Transcript G030036-00 - DCC

Thermal Noise in Advanced
LIGO Core Optics
Gregory Harry and COC Working Group
Massachusetts Institute of Technology
- Technical Plenary Session March 17-20, 2003
LSC Meeting - Livingston, LA
LIGO-G030036-00-R
1
Outline
• Sensitivity impact
• Coating thermal noise
• Relevance for material downselect
• Silica annealing
• Sapphire status
• Modeling (analytical and FEA)
• Direct thermal noise measurements
LIGO-G030036-00-R
2
Significance for sensitivity
BNS Range 120 Mpc
LIGO-G030036-00-R
BNS Range 200 Mpc
3
Coating thermal noise
Status
• Tantala/silica coating studied on silica
- fcoat = 2.8 +/- 0.7 10-4 (in modal Q measurements)
- tantala dominates loss
• Various other materials tried
- niobia/silica, tantala/alumina, alumina/silica
- none have consistently improved loss
• Some work on sapphire substrates
• Figure of merit developed
d fcoat ( Ypar/Ysub + Ysub/Yperp)
LIGO-G030036-00-R
4
Coating thermal noise
Recent results
• Special coating developed at SMA/Virgo
- tantala doped to reduce stress in coating
- fcoat = 2.8 10-4 undoped tantala/silica
- fcoat = 1.8 +/- 0.1 10-4 MIT
- fcoat = 1.5 +/- 0.7 10-4 Glasgow
- Young's modulus and optical absorption unchanged
• Annealed alumina/silica sample measured
- 2.1 +/- 0.6 10-4 Glasgow
- results pending at Syracuse
• Coated sapphire next (Glasgow)
LIGO-G030036-00-R
5
Coating thermal noise
Future plans
• New material being developed at SMA/Virgo
- index similar to tantala
- Young's modulus similar to silica
- working to get optical absorption down
• Further work with doped tantala/silica
• Correlate loss with coating stress
• Explore effects of annealing
• Measure coating thermal noise directly
LIGO-G030036-00-R
6
Material downselect
Substrate loss
-5
-5
BNS Range
GPaGPa
and f coat
= 1f10
BNS Range
vs vs
Q QforforYYcoat
=10100
and
coat =
coat = 1X10
Binary Neutron Star Inspiral Distance (Mpc)
210
200
190
180
170
160
150
140
130
Silica
Sapphire
120
0
50
100
150
200
250
300
350
Internal Q
LIGO-G030036-00-R
7
Material downselect
Coating loss
BNS Range
forvs
Y=
= 70
coat
BNS Range
Yfcoat
70GPa
GPa
BNS Range
Yfcoat
=200
200
BNS Range
for
vscoat
Y=
GPaGPa
210
210
Silica 200 million Q
Silica 200 million Q
Silica 130 million Q
Silica 130 million Q
Sapphire 200 million Q
Sapphire 200 million
190
Sapphire 60 million Q
Sapphire 60 million Q
180
180
200
190
BinaryNeutronStarInspiralDistance(Mpc)
BinaryNeutronStarInspiralDistance(Mpc)
200
170
160
150
140
130
120
0.0E+00 2.0E-05
LIGO-G030036-00-R
170
160
150
140
130
120
4.0E-05
6.0E-05
Coating
ff
Coating
8.0E-05
1.0E-04
0.0E+00
1.2E-04
2.0E-05
4.0E-05
6.0E-05
8.0E-05
Coating
f
Coating
f
1.0E-04
1.2E-0
8
Material downselect
Coating Young's modulus
-5
-5
BNS Range
withfvs
f
=
5
X
10
BNS Range
=
Y
5
for
10
coat coat
210
210
200
200
190
190
180
180
170
160
150
140
Silica 200 million Q
Silica 130 million Q
130
BinaryNeutronStarInspiral Distance(Mpc)
BinaryNeutronStarInspiral Distance(Mpc)
BNS Range
=for
1 -510
BNS Range
withvsfcoat
fYcoat
= 1 X 10-5
Sapphire 200 million Q
170
160
150
140
Silica 200 million Q
Silica 130 million Q
130
Sapphire 200 million Q
Sapphire 60 million Q
Sapphire 60 million Q
120
120
10
100
Coating Young's
modulus (GPa)
Coating Young's
modulus
LIGO-G030036-00-R
1000
10
100
1000
Young's modulus
(GPa)
CoatingCoating
Young's
modulus
9
Substrate thermal noise
Silica status
• Empirical understanding of
silica loss is developing
• Lossy surface layer limits Q
• Annealing can dramatically
increase Q
• High Q in polished sample
- 54 106
• High Q in flame drawn sample
- 200 106
• See silica discussion Thursday
afternoon
LIGO-G030036-00-R
10
Substrate thermal noise
Sapphire status
•
•
Thermal noise dominated by thermoelastic damping
Modal Q's typically about 200 106 (S. Rowan, V.
Mitrofanov, et al)
• Q's span 65 to 400 106 (K. Numata, P. Willems, et al)
• Low frequency dependence to loss unknown
• Anisotropy of loss not well understood
• See talk by G. Billingsley
LIGO-G030036-00-R
11
Substrate thermal noise
Recent Q results on sapphire
0.025
Two 40 kg samples measured
for Q at Caltech by Phil
Willems, 6 modes each
0.02
0.015
0.01
- white (“good”) sapphire
0.005
two degenerate modes
show high Q
Q1 = 200 106
Sets limit on
Q2 = 180 106
anisotropy of loss
}
- pink (“not”) sapphire shows
high Q of 260 106
LIGO-G030036-00-R
16
18
20
22
24
1/Q vs mode for both samples
• Results fit two parameter model
for single bulk f and surface
(barrel) f very well
12
Other loss sources
Bonding and charging
• Silicate bonding to silica suspension
• Bonded silica samples measured for Q at Glasgow and
Syracuse using different geometries
• Loss very high in bond region (f ~ 100 - 10-2)
• Calculations indicate will not effect thermal noise in advanced LIGO
(Syracuse sample bonded
in Glasgow)
•
Charging of optics
• Modeling and Q measurements suggest will not limit thermal noise
• Could be a source of other noise sources
• May need more study
LIGO-G030036-00-R
13
Thermal noise modeling
Analytical models we have
•
Non-modal, direct thermal noise calculation (Yu. Levin)
Better Paradigm than modal Q for thermal noise
•
Finite sized, uncoated mirrors (Liu and Thorne, Bondu et al)
•
Infinite sized, coated mirrors (Nakagawa / Gretarsson et al)
•
Anisotropic coatings (assuming isotropic layers)
fcoat+ = Ycoat / d (d1 f1 / Y1 + d2 f2 / Y2 )
•
Thermoelastic damping in coatings (M. Fejer, S. Rowan)
- sets limit on how low coating loss can be
- creates preferential matching of coatings and substrates
- see talk by Sheila Rowan later on Thursday
LIGO-G030036-00-R
14
Thermal noise modeling
Models we need
• Finite sized, coated mirrors
N. Nakagawa is thinking about this problem
FEA models indicate thermal noise goes down
• Multiple coatings on substrate
have secondary coating below first coating
mechanical impedance matching
one coating with low absorption, one with low loss
no one is thinking about this problem
• Anisotropic substrate
used for sapphire, may not be necessary
• Inhomogeneous loss distribution
probably better done by finite element analysis (FEA)
• (Coating thermal noise with Mexican hat beam)
not strictly necessary
LIGO-G030036-00-R
15
Thermal noise modeling
Finite element analysis
Code we have
OCEAN - coating, bonding, and surface loss Q
(D. Crooks, et al)
invaluable for coating and bonding loss efforts
I-DEAS – inhomogeneous, anisotropic modal Q, and
thermal noise (D. Coyne)
good agreement with Nakagawa theory, being used for initial LIGO
TAMA - inhomogeneous thermal noise (K. Numata, K. Yamamoto, et al)
shows thermal noise lower than Nakagawa theory for finite mirrors
What we need
A sharable version of TAMA code
Further development of most codes
LIGO-G030036-00-R
16
Direct measurement of thermal noise
Thermal Noise Interferometer (Caltech)
• Designed to measure
thermal noise in silica and
sapphire
• Silica mirrors in place
• Sapphire mirrors on hand
• Measured noise close to
Coating thermal noise
tantala/silica coating
thermal noise
• Development ongoing
• See TNI Technical Advisory
Committee session from
Tuesday
LIGO-G030036-00-R
17
Direct measurement of thermal noise
University of Tokyo experiments
• K. Numata thesis experiment
• Measure Brownian and thermoelastic
• BK7 glass for Brownian, f independent
• CaF2 for thermoelastic, good
agreement with theory
• Trying to measure coating noise
• K. Yamamoto thesis experiment
• Examined nonhomogeneous
loss
• Good agreement with Levin
LIGO-G030036-00-R
Thermal noise prospectus
What we need to do from here
• Reduce coating thermal noise to acceptable level
• Determine if high Q can be obtained in large, polished
silica optics
• Continue to study sapphire
• Further development of theories to turn Q’s into thermal
noise predictions
• Confirm thermal noise predictions with direct
measurements
19
Conclusions
• Thermal noise is a crucial problem in advanced LIGO
• Coating thermal noise reduction is proceeding
• Material downselect depends on many factors
• Silica and sapphire both are possible choices
• Work remains on thermal noise modeling
• Direct thermal noise measurements are beginning to
provide input
LIGO-G030036-00-R
20