5216
Research Article
Vol. 54, No. 16 / June 1 2015 / Applied Optics
Continuous-wave laser damage and conditioning of particle contaminated optics ANDREW BROWN,1 ALBERT OGLOZA,2 LUCAS TAYLOR,1 JEFF THOMAS,3
AND
JOSEPH TALGHADER1,*
1
Electrical Engineering, University of Minnesota, Minneapolis, Minnesota 55455, USA Naval Postgraduate School, 1 University Cir., Monterey, California 93943, USA 3 Electro Optics Center, Pennsylvania State University, 222 Northpointe Blvd., Freeport, Pennsylvania 16229, USA *Corresponding author: [email protected] 2
Received 15 January 2015; revised 11 May 2015; accepted 12 May 2015; posted 12 May 2015 (Doc. ID 232620); published 1 June 2015
This paper describes the physical processes that occur when high-power continuous-wave laser light interacts with absorbing particles on a low-absorption optical surface. When a particulate-contaminated surface is illuminated by high-power continuous-wave laser light, a short burst of light is emitted from the surface, and the particles rapidly heat over a period of milliseconds to thousands of degrees Celsius, migrating over and evaporating from the surface. The surviving particles tend to coalesce into larger ones and leave a relatively flat residue on the surface. The total volume of the material on the surface has decreased dramatically. The optical surface itself heats substantially during illumination, but the surface temperature can decrease as the material is evaporated. Optical surfaces that survive this process without catastrophic damage are found to be more resistant to laser damage than surfaces that have not undergone the process. The surface temperature of the conditioned surfaces under illumination is lower than that of unconditioned surfaces. These conditioning effects on particles occurred within the first 30 s of laser exposure, with subsequent laser shots not affecting particle distributions. High-speed photography showed the actual removal and agglomeration of individual particles to occur within about 0.7 ms. Elemental changes were measured using time-of-flight secondary ion mass spectroscopy, with conditioned residuals being higher in hydrocarbon content than pristine particles. The tests in this study were conducted on high-reflectivity distributed Bragg reflector coated optics with carbon microparticles in the size range of 20–50 μm, gold particles of size 250 nm, and silica 1 μm in size. © 2015 Optical Society of America OCIS codes: (140.3330) Laser damage; (140.3440) Laser-induced breakdown; (140.6810) Thermal effects. http://dx.doi.org/10.1364/AO.54.005216
1. INTRODUCTION More advanced and efficient continuous-wave (CW) lasers are bringing a renewal of interest in directed energy systems. Critical to the performance of these systems are the optics required for beam shaping and steering. These optics must operate under large irradiances on to the order of tens to hundreds of MWcm−2 and with very low absorption to prevent excessive heating and damage from occurring due to high operating power. While maintaining clean, low-absorption optics in controlled laboratory settings is feasible, high-energy-laser (HEL) systems are increasingly being called upon to perform in harsh environments where contamination is all but guaranteed. Unfortunately environmental contamination often increases absorption to the point where particle-induced laser damage creates a point of failure. When failure occurs it is unpredictable, with seemingly intact optics failing catastrophically without warning. In order for CW HEL systems to operate reliably in real-world conditions, particle-induced laser damage must be understood and methods developed to negate its effects. 1559-128X/15/165216-07$15/0$15.00 © 2015 Optical Society of America
CW laser damage of optics has been a known problem and area of research for several decades. The development of highpower chemical lasers in the 1970s and 1980s drove early research in the field. It was quickly recognized that the physical damage mechanisms of CW lasers were fundamentally different than that of pulsed lasers. Damage effects from ultrashort pulsed lasers are attributed to field-mediated effects whereas the longer operating time scales of CW lasers give rise to thermal effects causing localized melting and material delamination [1,2]. Microcracks created by thermal stress and shock damage are also considered as another source of damage [3]. While there exists a body of literature for CW damage of pristine optics, there is comparatively little work done for nonideal films with defects and less for contamination. Research for the Airborne Laser System (ABL) found that defects dominated damage thresholds with defective areas failing at less than a quarter of the normal damage irradiance [4]. Exposing optical components to irradiances beneath the damage threshold has been noticed to have a beneficial effect, conditioning intrinsic
Research Article film defects to survive higher irradiances [5], though to our knowledge this has not been investigated for environmental contaminants. Laser cleaning of particulates has been widely reported for pulsed lasers [6,7], though not for CW sources. A summary of the Boulder Damage Symposium states that real-time laser cleaning has been observed; however, it does not reference the sources [8]. The relatively weak bonding of particulate contamination to the surface of an optic has been assumed to reduce thermal coupling, negating particulate contamination effects especially in comparison to embedded particles and defects [8]. Though thermal coupling may be weaker for surface contamination, this study explores this type of contamination. This paper addresses the new case of particulate contamination and laser conditioning on the particles. In contrast with previous assumptions [8], particulate contamination is found to strongly reduce damage thresholds. CW laser conditioning occurring during an optic’s first exposure is found to greatly increase damage thresholds and reduce heating of contaminated optics. The physical transformation of particles due to conditioning is reported as well. Conditioning effects were limited to the first moment of an optic’s exposure with subsequent shots doing little to further particle transformation. 2. EXPERIMENTAL SETUP Laser damage testing was conducted at Penn State’s ElectroOptic Center (EOC) where a 17 kW CW ytterbium-doped IPG Photonics YLS-1700 fiber laser was used to test samples. The output of the laser was approximately Gaussian and was focused to produce a 1 mm spot size. Output irradiances used for testing ranged from 60 kWcm−2 to 3 MWcm−2 for the 1 mm spot size. All shots were 30 s in duration except when damage occurred. When damage occurred the laser was shut off manually to prevent the complete loss of the sample and melting of the aluminum optical mount, which could occur quickly due to light scattered from the damaged sample. The laser output wavelength was 1070 nm. The temperature at the surface of the optic was gathered during laser irradiation using a thermal camera with a range 0°C–250°C at 30 Hz. The spatial resolution of the thermal camera was approximately 1 mm with the maximum pixel temperature used to measure the temperature within the 1 mm beam spot. Temperatures measured were an average of the optic’s surface and any contamination. The temperature resolution of the thermal camera was a tenth of a degree. The optics tested are near-perfect blackbodies in the thermal IR so camera error is minimal [9]. Microscope images of the surface of the optic were collected before and after each laser shot using a 200× USB digital microscope. Further testing was done using a 175 W Nd:YAG laser and a highspeed camera with smaller spot sizes (200 μm) to produce similar irradiance levels to those used at Penn State’s EOC. The optics tested were high-reflectivity distributed Bragg reflectors (DBRs) fabricated by ATFilms on 1 0 0 fused silica substrates. Both hafnia/silica (HfO2 ∕SiO2 ) and tantala/silica (Ta2 O5 ∕SiO2 ) DBRs were tested. The hafnia/silica DBRs consisted of 89 layers with the top and bottom being silica. Tantala/silica DBRs consisted of 40 layers, the bottom layer being silica and the top tantala. The DBRs were optimized
Vol. 54, No. 16 / June 1 2015 / Applied Optics
5217
for maximum reflectivity at 1064 nm. The absorption of the DBRs was measured using hotothermal common-path interferometry (PCI). Hafnia/silica samples were measured at less than 7 ppm, and tantala/silica at less than 1 ppm. Clean DBR films were tested along with samples that were intentionally contaminated with 20–50 μm carbon particles. Carbon was selected to simulate real-world contaminants due to its large absorption and prevalence in organics. To contaminate the samples, pure carbon powder from SkySpring Nanomaterials, Inc. was mixed into a dilute solution with isopropyl alcohol (IPA) and dripped onto the optic until the entire surface was covered. The particles were allowed to settle for 2 min, and then the excess was blown off the surface with compressed nitrogen. Blowing the surface off before the alcohol was allowed to dry prevented particles from agglomerating. The density of particles varied across each optic with an average density of 130 particles per square millimeter. Initial testing with a PCI system has measured the average absorption of a 50 μm spot with carbon particles to be 100 ppm with a standard deviation of 140 ppm. Peak absorption values approached 2000 ppm. To maximize the use of samples, up to nine locations per 1 0 0 sample were tested. These locations were spaced 4 mm apart to prevent any crossover effects from other tests. Locations were not reused for multiple shots unless specifically to examine conditioning effects due to previous laser irradiation. 3. THERMAL STUDIES AND DAMAGE THRESHOLDS Our first laser damage studies used optics contaminated with gold and silica particles. Most of the incident laser power was reflected or scattered from the surface with relatively little heating or change of the damage thresholds of the optics. The scattering was observed to heat the aluminum optical mount by a greater amount than the surface heating caused by the absorption of the actual laser spot on the surface of the optic. To test effects of the absorbing contamination, pristine DBR coated optics were contaminated with carbon microparticles and were tested over a range of irradiances while surface temperature was measured. In every case, the samples contaminated with carbon experienced greater heating than clean samples for a given irradiance as seen for hafnia/silica in Fig. 1. This was to be expected given the added absorption of the optic due to the carbon’s presence. In general, higher irradiances caused greater sample heating, although sample-to-sample variations in the amount of carbon present on the surface affected this. A visible flash of light could be seen coming from the surface of the optic during the first instant of laser exposure, which will be discussed later in the paper. To verify that the differences in temperatures of contaminated optics were due to the density of contamination, contaminated samples were tested at multiple locations. The samples were then cleaned via drag wiping using IPA to remove most of the contaminants. Figure 2 shows microscope images of an optic’s surface before and after this cleaning. New locations on the cleaned samples were tested using the same irradiances. A large reduction in heating was noticed after cleaning as seen in Fig. 3. In addition to greater surface heating, failures occurred with contaminated optics that did not happen with pristine and
5218
Vol. 54, No. 16 / June 1 2015 / Applied Optics
Fig. 1. Maximum hafnia/silica sample temperatures reached during testing for clean optics (red triangles) and carbon-contaminated optics (green squares). Multiple locations on a pristine and contaminated optic were tested at different irradiances. Notice the greater surface heating experienced by the optic with carbon contamination.
drag-wipe cleaned optics. Of the contaminated hafnia/silica DBRs, a single failure of occurred at 1 MWcm−2 out of seven samples. Other testing locations on multiple samples survived as high as 3 MWcm−2 . Tantala/silica DBRs had a far lower contaminated damage threshold with the lowest failing at 60 kWcm−2 and 80% failing by 180 kWcm−2 . No pristine DBRs of either type were damaged during the course of testing despite being irradiated up to 3 MWcm−2 . Optics that had been drag wiped showed similar damage resistance to those of pristine optics. No tantalia/silica optics that had been cleaned failed despite being tested at 180 kWcm−2 . The smaller damage threshold of contaminated tantala/silica is in contrast with heating and absorption of the pristine films. In testing at 3 MWcm−2 hafnia/silica DBRs experienced heating of 42.6°C while tantala-silica DBRs only heated 1.8°C. At
Fig. 2. Carbon contamination on an optic (a) before and (b) after drag wiping.
Research Article
Fig. 3. Maximum temperatures observed on contaminated (green squares) and cleaned (red triangles) hafnia/silica DBR samples. Freshly contaminated locations were tested and compared to tests of locations that had been contaminated then cleaned via drag wiping. The removal of most of the contaminants by drag wiping causes a large reduction in absorption and heating.
this point any conclusions from this observation of a lower damage threshold in contaminated tantala/silica are premature and tangential to this paper, but we note that tantala has a significantly smaller bandgap relative to hafnia, which may provide a starting point for future investigations. When damage occurred, its onset was at the start of each test. No optics that survived the initial exposure failed later in the 30 s duration of the test. An exception to this was an optic that an airborne particle of dust was seen to settle on during the test. Once the dust particle had landed, the optic failed immediately. Once damage to an optic began, failure was catastrophic with the laser boring a hole millimeters deep within seconds. To test for conditioning effects, samples were tested twice at the same location and irradiance, and the two shots compared. This was done for both hafnia/silica and tantala/silica DBRs over a range of irradiances. A reduction in surface heating from the first to second shot was immediately apparent for contaminated samples as seen in Fig. 4. PCI tests of conditioned areas measured a reduction in absorption from approximately 100– 50 ppm showing the absorption of the sample had decreased from the first shot. This heating reduction was not present in the clean optics tested, implying that conditioning was affecting the particles and not the films. Reduction in heating due to sample conditioning was also noticed in the temperature versus time profiles collected during testing. Freshly contaminated samples would often reach an initial high temperature and then begin to cool toward a lower steady state value, all while constantly being irradiated. Conditioning occurred within the first milliseconds of the test, reducing absorption and allowing the optic to stabilize at a new reduced temperature. Figure 5 shows optics conditioned at a higher power irradiance show a greater difference between the initial high peak temperature and the conditioned steady state value. This real-time heating reduction was not seen in subsequent shots of optics that had already been conditioned. Conditioning was seen to have a strong effect on damage thresholds. Contaminated tantala/silica was conditioned over
Research Article
Fig. 4. Reduction in operating temperature between first (red triangles) and second shots (blue x’s) on contaminated optics. Test locations on carbon-contaminated samples were tested twice at a given irradiance with the maximum temperature obtained during the two 30 s tests recorded. Pairs of tests can be seen for each irradiance with the first shot reaching a higher surface temperature than the second shot. Conditioning occurring during the first shots reduced absorption and caused second shots to run cooler.
Vol. 54, No. 16 / June 1 2015 / Applied Optics
5219
Fig. 5. Temperature versus time of two repeat shots of carboncontaminated optics. Locations on carbon-contaminated optics were tested twice at a given irradiance. The surface temperature versus time was recorded for both shots. The first shot (red solid line) is seen to run at higher temperatures than the second shot (blue dashed line), particularly at the larger irradiance. Real-time conditioning can be seen as the first shot reaches an initial high temperature before conditioning reduces absorption and the optic begins to stabilize at a lower steady state value.
a series of nine shots to withstand 2000 kWcm−2 as compared to its normal first-shot damage threshold of under 180 kWcm−2 . In all the testing that occurred, no damage ever occurred at an irradiance that a sample had already once survived. 4. PARTICLE STUDIES Microscope surface images of the optics tested were taken before and after each laser shot. Changes in the position, size, and number of particles were noticed and the images were analyzed using a MATLAB script to quantify this change. When a sample was first exposed, many particles were removed from within the spot size of the beam. The particles that remained were transformed and agglomerated within the beam spot as seen in Fig. 6. Virtually all particle removal and agglomeration occurred on the first shot, with subsequent shots causing little to no change, as shown in Fig. 7. To find the minimum irradiance required to transform particles, the radius at which particles ceased to be removed or agglomerated was compared to the beam profile of the laser. For hafnia/silica samples the threshold was 17.7 kWcm−2 and for tantala/silica it was 15.1 kWcm−2 . Knowing that conditioning effects were occurring within the first laser shot, a high-speed camera was used to film
Fig. 6. Images (a) before and (b) after laser conditioning. The circular pattern of particle removal with some newly formed large particles is typical of the carbon microparticle removal and agglomeration process seen.
5220
Vol. 54, No. 16 / June 1 2015 / Applied Optics
Research Article
Fig. 8. High-speed photo images of carbon particle transformation process from top left: 0.17 ms; top right: 0.24 ms; bottom left: 0.6 ms; bottom right: 0.76 ms.
Fig. 7. Total histogram counts of carbon microparticles from seven tests with two laser shots. The number of small particles has decreased and large particles increased, corresponding to particle removal and agglomeration. Virtually all particle changes occur within the first test, with the second test doing little to change the distribution.
particles during their initial exposure to the laser. Particles were seen to brightly glow and move around the surface for 0.7 ms into the laser exposure seen in still images taken from the highspeed camera in Fig. 8. After this time the particles ceased to glow and had stabilized into their final agglomerated position. Particle motion appears to be random, although the limited number of camera frames capturing this motion makes it difficult to determine. To test if contaminant mass was being evaporated off the surface, the volume of surface contaminants before and after laser irradiation was calculated using microscope surface images and surface profilometry. Images taken before testing were analyzed using a MATLAB script to count particles and measure their area. The volume of the particles was calculated from their area using a spherical approximation. This approximation was justified as the sphericity of the pristine particles used to contaminate the samples was measured with a mean of 89.5% and a standard deviation of 6.7% using a Microtrac Bluewave laser diffraction particle analyzer. After laser exposure, particles had drastically changed shape, flattening and broadening among other less consistent changes, and volume could not be determined from the images taken. To measure the final volume of particles, three-dimensional scans were taken using a P-16 surface profilometer, as shown in Fig. 9, with data taken every 8 μm × 0.5 μm. Of the samples measured, the remaining volumes were all significantly reduced from the originals, with the remaining volumes varying between 3% and 76% of the
originals. The residue was mechanically fixed to the surface and did not move from the 1 μN force exerted by the profilometer tip. Carbon particles that had not undergone conditioning were more weakly bonded to the surface and could be moved by the profilometer tip. When viewed in a bright field microscope, the conditioned particles were noticed to have changed transparency, as seen in Fig. 10. To test for elemental changes, a conditioned sample was analyzed by two methods. Energy-dispersive x-ray spectroscopy was attempted, but static charging from the electron gun of the scanning electron microscope and the insulating nature of the sample prevented a clear image of the sample from being taken. Regions of interest could not be resolved, so a new measuring technique was required. Time-of-flight secondary ion mass spectroscopy (TOF-SIMS) was used. This measurement technique can analyze small areas, allowing elemental composition of a specific region to be gathered. Regions of interest were selected via an optical microscope and could be clearly resolved. Conditioned spots were measured with a higher hydrocarbon signal in comparison to the pristine carbon particles. To see if heating alone could cause similar particle transformation, pristine particles were heated in a rapid thermal
Fig. 9. Surface profilometer scan of a conditioned optic with contaminant residue. The residue covers a wider, shallower area than the carbon particles before laser irradiation. Surface scans were used to calculate the final volume of contaminant residue after laser conditioning.
Research Article
Fig. 10. Pristine particles on the right and conditioned particles on the left. The transparency change of the particles is typical of all laser shots.
annealing (RTA) system to 1100°C to see if similar conditioning effects could be achieved. No such changes were measured, implying that the particles are reaching higher temperatures during testing. 5. DISCUSSION From the studies conducted, one can describe the processes that take place when particles are irradiated and conditioned. Gold and silica particles absorb very little, and most incoming photons are reflected from the surface. With little additional absorption from the particles, surface heating of the optic is limited, and the damage threshold remains mostly unaffected. Contaminants such as these do not pose a high a risk to the optic itself, but the reflected light can heat optical mounts and damage rubber seals used to keep contaminants out of laser systems. In one test of a failed optic, a significant portion of the incident power was scattered from the damage spot on the optic’s surface onto the aluminum mount. Before the laser was shut down, the aluminum mount was heated past its melting point and destroyed. In preliminary tests, similar results were seen for sea salt contamination, with optical mounts heating more than the optic surface. In contrast to gold and silica particles, carbon particles strongly absorb incident photons, and large amounts of heat were generated. The density of the carbon contamination directly affects the amount of heating. This was verified by testing samples before and after drag wiping. Drag wiping was highly effective at removing carbon contamination and greatly reduced heating. The greater surface heating caused by contamination correlated to a reduction in damage thresholds for both hafnia/silica and tantala/silica DBRs. Central to the question of conditioning versus damage is what happens within the first milliseconds of laser exposure to a freshly contaminated sample. From the temperature profiles of contaminated samples it is clear that conditioning takes place within several camera frames of a sample’s first exposure to the laser. Video taken from the high-speed camera verifies the flash of carbon evaporating from the surface to take place in under 0.7 ms. Damage to samples also occurred within the first moments of testing. With both conditioning and damage occurring within this short initial time interval, the survival of
Vol. 54, No. 16 / June 1 2015 / Applied Optics
5221
an optic is determined before observers are even aware that the test has begun. When a laser first irradiates the sample, carbon particles quickly absorb the incoming energy. Heat transfer by conduction into the substrate is negligible in comparison to the incoming power, and the particles rapidly gain temperature. Within a millisecond, particles in the beam spot have reach thousands of degrees Celsius and can be seen in high-speed camera video moving and evaporating in a visible flash of light. Tests conducted with RTA at 1100°C were insufficient to cause similar conditioning effects, implying that particle temperatures during laser irradiation are far higher. Surface images and profile surface scans verify that during the evaporation process a significant amount of particle volume is removed from the surface. The remaining particles migrate on the surface and coalesce into large, flat, semi-transparent residues that are more firmly bonded to the surface than the original particles. TOF-SIMS analysis shows the residue to be higher in hydrocarbon content than the original carbon particles. What causes this particular result and what its impact may be are unknown and are topics for future research. The energy that is initially absorbed by the surface is determined by the combination of incident irradiance and contaminant density. If the energy absorbed is high enough and the thermal coupling from the contamination to the film is sufficient, a critical point is reached, and the energy transferred into the optic drastically increases regardless of ongoing particle conditioning, causing catastrophic failure. If the critical point is not reached by the time conditioning reduces absorption, the optic begins to cool and stabilize at a new steady state temperature beneath the initial peak temperature. Repeated shots to an already conditioned optic cause less heating because the absorption has already decreased. The reduction in operating temperature for shots after conditioning was observed in both maximum sample temperature as well as in the temperature versus time profiles gathered. Surface images taken during testing showed conditioned residue to be unaffected by repeated exposure. In CW laser systems where contamination is possible, it would be very beneficial to ramp or step up power levels before operating at higher power levels. This intentional conditioning will remove many contaminants and reduce the total absorption of the optics in the system before irradiances reaches levels that could cause catastrophic failures of unconditioned optics. This sort of conditioning was used on a contaminated tantala/ silica DBR to raise the damage threshold more than an order of magnitude from below 180 kWcm−2 to over 2 MWcm−2 . 6. CONCLUSION Carbon contamination of high-reflectivity DBR-coated optics was seen to increase sample heating and decrease damage thresholds under high-power CW laser irradiation. Real time conditioning of carbon contaminants was observed in a reduction in sample temperature during the first exposure, and reduced steady state temperatures for following shots. Highspeed video verified that particle transformations occurred within 0.7 ms during first exposure. Particles were removed from the surface of the optic by the laser with remaining particles becoming agglomerated into a relatively flat residue.
5222
Vol. 54, No. 16 / June 1 2015 / Applied Optics
Subsequent exposures to the laser did little to change the particle distribution. Conditioned particle residue appeared optically different and was measured to have a higher hydrocarbon content than nonconditioned carbon particles. Office of Naval Research (ONR) (N00014-12-1-1030). We would like to gratefully acknowledge Penn State’s EOC for access to their high-power fiber laser. We also thank ATFilms for providing all the high-reflectivity optics tested. Characterization of samples and carbon particles took place using the University of Minnesota Nanofabrication Center. TOF-SIMS elemental data was gathered by the Evans Analytical Group. REFERENCES 1. B. C. Stuart, M. D. Feit, S. Herman, A. M. Rubenchik, B. W. Shore, and M. D. Perry, “Nanosecond-to-femtosecond laser-induced breakdown in dielectrics,” Phys. Rev. B 53, 1749–1761 (1996).
Research Article 2. J. Palmer, “Continuous wave laser damage on optical components,” Opt. Eng. 22, 435–446 (1983). 3. H. Gong, C. Li, and Z. Li, “CW-laser-induced thermal and mechanical damage in optical materials,” Proc. SPIE 3578, 576–583 (1999). 4. A. Stewart, L. Bonsall, J. Bettis, J. Copland, K. Healey, G. Charlton, W. Hughes, and J. C. Echeverry, “Laser damage in multispectral optical coatings for the ABL,” Proc. SPIE 3578, 162–171 (1998). 5. H. Bercegol, “What is laser conditioning? A review focused on dielectric multilayers,” Proc. SPIE 3578, 421–426 (1998). 6. S. Shukla, S. Kudryashov, K. Lyon, and S. Allen, “Pulsed laser cleaning of sub- and micron-size contaminant particles from optical surfaces: cleaning versus ablation and damage,” Proc. SPIE 5991, 59910N (2005). 7. F. Y. Genin, K. Michlitsch, J. Furr, M. R. Kozlowski, and P. Krulevitch, “Laser-induced damage of fused silica at 355 and 1064 nm initiated at aluminum contamination particles on the surface,” Proc. SPIE 2966, 126–138 (1997). 8. R. Shah, J. Rey, and A. Stewart, “Limits of performance: CW laser damage,” Proc. SPIE 6403, 640305 (2007). 9. G. Cleek, “The optical constants of some oxide glasses in the strong absorption region,” Appl. Opt. 5, 771–775 (1966).
Research Article
Vol. 54, No. 16 / June 1 2015 / Applied Optics
Continuous-wave laser damage and conditioning of particle contaminated optics ANDREW BROWN,1 ALBERT OGLOZA,2 LUCAS TAYLOR,1 JEFF THOMAS,3
AND
JOSEPH TALGHADER1,*
1
Electrical Engineering, University of Minnesota, Minneapolis, Minnesota 55455, USA Naval Postgraduate School, 1 University Cir., Monterey, California 93943, USA 3 Electro Optics Center, Pennsylvania State University, 222 Northpointe Blvd., Freeport, Pennsylvania 16229, USA *Corresponding author: [email protected] 2
Received 15 January 2015; revised 11 May 2015; accepted 12 May 2015; posted 12 May 2015 (Doc. ID 232620); published 1 June 2015
This paper describes the physical processes that occur when high-power continuous-wave laser light interacts with absorbing particles on a low-absorption optical surface. When a particulate-contaminated surface is illuminated by high-power continuous-wave laser light, a short burst of light is emitted from the surface, and the particles rapidly heat over a period of milliseconds to thousands of degrees Celsius, migrating over and evaporating from the surface. The surviving particles tend to coalesce into larger ones and leave a relatively flat residue on the surface. The total volume of the material on the surface has decreased dramatically. The optical surface itself heats substantially during illumination, but the surface temperature can decrease as the material is evaporated. Optical surfaces that survive this process without catastrophic damage are found to be more resistant to laser damage than surfaces that have not undergone the process. The surface temperature of the conditioned surfaces under illumination is lower than that of unconditioned surfaces. These conditioning effects on particles occurred within the first 30 s of laser exposure, with subsequent laser shots not affecting particle distributions. High-speed photography showed the actual removal and agglomeration of individual particles to occur within about 0.7 ms. Elemental changes were measured using time-of-flight secondary ion mass spectroscopy, with conditioned residuals being higher in hydrocarbon content than pristine particles. The tests in this study were conducted on high-reflectivity distributed Bragg reflector coated optics with carbon microparticles in the size range of 20–50 μm, gold particles of size 250 nm, and silica 1 μm in size. © 2015 Optical Society of America OCIS codes: (140.3330) Laser damage; (140.3440) Laser-induced breakdown; (140.6810) Thermal effects. http://dx.doi.org/10.1364/AO.54.005216
1. INTRODUCTION More advanced and efficient continuous-wave (CW) lasers are bringing a renewal of interest in directed energy systems. Critical to the performance of these systems are the optics required for beam shaping and steering. These optics must operate under large irradiances on to the order of tens to hundreds of MWcm−2 and with very low absorption to prevent excessive heating and damage from occurring due to high operating power. While maintaining clean, low-absorption optics in controlled laboratory settings is feasible, high-energy-laser (HEL) systems are increasingly being called upon to perform in harsh environments where contamination is all but guaranteed. Unfortunately environmental contamination often increases absorption to the point where particle-induced laser damage creates a point of failure. When failure occurs it is unpredictable, with seemingly intact optics failing catastrophically without warning. In order for CW HEL systems to operate reliably in real-world conditions, particle-induced laser damage must be understood and methods developed to negate its effects. 1559-128X/15/165216-07$15/0$15.00 © 2015 Optical Society of America
CW laser damage of optics has been a known problem and area of research for several decades. The development of highpower chemical lasers in the 1970s and 1980s drove early research in the field. It was quickly recognized that the physical damage mechanisms of CW lasers were fundamentally different than that of pulsed lasers. Damage effects from ultrashort pulsed lasers are attributed to field-mediated effects whereas the longer operating time scales of CW lasers give rise to thermal effects causing localized melting and material delamination [1,2]. Microcracks created by thermal stress and shock damage are also considered as another source of damage [3]. While there exists a body of literature for CW damage of pristine optics, there is comparatively little work done for nonideal films with defects and less for contamination. Research for the Airborne Laser System (ABL) found that defects dominated damage thresholds with defective areas failing at less than a quarter of the normal damage irradiance [4]. Exposing optical components to irradiances beneath the damage threshold has been noticed to have a beneficial effect, conditioning intrinsic
Research Article film defects to survive higher irradiances [5], though to our knowledge this has not been investigated for environmental contaminants. Laser cleaning of particulates has been widely reported for pulsed lasers [6,7], though not for CW sources. A summary of the Boulder Damage Symposium states that real-time laser cleaning has been observed; however, it does not reference the sources [8]. The relatively weak bonding of particulate contamination to the surface of an optic has been assumed to reduce thermal coupling, negating particulate contamination effects especially in comparison to embedded particles and defects [8]. Though thermal coupling may be weaker for surface contamination, this study explores this type of contamination. This paper addresses the new case of particulate contamination and laser conditioning on the particles. In contrast with previous assumptions [8], particulate contamination is found to strongly reduce damage thresholds. CW laser conditioning occurring during an optic’s first exposure is found to greatly increase damage thresholds and reduce heating of contaminated optics. The physical transformation of particles due to conditioning is reported as well. Conditioning effects were limited to the first moment of an optic’s exposure with subsequent shots doing little to further particle transformation. 2. EXPERIMENTAL SETUP Laser damage testing was conducted at Penn State’s ElectroOptic Center (EOC) where a 17 kW CW ytterbium-doped IPG Photonics YLS-1700 fiber laser was used to test samples. The output of the laser was approximately Gaussian and was focused to produce a 1 mm spot size. Output irradiances used for testing ranged from 60 kWcm−2 to 3 MWcm−2 for the 1 mm spot size. All shots were 30 s in duration except when damage occurred. When damage occurred the laser was shut off manually to prevent the complete loss of the sample and melting of the aluminum optical mount, which could occur quickly due to light scattered from the damaged sample. The laser output wavelength was 1070 nm. The temperature at the surface of the optic was gathered during laser irradiation using a thermal camera with a range 0°C–250°C at 30 Hz. The spatial resolution of the thermal camera was approximately 1 mm with the maximum pixel temperature used to measure the temperature within the 1 mm beam spot. Temperatures measured were an average of the optic’s surface and any contamination. The temperature resolution of the thermal camera was a tenth of a degree. The optics tested are near-perfect blackbodies in the thermal IR so camera error is minimal [9]. Microscope images of the surface of the optic were collected before and after each laser shot using a 200× USB digital microscope. Further testing was done using a 175 W Nd:YAG laser and a highspeed camera with smaller spot sizes (200 μm) to produce similar irradiance levels to those used at Penn State’s EOC. The optics tested were high-reflectivity distributed Bragg reflectors (DBRs) fabricated by ATFilms on 1 0 0 fused silica substrates. Both hafnia/silica (HfO2 ∕SiO2 ) and tantala/silica (Ta2 O5 ∕SiO2 ) DBRs were tested. The hafnia/silica DBRs consisted of 89 layers with the top and bottom being silica. Tantala/silica DBRs consisted of 40 layers, the bottom layer being silica and the top tantala. The DBRs were optimized
Vol. 54, No. 16 / June 1 2015 / Applied Optics
5217
for maximum reflectivity at 1064 nm. The absorption of the DBRs was measured using hotothermal common-path interferometry (PCI). Hafnia/silica samples were measured at less than 7 ppm, and tantala/silica at less than 1 ppm. Clean DBR films were tested along with samples that were intentionally contaminated with 20–50 μm carbon particles. Carbon was selected to simulate real-world contaminants due to its large absorption and prevalence in organics. To contaminate the samples, pure carbon powder from SkySpring Nanomaterials, Inc. was mixed into a dilute solution with isopropyl alcohol (IPA) and dripped onto the optic until the entire surface was covered. The particles were allowed to settle for 2 min, and then the excess was blown off the surface with compressed nitrogen. Blowing the surface off before the alcohol was allowed to dry prevented particles from agglomerating. The density of particles varied across each optic with an average density of 130 particles per square millimeter. Initial testing with a PCI system has measured the average absorption of a 50 μm spot with carbon particles to be 100 ppm with a standard deviation of 140 ppm. Peak absorption values approached 2000 ppm. To maximize the use of samples, up to nine locations per 1 0 0 sample were tested. These locations were spaced 4 mm apart to prevent any crossover effects from other tests. Locations were not reused for multiple shots unless specifically to examine conditioning effects due to previous laser irradiation. 3. THERMAL STUDIES AND DAMAGE THRESHOLDS Our first laser damage studies used optics contaminated with gold and silica particles. Most of the incident laser power was reflected or scattered from the surface with relatively little heating or change of the damage thresholds of the optics. The scattering was observed to heat the aluminum optical mount by a greater amount than the surface heating caused by the absorption of the actual laser spot on the surface of the optic. To test effects of the absorbing contamination, pristine DBR coated optics were contaminated with carbon microparticles and were tested over a range of irradiances while surface temperature was measured. In every case, the samples contaminated with carbon experienced greater heating than clean samples for a given irradiance as seen for hafnia/silica in Fig. 1. This was to be expected given the added absorption of the optic due to the carbon’s presence. In general, higher irradiances caused greater sample heating, although sample-to-sample variations in the amount of carbon present on the surface affected this. A visible flash of light could be seen coming from the surface of the optic during the first instant of laser exposure, which will be discussed later in the paper. To verify that the differences in temperatures of contaminated optics were due to the density of contamination, contaminated samples were tested at multiple locations. The samples were then cleaned via drag wiping using IPA to remove most of the contaminants. Figure 2 shows microscope images of an optic’s surface before and after this cleaning. New locations on the cleaned samples were tested using the same irradiances. A large reduction in heating was noticed after cleaning as seen in Fig. 3. In addition to greater surface heating, failures occurred with contaminated optics that did not happen with pristine and
5218
Vol. 54, No. 16 / June 1 2015 / Applied Optics
Fig. 1. Maximum hafnia/silica sample temperatures reached during testing for clean optics (red triangles) and carbon-contaminated optics (green squares). Multiple locations on a pristine and contaminated optic were tested at different irradiances. Notice the greater surface heating experienced by the optic with carbon contamination.
drag-wipe cleaned optics. Of the contaminated hafnia/silica DBRs, a single failure of occurred at 1 MWcm−2 out of seven samples. Other testing locations on multiple samples survived as high as 3 MWcm−2 . Tantala/silica DBRs had a far lower contaminated damage threshold with the lowest failing at 60 kWcm−2 and 80% failing by 180 kWcm−2 . No pristine DBRs of either type were damaged during the course of testing despite being irradiated up to 3 MWcm−2 . Optics that had been drag wiped showed similar damage resistance to those of pristine optics. No tantalia/silica optics that had been cleaned failed despite being tested at 180 kWcm−2 . The smaller damage threshold of contaminated tantala/silica is in contrast with heating and absorption of the pristine films. In testing at 3 MWcm−2 hafnia/silica DBRs experienced heating of 42.6°C while tantala-silica DBRs only heated 1.8°C. At
Fig. 2. Carbon contamination on an optic (a) before and (b) after drag wiping.
Research Article
Fig. 3. Maximum temperatures observed on contaminated (green squares) and cleaned (red triangles) hafnia/silica DBR samples. Freshly contaminated locations were tested and compared to tests of locations that had been contaminated then cleaned via drag wiping. The removal of most of the contaminants by drag wiping causes a large reduction in absorption and heating.
this point any conclusions from this observation of a lower damage threshold in contaminated tantala/silica are premature and tangential to this paper, but we note that tantala has a significantly smaller bandgap relative to hafnia, which may provide a starting point for future investigations. When damage occurred, its onset was at the start of each test. No optics that survived the initial exposure failed later in the 30 s duration of the test. An exception to this was an optic that an airborne particle of dust was seen to settle on during the test. Once the dust particle had landed, the optic failed immediately. Once damage to an optic began, failure was catastrophic with the laser boring a hole millimeters deep within seconds. To test for conditioning effects, samples were tested twice at the same location and irradiance, and the two shots compared. This was done for both hafnia/silica and tantala/silica DBRs over a range of irradiances. A reduction in surface heating from the first to second shot was immediately apparent for contaminated samples as seen in Fig. 4. PCI tests of conditioned areas measured a reduction in absorption from approximately 100– 50 ppm showing the absorption of the sample had decreased from the first shot. This heating reduction was not present in the clean optics tested, implying that conditioning was affecting the particles and not the films. Reduction in heating due to sample conditioning was also noticed in the temperature versus time profiles collected during testing. Freshly contaminated samples would often reach an initial high temperature and then begin to cool toward a lower steady state value, all while constantly being irradiated. Conditioning occurred within the first milliseconds of the test, reducing absorption and allowing the optic to stabilize at a new reduced temperature. Figure 5 shows optics conditioned at a higher power irradiance show a greater difference between the initial high peak temperature and the conditioned steady state value. This real-time heating reduction was not seen in subsequent shots of optics that had already been conditioned. Conditioning was seen to have a strong effect on damage thresholds. Contaminated tantala/silica was conditioned over
Research Article
Fig. 4. Reduction in operating temperature between first (red triangles) and second shots (blue x’s) on contaminated optics. Test locations on carbon-contaminated samples were tested twice at a given irradiance with the maximum temperature obtained during the two 30 s tests recorded. Pairs of tests can be seen for each irradiance with the first shot reaching a higher surface temperature than the second shot. Conditioning occurring during the first shots reduced absorption and caused second shots to run cooler.
Vol. 54, No. 16 / June 1 2015 / Applied Optics
5219
Fig. 5. Temperature versus time of two repeat shots of carboncontaminated optics. Locations on carbon-contaminated optics were tested twice at a given irradiance. The surface temperature versus time was recorded for both shots. The first shot (red solid line) is seen to run at higher temperatures than the second shot (blue dashed line), particularly at the larger irradiance. Real-time conditioning can be seen as the first shot reaches an initial high temperature before conditioning reduces absorption and the optic begins to stabilize at a lower steady state value.
a series of nine shots to withstand 2000 kWcm−2 as compared to its normal first-shot damage threshold of under 180 kWcm−2 . In all the testing that occurred, no damage ever occurred at an irradiance that a sample had already once survived. 4. PARTICLE STUDIES Microscope surface images of the optics tested were taken before and after each laser shot. Changes in the position, size, and number of particles were noticed and the images were analyzed using a MATLAB script to quantify this change. When a sample was first exposed, many particles were removed from within the spot size of the beam. The particles that remained were transformed and agglomerated within the beam spot as seen in Fig. 6. Virtually all particle removal and agglomeration occurred on the first shot, with subsequent shots causing little to no change, as shown in Fig. 7. To find the minimum irradiance required to transform particles, the radius at which particles ceased to be removed or agglomerated was compared to the beam profile of the laser. For hafnia/silica samples the threshold was 17.7 kWcm−2 and for tantala/silica it was 15.1 kWcm−2 . Knowing that conditioning effects were occurring within the first laser shot, a high-speed camera was used to film
Fig. 6. Images (a) before and (b) after laser conditioning. The circular pattern of particle removal with some newly formed large particles is typical of the carbon microparticle removal and agglomeration process seen.
5220
Vol. 54, No. 16 / June 1 2015 / Applied Optics
Research Article
Fig. 8. High-speed photo images of carbon particle transformation process from top left: 0.17 ms; top right: 0.24 ms; bottom left: 0.6 ms; bottom right: 0.76 ms.
Fig. 7. Total histogram counts of carbon microparticles from seven tests with two laser shots. The number of small particles has decreased and large particles increased, corresponding to particle removal and agglomeration. Virtually all particle changes occur within the first test, with the second test doing little to change the distribution.
particles during their initial exposure to the laser. Particles were seen to brightly glow and move around the surface for 0.7 ms into the laser exposure seen in still images taken from the highspeed camera in Fig. 8. After this time the particles ceased to glow and had stabilized into their final agglomerated position. Particle motion appears to be random, although the limited number of camera frames capturing this motion makes it difficult to determine. To test if contaminant mass was being evaporated off the surface, the volume of surface contaminants before and after laser irradiation was calculated using microscope surface images and surface profilometry. Images taken before testing were analyzed using a MATLAB script to count particles and measure their area. The volume of the particles was calculated from their area using a spherical approximation. This approximation was justified as the sphericity of the pristine particles used to contaminate the samples was measured with a mean of 89.5% and a standard deviation of 6.7% using a Microtrac Bluewave laser diffraction particle analyzer. After laser exposure, particles had drastically changed shape, flattening and broadening among other less consistent changes, and volume could not be determined from the images taken. To measure the final volume of particles, three-dimensional scans were taken using a P-16 surface profilometer, as shown in Fig. 9, with data taken every 8 μm × 0.5 μm. Of the samples measured, the remaining volumes were all significantly reduced from the originals, with the remaining volumes varying between 3% and 76% of the
originals. The residue was mechanically fixed to the surface and did not move from the 1 μN force exerted by the profilometer tip. Carbon particles that had not undergone conditioning were more weakly bonded to the surface and could be moved by the profilometer tip. When viewed in a bright field microscope, the conditioned particles were noticed to have changed transparency, as seen in Fig. 10. To test for elemental changes, a conditioned sample was analyzed by two methods. Energy-dispersive x-ray spectroscopy was attempted, but static charging from the electron gun of the scanning electron microscope and the insulating nature of the sample prevented a clear image of the sample from being taken. Regions of interest could not be resolved, so a new measuring technique was required. Time-of-flight secondary ion mass spectroscopy (TOF-SIMS) was used. This measurement technique can analyze small areas, allowing elemental composition of a specific region to be gathered. Regions of interest were selected via an optical microscope and could be clearly resolved. Conditioned spots were measured with a higher hydrocarbon signal in comparison to the pristine carbon particles. To see if heating alone could cause similar particle transformation, pristine particles were heated in a rapid thermal
Fig. 9. Surface profilometer scan of a conditioned optic with contaminant residue. The residue covers a wider, shallower area than the carbon particles before laser irradiation. Surface scans were used to calculate the final volume of contaminant residue after laser conditioning.
Research Article
Fig. 10. Pristine particles on the right and conditioned particles on the left. The transparency change of the particles is typical of all laser shots.
annealing (RTA) system to 1100°C to see if similar conditioning effects could be achieved. No such changes were measured, implying that the particles are reaching higher temperatures during testing. 5. DISCUSSION From the studies conducted, one can describe the processes that take place when particles are irradiated and conditioned. Gold and silica particles absorb very little, and most incoming photons are reflected from the surface. With little additional absorption from the particles, surface heating of the optic is limited, and the damage threshold remains mostly unaffected. Contaminants such as these do not pose a high a risk to the optic itself, but the reflected light can heat optical mounts and damage rubber seals used to keep contaminants out of laser systems. In one test of a failed optic, a significant portion of the incident power was scattered from the damage spot on the optic’s surface onto the aluminum mount. Before the laser was shut down, the aluminum mount was heated past its melting point and destroyed. In preliminary tests, similar results were seen for sea salt contamination, with optical mounts heating more than the optic surface. In contrast to gold and silica particles, carbon particles strongly absorb incident photons, and large amounts of heat were generated. The density of the carbon contamination directly affects the amount of heating. This was verified by testing samples before and after drag wiping. Drag wiping was highly effective at removing carbon contamination and greatly reduced heating. The greater surface heating caused by contamination correlated to a reduction in damage thresholds for both hafnia/silica and tantala/silica DBRs. Central to the question of conditioning versus damage is what happens within the first milliseconds of laser exposure to a freshly contaminated sample. From the temperature profiles of contaminated samples it is clear that conditioning takes place within several camera frames of a sample’s first exposure to the laser. Video taken from the high-speed camera verifies the flash of carbon evaporating from the surface to take place in under 0.7 ms. Damage to samples also occurred within the first moments of testing. With both conditioning and damage occurring within this short initial time interval, the survival of
Vol. 54, No. 16 / June 1 2015 / Applied Optics
5221
an optic is determined before observers are even aware that the test has begun. When a laser first irradiates the sample, carbon particles quickly absorb the incoming energy. Heat transfer by conduction into the substrate is negligible in comparison to the incoming power, and the particles rapidly gain temperature. Within a millisecond, particles in the beam spot have reach thousands of degrees Celsius and can be seen in high-speed camera video moving and evaporating in a visible flash of light. Tests conducted with RTA at 1100°C were insufficient to cause similar conditioning effects, implying that particle temperatures during laser irradiation are far higher. Surface images and profile surface scans verify that during the evaporation process a significant amount of particle volume is removed from the surface. The remaining particles migrate on the surface and coalesce into large, flat, semi-transparent residues that are more firmly bonded to the surface than the original particles. TOF-SIMS analysis shows the residue to be higher in hydrocarbon content than the original carbon particles. What causes this particular result and what its impact may be are unknown and are topics for future research. The energy that is initially absorbed by the surface is determined by the combination of incident irradiance and contaminant density. If the energy absorbed is high enough and the thermal coupling from the contamination to the film is sufficient, a critical point is reached, and the energy transferred into the optic drastically increases regardless of ongoing particle conditioning, causing catastrophic failure. If the critical point is not reached by the time conditioning reduces absorption, the optic begins to cool and stabilize at a new steady state temperature beneath the initial peak temperature. Repeated shots to an already conditioned optic cause less heating because the absorption has already decreased. The reduction in operating temperature for shots after conditioning was observed in both maximum sample temperature as well as in the temperature versus time profiles gathered. Surface images taken during testing showed conditioned residue to be unaffected by repeated exposure. In CW laser systems where contamination is possible, it would be very beneficial to ramp or step up power levels before operating at higher power levels. This intentional conditioning will remove many contaminants and reduce the total absorption of the optics in the system before irradiances reaches levels that could cause catastrophic failures of unconditioned optics. This sort of conditioning was used on a contaminated tantala/ silica DBR to raise the damage threshold more than an order of magnitude from below 180 kWcm−2 to over 2 MWcm−2 . 6. CONCLUSION Carbon contamination of high-reflectivity DBR-coated optics was seen to increase sample heating and decrease damage thresholds under high-power CW laser irradiation. Real time conditioning of carbon contaminants was observed in a reduction in sample temperature during the first exposure, and reduced steady state temperatures for following shots. Highspeed video verified that particle transformations occurred within 0.7 ms during first exposure. Particles were removed from the surface of the optic by the laser with remaining particles becoming agglomerated into a relatively flat residue.
5222
Vol. 54, No. 16 / June 1 2015 / Applied Optics
Subsequent exposures to the laser did little to change the particle distribution. Conditioned particle residue appeared optically different and was measured to have a higher hydrocarbon content than nonconditioned carbon particles. Office of Naval Research (ONR) (N00014-12-1-1030). We would like to gratefully acknowledge Penn State’s EOC for access to their high-power fiber laser. We also thank ATFilms for providing all the high-reflectivity optics tested. Characterization of samples and carbon particles took place using the University of Minnesota Nanofabrication Center. TOF-SIMS elemental data was gathered by the Evans Analytical Group. REFERENCES 1. B. C. Stuart, M. D. Feit, S. Herman, A. M. Rubenchik, B. W. Shore, and M. D. Perry, “Nanosecond-to-femtosecond laser-induced breakdown in dielectrics,” Phys. Rev. B 53, 1749–1761 (1996).
Research Article 2. J. Palmer, “Continuous wave laser damage on optical components,” Opt. Eng. 22, 435–446 (1983). 3. H. Gong, C. Li, and Z. Li, “CW-laser-induced thermal and mechanical damage in optical materials,” Proc. SPIE 3578, 576–583 (1999). 4. A. Stewart, L. Bonsall, J. Bettis, J. Copland, K. Healey, G. Charlton, W. Hughes, and J. C. Echeverry, “Laser damage in multispectral optical coatings for the ABL,” Proc. SPIE 3578, 162–171 (1998). 5. H. Bercegol, “What is laser conditioning? A review focused on dielectric multilayers,” Proc. SPIE 3578, 421–426 (1998). 6. S. Shukla, S. Kudryashov, K. Lyon, and S. Allen, “Pulsed laser cleaning of sub- and micron-size contaminant particles from optical surfaces: cleaning versus ablation and damage,” Proc. SPIE 5991, 59910N (2005). 7. F. Y. Genin, K. Michlitsch, J. Furr, M. R. Kozlowski, and P. Krulevitch, “Laser-induced damage of fused silica at 355 and 1064 nm initiated at aluminum contamination particles on the surface,” Proc. SPIE 2966, 126–138 (1997). 8. R. Shah, J. Rey, and A. Stewart, “Limits of performance: CW laser damage,” Proc. SPIE 6403, 640305 (2007). 9. G. Cleek, “The optical constants of some oxide glasses in the strong absorption region,” Appl. Opt. 5, 771–775 (1966).
