1. Introduction

Ultra-high intensity laser-matter interaction offers numerous applications like particle acceleration, X-ray generation and inertial confinement fusion [1]. Proof-of-principle experiments for these applications have been carried out using flashlamp-pumped laser amplifiers. However, bringing these applications to practical use will demand lasers capable of delivering pulse energies up to kJ-levels at multi-Hz repetition rates with high wall-plug efficiency. Flashlamp-pumped laser systems are limited to low repetition rate operation because of inherent thermal issues coupled with poor electrical-to-optical efficiency. Diode pumped solid state laser (DPSSL) systems, using advanced gain media and cooling schemes, offer a promising approach to overcome the limitations imposed by flashlamp-pumped technology. Owing to their excellent electrical-to-optical efficiency, spectral overlap with the absorption lines in the gain medium, and high-brightness emission, laser diodes are ideal pump sources for achieving high overall efficiency, low thermal load and hence high pulse repetition rates. The choice of gain medium, amplifier geometry, thermal management, and extraction architecture are important aspects for development of a scalable high energy, high repetition DPSSL amplifier capable of delivering kJ-level energies. To date, results from various amplifier designs exploring different materials and geometries to amplify nanosecond pulses have been reported. The LUCIA project has recently reported 14 J pulses at 2 Hz with 13% efficiency based on an active mirror geometry using ceramic Yb:YAG [2]. The Mercury project implemented a room temperature gas cooled multi-slab geometry using Yb:S-FAP as a gain medium and was able to generate 550 W of average power (55 J at 10 Hz) with an efficiency of 7.6% [3]. The HALNA project utilised Nd:phosphate glass in a zig-zag optical architecture to generate 21.3 J at 10 Hz with 11.7% efficiency [4]. Recently, Körner et.al. reported 35% optical-to-optical efficiency for cryogenic cooled 1.1 J Yb:YAG amplifier at 1 Hz operation [5]. In previous publications about DiPOLE, we have presented a conceptual design for an energy scalable diode pumped, cryogenic gas cooled, multi-slab Yb:YAG amplifier [6,7] designed for amplifying nanosecond pulses to kilojoule energy levels, along with initial results for a prototype 10 J amplifier [8]. In this paper, we present development of the front-end capable of generating spectrally and temporally controlled pulses and a novel multi-pass architecture capable of supporting six passes, thus enabling operation of the cryogenic gas cooled multi-slab amplifier DiPOLE at 10 J and 10 Hz. We also present spatial and temporal characteristics and long-term shot-to-shot energy stability, validating the concept of a scalable multi-slab cryogenic gas cooled amplifier.

2. DiPOLE amplifier setup

Figure 1 shows a schematic diagram of the DiPOLE prototype laser system. Spectrally and temporally controlled pulses are generated in a fibre-based frontend system (IDIL Fibres Optiques, Lannion, France). This consists of a Yb-doped CW fibre oscillator delivering 15 mW narrow linewidth (70 kHz) output, tuneable from 1028.5 nm to 1031.2 nm (measured in air). Two phase modulators operating at RF frequencies of 2 GHz and 14 GHz provide optional spectral broadening of the output. Spectral broadening was not applied for the experiments described in this paper. An acousto-optic modulator (AOM) chops the CW output into 100 ns flat-top pulses at a repetition rate of 10 kHz. The chopped pulses are further amplified to 2 W peak power in a multi-stage fibre amplifier. Finally, an electro-optic modulator (EOM) controlled by an arbitrary waveform generator (AWG, Kentech Instruments, Wallingford, UK) shapes the output pulses to the desired temporal profile, varying from 2 ns to 10 ns in duration with a peak output power of 500 mW. The output from the fibre-frontend seeds a regenerative amplifier (HZDR Innovations, Rossendorf, Germany), which consists of a 1.5 mm thick, 10% doped Yb:YAG crystal in an active-mirror geometry, pumped by a fibre-coupled 940 nm laser diode operating at 1.8 ms pulse duration. The regenerative amplifier (RGA) cavity is designed to accommodate and amplify pulses up to 15 ns in duration. In its nominal operating regime the RGA amplifies pulses of duration 10 ns to an energy of ~1 mJ at a repetition rate of 10 Hz, with 12 round-trips. The Gaussian output beam from the RGA is expanded to 3 mm in diameter before further amplification in a thin-disk Yb:YAG multi-pass booster amplifier stage (HZDR Innovations, Rossendorf, Germany). This consists of a 2 mm thick, 2.5 at.% doped Yb:YAG crystal, arranged in an active mirror configuration, which is pumped by a 940 nm, 2 kW peak power, diode stack with a pulse duration of 1.2 ms. An image-relaying multi-pass architecture is used to double-pass the gain medium seven times [9]. The booster amplifier delivers up to 100 mJ pulses at 10 Hz with an M² value of 1.3. The 3 mm diameter booster output beam is then expanded by a x3 magnifying Galilean telescope. A 15 mm diameter Faraday isolator and Pockels cell (PC) are installed after the beam expander to protect the booster amplifier against feedback from the main cryogenic power amplifier and to improve the intensity contrast of the laser pulses. The beam is then expanded by a Galilean telescope by a factor of 8 to a diameter of 96 mm and spatially shaped, from circular Gaussian to square flat-top, using a 20 mm square serrated aperture and a vacuum spatial filter (VSF). This shaping process reduces the input energy to the main amplifier to 25 mJ.

 

Fig. 1 Schematic diagram of DiPOLE amplifier. E1, E2: Beam Expanders; VSF#1, VSF#2, VSF#3: Vacuum spatial filters; M2, M3, M4, M5, M6, M7, M8: monolithic flat mirrors; DM1, DM2: dichroic mirrors; LD-D1, LD-D2: pump sources; MA1, MA2, MA3 and MA4: mirror arrays, also shown are the mirror positions and passes for MA1 and MA4.

Download Full Size | PDF

The main DiPOLE cryogenic amplifier contains four ceramic YAG discs (Konoshima Chemical, Takuma, Japan), each with a diameter of 55 mm and a thickness of 5 mm. The discs consist of a 45 mm diameter Yb-doped inner region that is surrounded by a 5 mm wide, 0.25% Cr⁴⁺-doped absorptive cladding to minimise transverse amplified spontaneous emission (ASE) loss and prevent parasitic oscillations at high gain. The inner two discs have a higher Yb doping (2.0 at.%) than the outer two discs (1.1 at.%). Disc thickness and doping levels were chosen to equalise gain coefficient and heat load in the discs and to minimise ASE losses, and also to maximise optical efficiency by finding the best compromise between pump absorption and re-absorption losses due to the quasi 3-level nature of Yb:YAG [6].

The discs are held in aerodynamically shaped vanes and arranged in a stack with 1.5 mm gaps in-between discs. Helium gas at cryogenic temperature was forced through the gaps at a typical volume flow rate (VFR) of 35 m³/h and pressure of 10 bar. The helium gas was cooled by passing it through a liquid nitrogen heat exchanger and circulated by a cryogenic fan. The amplifier was pumped from both sides by two 940 nm diode laser sources [10] each delivering 20 kW peak power with variable pulse duration up to 1.2 ms, and repetition rate up to 10 Hz. The emission spectrum of the diode sources was less than 6 nm (FWHM) wide. The pump sources produced a 20 x 20 mm² square, flat-top beam profile at their image plane, which was aligned centrally in the amplifier head.

To enable efficient extraction of the stored energy in the main amplifier, a novel multi-pass architecture is employed. This incorporates a pair of 4f image-relaying Keplerian vacuum telescopes per pass, one on each side of the amplifier head. Plano-convex fused silica singlet lenses with effective focal length of 900 mm are used in each 1:1 image-relaying telescope, giving an effective F-number of 45, minimising aberrations and ensuring beam quality is maintained. The beams pass through the centre of the cryogenic amplifier at a maximum off-axis angle of 5 degrees in horizontal and vertical axes. The combination of mirror and mirror array M6/MA2 and M7/MA3 fold the beams towards the VSFs and propagate them parallel to the VSFs axes. To filter high spatial frequencies and prevent propagation of stray light, flat tantalum plate pinholes are installed at the focal plane of alternate telescopes. Each pinhole plate has a circular aperture of 2 mm in diameter; this is about 30 times the diffraction-limit of the beam. At the other side of the VSFs, the M8/MA4 and M5/MA1 mirror combination returns the beam on a new path before being re-injected into the amplifier at a different off-axis angle for the next pass. In order to minimise rotation of the beam caused by out-of-plane reflections, which would otherwise reduce pump overlap and extraction efficiency, out-of-plane direction changes are arranged to occur only in the M8/MA4 beam path where the greater mirror to mirror-array separation ensures a small angular change. As the extraction beam is angularly multiplexed through the amplifier head, the overlap with the pump beam reduces the amplified beam size to 18 x 18 mm². In the current DiPOLE configuration the seed beam passes six times through the main amplifier head.

3. Experimental results and discussion

As reported in [8], the energy scaled-down prototype DiPOLE amplifier acts as a variable gain amplifier depending upon the operational temperature. A maximum small-signal single-pass gain of 11 was recorded for 1.2 ms pump duration at 88 K. However, undesired effects such as enhanced ASE losses as well as initiation of longitudinal self-lasing occur at lower temperatures. Small-signal gain and ASE losses for the DiPOLE amplifier head are detailed in [8]. To achieve a trade-off between amplifier gain, ASE losses, and to avoid self-lasing, the amplifier was operated at 140 K. Initially, the amplifier was pumped for 1 ms duration (T_p) equal to the fluorescence life time (τ_p) of Yb:YAG. This condition (T_p = τ_p) offers a good compromise between the fluorescence efficiency (η_fl) and stored energy (E_st), where both are at 63% of the maximum possible value [6]. A total of 94.8% of the pump energy was absorbed in the amplifier head. To ensure efficient and stable heat transfer from the gain medium to the circulating helium gas turbulent flow was maintained across the surface of the YAG disc by operating at a VFR of 35 m³/hr at 10 bar pressure. More details on thermal management aspects can be found in [11].

To assess the level of thermally-induced wavefront distortion within the DiPOLE amplifier head theoretical single-pass wavefront predictions were compared with experimentally measured data. The predicted single-pass wavefront distortion through the amplifier was calculated from temperature maps, derived from a Conjugate Heat Transfer (CHT) model, and converted to equivalent optical path difference (OPD) using an appropriate value for the thermal coefficient of optical path length γ(T) given by Aggarwal et al [12]. Details of the CHT model are reported in [11]. Experimentally, the single-pass transmitted wave front was measured by propagating a collimated 1030 nm cw probe laser beam through the amplifier, initially with the pumps switched off. The measured wavefront was then subtracted from that measured with the pump sources switched on to assess the level of thermally-induced aberrations. The measurement was made 6 ms after the pump pulse, so that the probe laser did not experience any amplification, but the cumulative effect of thermal distortions, which have a long time constant, could still be measured as a change in wavefront. The probe laser was expanded and spatially shaped to fill the 20 mm x 20 mm active area of the amplifier and after passing through the amplifier was down collimated by a x5 de-magnifying telescope to fit onto a Shack-Hartmann wavefront sensor (Imagine Optic HASO3-First). Figure 2(a) shows the theoretically calculated single-pass wavefront map predicting a peak-to-valley (P-V) wavefront distortion of 0.13 µm. However, experimentally measured wavefront distortion was found to be 0.96 µm P-V as shown in Fig. 2(b), indicating a significantly higher level of thermally-induced distortion within the setup. The main reason for this discrepancy was traced to the choice of UV grade fused silica (FS) used for the windows in the amplifier vacuum vessel.

 

Fig. 2 Single-pass, thermally-induced transmitted wavefront distortion through the amplifier: 2(a) theoretical prediction; 2(b) measurement with high-OH FS vacuum windows; 2(c) measurement with low-OH vacuum windows.

Download Full Size | PDF

In this experiment, HPFS Standard Grade Corning 7980 synthetic amorphous fused silica was used. Despite its high UV transmission, this grade of fused silica has a significant OH content that exhibits an absorption band centred at 945 nm [13]. For a window thickness of 15 mm, this gave rise to approximately 0.3% absorption of pump radiation, corresponding to an absorbed power of ~0.6 W per window. Although the level of absorbed power is low, the poor thermal conductivity of fused silica and the absence of convection loss on the inside surface of the vacuum window resulted in a ~10 K rise in temperature at its centre, measured using an infrared camera (FLIR T400). Figure 2(c) shows the single-pass transmitted wavefront distortion measured after replacing the Corning 7980 FS windows with Heraeus Suprasil 3002 synthetic fused silica [14], which has an OH content of < 1 ppm. The experimentally measured P-V after replacement of the windows was 0.21 µm, close to the theoretically calculated value of 0.13 µm for single-pass wavefront distortion. This highlights the importance of correctly specifying passive optics when designing high average power laser systems.

After wavefront assessment, a pulsed beam from the front-end was injected into the main amplifier and synchronised to arrive at the trailing edge of the pump pulse so as to experience maximum gain. The wavelength was tuned at 1029.5 nm as this corresponds to the gain peak at 140K [8]. The beam was propagated through the amplifier six times, and an uncoated FS wedge was used to reflect part of the final output on to an energy meter (Gentec QE50S). A 25 cm long water-filled container acted as a beam dump for the experiment. The dependence of output energy on input energy for pump pulse duration of 1 ms at 10 Hz was then determined by increasing seed energy from 5 mJ to 25 mJ, the results of which are shown in Fig. 3. These are compared with theoretical predictions from a numerical model described previously [6]. Measured results agree well with the theoretical prediction and in both cases gain saturation within the amplifier occurs at around 9.5 J for a seed input of ~16 mJ.

 

Fig. 3 Dependence of the output energy on the seed input energy for 1 ms pumping at 10 Hz. The dotted line shows the theoretically predicted output of the amplifier.

Download Full Size | PDF

The spectral width of the output could not be accurately measured as it was below the resolution limit of the instruments at our disposal. The expected linewidth of the cw fibre oscillator is sub-MHz, whereas the transform limited bandwidth of a 10 ns pulse is several 10s of MHz. The temperature-dependent gain spectrum of cryo-cooled Yb:YAG is discussed in [8].

To maximise output energy the pump pulse was increased to 1.2 ms duration, the longest duration the diode pump sources are capable of, equivalent to total pump energy of 48 J. At 10 Hz, the output energy was 10.8 J at 1029.5 nm, corresponding to an optical-to-optical (O-to-O) conversion efficiency of 22.5%. Figure 4 shows the output energy from the amplifier as a function of the pump input, also shown is the corresponding optical-to-optical (O-to-O) efficiency of the system. The insert in Fig. 4 compares predicted theoretical performance for different numbers of passes based on a numerical model prediction [6] with the experimentally determined value. It should be noted that during this experiment, the actual pump duration was fixed at 1.2 ms to maintain a fixed heat load in the amplifier. The effective pump input energy was changed by altering the delay of the seed pulse relative to the pump pulse, the corresponding effective pump pulse duration is shown on the secondary x-axis. The ASE energy contrast (the ratio energy contained in the ASE pedestal to the total energy) was measured by blocking the seed and measuring the energy at the output of the amplifier. The energy measured was 2.5 mJ and hence the ASE energy contrast was 4000:1.

 

Fig. 4 DiPOLE laser performance at 10 Hz, 140 K (green curve), also shown is the optical-to-optical efficiency of the system (red curve). The insert shows the theoretical evolution of the output with reference to the number of passes for pump duration of 1.2 ms and seed input of 16 mJ.

Download Full Size | PDF

The ability to control the output temporal pulse profile is an important requirement from an application perspective. As mentioned in Section 2, the temporal profile of the seed is shaped via an EOM, and controlled by an AWG to a resolution of better than 200 ps. A square flat-top 10 ns temporal pulse profile was chosen for high energy operation, in order to achieve optimum efficiency for subsequent frequency conversion experiments, results from which will be reported separately. After several iterations, a near-flat-top profile was generated at the output, as shown in Fig. 5(a), also shown are temporal profiles at the output of the two preceding amplifier stages. In another experiment, the seed pulse was reduced to 2 ns and the output energy was limited to 4.5 J, thus minimising the risk of optical damage to critical components. Figure 5(b) shows the corresponding temporal profiles at different amplification stages for 2 ns pulse operation. It should be noted that for shorter pulse duration the temporal profiles of the booster and the main amplifier are similar, indicating that the amplifier is not fully saturated and more energy can still be extracted at 2 ns pulse duration. Other temporal profiles, such as an exponentially ramping profile, which is relevant for shock propagation studies [15], can also be generated by adjusting the temporal profile of the fibre-frontend.

 

Fig. 5 5(a) Temporal profiles for 10ns operation at different amplification stages, (1) is at the output of the regenerative amplifier, (2) is at the output of the booster amplifier and (3) is at the output of the main amplifier. 5(b) Shows temporal profiles of the amplifier for 2 ns pulse operation at different amplification stages.

Download Full Size | PDF

Figure 6 shows the far-field and near-field profiles measured at 10 Hz, 10.8 J operation. The far-field profile shown in Fig. 6(a) was compared to the predicted diffraction limited spot-size to obtain an estimation of output beam quality. A diffraction limited beam is defined as a beam having a perfect flat phase or wavefront which will lead to the minimum spot size and maximum peak intensity when focused, deviation from a flat wavefront will degrade the focal spot profile. A diffraction limited square shaped beam will show the same intensity distribution as that of a diffraction pattern generated by a square aperture of the same width, with a sinc² shaped far-field distribution whose FWHM width is given by θ = 0.886 λ/d. Thus for a diffraction limited square flat-top beam with d = 20 mm and λ = 1.03 µm the angular spread at FWHM will be 45.6 µrad. To estimate the angular spread at FWHM of the far-field profile shown in the Fig. 6(a), the camera was initially calibrated by collecting the diffraction pattern produced by a transmission grating with a 6 mm grating period, for which the angular spacing between adjacent diffraction orders is 171.6 µrad. A comparison between this value and the diffraction pattern collected by the camera gave an angular resolution of 4.9 µrad/pixel, thus allowing an estimation of the value of the angular spread at FWHM of 91.5 µrad and 86.6 µrad in x and y-axes, respectively from Fig. 6(a). Comparing these values with the diffraction limited beam shows that the far-field spot is approximately 2 and 1.9 times diffraction limited in x and y-axes, respectively. The near-field inhomogeneity shown in Fig. 6(b) can be attributed to the diffraction patterns formed due to the neutral density filters used in the diagnostics as well as inhomogeneity in the gain medium which is a subject of ongoing investigation.

 

Fig. 6 6(a) shows the far-field profile at 10 Hz, 10.8 J operation and 6(b) is the near-field profile of DiPOLE output at 10 Hz, 10.8 J operation, also shown is the cross-section profile in X-axis and Y-axis.

Download Full Size | PDF

To validate the long-term operation of the DiPOLE amplifier at the design baseline fluence for a 100 J upgrade, the 10 J laser was operated over 50 hours (achieved in runs extending from 4 to 6 hours) at 2 J/cm² fluence and 10 Hz, corresponding to an output energy of 7 J. Figure 7 shows the output energy of the system for 1.8 million shots with 0.85% rms overall energy stability. No damage or beam deterioration was observed during long-term operation of the DiPOLE amplifier. Note that during this experiment there was a single drop-out (close to 30h) in almost 2 million shots, this was traced back to a reduced booster output possibly due to an air-breakdown in the booster multi-pass.

 

Fig. 7 Output energy during long-term operation of laser. Data was collected in runs lasting up to 6 h, vertical lines denote start and finish of individual runs.

Download Full Size | PDF

Figure 8 shows a long-term (4.5 h) beam pointing stability measurement, again at 10 Hz, 7 J operation. The rms values for long term pointing stability were 17µrad and 13µrad, the range (peak-to-peak) were 114 µrad and 105 µrad for x-axis and y-axis, respectively.

 

Fig. 8 Beam pointing stability at 10 Hz, 7 J. operation.

Download Full Size | PDF

4. Conclusion

We report on the results obtained from a cryogenic gas cooled, multi-slab ceramic Yb:YAG amplifier designed for operation at 10 Hz repetition rate. At 140 K, the amplifier generated 10.8 J of energy in a 10 ns pulse at 1029.5 nm for a 48 J, 940 nm pump, corresponding to an optical-to-optical conversion efficiency of 22.5%. We also highlight the importance of correctly specifying passive optics to minimise thermally-induced wavefront distortion in high energy, high repetition rate systems. By comparing experimentally measured values of the far-field angular spread with calculated values for a diffraction limited beam the output beam is estimated to be approximately 2 and 1.9 times diffraction limited in x and y-axes, respectively. Finally, we have demonstrated long-term energy stability of 0.85% rms for the laser operating at 7 J, 10 Hz for over 1.8 million shots.

This gives confidence in the design of a 100 J system (DiPOLE100), currently being built at the Central Laser Facility based on the same cryogenic gas cooled, multi-slab Yb:YAG technology. This system, which is developed for and in collaboration with the HiLASE project [16], exploits the scalability of the DiPOLE architecture to higher pulse energies. It will deliver 100 J in a single beam and consists of a front-end system similar to the one described here, seeding a scaled-up amplifier, which boosts the energy by a factor of 10 and operates at the same repetition rate, pump and extraction fluence [17].