain directly at this wavelength.
Dye laser systems have been successfully used for Laser Guide Star assisted Adaptive Optics (LGS AO) correction, but they are typically high-maintenance and can be troublesome to operate at remote astronomical observatories. More reliable alternatives are becoming available with recent advances in nonlinear frequency conversion. Greater than P589~50W CW has been achieved based on the cavity-enhanced Sum-Frequency Generation (SFG) from two single-frequency 1064nm and 1319nm lasers  but such multiple-cavity systems remain complex. The return flux from a LGS can be noticeably increased by the re-pumping of the Sodium atom. In practice this can be done by including a ~10% fraction of the launched power at the D2b transition frequency, in this case, 1.71 GHz toward the blue with respect to the D2a centre line .
LGS AO systems installed at remote astronomical observatory locations naturally favour compact, highly reliable, maintenance- and alignment-free devices, in order to maximize useful observation time at these sites. Fibre-based oscillators and amplifiers clearly alleviate constraints on cavity alignment, as does (to a certain extent) the single-pass frequency conversion of their outputs (although, in this case, the generated visible powers remain typically limited). Indeed, up to P589~3.5W CW has been obtained from the single-pass SFG of two Lanthanoid-doped fibre amplifiers (1583nm and 938nm from Erbium and Neodymium, respectively), a result in fact limited by the available 938nm power . Other fibre solutions have been studied, and of direct interest here are certain Bi- and Yb-doped systems. Rulkov et al.  obtained P1178~6.4W (0.2nm linewidth) from a Bi-doped fibre cavity (producing P589~125mW in PP-MgO:LN). Using a linearly polarized Yb-doped fibre system, Sinha et al.  generated a P1150~90mW (frequency doubled in a PP-LN waveguide to P575~40mW at η575~67%,). Shirakawa et al.  recently demonstrated both the long-wavelength operation and power-scalability of such Yb-doped systems under MOPA configuration (using a cladding pumped, Raman fibre seeded, photonic-band-gap fibre), reaching P1156>30W and P1178~9.1W CW, yet with linewidth too broad for efficient cavity-enhanced frequency doubling. We note, however, that the single-pass frequency doubling process itself does not currently appear to be a limiting factor at the power levels required for LGS 589nm sources. Using CW single-line Yb-doped fibre systems as fundamentals, Sinha et al.  obtained P532~18.3W CW in truly stoichiometric PP-SLT (η532~24.4%, limited by thermal dephasing), and Südmeyer et al.  reached similar powers at remarkably high conversion efficiencies under intra-cavity SHG in LBO (P532 > 20W, at η532~88%).
Several authors indicate that Raman-fibre-based high-power systems are not suited for generating linewidths narrower than the ~100MHz natural linewidth of Brillouin scattered in-fibre processes, given both the relative Raman and limiting Brillouin gain coefficients of typical fibres, and the strong line-broadening due to four-wave-mixing observed in most Raman-enhanced fibres [6, 10]. Feng et al.  first reported 589nm generation via frequency doubling of a Raman fibre system, using a non-critically phase matched LBO to produce P589~10mW (Δλ589~0.7nm). Sharma  has single-pass frequency doubled a relatively broadband CW Raman fibre laser developed by IPG, using PP-MgO:LN to achieve P589~1.52W (η589~7.7%, Δλ589~0.8nm). Georgiev et al.  used a similar IPG system (Δλ1178~0.4nm) and a suitably short PP-LN sample to produce P589~3.03W (η589~12.3%, Δλ589~0.6nm), also CW. Finally, a more spectrally intense system (P1178~25W with linewidth ΔνFWHM, 1178~6-9GHz, depending on conditions) was engineered for ESO by Volius Ltd. and frequency doubled in our laser laboratories, in a 30mm long PP-KTP [14, 15] to achieve P589~4.2W at η589~22%, with ΔνFWHM, 589~45GHz. More recently, we obtained P589~5.25W after optimisation of the same set-up. In all cases, the spectral intensity of the 1178nm beam, combined with the spectral acceptance of the SHG process in the chosen material, significantly limited the visible conversion efficiencies. To the best of our knowledge, the Volius Ltd. system was, at the time, the most spectrally intense 20W-class Raman fibre system available, operating in this wavelength window.
The results outlined in the remainder of this paper follow directly from our recent important advances in the field of narrowband high-power Raman fibre amplifiers. This builds on previous work completed in our laboratories . Using a simple but extremely effective method for SBS suppression, high spectral intensities have been reached from our new systems , demonstrating continuous wave P1178>20W within a linewidth envelope ΔνFWHM<3.4MHz. The optical properties of this new source are particularly suited to high-efficiency cavity-enhanced frequency doubling, as will be shown in the following sections.
Finally, we note that the MOPA configuration of this Raman fibre system makes the frequency locking to the selected sodium transition reasonably trivial. The generated 589nm beam is compatible with modulator-induced sideband generation for inclusion of the D2b emission line, and we expect the IR linewidth can be readily fine-tuned (within the acceptance of the doubling cavity and without significant loss of conversion efficiency) for both optical re-pumping and LGS return flux enhancement. The result is in our opinion the most suitable, flexible and reliable technology for future LGS-AO systems.
2. Background, experimental set-up and procedure
We use an 1120nm Raman oscillator pumped, 1178nm DFB diode seeded, non-polarization-maintaining state-of-the-art SBS-suppressed CW narrowband Raman fibre amplifier as fundamental for the SHG . Not reported in this article, a similar scheme based on PM fibres is also progressing in collaboration with industry. Both the narrowband non-PM 1178nm amplifier and the 1120nm pump were built in-house. The pump is capable of P1120>80W CW emission, with linewidth Δλ~2nm .
We recover the amplified 1178nm elliptical polarisation to vertical linear (σ-polarised) with PER>25dB using a servo-controlled bulk-optic feedback loop, and couple the optically isolated, collimated beam into a bowtie-configured (modified, commercial) cavity, using two pairs of steering mirrors (MS1–S4) and a mode-matching lens, as schematically represented in Fig. 1. The cavity is locked using the well-established Pound-Drever-Hall method : the cavity length is adjusted by acting on the piezo-mounted mirror, MC2, based on a signal measured using a fast photodiode, PD2. The photodiode PD1 serves to monitor the intra-cavity IR power. Both the original cavity and associated electronics were purchased as off-the-shelf products from Toptica Photonics AG .
We use a thermally stabilised 3×3mm2 aperture, 30mm long, non-critically phase matched AR-coated LBO crystal within the cavity (TOP~43°C). It is placed between two curved mirrors (MC3, MC4, rC=46mm), and the generated 589nm beam is extracted through the mirror MC4 (R1178, σ>99.98%, T589, π>99.7%). The waist (e-2 diameter) in the crystal is 2ω0~48µm.
In order to achieve high conversion efficiencies, it is critical to minimize the (passive) cavity losses: the crystal end-facets are anti-reflection-coated for the 1178nm beam, and we estimate our total per-pass IR loss (excluding the transmission of the incoupling mirror MC1), to be δ ≤ 0.008. The intra-cavity power is further determined by the reflectivity of the selected input coupler (IC) and the overall conversion efficiency (active losses), amongst others. Good impedance matching to the enhanced fundamental mode is best achieved at low incident powers for a relatively high IC reflectivity (and vice versa). As the target (visible) output power is increased, so the reflectivity of the IC must decrease, in order to obtain best matching to this enhanced mode. Thus, it becomes possible to optimize the IC reflectivity for a desired (achievable) output power and efficiency, knowing all other cavity parameters. We start by experimentally investigating the effect of varying the IC reflectivity on conversion efficiencies as a function of incident IR power, and present the results in the next section.
3. Results and discussion
We experimentally investigate the quality of the conversion for a number of IC mirror reflectivities, and after optimising the coupling of the 1178nm beam to the fundamental cavity mode and tuning the control electronics at each power level, generate the graphs presented in Fig. 2.
The trend of output power with IC mirror reflectivity is as expected (and as described above): a lower output power is observed before rollover of the efficiency curves, at lower reflectivities. At higher incident powers the high-reflectivity curves tend to saturate in efficiency before rollover; at this point, the combination of cavity parameters is such that the enhanced power extraction causes a mismatch in the coupling impedance. Empirically we observe a saturation of the efficiency at, for example, an input coupler reflectivity of RIC=0.94 and for an input power around P1178~10W (respectively, for RIC=0.92 and P1178~15W). The lower reflectivity (RIC=0.90) does not saturate in efficiency at our available incident pump power levels, and we expect this IC to be most suited for powers in the range P1178~23–25W.
We achieve a maximum P589~14.5W CW with external efficiency η589~86%, and expect only slightly higher efficiencies using this IC at powers in the range P1178~18–20W. Figure 2(a). strongly suggests that provided the fundamental mode coupling quality and other remaining parameters be maintained, higher visible powers can be just as easily achieved, given higher pump powers and a suitably selected IC. We are currently investigating ways to efficiently boost the available 1178nm pump power.
The inset in Fig. 2(a). illustrates the spatial profile of the 589nm beam, measured here using a SID4 wavefront sensor from Phasics , at P589~8W and after strong reflective attenuation of the beam (α~-60dB). We note that at this power, the phase-front error within the 1/e2 diameter is better than 0.018λ (<11nm rms) and beam ellipticity (or ratio of its two major axes) is 0.996. The power handling of the device (together with suitable attenuation) prohibited us from completing further measurements at higher visible powers. This problem is currently being addressed, but we expect no significant degradation of the measured beam quality, at P589~15W or higher, given the high beam quality is inherent to the use of the frequency doubling cavity.
Using a Ge or Si photodiode (respectively with rise times τGe<10ns, τSi<1ns) in the appropriately attenuated beam paths (respectively, leaked through steering mirror MC4 and fractionally picked off from the visible output beam using a pair of uncoated 2° wedge prisms), we investigate the fast temporal instabilities of the visible output (for P589~8–10W), simultaneously with the IR fundamental (both locked to the cavity, and free-running). The signals are terminated at 50Ω and acquired through an oscilloscope (Agilent, 100MHz, 200Ms.s-1). We present the results related to the 1178nm elsewhere , and focus here on the stability of the visible power. Results are presented in Fig. 3.
We see that there are no intensity noise features in the temporal range [DC, 100 kHz-1]: the white noise suppression relative to its DC level is better than -60dB over this range. The resolution of the DC component itself, approximately 1kHz in width here, is limited by the sample length. A closer look at the DC component using a low-frequency spectrum analyser (HP3582A, with DC-25kHz operating range) in fact reveals that it is resolution-limited down to 6Hz. We use a similar device (HP8593E, with 9kHz - 22GHz operating range) to confirm the lack of spectral noise features over [100MHz-1GHz], and do not investigate system behaviour at other frequencies. As required for the described LGS AO application, the output power is also highly stable over longer timescales.
Finally, we simultaneously measure the incident IR and visible linewidths using a pair of appropriately configured scanning Fabry-Pérot OSA. Each device has a FSR~1GHz and finesse f>500 (using a presumed single-line source we measured fIR~720), and both units were manufactured by Toptica Photonics AG. Typical scans are presented in Fig. 4.
In practice, we first note that all pairs of IR and visible spectra are essentially indistinguishable with varying P589, and remain always around 7MHz in the visible (the FWHM589 of a best fitting Gaussian distribution is averaged to 7MHz over 10 scans, as measured over varying timescales and with minimal standard deviation). The same observation holds for the detail of the evident fast peaks, especially obvious in the visible scan. These observations are readily explained by the fast frequency jitter inherent in our DFB diode seed , and similarly, we are able to confirm this by scanning the FP OSA at various rates. Based on the additional information gained from the RF spectrum analyser results (white noise only, outside of the DC component on those scans), we consider this frequency jitter to be random.
Our visible linewidth envelope measurements are highly consistent with (fIR+ΔfIR) being frequency doubled to give (fVIS+2ΔfIR), whereas we know that the lineshape of the instantaneous visible spectrum (intensity) will follow the auto-convolution of the IR lineshape (in amplitude), which can be further broadened as a function of the spectral phase correlation of the IR light . (This spectral phase correlation explains amongst other effects, the previously presented  6-to-8-fold broadening of the FWHM linewidth under SHG to 589nm using the 1178nm Raman amplifier built by Volius Ltd, and we speculate that it has its source in the four-wave-mixing responsible for the observed line broadening). Consequently, we suspect that the quasi-instantaneous IR linewidth is significantly less (possibly an order of magnitude or better) than the 3.5MHz frequency instability (as measured over 230µs). Subsequently (and given this linewidth is well within both the natural SBS linewidth of the fibres used, and the acceptance of the doubling cavity) we might also expect to achieve both narrower IR and visible linewidths without noticeable changes to the overall system behaviour. This might be achieved via either additional frequency stabilisation of the emission frequency of the DFB diode, or via use of a different seed technology. We are currently investigating this latter method.
Based on additional reports [24, 25], we consider the generated (comfortably 10W-class) narrowband CW 589nm source to be also particularly suitable for spectroscopy and atom trapping and cooling. The result itself is not, however, limited only to Sodium: we further expect that the developed (possibly frequency-doubled, or -quadrupled) Raman fibre technology will produce, in a similar manner, any chosen IR, visible or even UV wavelength at a similar 10W-class power level (or higher), with the respective wavelength ranges being solely limited by available suitable seed- and pump-pairs.
4. Conclusions and scope
We present here what we believe to be to date, the highest power, narrowest linewidth and highest spectral density, Raman fibre amplifier based, frequency-doubled CW source at 589nm. It is in fact to the best of our knowledge, the highest power 589nm source obtained from any IR fibre system, to date. We achieve P589=14.5W, with external conversion efficiency η589~86% and visible linewidth Δν589, FWHM≤7MHz. We further note that higher external conversion efficiencies are only seldom reported elsewhere in the Literature, and that these presented here are at least equal to the majority of those obtained using narrow- or single-line Lanthanoid-doped fibre systems (e.g., Yb). We believe the developed compact laser technology is fully compatible with the requirements of most spectroscopic work and many medical applications amongst others, and based solely on pump- and seed-pair availabilities, might provide any wavelength in the IR, visible or UV, at CW powers comfortably in excess of 10W.
This particular source was developed for the purposes of LGS AO observations at ESO’s current ground-based telescopes, and already satisfies basic power and spectral requirements. Additional on-going and future work is aimed at continued development of PM and non-PM fibre Raman amplifiers, and at further increasing the available output power at 589nm. We also intend to demonstrate features allowing an increase in return flux from the generated LGS (without increasing projected power), amongst which is the Sodium re-pumping by addition of the D2b line. Finally, we are currently also investigating the coherent combination of a pair of seed-twinned Raman amplifiers, and the subsequent SHG of the combined beam. Technology transfer to industry is also ongoing as it is one of the missions of our public institute.
References and links
1. N. Ageorges and C. Dainty, Eds. Laser Guide Star Adaptive Optics for Astronom Kluwer (Academic Publishers, ISBN 0-7923-6381-7, 2000).
2. C. A. Denman, P. D. Hillman, G. T. Moore, J. M. Telle, J. E. Preston, J. D. Drummond, and R. Q. Fugate, “50-W CW Single Frequency 589-nm FASOR,” in Advanced Solid-State Photonics (TOPS), C. Denman and I. Sorokina, Eds., 98, OSA Trends in Optics and Photonics (Optical Society of America), paper 698 (2005).
3. C. Denman, P. Hillman, G. Moore, J. Telle, J. Drummond, S. Novotny, M. Eickhoff, and R. Fugate, “Two-Frequency Sodium Guidestar Excitation at the Starfire Optical Range,” presentation at the Center for Adaptive Optics Fall Retreat (2006).
4. D. M. Pennington, J. W. Dawson, A. Drobshoff, S. Payne, D. Bonaccini, W. Hackenberg, and L. Taylor, “Compact fibre laser for 589nm laser guide star generation,” Conference on Lasers and Electro-optics Europe, Cleo/Europe, 10.1109/CLEOE.2005.1568310, 532 (2005).
5. A. B. Rulkov, A. A. Ferin, S. V. Popov, J. R. Taylor, I. Razdobreev, L. Bigot, and G. Bouwmans, “Narrow-line, 1178nm CW bismuth-doped fibre laser with 6.4W output for direct frequency doubling,” Opt. Express 15, 9, 5473–5476 (2007). [CrossRef] [PubMed]
6. S. Sinha, C. Langrock, M. J. F. Digonnet, M. M. Fejer, and R. L. Byer, “Efficient yellow-light generation by frequency doubling a narrow-linewidth 1150nm ytterbium fibre oscillator,” Opt. Lett. 31(3), 347–349 (2006). [CrossRef]
7. A. Shirakawa, H. Murayama, K. Ueda, C. B. Olausson, J. K. Lyngsø, B. J. Mangan, and J. Broeng, “High-power Yb-doped solid-core photonic bandgap fibre amplifier at 1150–1200nm,” presented at Photonics West, SPIE, Fiber Lasers VI: Technology, Systems, and Applications (2009).
8. S. Sinha, D. S. Hum, K. E. Urbanek, Y. Lee, M. J. F Digonnet, M. M. Fejer, and R. L. Byer, “Room-Temperature Stable Generation of 19 Watts of Single-Frequency 532-nm Radiation in a Periodically Poled Lithium Tantalate Crystal,” J. Lightwave Technol. 26(24), 3866–3871 (2008). [CrossRef]
9. T. Südmeyer, Y. Imai, H. Masuda, N. Eguchi, M. Saito, and S. Kubota, “Efficient 2nd and 4th harmonic generation of a single-frequency, continuous-wave fibre amplifier,” Opt. Express 16(3), 1546–1551 (2008). [CrossRef]
10. L. J. Nilsson, Y. Jeong, and P. Dupriez, “Fibre-based 589nm Laser for Sodium Guide Star,” Southampton University report note, Guide star final report SPC FA8655-04-1-3065 (2006).
11. Y. Feng, S. Huang, A. Shirakawa, and K. Ueda, “589nm Light Source based on Raman Fiber Laser,” Jpn. J. Appl. Phys. 43(6A), L722–L724 (2004). [CrossRef]
12. L. B. Sharma, “1.52W Frequency Doubled Fibre-Based Continuous Wave Orange Laser Radiation at 590nm,” Rev. Laser Eng. 33(2), 130–131(2005).
13. D. Georgiev, V. P. Gapontsev, A. G. Dronov, M. Y. Vyatkin, A. B. Rulkov, S. V. Popov, and J. R. Taylor, “Watts-level frequency doubling of a narrow line linearly polarized Raman fibre laser to 589nm,” Opt. Express 13(18), 6772–6776 (2005). [CrossRef]
14. L. Taylor, Y. Feng, D. Bonaccini Calia, and W. Hackenberg, “Multi-watt 589-nm Na D2-line generation via frequency doubling of a Raman fibre amplifier: a source for LGS-assisted AO,” Proc. SPIE 6272, 627249 (2006). [CrossRef]
15. D. Bonaccini Calia, W. Hackenberg, S. Chernikov, Y. Feng, and L. Taylor, “AFIRE: fibre Raman laser for laser guide star adaptive optics,” Proc. SPIE 6272, 62721M (2006). [CrossRef]
16. Y. Feng, L. Taylor, and D. Bonaccini Calia, “Multiwatts narrow linewidth fibre Raman amplifiers,” Opt. Express , 16(15), 10927–10932 (2008). [CrossRef]
17. Y. Feng, L. Taylor, and D. Bonaccini Calia, “20W CW, 4MHz linewidth Raman fibre amplifier with SHG to 589nm,” Proc. SPIE , Fiber Lasers VI: Technology, Systems, and Applications, D. V. Gapontsev, D. A. Kliner, J. W. Dawson, and K. Tankala Eds., 7195, xvii-xviii (2009).
18. Y. Feng, L. Taylor, and D. Bonaccini Calia, “80W-class 1120nm Raman fibre oscillator,” in preparation
19. R. W. Drever, J. L. Hall, F. V. Kowalski, J. Hough, G. M. Ford, A. J. Munley, and H. Ward, “Laser phase and frequency stabilization using an optical resonator,” Appl. Phys. B 31(2), 97–105 (1983). [CrossRef]
22. S. Kraft, A. Deninger, C. Trück, J. Fortágh, F. Lison, and C. Zimmermann, “Rubidium spectroscopy at 778–780 nm with a distributed feedback laser diode,” Laser Phys. Lett. 2(2), 71–76 (2005). [CrossRef]
23. R. Holzlöhner, L. Taylor, Y. Feng, D. B. Calia, and W. Hackenberg, “Modeling of spectral broadening in second-harmonic generation,” in CLEO/Europe and IQEC 2007 Conference Digest, Optical Society of America, paper CD_25 (2007).
24. K. B. Davis, M. O. Mewes, M. R. Andrews, N. J. van Druten, D. S. Durfee, D. M. Kurn, and W. Ketterle, “Bose-Einstein condensation in a gas of sodium atoms,” Phys. Rev. Lett. 75(22), 3969–3973 (1995). [CrossRef]
25. E. Streed, A. Chikkatur, T. Gustavson, M. Boyd, Y. Torii, D. Schneble, G. Campbell, D. Pritchard, and W. Ketterle, “Large atom number Bose-Einstein condensation machines,” Rev. Sci. Instr. 77(2), 023106-023106-13 (2006). [CrossRef]