Optical parametric amplification, employing periodically poled KTiOPO4 as the gain medium, was used to amplify radiation emitted by a gain-switched laser diode. The pulses, which had durations between 20 ps and 2 ns, were amplified with up to 50 dB in a double stage set-up and reached pulse energies of 1 and 23 μJ, respectively.
© 2005 Optical Society of America
Compact and reliable sources generating microjoule pulses with durations between 10 ps and up to several nanoseconds with the possibility to choose the output wavelength would be useful for nonlinear microscopy, sensing and possibly other applications. Active or passive mode-locking of solid state lasers can provide optical pulses within this pulse length range. Increasing the pulse energy to μJ levels, however, requires substantial extension of the laser cavity length  or cavity dumping arrangements , which generally make the devices more complex and less reliable. A simpler and more compact alternative can be optical sources based on amplified gain-switched laser diodes. Gain-switched laser diodes can provide pulses from picoseconds to nanoseconds depending on the laser diode structure and the electrical feeding characteristics, which allows for adjustable pulse length generation. One important advantage of laser diodes is their flexibility in the operational wavelength. Moreover a diode’s repetition rate is only controlled by the frequency of the applied current pulses and can therefore easily be synchronized with other equipment. However, gain-switched diodes typically provide pulses with energies of tens of pJ and subsequent amplification of about 50 dB is needed to reach μJ levels. Various methods like regenerative, multipass or fiber amplifiers have been employed to boost the energy of the gain-switched laser diodes. Bado et al  used a low repetitive alexandrite regenerative amplifier to achieve 10 mJ pulses from the 1 pJ pulses emitted by a gain-switched laser diode at 755 nm. Further a Ti:sapphire regenerative amplifier operating at 10 kHz has been utilized to amplify picosecond pulses emitted from a gain-switched laser-diode around 810 nm to μJ pulses . Compared to the compactness of a gain-switched diode, regenerative amplifiers are complex setups, which employ intracavity intensity modulators and demand precise timing for the in-coupling of the seed and out-coupling of the amplified signal. Optical fiber amplifiers with high amplification ratios have been used to amplify pulse trains from gain switched laser diodes . In this case, the buildup of amplified spontaneous emission along the amplifier stages has to be dealt with by either employing optical filters, modulators or both . In any case, the usable wavelength ranges are rather limited in the bulk regenerative amplifiers as well as fiber amplifiers. Finally, tapered amplifiers can be employed to amplify gain-switched laser diodes . However, the optical isolators needed to prevent back reflections complicate the setup, the achievable amplification is relatively moderate and the acceptable peak power is limited.
An alternative to laser amplifiers is to employ optical parametric amplification (OPA), which provides a very efficient way to amplify pulsed optical signals. At the same time the heat load is substantially lowered, which potentially allows the generation of energetic pulses at high repetition rates. Additionally, since no electro-optic modulators are needed, the set-up will be simpler. By furthermore selecting quasi-phase-matched (QPM) crystals as the gain medium, any wavelength within the crystal’s transparency range can be amplified, by merely choosing the appropriate QPM period. Employing periodically poled KTiOPO4 (PPKTP) as the amplifying material in OPAs is very attractive due to its high damage threshold and its large nonlinearity, which results in substantial parametric gain even for a single pass OPA. PPKTP has been used extensively for amplification of nanosecond and femtosecond pulses [8–11].
In this work we demonstrate high-gain amplification of a gain-switched laser diode operating at 972 nm utilizing PPKTP. By employing a double stage set-up signal gains of 50 dB were achieved, which resulted in pulses with up to 20 μJ. Since the duration of the pulses emitted by the laser diode could be controlled electronically, we report here on the OPA performance with the shortest pulses, which had durations of 20 ps, as well as the longest pulses, which were 2 ns long, the length of the latter being limited by the pump pulse duration.
The experimental set-up for the two-stage OPA can be seen in Fig. 1. It consisted of a long gain-channel laser diode, which was used as the seed, a frequency-doubled, actively Q-switched Nd:YAG laser, serving as the pump and two PPKTP crystals, performing as the gain material. A pulse generator (SRS DG535) provided the synchronization signals, which controlled the repetition rate and the timing of the seed and the pump pulses. The signal from the pulse generator was used to trigger a home-built high-speed electric current switch, which could generate 1 A current pulses into a 50 Ω load. The pulse duration could be changed to any value between 1 and 6 ns. The rise and fall times were also adjustable and could be shorter than 100 ps. After proper attenuation the electrical pulses were fed into the laser diode through a bias-tee and care was taken to ensure impedance matching. The front surface of the laser diode, which had a 2 mm-long gain channel, was AR-coated, whereas the back surface was highly reflective. In CW operation the laser generated up to 250 mW of polarized radiation at 972 nm in a single-transversal mode. The threshold current for CW operation was measured to be 36.3 mA. Gain switching of the diode was achieved by using 240 mA current pulses combined with a small constant current bias (10 – 35 mA). For the shortest electrical pulses the DC bias was primarily used to offset negative-polarity ripples in the electrical signal, while for the longer pulses it could also be used to inject higher concentrations of quasi-equilibrium charge carriers, which greatly increased the damping of the relaxation oscillations at the beginning of the optical pulse, thus making the pulses much smoother. Fig. 2. shows the relaxation oscillations, which occurred when the pulses were superimposed with a constant current slightly smaller than the threshold. Since the impedance matching between the detector and the transmission line was not perfectly optimized, some artifacts with negative amplitude appeared. Depending on the exact shape of the electrical pulses, the diode emitted pulses with approximate energies between 10 and 200 pJ and durations between 20 ps and 2 ns, respectively. The repetition rate of the experiment was set to 20 Hz, which is the highest possible with this particular pump laser. By employing a different pump laser, however, this value could be increased considerably due to the good thermal properties of KTP and the possibility to operate the diode at more than 30 MHz without changing the properties of the emitted pulses. Due to the astigmatism of the seed beam, which is typical for laser diodes, appropriate aspherical focusing optics was employed to reshape the beam to a circular, collimated beam.
The gain medium of the OPA were two 7 mm long PPKTP crystals, which had a QPM period of 9.01 μm. In order to fulfill the phase matching condition in the PPKTP samples at the seed wavelength, the crystals were angle-tuned by rotation and operated in a noncollinear configuration . Both OPA stages were pumped by the same actively Q-switched, frequency-doubled Nd:YAG laser. It produced 4.5 ns long pulses at a wavelength of 532 nm and a repetition rate of 20 Hz in a non-diffraction limited beam (M2 = 7). The measured temporal jitter between the seed and the pump of 280 ps was mainly due to the timing uncertainty of the Nd:YAG laser’s Q-switch. A dichroic mirror was used to achieve spatial overlap between the seed and the pump beams, which were focused into the first crystal to radii of 40 μm and 120 μm, respectively. The tight seed focus was needed to achieve sufficient intensity for the OPA to discriminate against quantum noise. The relatively large M2 value of the pump, however, prevented proper mode-matching in the first OPA stage and should be considered for further efficiency improvements in the future. After the first crystal the signal was separated from the remainder of the pump and used to seed the second OPA. The seed and the pump beam for the second stage were both focused into the second crystal having the same radius of 120 μm and a delay line in the pump path ensured the precise timing for the second OPA.
3. Results and discussion
The temporal traces of the two seed pulse shapes used in this experiment are shown in Fig. 3. The traces were detected by employing a 2 GHz InGaAs photodiode and a 1 GHz analog bandwidth oscilloscope, which limited the shortest observable pulse length to ~ 400 ps. The shortest pulse possible to generate with this laser diode consisted of a single isolated relaxation oscillation. The spectrum of the ps seed pulse is shown in Fig. 4 together with the spectrum of the CW fluorescence, which was generated by the laser diode when operated below lasing threshold. The spectrum of the ps seed had a FWHM bandwidth of 7 nm, which should be compared to the FWHM spectral bandwidth of 40 nm for the fluorescence, and it did not contain appreciable mode structure. This indicates that the laser gain increased rapidly during the short electrical pulse and then was suddenly quenched by the relaxation oscillation. When the relaxation oscillation is of the order or shorter than the laser diode cavity roundtrip time (~ 20 ps in our case), the Fabry-Perot cavity modes will not develop in the spectrum. The longer, nanosecond optical pulse emitted by the laser diode essentially followed the shape of the electrical pulse, except for the first relaxation oscillation peak. For comparison, the spectrum in this case also contained a number of well-defined Fabry-Perot modes (see Fig. 7).
The first stage of the OPA served as a pre-amplifier. Both the ps seed and the ns seed could be amplified by approximately 40 dB, see Fig. 5. However, the onset of significant optical parametric generation (OPG), i.e. amplification of noise, limited the pump energy, which could be used in the first OPA stage. The OPG process decreases the spatial and spectral quality of the signal and should therefore be avoided. The rapid increase of the OPG can be seen in Fig. 6. Here the ratio between the OPG amplitude, which was generated without any seed present, and the OPA amplitude is plotted versus the pump energy. A factor facilitating the buildup of OPG in the first stage is the difference in the foci of the seed and the pump, since noise will be amplified in the absence of the seed. The same is true for the temporal buildup of the OPG signal in the pump pulse areas free from the seed signal. Actually, the OPG generation was suppressed over the seed pulse. This is evident when the nanosecond seed pulses were used. In Fig. 7 we show the spectrum of the nanosecond pulse-seeded OPA and the spectrum of the OPG when the nanosecond seed was blocked. The amplitude of the pure OPG signal was 5% of the OPA signal. In the seeded OPA the pump depletion decreased the pump power below the OPG threshold. As a result the OPA spectrum retained the same width as that of the seed with the longitudinal modes of the laser diode clearly visible. In the picosecond pulse seeded OPA the pump was almost completely depleted at the temporal position of the seed, as we will show below. Thus, the OPG signal, which does arise at higher pump energies, will be confined only to the pump pulse areas adjacent to the seed pulse. Thus OPG could be completely avoided in the picosecond pulse-seeded OPA by using a pump laser with shorter pulse lengths.
Despite of the pump energy limitation in the first stage, considerable gain was achieved for both seed shapes, 34 dB and 38 dB for the ps and the ns seed pulses, respectively. The purpose of the second stage was to boost the signal energy further and an additional 15 dB and 12 dB were added to the ps and the ns signal, respectively. The signal energies generated in the first stage were large enough to saturate the gain in the second stage, which explains the lower gain in the second stage.
Pumping the first stage with the maximum energy of 270 μJ and the second stage with 400 μJ the 2 ns long pulse was amplified to 23 μJ. Another 19 μJ were generated at the wavelength of the idler at 1175 nm. In the picosecond pulse-seeded OPA signal pulses with energies of 1 μJ were generated. In this case the associated idler energy was 0.8 μJ. Considering that the 4.5 ns pump pulse energy for the second OPA stage was 400 μJ and the seed pulse length is ~ 20 ps, the pump should be locally almost completely depleted by the amplification. However, it was hard to verify this experimentally due to the limited temporal resolution of the oscilloscope. The peak intensities for the ps and ns OPA signal reached 55.6 kW and 11.5 kW, respectively.
The pulse length of the picosecond signal was measured using a noncollinear second harmonic autocorrelation setup. The frequency doubling crystal was a 2.3 mm-long PPKTP sample, which had a QPM period of 6.78 μm. One of the measured traces can be seen in Fig. 8, where a Gaussian pulse with a FWHM of 25 ps has been fitted. The background level in the trace originated from the comparably long OPG pulse, which surrounds the amplified signal both temporally and spatially as mentioned above. Since the spectral bandwidths of the frequency doubling crystal were narrower than the spectral width of the OPA pulse, several autocorrelation traces were measured, using different noncollinear angles in the frequency doubling crystal in order to access different spectral regions. This approach is analogous to the one reported in Ref. 13 and employed in commercial frequency resolved optical gating devices. All measurements resulted in the same amplified signal duration of 18 ps assuming a Gaussian pulse shape. This pulse length actually confirms our previous inference from the picosecond seed spectrum in Fig. 4, that all the radiation from the relaxation oscillation is created during one single round trip through the diode. From the background level of the autocorrelation trace we estimated the amplitude of the OPG signal present around the 1 μJ OPA pulse. A simulation of an autocorrelation trace for a 18 ps Gaussian OPA pulse superimposed on a 2 ns Gaussian OPG pulse, gave an OPG/OPA amplitude ratio of 0.05.
In conclusion we demonstrated two-stage optical parametric amplification in PPKTP of a gain-switched laser diode generating pulses at 972 nm with adjustable pulse length. Signal gains of ~ 50 dB were obtained in both the picosecond and nanosecond pulse regimes, generating signal pulses with energies of up to 20 μJ and 1 μJ for the pulse length of 2 ns and 18 ps, respectively, by using nanosecond pump pulses with total energies of 670 μJ. The corresponding peak intensities for the ps and ns OPA signals were 55.6 kW and 11.5 kW, respectively. The efficiency and the amplification factor can be further improved by using a diffraction limited pump laser with pulse lengths that better match the seed duration. Due to the high nonlinearity of PPKTP the pump pulse energies required for such amplification are low enough to be generated by amplifying low-jitter microchip lasers with hybrid Q-switching  or using a pulsed diode laser-seeded fiber amplifier. Finally, it should be mentioned that the spectrum of gain-switched laser diodes can be externally controlled even in the picosecond pulse regime, as demonstrated in Ref. 15. Coupled with the wavelength agility of the OPA this opens up the prospect for compact, tunable picosecond sources with high peak powers.
We would like to acknowledge the Göran Gustafsson foundation and the Carl Trygger foundation for partial support of this work.
References and links
1. S. H. Cho, F. X. Kärtner, U. Morgner, E. P. Ippen, J. G. Fujimoto, J. E. Cunningham, and W. H. Knox, “Generation of 90-nJ pulses with a 4-MHz repetition-rate Kerr-lens mode-locked Ti:Al2O3 laser operating with net positive and negative intracavity dispersion,” Optics Lett. 26, 560–562 (2001). [CrossRef]
2. M. S. Pshenichnikov, W. P. De Boeij, and D. A. Wiersma, “Generation of 13-fs, 5-MW pulses from a cavity-dumped Ti:sapphire laser,” Optics Lett. 19, 572–574 (1994). [CrossRef]
3. P. Bado, M. Pessot, J. Squier, G. A. Mourou, and D. J. Harter, “Regenerative amplification in alexandrite of pulses from specialized oscillators,” IEEE J. Quantum Electron. 24, 1167–1171 (1988). [CrossRef]
4. F. Balembois, M. Gaignet, P. Georges, A. Brun, N. Stelmakh, and J. M. Louritoiz ,“Tunable picosecond blue and ultraviolet pulses from a diode-pumped laser system seeded by a gain-switched laser diode,” Appl. Opt. 37, 4876–4880 (1998). [CrossRef]
5. P. Dupriex, A. Piper, A. Malinowski, J. K. Sahu, M. Ibsen, Y. Jeong, L. M. B. Hickey, M. N. Zervas, J. Nilsson, and D. J. Richardson, “321 W average power, 1 GHz, 20 ps, 1060 nm pulses fiber MOPA source,” presented at OFC/NFOEC, Anaheim, USA, 7-11 March 2005.
6. C. C. Renaud, H. L. Offerhaus, J. A. Alvarez-Chavez, J. Nilsson, W. A. Clarkson, P. W. Turner, D. J. Richardson, and A. B. Grundinin, “Characteristics of Q-Switched Cladding-Pumped Ytterbium-Doped Fiber Lasers with Different High-Energy Fiber Design,” IEEE J. Quantum Electron. 37, 199–206 (2001). [CrossRef]
7. R. Häring, T. Schmitt, A. R. Bellancourt, F. Lison, K. Lauritsen, R. Erdmann, and W. Kaenders, “10 W peak power from a gain-switched picosecond all-semiconductor laser,” Proc. SPIE 5707, 302-308 (2005). [CrossRef]
8. A. Fragemann, V. Pasiskevicius, G. Karlsson, and F. Laurell, “High-peak power nanosecond optical parametric amplifier with periodically poled KTiOPO4,” Opt. Express 11, 1297–1302 (2003). [CrossRef] [PubMed]
9. F. Rotermund, V. Petrov, F. Noack, V. Pasiskevicius, J. Hellström, and F. Laurell, “Efficient femtosecond traveling-wave optical parametric amplification in periodically poled KTiOPO4,” Opt. Lett. 24, 1874–1876, (1999). [CrossRef]
10. V. Petrov, F. Noack, F. Rotermund, V. Pasiskevicius, A. Fragemann, F. Laurell, H. Hundertmark, P. Adel, and C. Fallnich, “Efficient all-diode-pumped double stage femtosecond optical parametric chirped pulse amplification at 1-kHz with periodically poled KTiOPO4,” Jpn. J. Appl. Phys. 42, L1327–L1329 (2003). [CrossRef]
11. I. Jovanovic, J. R. Schmidt, and C. A. Ebbers, “Optical parametric chirped-pulse amplification in periodically poled KTiOPO4 at 1053 nm,” Appl. Phys. Lett. 83, 4125–4127, (2003). [CrossRef]
12. A. Fragemann, V. Pasiskevicius, and F. Laurell, “Broadband Optical Parametric Amplification in the Mid-Infrared Spectral Region with Periodically Poled KTiOPO4,” presented at CLEO, Baltimore, USA, 22–27 May 2005.
13. P. O’Shea, M. Kimmel, X. Gu, and R. Trebino, “Highly simplified device for ultrashort-pulse measurement,” Optics Lett. 26, 932–934 (2001). [CrossRef]
14. B. Hansson and M. Arvidsson, “Q-switched microchip laser with 65 ps timing jitter,” Electron. Letters 36, 1123–1124 (2000). [CrossRef]
15. O. Smolski, J. Jiang, C. Roychoudhuri, E. Portnoi, G. Venus, and J. Bullington, “Tunable picosecond pulses from gain-switched grating coupled surface emitting laser,” Proc. SPIE 4652, 59–62 (2002). [CrossRef]