June 1, 2006GWADW 2006 1

LIGO Core Optics:a decade of development and experience

W. Kells

LIGO Laboratory, Caltech

With acknowledgement of entire LIGO team for its optical development

June 1, 2006GWADW 2006 2

LIGO “Core” Optics

6 (4 test masses; splitter; recycling mirror) large optics which form high power cavities.» 11Kgm ( 25cm, h 10cm)» Low loss, low distortion fused silica

Designed (epoch ’94-97) to achieve science requirements: with 6Watt input

» Extensive simulations» Protracted “pathfinder” fabrication test pieces» Transition from 535 to 1064nm

– Valuable lessons learned from Caltech 40m prototype interferometer

June 1, 2006GWADW 2006 3

Major early concerns

Fabrication tolerances: match of optical modes ROC of mirrors arm imbalance: excessive “contrast defect” to dark port reflectivity, loss

» Coating stability and uniformity

Thermal lensing: effect on recycling cavity “point design”» Long term contamination build up on HR surfaces

» Uncertain residual Silica bulk absorption.

Static charging on suspended dielectric TMs Inherent unstable recycling cavity design

» Hypersensitivity to polish, coating, homogeneity errors

Effective loss of long cavities with figure distortions» Essential target of “FFT” studies

» Coupled with coating reflectivity tolerance: rifo >/< 0 (point design recycling)

June 1, 2006GWADW 2006 4

Optical Loss Expectations

Goal: 30 based on older polish/coating information Pathfinder development & fabrication proved much better:

» Micro roughness rms <0.28 nm prompt loss ~(rms)2 <10 ppm

» Super polished substrate 2 - 3x lower rms

Simulation (FFT) with Fab. Data: Figure= modal distortion Roughness= loss Low absorption= cold “start up” Witness sample reflectivity

Simulated G (at least: CR field not affected

by degenerate recycling) far exceeds goals Consistent with Advanced ligo requirements

CRRCG

H1 ETMy polished surface PSD

localized roughness

Global surface metrology

FFT mirror map

June 1, 2006GWADW 2006 5

Scatterometer studies

Observed interferometer gains lower than Sim. predictions.– Consistent with 50-70ppm avg. additional loss per TM.– Consistent with “visibilities” (resonant reflectivity defect) of individual arms

In situ studies: Some HR surfaces viewable @ 3 angles:

Angular dependence more isotropic,

“point like” than metrology prediction

In situ observed scatter ~70 ppm mirror ~same level, character for every TM

independent of history/cleaning.

FFT map representation

H1 ETMy

roughness

k-1

k-2

Scatterometer port 5.5 10-8 Sr

ITM Main arm beam

June 1, 2006GWADW 2006 6

In Situ Optics Performance

~41, which is:» Consistent with measured arm visibilities

» Consistent with total arm loss

dominated by prompt scatter.

» Scatterometer data extrapolated

to absolute loss

» Consistent with lower than anticipated

contrast defect ( and small FFT dependence on maps)

CAVITY V TITM TWITNESS Scatter

2k X .0222 .0277 .0283 0.85

2k Y .0211 .0272 .0281 7

4k X .0241 .0279 .0275 7.5

4k Y .0214 .0263 .028 8.8

CRRCG

Replaced ITM

June 1, 2006GWADW 2006 7

Homogeneous roughness ?

Expect isotropic glow from “homogeneous” polish roughness» Find: “point” defect scatter dominates

» Bench scans (1064nm) also show excess

Resonant arm, Gaussian illuminated ETM

Reference calibration:known cavity loss

Is it just dust ??

June 1, 2006GWADW 2006 8

Analysis of the “Globular Cluster”

Cleanest point scatter image: 2k ETMy:» Grab video stills for detailed analysis:

This point defect background~same for all optics. Diffuse (micro roughness) background contributes < 1/3 of total scatter. Other blemishes don’t dominate total (?) Puzzle: Why these point defects missed in Lab. QA?

Defocused Focused

June 1, 2006GWADW 2006 9

Coatings sensitive to handling For several years Hanford 2k performed poorly

» X arm visibility (resonant reflectivity) poor

» Ugly recycling cavity “mode” pattern

» Excess dark port contrast

» More dramatic: unlocked arm cavity

Found: AR coating anomaly» Hypothesis: extended harsh cleaning of

surfaces had etched coating layers.

Lesson: coating sensitivity to thickness

change (confirmed by model).

1r

Bench scan of removed ITM

June 1, 2006GWADW 2006 10

Contamination & thermal lensing

~7 years of installed Core optics» No evidence of accumulating contamination (scattering or absorbing)

– Routine full lock only ~5 yrs. High power only 1-2 yrs.– Some optics >6 yrs hanging have no evidence of HR absorption >1ppm (design)– Net scatter loss seems independent of TM installation epoch (though high !)

» Residual absorption has been found consistent with materials/Fab. expected.– As anticipated by simulations, this level essentially only affects SB fields– Bulk silica absorption not controlled sufficiently for “point” thermal design.– “TCS” system required for compensating residual variations.

This typical experience: extrapolates well to Adv. LIGO !» Outstanding discrepancy: installed TM scatter loss far too high

– Assumed either treatable “dust” issue; or adjustment of coating process

However also contamination accidents– High power operation revealed >10x residual coating absorption– Unique to pair of ITMs: no evidence in other Hanford optics. When ??

June 1, 2006GWADW 2006 11

Contamination in LIGO I TMs Goal: corroborate in situ performance with bench tests

» Many LIGO COC optics studied– Comparisons establish “typical” from anomalous

» Absolute calibration to various reference mirrors.

Components of “loss test” cavity

Example: What is anomalous contaminanton H1 ITMs?

Absorption is lumpy but not point like Scatter also anomalous and correlates well spatially with absorption Easily removed by surface cleansing Fine absorbing dust, sucked in during vent?

Normalized Correlation = 0.5

Mean Abs.= 11.8ppm

June 1, 2006GWADW 2006 12

Conclusions, Direction

LIGO I optical performance meets design. “As built” expectations far exceeded design.

» Can be of significant concern for Advanced LIGO, which

has initially assumed at least duplicating “as built” performance

• Design OTF tests to understand anomalous scatter:• “Frozen” in the coatings ?

• Surface contamination (~ common to all installed optics !)

• Also apparent cleaning streaks/defects: significant in terms of loss?

June 1, 2006GWADW 2006 13

Expectations vs Performance

Expectations: FFT simulations.» Design era (c ’96 ’98). Remarkable agreement with current operations.

» “As built” simulations based on bench measurements of actual fabricated optics (c summer ’03)

In Situ measurements (here, fullest story: H1 interferometer)» Scatterometer sampling of in lock beam scatter from HR surfaces.

» Arm visibilities (~’00 culminating 11/02)

» Operational performance (recycling gains, contrast defect) (to present).

» Comparison with super polished H2 ETMs.

» Detailed study of HR surface (Image analysis) “beam spots” (10/03)

June 1, 2006GWADW 2006 14

“As built” FFT Simulation

FFT simulation of H1 with no free parameters:» “Cold” state: no thermal lens (little effect on CR light)

» ~ 92 (observed ~ 41)

FFT uses measured distortion maps,all HR interfaces» minor effect on FFT

» ~13% for full as built simulation. Negligible for loss matched case.

» Consistent with very good ifo contrast defect– 6 10-4 for H1– 3 10-5 for L1

Other in situ observations (e.g H1 arm visibilities) are consistent with arm loss needed to “match” observed .

CRRCG

CRRCG

CRRCG