1. Introduction

Pulsed laser systems producing pulses with high energies at high repetition rates are an essential tool both in science and for industrial applications. Because the scaling of these lasers towards higher pulse energies is often limited by the increasing average power, operating it in a burst mode may be an attractive alternative in some special cases. In such a scheme pulses are generated at a very high repetition rate within a suitable short time window, followed by thermal relaxation in an idle period, before the next group of pulses – i.e. the next burst – is launched.

These lasers perfectly fit to applications in aerodynamics and combustion science, such as the investigation of fast processes like high speed flows, using planar Mie scattering, particle image velocimetry (PIV), planar Doppler velocimetry (PDV), planar laser induced fluorescence, or coherent anti-Stokes Raman analysis [1–3]. Applications of burst-mode lasers in materials processing offer new possibilities. The unique time structure of a burst containing ultra-short pulses allows to reach higher ablation ratios due to the high peak power in combination with the high average power during a burst [4–6]. A third kind of potential applications is connected with the match of the laser’s temporal structure to the operation mode of particle accelerators. Hence, burst-mode lasers are predestined to be used in photo injectors or as an interaction laser source [7]. Bursts with single-pulse durations in the fs range and single-pulse energies in excess of 10 or even 100 mJ are often demanded by these applications.

During the past decade, several systems aiming at the generation of high energy pulse bursts based on Nd³⁺-doped materials have been demonstrated [2,3,8,9]. In these systems, the power amplifiers are based on flash lamp pump technology, significantly limiting the efficiency and the repetition rate. Other approaches based on diode pumped fiber lasers were presented [10,11]. They allow the generation of high repetition rate bursts containing ultra-short pulses. Nevertheless, there are limitations in average power during a burst owing to mode instabilities [10]. Inherently, pulse energy is also limited with the fiber approach. Recently higher energies of more than 44 mJ were achieved by using diode pumped Yb:YAG thin disk amplifiers [12].

In this work we present a novel amplifier scheme for the generation of bursts from a laser system using Yb:CaF₂ as the active medium. Compared to Nd³⁺ based materials, Yb³⁺ doped media offer a lower quantum defect and a higher fluorescence lifetime in many host materials. These advantages come at the cost of comparably low emission and absorption cross sections and a quasi-three level behavior at room temperature. Therefore the active medium in our amplifiers is cooled to 100 K, depopulating the lower laser levels and hence allowing a four-level scheme. The operation at cryogenic temperatures results further in higher peak cross sections and advantages in mechanical properties [13]. In contrast to many Yb³⁺-doped crystals, Yb:CaF₂ shows no significant reduction in amplification bandwidth at cryogenic temperatures [14]. This is due to the inhomogeneous broadening caused by clustering of the Yb³⁺-ions in the crystal lattice [15].

The use of Yb:CaF₂ in a burst-mode scheme offers extraction at higher fluences since the total extraction fluence is distributed over several pulses all arriving during the upper laser level lifetime of 1.9 ms [16]. As a consequence, laser induced damage is no longer the limiting factor for extraction efficiency as in single pulse nanosecond amplifiers [17].

The system described in this paper utilizes a combination of Yb:CaF₂ as the active medium, cryogenic cooling, and the burst mode operation scheme to efficiently generate trains of high peak power pulses. The results presented here have been measured with the first amplifier of the two stage amplifier system which is currently under construction. This system is designed to deliver bursts with more than 5 J output energy and single pulse energies of more than 100 mJ [18]. The successful demonstration of the described amplification scheme represents an important milestone proving the huge potential of this approach.

2. Setup

The seed oscillator of the laser system (Satsuma, Amplitude Systems) generates pulses with 350 fs duration and 4.1 µJ energy at a repetition rate of 1 MHz. From this quasi-continuous pulse train bursts are picked out with a Pockels cell, equipped with a driver capable of MHz switching (Bergmann Messgeräte Entwicklung KG). Herewith it is also possible to reduce the internal repetition rate of the burst, because every single pulse can be switched individually. A burst may contain more than 2000 pulses. The individual pulses are further stretched to a duration of 50 ps in an Oeffner type stretcher [19], after which individual pulses have an energy of about 3 µJ.

The setup of the subsequent amplifier, as shown in Fig. 1, is based on a relay imaging design. The image plane is situated on the dichroic mirror (DM) close to the active medium. This plane is imaged onto the spherical mirror SM2 (focal length f = 625 mm) by the 3”-diameter spherical mirror SM1 (f = 1250 mm). From SM2 the beam is imaged back into the first image plane. Within this system each laser pulse performs 5 round trips corresponding to 10 amplification passes before it is coupled out. Due to the off-center position of the input beam, input and output are spatially separated. Afterwards, the beam is coupled once more into the system for 5 more roundtrips by another imaging telescope with higher focal length formed by SM3 (f = 1500 mm) and SM4 (f = 750 mm). The outer imaging system introduces a displacement of the beam which again allows to separate the output from the input.

 

Fig. 1 20-pass amplifier setup: DM dichroic mirror; L1-L3 lenses; M1-M4 turning mirrors; SM1 spherical mirror f = 1250 mm; SM2 spherical mirror f = 625 mm; SM3 spherical mirror f = 1500 mm; SM4 spherical mirror f = 1500 mm.

Download Full Size | PDF

The active medium is a Yb:CaF₂ crystal with 13 mm thickness and 10 mm square aperture and a doping level of 2.3 at. % (Hellma Materials GmbH). The crystal is cooled to 100 K using a cryostat (Janis Research Company, LLC 500UC), connected to a high vacuum chamber allowing to achieve pressures on the order of 10⁻⁷ mbar. This is necessary to avoid condensations on the crystal surface. The dichroic mirror directly behind the laser material is also placed in vacuum. It separates the extracting laser beam from the pump beam.

As the pump source, a 2.5 kW laser diode module (Jenoptik AG) was used. It contains 25 diode bars together with a beam compaction optic in a sealed housing. The diode driver (Lastronics GmbH) produced temporally rectangular pulses with rise and fall times of less than 100 µs. The pump beam was focused into the laser medium using lenses L1-L3. Its position can be adjusted by Mirror M4. At the excitation wavelength of 940 nm about 70% of the incident pump power was absorbed.

3. Results

In a first test, the amplifier was seeded with a tunable continuous wave (CW) oscillator with 200 mW output power to investigate the broadband amplification capability of the cryogenically cooled Yb:CaF₂. To avoid gain saturation the number of roundtrips was reduced to eight. The gain was determined by measuring the time dependent signal from a photo diode and comparing the peak output with the background from the transmitted and hence unamplified beam. This method additionally accounts for losses in the amplifier setup. The residual absorption of the material was calculated from the cross sections given in [20] and taken into account for the signal height of the transmitted beam.

The wavelength dependent gain for a pump duration of 2 ms at maximum pump power is shown in Fig. 2. Gain was observed for seed wavelengths between 1017 nm and 1067 nm. The highest amplification corresponding to a single-pass gain of 1.6 was achieved in a plateau around 1030 nm. Towards shorter wavelengths the measurement was limited by the tuning range of the oscillator. The measured values are well described by a simplified model using the calculation of the small signal gain for a uniformly excited medium. This is calculated from the absorption and emission cross sections σ_a and σ_e with a constant inversion β:

 

Fig. 2 Measured small signal gain in single pass as function of wavelength. The dashed line represents the calculated gain for the amplifier in case of 21% inversion based on cross sections given in [20]. Measurement done at 1 Hz.

Download Full Size | PDF

Here, N_dop is the doping concentration and d the thickness of the material. The best match of measurement and model is achieved for β = 0.21 as shown by the dashed line in Fig. 2.

To generate high energy pulses with short duration, the seed from the fs-frontend was used. In Fig. 3 the results for amplification of 1000, 1500 and 2000 pulses within one burst with an internal repetition rate of 1 MHz are shown. The pump pulse duration covers a period τ₀ of 600 µs prior to the first pulse and the burst duration. The time τ₀ was adjusted to achieve constant gain at maximum output energy. With an increasing number of pulses the available output energy was increased from 617 mJ for 1000 pulses to 915 mJ for 1500 pulses. In the case of 2000 pulses the output energy was limited by the damage threshold of mirror M2, i.e. we were not able to operate the amplifier at the maximum diode current in this configuration. As a consequence, the highest efficiency, defined as the ratio of extracted energy to absorbed energy, was achieved for 1500 pulses. It amounts to 25%.

 

Fig. 3 Left: Total output energy of the amplifier as a function of the pump power for different number of pulses per burst n. The repetition rate within the burst is fixed at 1 MHz. The pump pulse starts 600 µs before the laser pulse arrival and lasts until the end of the burst. Right: Efficiency calculated as the ration of extracted energy to absorbed pump energy. Pump light absorption was calculated to be 70% using the cross sections in combination with the spectral distribution of the diode module output. Measurements done at 1 Hz.

Download Full Size | PDF

Although the pump profile was not homogenized, the output beam profile of the amplifier was Gaussian like without any hot spots as shown in Fig. 4. The negative thermal lens generated by the material (c.f. dn/dT values given in [13]) produced a divergent output beam. For high output energies of more than 300 mJ in the case of 1000 pulses the profile became slightly distorted (cf. Fig. 4 right), which might be attributed to thermal stress.

 

Fig. 4 Beam profiles, from left to right: Fluorescence distribution in the crystal at full power (measured in image plane in front of SM2); Seed beam in the crystal plane; Propagated output profile at 300 mJ in 1000 pulses; Propagated output profile at 600 mJ in 1000 pulses. Profiles measured in 1 Hz operation.

Download Full Size | PDF

Measurements were done at repetition rates from 1 Hz to 10 Hz, with always a 1 ms burst length. Up to a repetition rate of 5 Hz no significant change in the amplifier performance could be observed. For higher repetition rates, however, a slight drop of about 10% in the output energy occurred, which may be attributed to the generation of a thermal gradient in the material. Hence the lower emission cross section at higher temperature results in lower gain.

The delay between pump and seed pulses was optimized to give an equal energy distribution in the output pulse train while pumping at maximum diode current. At this point of operation, the depletion of the inversion due to spontaneous emission and energy extraction is equal to the inversion generated due to the pump beam, which results in a constant inversion and gain over the burst duration. The stability of the temporal energy distribution as shown in Fig. 5 left is limited due to instabilities of the seed laser. Such a variation in the temporal structure is then directly transferred to the output beam. The spectral width of the input is reduced from 4.3 nm to 4 nm, while the central wavelength is shifted from 1031 nm to 1031.5 nm. The output spectrum corresponds to 390 fs Fourier limited pulse length, calculated by a Fourier transformation of the spectrum assuming a flat phase.

 

Fig. 5 Left: Temporal energy distribution of the output pulse train for 1000 pulses in 1 ms at maximum output energy. High frequency modulations originate from the seeding fs front end laser system. Right: Integrated input and output spectrum of the burst. Measurements done at 1 Hz.

Download Full Size | PDF

To reach higher single-pulse energies the intra-burst repetition rate for the 1 ms burst was reduced stepwise, which is also accompanied by a reduction of the total seed energy. To ensure a uniform distribution of the energy over the pulses within the burst, the time τ₀ was increased (cf. Table 1). As a consequence, the total output energy was nearly preserved for down to 250 pulses per burst, corresponding to a single-pulse energy of 2.01 mJ. The performance at different intra-burst repetition rates is summarized in Table 1. Further reduction of the pulse repetition rate lead to laser induced damage of the material.

Table 1. Performance of the amplifier at 1 Hz with different intra burst repetition rates

View Table

4. Conclusion

We have demonstrated a new approach for the generation of high repetition rate, high energy ultra-short laser pulses as a burst. The described laser amplifier is based on Yb:CaF₂ as the active medium cooled down to 100 K. A novel multi-pass relay imaging scheme provides a good overlap of the extraction laser beam with the pump light distribution. During a burst with a duration of up to 1.5 ms an averaged output power of more than 600 W was demonstrated. This corresponds to an efficiency of up to 25% of extracted energy related to absorbed pump energy, which is the highest efficiency that has been demonstrated so far for a pulsed Yb:CaF₂ laser amplifier [21,22]. The burst’s internal repetition rate was tuned between 250 kHz and 1 MHz with only minor impact on the output power, leading to a maximum single pulse energy of up to 2 mJ. The output spectrum supports a pulse duration of 390 fs Fourier limited. Laser operation at a repetition rate of up to 10 Hz was shown. Up to 5 Hz only minor impact on the Gaussian beam profile and output power was observed. Further development will include an additional amplifier and a grating compressor.