High brightness compact microchip-seeded MOPA system was realized. Implementing a microchip preamplifier stage acting as gain aperture element lead to excellent output beam quality with M2 = 1.4. At maximum amplification level, 235 mJ (0.4 GW) of output energy (power) was measured. Analysis of the effect of the preamplifier showed that this element increases the available beam intensity by two orders of magnitude without significant increase in system footprint. Final beam brightness was 18 PW/sr.cm2.
© 2018 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
Since the invention of the first laser in the 1960’s, pulsed systems peak power has been increasing in a tremendous way thanks to several technological breakthroughs. Only one year after first laser demonstration, Q-switched oscillator was experimentally achieved  generating pulses of about 1MW in 100 ns, opening the way to applications such as material processing or nonlinear optics . The development of Master Oscillator Power Amplifier (MOPA) scheme lead to a continuous increase of available energy, particularly with neodymium-based lasers routinely delivering Joule-level pulses and even kJ to MJ for the largest Nd:glass systems. In parallel, the development of mode-locking technique  opened the race toward ultra-short pulses generation. As pulses shorter than one picosecond became available, multi-MW peak power beams could be generated despite modest energy levels. A new breakthrough came in the 1980’s with the invention of Chirped Pulse Amplification (CPA) , allowing amplification of sub-ps and fs pulses to high energy levels, currently several Joules to few 100 J, making PW peak power systems available. All these systems are now commonly used in practical applications such as material engineering, surgery and defense, or for basic science studies such as particle acceleration, fusion science or ultrafast chemistry. These technology developments and the spreading of related applications lead to two antagonist demands: on one hand, more energy and power was required leading to larger and larger systems, on the other hand practical applications request compactness in a general trend of technology miniaturization.
On the latter aspect, the development of giant-pulse microchip lasers can be seen as another technological breakthrough in the field of pulsed lasers . Based on passive Q-switching, these laser-diode-pumped solid-state lasers are extremely compact and simple in design. They can generate multi-MW output power , with pulse duration ranging from nanosecond to a few hundreds of picoseconds, in IR, visible and UV domains  making them ideal candidates for a broad range of applications such as THz-wave generation [8,9] or laser-induced breakdown [10, 11].
Due to the aforementioned qualities, the demand for higher energy microchip lasers has been rising. Such a compact sub-nanosecond system would indeed be useful for applications such as laser peening, skin treatment or spatial engine ignition. In these cases, the required energy would be comprised between 100 mJ and 1J. In a more prospective vision, multi-J systems could be used as a very compact pump source for ultra-high intensity systems, whose current large footprint is partly due to large pumping units. However, due to the small volume of the microchip laser medium, the achievable stored energy is limited and thus the available output pulse energy hardly exceeds a few mJ. In order to reach the 100 mJ milestone, we developed a microchip-seeded MOPA system. The challenging aspect of such a development is to increase the energy while preserving good beam quality and system compactness. In order to meet these requirements, we designed a multi-stage system relying on compact elements. A microchip laser delivering 2 mJ sub-nanosecond pulses was used as a master oscillator. In order to maximize amplification and ensure optimal beam quality, this beam was then preamplified and spatially cleaned, both actions being realized in a single ultra-compact element, the microchip preamplifier. Finally, we used a compact rod-type power amplifier in double-pass configuration, this last stage providing most of the final required energy. Using this setup, we report microchip laser amplification up to 235 mJ, delivered in 600 ps at 10 Hz repetition rate. The final beam quality factor was M2 = 1.4, ensuring excellent focusing properties. The system was arranged such as fitting on a 45 cm x 30 cm optical breadboard, making it easy to displace and implemented into compact experimental setups. In what follows, we first give a detailed description of our system. We will then present experimental results and discuss the system performances and current limitations.
2. Experimental setup
The master oscillator was a passively Q-switched microchip laser. The laser medium consisted in a 1.1-at.%-doped, 5-mm-long Nd3+:YAG crystal, coated for high reflection (HR) at 1064 nm and high transmission (HT) at 808 nm on one side, while the opposite surface was coated for HT at 1064 nm and HR at 808 nm. A flat-flat mirror coated for 50% reflection at 1064 nm was used as output coupler (OC). Q-switching was triggered by a 4-mm-long Cr4+:YAG saturable absorber with initial transmission T = 30%, inserted between the laser medium and the OC. This oscillator was end-pumped by a 120 W 808 nm fiber-coupled diode laser. The pump beam at the output of the 600-μm-core fiber was focused in the laser medium by using a collimator composed of two aspheric lenses. These lenses were chosen such as producing a roughly top-hat focal spot with a 1.2 mm full width at half maximum (FWHM) located at the output plane of the Nd3+:YAG crystal. The pump beam diameter variation over the crystal length was less than 10%. The diode laser was operated at 100 Hz repetition rate, with pulse duration of 250 μs. The oscillator was maintained at a constant temperature of 25 °C by thermoelectric cooling. This arrangement can be seen on the top right of Fig. 1.
The generated 1064 nm laser beam was then transported to the preamplification stage. The micro-preamplifier was simply a 4-mm-long 1.1-at.%-doped Nd3+:YAG crystal, end-pumped at 100 Hz by diode laser identical to the one used for oscillator. In this case however, the pump beam was focused with a diameter of 0.8 mm such as matching the input laser mode size. Precise justification of this value will be given in the results section. The crystal had identical coating as the one used for oscillator, allowing double-pass pump absorption as well as double-pass beam amplification. No cooling was applied to this element. The corresponding setup is depicted on bottom right of Fig. 1.
The beam was extracted by polarizing beam splitter (PBS) after rotating the polarization by double-pass through a quarter-wave plate (QWP). An optical isolator was installed for protecting the oscillator and preamplifier from any parasitic light emitted by the main amplifier and ensuring stable operation. The preamplified beam was then expanded by a factor 1.75 using a combination of f = −100 mm concave lens and f = 175 mm convex lens.
The main amplifier consisted in a 5-mm-diameter and 126-mm-long 0.6-at.%-dopped Nd3+:YAG rod. Pumping was made by five diode laser arrays exciting the rod on its full length and symmetrically distributed around the rod, ensuring gain homogeneity. The total power delivered by diode laser arrays was 6 kW at maximum. With a pump pulse duration of 250 μs, the maximum pump energy was then 1.5 J. The diode lasers and the rod were cooled by deionized water at 26°C provided by a recirculating chiller. High reflection mirror at normal incidence was used at the rod output for double-pass amplification. QWP was inserted between the rod end the mirror in order to compensate for thermal birefringence and allowing extraction of the amplified beam by PBS after second pass. The arrangement of these elements is described in Fig. 1.
In order to ensure that pump energy is fully used in both preamplifier and main amplifier, the respective pump pulses were triggered by oscillator driver with two independent internal negative delays. These delays were chosen such as oscillator beam was injected at the very end of the pump pulse. The amplification level was found to be insensitive to delay variation in a range of ± 10 microseconds, ensuring that oscillator Q-switching time jitter (few microseconds) is not an issue.
3. Experimental results
As a first step, the output beam of the oscillator was characterized. The cavity design presented above lead to an output energy of 2 mJ, while pulse duration was 410 ps, which is consistant with previous experiments using similar setup [12, 13]. In order to increase the Q-switching threshold and then the output energy, the pump beam was focused to a rather large diameter (see previous section), with a cross-section exceeding that of the TEM00 mode. The short pulse duration results from the short length of the cavity  which was only 9 mm long. This short length reduces the diffraction losses of high divergence modes. As a consequence, both high energy and short pulse duration are obtained at the cost of higher-order modes oscillation. Experimentally, characteristic secondary ring-shaped emission was observed in the intensity profile (blue curve Fig. 2). The central lobe can be satisfactorily fitted by a Gaussian curve whose half-width at 1/e2 is defined as w0. One can then see that input beam radius at 1/e2 is about 2w0. As a result, injection into the amplifier becomes a problem. Due to finite aperture of the laser rod, a large beam size will produce Fresnel rings due to diffraction of outer lobes. On the other hand, reducing beam size such as keeping the outer lobes clear from rod edge will lead to a small central lobe diameter reducing extraction efficiency and increasing peak intensity (which can cause damage). In order to address this issue, we added a beam cleaning stage. As system compactness is a requirement in our work, space consuming systems such as spatial filtering were not adapted. Instead, we chose to improve the beam quality by gain spatial modulation through compact pre-amplifier. The beam-shaping effect of inhomogeneous gain distribution has been evidenced by previous work  on continuous wave (CW) Nd:YVO4 multi-stage MOPA system, in the continuity of gain guiding experiments initially performed in the case of fiber lasers . In this case however, due to the low saturation power of the material, gain is saturated by the input beam and the effect is then strongly dependant on the input beam profile as well as thermal gradients introduced by CW pumping over long propagation distance. In the case of Nd:YAG, saturation fluence is 667 mJ/cm2. For an input beam of 0.5 mm radius, saturation energy is about 5 mJ. It is then possible to operate in non-saturated gain regime where amplification factor does not depend on the input beam profile. The cleaning effect is entirely determined by the gain value and distribution, that is, pump energy and distribution. In this regime, we can think of it as a gain aperture.
As shown in Fig. 2, the pump beam profile in preamplifier was adjusted such as matching the desired Gaussian pulse and leave the side lobes outside the pumped area. The key parameter is then the overlap integral:Fig. 3(a) where the normalized measured profiles for different preamplifier gain values are plotted.
From recorded beam profiles, it is possible to quantify this beam quality improvement. If we name I0 the peak intensity and Iside the side lobe peak intensity, we can plot the ratio Iside/I0 as shown in Fig. 4(b). This operation was done for secondary peak on the left (green curve) and on the right (blue curve) of the main peak. For the oscillator profile, the side lobes intensity represents between 20% and 25% of the main peak intensity. We can see that even for small gain value (< 1.5) this ratio rapidly falls under 10%. In this case, the 1/e2 width is no more affected by the side lobes. For maximum gain (G ≈3), the side lobe intensity is less than 5% of the main peak which can be considered as negligible. As a further characterization of beam cleaning effect, we measured beam quality for oscillator beam and for amplified beam. In both cases, beam longitudinal profile was measured by translating a CMOS camera along the beam focused by a 300-mm-focal length concave lens. The recorded caustics where fitted by beam analysis software to evaluate the M2 beam quality factor as shown in Fig. 3(b). Due to the presence of higher order modes, the oscillator beam had strong divergence resulting in a beam quality factor of M2 = 3. For optimum pumping conditions corresponding to a gain of about 3, we can see that divergence strongly decreases and beam quality factor drops to M2 = 1.3, which makes the beam close to Gaussian quality.
The second part of the experiments consisted in studying the effect of the preamplifier stage on beam amplification by the main laser head. The amplifier was first independently characterized. Particularly, the small signal gain has been measured by amplifying a 0.1 mJ seed signal with different pumping energies. We obtained the value g0[cm−1] = 0.17Epump[J] - 0.03, which will be used for further calculations. This low energy beam was also used for characterizing the gain homogeneity across the rod, on a range covering 80% of the rod aperture. It has been found that the small signal gain inhomogeneity is about 3%.
In order to evaluate the benefits of the pre-amplifier stage, main amplifier performance was evaluated for two input beam conditions: low beam quality low energy seeder from the oscillator, and high-quality high-energy seeder from the pre-amplifier. In both cases, the amplification was measured with a pump pulse duration of 250 μs and its energy was controlled by varying the driving current. The delay between the seed pulse and the pump pulse of the amplifier was set such as the seed pulse reaches the amplifier and the end of the pump pulse, when the gain is maximum. In order to take into account losses at seeder injection in the amplifier, gain was calculated as the ratio between the output energy of the pumped amplifier to the output energy of the unpumped amplifier. By doing so, we consider only the input beam energy fraction effectively injected into the amplifier. This value was 1.7 mJ in the case of the oscillator alone and 4.5 mJ when adding the preamplifier. The measured output energy and gain as a function of the pump energy is are depicted in Fig. 5(a) with blue triangles and red squares corresponding to pre-amplifier off and on respectively.
In both cases, we can observe an exponential growth of the output energy followed by gain saturation as the output energy increases proportionally to the pump energy. However, without preamplification, this saturation occurs at pump energy higher than 1.2 J whereas turning on the preamplifier reduces this transition value to about 0.8 J. Energy extraction is thus more efficient in the latter case, which can be seen in the output energy values: without preamplifier, the maximum output energy is 180 mJ whereas pre-amplified beam allows output energy to reach 235 mJ. This corresponds to 35% increase of available output energy when using preamplifier. However, the most considerable effect can be observed on beam profile, which is shown in Figs. 5(b) and 5(c) for maximum amplification with preamplifier off and on respectively. Without preamplification, one can observe strong ring pattern on the beam, extending on a wide area and with strong intensity modulation in the central part of the beam. This chaotic profile results from diffraction of the side lobes of the input beam. As the relative intensity level of main and side lobes are comparable, strong interference is generated and amplified after propagation through the laser head. On the contrary, when amplifier is turned on, the beam shape is preserved during amplification, due to the negligible value of side lobes compared to the main peak as shown before. For maximum pumping conditions, the M2 value of the beam without preamplification was found to reach M2 = 10 whereas preamplified beam had a value of M2 = 1.4, only slightly increased compared to the input beam.
In this section, we would like to discuss the benefits of preamplification as well as the limitations of the current system. As shown by the results, preamplification leads to a double enhancement of the system output performance: higher output energy and great improvement of beam quality. Concerning the first aspect, the main reason is the higher energy injected into the amplifier. The simplest (no-loss) model to describe pulse amplification is given by  in which the single-pass amplifier gain and extraction efficiency are respectively given by:Fig. 6 for single (blue line) and double-pass (red line). For the input energy corresponding to oscillator only, the calculation gives an output energy of 190 mJ. An input energy corresponding to the preamplified beam leads to an output energy of 240 mJ. We can see that the calculated values are close to the measured ones, the overestimation of a few percents being explained by the no-loss nature of the model (especially ASE contribution is ignored). As we mentioned before, the experimental data of Fig. 5(a) shows that amplification is always saturated for maximum pumping in both cases.
In Fig. 6, we can also observe this saturation effect, this time as a function of input energy. For energies lower than 1 mJ, a slight increase of the input energy induces a strong difference on output beam. For energies higher than 2 mJ, output energy growth tends to saturate. Thus, the observed 35% enhancement of output energy when turning preamplifier on was triggered by a 164% increase of input beam energy. It is then obvious that we reached the limits of this amplifier in terms of extraction efficiency and further increase of seeder energy (for example by adding a second preamplification stage) would lead to very low benefits. However, the second aspect of preamplifier action, ie beam cleaning effect, is the main reason of system strong improvement. As it was shown in previous section, the profile of amplified beam was dramatically improved when the preamplifier was turned on. As the output beam is close to Gaussian, one can expect good focusing properties, which is a requirement of many laser users. Laser intensity in the focal spot is not a relevant number for our system, since it depends on the focusing optics. Instead, we propose to use the brightness as an indicator of the focusing quality of our laser. The general definition of an optical source brightness as a radiometric quantity is given by B = P/AΩ where P is the emitted power, A is the area of emitting surface and Ω is the solid angle of emitted radiation. As it has been shown in previous works , we can transpose this definition to a pulsed laser beam, P being the laser peak power, A the mode area and Ω the solid angle of divergence. If w is the mode radius and θ the half-angle beam divergence, we have Ω = πθ2, S = πw2 and by definition M2 = πwθ/λ where λ is the laser wavelength and brightness can then be rewritten as:Fig. 7.
Thanks to strong beam quality improvement, we can see that the brightness of the preamplified beam is increased by two orders of magnitude compared to the initial oscillator beam. Taking the maximum values, we have B = 418 TW/sr.cm2 without preamplification and B = 18 PW/sr.cm2 when using preamplified beam. From a user perspective, this means that the intensity being proportional to brightness , turning preamplifier on increases the available intensity by two orders of magnitude, for the same focusing conditions and similar beam energy. Table 1 summarizes the system performances.
The main limitation of the system is the relatively low repetition rate, which has been set to 10 Hz in the current experiments. For repetition rates higher than few tens of Hz, strong thermal lensing has been observed in the main amplifier. This thermal focusing effect is enhanced by the double-pass configuration, leading to destruction of optics after amplifier, well below maximum output energy. This thermal lensing was experimentally confirmed by recording spot size of low-energy probe beam after amplifier pumped at different repetition rates. As no significant variation could be observed for 10 Hz repetition rate, this value was chosen for safe operation of the system.
Comparing beam size after preamplifier [Fig. 3(d)] and after main amplifier for maximum gain [Fig. 5(c)] one can still observe significant size reduction, although initial beam waist was located inside the amplifier and beam divergence was expected in far field. Since thermal lens effect was measured with continuous beam, we cannot rule out a transient self-focusing effect occurring during the pump pulse. Further investigation will be made on that aspect. It is also worth noticing that the input beam power exceeds the critical power Pc for self-focusing. The beam power varies then from about 3Pc at amplifier entrance to 90Pc after double-pass amplification. Hence, accumulated phase shift along propagation distance is also a possible candidate for explaining the observed beam size reduction.
Distributed Face Cooling (DFC) structured medium that we initially developed as compact high-power oscillator  could be an interesting alternative compared to our current rod-type amplifier. As DFC substantially relax heat load in the medium , we think it would allow MOPA operation at 100 Hz repetition rate or higher. Furthermore, the shorter amplification distance and larger clear aperture (allowing larger input beam) would also reduce the effect of self-focusing. Finally, the footprint of such an amplifier would be strongly reduced compared to the rod type laser head, which is the most space consuming element in our current system.
High brightness compact MOPA system has been developed. For ensuring a reduced footprint as well as excellent beam quality, we implemented a microchip preamplifier module acting simultaneously as a gain medium and a spatial cleaning device. Combined with a compact rod-type power amplifier, 235 mJ beam could be obtained, delivered in 600 ps corresponding to a peak power of 0.4 GW. The initial microchip peak power of 4.6 MW was then multiplied 74. The dramatic improvement of beam quality due to preamplifier made the initial microchip brightness of 51 TW/sr.cm2 rise up to 18 PW/sr.cm2 after amplification, meaning 350 times brighter beam. Such a tool could satisfy the demand of energetic high quality sub-nanosecond pulses with reasonable size system. Using these results as a basis, we aim at further size reduction and reaching the Joule level in near future.
ImPACT Program of Council for Science, Technology and Innovation (Cabinet Office, Government of Japan).
References and links
1. F. J. McClung and R. W. Hellwarth, “Giant Optical Pulsations from Ruby,” Appl. Opt. 1(S1), 103–105 (1962). [CrossRef]
2. P. A. Franken, A. E. Hill, C. W. Peters, and G. Weinreich, “Generation of Optical Harmonics,” Phys. Rev. Lett. 7(4), 118–119 (1961). [CrossRef]
3. L. E. Hargrove, R. L. Fork, and M. A. Pollack, “Locking of He-Ne laser modes by synchronous intracavity modulation,” Appl. Phys. Lett. 5(1), 4–5 (1964). [CrossRef]
4. D. Strickland and G. Mourou, “Compression of amplified chirped optical pulses,” Opt. Commun. 56(3), 219–221 (1985). [CrossRef]
5. T. Taira, Y. Matsuoka, H. Sakai, A. Sone, and H. Kan, “Passively Q-switched Nd:YAG microchip laser over 1-MW peak output power for micro drilling,” Conference on Lasers and Electro-Optics CLEO 2006, Long Beach, CA, USA, May 21–26, CWF6 (2006).
7. R. Bhandari, N. Tsuji, T. Suzuki, M. Nishifuji, and T. Taira, “Efficient second to ninth harmonic generation using megawatt peak power microchip laser,” Opt. Express 21(23), 28849–28855 (2013). [CrossRef] [PubMed]
8. S. Hayashi, T. Shibuya, H. Sakai, T. Taira, C. Otani, Y. Ogawa, and K. Kawase, “Tunability enhancement of a terahertz-wave parametric generator pumped by a microchip Nd:YAG laser,” Appl. Opt. 48(15), 2899–2902 (2009). [CrossRef] [PubMed]
9. K. Nawata, S. Hayashi, H. Ishizuki, K. Murate, K. Imayama, Y. Takida, V. Yahia, T. Taira, K. Kawase, and H. Minamide, “Effective Terahertz Wave Parametric Generation Depending on the Pump Pulse Width Using a LiNbO3 Crystal,” IEEE Trans. Terahertz Sci. Technol. 7(5), 617–620 (2017). [CrossRef]
10. M. Tsunekane, T. Inohara, A. Ando, N. Kido, K. Kanehara, and T. Taira, “High Peak Power, Passively Q-switched Microlaser for Ignition of Engines,” IEEE J. Quantum Electron. 46(2), 277–284 (2010). [CrossRef]
11. H. H. Lim and T. Taira, “Sub-nanosecond laser induced air-breakdown with giant-pulse duration tuned Nd:YAG ceramic micro-laser by cavity-length control,” Opt. Express 25(6), 6302–6310 (2017). [CrossRef] [PubMed]
13. R. Bhandari, T. Taira, A. Miyamoto, Y. Furukawa, and T. Tago, “> 3 MW peak power at 266 nm using Nd:YAG/ Cr4+:YAG microchip laser and fluxless-BBO,” Opt. Mater. Express 2(7), 907–913 (2012). [CrossRef]
14. X. Yan, Q. Liu, X. Fu, D. Wang, and M. Gong, “Gain guiding effect in end-pumped Nd:YVO4 MOPA lasers,” J. Opt. Soc. Am. B 27(6), 1286–1290 (2010). [CrossRef]
15. A. E. Siegman, Y. Chen, V. Sudesh, M. C. Richardson, M. Bass, P. Foy, W. Hawkins, and J. Ballato, “Confined propagation and near single-mode laser oscillation in a gain-guided, index antiguided optical fiber,” Appl. Phys. Lett. 89(25), 251101 (2006). [CrossRef]
16. L. M. Frantz and J. S. Nodvik, “Theory of Pulse Propagation in a Laser Amplifier,” J. Appl. Phys. 34(8), 2346–2349 (1963). [CrossRef]
17. P. Shukla, J. Lawrence, and Y. Zhang, “Understanding laser beam brightness: A review and new prospective in material processing,” Opt. Laser Technol. 75, 40–51 (2015). [CrossRef]
18. A. Kausas, L. Zheng, and T. Taira, “Structured laser gain-medium by new bonding for power micro-laser,” Proc. SPIE 10082, 100820Z (2017). [CrossRef]
19. L. Zheng, A. Kausas, and T. Taira, “Drastic thermal effects reduction through distributed face cooling in a high power giant-pulse tiny laser,” Opt. Mater. Express 7(9), 3214–3221 (2017). [CrossRef]