112: Astronomy The Physics of Stars
Class 17 Notes: Core Collapse Supernovae
Our topic today is the post-main sequence evolution of massive stars, culminating in their
deaths via supernova explosion. Supernovae of this type are called core collapse supernovae,
to distinguish them from supernovae that occur due to accretion onto a white dwarf that
pushes it above the Chandresekhar limit.
Before beginning, I will warn you that we are now entering into an area of active research
where there are still very signicant uncertainties. What I will tell you is the best of our
understanding today, but signicant parts of it may well turn out to be wrong. I will try
to highlight the areas where what I say is least certain, and I will point out a couple of
places where statements asserted quite condently in your textbook have turned out to be
incorrect.
I. Post-main sequence evolution
A. Mass Loss
One important eect that distinguishes the evolution of massive stars from that
of lower mass stars is the importance of mass loss, both on the main sequence
and thereafter. Low mass stars do not experience signicant mass loss before the
AGB phase, but massive stars, as we have already seen, can lose mass while still
on the main sequence, and can lose even more after they leave it.
Like other aspects of stellar mass loss, the exact mechanisms are not understood.
Very massive stars, those above 85 M or so, lose mass in a rapid and unstable
manner. We have already encountered one star like this: Carinae. This is an
example of a type of star called a luminous blue variable, or LBV.
[Slide 1 Car]
As these processes reduce the stars mass, its atmosphere becomes less and less
dominated by hydrogen, eventually reaching X 0.1 or even less. We see these
stars are somewhat lower mass (but still very massive) stars whose atmospheres
are dominated by helium rather than hydrogen. These are called Wolf-Rayet
stars, and they are eectively the bare cores of massive stars. Stars from 10 85
M skip the LBV phase and go directly to the Wolf-Rayet phase.
Stars become WRs while they are still on the main sequence, i.e. burning hydrogen in their centers. Stars in this case are called WN stars, because they are
Wolf-Rayet stars that show large amounts of nitrogen on their surfaces. The nitrogen is the product of CNO cycle burning, which produces an equilibrium level
of nitrogen above the amount that the star began its life with.
WR stars continue to lose mass rapidly, often producing spectacular nebulae that
1
look like planetary nebulae. They shine for the same reason: the expelled gas is
exposed to the high energy radiation of the star, and it oresces in response.
[Slides 2, 3 WR nebula NGC 2359, WR 124]
Mass loss continues after the star exhausts H and begin burning He at this point
the surface composition changes and we begin to see signs of 3 burning. These
are WC stars. The continuing mass loss removes the enhanced nitrogen from the
CNO cycle, and convection brings to the surface the result of 3 burning, which
is mostly carbon. Very rarely, we see WR stars where the carbon is being blown
o, and the surface is dominated by oxygen.
The mass loss can be quite dramatic 100 M stars are thought to get down to
nearly 30 M by the time they evolve o the main sequence.
B. Movement on the HR diagram
While these stars show dramatic mass loss, their luminosities do not evolve all
that much as they age. That is for the reason we mentioned last time in the
context of low mass stars luminosity evolution: the role of radiation pressure.
The luminosity varies as L 4 4 , and is in turn given by the Eddington
quartic:
2
M
4 4 + 1 = 0.
0.003
M
For very massive stars, the rst term is dominant, so is roughly constant, and
L is too. This is simply a reection of the fact that very massive stars are largely
supported by radiation pressure. As a result, their luminosity is equal to the
Eddington luminosity, which depends only on total mass, not on composition.
[Slide 4 Meynet & Maeder tracks]
This non-evolution of the luminosity continues to apply even after these stars
leave the main sequence. As the stars develop inert ash cores and burning shells
like lower mass stars, they cannot increase in luminosity, but they can increase
in radius and go to lower eective temperature. The net eect is that they move
along nearly horizontal tracks on the HR diagram. The slide shows the latest
Geneva models.
As you can see, the luminosities increase less and less for stars of higher and higher
masses, and instead they evolve at constant luminosity. Thus massive stars never
have a red giant phase, since that would require an increase in luminosity.
C. Internal structure: the onion model
The internal structure of a massive star near the end of its lifetime comes to
resemble an onion. In the center is an ash core, with the type of ash depending
on the stars evolutionary state. At rst it is helium, then carbon, etc., until at
last the core is composed of iron. The temperature is high enough that the core
is never degenerate until the last stages in the stars life, when it consists of iron.
2
Above the core is a burning shell where the next lowest element Z element in the
burning chain burns. Thus above an iron core is a silicon burning layer, above
a helium core is a carbon burning layer, etc. Outside that shell, there is inert
material composed of the same element. As one moves further outward in the
star, one encounters the next burning shell where a lower Z element burns, and
so forth until one reaches the hydrogen burning layer and the hydrogen envelope,
if any is left, above it.
[Slides 5, 6 schematic representation and numerical computation from Heger et
al. of onion structure]
The onion structure grows until the star develops an iron core. Since iron is at
the peak of the binding energy curve, it cannot be further burned. A star with an
iron core and an onion structure around it is known as a supernova progenitor.
II. Supernovae
A. Evolution of the Core
Now consider what happens in the core of a supernova progenitor. The iron core
is much like the helium core that we discussed in the context of lower mass stars:
it has no nuclear reactions, so it becomes isothermal. If it gets to be more than
roughly 10% of the stellar mass, it will exceed the Schnberg-Chandrasekhar limit
o
and begin contracting dynamically. If it becomes degenerate, degeneracy pressure
can slow the collapse, but if the core exceeds the Chandrasekhar mass of 1.4 M ,
electron degeneracy pressure cannot hold it up. The core must therefore contract
once it exceeds a certain minimum mass.
Contraction creates two instabilities. First, at the high pressures found in the
core, heavy nuclei can undergo reactions of the form
I (A, Z ) + e J (A, Z 1) + e ,
i.e. nucleus I captures a free electron, which converts one of its protons into a
neutron. Well discuss why these reactions happen in a few moments. Reactions
of this sort create an instability because removing electrons reduces the number
of electrons, and thus the degeneracy pressure. The loss of pressure accelerates
collapse, raising the pressure again and driving the reaction to happen even faster.
Second, since the gas is degenerate, its pressure is unrelated to its temperature.
As it collapses, its temperature rises, but this does not halt the collapse because
it doesnt raise the pressure. Once the temperature exceeds about 6 7 109 K,
photons are able to start photodisintegrating iron via the reaction
56
Fe + 100 MeV 13 4 He + 4n.
As the equation indicates, the reaction is highly endothermic, absorbing about 100
MeV from the radiation eld, or about 2 MeV per nucleon, each time it happens.
3
In eect, all the energy that was released by burning from He to Fe is now given
back. The loss of thermal energy also accelerates collapse, which leads the core
to contract more, which accelerates the reaction, etc.
Both of these are ionization-like processes, and they serve to keep a < 4/3, in the
unstable regime where collapse cannot be halted. Once the collapse proceeds far
enough, an even more endothermic reaction can take place when photons begin
to disintegrate helium nuclei:
4
He + 27 MeV 2p + 2n.
This reaction absorbs 27 MeV each time it happens, or 6 7 MeV per nucleon.
The disintegration of He creates a population of free neutrons and protons. Normally free neutrons spontaneously decay into a proton plus an electron plus a
neutrino:
n p + e + e .
The free neutron lifetime is 614 seconds, and the reaction is exothermic (as it
must be, since it is spontaneous). The energy released can be determined just by
the dierence in mass between a proton, mp = 1.67262 1024 g, and a neutron,
mn = 1.67493 1024 g:
E = (mn mp )c2 = 1.3 MeV.
However, conditions in the core are very dierent from those in free space. The
electrons are highly relativistically degenerate. Consider what this means energetically. Back at the beginning of the class, we showed that, for a population of
degenerate electrons, they occupy all quantum states up to a maximum momentum
1/3
3h3 n
,
p0 =
8
where n is the number density of electrons. If a new electron were to be created
by the decay of a neutron, it would have to go into an unoccupied quantum state,
and the rst available state has a momentum a just above p0 . The corresponding
energy is
1 /3
3h3 n
E0 = pc =
c
8
in the limit where the electrons are highly relativistic. If we compare this to the
energy E that is released by neutron decay, we nd that E0 becomes equal to
E when
8 E 3
n=
= 9.6 1030 cm3
3
ch
If we have one electron per two nucleons, the average for elements heavier than
hydrogen, the corresponding mass density is
1
= nmp = 8.0 106 g cm3 .
2
4
Once the density exceeds this value, it is no longer energetically possible for free
neutrons to undergo spontaneous decay. Instead, the opposite is true, and the
reverse reaction
p + e n + e
begins to occur spontaneously. Each such reaction requires 1.3 MeV of energy, and
even further reduces the degeneracy pressure of the electrons. Electron capture
by heavy elements earlier on in the collapse occurs for very similar reasons.
The collapse is only halted once another source of pressure becomes available:
at suciently high density, the neutrons become degenerate. The structure of
this degenerate neutron matter is not well understood, and is a subject of active
research, but the bottom line of what we understand seems to be that the collapse
is halted once the density reaches around 1015 g cm3 . The radius of the core
at this point is about 40 km, although as the neutron star cools o it eventually
shrinks to about 10 km. The density of 1015 g cm3 in the core is roughly the
density of an atomic nucleus, so the core at this point is a giant atomic nucleus,
several km is diameter, with the mass of the Sun.
B. Explosion Mechanism and Energy Budget
All of these processes occur in the core on dynamical timescales. The initial
iron core is of order 5, 000 10, 000 km in radius, and the mass is of order a
Chandrasekhar mass, about 1.5 M , so the dynamical time is
1
1 second.
tdyn
G
Thus the core collapses on a timescale that is tiny compared to the dynamical
time of the star as a whole the outer envelope of the star just sits there while
the core collapses.
The collapse of the iron core causes the material above it to begin falling, and
the exact sequence of events thereafter is somewhat unclear. Your book gives the
impression that this is a solved problem, but your book is wrong on this point.
Exactly how supernovae work is far from clear. Nonetheless, we can give a rough
outline and make some general statements.
First of all, we can gure out the energy budget. Ultimately what drives everything is the release of gravitational potential energy by the collapse of the iron
core. It is this sudden energy release that explodes the star. The core has an initial mass of Mc 1.5 M , and an initial radius Rc 104 km. The nal neutron
core has a comparable mass and a radius of Rnc 20 km. Thus the amount of
energy released is
2
Egrav GMc
1
1
Rc Rnc
2
GMc
3 1053 erg.
Rnc
Of this, the amount that is used to convert the protons and electrons to neutrons
is a small fraction. Each conversion ultimately uses up about 7 MeV, so the total
5
nuclear energy absorption is
Enuc = 7 MeV
Mc
Egrav
2 1052 erg
.
mH
15
Thus only 10% percent of the energy is used up in converting protons to
neutrons. The rest is available to power an explosion.
Similarly, some of the energy is required to eject the stellar envelope. The binding
energy of the envelope to the core is roughly
Ebind =
Egrav
GMc (M Mc )
5 1051 erg
.
Rc
60
Thus only a few percent of the available energy is required to unbind the envelope.
The remaining energy is available to give the envelope a large velocity, to produce
radiation, and to drive nuclear reactions in the envelope. We dont have a good
rst-principles theory capable of telling us how this energy is divided up, but we
can infer from observations.
The observed speed of the ejecta is around 10,000 km s1 , so the energy required
to power this is
Egrav
1
.
Ekin = (M Mc )v 2 1051 erg
2
300
Finally, the observed amount that is released as light is comparable to that released in kinetic energy:
Erad 1051 erg
Egrav
.
300
Both of these constitute only about 1% of the total power.
So where does the rest of the energy go? The answer is that it is radiated away
too, but as neutrinos rather than photons. The neutrinos produces when the
protons in the core are converted into neutrons dont escape immediately, but
they do eventually escape, and they carry away the great majority of the energy
with them. Only about 1% is used to power everything else, but that is more
than enough to blow up the star.
Getting that 1% number from rst principles, and more generally understanding
the mechanism by which the energy released in the core is transferred into the
envelope of the star, is one of the major problems in astrophysics today. We have
a general outline of what must happen, but really solving the problem is at the
forefront of numerical simulation science.
Heres what we know in general outline: as long as the collapsing core has a pressure set by relativistic electrons, its adiabatic index is a = 4/3. As it approaches
nuclear density and more of the electrons convert to neutrons, it initially experiences an attractive nuclear force that pulls it together, and this has the eect of
6
pushing a even lower, toward 1, and accelerating the collapse. Once the densities get even higher, though, the strong nuclear force becomes repulsive, and a
increases to a value
4/3.
This is sucient to halt collapse of the core, and from the perspective of the
material falling on top of it, it is as if the core suddenly converted from pressureless
foam (a < 4/3) to hard rubber (a > 4/3). The infall is therefore halted suddenly,
and all the kinetic energy of the infalling material is converted to thermal energy.
This thermal energy raises the pressure, which then causes the material above the
neutron core to re-expand it bounces. The bounce launches a shock wave out
into the envelope.
The bounce by itself does not appear to be sucient to explode the star. The
shock wave launched by the bounce stalls out before it reaches the stellar surface.
However, at the same time all of this is going on, the core is radiating neutrinos
like crazy. Every proton that is converted into a neutron leads to emission of a
neutrino, and the collapsing star is suciently dense the that neutrinos cannot
escape. Instead, they deposit their energy inside the star above the core, further
heating the material there and raising its pressure.
The neutrinos are thought to somehow re-energize the explosion and allow it to
nally break out of the star. Heres an example of a computer simulation where
this happens:
[Slide 7 Adam Burrows supernova simulation]
However, there are lots of details missing, and it turns out to be much easier to
explode stars in 2D (which this simulation uses) than in 3D. The problem is not
solved.
C. Nucleosynthesis
The shock propagating outward through the star from the core heats the gas up
to 5 109 K, and this is hot enough to induce nuclear burning in the envelope.
This burning changes the chemical composition of the envelope, creating new
elements. Much of the material is heated up enough that it burns to the iron
peak, converting yet more of the star into iron-like elements.
I say iron-like because the initial product is not in fact iron. The reason is that
the most bound element, 56 Fe, consists of 26 protons and 28 neutrons, so it has
two more neutrons than protons. The fuel, consisting mostly of elements like 4 He,
12
C, 16 O, 28 Si, all have equal numbers of protons and neutrons. Thus there are
not enough neutrons around to pair up with all the protons to make 56 Fe.
Converting protons to neutrons is via decays is possible, and in fact it is the
rst step in the pp-chain. However, as we learned studying that reaction, decays
are slow, and in the few seconds that it takes for the shock to propagate through
the star, there is not enough time for them to occur.
7
The net result is that the material burns to as close to the iron peak as it can get
given the ratio of protons to neutrons available. This turns out to be 56 Ni. This
is not a stable nucleus, since it is subject to decay, but the timescale for decay
is much longer than the supernova explosion goes on for, so no beta decays occur
until long after the nucleosynthetic process is over.
Not all the material in the star is burned to the iron peak. As the shock wave
propagates through the star it slows down and heats things up less. The net
results is that material further out in the star gets less burned, so the supernova
winds up ejecting a large amounts of other elements as well. Calculating the exact
yields from rst principles is one of the goals of supernova models.
D. Observations
1. Light Curves
When a supernova goes o, what do we observe from the outside? The rst
thing, which was only seen for the rst time a couple of years ago, is a bright
ultraviolet ash from the shock breaking out of the stellar surface. We saw
this because Alicia Soderberg, a postdoc at Princeton (as was I at the time)
got very lucky. She was using an x-ray telescope to study an older supernova
in a galaxy, when she saw another one go o. The telescope was observing
the star as it exploded, and it saw a ash of x-rays as the shock wave from
the deep interior of the star reached the surface.
[Slide 8 Soderberg image]
After the initial ash in x-rays, it takes a little while before the optical emission reaches its peak brightness. That is because the expanding material
initially has a small area, and most of that emission is at wavelengths shortward of visible. As the material expands and cools, its optical luminosity
increases, and reaches its peak a few weeks after the explosion. After that
it decays. The decay can initially take one of two forms, called linear or
plateau, but after a while they all converge to the same slope of luminosity
versus time.
[Slide 9 light curve image]
This slope can be understood quite simply from nuclear physics. As we
mentioned a moment ago, the supernova synthesizes large amounts of 56 Ni.
This nickel is unstable, and it undergoes the decay reaction
56
28 Ni
56
27 Co
+ e+ + e +
with a half-life of 6.1 days. This is short enough that most of the nickel decays
during the initial period of brightening or shortly thereafter.
However, the resulting
reaction:
56
Co is also unstable, and it too undergoes a decay
56
27 Co
8
56
26 Fe
+ e+ + e + .
This reaction has a half-life of 77.7 days, and it turns out to be the dominant
source of energy for the supernova in the period from a few tens to a few
hundreds of days after peak. The expanding material is cooling o, and this
would cause the luminosity to drop, but the radioactive decays provide a
energy source that keeps the material hot and emitting.
By computing the rate of energy release as a function of time via the decay
of cobalt-56, we can gure out how the luminosity of the supernova should
change as a function of time. Radioactive decays are a statistical process, in
which during a given interval of time there is a xed probability that each
atom will decay. This implies that the number of cobalt-56 decays per unit
time that occur in a particular supernova remnant must be proportional to
the number of cobalt-56 atoms present:
dN
= N.
dt
Here N is the number of cobalt-56 atoms present and is a constant. The
equation simply asserts that the rate of change of the number of cobalt-56
atoms at any given time is proportional to the number of atoms present at
that time.
This equation is easy to integrate by separation of variables:
dN
= dt
N
N = N0 et ,
=
where N0 is the number of atoms present at time t = 0. The quantity is
known as the decay rate. To see how it is related to the half-life 1/2 , we can
just plug in t = 1/2 :
1
N0 = N0 e1/2
2
For
56
=
=
ln 2
.
1 /2
Co, = 0.0039 / day.
While radioactive decay is the dominant energy source, the luminosity is
simply proportional to the rate of energy release by radioactive decay, which
in turn is proportional to the number of atoms present at any time, i.e. L N .
This means that the instantaneous luminosity should follow
L et
=
log L = (log e)t + constant.
Thus for the cobalt-56-powered part of the decay, a plot of log L versus time
should be a straight line with a slope of
(log e) = 0.0017 day1 .
An excellent test for this model was provided by supernova 1987A, which went
of in 1987 in the Large Magellanic Cloud, a nearby galaxy. The supernova
9
was observed for more than ve years after the explosion, and as a result we
got a very good measure of how its luminosity dropped. We can see a clear
period when the slope follows exactly what we have just calculated. Once
enough of the 56 Co decayed, other radioactive decays with longer half-lives
took over.
[Slide 10 light curve of SN1987A]
The eect is even more prominent in type Ia supernovae, which are produced
when a white dwarf is pushed over the Chandrasekhar limit. In that case the
nuclear reaction burns a much larger fraction of the star to 56 Ni, so its decay
into cobalt and then iron completely dominates the light curve.
2. Neutrinos
Supernova 1987A also provided strong evidence for another basic idea in supernova theory: that supernovae involve the neutronization of large amounts
of matter, and with it the production of copious neutrino emission. The rst
detection of supernova 1987A was not its light. The shock wave takes some
time to propagate through the star after the core collapses. The neutrinos,
however, escape promptly, and on February 23, 1987 the Kamiokande II neutrino detector in Japan and the IMB detector in Ohio both measured a burst
of neutrinos that arrived more than three hours before the rst detection of
visible light from the supernova. Burst is perhaps too strong a word, since
the total number of neutrinos detected was 20 neutrinos are hard to measure! Nonetheless, this was vastly above the noise level, and provided the
rst direct evidence that a supernova explosion involves release of neutrinos.
3. Historical importance
A brief aside: because of their brightness and the long duration for which they
are visible, supernovae played an important part in the early development of
astronomy, and in the history of science in general. In November of 1572,
a supernova went o that was, at its peak, comparable in brightness to the
planet Venus. For about two weeks the supernova was visible even during the
day. It remained visible to the naked eye until 1574.
The 1572 supernova was so bright that no one could have missed it. One of
the people to observe it was the Dane Tycho Brahe, who said On the 11th
day of November in the evening after sunset, I was contemplating the stars in
a clear sky. I noticed that a new and unusual star, surpassing the other stars
in brilliancy, was shining almost directly above my head; and since I had, from
boyhood, known all the stars of the heavens perfectly, it was quite evident
to me that there had never been any star in that place of the sky, even the
smallest, to say nothing of a star so conspicuous and bright as this. I wqs so
astonished of this sight that I was not ashamed to doubt the trustworthyness
of my own eyes. But when I observed that others, on having the place pointed
out to them, could see that there was really a star there, I had no further
10
doubts. A miracle indeed, one that has never been prevoiously seen before
our time, in any age since the beginning of the world.
[Slide 11 plate from Tychos Stella Nova]
Tycho was so impressed by the event that he wrote a book about it and
decided to devote his life to astronomy. He went on to make the observations
that were the basis of Keplers Laws. Kepler himself saw another supernova
in 1604. The supernovae played a critical role in the history of science because
they provided clear falsication of the idea that the stars were eternal and
unchanging, which had dominated Western scientic thought since the time
of the ancient Greeks. Previous variable events in the sky, such as comets,
were taken to be atmospheric phenomena, and there was no easy way to
disprove this. With the supernovae, however, they persisted long enough to
make parallax observations possible. The failure to detect a parallax for the
supernovae provide without a doubt that they were further away than the
moon, in the supposedly eternal and unchanging realm outside the terrestrial
sphere.
Unfortunately for us, Tychos supernova was the last one to go o in our
galaxy (unless one went o on the far side of the galactic center, where we
wouldnt be able to see it due to obscuring dust). A number of astronomers
would very much like there to be another one, since astronomical instrumentation has improved a bit since Tychos day...
III. Supernova Remnants
The material ejected into space by a supernova into space slams into the interstellar
medium, the gas between the stars, at a velocity up from a few to ten percent of the
speed of light. When this collision happens, it creates a shock in the interstellar medium
that heats interstellar gas to temperatures of millions of K. The shocked bubble lled
with hot gas is known as a supernova remnant, and such remnants can be visible for
many thousands of years after the supernova itself fades from view.
The association of these structures with supernovae can be demonstrated quite clearly
by looking with modern telescopes at the locations of historical supernovae. For example, remnants have been identied for both Tychos and Keplers supernovae, and
another for the Crab supernova (named after the constellation where it is located).
The Crab supernova was recorded in 1054 by Chinese astronomers no one in Europe
at the time was paying attention to the sky, or if they were, they didnt bother to write
it down.
[Slides 12 - 14 Tychos SNR, Keplers SNR, and the Crab SNR]
We can understand the structure of a supernova remnant using a simple mathematical
argument made independently by L. I. Sedov in the USSR and G. I. Taylor in the
UK. These authors discovered the solution independently because Taylor discovered it
while working in secret on the British atomic bomb project, which was later merged
11
with the American one. It turns out that the problems of a supernova exploding in the
interstellar medium and a nuclear bomb exploding in the atmosphere are quite similar
physically. Sedov published his solution in 1946, just after the end of Wolrd War II,
while Taylors work was still secret.
Consider an idealized version of the supernova problem. An explosion occurs at a point,
releasing an energy E . The explosion occurs inside a medium of constant density ,
and we assume that the energy of the explosion is so large that the pressure it exerts is
vastly greater than the pressure in the ambient material, so that the ambient gas can
be assumed to be pressureless. This is a very good approximation for both supernovae
and nuclear bombs. We would like to solve for the position of the shock front r as a
function of time t.
The mathematical argument used to solve this relies on nothing more than fancy
dimensional analysis. Consider the units of the given quantities. We have the energy
E , density , radius r, and time t, which have units as follows:
[r]
[t]
[ ]
[E ]
=
=
=
=
L
T
M L3
M L2 T 2 .
Here L means units of length, T means units of time, and M means units of mass.
Thus a density is a mass per unit volume, which is a mass per length cubed. Energy
has units of ergs of Joules, which is a mass times an acceleration times a distance, and
acceleration is distance per time squared.
We want to have a formula for r in terms t, , and E . It is clear, however, that there
is only one way to put together t, , and E such that the nal answer has the units of
length! The mass must cancel out of the problem, so clearly the solution must involve
E/. This has units
E
= L5 T 2 .
We want to obtain something with units of length, so clearly the next step is to cancel
out the T 2 by multiplying by t2 . This gives
E2
t = L5 .
Finally, to get something with units of L and not L5 , we must take the 1/5 power.
Thus, the radius of the shock as a function of time must, on dimensional grounds, be
given by
1/5
E
r=Q
t2/5 ,
where Q is a dimensionless constant. Similarly, the shock velocity as a function of time
12
must follow
dr
2
E
v=
=Q
dt
5
1/5
t3/5 .
Actually solving the equations of uid dynamics shows that
Q=
75
16
(a 1)(a + 1)2
3a 1
1/5
,
where a is the adiabatic index of the gas into which the shock propagates.
Taylor used this solution to deduce the energy of the rst atomic explosion at Trinity
using nothing but photos of the blast wave at dierent times that has been published
in newspapers and magazines. When he published the result in 1950, a number of
people were not happy.
For supernovae we generally cant see them expand the expansion takes too long.
However, we can obtain a relationship we can test between the temperature of the
shocked material and the radius of the remnant. At the shock the kinetic energy of
the expanding gas is converted into heat, so the temperature at the shock, which is a
measure of internal energy per unit mass, is simply proportional to the kinetic energy
per unit mass, which varies as v 2 . Thus we have
2
Tshock v
E
2/5
t6/5 .
Now let us rewrite this in terms of the radius. Solving the rst equation for t, we have
t
E
1/2
r 5/2 ,
and plugging this into our equation for the shock temperature, we have
Tshock
E 3
r.
Thus the temperature of supernova remnants should decrease as the third power of
their size, assuming roughly constant energy and ISM density. Small remnants such as
Keplers, Tychos, and the Crab are visible in x-rays, but the rapid temperature drop
with size ensures that, once they expand signicantly, they cool o too much to be
visible in x-rays.
13