2.1. Beamline A

Beamline A consists of a fiber front-end that includes a pulse stretcher, pulse picker, and optical isolator. The front-end produces laser pulses with an energy of $1\ \mu {\rm J}$ at a repetition rate of 1.75 kHz. These pulses are further amplified in a regenerative amplifier to an energy around 150 mJ, then in a linear amplifier to an energy around 0.9 J. The amplified pulses are compressed in a grating pulse compressor to below 3 ps.

2.2. Beamline B

Beamline B^{[Reference Chyla, Miura, Smrz, Severova, Novak, Nagisetty, Endo and Mocek15]} starts with an Yb-doped fiber oscillator operating at a 50 MHz repetition rate, delivering an output power of 2 W at a center wavelength of 1030 nm with a bandwidth of more than 20 nm. Pulses from the fiber laser are stretched by a Martinez-type stretcher up to 500 ps. A gold-coated grating with a groove density of $1740\ {\rm l\ mm}^{-1}$ was employed, providing a group delay dispersion of $1.41\times 10^{8}\ {\rm fs}^{2}/{\rm rad}$. After the pulse stretching, pulses are coupled into the regenerative amplifier cavity, which contains an ${\rm Yb}^{3+}{:}{\rm YAG}$ thin disk, a thin film polarizer (TFP), a Pockels cell with a $10\ {\rm mm} \times 10\ {\rm mm} \times 22\ {\rm mm}$ BBO crystal, high-reflection mirrors, and a quarter-wave plate. Pulses from the stretcher are captured in the amplifier cavity when a quarter-wave voltage of more than 10 kV is applied to the Pockels cell, and they travel inside the cavity as long as the high voltage is applied. After amplification, the amplified pulse is ejected through the TFP and passes through an optical isolator consisting of a half-wave plate and a Faraday rotator. Afterwards, the amplified beam is diverted from the input beam by the polarizing beam splitter (PBS) and sent to the pulse compressor. A schematic of the current status of Beamline B is shown in Figure 3.

Figure 3. Schematic of the current status of Beamline B. Shown are flat mirrors (M), convex mirrors (XM), concave mirrors (CM), lenses (L), half and quarter waveplates ($\lambda /2, \lambda /4$), PBS, and a thin film polarization beam splitter (DP).

In order to reduce thermally induced stress between the heatsink and the gain media, CuW is adopted as the heatsink material, because it has similar linear expansion coefficient to YAG. The ${\rm Yb}^{3+}{:}{\rm YAG}$ crystal, doped to 7 at.%, has thickness of 0.22 mm and is soldered on the CuW heatsink using gold-tin solder. The mounted disk is pumped by a fiber-coupled diode laser module delivering an optical power of up to 1 kW at a wavelength of 969 nm. The pump spot size on the disk was set to 4.8 mm to achieve an output of 45 mJ, and the amplifier cavity was designed so that the cavity mode was matched to the designed pump spot size. The optical-to-optical efficiency was close to 20%. The amplified laser pulses had a bandwidth of 1.5 nm, so they could be compressed down to 1 ps.

The laser cavity will later be upgraded with a second thin-disk head to reach an output energy of 100 mJ. Additionally, the Martinez-type stretcher will be replaced with a fiber-chirped Bragg grating stretcher that allows better control of dispersion and is more stable and compact. Finally, the amplified pulses will be directed to a second regenerative amplifier that will be constructed in 2014.

2.3. Beamline C

Beamline C^{[Reference Smrz, Miura, Chyla, Endo and Mocek16]} is aimed at achieving a pulse energy of 5 mJ at a 100 kHz repetition rate. In order to meet these requirements, an intense study has been conducted to develop a high repetition rate regenerative amplifier. The target specifications of Beamline C will be achieved after completing three milestones determined by the pulse energies, namely 0.5, 2, and 5 mJ. The experimental setup for reaching the first milestone (0.5 mJ) is shown in Figure 4. The regenerative amplifier is seeded by an ${\rm Yb}^{3+}$-doped fiber oscillator, as in Beamline B, but the pulses are stretched by a chirped volume Bragg grating (CVBG) up to $\sim $160 ps. The dispersion of the CVBG is $60\ {\rm ps\ nm}^{-1}$. The FWHM spectral bandwidth and the clear aperture are $2.2 \pm 0.5\ {\rm nm}$ and $8\ {\rm mm} \times 8\ {\rm mm}$, respectively. The dimensions of the BBO crystal for the Pockels cell are $5\ {\rm mm} \times 5\ {\rm mm} \times 25\ {\rm mm}$ and its quarter-wave voltage is 5.2 kV. An ${\rm Yb}^{3+}{:}{\rm YAG}$ thin-disk with a free aperture of 8 mm and a thickness of $220\ \mu {\rm m}$ is installed in the laser head. The disk is pumped by a 1 kW fiber-coupled diode laser module at a wavelength of 969 nm with a pump spot diameter of 2.8 mm. A schematic of the beamline is shown in Figure 4.

Figure 4. Schematic of the current status of Beamline C. Shown are an optical isolator (OI), $\lambda /2, \lambda /4$, PBS, and a CVBG stretcher and compressor.

The observed output energy was $830\ \mu {\rm J}$ at a 100 kHz repetition rate. The low pulse energy enabled compression in a highly efficient CVBG compressor. The compressed pulse energy and the efficiency of the CVBG were $730\ \mu {\rm J}$ and 88%, respectively. The output pulse had a spectral bandwidth of 1.2 nm and was compressed only to 4 ps pulse duration because the CVBG did not account for the dispersion of material in the path of the beam. By adding an additional diffraction grating compressor, the duration of the compressed pulses was decreased below 2 ps. The output energy will be increased by using more intense pump light and by modifying the thin-disk head and the cavity. Technical difficulties connected to further development of all the mentioned thin-disk beamlines are connected mostly to thermal management of the thin-disk modules and the availability of pump modules at a wavelength of 969 nm.

Table 1. Status of kW-class Thin-disk Beamlines.

The current status of thin-disk beamlines is indicated in Table 1.

3.1. Front end

The front end starts with a temperature-stabilized tunable CW fiber oscillator. The wavelength of the oscillator is matched to the peak of the gain curve of the cryogenically cooled amplifiers. The CW beam is then temporally shaped in an acousto-optic (A-O) modulator to limit the repetition rate to 10 kHz and subsequently shaped by an electro-optic (E-O) modulator to produce 2–10 ns pulses with a semi-triangular shape. The temporal resolution of the shaper is below 200 ps. The pulses are further phase modulated by 2 and 4 GHz modulators to increase the bandwidth of the pulses and prevent stimulated Brillouin scattering (SBS) and stimulated Raman scattering (SRS) in the amplifier chain. Then the repetition of the pulses is decreased by a pulse picker and pulses are amplified in a thin regenerative disk amplifier to $\sim $1 mJ. The Gaussian beam coming from the regenerative amplifier is then spatially shaped to a square cross-section super-Gaussian profile in a beam shaper. Then it is further amplified to $\sim $100 mJ in a multi-pass booster amplifier. The booster amplifier preserves the square super-Gaussian beam profile that is injected into the 10 J main pre-amplifier.

3.2. 10 J main pre-amplifer

The 10 J main pre-amplifier is based on a multi-slab design. It consists of four circular ${\rm Yb}^{3+}{:}{\rm YAG}$ slabs with two doping levels of ${\rm Yb}^{3+}$ (1.1, 2.0 at.%). The different doping levels are needed to uniformly divide the heat load among the slabs. The volume of each circular slab is diameter in 45 mm with a thickness of 5 mm and the pumped area is square $23\ {\rm mm} \times 23\ {\rm mm}$. The pump beam is homogenized light from diode stacks operating at 939 nm and producing $700\ \mu {\rm s}$ long laser pulses at a repetition rate of 10 Hz. The ${\rm Yb}^{3+}{:}{\rm YAG}$ is clad with a 5 mm ${\rm Cr}^{4+}{:}{\rm YAG}$ absorber (absorption coefficient $6\ {\rm cm}^{-1}$) that prevents amplified spontaneous emission (ASE) and parasitic oscillations. The amplifier is cooled by forced helium gas flow and operates between 150 and 170 K.

The extraction scheme of the multi-pass amplifier is shown in Figure 6. The beam is injected into the amplifier through a dichroic mirror and then image-relayed by a spatial filter ($f=1\ {\rm m}$) to a back reflector and back to the amplifier head. There is one spatial filter on each side of the amplifier head. Each pass is propagated by a set of separate mirrors. A deformable mirror is placed in the amplifier after the third pass. After seven or eight passes, the beam is ejected from the amplifier with a pulse energy of 7 J in the beam and a size of $20\ {\rm mm} \times 20\ {\rm mm}$.

Figure 6. Schematic of the 10 J cryogenic multi-slab amplifier. It consists of ${\rm Yb}^{3+}{:}{\rm YAG}$ ceramic slabs in the laser head (Yb:YAG), dichroic beam splitters (DBSs), lens arrays (LAs), vacuum spatial filters (VSFs), and homogenized pump diode laser modules (PDs).

Figure 7. Schematic of the 100 J cryogenic multi-slab amplifier. It consists of Yb:YAG, lenses (L), VSF, and PD.

3.3. 100 J power amplifier

The 100 J power amplifier is also based on the multi-slab design. It consists of six square ${\rm Yb}^{3+}{:}{\rm YAG}$ slabs with three doping levels of ${\rm Yb}^{3+}$ (0.4, 0.6, 1.0 at.%). The volume of each slab is $100\ {\rm mm} \times 100\ {\rm mm} \times 8.5\ {\rm mm}$ and the square pumped area is around $75\ {\rm mm} \times 75\ {\rm mm}$. The parameters of the pump light are similar to the 10 J amplifier. The ${\rm Yb}^{3+}{:}{\rm YAG}$ is clad with a 10 mm wide ${\rm Cr}^{4+}{:}{\rm YAG}$ absorber (absorption coefficient $3\ {\rm cm}^{-1}$) that prevents ASE and parasitic oscillations.

The extraction scheme of the multi-pass amplifier is shown in Figure 7. The beam is injected into the amplifier at a range of angles around $5^\circ $, so any overlap with the pump beam outside the amplifier head is avoided and no dichroic mirrors are used. Each additional pass is image-relayed by a dedicated spatial filter ($f=3\ {\rm m}$) back to the amplifier head. After the first pass, a deformable mirror is implemented to prevent degradation of the wavefront on subsequent passes. After four passes, the beam is ejected from the amplifier with a pulse energy of up to 120 J and a beam size around $75\ {\rm mm} \times 75\ {\rm mm}$. The average fluence in the amplified beam is $2\ {\rm J\ cm}^{-2}$. The amplifier is cooled by forced helium gas flow and operates between 150 and 170 K.

The current status of the multi-slab laser system is indicated in Table 2.

Table 2. Status of kW-class Multi-slab Beamline.

3.4. Numerical modeling

The HiLASE team has undertaken extensive energetics, thermal and fluid-mechanical modeling in order to optimize various amplifier parameters.

Figure 8. (a) Schematic of the laser slab with dimensions in mm, dashed line shows the spot of the pump beam in the ${\rm Yb}^{3+}{:}{\rm YAG}$ part of the slab that is clad by ${\rm Cr}^{4+}{:}{\rm YAG}$. (b) Transverse heat load of the slab used for the calculations (assumed constant in the longitudinal direction).

For energetics modeling, we have developed a MATLAB code^{[Reference Sawicka, Divoky, Novak, Lucianetti, Rus and Mocek20]} for the evaluation of stored energy in the laser amplifier that includes ASE. This model calculates the pump energy absorbed by the gain medium, excited ion density, ASE and heat sources in the ${\rm Yb}^{3+}{:}{\rm YAG}$ laser slabs by solving the rate equations in discrete time steps. During each step, the absorption of the pump radiation and spontaneous emission are calculated independently. In the absorption phase, the energy from a polychromatic pump source is absorbed in the medium in accordance with a probability proportional to the number of unexcited active ions and the absorption cross-section dependent on the pump wavelength. In the ASE phase, spontaneously emitted photons with random polarization are generated by the Monte Carlo method using the excited ion density inside the slab as the probability distribution. Their wavelength distribution is based on a probability density function derived from the emission cross-section. Rays containing thousands of photons are traced through the medium and amplified proportionally to the population inversion and the path length in each cell. All slabs of the amplifier are included in the model so the propagation of the rays among the slabs can be modeled. All surfaces can be treated as Fresnel reflecting, AR coated, or absorbing. Side faces can also include scattering. The slab model includes the active ${\rm Yb}^{3+}{:}{\rm YAG}$ core and the absorbing ${\rm Cr}^{4+}{:}{\rm YAG}$ cladding. A schematic of the slab and the calculated heat deposition are shown in Figure 8. Results were obtained for a slab size $100\ {\rm mm} \times 100\ {\rm mm}$, pump size $75\ {\rm mm} \times 75\ {\rm mm}$, ${\rm Cr}^{4+}{:}{\rm YAG}$ cladding 10 mm, absorption coefficient $3\ {\rm cm}^{-1}$, pump duration 1 ms, intensity $5\ {\rm kW}\ {\rm cm}^{-2}$, and temperature 160 K.

The wavelength-resolved absorption and emission cross-sections and lifetime on the upper laser level for a given temperature were obtained experimentally ^{[Reference Jambunathan, Koerner, Sikocinski, Divoky, Sawicka, Lucianetti, Hein and Mocek21]}. The cross-section and lifetime measurement station supports spectral resolution down to 15 pm and temporal resolution down to 100 ps. An example of the measurement of the absorption and emission cross-section of ${\rm Yb}^{3+}{:}{\rm YAG}$ at a temperature of 160 K is shown in Figure 9.

Figure 9. Measured absorption and emission cross-sections of the ${\rm Yb}^{3+}{:}{\rm YAG}$ at a temperature of 160 K.

A three-dimensional finite-element method (FEM) using Comsol Multiphysics software was chosen to model the thermal and stress effects in the amplifiers. The sources of heat were calculated in the ASE code. The lateral surfaces of the slabs are assumed to be cooled by flowing helium gas at 160 K. The spatially resolved heat transfer coefficient was derived from a two-dimensional model of a turbulent flow of helium gas at 160 K using the standard $k\varepsilon $ model^{[Reference Launder and Spalding22]} together with Kays–Crawford heat exchange in the turbulent boundary layer ^{[Reference Kays23, Reference Schlichting and Gersten24]}. The slab was assumed to have no thermal contact with its 2 cm thick Invar holder; hence all the heat is removed by convection through the faces. From the temperature and stress maps of the slab (Figure 10), optical path difference (OPD) and birefringence depolarization losses (Figure 11) were calculated for a single slab according to a prior approach^{[Reference Slezak, Lucianetti, Divoky, Sawicka and Mocek25]}. The gradual decrease of heat exchange efficiency in the direction of gas flow, caused by coolant heating, is the reason for the loss of left–right symmetry of the temperature, stress, depolarization, and OPD maps.

Figure 10. (a) Transverse distribution of temperature and (b) transverse distribution of the $xy$ stress component in a longitudinal cut in the center of the laser slab.

Figure 11. (a) Depolarization of the beam at the output of the amplifier (after four passes through six laser slabs) caused by stress-induced birefringence. (b) Stress- and temperature-induced OPD after a single pass through the laser head (after one pass through six laser slabs).

A beam propagation model of the 100 J power amplifier was created in MIRÓ using a Fresnel diffraction integral for propagation and the Frantz–Nodvik equation for amplification. The model was used to estimate beam aberrations, taking into consideration only the thermal OPD. The results of the beam intensity and OPD are shown in Figure 12. The resulting wavefront was corrected numerically. The reliability of the wavefront correction code was verified experimentally in a slab simulator^{[Reference Pilar, Divoky, Sikocinski, Slezak, Lucianetti, Kmetik, Bonora and Mocek26]}.

Figure 12. (a) Beam profile and (b) OPD of the beam at the output of the 100 J multi-slab system calculated in MIRÓ. Dashed square indicates the position of the laser beam.

The numerical model for wavefront correction calculates influence functions from a plate equation describing the bending of the thin facesheet for each individual actuator of the deformable mirror. The deformable mirror consists of a continuous gold facesheet (size of $100\ {\rm mm} \times 100\ {\rm mm}$, thickness of 1 mm) on which lateral forces are applied by piezoelectric stack actuators. The actuators form an equidistantly spaced rectangular array of $6 \times 6$ actuators and are capable of push/pull operation. The deformation of the mirror is computed as a superposition of the influence functions and the algorithm minimizes the rms OPD value.

The OPD after subtraction of defocus and tilt and the OPD corrected by the deformable mirror are shown in Figure 13. The rms value of the OPD was reduced from 111 nm down to 31 nm.

Figure 13. (a) Output OPD calculated in MIRÓ and shown in Figure 12(b) after subtraction of tilt and defocus. (b) Residual OPD after correction by the deformable mirror with 36 actuators.

4.1. Laser-induced damage threshold testing

First, the laser-induced damage threshold (LIDT) automated experimental station would be introduced. The station design allows one to measure the LIDT under a wide range of laser parameters: from the irradiation of small spots with 1–2 ps laser pulses at various wavelengths and a 1 kHz repetition rate to the irradiation of large spots with 2–10 ns laser pulses at 1030 nm with a 10 Hz repetition rate. The main advantage of this station is real-time monitoring of laser damage with an acquisition frequency of up to 1 kHz. This station will allow the determination of the damage occurrence, as well as following the damage growth and damage threshold variation under repetitive irradiation. A schematic of the station is shown in Figure 15. The particular design of the LIDT station permits investigations of samples of up to 1 kg with surface sizes of $100\ {\rm mm} \times 100\ {\rm mm}$.

Figure 15. Schematic of the LIDT measurement station: (1) high-speed shutter, (2) beam positioning and focus, (3) beam diagnostics, (4) scattered light damage detection and fluorescence collector, (5) slow-motion camera, (6) interference damage detection; (7) XYZ tower, (8) beam dump. It uses laser pulses from Beamline A (L1A), Beamline B (L1B), and the Multi-slab (L2) laser system.

4.2. Mid-IR optical parametric generator

For the investigation of laser–material interactions and processing, as well as the thin-disk and multi-slab systems, a mid-IR pulse source with a high repetition rate and an average power of 10 W ^{[Reference Novak, Miura, Severova, Smrž, Endo and Mocek27]} is being developed. The proposed scheme of the wavelength conversion setup (Figure 16) uses the picosecond output from the thin-disk regenerative amplifier as the pump beam. The seed beam for the wavelength conversion setup comes from the optical parametric generator (OPG) and the signal is tunable between 1.6 and $2.1\ \mu {\rm m}$. Only the signal beam is amplified by the optical parametric amplifier (OPA) chain. Idler beams having wavelengths from 2 to $3\ \mu {\rm m}$ can be extracted from the last parametric amplifier.

Figure 16. Schematic of the mid-IR parametric generator and amplifier. It consists of the thin-disk laser system, beam splitters (BS), mirrors (M), dichroic mirrors (DM), an OPG, and OPA.

4.3. EUV light generation

For EUV generation, powerful ${\rm CO}_{2}$ lasers are used to evaporate tin (Sn) droplets to generate a plasma that emits light at 13.5 nm. The ${\rm CO}_{2}$ laser provides a much higher average power and higher conversion efficiency to UV light, but the laser footprint and plasma size are large^{[Reference Endo28]}, so it is suited for high-volume manufacturing. There is also a need for a small-scale EUV source with very high brightness for method and component testing. This compact source requires a stable beam with a higher beam quality than the beam quality provided by ${\rm CO}_{2}$ lasers but at the same time the average power output required is only in the region of 1 kW. For this reason, we are constructing an EUV generation station using a thin-disk amplifier with a high average power. This source will be employed to study processes during EUV light generation.