Open Access
Add to BibSonomyAbstract
We report on the generation of 17.6W of visible radiation at 650nm using four-wave-mixing in an endlessly single-mode silica fiber. The conversion efficiency was as high as ~30%. This high efficiency could be obtained by exploiting the natural absorption of silica for the mid-infrared radiation >2.5µm. In a separate experiment 1.6W of mid-IR radiation at 2570nm were generated simultaneously with 14.4W at 672nm. These power levels of picosecond red radiation are among the highest reported so far for a diffraction limited beam quality in this wavelength region.
©2012 Optical Society of America
1. Introduction
Sources in the visible and mid-infrared (mid-IR) wavelength regions are becoming very attractive due to the growing number of emerging applications for them in the industrial, biomedical, military, and/or civil engineering areas. These include Raman spectroscopy [1], combustion monitoring [2], environmental monitoring, gas sensing [3], active thermography [4], semiconductor processing [5], and counter-measures, just to mention a few. Furthermore, the recent popularity gained by Thulium/Holmium doped fibers [6] as active materials for mid-IR radiation in the 2µm region has evidenced the urgent need for suitable high-power seed sources in this wavelength range. However, the availability of coherent sources in the visible and mid-IR regions is somewhat scarce and the few available ones tend to be very complex and expensive.
There are several approaches to generate coherent visible radiation, the most conventional of which is the direct development of visible lasers such as, for example, semiconductor [7], organic dye [8], or quantum well [9] lasers. However, even though these technologies are widespread and, in some cases, well established, they are not suitable for many of the applications mentioned above due to their relatively low output powers. A different approach to obtain visible radiation is frequency conversion, in which energy at an initial non-visible wavelength (normally near-IR) is transformed into visible light via some non-linear process. For example, the use of nanowires of non-linear optical materials illuminated by near-IR sources can lead to the generation of visible radiation via second harmonic generation [10].
Similarly, the generation of mid-IR radiation can be done either directly or by frequency conversion. Thus, the direct generation of mid-IR radiation is almost exclusively based on the use of solid-state lasers fabricated using exotic materials (antimony lasers or lead-salt diode lasers, for example), or rare-earth doped materials (such as Thulium and Holmium) [11]. On the other hand, there are a wide variety of frequency conversion schemes and non-linear materials for the generation of mid-IR radiation. One popular approach is to use difference frequency mixing in non-linear crystals exhibiting second order non-linearity, such as periodically-poled LNbO3 [12], for example. Another widespread technique is the use of optical parametric oscillators or generators [13]. Further approaches to achieve frequency conversion to the mid-IR are the use of photonic band-gap devices, cascaded Raman generation [14], and cascaded parametric oscillators [15].
Most of the sources listed above are, generally, quite expensive and complex. Furthermore, very few of those sources, particularly those based on frequency conversion, use an optical fiber for the generation of new wavelengths. This is mainly because most of the methods listed above rely on the use of second order non-linearity, which is extremely small in conventional fibers [16]. However, the combination of the interesting characteristics of visible/mid-IR radiation with the flexibility and convenience provided by optical fibers can be extremely attractive for many applications. This can be achieved by exploiting the third order non-linearity of optical fibers in the form of four wave mixing. Thus, a technique to generate visible or mid-IR radiation in Large Mode Area (LMA) optical fibers has already been reported in [17] and [18]. This technique exploits degenerate four-wave-mixing (FWM) in a conventional endlessly single-mode silica fiber (ESF). This process, given the appropriate pump wavelength, simultaneously generates coherent narrowband radiation in the two wavelength regions of interest. This approach is ideal for developing compact and cost-effective sources for CARS microscopy [19], where relatively low average powers are required. However, under high power operation, the pulses coming out of the system are typically not clean, exhibiting multiple peaks due to successive energy back-conversion processes. This also limits the conversion efficiency and the maximum peak power of the output pulses, which can seriously compromise the applicability of the technique.
In this paper we first introduce a refinement of the previous technique in which all the problems mentioned above are simultaneously solved. This new idea consists in the attenuation of one of the FWM partners. If this attenuation is high enough, it will prevent the energy from being back-transferred from the FWM partners to the pump. Thus, the energy flow will always go from the pump to one of the FWM partners (the one not being attenuated). This will result in smooth output pulses, in improved conversion efficiency and in higher output peak powers. This attenuation can be artificially induced or it can be obtained by exploiting the absorption bands of a gas/material. Thus, for example, it is particularly attractive to exploit the strong attenuation peak typically present in silica glass above 2.7μm to increase the conversion efficiency in the visible range. This theoretical study also explains why the conversion efficiency obtained in our previous work (~35%) [17] was much higher than originally expected (without considering the mid-IR absorption).
Additionally, using this refined technique, we report on the generation of high-power visible and mid-IR diffraction-limited radiation. Thus, in separate experiments, 1.6W at 2570nm and 17.6W at 650nm could be demonstrated. This last value is, to the best of our knowledge, among the highest powers reported so far for diffraction limited red radiation. Thus, this is a very promising approach for generating high-power picosecond pulses with wavelength diversity.
The manuscript is divided as follows: in the second section a refinement of the generation technique presented in [17] will be discussed and in section 3 the high-power experimental setup and results will be presented. Finally some conclusions will be drawn.
2. Improved parametric generation of visible or mid-IR radiation
As detailed in [17], using endlessly single-mode fibers pumped in the normal dispersion regime has allowed for the first time the efficient generation of signal and idler photons very far away from the pump wavelength. The reason for this is that the mode field diameter of the light in endlessly single-mode fibers is almost independent of the wavelength, thus allowing a very good overlap of the electric fields even at widely spaced wavelengths. Moreover, pumping in the normal dispersion regime far from the zero dispersion wavelength (ZDW) typically implies that the phase matching condition [16] will only be fulfilled in a relatively narrow frequency band far away from the pump wavelength. Thus, with the technique presented in [17] it is possible to generate narrowband radiation in the visible and mid-IR wavelength ranges when pumping an endlessly single-mode fiber with a pump wavelength falling within the amplification band of Yb-doped fibers. Furthermore, this technique is inherently tunable (by changing the pump wavelength) and, by using large-mode area (LMA) ESFs, scalable to high powers.
In order to theoretically study the parametric generation of the light along the endlessly single-mode fiber, we have developed a simulation program based on the non-linear Schrödinger equation (NLSE) [16]. In this simulation tool the propagation of pump, signal and idler photons is modeled using coupled NLSEs.
The phase-matching diagram for a commercially available 15µm-core endlessly single-mode fiber (LMA-15 from NKT-Photonics) when pumped with 250ps picosecond pulses of 120kW peak power can be observed in Fig. 1 . The ZDW in this fiber is ~1250nm. As can be seen, when pumping with a wavelength of 1064nm, the fiber will be able to generate narrowband radiation around 650nm and 2929nm.
Fig. 1 Phase-matching diagram for the commercially available ESF LMA-15. When pumped at 1064nm this fiber will generate narrowband radiation at 650nm and at 2929nm.
However, when calculating the conversion efficiencies and output pulse shapes of such a system (see Fig. 2 ), some problems of the approach can be recognized. The most apparent one is that the output pulse shapes are not smooth, but present strong oscillations (Fig. 2(a)). Another problem is that the peak power of the output pulses is limited. This limitation is ~40% of the pump peak power for the visible wavelength in this case. Furthermore, as can be seen in Fig. 2(b), the energy conversion efficiency after 0.5m LMA-15 fiber is just ~15% for the visible wavelength and ~2% for the mid-IR wavelength.
Fig. 2 (a) Pulse shapes at the pump (blue), signal (red) and idler (green) wavelengths after ~0.4m LMA-15 fiber, and (b) evolution of the energy conversion efficiency along the fiber length.
All these limitations are rooted in the energy back-conversion process that takes place during parametric generation once that the signal and idler waves have reached a certain level [16]. This is schematically illustrated in Fig. 3 . At the beginning of the fiber all the energy/photons (represented by the filled circles in Fig. 3) are injected at the pump wavelength (center blue arrows in Fig. 3), as represented in Fig. 3(a). As this light starts to propagate through the fiber, the parametric generation of photons at the signal (left red arrow) and idler wavelength (right blue arrow) starts taking place. By virtue of FWM two pump photons are annihilated and a signal and idler photons are created (Fig. 3(b)). This process continues until the signal and idler waves reach a certain level. At that moment the direction of energy transfer changes, and signal and idler photons start being annihilated and pump photons are created (Fig. 3(c)).
Fig. 3 Schematic representation of the energy back-conversion process: (a) at the beginning all the energy/photons (represented by the circles) are injected at the pump wavelength (blue central arrow); (b) as the light propagates down the fiber FWM starts to transfer energy to the signal (left red arrow) and idler (right green arrow) wavelengths; (c) this processes continues until the signal and idler waves are intense enough, moment at which the direction of energy transfer is reversed and the energy flows back to the pump wavelength.
This process explains the oscillations observed in the output pulses in Fig. 2(a), since they are the result of successive conversion and back-conversion processes. Additionally, the back-conversion process is also responsible for the limitation in the peak power of the output signal and idler pulses. Moreover, the fact that the energy periodically flows back into the pump as the light travels along the fiber also reduces the overall energy conversion efficiency. Thus, if the back-conversion could be prevented from happening, all these limitations would be simultaneously overcome. One way of achieving this is to introduce propagation losses at either the signal or the idler wavelength, as illustrated in Fig. 4 . The basic idea behind this is that during the back-conversion process one photon at the signal wavelength and one photon at the idler wavelength have to be annihilated to generate two pump photons. However, if there are no signal or idler photons, then the back-conversion cannot take place, i.e. it will be forbidden, and the energy flow will only go from the pump in the direction of either the signal or the idler (i.e. the one not being attenuated), as shown in Fig. 4(c). These strong propagation losses can be natural (material absorption) or artificial (doping, filters, etc). In this work we exploit the strong natural absorption of silica for wavelengths >2.7µm.
Fig. 4 Schematic representation of the method employed to avoid energy back-conversion: (a) at the beginning all the energy/photons (represented by the circles) are injected at the pump wavelength (blue central arrow); (b) as the light propagates down the fiber, FWM starts to transfer energy to the signal (left red arrow) and idler (right green arrow) wavelengths. However, the photons at the idler wavelength are lost due to the high propagation losses introduced (yellow arrows); (c) this way, since there is a strong imbalance in the number of photons at the signal and idler wavelengths, the energy back-conversion process is mitigated.
If strong propagation losses (500dB/m) at the idler wavelength are included in the simulations of the LMA-15 fiber (while all the other parameters remain the same as in Fig. 2), some significantly improved results can be obtained, as shown in Fig. 5 . There it can be seen that the output pulse at the signal wavelength (in red) has a smooth shape, and that its peak power is much higher than before (Fig. 5(a)). Moreover, in Fig. 5(b) it can be observed that the conversion efficiency for the signal wave has increased from 15% to 35%. These significant improvements come, however, at the cost of a very poor efficiency in the generation of the mid-IR idler wave. However, since most applications require either the generation of the signal or the idler wave, the loss of efficiency in the generation of the FWM-partner should not be of concern.
Fig. 5 (a) Pulse shapes at the pump (blue), signal (red) and idler (green) wavelengths after ~0.4m LMA-15 fiber, and (b) evolution of the energy conversion efficiency along the fiber length. In this case a propagation loss of 500dB/m at the idler wavelength has been considered in the simulations.
3. Experimental results
The experimental setup used in this work is schematically illustrated in Fig. 6 , which is very similar to that described in [20]. The system is seeded by a passively Q-switched microchip laser (Nd:YVO4/SESAM) delivering 250ps pulses at 1064nm with a variable repetition rate between 1 and 3MHz and a pulse energy of ~50nJ. This seed is sent in a double-pass amplifier consisting of a 40µm core diameter Yb-doped photonic-crystal fiber (PCF). This amplified seed is then launched into a large-pitch fiber amplifier [21] for further amplification. Thus, at the output of the system a maximum of ~110W at 1064nm can be obtained at repetition rates ranging from 1MHz to 3MHz. The output pulses have a maximum output energy of ~100µJ and a maximum peak power of ~400kW. The output of the system is monitored using a camera, an optical spectrum analyzer and a fast photodiode.
Fig. 6 Schematic diagram of the experimental setup. The system is seeded by a microchip laser (MCL) emitting 250ps pulses at a variable repetition rate from 1 to 3MHz. OI: optical isolator; HWP: half-wave plate; PH: pin-hole; FR: Faraday rotator; BPF: band-pass filter; PD: photodiode; OSA: optical spectrum analyzer; Cam: camera.
This high-power emission at 1064nm is then coupled into an ESF to generate FWM. Two commercially available fibers have been used: the LMA-15 (with a 15µm core) and the LMA-10 (with a 10µm core), both from NKT Photonics. Typical coupling efficiencies into these fibers are around 60%. The fibers had to be specially prepared to withstand the very high peak and average powers of the pump. This preparation consisted in completely removing the acrylate coating over the whole fiber length and in protecting the end-facets with end-caps. The end-caps consisted of ~400µm long pieces of 105µm core diameter multimode fibers. In order to characterize the parametric generated waves, the output of the ESF is directed towards a prism which spatially separates the different spectral components. Thus, the output power at the signal and pump wavelengths can be independently measured. Unfortunately our prism absorbs the mid-IR signal, so we had to use a long-pass Germanium filter (with a cut-off of ~2µm and ~40% transmissivity) placed in front of the output of the ESF to measure the mid-IR power content. Apart from these power measurements, the output spectrum and the pulse shapes at the signal wavelength were monitored as well.
For the first experiment a 27cm long piece of LMA-10 ESF was used. In this fiber the signal is emitted at 672nm and the idler at 2570nm. The relatively short emission wavelength of the mid-IR wave ensures that it will not suffer very high absorption by Silica over the 27cm long fiber. Thus, by coupling 60W pump power at 1064nm with 3MHz repetition rate, it was possible to obtain as much as 14.4W of signal power and, simultaneously, 1.6W of idler power (see Fig. 7 ). Thus the conversion efficiencies are ~23% and ~2.6% in the visible and mid-IR, respectively.
Fig. 7 Experimental results using 27cm of LMA-10 fiber. The fiber was pumped at 1064nm with a maximum of 60W pump power (250ps pulses at 3MHz repetition frequency).
In order to further increase the output power, a larger fiber was employed. Thus, for this second experiment 57cm of the LMA-15 were used. As shown in Fig. 1, in this fiber the signal wavelength is 650nm and the idler wavelength is 2929nm. Due to the long wavelength of the idler, it is expected that it would undergo much higher propagation losses than in the previous experiment with the LMA-10. Therefore, according to the theoretical considerations discussed in section 2, the conversion efficiency for the signal wave should increase and for the idler wave should decrease. This is confirmed in Fig. 8(a) , which shows a maximum power of 17.6W at 650nm when pumping with 60W at 1064nm (250ps at 1.5MHz). This translates into a ~30% conversion efficiency which, as expected, is higher than in the previous case. According to simulations, this conversion efficiency into the signal wave is compatible with a ~200dB/m attenuation of the idler wave. Unfortunately our Germanium filter was damaged during this experiment and we could not record the output power of the idler wave. However, a previous experiment with the same fiber but with 2MHz repetition frequency gave a maximum value of 0.54W for 40W coupled pump power. This implies a conversion efficiency of ~1.2% in the mid-IR, again satisfying the expectation of a reduced efficiency in comparison with the previous experiment. Additionally, in Fig. 8(b) the evolution of the output pulses at the signal wavelength with increasing output power can be seen. As predicted by numerical simulations, as the pump power is raised the peak power of the output pulses increases and the pulses become longer. Thus, the output pulse duration evolves from 77ps at 2.9W all the way up to 208ps at 17.6W. This implies that the pulses in the visible region reached a maximum peak power of ~56kW.
Fig. 8 Experimental results using 57cm of LMA-15 fiber. The fiber was pumped at 1064nm with a maximum of 60W pump power (250ps pulses at 1.5MHz repetition frequency): (a) output power at the signal wavelength as a function of the coupled pump power, (b) evolution of the output pulses at 650nm with increasing output power.
Finally, Fig. 9 shows the evolution of the spectrum at the output of the LMA-15 for increasing output power. As can be seen, the parametric generation is very clean, i.e. there is only energy around the pump and signal wavelengths. Additionally, it can be observed that for relatively low powers (up to ~3W) the spectrum at the signal wavelength is extraordinarily narrow, but it quickly broadens as the output power increases. This broadening is due to non-linear effects such as SPM and Raman scattering [16], and it is an indirect proof of the high peak powers generated in this experiment at 650nm (~50kW). At 16.1W the spectrum is broadened to a FWHM of 1.2nm, which should still be unproblematic for most applications. The spectrum at the maximum power of 17.6W was nearly indistinguishable from that at 16.1W.
Fig. 9 Evolution of the spectrum at the output of the 57cm of LMA-15 fiber for increasing output powers.
4. Conclusion
We have reported on the efficient generation of high-power visible and mid-IR radiation exploiting FWM in an endlessly single-mode fiber. Additionally, a refinement of the parametric generation technique has been presented. This refinement is based on the attenuation of one of the FWM-partners which leads to a substantially increased conversion efficiency of the other FWM partner together with smooth output pulses and higher peak powers. This refinement in conjunction with a pump laser delivering 110W output power at 1064nm (with 250ps pulses and a variable repetition rate of 1MHz to 3MHz), has allowed us obtaining a maximum output power of 17.6W at 650nm with a conversion efficiency of 30%. Additionally, in a different experiment ~14.4W at 672nm and ~1.6W at 2570nm could be simultaneously generated. In conclusion, the simplicity, efficiency and high-power scalability make this technique a very attractive alternative for the generation of high-power picosecond pulses in the visible and mid-infrared regions. This together with the inherent tuneability capabilities of the technique (with just a moderate tuning of the pump wavelength a wide range of wavelengths in the visible and, especially, in the mid-infrared regions can be swept), make this approach a very promising alternative to generate a source for a wide range of applications including spectroscopy, CARS, and LIDAR.
Acknowledgments
The research leading to these results has received funding from the German Federal Ministry of Education and Research (BMBF) under contract 13N10773, the European Research Council under the European Union's Seventh Framework Programme (FP7/2007-2013) and the Thuringian Ministry of Education, Science and Culture under contract PE203-2-1 and contract B514-10061.
References and links
1. S. Shim and R. A. Mathies, “Generation of narrow-bandwidth picosecond visible pulses from broadband femtosecond pulses for femtosecond stimulated Raman,” Appl. Phys. Lett. 89(12), 121124 (2006). [CrossRef]
2. M. G. Allen, “Diode laser absorption sensors for gas-dynamic and combustion flows,” Meas. Sci. Technol. 9(4), 545–562 (1998). [CrossRef] [PubMed]
3. M. Seiter and M. W. Sigrist, “Trace-gas sensor based on mid-IR difference-frequency generation in PPLN with saturated output power,” Infrared Phys. Technol. 41(5), 259–269 (2000). [CrossRef]
4. S. M. Shepard, “Introduction to active thermography for non-destructive evaluation,” Anti-Corrosion Methods and Materials 44(4), 236–239 (1997). [CrossRef]
5. S. Kim, P. Klimecky, J. B. Jeffries, F. L. Terry Jr, and R. K. Hanson, “In situ measurements of HCl during plasma etching of poly-silicon using a diode laser absorption sensor,” Meas. Sci. Technol. 14(9), 1662–1670 (2003). [CrossRef]
6. P. F. Moulton, G. A. Rines, E. V. Slobodtchikov, K. F. Wall, G. Frith, B. Samson, and A. L. G. Carter, “Tm-Doped Fiber Lasers: Fundamentals and Power Scaling,” IEEE J. Sel. Top. Quantum Electron. 15(1), 85–92 (2009). [CrossRef]
7. M. G. Allen, “Diode laser absorption sensors for gas-dynamic and combustion flows,” Meas. Sci. Technol. 9(4), 545–562 (1998). [CrossRef] [PubMed]
8. F. J. Duarte, “Organic dye lasers: brief history and recent developments,” Opt. Photon. News 14(10), 20–25 (2003). [CrossRef]
9. C. Lindström, R. D. Burnham, D. R. Scifres, T. L. Paoli, and W. Streifer, “One watt CW visible single-quantum-well lasers,” Electron. Lett. 19(3), 80–81 (1983). [CrossRef]
10. Y. Nakayama, P. J. Pauzauskie, A. Radenovic, R. M. Onorato, R. J. Saykally, J. Liphardt, and P. Yang, “Tunable nanowire nonlinear optical probe,” Nature 447(7148), 1098–1101 (2007). [CrossRef] [PubMed]
11. B. Peng and T. Izumitani, “Optical properties, fluorescence mechanisms and energy transfer in Tm3+, Ho3+ and Tm3+ -Ho3+ doped near-infrared laser glasses, sensitized by Yb3+,” Opt. Mater. 4(6), 797–810 (1995). [CrossRef]
12. D. G. Lancaster, D. Richter, R. F. Curl, F. K. Tittel, L. Goldberg, and J. Koplow, “High-power continuous-wave mid-infrared radiation generated by difference frequency mixing of diode-laser-seeded fiber amplifiers and its application to dual-beam spectroscopy,” Opt. Lett. 24(23), 1744–1746 (1999). [CrossRef] [PubMed]
13. M. Schellhorn, M. Eichhorn, C. Kieleck, and A. Hirth, “High repetition rate mid-infrared laser source,” C. R. Phys. 8(10), 1151–1161 (2007). [CrossRef]
14. P. T. Rakich, Y. Fink, and M. Soljacić, “Efficient mid-IR spectral generation via spontaneous fifth-order cascaded-Raman amplification in silica fibers,” Opt. Lett. 33(15), 1690–1692 (2008). [CrossRef] [PubMed]
15. D. G. Lancaster, “Efficient Nd:YAG pumped mid-IR laser based on cascaded KTP and ZGP optical parametric oscillators and a ZGP parametric amplifier,” Opt. Commun. 282(2), 272–275 (2009). [CrossRef]
16. G. P. Agrawal, Nonlinear Fiber Optics (Academic Press, 1995).
17. D. Nodop, C. Jauregui, D. Schimpf, J. Limpert, and A. Tünnermann, “Efficient high-power generation of visible and mid-infrared light by degenerate four-wave-mixing in a large-mode-area photonic-crystal fiber,” Opt. Lett. 34(22), 3499–3501 (2009). [CrossRef] [PubMed]
18. L. Lavoute, J. C. Knight, P. Dupriez, and W. J. Wadsworth, “High power red and near-IR generation using four wave mixing in all integrated fibre laser systems,” Opt. Express 18(15), 16193–16205 (2010). [CrossRef] [PubMed]
19. M. Baumgartl, M. Chemnitz, C. Jauregui, T. Meyer, B. Dietzek, J. Popp, J. Limpert, and A. Tünnermann, “All-fiber laser source for CARS microscopy based on fiber optical parametric frequency conversion,” Opt. Express 20(4), 4484–4493 (2012). [CrossRef] [PubMed]
20. A. Steinmetz, F. Jansen, F. Stutzki, R. Lehneis, J. Limpert, and A. Tünnermann, “Sub-5-ps, multimegawatt peak-power pulses from a fiber-amplified and optically compressed passively Q-switched microchip laser,” Opt. Lett. 37(13), 2550–2552 (2012). [CrossRef] [PubMed]
21. 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]
