Report from UWA

The Big Lift:

First full scale isolator is

lifted out of assembly tent and

successfully installed in ITM tank 16 Aug 2006

UWA Optics Team: Ju Li, Chunnong Zhao, Sascha Schediwy, Pablo Barriga, Slavek Gras, Jerome Degallaix, Yan Zewu, Fan Yaohui, S Sunil, Jean-Charles Dumas, David Blair

Light Scattering in LIGO Fused Silica Sample

• As arranged in last telecon Helena sent a sample of

fused silica for UWA to measure light scattering in the

bulk.

• We have measured one plane using our ASM machine.

(Automatic scattering measuring machine…tomography

using single beams.)

• Result: very high scattering…see next slide…74ppm/cm

Parametric Instability Studies at UWA

1. Quantifying PI taking into account as many uncertainties as possible.

2. Suppressing PI using ring damping3. Suppressing PI using active feedback4. Experimental studies at Gingin

Three new results on Parametric Instability

1. Tilts and Spot Size: Both tilt fluctuations and spot size variations can tune PI. Tilt induced degeneracy breaking between HG and LG modes can introduce further unstable modes.

2. Ring damper suppression of PI. Preliminary results for AdvLIGO with sapphire test masses (easiest case, fewer modes) show stable operating windows with 5% increased thermal noise, 2.3% degraded ns-ns inspiral performance.

3. Enhanced optical spring effects allow short tranquiliser cavities for parametric instability control to use lower power and lower finesse.

Higher Order Modes

Perfectly aligned cylindrically symmetric cavity with finite size mirrors it can only sustain LG modes.

HG modes can be expressed as linear combinations of LG modes, but the frequencies are degenerate: no contribution to PI

Tilt fluctuations break the degeneracy: additional modes potentially contribute to PI. We are quantifying this effect.

Pablo Barriga

Frequency change with spot/mirror size ratio for modes of order 6

0

500

1000

1500

2000

2500

3000

0.3 0.32 0.34 0.36 0.38 0.4 0.42 0.44

Spot Size/Mirror Size Ratio

D F

req

ue

ncy

[Hz]

LG06 LG14 LG22 LG30

Spot Size TuningThermal lens induced changes in spot size will change the mode gap and the resonant conditions for Parametric Instabilities.(Edge effects)

A change of 1% in the spot/mirror ratio produces typically ~30Hz change of frequency. Sufficient to cause significant parametric gain modulation..

Advanced LIGO

Pablo Barriga

Frequency change with mirror tilt

0

10

20

30

40

50

60

70

80

0 200 400 600 800 1000 1200

Mirror tilt (nrad)

D F

requ

ency

(Hz)

HG04 HG40 HG05 HG50

Frequency Gap Tilt EffectsThe frequency change will depend on the orientation of the mode profile in relation to the tilt. For PI only the frequency gap between TEM00 and HOM is significant.

Therefore the frequency gap between fundamental and higher order modes will also depend on this relative orientation.

HG04

HG50

HG40

HG05

Frequency change with ETM mirror tilt for modes of order 4

18370

18375

18380

18385

18390

18395

18400

0 200 400 600 800 1000 1200

Mirror tilt (nrad)

Fre

quen

cy (

Hz)

LG04 LG12 LG20 HG40 HG31 HG22

Frequency Shift for LG Modes of Order 4

Transverse mode frequency variation with mirror tilt.

Frequency difference between fundamental mode and modes of order 4.

LG04

LG12

LG20

HG40

HG31

HG22

Conclusion: Tilts and Spot Size

• Effects are small and are unimportant assuming spot size is well controlled and tilts are within AdvLIGO specs…10nrad

• However spot size is important in obtaining accurate estimates of PI

Ring Damper

• We have done the ring damper modelling for sapphire because it involves fewer modes and we can get results in less time.

• Fused silica now underway.

• Choose to use a standard optical coating for the ring damper as it is a well understood vacuum compatible material.

Optical coatingOptical Coating Strip

Test mass radius r = 0.157mThickness d = 0.13m

various positions,different dimensions:-thickness-widthdifferent properties:-loss angle (thermal noise)

-tantala/silica layers-loss freq. dependent

Ring damper model

PI and Ring Damper calculation proceedure

1. High resolution FEM model (half-plane symmetry model) with coatings calc freq and mode shape of all acoustic modes to 160kHz (coating has small effect on mode shape, frequency and mode effective mass. (some non-symmetric modes omitted)

2. Low resolution FEM(full cylinder model with inertia relief) to calculate thermal noise as seen by Gaussian beam due to substrate and coatings losses. (confirmed no resolution induced errors)

3. Use Matlab code eigenvalue method to calculate diffraction losses for different ROC configurations.

4. Calculate optical mode shapes analytically corresponding to infinite mirror approximation

5. Run Matlab Parametric Instability Code using optical modes to 9th order, to calculate overlap parameter as function of radius of curvature.

6. Create histogram of number of unstable modes, R-values etc.

7. Repeat for different ring dampers. Each ring damper is designed so that it increases the brownian thermal noise by a given percentage at 100Hz referred to the Brownian thermal noise for ROC of 2076km beam spot radius 6cm.

8. Use bench code to estimate binary inspiral range. (Switch off residual gas and suspension TN and gravity gradient noise). Photothermal noise and quantum noise terms at default values in Bench Code.

Effect of the coating losses on parametric gain

Comparison of R gain for different RoC, with PRM (red) and without PRM (blue)

Power recycling mirror /coating

Parametric Instability control

•Ring damper 3cm x 20microns, 5 x 10-3 loss angle sufficient to create stable windows (worst case model, no suppression of instability by arm imbalance).•Brownian thermal noise degraded 5%, inspiral range degraded 1.5%..see next slidestable windows

ALS, DTN=5%

About 30% 0f phase space is instability free

NS/NS binary inspiral range vs. ETM RoC (ITM fixed)

AdvLIGO

It seems that the NS/NS range degradation due to the ring damper remains constant for different RoC_ETM.

6.5,6.3 6.37,6.27 6.24,6.18 6.12,6.09 6.01,6.01 5.90,5.93Spot sizes

ALS with a ring vs AdvLIGO @ 2076 m

ALS AdvLIGO

212.03Mpc (with strip) 178.17 Mpc (no strip)

We are surprised that sapphire with ring damper is better than the Adv LIGO number we get from default settings of bench.

It would be very useful have independent verification of these results!

ω

0

ω

0

Unstable - Instability

ω0 – Ωm ω0

Ωm

ω0 + Ωm ω0

Ωm

ω

0

ω

0

Stable – Cold Damping

l

cFSR

2

Stable Unstable

Frad

Enhanced Optical SpringEnhanced Optical SpringParametric Instability Parametric Instability

Sascha Schediwy - schediwy@physics.uwa.edu.au

Closed Loop TF’s

Sascha Schediwy - schediwy@physics.uwa.edu.au

Enhanced Optical SpringEnhanced Optical SpringParametric InstabilityParametric Instability

Background information for next slide:--Optical spring Q change arrises due to the phase lag introduced by a non-zero cavity storage time. ●Next slide shows Q-1 vs. cavity offset for 10cm Nb

cavity. ● Cavity is locked with PDH feedback servo.● Servo uses analogue f-1 filter with corner frequency fc.

● Servo electronics introduces phase lag.● Optical spring instability or damping enhanced ~20

times.

Niobium resonator fmech = 750Hz

-20

-10

0

10

20

30

40

10 100 1000 10000 100000

Fequency (Hz)

Lo

g M

ag (

dB

)

fc=300Hz

fc=3000Hz

fmech

-180

-120

-60

0

60

10 100 1000 10000 100000

Frequency (Hz)

Ph

ase

(deg

)fc=300Hz

fc=3000Hz

Enhanced Optical SpringEnhanced Optical SpringParametric InstabilityParametric Instability

-1.0E-05

-5.0E-06

0.0E+00

5.0E-06

1.0E-05

1.5E-05

2.0E-05

2.5E-05

3.0E-05

-100000 -80000 -60000 -40000 -20000 0 20000 40000 60000 80000 100000

Offset (Hz)

Invers

e Q

(-)

fc @ 300Hz, -74deg

fc @ 3000Hz, -19deg

CriticallyUnstable Region

Cold Damping Region

Unstable Region

-74deg

-19deg

no enhancement

Measurement

Theory

Sascha Schediwy - schediwy@physics.uwa.edu.au

Enhanced Optical SpringEnhanced Optical SpringParametric Instability Parametric Instability

Formulae to calculate the change in the quality factor Qnew’ for the enhanced optical spring from kopt and the servo introduced phase lag θ:

Sascha Schediwy - schediwy@physics.uwa.edu.au

'' optRopt kk '' opt

Iopt kk

sincossincos

sincos

'

ikiik

ikik

ekk

Iopt

Ropt

Iopt

Ropt

ioptopt

222

1

1

1

16

p

offsetp

offsetinRopt

f

ff

f

Rc

Pk

3222

1

1

1

32

p

offsetp

offsetminIopt

f

ff

ff

Rc

Pk

'

4'

22

Iopt

effmopt k

mfQ

'

11

'

1

optmechnew QQQ where:

andwith

and

Above spring constant formulae derived from 1st principles and also from:Braginsky, V.B. and S.P. Vyatchanin,. Physical Letters A, 2002. 293: p. 228-234.

Sheard, S. Gray, M. Mow-Lowry, C. and McClelland, D. Physical Review A, 2004 69: 051801(R).

ConclusionConclusion

1. Short Optical Cavity Tranquiliser looks much more practical as strong optical damping can be achieved for relatively low finesse.

2. Need to confirm noise performance

3. Ring damper looks very promising but difficult to completely elliminate instability without serious noise price.

4. Thermal tuning: spot size effects not a problem

5. Need to repeat analysis for fused silica.

IFO:

Sapphire ETM/ITM:r = 0.16m d = 0.13m

Spot size:w = 6.0cm

arm length:L=4km

diffraction loss (per bounce @ 2076m):3.81ppm

Unstable modes (R>=1): 380

AL - IFO modeled

Brownian Noise at 2076 m (FEM result ):

ITM ETMCoating 2.7242e-21 4.1503e-21Substrate 1.4448e-21 1.4448e-21Ring 1.2063e-21 1.2063e-21Total 7.5304e-21

Ring damper model (all TMs) :

Strip width: 3 cmStrip thickness: 20µmLoss angle: chosen in such way that the ifo Brownain thermal noise @ 2076 would increase by 5%. Φ = 3.5909e-3

Coating info:ITM T=5% -> 8 layers of Ta2O5 +8 layers SiO2ETM T=5ppm -> 18 layers of Ta2O5 +18 layers SiO2

Notes:

-100% , 90% and 75% corresponds to the coating TN. I assumed that future coating may be better than the present ones by these factors: used in the Bench code.

-If Total Brownian TN @ 2076 m is equal 7.5304e-21 (without a strip) than by adding a strip with loss = 3.5909e-3 it causes 5% Brownian thermal noise degradation .If Coating TN was reduced to 90% for both ITM and ETM mirrors than the strip with the same loss value as above would cause DT=~14.5%. If coating TN was 75% smaller than DT increases to ~32.3%.

According to NS/NS results (slide 3), inspiral range does not follow this pattern. This range would be only decrease by ~1.7% for all cases (0%,10%,25%). I’m not sure why it is so… maybe it’s due to the high thermoelastic noise? As seen in the slide 5 this noise is much higher than the strip one. If this is true, than silica test mass with much lower thermoelastic noise would be more susceptible to the strip noise (I think)Slide 4 shows only NS/NS range for not reduced coating TN (100%).