Black Holes and String Theory

Hussain Ali Termezy

Submitted in partial fulfilment of the requirements

for the degree of Master of Science of Imperial College London

September 2012
Contents

1 Black Holes in General Relativity 2

1.1 Black Hole Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2 Black Hole Thermodynamics . . . . . . . . . . . . . . . . . . . . . . . . . . 5

2 String Theory Background 19

2.1 Strings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.2 Supergravity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

3 Type IIB and Dp-brane solutions 25

4 Black Holes in String Theory 33

4.1 Entropy Counting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
Introduction
The study of black holes has been an intense area of research for many decades now, as
they are a very useful theoretical construct where theories of quantum gravity become
relevant. There are many curiosities associated with black holes, and the resolution of
some of the more pertinent problems seem to require a quantum theory of gravity to
resolve. With the advent of string theory, which purports to be a unified quantum theory
of gravity, attention has naturally turned to these questions, and have remarkably shown
signs of progress. In this project we will first review black hole solutions in GR, and then
look at how a thermodynamic description of black holes is made possible. We then turn
to introduce string theory and in particular review the black Dp-brane solutions of type
IIB supergravity. Lastly we see how to compute a microscopic account of the Bekenstein
entropy is given in string theory.

1
Chapter 1

Black Holes in General Relativity

1.1 Black Hole Solutions

We begin by reviewing some the basics of black holes as they arise in the study of general
relativity. Let us first consider the Schwarzschild solution where we encounter our simplest
example.
Schwarzschild Solution
The Schwarzschild metric is, by Birkhoff’s theorem, the unique, spherically symmetric
solution to the vacuum Einstein equations, Rµν = 0, with mass distribution M , given in
coordinates (t, r, θ, φ)

2GN M −1 2
   
2 2GN M 2
ds = − 1 − dt + 1 − dr + r2 dΩ2 (1.1)
r r

where dΩ2 = dθ2 + sin2 θdφ2 . This metric is valid outside and up to the surface of the
spherically symmetric source (e.g. star); is asymptotically flat, i.e. is Minkowski in the
limit r → ∞; and admits the timelike Killing vector k µ = (1, 0, 0, 0), i.e. is stationary (in
fact static).
The metric has two singular points (i.e. at which det gµν = 0 or parts of the metric
blow up): one at r = 0 and the other at r = 2GN M . Singularities can be due to a
failure of one’s coordinate system to cover all points on the manifold and are removable
by changing system, but can be also feature of the spacetime itself. In the latter case, this
means that there will be geodesics which cannot be continued through all values of their
affine parameter, hence it is said to suffer from geodesic incompleteness. We can test for
such singularities by building scalars from the curvature tensor, which if blow up in the
limit of the singularity, imply geodesic incompleteness of the spacetime. (However, the
implication doesn’t run the other way).
Let us characterise the Schwarzschild singularities. It can be shown that

48GN M 2
Rµνρσ Rµνρσ = . (1.2)
r6
This is the Kretschmann scalar which blows up as r → 0, hence we conclude that r = 0 is a
spacetime singularity. However, it turns out that r = 2GN M doesn’t react similarly with

2
any such scalar. This can be seen by changing coordinates to those of Kruskal-Szekeres
 1/2  
r t
u= −1 er/4GN M cosh (1.3)
2GN M 4GN M
 1/2  
r r/4GN M t
v= −1 e sinh (1.4)
2GN M 4GN M
so that the metric takes the form

32G3N M 3 −r/2m
ds2 = e (−dv 2 + du2 ) + r2 dΩ2 (1.5)
r
where r is given implicitly by
 
2 2 r
u −v = − 1 er/2m . (1.6)
2GN M
This is the maximal analytic extension to the Schwarzschild metric. In particular, the
metric now covers the region between r = 2GN M and r = 0, and we see that the surface
r = 2GN M , (now at v = ±u), is no longer singular, though far from unremarkable.
A handy perspective of spacetime is offered by a Carter-Penrose or conformal diagram,
since the diagram is obtained by conformally transforming the metric, i.e. gµν → gµν 0 =
2 µ
Ω gµν , Ω = Ω(x ). The diagram’s significance lies in ‘compacting’ an otherwise infinite
spacetime picture into a finite one whilst maintaining its causal structure. The CP diagram
for Schwarzschild is

Figure 1.1: (eternal) Schwarzschild space-time: red (black) lines are constant in r (t), the
central diagonals are event horizons (r = 2GN M ), and the top and bottom are singular-
ities. Only (t, r) plane drawn, so each point is a two-sphere. BH and WH mean black
hole and white hole. =+ , =− mean future and past null infinity respectively, and i+ , i−
mean future and past timelike infinity whilst i0 is spatial infinity. Labels are left-right
symmetric.

Interpretation of the diagram and event horizon- example of null hypersurface. Cosmic
censor
Reissner-Nordström Solution
The next black hole solution we need to consider is the Reissner-Nordström solution, which
is also static and incorporates charge. It is a solution to Einstein-Maxwell (EM coupled

3
to GR) action and will play a key role when we come to address string theory. The metric
is given by
−1
Q2 Q2
  
2 2GN M 2 2GN M
ds = − 1 − + 2 dt + 1 − + 2 dr2 + r2 dΩ2 (1.7)
r r r r
p
The solutions for which the metric is singular are r = r± = GN M ± G2 M 2 − Q2 . We
will be interested in the special case of the extremal black hole given by |Q| = GN M , and
the two horizons coincide. This will link to a similar relation occuring in supersymmetry:
the saturation of the BPS bound. Rewriting this using the coordinate r = ρ + Q gives1

Q −2 2 Q 2 2
   
2
ds = − 1 + dt + 1 + dρ + ρ2 dΩ2 (1.8)
ρ ρ

in Cartesian-like coordinates, ρ = |x̄3 |, x̄3 = (x, y, z), dx̄23 = dρ2 + ρ2 dΩ2 we can write this
as
ds2 = −H −2 dt2 + H 2 dx̄23 (1.9)
where H = 1 + |x̄Q3 | . The function, H, is harmonic in the three-dimensional Euclidean
space and belongs to the Majumdar-Papapetrou class of solutions which solve the Einstein
equations
N
X Qi
H =1+ (1.10)
|x̄3 − x̄3,i |
i=1

These are in fact supersymmetric solutions (c.f. brane solutions later) where the gravi-
tational attraction is cancelled by electrostatic repulsion. Going back to (1.8) the near
horizon geometry can be examined close to ρ = 0, this becomes

ρ2 2 Q2 2
ds2 = − dt + 2 dρ + Q2 dΩ2 (1.11)
Q2 ρ

writing z = Q ln ρ we have an AdS2 × S 2 geometry

ds2 = −e−2z dt2 + dz 2 + Q2 dΩ2 (1.12)

ADM mass
In d-dimensions, for gµν = ηµν + hµν
(d)
16πGN M
h00 ' (1.13)
(d − 2)ωd−2 rd−3

n+1
(d − 2)ωd−2
Z
n 2π 2
M= , ωn = dω = (1.14)
Γ n+1
(d)


16πGN Sn 2

4
 Law Thermodynamics Black Holes
T constant throughtout body κ constant over horizon of stationary black
Zeroth
in thermal equlilbrium hole
κ
First dE = T dS + work terms dM = 8πG N
dA + ΩH dJ + ΦH dQ
Second δS ≥ 0 in any process δA ≥ 0 in any process
Impossible to achieve T = 0 Impossible to achieve κ = 0 by a physical
Third
by a physical process process

1.2 Black Hole Thermodynamics

The four laws of black hole mechanics are, using [8]: Therefore there is an explicit analogy
suggesting that black holes may in fact be thermodynamic objects: S ↔ A/8πα, E ↔ M ,
and T ↔ ακ. Entropy doesn’t scale as volume. The Planck-Nernst version of the third
law, i.e. that S → 0 as T → 0 is not satisfied, since e.g. extremal black holes have an
area with κ = 0. This is not concerning since, in any case, the third law is not seen
to be ‘fundamental’ in the sense that the other laws are, and there exist examples of
simple quantum systems violating it: see [9]. The stumbling block preventing us making
an identification between these sets of laws is that black holes only absorb and emit no
radiation, i.e. T is not physically given by κ (hence S is not given by A). This problem
was in fact resolved, and the identification made precise, by Hawking in 1974 who showed
that thermal emission did indeed occur with respect to black holes at temperature
~κ A
T = =⇒ S =
2π 4~
as a blackbody. This is a very nice result, but the rabbit hole gets deeper. Problems
with entropy: what are the microstates, since we know the origin of temperature and
things thrown in vanish (overall entropy decrease- however, black hole radiation has as-
sociated and entropy and infalling mass causes area to increase); how to define entropy
(entanglement entropy);
Generalised second law (total entropy never decreases)

Stotal = S + SBH δStotal ≥ 0 (1.15)

Information problem- will come to later. These problems are related see [11]. GSL seems
to work well challenges to and protection of [10] We prove the second
The Second Law: Hawking’s area theorem
We need to introduce some machinery in order to arrive at proof, due to Hawking [5]

Definition 1.1. Given a manifold M be an oriented manifold of dimension n and S

another with dimension p < n. An embedding is given by a one-to-one map φ : S → M
and any point p ∈ S has an open neighbourhood, O, such that φ−1 : φ [O] → S is smooth.
We say φ [S] is an embedded submanifold of M . An embeddded submanifold of dimension
n − 1 is called a hypersurface.
1
following [3]

5
A tangent vector t ∈ Tp (M ) is said to be tangent to φ [S] at p if it is the tangent to a
curve in φ [S] at p. Given such a t, the normal to φ [S] at p, l ∈ Tp∗ (M ) if lµ tµ = 0.
A hypersurface in a Lorentzian manifold is said to be timelike, null or spacelike if any
normal l to it is spacelike (gµν lµ lν > 0), null (gµν lµ lν = 0) or timelike (gµν lµ lν < 0)
respectively.
Let S be a smooth function of coordinates xµ and consider the hypersurface S(x) = 0,
denoted by H, with normal given by lµ = f (x)(g µν ∂ν S) where f is an arbitrary non-zero
function. A tangent vector, t, to H by definition has t · l = 0 and therefore l is itself
tangent as l · l = 0 from above. Hence we may write:
dxµ
lµ =
dλ
In fact, xµ above will be null geodesics, which we now show:

lρ ∇ρ lµ = (lρ ∂ρ f )g µν ∂ν S + f g µν lρ ∇ρ ∂ν S = (l · ∂ ln f )lµ + f g µν lρ ∇ν ∂ρ S
dxρ ∂ d
=( ln f )lµ + f lρ ∇µ (f −1 lρ ) = ( ln f )lµ + lρ ∇µ lρ − (∂ µ ln f )l2
dλ ∂xρ dλ
d 1 d 1
= ( ln f )lµ + ∇µ (lρ lρ ) − (∂ µ ln f )l2 = ( ln f )lµ + ∂ µ l2 − (∂ µ ln f )l2
dλ 2 dλ 2

Note l2 |H = 0 does not mean ∂µ l2 |H = 0, so

d 1
lρ ∇ρ lµ |H = ( ln f )lµ + ∂ µ l2 |H
dλ 2
But,
d 2
l2 |H = const ⇒ l = 0 ⇒ tµ ∂µ l2 |H = 0 ⇒ ∂µ l2 |H ∝ lµ
dλ
for any tµ tangent on H. So, finally, lρ ∇ρ lµ |H = κlµ , κ ∈ R and can be set to vanish by
choosing f in which case λ is an affine parameter.
Hence we see that the xµ (λ)’s, λ affine, are geodesics which we will call the generators
of H.
By Frobenius’ theorem, a vector field, lµ , is hypersurface orthogonal iff

l[α ∂β lγ] = 0

In particular, given a Killing vector ξ, we define a Killing horizon of ξ as a null hypersurface

such that ξ|H is normal, i.e. ξ = f l, for l normal to H in the affine parametrisation. Hence
we obtain a formula for surface gravity:
1
κ2 = − (∇µ ξ ν )(∇µ ξν )
2 H

and κ is constant on orbits of ξ.

Killing horizon coincides with event horizon for static spacetimes. Simple example of
bifurcate horizon is Rindler.

6
 Having acquainted ourselves with hypersurfaces we can now go on to define what
exactly we mean by a black hole, which we have, up to now, regarded as a ‘region of no
escape’, but can also think of as the impossibility of escaping to =+ . Before we can arrive
at that, we will need to write down a flurry of definitions which will also be needed in
understanding the statement of the area theorem.
The first definition that will concern us will allow us to distinguish between fu-
ture(past), i.e. forwards (backwards) in time, as we have in special relativity, in a general
setting. (Note, there are spacetimes which do not admit such a demarcation continuously,
but we will not refer to them).

Definition 1.2. Given (M, g), we say this spacetime is time orientable, i.e. has a time
orientation, if there exists a smooth nowhere vanishing timelike vector field, T , on M .
Given a causal (i.e. timelike or null) vector, V , we say that V is future directed if V · T < 0
and past directed if V ·T > 0, and a curve λ whose tangent is such a V , is also, respectively,
called future or past directed.

Definition 1.3. A future (past) directed causal curve λ : I → M , where I an open inter-
val in R, is said to have a future (past) endpoint, p ∈ M , if for each of its neighbourhood’s,
O, there is a t0 ∈ I such that for all t > t0 we have λ(t) ∈ O.

If such a point does not exist, we say the curve is future (past) inextendible.

Note that a curve (as above) cannot have more than one future or past endpoint, and
that the point p need not lie on it. This definition allows us to separate cases where a
curve terminates due to not being defined to extend further, or e.g. due to ‘hitting’ a
singularity.
The next few definitions are crucial for understanding the notion of ‘predictability’ of
general spacetimes.

Definition 1.4. Given S ⊂ M , define the future domain of dependence of S as the set

D+ (S) = {p ∈ M Every past, inextendible, causal curve through p intersects S}

The past domain of dependence of S, D− (S), is defined similarly, replacing “future” by

“past”. If S is such that
D(S) =: D+ (S) ∪ D− (S) = M
then S is said to be a Cauchy surface, and (M, g), a globally hyperbolic spacetime.

A globally hyperbolic spacetime is a particularly powerful concept as it allows one to

determine uniquely solutions to hyperbolic differential equations globally from the initial
data on S, where S is a Cauchy surface.
The next definition needed in this discussion will use the idea of an ‘asymptotically
flat’ spacetime. Whilst it is possible to formulate a precise definition of the same, so
we know exactly what we want‘at infinity’, we omit it here and refer the reader to pg.
276 of [8] where it is given in full. The main difficulty in defining an asymptotically flat
spacetime is that it is not clear what we mean by taking ‘limits to infinity’ in general

7
coordinate systems such as to retain coordinate independence. The solution, roughly, is
that we can conformally embed our spacetime (M, gµν ) into another (M f, geµν ) where the
latter has properties similar to Minkowski space.
Given an asymptotically flat spacetime, (M, gµν ), define the causal past of the subset
=+ ⊂ M by J − (=+ ). Define also the [topological] closure of this by J¯− (=+ ) and its
boundary by J˙− (=+ ) = J¯− (=+ ) − J − (=+ ).

Definition 1.5. (Wald).

Given an asymptotically flat spacetime (M, g) and its associated conformal embedding
(Mf, ge), we say that the former is strongly asymptotically predictable if an open region
V ⊂ M̃ satisfying M ∩ J − (=+ ) ⊂ Ve , with the closure taken in M
e f exists such that (Ve , ge)
is globally hyperbolic.

Note the closure manifests the utility of the embedding we spoke of earlier, in particular
i0 ∈ Ve . The crucial point here is that J − (=+ ) needn’t contain the whole of M , i.e. there
can exist regions in M not visible to J − (=+ ), so that a black hole can be defined as

Definition 1.6. A strongly asymptotically predictable spacetime (M, g) is said to contain

a black hole if M * J − (=+ ) and the (future) event horizon is given by H+ = M ∩ J˙− (=+ )

It is relevant here to mention that a globally hyperbolic spacetime forbids observers

outside or on the horizon from detecting a black hole in finite time, meaning that naked
singularities, i.e. ones singularities not ’covered’ by a black hole, are precluded- apart
from say an initial singularity. This point forms the basis of the cosmic censorship hypoth-
esis, which roughly states: All ‘physical’ spacetimes are globally hyperbolic, i.e. nature
forbids naked singularities. Note this has not been proven to date and is one of the major
outstanding problems in classical GR.
Properties of future event horizons:
1. null hypersurface

2. achronal, i.e. no two points on the horizon can be connected by a timelike curve

3. does not contain i0 or =−

4. can have past endpoints

5. cannot have future endpoints (Penrose’s theorem)

Having introduced null hypersurfaces, we next need the idea of congruences of curves:
a family of curves, with exactly one passing each point. We call it a geodesic congruence
if the curves are geodesics. In particular, the integral curves generated by a continuous
vector field are a congruence of curves. In the following, we will be following [2] closely.
We will be concerned with null geodesic congruences, so to start, consider the null tangent
field in the affine parametrisation, ξ α , generating a congruence. We can define the tensor

Bµν = ∇µ ξν

8
so that Bµν ξ ν = Bµν ξ µ = 0. Consider a ‘displacement’ vector η α , that can be chosen to
commute with ξ (by appropriately choosing coordinates, c.f. commutability of coordinate
basis) so that
Lξ η µ = 0 ⇒ ξ · ∇η = η · ∇ξ = B µν η ν . (1.16)
B µν η ν will thus measure geodesic deviation, i.e. the failure to parallel transport η α . Hence,
given a geodesic, we can describe geodesics nearby by a displacement vector, however such
a displacement vector will not be unique since adding a multiple of ξ will give another
such vector. To fix this, we impose the conditions

η·ξ =0=η·n n · ξ = −1 n2 = 0

where n is another vector that has been introduced for the following reason: the space
of vectors orthogonal to ξ also includes ξ as it is null, hence the condition η · ξ = 0 is
insufficient to fix the gauge. Thus we choose a vector, n, not orthogonal to ξ, so that the
condition η · n will uniquely specify the displacement vectors in the two-dimensional space
we are interested in. Note the choice n · ξ = −1 is arbitrary. Consistency in our choice
requires that t · ∇n = 0- (parallel transport). Write a projector P µν = δνµ + tµ nν + nµ tν , so
that in particular P µν η ν = η µ . Hence, P projects into the two-dimensional tangent space
spanned by the η vectors. Project B into the two-dimensional so that it is orthogonal to n
as well, i.e. so there are no components in the direction of n. Decompose B̂ into algebraic
irreducible parts. B̂ µν = P µρ B ρσ P σν
It is readily checked that B̂ also satisfies a relation similar to (1.16) i.e.

B̂ µν η ν = t · ∇η µ (1.17)

Define the expansion, shear and twist, respectively, as:

θ = B̂ µν (1.18)
1
σ̂ µν = B̂(µν) − P µν B̂ ρρ (1.19)
2
ω̂ µν = B̂[µν] (1.20)

which (algebraically) decomposes B̂ as

1
B̂ µν = θP µν + σ̂ µν + ω̂ µν (1.21)
2
A relation which will be of great use to us is the Raychaudhuri equation (for null geodesics),
which is also of key importance in establishing the singularity theorems of Hawking and
Penrose. We arrive at it from the following considerations. First,
dθ dxµ ∂
= θ = ξ · ∇θ = ξ · ∇B̂ µµ
dλ dλ ∂xµ
using ξ · ∇ξ = 0 and ξ 2 = 0

B̂ µν = B µν + ξ µ (nσ B σν + nσ B σρ nρ ξν ) + (B µρ nρ )ξν (1.22)

9
Since B µν ξµ = 0, we have (again with ξ · ∇ξ = 0 and ξ 2 = 0),

B̂ µµ = B µµ = B µν P νµ

Recall that ξ · ∇n = 0. This means ξ · ∇P νµ = 0. Hence,

dθ
= ξ · ∇(B µν P νµ )
dλ
= P νµ ξ · ∇B µν
= P νµ ξ ρ ∇ρ ∇ν ξ µ
= P νµ ξ ρ ∇ν ∇ρ ξ µ + P νµ ξ ρ [∇ρ , ∇ν ] ξ µ

Note ∇(ξ · ∇ξ) = (∇ξ)(∇ξ) + ξ∇∇ξ = 0 =⇒ ξ∇∇ξ = −(∇ξ)(∇ξ) = −BB, and

[∇µ , ∇ν ] ξ α = Rαβµν ξ β

dθ
= −P νµ B µρ Bνρ + P νµ ξ ρ Rµσρν ξ σ
dλ
= −P νµ B µρ B ρν + δµν ξ ρ Rµσρν ξ σ
= −B̂ µρ B̂ ρµ − Rσρ ξ σ ξ ρ

where the last equality holds due to the trace, and made manifest by e.g. copious use of
the identities.
Using P µν Pνµ = trδ + 4(n · ξ) + 2(n · ξ)(n · ξ) = 4 − 4 + 2 = 2, and P µν B̂νµ = trB̂ = θ,
which again is obtained from the identities. This gives
1
P µν σνµ = P µν B̂νµ − θP µν Pνµ = θ − θ = 0
2
Hence we can write
1 1
−B̂ µρ B̂ ρµ = −( θP µρ + σ̂ µρ + ω̂ µρ )( θP ρµ + σ̂ ρµ + ω̂ ρµ )
2 2
1 1 1
= −( θ2 P µρ P ρµ + θP µρ σ̂ ρµ + θσ̂ µρ P ρµ + σ̂ µρ σ̂ ρµ + ω̂ µρ ω̂ ρµ )
4 2 2
1 2 µρ µρ
= −( θ + σ̂ σ̂µρ − ω̂ ω̂µρ )
2
where antisymmetry has been used to write the last line in a prettier way. Finally, we
can write Raychadhuri’s equation:
dθ 1
= − θ2 − σ̂ µρ σ̂µρ + ω̂ µρ ω̂µρ − Rσρ ξ σ ξ ρ (1.23)
dλ 2
In particular, let us consider this equation given the weak energy condition, Tµν ξ µ ξ ν ≥ 0,
and the expansion of θ of the null geodesic generator of a null hypersurface, so that
dθ 1 1 d −1 1 1
≤ − θ2 − 8πTµν ξ µ ξ ν ≤ − θ2 ⇐⇒ θ ≥ ⇐⇒ θ−1 ≥ θ0−1 + λ (1.24)
dλ 2 2 dλ 2 2

10
where we have used the positivity of σ̂ 2 , the vanishing of omega (since ξ here is normal
to the null hypersurface andusing (1.22) with Frobenius’ theorem) and that ξ is null in
Einstein’s equation in the first inequality. Here θ0 is the initial value of θ. Suppose
that θ0 < 0, then θ → −∞ within an affine length of 2/|θ0 |. This is easy to see, since
θ ≤ θ0 /(1 + (λθ0 /2)) and the denominator is zero at λ = 2/|θ0 |. This tells us that if
there is convergence in the congruence at any point, then there will exist caustics which
we can think of as singularities in the congruence. Conjugate points? θ = 0 for stationary
spacetimes.
We define the area element: a = εµνρσ ξµ nν ηρ0 ησ00 then
da
= ξ · ∇a = εµνρσ ξµ nν (ξ · ∇(ηρ0 )ησ00 + ηρ0 ξ · ∇ησ00 ) = εµνρσ ξµ nν B̂ρλ (ηλ0 ησ00 − ηλ00 ησ0 ) (1.25)
dλ

4εµνρσ Bρλ ξ[µ nν ηλ0 ησ]

00
= εµνρσ Bρλ (ξλ n[µ ην0 ησ]
00 0
+ nλ ξ[µ ηnu 00
ησ] 00 0
+ ξ[µ nν ησ] 0 00
ηλ − ξ[µ nν ησ] ηλ )
00 0 0 00
= εµνρσ Bρλ (ξ[µ nν ησ] ηλ − ξ[µ nν ησ] ηλ )) (Using B · ξ = B · n = 0)
=ε µνρσ
ξµ nν B̂ρλ (ηλ0 ησ00
− ηλ00 ησ0 ) (equation (1.25))
4
= εµνλσ εµνρσ Bρλ ξµ1 nµ2 ηµ0 3 ηµ004 εµ1 µ2 µ3 µ4
4!
4!
= δλρ Bρλ ξµ1 nµ2 ηµ0 3 ηµ004 εµ1 µ2 µ3 µ4
4!
da
=⇒ = θa (1.26)
dλ
Theorem 1.1 (Hawking’s area theorem). Let (M, g) be a strongly predictable asymptoti-
cally flat spactime satisfying Tµν ξ µ ξ ν ≥ 0 for all null ξ µ . The area, A, of the future event
horizon doesn’t decrease with time.
Proof. The requirement of a strong asymptotically flat spacetime is equivalent to the
requirement that a family of Cauchy surfaces, Σ(λ), exist such that Σ(λ0 ) ⊂ D+ (Σ(λ)).
We know that H+ is a null hypersurface, so we can use the affine parameter, λ if its
generator to parametrise the Cauchy surfaces, so that the area of the event horizon is
given by A(λ) = H+ ∩ Σ(λ)- i.e. the area of the horizon at affine point λ. Hence we seek
to prove A(λ0 ) ≥ A(λ) for λ0 > λ.
In order to do this, it suffices to consider the area elements above. Recall
da
θ≥0⇒ = θa ≥ 0 (1.27)
dλ
so that all we need to show is θ ≥ 0 everywhere on H+ . Suppose it’s not. Then θ → −∞
in finite affine parameter. Since generators of the horizon do not have future endpoints,
by theorem 4.5.12 in Hawking and Ellis, which states that if a point r lies in the interval
of two points p and q, such that r is a ’conjugate’ point to p along a geodesic γ, then
there is a timelike curve connecting p to q. Here r conjugate to p along γ means that
θ → −∞ at r. But if there is such a timelike curve for the generators of the horizon then
the achronicity property of the horizon is violated. Hence θ ≥ 0. Hence
Z
dA
≥ θda ≥ 0 (1.28)
dλ

11
and the proof is complete.

Hawking Radiation
Particle creation in curved spacetime
In order to show that black holes radiate, and are indeed thermodynamic in their own
right, we will adopt the approach of Unruh, which captures the main ideas of the original
caluclation by Hawking [6], and has the advantage of not relying on transplankian modes.
The latter used what is known as the geometrical optics approximation to tackle the
question. These are semi-classical calculations wherein the black hole is treated as classical
and the fields are quantum mechanincal, i.e. gravity is not quantised. This however does
not appear to detract from the validity of the result, since curvature is small in the region
where these particles are created.
Consider the generally covariant Klein-Gordon equation2 :

( − m2 )fi = 0

which has solutions {fi }. For a pair of solutions, we introduce the “inner product”3
(endowing the space of solutions with a symplectic structure):
Z
(f1 , f2 ) ≡ i dΣµ (f2∗ ∂µ f1 − f1 ∂µ f2∗ )
↔
Z
≡ i dΣµ (f2∗ ∂µ f1 )

where dΣµ is the volume element pointing normal to a Cauchy hypersurface, i.e. we
consider globally hyperbolic spacetimes. This will guarantee that the space of global
solutions corresponds is determined by inital data on a Cauchy surface. It is not hard to
show that this definition is independent of the choice of hypersurface, i.e. that

(f1 , f2 )Σ1 = (f1 , f2 )Σ2

To do this, take two hypersurfaces Σ1 and Σ2 where functions on these surfaces vanish at
spatial infinity (in case they’re non-compact). Let V be the four-volume bounded by these
hypersurfaces and if necessary, time-like surfaces on which the functions vanish. Then we
have
↔ ↔
I Z
∗
(f1 , f2 )Σ1 − (f1 , f2 )Σ2 = i µ
dΣ (f2 ∂µ f1 ) = i dV ∇µ (f2∗ ∂µ f1 )
Z∂V V

=i dV ∇ (f2 ∂µ f1 − f1 ∂µ f2∗ )
µ ∗

ZV Z
=i dV (f2∗ f1 − f1 f2∗ ) = i dV m2 (f2∗ f1 − f1 f2∗ )
V V
=0
2
In general we will have ( − m2 − ξR)fi = 0 and we take the “minimal coupling” prescription ξ = 0,
although this is not essential.
3
Not positive definite since (f, f ) = −(f ∗ , f ∗ ).

12
where dV is the four-volume element, and in the second line the quadratic terms cancel
the Klein Gordon equation gives the final step. Talk about volumes- see jacobson. We
can choose the basis {fi , fi∗ } such that the following relations are satisfied,
(fi , fj ) = δij , (fi∗ , fj∗ ) = −δij , (fi , fj∗ ) = 0 (1.29)
A solution of the Klein-Gordon (KG)equation can be expanded as:
X
ϕ(x) = (ak fk (x) + a∗k fk∗ (x))
k

which gives the quantized field

(âk fk (x) + â†k fk∗ (x))

X
ϕ̂(x) = (1.30)
k

where hats are placed to denote operators. We can write,

âi = (ϕ̂, fi ) â†i = −(ϕ, fi∗ ) (1.31)

and the commutation relation, [âi , â†j ] = δij - all else vanish. In flat space, the field
operator, ϕ, can be decomposed into Fourier modes
X 1
ϕ̂ = √ (âk eik·x−iωt + â†k e−ik·x+iωt )
k
2ωV
p
where ω = |k|2 + m2 , V is the volume on which ϕ is defined, the vacuum state is defined
as
âk |0 i = 0
i.e. the state annihilated by all âk ’s, and the â†k ’s create particles. The above expansion
gives f ∼ e−iωt - the positive frequency modes. This is a canonical choice that we make
in Minkowski spacetime: we have a natural split of positive and negative norm solutions
whence we can define the vacuum uniquely (using Poincaré invariance) and hence the
Fock space, no matter the frame in which t is taken to be the time coordinate. Explicitly,
the positive Fourier modes u(t, x)k are the set of orthonormal functions satisfying the KG
equation, with
ik µ ∂µ uk = ωk uk (1.32)
and k µ = (1, 0, 0, 0) is a timelike Killing vector, and we can go on to define our Fock space
etc.
However, splitting positive and negative frequency modes in curved spacetime is co-
ordinate dependent and since, prima facie, there is no rule for choosing from the many
bases, {fi , fi∗ }, there is no unique vacuum. This means it is unclear how to define a parti-
cle in curved spacetime, rather such a definition is coordinate dependent. We would like
to understand this ‘non-uniqueness’ further. A Fock space can be constructed from (1.30)
by taking the ‘vacuum’ as the state satisfying4
âk |0 ia = 0 h 0|0 ia = 1
4
The inner product, h | i, is positive definite

13
for all k, and Fock space, H, has basis {|0 ia , a†i |0 ia , a†i a†j |0 ia , . . . }. Consider now two bases
of complex solutions to the KG equation: {fi , fi∗ }, taken as before, with time coordinate
t and vacuum |0 ia ; and {hi , h∗i }, with time coordinate t0 and vacuum |0 ib , analogously
defined, so we can expand as

(âk fk (x) + â†k fk∗ (x)) = (b̂k hk (x) + b̂†k h∗k (x))
X X
ϕ̂(x) = (1.33)
k k

Since we have two complete bases, we can write

X X
fi = αij hj + βij h∗j fi∗ = ∗ ∗
αij ∗
hj + βij hj (1.34)
j j

These expressions go by the name of a Bogolubov transformation, and the coefficients αij ,
βij are called Bogolubov coefficients, which, using (1.29) satisfy
X
∗ ∗
αik αjk − βik βjk = δij (1.35)
k

and we can write operators using (1.31) as

∗ † ∗ †
X X
∗
âi = αij b̂j + βij b̂j b̂i = αij âj − βij âj (1.36)
j j

Of particular use is the particle number operator, defined for the fi -th mode

Nia = a†i ai (1.37)

where the superscript labels the basis. Also,

h 0|Nia |0 ib = h 0|a†i ai |0 ib = ∗ †
X X
h 0|(bj βji )(βik bk )|0 ib = |βji |2 (1.38)
j,k j

where we have used the commutation relation for the b-operators to pick up a delta in the
last step.
This poses the following problem: an observer using operators âi will not regard the
state |0 ib of an observer using operators b̂i as the vacuum, rather will find particles therein.
Thus, the vacuum in one setting appears to have a multitude of particles when observed
in another. In cases where the spacetime is asymptotically flat, it may be suggested to
use the natural particle definitions of this region to bypass the above ambiguity. However,
modes travelling through curved regions may be found to contain particles despite not
starting out with any. This is the particle creation effect and it occurs when β 6= 0.
Unruh Effect
Having established the phenomena of particle creation in curved spacetime, let us now
make contact with black holes and radiation. Before we can do that though, we need to
first review particle creation in Minkowski spacetime as seen by an accelerated observer,
i.e. Rindler spacetime, where we will find that the Minkowski vaccum appears thermal:

14
the Unruh effect. We will work in (1+1)-dimensional spacetime for simplicity, but this
easily generalises.
The metric in two-dimensional Rindler space is related to the Minkowski metric as

ds2 = e2aξ (−dη 2 + dξ 2 ) = −dt2 + dx2 (1.39)

where x = a1 e2ξ cosh(aη) and t = a1 e2ξ sinh(aη) gives the relation to Minkowski space.
These coordinates only cover the region x > |t|- ‘right wedge’. By defining the null
coordinates u = η + ξ and v = η − ξ we see

ds2 = −2e2aξ dudv

where the patch covered by these coordinates is u < 0, v > 0. If we further define U and
V as exponentials of u and v respectively, we see there is a close analogy existing between
transformations taking us from Minkowski to Rindler, and Schwarzschild to Kruskal: A
spacetime with a horizon usually displays such an exponential relationship between two
coordinate systems. The horizon in Rindler space is an acceleration horizon, u = 0 = v,
which chops up our spacetime into wedges (right and left) such that they are causally
disconnected. The horizon is in fact a bifurcate Killing horizon so we can use this to
capture the essential aspects of an arbitrary Killing horizon.
We have Cauchy surfaces given by η = const, so the Rindler wedge is globally hyper-
bolic, but it does not cover all of Minkowski space. Hence we are left with the problem
that these Cauchy surfaces are not Cauchy in Minkowski space, so modes defined on a
Rindler patch do not provide a complete set into which a general Minkowski solution can
be expanded. Thus we seek to seek to extend Rindler coordinates and in fact it is sufficient
to cover the region |x| > |t|, i.e. right and left wedges. To do this, note that η gives us
the timelike Killing vector
∂ ∂ ∂
= ax + at (1.40)
∂η ∂t ∂x
which is equally good in both left and right wedges (it will run backwards to time in the
left wedge (t ∼ − sinh η))5 .
The KG-equation (massless for simplicity) has postive frequency solutions6 :
(
e−iωη+ikξ right wedge
f R (k) = (1.41)
0 left wedge
(
0 right wedge
f L (k) = (1.42)
eiωη+ikξ left wedge
where we have explicitly indicated the support for our solutions, and |ω| = k.. We can
write the field by expanding as
†R ∗R †L ∗L
X
ϕ̂ = b̂R R L L
k fk + b̂k fk + b̂k fk + b̂k fk (1.43)
k
5
So we are using a boost timelike Killing vector field, as opposed to a time translation one.
6 √
use ∇2 φ ∼ ∂( gg∂φ). Note also that the two dimensional case is special

15
 Our task is to find the particle content as seen by a Rindler observer. Thus, we need to
expand Minkowski in a basis of Rindler and then compute the Bogolubov transformation in
order to achieve this. Bogolubov will relate Minkowski vacuum to Rindler whose vacuum
is |0iR = |0iR ⊗ |0iL so that it covers the full Cauchy slice in Minkowski. Recall the
Minkowski expansion is given by

ĉk hk + ĉ†k h∗k

X
ϕ̂ = (1.44)
k

Unruh essential insight: work with any complete set of Minkowski modes (will give same
vacuum). Any positive freq. set will do so long as it is (a) pos. freq. (b) well defined on
timeslice (c) can be analytically continued. Now writing, which is easily verifiable
(
iω/a f R (k) right wedge
(a(x − t)) = (1.45)
e−πω/a f ∗L (−k) left wedge

we can see the similarity with Schwarzschild in the geometrical optics approach. Then

h(k) = N (f R (k) + e−πω/a f ∗L (−k)) (1.46)

where N is a normalisation constant. Bogolubov transformation is then (dropping left

and right notation)
b = N (c + e−πω/a c† ) (1.47)
hence we find for N
1
b= p (eπω/a c + e−πω/a c† ) (1.48)
2 sinh(πω/a)
Number operator gives: (note c’s kill mink. vacua)
1
hN iR = h0|b† b|0iM = (1.49)
e2πω/a −1
and Planck distribution.
a ~
kT = (1.50)
2π c
replacing dimensionality constants. Note that essentially what happened was that horizon
split region; frequency mismatch; observer can see pair production at horizon without
violating energy cons.
Black Hole Thermality
In order to analyse particle creation by a black hole7 , we can use our knowledge of
Minkowski to Rindler, to relate Schwarzschild to Kruskal, although black holes occur
naturally arise due to gravitational collapse, and an eternal black hole radiating is some-
what unphysical nonetheless provides a useful way of approaching the problem. So we
work analogously to the Rindler case. We quantise the right hand (region I) of Kruskal by
taking the Cauchy surface H− ∪=− and finding positive modes here. There are two Killing
7
this follows [17].

16
vectors at H− , one given by Kruskal-Szekeres, and the other by Schwarzschild time. We
choose affine (akin to geometrical optics) which is given by Kruskal-Szekeres. For =− take
the null past coordinate. Then we seek to relate the positive frequency solutions from the
past to the future region H+ ∪ =+ by computing the Bogolubov transformation. Again,
by analogy with Rindler, we do this by extending our solutions to other regions. In the
end we find
1
T = (1.51)
8πM
Information Paradox
The radiation Hawking found in 1974, and we have been discussing, raises very big prob-
lems which theorists have been trying to resolve ever since. Here we discuss what is known
as the ‘information’ paradox, or ‘loss of quantum coherence.’ This is

To begin, Hawking discovered that black holes radiation is exactly thermal, in the semiclas-
sical approximation (which is not exact, in particular backreaction effects- backscattering
of states affecting the geometry around the black hole- give rise to greybody factors 8 that
are deviations away from perfect thermality), and independent of the details of the col-
lapsing object, i.e. the initial state, characterised by a variety of parameters, collapses to
a black hole described by only a few. This last statement is the no-hair conjecture which
loosely reads: stationary black holes are characterised by their mass, angular momentum
and (electric or magnetic) charge- where these quantities are in fact global charges that
can be measured at infinity. There already appears to be loss of information. However,
the initial state is just hiding behind the horizon and part of a quantum system along with
the outgoing radiation, so everything seems to be alright at this point. But let’s probe
this situation deeper.
First let us recall some basic facts about states in quantum mechanics. A basic principle
in quantum theory is known as the ‘quantum xerox principle’, or ‘no-cloning theorem’.
make a duplicate of state, and (schematically) goes like this. Suppose we have a state that
we insert into a machine which outputs the state and its duplicate

|ψi → |ψi × |ψi (1.52)

then
|ψ1 i + |ψ2 i → (|ψ1 i + |ψ2 i) × (|ψ1 i + |ψ2 i)
where |ψi = |ψ1 i + |ψ2 . This is wrong by the rule of linear evolution of state vectors which
asserts
|ψ1 i + |ψ2 i → |ψ1 i × |ψ1 i + |ψ2 i + |ψ2 i (1.53)
so we see if one copy of information were to be outside a black hole we cannot have the same
inside and vice versa. Recall also the notion of pure and mixed state: the former being
described by linear combinations of state vectors, and the latter a statistical ensemble of
pure states, e.g. an entangled state, which we can describe by a density matrix ρ whence
8
Here we are using the word ‘thermal’ to denote a planckian spectrum of radiation. Greybody factors
are then a feature of any warm body whose wavelength is similar to its size.

17
the von-Neumann entropy, S = −tr ρ ln ρ, can be computed. What we need to take home
is: “unitarity demands pure states do not evolve to mixed states”.
By Stefan’s law, as black holes radiate they will lose mass, at a rate dM/dt ∼ M −2 ,
which gives a lifetime of τ ∼ M 39 , although what exactly happens to the black hole at
late times (when the hole is Planck sized) requires a quantum theory of gravity to tell us.
As the black hole evaporates we have three possibilities

1. we are left with a singularity

2. we are left with a stable remnant

3. the black hole radiates away all its mass and disappears.

Option 1 is not satisfactory as it means the theory is incomplete since quantum mechanics
will not have explained the singularity problem. Option 2 seems to imply an infinite
number of possible states since black holes can be arbitrary sized, so an arbitrary amount
of information is stored in this remnant. This doesn’t seem to be correct as an infinite
number of states within a finite bounded region is very problematic, e.g. how could we keep
the remnant stable? Option 3 means the black hole has to somehow get its information
out before it vanishes in order for information loss not to occur. The last option presents
two main difficulties, which are in essence the information paradox:

1. the initial state is uncorrelated with the outgoing radiation: how does it transfer
information? If the initial state starts of pure, how does it end in a mixed state?10

2. by a Schwinger process, we can create an entangled pair (in mixed state). If this
pair straddles the horizon, where one falls in and the other shoots off to infinity,
then when the black hole disappears, the quanta at infinity will be in mixed state
with nothing to mix with!

These two problems have been vigorously attacked, but the information seems to be lost
unless we change fundamental notions like locality. Recently however, there has come
a proposal from string theory, developed by Mathur and others [23][24], which seems to
have a solution in what is called ’fuzzballs’. The key point here is that the horizon is not
empty space: the matter making the hole exists up to the horizon, so that information is
radiated out.

9
this is longer than the age of the universe for a solar mass sized black hole!
10
This is a point in principle not in practice where information is often lost.

18
Chapter 2

String Theory Background

2.1 Strings
A string is a one dimensional object sweeping out a world-sheet in space time akin to the
point particle’s world-line. Hence, to write an action describing the string we consider
the relativistic point particle first. We will see that the action for the point particle is
proportional to the length of its world-line; generalising, the string action should thus be
proportional to the world-sheet area.
The action for a (massive) point particle in D dimensions is,
Z p
S = −m dτ −Ẋ 2 (2.1)

dX µ
Here we have used Ẋ µ = and Ẋ 2 = ηµν Ẋ µ Ẋ ν . The vector X µ (τ ) describes the
dτ
position of the particle along its world-line, parametrised by τ which we take to be proper
time. Also, we are working in D-dimensional Minkowski space R1,D−1 with signature
ηµν = diag(−1, +1, +1, ..., +1).

Varying, we have:
" # " # τ1
τ1
−Ẋµ Ẋµ δX µ
Z
∂ µ
δS = m dτ p δX + m p (2.2)
τ2 ∂τ −Ẋ 2 −Ẋ 2 τ2
| {z }
surface term
Where the ‘surface’ term vanishes by imposing suitable boundary conditions. This
yields the equation of motion:

∂ −mẊµ
p =0 (2.3)
∂τ −Ẋ 2

There are a couple of unsatisfactory things with the above action. First off, there is a
square root which likes to make life hard. Second, it is clear this action is useless to
describe massless particles. Let us do better by writing an action which holds for massless
particles too, and is easier to deal with. Consider
Z
1
S= dτ (e−1 Ẋ 2 − em2 ) (2.4)
2

19
where e = e(τ ).
The equations of motion, varying with respect to X µ and then by e, are:

∂
(−e−1 Ẋ µ ) = 0 (2.5)
∂τ
Ẋ 2 + e2 m2 = 0 (2.6)
We can show equivalence with the previous action by solving (2.6) for e,
p
−Ẋ 2
e=
m
from which we recover [the equation of motion] (2.3) by substituting into (2.5). Plugging
this expression for e in the action (2.4) gets us back (2.1), the action we started with.
The action has symmetry under both Poincaré transformations and reparametrisations.
Recall the first of these is given by the transformation

X µ → X 0µ = Λµν X ν + cµ

where cµ is a constant translation and Λ is a Lorentz matrix satisfying ηαβ Λαµ Λβν = ηµν .
The second can be shown by realising that e(τ ) in fact acts as an ‘einbein’. Writing
√
e = −gτ τ , where gτ τ is the 1d metric on the particle’s world-line and gτ τ g τ τ = 1, we get
reparametrisation invariance by

√ p p
e(τ )dτ = gτ τ dτ = gτ τ dτ 2 ≡ gτ̃ τ̃ dτ̃ 2 = e(τ̃ )dτ̃

∴ dτ̃ /dτ = e/ẽ, where ẽ ≡ e(τ̃ ). Since dX/dτ = (dX/dτ̃ )(dτ̃ /dτ ) we have invariance as
required, i.e.
Z  2
1 −1 dX
S= dτ̃ (ẽ − ẽm2 ) (2.7)
2 dτ̃
Nambu-Goto Action
In order to arrive at an action for a string, we parametrise its world-sheet by two coordi-
nates, σ a = (τ, σ), timelike and spacelike respectively where −∞ < τ < ∞ and 0 ≤ σ ≤ π.
The world-sheet is mapped onto the D-dimensional Minkowski spacetime background, or
target space, via X µ (σ, τ ). Strings can be open or closed, for which we have a curved
world-sheet in the former case and a curved cylinder in the latter. Our action- in keep-
ing with our earlier comments about proportionality with the world-sheet area- is hence
written,
Z Z q
S = −T dA = −T dτ dσ −det (ηµν ∂a X µ ∂b X ν )
Z p
= −T d2 σ −det γab (2.8)

20
where γab is the induced metric on the world-sheet and T is for tension.
1
T = , α0 = ls2
2πα0
where we have defined the “string length” ls and α0 is the “universal Regge slope”. It is
easily checked that all of the above is dimensionally consistent. This is the basic scale of
the theory. Alternatively, by evaluating the determinant, we can write
Z q
2
S = −T d σ −(Ẋ · X 0 )2 + Ẋ 2 X 02

This is the “Nambu-Goto action”. Again, as with the point particle action (2.1), we have a
square root which is bothersome (in particular for purposes of quantisation). Thankfully,
we have an alternative, (classically) equivalent action:
Polyakov action
√
Z
1
S=− d2 σ −gg αβ ∂α X µ ∂β X ν ηµν (2.9)
4πα0
where gαβ is the worldsheet metric. The symmetries of this action include those of Nambu-
Goto, but with an extra addition: Weyl invariance.

gαβ → e2φ(σ) gαβ

Conformal gauge
gαβ = e2φ(σ) ηαβ
In this gauge, the action becomes
Z
1
S=− d2 σ ∂α X · ∂ α X
4πα0
variation:
Z Z π τ =τf
1 1
δS = d2 σ (∂α ∂ α X) · δX + dσ Ẋ · δX (2.10)
2πα0 2πα0 0 τ =τi
Z τf σ=π
1
− dτ X 0 · δX (2.11)
2πα0 τi σ=0

In order for the last term to vanish- like the second term above- and equation of motion
reduce to
∂α ∂ α X µ = 0 (2.12)
we need
X 0µ δXµ σ=0,π
=0
we have boundary conditions (not unique but physically relevant):
“Neumann” boundary conditions:

X 0µ σ=0,π
=0 (2.13)

21
ends are freely placed in spacetime. “Dirichlet” boundary conditions:

δX µ |σ=0,π = 0 (2.14)

→ end points fixed at X µ = cµ where cµ is constant. Considering Neumann boundary

conditions (at both ends) for some coordinates X a = 0, ..., p, and Dirichlet (at both ends)
for X I = p + 1, ..., D − 1 gives

SO(1, D − 1) → SO(1, p) × SO(D − p − 1)

so that the ends of the (open) string are fixed in a (p + 1)-dimension hypersurface- a
D-brane or Dp-brane where p is the spatial dimension of the brane. This hypersurface is
in fact a dynamical object due to the momentum from open strings ending on it. Note
that the equation of motion (2.12) is just the two-dimensional Laplace equation which has
solutions
µ µ
X µ = X+ (σ + ) + X− (σ − ) (2.15)
µ µ
where σ ± = τ ± σ are lightcone coordinates and X+ and X− are left- and right-moving
waves.
D-brane action: Born-Infeld action
Z q
S = −TDp dp+1 ξ −det(γµν + 2πα0 Fµν ) (2.16)

generalisation on Nambu-goto, ξ runs from zero to p. TDp is D-brane tension. Fµν =

2∂[µ Vν] field strength, Vµ is BI vector.

1
Tp = (2.17)
gs (2π)p lsp+1
Dp-branes couple to p+1 RR gauge fields just like electric charge to gauge potential
in Maxwell theory. D-branes exert zero force (see remark about Majumdar-Papapetrou).
Note here along with gravitational have dilatonic attraction. Low-energies massless modes
are relevant. For Q D-branes, weak field theory is U (Q) super Yang-Mills. Tension propor-
tonal to 1/gs means d-branes non-perturbative excitations of string theory becoming heavy
in weak coupling limit. We have introduced the coupling constant which counts loops in
string amplitudes (the interaction1 of strings an S-matrix summing over topologies). Note
2
gopen = gs . D-branes preserve 1/2 supersymmetry. Note, D(-1) is instanton. Not all type
II branes are Dirichlet, e.g. NS5 brane, F1. Each electric p-brane has magnetic dual (in
10 dimensions):
? dC p+1 = dC̃ 7−p (2.18)
So far we have only described the bosonic string. When we allow for fermions we have
what is known as a superstring theory, due obviously to the supersymmetry induced and
is realised by the inclusion of spinors. The number of consistent superstring theories is
then constrained to five, all of which are in ten dimensions: SO(32) Heterotic, E8 × E8
Heterotic, Type I, Type IIA and Type IIB. These theories are all in fact related by duality
1
interactions occur by joining and splitting of the string.

22
transformations. In addition, there is an eleven-dimensional theory, M-theory, which is
approached by the type IIA and E8 × E8 Heterotic theories at strong coupling. However,
this last theory is not well understood past its low-energy limit, (N=1 eleven dimensional
supergravity) and no strings are found there either. We will only concern ourselves with
the type II theories (N = 2 supersymmetry) in ten dimensions.
Type IIA: obtained from 11D sugra by KK, non-chiral NS-NS sector:

gµν Bµν Φ (2.19)

RR-sector
C (1) C (3) C (5) C (7) (2.20)
Type IIB:chiral, not from reduction, related by dualities-will not go into.

C (0) C (2) C (4) C (6) C (8) (2.21)

Similarly we have the Type IIA non-chiral fermions

ψµ λ (2.22)

and IIB
ζµi(∓) , χi(±) (2.23)
D-branes and BPS states, the (extended) supersymmetry algebra is given schematically

{Q, Q} ∼ (CΓ) · P + (CΓ(p) ) · Zp (2.24)

where Γp is a totally antisymmetric product of Dirac matrices (see later) and Zp is a p-form
‘central’2 charge, and P is the momentum vector. This is the charge for what forms do
these couple to-like EM. If we now sandwich the above anti-commutator between states
we see (in the rest frame and dropping indices):

0 ≤ hphys|{Q, Q}|physi = (M − a|Z|) (2.25)

⇒ M ≥ a|Z| (2.26)

Equation (2.26) is known as the Bogomolny-Prasad-Sommerfield (BPS) bound. Saturated

bound (BPS state) means some generators annihilate states, number of states doesn’t
change under adiabatic variations. D-branes are BPS with 28 states instead of maximal
216 .
2.2 Supergravity
Supergravity theories describe interactions of massless fields, i.e. the low-energy dynamics,
of a given string theory, and are the supersymmetric extensions of general relativity.
Action in d=10 dimensions for the ns-ns sector of the supergravity

d √
Z  
1 −2Φ 2 1 2
S = 2 d x −ge R − 4(∂Φ) + H (2.27)
2κ 2 · 3!
2
Although does not commute with Lorentz transformations

23
note that indices not shown means complete antisymmetrisation. Phi is known as the
dilaton, and its vacuum expectation value gives the string coupling: gs = eφ0 Low energy
effective NS-NS sector common to all sugras where H = dB2 and d = 10, 26. This action
is written in the string frame, with string metric gµν but we can also write the action
4
under a conformal rescaling, gµ = e d−2 Φ gEµν , where gEµν is the Einstein frame metric.
We will be working mostly in the string frame. Varying the action is easy but note the
metric isn’t as straightforward as usual since we have coupling with the dilaton, but the
field equation is obtained essentially by integrating by parts twice. It is given in full in
section 4.2 of Ortin.
∇µ (e−2Φ Hµνρ ) = 0 (2.28)
1 1
∇2 Φ = (∂Φ)2 + R + H 2 (2.29)
4 48
1
Rµν = 2∇µ ∂ν Φ + Hµρσ Hνρσ (2.30)
4

24
Chapter 3

Type IIB and Dp-brane solutions

Type IIB supergravity contains a self-dual five-form field strength which makes writing
a classical action for it problematic, although can be dealt with, e.g. see [29], and we
will write one down after to emphasise that it is these equations which are primary.
It is best therefore to work at the level of the equations of motion. Note, here F5 =
∂C4 + 43 B2 ∂C2 − 34 C2 ∂B2 , F5+ = ∂C4 , H = ∂B2 , H̃ = ∂C2 , and C0 = `. For the metric,
we have
 
9 ρσ 1 2Φ 1 2
Rµν = 2∇µ ∂ν Φ − Hµ Hνρσ − e ∂µ `∂ν ` − gµν (∂`)
4 2 2
 
9 2Φ e ρσ e 1 e − `H) + 25 e2Φ Fµρσλκ Fνρσλκ (3.1)
2
+ e (H − `H)(µ (H − `H)ν)ρσ − gµν (H
4 6 6

the gauge fields give,

10
∇µ (`2 + e−2Φ )∂B2 − `∂C2 µσλ = Fµνρσλ (∂C2 )µνρ
 
(3.2)
3
10
∇µ [∂C2 − `∂B2 ]µσλ = − Fµνρσλ (∂B2 )µνρ (3.3)
3
F5+ = ∗F5+ (3.4)
and finally, the varying the scalars yields,
3
∇2 ` = − ∂B2 (∂C2 − `∂B2 ) (3.5)
2
1 3
∇2 Φ = (∂Φ)2 + RG + (∂B2 )2 (3.6)
4 16
Type IIB supergravity action:

√
Z    
10 −2Φ 2 3 2 1 2 3 2 5 2
SIIB = d x −g e −R + 4(∂Φ) − (∂B2 ) − (∂`) − (∂C2 − `∂B2 ) − F5 +SCS
4 2 4 6
(3.7)
where SCS is the Chern-Simons part of the action.
We now turn to the subject of black-brane solutions to the Type IIB theory. Recall that
the symmetry for BPS D-branes is given by the broken Lorentz symmetry SO(d, 1) →

25
SO(d − p) × SO(p, 1). The factors describe the symmetry transverse and parallel to
the brane respectively, and since we also have translational symmetry along the brane,
the Lorentz group can be upgraded to a Poincaré4 symmetry. We will also have some
preserved supersymmetry which we will discuss later. For now, note that the solutions we
will look at are analogues of the extremal Reissner-Nordström solution1 and we will not
be discussing non-BPS systems. We seek to verify the Dp-brane solutions table 3 (where
|x⊥ | = r): These solutions were proposed in [30]. We see that the dilaton is constant only

−1/2 1/2
ds2 = Hp (r)dx2k + Hp (r)dx2⊥
(3−p)/4
eΦ = gs Hp
C01...p = −(1 − Hp−1 )
r
7−p
Hp = 1 + rp

Table 3.1: Dp-brane solutions

for p = 3. The constant one in the harmonic function, Hp , is the asymptotically flat part
of the geometry. We can see that, for p < 3, as r → 0 the coupling becomes large such
that the system is now in the nonpertubative regime, and the solution is not reliable. The
near horizon geometry for the D3-brane is AdS5 × S 5 . The singularity and the horizon for
solutions p 6= 3 is at r = 0. This may seem like a naked singularity, but we demand that
causal geodesics are not able to hit the horizon in finite affine parameter; however, for
p = 6 this is actually a naked singularity. For p = 3, examining the curvature invariants
gives that the spacetime everywhere nonsingular, so that there are two asymptotically flat
regions separated by the horizon.
For black Dp-brane solutions, we need only retain the R-R field Cp+1 with field strength
Fp+2 . The NS-NS two-form is also dropped. We are left with the following equations of
motion for the only three cases we need to verify, since the others are related by dualising:
p = 3 brane:
25
Rµν = Fµαβσρ Fναβσρ (3.8)
6
F5 = ∗F5 (3.9)
RG = 0 (3.10)

p = 1 brane:
 
9 1
Rµν = 2∇µ ∂ν Φ − e2Φ Gµν H 2
4 6
RG
∇2 Φ = (∂Φ)2 + (3.11)
4
∇µ H = 0
1
One reason for this is that in the extremal case we need not worry about Hawking radiation which
causes quantum mechanical instabilities for the black p-brane solution.

26
p = −1 brane2 :

e2Φ
 
1
Rµν = 2∇µ ∂ν Φ − ∂µ `∂ν ` − Gµν (∂`)2 (3.12)
2 2
∇2 ` = 0

We will concentrate on the p = 3 brane and remark that the verification follows similarly
for the other cases. To compute the Ricci we use the vielbein:

ea = Hp−1/4 dxa a = 0, ..., p

eA = Hp1/4 dxA A = p + 1, ..., 9
1 1
−→ dea = − Hp−5/4 (∂C Hp ) dxC ∧ dxa = − Hp−5/4 (∂C Hp ) eC ∧ ea
4 4
1 1
deA = Hp−3/4 (∂A Hp ) dxC ∧ dxA = Hp−5/4 (∂C Hp ) eC ∧ eA
4 4
Cartan’s first equation reads
deµ̃ = −wµ̃ν̃ ∧ eν̃ (3.13)
where µ̃ is the vielbein index from 0 to 9. So

dea = −(wab ∧ eb + waB ∧ eB )

deA = −(wAb ∧ eb + wAB ∧ eB )

suggests only non-zero components: (wab = 0 by antisymmetry)

1
waB = − Hp−3/2 (∂ B Hp ) dxa
4
Ba
:= −w
1
wAB = Hp−1 ∂ [B Hp dxA]
2
1
−→ dw B = − ∂C (Hp−3/2 ∂B Hp ) dxC ∧ dxa
a
4
1
dwAB = δBD ∂C (Hp−1 ∂ [D| Hp )dxC ∧ dx|A]
2
Cartan’s second equation for the curvature two-form is

Rµ̃ν̃ = dwµ̃ν̃ + wµ̃ρ̃ ∧ wρ̃ν̃ (3.14)

1
= Rµ̃ν̃ σ̃ρ̃ eσ̃ ∧ eρ̃ (3.15)
2
2
see Ortin section 19.2.7 for a full discussion of the D(-1)-brane solution

27
So we have the components:
1 −3
Rab = waC ∧ wCb = − H ∂C Hp ∂ C Hp dxa ∧ dxb
16 p
1 −3
=− H (∂Hp )2 dxa ∧ dxb
16 p

RAB = dwAB + wAC ∧ wCB

1 1
= δBD ∂C (Hp−1 ∂ [D| Hp ) dxC ∧ dx|A] + Hp−2 δBD δCE ∂ [E Hp dxA] ∧ ∂ [D Hp dxC]
2 4

RaB = dwaB + waC ∧ wCB

1 1
= − ∂C (Hp−3/2 ∂B Hp ) dxC ∧ dxa − δBD Hp−5/2 ∂C Hp dxa ∧ ∂ [D Hp dxC]
4 8

RAb = dwAb + wAC ∧ wCb

1 1
= ∂C (Hp−3/2 ∂ A Hp ) dxC ∧ dxb + Hp−5/2 ∂ [C Hp dxA] ∧ ∂C Hp dxb
4 8
using equation (3.15), we can read off the Riemann tensor components:
1 −5/2
Rabcd = −(δca ηbd − δda ηbc )
H ∂B Hp ∂ B Hp
16 p
1 1 1
RAbCd = ηbd ∂C (Hp−3/2 ∂ A Hp ) + ηbd δC A −5/2
Hp ∂E Hp ∂ E Hp − ηbd Hp−5/2 ∂C Hp ∂ A Hp
4 16 16
1 1 1
RaBCd = − δda ∂C (Hp−3/2 ∂B H) + δda Hp−5/2 ∂C Hp ∂B Hp − δBC δda Hp−5/2 ∂E Hp ∂ E Hp
4 16 16
A 1 A −1/2 −1 1 −1/2 −1 A
R BCD = δ[D Hp ∂C] (Hp ∂B Hp ) − δB[D Hp ∂C] (Hp ∂ Hp )
2 2
1 A −5/2 1
− δ[D Hp ∂C] Hp ∂B Hp + δB[D Hp−5/2 ∂C] Hp ∂ A Hp
8 8
1 A A −5/2
− (δC δBD − δD δBC )Hp ∂E Hp ∂ E Hp
16
and so the Ricci:
2(p − 1) −5/2 1
Rbd = −ηbd Hp ∂B Hp ∂ B Hp + ηbd Hp−3/2 ∂C ∂ C Hp (3.16)
16 4

2(p − 1) −5/2 28 − 12p −5/2

RBD = δBD Hp ∂C Hp ∂ C Hp + Hp ∂D Hp ∂B Hp
16 16
1 p − 3 −3/2
− δBD Hp−3/2 ∂C ∂ C Hp + Hp ∂ B ∂ D Hp (3.17)
4 2
Finally, the Ricci scalar is
−(p − 3)(p + 1) −5/2 2p − 7 −3/2
R = Rµ̃µ̃ = Raa +RAA = Hp ∂ B Hp ∂ B Hp + Hp ∂B ∂ B Hp (3.18)
4 2

28
 −3/2
The case of p = 3 reduces to R = − 12 H3 ∂B ∂ B H3 which indeed vanishes for Hp har-
monic.

For the field strength we have, by symmetry (Poincaré4 × SO(6)), the only surviving
components are FXuvwz and FU V W XY . We can write
1
FXuvwz = εuvwz F̃X dxX ∧ dxu ∧ dxv ∧ dxw ∧ dxz
4!
1
= εuvwz εuvwz F̃X dxX ∧ dx0 ∧ dx1 ∧ dx2 ∧ dx3
4!
4!
= F̃X dxX ∧ dx0 ∧ dx1 ∧ dx2 ∧ dx3
4!
= F̃X dxX ∧ (w.s) vol. form

Bianchi identity dF = 0 gives:

dF̃ = 0 ⇒ F̃ = dK
where K = K(x⊥ ) is a scalar. So,
1 X
∗F =  F̃X dxM ∧ dxN ∧ dxQ ∧ dxS ∧ dxT
5! M N QST
1
=√ εXU V W Y Z gU M gV N gW Q gY S gZT F̃X dxM ∧ dxN ∧ dxQ ∧ dxS ∧ dxT
−g5!
1
= 1/2 εXU V W Y Z Hp5/2 F̃X dxU ∧ dxV ∧ dxW ∧ dxY ∧ dxZ
Hp 5!
1
= εXU V W Y Z Hp2 F̃X dxU ∧ dxV ∧ dxW ∧ dxY ∧ dxZ
5!
By the self duality of the five-form field strength, we have d ∗ F = 0, so:
1 XABCDE
d∗F = ε d(Hp2 ∂X K) ∧ dxA ∧ dxB ∧ dxC ∧ dxD ∧ dxE
5!
1
= εXABCDE ∂Y (Hp2 ∂X K) dxY ∧ dxA ∧ dxB ∧ dxC ∧ dxD ∧ dxE
5!

Using a harmonic ansatz K(r) = (Hp )q we see d ∗ F = 0 for q = −1. Hence we have:

F = dHp−1 ∧ (w.s.) vol. form

Finally, we show that (3.8) is satisfied.

Fap1 p2 p3 M Fbp1 p2 p3 M ∼ (H −1/4 · H · H −2 )2 ηbd εac1 c2 c3 εdc1 c2 c3 ∂ C H∂C H (3.19)

−5/2 C
∼ ηab H ∂ H∂C H (3.20)

Since the field strength, F , is only determined up to some constant factor, we needn’t be
precise. Then
Rab − Fac1 c2 c3 C Fbc1 c2 c3 C ∼ ηab ∂ 2 H (3.21)

29
and again for H harmonic, our equations are satisfied.

RAB FAM1 M2 M3 M4 FBM1 M2 M3 M4 + FAm1 m2 m3 m4 FBm1 m2 m3 m4 (3.22)

−1/4
∼ (H · H · H −2 )2 X... Y ...
A ∂X HB ∂Y H + H −5/2
∂A H∂B H (3.23)
∼ δAB H 5/2 ∂C H∂ C H + H −5/2 ∂A H∂B H (3.24)

so we see (R − F F ) ∼ ∂∂H as required. The fact that the same function H appears in
the field strength as well as both ‘parts’ of the metric is no coincidence. We now show
that this is in fact necessary- for the particular case of the three-brane, but others follow
similarly- due to supersymmetry. We will be following [33] closely. We start by the making
the following ansatz

xµ = {xm , y M } Poincaré4 × SO(6) split (3.25)

2 2X m n 2Y M N
ds = e ηmn dx dx + e δM N dy dy (3.26)
(4) 1 (4)
Cabcd = −√ abcd eZ C0123 = −eZ (3.27)
−gq
p
Here X, Y, Z depend only on y M , and by SO(6) symmetry, only on r = δM N y M y N .
Similarly φ = φ(r) is the dilaton ansatz. Note also the determinant of the ‘parallel’ space
metric gq .
Supersymmetry preservation requires that there exists Killing spinor κ satisfying
 
˜ i µ1 ...µ5
δψµ = ∇µ + 2 Γ Γµ Fµ1 ...µ5 κ = 0 (3.28)
4 · 5!
δλ = iΓµ κ∗ Pµ = 0 (3.29)

Here Pµ = (∂µ Φ)(1 − Φ∗ Φ)−1 and ∇ ˜ µ = ∂µ + 1 ωµα̃β̃ Γ . Here the vielbein indices are
4 α̃β̃
given, using the same split, as α̃ = {a, A}. The vielbeins and spin connection are, (similar
to our previous calculation)

ema = eX δm
a A
eM = eY δM
A
(3.30)
(ωm )aB =e X−Y a
∂B Xδm (ωM )BA = 2∂[B Y δA]M (3.31)

A basis for the gamma matrices preserving our symmetry is

Γµ̃ = {γa ⊗ 1, γ5 ⊗ ΣA }

where Γµ = eµν̃ Γν̃ , Γµ1 ...µn = Γ[µ1 Γµ2 . . . Γµn] , where anitsymmetrisation is taken with
weight unity and γa and ΣA are the Dirac matrices for d = 4 and d = 6 respectively. We
will also need
γ5 = iγ 0 γ 1 γ 2 γ 3 and Γ7 = −iΣ4 Σ5 Σ6 Σ7 Σ8 Σ9
so γ52 = 1 = Γ27 . The most general spinor respecting SO(1, 3) × SO(6) is written

κ(x, y) = ζ ⊗ η

30
where ζ is a constant spinor and η = η(r) for SO(1, 3) and SO(6) respectively. We
decompose these into their chiral eigenstates using (1±γ5 ) and (1±Γ7 ), again respectively.
Note also the chirality condition
Γ11 κ = κ Γ11 =: γ5 ⊗ Γ7
hence the chiralities are correlated. Let us plug these facts into our transformation rule,
and also note the identities
{Γa , ΓA } = 0 (3.32)
Γµ1 ...µn α = Γµ1 ...µn Γα − nΓ[µ1 ...µn−1 η µn ]α (3.33)
1
Γµ1 ...µn = − (−1)n(n−1)/2 µ1 ...µ10 Γµn+1 ...µ10 Γ11 (3.34)
(10 − n)!
hence,
1 i
δψm = (∂m + ωmaB (Γa ΓB − ΓB Γa )κ + 2 (Γn1 n2 n3 n4 P Γm Fn1 n2 n3 n4 P )κ
4 4 · 5!
i
+ 2 (ΓN1 N2 N3 N4 N5 Γm FN1 N2 N3 N4 N5 )κ (3.35)
4 · 5!
Take this expression term by term:
1 aB 1 1 1
ω (Γa ΓB − ΓB Γa ) = (ωm )aB Γa ΓB = eX−Y ∂B Xδm a
Γa ΓB = eX−Y ∂M XδB
M a
δm Γa ΓB
4 m 2 2 2
1
= eX−Y ∂M X(e−X eY eam eM
B )Γa Γ
B
2
1
= γa ⊗ ΣM γ5 ∂M X
2
where the vielbeins have been inserted to convert to world indices. Next consider the term
i i
2
(Γn1 n2 n3 n4 P Γm Fn1 n2 n3 n4 P ) = 2 Γn1 n2 n3 n4 ΓP Γm Fn1 n2 n3 n4 P
4 · 5! 4 · 5!
5!
=− 2 iΓ0 Γ1 Γ2 Γ3 ΓP Γm gq−1 ∂P eZ
4 · 5!
1
= − γ52 ΣP γm gq−1 ∂P eZ
16
1
= − ΣP γm gq−1 ∂P eZ (3.36)
16
and the final term goes as
i i
ΓN1 N2 N3 N4 N5 Γm FN1 N2 N3 N4 N5 = − 2 n1 ...n4 N1 ...N6 Γn1 ...n4 N6 Γ11 Γm (?FN1 ...N5 )
42 · 5! 4 · 5!5!
1
= − ΣP γm gq−1 ∂P eZ (3.37)
16
where the last line is obtained by playing with epsilon identities and reducing to the
previous term. Altogether,
1 1
δψm = ∂m κ + γm ⊗ ΣM (γ5 ∂M X − e−4X ∂M eZ )κ (3.38)
2 4

31
The next component we need to consider is
1 i
δψM = (∂M + ωMAB ΓAB )κ+ 2 (ΓP n1 n2 n3 n4 ΓM FP n1 n2 n3 n4 +ΓN1 N2 N3 N4 N5 ΓM FN1 N2 N3 N4 N5 )κ
4 4 · 5!
(3.39)
This essentially goes as before, but we use ΓAB = 2ηAB − ΓBA in the (ωM )AB ΓAB term.
We arrive at
1 1 1
δψM = ∂M κ + ∂M Y κ − γ5 ⊗ ΣP ΣM (γ5 ∂P Y + e−4X ∂P eZ )κ (3.40)
2 2 4
Let us analyse our results. Starting with

δλ = iΓµ κ∗ Pµ = 0 (3.41)

we see that this is satisfied for Φ = constant. In equation (3.38), the derivative on κ
vanishes by independence from the parallel directions. Using (1 − γ5 )ζ = 0 we see that
Z = 4A. Finally, equation (3.40) is solved by Y = −X, using (1 − Γ7 )η0 = 0, where η0
is a constant spinor and ∂κ = 21 ∂Y κ. The final solution therefore is κ = eX/2 ζ ⊗ η0 . We
have verified that in fact eX/2 = H 1/8 so that in the end we have

κ = H 1/8 ζ0 ⊗ ηo (3.42)

We can write
7−p
√
  
 r 7−p
p rp 7−p
Hp = 1 + , = (2 π)5−p Γ gs Np (3.43)
r ls 2

Also for the field strength,

Fp+2 = Q ? ω8−p (3.44)
so that integrating over an (8 − p)-sphere gives Q as the charge. But for the p = 3 case
we need to incorporate duality, so

F5 = Q(ω5 + ?ω5 ) (3.45)

32
Chapter 4

Black Holes in String Theory

4.1 Entropy Counting

The entropy counting of black holes, i.e. a statistical mechanical counting of the mi-
crostates giving rise to the entropy, in string theory has been one of the main successes
to date for this particular programme for quantizing gravity, first achieved by Strominger
and Vafa in [34] following earlier papers examining the subject such as Susskind [36], Sen
[37] and others. The calculation was for an extreme, non-rotating five dimensional black
hole and was simplified by Callan and Maldacena in [35]. Following this breakthrough,
calculations were performed for non-extremal [39], spinning [38], four dimensional [40] and
far-from-extremal [41] black holes. Let us discuss in some more detail how black holes have
been treated in string theory.

As early back as the ’60’s, people were considering the idea of black holes as funda-
mental strings. This approach was made more robust by Sen and Susskind, as we’ve
already mentioned, and led to what is known as the correspondence principle, formulated
by Horowitz and Polchinski in [41]. The principle asserts that since we have an infinite
tower of massive states for the quantized string, and since for a Schwarzschild black hole,
the radius is proportional to the mass of the hole, then for large enough black hole masses-
enough that the radius is larger than the string scale- there will be a string state with
equivalent mass such that its length scale will be smaller than its Schwarzschild radius,
thus forming a black hole. (Recall in four dimensions G ∼ gs2 ls2 so that G increases as
we turn up the string coupling, and spacetime becomes flat by in the null limit). If we
consider here the low energy effective action, then we know the fundamental string couples
to the NS gauge field and is electrically charged. Susskind proposed this should hold the
other way: black holes at weak coupling should be described by strings. The transition
point is known as the ‘correspondence point.’ Should this be so, the entropies of the two
pictures should be comparable. In fact this turns out to be the case, but works modulo
exact coefficients.

Another approach, that we have been alluding to up till now, is given by exploiting the
supersymmetry of string theory. The advantage here is that we need not renormalise mass
since we use the BPS property of the states, the degeneracy of which does not change at
least for ≥ 16 supercharges, so that we compute this degeneracy in the weak coupling
limit whence we find Smicro (microscopic entropy) using the logarithm and then turn up

33
the coupling so that the state forms a black hole and we can use the Bekenstein-Hawking
entropy formula. The focus is then shifted from strings to the nonperturbative solitonic
states coupling to the RR sector, the Dp-branes we’ve been looking at, since they are BPS
preserving 1/2 supersymmetry, valid for small curvature1 . We do not need the analogue
of the correspondence point since the mass is directly related to the charge. When we
compactify the theory down to d-dimensions, and wrap the brane around the compactfied
directions, the system will appear as a d-dimensional pointlike object.
Finally note that D-branes aren’t the end of the story: M-theory uses its symmetry
to map between black hole systems and thereby provides another method for entropy
calculations. In addition there have been recent developments in the subject such as the
attractor mechanism and AdS/CFT correspondence.
Let us expand further on this BPS (supersymmetric) approach by seeing how it works.
First off, how do we put BPS branes together so that we can make a black hole. This
is given by the ‘intersection rules’, as is known in the literature. For supersymmetric
intersections, we will get an extremal black hole. Here, we will only be looking at Dp-
branes, and supersymmetry along with duality relations give (for p ≤ 6, and restriciting
to pairwise intersections)2 :

Dm k Dm + 4(m), m = 0, 1, 2 −→ Dp ⊥ Dp0 (m), p + p0 = 4 + 2m, W k Dp (4.1)

where the brackets tell us the dimension of the intersection. In this section we will study
the case of the d = 5 black hole with three charges. For d > 5, the black hole horizon
has a non-zero radius only by including higher order corrections to the curvature tensor
in Einstein-Hilbert action. We choose d = 5 here as it is the simplest example to consider,
and take the approach described by [35]. The system we consider is D1 − D5 − P P , where
P P is a gravitational wave on D1.
We begin by compactification of Type IIB on a T 5 so that the noncompact directions
are (x0 , . . . , x4 ), then wrap the D5 on the whole of T 5 ; the D1 on x5 with momentum
in this direction also. This system preserves 1/8 of the supersymmetries of the vacuum.
Harmonic function rule, gives ansatz for metric, p+1 form potential, and dialton (based on
solutions describing the extreme branes independently), proposed in for example [42][43].
It goes as

1. take individual solutions, e.g. table 3, smear over relative transverse directions

2. metric remains diagonal and is a superposition of the (metric) solutions

3. do the same to the dilaton

4. sum for the form potentials

A system for d=5 is given by D1-D5-W: regular. To see how the harmonic function
rule works, it is instructive to first ignore the gravitational wave and concentrate on the
1
small compared to 1/ls in the string-frame. This is because the low energy apporximation is only
valid here.
2
they can intersect at angles to, but this is more complicated.

34
 0 1 2 3 4 5 6 7 8 9
D1 − − ∼ ∼ ∼ ∼ · · · ·
D5 − − − − − − · · · ·
W − → ∼ ∼ ∼ ∼ · · · ·

Table 4.1: D1-D5-PP system: the symbols −, ·, ∼ represent extended, pointlike and
smeared directions of the brane, whilst → indicates the direction in which the wave moves.

D1-D5 system. Smearing the dependence in some directions by harmonic rule to constrain
the dependence of the solution.
ds2 = HD1 (r)−1/2 HD5 (r)−1/2 (−dt2 + dx21 ) + HD1 (r)1/2 HD5 (r)1/2 (dr2 + dΩ23 )
4
X
+ HD1 (r)1/2 HD5 (r)−1/2 dxk dxk (4.2)
k=2

and dilaton
HD5
e−2Φ = (4.3)
HD1
The black hole just constructed has zero entropy, since after compactifying, we can com-
pute its area (in a calculation like the one below) and find
 r 
rD1 rD5
A = lim 2π r3
2
=0 (4.4)
r→0 r4
So in order to get a solution with non-zero horizon we need a BPS superposition with
another object. The solution is the D1-D5-PP described above. To put a wave on the
system, we boost the solution to obtain
ds2 = −λ−2/3 dt2 + λ1/3 (dr2 + r2 dΩ23 ) (4.5)
 r 2
i
Y
λ= Hi Hi = 1 + , i = D1, D5, W
r
i
Note that this reduces to the d = 5 RN black hole in the case rD1 = rD5 = rW , although
this equality is not attained in general. Using arraying, we can write3

2 gs ND1 `6s 2 gs2 NW `8s

rD1 = , rD5 = gs ND5 `2s , 2
rW = (4.6)
V R2 V
For an arbitrary black hole in d dimensions the entropy formula is (note conventions
~ = kB = c = 1)
Ad
S=
4Gd
Newton’s constant in d < 10 dimensions is obtained using the ten dimensional one, as (in
the case of d = 5)
πgs2 `8s √
 
G10
Gd = 10−d
= , G10 = ( 2π)6 gs2 `8s
(2π) V10−d d=5 4RV
3
As we remarked earlier, in order for the curvature invariants to be small, we need rD1,D5,W  ls .

35
The area of the d = 5 dimensional black hole is given by the 3-sphere volume = 2π 2 3
ri 2


r .
ri 2
The horizon of this black hole is r = 0. Close to this limit, Hi = 1 + r → r we
have for the angular piece of the metric
 2 r 2 r 2 1/3
rD1
1/3 2
lim λ r dΩ23 = D5 W
r2 dΩ23
r→0 r6
2 2 2 1/3
dΩ23


= rD1 rD5 rW

2 r 2 r 2 )1/6 , whence we compute

Hence we can read off the horizon radius as rH = (rD1 D5 W
the entropy as

A5 2π 2 rH 3 2 r 2 r 2 )1/2
2πV R(rD1 D5 W
S= = =
4G5 4(πgs2 `8s /4RV ) gs2 `8s
1/2
2πV R ND1 ND5 NW gs4 `16

s
= 2 8
g s `s V 2 R2
p
= 2π ND1 ND5 NW

The ADM mass can be written M = MD1 + MD2 + MW due to the BPS condition
where charges are additive, and can also be made apparent by expanding the coefficient
of dt2 in the limit r → ∞
  r 2 −2/3 2  ri 2
  2
2 rD1 2
rD5 2 
rW
−2/3 i
Hi = 1+ = 1− + . . . =⇒ gtt ∼ − 1 − + +
r 3 r 3 r2 r2 r2
πri2
and Mi = 4G5 (c.f. RN BH) so that

ND1 R ND5 RV NW
M= 2
+ 6
+
gs `s gs `s R

We now need to count the states of the system described, i.e. give a statistical deriva-
tion of the entropy. To do this in detail is beyond the scope of this dissertation, however
we will outline the approach given in [35]. Recall that the configuration of the N = 1,
d = 5- preserving 4 of the 32 type IIB supercharges- system we are studying: Type IIB
theory on T 5 = T 4 × S 1 such that ND5 D5 branes are wrapped on the whole of T 5 ; ND1
D1 branes wrap the S 1 of length 2πR and momentum NW /R is carried along the S 1 . The
system is in a bound state with zero binding energy. Recall that we add a gravitational
wave since the D1-D5 system itself has zero entropy, probably because it’s in the ground
state, thus we seek excited BPS states realised by adding a gravity wave to the D1-brane.
More specifically, we consider plane fronted gravitational waves with parallel rays (pp-
waves), which carry either left or right moving momentum and are 1/2 BPS states. In
the world volume theory right movers are in ground state, left movers carry NW modes.
The entropy then counts how many states this momentum can be distributed between
excitations of the system. We look for massless excitations since the wave moves at the
speed of light, and otherwise the BPS mass formula is violated.

36
 We have not yet identified the degrees of freedom carrying the momentum, so we first
do this. Since, our symmetry group is SO(1, 1) × SOk (4) × SO⊥ (4), we cannot carry
momenta on the parallel directions of the (rigid) D-branes. However, excitations on the
brane are described by massless open strings, bosonic and fermionic, so that we can make
them carry the momentum. Several types of open strings can feature: (1,1) i.e. open
strings from D1 to D1, and similarly (5,5), (1,5) and (5,1). The last two are distinguished
by orientation of the string. The total momentum, NW /R is carried by bosonic and
fermionic strings in quanta of 1/R.
The maximum number of massless strings will give us the highest entropy, but exciting
some causes others to become massive. Now, take R large and dimensionally reduce on T 4
so that the theory is 1+1 dimensional and the energy carried by individual excitations is
small. The theory is (4,4) superconformal, i.e. four left and four right moving supersym-
metric generators. The (1,1) and (5,5 )strings are gauge bosons of the U (N1 ) and U (N5 )
gauge groups. Similarly, the (1,5) and (5,1) strings are fundamental and antifundamental
of U (N1 ) × U (N5 ). We are interested in the IR limit Higgs branch, as opposed to the
Coulomb branch of the theory, essentially because the latter does not allow the bound
state we’re after. Here, we can drop the (1,1) and (5,5) strings from our counting since
the Higgs field makes the vector multiplets describing them massive. Counting the number
of massless degrees of freedom now gives 4Q1 Q5 bosonic and 4Q1 Q5 fermionic, and so the
total central charge, characterising the conformal theory is given by
1 1
c = nbose + nfermi = (1 + )4ND1 ND5 (4.7)
2 2
where bosons contribute one and fermions a half. The degeneracy is given by Cardy’s
formula r  r 
1 c
d(c, NW ) ∼ exp πcEL = exp 2π ER (4.8)
3 6
and plugging the total energy E = NW /R and volume L = 2πR in gives
p
Smicro = log d(c, NW ) = 2π ND1 ND5 NW (4.9)

37
Bibliography

[1] S.M. Carroll, “Lecture Notes on General Relativity”, gr-qc/9712019.

[2] P.K. Townsend, “Black Holes”, gr-qc/9707012.

[3] N. Lambert, “Advanced General Relativity”, Lecture notes, Dept. of Mathematics,

King’s College London.

[4] E. Papantonopoulos, “Physics of Black Holes: A Guided Tour”, Lecture Notes in

Physics 769, Springer, Berlin, 2009.

[5] S. Hawking, “Black Holes in General Relativity”, Commun. Math. Phys. 25, 152-166,
(1972).

[6] S. Hawking, “Particle Creation by Black Holes”, Commun. Math. Phys. 43, 199-220,
(1975)

[7] S. Hawking and G. Ellis, “The large scale structure of space-time”, Cambridge Uni-
versity Press, 1973.

[8] R.M. Wald, “General Relativity”, The University of Chicago Press, 1974.

[9] R.M. Wald, “The ‘Nernst Theorem’ and Black Hole Thermodynamics”, gr-qc/9704008.

[10] C. Eling and J.D. Bekenstein, “Challenging the generalized second law”, gr-
qc/0810.5255.

[11] J.B. Bekenstein,“Do we understand black hole entropy?”gr-qc/9409015

[12] G. Compère, “An introduction to the mechanics of black holes”, gr-qc/0611129.

[13] L.H. Ford, “Quantum Field Theory in Curved Spacetime”, gr-qc/9707062.

[14] T. Jacobson, “Introduction to Quantum Fields in Curved Spacetime and the Hawking
Effect”, gr-qc/0308048.

[15] S.F. Ross, “Black Hole Thermodynamics”, hep-th/0502195

[16] P. Duffell, “Quantum Field Theory in Curved Spacetime”, New York University, 2009.

[17] C. Krishnan, “Quantum Field Theory, Black Holes and Holography”, hep-
th/1011.5875

38
[18] D. Biggati and L. Susskind, “TASI lectures on the Holographic Principle”, hep-
th/0002044

[19] S.D. Mathur, “The information paradox: a pedagogical introduction”, hep-

th/0909.1038

[20] J. Preskill, “Do Black Holes Destroy Information?”, hep-th/9209058

[21] D. Page, “Black Hole Information”, hep-th/9305040

[22] L. Smolin, “The black hole information paradox and relative locality”, gr-
qc/1108.0910

[23] K. Skenderis and M. Taylor, “The fuzzball proposal for black holes”, hep-th/0804.0552

[24] S.D. Mathur,“The fuzzball proposal for black holes: an elementary review”, hep-
th/050205

[25] D. Tong, “Lectures on String Theory”, hep-th/0908.0333.

[26] R.J. Szabo, “An Introduction to String Theory and D-Brane Dynamics”, Imperial
College Press, London, 2004.

[27] K. Becker, M. Becker, J.H. Schwarz, “String Theory and M-Theory: A Modern
Introduction”, Cambridge University Press, 2007.

[28] T. Ortı́n, “Gravity and Strings”, Cambridge University Press, 2004.

[29] E. Bergshoeff, H.J. Boonstra and T. Ortı́n, hep-th/9508091.

[30] G.T. Horowitz and A. Strominger, “Black strings and p-branes”, Nucl. Phys. B360,
197 (1991).

[31] A.W. Peet, “TASI lectures on Black Holes in String Theory”, hep-th/0008241

[32] A.W. Peet, “The Bekenstein Formula and String Theory (N-brane Theory)”, hep-
th/9712253

[33] M.J. Duff and J.X. Lu, “THE SELF-DUAL TYPE IIB SUPERTHREEBRANE”,
Nucl. Phys. B357 (1991) 534

[34] A. Strominger and C. Vafa, “Microscopic Orgin of the Bekenstein-Hawking Entropy”,

hep-th/9601029

[35] C.G. Callan and J.M Maldacena, “D-brane Approach to Black Hole Quantum Me-
chanics”, hep-th/9602043

[36] L. Susskind, “Some Speculations about Black Hole Entropy in String Theory” , hep-
th/9309145

[37] A. Sen, “EXTREMAL BLACK HOLES AND ELEMENTARY STRING STATES”,

hep-th/9504147

39
[38] J.C. Breckenridge, R.C. Myers, A.W. Peet and C. Vafa, “D-branes and Spinning
Black Holes”, hep-th/9602065

[39] Some relevant papers include:

C.G. Callan and J.M Maldacena, “D-brane Approach to Black Hole Quantum Me-
chanics”, hep-th/9602043.
G.T. Horowitz and A. Strominger, “Counting States of Near-Extremal Black Holes”,
hep-th/9602051.
J.C. Breckenridge, D.A. Lowe, R.C. Myers, A.W. Peet, A. Strominger and C. Vafa,
“Macroscopic and Microscopic Entropy of Near-Extremal Spinning Black Holes”, hep-
th/9603078.

[40] See for example: J.M. Maldacena and A. Strominger, “Statistical Entropy of Four-
Dimensional Extremal Black Holes”, hep-th/9603060

[41] G.T. Horowitz and J. Polchinski, “A Correspondence Principle for Black Holes and
Strings”, hep-th/9612146

[42] A.A. Tseytlin, “Harmonic superpositions of M-branes”, hep-th/9604035

[43] J.P. Gauntlett, D.A. Kastor, J. Traschen, “Overlapping Branes in M-Theory”, hep-
th/9604179

[44] G.T. Horowitz, “The Origin of Black Hole Entropy in String Theory”, gr-qc/9604051

[45] G.T. Horowitz, “Quantum States of Black Holes”, gr-qc/9704072

[46] J.M. Maldacena, “Black holes and D-branes”, hep-th/9705078

[47] T. Mohaupt, “Black Holes in Supergravity and String Theory”, hep-th/0004098

[48] D. Klemm, “Black holes and singularities in string theory”, hep-th/0410040

[49] K. Skenderis, “Black holes and branes in string theory”, hep-th/9901050

[50] S.R. Das and S.D. Mathur, “The Quantum Physics of Black Holes: Results from
String Theory”, gr-qc/0105063

[51] A. Sen, “An Introduction to Non-perturbative String Theory”, hep-th/9802051

[52] K.S. Stelle, “BPS Branes in Supergravity”, hep-th/9803116

40