of 16
Nobel Lecture: LIGO and gravitational waves II
*
Barry C. Barish
California Institute of Technology, Pasadena, California 91125, USA
(published 18 December 2018)
DOI:
10.1103/RevModPhys.90.040502
CONTENTS
I. Introduction
1
II. The LIGO Laboratory
2
III. The LIGO Scientific Collaboration (LSC)
3
IV. Initial LIGO Performance and Science
4
V. Advanced LIGO
5
VI. The Black Hole Merger Event (GW150914)
6
VII. More Black Hole Mergers
8
VIII. Science Implications of the Observed
Black Hole Mergers
9
IX. Virgo Detects Gravitational Waves
9
X. Neutron Star Binary Merger
10
XI. Future Directions for LIGO and Gravitational Waves
12
XII. The Future
14
XIII. Final Remarks and Reflections
15
XIV. LIGO
Some Key Members
16
Acknowledgments
16
References
16
I. INTRODUCTION
The observation of gravitational waves (see Fig.
1
)inthe
Laser Interferometer Gravitational-wave Observatory
(LIGO) was announced on 11 February 2016 (
Abbott
et al.
,
2016a
), 100 years after Einstein proposed the existence of
gravitational waves (
Einstein, 1916
,
1918
). This observation
came after more than 50 years of experimental efforts to
develop sensitive enough detectors to observe the tiny
distortions in spacetime from gravitational waves. The
Nobel Prize for 2017 was awarded to Rainer (
Rai
)
Weiss, Kip Thorne and myself
for decisive contributions
to the LIGO detector and the observation of gravitational
waves
. In fact, the success of LIGO follows from decades
of R&D on the concept and techniques, which were covered
in Rai Weiss
Nobel Lecture, followed by the design,
construction and evolving the LIGO large-scale interferom-
eters to be more and more sensitive to gravitational waves.
This work has been carried through the LIGO Laboratory
and the scientific exploitation through the LIGO Scientific
Collaboration, having more than 1000 scientists, who author
the gravitational wave observational papers. In addition,
many others made important contributions to the science of
black holes, numerical relativity, etc.
In these three lectures, Rai, Kip and I tell the story of LIGO
and gravitational waves in three parts. Rai covers the physics
of gravitational waves, the experimental challenges and some
of the pioneering interferometer work. He highlights the
experimental challenges and some important early innovations
that were proposed, tested at small scale and have been
incorporated in the LIGO interferometers. In this lecture, I
describe the LIGO project and the improvements that led to
detection of merging black holes in Advanced LIGO. I also
describe some key features of the interferometers, some
implications of the discoveries, and finally, how we envision
the evolution of LIGO over the coming decade. Kip will talk
FIG. 1.
Figure
1
is the gravitational wave discovery figure that
was observed by members of the LIGO Scientific Collaboration
within minutes after the event
was recorded in Advanced LIGO.
Each of the three figures shows the detected
strain
signals in
units of
10
21
vs time. The top trace is the observed waveform
detected in the Hanford, Washington interferometer, and the
middle trace is the observed waveform in Livingston, Louisi-
ana. The two signals are almost identical, but are shifted by
6.9 msec and are superposed in the bottom trace. From the
Caltech/MIT/LIGO Lab.
*
The 2017 Nobel Prize for Physics was shared by Rainer Weiss,
Barry C. Barish, and Kip S. Thorne. These papers are the text of the
address given in conjunction with the award.
REVIEWS OF MODERN PHYSICS, VOLUME 90, OCTOBER
DECEMBER 2018
0034-6861
=
2018
=
90(4)
=
040502(16)
040502-1 © 2018 Nobel Foundation, Published by the American Physical Society
about some early personal history, theoretical advances that
were crucial to making and interpreting the LIGO, and finally,
his vision of the future opportunities in this new field of
science.
Our three talks can be read as a series having some overlaps,
or can each be read individually. In our lectures, we single out
some individuals who played key roles in the discovery of
gravitational waves, but by necessity, we have left out many
other very important contributors.
II. THE LIGO LABORATORY
I became Principal Investigator of LIGO in early 1994 and,
consequently, my part of our story begins at that point. A few
months earlier, I had become
available,
due to the unfortu-
nate cancellation of the Superconducting Super Collider
(SSC) in Texas by the U.S. Congress (
Riordan, Hoddeson,
and Kolb, 2015
). The SSC was the realization of a long
process by the U.S. and world particle physics communities,
conceptually and technically, to develop a facility that would
open particle physics to a new energy regime, where there
were strong arguments that science beyond the standard model
of particle physics would become accessible. The Higgs
phenomenon had been proposed to explain the origin of mass
for the elementary particles, and the search for the associated
Higgs particle was to be the first major goal of the SSC. It was
eventually discovered at the Large Hadron Collider at the
European Organization for Nuclear Research (CERN)
Laboratory in Geneva, Switzerland that has some of the
capability of the SSC.
The SSC was designed to have two complementary
detectors with different features that would complement each
other and confirm discoveries. I was co-spokesperson with
Bill Willis of Columbia University for the Gammas,
Electronics, Muons (GEM) detector, one of two large
detector facilities. We had spent several years in the early
1990s developing concepts, technologies and design for
GEM, and in 1993, the conceptual design for GEM had
been successfully reviewed. We were just embarking on the
final technical design and preparing for the beginning of
construction when the SSC was canceled by Congress. This
was obviously extremely disruptive to all involved in the
SSC. From GEM, a significant contingent of the physicists
joined the CERN Large Hadron Collider experiments, and
many of our technical developments were incorporated in the
CMS and Atlas detectors at CERN.
I had decided not to join the CERN experiments, at least
not right away, because I preferred to take my time to decide
what I wanted to do next. In fact, I was extremely busy on
all the tasks involved in closing down the GEM facilities,
and especially, in helping many colleagues I had hired to the
SSC find jobs. However, I was approached by the California
Institute of Technology (Caltech) to become the LIGO
Principal Investigator in early winter 1994. My previous
roles on LIGO had been advisory, as I had been a strong
supporter for developing the experimental gravitational
wave effort at Caltech. I agreed to take over the LIGO
leadership, and my first task was to strengthen the LIGO
team and to evaluate and revise the LIGO proposal to the
National Science Foundation (NSF). The NSF was on a tight
schedule to make a decision whether to move forward with
LIGO. They wanted an external review of the revised
proposal by early summer 1994 and a discussion at the
National Science Board later in the summer. They needed a
decision whether to go forward with the LIGO construction
project by fall.
To organize the LIGO effort, I created a new structure for
the project built around the LIGO Laboratory that unified the
Caltech and Massachusetts Institute of Technology (MIT)
efforts. We divided responsibilities and created an overall
concept for developing the two distant instruments in
Hanford, Washington and Livingston, Louisiana. The decision
to build LIGO at those two sites had already been made,
following an evaluation by the NSF of about 20 proposals.
After becoming LIGO PI, my first priority was to make
some key hires to strengthen the project, especially from the
SSC where some extremely talented individuals had become
available. Most importantly, I hired Gary Sanders as project
manager, the same position he had with me for GEM at the
SSC. Together, we rapidly filled out key positions on LIGO
with other senior members, including Albert Lazzarini as
integration manager, John Worden for the large vacuum
system, Jay Heefner (now deceased) and Rolf Bork to develop
digital controls, Dennis Coyne as chief engineer, and others to
fill out the initial team.
The next task was to integrate the new members of LIGO
with the very talented existing LIGO staff members that
included Stan Whitcomb, Robbie Vogt, Bill Althouse, Mike
Zucker, Fred Raab at Caltech, and the MIT group under Rai
s
leadership that included David Shoemaker and Peter Fritschel.
The combined Caltech-MIT effort became the LIGO
Laboratory, which was to be the organization responsible
for the construction and operation of LIGO. LIGO Laboratory
is jointly operated by Caltech and MIT through a Cooperative
Agreement between Caltech and NSF. LIGO Laboratory
includes LIGO Hanford and Livingston Observatories,
Caltech and the MIT LIGO facilities. There are currently
178 staff: scientific (including academic staff, postdocs,
grad students, engineers, and technicians), and admini-
strative support staff. I stepped down from being LIGO
Director in 2006 and Jay Marx became the second
LIGO Lab Director. Jay very capably led further improve-
ments of initial LIGO to design sensitivity, as well as the
early developments of Advanced LIGO. In 2011, Dave Reitze
became LIGO Executive Director and has done a superb job of
leading LIGO through the construction and commissioning of
Advanced LIGO, and most importantly, the first gravitational
wave detections.
In early 1994, we rapidly revised and re-costed the LIGO
construction proposal to NSF to account for the larger team,
increasing the planned staffing for the Hanford and Louisiana
sites, and incorporating a more ambitious technical infra-
structure. The plan was to make the initial implementation
more robust, as much as possible to be able to accommodate
building a second improved version (Advanced LIGO) within
the same infrastructure. The total cost of the increases
amounted to about $100M, bringing the total Initial LIGO
construction costs to almost $300M. At the time, this
corresponded to our asking for funding for the largest project
the NSF had ever undertaken.
Barry C. Barish: Nobel Lecture: LIGO and gravitational waves II
Rev. Mod. Phys., Vol. 90, No. 4, October
December 2018
040502-2
The NSF conducted the external review of our revised
proposal in late spring 1994 and we received a very
encouraging and strongly positive review. Before making a
decision, however, Kip Thorne and I were invited to present
LIGO to the National Science Board (NSB). This was very
unusual at the NSF, as the NSB does not normally interact
directly with proponents. At the NSB meeting, Kip presented
the theoretical underpinnings of gravitational waves, as well
as giving a description of the physics we would be able to
produce with detections. I presented our plans for the project,
which involved first building and exploiting Initial LIGO,
which would be based as much as possible on technologies
we had already demonstrated. This approach was hardly
conservative, because LIGO was such a huge extrapolation
from the R&D prototypes. It was a factor of a 100 in size and
at least as large a factor in technical performance. A key
feature of our plan was to initiate an ambitious R&D program
to develop and test Advanced LIGO technologies, immedi-
ately following the completion of Initial LIGO construction.
This would be carried out by keeping the key technical staff
that had developed Initial LIGO. This was an unusual request,
because Advanced LIGO was only a strategic concept and had
not been proposed at this stage. Following our presentations to
the NSB, we received formal approval at the full requested
funding for Initial LIGO, with a commitment to support the
crucial R&D program for Advanced LIGO.
The basic scheme for LIGO was to use a special high power
stabilized single-line laser (neodymium-doped yttrium alumi-
num garnet = Nd:YAG) that entered the interferometer and
was split into two beams transported in perpendicular direc-
tions. The LIGO vacuum pipe is 1.2 m in diameter and is kept
at high vacuum (
10
9
torr). The
test
masses are very high-
quality mirrors that are suspended, in order to keep them
isolated from the Earth. They are made of fused silica and
hung in a four-stage pendulum for Advanced LIGO. In the
simplest version of the interferometer, the equal length arms
are adjusted such that the reflected light from mirrors at the
far ends arrives back at the same time, and inverting one,
the two beams cancel each other and no light is recorded in the
photodetector. This is the normal state of the interferometer
working at the
dark port.
Many effects make the beams not
completely cancel, and the actual optical configuration is more
sophisticated.
When a gravitational wave crosses the interferometer, it
stretches one arm and compresses the other, at the frequency
of the gravitational wave. Consequently, the light from the two
arms returns at slightly different times (or phase) and the two
beams no longer completely cancel. This process reverses
itself, stretching the other arm and squeezing the initial arm at
the frequency of the gravitational wave. The resulting fre-
quency and time-dependent amount of light is recorded by a
photosensor and recorded as the waveform from the passage
of a gravitational wave. The experimental challenge is to make
the interferometer sensitive to the incredibly tiny distortions of
spacetime that come from a gravitational wave, while at the
same time, suppressing the various background noise sources.
The spacetime distortions from the passage of an astrophysi-
cal source are expected to be of the order of
h
¼
Δ
L=L
10
21
, a difference in length of a small fraction of the size of a
proton. In LIGO, we have made the length of the interferometer
arms as long as is practical, in our case 4 km, and this results
in a difference in length we must be sensitive to that is still
incredibly small, about
10
18
m. For reference, that is about
1000 times smaller than the size of a proton! If that sounds very
hard, it is!! Skipping the details, what enables us to achieve this
precision is the sophisticated instrumentation that reduces
seismic and thermal noise sources, effectively making the
statistics very high by having many photons traversing the
interferometer arms.
The initial version of LIGO was constructed during the
period from 1994 to 1999, employing technologies that
represented a balance between being capable of achieving
sensitivity levels where the detections of gravitational waves
might be
possible,
and using techniques that we had fully
demonstrated in our laboratories. LIGO was a huge extrapo-
lation from the 30 m prototype interferometer (
Shoemaker
et al.
, 1988
) in Garching, Germany and the 40 m prototype
(
Abramovici
et al.
, 1996
) at Caltech interferometers that
preceded it, and especially considering the very large NSF
investment, we needed to be confident of technical success.
In reality, from the best theoretical estimates at the time, we
anticipated that we would likely need to achieve sensitivities
well beyond those of Initial LIGO before achieving detections.
So, developing the techniques and building Advanced LIGO
was always an integral feature of our plans.
III. THE LIGO SCIENTIFIC COLLABORATION (LSC)
LIGO Laboratory, even after the strengthening in 1994,
was relatively small for building such a large, ambitious and
challenging construction project. The key hires that were
added to the original Caltech and MIT LIGO teams were
focused in areas needed for the construction of this
sophisticated and challenging project. There were other
crucial areas where we were weak. These were basically in
areas that would be needed to extract the science, building
computing facilities, data analysis infrastructure, search
algorithms and pipelines, etc. In addition, there was exper-
tise in hardware areas outside Caltech and MIT that could
be strengthened by involving expertise from the larger
worldwide community.
Gary Sanders and I both had worked within high energy
physics collaborations. We appreciated the value of such
collaborations, but believed LIGO needed a different model.
By 1997, in the middle of LIGO construction I made a
proposal to the NSF through a review committee chaired by
Boyce McDaniel of Cornell, who knew how high energy
collaborations worked. Our model for LIGO was somewhat
different. We wanted to create a collaboration for LIGO that
was focused on the science, an
open
collaboration where
individuals or groups could join if they could make significant
contributions to LIGO science. They would not be expected to
necessarily provide resources or hardware, as is done in high
energy physics collaborations. In order to make joining the
LSC as attractive as possible, we took a further step to insure
that Caltech and MIT LIGO Laboratory scientists would not
have a strong built-in advantage to do LIGO science. The step
was to make LIGO Laboratory individual scientists join the
LSC to do their science and that the science component of
Barry C. Barish: Nobel Lecture: LIGO and gravitational waves II
Rev. Mod. Phys., Vol. 90, No. 4, October
December 2018
040502-3
LIGO would be carried out through that organization. This
concept was endorsed by the McDaniel committee, the NSF
approved, and we initiated a fledgling collaboration.
It took some effort, but I managed to convince the Caltech
and MIT administrations and LIGO Laboratory scientific staff
of this plan. I asked Rai Weiss to become the first spokesperson
of the LSC and his credibility and approach helped to get the
collaboration off the ground. Rai was succeeded by Peter
Saulson of Syracuse, who also had a long history in LIGO, was
much respected, and was the first spokesperson from outside
Caltech and MIT. He was subsequently followed by David
Reitze (now LIGO Laboratory Executive Director), Gabriela
Gonzalez of Louisiana State University (LSU) and now David
Shoemaker (MIT). As the LSC matured, it has become more
democratic in involving collaborators, publication policies, etc.
The LSC has all the responsibilities to produce the science
with calibrated strain data. The data pipelines have been
developed through the LSC and the various analysis groups
on different science are organized and run by the LSC. LSC
members also participate in some hardware detector areas
and future detector R&D and planning. Also, LSC members
participate in calibration of the data and have played a major
role in detector characterization of the data. We have a
program for young scientists to spend time at the LIGO sites,
a
fellows
program.
The LSC has been extremely successful, the proof being
the strong and effective role it played in making the discovery
of gravitational waves, analyzing the data, interpreting the
results, and in writing up and presenting the results. The
strong and effective LSC role has continued to the present and
the announcement of detection of a neutron star binary merger
and all the follow-up measurements with various astronomical
instruments.
The LIGO Scientific Collaboration and its 1200 members
deserve shared credit for the discoveries reported here,
resulting in the Nobel Prize, which must be given to no more
than three individuals. The LSC has grown to more than
1200+ scientific collaborators, from 108 institutions and 18
countries! We conducted a study last year to figure out how to
evolve the LSC for the gravitational wave observational era
we are entering. We expect to institute some changes by the
time of the next data run schedule to begin about the end of
this calendar year, and we fully expect the LSC to play the
central role in producing the future science with LIGO.
IV. INITIAL LIGO PERFORMANCE AND SCIENCE
We had a two-step concept for LIGO, since beginning
the project. Initial LIGO (
Abbott
et al.
, 2009
), as much as
possible, was based on demonstrated methods and technol-
ogies, while the second stage, Advanced LIGO, was to
achieve significantly improved sensitivities, through imple-
menting methods and technologies that we would develop
through an ambitious post construction R&D program. It was
from that perspective that we proposed to the NSF that while
we would be commissioning, running and learning from
Initial LIGO, we would be funded to simultaneously carry
out an ambitious R&D program to develop the techniques that
would improve LIGO in a second step to a sensitivity where
detections would become
probable.
The NSF approved that
plan and funded what was to become a successful R&D
program that began in about 2000 and led to the Advanced
LIGO concept that was proposed and approved by the NSF in
2003. The actual project funding was awarded several years
later. I emphasize that our being able to carry out this R&D
and design program, keeping key individuals who developed
and built Initial LIGO, was crucial to our eventual success in
detecting gravitational waves with Advanced LIGO. Another
important point is that the Initial LIGO infrastructure was
designed such that the interferometer subsystems could be
evolved or replaced inside the same infrastructure (vacuum
vessels).
After the completion of construction of Initial LIGO (see
Fig.
2
), we began commissioning and rapidly achieved better
sensitivity than any previous gravitational wave detectors. As
a result, we embarked on our first gravitational wave search
data run. We did not detect gravitational waves, but we did set
new astronomical limits on a variety of possible gravitational
wave sources. Following the first data run, we made some
technical improvements that reduced the background noise
levels, both planned improvements and others from what we
had learned in the first data run, and then we embarked on a
second data run. Again, we did not detect gravitational waves,
and again we achieved new limits and published them on
various possible sources. All together, we repeated this basic
cycle for over a decade, improving sensitivity and taking data
for a total of six data runs at ever-increasing sensitivity (see
Fig.
3
). For the final data runs, the interferometer sensitivities
achieved very close to our original Initial LIGO design goals.
We searched for gravitational waves from several potential
sources: mergers of binary black holes, a black hole and a
FIG. 2.
The LIGO Interferometers in Hanford, Washington and Livingston, Louisiana. From the Caltech/MIT/LIGO Lab.
Barry C. Barish: Nobel Lecture: LIGO and gravitational waves II
Rev. Mod. Phys., Vol. 90, No. 4, October
December 2018
040502-4
neutron star, and binary neutron star systems. We also
searched for signals from continuous sources, such as known
and unknown pulsars, possible stochastic background signals,
and unmodeled signals from some new source we had not
specifically targeted. Unfortunately, even with the impressive
improvements in interferometer sensitivity, we did not detect
gravitational waves. The resulting limits on various sources
of gravitational waves were constraining for some of the
published models of gravitational wave production from
astrophysical phenomena.
The final Initial LIGO searches for black hole binary
systems were performed in collaboration with the Virgo
detector (
Aasi
et al.
, 2013
). Although we had not detected
signals, we were cautiously confident that the technical
improvements envisioned for Advanced LIGO would be
sufficient to finally achieve detection.
V. ADVANCED LIGO
By about 2004, the improved technologies developed
for Advanced LIGO (
Advanced LIGO, 2015
) were mature
enough to propose Advanced LIGO to the National Science
Foundation. After reviewing the proposal, the NSF continued
to support the technical developments in parallel with the
continued running of Initial LIGO. The Advanced LIGO
project received major funding through the NSF Major
Research Equipment and Construction (MREFC), leading
us to end the Initial LIGO scientific program and begin
construction of Advanced LIGO. Additional significant con-
tributions to Advanced LIGO included a prestabilized laser
system from the Max Planck Institute (Germany), test mass
suspension systems from the Science and Technology
Facilities Council (UK), and thermal compensation wavefront
sensors and interferometer control components from the
Australian Research Council.
The basic goal of Advanced LIGO is to improve the
sensitivity from Initial LIGO by at least a factor of 10 over
the entire frequency range of the interferometer (see Fig.
4
).
It is important to note that a factor of x10 improvement in
sensitivity increases the distanc
e we can search by that factor,
since we measure an amplitude. It thus increases the volume
of the Universe (or rate for most sources) searched for by a
factor of x1000. (The sensitivity to most sources is propor-
tional to the volume we search.) Therefore, there is a very
high premium in LIGO on increasing the range we can
search, and consequently, we spend a good fraction of our
time improving the sensitivity, rather than taking very long
data runs.
FIG. 3.
The evolving improved sensitivity of Initial LIGO.
FIG. 4.
Advanced LIGO sensitivity goal.
Barry C. Barish: Nobel Lecture: LIGO and gravitational waves II
Rev. Mod. Phys., Vol. 90, No. 4, October
December 2018
040502-5
The Initial and Advanced LIGO gravitational wave detec-
tors are Michelson interferometers with 4 km long arms. Both
use Fabry-Perot cavities to increase the interaction time with a
gravitational wave, and power recycling to increase the
effective laser power. Signal recycling at the output (dark)
port is a new feature of Advanced LIGO, and this changes the
control and readout systems. Signal recycling enables tuning
the sensitivity response to the physics goals, presently black
hole and neutron star mergers. For Advanced LIGO, the
design sensitivity is moved to lower frequencies (e.g. 40 Hz
down to 10 Hz).
The improved seismic isolation system uses both passive
and active isolation, and the improved test mass suspensions
use a quadruple pendulum. Higher laser power, larger test
masses and improved mirror coatings have been incorporated.
The Advanced LIGO interferometers are installed in the same
infrastructure, including the same vacuum system as used for
Initial LIGO.
The Advanced LIGO laser is a multistage Nd
YAG laser.
Our goal is to raise the power from
18
W in Initial LIGO to
180 W for Advanced LIGO, improving the high frequency
sensitivity, accordingly. The prestabilized laser system con-
sists of the laser and a control system to stabilize the laser
in frequency, beam direction, and intensity. For the results
presented here, due to stability issues, heating and scattered-
light effects, the laser power has only modestly been
increased. We plan to bring the power up systematically in
steps, studying these effects for the next data runs.
The key improvement in Advanced LIGO that enabled the
detection of the black hole merger was implementing active
seismic isolationand a quadruplesuspension system (see Fig.
5
).
The multiple suspension system moved all active compo-
nents off the final test masses and gives better isolation. Initial
LIGO used 25-cm, 11-kg, fused-silica test masses, while for
Advanced LIGO the test masses are 34 cm in diameter to
reduce thermal noise contributions and are 40 kg, which
reduces the radiation pressure noise to a level comparable to
the suspension thermal noise. The test mass is suspended by
fused-silica fibers, rather than the steel wires used in initial
LIGO. The complete suspension system has four pendulum
stages, increasing the seismic isolation and providing multiple
points for actuation.
The active seismic isolation senses motion and is com-
bined with the passive seismic isolation using servo tech-
niques to improve the low frequency sensitivity by a factor
of x100. Since the rate for gravitational events scales with
the volume, this improvement increases the rate for events
by
10
6
. This improvement enabled Advanced LIGO to make
a detection of a black hole merger in days, while Initial
LIGO failed to detect gravitational waves in years of
data taking.
VI. THE BLACK HOLE MERGER EVENT (GW150914)
The observation of the first black hole merger by Advanced
LIGO (
Abbott
et al.
, 2016b
,
2016c
,
2016d
,
2016e
) was made
on 14 September2015. Figure
1
shows the data, and Fig.
6
reveals the key features of the observed compact binary
merger. These are a result of the analysis of the observed
event shown in Fig.
1
. At the top of Fig.
6
, the three phases of
the coalescence (inspiral, merger, and ringdown) are indicated
above the waveforms. As the objects inspiral together, more
and more gravitational waves are emitted and the frequency
and amplitude of the signal increases (the characteristic chirp
signal). This is following by the final merger, and then, the
merged single object rings down. The bottom pane shows on
the left scale that the objects are highly relativistic and are
moving at more than 0.5 the speed of light by the time of the
final coalescence. On the right side, the scale is units of
Schwarzschild radii and indicates that the objects are very
compact, only a few hundred kilometers apart when they enter
our frequency band.
By comparing and fitting our waveform to general rela-
tivity, we have concluded that we have observed the merger of
FIG. 5.
Advanced LIGO multistage suspension system for the test masses with active-passive seismic isolation.
Barry C. Barish: Nobel Lecture: LIGO and gravitational waves II
Rev. Mod. Phys., Vol. 90, No. 4, October
December 2018
040502-6
two heavy compact objects (black holes), each
30
times the
mass of the Sun, and going around each other separated by
only a few hundred kilometers and moving at relativistic
velocities.
On the top of the figure, we see the three different phases
of the merger: the inspiral phase of the binary black holes
system, then their merger of the two compact objects, and
finally, there is a ringdown. The characteristic increasing
frequency and amplitude with time (the so-called chirp signal)
can be seen in the inspiral phase. The largest amplitude is
during the final merger, and finally, there is a characteristic
ringdown frequency. The two black holes inspiral and merge
together due to the emission of gravitational radiation coming
from the accelerations.
The bottom of Fig.
6
is even more revealing. The right-hand
axis is basically in units of about 100 km. That means the
separation between the two merging objects at the beginning
was only
400
km, and at the end
100
km. Therefore, these
30
solar mass objects are confined in a volume only about
twice the size of Stockholm, yet the final black hole has
60 times the mass of our Sun, or
10
×
10
6
times as massive as
the Earth. From the axis on the left, we see that when we first
observed this event, the two objects were moving at about
three-tenths the speed of light and that increases to over half
the speed of light by the time of the final merges!
In order to be confident that what we observed was a real
event and not a background fluctuation, we directly measure
the background probability by comparing coincidence time
slices for the two detectors, both in time (e.g.

10
ms) and out
of time! Since GW150914 occurred only days after we began
data taking with Advanced LIGO, this required taking about
one month of data before we could quantitatively establish the
probability that the event was real. In other words, in addition
to searching for coincidence in-time signals, we look for
coincidences between all the out-of-time slices during our data
taking. These background slices could not have come from
any physical phenomena traveling at relativistic speeds, like
gravitational waves. The total number of time slices we
compared was equivalent to in-time background levels equiv-
alent to over 67 000 years of data taking. Taking into account
the different event classes in our search, we reduce the limit on
the false alarm rate to 1 in over 22 500 years. This corresponds
to a probability that our observed event is accidental to
<
2
×
10
6
, establishing a significance level of
4
.
6
σ
. I empha-
size that the measured significance level is set from the
number of bins compared from 16 live days of data taking.
This represents a lower limit on the actual significance of
GW150914.
Figure
7
shows the statistical significance as described
above for the GW150914 event, compared to the measured
background levels, under two different assumptions. The
horizontal axis is a measure of the significance of the events
FIG. 7.
The statistical significance of the event as a generic transient event on the left, and directly searching for events from a merger
on the right.
FIG. 6.
The physics interpretation of the observed event as a
binary black hole merger.
Barry C. Barish: Nobel Lecture: LIGO and gravitational waves II
Rev. Mod. Phys., Vol. 90, No. 4, October
December 2018
040502-7
and the vertical access is the rate. The left-hand plot shows the
observed GW150914 event at the level of one event level, and
having statistical significance
σ
>
4
.
6
, as described above.
This plot assumes a generic signal shape for the event. The
right-hand plot shows a significance of
>
5
σ
, when a binary
coalescence form is assumed. Note that the second most
significant event in these data is about
2
σ
, which may well
also be a binary black hole merger, but at this early stage in
LIGO, we are only declaring
5
σ
events as gravitational wave
binary mergers.
The shapes of the waveforms that describe the merger,
coalescence and ringdown reveal the detailed parameters of
the merger. The orbits decay as the two black holes accelerate
around each other and emit energy into gravitational waves
determined by the
chirp mass,
as defined below, at leading
order in the strength of the binary
s gravity,
¼
ð
m
1
m
2
Þ
3
=
5
M
1
=
5
c
3
G

5
96
π
8
=
3
f
11
=
3
_
f

3
=
5
.
The next order terms enable the measurement of the mass
ratios and spins, the redshifted masses are directly measured,
and the amplitude is inversely proportional to the luminosity
distance. Orbital precession occurs when spins are misaligned
with the orbital angular momentum. GW150914 shows no
evidence for precession. The sky location is extracted from the
time delay between detectors and the differences in the
amplitude and phase in the detectors.
Using numerical simulations to fit for the black hole merger
parameters, we determine that the total energy radiated into
gravitational waves is
ð
3
.
0

0
.
5
Þ
M
c
2
. The system reached
a peak luminosity of
3
.
6
×
10
56
ergs
=
s, and the spin of the
final black hole
<
0
.
7
of the maximal black hole spin.
The main parameters of the black hole merger are sum-
marized in Table
I
.
With only two detectors we cannot locate the direction very
well, but comparing the time, amplitude and phase in the
Livingston and Hanford interferometers, we are able to locate
the gravitational wave as coming up from the Southern
Hemisphere within an area of about 600 square degrees.
Our most recent observations, discussed later, include Virgo
and with three detectors we determine the sky location to tens
of square degrees.
VII. MORE BLACK HOLE MERGERS
The first Advanced LIGO data run (O1) continued for four
months, from September 2015 to January 2016. Our second
data run (O2) ran from December 2016 to the end of August
2017. Much like the evolution described for Initial LIGO, we
are making improvements between Advanced LIGO data runs
and expect to achieve design sensitivity within a few years. We
are actively searching for other signals, besides binary merges,
including bust signals from phenomena such as supernova
explosions or gamma ray bursts, continuous wave signals
from spinning neutron stars (pulsars), stochastic background
signals, etc. So far, we have only detected binary merger
signals, but hope to detect others as our sensitivity improves.
We search for transient or burst events in two different
ways. First, we use an unmodeled search for a power excess,
using a wavelet technique, which makes no assumptions about
the expected binary merger waveform. The second search
method employs a matched template technique using binary
merger waveforms. We have populated the available mass 1 vs
mass 2 plane with templates that describe the waveform,
calculated using post-Newtonian calculations and/or numeri-
cal relativity, as required. There are several hundred thousand
templates, which are each multiplied by the observed noise in
each interval of time (matched template technique).
Since announcing our observation of GW150914, we have
reported (
Abbott
et al.
, 2016f
,
2017a
) several more observa-
tions of black hole mergers. As shown in Fig.
8
, we have so far
reported four black hole merger events having high statistical
significance (
>
5
σ
), and one of
2
σ
, which we do not declare as
an event. However, all the characteristics of that event are
similar to the other black hole mergers, except that it is farther
away, making the signal weaker and the signal to noise less
significant.
It is interesting that we have only one such event candidate,
having marginal signal to noise. This is a result of the
extremely steep cutoff of the noise background when requir-
ing a coincidence between the two sites, as is evident by the
extremely sharp fall-off of the background noise in Fig.
7
.
We can conclude that there is very little correlation between
the noisy events observed at the two sites. This was not
necessarily anticipated, as we worried about correlated noise.
However, this result is both more convincing, in terms of
the events we have observed and bodes very well for the
cleanliness of events in future observations by using the
coincidence technique.
TABLE I. The main parameters of the black hole merger.
Primary black hole mass
36
þ
5
4
M
Secondary black hole mass
29
þ
4
4
M
Final black hole mass
62
þ
4
4
M
Final black hole spin
0
.
67
þ
0
.
05
0
.
07
Luminosity distance
410
þ
160
180
Mpc
Source redshift,
z
0
.
09
þ
0
.
03
0
.
04
FIG. 8.
Event characteristics of reported black hole mergers in
LIGO.
Barry C. Barish: Nobel Lecture: LIGO and gravitational waves II
Rev. Mod. Phys., Vol. 90, No. 4, October
December 2018
040502-8
For the black hole merger events shown in Fig.
8
, notice
that the amount of time and number of cycles observable in
LIGO is very dependent on the mass of the black hole system.
The heaviest black hole we observed was the first one and
we only observed a few cycles, while for the lightest one
(GW151226) we have many cycles. Also, I repeat that the
candidate event, LVT151-12, has characteristics completely
consistent with the other events, but is about twice as far away
and therefore has lower statistical significance.
VIII. SCIENCE IMPLICATIONS OF THE OBSERVED
BLACK HOLE MERGERS
Gravitational waves represent a completely new way to
view the Universe. We have every reason to expect that we
will discover new phenomena and learn
new
astrophysics
from gravitational waves. This has already been realized from
the very first observations of gravitational events.
Conclusions from the first observations of black hole
merger observations from gravitational waves include the
following:
Stellar binary black holes exist.
They form into binary pairs.
They merge within the lifetime of the Universe.
Masses (
M>
20
M
) are considerably heavier than what
was known or expected of stellar mass black holes.
The fact that the observed black hole mergers in LIGO are so
massive has opened the question of how they were produced.
LIGO is strongly biased to detecting heavy masses, because
the stronger signal provides sensitivity to greater distance and
the volume observed grows like the cube of the distance.
Nevertheless, the creation of such large masses needs explan-
ation. If these stellar mass black holes come from the gravita-
tional collapse of heavy stars, it requires special conditions for
the parent star to survive, such as low-metallicity regions of the
Universe. Another possibility is that the heavy black holes were
produced in dense clusters, and a third is that they are primordial
and may even be associated with the dark matter; see Fig.
9
.
The next challenge will be to distinguish between these or
other possibilities for the origin of such heavy stellar black
holes. More events will give us distributions of masses and
other parameters, while larger signal/noise events will enable
determining the other feature of these mergers. For example,
are the spins of the merging black holes aligned, antialigned or
is there no correlation in the spins?
Our other major scientific goal is to test general relativity in
the important regime of strong field gravity. The merger of
black holes presents just such a new
laboratory
for these
studies. We can compare the observed waveform with the
predictions of general relativity looking for deviations. So far,
all tests are in good agreement with the theory.
The first test is just how well general relativity fits the
details of our data, or if there are any other features in the data.
At the present level, the deviations we see are all consistent
with the instrumental noise in our interferometers.
We have tried to quantify our level of agreement by adding
a dispersion term, which would indicate that some compo-
nents might not travel at the speed of light and are dispersed.
The large distance to the black hole merger events provides
good sensitivity for such tests. From Fig.
10
we see that very
little dispersion is allowed. The limit
α
¼
0
corresponds to an
upper limit on the mass of gravitons (a hypothetical particle
mediating the force of gravity). Assuming that gravitons are
dispersed in vacuum like massive particles, then the bound on
graviton mass is
M
g
<
7
.
7
×
10
23
eV
=c
2
More stringent tests of general relativity will be possible as
we collect more events, and in particular we can test alternate
formulations of general relativity.
IX. VIRGO DETECTS GRAVITATIONAL WAVES
Finally, very recently we reported (
Abbott
et al.
, 2017b
)
one more black hole merger event (GW170814) and for the
first time this was also observed in the Virgo detector near Pisa
in Italy (see Fig.
11
). Virgo is a collaboration of France, Italy,
Netherlands, Poland and Hungary. The founders of Virgo were
FIG. 9.
Masses of observed black hole mergers in LIGO.
FIG. 10.
Tests of the inclusion of a dispersion term in general
relativity. Upper limits are shown on the presence of a dispersion
term.
Barry C. Barish: Nobel Lecture: LIGO and gravitational waves II
Rev. Mod. Phys., Vol. 90, No. 4, October
December 2018
040502-9
Adalberto Giazotto and Alain Brillet and the present leader-
ship is Federico Ferrini as the European Gravitational
Observatory (EGO) director and Jo van de Brand as Virgo
spokesperson. Virgo is a 3 km interferometer, similar to LIGO,
but with some technical differences.
Adding Virgo, in this detection, not only gives independent
confirmation of the LIGO black hole merger detections,
but improves markedly the ability to triangulate. This is a
precursor to also adding KAGRA in Japan and LIGO-India
detectors to the network. The combination will give good sky
coverage and more than an order-of-magnitude improvement
in locating the direction of the source. This can be seen in
Fig.
12
, where the sky location of this last observed event is
dramatically better than the previous observed black hole
merger events.
The first three-way coincidence enabled a test of general
relativity that could not be carried out with LIGO alone,
because the two LIGO detectors are almost coaligned.
Including the nonaligned Virgo data, the polarization of the
gravitational waves can be studied to determine whether the
waves are transverse as predicted in general relativity.
GW170814 enabled the first tests for nontransverse polar-
izations. The first such analysis was performed using polar-
izations that are forbidden by general relativity and these
were disfavored in the analysis. More precise tests and
analysis will be possible in the future using events detected
in both LIGO and Virgo.
X. NEUTRON STAR BINARY MERGER
Two weeks after the Nobel announcement in October
and almost two months before this lecture, we announced
the first observation of a merger of a neutron star binary
system (see Fig.
13
)(
Abbott
et al.
, 2017c
). This was also the
first gravitational wave event to have electromagnetic counter-
parts observed in a large variety of astronomical instruments
(see Fig.
15
). This event initiates the eagerly sought after new
field of multimessenger astronomy.
On 17 August 2017, LIGO and Virgo registered a gravi-
tational wave signal from the inspiral of two neutron stars.
Many of us expected the first source we would observe in
LIGO would be from the merger of binary neutron stars,
because there is a known rate of detected binary neutron stars
in our own galaxies by radio telescopes. Although accurately
predicting the rate of mergers detectable by LIGO from these
data is not very precise, we knew it to be in the range of
detectability by LIGO. No such definite predictions were
possible for binary black hole mergers or for black hole
neutron star mergers (which we have not yet detected).
Neutron stars are very dense nuclear matter and one of the
main goals studying such describe superdense matter by a
relationship called the
equation of state.
The neutron stars
mass and equation of state determines size and tidal defor-
mation from the gravitational pull of the companion neutron
star. We cannot determine the equation of state from this event,
but this is a future goal from these detections.
From our estimation of the parameters, the masses of the
two compact objects turn out to be between 0.86 and 2.26
solar masses as shown in Fig.
14
. In fact, restricting the spins
to the range for binary neutron stars limits the mass
FIG. 11.
The Virgo Suspended Mass Interferometer in Cascina,
Italy.
FIG. 12.
Black hole merger GW170814 showing LIGO-Hanford, LIGO-Livingston and Virgo signals on the left, and the reconstructed
direction of the gravitational wave on the right.
Barry C. Barish: Nobel Lecture: LIGO and gravitational waves II
Rev. Mod. Phys., Vol. 90, No. 4, October
December 2018
040502-10
determination to between 1.17 and 1.60 solar masses, con-
sistent with being neutron stars. The distance to the source is
well determined from the amplitude of the merger gravita-
tional wave signal to be about 40 Mpc (
130
×
10
6
light years).
This event occurred only a few days after Virgo had joined
the gravitational wave network and greatly improved locali-
zation for the black hole merger event discussed above.
The localization of the neutron merger event, including
Virgo, is an oblong about 2 deg across and 15 deg long,
covering about 28 square deg.
The Fermi satellite observed a gamma ray burst
GRB170817A in the same region of the sky and the triggered
follow-ups identified the fading light from the event from
near NGC 4993, first by the 1 m Swope optical telescope
(
LIGO
et al.
, 2017
). The gravitational wave observation of the
neutron star merger provides a new independent method of
measuring the Hubble constant, the rate of expansion of the
Universe, using a method proposed by Bernard Schutz, called a
standard siren
measurement (see Fig.
16
)(
Schutz, 1986
).
The idea is to use the distances of galaxies determined just from
the gravitational-wave observations, by determining the lumi-
nosity (e.g. distance) directly from the observations. Standard
sirens are compact binary systems, consisting of neutron stars
or black holes, whose gravitational waveform as the compact
objects inspiral toward merger carries information about the
distance of the source, as well as the masses of the compact
objects and other parameters of the system. In this case, we use
the precise information of which galaxy the event occurred in
from the optical observations.
GW170817 represents the first joint detection of gravita-
tional and electromagnetic waves from the same astrophysical
source. All the data agree with the hypothesis that the source is
the merger of two neutron stars in the host galaxy NGC4993
in the constellation of Hydra. Analysis of the waveform of
GW170817 yielded a distance estimate of about 44 Mpc,
assuming that the sky position of GW170817 was exactly
coincident with its optical counterpart. The gravitational wave
distance estimate is completely independent of the cosmic
distance ladder derived from electromagnetic observations, so
that future measurements with more statistics from gravita-
tional waves can help distinguish the uncertainties between
the electromagnetic observations.
The observation in LIGO and companion observation of a
short gamma ray burst opened a full astronomical campaign
to observe this event in different wavelengths and devices,
including the large neutrino detectors (see Fig.
17
). I will not
summarize those results here (
Abbott
et al.
, 2017d
,
2017e
,
2017f
), as that is not the subject of this Nobel Lecture, but the
rich observations gave strong support to the
kilonova
picture
of a binary neutron star merger, as well as that neutron star
binary mergers are a significant source of the heavy elements
in nature. Even more exciting is the fact that the long
anticipated idea of doing multimessenger astronomy
using
the complementary information from electromagnetic, neu-
trinos and gravitational waves to study the same phenomena
has become a reality.
FIG. 13.
The first observation of binary neutron star merger.
Time-frequency representations of gravitational wave event
GW170817, observed by the LIGO-Hanford (top), LIGO-Living-
ston (middle), and Virgo (bottom) detectors.
FIG. 14.
The main parameters of GW170817.
Barry C. Barish: Nobel Lecture: LIGO and gravitational waves II
Rev. Mod. Phys., Vol. 90, No. 4, October
December 2018
040502-11