UPDATE: Study of scattering points on LIGO mirrors Lamar Glover (Cal State LA) Riccardo DeSalvo (Cal State LA) Innocenzo Pinto (Unisannio) DCC Number G1600163 Special Thanks to Travis Sadecki and Rick Savage of LIGO Hanford for providing recent images
Brief Recap of Study Subject Matter LIGO mirrors present unexpectedly large scattering losses. Previously, measurements were made in situ by Bill Kells and on samples by Josh Smith at Cal State Fullerton and the Syracuse group. Newer photos were provided by Travis Sadecki and Rick Savage from LIGO Hanford with updated, better optimized equipment in Nov. 2015. (i.e. images from CCD cameras with higher resolution, made for infrared photography)
Brief Recap of Study Subject Matter (Cont’d) At Cal State LA, we are continuing our analysis of the scattering points detected in the Advanced LIGO mirrors In astronomy, looking where there appear to be no stars, the beginning of the universe was seen In LIGO mirrors, the problem of scatterers may be understood looking away from the large scatterers, in the depth of the coating layer stack
Brief Recap of Study Subject Matter (Cont’d) The largest scatterers are likely accidental Dirt on the mirror Impurity-induced defects during deposition A large number of scatterers were observed where none would be expected
Method Review Individual scatterers were identified using astronomical algorithms for stars in galaxies Extract apparent amplitude and position of each scatterer Fit the stored beam position and profile to determine the illumination power on each scatterer. Re-normalize the amplitude of each scatterer (dividing by the illumination power) to extract actual scattering power of each point.
Start with a low exposure image Inspect with increasing large exposure photo 0206 @ 0.000125 sec sum 1st Run: 101 scatterers 101 2nd Run: 17 scatterers 118 3rd Run: 16 scatterers 134 photo 0224 @ 0.01 sec 1st Run: 5,680 scatterers 5.680 2nd Run: 6,126 scatterers 11,806 photo 0225 @ 0.0125 sec 1st Run: 6,216 scatterers photo 0240 @ 0.4 sec (beware of saturation) 1st Run: 183,522 scatterers 183,522 2nd Run: 180,267 scatterers 363,789
Initializing the analysis software Low exposure. An obvious cluster is visible It is probably due to accidental sources (dirt?) Original Photo :
Daophot mechanism: Point Spread Templates “daophot” creates a point spread template from selected sources within the image. The template is applied to search scatterers in the image Magnitude data is derived for each identified source. Point Spread Template for an exposure time of 1.25 x 10-4 s
Types of Scatterers “Well-Mannered”: Used in template construction Few Not “Well-Mannered”: Could it be due to dirt buried into layers?
Identifying/subtracting scatterers Original Photo in “ds9” 256 100 Resulting Photo from first run of “daophot” package 256 Even at this low exposure few scatterers are VERY bright Saturate CCD After extracting the bright point sources, the pixel scale is expanded The gray CCD background becomes obvious. Black dots mark the place where bright scatterers have been excised “dirt” is ignored by daophot and is less bright
Pixel Intensity Pixel amplitude histogram of original Image 100 200 Residual pixel amplitude histogram after scatterer extraction In a good image Daophot is very effective in extracting scatterers Width of residuals (2-3 pixels FWHM) illustrates Daophot’s good resolution -50 50
Pixel Intensity @ Exposure Time = 1.25x10-4 sec 100 250 Pixel amplitude histogram of original Image Original Image 50 -50 Residual pixel amplitude after “daophot” Process Image after “daophot” Process
Pixel Intensity @ Exposure Time = 0.01 sec Pixel amplitude histogram of original Image Original Image 100 250 50 -50 Residual pixel amplitude after “daophot” Process Width of residuals widened by saturation Still a few pixels FWHM, + small tails Image after “daophot” Process
Point Spread Templates Run 1 Run 2 Run 3 T = 1.25 x 10-4 s Point Spread Template for different runs T = 0.01 s Template quality worsens with: Increased exposure time High run number T = 0.0125 s T = 0.4 s
Pixel Intensity @ Exposure Time = 0.0125 sec Pixel amplitude histogram of original Image Original Image 100 250 FWHM still good Tails grow with exposure and saturation 50 -50 Residual pixel amplitude after “daophot” Process Image after “daophot” Process
Efficiency of Daophot’s cycles 200 Large amplitude scatterers picked up first Weaker pixels plucked from background noise 100 Scatterer light intensity @ Exposure Time = 0.01 sec 2000 7000
Scatterer light intensity @ Exposure Time = 0.01 sec Adding up cycles 200 80 Run 1 Run 2 1000 10000 20000 2000 Run 1 + 2 10000 20000 100 5 1 Scatterer light intensity @ Exposure Time = 0.01 sec
Scatterer light intensity @ Exposure Time = 0.0125 sec 70 40 5000 20000
The origin of scatterer apparent sizes Local illumination level (correctable) Physical scatterer size inside host layer Depth of host layer inside dielectric coating stack (All scatterers are nanometer scale, diffraction limited)
1: Crystallite Formation in a single evaporated layer Growth direction Side View Top View Growth Direction As mirror layers are formed, crystallites appear at different depths. Those that appear earlier in the depositing process grow more Create larger scatterers
1: Crystallite Formation in a single evaporated layer Deposition starts as a glassy layer As layer grows, crystallites form in the glassy matrix. Slowly grow. Columnar growth ensues Dielectric mirror layers ~ 0.1 µm. Well inside glassy matrix region Where there “appear” to be NO crystallites Growth Direction Mannan.Ali@physics.org Chapter 4: Thin Film Deposition Web: http://members.xoom.com/MannansZone/thesis.html
Scattered light detected by CCD 2: Depth in stack effect Two depth attenuation components: Attenuation of impinging light due to reflections through depth of scattering layer Further attenuation from reflection of Scattered light Scattered light detected by CCD Complex effect to simulate
Analysis of spatial distribution From previous analysis of initial LIGO mirror (New mirror still to be completed) Uniform scatterer density Uniform scattering amplitude across radius
Number of Scatterers vs. Exposure Time photo 0206 @ 0.000125 sec 101 scatterers photo 0224 @ 0.01 sec 5,680 scatterers photo 0225 @ 0.0125 sec 6,216 scatterers photo 0240 @ 0.4 sec 183,522 scatterers (saturation) Number increases almost linearly with exposure Exploring smaller scatterers But also deeper in the dielectric coating layers ! !
Discussion of crystal number and distribution intensity How can so many scatterers exist, and be so uniformly distributed? Cannot be “dirt” ! Only a thermodynamical source like the nucleation theory can explain the observations Image: Old Ligo
Competing forces in nucleation Ordered volume inside crystal Disordered outside Inside crystallite bonds are formed Energy gain:
Competing forces in nucleation In disordered surface Many more bonds are frustrated than in the glassy matrix => Large Energy loss
Competing forces in nucleation This costs energy Here you gain energy Ordered volume Disordered surface Gain (+) Loss (-)
Grain Critical radius As a result of the competition between area ΔGs and volume ΔGv ΔG has a maximum Nucleation Energy ΔG* Crystallites have a critical size
Nucleation speed An activation energy means that the nucleation probability depends on the deposition temperature
Effects of nucleation size critical size Below r* there is energetic advantage to disappear Above r* there is energetic advantage to grow Nanolayering depresses number and growth of crystallites Nanolayering below r* may even inhibit crystallite generation
Is the surface of Crystallites the source of mechanical losses? Surface is most disordered More frustrated bonds exist on the surface than inside the glassy matrix ! ! Large probability of bonds flipping between energy states during vibrations Typical mechanical energy loss mechanism
Effects of depressing crystallites There is circumstantial experimental evidence that the surface of crystals is a site where large energy losses happens. Observed when over-annealing generates X-ray-detectable crystallites Reduced number and size of crystallite may reduce thermal noise, as well as light scatter.