2.1

Manufacturing of optical coatings and particle mitigation schemes

Details on the manufacturing of laser-optics have been reported previously [9]. Standard optics (Sample Ref. #015 and #042) were manufactured by coating super-polished fused silica substrates via conventional ion beam sputtering (IBS) to produce anti-reflective (AR) optical coatings at 355 nm. The coating was realized with four layers of aluminum oxide and silicon dioxide.

Furthermore, two different approaches to reduce the number of damage precursors in the optical coating were tested: ion etching with a secondary ion source (strategy 1) and laser etching with a laser source (strategy 2). Both approaches for particle mitigation were first tested purely as a substrate pretreatment (samples #007 to #013 for ion etching and samples #018 to #034 for laser etching, respectively) and then also applied during coating in the IBS coating machine.

2.2

Test strategy

We decided to test the manufactured optical coatings with large area raster scans on an area of 60 mm² using two different laser fluences: 15 J/cm² and 25 J/cm². The setup used for laser-induced damage testing at DLR was described previously. [10] Briefly, the third harmonic (355 nm) of a pulsed Nd:YAG laser (Innolas DPSS 500, pulse repetition rate of 100 Hz, 8.2 ns pulse duration at 355 nm (FWHM)) was focused to the optics under test giving a near-Gaussian beam profile with a typical beam diameter (1/e²-diameter of a Gaussian fit) of 300 μm. The laser light was irradiated line-by-line with a number of 10 pulses per effective Gaussian beam area by adjusting the forward feed of the sample holder mounted to a bidirectional translation stage. The high scanning speed (meaning a low number of laser pulses per area) in combination with an online damage detection system minimizes damage growth due to the subsequent irradiation of laser-induced damages. Before and after each raster scan, the optical surface was investigated via differential interference (“Nomarksi”) and dark field microscopy using an Olympus BX61 microscope at 200x magnification. Due to the large surface area and the high magnification the stitched micrographs had a fairly large size of ~4 GPx (“giga pixels”), which corresponds to the size of approximately 167 image files from modern digital cameras (assuming an image size of 24 mega pixels). The large size of the images requires an efficient and automatic data evaluation. The test strategy is visualized in Figure 2.

Figure 2:

Test strategy for the coating optimization.

2.3

Data evaluation

In order to compare the number of damages created in the raster scans of the different manufactured laser optics, we developed a script written in the programming language Python that counts the number of damages and evaluates their size distribution.

For the evaluation we decided to use dark field microscopic images, since such micrographs taken with well-defined settings (e.g. illumination time of 1 s) were found to have more reproducible brightness distributions than Nomarski micrographs. For the comparison of identical positions in the micrographs, dark field images taken before and after irradiation were aligned. This was achieved via a coordinate transformation that rotates and translates the images to match the pixel coordinates of markers written to just below the surface of each optical substrate via laser beam writing. The micrographs were saved in a single stitched image file of 4GPx and binned to a size of ~1GPx (2×2-pixel binning).

In order to detect laser-induced damages, an algorithm is used to look for connected pixels (connected horizontally, vertically and diagonally; meaning that each pixel has 8 neighbors) with an intensity above a specific brightness threshold. This evaluation is performed for dark field micrographs taken before irradiation (counting the number of “defects”) and after laser irradiation (counting the combined number of “defects and damages”).

In principle, one could assume that the number of “damages” can then be calculated by subtracting the number of “defects” (micrograph prior to irradiation) from the number of “defects and damages” (micrograph after irradiation). The issue with this approach is that - in particular for raster scans with a low laser fluence - it can easily be obscured from effects such as laser-induced cleaning (e.g. cleaning residuals from the cleaning bath [10]) or a large number of laser-induced damages induced by a single small dust particle, which can be encountered even in a clean room environment. Figure 3 shows examples of such low probability but sometimes nonetheless important effects.

Figure 3:

Challenges to classify laser-induced damages.

In order to find an easily accessible metric to assess the damage resistance of the optical coating neglecting such effects, we decided to additionally look for “damages without detected precursor”. Here, “damages without visible precursor” mean damages that were generated in an area on the optical surface without any defect in its vicinity prior to the raster scan (the minimum distance was defined to ~50 μm). In practice, this was achieved by a masking algorithm. This algorithm detects defects in the dark field images taken before irradiation (left panel of Figure 4). The area around any detected defect is then not taken into consideration when counting the number of damages in the dark field micrograph taken after laser irradiation (right panel of Figure 4).

Figure 4:

Masking of a polishing residual and a defect in the data evaluation.

The final results of the Python script are then a list with the number and size of objects detected prior to laser irradiation (“the number of defects”), a list with the number and size of objects detected after irradiation (“the number of defects and damages”) and the number and size of “damages without visible precursor” obtained using the masking algorithm. In addition to that, an image file comparing microscopic images at a specific position (taken before and after irradiation) is generated for each object. Examples of these automatically generated images are provided in Figure 5.

Figure 5:

Examples of laser-induced damages. The top picture of Figure 5 shows an example of a laser-induced damage without visible precursor, whereas the damage in the lower picture was generated near a visible defect. Images labeled with “DF before” and “DF after” are dark field micrographs taken before and after the raster scans, respectively. Images labeled “DF after C” show the result of the image analysis in which the connected pixels with an intensity above a specific threshold are colorized. Finally, the images labeled “BF, after” are Nomarski micrographs taken after irradiation.

2.4

Laser-induced contamination measurements

Space optics are typically intended to be used in vacuum and should thus be resistant to laser-induced contamination (LIC), which may originate from the outgassing of volatile components, e.g. from glues, insulation materials or electronic boards. We have recently developed an ISO technical report [11] describing the test methodology for LIC measurements.

We compared the susceptibility of standard IBS coatings to optics that were optimized for laser-induced damage (using ion etching with conditions similar to optics #027) in a test setup as depicted in Figure 6.

Figure 6:

Schematic of the experimental setup for laser-induced contamination testing at 355nm.

In these tests, the sample is placed in a vacuum chamber (<10^-6 mbar base pressure) and is irradiated for ~20 hours with the 355 nm, nanosecond-pulsed laser system also used for LIDT testing. As contaminant, the glue Armstrong A-12 (Loctite, Henkel Corporation, USA) on alumina foil was tested. The contaminant was heated to 100°C to enhance the outgassing and thus to speed up the deposit formation. The contamination growth was monitored in-situ by detecting the laser-induced fluorescence with a long-distance microscope and an electron multiplying CCD detector. Additional LIC tests have also been performed at the fourth harmonic of the Nd:YAG laser at 266 nm. The 1/e² beam diameter was ~440 nm for LIC tests at 266 nm and ~620 nm for LIC tests at 355 nm.

3.1

Successful optimization process of the optical coating manufacturing

Figure 7 shows the number of damages (more precisely the number of “damages without detected precursor” as defined in section 2.3) for large-area raster scans with an energy density of 25 J/cm² (panel A) and 15 J/cm² (panel B). In the beginning we tested applying the particle mitigation techniques only to the substrate and not during the IBS sputtering process. Substrates of samples #007 to #013 were treated via ion etching and we varied the ion current (100 -300 mA), the ion energy (100-400 eV) and the incidence angle of the secondary ion source. Substrates of samples #018 to #022 were treated via laser etching and we varied the laser fluence and the scanning speed of the galvano scanner directing the etching laser beam to the sample. However, both methods showed no significant reduction of the damage density compared to the reference samples.

Figure 7:

Improvement of the detected damage density during the optimization process of the particle mitigation scheme.

In contrast to this, optics in which the particle mitigation techniques were not only applied to the substrate, but additionally also during the IBS coating, a strong reduction in the damage density was obtained for both the ion etching (samples #023 to #027) as well as for the laser etching (#029 to #034). For specific configurations (sample #027 and #034) optics could be obtained that showed no laser-induced damage with a diameter greater than 4.5 μm when tested in a raster scan at 25 J/cm². It should be noted, that this improvement of the optical coatings is not only visible in the “damage density without visible precursor” as indicated in Figure 7, but also in the number of detected “defects and damages”. For the interested reader, these results are provided in the appendix. Additionally, the improvement of the damage threshold has also been demonstrated in ramped raster scan tests as reported earlier. [9]

3.2

Laser-induced contamination testing of the optimized coating

It has recently been shown, that nano-sized absorbers embedded in optical coatings can strongly enhance the formation of laser-induced deposits. [12] It has thus been a hope that laser optics manufactured using ion etching (with parameters similar to sample #027) would be less susceptible to LIC than optics coated via conventional ion beam sputtering.

This was tested in LIC tests with the epoxy adhesive Armstrong A12 (~600 mg per LIC test) at energy densities of 0.5 and 1 J/cm² and for the two wavelengths 355 nm and 266 nm. The test duration was ~20 hours and the contaminant was heated to 100°C in the vicinity of the optics under test. Table 1 shows the height profiles of LIC deposits after tests with a laser fluence of 0.5 J/cm². Similar results for 1.0 J/cm2 are given in Table 3 in the appendix.

Table 1:

Height profiles of laser-induced deposits generated in LIC tests with Armstrong A12 with a laser fluence of 0.5 J/cm².

Several interesting findings can be deduced from Table 1:

In principle, the formation of any laser-induced deposit should be considered as a laser-induced surface damage, defined in the ISO-21254 standard as “any permanent laser radiation-induced change of the surface characteristics of the specimen…”. Indeed, the deposits as shown in Table 1 will induce a significant transmission loss (a few percent) of the AR coating and modify properties of the transmitted laser beam (e.g. the beam profile and the wavefront).

The continued irradiation of a laser-induced deposit can however also lead to the formation of a plasma and thus to an additional modification of surface or bulk material. [17, 18] In fact, this was observed in one LIC test on an improved laser optics at 355 nm resulting in a contamination-induced damage at a laser fluence of 1 J/cm² (all other optics remained undamaged during LIC testing). From in-situ fluorescence microscopy it can be concluded that this damage was formed in the center of the deposit after ~10 hours of irradiation (see Figure 8).

Figure 8:

Formation of a laser-induced contamination and subsequently a contamination-induced damage monitored via in-situ fluorescence microscopy.

Figure 9:

Microscopic images of the contamination/laser-induced damage.

A subsequent inspection of the optical sample with Nomarski and raster electron microscopy showed that material was melted and distributed over the irradiated area (see Figure 9). This morphology can only be explained by the formation of a laser-induced plasma that locally heated the material above the melting point.