Gravitational Waves, Volume 2:

Astrophysics and Cosmology Maggiore

Visit to download the full and correct content document:
https://ebookmass.com/product/gravitational-waves-volume-2-astrophysics-and-cosm
ology-maggiore/
More products digital (pdf, epub, mobi) instant
download maybe you interests ...

Relativity Made Relatively Easy. Volume 2: General

Relativity and Cosmology Andrew M. Steane

https://ebookmass.com/product/relativity-made-relatively-easy-
volume-2-general-relativity-and-cosmology-andrew-m-steane/

Dipped In Sunshine (Surfing The Waves Book 2) Fearne

Hill

https://ebookmass.com/product/dipped-in-sunshine-surfing-the-
waves-book-2-fearne-hill/

Human-Animal Relationships in San and Hunter-Gatherer

Cosmology, Volume I: Therianthropes and Transformation
1st ed. 2020 Edition Mathias Guenther

https://ebookmass.com/product/human-animal-relationships-in-san-
and-hunter-gatherer-cosmology-volume-i-therianthropes-and-
transformation-1st-ed-2020-edition-mathias-guenther/

Over and Under the Waves Kate Messner

https://ebookmass.com/product/over-and-under-the-waves-kate-
messner/
Human-Animal Relationships in San and Hunter-Gatherer
Cosmology, Volume II: Imagining and Experiencing
Ontological Mutability 1st ed. 2020 Edition Mathias
Guenther
https://ebookmass.com/product/human-animal-relationships-in-san-
and-hunter-gatherer-cosmology-volume-ii-imagining-and-
experiencing-ontological-mutability-1st-ed-2020-edition-mathias-
guenther/

A Modern Introduction to Classical Electrodynamics 1st

Edition Michele Maggiore

https://ebookmass.com/product/a-modern-introduction-to-classical-
electrodynamics-1st-edition-michele-maggiore/

Piper Makes Waves Elizabeth Doyle Carey

https://ebookmass.com/product/piper-makes-waves-elizabeth-doyle-
carey/

A Mysterious Universe: Quantum Mechanics, Relativity,

and Cosmology for Everyone Suhail Zubairy

https://ebookmass.com/product/a-mysterious-universe-quantum-
mechanics-relativity-and-cosmology-for-everyone-suhail-zubairy/

CORe JaVa Volume 1 Fundamental volume 2 twelfth edition

Cay

https://ebookmass.com/product/core-java-volume-1-fundamental-
volume-2-twelfth-edition-cay/
GRAVITATIONAL WAVES
Gravitational Waves

Volume 2
Astrophysics and Cosmology

Michele Maggiore
Département de Physique Théorique
Université de Genève

3
3
Great Clarendon Street, Oxford, OX2 6DP,
United Kingdom
Oxford University Press is a department of the University of Oxford.
It furthers the University’s objective of excellence in research, scholarship,
and education by publishing worldwide. Oxford is a registered trade mark of
Oxford University Press in the UK and in certain other countries
© Michele Maggiore 2018
The moral rights of the author have been asserted
First Edition published in 2018
Impression: 1
All rights reserved. No part of this publication may be reproduced, stored in
a retrieval system, or transmitted, in any form or by any means, without the
prior permission in writing of Oxford University Press, or as expressly permitted
by law, by licence or under terms agreed with the appropriate reprographics
rights organization. Enquiries concerning reproduction outside the scope of the
above should be sent to the Rights Department, Oxford University Press, at the
address above
You must not circulate this work in any other form
and you must impose this same condition on any acquirer
Published in the United States of America by Oxford University Press
198 Madison Avenue, New York, NY 10016, United States of America
British Library Cataloguing in Publication Data
Data available
Library of Congress Control Number: 2017943894
Set ISBN 978–0–19–875528–9
Volume 1 978–0–19–857074–5
Volume 2 978–0–19–857089–9
only available as part of a set
DOI 10.1093/oso/9780198570899.001
Printed and bound by
CPI Litho (UK) Ltd, Croydon, CR0 4YY
Links to third party websites are provided by Oxford in good faith and
for information only. Oxford disclaims any responsibility for the materials
contained in any third party website referenced in this work.
Preface to Volume 2

In the preface to Vol. 1 we wrote: “The physics of gravitational waves

is in a very special period. [...] As a result of these experimental ef-
forts, there are good reasons to hope that the next decade will witness
the direct detection of gravitational waves and the opening of the field of
gravitational-wave astronomy and, possibly, cosmology.” The writing of
Vol. 2 took 10 more years, and during that time these hopes have indeed
been fulfilled. The first direct detection of gravitational waves (GWs)
took place in September 2015, when the two detectors of the LIGO
Observatory, at the very beginning of the first science run of advanced
LIGO, observed the signal from a coalescing black-hole binary. The re-
sult was announced by the LIGO and Virgo collaborations in February
2016. For this discovery Reiner Weiss, Barry Barish and Kip Thorne
were awarded the 2017 Nobel Prize for Physics “for decisive contribu-
tions to the LIGO detector and the observation of gravitational waves”.
Several more binary black-hole coalescences have now been observed, in-
cluding a triple coincidence between the two LIGO interferometers and
the Virgo interferometer. Furthermore, in August 2017, the GWs from
the coalescence of a neutron star binary were observed by LIGO and
Virgo, in coincidence with a γ-ray burst detected by Fermi-GBM and
by INTEGRAL. The electromagnetic counterpart was then identified
and observed by several telescopes in all bands of the electromagnetic
spectrum. Thus, the GW window has been opened, and we are starting
to look through it. With the planned improvements in the sensitivi-
ties of ground-based interferometers, we expect that GWs from astro-
physical sources will soon be observed routinely while, in the near and
mid-term future, we expect that several other detectors, such as pul-
sar timing arrays, the planned LISA space interferometer, and possibly
third-generation ground-based detectors such as the Einstein Telescope,
will explore the Universe up to cosmological distances. Thus, we are
indeed in a very exciting period for GW astrophysics and cosmology.

The first volume of this book dealt with the theory of GWs, in Part I,
and with experiments, in Part II. This second volume is devoted to
what we can learn from GWs, in astrophysics (Part III) and in cosmol-
ogy (Part IV). Our main aim has been to systematize a large body of
theoretical and observational developments that have taken place in GW
physics over the last few decades. Even though we expect the field to
evolve rapidly, thanks to new experimental discoveries as well as to the-
oretical advances, we believe that most of these methods will still form
the backbone of our understanding for many years to come, and will be
xiv

the basis for future developments. As with Vol. 1, we have typically tried
to rederive afresh all theoretical results, trying to give a coherent and
consistent picture of the field, and clarifying or streamlining the existing
derivations whenever possible. An Errrata web page will be maintained
at http://theory.physics.unige.ch/˜maggiore/home.html

Various people have read and commented on the drafts of some chap-
ters. I am particularly grateful to Alessandra Buonanno, Emanuele
Berti, Thibault Damour, Ruth Durrer, Valeria Ferrari, Stefano Foffa,
Kostas Kokkotas, Michael Kramer, Martin Kunz, Georges Meynet, An-
drea Passamonti, Eric Poisson, Luciano Rezzolla, Alberto Sesana and
Riccardo Sturani, for their careful reading of selected chapters, and to
Camille Bonvin, Daniela Doneva, Alex Kehagias and Vittorio Tansella
for useful comments.
I also wish to thank Sonke Adlung, Harriet Konishi and the staff of
Oxford University Press for their competent and friendly help, and Mac
Clarke for the careful and competent copyediting.

Geneva, December 2017

Stellar collapse
10
We begin our study of astrophysical sources of gravitational waves (GWs)
10.1 Historical Supernovae 4
with a chapter devoted to stellar collapse. From the point of view of GW
physics, this is interesting for at least two different reasons. First, stel- 10.2 Properties of Supernovae 10
lar collapse through supernova explosion is itself a potentially interesting 10.3 The dynamics of core
collapse 21
source of GW production. Second, stellar collapse can leave as remnant
10.4 GW production by
a compact object, such as a white dwarf (WD), a neutron star (NS) or self-gravitating fluids 35
a stellar-mass black hole (BH). As we will see in detail in later chapters, 10.5 GWs from stellar collapse 46
these compact objects are among the most interesting sources from the
10.6 Complements 66
point of view of GW astrophysics.
Supernovae (SNe) are among the most spectacular events in the Uni-
verse. As we will discuss in detail in this chapter, there are two very
different mechanisms leading to SN explosions: (1) The gravitational
collapse of the core of a star, once the nuclear fuel that feeds the ther-
monuclear reactions inside the core is exhausted. As we will see in Sec-
tion 10.2.1, depending on the properties of the progenitor, this leads to
events classified as type Ib, Ic or type II SNe, and leaves behind a com-
pact remnant, usually a neutron star (which is sometimes observable as a
pulsar) or possibly a black hole (BH).1 (2) The thermonuclear explosion 1
There are also other mechanisms that
of a white dwarf that accretes mass from a companion, going beyond its can trigger the explosion, as in electron-
capture SNe, as well as a more refined
Chandrasekhar limit. This gives rise to type Ia SNe. In this case the classification of SN type, as we will dis-
star that explodes is dispersed in space and its remnant is not a compact cuss in more detail later in the chapter.
object. A typical core-collapse SN can release an energy ∼ 1053 erg. Of It is also in principle possible that the
this energy, 99% is emitted in neutrinos, about 1% goes into kinetic en- collapse produces no visible SN event
and all the matter is swallowed by the
ergy of the ejected material, and less than 0.01%, i.e. about 1049 erg, is final BH.
released in photons. The corresponding peak luminosity in photons can
be of order a few times 109 L or higher. Thus, a typical core-collapse
SN at its peak has an optical luminosity that rivals the cumulative light
emitted by all the stars in its host galaxy. As we will see in detail in this
chapter, type Ia SNe have similar electromagnetic luminosities. Such
events, when they take place in our Galaxy (barring obscuration from
dust in the Galactic plane beyond a few kpc), can lead to truly impres-
sive optical displays that have been observed by mankind since ancient
times. A number of milestones in astronomy are associated with nearby
SN explosions. Tycho’s SN in 1572 and Kepler’s in 1604 marked a new
epoch in astronomy, while recent discoveries such as the pulsar in the
remnant of the 1054 Crab SN and the detection of neutrinos from the
1987A SN in the nearby Large Magellanic Cloud rank among the mile-
stones of modern astronomy. We therefore find it appropriate to begin
this chapter with an introduction to historical SNe.

Gravitational Waves, Volume 2: Astrophysics and Cosmology. Michele Maggiore.

c Michele Maggiore 2018. Published in 2018 by Oxford University Press.
DOI 10.1093/oso/9780198570899.001.0001
4 Stellar collapse

Table 10.1 Summary of the historical SNe, and the source of their records, adapted
from Green and Stephenson (2003). Remnants updated from Ferrand and Safi-Harb
(2012). (The identifier of the remnants, as for example in G120.1+01.4, refers to its
Galactic coordinates). We have classified SN 1604 as type Ia, following for example
Blair et al. (2007), Reynolds et al. (2007) and references therein. The nature of the
Crab SN (core collapse or electron capture) is debated; see the Further Reading.

Date Length of Remnant Classification Historical records

visibility Chinese Japanese Korean Arabic European

AD1604 (Kepler’s) 12 months G004.5+06.8 type Ia few – many – many

AD1572 (Tycho’s) 18 months G120.1+01.4 type Ia few – two – many
AD1181 6 months G130.7+03.1 ? – few few – – –
AD1054 (Crab) 21 months G184.6−05.8 debated many few – one –
AD1006 3 years G327.6+14.6 type Ia many many – few two
AD393 8 months G347.3−00.5 – one – – – –
AD386? 3 months G011.2−00.3 – one – – – –
AD369? 5 months – – one – – – –
AD185 8 or 20 months G315.4−02.3 type Ia ? one – – – –

10.1 Historical Supernovae

The reader eager to plunge into more
technical issues can simply skip this
section
2
Novae arise when a white dwarf (WD)
in a binary system accretes mass from
a companion at a rate of about 10−9 −
10−8 M /yr. The material accreted
is rich in hydrogen and accumulates
on the surface of the WD, where it
is compressed by further accretion and
heated. When a layer of about 10−5 −
10−4 M of hydrogen has been ac-
creted, a runaway thermonuclear reac-
tion takes place, releasing an energy of
order 1045 erg, which is therefore sev-
eral orders of magnitude smaller than in
SN explosions. The WD then gradually
cools down and goes back to its quies-
cent state, where the accretion process
can start again, leading to recurrent
nova explosions. A clear distinction be-
tween classical novae and supernovae
was first made by Baade and Zwicky
in 1934.
As we will see in Section 10.2.1, in a galaxy like ours the estimated rate
of SN explosions is of order of two per century. However, in the Galactic
plane the extinction of visible light significantly limits our view; for
instance, the extinction of visible light from the Galactic center is about
30 magnitudes. Thus, a Galactic SN visible to the naked eye is a very
rare event. However, when a nearby star becomes a SN, the effects are
quite spectacular. In the historical period, this has happened a handful
of times.
Evidence for the oldest SNe observed by mankind is based on his-
torical records. Since a Galactic SN remains visible for months, and
sometimes up to a few years, one can focus on historical records of stars
that suddenly appeared and remained visible for at least four months,
in order to eliminate most novae2 and the possibility of confusion with
comets (comets being eliminated also because of their evident proper
motion).
Table 10.1 gives a summary of SNe that have been seen by the naked
eye in historical times, and for which we have written records, in the
last 2000 years.
SN 185
The oldest recorded SN is SN 185, which was first observed on Decem-
ber 7, 185 at the imperial observatory of Lo-Yang, in central China.
The Chinese astronomers recorded its appearance and its gradual fad-
ing, providing the oldest plausible historical account of a SN. It finally
disappeared after either 8 or 20 months (depending on the interpre-
tation of a sentence in the record to mean “next year” or “year after
next”). The identification of the recorded position of the event with
 10.1 Historical Supernovae 5

the region between α and β Cen suggests that the remnant of SN 185
could be identified with the SN remnant RCW 86 (G315.4−02.3), at an
estimated distance of about 2.8 kpc. Recent observations with Chandra
and XMM-Newton have indeed strengthened the case for this identifica-
tion. The study of its X-ray synchrotron emission has in fact suggested
an estimate for the age of this remnant consistent with the explosion
date, within the uncertainties of the determination. The regular shape
of the remnant shell, together with the absence of a pulsar, would point
toward a type Ia SN, which is also suggested by the Chandra X-ray
observations.
Another possible remnant candidate has been proposed, G320.4−01.2
(RCW 89), which contains the pulsar PSR B1509-58, and therefore
would corresponds to a core-collapse SN. The timing measurement of
this pulsar are also consistent with the age of SN 185.

SN 1006

A number of other possible SNe were recorded before AD 1000; see

Table 10.1. However, the first event for which we have extensive histor-
ical records is a SN that exploded in AD 1006 in the direction of the
Lupis constellation, and was recorded in China, Japan, Europe, Egypt
and Iraq. It disappeared a first time from view after 17 months, and
remained occasionally visible at dawn for 3 years.
The identification of the likely remnant of this SN came in 1965, by
searching radio catalogues covering part of the region of sky suggested
by the historical records. The remnant is now identified with the radio
source PKS 1450-51 (whose Galactic coordinates are G327.4+14.6). Its
distance from us, as we now infer from the remnant, was about 2.2 kpc.
From the spectrum of the remnant it is believed that it was a type Ia
SN. Using the standard luminosity of type Ia SNe together with the
known distance to the remnant (and plausible assumptions about ab-
sorption) allows us to estimate that it reached an apparent visual mag-
nitude V ' −7.5.3 By comparison, the full Moon has a maximum appar- 3
The standard astronomical notions of
ent magnitude V ' −12.94, and the limiting brightness for an object to luminosity, magnitudes and color index
of stars are recalled in the Complement
be visible with the naked eye when the Sun is high is V ' −4, dropping Section 10.6.
to V ' −2.5 when the Sun is less than 10 degrees above the horizon.
Comparing with the visual magnitudes given in Table 10.5 on page 70,
we see that, at its maximum brightness, SN 1006 was by far the brightest
object in the night sky after the Moon, and easily visible during day-
time. The historical records confirm that it was indeed visible during
the day, and it was even bright enough to read at night by its light! The
most detailed reports by far are from Chinese astronomers, who deter-
mined the position of this “guest star” to good accuracy. Two accounts
of this SN appeared also in Europe, in chronicles of the monasteries of
St. Gallen, in Switzerland, and of Benevento, in Italy.
6 Stellar collapse
4
The subject is quite controversial
among historians of science. Some Eu-
ropean documents have been suggested
as referring to SN 1054, but look quite
imprecise, and overall the evidence that
they actually refer to the SN is not very
convincing. It has been suggested that
the lack of written documents was due
to censorship from the Church (given
that the chronicles of the epoch were
compiled in monasteries), possibly con-
nected with the fact that the SN ap-
peared just at the time of the East-
West Schism that gave rise to the sepa-
ration of the Church of the Roman Em-
pire into what later became the Roman
Catholic Church and the Eastern Or-
thodox Church. In fact, the date of
the excommunication of the Patriarch
of Constantinople Michael I Cerularius
is July 16, 1054. The usually accepted
date for the appearance of the SN is
July 4, so by July 16 it was just near
its maximum luminosity.
There have also been suggestions that
this SN is depicted, together with the
Moon, in cave paintings from native
americans in Arizona. It has been pro-
posed that the paintings represents a
conjunction between the Moon and the
SN, made possible by the fact that soon
after it appeared, on July 5, the SN
as seen from Earth was indeed on the
plane of the ecliptic and in a direction
close to that of the Moon. However,
the dating of the paintings is quite im-
precise (between the 10th and 12th cen-
turies), and only one of them shows the
crescent moon with the correct orienta-
tion in relation to the SN, so it is diffi-
cult to make definite statements.
SN 1054 (the Crab supernova)
Shortly after, in 1054, the Crab SN exploded and was recorded by
Arab, Chinese and Japanese astronomers. It gave rise to the Crab neb-
ula, whose remnant neutron star we observe today as the Crab pulsar,
PSR B0531+21. Its distance from us is 2.0 ± 0.5 kpc.
The identification of the remnant of the 1054 SN with the Crab nebula,
based on the expansion speed of the cloud measured spectroscopically, as
well as on various concordant historical records, is quite firm. The first
suggestion that the Crab nebula was the remnant of SN 1054 was due
to Hubble. In 1928, when the nature of nebulae was still unknown, he
observed that the Crab nebula was expanding radially and deduced that
it was most probably the remnant of an explosion. From the expansion
rate he could infer that the explosion took place nine centuries earlier,
consistent with SN 1054. This analysis was refined and strengthened by
Mayall and Oort in 1939 and 1942. From the analysis of the historical
records, it is believed that it reached an apparent magnitude between
−7 and −4.5 and again it was visible during daylight. It remained
visible to the naked eye for about a year and a half. It is puzzling that,
for such an impressive phenomenon, there is no unambiguous European
record, particularly considering that, half a century earlier, SN 1006 was
recorded in Europe, so European chroniclers of that epoch did not lack
interest in such phenomena.4
Today, after almost 1000 years, the Crab nebula still has a luminosity
of 8 × 104 L , mostly in the form of polarized synchrotron radiation,
indicating the presence of highly relativistic electrons spiraling around
magnetic field lines. The source of this energy, so long after the explo-
sion, remained a puzzle until the discovery of the pulsar. As we will
discuss in Section 11.1.1, ultimately, this huge luminosity comes from
the pulsar’s spindown. The Crab nebula was first detected in radio
waves in 1963 and in X-rays in 1964. The discovery of the pulsar inside
the Crab nebula, in 1968, was a milestone in astronomy and made the
Crab the most studied and famous SN remnant, together with the recent
SN 1987A.
SN 1181
The next visible Galactic SN appeared in AD 1181, and was recorded in
China and Japan. It remained visible for six months. The SN remnant
3C58, with Galactic coordinates G130.7+03.1, at a distance of about
3.2 kpc, has been proposed as the remnant of this supernova. Within
this nebula has been discovered the radio and X-ray pulsar J0205+6449.
Recent work, however, has questioned this identification of the rem-
nant. If the identification were correct, its young age of less than 900
years would suggest a rapid expansion of the nebula, while the observed
expansion seems considerably lower, and rather suggests that the SN
remnant 3C58 could be at least 2700 years old.
 10.1 Historical Supernovae 7

SN 1572 (Tycho’s supernova)

The next SN visible to the naked eye appeared after about four centuries,
in AD 1572, in the direction of the constellation of Cassiopea, and was
observed in Asia and Europe. The SN was probably first detected in
Europe on November 6 (or possibly a few days earlier) by the abbot of
Messina and by other observers in Europe. Very detailed observations
of it were made by the Danish astronomer Tycho Brahe, who realized
that it was a new star and begun observing it on November 11.
He carefully followed it over the following year, establishing that its
position was fixed, and determining it to within a few arcminutes. He
was also the only astronomer to monitor carefully the decline of its
brightness in the following months. This SN is thus often referred to
as Tycho’s SN.5 It remained visible for 18 months. The fact that its 5
As Tycho reports, “On the 11th day
position remained fixed, without a daily parallax with respect to the of November in the evening after sun-
set, I was contemplating the stars in a
fixed stars, allowed Brahe to infer that the object was far beyond the clear sky. I noticed that a new and un-
Moon; furthermore, the absence of apparent proper motion over a period usual star, surpassing the other stars in
of months showed him that it was much further away than the planets, brilliancy, was shining almost directly
and was indeed a star. This was a highly non-trivial conclusion for above my head; and since I had, from
boyhood, known all the stars of the
the epoch, which contradicted the accepted Aristotelian doctrine that heavens perfectly, it was quite evident
changes in the sky could only take place in the sublunar region, and to me that there had never been any
ultimately led to the notion of the immutability of the heavens being star in that place of the sky, even the
abandoned. It also opened the way to modern astronomy, revealing the smallest, to say nothing of a star so con-
spicuous and bright as this. I was so
importance of performing accurate astrometric measurements. astonished of this sight that I was not
Chinese observers recorded that it was visible in daylight, while Ko- ashamed to doubt the trustworthiness
rean documents compared its brightness with that of Venus. This sug- of my own eyes. But when I observed
that others, on having the place pointed
gests that at its maximum the SN was brighter than the threshold
out to them, could see that there was
V ' −4 for visibility during daylight, but not by much. The remnant is really a star there, I had no further
identified with the radio and X-ray source 3C10 (G120.1+01.4).6 Today doubts. A miracle indeed, one that has
the remnant is a shell of angular diameter 8 arcmin (by comparison, the never been previously seen before our
time, in any age since the beginning of
full Moon has an angular diameter between approximately 29 and 34
the world.” (From Burnham, 1978).
arcmin, depending on its exact distance from Earth along its orbit).
Historical records of the light curve suggest a type Ia SN. No compact
remnant has been detected at its center, which also points toward a 6
Beautiful images of this and
type Ia SN. Finally, the classification as type Ia has been beautifully other SN remnants in X-rays
confirmed thanks to the detection of a light echo from the SN, due to have been taken by Chandra, see
scattering and absorption/re-emission of the SN flash by interstellar dust http://chandra.harvard.edu/photo/2011
/tycho/
near the remnant. In this way it has been possible to observe some of
the light emitted by the SN near maximum brightness, which has arrived
more than four centuries after the direct flash, and it has been possible
to perform a spectroscopic analysis. The spectrum has confirmed that
SN 1572 was a typical type Ia SN. Its distance from us has usually
been quoted in the range 2.3 − 2.8 kpc, although the absolute brightness
inferred from the light echo, together with the historical information
that it had V ' −4, rather gives a distance 3.8+1.5
−1.1 kpc.
A search for the companion star from which the white dwarf accreted
mass, going beyond the Chandrasekhar limit, has revealed a G2 star
in the direction of the center of the nebula, and at a distance between
8 Stellar collapse
Fig. 10.1 The light curve of the SN
of AD 1604 from European (◦) and
Korean (•) observations, with a Eu-
ropean upper limit on October 8.
From Stephenson and Green (2002).
2.5 and 4.0 kpc, so compatible with the distance to the nebula. This
star has a peculiar phase-space velocity, more than three times as large
as that of any other star in this region, which could be interpreted as
arising from the kick received by the companion in the explosion. Other
interpretations, in terms of a star unrelated to the SN and belonging to
the thick disk population, are, however, possible.
SN 1604 (Kepler’s supernova)
The next SN visible to the naked eye appeared in 1604, and was first ob-
served on October 9 in Italy, by astronomers in Verona and in Cosenza.
It remained visible until October 1605, and was extensively followed
by European, Chinese and Korean astronomers. Johannes Kepler was
among those who observed it. Due to poor weather conditions in Prague,
he could start observing it only on October 17. However, his treatise De
Stella Nova in Pede Serpentarii (“On the new star in the foot of Ophi-
uchus”, Serpentarius being a less familiar name for the constellation
Ophiuchus) was the most important European description, in particu-
lar for the accurate positional measurements, so this SN is also called
Kepler’s SN. The most accurate positional measurements are in gen-
eral from European astronomers, with an accuracy of about 1 arcmin
(compared with a precision of about 1 degree from Chinese and Korean
astronomers). Its luminosity curve can be inferred both from European
observations and from Korean documents recording its luminosity on a
day-by-day basis for the first six months, and matching the European
observations. The SN was detected about 20 days before maximum lu-
minosity, thanks also to the fortuitous circumstance that at that time
Mars and Jupiter happened to be in conjunction. This conjunction was
being carefully observed, and the SN appeared at just 3 degrees NE from
the position of these two planets (and 6 degrees East of Saturn). We
therefore have a detailed record of the luminosity curve from appearance,
thorough maximum luminosity and fading; see Fig. 10.1. According to a
reconstruction of the luminosity curve due to Baade, it reached a max-
imum visual magnitude V ' −2.25, so it was not as spectacular as
SN 1006, with its inferred V ' −7.5, the Crab (maximum V between
−7 and −4.5) or Tycho’s SN (maximum V ' −4). Still, in the night
sky it was brighter than any star, and among the planets only Venus
was significantly brighter than the SN. When the SN appeared it was as
bright as Mars, and then surpassed Jupiter in a few days. The fact that
this SN was relatively high with respect to the Galactic plane reduced
significantly the obscuration by dust, allowing the SN to be visible by
the naked eye.
The remnant (whose Galactic coordinates are G004.5+06.8) was iden-
tified by Baade in 1943, and has been studied at infrared, optical, radio
and X-ray wavelengths. Its distance from us is quite uncertain, with
estimates ranging from about 3 kpc to more than 6 kpc. It is located in
the direction of the Galactic center, so, taking 4 kpc as an estimate of its
distance, it is roughly half-way from the Galactic center. It is relatively
 10.1 Historical Supernovae 9

high above the Galactic plane, at a vertical height z ' 590 (d/5 kpc) pc,
where d is its distance from us.
The classification of this SN has been controversial for decades. Baade
suggested that the historical light curve was consistent with a type Ia,
but his claim was later questioned. The presence of dense circum-
stellar material implies significant mass loss from the precursor star,
which is more suggestive of a core-collapse SN (although there are a few
examples of circumstellar material around spectroscopically confirmed
type Ia SNe). Since the 1990s, however, evidence accumulated in favor
of a type Ia SN. Analyses of X-ray spectra with Exosat, ASCA, XMM-
Newton and, more recently, Chandra indicate an overabundance of Fe
and Si in the ejecta and relatively little oxygen emission. Such a com-
position is inconsistent with a core-collapse SN, and is instead typical of
a type Ia SN. Furthermore, no neutron star has been detected near its
center, either in radio or in X-rays. In X-rays, the limit on the flux from
a hypothetical central compact object shows that, if it were present,
its intrinsic X-ray luminosity would be more than 100 times smaller
than that of the compact remnant of a typical core-collapse SN such as
Cas A. Infrared studies with Spitzer point again toward the conclusion
that this was a type Ia SN, although with significant circumstellar ma-
terial. Recent hydrodynamical simulations show that the morphology
of the remnant and its expansion characteristics can be reproduced by
a model of a type Ia SN originating from a white dwarf that accreted
mass by wind accretion from an asymptotic giant branch star with mass
(4 − 5)M .

The population of Galactic SN remnants

Another interesting Galactic SN remnant is Cas A, which is observed as
a bright source in radio and X-rays. At visible wavelengths this remnant
displays a ring of expanding knots. From their expansion velocity, and
assuming no deceleration, one can estimate that this SN exploded around
AD 1671. Its distance from us is estimated at 3.4+0.3 −0.1 kpc. It is quite
puzzling that for such a recent and relatively nearby SN there is no
historical record.
Galactic SNe of course occur at a much higher rate than SNe visible
to the naked eye, given that, for most lines of sight close to the Galactic
plane, obscuration by dust limits our view in visible light to a few kpc.
As of 2017, the catalog of Galactic SN remnants lists 381 objects.7 The 7
See http://www.physics.umanitoba.
remnant age is often uncertain or undetermined, but O(20 − 30) are ca/snr/SNRcat, which is regularly
updated.
estimated to be less than 2000 yr old. As we will see in Section 10.2.1,
typical estimates for the rate of SN explosions in our galaxy are of order
2 SNe per century. According to this estimate, there should be about
40 remnants younger than 2000 yr (raising to about 60 if we take 3
SNe per century as an estimate of the rate). This is not inconsistent
with the number of known young remnants, although it suggests that
more young remnants have still to be discovered. Comparing with the
10 Stellar collapse
8
Needless to say, by this we mean that
the signal from the SN reached Earth
around 1900. The actual explosion took
place about 25,000 yrs ago, given the
time taken by light to reach us from the
Galactic center. The same applies to all
the other SN explosion dates mentioned
above.
9
Or, the other way around, Shapley
in 1919 used this SN as an argument
against the island universe hypothesis
(i.e. the existence of other galaxies,
separated from ours by huge distances),
arguing that SN 1885A in Andromeda
would have M = −16, which was “out
of the question”. The typical abso-
lute magnitude of most novae is around
−8.8.
10
Actually, even before the discovery of
the neutron, in 1931, neutron stars were
somewhat anticipated by Lev Landau,
who wrote about stars where “atomic
nuclei come in close contact, forming
one gigantic nucleus”.
11
“With all reserve we advance the
view that supernovae represent the
transition from ordinary stars into neu-
tron stars, which in their final stages
consist of closely packed neutrons”,
W. Baade and F. Zwicky, Meeting of
the American Physical Society, Decem-
ber 1933, published in Phys. Rev. 45
(1934) 138.
12
See http://graspa.oapd.inaf.it/
asnc.html for an on-line version.
7 SNe in Table 10.1 for which convincing historical records exist, we can
estimate that about 10 − 20% of the galactic SNe exploding in the last
2000 years have been recorded in writing.
Among the known Galactic SN remnants, the record for the youngest
is currently held by G001.9+00.3. This remnant is located in the direc-
tion of the Galactic center, at a distance from the center of about 300 pc
in projection. From its expansion rate, assuming no deceleration, the
explosion date is estimated as the year 1855 ± 11. Since some decelera-
tion must have occurred, this is actually an upper bound on its age, and
the explosion date was most likely around 1900.8 The remnant was first
observed by VLA (Very Large Array) radio observations, and in X-rays
by Chandra.
10.2 Properties of Supernovae
Modern research on SNe began in 1885 when Hartwig discovered the
first extragalactic SN, in the Andromeda galaxy. It was first interpreted
as a nova. However, when in 1919 Lundmark estimated the distance to
Andromeda, it became clear that Hartwig’s “nova” had actually been
O(103 ) times brighter than a classical nova.9 However, it was only with
the pioneering paper by Baade and Zwicky in 1933 that the notion of SNe
was put forward and a first clear distinction was made between novae
and supernovae. In 1933, just one year after Chadwick’s discovery of the
neutron, Baade and Zwicky put forward the idea of a neutron star10 and
further speculated that SNe are an end-state of stellar evolution, with
the source of energy being provided by the gravitational energy released
by the collapse of the progenitor star to a neutron star.11
In the following years systematic searches, performed particularly by
Zwicky at Palomar Mountain, led to the discovery of 54 SNe up to 1956,
and 82 more from 1956 to 1963. In recent years, largely stimulated by
the importance of type Ia SNe for determining the expansion rate of
the Universe, intensive searches have dramatically increased the number
of detected SNe. Browsing the Asiago Supernova Catalog12 one finds
that in the period 1980–1989 were observed O(200) SNe, in the period
1990–1999 were found over 900 SNe, and in the period 2000–2009 were
observed O(3500) SNe, obviously all of them extragalactic. As of 2017
the total number of SNe in the catalog is about 9000, and is increasing at
a rate of about two SNe per day. This large sample, combined with the
improved quality of the observations, has allowed remarkable progress
in our understanding of the properties of SNe. Future large synoptic
surveys are expected to detect possibly O(105 ) SNe per year, at redshifts
z ∼ 0.5 − 1.
Much progress has also been made in recent years in identifying the
progenitor stars of local SN explosions, by comparisons with archival
data from HST or ground-based telescopes. Currently O(20) progenitors
have been identified. To draw statistical conclusions from this sample,
most of the information comes from progenitors at a distance below
 10.2 Properties of Supernovae 11

28 Mpc, although objects outside this distance, such as the massive

progenitor of SN2005gl at 60 Mpc, have also been identified.
Nowadays we understand that there are two basically different mech-
anisms behind SN explosions: either the core collapse of a star, powered
by the energy released in the gravitational collapse (just as proposed by
Baade and Zwicky) or the thermonuclear explosion of a white dwarf in
a binary system that accretes mass from a companion, going beyond its
Chandrasekhar limit.13 However, as we now discuss, a classification of 13
As we will later discuss, another pos-
SNe that is more direct from an observational point of view is based sibility that has been proposed in the
literature is a “double-degenerate” sce-
rather on the spectrum of the light emitted at maximum luminosity. nario, in which the thermonuclear ex-
plosion results from the merger of two
white dwarfs.
10.2.1 SN classification
The first distinction, proposed by R. Minkowski in 1941, was between
type I and type II SNe. Type I are defined by the fact that their spec-
trum near maximum luminosity does not show evident H lines, while
type II near their peak luminosity have prominent hydrogen lines. Sub-
sequently, in the late 1980s it was realized that type I SNe did not
constitute a homogeneous class, and that it was necessary to distin-
guish between two fundamentally different classes, which were denoted
as type Ia and type Ib. Type Ia SNe are defined as type I SNe that,
in their spectrum taken near maximum luminosity, show a deep absorp-
tion trough, typically around 6000–6150 Å, produced by blue-shifting of
two Si II lines with rest-frame wavelengths λ = 6347 and 6371 Å (often
collectively denoted as λ6355). The significance of these Si lines, as well
as lines of other intermediate-mass elements such as Ca, Mg, S and O
that are present in the spectrum of type Ia SNe at maximum light, is
that the part of the ejecta that is exposed shortly after the explosion
has undergone thermonuclear processing, leading to the production of
intermediate-mass elements. This thermonuclear processing has, how-
ever, been incomplete, since Si, O, etc. have not been burnt further to
eventually produce elements of the iron group.
The mean velocities of the ejecta, obtained from the blueshift of the
spectral lines, are of order 5000 km/s, with peak velocities as large as
20,000 km/s ' c/15. The typical spectrum of a type Ia SN is shown
in the upper curve of the left panel in Fig. 10.2. It shows strong Si II
absorption at about 6150 Å, while H lines are absent. At late times, say
after more than 4 months, the typical spectrum of a type Ia SN changes,
and rather than being dominated by lines of intermediate-mass elements,
it is now dominated by a blend of dozens of Fe emission lines, mixed with
some Co lines. This is a consequence of the fact that, with time, the
initially very dense ejecta become more and more transparent because
of their expansion or, in other words, the photosphere (i.e. the surface
from where the photons can start to free stream toward us) recedes,
exposing the products of thermonuclear burning in the inner core.
Type Ib SNe, in contrast, are defined as type I SNe whose spectrum
near maximum brightness does not show the Si II absorption line that
defines type Ia SNe. In fact, it is useful to distinguish further between
12 Stellar collapse

max. +10 months

Ia

Ib

II

Fig. 10.2 Left panel: Spectra of basic SN types near maximum luminosity. Type II
are defined by a clear signature of Hα. Type Ia is defined by a lack of H lines and
a strong Si II absorption at about 6150 Å. Type Ib/c lack both H and Si lines, but
type Ib has He lines, which are not present in type Ic. Ten months later (right panel)
SN Ia show strong emissions of [FeII] and [FeIII], SN Ib/c are dominated by [CaII]
and [OI]. These same lines plus strong Hα emission are typical of SN II at late times.
From Cappellaro and Turatto (2001).

type Ib and type Ic SNe. Neither type shows H or Si lines near maximum
H lines ?
luminosity. However, near maximum luminosity, type Ib SNe show He
no yes lines, while type Ic SNe have neither H nor He lines. At late time, the
spectra of type Ib and type Ic become very similar, indicating that the
Si II lines ? plateau ? distinction among Ib and Ic (sometimes collectively denoted as Ibc) is
less fundamental than the distinction between type Ia and type Ibc.
yes no no yes
Type II SNe are defined by the fact that they show prominent H lines
He lines ? in the spectrum taken near maximum luminosity. They are further di-
no yes vided into two classes, depending not on further spectral features, but
II b rather on the decline with time of their light curves. Those that, after
Ia Ic Ib II L II P reaching maximum luminosity, show a plateau of almost constant lumi-
thermonuclear
small
core collapse
large nosity for about 2–3 months are called II-P, while those that show a
explosion envelope
decline approximately linear in magnitude (and exponential in luminos-
ity, recalling that the magnitude is obtained from the logarithm of the
Fig. 10.3 A decision tree for the luminosity; see the Complement Section 10.6) are called II-L. A compar-
classification of SNe, based on spec- ison between the spectra of type Ia, Ib and II SNe, both at maximum
tral features near maximum light luminosity (left panel) and after 10 months (right panel), is shown in
and, for type II SNe, on the tem-
Fig. 10.2. A decision tree based on these criteria is shown in Fig. 10.3.
poral evolution of the light curve.
The separation into type I and type II, however, does not really reflect
the underlying explosion mechanisms. Rather, type Ia SN are believed
 10.2 Properties of Supernovae 13

to originate from the thermonuclear explosion of a carbon–oxygen white

dwarf (WD) in a binary system. In the simplest and commonly accepted
scenario, called single-degenerate, a white dwarf (WD) accretes matter
from a companion such as a red giant (or even a main sequence star for
a sufficiently close binary system) until its mass goes beyond the Chan-
drasekhar limit MCh ' 1.39M . Then the star undergoes a disruptive
thermonuclear explosion, and is dispersed in the external space in the
form of a gaseous nebula. No compact remnant is formed in the process.
An alternative scenario, called double-degenerate, rather involves the co-
alescence of two WDs, which again results in a thermonuclear explosion,
and no compact remnant.
In contrast, types Ib, Ic, II L and II P all correspond to core collapse
SNe, i.e. to the gravitational collapse of the inner part of the star. When
a star has exhausted the nuclear fuel in its core, it is no longer able to
balance its self-gravity by the pressure generated by the thermonuclear
reactions, and a complicated dynamics begins. As we will discuss in
detail in Section 10.3, this can lead to a SN explosion, with ejection of
the external layers of the stars, while the core collapses to a NS or a BH.
The difference between the various types of core collapse can be traced
to the mass and size of the progenitor star, and to the size and com-
position of its envelope. In the early phase of the explosion the ejecta
are very dense and opaque to electromagnetic radiation. This implies
that the light emitted in this early stage, including the spectrum near
maximum luminosity, only probes the outermost layers and is therefore
mostly sensitive to the size and composition of the envelope.
The mass of the H envelope of a star can vary, depending on many
factors. For instance, stars can lose their envelope through strong stel-
lar winds, as in the Wolf–Rayet stars.14 Furthermore, when a star is a 14
Wolf–Rayet (WR) stars originate
member of a close binary system, the interaction with the companion from luminous OB stars with strong
and broad emission lines and an at-
can have important effects on its envelope; the two stars can undergo a mosphere made mostly of He, while
phase of a common envelope evolution, or material can be transferred H is deficient or totally absent. It
from a star to the other through Roche-lobe overflow. It is generally is believed that they are very massive
accepted that type II SNe derive from the explosion of stars with mass evolved stars, whose H layer has been
blown away by strong stellar winds,
in the range 10 − 30 M . In a typical SN of type II-P the progenitor is a produced by radiation pressure. As a
red supergiant whose H envelope has a large mass, of order 5 − 10 M . result, WR stars are continually eject-
The energy of the explosion completely ionizes the H envelope. The gas ing gas at high speed, with line-of-sight
then enters a stage of prolonged recombination, which maintains an ap- velocities that can be over 2000 km/s.
The resulting Doppler broadening is re-
proximately constant temperature, typically between 5000 and 8000 K, sponsible for the observed broadness of
and the photons emitted during recombination produce the plateau of their emission lines. These winds are
type II-P SNe. The duration of this plateau depends on the mass of the also responsible for a surface composi-
H envelope. As the photosphere recedes, the dominant mechanism for tion that shows the presence of prod-
ucts of the nuclear burning in the inner
photon emission becomes the decay of various radioactive nuclei newly regions.
synthesized in the explosion, which at first can extend the plateau for
a brief time. Then, the SN enters in the phase where its luminosity is
powered uniquely by radioactive decays. Since radioactive decays follow
an exponential law N (t) = N0 e−t/τ (where N (t) is the number of nuclei
at time t, and τ is related to the half-life τ1/2 by τ1/2 = τ ln 2), the corre-
sponding luminosity L decreases exponentially, with a slope determined
14 Stellar collapse
15
Hard X-rays and γ-rays were also
directly detected from the type II
SN1987A in the Large Magellanic
Cloud. The 56 Ni decay chain also pow-
ers the light curve of type Ia SNe. For
the type Ia SN2014J, which is at a dis-
tance of just 3.3 Mpc, it has been pos-
sible to detect with the INTEGRAL
satellite the two main γ-ray lines due
to the 56 Co decay; see Churazov et al.
(2014) and Diehl et al. (2015).
by the half-life of the dominant process. Thus in this stage log L, and
hence the negative of the apparent magnitude, decreases linearly.
Type II-L supernovae are due to the core collapse of stars with an H
envelope of about 1−2 M . In this case, because of the smaller envelope
mass, the velocity of expansion of the H-rich envelope is larger, and the
SN enters directly into the phase dominated by radioactive decays, with
little or no plateau, and a linear decrease of the magnitude. Intermediate
cases exist between typical type II-P and typical type II-L supernovae,
depending on the size of the H envelope.
In the phase powered by radioactive decays an important role is played
by 56 Ni, which is the main product of burning at nuclear statistical equi-
librium in the conditions of temperature and density typically present
in SNe. In particular, 56 Ni produces 56 Co through electron capture
56
Ni + e− → 56
Co + νe + γ , (10.1)
with a half life of 6.1 days, emitting photons with energies of 750, 812 and
158 keV. The main source of luminosity later becomes the radioactive
decay of 56 Co to 56 Fe, which goes mostly through electron capture (81%)
56
Co + e− → 56
Fe + νe + γ , (10.2)
and, to a lesser extent (19%), through β-decay 56 Co → 56 Fe+e+ +νe +γ.
The process has a half-life of 77.7 days and the photons from electron
capture are emitted at energies of 1.238 MeV and 847 keV. These γ-
ray photons are down-scattered and thermalized in the ejecta until they
emerge as optical or near-infrared photons.15 The subsequent important
radioactive decay is that of a different isotope of cobalt, 57 Co, which is
initially less abundant but has a longer lifetime, τ1/2 ' 271 days.
In general, because of the very high expansion velocity of the ejecta,
the lines of type II SNe show significant Doppler broadening. However,
approximately 10–15% of type II SNe show, on top of the broad lines,
some narrow lines, and are then classified as type IIn, where “n” stands
for narrow. They are interpreted as type II SNe with significant cir-
cumstellar material, which is most likely due to episodes of strong mass
losses before the SN explosion. When the ejecta hits this circumstellar
material, some of its kinetic energy is converted to radiation, and pro-
duces lines whose broadening rather reflects the much smaller velocity
of the circumstellar material.
The close relation between type Ib and type II SNe is clearly seen by
comparing the left and right panels of Fig. 10.2. On the left panel, which
shows the spectra near maximum luminosity, the spectra of type Ib and
type II SNe look very different. However, after about 10 months, we
see that the spectra of type Ib SNe are similar to those of type II,
with strong lines from neutral oxygen and singly-ionized calcium. In
contrast, the spectrum of a type Ia SN remains very different, and after
10 months is rather dominated by iron lines. The meaning of these
patterns can be understood by observing that the light emitted near
maximum luminosity only carries informations about the external layers
 10.2 Properties of Supernovae 15

of the stars, given that the optical depth of the ejecta at that time is
high. After 10 months we are instead receiving light free-streaming from
deeper regions of the ejecta, and we find that type Ib and type II SNe
now look very similar. The natural interpretation of these facts is that
type Ib SNe are the result of the core collapse of stars that have lost their
H envelope, because of stellar winds or interaction with a companion.
This explains the lack of H lines, which instead dominate the spectrum at
maximum luminosity in type II SNe, and the similarity of the spectra at
later times. This interpretation is further supported by the existence of
cases intermediate between type Ib and type II; the missing link was first
provided by SN 1993J, whose spectrum evolved from that of a type II to
type Ib in just a few weeks. This is interpreted as the result of the core
collapse of a star whose H envelope had a rather small mass, of about
0.2 M . There are other examples of this type, so SNe that quickly
evolve from type II to type Ib are sometimes classified as members of a
new class, called type IIb. In Fig. 10.3 we represent this class with an
arrow connecting type II-L with type Ib.
Type Ic SNe also fit well in this scheme. Again their spectrum at late
time is the same as that of type Ib and type II, showing that they are
core-collapse SNe. The lack of both H and He lines can be understood on
recalling that a massive star, during its lifetime, goes through different
stages of nuclear burning, which results in a onion-like structure. In
particular, the most external layers of the star consist of a H envelope,
on top of a He envelope. Type Ic events are then naturally interpreted
as the core collapse of stars that have lost both their H and their He
envelopes.16 16
Likely type Ic have some Helium too
Type Ib, Ic and II SN are then collectively denoted as core-collapse in their outer layer, but this helium pro-
duces no line due to the lack of favor-
SNe, and can be thought as a sequence Ic–Ib–IIb–IIL–IIP, from smaller able physical conditions, see Dessart,
to larger H and He envelopes. Type Ic, Ib and IIb SNe are also collec- Hillier, Li and Woosley (2012).
tively called “stripped-envelope SNe”. An interesting subclass of type Ic
SNe, denoted type Ic-BL (for “broad-line”) consists of type Ic SNe with
unusually broad absorption lines, and photospheric velocities in excess
of 20,000 km/s.
The difference between core-collapse and type Ia SNe is further high-
lighted by the fact that core-collapse SNe have only been observed in
spiral galaxies, near sites of recent star formation. This again indicates
that their progenitors are massive short-lived stars. Type Ia SNe, in
contrast, have been observed both in spiral and in elliptical galaxies.
The latter show little or no sign of recent star formation, showing that
type Ia SNe are not related to massive stars, since these have a short
life-time. Rather, their progenitors must be low-mass stars.

10.2.2 Luminosities
Luminosity of Type Ia SNe
The thermonuclear explosion of type Ia SNe takes place when the mass
of the WD goes beyond a fixed threshold, given by the Chandrasekhar
mass. As a result, their intrinsic luminosity, to a first approximation, is
16 Stellar collapse
17
We quote the value from P. Astier et
al. [The SNLS Collaboration] (2006).
The dependence on h0 comes from the
fact that, for a given flux F , the lu-
minosity is L = 4πF d2L , where dL (z)
is the luminosity distance. For SNe
at cosmological distances dL ∝ h−1 0 ,
so L ∝ h−2 0 . Then, from M =
−(5/2) log10 L+const. (see eq. (10.155)
in the Complement Section 10.6) we
get M = M (h0,ref ) + 5 log10 h0 /h0,ref ,
where h0,ref is the reference value cho-
sen for the reduced Hubble constant h0 .
18
The exact numerical values of the co-
efficients in this relation depend on the
sample of type Ia SNe used. Here we
have used the numbers from the origi-
nal paper by Phillips (1993), obtained
however with a sample that is quite lim-
ited compared with more recent ones.
19
The discovery of the accelerated ex-
pansion using type Ia SNe has been
awarded with the 2011 Nobel Prize
in Physics to Saul Perlmutter, Brian
Schmidt and Adam Riess.
uniform. Type Ia SNe rise to maximum luminosity in a period of about
20 days, reaching a peak value of the absolute blue magnitude MB,peak
whose average over several type Ia events is given by17
hMB,peak i ' −19.31 ± 0.03 + 5 log10 (h0 /0.70) . (10.3)
Using eqs. (10.156) and (10.161) in the Complement Section 10.6, for
the typical peak luminosity in the blue band of a type Ia SN we get
LB,peak ' 8 × 109 LB, . (10.4)
This huge luminosity allows us to see type Ia SNe at cosmological dis-
tances. Then, the fact that their intrinsic peak luminosity is, to a first
approximation, quite uniform, makes type Ia SNe potential standard
candles for cosmological purposes. Actually, the observed spread in peak
magnitudes of type Ia SNe by itself would still be too large for them to be
used as accurate standard candles in cosmology. However, the residual
differences in the intrinsic luminosities of type Ia SNe can be corrected
thanks to empirical relations between the peak luminosity and the shape
of the light curve, with faster declining SNe being fainter. The original
correlation, known as the Phillips relation (or the Pskovskii–Phillips re-
lation), is expressed in terms of ∆m15 (B), the decline rate in B-band
magnitude after 15 days, and has the form18
(MB )peak = −21.727 + 2.698 ∆m15 (B) , (10.5)
with similar linear correlations in the V - and I-bands. Many refinements
of this idea have been developed. For instance, the correlation between
peak luminosity and the shape of the light curve can be expressed either
using a “stretch parameter”, defined introducing a stretch in the time
axis relative to a standard luminosity template (Perlmutter et al. 1997),
or using a multi-parameter fit in multiple colors (Riess et al. 1996),
which also makes use of the fact that fainter type-Ia SNe appear redder
than brighter objects. With these corrections the dispersion MB can be
reduced to about ∆MB ' 0.1 and, as a result, the distance to type Ia
SNe can be deduced to about 10% accuracy. This has allowed the use
of type Ia SNe at cosmological distances as accurate probes of the ex-
pansion of the Universe. In this way it is has been shown that in the
recent cosmological epoch the expansion of the Universe is accelerating,
providing clear indication in favor of the existence of a dark energy com-
ponent, that dominates the total energy budget of the Universe in the
recent cosmological epoch.19
Luminosity of core-collapse SNe
In contrast, the luminosities of typical core-collapse SNe span a broad
range of absolute peak magnitudes, from about −15 to −20.5. We see,
from eq. (10.156) in the Complement Section 10.6, that a difference in
magnitude ∆M ' 5 corresponds to a factor 100 in luminosity. Excep-
tionally, core-collapse SNe with magnitude up to about M = −22 have
been observed.
 10.2 Properties of Supernovae 17

Fig. 10.4 Comparison of the absolute R-band light curve of various SNe. The
curve labeled SN 1998dh is a typical SN Ia. All other curves represent core-
collapse SNe, except possibly the most luminous SN 2006gy, which could be
a pair-instability SN. From Smith et al. (2007).

Contrary to type Ia SNe, there does not seem to be a correlation,

such as the Phillips relation, between the decline rate of the light curve
and the peak absolute magnitude. From systematic studies of R-band
photometry,20 taking into account the host galaxy extinction, which is 20
See the Complement Section 10.6 for
often significant, one finds that the populations of type Ib and type Ic definitions of the different color bands.
SNe are statistically indistinguishable, with average values of the peak
magnitude in the red band

hMR,peak i = −18.2 ± 0.9 . (10.6)

Individual type Ib and Ic SNe, however, can have fluctuations in the
peak absolute magnitude of approximately ±2 magnitudes around these
average values. Type Ic-BL SNe have a somewhat higher luminosity,
with a red-band peak magnitude
hMR,peak i = −19.0 ± 1.1 , (10.7)
and a distribution scattered by about ±1 mag around the mean.
Figure 10.4 shows the R-band luminosity curves of various SNe, dis-
playing a typical type Ia SN (SN 1998dh) together with a variety of
core-collapse SNe. We see the typical plateau of a type II-P SN in the
light curve of SN 1999em, while SN 1987A, in the Large Magellanic
Cloud, the best observed supernova ever, is considered an extreme case
of type II-P, with a steady increase in magnitude lasting about 3 months,
and a low peak magnitude. As we will see in Section 10.3.1, the low lu-
minosity of SN 1987A is due to the fact that, at the time of explosion, its
18 Stellar collapse

progenitor was a blue supergiant, rather than a red supergiant. Among

the other curves in the figure, SN 1994W is a type IIn SN that is pow-
ered by a strong interaction with its circumstellar material. Particularly
noteworthy is the upper curve, which refers to SN 2006gy, a most re-
markable SN explosion. This type IIn SN exploded in Sept. 2006 in
the galaxy NGC 1260, at a distance of about 73 Mpc. It reached a
peak visual magnitude of about −21.8, and stayed brighter than −21
for about 100 days. It was also characterized by a very slow rise, reach-
ing its maximum luminosity in about 70 days, instead of the 20 days of
a typical SN. Its total radiated electromagnetic energy is estimated at
∼ 1051 erg, two orders of magnitudes larger than typical core-collapse
SNe. A yet more luminous supernova, SN 2005ap, has been recorded,
about twice as bright as SN 2006gy at peak luminosity, although not
as energetic overall, since its light curve declined in a few days. A few
other ultra-luminous SNe with comparable radiated energy have been
found in recent years. The mechanism that powers these ultra-luminous
events is still debated. A possible explanation is that SN 2006gy is a
pair-instability SN, discussion of which we defer to page 33.

10.2.3 Rates
For the purposes of direct observation with GW detectors we are par-
ticularly interested in the rate of SNe in our Galaxy (although ‘third-
generation’ GW interferometers such as the Einstein Telescope could
detect SNe at extragalactic distances, and SNe at cosmological dis-
tances are also potentially interesting sources of stochastic backgrounds
of GWs). However, most information on SN rates comes from observa-
tions of SNe in other galaxies. To translate the information obtained
from other galaxies into predictions for the SN rate in our Galaxy, the
rates must be normalized to some quantity correlated to the stellar mass
of the galaxy. The most commonly used quantity is the B-band lumi-
nosity of the galaxy in question, which is a measure of the stellar mass
of the galaxy, at least for galaxies of the same morphological type. The
classical units is the SNu (SN unit in B-band), defined as the number of
SN events per century, per 1010 LB, , where LB, is the solar luminosity
in the B-band. When one wishes to emphasize that the normalization
has been performed to the B-band luminosity, this unit is also denoted
by SNuB. Another useful normalization is to the far-infrared (FIR) band
luminosity, and the corresponding unit is denoted by SNuIR. Its useful-
ness is that the FIR luminosity of a galaxy is generated by dust heated
by massive stars and is therefore proportional to the star formation rate.
To transform from SNu to rates in our Galaxy we need its B-band lu-
minosity, which is
LB,Gal ' 2.3 × 1010 LB, , (10.8)
so the rate of supernova explosion in our Galaxy, expressed in terms of
the SNu unit, is about 2.3 SNu. Recent determination of the SN rates
using the existing databases of extragalactic SNe, for the different types
of SNe and for different classes of host galaxies, are shown in Table 10.2.
 10.2 Properties of Supernovae 19

Table 10.2 The local SN rates, per galaxy type, in SNu. As usual, h0 =
H0 /(100 km s−1 Mpc−1 ). From Cappellaro and Turatto (2001).

Galaxy SN type
type Ia Ib/c II All

E-S0 0.32 ± .11 h20 < 0.02 h20 < 0.04 h20 0.32 ± .11 h20
S0a-Sb 0.32 ± .12 h20 0.20 ± .11 h20 0.75 ± .34 h20 1.28 ± .37 h20
Sbc-Sd 0.37 ± .14 h20 0.25 ± .12 h20 1.53 ± .62 h20 2.15 ± .66 h20

All 0.36 ± .11 h20 0.14 ± .07 h20 0.71 ± .34 h20 1.21 ± .36 h20

We see that core-collapse SNe are not observed in elliptical galaxies,

consistently with the fact that their progenitors are massive, short-lived
stars. In contrast, the rate of type Ia SNe is constant, within the error,
from elliptical galaxies to late spirals. Taking our Galaxy to be type
Sb-Sbc (which is accounted for by performing an interpolation of the
rates on a grid of galaxy types), using eq. (10.8) and setting h0 = 0.70,
one finds that the expected rates RG in our Galaxy are

RG (type Ib + Ic + II) = (1.7 ± 1.0) events/century , (10.9)

and
RG (type Ia) = (0.5 ± 0.2) events/century . (10.10)
Probability of Milky Way SN
100%
Thus, overall we get 80%
1/Century
2/Century
3/Century
60% 5/Century
RG (all) = (2.2 ± 1.2) events/century , (10.11)
40%

i.e. approximately between one and three Galactic SNe per century. 20%
This rate would imply that there should be about 44 ± 24 SN remnants 0%
Next 10 Years Next 20 Years Next 30 Years
in our Galaxy younger than 2000 years, which is consistent with the
O(20 − 30) that are presently known; see the discussion on page 9. Fig. 10.5 Probabilities for one or
Somewhat higher rates are predicted using the five historical SNe ob- more SNe in the Galaxy over dif-
served in the last 1000 yr, and generating with a Monte Carlo simula- ferent time spans, depending on the
tion a number of SN events distributed in the Galaxy, to see how many assumed SN rate. From Kistler,
events are necessary in order to have five events with apparent magni- Yuksel, Ando, Beacom and Suzuki
tude mV < 0. Using a realistic model of the Galaxy, which includes the (2011).
thin and thick disks as well as the stellar halo, plus a realistic distribu-
tion of dust to compute the extinction, one finds that about 39 events
would be needed to produce the five visible historical SNe in the last
1000 yr, corresponding to a rate21 21
See Tammann, Loeffler and Schroder
(1994).
RG (all) = (3.9 ± 1.7) events/century . (10.12)

This is higher than the value given in eq. (10.11), but consistent within
the errors, especially in view of the uncertainties in the modelization
used and of the low statistics of historical SNe.
20 Stellar collapse

22 An upper bound on the Galactic SN rate can be obtained from the

See Diehl et al. (2006).
absence of neutrino bursts of Galactic origin at neutrino detectors such
as Baksan, Mont Blanc, IMB and (Super)-Kamiokande. The detection
of neutrinos from SN 1987A in the Large Magellanic Cloud shows that
neutrinos from a Galactic core-collapse SNe would be clearly detected.
20 The absence of any detection of neutrino bursts in the last 30 years puts
an upper bound
15
Probability density fr' #r'$

RG (type Ib + Ic + II) < 7.7 events/century (90% c.l.) . (10.13)

10
A direct estimate of the rate of core-collapse SNe in our Galaxy is ob-
tained from the γ-rays produced by the radioactive decay of 26 Al. It
5
can be shown that a substantial fraction of 26 Al is of Galactic origin.
The present-day equilibrium mass in 26 Al produced by ongoing nucle-
0
0 5 10 15 20 25
osynthesis throughout the Galaxy can be estimated from this γ-ray flux,
distance r' !kpc"
making some assumptions on its three-dimensional spatial distribution.
Fig. 10.6 The probability distribu-
Since 26 Al is mostly produced by massive stars, from this one can infer
tion for explosion of a Galactic
that the rate of core-collapse SNe in the Galaxy is equal to22
SN as a function of the distance
from Earth, in kpc. From Ahlers,
RG (type Ib + Ic + II) = (1.9 ± 1.1) events/century , (10.14)
Mertsch and Sarkar (2009).
in excellent agreement with eq. (10.9). Given a rate, the corresponding
prediction for the probability of observing at least one galactic event
7
during a given time-span is given by the Poisson distribution for that
M81,M82
NGC 253
M83

NGC 6946
Andromeda, M33

rate. These probabilities are shown in Fig. 10.5 for different choices of
MW
Supernovae per decade (1999-2008)

6
2008iz the rate, ranging from 1 event per century up to a 5 per century, and
5
2008bk 2008ax for different time spans.
4
2005af 2005cs
Beside the rate of Galactic SNe, another relevant issue, particularly for
3
2004dj 2008S 2003gd 2005at 2008?
GW astronomy, is what is the distribution of distances from the Earth
2
2004am 2004et 2002bu 1999gi 2007gr
at which we can expect such events. This question can be addressed
1 NGC
2002hh1999gq 2002ap 1999em1999ev
by performing a Monte Carlo calculation with a large number of sources
300-T 2002kg

0 drawn from a probability distribution that encodes our knowledge of the

0 2 4 6 8 10
Distance [Mpc] distribution of supernova remnants in the Galaxy. To obtain a realis-
tic distribution it is also necessary to take into account the spiral arm
Fig. 10.7 Estimates of the core- structure of the Galaxy. The result, shown in Fig. 10.6, indicates that
collapse supernova rate in the the distribution of events as a function of distance from the Earth is
nearby universe, based on that
very broad, and distances between, say, 5 and 18 kpc can be considered
expected from 22 supernovae ob-
as typical.
served in 1999–2008 (bins), com-
pared with the theoretical pre- Both for GW astronomy and for neutrino astronomy it is important
diction using B-band luminosity to know what distances a detector should reach, in order to expect at
(dashed line). From Kistler, Yuksel, least one event per year. Figure 10.7 shows the number of core-collapse
Ando, Beacom and Suzuki (2011). SNe actually detected, in the local universe, at distances up to 10 Mpc,
during the period 1999–2008, while the dashed line gives the prediction
obtained from the rates in Table 10.2, after assigning the appropriate
blue luminosity and Hubble type to the approximately 40 major galaxies
within 10 Mpc. First of all one observes that the theoretical prediction
is lower by a factor of approximately 2–3, compared with the number
of observed events. Furthermore, the sample of detected galaxies is
likely incomplete, because SN surveys under-sample small galaxies and
 10.3 The dynamics of core collapse 21

the Southern hemisphere, so the actual number of SN events that oc-

1.5
curred in this period within 10 Mpc could be even larger. A possible Observed SNe
explanation for this discrepancy is that the B-band luminosity is not an Expected Rates:
1.2
accurate indicator of the number of high-mass stars, and hence of the UV
B-band

core-collapse SN rate. This is due to the fact that different galaxies can Combined

RSNH<DL @yr-1D
have different dust obscuration in the B-band, and that B-band light 0.9

receives contribution from both high-mass and low-mass stars. In any

case, using the actually observed SNe shown in Fig. 10.7, we see that 0.6

the core-collapse SN rate reaches 1 event per year at a distance of about

6 Mpc, while at 10 Mpc we can expect 2 events per year. We also observe 0.3

that all the 22 SNe observed in this period were core-collapse SNe. This

5Mt 90%

5Mt 10%
HK 10%
HK 90%
sets a bound on the ratio of type Ia to core-collapse events in the local 0
0 2 4 6 8 10
Distance @MpcD
universe of order 0.2 (at 90% c.l.). The same information is shown, in
terms of the cumulative core-collapse SN rate as a function of distance,
Fig. 10.8 The cumulative rate of
in Fig. 10.8.
core-collapse supernovae in the
The rate of core-collapse SNe at even larger distances is shown in the
nearby universe, based on that ex-
top panel of Fig. 10.9 (denoted there by RCCSN ). On the left side of the pected from 22 supernovae observed
figure the rate is plotted against the distance (in Mpc), while beyond in 1999–2008, compared with the
z = 0.1 the rate is shown against the redshift. Since massive stars are theoretical prediction using B-band
short-lived, the rate of core-collapse SNe should follow the star formation emission from 451 galaxies (dashed),
rate. The central solid line gives the prediction obtained using a fiducial UV emission from 315 galaxies (dot-
cosmic star formation history, with the upper and lower lines giving the ted), and the 589 galaxies of the
uncertainty range of the theoretical model. The bottom panel shows the combined catalog (solid). The ver-
ratio of type Ia to core-collapse events. In the local region we have an tical bands corresponds to the reach
upper bound (shown by the lines at 90% and 99% c.l.), coming from of different possible neutrino detec-
tors. From Kistler, Haxton and
the fact that the 22 SNe observed in the local region in 1999–2008, and
Yüksel (2013).
shown in Fig. 10.7, were all core-collapse SNe. At cosmological distances
this ratio takes values between 0.2 and 0.4, consistent with the values
reported in Table 10.2.

10.3 The dynamics of core collapse Kistler et al. (2008)

Smartt et al. (2008)
RCCSN [yr Mpc ]
-3

-3
10 Dahlen et al. (2008)

10.3.1 Pre-SN evolution

-1

-4
10
The evolution of a massive star is governed by the competition between Cappellaro et al. (1999)
Dahlen et al. (2004)
the gravitational contraction under its own gravity, on the one hand, and Cappellaro et al. (2005)
Botticella et al. (2008)
the pressure exerted by the radiation, by the thermal gas and, in the 1
Ia/CC ratio

10 100
99%
late stages of its evolution, by the partially degenerate electron gas, on 90%

the other. Energy is lost mainly to radiation during the core H- and He- 0.1
10 100 0.5 1
burning phases, and to neutrino emission afterwards, and this energy is distance D [Mpc] redshift z
provided by the thermonuclear reactions taking place in the inner parts
of the star. These reactions are very sensitive to the temperature in Fig. 10.9 Evolution of the core-
the core. High temperatures are necessary to overcome the Coulomb collapse supernova rate (top panel)
repulsion of the positively charged nuclei and bring them close enough. and the ratio of Type Ia to core-
At that point the attractive part of the short-range strong force between collapse SNe (bottom panel). From
nucleons can take over and create a bound nuclear state. The higher Horiuchi, Beacom and Dwek (2009).
the electric charge of the nuclei, the higher is the Coulomb barrier to
be overcome. When a protostellar cloud collapses under its own self-
22 Stellar collapse
23
A more detailed discussion of the
value of this minimal mass, as well as
its dependence on metallicity, is given
on page 30. The notion of star metallic-
ity is reviewed in the Complement Sec-
tion 10.6.
24
See e.g. Padmanabhan (2001), Sec-
tion 3.4.3, or Carroll and Ostlie (2007),
Section 13.1, for derivations and de-
tailed discussions.
gravity, the first thermonuclear reaction that can ignite is the fusion of
deuteron and proton to give 3 He, D + p → 3 He + γ, and afterwards the
fusion of protons into 2 He nuclei. If the mass of the protostellar cloud
is smaller than about 0.08M , the central temperature is never large
enough to allow for the ignition of stable thermonuclear reactions, and
radiation is only emitted as the cloud contracts and cools down. The
outcome is then a planet or a brown dwarf. If instead M > ∼ 0.08M
the star starts burning steadily H to He in its core. For a star of mass
M ' M this phase lasts about 1010 yr, and in this stage the star is
on the main sequence of the Hertzsprung–Russell (HR) diagram. More
generally, in a large range of masses, the evolutionary time-scale on the
main sequence can be approximately reproduced by
tevol ' 7.3 × 109 (M/M )−5/2 yr . (10.15)
When the H fuel in the core is nearly exhausted the star enters a
complicated evolutionary phase, which depends strongly on its mass.
We focus here on massive stars, M > ∼(6 − 9)M , since a value in this
range is believed to be the minimum progenitor mass necessary for a SN
explosion.23 When the H fuel is almost exhausted the He core can no
longer support itself, and begins to contract. As a result, the temper-
ature at the edge of the core increases sufficiently that the H in a shell
around the core can again undergo fusion into He. At this stage the
star consists of an inert He core surrounded by a burning H shell. The
core no longer produces energy, so it has a vanishing luminosity. Since
the luminosity produced at radius r is proportional to the temperature
gradient dT (t)/dr, the vanishing of the luminosity implies that the He
core is isothermal. However, the luminosity produced by the H shell is
larger than the one that was produced during core-H burning. Part of
this energy goes into a slow expansion of the envelope, and as a result
the star becomes larger and redder.
The burning H shell continues to produce He, and these “nuclear
ashes” increase the size of the He core. This phase of the evolution ends
when the He core is no longer able to support the weight of the matter
over it. Indeed, for a star of mass M with an isothermal core of mass
qM and an envelope of mass (1 − q)M , there is a maximum value of
q beyond which the isothermal core can longer sustain the weight of
the layers above it. This critical value, known as the Chandrasekhar–
Schönberg limit, is given by qmax ' 0.37(µenv /µic )2 , where µenv and
µic are the mean molecular weights in the envelope and in the core,
respectively.24 For a star with solar-mass composition, qmax ' 0.08. For
low-mass stars at this stage the electron gas starts to be degenerate, and
electron degeneracy pressure increases the value of qmax . For instance,
for a 1 M star the resulting value is qmax ' 0.13. In contrast, for
a massive star, say M > ∼ 5M , the effect of electron degeneracy in the
core at this stage is still negligible, so when the mass in the He core
becomes about 8% of the total mass of the star the He core begins to
contract, on a time-scale much shorter than the typical time-scale of the
main-sequence evolution. For a star with M = 5M , this phase lasts
 10.3 The dynamics of core collapse 23

about 3 × 106 yr. The contraction raises the temperature in the core,
and when the inner core temperature reaches a value T ∼ 108 K, He
burning ignites and stops the contraction of the core. At this stage the
star has two sources of nuclear energy. An inner core, where He is burnt
to carbon and oxygen, and a shell where H is burnt to He. The two
are separated by a He-rich region, where the temperature is not large
enough to burn He.
Carbon is produced through the triple-alpha process in which two α
particles, i.e. nuclei of 42 He, collide to produce beryllium,
4
He + 4 He ↔ 8 Be . (10.16)

The 8 Be nucleus is unstable and will decay very quickly back to two α
particles, unless it reacts immediately with a third α particle to produce
C,
4
He + 8 Be → 12 C + γ . (10.17)
In this sense, the triple-alpha process is really a three-body collision.
After sufficient C has been produced, in the He-burning core there are
the conditions for producing oxygen, through
12
C + 4 He → 16
O+γ. (10.18)

When most of the He fuel has been burnt, thermonuclear reactions can
no longer support the C–O core, which contracts. Then a He shell around
it ignites.
For masses in the range (6 − 9)M the subsequent evolutionary sce-
nario is not so well known, with a final output that can be (from the lower
to the upper mass range), a CO white dwarf, a NeMgO white dwarf, an
electron-capture SN before the neon photo-disintegration phase, and fi-
nally nuclear burning all the way up to the production of an iron core.25 25
See Woosley and Heger (2015), and
For a star with a mass higher than a critical value Mup in the range the discussion on page 31.
(7 − 9)M , after further contraction the burning of C in the core begins.
The theoretical prediction for the critical mass depends on issues such
as the treatment of convection. For M > Mup the star now has a core
where C is burnt, a burning He shell and an outer H-burning shell,
separated by inert regions. C burns through the direct fusion of two 12 C
nuclei into a highly excited state of 24 Mg, which then decays into various
channels, including decay to 20 Ne. After C burning, the composition of
the core is mostly dominated by the 16 O produced during He burning
and by 20 Ne and 24 Mg produced by C burning. Even though O has a
lower Coulomb barrier, neon burning ignites first through the process
20
Ne + γ → 16 O + 4 He, induced by high-energy photons from the tail of
the Planckian distribution, and later O burning ignites.
Ne ignites for stars with a mass above a critical value that is some-
what uncertain, but is estimated to be in the range (10 − 12)M . After
the ignition of Ne the star goes through the subsequent stages of O and
silicon burning. Si burning eventually produces a core of elements of the
iron group, mostly 54 Fe, 56 Fe and 56 Ni. These nuclei have the highest
binding energy per nucleon, so no further energy can be extracted by
24 Stellar collapse

fusion of these elements. Their relative abundance at a given temper-

ature and density is determined not only by their binding energy per
nucleon, but also by the ratio of neutrons to protons in the core. For
temperatures T ∼ < 1010 K, the equations of nuclear statistical equilibrium
favor the most tightly bound nucleus for a given neutron excess (i.e. the
relative excess of neutrons of protons). For a composition with approxi-
mately an equal number of neutrons and protons this is 56 Ni while for a
neutron excess η ' 0.07 it is 56 Fe, and for even higher η it shifts toward
heavier isotopes.
Neutrino losses also play an important role in the dynamics of the star
in these stages. When the temperature exceeds 5 × 108 K, thus after He
burning, neutrino losses due to pair annihilation become the dominant
mechanism for the energy loss of the star, more important than radiative
diffusion and convection. When the Si-burning phase is reached neutrino
losses due to electron capture also become important. Neutrino produc-
tion through pair annihilation is very sensitive to temperature, with a
loss term roughly proportional to T 9 . As we go through the process
of exhausting a nuclear fuel, core contraction and ignition of the next
nuclear fuel, the core temperature rises, and neutrino losses raise very
rapidly. This leads to an acceleration of the burning process. As we see
Table 10.3 Core temperature, core from Table 10.3, for a star with M = 15M the main-sequence lifetime
density and duration of various core- of steady H burning lasts 11 Myr, He burning 2 Myr, and C burning
burning stages of a star with M = 2000 yr, while Si burning is completed in just about 18 days! The corre-
15M . From Table I of Woosley,
Heger and Weaver (2002). sponding values for a star with M = 25M are given in Table 10.4. Thus,
the internal structure of the star evolves on a time-scale much shorter
Fuel T (K) ρ (g/cm3 ) τ that that needed to propagate these changes to the surface. After C
burning has started, the surface luminosity and temperature basically
H 3.5 · 107 5.8 11 Myr have no time to adjust to the changes in the interior and do not evolve
He 1.8 · 108 1.4 · 103 2 Myr
C 8.3 · 108 2.4 · 105 2 kyr
significantly, until the star explodes as a SN.
Ne 1.6 · 109 7.2 · 106 0.7 yr During its supergiant phase, when the star has an extended envelope
O 1.9 · 109 6.7 · 106 2.6 yr supported by He burning (either in a shell or in the core), there are two
Si 3.3 · 109 4.3 · 107 18 d possible solutions for the equations of stellar structure, a red supergiant
or a blue supergiant, separated by thermally unstable solutions. Red
supergiants have a convective envelope, lower temperatures and a much
larger radius than blue supergiants. Whether a star evolves to a blue
Table 10.4 As in Table 10.3, for M = or to a red supergiant depends on its mass, metallicity and mass loss
25M . From Table I of Woosley, Heger rate. In general, for a star to spend most of its core He-burning phase as
and Weaver (2002).
a blue supergiant directly after the main-sequence phase, it must have
M∼ < 20M , low metallicity and low mass losses. Whether the progenitor
Fuel T (K) ρ (g/cm3 ) τ
of a SN was a red or a blue supergiant has important implications for
H 3.8 · 107 3.8 6.7 Myr the luminosity of the SN explosion. Since a blue supergiant is more
He 2.0 · 108 0.8 · 103 0.8 Myr compact, its envelope sits in a deeper gravitational well and a larger
C 8.4 · 108 1.3 · 105 0.5 kyr fraction of the energy of the explosion goes into ejecting the envelope
Ne 1.6 · 109 3.9 · 106 0.9 yr
O 2.1 · 109 3.6 · 106 0.4 yr
of the star out of this potential well. As a result, the electromagnetic
Si 3.6 · 109 3.0 · 107 0.7 d luminosity released in the SN explosion of a blue supergiant is lower.
A clear example of this is provided by the explosion of SN1987A in the
Large Magellanic Cloud. As we see from Fig. 10.4 on page 17, SN1987A
was a sub-luminous supernova. The sub-luminous nature of SN1987A is
 10.3 The dynamics of core collapse 25

explained by the fact that its progenitor was a blue supergiant. In fact,
in this case we know the progenitor, which was catalogued as Sk 202-69,
and it is indeed known that at the time of the explosion it was a 12-th
magnitude blue supergiant, with M ' 20M . Actually, observations of
the circumstellar material show that the star was a red supergiant until
about 30,000 years before the explosion. The transition from red to blue
supergiant shortly before the explosion could take place in a single star
by a combination of low metallicity and fast rotation, resulting in an
envelope richer in He. Such a heavier envelope favors a blue solution.
Another possibility is that the progenitor was originally in a binary
system, with a red supergiant of about (16 − 18)M and a companion
with a mass of order 3M , probably a main-sequence star. In this
scenario the red supergiant expands and engulfs the companion star. In
the following phase of common envelope evolution part of the common
envelope is ejected while the main sequence star is tidally disrupted.
The resulting larger mass of the envelope drives the transition from red
to blue supergiant.

10.3.2 Core collapse and neutrino-driven delayed

shock
The collapse of the core
After the burning of Si in the core is completed, the star has an onion-
like structure with a Fe–Ni core (usually generically referred to simply
as the iron core), surrounded by a shell of Si burning and further con-
centric shells of O burning, Ne burning, C burning, He burning and H
burning, separated by inert regions. The pressure in the core at this
stage is dominated the degeneracy pressure of the relativistic electrons,
and stability is only possible as long as the mass of the core is below the
Chandrasekhar mass MCh . However, the burning of Si in a shell outside
the core keeps producing more Fe–Ni “ashes”, which increase the mass of
the core. When the mass of the partially degenerate Fe–Ni core exceeds
the critical threshold, the core begins to collapse. The Chandrasekhar
mass is given by
MCh ' 5.828 Ye2 M , (10.19)
where Ye is the number of electrons per baryon,
ne
Ye = , (10.20)
nB
and ne and nB are the number densities of electrons and baryons, re-
spectively. If we take equal number densities for protons and neutrons
(whether free or bounds in nuclei), we have nB = np + nn = 2np . Fur-
thermore, by charge neutrality np = ne , so we get Ye = 1/2, which gives
MCh ' 1.457M . This value is reduced to 1.42M by general-relativistic
corrections. Actually, because of an excess of neutrons over protons, Ye
is somewhat smaller than 1/2, and across the iron core typically ranges
from a value Ye ' 0.42 at the center to a value Ye ' 0.48 at the edge.
26 Stellar collapse

Taking an average value Ye ' 0.45 gives MCh ' 1.18M , which is then
reduced to 1.15M by general-relativistic corrections. This is the critical
mass for a completely degenerate electron gas, i.e. for a gas where the
electron temperature is negligible compared with the Fermi energy. In
the iron core, however, the electrons are only partially degenerate and
a significant contribution to the pressure still comes from the electron
temperature, raising the critical mass to a value
"  2 #
πkT
MCh (T ) ' MCh (T = 0) 1 + , (10.21)
F

where F is the Fermi energy,

 1/3
ρ
F ' 1.1 Ye1/3 MeV . (10.22)
10 g/cm3
7

During Si-shell burning, the mass of the core increases because of the
production of Fe–Ni ashes. At the same time the critical value (10.21) de-
creases because the large energy losses due to neutrino emission lower the
value of the T -dependent term. This finally brings the iron core beyond
the critical value, and the collapse begins. For a star with M = 15M
the mass of the core at time of collapse is about Mcore ' 1.34 M , while
heavier stars collapse with heavier iron cores. As a result, one can ex-
pect that more massive stars produce as remnants more massive neutron
stars, although the mass of the remnant also depends on the subsequent
collapse dynamics, as we will discuss in detail in Section 10.3.3.
The collapse is then accelerated by two instabilities. First, as the
density in the collapsing core increases, electrons reach Fermi energies
of the order of MeV, and the process of electron capture by heavy nuclei
becomes energetically favorable. This produces neutron-rich nuclei via
inverse β-decay and removes electrons. Therefore Ye decreases, and con-
sequently the electron pressure that was trying to oppose the collapse
also decreases, or, in other words, the maximum mass that can be sta-
bilized by electrons decreases. Indeed, we see from eq. (10.21) that the
term MCh (T = 0) is proportional to Ye2 while the T -dependent term is
4/3
proportional to MCh (T = 0)/2F and therefore to Ye , so both terms
decrease. A second instability is due to photo-disintegration. At the
temperatures now present in the core, which are of order 1010 K (so
kB T ∼ MeV), photons are sufficiently energetic to destroy even some of
the strongly bounds nuclei of the iron group, with processes such as
56
Fe + γ → 13 4 He + 4n . (10.23)

This process is highly endothermic, since the 56 Fe nucleus has higher

binding energy per nucleon than the α particles. It therefore reduces
the thermal energy of the core, and decreases the T -dependent term in
eq. (10.21). It should be noted, however, that due to the exponentially
large number of excited nuclear states available at large temperatures,
the partition function still favors the presence of bound heavy nuclei, and
 10.3 The dynamics of core collapse 27

prevents their complete disintegration into α particles, or even further

into free nucleons.
Thus, the collapse accelerates; in the inner core the collapse is homol-
ogous, i.e. the collapse velocity is proportional to the radial distance,
until a radius where the velocity exceeds the local sound speeds. Beyond
this radius the collapse can no longer continue to increase linearly with
distance, but still this outer core is supersonic. In about one second,
the core collapses from the typical size of a white dwarf, i.e. about the
Earth’s radius, down to a radius of about 50 km.
The next remarkable step takes place when the density reaches a value
ρ ' 1012 g/cm3 . Then the neutrinos, which are being copiously produced
by electron capture and, to some extent, by the β-decays of some of the
nuclei, become trapped. Neutrinos in fact scatter coherently over the
nuclei, and their diffusion time becomes larger than the typical collapse
time-scale.

Bounce and neutrino-driven delayed shock

Eventually the whole core reaches nuclear densities. At this point the
whole collapsing core can be seen as a single gigantic nucleus. At such
a density the strong interaction between nucleons becomes repulsive.
When the density reaches the value ρ ' 8 × 1014 g/cm3 , approximately
three times larger than the density of an atomic nucleus, the nuclear
matter bounces back, sending a shock wave into the outer core, which
is still falling inward at supersonic speed.
Originally it was thought that this bounce would have caused the
explosion of the star. Nowadays, numerical simulations have conclusively
shown that this “prompt” bounce-shock is not sufficient, by itself, to
drive the SN explosion. The reason is that when the shock wave meets
the infalling outer core most of its energy is lost, and goes into heating
the outer core. The violence of the encounter is such that the iron of the
outer core is completely disintegrated into protons and nucleons. This
removes from the shock about 1051 ergs for each 0.1M of the outer
core. Furthermore, this change in the composition of the outer core
induces further energy losses, because the capture rates of electrons on
free protons, through inverse β-decay e− + p → n + νe , is larger than on
heavy nuclei, because of blocking from nuclear-shell effects. A fraction
of the neutrinos produced in this way leaves the star quickly, producing
a neutrino burst, which carries away further energy that could otherwise
have contributed to the explosion. The shock then stalls, and after about
10 ms from the core bounce one is left with an almost stationary shock
wave, at a radial distance of 100–200 km, accreting the infalling matter
of the outer core at a rate of about (1 − 10)M s−1 .
A second energy source is therefore needed at this stage to drive the
explosion. While there is a consensus on the fact that the prompt explo-
sion mechanism does not work, the details of the mechanism that does
finally drive the explosion are not yet known. It is, however, clear that
a crucial role in reviving the shock is played by the neutrinos. In the
28 Stellar collapse
26
This is basically related to the fact
that, as we will discuss on page 31,
stars in this mass range explode when
they have a Ne–O core, and at the edge
of such a core the density gradient is
very steep. In contrast, heavier stars
explode after burning their fuels all the
way to iron, and therefore have an iron
core. The profile of iron cores is much
flatter, leading to larger infall and ram
pressure from the external layers. Iron
cores are therefore much harder to ex-
plode than Ne–O cores.
inner core region, because of the high density, neutrinos cannot stream
freely. Therefore, just as a star has a photosphere that delimits the re-
gion where the photons can begin to stream freely to the exterior, in
this case a “neutrinosphere” develops below the shock front. Neutrinos
deposit energy in the matter between the nascent neutron star in the
inner core and the photosphere, through scattering processes such as
νe + n → e− + p and ν̄e + p → e+ + n. Since most of the gravitational
energy liberated in the collapse goes into neutrinos, these processes are
quite significant, and heat up the matter below the shock front, gen-
erating a further pressure that can revive the shock, on a time-scale of
0.1 s. In fact, to revive the shock, it is sufficient that a few percent of the
total neutrino energy is converted into thermal energies of the nucleons,
leptons and photons. If this happens, the shock wave resumes and lifts
the external layers of the star in a SN explosion.
Hydrodynamical instabilities
While the scenario just described is theoretically appealing and convinc-
ing, a proof that this is what actually happens must rely on numerical
studies. Simulations of SN explosions are, however, of the outmost com-
plexity. Present simulations unavoidably make important approxima-
tions of some sort, such as Newtonian rather than general-relativistic
gravity and spherical symmetry instead of full three-dimensional simu-
lations, but the outcome turns out to depend strongly on details. Spher-
ically symmetric simulations, even including accurate hydrodynamic
codes, neutrino transport and an extended set of neutrino reactions,
fail to obtain SN explosions unless the progenitor is in the mass range
∼ (7 − 9)M .26 For larger masses, this points toward the importance of
instabilities in a full three-dimensional setting. Rayleigh–Taylor finger-
ing instabilities have been considered as a possible boost for neutrino
luminosity. More generally, since the heating from neutrinos is strongly
dependent on the distance from the center, convective instabilities are
expected to play an important role. This is also confirmed by the de-
tailed observations that have been made possible by the explosion of
SN 1987A in the Large Magellanic Cloud, which have shown that ra-
dioactive nuclei produced deep in the core of the star have been mixed
with the hydrogen envelope of the star. Three-dimensional numerical
simulations have indeed shown that violent convective overturns appear.
To assess whether this really leads to SN explosion, it is necessary to
implement accurate energy-dependent neutrino transport codes, which
in the full three-dimensional setting is beyond present computer power.
However, three-dimensional simulations have revealed the existence of
non-radial instabilities that might be very important for the explosion
mechanism. In particular it has been found that, under the effect of
accretion from the infalling material, the shock front displays a non-
radial instability termed SASI (standing accretion shock instability),
which in a decomposition in spherical harmonics is especially evident in
the l = 1, 2 mode, i.e. it produces a dipole and a quadrupole deformation
 10.3 The dynamics of core collapse 29

of the accretion shock front. This causes a back-and-forth sloshing of

the shock front, which transfers energy outward more efficiently than
convection, and ultimately can lead to an asymmetric explosion. At the
same time, such a highly anisotropic explosion can result in a kick of the
newborn neutron star, with velocities that in some numerical simulations
have reached 1000 km/s, in agreement with the measured velocities of
young pulsars.
Magnetohydrodynamic instabilities might also play an important role
in driving the explosion. The rotational energy stored in the nascent
proto-neutron star is huge,
  2  
52 MNS RNS 1 ms
Erot ∼ 2.4 × 10 erg . (10.24)
1.5M 10 km PNS
Magnetohydrodynamic processes can in principle extract efficiently part
of this rotation energy, and explode the star. The numerical investi-
gation of these mechanisms is, however, very challenging, since it re-
quires full three-dimensional simulations, accurate neutrino transport
codes and high numerical resolution.

Asymmetric explosions and pulsar kicks

The proper motion of a pulsar can be measured using pulsar timing
or interferometers. To translate the proper motion into a velocity one
also needs to infer the pulsar’s distance. Unless annual parallaxes are
available, the distance is estimated using the dispersion measure, defined
in eq. (6.37), together with a model of the Galactic electron density.
From the study of 73 young pulsars (with ages less than 3 Myr) one
finds that the average value of their inferred three-dimensional velocity
is
hv3D i = (400 ± 40) km/s , (10.25)
while the average value of their two-dimensional velocity in the plane
transverse to the line of sight is

hv2D i = (307 ± 47) km/s . (10.26)

The two-dimensional velocity distribution is shown in Fig. 10.10. Ob-
serve that some pulsars have velocities larger than 1000 km/s. In partic- Fig. 10.10 The distribution of two-
ular, PSR B2011+38 and PSR B2224+64 have two-dimensional speeds dimensional velocities for young pul-
of about 1600 km/s. These proper-motion velocities are statistically sars. From Hobbs, Lorimer, Lyne
and Kramer (2005).
much higher than those of their progenitor stars, the massive OB stars.
This suggests that pulsars might get a significant kick at birth, during
the SN explosion. Various mechanisms have been proposed to explain
these kicks. In an asymmetric explosion the ejecta exert on the proto-
neutron star a direct “contact” force through hydrodynamical interac-
tions during the explosion, which, however, can hardly produce a kick of
about 200 km/s. More importantly, numerical simulations of supernova
explosions suggest that the ejecta also exert a long-lasting asymmetric
gravitational acceleration on the proto-neutron star, due to the fact that
30 Stellar collapse
27
This is due to the fact that “metals”
(in the sense of elements other than H
and He in the chemical composition of
the star, as discussed in the Comple-
ment Section 10.6) increase the opac-
ity, i.e. have a larger scattering cross-
section on photons. This increases the
momentum transfer from the photons
to the matter and therefore the stellar
winds.
28
See Poelarends, Herwig, Langer and
Heger (2008) and Fryer et al. (2011).
29
See van Kerkwijk, Bergeron and
Kulkarni (1996), and Kilic, Prieto,
Brown and Koester (2007). The value
of the mass is determined spectroscopi-
cally. From the broadening of the spec-
tral line one can determine the surface
gravity of the star, and therefore the
ratio M/R2 . Using the mass-radius re-
lation for a WD one can then infer sep-
arately R and M . Actually, the mass-
radius relation depends on the chem-
ical composition of the WD. A WD
with such a low mass is the remnant
of a star that never burned He, so
it is self-consistent to use the mass-
radius relation for a He WD. Theoret-
ically, the mass-radius relation is bet-
ter known at zero temperature, but one
must also include corrections due to the
WD temperature (which is also deter-
mined spectroscopically). The theoret-
ical uncertainty that enters in model-
ing the temperature dependence of the
mass-radius relation dominates the er-
ror on the mass reconstruction, which
is of order 0.02M .
the gravitational force exerted by ejecta that leave faster in one direc-
tion does not compensate that of the slower and typically denser ejecta
leaving in the opposite direction. This mechanism can build velocities
larger than 1000 km/s. As we will study in Section 14.3.5, even-higher
recoil velocities can be imparted to the remnant in the coalescence of
spinning compact binaries.
10.3.3 The remnant of the collapse
An important question is what remnant, if any, is left behind by a SN
explosion. The answer depends on the mass of the collapsing core at the
time of explosion and on the amount of material that falls back after the
explosion. The mass of the collapsing core in turn depends on the initial
mass Mprog of the progenitor star (more precisely, on the mass that
the progenitor had at the time when it first reached the main sequence,
known as the zero-age main sequence, or ZAMS) and on the history of
its mass loss. For single stars the only mass loss mechanism is provided
by stellar winds, which are functions of the star’s metallicity. Stars with
low metallicity have weaker stellar winds and retain most of their initial
mass, while stars with solar-type metallicities have stronger winds and
hence larger mass losses.27 For stars in a binary system the situation
is complicated by the various possible mass loss mechanisms due to the
interaction with the companion.
min
The exact value of the minimal progenitor mass Mprog for which the
core collapses and a SN explosion takes place has been subject to much
investigation. Earlier work suggested that it is around 9M , indepen-
dent of metallicity. More recent studies28 suggest that the transition
indeed takes place around 9M for solar metallicities Z = Z ' 0.016,
but it drops to about 6.3M at Z = 10−3 Z , with a possible fit
9.0 + 0.9 log10 (Z/Z ) if Z > 10−3 Z

min
Mprog = (10.27)
6.3M if Z < 10−3 Z .
For masses below this critical value there is no SN explosion and the
remnant is a white dwarf (WD), with a composition that depends on
the elements that the progenitor star has been able to burn. For the
remnant, we can distinguish the following possible regimes (taking into
account that the values of the masses separating these regimes are still
somewhat uncertain, and should only be taken as indicatives).
< 7M , WD remnant. On the lower end of the mass distri-
(a) Mprog ∼
bution one has WDs whose spectrum indicates that they are made of
He, and have masses Mremn ∼ < 0.45M . It is believed that their pro-
genitors were members of binary systems that lost their envelope to a
companion before reaching the tip of the giant branch phase, i.e. be-
fore He burning started. As a result the central pressure was relieved
and these stars never burned He to C and O. The lightest known WD
has a mass M ' 0.17M .29 If the envelope of the progenitor is not
stripped by binary interaction, the star can go through successive stages
 10.3 The dynamics of core collapse 31

of nuclear burning. As we saw in Section 10.3.1, for progenitor masses

below a critical value Mup , which for definiteness we take here to be
' 7M , the star is only able to burn He, producing C and O. Thus,
typical WDs are made of carbon and oxygen, and are called CO WDs.
The corresponding remnant mass Mremn increases with the mass of the
progenitor, from typical values of order 0.6M up to the Chandrasekhar
mass MCh ' (1.3 − 1.4)M .

(b) 7M ∼ < Mprog ∼< 9M : NeO WD or electron-capture SNe. For pro-

genitor masses in this range the star reaches the stage where C is burnt
to O and Ne. Then a degenerate Ne–O core forms. The subsequent
dynamics depends on the mass losses. If sufficient material is expelled,
the star will end up as a NeO WD. Otherwise, the star undergoes core
collapse. Electron degeneracy is reached before the ignition of Ne, and
the collapse of the core is triggered by electron capture on Mg and Ne,
which reduces the degeneracy pressure. Therefore we do not have the
collapse of an iron core, as in more massive stars, but rather of a Ne–O
core (or, more precisely, of a Ne–O–Mg core) and the resulting SN explo-
sion is called an electron-capture SN. The remnant of the collapse in this
case is a neutron star (NS). The mass of the collapsing core is close to
the Chandrasekhar mass, which for a Ne–O core is MCh ' 1.38M , and
the mass of the NS remnant will then be equal to the Chandrasekhar
mass minus the mass of the ejected material. Simulations of the collapse
predict ejected masses typically around 0.02M (although values in the
range (0.01 − 0.20)M could be possible), leading to a typical mass of
the NS remnant of about 1.36M .
In typical models30 one finds that, for solar metallicities, electron- 30
See e.g. Poelarends, Herwig, Langer
capture SNe takes place in a window of progenitor masses ∆Mprog = and Heger (2008).
0.25M , near Mprog ' 9M . Convolving this result with the mass dis-
tribution of progenitor stars given by the Salpeter initial mass function,
one finds that electron-capture SNe should be about 4% of all SNe in
the local Universe. With more extreme choices of the mass loss rate
the window of progenitor masses can widen up to ∆Mprog = 1.4M ,
and the fraction of electron-capture SNe can rise to 20%. Furthermore,
for lower metallicities, the window for electron-capture SN widens and
shifts toward lower masses.

(c) 9M ∼ < Mprog ∼< 10.3M . For stars in this regime the dynamics can
be quite complicated. Ne and O ignite off-center (while C ignites cen-
trally). The off-center ignition results in so-called “convectively bounded
flames”. Si can also ignite off-center, possibly leading to a localized de-
flagration. These complications make it difficult to realistically model
the evolution in this mass range. Unless an electron-capture SN takes
place, the evolution proceeds to the formation of an iron core.

(d) 10.3M ∼< Mprog ∼

< (30 − 40)M : core collapse to NS or BH. Starting
from a mass around 10.3M , all fuels ignite centrally. These stars burn
32 Stellar collapse

the nuclear fuels all the way up to iron and form an iron core that
undergoes collapse, giving rise to the classic core-collapse SNe. As we
discussed in Section 10.3.2, the SN explosion is most probably driven
by instabilities, such as convective instabilities or the standing accretion
shock instability (SASI), that enhance the conversion of the neutrino
energy into kinetic energy of the nucleons, leptons and photons. The
material accelerated in this way moves outward, pushing the layers above
it and producing the explosion. In the process, because of the work
done lifting the upper layers, some of the material belonging to the
shock front loses enough energy to be decelerated below the local escape
velocity. This material will then fall back onto the newborn proto-NS,
adding to the mass of the remnant. For progenitor masses below 11M
the amount of fallback is believed to be negligible. In contrast, for
progenitors heavier than about 11M it can become very important.
Most of the fallback takes place in the first 10–20 s after the explosion.
The amount of fallback depends on the explosion mechanism and on
details of the stellar structure, so it gives rise to the main theoretical
uncertainty in the computation of the remnant mass for stars in this mass
range. In general, however, we can consider three possible scenarios: (1)
A successful SN explosion is launched, with most of the star’s material
being ejected, although some material falls back onto the the proto-NS,
increasing the mass of the NS remnant. (2) A successful SN explosion is
launched, but the amount of material that falls back is sufficient to push
the proto-NS beyond its Chandrasekhar limit. In this case, on a time-
scale ranging from minutes to hours, we have black hole (BH) formation
by fallback. (3) The explosion is not sufficiently energetic to eject the
external layers. This gives rise to a failed SN, and all the material of
the star will collapse, giving rise to the formation of a proto-NS followed
by BH formation by rapid accretion. In this case BH formation takes
place within a few seconds after core collapse. The mass of the resulting
BH depends on whether the total mass of the star at collapse goes into
the BH. The other possibility is that, even for failed SNe, a considerable
fraction of the H envelope, and possibly all of it, is anyhow stripped
because it is extremely loosely bound. In that case the mass of the
resulting BH is rather given by the mass of the He core of the pre-
collapse star. Cases (2) and (3) are referred to as the collapsar model
for BH formation.
Numerical simulations indicate that, for stars of solar metallicity, pro-
genitor masses below approximately 15M always give rise to a NS,
while for heavier progenitors the remnant can be either a NS or a BH,
depending on details of the explosion. As we saw above, the collapse
of stars with Mprog ∼ < (9 − 11)M is expected to produce NS remnants
with masses narrowly peaked around 1.36M . In contrast, because of
the role of fallback, the mass distribution of NS created in the collapse
of stars with Mprog > ∼ (9 − 11)M is broader, and can cover the range
[1.2 − 2.0]M .
Observationally, a few NS are known with a mass around 2M . One
is the millisecond pulsar PSR J1614-2230. This pulsar is a member of
 10.3 The dynamics of core collapse 33

a binary system. This allows a precise determination of its mass, which

turns out to be MNS = (1.97 ± 0.04) M .31 The formation mechanism 31
See Demorest, Pennucci, Ransom,
for systems such as this binary millisecond pulsar suggests that this NS Roberts and Hessels (2010). The mea-
sure is based on the observation of
can have accreted only O(10−2 )M from the companion. Thus, unless Shapiro time delay from the pulsar. As
different mechanisms of mass transfer from the companion can be found, we see from eq. (6.95) of Vol. 1, the
such a high value for a NS mass can only be explained by fallback at range r and shape s of the Shapiro time
birth. A second example is given by PSR J0348+0432, a pulsar with delay determine the mass of the com-
panion and the inclination of the orbit,
a spin-down period of 39 ms and a mass MNS = (2.01 ± 0.04) M .32 respectively. Furthermore, for a spec-
This pulsar is also a member of a binary system. Its companion is a troscopic binary the observation of the
very low-mass WD, with a mass M = 0.172 ± 0.003M , and the orbital Doppler shift of the lines of the com-
period is only 2.46 hr. This system is therefore particularly interesting panion (which in this case turns out to
be a WD with M = (0.500±0.006)M )
since it is very relativistic, and at the same time the large pulsar mass fixes the Keplerian mass function, i.e.
creates a strong gravitational field, so it can be used to test relativistic a combination of the masses of the
gravity in strong fields. two stars. Having fixed the compan-
Numerical simulations suggest that, while for Mprog ∼ < 15M all core ion mass with Shapiro time delay, one
can then obtain the pulsar mass.
collapses produce NS, for heavier progenitors there is no obvious corre- 32
lation between the mass of the progenitor and whether the star collapses See Antoniadis et al. (2013).
to a NS, to a BH by fallback or to a BH by direct collapse. This is basi-
cally due to the fact that the stellar structure varies non-monotonically
with the mass, and even small changes in the progenitor mass can give
rise to important differences in the explosion. Failed explosions with
BH formation have been obtained for progenitors below 20M , while
successful SNe with NS formation are found also between 20 and 40
M .33 However, independently of whether the remnant is a BH or a NS, 33
See Ugliano, Janka, Marek and Ar-
the remnant mass is an almost monotonic function of the mass enclosed cones (2012) and Janka (2012).
below the O–Si burning shell.

(e) (30 − 40)M ∼ < Mprog ∼< 95M . In this regime the main theoreti-
cal uncertainty comes from mass losses of the progenitors due to stel-
lar winds. Mass losses increase with the initial progenitor mass. For
Mprog >∼ (30 − 40)M the whole H envelope can be lost and the He core
is uncovered before the star collapses, leading to a Wolf–Rayet star.
In the WR stage the mass losses due to stellar winds are particularly
strong, and their theoretical modelization is quite uncertain. For suffi-
ciently large mass loss rate the star gives rise to a NS remnant, otherwise
the star collapses to a BH.

(f ) 95M ∼ < Mprog ∼ < 130M : pulsational pair-instability SNe. For these
extremely massive stars, and for low metallicities, a mechanism different
from iron core collapse can set in, and can lead to a “pair-instability
supernova”, which is an extremely luminous type of SN explosion. The
instability arises because, after burning He and C, if the resulting O
core is heavier than (40 − 50)M , one reaches a central temperature
Tc of order 109 K at relatively low density, and in these conditions it
is possible to efficiently create electron–positron pairs from high-energy
photons before the ignition of O. Part of the thermal energy of the star is
therefore drained and transformed into the rest mass of the e± pairs. As
a result the pressure drops and rapid contraction takes place. This leads
34 Stellar collapse

to a temperature increase and, since the rate of pair production increases

as Tc9 , we get a runaway collapse followed by a thermonuclear explosion
due to explosive ignition of O. This can eject several solar masses of
material but, for stars that at birth had a mass in the approximate
range (95 − 130)M , it does not unbind the star. The residual core
contracts and, on a time-scale ranging from a few days to a few decades,
it starts burning its nuclear fuel again. For progenitor masses in the
range (95 − 130)M , at the time of death the He core has a mass in
the range (40 − 60)M , if it has not lost much mass to stellar winds.
If, after this first pulse, the He core mass remains above about 40M ,
the star encounters the pair instability again. In this way the star can
go through a series of pulses, and one has a “pulsational pair-instability
SN”. Later pulses are more energetic than the first one, since the first
pulse lost part of its energy in expelling the H envelope (and possibly
part of the He envelope). Then, the ejecta of the second and later pulses
catch up with the first shell. In the resulting shock all the kinetic energy
is dissipated in radiation, since the collision takes place at large radii,
where adiabatic losses due to the expansion are negligible. This should
be compared with the classic iron-core collapse where, as we saw, only
about 1% of the kinetic energy is transformed into radiation. As a result,
one can estimate that ∼ 1050 erg can be radiated in light, about an order
of magnitude more than in a typical SN explosion. It is possible that
this mechanism explains superluminous SN events.
In this way a star can go through a series of SN explosions, until the
core starts burning Si to Fe, and in the end the core will undergo the
standard iron-core collapse, and the remnant will typically be a BH.
Observe that the pair-instability mechanism hinges on the possibility of
a He core with mass over 40M . Most probably this cannot be achieved
for stars of solar metallicity, because of mass loss to stellar winds. Thus,
pair-instability SNe are most probably restricted to low-metallicity stars.

(g) 130M ∼ < Mprog ∼

< 260M : pair-instability SNe. Stars with low metal-
licity, and particularly Pop III stars, can have masses in this range, since
for low Z these large masses can be preserved against stellar winds. For
progenitors masses approximately in the range (130 − 260)M (corre-
sponding to He cores between 60 and 137M ) the thermonuclear ex-
plosion that follows the pair-instability collapse is sufficiently energetic
to completely disrupt the star, leaving no remnant behind. This gives
rise to events with explosion energies in excess of 1053 erg, and the
production of more than 50M of 56 Ni, possibly the biggest stellar ex-
plosions in the Universe. These events have sometimes been referred to
as hypernovae, although nowadays they are more commonly called pair-
instability (or pair-capture) SNe. These events also emit a huge energy
in neutrinos, of order (0.01 − 0.03)M c2 ∼ 1055 erg. A number of ultra-
bright SNe explosions (such as SN 2006gy, whose light curve is shown
in Fig. 10.4) have been suggested as candidates for pair-instability SNe,
although other explanations might be possible for their extreme lumi-
nosities, such as the interaction of ejected material with a dense circum-
 10.4 GW production by self-gravitating fluids 35

stellar medium, or the extraction of extra energy from the spin-down of

a highly magnetized proto-NS. In our Galaxy, the star η-Carinae is a
potential candidate for a future pair-instability SN.
34
Pair-instability SNe also have pecu-
(h) Mprog > liar nucleosynthesis signatures, which
∼ 260M . Finally, for even heavier stars the nuclear burning should be reflected in the chemical com-
of the core is unable to halt the collapse induced by the pair instability position of some very old low-mass
and reverse it with a thermonuclear explosion. Thus the star collapses stars. At the moment there are not
directly to a BH, with a mass comparable to that of the progenitor many observations of very old low-
mass stars bearing such signatures.
star. If progenitors with these masses were an important component of
See, however, Aoki, Tominaga, Beers,
Pop III stars, they will have produced a population of BHs with masses Honda and Lee (2014) for one possible
of several hundred solar masses.34 case.

10.4 GW production by self-gravitating

fluids
In this section we will first of all set up the relevant equations for the
generation of GWs by a self-gravitating fluid, adapting to this case the
general formalism for GW production, developed in particular in Chap-
ters 3 and 5 of Vol. 1. Although we are developing it here in the context
of GWs from stellar collapse, this formalism is of more general interest.

10.4.1 Energy–momentum tensor of a perfect fluid

In general relativity (GR) a perfect fluid is defined by the condition
that, at every point in space-time, there is a locally inertial frame (i.e. a
frame where, in a sufficiently small region around the space-time point P
considered, the metric is flat), comoving with the fluid element (so that
in this frame the fluid velocity vanishes at the point P), where the fluid
is isotropic. In this frame the energy–momentum tensor of the fluid at
the point P must therefore satisfy T 0i = 0 and T ij ∝ δ ij , i.e.
T 00 = ρc2 , T 0i = 0 , T ij = pδ ij , (10.28)
so (in this frame) the energy density of the fluid is ρc2 and p is the
pressure. To find the form of T µν in a frame that is still locally in-
ertial, but not necessarily comoving with the fluid, we introduce the
four-velocity field uµ at the space-time point P, defined by requiring
that (1) in the comoving locally inertial frame of the fluid uµ takes the
value u0 = c, ui = 0; (2) it transforms as a four-vector under Lorentz
transformations of the locally inertial frame constructed at P.
Observe that the normalization condition on uµ can be written in a
Lorentz-invariant form as
ηµν uµ uν = −c2 . (10.29)
In terms of this four-vector the rest-frame expression for the energy–
momentum tensor (10.28) can be written compactly as
 p
T µν = ρ + 2 uµ uν + p η µν . (10.30)
c
36 Stellar collapse

If we now define ρ and p as the Lorentz-invariant quantities defined by

35
Note, however, that the physical en-
ergy density is −T00 , i.e. it is the (00)
component of a tensor with one up-
per and one lower index (this defini-
tion will carry over also to curved space;
see Section 18.3). The Lorentz scalar
ρ(x)c2 coincides with the physical en-
ergy density −T00 (x) at a point x only
in a frame where uµ (x) = (c, 0, 0, 0).
Otherwise, in a generic frame (but still
in flat space),
−T00 = ρ(u0 )2 + p[(u0 )2 /c2 − 1] .
(10.31)
their values in the comoving locally inertial frame, the right-hand side
of eq. (10.30) is a Lorentz tensor, and therefore eqs. (10.29) and (10.30)
give the generalization of the energy–momentum tensor of a perfect fluid
for a generic locally inertial frame (so in particular in flat space).35
Of course, uµ is a function of the space-time point P used in the
construction, so it is actually a four-vector field uµ (t, x). At the level of
the particles making up the fluid, uµ corresponds to the average, over
many particles in a small volume element, of the particle four-velocities,
defined as the derivative of the particle trajectory xµ0 (τ ) with respect to
proper time τ , rather than with respect to coordinate time. In fact, for
a particle on a trajectory xµ0 (τ ), proper time τ is defined by
c2 dτ 2 = −ηµν dxµ0 (τ )dxν0 (τ ) , (10.32)
and therefore
dxµ0 dxν0
ηµν = −c2 . (10.33)
dτ dτ
Comparison with eq. (10.29) shows that uµ is the average of dxµ0 /dτ .
The coordinate velocity of the particles is rather v i ≡ dxi0 /dt. Writing
v i = (dxi0 /dτ )(dτ /dt) and observing that u0 is the average of cdt/dτ ,
for the corresponding velocity fields we have

cui (t, x)
v i (t, x) = . (10.34)
u0 (t, x)
Using
−1/2
v2

dt
= 1− , (10.35)
dτ c2
we have −1/2
v 2 (t, x)

0
u (t, x) = c 1 − , (10.36)
c2
and, from eq. (10.34),
−1/2
v 2 (t, x)

ui (t, x) = 1 − v i (t, x) . (10.37)
c2

Thus, in terms of v i , we can rewrite eq. (10.30) as

v2
 
1
T 00 = ρc 2
+ p , (10.38)
1 − (v 2 /c2 ) c2
ρc2 + p v i
T 0i = , (10.39)
1 − (v 2 /c2 ) c
(ρc2 + p) v i v j
T ij = pδ ij + . (10.40)
1 − (v 2 /c2 ) c2

These expressions could also have been obtained directly from eq. (10.28)
by performing a Lorentz boost with velocity v i .
 10.4 GW production by self-gravitating fluids 37

For non-relativistic matter, the pressure p is O(ρv 2 ) and is therefore

suppressed with respect to ρc2 by a factor O(v 2 /c2 ), where v 2 is the
average of the square velocity of the particles in the fluid at the point
under consideration. Then, to leading order in the fluid velocity field
v/c, eqs. (10.38)–(10.40) reduce to36 36
Note that the condition p = O(ρv 2 )
valid for non-relativistic matter implies
T 00 = ρc2 1 + O(v 2 /c2 ) ,
 
(10.41) that, as far as the powers of v are con-
0i i 2 2 cerned, the terms pδ ij and ρv i v j in
 
T = ρcv 1 + O(v /c ) , (10.42)
eq. (10.43) are of the same order. In the
ij ij i j 2 2
 
T = (pδ + ρv v ) 1 + O(v /c ) . (10.43) rest frame of the fluid at the point P , v i
vanishes, while p is proportional to the
In curved space the generalization of eqs. (10.29) and (10.30) to a generic average of the square velocity, which is
frame, not necessarily locally inertial, is simply obtained by constructing a scalar, so in the rest frame at P only
an object that transforms as a tensor under general coordinate transfor- pδ ij survives, giving eq. (10.28), while
at a generic point such that hvi i2 ∼
mations and that, in a locally inertial frame where gµν = ηµν , reduces hv 2 i the two terms are of the same or-
to its flat-space counterpart. This is obtained by writing der. However, it is clear that the term
pδ ij cannot contribute to GW produc-
 p
T µν = ρ + 2 uµ uν + p g µν , (10.44) tion, since it is isotropic, and its contri-
c bution to hTT ij vanishes when we per-
form the projection onto the transverse-
with ρ and p now defined as the diffeomorphism-invariant quantities traceless part, as we will indeed check
explicitly in eq. (10.76). Therefore, the
equal to their values in the locally inertial frame. Similarly, uµ is now
leading contribution of T ij to GW pro-
defined as the four-vector under general coordinate transformation that, duction comes from the term ρv i v j .
in the comoving locally inertial frame, takes the value (u0 = c, ui = 0).
Thus, it now satisfies
gµν uµ uν = −c2 . (10.45)
Equations (10.44) and (10.45) provide the general expression for the
energy–momentum tensor of an ideal fluid in curved space.
If the fluid is not ideal, in the comoving locally inertial frame it will
still satisfy T 0i = 0, since this condition can be taken as the definition of
a comoving frame, i.e. of being at rest with respect to the fluid motion,
at point P. However, if the fluid is not isotropic, T ij will no longer be
proportional to δ ij . Thus, in the comoving locally inertial frame we now
have 37
The energy–momentum tensor for a
T 00 = ρc2 , T 0i = 0 , T ij = pδ ij + Σij , (10.46) fluid in curved space will be further
where Σij is a symmetric and traceless tensor (since any trace part can discussed in Part IV. Let us antici-
pate, from Section 18.3, that in curved
be reabsorbed into the definition of p). The tensor Σij is called the space the physical energy density will
anisotropic stress tensor, and describes the effect of viscosity. In a frame be given by −T00 (which, in a generic
that is still locally inertial but not necessarily comoving, eq. (10.30) metric, is of course not the same as
T 00 ). Let us also mention that, in a
generalizes to
cosmological setting, for a generic fluid
 p
T µν = ρ + 2 uµ uν + pη µν + Σµν , (10.47) in curved space we will rather use a dif-
c ferent convention, and we will define ρ
µν
where Σ is defined as the Lorentz tensor that, in the comoving frame, from −T00 = ρc2 in any frame (and not
just in the locally inertial frame comov-
has Σ00 = Σ0i = 0, while its space-space components have the values ing with the fluid element), so that ρ
Σij given in eq. (10.46). Similarly eq. (10.44) is generalized by adding to will be the component of a tensor with
the right-hand side the quantity Σµν , now defined as the tensor under an upper and a lower index, rather than
general coordinate transformations that reduces to Σij in the locally a scalar; cf. Note 10 on page 374. As we
will see in Section 18.3.2, in cosmolog-
inertial comoving frame, so that37 ical perturbation theory the difference
 p between these two definitions of ρ only
T µν = ρ + 2 uµ uν + pg µν + Σµν . (10.48) shows up at second order.
c
38 Stellar collapse

10.4.2 GW production from gravitating Newtonian

fluids

In principle, once we have T 00 (t, x) = ρ(t, x)c2 , the emission of GWs

to lowest order in v/c could be computed simply plugging it into the
quadrupole formula (3.59) of Vol. 1. This however requires the compu-
tation of the second time derivative of the quadrupole moment. When
ρ(t, x) is obtained from numerical simulations, as for SN explosions, this
requires replacing the derivative by finite differences, and introduces
significant numerical noise, particularly at high frequencies. To make
things worse, in the computation of the quadrupole moment the density
ρ is weighted with a factor r4 (a factor r2 coming from d3 x = r2 drdΩ
and a factor r2 coming from the term xi xj in the definition of the second
mass moment). Thus, a large contribution to the quadrupole moment
comes from the low-density outer layers of the star, where the low val-
ues of ρ is compensated by the large value of r4 . However, even if the
contribution of the outer layers to the quadrupole moment is large, their
contribution to its second time derivative is suppressed because the mo-
tion in the outer low-density layers is slower and less coherent, and in
fact the main contribution to the GW signal comes from the inner re-
gion. Thus, when taking the second time derivative, the contribution
of the external regions should undergo partial cancellations due to the
less coherent nature of the matter flow (compare with Section 4.3.2 of
Vol. 1, where we discussed GW radiation produced by coherent and by
incoherent motions). The numerical error in the evaluation of the sec-
ond time derivative can, however, spoil these delicate cancellations (even
more considering that in the outer region the numerical grid is coarser)
and introduces significant numerical artifacts.
A better procedure is to use the space-space components of the energy–
momentum tensor as a source for hTT ij . Recall indeed from eq. (3.26)
that, in the flat-space treatment of Chapter 3, hTT ij was originally given
by a spatial integral of the space-space components T ij , evaluated at re-
tarded time. Performing a multipole expansion of this expression, we got
eq. (3.34). Only upon repeated use of energy–momentum conservation
was this expression transformed into a second time derivative of T 00 .
In our case, to avoid numerical problems with the time derivatives, it is
clearly convenient not to perform these manipulations, and start directly
from the expression of hTT ij in terms of the space-space components of
the energy–momentum tensor.
However, a complication inherent to the use of the space-space com-
ponents is that to compute hTT ij for a gravitating system we cannot
just naively use the flat-space formulas of Chapter 3. In fact, from
eqs. (10.41) and (10.43) we see that T ij is suppressed with respect to
T 00 by a factor O(v 2 /c2 ). To get correctly all terms of the same or-
der we must therefore compute the (i, j) component of the source term
to order v 2 . To this order, the Newtonian potential U also contributes
since, by the virial theorem, U ∼ v 2 . We have studied these issues in full
generality in Chapter 5. However, in order to get the correct Newtonian
 10.4 GW production by self-gravitating fluids 39

result it is not necessary to go through the full formalism of Chapter 5.

We can simply recall, from Section 5.2, that the Einstein equations can
be recast in the exact form
16πG αβ
2hαβ = τ , (10.49)
c4
where 2 is the flat-space d’Alembertian, hαβ ≡ (−g)1/2 g αβ − η αβ , and
we have fixed the gauge ∂β hαβ = 0. The tensor τ αβ is given by

τ αβ = (−g)(T αβ + tαβ
LL ) + ∂µ ∂ν χ
αβµν
, (10.50)

where tαβ
LL is the Landau–Lifshitz energy–momentum pseudotensor, which
is quadratic in the derivatives of hαβ and is given in eq. (5.75). In the
last term, χαβµν = (c4 /16πG)(hαµ hβν − hµν hαβ ). As a consequence of
∂β hαβ = 0, τ αβ satisfies the exact conservation equation

∂µ τ µν = 0 , (10.51)

with an ordinary rather than covariant derivative; see eq. (5.82).

We now compute the lowest-order term in the post-Newtonian ex-
pansion of τ αβ in the near-source region. This can be done using the
Newtonian form of the metric perturbation,

h00 = −2φ , hij = −2φδij , h0i = 0 ; (10.52)

see eq. (5.11). Writing gµν = ηµν + hµν , in the limit of small hµν we
found in Section 5.2 that hαβ ' −hαβ + 12 η αβ h. Using the Newtonian
expression for hµν we find that the Newtonian expression for hµν is

h00 = 4φ , h0i = hij = 0 . (10.53)

We can now plug this expression for hαβ into the Landau–Lifshitz energy–
momentum pseudotensor given in eq. (5.75). To lowest order, the factors
gµν in terms such as gλµ g νρ ∂ν hαλ ∂ρ hβµ can simply be replaced by ηµν ,
and similarly the factors −g can be replaced by 1. Furthermore, we
recall that in the near region a time derivative brings one more power of
v/c than spatial derivatives, ∂0 ∼ (v/c)∂i . Therefore, at leading order
we can neglect all terms involving ∂0 and keep only spatial derivatives.
We express the result in terms of the Newtonian potential U (defined so
that U > 0), related to φ by U = −φc2 . Observe that U satisfies the
Poisson equation
∇2 U = −4πGρ ; (10.54)
see eqs. (5.17) and (5.21), with (0) T 00 = ρc2 . Then, using eq. (10.53), a
simple calculation gives χαβµν = 0 and
38
Observe that tij ij ij
 
ij 1 i j 1 ij k LL = −t , where t
tLL = ∂ U ∂ U − δ ∂k U ∂ U . (10.55) is the tensor defined in eq. (5.224). In
4πG 2 terms of tij
LL , eq. (5.223) reads

Note that tij

LL is of the same order as the space-space components T
ij
of d2 z i
Z
ma 2a = − dSj tij
LL .
38
the fluid energy–momentum tensor. In fact, if L is the typical length- dt Sa
40 Stellar collapse

scale of variation of U , we have

1 U
∂i U ∂j U ∼ U. (10.56)
G GL2
Since ∇2 U = −4πGρ, we have U/(GL2 ) ∼ ρ. Furthermore, because of
the virial theorem, U ∼ v 2 , and therefore
1
∂i U ∂j U ∼ ρv 2 . (10.57)
G
We can similarly verify that also t00 0i 2
LL and tLL are of order ρv . However,
00 0i
since T and T are of order ρ and ρcv, respectively, the contribution
of the Landau–Lifshitz pseudotensor to τ 00 and τ 0i is subleading. Thus,
to leading order,

τ 00 = ρc2 , (10.58)
τ 0i = ρcv i , (10.59)
 
1 1
τ ij = pδ ij + ρv i v j + ∂ i U ∂ j U − δ ij ∂k U ∂ k U . (10.60)
4πG 2
The result agrees with what we already found in eqs. (5.111)–(5.113),
39
Observe that V in eqs. (5.111)– where we also gave the result up to 1PN order for τ 00 and τ 0i .39
(5.113) is the potential given by the To understand the physical meaning of the U -dependent term in τ ij
retarded integral (5.39). However, to
lowest order in v/c it reduces to U , as
we can apply the conservation equations (10.51) to the Newtonian ex-
we can see observing that to lowest or- pression for τ ij given in eqs. (10.58)–(10.60). The equation ∂µ τ µν = 0
der σ = (1/c2 )(T 00 + T ii ) ' T 00 /c2 , with ν = 0 gives the continuity equation
and using the expansion (5.40) in the
retarded integral. ∂t ρ + ∂i (ρv i ) = 0 , (10.61)

while the equation with ν = i gives the Euler equation in a gravitational

potential,
ρ(∂t v i + v j ∂j v i ) = −∂ i p + ρ∂ i U , (10.62)
where we have used eq. (10.54). Thus, we see that the term in τ ij that
depends on U correctly gives the Newtonian force in the Euler equation.
The quadrupole radiation can now be computed from
16πG ij
2hij = τ , (10.63)
c4
The problem is now formally identical to that studied in Section 3.1 of
Vol. 1, with h̄µν replaced by hµν (which for hµν  1 is the same as
replacing hµν by −hµν , since in the weak-field limit hµν ' −h̄µν ) and
Tµν replaced by −τµν [because of the different relative sign between the
right-hand sides of eqs. (10.63) and (3.3), which indeed reflects the fact
that hµν reduces to −h̄µν in the weak-field limit]. Therefore eq. (3.34)
now becomes, for the term corresponding to quadrupole radiation,
1 4G
Z
hTT
ij (t, x) = Λ ij,kl (n̂) d3 x0 τ kl (t − r/c, x0 ) , (10.64)
r c4
where, as usual, r = |x|, n̂ = x/r is the unit vector in the direction (θ, φ),
and the Λ tensor is the projector that extracts the TT part, introduced
Another random document with
no related content on Scribd:
The Project Gutenberg eBook of L'immoraliste
This ebook is for the use of anyone anywhere in the United States
and most other parts of the world at no cost and with almost no
restrictions whatsoever. You may copy it, give it away or re-use it
under the terms of the Project Gutenberg License included with this
ebook or online at www.gutenberg.org. If you are not located in the
United States, you will have to check the laws of the country where
you are located before using this eBook.

Title: L'immoraliste

Author: André Gide

Release date: February 27, 2024 [eBook #73058]

Language: French

Original publication: Paris: Mercure de France, 1928

Credits: www.ebookgratuits.com and Laurent Vogel (This book was

produced from scanned images of public domain material
from the Google Books project.)

*** START OF THE PROJECT GUTENBERG EBOOK

L'IMMORALISTE ***
 ŒUVRES
DE

ANDRÉ GIDE
II

L’IMMORALISTE

PARIS
MERCVRE DE FRANCE
XXVI, RVE DE CONDÉ, XXVI

MCMXXVIII
 IL A ÉTÉ TIRÉ :

89 exemplaires sur vergé d’Arches numérotés à la presse de 1 à 89.

550 exemplaires sur vergé pur fil Lafuma numérotés de 90 à 639.

Tous droits réservés.

L’IMMORALISTE
Je te loue, ô mon Dieu ! de ce que tu m’as fait
créature si admirable.

Psaumes, C X X X I X , 14.
 PRÉFACE

Je donne ce livre pour ce qu’il vaut. C’est un fruit plein de cendre

amère ; il est pareil aux coloquintes du désert qui croissent aux
endroits calcinés et ne présentent à la soif qu’une plus atroce
brûlure, mais sur le sable d’or ne sont pas sans beauté.
Que si j’avais donné mon héros pour exemple, il faut convenir
que j’aurais bien mal réussi [1] ; les quelques rares qui voulurent bien
s’intéresser à l’aventure de Michel, ce fut pour le honnir de toute la
force de leur bonté. Je n’avais pas en vain orné de tant de vertus
Marceline ; on ne pardonnait pas à Michel de ne pas la préférer à
soi.
[1] Il a paru en juin 1902 une édition petit in-8o de ce
livre, tirée à 300 exemplaires sur vergé d’Arches.

Que si j’avais donné ce livre pour un acte d’accusation contre

Michel, je n’aurais guère réussi davantage, car nul ne me sut gré de
l’indignation qu’il ressentait contre mon héros ; cette indignation, il
semblait qu’on la ressentît malgré moi ; de Michel elle débordait sur
moi-même ; pour un peu, l’on voulait me confondre avec lui.
Mais je n’ai voulu faire en ce livre non plus acte d’accusation
qu’apologie, et me suis gardé de juger. Le public ne pardonne plus,
aujourd’hui, que l’auteur, après l’action qu’il peint, ne se déclare pas
pour ou contre ; bien plus, au cours même du drame on voudrait qu’il
prît parti, qu’il se prononçât nettement soit pour Alceste, soit pour
Philinte, pour Hamlet ou pour Ophélie, pour Faust ou pour
Marguerite, pour Adam ou pour Jéhovah. Je ne prétends pas,
certes, que la neutralité (j’allais dire : l’indécision) soit signe sûr d’un
grand esprit ; mais je crois que maints grands esprits ont beaucoup
répugné à… conclure — et que bien poser un problème n’est pas le
supposer d’avance résolu.
C’est à contre-cœur que j’emploie ici le mot « problème ». A vrai
dire, en art, il n’y a pas de problèmes — dont l’œuvre d’art ne soit la
suffisante solution.
Si par « problème » on entend « drame », dirai-je que celui que
ce livre raconte, pour se jouer en l’âme même de mon héros, n’en
est pas moins trop général pour rester circonscrit dans sa singulière
aventure. Je n’ai pas la prétention d’avoir inventé ce « problème » ; il
existait avant mon livre ; que Michel triomphe ou succombe, le
« problème » continue d’être, et l’auteur ne propose comme acquis ni
le triomphe, ni la défaite.
Que si quelques esprits distingués n’ont consenti de voir en ce
drame que l’exposé d’un cas bizarre, et en son héros qu’un malade ;
s’ils ont méconnu que quelques idées très pressantes et d’intérêt
très général peuvent cependant l’habiter — la faute n’en est pas à
ces idées ou à ce drame, mais à l’auteur, et j’entends : à sa
maladresse — encore qu’il ait mis dans ce livre toute sa passion,
toutes ses larmes et tout son soin. Mais l’intérêt réel d’une œuvre et
celui que le public d’un jour y porte, ce sont deux choses très
différentes. On peut sans trop de fatuité, je crois, préférer risquer de
n’intéresser point le premier jour, avec des choses intéressantes —
que passionner sans lendemain un public friand de fadaises.
Au demeurant, je n’ai cherché de rien prouver, mais de bien
peindre et d’éclairer bien ma peinture.
 A HENRI GHÉON
son franc camarade
A. G.
 (A Monsieur D. R., président du
conseil.)

Sidi b. M. 30 juillet 189.

Oui, tu le pensais bien : Michel nous a parlé, mon cher frère. Le

récit qu’il nous fit, le voici. Tu l’avais demandé ; je te l’avais promis ;
mais à l’instant de l’envoyer, j’hésite encore, et plus je le relis et plus
il me paraît affreux. Ah ! que vas-tu penser de notre ami ? D’ailleurs
qu’en pensé-je moi-même ? Le réprouverons-nous simplement, niant
qu’on puisse tourner à bien des facultés qui se manifestent
cruelles ? — Mais il en est plus d’un aujourd’hui, je le crains, qui
oserait en ce récit se reconnaître. Saura-t-on inventer l’emploi de
tant d’intelligence et de force — ou refuser à tout cela droit de cité ?
En quoi Michel peut-il servir l’État ? J’avoue que je l’ignore… Il lui
faut une occupation. La haute position que t’ont value tes grands
mérites, le pouvoir que tu tiens, permettront-ils de la trouver ? —
Hâte-toi. Michel est dévoué : il l’est encore ; il ne le sera bientôt plus
qu’à lui-même.

Je t’écris sous un azur parfait ; depuis les douze jours que Denis,
Daniel et moi sommes ici, pas un nuage, pas une diminution de
soleil. Michel dit que le ciel est pur depuis deux mois.
Je ne suis ni triste, ni gai ; l’air d’ici vous emplit d’une exaltation
très vague et vous fait connaître un état qui paraît aussi loin de la
gaîté que de la peine ; peut-être que c’est le bonheur.
Nous restons auprès de Michel ; nous ne voulons pas le quitter ;
tu comprendras pourquoi, si tu veux bien lire ces pages ; c’est donc
ici, dans sa demeure, que nous attendons ta réponse ; ne tarde pas.
 Tu sais quelle amitié de collège, forte déjà, mais chaque année
grandie, liait Michel à Denis, à Daniel, à moi. Entre nous quatre une
sorte de pacte fut conclu : au moindre appel de l’un devaient
répondre les trois autres. Quand donc je reçus de Michel ce
mystérieux cri d’alarme, je prévins aussitôt Daniel et Denis, et tous
trois, quittant tout, nous partîmes.
Nous n’avions pas revu Michel depuis trois ans. Il s’était marié,
avait emmené sa femme en voyage, et, lors de son dernier passage
à Paris, Denis était en Grèce, Daniel en Russie, moi retenu, tu le
sais, auprès de notre père malade. Nous n’étions pourtant pas
restés sans nouvelles ; mais celles que Silas et Will, qui l’avaient
revu, nous donnèrent, n’avaient pu que nous étonner. Un
changement se produisait en lui, que nous n’expliquions pas encore.
Ce n’était plus le puritain très docte de naguère, aux gestes
maladroits à force d’être convaincus, aux regards si clairs que
devant eux souvent nos trop libres propos s’arrêtèrent. C’était…
mais pourquoi t’indiquer déjà ce que son récit va te dire ?
Je t’adresse donc ce récit, tel que Denis, Daniel et moi
l’entendîmes. Michel le fit sur sa terrasse où près de lui nous étions
étendus dans l’ombre et dans la clarté des étoiles. A la fin du récit,
nous avons vu le jour se lever sur la plaine. La maison de Michel la
domine, ainsi que le village dont elle n’est distante que peu. Par la
chaleur, et toutes les moissons fauchées, cette plaine ressemble au
désert.
La maison de Michel, bien que pauvre et bizarre, est charmante.
L’hiver, on souffrirait du froid, car pas de vitres aux fenêtres ; ou
plutôt pas de fenêtres du tout, mais de vastes trous dans les murs. Il
fait si beau que nous couchons dehors sur des nattes.
Que je te dise encore que nous avions fait bon voyage. Nous
sommes arrivés ici le soir, exténués de chaleur, ivres de nouveauté,
nous étant arrêtés à peine à Alger, puis à Constantine. De
Constantine un nouveau train nous emmenait jusqu’à Sidi b. M. où
une carriole attendait. La route cesse loin du village. Celui-ci perche
au haut d’un roc comme certains bourgs de l’Ombrie. Nous
montâmes à pied ; deux mulets avaient pris nos valises. Quand on y
vient par ce chemin, la maison de Michel est la première du village.
Un jardin fermé de murs bas, ou plutôt un enclos l’entoure, où
croissent trois grenadiers déjetés et un superbe laurier-rose. Un
enfant kabyle était là, qui s’est enfui dès notre approche, escaladant
le mur sans façon.
Michel nous a reçus sans témoigner de joie ; très simple, il
semblait craindre toute manifestation de tendresse ; mais sur le
seuil, d’abord, il embrassa chacun de nous trois gravement.
Jusqu’à la nuit nous n’échangeâmes pas dix paroles. Un dîner
presque tout frugal était prêt dans un salon dont les somptueuses
décorations nous étonnèrent, mais que t’expliquera le récit de
Michel. Puis il nous servit le café qu’il prit soin de faire lui-même.
Puis nous montâmes sur la terrasse d’où la vue à l’infini s’étendait,
et, tous trois, pareils aux trois amis de Job, nous attendîmes,
admirant sur la plaine en feu le déclin brusque de la journée.
Quand ce fut la nuit, Michel dit :
 PREMIÈRE PARTIE

Mes chers amis, je vous savais fidèles. A mon appel vous êtes
accourus, tout comme j’eusse fait au vôtre. Pourtant voici trois ans
que vous ne m’aviez vu. Puisse votre amitié, qui résiste si bien à
l’absence, résister aussi bien au récit que je veux vous faire. Car si
je vous appelai brusquement, et vous fis voyager jusqu’à ma
demeure lointaine, c’est pour vous voir, uniquement, et pour que
vous puissiez m’entendre. Je ne veux pas d’autre secours que celui-
là : vous parler. Car je suis à tel point de ma vie que je ne peux plus
dépasser. Pourtant ce n’est pas lassitude. Mais je ne comprends
plus. J’ai besoin… J’ai besoin de parler, vous dis-je. Savoir se libérer
n’est rien ; l’ardu, c’est savoir être libre. — Souffrez que je parle de
moi ; je vais vous raconter ma vie, simplement, sans modestie et
sans orgueil, plus simplement que si je parlais à moi-même.
Écoutez-moi :

La dernière fois que nous nous vîmes, c’était, il m’en souvient,

aux environs d’Angers, dans la petite église de campagne où mon
mariage se célébrait. Le public était peu nombreux, et l’excellence
des amis faisait de cette cérémonie banale une cérémonie
touchante. Il me semblait que l’on était ému, et cela m’émouvait moi-
même. Dans la maison de celle qui devenait ma femme, un court
repas vous réunit à nous au sortir de l’église ; puis la voiture
commandée nous emmena, selon l’usage qui joint en nos esprits, à
l’idée d’un mariage, la vision d’un quai de départ.
Je connaissais très peu ma femme et pensais, sans en trop
souffrir, qu’elle ne me connaissait pas davantage. Je l’avais épousée
sans amour, beaucoup pour complaire à mon père, qui, mourant,
s’inquiétait de me laisser seul. J’aimais mon père tendrement ;
occupé par son agonie, je ne songeai, en ces tristes moments, qu’à
lui rendre sa fin plus douce ; et ainsi j’engageai ma vie sans savoir
ce que pouvait être la vie. Nos fiançailles au chevet du mourant
furent sans rires, mais non sans grave joie, tant la paix qu’en obtint
mon père fut grande. Si je n’aimais pas, dis-je, ma fiancée, du moins
n’avais-je jamais aimé d’autre femme. Cela suffisait à mes yeux pour
assurer notre bonheur ; et, m’ignorant encore moi-même, je crus me
donner tout à elle. Elle était orpheline, elle aussi, et vivait avec ses
deux frères. Marceline avait à peine vingt ans ; j’en avais quatre de
plus qu’elle.
J’ai dit que je ne l’aimais point ; du moins n’éprouvais-je pour elle
rien de ce qu’on appelle amour, mais je l’aimais, si l’on veut entendre
par là de la tendresse, une sorte de pitié, enfin une estime assez
grande. Elle était catholique et je suis protestant… mais je croyais
l’être si peu ! Le prêtre m’accepta ; moi j’acceptai le prêtre ; cela se
joua sans impair.
Mon père était, comme l’on dit, « athée » ; du moins je le
suppose, n’ayant, par une sorte d’invincible pudeur que je crois bien
qu’il partageait, jamais pu causer avec lui de ses croyances. Le
grave enseignement huguenot de ma mère s’était, avec sa belle
image, lentement effacé en mon cœur ; vous savez que je la perdis
jeune. Je ne soupçonnais pas encore combien cette première
morale d’enfant nous maîtrise, ni quels plis elle laisse à l’esprit.
Cette sorte d’austérité dont ma mère m’avait laissé le goût en m’en
inculquant les principes, je la reportai toute à l’étude. J’avais quinze
ans quand je perdis ma mère ; mon père s’occupa de moi, m’entoura
et mit sa passion à m’instruire. Je savais déjà bien le latin et le grec ;
avec lui j’appris vite l’hébreu, le sanscrit, et enfin le persan et,
l’arabe. Vers vingt ans, j’étais si chauffé qu’il osait m’associer à ses
travaux. Il s’amusait à me prétendre son égal et voulut m’en donner
la preuve. L’Essai sur les cultes phrygiens, qui parut sous son nom,
fut mon œuvre ; à peine l’avait-il revu ; rien jamais ne lui valut tant
d’éloges. Il fut ravi. Pour moi, j’étais confus de voir cette supercherie
réussir. Mais désormais je fus lancé. Les savants les plus érudits me
traitaient comme leur collègue. Je souris maintenant de tous les
honneurs qu’on me fit… Ainsi j’atteignis vingt-cinq ans, n’ayant
presque rien regardé que des ruines ou des livres, et ne connaissant
rien de la vie ; j’usais dans le travail une ferveur singulière. J’aimais
quelques amis (vous en fûtes), mais plutôt l’amitié qu’eux-mêmes ;
mon dévouement pour eux était grand, mais c’était besoin de
noblesse ; je chérissais en moi chaque beau sentiment. Au
demeurant, j’ignorais mes amis, comme je m’ignorais moi-même.
Pas un instant ne me survint l’idée que j’eusse pu mener une
existence différente ni qu’on pût vivre différemment.
A mon père et à moi des choses simples suffisaient ; nous
dépensions si peu tous deux, que j’atteignis mes vingt-cinq ans sans
savoir que nous étions riches. J’imaginais, sans y songer souvent,
que nous avions seulement de quoi vivre ; et j’avais pris, près de
mon père, des habitudes d’économie telles que je fus presque gêné
quand je compris que nous possédions beaucoup plus. J’étais à ce
point distrait de ces choses, que ce ne fut même pas après le décès
de mon père, dont j’étais unique héritier, que je pris conscience un
peu plus nette de ma fortune, mais seulement lors du contrat de mon
mariage, et pour m’apercevoir du même coup que Marceline ne
m’apportait presque rien.
Une autre chose que j’ignorais, plus importante encore peut-être,
c’est que j’étais d’une santé très délicate. Comment l’eussé-je su, ne
l’ayant pas mise à l’épreuve ? J’avais des rhumes de temps à autre,
et les soignais négligemment. La vie trop calme que je menais
m’affaiblissait et me préservait à la fois. Marceline, au contraire,
semblait robuste ; et qu’elle le fût plus que moi, c’est ce que nous
devions bientôt apprendre.

Le soir même de nos noces, nous couchions dans mon

appartement de Paris, où l’on nous avait préparé deux chambres.
Nous ne restâmes à Paris que le temps qu’il fallut pour
d’indispensables emplettes, puis gagnâmes Marseille, d’où nous
nous embarquâmes aussitôt pour Tunis.
Les soins urgents, l’étourdissement des derniers événements
trop rapides, l’indispensable émotion des noces venant sitôt après
celle plus réelle de mon deuil, tout cela m’avait épuisé. Ce ne fut que
sur le bateau que je pus sentir ma fatigue. Jusqu’alors chaque
occupation, en l’accroissant, m’en distrayait. Le loisir obligé du bord
me permettait enfin de réfléchir. C’était, me semblait-il, pour la
première fois.
Pour la première fois aussi, je consentais d’être privé longtemps
de mon travail. Je ne m’étais accordé jusqu’alors que de courtes
vacances. Un voyage en Espagne avec mon père, peu de temps
après la mort de ma mère, avait, il est vrai, duré plus d’un mois ; un
autre, en Allemagne, six semaines ; d’autres encore ; mais toujours
des voyages d’études ; mon père ne s’y distrayait point de ses
recherches très précises ; moi, sitôt que je ne l’y suivais plus, je
lisais. Et pourtant, à peine avions-nous quitté Marseille, divers
souvenirs de Grenade et de Séville se ravivèrent, de ciel plus pur,
d’ombres plus franches, de fêtes, de rires et de chants. Voilà ce que
nous allons retrouver, pensai-je. Je montai sur le pont du navire et
regardai Marseille s’écarter.
Puis, brusquement, je songeai que je délaissais un peu
Marceline.
Elle était assise à l’avant ; je m’approchai, et, pour la première
fois vraiment, la regardai.
Marceline était très jolie. Vous le savez ; vous l’avez vue. Je me
reprochai de ne m’en être pas d’abord aperçu. Je la connaissais trop
pour la voir avec nouveauté ; nos familles de tout temps étaient
liées ; je l’avais vue grandir ; j’étais habitué à sa grâce… Pour la
première fois je m’étonnai, tant cette grâce me parut grande.
Sur un simple chapeau de paille noire elle laissait flotter un grand
voile. Elle était blonde, mais ne paraissait pas délicate. Sa jupe et
son corsage pareils étaient faits d’un châle écossais que nous
avions choisi ensemble. Je n’avais pas voulu qu’elle s’assombrît de
mon deuil.
Elle sentit que je la regardais, se retourna vers moi… Jusqu’alors
je n’avais eu près d’elle qu’un empressement de commande ; je
remplaçais, tant bien que mal, l’amour par une sorte de galanterie
froide qui, je le voyais bien, l’importunait un peu ; Marceline sentit-
elle à cet instant que je la regardais pour la première fois d’une
manière différente ? A son tour, elle me regarda fixement ; puis, très
tendrement, me sourit. Sans parler, je m’assis près d’elle. J’avais
vécu pour moi ou du moins selon moi jusqu’alors ; je m’étais marié
sans imaginer en ma femme autre chose qu’un camarade, sans
songer bien précisément que, de notre union, ma vie pourrait être
changée. Je venais de comprendre enfin que là cessait le
monologue.
Nous étions tous deux seuls sur le pont. Elle tendit son front vers
moi ; je la pressai doucement contre moi ; elle leva les yeux ; je
l’embrassai sur les paupières, et sentis brusquement, à la faveur de
mon baiser, une sorte de pitié nouvelle ; elle m’emplit si violemment,
que je ne pus retenir mes larmes.
— Qu’as-tu donc ? me dit Marceline.
Nous commençâmes à parler. Ses propos charmants me
ravirent. Je m’étais fait, comme j’avais pu, quelques idées sur la
sottise des femmes. Près d’elle, ce soir-là, ce fut moi qui me parus
gauche et stupide.
Ainsi donc, celle à qui j’attachais ma vie avait sa vie propre et
réelle ! L’importance de cette pensée m’éveilla plusieurs fois cette
nuit ; plusieurs fois je me dressai sur ma couchette pour voir, sur
l’autre couchette plus bas, Marceline, ma femme, dormir.
 Le lendemain, le ciel était splendide ; la mer calme à peu près.
Quelques conversations point pressées diminuèrent encore notre
gêne. Le mariage vraiment commençait. Au matin du dernier jour
d’octobre, nous débarquâmes à Tunis.

Mon intention était de n’y rester que peu de jours. Je vous

confesserai ma sottise : rien dans ce pays neuf ne m’attirait que
Carthage et quelques ruines romaines : Timgat, dont Octave m’avait
parlé, les mosaïques de Sousse et surtout l’amphithéâtre d’El Djem,
où je me proposais de courir sans tarder. Il fallait d’abord gagner
Sousse, puis de Sousse prendre la voiture des postes ; je voulais
que rien d’ici là ne fût digne de m’occuper.
Pourtant Tunis me surprit fort. Au toucher de nouvelles
sensations s’émouvaient telles parties de moi, des facultés
endormies qui, n’ayant pas encore servi, avaient gardé toute leur
mystérieuse jeunesse. J’étais plus étonné, ahuri, qu’amusé, et ce qui
me plaisait surtout, c’était la joie de Marceline.
Ma fatigue cependant devenait chaque jour plus grande ; mais
j’eusse trouvé honteux d’y céder. Je toussais et sentais au haut de la
poitrine un trouble étrange. Nous allons vers le sud, pensai-je ; la
chaleur me remettra.
La diligence de Sfax quitte Sousse le soir à huit heures ; elle
traverse El Djem à une heure du matin. Nous avions retenu les
places du coupé. Je m’attendais à trouver une guimbarde
inconfortable ; nous étions au contraire assez commodément
installés. Mais le froid !… Par quelle puérile confiance en la douceur
d’air du Midi, légèrement vêtus tous deux, n’avions-nous emporté
qu’un châle ? Sitôt sortis de Sousse et de l’abri de ses collines, le
vent commença de souffler. Il faisait de grands bonds sur la plaine,
hurlait, sifflait, entrait par chaque fente des portières ; rien ne pouvait
en préserver. Nous arrivâmes tout transis, moi, de plus, exténué par
les cahots de la voiture, et par une horrible toux qui me secouait
encore plus. Quelle nuit ! — Arrivés à El Djem, pas d’auberge ; un
affreux bordj en tenait lieu : que faire ? La diligence repartait. Le
village était endormi ; dans la nuit qui paraissait immense, on
entrevoyait vaguement la masse informe des ruines ; des chiens
hurlaient. Nous rentrâmes dans une salle terreuse où deux lits
misérables étaient dressés. Marceline tremblait de froid, mais là du
moins le vent ne nous atteignait plus.
Le lendemain fut un jour morne. Nous fûmes surpris, en sortant,
de voir un ciel uniformément gris. Le vent soufflait toujours, mais
moins impétueusement que la veille. La diligence ne devait repasser
que le soir… Ce fut, vous dis-je, un jour lugubre. L’amphithéâtre, en
quelques instants parcouru, me déçut ; même il me parut laid, sous
ce ciel terne. Peut-être ma fatigue aidait-elle, augmentait-elle mon
ennui. Vers le milieu du jour, par désœuvrement, j’y revins,
cherchant en vain quelques inscriptions sur les pierres. Marceline, à
l’abri du vent, lisait un livre anglais qu’elle avait par bonheur
emporté. Je revins m’asseoir auprès d’elle.
— Quel triste jour ! Tu ne t’ennuies pas trop ? lui dis-je.
— Non, tu vois : je lis.
— Que sommes-nous venus faire ici ? Tu n’as pas froid, au
moins.
— Pas trop. Et toi ? C’est vrai ! tu es tout pâle.
— Non…
La nuit, le vent reprit sa force… Enfin la diligence arriva. Nous
repartîmes.
Dès les premiers cahots je me sentis brisé. Marceline, très
fatiguée, s’endormit vite sur mon épaule. Mais ma toux va la
réveiller, pensai-je, et doucement, doucement, me dégageant, je
l’inclinai vers la paroi de la voiture. Cependant je ne toussais plus,
non : je crachais ; c’était nouveau ; j’amenais cela sans effort ; cela
venait par petits coups, à intervalles réguliers ; c’était une sensation
si bizarre que d’abord je m’en amusai presque, mais je fus bien vite
écœuré par le goût inconnu que cela me laissait dans la bouche.
Mon mouchoir fut vite hors d’usage. Déjà j’en avais plein les doigts.
Vais-je réveiller Marceline ?… Heureusement je me souvins d’un
grand foulard qu’elle passait à sa ceinture. Je m’en emparai