Transcript Gravitational Wave Astronomy: Past, Present, and Future

Gravitational Wave Astronomy:
Past, Present, and Future
~or~
How To Get Lots of Science
Out of Null Results
Jeffrey Silverman
Ay 250: Transient Universe
April 24, 2007
Introduction to Gravity Waves (GW)




E-M radiation observations biased towards
“hot” objects.
GW observations biased toward compact,
massive, rapidly moving objects.
GW astronomy will open “a whole new
window” on the sky (much like the advent
of radio and X-ray astronomy).
GWs probe places with extreme gravity:
• Usually opaque to E-M radiation.
• Great places to test GR and other gravitational
theories.
2
The Physics of GWs





Like Maxwell’s Equations (E-M waves),
Einstein’s field equations have radiative
solutions.
In Einstein’s formulation, GWs propagate at c
(i.e. the graviton has zero mass).
GWs are produced by accelerated
mass-energy (like E-M
waves produced by
accelerated charges).
Also like E-M waves, flux
falls as r-2.
GW exist in all theories of
gravity.
3
c. 1947, Princeton University
More Physics of GWs



The first non-vanishing radiative multipole is the
quadrupole.
There are two independent polarizations separated by
45˚: h (“h-cross”) and h+ (“h-plus”).
In general, GWs consist of a linear combination of the
two states.
h
h+
LIGO website
Wikipedia


Wikipedia
Einstein never took GWs seriously, he thought their effect
was just too small to ever be detected.
Arthur Eddington commented, “Gravitational waves
4
propagate at the speed of thought.”
What Will We See?


Strength of GWs best expressed as a
dimensionless quantity, the strain:
h ≡ DL / L
(i.e. the fractional length change).
Assuming GWs couple only to the quadrupole
moment:
non  symm .
 EKin.
G
2


c
GQ

h~ 4 ~
2
c r
c r




If the energy ~ Mc2 then:
h~
510-14
/ (r / pc)
5
What Will We See?







1M of GWs  h ~ 510-14 / (r / pc)
Galactic center (8kpc)  h ~ 10-17
Virgo cluster (17Mpc)  h ~ 10-21
Hubble distance (c/H0~4Gpc)  h ~ 10-23
This OoM estimate is optimistic; most
sources will radiate much less than 1M in
GWs.
Notice strain goes as r-1 (flux goes as r-2).
Note that the strain is tiny NOT because the
radiated energy is small (it’s huge), but
because space-time is a “stiff medium”.
6
What Else Will We See?



GW sources cannot be much smaller
than a Schwarzschild radius: 2GM / c2
They also cannot emit strongly at
periods shorter than the light travel
time around the circumference:
P  2pRSch / c  4pGM / c3
This implies frequencies (f = P-1):
f  c3 / 4pGM ~ 1.6104 Hz (M / M)
7
(Indirect) Evidence of GWs




PSR B1913+16 was the first binary pulsar
to be discovered (Hulse & Taylor 1975).
Observed for over 30 years.
Weak radio source (1 mJy at 1400 MHz).
Joseph H. Taylor Jr. and Russell A. Hulse
shared the Nobel Prize in Physics in 1993:
"for the discovery of a new
type of pulsar, a discovery
that has opened up new
possibilities for the study of
gravitation."
c. 1993, Nobel Foundation
8
c. 1993, Nobel Foundation
GWs Are Out There!



Orbital parameters for PSR B1913+16 are known
to extremely high accuracy (both relativistic and
non-relativistic measurables).
Binary should emit energy as GWs
 system loses energy
 orbit should shrink
 the period should decrease
GR says the rate of period decrease is:
(Weisberg &
Taylor 2004)

Using the measured values and correcting for the
relative acceleration between the solar system
and the binary:
P
measured
Pcorrected
 1.0013 0.0021
9
Building Up Our Confidence



The dominant
dissipation in the
binary is energy loss
by GWs (not mass
loss or tidal drag).
No GWs directly
detected yet.
However, the HulseTaylor PSR has
convinced us they
exist (and that we
understand them
relatively well).
10
Weisberg & Taylor 2004
Orbital Shrinkage

The two NSs will merge in about 300 Myr.
11
Weisberg et. al, 1981
Sources of GWs I

Coalescing Compact Binaries
• Binary companions will eventually
merge (since orbit is shrinking).
• Can consist of NS/NS (e.g. HulseTaylor), BH/BH, or BH/NS.
• Small sizes, large masses, and huge
orbital velocities  efficient GW
emission.
• Signal will look like a chirp (a technical
term).
12
Coalescing Compact Binaries

To first order, a chirp signal is
described by its amplitude and change
in frequency over time:
A  M c5 3 f
23
r 1
5 3 11 3

f  Mc f
where Mc is the “chirp mass”:
Mc

35

M 1M 2 

M 1  M 2 1 5
Measure f, f-dot, and A, solve for Mc
and get r (distance to source).
13
Coalescing Compact Binaries

Can learn much from the exact waveform:
• Harmonic content  eccentricity of orbit
• Overall modulation  mass ratio of the two objects and
spin-orbit coupling (i.e. frame-dragging)
• Higher-order corrections  mass and spin of constituents
• h versus h+  orbital inclination
• Timing of end point (merger)  equation of state of nuclear
matter
14
Sigg 1998
Coalescing BHs

The BH/BH chirp waveform has three
phases:
• Inspiral (BHs approach each other)
• Merger (actual coalescence of BHs)
• Ringdown (new BH relaxes from excited
state)
15
MPI for Gravitational Physics/W.Benger-ZIB
Sources
Sources of
of GWs
GWs III

Normal Binary Stars
• Orbital periods  1 hour  fGW  10-3 Hz
• Can only be detected in space (discussed
later)
16
LISA website
Sources of GWs III

Core Collapse Supernovae (CC SNe), if they explode
asymmetrically:
• Some explosions are asymmetric.
• How asymmetric, we don’t know very well.
• Initial neutrino emission drains much of the energy that
could go into GWs.
• Possibly ~10-3 Mc2
could go into GWs.
• If we observe a CC
SN in E-M radiation
and GWs, we can
compare the
propagation speed
of GWs to c.
17
LISA website
Sources of GWs IV

(Super) Massive BHs
• M  105 M BH swallowing a nearby object
(especially another Massive BH).
• Can only be detected in space (discussed
later).
18
LISA website
Sources of GWs V

Isolated PSRs:
• Asymmetric about their rotation axis 
nonzero quadrupole moment.

Stochastic Background
• Could be caused by density fluctuations in
the early universe (like a GW CMB).
• If measured, it links us to the Planck Era
and can discriminate between different
cosmological models!
• Most models predict that the strain would be
quite small.
19
Joseph Weber (1919 – 2000)




Father of GW astronomy
Started on GW detection in 1958 (Weber
1960)
Built first detector in 1966 at Univ. of MD
“Resonant Mass” detector (or “Weber Bar”)
• Two 1.5-ton Al cylinders (i.e. bars) with
piezoelectric crystals glued on.
• When squeezed (by a passing GW) the crystals
develop electric voltages.
• Strung many crystals together to amplify the
signal.
• Bars were placed at separate locations sift out
random noise.
• A GW could excite the fundamental longitudinal
mode of the bar, ~1657 Hz, which would induce a
voltage across the bar.
20
Direct “Detection” of GWs





Weber claimed extremely strong GW detections, ~1 per
month, from the late 60’s through the 70’s.
No one could explain the large amplitudes theoretically.
Other groups built bigger and more sensitive Weber Bars,
but could not reproduce his results.
By 1975, nearly everyone in the
field agreed that Weber’s detections
were simply noise.
Through the early 80’s, better
Weber bars were built using better
digital electronics and cryogenic
cooling techniques (to decrease
noise), but none successfully
detected GWs.
21
AIP Emilio Segrè Visual Archives
Using Lasers to See GWs






As early as 1956, laser interferometry was
proposed to search for GWs.
Even Weber suggested this in the late 60’s.
If a laser is bounced between two mirrors,
the distance between them can be accurately
measured.
If the separation is large compared to the
GW, then it will appear as a plane wave.
The GW will stretch and compress the spacetime distance between the mirrors.
This tiny change in distance can be detected
using interferometry.
22
Michelson Interferometers



Most designs are based on the Michelson
interferometer with response function
A(W)  sinc(WL/c)
To effectively increase L, can bounce the light many
times (off-axis) before detection, but this degrades
the signal.
Instead, use a Fabry-Perot cavity:
• Partially transmitting input mirror
• Highly reflective rear mirror
• Adjust length to multiple of laser wavelength and
cavity becomes resonant
• This increases power without degrading signal!

Response becomes A(W)  sinc(WL/c)FPI(WL/pc)
where FPI(x) = |t1 / 1-r1r2eix|2
L
L
23
Sigg 1998
More Michelson Interferometers

Recycle lost input power with a Power-Recycling mirror:
• Place another partially transmitting mirror at the input.
• Form another resonant cavity.
• Recycle light that would be lost back out the input.

Make signal resonant with a Signal-Recycling mirror:
• Place another partially transmitting mirror to the antisymmetric port.
• Can shape interferometer response around a narrow
frequency band.



If both are used, it’s called Dual-Recycled.
Response becomes A(W)  sinc(WL/c)FPI(WL/pc)GR
where GR is the total gain from the recycling cavity(ies).
All combinations of recycling and Fabry-Perot arms
can be used.
24
Sigg 1998
The LIGO Project



Laser Interferometer Gravitational wave
Observatory
Collaboration between Cal Tech and MIT.
Two widely separated sites under common
management (to make coincidence
measurements)
Hanford, WA (LIGO website)
25
Livingston, LA (LIGO website)
The LIGO Project






Interferometers with 4km arms.
Can operate several interferometers simultaneously.
Hanford site has 4km and 2km arm detectors.
Well collimated lasers.
Vacuum of 10-9 torr H and 10-10 torr other gases.
Lifetime of at least 20 years.
One 4km arm (LIGO website)
26
Beam splitter (LIGO website)
The LIGO Interferometer




Power-recycled Fabry-Perot
interferometer.
Sensing and control system
Seismic
isolation
system (not
shown)
Suspensions
27
Interferometer
(Abbott
al. 2007)
Interferometer
(LIGOet
website)
The LIGO Mirrors





~10 kg cylinders 25cm in
diameter and 10cm thick.
Made of high-purity fused
silica.
Permanent magnets glued
to the back to control
longitudinal and angular
orientation.
Permanent magnets glued
to the side to control
sideways motion.
Coil drivers mounted on
suspension cage to adjust
force on magnets/mirrors
(by changing current in
the coils).
28
Mirror and Cage (LIGO website)
The LIGO Laser



Solid state diode-pumped Nd:YAG
laser.
Operates at 10W and 1064nm.
Diffraction
limited
beam of
width ~30
to 40mm.
Part of the laser setup
(LIGO website)
29
The LIGO Scientific Collaboration!!!
30
Sources of Noise From the Laser


Shot noise:
fluctuations in the
number of
photons in the
beam
Light amplitude
and laser
frequency noise:
the laser’s beam
isn’t perfect
31
Sigg 1998
Sources of Noise From the Laser


Scattered light:
light can scatter
into and out of the
beam path (back
scatter is why
fiber optics aren’t
used)
Beam jitter: the
laser’s output
angle and position
isn’t perfect
Sigg 1998
32
Sources of Noise From the Laser

Residual gas
column density
fluctuations: if the
vacuum isn’t
perfect, changes
in gas density will
change the index
of refraction
33
Sigg 1998
Seismic Noise





Affects mirror motion directly
The Earth is in constant motion from
volcanoes, ocean
waves, wind, and
lunar tidal forces
Strongest from
0.1Hz to 10Hz
Huge source of
noise on Earth at
low frequencies
Absent in space
Sigg 1998
34
Thermal Noise in Suspensions



Also affects mirror motion directly
Heat excites motion in suspension
elements
Damped by
suspending mirrors
as pendulums
hung with thin
wire.
35
Sigg 1998
Thermal Noise in Mirrors




Again affects mirror motion directly
Heat excites normal modes in the
mirrors
themselves
Frequencies are
more toward kHz
range (sorta)
Can be modeled
well if the mirrors
are fabricated
accurately
Sigg 1998
36
Radiation Pressure Imbalance


Number of photons hitting a mirror will
fluctuate based on photon counting
statistics
The recoil of these
photons will
introduce a small
force on the
mirrors
37
Sigg 1998
Gravity Gradient Noise



Any mass place near a mirror will exert
a gravitational force on the mirror
The Earth’s
internal seismic
waves and density
fluctuations in the
atmosphere are
the main concern
Sets the ultimate
limit in sensitivity
for Earth-based
detectors
38
Sigg 1998
Other (Extreme) Sources of Noise




Radiometer noise
Electric field fluctuations
Magnetic field fluctuations
Cosmic ray muons
39
Noise Summary


Red curve is LIGO goal (current LIGO).
Blue curve is theoretical limit on Earth.
40
Sigg 1998
Noise Summary
41
LIGO website
Noise Summary (on Earth)




Red curve is Initial (current) LIGO.
Blue curve is theoretical limit on Earth.
Can possibly (one
day) model or
eliminate noise
from all sources
and get down to
blue curve on
Earth. (A LIGO)
Or, you can go into
space!!! (LISA)
42
Sigg 1998
LIGO Results: Fourth Science Run (S4)





Most recently released data set.
22 Feb 2005 to 23 Mar 2005.
Used all three LIGO interferometers
(2km and 4km at Hanford, 4km at
Livingston) and GEO 600 in Germany
(more on GEO later).
~570 hours of observation time on
each LIGO interferometer.
~402 hours when all three were
collecting data simultaneously.
43
LIGO Results: Stochastic Background

A stochastic background of GWs has been
predicted from many possible sources:
• Amplification of quantum vacuum fluctuations
(Grishchuk 1975, Grishchuk 1997, Starobinsky 1979)
• Pre-Big-Bang models (Gasperini & Veneziano 1993, Buonanno et
al. 1997)
•
•
•
•
•

Phase transitions (Kosowky et al. 1992, Apreda et al. 2002)
Cosmic strings (Caldwell & Allen 1992, Damour & Vilenkin 2005)
Rotating NSs (Regimbau 2001)
Supernovae (Coward et al. 2002)
Low-mass X-ray binaries (Cooray 2004)
Despite much searching, none have been
discovered thus far.
44
LIGO Results: Stochastic Background




192 second long intervals of data with 1/32
Hz frequency resolution.
S4 results are beginning to differentiate
between different cosmological models.
S5 (and A LIGO) will
probe even
more parameter
space.
Results are a
90% upper limit
on the GW
background.
45
Abbott et al. 2006
LIGO Results: Stochastic Background




10 trials were performed with artificially
injected amplitudes.
Left (gray) error bars denote the
theoretical errors.
Right (black)
error bars denote
the standard
errors over the
10 trials.
Seem to recover
injected signal
quite accurately!
Abbott et al. 2006
46
LIGO Results: GWs From Single PSRs







Appeared on astro-ph on 4 Apr 2007.
95% upper limits on GW amplitudes for 78 PSRs.
Tightest strain upper limit is 3.210-25
Strain is directly related to e, a PSR’s equatorial
ellipticity.
Smallest ellipticity is
8.5  10-7
Strange quark stars
or hybrid stars have
e ~ 10-5
More conventional
NS EOSs have
e ~ 10-7
47
Abbott et al. 2007
LIGO Results: Transient GW Bursts



Appeared on astro-ph on 6 Apr 2007.
90% upper limit on mean rate of GW bursts:
0.15 per day
Assuming isotropic emission of GWs (for OoM):
EGW



r 2c3
2
2pf 0 2 hmin

4G
hmin = best
sensitivity and
assume 50%
efficiency
r ~ 10kpc 
810-8 M in GWs
r ~ 16Mpc 
0.2 M in GWs
48
Abbott et al. 2007
LIGO Results: Transient GW Bursts


Monte Carlo was run to test efficiency.
Statistical errors comparable to size of
symbols.
49
Abbott et al. 2007
LIGO Results: Transient GW Bursts




CC SN model with non-spinning 11M
progenitor (and 50% detection efficiency)
visible to ~0.2kpc.
CC SN model with spinning 15M
progenitor (and 50% detection efficiency)
visible to ~8kpc.
A pair of merging 10M BHs visible to
~1.5Mpc.
A pair of merging 50M BHs visible to
~60Mpc.
50
The Other Guys

TAMA300 (http://tamago.mtk.nao.ac.jp/)
• Tokyo, Japan and started in 1995
• 300m dual-recycled Fabry-Perot
• Test-bed for new interferometer technologies

GEO 600 (http://geo600.aei.mpg.de/)
• Hannover, Germany (British/German) and started
in 2002
• 600m dual-recycled Fabry-Perot
• Comparable sensitivity to LIGO
• Between 2002 and 2006, participated in several
data runs in coincidence with LIGO

VIRGO (http://www.virgo.infn.it/)
• Pisa, Italy (Italian/French) and started in 2003
• 3km power-recycled Fabry-Perot
• Comparable sensitivity to LIGO
51
The Future of LIGO: S5





Fifth Science Run began in Nov
2005.
Expected to continue into late
2007.
Goal of collecting a full year of
coincident LIGO data.
Uses all 3 LIGO interferometers and
GEO 600 as well.
First science run at design
sensitivity.
52
The Future of LIGO: Einstein@Home





http://einstein.phys.uwm.edu/
Based on SETI@Home (Mo will talk
about this).
Uses idle computer time to search for
GWs from PSRs in data from LIGO and
GEO.
Supported by the American Physical
Society (APS) and by a number of
international organizations.
Currently searching 840 hours of data
from S5.
53
The REAL Future of LIGO







Advanced LIGO (A LIGO).
Sensitivity increase by ~10x.
Lowest frequency goes from ~40 Hz to
~10 Hz.
During first several hours of operation
data will exceed S5.
Observe binary NSs ~15x further 
event rate 3000x greater.
NS-BH binaries visible to 650Mpc.
BH-BH binaries visible to z=0.4.
54
Advanced LIGO




Laser power increased from 10W to
~200W.
Test masses larger in diameter (~34cm)
to reduce internal thermal noise.
Test masses more massive (~40kg) to
reduce radiation pressure noise.
Test masses suspended by fused silica
fibers to reduce suspension thermal
noise.
55
Advanced LIGO



Performance dominated at most frequencies
by quantum noise of sensing test mass
positions.
Should take this architecture to its technical
endpoint; it is as sensitive as one can make
an interferometer based on a Fabry-Perot
Michelson configuration.
Further advances will come from R&D that is
just beginning:
• cryogenic optics and suspensions
• purely reflective optics
• readouts that fully exploit the quantum nature of
the measurement

These developments will help instruments in
the second decade of this century.
56
Advanced LIGO…Someday






2008: Receipt of funding for the fabrication
and construction project.
2009: Delivery of first interferometer
hardware to the observatory staging facilities.
2011 (early): Decommissioning of initial LIGO
at Livingston and simultaneous start of
installation of Advanced LIGO there.
2011 (late): Decommissioning of initial LIGO
at Hanford and simultaneous start of
installation of Advanced LIGO there.
2013: Both observatories in operation.
GEO 600 is a full partner in Advanced LIGO,
participating at all levels in the effort.
57
LISA: GWs IN SPACE!!!!






Laser Interferometer Space Antenna
Space-based GW observations first
proposed in 1977
PLANNED LAUNCH DATE: 201X
MISSION DURATION: Five years for
nominal mission (10 years extended
mission)
Jointly sponsored by the European
Space Agency (ESA)
Will measure the (change in) distance
between test masses separated by 5
million km with a precision of 10pm.
58
The LISA Spacecraft(s)





3 identical cylinders 1.8m in diameter and 0.5m high.
Each cylinder supports a Y-shaped tubular structure
that contains two instruments.
A cover across the top (not shown) will protect the
detector from photon pressure from the Sun and
variable solar B field (two external sources of noise).
The Y-shaped structure is gold-coated and suspended
by stressed-fiberglass
bands to thermally isolate
it from the spacecraft.
Two radio antennas (not
shown) will be mounted to
each cylinder for
communication with
59
Earth.
LISA website
The LISA Interferometer





Each spacecraft transmits light to the other two
spacecraft.
Rather than reflecting the received light, each
spacecraft transmits a new beam back.
Each spacecraft then compares the signals to
measure the variation in distance between them.
Thus, each cylinder acts as both the light source and
mirrors at the end of a
Michelson interferometer
arm  it behaves like
three independent
interferometers.
This allows GW
polarizations to be
60
detected.
LISA website
The LISA Payload

The Y-payload includes two identical
instruments, each including:
• 30cm telescope for transmission and reception of
laser signals
• an optical bench that contains interferometer
optics
• an inertial sensor
• a test mass shielded from non-gravitational
disturbances
• a capacitor plate
arrangement for
measuring the position
of the spacecraft with
respect to the test
61
mass
LISA website
The LISA Laser
1W at 1mm
LISA website
62
The LISA Accelerometer
The test mass is in the center,
it’s surrounded by capacitor plates.
LISA website
63
The LISA Test Masses & Thrusters
4cm highly polished cubes



To keep test masses floating
freely, distance between
them and surrounding
spacecraft is constantly
monitored.
If there’s a shift,
microthrusters fire to move
the spacecraft into position.
Minimize drifting by adding
biases in initial velocity and
position (Povoleri 2006).
Optical bench with test mass
64
mNewton thruster, accurate to 10nm
LISA’s Orbit




Each spacecraft will be in an Earth-like orbit
around the Sun.
They will maintain an equilateral triangular
configuration with center 20° behind the Earth
and side length 5 million km.
The plane of the
formation is tilted 60°
below the ecliptic.
The changing
orientation makes it
possible to determine
the direction of a GW
source.
65
LISA website
LISA’s Sources of Noise

Internal sources:
• E field generated by the spacecraft computer
acting on the test masses
• Effects from residual gas pressure near the test
masses
• Thermal radiation by the electrodes used to
measure the spacecraft position

External sources:
•
•
•
•

Solar wind buffeting
Spacecraft drift
Test mass charging
Interference from the interplanetary magnetic
field
Any of these disturbances may cause
movement of the test masses or changes in
the distance of the interferometer arms.
66
What Will LISA See?



Low-frequency GWs that will never be
detectable by terrestrial detectors (due
to gravity gradient noise).
Galactic compact binaries long before
coalescence.
Extragalactic
(S)MBH
binaries in
the final
months of
coalescence.
67
LISA website
What Might LISA See?


If SNe Ia occur when WD binaries coalesce, LISA will
determine the directions to, and time of collision, for
the next 500 SNe Ia in our galaxy.
Close encounters between SMBHs and compact or
MS stars will emit GWs:
• the encounters will
produce transient
bursts of GWs
• Models show an
event rate of ~15/yr
in our galaxy and
~3/yr in the Virgo
Cluster detectable
by LISA.
Rubbo 2006
68
What’s LISA’s Sweetspot?







GW observations  calculated energy of GWs (MGW)
Let dMGW be the error in MGW
Consider a good detection if dMGW / MGW  5%
Monte Carlo simulations run as a function of MGW
and distance
Best masses are
around Mz = 106 ±
(a few) M where Mz
= M(1+z)
Mass range shrinks
with redshift
Event rates also
calculated: 1-5
events/yr @ z~2-3
z = 0.5
69
Hughes 2005
GWs: Perpetually The Wave of the Future






The theory of GWs is pretty stable and much is wellunderstood.
There are places for improvement (post-Newtonian
and other forms of gravity, numerical simulations of
signal waveforms, further rate calculations, etc.).
Although there’s no direct detections yet, current
detectors have placed quite strong limits on many
astrophysical sources of GWs.
New technologies and upgrades of older detectors
are coming (relatively) soon.
Massive upgrades and space-based observations are
in the further future.
“In astronomy, it’s always the next telescope that’s
gonna solve all your problems.”
70
--James Graham
Gravitational Wave Astronomy
71