Mirror Cleaning
Michele Punturo
INFN Perugia
1
Thermal Transient
• A large thermal transient affects the Virgo operation
Sidebands amplitude
Locking steps
2
Mirror Absorption
• We attributed to the excess of absorption in the Virgo
input mirrors the responsibility of this thermal transient
• This has been possible thanks to the mirror thermal mode
technique
1. We measure the mirror temperature increase during the lock
procedure
2. We compare it with a FEM analysis, miming also the power
injected during all the locking procedure
3. We adjust the hypothetical absorption of the mirror, to match the
measured temperature increase
3
Resonant mode technique
• Obviously the resonant frequencies of a body depend on
the temperature of the body
• For a Virgo mirror we evaluated this dependence with a
ANSYS based FEM (2004, F.Travasso PhD thesis)
Frequency Variation [Hz]
5
4
3
2
1
Drum mode:
0.61 Hz/K
ButterxM
ButtertM
DrumM
0
-1
-2
-3
Butterfly mode(s): 0.28 Hz/K
-4
282 284 286 288 290 292 294 296 298 300 302
Temperature [K]
4
Resonant mode “calibration”
• To increase our confidence in this method we crosschecked the mirror temperature increase (due to the
environmental fluctuation) measured through the resonant
mode technique with the Super Attenuator temperature,
measured by traditional thermometers
• This is just an
example: the filters T
variation could be
both smaller and
larger than the mirror
fluctuation (different
height, presence of
the separating roof),
but it is always similar
0.4
0.3
 T [K]
0.2
0.1
0.0
dTWE10 Em_TESUWE10
dTWE08 Em_TESUWE08
dTWE06 Em_TESUWE06
dTWE04 Em_TESUWE04
dTWE02 Em_TESUWE02
dTWEF760s Em_TESUWE11 60s average
dTWEn drum mode
dTWE2n batterfly mode
-0.1
000
400
800
200
600
000
400
240
326
412
499
585
672
758
840
840
840
840
840
840
840
GPStime [s]
1 day
5
Expected T increase
• The expected temperature increase, due to the beam illumination, is
evaluated through a Matlab based FEM (confirmed by ANSYS & Comsol
FEMs)
80
rings
• Hypothesis:
• We know the laser power injected in the ITF (and we rescale it with the
power read by B5)
• We believe to the 0.7ppm/cm of substrate absorption (cross-checked with a
measurement in Lyon on a spare substrate)
• We concentrate all our “ignorance” on the surface (coating) absorption
(expected to be 1.3ppm)
• The environmental temperature is constant during the locking procedure
6
Measured excess
• NI: (ppm)
4.5 ±0.50 (stat)
± 0.23 (laser power)
+- 0.15 (finesse
fluctuation)
+- 0.38 (calibration)
• WI: Measurement
difficulties: results
ranging from 7ppm to
16ppm
7
Increase our confidence
• The confidence in the resonant mode is critical in this evaluation.
– Could we find an independent confirmation?
• Etalon effect in the input mirrors
AR
HR
HR
B7/B8 phd
LB
r2
dn
dT
r3
0
0
AR coating
LA
0.02
P/P
r1
varPot ( x)
0.04
0.06
0.067846 0.08
0
0
0.2
0.4
0.6
0.8
1
1
x
T
8
… Increase our confidence
• dn/dT depends on the particular kind of FS, but the
order of magnitude is correct (0.8-1 × 10-5)
• An incertitude of 15% is directly transferred in a similar
error in the temperature evaluation.
• Finally, we can confirm that an excess of absorption in
the input mirrors is present (and that the evaluation of
the absorbed ppm is reliable).
• Possible sources of absorption:
• Substrate
• Coating
• Pollution
9
How to clean the mirrors?
• The “natural” way should be to dismount the
mirrors
– Rejected, because we cannot accept the dead time
(~1 month) for the commissioning
• First contact polymer
– http://www.photoniccleaning.com/
Credit to L.Pinard (LMA)
Clear Polymer Applied
with A Pump Spray
Applying First Contact
Polymer with a Brush
10
Cleaning performance demonstration
Dirty Telescope Mirror
Polymer Sprayed Right Over The Contaminates
No Prior Cleaning of Any Kind.
Credit to L.Pinard (LMA)
Telescope mirror
After removing
the film
11
“Quantitative” evaluation of the performances
50 mm mirror after classical cleaning
Scattering 6 ppm
Absorption : 1.91 ppm
Credit to L.Pinard (LMA)
12
...“Quantitative” evaluation of the performances
50 mm mirror, with particles
Scattering 25 ppm
Absorption : 5 ppm
Credit to L.Pinard (LMA)
13
...“Quantitative” evaluation of the performances
50 mm mirror, after puting and
removing a film of ‘First Contact'
Scattering 6-6.5 ppm - idem after cleaning
Absorption : 1.73 ppm
The Film has ’cleaned’ the mirror
Credit to L.Pinard (LMA)
14
Virgo Mirrors Cleaning
http://virgo.pg.infn.it/~punturo/#2007
15
Effects of the cleaning
• No bad effects due to the cleaning procedure
– Mirrors surface looks really cleaner
– Any more translucent reflection
• But no substantial improvement (except a 25%
increase of the sidebands amplitude)
16
Credits: G.Vajente
Thermal transient
17
And what about the absorption?
18
T instability and absorption measurement
• Mirror T instability obstacle the measurement of the mirror
absorption
– Zoom of the previous plot
Until 10/12/2007
-1.0
dTWIn
B8DCnorm
1.90
1.88
1.86
dTWI [K]
-1.2
1.84
1.82
-1.3
1.80
1.78
-1.4
1.76
-1.5
B8_DC/B5_DC
-1.1
1.92
1.74
-1.7
2.04
dTNIn
B7DCnorm
2.02
2.00
dTNI [K]
-1.9
1.98
-2.0
1.96
1.94
-2.1
1.92
-2.2
B7_DC/B5_DC
-1.8
1.90
800 200 600 000 400 800 200 600 000 400 800 200 600 000 400 800 200 600 000
724 811 897 984 070 156 243 329 416 502 588 675 761 848 934 020 107 193 280
879 879 879 879 880 880 880 880 880 880 880 880 880 880 880 881 881 881 881
GPStime[s]
19
T instability and absorption measurement
• Mirror T instability obstacle the measurement of the mirror
absorption
– Zoom of the slide #8 plot
Until 10/12/2007
-1.0
dTWIn
B8DCnorm
1.90
1.88
1.86
dTWI [K]
-1.2
1.84
1.82
-1.3
1.80
1.78
-1.4
1.76
-1.5
B8_DC/B5_DC
-1.1
1.92
1.74
-1.7
2.04
dTNIn
B7DCnorm
2.02
2.00
dTNI [K]
-1.9
1.98
-2.0
1.96
1.94
-2.1
1.92
-2.2
B7_DC/B5_DC
-1.8
1.90
800 200 600 000 400 800 200 600 000 400 800 200 600 000 400 800 200 600 000
724 811 897 984 070 156 243 329 416 502 588 675 761 848 934 020 107 193 280
879 879 879 879 880 880 880 880 880 880 880 880 880 880 880 881 881 881 881
GPStime[s]
20
Locking warming up
• In this “general” T fluctuation of the order of 0.5K at the mirror
level, we must extract the warming up due to the mirror due to
the laser, that should be of the order of 10-40mK
-1.98
0.05
0.05
-1.42
-1.99
0.04
0.04
-1.44
dTNIn
V1:Pr_B5_DC_mean
-1.46
-1.48
0.02
B5_DC [W]
dT_WI [K]
dT_NI [K]
-2.01
-2.02
0.03
0.03
0.02
-1.50
-2.03
0.01
dTWIn
V1:Pr_B5_DC_mean
0.01
-1.52
-2.04
0.00
-1.54
-2.05
0
50000
100000
150000
Time [s]
0
200000
50000
250000
0.00
100000
300000
150000
Time [s]
200000
250000
300000
21
B5_DC [W]
-2.00
Locking warming up extraction
• Qualitatively is easy, but the numerical evaluation
can be affected by a large error
– Long-1.1relaxation time after the perturbation:
Data: Data6_dTWI
Model: ExpDec1
Equation: y = A1*exp(-x/t1) + y0
-1.2
Weighting:
y
No weighting
Chi^2/DoF
= 0.00024
R^2
= 0.97221
y0
-1.48229
±0.00136
A1
0.34628
±0.0026
t1
123100.49853 ±2228.18165
dTWI [K]
-1.3
-1.4
-1.5
dTWI
ExpDec1 fit of Data6_dTWI
-1.6
0
100000
200000
300000
400000
500000
600000
Time [s]
22
Matching with the FEM
• No real access to the overall Temperature trend
• FEM can be run just with the Tambient=constant hypothesis
• Matching with the real data performed lock-by-lock
293.020
-1.990
293.018
Tsim
293.016
dTNIn
-1.995
-2.020
-2.000
Tsim
293.018
293.010
293.016
293.020
293.008
293.014
293.018
-2.025
-2.030
-2.005
293.006
293.012
293.016
dT (measured)
293.020
293.012
-2.030
293.004
293.010
293.014
Tsim
293.002
293.008
293.012
293.000
293.006
293.010
0
-2.035
dT (measured)
-2.010
-2.035
-2.015
50000
100000
150000
200000
250000
-2.040
300000
X Axis Title
293.004
293.008
-2.040
293.002
293.006
-2.045
-2.045
293.000
293.004
0
50000
100000
293.002
150000
200000
250000
dT (measured)
T (simulated)
T (simulated)
T (simulated)
293.014
Pre-VSR1 evaluation: 6ppm
VSR1 evaluation:
NI (ppm): 4.5 ±0.50 (stat) ± 0.23
(laser power) ± 0.15 (finesse
fluctuation) ± 0.38 (calibration)
5.4ppm
5.4ppm
5.4ppm
Error to be defined!
300000
time [s]
293.000
-2.050
0
50000
100000
150000
200000
time [s]
250000
300000
350000
23
Matching with the FEM: WI
293.020
Pre-VSR1 evaluation: 16ppm
VSR1 evaluation:
WI (ppm): 7? ± ?? (difficulties to
measure it… same difficulty now)
-1.41
Tsim
293.016
-1.42
-1.43
Tsim
293.020
293.008
-1.44
293.016
dT [measured]
293.012
11ppm
-1.455
293.004
-1.45
293.000
293.020
0
50000
100000
150000
200000
250000
300000
293.008
-1.46
-1.460
350000 -1.45
Time [s]
dT (measured)
293.012
3.4ppm (T decrease during
the lock)
293.016
-1.46
293.004
11ppm
T (simulated)
-1.465
293.012
-1.47
293.000
0
293.008
50000
100000
150000
200000
250000
300000
Tsim
Time [s]
dTWIn
-1.48
dT [measured]
T (simulated)
T [simulated]
dTWIn
293.004
-1.49
293.000
0
50000
100000
150000
Time [s]
200000
250000
300000
24
Conclusions
• Despite the better look of the mirror surface,
no substantial decrease of the thermal
transient and of the absorption occurred
• What kind of pollution (if any) contaminated
the mirror?
• What other investigations cn be performed to
understand better?
• What to do if TCS is not enough?
25