Nanosecond single and multiple pulse laser damage studies on HfO2/SiO2 high-reflective coatings were performed at 1064 nm. The evolution of LIDT and 100% damage probability threshold under multiple irradiations revealed that fatigue effects were affected by both laser fluence and shot numbers. And the damage probability curves exhibiting different behaviors confirmed experimentally that this fatigue effect of the dielectric coatings was due to material modification rather than statistical effects. By using a model assuming Gaussian distribution of defect threshold, the fitting results of LID probability curves indicated the turning point appeared in the damage probability curves under large shot number irradiations was just the representation of the existence of newly created defects. The thresholds of these newly created defects were exponential decrease with irradiated shot numbers. Besides, a new kind of damage morphologies under multiple shot irradiations were characterized to further expose the fatigue effect caused by cumulative laser-induced material modifications.
© 2013 Optical Society of America
The optics used in laser applications are always subjected to multiple pulse irradiations. Thus, investigations on single and multiple pulse laser-induced damages in optical coatings are of high practical importance for high-power laser applications [1–3]. In many materials, the damage threshold subjected to multi-pulse exposure is lower than that generated under single-pulse illumination [4, 5]. Wagner et al. have showed the case of bulk damage in KTP crystals tested with nanosecond pulses at 1064 nm, what is usually interpreted as material modification could be understood as a consequence of measurement statistics without any modification of the material . Then it’s important for us to identify whether the decrease of the S-on-1 damage threshold for an increasing pulse number is due to statistical test without any modification of the material or to real fatigue . Real fatigue implies long-living light-induced material modification, which weaken the materials resistance to subsequent pulse exposures. And usually accumulation of color centers has been proposed as the underlying physical mechanism to explain the laser-induced material modification under irradiation . For dielectric films, to understand the multiple pulses damage mechanism and damage initiators, we must clarify the so-called fatigue is caused by material modification or not.
In this paper, the specific damage behaviors of dielectric coatings irradiated with repetitive laser frequency are analyzed. The results have demonstrated experimentally that the decrease of LIDTs of the dielectric coatings under nanosecond multiple pulse irradiations at 1064 nm is due to real fatigue, i.e., laser-induced material modification. Today, in the nanosecond pulse width region at 1064 nm, defects whether laser-induced defects or intrinsic defects with irreversible changes are usually attributed to laser-induced damages of optical films under multiple irradiations. By using a model assuming Gaussian distribution of defect threshold, the information on the evolution of defects with shot numbers is highlighted. Researchers have done much work on the nanosecond single laser-induced damages in HfO2/SiO2 multilayer at 1064nm and found nodular defects are the most critical defects [9–11]. Because of the well vacuum system and the improvement of polishing and cleaning process of substrates, the damages of nodular injections from the substrate in our multilayer coatings are extremely rare. And the typical damage morphologies of our coatings appear at the close-threshold laser fluence are the relatively shallow pit surrounded with plasma scald for single pulse irradiation and the non-pit damage morphology for multiple pulse irradiations respectively. The initiators that induce damages under single- and multi-shot are identified by a series of measurements. And the influences of shot numbers and laser fluence on damage morphologies and damage depth are investigated specifically to further discuss the fatigue effect caused by cumulative laser-induced material modifications.
2. Sample preparation and damage tests
2.1. Sample preparations
The HR coatings were deposited on BK7 substrates with a diameter of 50 mm via e-beam evaporation. The starting materials were hafnia and silica. All coatings runs used the same deposition technology and were in the same coating chamber. The coating stack was [Substrate/(HL)^12H4L/Air], where H denotes the quarter wavelength optical thickness (QWOT) of HfO2 and L denotes the QWOT of SiO2 . The total thickness of the film was about 5 μm. The reflectance of the HR coatings at 1064 nm was more than 99.5% at an angle of incidence of 0°.
2.2. Experimental setup
The laser damage test apparatus used in this study is shown schematically in Fig. 1. A Nd:YAG laser 12ns operated at 1064nm in single longitudinal mode with up to a 5Hz repetition rate. In our tests, the 1/e2 spot diameters in x and y axis measured via the knife-edge method were 260μm and 300μm. The on-line damage detection setup comprised of a microscope focused on the tested area and a CCD to determine whether the radiation sites were damaged or not. A visible He-Ne laser was used as the illumination source. The damage judgment was made with the comparison between the test area before and after laser irradiation. 20 sites for each energy density were tested and the fraction of sites damaged was recorded. If a site damaged at early shot numbers, the laser irradiations would not stop unless the damage growth was very catastrophic. And the site spacing was 1.5 mm which was about five times greater of the laser spot diameter. It was sufficiently high to avoid the subsequent pulses influencing the test result on the neighboring sites. The procedure was repeated for other fluences until the range of fluence was sufficiently broad to include points of zero damage probability and points of 100% damage probability to develop a plot of damage probability versus fluence. The relative error of damage probability was about ± 15% mainly due to the uncertainty of the nonuniformity among the samples (3%), the measurement of laser spot area (5%), and the fluctuation of laser energy (5%).
2.3. Analysis methods
The damage morphologies were observed by a Leica microscope. A Zeiss dual beam microscope that combined the functions of a scanning electron microscope (SEM) and a focused ion-beam (FIB) was used to obtain the details of the damage morphologies and inner structures. An atomic force microscopy (AFM) and a step profiler with a 2 µm pinhead radius were also employed in mapping damage depth.
3. Experimental results and discussions
3.1 Experimental results
Some S-on-1 damage probability curves, with S ranging from 1 to 1000, are shown in Fig. 2(a), which obviously show an increase in damage probability and a clear decrease in damage threshold with an increasing number of shots. Each of the shown curves is acquired in an independent measurement. Figures 2(b) and 2(c) present the evolution of LIDT and 100% damage probability threshold versus shot number derived from Fig. 2(a) .
The decrease of LIDT and 100% damage probability threshold with the increase of shot numbers is apparent, revealing distinct so-called fatigue effect. After shot 3000, both the LIDT and the 100% damage probability threshold tend toward stabilization. In detail, the stable value of the LIDT and the 100% damage probability threshold are 16.5 J/cm2 and 95 J/cm2, respectively. For LIDT, the decrease within the initial 10 pulses is not obvious relative to the whole decrease segment. But for the 100% damage probability threshold, most of the changes occur within these initial 10 pulses. This phenomenon suggests that for HfO2/SiO2 HR coatings the material irradiated with low laser fluence need more shots to accumulative laser-induced modifications. However, the first few shots can make the damage probability increase and the 100% damage probability threshold decrease notably when laser fluence is high. This result suggests that for fixed shot numbers the cumulative laser-induced damage is greater at high fluence than that at low fluence. Because of the influence of laser fluence and shot numbers on fatigue effect, the damage probability curves under different shot numbers manifest different damage behaviors.
The damage probability curves under different shot numbers are analyzed specifically (Figs. 3 and 4). The bold squares denote the experimental data and the solid lines represent the fitted curves. For shot numbers from 1 to 10, the blue lines in Fig. 3 show that the damage probability curves increase at a near-linear trend. For shot numbers larger than 10, as the results in our test, a turning point where the slope of the damage curve changes abruptly appears in the damage probability curves as shown in Figs. 4(a)-4(c). When the incident laser fluence is greater than the value of the turning point the damage probability increases sharply. The occurrence of turning point indicates that the presence or absence of laser-induced damage is closely related to both shot numbers and laser fluences. The phenomenon exposes the effect of multiple pulse laser-induced material modifications experimentally. The reason may be due to the influence of newly created defects, namely, the laser-induced defects or intrinsic defects with irreversible changes under multiple pulse irradiations [13, 14].
The evolution of defect threshold and density versus shot number has important implications for knowing the influence of repetitive laser irradiations on the damage mechanism. To confirm the defect information, we used the model assuming Gaussian distribution of defect threshold that developed by Krol et al. [15, 16] to determine the defect thresholds and densities, linked to the shape and slope of the damage probability curves. The specific simulation process was introduced in . When shot number is low, the damage probability is closer to a liner increase as the blue curves shown in Fig. 3. The associated best-fit curves with the model are the red lines and show the existence of only one class of defects. When shot number N = 1, best-fit curve is calculated with the following parameters: T1 = 110J/cm2, d1 = 180/mm2, ΔT1 = 65 J/cm2. The threshold standard deviation ΔT1 is very high, so the damage probability exhibits a near-linear trend. The calculated results for shot number N = 10 is T1 = 105J/cm2, d1 = 410/mm2, ΔT1 = 50 J/cm2. For large shot numbers, the associated best-fit curves with the model are plotted in Fig. 4. We note the existence of two classes of defects. As Table 1 shows, when shot number N = 60, one class of defects has a LIDT T1 of 37J/cm2, a density d1 of 9/mm2 and a threshold standard deviation ΔT1 of 10 J/cm2; the properties of the other class are a LIDT T2 of 94J/cm2, a density d2 of 360/mm2 and a threshold standard deviation ΔT2 of 5 J/cm2. The defect ensembles extracted from the probability curves for shot number N = 200 and 1000 are also listed in Table 1. The influence of shot numbers on the defect thresholds and densities are pronounced. When shot number increases, the thresholds of both the two classes of defects decrease and the class of defects with higher threshold decreases more. Corresponding to the single class of defects shown in Fig. 3, the simulated results in Fig. 4 confirm that material properties are modified by the preceding pulses in multiple pulse mode, and multiple pulse laser irradiations can induce new defects or make irreversible changes of intrinsic defects. It’s clearly the fatigue effect of the dielectric coatings at 1064nm is due to material modification under multiple pulse irradiations.
Figure 5 presents the fluence of the slope change point versus shot numbers. Actually, the turning point mainly corresponds to the threshold of the second class of defects. It’s distinct the turning point decreases with an increasing number of shots and follows an exponentially drop way with shot numbers. Hundreds of shot numbers later, the turning point will no longer change. In our test, when shot numbers larger than 300, the turning point tends to a stable value of about 64 J/cm2. This phenomenon indicates laser fluence also plays an important role in material accumulation. When laser fluence lower than a certain value, the subsequent shot numbers have small effect on damage generation. The decrease of turning point with increasing shot numbers is also one manifestation of accumulation. From the investigation on the multi-shot damage probability curves, we can see the incubation phenomenon is pronounced. This incubation phenomenon implies laser-induced modifications to the coating material that weaken it to subsequent pulse exposures when laser fluence reaches a certain value.
3.2 Morphology and depth of damage sites
Morphology observations enable precise viewing of damage characteristics and identification of the damage initiators of coatings. Our setup with in situ damage detection recorded the pulse number at which the irradiation sites were damaged. For a constant shot number and laser fluence, different irradiated sites may be damaged at different pulse numbers, therefore, the damaged sites we compared in this section were all damaged at the last few shots unless noted otherwise.
The damage morphologies for 1064 nm HR coatings at different fluence and shot numbers are very different. Figure 6 shows the damage sites at laser fluence of (a) 62 J/cm2 and (b) 206 J/cm2 in 1-on-1 case tested by optical microscopy. For single pulse irradiation, as shown in Fig. 6, the typical damage morphologies are central pits surrounded by large areas of plasma scald. Figure 7 shows the high-resolution SEM images of Fig. 6, which reveals detailed information of the two damage sites shown in Fig. 6. From Fig. 6 and Fig. 7 we can see that the diameter of the scald areas and the number of the central damage pits vary with laser fluence. When laser fluence is close to LIDT, the numbers of damage pits at center are always one or a few. In this case the color contours of the plasma scalding area are relatively smooth as shown in Fig. 6(a). When laser fluence increase, the numbers of central damage pits of most damage sites increase (Fig. 7(b)) as well. Meanwhile the color contours of the plasma scalding area become no more smooth because of the interaction effect of the plasma spark induced by these different damage pits at the center. When laser fluence increases, the part of the spot size where laser fluence is greater than precursors damage thresholds increases. As a result, more precursors in the test site with larger laser fluence irradiated are revealed, namely the numbers of central damage pits increase and in turn the overall size of the damage site increases. Because of the nonuniform defect distributions, even if the laser fluence is large, the numbers of central damage pits are not always large for all the tested sites. Our tested results show that for single pulse irradiation all the damage sites have one or a large number of micron-sized pits in the center. And there is also a submicron sized hole in the central of each damage pit as shown in Fig. 7(a), which suggests that the damage sites are induced by submicron absorbing defects.
To position the bottom of the damage pits accurately, the pits are measured via AFM, as shown in Fig. 8. Figures 8(a) and 8(b) illustrate the depth information of the damage pits in the two damage sites shown in Fig. 6, respectively. The damage depth may be related to the electric intensity distribution in the multilayer thin films. The highest and the second highest peaks of intensity are observed at the second and fourth interfaces, corresponding to the depths of 882 nm and 1204 nm. The AFM results for damage pits in 1-on-1 case are approximately 1204 nm, corresponding to the depth of the fourth interface as shown in Fig. 8, no matter what the laser fluence are. A higher electric field at the interface may play an important role in inducing damages , and we will discuss this in detail latter.
The damage morphology enables us to further understand the underlying damage mechanism, i.e., fatigue effect for material under multiple irradiating pulses. The typical damage morphologies induced by multiple pulses at different laser fluences are obtained. For multiple pulse irradiations, one kind of special damage morphologies appears at close-threshold laser fluence. First we investigate the damage sites caused by cumulative laser-induced material modifications operated at fluences much below the 1-on-1 damage threshold. Figure 9(a) illustrates the representative morphology at fluence of 25 J/cm2 for shot numbers N = 1000. Through the tested results of Nomarski microscopy and SEM, the damage exhibits large area of plasma scald. However, there is not any other damage characteristic such as damage pit within the damage site is found. And the depth information tested through step profiler shows the maximum depth of this kind of damage sites is range from about 20 nm to 100 nm which is smaller than the depth of the outermost layer. As we have discussed above, the typical damage morphologies for single shot damages may be induced by submicron sized absorbing defects at the relatively shallow interfaces and the effect of higher electric field in the coatings. However, the precursor of the non-pit damage morphology as shown in Fig. 9(a) may be the submicron sized defects within the outermost layer. Then we only see plasma scald on the surface but not observable pit. Figure 9(b) illustrates another representative kind of damage morphologies for multiple pulse irradiations. The damage site is irradiated at laser fluence of 87 J/cm2 with shot numbers N = 300. When laser fluence is larger than a certain value, some damage sites begin exhibit pits in the center just like the 1-on-1 damage morphologies as shown in Fig. 9(b). Here the certain values are different for different irradiated shot numbers, but are all below the 1-on-1 damage threshold. When laser fluence is high enough, all damage sites will exhibit pits in the center. For multiple pulse irradiations, besides these two typical damage morphologies, if the damage does not occur at the last shot and grow during the subsequent pulses, delamination may happen as shown in Fig. 9(c). Figure 9(c) presents a damage site with irradiated laser fluence of 147 J/cm2 and shot numbers N = 10 that damaged on shot number 1. We can see the delamination is not a circle, there are several reasons may explain about it. First our laser spot is not very circular, the 1/e2 spot diameters in x and y axis are 260μm and 300μm respectively. Then the nonuniform defect distributions may also influence the shape of the delamination. And delamination is mainly caused by mechanical stress, but the mechanical stress is always not the same in all directions. In this paper, damage growth is not our concern.
Because the other typical damage morphologies for multiple pulse irradiations also exhibit large area of plasma scald around the central damage pits as shown in Fig. 9(b). And the damage morphologies as shown in Fig. 9(a) can only be distinguished plasma scald. Then when the damage shows pits in the center, besides the absorbing defects at the relatively shallow interfaces, the defects within the outermost layer may play a role in inducing this kind of damage morphologies simultaneously. As we have discussed above, a turning point appears in the damage probability curves for large shot number irradiations. When the incident laser fluence is greater than the value of the turning point the damage probability increases sharply. An interesting phenomenon for damage morphologies with incident fluences before or after the turning point is found. The typical damage morphology as shown in Fig. 9(a) is found only appears at laser fluence before the turning point. The reason may be that when laser fluence is high enough, the second class of defects, i.e., the absorbing defects at the interfaces have much higher possibility to induce damage under high laser fluence. Then the damages all exhibit pits in the center when laser fluence is higher than the turning point. However, as we have discussed above, this kind of damage sites may contain the damages induced by defects within the outermost layer as well. In addition, both lateral and vertical directions of these no damage pit sites will get larger with fixed shot numbers and increasing laser fluence. Figure 10 illustrates the damage morphology and depth information of a damage site at fluence 74.5 J/cm2 for shot numbers N = 100. The diameter of this damage site is about 400 μm with maximum depth of the plasma scald is about 80 nm. Besides these large shot number irradiations, for low shot numbers such as N = 5 (except N = 1), there are also damage sites exhibiting only plasma scald. And the damages without pits usually appear at fluence close to the damage threshold. In addition, a few damage sites at high fluence also show this kind of damage morphologies when low shot number irradiations. The possibility is that the number of absorbing defects at interfaces which can induce damages is not high enough with low shot numbers irradiating.
For multiple pulse irradiations, to identify the sources responsible for the formation of the damage pits, FIB cross-sections are applied on these pits. Figure 11 shows the SEM and FIB tested results of the damage site in Fig. 9(b). The inset in Fig. 11(b) is taken in the intermediate process of the FIB cut. A smaller submicron sized hole below the wide upper part is found and further FIB is taken as the main image shown in Fig. 11(b). It’s clearly that the pit-diameter is varied with lodging depth and the lateral shape is much similar with the crater referred to as a complex crater . The smaller submicron-sized hole in center of the pit as shown in Fig. 11(a) actually corresponds to a narrow “channel” as shown in Fig. 11(b). In fact, the damage pit contains two parts: a narrow channel and a wide upper part. And the depth tests show the bottoms of the wide upper part correspond to the fourth interface for all our damage pits as shown in Fig. 8 and Fig. 11. Because of the very small diameter of the narrow channel as shown in Fig. 7(a) and the certain depth of the narrow channel, the AFM is hard to test the part of the depth of the narrow channel. Then the test results of AFM shown in Fig. 8 only show the depth of wide upper part which are correspond to the fourth interface. The FIB results resolve this problem, as shown in Fig. 11(b). The bottom of the wide upper part of this damage pit corresponds to the fourth interface, and the bottom of the narrow channel corresponds to the seventh interface. The analysis of other damage sites finds that the bottoms of wide upper parts are all located at the fourth interface, while the bottom of the narrow channels namely the location of the absorbing defects are random distributed at the relatively shallow interfaces but always at or below the fourth interface.
The reason of pit-diameters variation with lodging depth may be involving a combination of melting and fracture. Initially the submicron sized absorbing defect absorb inside and around the defect cause melting and superheating the material within the channel volume. This process is accompanied by a rapid internal-pressure buildup and because of the higher electric fields at the second and fourth interfaces, fracture along the fourth interface takes place, followed by ejection of both the fractured portion and the molten material from within the channel. And the damage pit exhibits a narrow channel and a wide upper part at last. These results are also applied to the situation of single pulse irradiation, but the absorbers will undergo cumulative laser-induced modifications when irradiated with multiple pulses.
The damage probability curves, damage morphologies and damage depth under single- and multiple-shot irradiations for HfO2/SiO2 high-reflective coatings at 1064nm in nanosecond were investigated to discuss the fatigue effect and damage mechanism. The damage probability curves illustrated obviously fatigue effect, i.e., material modification. On the one hand, damage threshold decreased and damage probability increased with an increasing number of shots. On the other hand, for large shot number irradiations a turning point appeared in the damage probability curves comparing with the damage probability curves increased at a near-linear trend for low shot number irradiations. It meant another class of defects appeared. For single pulse irradiation, the typical damage morphologies were central pits surrounded by large areas of plasma scald. For multiple pulse irradiations, damage morphology exhibiting only large area of plasma scald appeared. Both the depth of the two typical damage morphologies under single and multiple pulse irradiations were measured to discuss the different damage initiators and cumulative laser-induced material modifications.
This work is supported by the National Natural Science Foundation of China under Grant No. 11104293 and No. 61308021.
References and links
2. K. R. Manes, H. G. Ahlstrom, R. A. Haas, and J. F. Holzrichter, “Light-plasma interaction studies with high-power glass laser,” J. Opt. Soc. Am. 67(6), 717–726 (1977). [CrossRef]
3. A. E. Chmel, “Cumulative effect in laser-induced damage of optical glasses: a review,” Glass Phys. Chem. 26, 49–58 (2000).
4. A. E. Chmel, “Fatigue laser-induced damage in transparent materials,” Mater. Sci. Eng. B 49(3), 175–190 (1997). [CrossRef]
5. F. R. Wagner, A. Hildenbrand, J. Y. Natoli, and M. Commandré, “Multiple pulse nanosecond laser induced damage study in LiB3O5 crystals,” Opt. Express 18(26), 26791–26798 (2010). [CrossRef] [PubMed]
6. F. R. Wagner, A. Hildenbrand, H. Akhouayri, C. Gouldieff, L. Gallais, M. Commandre, and J. Y. Natoli, “Multipulse laser damage in potassium titanyl phosphate: statistical interpretation of measurements and the damage initiation mechanism,” Opt. Eng. 51(12), 121806 (2012). [CrossRef]
7. F. R. Wagner, C. Gouldieff, and J. Y. Natoli, “Contrasted material responses to nanosecond multiple-pulse laser damage: from statistical behavior to material modification,” Opt. Lett. 38(11), 1869–1871 (2013). [CrossRef] [PubMed]
8. F. R. Wagner, G. Duchateau, A. Hildenbrand, J.-Y. Natoli, and M. Commandré, “Model for nanosecond laser induced damage in potassium titanyl phosphate crystals,” Appl. Phys. Lett. 99(23), 231111 (2011). [CrossRef]
9. F. Y. Génin and C. J. Stolz, “Morphologies of laser-induced damage in hafnia-silica multilayer mirror and polarizer coatings,” Proc. SPIE 2870, 439–448 (1996). [CrossRef]
10. F. Y. Génin, C. J. Stolz, and M. R. Kozlowski, “Growth of laser-induced damage during repetitive illumination of HfO2/SiO2 multilayer mirror and polarizer coatings,” Proc. SPIE 2966, 273–282 (1997). [CrossRef]
13. L. A. Emmert, M. Mero, and W. Rudolph, “Modeling the effect of native and laser-induced states on the dielectric breakdown of wide band gap optical materials by multiple subpicosecond laser pulses,” J. Appl. Phys. 108(4), 043523 (2010). [CrossRef]
14. M. Mero, J. Liu, W. Rudolph, D. Ristau, and K. Starke, “Scaling laws of femtosecond laser pulse induced breakdown in oxide films,” Phys. Rev. B 71(11), 115109 (2005). [CrossRef]
15. H. Krol, L. Gallais, C. Grezes-Besset, J. Y. Natoli, and M. Commandre, “Investigation of nanoprecursors threshold distribution in laser-damage testing,” Opt. Commun. 256(1-3), 184–189 (2005). [CrossRef]
17. W. W. Liu, C. Y. Wei, S. L. Chen, Z. Fang, K. Yi, and J. D. Shao, “Statistical study of single and multiple pulse laser-induced damages of HfO2/SiO2 AR coatings at 1064nm ,” Opt. Commun. 301–301, 12–18 (2013).
18. Y. G. Shan, H. B. He, Y. Wang, X. Li, D. W. Li, and Y. A. Zhao, “Electrical field enhancement and laser damage growth in high-reflective coatings at 1064 nm,” Opt. Commun. 284(2), 625–629 (2011). [CrossRef]
19. S. Papernov and A. W. Schmid, “Two mechanisms of crater formation in ultraviolet-pulsed-laser irradiated SiO2 thin films with artificial defects,” J. Appl. Phys. 97(11), 114906 (2005). [CrossRef]