Significant improvement in the polishing process of fused silica optical components has increased their lifetimes at 351 nm. Nevertheless, for large laser facilities like the LaserMegaJoule (LMJ), zero defect optical components are not yet available. Therefore, a damage mitigation technique has been developed to prevent the growth of the laser-initiated damage sites. Because of the difficulty to produce mitigated sites with sufficiently large depth, the initial morphology of damage to mitigate is a critical issue. The aim of this work is to determine laser parameters (pulse duration, fluence) which permit us to initiate damage sites in accordance with our mitigation process. Confocal microscopy is used to observe damage sites that have sub-surface cracks and consequently to measure precisely the diameter and the depth of the area to mitigate.
©2009 Optical Society of America
For high power lasers like the Laser MegaJoule (LMJ)  in construction near Bordeaux in France or the National Ignition Facility (NIF)  in the USA, the lifetime of optical components must be increased. Although the polishing techniques of optical components have been considerably improved, nanometric cracks present under the surface can still initiate damage sites at nominal fluences . Laser irradiation of these weak points leads to the irreversible apparition of micron-sized damage sites. For fused silica components, these damage sites grow exponentially [4–6] under subsequent irradiation at lower fluence and make the component unusable. To avoid damage site growth, one of the most promising methods uses a CO2 laser operating at a 10.6 µm wavelength to locally melt the silica surface . This technique was also applied by Mendez and coworkers  for the repair of damage sites caused on the surface of mirrors and has been found to remove damage pits up to a depth of 0.5 µm. Bass, Guss and coworkers have also demonstrated [9–10] the possibility to mitigate successfully the damage growth of sites as large as 500 µm with cracks extending to 200 µm in depth.
Accordingly a complete process to improve the laser damage resistance of 3ω optics was developed by R. Prasad and coworkers . This process, called stabilization, includes different steps that consist in finding the defects and mitigating their subsequent growth before installing optical components inside high-power laser facilities. Since there is no non-destructive technique for efficiently detecting these nanometric defects, the revelation stage is obtained by irradiating the component with a table-top laser whose parameters (wavelength, fluence) are equivalent to the LMJ/NIF laser beam. After this step, all the component weaknesses that are likely to initiate damages will have been revealed, at least in theory. Then these damage sites are detected and mitigated. This mitigation procedure is efficient only if all the fractured material surrounding the damaged crater is completely melted. Otherwise, damage reappears at the same position during the following laser shots and the mitigation procedure is useless .
This stabilization process, although it is currently running, still needs to be improved. This optimization needs to know precisely the morphology (diameter and depth of the cracks present under the surface) of the created damage sites. For this, we have used a non-destructive observation technique based on confocal microscopy. Our aim is first to define much more precisely the damage sites in order to apply correct CO2 mitigation parameters and secondly to find an initiation procedure which produces much smaller damage sites that will be easier to mitigate. In Section 2, we describe our experimental approach. We study the influence of different laser parameters (pulse duration and fluence of irradiation) on the morphology of the initiated damage sites in Section 3. Then the effect of iterative procedure is related in Section 4, and the results are discussed in Section 5.
2. Experimental procedure
On the LMJ facility, final optical components (diffraction gratings and windows of the experimental chamber) will be exposed to mean fluences as high as 8 J/cm2 and local peak fluences of 14 J/cm2 at 3 ns pulse duration and at the wavelength of 351 nm. All the component weaknesses that could initiate damage on LMJ up to 14 J/cm2 must be detected. To considerably increase optics lifetime, the revelation procedure has to reach this range of fluence.
The influence of the different laser parameters is estimated through an experimental approach in 4 steps:
• Damage site revelation with an excimer laser.
• Observation and characterization of each damage site with a confocal laser scanning microscope (CLSM).
• Mitigation of each damage site with a CO2 laser.
• Observation and characterization of each mitigated site with a CLSM.
In order to compare results obtained with the different parameters, this study has been realized with identical samples, i.e., synthetic fused silica polished by SESO, 50 mm in diameter, and 5 mm thick.
2.1 Damage sites revelation with an excimer laser
Our revelation facility is based on an excimer laser (GSI LUMONICS PM848) operating at 351 nm  and is similar to the laser used by Prasad and coworkers . Its spatial beam profile is rectangular with a top-hat shape. Fluence modulations are less than 10% of the average value and the repetition rate is variable up to 200 Hz. The pulse duration is around 16 ns. The maximum energy available is approximately 130 mJ per pulse, and consequently the laser beam size in the sample plane has been adapted so as to meet our needs in term of maximal fluence. Figure 1 illustrates the beam setting used in this study.
With this beam, whose size in the sample plane is 485×260 µm2, a raster-scan mode is used to realize the homogeneous irradiation of the 400×400 mm2 LMJ components. The rectangular beam shape permits us to avoid beam overlap. Thus, a complete scan of a full LMJ optic lasts around 3 hours.
Our laser cavity is made of an off-axis resonator providing a beam with reduced divergence (θ ~0.3 mrad). In the normal configuration of the laser cavity, the output pulse profile corresponds to the succession of 2 cavity round trips (around 8 ns for each). Consequently, the typical temporal profile is composed of two consecutive peaks as shown in Fig. 2.
The exact knowledge of the laser pulse duration is necessary to scale the real fluence corresponding to the value at 3 ns. But, because of the particular shape of the temporal profile, the definition of the pulse duration is a critical point. In our case, the Full Width Half Maximum pulse duration (τFWHM) is equal to 14.5 ± 0.2 ns and the effective pulse duration (τeff) defined as the ratio of total energy to peak power is equal to 15.7 ± 0.3 ns. We can notice that, in this configuration, both values are close.
To investigate the influence of pulse duration on damage revelation, we need to have access to different values of this parameter. To do that, the pulse duration of our excimer laser can be modified by tilting the exit mirror of the cavity. After the 1st round trip, most of the energy of the laser pulse is not reflected in the direction of the rear optic and exits the cavity. Figure 3 represents the temporal profile in this configuration.
We can see clearly that the second peak (due to the second round trip in the cavity) has disappeared. In this case, the Full Width Half Maximum pulse duration is 7.1±0.2 ns and the effective pulse duration is 12±0.3 ns. The important difference between the two values is due to the energy available at the end of the pulse. This energy is not taken into account by the FWHM calculation method.
In order to rescale the real fluence of irradiation at 3ns, a scaling law is necessary. Previous experimental works based on Gaussian pulse shapes  have shown that the use of a scaling law in τ 0.5 is coherent to rescale laser induced damage threshold fluence for nanosecond pulses. Recently, Carr and co-workers  have developed a diffusive model to predict the effect of a pulse shape change from Gaussian to flat-in-time. But our pulse shape (see Fig. 2 and Fig. 3) does not permit us to apply directly this model. Nevertheless, very good agreement for damage density measurement was obtained among different table-top lasers with different pulse shapes (including our laser), by applying this scaling law with the effective pulse duration . Consequently, we will use the effective pulse duration (15.7 ns or 12 ns) to rescale our fluences at 3ns with a law in τ 0.5.
2.2 Observation and characterization of a damage site with a CLSM:
The presence of cracks under the damage site has already been observed using Scanning Electron Microscopy . Such an approach is limited by the destructive character of the technique. The sample must be broken and coated with metal before any observation can be made. Non-destructive techniques have also been investigated: we can cite the Optical Coherent Tomography  developed by Guss and coworkers to image the laser damage sites during CO2 mitigation. In our case, we have chosen to investigate another non-destructive technique based on confocal microscopy (CLSM).
A CLSM is an integrated system consisting of fluorescence or reflected microscope, a laser light source, a scan head and a computer. The configuration of the CLSM used for our characterizations is shown in Fig. 4. In practice in a CLSM, the laser beam is focused to a spot in the sample through the microscope objective (objective HCX PL APO CS with a magnification of 63.0x and a Numerical Aperture of 1.4). Fluorescence or reflective light emanating from this spot is collected by the same microscope objective, and sent to a photodetector (PMT). In our case, measurement is realized in reflective mode with a laser at 458 nm wavelength. By putting an aperture diaphragm (pinhole set as airy 1) in front of this PMT in a conjugate focal plane of the objective we can collect the light coming only from this point. The light originating from above or below the focus point does not reach the detector. For each image, resolution in the three directions and confocal volume are given in the caption.
As illustrated in Fig. 5, to generate an image, the laser beam is scanned across the object by a raster scanning mechanism based on two vibrating mirrors driven by galvanometric motors. These mirrors deliver the laser beam excitation to the specimen. The first mirror controls scanning along the X axis the other along the Y axis, in order to sweep all the field. Speed and angular extent of deflection of mirrors are controlled by the regulation of scanning speed and field cover (to obtain a close-up).
To obtain a three-dimensional view of the specimen the objective is moved along the Z axis with a nanometric step-size by a piezo-electric crystal, in order to realize several images. This stack of images is illustrated in the right part of the Fig. 5. Finally, the computer can generate a three-dimensional picture of a specimen by assembling a stack of these two-dimensional images from successive focal planes. During this operation, the number of stacks and the size of the two-dimensional images can be chosen.
As examples of CLSM characterization, Figs. 6 and 7 represent images of typical morphologies of a damage site revealed on the rear face of a silica component. In Fig. 6, the top view of a damage site, as can be seen on left image, can be done including all the stacks so that it represents not only what is on the surface but also the cracks under the surface. The side view, as can be seen in the central image, is also available by choosing the number of slices. The right image is a three dimensional picture, from a given viewpoint.
These two examples highlight morphologies of damage sites close to those described previously by Carr and coworkers : Fig. 6 is an example of a “pansy” damage and Fig. 7 represents a “mussel” damage. Nevertheless, we can affirm that in our configuration (fluence, pulse duration) the majority of the detected damage sites can be classified in the second category.
Concerning the structure of each damage site, Figs. 6 and 7 demonstrate that the sub-micron cracks under the damage site described by Wong and coworkers  and observed using Scanning Electron Microscopy  are also clearly identifiable using CLSM.
The ratio d/h was chosen as a parameter representative of these two particular morphologies. d is the larger diameter of the damage site including all cracks visible on the surface and h is the depth of the damage site defined as the depth of the deepest discernible crack. In Section 3, we will report this ratio for different laser damage sites.
2.3 Mitigation of damage sites using a CO2 laser
The third part of our experimental procedure consist in mitigating each revealed and characterised damage site, so as to eliminate all the damaged material, including cracks present under the surface. Since the damage sites were tens of microns deep, the sophisticated techniques of mitigation, developed by Bass, Guss and coworkers [9–10] were not necessary. Our mitigation facility is based on a CO2 laser (Synrad Firestar V20) operating at 10.6 µm with a 20 W maximum power. It allows the localized irradiation of each damage site on samples. A phase mirror changes the linear polarization into circular for getting symmetric craters. A lens mounted on a z translation stage permits us to focus the beam on the sample, with a diameter adjustable from 200 µm to 800 µm measured at 1/e2. A complete description of this facility is available in reference .
2.4 Characterization of the mitigated site:
After mitigation, the circular crater created on the silica surface has dimension closely related to the CO2 laser parameters. To analyse each crater shape, we perform an observation and characterization with the same CLSM previously used after the damage initiation step. During this last step we verify that all the previous cracks have disappeared and that there is no new crack that could have been propagated. Figure 8 is an illustration of such verification. The left part of the figure shows a conventional microscopy image, where no residual cracks are observable on the surface, and on the right the confocal microscopy image permits us to be sure that the mitigation process is correct.
3. Effect of the pulse duration and the fluence on the morphology of the damage sites
To choose the best method of defect revelation, we analyse the morphology of the damage sites initiated under different conditions. For this purpose, we have used several silica samples. On each sample, an area of 9 cm2 was irradiated by the laser described in Section 2.1 at given fluence and pulse duration. The observations of each damage site with a CLSM as described in Section 2.2 gave the diameter and the depth for all the laser damages. Then some laser damage sites were mitigated and observed once again with confocal microscopy.
3.1 Impact of the pulse duration on the size of the damage sites
We have compared the morphology of different damage sites revealed with the two excimer laser configurations described in Section 2.1. The first sample was irradiated at 28 J/cm2 in the 12 ns configuration and the second one at 32 J/cm2 in the 15.7 ns configuration. According to the temporal scaling law in τ 0.5, the damage density on both samples corresponds to an irradiation at 14 J/cm2 with 3 ns pulse duration. The results for the parameters defined by confocal microscopy observation (depth vs. diameter) are summarized in figure 9.
First, we can notice that damage sites that were created with the same laser parameters are very different in size for both configurations. This large distribution can be explained by the great variability of original defect morphologies such as nanometric cracks sizes. Nonetheless, for all damage sites, the aspect ratio d/h, is between 2 and 4. These results are coherent with previous work . Thanks to this observation, we can define the following rule of thumb for an industrial process: laser damage sites have depth less than half the diameter of the damage site. For an industrial process, a classical observation with a camera will thus be sufficient to work out the mitigation parameters to apply.
The second point to be noticed is that the distribution of damage sizes shifts to lower values when the pulse duration becomes shorter. This result is coherent with other works realized in the range 1–10ns . Consequently, using the excimer laser in its 12 ns configuration is a good way to reveal damage sites as small as possible. With the 12 ns configuration, we can mitigate many more damage sites with our current mitigation procedure .
3.2 Effect of the irradiation fluence on the damage site morphology
Now to evaluate the laser fluence effect we use exclusively the 12 ns configuration that produces smaller damage sites. A new silica sample was irradiated at the fluence of 23 J/cm2 in raster-scan mode that is much less than that of 28 J/cm2. According to the temporal scaling law in τ 0.5, this fluence is equivalent to 11.4 J/cm2 with a 3 ns pulse duration. Again each laser damage site on this new sample was observed with confocal microscopy. Figure 10 shows the damage sites sizes (diameter and depth) for these two fluences.
We can see clearly in this figure that the aspect ratio d/h is, once again, between 2 and 4 for all damage sites whatever the fluence. As we might expect, the distribution of damage sizes shifts to lower values when the fluence becomes smaller. This result is consistent with the model described by A.M. Rubenchik and M.D. Feit .
3.3 CO2 mitigation of real damages sites
A preceding study  concentrated on mitigated damage resistance under laser irradiation. According to the study, if the stabilization crater is not deep enough to remove all the damaged silica, fractures propagate from the previous damage under only one irradiation at 8 J/cm2. CLSM permits us to compare damage depth before and after mitigation.
Figure 11 shows CLSM characterizations before and after CO2 mitigation of a real damage site. The damage diameter is 108 µm and its depth is 20 µm without the cracks and 31 µm including the deepest discernible crack. The mitigation crater is 21 µm deep, which is sufficient to mitigate the main damage, but not all surrounding fractures. These fractures, still visible after mitigation, are smoothed due to the CO2 laser melting. The noticeable presence of lateral fractures could be explained by a mitigation crater not perfectly centred on the damage. Nevertheless, their presence as well as the deeper cracks indicates that the mitigation was not deep and not large enough. Consequently, this damage is not correctly mitigated and will grow under a new laser irradiation as mentioned previously.
Figure 12 shows CLSM characterizations for another real damage site, again before and after CO2 mitigation. The damage diameter is 70 µm and its depth is about 15 µm without the cracks and 20 µm including the deepest discernible crack. The mitigation crater is 22 µm deep, which is sufficient to mitigate both the main damage and all surrounding fractures. Consequently, the damaged silica has been correctly and totally removed with a mitigation crater slightly deeper. This result indicates that cracks do not propagate under CO2 laser irradiation.
Our current mitigation procedure  generates CO2 craters as deep as 20 µm with suitable laser damage resistance. Moreover, our CLSM observations show that this depth is sufficient to repair all damage sites of depth less than 20 µm, provided the CO2 laser is correctly centered. Still, for damage sites deeper than 20 µm, some cracks remain under the surface after the mitigation. Consequently, their damage growth is not stopped by our procedure, which needs further improvement.
4. Discussion and summary
We have demonstrated that we are able to correctly mitigate damage sites growth as long as the damage sites are less deep than 20 µm. In the normal configuration of our excimer laser (effective pulse duration of 16 ns), defects revealed under nominal LMJ running conditions generate some damage sites much deeper than 20 µm. By reducing the effective pulse duration to 12ns, we have found that damages sites created under equivalent conditions are smaller. With this configuration most of damages sites are successfully mitigated but a few of them are still too deep.
As it is not possible to further reduce the pulse duration of our excimer laser, we need to improve the effectiveness of our stabilization process. More variations of the CO2 laser parameters are under study to find a new process in order that deeper damage could be successfully mitigated. Another way to improve the process concerns an iterative procedure with fluences increasing up to 14 J/cm2 (equivalent pulse duration of 3 ns). We have demonstrated that the distribution of damage sizes shifts to lower values when the fluence gets smaller. Thus, it can be expected that a nanometric crack leads to a smaller damage when it is irradiated at a smaller fluence. To explore this possibility, a preliminary study with our excimer laser in the short pulse configuration compares the morphology of the damage sites revealed with the two following procedures applied to two similar samples. The area concerned on each sample is about 9 cm2.
Procedure on sample 1: One raster scan at 14 J/cm2
Procedure on sample 2: Successive raster scans at 4, 6, 8, 10, 12 and 14 J/cm2
After each raster-scan, the sample is observed by regular microscopy, and each new damage site is characterized. Results show that the two procedures are similar in term of damage site revelation. The same optical component weaknesses are revealed whatever the irradiation procedure as soon as the maximum fluence reached is 14 J/cm2 (3 ns equivalent pulse).
In the case of the sample 2, the three first raster scans (respectively at 4, 6 and 8 J/cm2) do not reveal damage sites. The three following raster scans at 10, 12 and 14 J/cm2 lead to the revelation of new damage sites at each fluence. With this iterative procedure, 80% of the damage sites have a diameter less than 40 µm compared to 50% when the irradiation is directly realized at the maximum fluence of 14 J/cm2. So we conclude that damage sites revealed with the iterative procedure are smaller. Nevertheless, such an iterative procedure with increasing fluences seems to have a drawback: each damage site must be detected and mitigated as soon as it appears. If this is not done, the growth of the considered damage site during the subsequent irradiation at higher fluence becomes critical and leads to a more complicated mitigation .
Plane fused silica samples were irradiated by an excimer laser in three configurations by varying the effective pulse durations and fluences. The laser induced damage sites were then mitigated with a CO2 laser. Confocal microscopy was used to characterize the sites before and after mitigation. The damage sites were correctly mitigated only if the mitigated site was big enough to include all surrounding fractures on the damage site created by the excimer laser. Those fractures are not easily detectable by conventional microscopy. Confocal microscopy is a powerful tool for the development of optic stabilization process. We are currently working on the UV-laser irradiation process and the mitigation process in order to improve the effectiveness of our stabilization process. Our goal is to have zero defect optical components for Laser MegaJoule operation.
References and links
1. M. L. André, “The French MégaJoule Project (LMJ),” Fusion Eng. and Design 44, 43–49 (1999). [CrossRef]
2. C.A. Haynamet al, “National Ignition Facility laser performance status,” Appl. Opt. 46(16), 3276–30303 (2007) [CrossRef]
3. H. Bercegol, P. Grua, D. Hebert, and J-P. Morreeuw, “Progress in the understanding of fracture related laser damage of fused silica,” Proc. SPIE 6720, 672003 (2007). [CrossRef]
4. M. A. Norton, L. W. Hrubesh, Z. Wu, E. E. Donohue, M. D. Feit, M. R. Kozlowski, D. Milam, K. P. Neeb, W. A. Molander, A. M. Rubenchik, W. D. Sell, and P. Wenger, “Growth of laser initiated damage in fused silica at 351nm,” Proc. SPIE 4347, 468 (2001). [CrossRef]
5. G. Razé, J.M. Morchain, M. Loiseau, L. Lamaignère, M. Josse, and H. Bercegol, “Parametric study of the growth of damage sites on the rear surface of fused silica windows,” Proc. SPIE 4932, 127–135 (2003). [CrossRef]
6. S. G. Demos, M. Staggs, and M. R. Kozlowski, “Investigation of processes leading to damage growth in optical materials for large aperture lasers,” Appl. Opt. 41(18), 3628–3633 (2002). [CrossRef]
7. L. W. Hrubesh, M. A. Norton, W. A. Molander, E. E. Donohue, S. M. Maricle, B. M. Penetrante, R. M. Brusaco, W. Grundler, J. A. Butler, J. W. Carr, R. M. Hill, L. J. Sumers, M. D. Feit, A. Rubenchik, M. H. Key, P. J. Wegner, A. K. Burnham, L. A. Hackel, and M. R. Kozlowski, “Methods for mitigating surface damage growth on NIF final optics,” Proc. SPIE 4679, 23–33 (2002). [CrossRef]
8. E. Mendez, K. M. Nowak, H. J. Baker, F. J. Villareal, and D. R. Hall, “Localized Co2 laser damage repair of fused silica optics,” Appl. Opt. 45(21), 5358,5367 (2006) [CrossRef]
9. I. L. Bass, G. M. Guss, and R. P. HAckel, “Mitigaztion of laser damage growth in fused silica with a galvanomter scanned CO2 laser,” Proc. SPIE 5991, 599010C (2006).
10. G. Guss, I. L. Bass, V. Draggoo, R. Hackel, S. Payne, M. J. Lancaster, and P. Mak, “Mitigation of growth of laser initiated surface damage in fused silica using a 4-6µm wavelength laser,” Proc. SPIE 6403, 64030M (2007). [CrossRef]
11. R. R. Prasad, J. R. Bruere, J. Halpin, P. Lucero, S. Mills, M. Bernacil, and P. Hackel, “Design of a production process to enhance optical performance of 3ω optics,” Proc. SPIE 5273, 296–302 (2003). [CrossRef]
12. S. Palmier, L. Gallais, M. Commandré, P. Cormont, R. Courchinoux, L. Lamaignère, J-L Rullier, and P. Legros “Optimization of a laser mitigation process in damaged fused silica Applied Surface Science255(10), 5532–5536 (2009).
13. B. Bertussi, H. Piombini, D. Damiani, M. Pommies, X. Leborgne, and D. Plessis, “SOCRATE: an optical bench dedicated to the understanding and the improvement of laser conditioning process,” Appl. Opt. 45(33), 8506–8516 (2006). [CrossRef]
14. B. C. Stuart, M. D. Feit, A. M. Rubenchik, B. W. Shore, and M. D. Perry, “Laser-Induced Damage in Dielectrics with Nanosecond to Subpicosecond Pulses,” Phys. Rev. Lett. 74(12), 2248–2251 (1995). [CrossRef]
15. C. W. Carr, J. B. Trenholme, and M. L. Spaeth, “Effect of temporal pulse shape on optical damage,” Appl. Phys. Lett. 90, 041110 (2007) [CrossRef]
16. L. Lamaignére, T. Donval, M. Loiseau, S. Bouillet, R. Courchinoux, J-C. Poncetta, B. Bertussi, and H. Bercegol, “Presentation and comparison of damage test procedures for fused silica and KDP crystals,” Proc. SPIE 7132, 71320X (2008). [CrossRef]
17. C. W. Carr, M. J. Matthews, J. D. Bude, and M. L. Spaeth, “The effect of laser pulse duration on laser-induced damage in KDP and SiO2,” Proc. SPIE 6403, 64030K (2006). [CrossRef]
18. G. Guss, I. Bass, R. Hackel, C. Mailhiot, and S. Demos, “In situ monitoring of surface post processing in large-aperture fused silica optics with optical coherent tomography,” Appl. Opt. 47(25), 4569–4576 (2008). [CrossRef]
19. J. Wong, J. L. Ferriera, E. F. Lindsey, D. L. Haupt, I. D. Hutcheon, and J. H. Kinney, “Morphology and microsctructure in fised silica induced by high ultraviolet 3ω (355nm),” J. Non Cryst. Solids 352, 255–272 (2006). [CrossRef]
20. M. Norton, J. Adams, C. Carr, E. Donohue, M. Feit, R. Hackel, W. Hollingsworth, J. Jarboe, M. Matthews, A. Rubenchik, and M. Spaeth., “Growth of laser damage in fused silica: diameter to depth ratio,” Proc. SPIE 6720, 67200H (2007). [CrossRef]
21. A. M. Rubenchick and M. D. Feit, “Initiation, Growth and Mitigation of UV Laser Induced Damage in Fused Silica,” Proc. SPIE 4679, 79–95 (2002). [CrossRef]
22. B. Bertussi, P. Cormont, S. Palmier, G. Gaborit, L. Lamaignère, P. Legros, and J-L. Rullier, “Effect of the temporal pulse duration on the initiation of damage sites on fused silica surfaces,” Proc. SPIE 7132, 71320A (2008) [CrossRef]