Generation of Scratches and Their Effects on Laser Damage Performance of Silica Glass

Scratches are deleterious to precision optics because they can obscure and modulate incident laser light, which will increase the probability of damage to optical components. We here imitated the generation of brittle and ductile scratches during polishing process and endeavored to find out the possible influence of scratches on laser induced damage. Brittle scratches can be induced by spiking large sized abrasives and small abrasives may only generate ductile scratches. Both surface roughness and transmittivity are degraded due to the appearance of brittle scratches while ductile scratches make little difference to surface roughness and transmittance. However, ductile and brittle scratches greatly increase the density of damage about one order of magnitude relative to unscratched surface. In particular, ductile scratches also play an unignorable role in laser induced damage, which is different from previous knowledge. Furthermore, ZrO2 and Al2O3 polished surfaces appear to perform best in terms of damage density.

Scientific RepoRts | 6:34818 | DOI: 10.1038/srep34818 Methods Samples used in the experiments were 50 mm in diameter and 5 mm thick and no obvious scratches were on the surface. The samples were polished with a polyurethane pad adhered onto a synthetic tin plate installed onto a lapping machine (FD-380XL, Fonda, China). The platen can rotate with respect to the central axis. The samples were located in a separator. Both the separator and the platen were driven independently. The polishing slurry was fed continuously at a flow rate of 10 mL/min. A dead load of 2.9 N was applied onto the backside of workpiece. The polishing time usually lasted 30 min and in some cases where scratches did not appear after 30 min lapping, the time was prolonged. Various combinations of slurries and polishing pads 20 (Universal Photonics Inc., USA) were used in our experiments to generate scratches on glass surfaces in order to find out likely effect of polishing pad and polishing compound on laser damage performance. The details are tabulated in Table 1.  Abrasive sizes of polishing compounds were examined with a particle size analyzer (Mastersizer 3000, Malvern, UK). The morphology of the abrasives was inspected with a scanning electron microscope (Helios Nanolab 650, FEI, USA). The surface roughness was evaluated with an optical profiler (NewView 7200, Zygo, USA) and the transmittivity was tested with a spectrometer (Lambda 950, Perkin-Elmer, USA) over the range of 300 nm~1100 nm.
Damage density test was performed on a Nd:YAG laser damage testing system (Laser Zentrum Hannover e.V., Germany). The Gaussian laser pulse (8 ns@355 nm, beam waist 800 μ m) was focused onto the rear surface of samples and the repetition rate was 10 Hz. The damage test protocol adopted was raster scan. The stage of sample holder moved at a certain speed so that each pulse overlapped with the pulses adjacent to them at FWHM (Full-Width-at-Half-Magnitude) to ensure that the scanned area was irradiated at nearly the same fluence. The sample surface was divided into 3~5 sub-regions which were illuminated at different laser fluence. Each sub-region was 10 mm × 10 mm in dimension. The detailed testing layout can be found elsewhere 21,22 .
The same area was monitored with an optical microscopy 500× (VHX-2000, Keyence, Japan) and stitched each image to form a large image prior to and following laser damage testing. If no cracks were found under high magnification, the scratches were viewed to be ductile, which are usually light color in the images. On the other hand, brittle cracks scatter light strongly, which will be dark in the image. In this way, the fractions of ductile and brittle scratches can also be quantified.

Results and Discussion
Abrasive size, Surface roughness, Scratches, and Transmittivity. The abrasives used in our experiments were observed with SEM and size analyzer and the particle diameter is found a bit different from the size provided by the manufacture. The size of SiC agrees well with the testing results while other abrasives do not. The nominal size of CeO 2 , ZrO 2 , and Al 2 O 3 are all 0.3~0.5 μ m, but the size analyzer suggests that the size all lies in 3~4 μ m (D50). The reason may be the accumulation of micro-particles when the size is under 1 μ m. Small particles are prone to agglomeration due to high relative surface area and high surface energy of small particles. Hence the abrasives were observed using SEM to verify our conjecture. We can understand that the SiC(W7, 7 μ m) and SiC(W40, 40 μ m) are dispersed very well whilst CeO 2 , Al 2 O 3 , and ZrO 2 show apparent agglomeration (Fig. 1).
The surface roughness of each sample is listed out in Table 1 along with transimittivity and scratches. The surface was examined with optical microscopy (500× magnification) to find out whether scratches occur on the surface and whether the scratches are ductile or brittle. For the samples with obvious scratches, we quantified the scratches. The images were first binarized into white-black images and then the ratio of the scratch pixels to the whole pixels was considered to the quantity of scratches by using a software package ImageJ 23 . Comparing the results of scratches, we can see that CeO 2 and Al 2 O 3 did not induce scratches under normal polishing conditions, but ZrO 2 sometimes may result in slight scratches which was also reported by other researchers 3 . It is known that complex chemical reactions between glass and CeO 2 occur and a hydrated layer covers the surface of polished glass during the polishing process while only mechanical abrasion dominates the removal mechanism of glass (a) surface roughness of sample A is 1.19 nm without scratches; (b) surface roughness of sample C is 1.08 nm with slight ductile scratches and the depth of scratches is ~20 nm; (c) roughness of sample E is over 5.19 nm with much deeper brittle and ductile scratches and the depth of the scratches is over 300 nm; (d) sample E has a lower transmittance than sample A&C over the UV-Vis-IR band; (e) the surface roughness and transmittance are strongly affected by brittle scratches.
when ZrO 2 used as polishing compound 24 . It is the chemical reactions that accelerate the material removal rate during the polishing of glass, 1 μ m/h for CeO 2 abrasive versus 0.33 μ m/h for ZrO 2 . After spiking SiC a kind of harder abrasive than CeO 2 , many scratches get visible (Fig. 2). The surface contains numerous ductile scratches on the surface polished with CeO 2 doped with 7 μ m SiC and the density of scratches gets high with increasing the concentration of SiC. However, almost no brittle scratches were found on the surface. On the other hand, there are a number of ductile scratches as well as some brittle ones on the surface processed by CeO 2 plus 40 μ m SiC. Likewise, the scratches including brittle and ductile become denser with increasing the concentration of SiC. Brittle scratches appear when 40 μ m SiC abrasives were infiltrated into CeO 2 slurry because increasing the size of SiC abrasives will decrease the number of abrasives bearing the downward load and therefore the load on a single abrasive will increase accordingly which will lead to brittle fractures when the load is in excess of a critical load to induce brittle fractures in fused silica.
Comparing the images of surface morphology and surface roughness in Fig. 3 and Fig. 2, it can be found that surface roughness is all ~1 nm except for the cases with 40 μ m SiC since brittle scratches severely deteriorate surface quality and thereby surface roughness. The transmittivity spectra show that brittle scratches (Sample E) also strongly lower the transmittivity of fused silica in UV band (351/355 nm) from because they can scatter the incident light strongly and weaken the intensity of transmitted light (Fig. 3). Ductile scratches make much trivial difference to surface roughness and transmittivity as compared to brittle scratches and the transmittivity is only slightly reduced at 351/355 nm. Thus brittle scratches must be eradicated for precision optical components as only ~1.3% brittle scratches can increase surface roughness from ~1 nm to > 4 nm and reduce the transmittance from > 93% to < 90%, which is undesirable in high power/energy laser systems.
Damage performance. The samples were scanned with various energy fluences with 355 nm, 8 ns pulsed laser. The fluence was then converted to 3 ns with empirical rule 12,25 , and the fluence in the paper is all the converted one, i.e. 3 ns. The damage density was then extracted by comparing the defect density before and after testing. Each sample was scanned 3~6 regions so that damage density with fluence can be plotted. From the damage density results, it is clear that surfaces full of scratches are more sensitive to laser fluence. There are more mechanical defects at the intersecting points, e.g. micro-deformation of glass, micro-cracks, etc. and these defects will affect the laser damage performance of fused silica. Therefore, there is a higher probability that the laser induced ablation will be severer than other area. ZrO 2 & Al 2 O 3 -polsihed samples perform better than other samples. The causes for the noticeable difference may be that Al 2 O 3 and ZrO 2 are not absorptive at 355 nm and with few scratches while CeO 2 can greatly absorb the incident 355 nm laser, deposit the absorbed laser energy, heat the (c) sample D after damage testing, from which it is clean that both brittle and ductile scratches can cause laserinduced damage. area locally and finally damage the fused silica sample. From the Fig. 4, field 1 & 2 were seriously damaged after laser illumination and there are brittle scratches in these two fields before damage testing. In spite of only ductile scratches in field 3 & 4, damage happened after raster-scan testing in the fields. Our results indicate that ductile scratches can also be damage precursors and can trigger damage to fused silica, which is different from previous results that ductile deformation may not be harmful to optical components in high power laser systems 18,19 .

Conclusions
The artificial scratches were investigated to find out their possible effects on surface quality and laser damage. Various abrasives frequently used in optical manufacturing community were experimented. The results show that CeO 2 is more efficient than Al 2 O 3 and ZrO 2 in polishing fused silica and CeO 2 , Al 2 O 3 and ZrO 2 are all capable of polishing out a smooth surface (surface roughness RMS ~ 1 nm). Adding SiC into CeO 2 slurry will result in ductile and/or brittle scratches on polished surfaces, which depends on the size of abrasives added. Larger size will bring about ductile and brittle scratches while smaller abrasives may generate ductile scratches. Increasing the concentration of SiC will definitely raise the density of scratches. Furthermore, ductile scratches are found to have limited influence on surface roughness and transmittance while brittle scratches impact onto surface roughness and transmittance. From damage density results, it is found that ZrO 2 and Al 2 O 3 perform best in damage density and surfaces with numerous scratches usually damaged severely. Both ductile and brittle scratches can be precursors to laser damage and initiate catastrophic damage to optical components.