1. Introduction

Short pulse (sub-ns) laser systems utilize chirped pulse amplification [1] (CPA) in order to safely amplify to high energies. This technique has led to a proliferation of diffraction grating based pulse compressors which operate in vacuum. For picosecond duration pulses multilayer dielectric (MLD) diffraction gratings operated in reflection demonstrate both the highest efficiency and highest damage threshold [2]. The Advanced Radiographic Capability (ARC) [3] on the National Ignition Facility (NIF) at the Lawrence Livermore National Laboratory (LLNL) will rely on pulse compressors with meter-size gratings and focusing optics to generate high peak power and irradiance on target. The ARC laser system will implement CPA on four beamlines of the NIF [4] to produce eight λ = 1053 nm high energy petawatt-class laser (HEPL) beamlets with a total energy up to 13.6 kJ and duration of 1-50 ps. The maximum energy will depend on the pulse duration and is limited by optical damage of the final compressor grating and downstream optics [5]. The ARC MLD gratings are comprised of etched SiO₂ grooves on top of a MLD coating. These are costly to fabricate and will be exposed to among the highest laser intensities in the beam path, making an understanding of their operational limits of particular interest. Previous laser damage assessments of MLD diffraction gratings have been based on catastrophic laser damage tests performed on small areas (~10⁻³ cm²) in air [6–10]. In order to test the MLD gratings and other ARC optics we have recently developed an in-vacuum laser damage test station (VLDTS) [11] that will enable testing of much larger areas (~20 cm²) under use conditions.

We have previously developed laser damage measurement techniques for estimating the performance of meter-scale NIF optics that are subjected to nanosecond pulse durations [12–14]. In the nanosecond regime, λ = 351 nm laser damage in transmissive fused silica optics is driven primarily by laser light absorption in defects such as residual polishing materials, surface irregularities and scratches. Hence, damage performance varies significantly over the surface of the sample and probability curves generated from small beam testing can depend on the test laser focal spot size [15]. As a result, laser damage performance qualification is done by measuring the cumulative density of damage precursors ρ(Φ), which is the areal density of sites that will cause laser damage up to a fluence Φ. In this work we demonstrate the extension of the ρ(Φ) methodology to picosecond duration pulses by high-fluence performance testing of MLD diffraction gratings.

Determining scaling of the laser damage with pulse duration (τ) is critical for establishing the maximum laser power for an adjustable pulse duration system such as ARC. It is well understood that for long pulse durations (~ns) optical damage is primarily driven by thermal diffusion into the material: a process which scales roughly as τ^0.5 [16, 17]. In dielectrics, as the pulse duration shortens to the timescale of energy transfer to the lattice, the damage mechanism becomes dominated by electron dynamics [18]. Stuart [19] observed that this transition occurred at pulse widths less than a few tens of picoseconds. For sub-picosecond duration pulses, damage in dielectrics is considered to be driven by the joule heating of valence-band electrons generated from multi-photon and avalanche ionization. Much of the study of picosecond laser damage physics has been motivated by advancements in laser machining [20]. For this work, we have measured laser damage in MLD gratings with pulse durations ranging from 1 to 30 ps: a range of pulse widths where the scaling is not well understood due to the many material-dependent condensed matter processes which are relevant.

Operating MLD coated optics in vacuum chambers can expose them to vapor-phase organic contaminants (VOCs), chemically remove oxygen from the MLD coating stack via laser assisted reduction reactions, and also deform the MLD coating stack from outgassing of trapped water vapor. Each of these processes depends on the environmental conditions as well as the characteristics of the MLD coating stack and has the potential to alter its performance. This has motivated recent testing of MLD gratings in vacuum with pulse durations 10 ps and below [21–23]. Additionally measurement of damage growth of gratings at 15 ps has been performed in situ using ~1 mm resolution inspection [24]. VOCs typically originate from machining oils (used during the fabrication of vacuum chamber and mechanical parts), resins and plasticizers (used in o-ring seals, cables, and similar items that contain polymers), trapped lubricants in motors and opto-mechanical devices, and surfaces with vacuum incompatible structure (e.g. anodized aluminum and cast metals). When under vacuum, these contaminants volatilize and condense on previously clean surfaces. Previous experiments have shown that VOCs can reduce the nanosecond laser damage resistance in dielectrics [25–28], but damage testing with sub-picosecond duration pulses showed no effect from this type of organic exposure [29, 30] where it was determined that in-vacuum damage was driven primarily by oxygen deficiencies. Additionally, the cleanliness of these vacuum test chambers is typically qualified by the use of online mass spectrometry, a high-cost method which has limited sensitivity, and requires pressures below 10⁻⁶ torr. We present, to our knowledge, the first study of the effect of controlled VOC exposure on picosecond laser damage of diffraction gratings.

2. Experimental method

2.1. In-Vacuum Laser Damage Test Station (VLDTS)

Figure 1 is a schematic diagram of the VLDTS used for high-fluence performance assessment of optic samples. The optical damage is generated from a λ_o = 1053 nm laser system which delivers short pulses (0.4-60 ps) with energies up to 20 mJ. The pulse duration is varied by adjusting the slant of the compressor and shapes are similar to that shown by Jovanovic [31]. The laser architecture relies on optical parametric chirped pulse amplification (OPCPA). The laser temporal and spatial characteristics were measured before each experiment. Pulse durations 3 ps and below are measured with a single-shot Frequency Resolved Optical Gating (FROG) diagnostic. Pulse durations above 3 ps are measured with a scanning cross correlator using a transform limited probe beam. The laser pulses are directed off two thin film polarizers (TFP) to obtain high polarization purity. A lens (L1) focuses the beam onto the sample in an F# ~150 geometry producing a 175 µm full width half max (FWHM) focal spot. The sample which is typically 2” diameter is positioned with a vacuum compatible motorized x-y stage located in a 60 cm diameter vacuum chamber. The fluence in the sample plane is estimated using both primary and on-shot fluence diagnostics. The beam is directed to the primary fluence diagnostic with a flipper mirror (FM). The focal spot is characterized with a 12-bit CCD camera (CCD1). The beam energy is attenuated with wedge windows and neutral density filters placed downstream in order to prevent distortion from short pulse nonlinear effects. A fluence calibration factor is determined from the measured focal spot and used to convert pulse energy to peak fluence at the sample plane. The on-shot fluence diagnostic measures the energy and focal spot of the beam transmitted through a 5% beam splitter (BS) located before the focusing lens. Before each measurement, a NIST calibrated energy meter was placed in the main beam path to calibrate the on-shot pyroelectric energy meter (EM). The focal spot is monitored in real-time from an identical lens (L2) focusing the beam onto a 12-bit CCD camera (CCD2). The reference energy and focal spot image of each laser shot incident on the test sample is recorded. We estimate the total random error in the fluence measurement to be 9%, due to random changes in the fluence measurements (5%) and the energy calibration of the reference beam (7%). The surface of the sample is monitored online using a long-distance white light microscope with ~10 µm resolution providing real-time detection of laser damage. For precision analysis, the entire sample area is scanned before and after the test with an automated step-scan microscope with a 5 × objective to obtain an image with 2 µm resolution [17].

 

Fig. 1 Schematic diagram of the VLDTS. Picosecond duration laser pulses are focused by a lens (L1) onto a sample contained in a vacuum chamber. The target surface is imaged with a white light microscope. The on-target laser fluence is estimated from calibrated diagnostics.

Download Full Size | PDF

2.2. Vacuum cleanliness

The vacuum chamber and all mechanical parts inside were assembled following proven vacuum engineering practices established for the NIF [32]. This includes gross cleaning, precision cleaning, and baking. Certified low-outgassing materials, lubricants, and adhesives were used where necessary. The chamber cleanliness was measured from swipe tests, mass spectroscopy, and optical transmission measurements on sol-gel coated fused silica witness optics. These witness optics are coated with sol-gel optimized for transmission at 526 nm and exposed to the chamber while under vacuum for 120 hours. During this exposure, VOCs are absorbed into the pores of the sol-gel coating, reducing the optical transmission through the optic. The change in transmission (ΔT) at the optimized wavelength is measured with a spectrophotometer with a sensitivity of ΔT = +/−0.01%, which is equivalent to a mono-layer of an organic compound such as dioctyl-phthalate. The maximum ΔT occurs when the pores are saturated (ΔT ~3.5%). This is a low cost method to accurately measure the VOC accumulation in vacuum chambers.

After the vacuum chamber and components were assembled, a sol-gel witness sample exposure to vacuum for 120 hours resulted in a ΔT = 0.16%. During this time, a surface-acoustic wave (SAW) sensor, which was used to measure the real-time deposition rate of contaminants, showed that a steady-state mass load of 7 µg/cm² was reached within the 120 hour test. After 1 month of vacuum chamber operation, a sol-gel witness sample yielded ΔT = 0.09%, showing that residual VOCs were still present in the chamber. To reduce VOCs further, a thin layer of baked technical grade 3 silica gel getter material was installed in the test chamber. A subsequent measurement was at the limit of the measurement accuracy (ΔT = 0.01%) indicating successful mitigation of the VOCs. The grating witness samples were cleaned using a process developed by Britten [33] and stored in a dry nitrogen box before testing.

To determine the effects of trace amounts of organic contamination on the laser damage tests we exposed test samples to a controlled amount of VOC. A drop of liquid contaminant was placed on a temperature controlled copper crucible and contained in a small chamber attached to the test chamber shown in Fig. 1. Dibutyl-phthalate (DBP) – a common plasticizer which has been observed to outgas from polypropylene used in optic mounts [34] – was chosen as the controlled contaminant in these experiments. The DBP is set to a fixed temperature and the valve is opened, exposing the chamber. A 4 mm diameter orifice is installed in the vacuum line between the vacuum chamber, and the turbomolecular pump (TMP) and roughing pump (RP), to control the gas flow rate out of the chamber and maintain base pressure at ~5x10⁻⁵ torr. The deposition rate of DBP onto the sample surface can be modeled using the DBP temperature, vacuum pressure and surface area inside the chamber. A sol-gel witness sample is placed in the chamber during the exposure to determine the total DBP accumulation.

2.3. Test procedure

We have developed a two-step approach to characterize the performance of MLD grating witness samples consisting of R-on-1 and ρ(Φ) measurements. All fluences reported here are in the plane normal to the incident beam propagation and all grating samples are tested at an incidence angle of 76.5° [35] with S-polarization.

An R-on-1 test was performed by focusing the test laser onto one location of the sample. The test site was exposed to the 10 Hz laser while the energy was increased until a change in the sample surface (i.e. damage) was observed with the in situ microscopy. Flexible and repeatable fluence ramps (for example, 0.5 J/cm² fluence steps and 10 shots/step) can be delivered to the sample using the computer controlled λ/2 waveplate and shutter. The damage initiation event is typically assigned to the maximum peak fluence recorded during the last shots of the ramp sequence. The test was repeated on multiple sample sites and the damage probability was determined from the cumulative distribution of the damage initiation fluences. The R-on-1 test results enable qualitative comparisons of different grating samples and test conditions and guide the subsequent ρ(Φ) measurements. In addition, R-on-1 test sites provide useful fiducials, or alignment markers, to define the area for raster scanning. The R-on-1 test method is commonly used because it is more representative of laser-conditioned optics than ISO standard 1:1 and N:1 test methods [6]. However, these small area tests typically yield substantially higher damage initiation fluences than what is observed on full scale (large area) optics in operation.

To determine ρ(Φ) a 1 × 1 cm² area is raster scanned using the small area beam. During the raster scan the sample was moved at a speed of 1 mm/s and the test laser was operated continuously at 10 Hz resulting in a 100 µm separation between each shot. As a guide for these tests, the laser energy was set such that the fluence does not exceed a value 10% lower than the lowest fluence damage site from the 20 site R-on-1 test. Typically this fluence level generates an ideal number of damage sites on a sample for the post-processing analysis. The sample position and laser energy for each shot was recorded during the raster and used to generate a fluence map assuming a constant fluence calibration factor. A corresponding damage map was generated from a detectible change in the sample surface as measured from the offline microscopy. The estimated fluence incident on each damage site was determined by registration of the fluence and damage maps using the alignment markers. The damage density vs. fluence was determined by binning the fluence map values, and calculating the area exposed and number of initiated damage sites per fluence bin.

The ρ(Φ) is characteristic of the optic and is consistent with different test beam sizes, making it essential for estimating operational performance. In addition, large area testing is required for detection of sparse defects with lowered damage resistance. Raster scanning of a focused beam can be used to simulate a large-area beam damage test using a low energy test laser. For these grating samples, a single shot large area test capability with 1 cm² area would require a ~4000 × increase in laser energy than was used in these tests. However, using a Gaussian focus requires overlapping consecutive laser shots on the sample in order to expose the entire sample area to a minimum fluence. This beam overlap during the raster results in a laser conditioning effect from the fluence ramping. Additionally this overlap may over-estimate the damage fluence as it is assumed that damage occurred from the highest fluence pulse incident on the damage site, as is the case for the R-on-1 procedure.

3. Results

3.1. Damage probability versus pulse duration and vacuum environment

Determining the scaling of the laser damage resistance with pulse duration is critical for predicting operational limits of a variable pulse duration laser system. It can also provide information on what mechanisms are contributing to damage. We have carried out several R-on-1 tests in order to determine the effects of pulse duration in both air and clean vacuum environments on MLD diffraction gratings. Figure 2(a) shows the measured damage probability as a function of incident laser fluence performed using 1 ps, 10 ps, and 30 ps duration pulses on the same grating sample in air and vacuum (closed and open symbols, respectively). For each of these pulse durations these measurements indicate higher damage resistance in clean vacuum than in air. These results typically follow an “S” curve due to the statistical nature of the measurement. Consequently, the lowest fluence which caused damage, commonly chosen as the damage threshold, can have the most variation from identical tests on the same sample and depends on the total area tested. For comparing R-on-1 measurements we have chosen the fluence which corresponds to a 50% damage probability since we have found it to be the most repeatable metric.

 

Fig. 2 (a) Results from R-on-1 measurements of an ARC grating sample made in air (solid markers) and clean vacuum (hollow markers) using 30 ps (red squares), 10 ps (green triangles), and 1 ps (blue circles) duration pulses. (b) The 50% R-on-1 damage probabilities are shown as a function of pulse duration performed in air (solid squares) and vacuum (hollow squares). The lines are power fits to the air and vacuum data, which both scale as τ^{0.22 ± 0.03}.

Download Full Size | PDF

The 50% damage probability fluences from these measurements are plotted in Fig. 2(b) as a function of pulse duration. The 30 ps measurement in air yielded a 50% damage probability fluence of 5.6 J/cm². The chamber was evacuated to a pressure <1 × 10⁻⁴ torr and an R-on-1 test was performed showing an increase to 6.3 J/cm². The corresponding 50% damage probabilities in air and vacuum for 10 ps duration pulses were 4.5 J/cm² and 4.7 J/cm², respectively. Additionally, test results for 1ps pulses show 50% damage probabilities of 2.6 J/cm² in air and 2.9 J/cm² in vacuum. The error bars represent the estimated random error in the fluence and pulse duration measurements. The lines are power fits to the data points in order to estimate the damage scaling with pulse duration over this range. The laser damage results from the in-air measurements and the in-clean-vacuum measurements have identical pulse-width scaling (τ^{0.22 ± 0.03}). This pulse-width scaling result is similar to previous in-air HEPL MLD grating damage test measurements [6, 22]. The ratio of the fit functions indicates a 10% reduction of the damage fluence when tested in air for the pulse durations ranging from 1 to 30 ps. A similar reduction in the damage fluence in air was also recently observed to occur in femtosecond damage testing of metallic gratings [36]. These are, to our knowledge, the first direct comparisons between air and vacuum damage fluence of gratings with picosecond duration pulses using high sensitivity VOC qualification of the test chamber.

3.2. Effects of pulse duration on damage morphology

Figure 3 shows the morphology of damage sites from the R-on-1 tests performed in the previous section. Figure 3(a) shows a SEM image of a typical 30 ps in-vacuum damage site. Damage is characterized by many small sites located within the focal spot footprint. The extent of the overall damage area is elliptical due to the 76.5° laser incidence angle. Figure 3(d) is a SEM image of a small region of a 30 ps vacuum test site showing scattered damage sites with sections of the grating lines (top layer) removed, and an intact MLD coating stack below. Figure 3(b) is a SEM image of a typical 10 ps in-vacuum damage site. Many small damage sites are also present along with larger areas of more extensive damage than observed using 30 ps pulses. A small region of the 10 ps damage site (Fig. 3(e)) shows a larger area where sections of entire grating lines have been removed and also partial grating lines remain. A 1 ps in-vacuum damage site (Fig. 3(c)) shows the most catastrophic damage to occur at this pulse duration. A large area of the sample has complete damage up to the focal spot FWHM. A close-up view of this damage site shows the most grating debris occurs for this pulse duration. A comparison between 1 ps damage sites tested in-vacuum (Fig. 3(g)) and in air (Fig. 3(h)) shows more grating debris near the edge of the damage for the in-air measurement. This suggests that grating damage debris is freely ejected from the surface under vacuum conditions, however when air is present the debris can be re-deposited onto the surface. This effect was also observed in laser ablation of metal [37] and may explain increased laser ablation rate in air at low fleunces. The trapping of debris can effectively lower the damage resistance measured with the R-on-1 technique as this includes multiple processes: laser conditioning, damage initiation and damage growth. Single shot test methods are required to further investigate this effect. Debris re-deposition onto the sample can result in both grating line absences and added nodes to the grating structure that are known to cause electric-field enhancement in the material [38] and reduce the laser damage resistance.

 

Fig. 3 SEM imaging of R-on-1 test sites on a grating sample in vacuum using (a) 30 ps, (b) 10 ps, and (c) 1 ps duration pulses, respectively. Higher resolution images are shown for areas of interest in (d), (e), and (f) to highlight the salient morphological features. The region near the edge of 1 ps damage sites tested in (g) vacuum and (h) air shows more debris present for the in-air testing.

Download Full Size | PDF

These observed differences in morphologies suggest that different damage mechanisms are in effect within this range of pulse durations. In the damage process, it is understood that conduction band electrons are generated in an area on the grating groove with highest electric field and these electrons are rapidly heated by the incident laser pulse. It is possible that in the 1 ps case the damage is mostly limited to part of the grating structure, producing smaller grating debris fragments. For longer pulse durations the process becomes more complex as the valence electrons have time to either recombine through or transfer energy to the ion lattice.

In addition to the observed debris re-deposition, there are other factors which may contribute to the reduction of the R-on-1 damage resistance in an air environment. Air surrounding the SiO₂ grating lines may alter the plasma dynamics preceding material damage. When damage occurs, a shockwave could propagate through air and damage or weaken nearby grating lines. Particles suspended in air could accumulate on the grating surface and become defects.

3.3. Vapor Phase Organic Contamination (VOC)

It is critical to perform laser damage characterization of MLD gratings in a test environment consistent with their use environment. We have placed sol gel witness optics in vacuum chambers both on-site and in other HEPL systems in order to determine what range of VOCs are common to vacuum chambers that house pulse compression gratings. The sol-gel witness samples are exposed to each vacuum chamber for approximately 120 hours and the corresponding change in transmission is measured, as described in Section 2.2. The results of the relative VOC contamination of the ARC compressor vessel, the VLDTS chamber and a reactive ion beam etch (RIBE) chamber are shown in Fig. 4(a) . The contamination levels of a few compressor vacuum chambers located outside of NIF varied from 0.5 to 2.4%. The large range in the measured values is likely due to varying vacuum cleanliness standards. Measurements of the VLDTS chamber made over several months of operation show a ΔT varying from 0.01 to 0.04%, corresponding to levels up to a mono-layer deposit of typical VOCs. The lowest VOC level was measured in a RIBE chamber resulting in a value at the measurement limit (ΔT = 0.01%). We believe that this low level of contamination was due to the continuous removal of hydrocarbons by the oxygen RF plasma.

 

Fig. 4 (a) VOC outgassing measurements of the ARC compressor vessel, and VLDTS shown as the change in transmission of sol-gel witness samples. A lower ΔT indicates less contamination. Results are compared with an ultra-clean RIBE chamber. (b) 50% R-on-1 30 ps damage probability fluence (red triangles) from an ARC grating sample as a function of VOC exposure by introduction of dibutyl-phthalate.

Download Full Size | PDF

Figure 4(b) shows results of 30 ps R-on-1 in-vacuum measurements of an ARC grating sample with controlled exposure to DBP contamination. Each point represents the 50% damage probability from a 20 site R-on-1 test. The measurement with the lowest contamination was obtained in clean vacuum, without DBP exposure. The silica gel getter material was removed from the chamber for the subsequent R-on-1 tests performed on the sample with increasing exposure to DBP. The highest contamination level was obtained by exposing the test chamber and sample under vacuum to DPB held at 10 °C for a period of ~15 hours yielding a 0.57% ΔT in the sol-gel witness sample. A significant change in the laser damage was not observed to occur over this range of DBP exposure. It is important to note that these results do not rule out effects of these VOC exposures on a single-shot laser system (such as ARC) as the fluence ramping effect of the R-on-1 test may laser clean the surface hydrocarbon contamination. Further testing using different damage test methods such as 1-on-1 are needed to address this issue.

3.4. Damage Density

For high-fluence operational assessment of grating samples, we have studied the onset of damage with low fluence raster scanning of 1 cm² areas. Figure 5(a) shows a pre-raster offline microscopy image of a 1.3 mm × 0.9 mm region of grating sample. Several small defects are observed. The white ellipse represents the beam FWHM projected onto the sample plane centered on the defect. A 1 cm² area of this grating was raster scanned at fluences below the 5% R-on-1 damage probability with a 1 ps laser pulse using the method described in section 2. The laser is incident from the right side of the images. Figure 5(b) shows the offline microscopy image of the same area as Fig. 5(a) after the raster scan. Here one defect has seeded grating damage which extends in the direction of the laser propagation. Closer inspection of this damage site with an SEM (Fig. 5(c-d)) shows three different damage morphologies. First, a ~10 µm diameter pit exists in the MLD stack which is characteristic of a buried nodular inclusion which has ejected from the coating [39, 40]. Second, grating lines are damaged in a diameter twice that of the defect as seen in Fig. 5(d). And third, a 10 µm × 100 µm comet-like damage crater, shown in Fig. 5(c), extends horizontally along the direction of propagation of the incident laser light and appears to originate from the defect edge. Due to the multiple-shot exposure from beam overlap during the raster scan, it is not known when these nodules eject through the coating and when damage occurs to the grating lines.

 

Fig. 5 High resolution microscopy images of a sparse defect in ARC MLD grating sample (a) before and (b) after raster damage testing at 1 ps in vacuum. The laser was incident from the right side of these images. The laser focal spot (FWHM) projected on the sample plane is drawn in white. Several features are observed in the pre-raster image (a), but in this case damage occurs on the largest one. SEM images (c-d) show more detail of the damage observed in (b). Here the defect appears to be a nodule inclusion which has ejected and caused significant damage to the MLD coating and the nearby grating structure.

Download Full Size | PDF

Figure 6(a) shows 30 ps R-on-1 and raster damage test results performed in vacuum on an ARC grating sample. Two separate 20 site R-on-1 tests (blue squares and green triangles) were performed at different locations on the sample. The lowest fluence (5% damage probability) was 5.7 J/cm². An area of 1 cm² was then raster scanned at low fluence (<5.1 J/cm²) in order to measure the density of damage initiations (red circles). The resulting ρ(Φ) data shows a steep scaling of the density with fluence and indicates that a beam with uniform fluence of 4.8 J/cm² would cause 76 damage site initiations in an area of 1 cm².

 

Fig. 6 a) R-on-1 damage probability (blue squares, and green triangles) and ρ(Φ) (red circles) measurements of a MLD grating sample measured with 30 ps pulses in vacuum. b) SEM image of a typical 30 ps damage site from the ρ(Φ) measurement. The two R-on-1 measurements were performed on different areas of the sample to show variation across the sample.

Download Full Size | PDF

This data provides a significant improvement of the damage performance estimate over the commonly used small-area R-on-1 test. However, estimating full scale performance of an ARC grating to the level of a few damage sites on a meter size grating requires extrapolation of the ρ(Φ) data over many orders of magnitude. As a result, further testing of even larger sample areas is required to study the lower density portion of the ρ(Φ) curve. The damage sites observed in this raster scan were mostly seeded by pre-existing defects. An SEM image of a typical raster scan damage site is shown in Fig. 6(b). The laser was incident from the right side of the image. Here there are also three different damage morphologies observed. First, a crater extends deep into the MLD stack and appears to be from an ejected nodule. Second, an area with twice the diameter of the crater has grating line damage as was the case for the 1 ps raster damage sites. Third, a small area on the edge of the crater opposite of the incidence laser shows damage to the MLD stack. This damage appears to be melting from laser heating and is further indication of different damage mechanisms occurring for 1 ps and 30 ps pulses. The large comet-like craters in the 1ps damage sites were not observed at 30 ps. Further testing is needed to determine the damage growth rate in order to estimate grating lifetime under high fluence operation.

4. Summary

We have found that commonly used small-area, test to failure techniques such as R-on-1 do not predict picosecond duration laser performance limits of MLD diffraction gratings. To thoroughly assess the damage precursor characteristics, we have measured the damage density (ρ(Φ)) from picosecond laser irradiation using the raster-scan technique. The ρ(Φ) measurements from large area testing reveal performance limiting defects not observed by R-on-1 testing. SEM inspection of several 1 ps and 30 ps low fluence initiated damage sites indicate they are primarily from ejected nodular inclusions within the MLD stack. We utilize the R-on-1 test method to evaluate the laser damage resistance with different vacuum environments and pulse durations. Over the pulse duration range tested (1-30 ps) the damage threshold scales as τ^{0.22 ± 0.03} _. Measurements show that testing in air has a 10% lower R-on-1 damage fluence on ARC grating samples as compared to clean vacuum. The damage morphologies observed using high resolution metrology depend on pulse duration and pressure. Additionally we observed larger damaged areas from 1 ps pulse durations than 30 ps for both the R-on-1 and raster scanning tests. We have performed, to our knowledge, the first damage test measurements in this pulse-width regime on diffraction gratings with controlled exposure of organic contamination. The amount of VOC exposure was qualified with a simple, yet high sensitivity technique. No change in the R-on-1 damage probability was observed over a VOC exposure range exceeding that expected for the ARC compressor vessel. Additional damage tests are needed in order to determine if conditioning had an effect on the VOC exposure measurements. Thorough ρ(Φ) is critical for determining operational limitations of high energy short pulse laser systems that utilize MLD compressor gratings.