Transcript SeaUnivStudentsTalk - LIGO Hanford Observatory

Lasers and Optics of
Gravitational Wave Detectors
Rick Savage
LIGO Hanford Observatory
1
GW detector – laser and optics
Power Recycled
Michelson
Interferometer
with Fabry-Perot
Arm Cavities
end test mass
4 km (2 km) Fabry-Perot
arm cavity
recycling
mirror
input test mass
Laser
signal
beam splitter
2
Closer look - more lasers and optics
3
Pre-Stabilized Laser System

Laser source

Frequency
pre-stabilization
and actuator for
further stab.
Compensation
for Earth tides


Power stab. in
GW band

Power stab. at
modulation freq.
(~ 25 MHz)
4
Initial LIGO 10-W laser



Master Oscillator Power Amplifier
configuration (vs. injection-locked
oscillator)
Lightwave Model 126 non-planar
ring oscillator (Innolight)
Double-pass, four-stage amplifier
» Four rods - 160 watts of laser
diode pump power

10 watts in TEM00 mode
5
LIGO I PSL performance




Running continuously since
Dec. 1998 on Hanford 2k
interferometer
Maximum output power has
dropped to ~ 6 watts
Replacement of amplifier
pump diode bars had restored
performance in other units
Servo systems maintain lock
indefinitely (weeks - months)
6
Frequency stabilization

Three nested control loops
»
»
»

20-cm fixed reference cavity
12-m suspended modecleaner
4-km suspended arm cavity
Ultimate goal: Df/f ~ 3 x 10-22
7
Power stabilization

In-band (40 Hz – 7 kHz) RIN
» Sensors located before and
after suspended modecleaner
» Current shunt actuator amp. pump diode current

RIN at 25 MHz mod. freq.
» Passive filtering in 3-mirror
triangular ring cavity (PMC)
» Bandwidth (FWHM) ~ 3.2 MHz
3e-8/rtHz
8
Earth Tide Compensation



Up to 200 mm over 4 km
Prediction applied to ref.
cav. temp. (open loop)
End test mass stack
fine actuators relieve
uncompensated residual
100mm
prediction
residual
9
Concept for Advanced LIGO laser
power supplys
control
PC
diodeboxes

10 x 30W
@ 808 nm
10 x 30W
@ 808 nm
10 x 30W
@ 808 nm
2 x 30W
@ 808 nm
10 x 30W
@ 808 nm
output
Slave II
Master


Being developed by
GEO/LZH
Injection-locked, endpumped slave lasers
180 W output with
1200 W of pump light
Slave I
Pound-Drever-Hall Locking - Electronic
oscillator, mixer, phase-splitter, servo
10
Brassboard Performance



LZH/MPI Hannover
Integrated front end based on
GEO 600 laser – 12-14 watts
High-power slave – 195 watts
M2 < 1.15
11
Concept for Advanced LIGO PSL
high
power
stage
I5
PS S3
medium
power
stage
ILS1
suspended
modecleaner
PS S2
spatial
filter
cavity
PM C 1
PS S1
I2
NPRO
1
2
3
I1
ILS2
FS S
AOM
I3
4
FS S- A1
Diagnostic
reference
cavity
tidal feedback
FS S- A2
I4
PM C 2
12
Core Optics – Test Masses

Low-absorption fused silica substrates
» 25 cm dia. x 10 cm thick, 10 kg




Low-loss ion beam coatings
Suspended from single loop of music wire (0.3 mm)
Rare-earth magnets glued to face and side for
orientation actuation
Internal mode Qs > 2e6
13
LIGO I core optics
Caltech data






RITM ~ 14 km (sagitta ~ 0.6 l) ; RETM ~ 8 km
Surface uniformity ~ l/100 over 20 cm. dia. (~ 1 nm rms)
“Super-polished” – micro-roughness < 1 Angstrom
Scatter (diffuse and aperture diffraction) < 30 ppm
Substrate absorption < 4 ppm/cm
Coating absorption < 0.5 ppm
14
Adv. LIGO Core Optics

LIGO recently chose fused silica over sapphire
» Familiarity and experience with polishing, coating, suspending,
thermally compensating, etc. – less perceived risk


Other projects (e.g. LCGT) still pursuing sapphire test masses
Thermal noise in coatings expected to be greatest challenge
fused silica
38 cm dia., 15.4 cm thick, 38 kg
sapphire
15
Processing, Installation and Alignment
Experience indicates
that processing and
handling may be
source of optical loss
gluing
vacuum baking
wet cleaning
suspending
balancing
transporting
16
Thermal Issues
» ~ 25 kW for initial LIGO
» ~ 600 kW for adv. LIGO

Substrate bulk absorption
» ~ 4 ppm/cm for initial LIGO
» ~ 0.5 ppm/cm ($) for adv. LIGO

Coating absorption
» ~ 0.5 ppm for initial & adv. LIGO

radius
Thermo-optic coefficient
»

Surface absorption
Circulating power in arm cavities
depth

dn/dT ~ 8.7 ppm/degK
Thermal expansion coefficient
» 0.55 ppm/degK

“Cold” radius of curvature of optics
adjusted for expected “hot” state
Bulk absorption
17
Thermal compensation system
CO2
Laser
?
Over-heat
Correction
Under-heat
Correction
ZnSe
Viewport
ITM
ITM
Inhomogeneous
Correction
PRM
Adv. LIGO
concept
Compensation
Plates
ITM
SRM
18
Coating vs. substrate absorption
Surface distortion
Optical path difference
0
x 10
-6
Optical Path Difference in Transmission for 1 W absorbtion
0
x 10
-7
Surface Distortion in for 1 W absorbtion
Coating
Substrate
-0.2
-0.2
-0.4
-0.4
-0.6
-0.6
OPD (m)
OPD (m)
Coating
Substrate
coating
-0.8
-0.8
-1
-1
-1.2
-1.2
-1.4
substrate
0
0.02


0.04
0.06
0.08
radius (m)
0.1
substrate
0.12
0.14
-1.4
coating
0
0.02
0.04
0.06
0.08
radius (m)
0.1
0.12
0.14
OPD almost same for same amount of power absorbed in coating or substrate
Power absorbed in coating causes ~ 3 times more surface distortion than same
power absorbed in bulk
19
Summary


LIGO utilizes 10-W solid state lasers
» Relative frequency stability ~ 10-21/rtHz
» Relative power stability ~ 10-8/rtHz
» Advanced LIGO lasers: similar requirements at 200 watt power level
LIGO test masses (mirrors) 25 cm dia., 10 cm thick fused silica
» Surface uniformity ~ l/100 p-v (1 nm rms) over 20 cm diameter
» Coating absorption < 1 ppm, bulk absorption ~ few ppm/cm
» Active thermal compensation required to match curvatures of optics
» Non-invasive measurement techniques required for characterizing
performance of optics
20
Anomalous absorption in H1 ifo.


ITMY

Negative values imply annulus
heating
Significantly more absorption
in BS/ITMX than in ITMY
How to identify absorption site?
ITMX
TCS power is absorbed
in HR coatings of ITMs
21
Need for remote diagnostics

Water absorption in viton spring seats makes
vacuum incursions very costly.
» Even with dry air purge, experience indicates
that 1-2 weeks pumping required per 8 hours
vented before beam tubes can be exposed to
chambers

Development of remote diagnostics to determine
which optics responsible of excess absorption
22
Spot size measurements
ITMX




BeamView CCD cameras in ghost
beams from AR coatings
Lock ifo. w/o TCS heating
Measure spot size changes as ifo.
cools from full lock state
Curvature change in ITMX path
about twice that in ITMY path
ITMY
23
Arm cavity g factor changes



Again, lock full ifo. w/o TCS heating, break lock, lock single arm and
measure arm cavity g factor at precise intervals after breaking lock
g factor change in Xarm larger than Yarm by factor of ~ 1.6
Calibrate with TCS (ITM-only surface absorption)
24
Results and options

ETM surface


Beamsplitter not significant
absorber
ITMX is a significant absorber
~ 25 mW/watt incident
ITMY absorption also high
~ 10 mW/watt incident
»

Factor of ~5 greater than
absorption in H2 or L1 ITMs
Options
» Try to clean ITMX in situ
» Replace ITMX
» Higher power TCS system


From analysis by K. Kawabe
30-watt TCS laser was tested
Eventually ITMX was
replaced and ITMY was
cleaned in-situ
25
Origin of G-factor measurement technique


Simple question:
“For a resonant optical cavity, can the Pound-Drever-Hall locking signal
distinguish between frequency and length variations?”
i.e. does
Df DL
f


L
Of course!
Or does it?
26
High-frequency response of optical cavities

Dynamic resonance of light in Fabry-Perot cavities
(Rakhmanov, Savage, Reitze, Tanner 2002 Phys. Lett. A, 305 239).
27
High frequency length response

Peaks in length
response at multiples
of FSR suggest
searches for GWs at
high frequencies.

HF response to GWs
different than length
response
Different antenna
pattern, but still
enhancement in
sensitivity
LIGO band

1FSR
2FSR
28
High frequency response to GWs




Long wavelength approximation not
valid in this regime
Antenna pattern becomes a function
of source frequency as well as sky
location and polarization
All-sky-averaged response about a
factor of 5 lower than at low freq.
Significant sensitivity near multiples
of 37.5 kHz (arm cavity FSR)
Movie (by H. Elliott): Antenna pattern for one source
polarization as source frequency sweeps from 22 to 36 kHz
29
G-factor Measurement Technique

Dynamic
resonance of light
in Fabry-Perot
cavities (Rakhmanov,
Savage, Reitze, Tanner
2002 Phys. Lett. A, 305
239).

Laser frequency to
PDH signal
transfer function,
Hw(s), has cusps at
multiples of FSR
and features at
freqs. related to the
phase modulation
sidebands.
30
Misaligned cavity
Features appear at frequencies related to higher-order transverse modes.
 Transverse mode spacing: ftm = f01- f00 = (ffsr/p) acos (g1g2)1/2
 g1,2 = 1 - L/R1,2
 Infer mirror
curvature changes
from transverse mode
spacing freq. changes.
 This technique
proposed by F. Bondu,
Aug. 2002.

Rakhmanov, Debieu,
Bondu, Savage, Class.
Quantum Grav. 21 (2004)
S487-S492.
31
H1 data – Sept. 23, 2003
• Lock a single arm
• Mis-align input
beam (MMT3) in
yaw
• Drive VCO test
input (laser freq.)
• Measure TF to
ASPD Qmon or Imon
signal
2ffsr- ftm
• Focus on phase of
feature near 63 kHz
32
Data and (lsqcurvefit) fits.
ITMx TCS annulus heating  decrease in ROC (increase in curvature)
R = 14337 m
R = 14096 m
Assume metrology value for RETMx = 7260 m
Metrology value for ITMx = 14240 m
33