The Reissner-Nordström Metric: Jonatan Nordebo March 16, 2016
The Reissner-Nordström Metric: Jonatan Nordebo March 16, 2016
Jonatan Nordebo
Abstract
A brief review of special and general relativity including some classi-
cal electrodynamics is given. We then present a detailed derivation of
the Reissner-Nordström metric. The derivation is done by solving the
Einstein-Maxwell equations for a spherically symmetric electrically
charged body. The physics of this spacetime is then studied. This
includes gravitational time dilation and redshift, equations of motion
for both massive and massless non-charged particles derived from the
geodesic equation and equations of motion for a massive charged par-
ticle derived with lagrangian formalism. Finally, a quick discussion of
the properties of a Reissner-Nordström black hole is given.
1
Contents
1 Introduction 3
2
1 Introduction
In 1915 Einstein completed his general theory of relativity. It did not take
long before the first non-trivial exact solution for the Einstein field equa-
tions was found by Karl Schwarzschild in 1916 which corresponds to the
gravitational field of a spherically symmetric object [1]. In the same year
Hans Reissner generalized Schwarzschild’s solution to include an electri-
cally charged object [2]. Gunnar Nordström later (independently of Reiss-
ner) arrived at the same solution [3] now known as the Reissner-Nordström
metric.
The Reissner-Nordström metric is a famous solution to the Einstein
field equations. It describes the spacetime geometry around a spherically
non-rotating charged body. The universe at large appear to be electrically
neutral, so it is highly unlikely to find a macroscopic object that possess
a considerable amount of net charge. The Reissner-Nordström solution is
therefore not relevant to realistic situations in astrophysics. It does however
contribute in understanding the fundamental nature of space and time.
Also, the Reissner-Nordström metric is a more general solution than the
Schwarzschild metric, so by simply putting the electrical charge to zero we
obtain the Schwarzschild solution (which has plenty of practical applica-
tions).
In this paper we will first give a short introduction to general relativ-
ity. Our main goal is then to present a detailed derivation of the Reissner-
Nordström metric (which is often overlooked in many textbooks) without
assuming a static spacetime. We then proceed to study some of the physics
in this spacetime such as:
3
(1) The laws of physics take the same form in all inertial reference frames.
where
1
γ≡q .
v2
1− c2
4
If we look at an infinitesimal spacetime interval (which of course also is an
invariant) we have
ds2 = c2 dt2 − dx2 − dy 2 − dz 2 .
Note that it is also possible to choose the interval to be
ds2 = −c2 dt2 + dx2 + dy 2 + dz 2 .
The choice of sign is arbitrary and has no physical implication as long as
one is consistent. Throughout this paper we will use the sign convention of
the former infinitesimal interval, i.e. (+, −, −, −).
If we let (x0 , x1 , x2 , x3 ) ≡ (ct, x, y, z) we can write the infinitesimal in-
terval as
X3 X 3
ds2 = ηαβ dxα dxβ ,
α=0 β=0
where ηαβ is the metric tensor. When we use a Cartesian coordinate system,
the metric tensor is given by
1 0 0 0
0 −1 0
0
ηαβ = ,
0 0 −1 0
0 0 0 −1
and is known as the Minkowski metric. If we also use Einstein’s summation
convention, which implies summation over an index that appears both as
a subscript and a superscript, we can write the infinitesimal interval as
3 X
X 3
2
ds = ηαβ dxα dxβ ≡ ηαβ dxα dxβ . (2.1)
α=0 β=0
5
2.1 4-vectors
In special relativity a 4-vector is an object with four components that trans-
form in a specific way under Lorentz transformations. More specifically,
the components of a contravariant 4-vector V α transforms according to
V α = Λαµ V µ
Vα = Λαµ Vµ .
V α = η αµ Vµ
Vα = ηαµ V µ .
η αµ ηµβ = δβα ,
Vα = (V0 , V1 , V2 , V3 ) = (V 0 , −V 1 , −V 2 , −V 3 ).
Raising and lowering indices does not only apply to vectors, but to tensors
of any rank.
With the metric tensor we can define an invariant scalar product for
4-vectors according to
V 2 = ηαβ V α V β = η αβ Vα Vβ = V α Vα .
6
Consider now a moving particle. The proper time τ (i.e. the time mea-
sured by a clock following the particle) is independent of the coordinate
system used. The infinitesimal interval is given by
ds 1p 2
dτ = = dt − dx2 − dy 2 − dz 2
cs c
2 2 2 r
1 dx dy dz v2 dt
= 1− − − = dt 1 − 2 = .
c dt dt dt c γ
dxα
α dt dt dr
u ≡ = c , = γ(c, v),
dτ dτ dτ dt
where m is the rest mass of the particle. Let F ≡ dp/dt be the 3-force. The
kinetic energy of the particle is then given by
Z Z Z Z
dp
Ek = F · dx = · dx = v · dp = v · d(mγv).
dt
Using integration by parts we obtain
Z Z
2 2 m
Ek = mγv − mγv · dv = mγv − γd(v 2 )
2
mc2
= mγv 2 + − E0 = γmc2 − E0 ,
γ
pα = (E/c , p).
7
Finally we will derive the so-called relativistic energy-momentum relation.
Since the inner product of 4-vectors is invariant, we must have that
pα pα = p0α p0α .
E 2 = p2 c2 + m2 c4 ,
F = q (E + v × B) .
∂ρ
+ ∇ · J = 0. (2.6)
∂t
This can easily be seen by considering a volume V with charge q. The net
current that flows into V is
ZZ ZZZ
dq
I= =− J · dS = − (∇ · J)dV,
dt
∂V V
8
where we used the divergence theorem in the last step. The derivative of q
can on the other hand be written as
ZZZ ZZZ
dq d ∂ρ
= ρdV = dV.
dt dt ∂t
V V
∂A
E = −∇Φ − (2.7)
∂t
B = ∇ × A. (2.8)
∂ ρ
∇2 Φ +
(∇ · A) = − (2.9)
∂t 0
1 ∂2A
1 ∂Φ
∇2 A − 2 2 − ∇ ∇ · A + 2 = −µ0 J. (2.10)
c ∂t c ∂t
Though these equations look a bit more complicated, we have reduced the
number of components to solve for from 6 to 4. Also, the potentials are
not physically meaningful quantities that can be measured; the electric and
magnetic field are. In other words, the potentials are not uniquely deter-
mined by Maxwell’s equations. We can therefore make the replacements
A → A + ∇ψ
Φ → Φ − ∂ψ/∂t,
and it will not affect the electric and magnetic field. For example, the elec-
tric field will not change since
∂A0
0 0 ∂ψ ∂A ∂
E = −∇φ − = − ∇φ − ∇ − + ∇ψ
∂t ∂t ∂t ∂t
∂A
= −∇φ − = E.
∂t
To be able to make such replacements is know as gauge freedom, and these
transformations are known as gauge transformations. An often used gauge
is the so-called Lorenz gauge, which allows one to choose A and Φ so that
they satisfy the condition
1 ∂Φ
∇·A+ = 0.
c2 ∂t
9
Using the d’Alembertian operator defined by ≡ ∇2 − c−2 ∂ 2 /∂t2 and the
Lorenz gauge, equation (2.9) and (2.10) can now be written as
Φ = −ρ/0
A = −µ0 J.
∂α Aα = 0.
The first one of these corresponds to equations (2.2) and (2.5) while the
second one is equivalent to (2.3) and (2.4).
The equation for charge conservation can also be written in a compact
way. We simply do this by taking the 4-divergence of the 4-current and set
it equal to zero:
1∂ ∂ρ
j α,α = , ∇ · (cρ, J) = + ∇ · J = 0.
c ∂t ∂t
10
This is exactly the same as equation (2.6).
Another object, which will be very central in general relativity, is the
electromagnetic stress-energy tensor given by [6]
αβ 1 1 αβ µν αµ β
T = η F Fµν − F F µ , (2.16)
µ0 4
which describes the flow of electromagnetic energy and momentum in space-
time. Properties for the stress-energy tensor is that it is symmetric (T αβ =
T βα ) and traceless:
α αβ 1 1 αβ µν αµ β
T ≡ T α = ηαβ T = ηαβ η F Fµν − ηαβ F F µ (2.17)
µ0 4
1 1 αβ µν αµ
= ηαβ η F Fµν − F Fαµ = 0, (2.18)
µ0 4
since ηαβ η αβ = η αα = 4.
Note: in this section we let greek indices have the range 1, 2, . . . , n. In later
sections where we apply this to general relativity we will let greek indices
11
have the range 0, 1, . . . , n − 1, where the 0-component corresponds to the
time-component.
A differentiable manifold is a specific type of manifold. It essentially
just means that we are able to do differential calculus on the manifold.
More specifically, the partial derivatives ∂x0α /∂xµ and ∂xα /∂x0µ exist. These
partial derivatives now specify the transformation between the two coor-
dinate systems:
∂x0α µ ∂xα 0µ
x0α = x , x α
= x .
∂xµ ∂x0µ
From now on we will only work with differentiable manifolds and we can
therefore use the two synonymously.
A contravariant vector transform as
∂x0α µ
V 0α = V . (3.1)
∂xµ
while a covariant vector transform as
∂xµ
Vα0 = Vµ . (3.2)
∂x0α
From the transformation rules of contravariant and covariant vectors we
can generalize this to a tensor of any rank. Consider the direct product
T αβ ≡ V α Uβ of one contravariant vector and one covariant vector. This
object will transform as
0α ν
0α ∂x µ ∂x ∂x0α ∂xν µ
T β = V Uν = T . (3.3)
∂xµ ∂x0β ∂xµ ∂x0β ν
g αβ gβγ = g αγ = δγα .
12
will have the signature (+, −, −, −). A manifold with a metric with only
positive signs is called Riemannian. A generalization of the Riemannian
manifold is the pseudo-Riemannian manifold that do not need to have a sig-
nature with only positive signs. The signature used in describing spacetime
is clearly not of only positive signs, and is therefore a pseudo-Riemannian
manifold.
The metric tensor can be used to raise and lower indices:
gαβ T µβν = T µαν . (3.4)
A tensor with only up-indices is called contravariant while a tensor with
only down-indices is called covariant. If it has at least one up and one
down index it is called mixed. Furthermore, a tensor with m up indices
and n down indices will be denoted as an (m, n)-tensor.
A summation over an upper and lower index is called a contraction and
will yield a new tensor of lower rank. For example, consider a tensor T αβ γ .
If we do a contraction of α with β we get a new tensor
Rγ ≡ T αα γ = gαβ T αβγ .
It is straightforward to show that Rγ satisfy the transformation law
∂x0γ µ
R0γ = R .
∂xµ
The sum or difference of two tensors is a tensor of the same type. This
can easily be seen if we look at a specific example of a sum between two
second rank tensors. Consider the object Rαβ ≡ S αβ + T αβ , where S and T
are tensors. It will transform as
∂xα ∂x0ν 0 µ 0µ ∂xα ∂x0ν 0 µ
Rαβ = S αβ + T αβ =
S ν + T ν ≡ R ν.
∂x0µ ∂xβ ∂x0µ ∂xβ
It is then obvious that this generalizes to a sum or difference between two
tensors of any rank. It is now easy to convince oneself that any equation be-
tween two tensors with the same upper and lower indices will be invariant
under coordinate transformations.
13
Because of the second term on the right hand side, ∂V α /∂xβ does not trans-
form as a tensor. A derivative that do yield another tensor when operating
on a tensor is the covariant derivative, denoted by ∇µ (or by use of a semi-
colon). The covariant derivative of a contravariant vector V α is defined
by
∇β V α ≡ V α;β ≡ ∂β V α + Γαβµ V µ , (3.5)
where Γαβµ is the Christoffel symbols (also known as affine connections) which
can be thought of as a "correction" term. We claimed that ∇β V α is a tensor,
so it must transform as
∂xµ ∂x0α
∇0β V 0α = ∇µ V ν .
∂x0β ∂xν
From this we can work out how the Christoffel symbols must transform.
The both sides in the above equation can be expanded with the use of equa-
tion (3.5), then we transform V 0 to V :
∇β Vα ≡ Vα;β = ∂β Vα − Γµαβ Vµ ,
14
general, changing the order of covariant differentiation changes the result.
Properties such as linearity
∇(T + S) = ∇T + ∇S
still holds.
If one require that the affine connection is torsion free, i.e. that it is sym-
metrical in its lower indices (Γαβγ = Γαγβ ), and that the covariant derivative
of the metric tensor is zero everywhere (a property called metric compati-
bility), then the affine connection is unique [8]. Properties that directly fol-
lows from this is that the covariant derivative of the inverse metric is zero
(∇γ g αβ = 0) and that it commutes with rasing and lowering indices:
1
Γλαβ = g λµ (∂α gβµ + ∂β gµα − ∂µ gαβ ). (3.9)
2
∇β Vα = ∂β Vα − Γµαβ Vµ .
15
The three terms in the above expression can be written as
∂γ (∇β Vα ) = ∂γ ∂β Vα − ∂γ Γµαβ Vµ − Γµαβ (∂γ Vµ )
Γµαγ (∇β Vµ ) = Γµαγ ∂β Vµ − Γνµβ Vν
16
In a similar way as we got equation (3.10), one can show that the fol-
lowing holds for a second rank covariant tensor:
With the use of the product rule and the above equation we obtain
Using the above equation together with (3.12) (but with ∇γ Vβ instead of
Tγβ ) one can show that the Riemann tensor must satisfy the relation [9]
From the cyclic identity of the Riemann tensor, it can be showed that the
Ricci tensor is symmetric. The trace of the Ricci tensor is known as the Ricci
scalar:
R ≡ Rαα = g αβ Rαβ . (3.15)
Consider now the Bianchi identity given by equation (3.13). If we con-
tract µ with β and use that Rαβγδ = −Rαβδγ we obtain
we get
∇δ R − 2 ∇β Rβδ = 0.
17
The above equation can be written in an equivalent form as
µν 1 µν
∇µ R − g R = 0.
2
This expression leads us to define a new tensor, which is of great impor-
tance in general relativity, called the Einstein tensor:
1 αβ
Gαβ ≡ Rαβ − g R.
2
The Einstein tensor is symmetric (since both Rαβ and g αβ are symmetric)
and divergence-less.
18
We will only consider the case when the tensor is a vector. For a contravari-
ant vector V α , the above equation reads
19
There is great similarity between the equivalence principle and that a curved
Riemannian manifold appears locally flat. Because of this resemblance one
may expect that spacetime in general relativity can be described with a
pseudo-Riemannian manifold.
In a locally inertial Cartesian coordinate system with coordinates Lα the
metric is given by
ds2 = ηαβ dLα dLβ .
By using that
∂Lα µ
dLα = dx
∂xµ
for any change to the arbitrary coordinates xα , we have that
where we defined
∂Lµ ∂Lν
g̃αβ ≡ ηµν .
∂xα ∂xβ
Locally in the coordinate system of Lα the equations of motion of a free
particle is
d2 Lα
= 0,
dλ2
where λ is a parameter (for massive particles this parameter can be taken to
be the proper time, but not for massless particles such as photons). Chang-
ing to the coordinates xµ and using the chain rule we can write the equa-
tions of motion as
d2 xα dxµ dxν
2
+ Γ̃αµν = 0,
dλ dλ dλ
where we defined
∂xα ∂ 2 Lσ
Γ̃αµν = .
∂Lσ ∂xµ ∂xν
This has exactly the same form as the geodesic equation derived in section
3.3. In fact, one can show that g̃αβ and Γ̃αµν has the exact same relation to
each other as gαβ and Γαµν has (as derived in section 3.1) [5]. So in general
relativity, where we treat spacetime as a 4-dimensional pseudo-Riemannian
manifold, we can express the equations of motion geometrically. That is,
a particle in free fall will follow a path that is a geodesic, given by the
geodesic equation.
20
(1) The equation holds true in absence of gravitation (i.e it holds true in
special relativity).
(2) It is a tensor equation (i.e it preserves its form under a general coord-
inate transformation).
By the equivalence principle one can write down an equation that holds
in a locally inertial coordinate system and then make a general coordinate
transformation to find the corresponding equation in that coordinate sys-
tem. With the principle of general covariance finding the equations that
holds for all coordinate systems is much simpler. It can be seen that it
follows from the equivalence principle by considering any equation that
satisfy condition (1) and (2). Since the equation is generally covariant it
preserves its form under a general coordinate transformation, so if it is true
in any coordinate system it is true in all coordinate systems. The equiva-
lence principle tells us that at every point in spacetime there exists locally
inertial coordinate systems in which the effects of gravity are absent. Since
we assumed that our equation holds in special relativity (i.e. no gravity)
and therefore holds in these locally inertial systems, it must hold in all co-
ordinate systems.
Note that any equation can be made generally covariant by working out
what it looks like in arbitrary coordinate systems. So in it self the principle
of general covariance has no physical meaning.
Our method now to find equations that are valid in a general gravita-
tional field is by simply take the valid equations (and definitions) of special
relativity and replace partial derivatives (with respect to coordinates) by
covariant derivative and the Minkowski metric ηαβ by the general metric
tensor gαβ .
F αβ,β = −µ0 j α
Fαβ,γ + Fβγ,α + Fγα,β = 0
21
Maxwell’s equations are
F αβ;β = −µ0 j α
Fαβ;γ + Fβγ;α + Fγα;β = 0.
T ≡ T αα = gαβ T αβ = 0.
22
So the geodesic equation can be approximated with
d2 xα dt 2
α
+ Γ00 c = 0.
d2 τ dτ
d2 xα 1 αµ dt 2
− g g00,µ c = 0. (4.3)
d2 τ 2 dτ
The weak field approximation allows us to write the metric tensor as the
Minkowski metric plus a small perturbation term:
We will ignore all terms higher than first order. This means that all products
of h (or its derivatives) with h (or its derivatives) are neglected. With the
condition δγα = g αβ gβγ , we must have that
g αβ = η αβ − hαβ ,
where hαβ = η αµ η βν hµν . To first order, and from the constancy of the
Minkowski metric we see that
d2 xα dt 2
1 αµ
= η h00,µ c .
dτ 2 2 dτ
d2 xα d2 xα dt 2
= .
dτ 2 dt2 dτ
d2 xi 1
2
= − c2 δ ij h00,j ,
dt 2
23
where i and j can be 1,2 or 3. But the left hand side of this equation is just
the usual components of the 3-acceleration. h00,j is just the gradient of h00 ,
so comparing with equation (4.2) we find that
2Φ
h00 = ,
c2
and since g00 = η00 + h00 we conclude that
2Φ 2GM
g00 = 1 + 2
=1− 2 .
c c r
Thus the 00-component of the metric tensor seems to be closely related to
the Newtonian potential energy.
8πG
Gαβ = Tαβ , (4.4)
c4
where G is Newton’s gravitational constant, Tαβ is the stress energy tensor
and Gαβ is the Einstein tensor defined by
1
Gαβ ≡ Rαβ − R gαβ . (4.5)
2
In total there are 4×4 = 16 equations. But since Gαβ (and Tαβ ) is symmetric
this reduces to 10 independent equations. As we found earlier in section
3.2, the Einstein tensor satisfy Gαβ ;α = 0, which reduces the number even
further to 10 − 4 = 6 independent equations.
Another useful form of Einstein’s field equations can be obtained if we
take the trace on both sides. The trace of Gαβ is
αβ αβ 1
g Gαβ = g Rαβ − R gαβ = R − 2R = −R
2
8πG
R=− T.
c4
This allows us to write Einstein’s field equations in the equivalent form:
8πG 1
Rαβ = 4 Tαβ − T gαβ . (4.6)
c 2
24
Einstein’s field equations together with the geodesic equations, which gov-
ern how particles only influenced by gravity move, build up the core of
general relativity.
It should be noted that Einstein’s field equations can not be derived
from any underlying principle. Nevertheless, one can make reasonable ar-
guments that this is a good candidate. One of these is that it reduces to the
Poisson equation
∇2 Φ = 4πGρ,
where Φ is the gravitational potential and ρ is the mass density when con-
sidering the 00-component in a weak-field approximation [7].
For curved spacetime (but still spherically symmetric) this can be general-
ized to
One could make the assumption that A and B is independent of time, but
we will keep the time-dependence for now as we will find later that A and
B must necessarily be independent of time.
Recall Einstein’s field equations:
8πG 1
Rαβ = 4 Tαβ − T gαβ .
c 2
25
Since we are working in vacuum (no matter) with an electromagnetic field
we will use the electromagnetic stress-energy tensor given by
1 1 µν νµ
Tαβ = gαβ Fµν F − gβν Fαµ F . (5.1)
µ0 4
8πG
Rαβ = Tαβ .
c4
Finally we also need the source-free Maxwell equations given by
F αβ;β = 0 (5.2)
Fαβ;γ + Fβγ;α + Fγα;β = 0. (5.3)
The calculation of the Ricci tensor is done by first calculating the Christof-
fel symbols with
1
Γλαβ = g λµ (∂α gβµ + ∂β gµα − ∂µ gαβ ),
2
and then computing the Ricci tensor with
Ȧ
Γ000 =
2Ac
Ḃ Ḃ
Γ101 = Γ110 = , Γ011 =
2Bc 2Ac
A0 A0
Γ001 = Γ010 = , Γ100 =
2A 2B
1 B0
Γ212 = Γ221 = , Γ111 =
r 2B
1 r
Γ313 = Γ331 = , Γ122 =−
r B
r sin2 θ
Γ323 = Γ332 = cot θ, Γ133 =−
B
Γ233 = − sin θ cos θ
26
where a dot represent differentiation w.r.t. t, and a prime w.r.t. r. For the
Ricci tensor all non-zero components are:
!
A0 A0 B 0 A00 A0
B̈ Ḃ Ȧ Ḃ
R00 = − + + + − + − (5.4)
4B A B 2B Br 2Bc2 4Bc2 A B
!
A0 A0 B 0 A00 B0
B̈ Ḃ Ȧ Ḃ
R11 = + − + − − − (5.5)
4A A B 2A Br 2Ac2 4Ac2 A B
0
B0
r A 1
R22 = − − − +1 (5.6)
2B A B B
R33 = R22 sin2 θ (5.7)
Ḃ
R01 = R10 = . (5.8)
Brc
This is as far as we can get in generalizing a spherically symmetric gravita-
tional field. To determine A and B any further we need to invoke Einstein’s
field equations, and that means we must specify the stress-energy tensor.
In our case the stress-energy tensor was given in terms of the metric tensor
and the electromagnetic tensor Fαβ . From the spherical symmetry we have
that the electric field can only have a radial component. Also, this radial
component must not depend on θ or φ, so we have that
Er = E1 = E1 (t, r) = c F01 = −c F10 .
All the other components are zero since there are no currents or magnetic
monopoles. In matrix form we have
0 Er /c 0 0
−Er /c 0 0 0
Fαβ =
0
.
0 0 0
0 0 0 0
The components of the stress-energy tensor can now be computed with
equation (5.1). Consider the first term in the parenthesis. Carrying out the
summation gives
1 1 1
gαβ Fµν F µν = gαβ Fµ0 F µ0 + Fµ1 F µ1 = gαβ F10 F 10 + F01 F 01
4 4 4
1 01
1
= gαβ 2F01 F = gαβ F01 F 01 .
4 2
For the second term we get
gβν Fαµ F νµ = gβν Fα0 F ν0 + gβν Fα1 F ν1 = gβ1 Fα0 F 10 + gβ0 Fα1 F 01 ,
and we can write equation (5.1) as
1 1
Tαβ = gαβ F01 F 01 − gβ1 Fα0 F 10 − gβ0 Fα1 F 01 .
µ0 2
27
The components the stress-energy tensor can now easily be obtained. We
have:
1 1 01 01
T00 = g00 F01 F − g00 F01 F
µ0 2
1 1
=− g00 F01 F 01 = − A F01 F 01
2µ0 2µ0
1 1 01 01
T11 = g11 F01 F − g11 F01 F
µ0 2
1 1
=− g11 F01 F 01 = B F01 F 01
2µ0 2µ0
1 1 2
T22 = g22 F01 F 01 = − r F01 F 01 (5.9)
2µ0 2µ0
1
T33 = g33 F01 F 01 = T22 sin2 θ.
2µ0
A0 B 0 ∂
0= + = ln(AB).
A B ∂r
This means that the product AB must be constant with respect to r. We can
write this as
AB = f (t),
where f (t) is some function that does not depend on r. We will however
show later that f must equal unity, but we will keep it as it is for now. As
we now have the relation g00 = −f /g11 (since A = g00 and B = −g11 ), we
can easily show that
28
We will now solve Maxwell’s equations. Equation (5.3) does not give us
any more information because of the non-existence of magnetic monopoles,
and it can be seen that it is directly satisfied by considering the case α = 0,
β = 1 and γ = 0:
Similarly for the other cases it can be showed that Fαβ;γ + Fβγ;α + Fγα;β is
identically zero. Equation (5.2) will however give us a more explicit form
of Er . Using equation (3.6) for the covariant derivative of a second rank
contravariant tensor, equation (5.2) reads
0 = ∂0 F 10 + Γ1µβ F µβ + Γβµβ F 1µ .
since Γ101 = Γ110 = 0 (or since F 01 = −F 10 together with Γ101 = Γ110 ). The
third term also vanish
since all the Christoffel symbols in the parenthesis equals zero. Thus we
simply end up with
0 = ∂0 F 10 .
This of course implies that F 10 , and therefore Er , must not depend on time.
That is, we have
Er = Er (r).
Using α = 0 in equation (5.10) we obtain
Similarly as before the second term vanish but in this case the third term
does not. We have that
29
Equation (5.11) now reads
∂ 01 2 01
0= F + F
∂r r
which is an ordinary first order differential equation with the (easily checked)
solution
const.
F 01 = ,
r2
which let us write
const.
Er = .
r2
By the Gauss’s flux theorem we can conclude that the constant must equal
Q/4π0 , and we have
Q
Er = .
4π0 r2
This is not an unfamiliar expression. It is really Coulomb’s law, although
one must of course remember that r is just the one of our chosen coordi-
nate and do not necessarily measures the "real" radial distance when in a
Reissner-Nordström spacetime.
We are now close to get the final form of the Reissner-Nordström metric.
We only need to get a more explicit form of A and B in terms of r. This can
now be done by considering one of Einstein’s field equations, namely
8πG
R22 = T22 .
c4
For the left hand side we have
0
B0
r A 1 1 ∂
R22 = − − − +1=− (rA) + 1
2B A B B f ∂r
1 ∂ 1 8πG 1
− (rA) + 1 = r2 Er2 .
f ∂r f c4 2µ0 c2
∂ GQ2
(rA) = f − .
∂r 4πc6 µ0 20 r2
C(t) GQ2
A=f+ + ,
r 4π0 c4 r2
30
where C(t) is a function that may depend on time. When Q = 0 the metric
must reduce to the Schwarzschild metric. And as we showed in section 4.4,
when gravity is weak (i.e. when r is large) the metric tensor component
g00 must approach 1 − 2GM/c2 r. So at this limit, if the geodesics of the
metric should agree with the motion of Newtonian gravity, we must have
that f = 1 (which implies that AB = 1) and that C(t) = −2GM/c2 ≡ −rs . It
is worth noting that from the relation AB = f (t) one
p could directly get rid
of f (t) by redefining the time coordinate as dt → f (t)dt.
The constant rs is commonly known as the Schwarzschild radius. If we
also define
2 GQ2
rQ ≡ ,
4π0 c4
A and B can finally be written as
2
rs rQ
A=1− + 2
r r
!−1
2
1 rs rQ
B= = 1− + 2 ,
A r r
P
Br ∝ ,
r2
which corresponds to the electromagnetic tensor component
F23 = r2 Br sin θ.
31
The diagonal components of the stress-energy tensor would also get one
extra term containing Br2 . The implication of this is that we could make
our "no magnetic charge"-solution to a "with magnetic charge"-solution by
simply replacing Q2 by Q2 + P 2 /c2 , if one uses SI-units where the magnetic
charge is measured in ampere·meters (A · m).
This discussion about magnetic monopoles was just a quick add-on to
the Reissner-Nordström metric but we will not use it any further. That is,
in the following sections we will use P = 0.
c2 2
0 = ds2 = A dt2 − dr .
A
Using λ as an parameter for the geodesic the above can be written as
2 2
2 dt 2 dr
A =c .
dλ dλ
32
If we let λ1 correspond to t1 and λ2 to t2 , solving for dt/dλ and integrating
we obtain Z λ2
1 dr
t2 − t1 = c dλ.
λ1 A dλ
Let now t01 be the coordinate time when the electromagnetic wave has os-
cillated exactly one period after first emission at r = r1 . Similarly, let t02
be the coordinate time when the electromagnetic wave has oscillated one
period after receiving the signal at r = r2 . Since the integral above does not
depend on t we have that
t2 − t1 = t02 − t01 ,
or equivalently as
From this we see that the coordinate time period at the point of emission
and the coordinate time period at the point of reception are equal. A real
clock however measures the proper time, so with equation (5.12) the peri-
ods in proper time are
s
2
rQ
rs
∆τ1 = ∆t1 1 − + 2
r1 r1
and s
2
rQ
rs
∆τ2 = ∆t2 1− + 2.
r2 r2
The proper frequency of the electromagnetic wave is simply the reciprocal
of the proper period and since ∆t1 = ∆t2 we must have the relation
!1/2 !−1/2
2
rQ 2
rQ
rs rs
f2 = f1 1− + 2 1− + 2 , (5.13)
r2 r2 r1 r1
which means that f2 will be smaller than f1 and the wave is said to be
redshifted. Letting r1 > r2 (traveling radially inwards) the result would
instead be that f2 is larger than f1 and the wave is said to be blueshifted.
33
5.2 The Geodesic Equation
We will now look at the equations that describes the motion of a massive
non-charged freely falling particle or a photon in a Reissner-Nordström
spacetime. Both the particle and the photon will follow a geodesic, so find-
ing out the path they take is the same as solving the geodesic equation. The
massive particle will follow a timelike geodesic while the photon follows a
null geodesic. Note: in this and the following sections we put c = 1. We do
this because it let us skip a lot of tedious work remembering to use c when
dealing with the t-component.
Let xα = xα (λ) be a curve parameterized by λ. For xα to be a geodesic
it must satisfy the geodesic equation given by
d2 xα µ
α dx dx
ν
+ Γ µν = 0. (5.14)
d2 λ dλ dλ
For the massive particle it would be natural to have the geodesic parame-
terized by the proper time, and we will actually do this later. But since the
photon follows a null geodesic it can not be parameterized by the proper
time so we will stick with the more general parameter λ so that we can treat
both the massive particle and the photon as far as possible.
Using all the non-zero Christoffel symbols written down in the begin-
ning of this chapter we have that for α = 0, equation (5.14) reads
d2 t A0 dt dr
+ = 0. (5.15)
dλ2 A dλ dλ
Considering α = 1 we obtain
d2 r A0 dt 2 B 0 dr 2
0= 2 + +
dλ 2B dλ 2B dλ
2
r sin θ dφ 2
2
r dθ
− − . (5.16)
B dλ B dλ
For α = 2 and α = 3 we get
2
d2 θ 2 dθ dr dφ
0= 2 + − sin θ cos θ (5.17)
dλ r dλ dλ dλ
2
d φ 2 dφ dr dφ dθ
0= 2 + + 2 cot θ . (5.18)
dλ r dλ dλ dλ dλ
By the spherical symmetry we must of course have that the trajectory is
in a plane. We can therefore, without any loss of generality, put θ = π/2.
This means that the derivatives of θ vanish and equation (5.17) is instantly
satisfied. With this simplification equation (5.16) and (5.18) now become
d2 r A0 dt 2 B 0 dr 2 r dφ 2
+ + − =0 (5.19)
dλ2 2B dλ 2B dλ B dλ
34
and
d2 φ 2 dφ dr
+ = 0, (5.20)
dλ2 r dλ dλ
respectively. If we divide the above equation (5.20) with dφ/dλ and note
that −1 2
dφ d φ d dφ
2
= ln
dλ dλ dλ dλ
and
2 dr d
ln r2
=
r dλ dλ
we obtain
d 2 dφ
ln r = 0.
dλ dλ
This means that the quantity inside the logarithm is a constant of motion
and we put it equal to a constant L:
dφ
r2 = L. (5.21)
dλ
If we compare this with the Newtonian theory, L correspond to angular
momentum per unit mass. So this equation basically states that angular
momentum is conserved.
In a similarly way, we can divide equation (5.15) by dt/dλ and obtain
d dt
ln A = 0,
dλ dλ
which again means that the quantity inside the logarithm must be a con-
stant which we will denote by e:
dt
A ≡ e. (5.22)
dλ
The interpretation of the constant of motion e will be made later. We can
now use equation (5.21) and (5.22) in equation (5.19) and obtain
2
d2 r e 2 A0 B0 L2
dr
+ + − = 0.
dλ2 2BA2 2B dλ Br3
35
The expression inside the square bracket must be a constant
2
dr e2 L2
B − + 2 = const ≡ −e20 ,
dλ A r
where we implicitly defined e0 (it will be clear later why we defined it this
way). Since B = 1/A, an equivalent form to the above is
2
L2
dr
= e2 − A e20 + 2 . (5.23)
dλ r
We now have dr/dλ as a function of r alone. But we are also able to express
dr/dφ as a function of r. If we divide the above equation with φ̇2 = L2 /r4
we get
2
r4 k2 r2 e20
dr 2
= 2 −r A 1+ 2 . (5.24)
dφ L L
Recall that A was defined as
!
2
rs rQ
A= 1− + 2 ,
r r
so we have that
r 2 e2 r2 e2
r2 A 1 + 20 = r2 − rrs + rQ 2
1 + 20
L L
!
2 2
2
rQ e0 rs e2 e2
= rQ − rs r + 1 + 2 r2 − 20 r3 + 02 r4 .
L L L
Substituting this into equation (5.24) we finally get
!
2 2 e2
dr 2
rQ 0 rs e2 1
r2 + 20 r3 − 2 e20 − e2 r4 (5.25)
= −rQ + rs r − 1 + 2
dφ L L L
We have now obtained dr/dφ in terms of r. But as will be shown below this
can be simplified if we look at two special cases, namely null geodesics and
timelike geodesics corresponding to paths taken by massless and massive
particles, respectively.
Remember that when c = 1 and θ = π/2, the metric is given by
1 2
ds2 = dτ 2 = A dt2 − dr − r2 dφ2 .
A
From equation (5.21), (5.22) and (5.23) we have that
L2 2
dφ2 = dλ
r4
e2
dt2 = 2 dλ2
A
L2
2 2 2
dr = e − A e0 + 2 dλ2
r
36
which let us write the differential proper time simply as
dτ 2 = e20 dλ2 .
From this we see that for null geodesics e0 must equal zero, and for timelike
geodesics e0 must be larger than zero. So for timelike geodesics parameter-
ized with the proper time, e0 simply equals unity and equation (5.25) reads
!
2 2
rQ
dr rs 1
= −rQ + rs r − 1 + 2 r2 + 2 r3 − 2 1 − e2 r4 ,
2
(5.26)
dφ L L L
This can be considered as an energy equation where the first two terms on
the R.H.S. correspond to the kinetic part and the third term to the potential
part. In Newtonian mechanics the total energy E of a particle with mass m
that is influenced only by gravity from a spherically symmetric object with
mass M satisfies
2E 2GM L2 2GM
= ṙ2 + r2 φ̇2 − = ṙ2 + 2 − .
m r r r
By comparing the two above equations we can see that e2 − e20 corresponds
to the total energy per unit mass. But since E is the Newtonian total energy
(i.e. no rest energy) we can make the interpretation that e2 corresponds
to the total relativistic energy per unit mass while e20 corresponds to the
rest energy per unit mass of the particle. So our conclusion earlier that
e0 = 1 (when using τ as the parameter) for massive particles and e0 = 0 for
massless particles now have an intuitive understanding. The big difference
from Newtonian mechanics is that general relativity introduces an extra
term which depend on r−3 which get dominant when r is small. When
r is large (i.e. for weak gravity) this extra term can be neglected and the
equation reduces to the Newtonian one.
37
5.2.2 Circular Orbits of Photons
One interesting special case is that photons (or other massless particles) can
dr d2 r
be in a circular orbit. For circular orbits we must have dλ = 0 and dλ 2 = 0.
From equation (5.19) we must have that
2 2
A0
dt dφ
−r =0
2 dλ dλ
From the two expressions above we must therefore have that rA0 = 2A,
which explicitly reads
!
2 2
rs 2rQ rs rQ
− 2 =2 1− + 2 .
r r r r
3
0 = r2 − rs r + 2rQ
2
2
with the solution s 2
3rs 3rs 2.
r± = ± − 2rQ
4 4
We have three different cases depending on the value of the expression
inside the square root. These three cases correspond to the existence of two,
one or zero real-valued solutions. The only possible orbit for the photon is
however the solution given by r+ , as will be clear when discussing black
holes and event horizons later in section 5.4. The reason is that r− will be
inside of the event horizon except in the case when 2rQ = rs , then the r−
photon orbit and the event horizon are at the same radius.
1
L = gαβ ẋα ẋβ + qAα ẋα
2
38
where q is the charge per unit mass of the particle, Aα the four-potential
and a dot represents a differentiation with respect to the proper time. If we
had no electromagnetic field present the Lagrangian would only consist of
the first term. In flat spacetime this first term would be the kinetic part
(using ηαβ instead of gαβ ), while in a gravitational field it also contains the
gravitational potential energy. The second term arises because we have
potential energy from the electromagnetic field, and splitting it up in its
timelike and spacelike coordinates we have
where uα is the 4-velocity. The first term in this expression clearly corre-
sponds to the static electric potential energy while the second term corre-
sponds to a "magnetic potential energy". If one make a gauge transforma-
tion of the electric and magnetic potentials the Lagrangian will change by
an addition of a total derivative of a function. This will however not change
the equations of motion.
Since we only have a radial electric field and no magnetic field the only
non-zero component of the four-potential is
Q Q̃
A0 = ≡
4π0 r r
and the Lagrangian now becomes
1h 2 i q Q̃
L= A ṫ − A−1 ṙ2 − r2 θ̇2 − r2 sin2 θ φ̇2 + ṫ.
2 r
The motion of the particle is determined by solving the Euler-Lagrange
equations given by
d ∂L ∂L
α
− α = 0.
dτ ∂ ẋ ∂x
Since the Lagrangian does not explicitly depend on t or φ we see from the
above equation that ∂L/∂ ṫ and ∂L/∂ φ̇ are both constants:
∂L q Q̃
= A ṫ + ≡ e = const (5.28)
∂ ṫ r
∂L
= r2 φ̇ ≡ L = const. (5.29)
∂ φ̇
The only difference between equation (5.28) and the equation for e in the
case of a massive non-charged particle is the "extra" term q Q̃/r which clearly
corresponds to a static electric potential energy. As before, the particle must
move in a plane and we can put θ = π/2. So for the θ-component (α = 2) the
39
Euler-Lagrange equation is immediately satisfied and for the r-component
(α = 1) we get
d ∂L ∂L
0= −
dτ ∂ ṙ ∂r
1 q Q̃
= −A−1 r̈ − A0 ṫ2 + A−2 A0 ṙ2 − 2rφ̇2 − 2 ṫ
2 r
which is the same as equation (5.19) except for the last term containing Q̃.
We will however not use the above equation. Instead we will use the metric
which, when θ = π/2, is given by
40
where
1 rs
gtt = − =1− .
grr r
The metric becomes singular at r = rs because when r → rs we have that
grr → ±∞. Another singular point is clearly r = 0. Note that for any spher-
ically symmetric object (e.g. a star) with a radius larger than rs there will be
no singularity. This is because the Schwarzschild solution is only valid out-
side the object where there is vacuum, which imply that all components of
the stress-energy tensor are zero. Inside the object the stress-energy tensor
does not vanish and one would obtain a solution that does not have any
singularities.
Consider now an object that do have a radius less than rs . Then the
Schwarzschild solution does hold at r = rs and we do have a singularity
at r = rs . An object with the property that its radius is less than rs is called
a black hole. This name is justified (as will be shown later) by the fact that
no massive particle nor light can escape if at a distance closer than rs . This
"boundary" in spacetime is called an event horizon and it marks the surface
for which events inside of it can not affect the outside.
As discussed in section 5.1, gravitational time dilation and redshift will
appear in a gravitational field. When dealing with black holes a few in-
teresting things happen. To an observer, a clock near the black hole will
appear to run slower than a clock further away. By equation (5.12) (letting
rQ = 0), one can see that the time dilation will be infinite for a clock falling
towards the black hole as it approaches the event horizon. This leads to
that it would require an infinite time to reach the event horizon for an object
falling towards the black hole, as seen from an outside observer. Also, from
equation (5.13) one can see that the redshift of an electromagnetic wave
traveling outwards goes to infinity when the point of emission approaches
the event horizon.
Up till now we have worked with the spherical coordinates (t, r, θ, φ).
But what if we were to choose a different coordinate system, would there
still be any singularities? As it turns out, the singularity at r = rs can
actually be made to disappear if one uses the right set of coordinates, and
one can see that it would not take an infinite amount of proper time for an
object to fall past the event horizon. To see that r = rs is merely an apparent
singularity and that not even light can escape from a black hole, we will use
the so-called Lemaître coordinates λ and ρ which differentials transform as
r
rs rs −1
dλ = dt + 1− dr
r r
r
r rs −1
dρ = dt + 1− dr.
rs r
41
Taking the difference between the two above equations we have
r r r
r rs rs −1 r
dρ − dλ = − 1− dr = dr. (5.30)
rs r r rs
where
2
1 rs rQ
gtt = − =1− + 2.
grr r r
42
A quick inspection suggest that possible event horizons should occur when
2 /r 2 which yields the quadratic equation
0 = gtt = 1 − rs /r + rQ
0 = r2 − rs r + rQ
2
Consider the situation when rs > 2 rQ . In this case there is two coor-
dinate singularities at r+ and r− . Note that these two singularities occur
because of our choice of coordinate system. As in the Schwarzschild metric
one can choose a coordinate system in which there exist no singularities
except at r = 0 (which still is a true singularity). The metric in this case can
43
be divided into three regions:
Region 1: r+ < r < ∞
Region 2: r− < r < r+
Region 3: 0 < r < r−
An object coming from region 1 and falling into region 2 would have the
same experience as when crossing the event horizon in the Schwarzschild
black hole. For an outside observer the infalling object would be infinitely
redshifted and it would never reach the event horizon. The proper time for
reaching the event horizon for the object would however be finite. Once
inside region 2 all massive particles and photons necessarily move in the
direction of decreasing r. This unavoidable decrease in r does however stop
when reaching region 3, and the object is therefore not doomed to end up at
the singularity at r = 0. If now moving back (increasing r) to region 2 again,
the object can only move in direction of increasing r and ultimately come
out beyond the horizon at r = r+ [8]. However, it is highly speculative if
this journey through the black hole is physically real.
Consider now the situation when rs < 2 rQ . In this case there is no
singularities when r > 0, and therefore no event horizons. The singular-
ity at r = 0 does still exist, which means that there is no event horizon
preventing someone far away to directly observe this singularity. A singu-
larity with this property (i.e. no event horizon "hiding" it) is called a naked
singularity. It is widely believed, but not proven, that no naked singular-
ity (except maybe the one occurring in the Big Bang model) exist in the
universe [11]. This assumption is called the weak cosmic censorship hypoth-
esis and was formulated by Roger Penrose in 1969. So this solution when
rs < 2 rQ is therefore usually considered to be unphysical.
Lastly, when rs = 2 rQ there exist only one horizon, and the black hole
is called extremal. The event horizon is located at r = rs /2, and in this
case gtt is positive on both sides which means that an observer inside the
event horizon does not necessarily move towards the singularity at r =
0. However, this extremal black hole with only one horizon seems to be
unstable since adding any nonzero mass would turn it to the "regular" case
when rs > 2 rQ .
44
It was later shown how this gravitational field give rise to phenomena such
as gravitational time dilation and redshift. With the use of the geodesic
equations we derived the equations of motion for both massive and mass-
less non-charged particles while we used the lagrangian formalism to de-
rive the equations of motion for massive charged particles. We also found
out how these equations corresponds to the well known equations derived
from Newtonian mechanics. Lastly we discussed the properties of a Reissner-
Nordström black hole. Depending on the relative value of rs and rQ there is
two, one or zero event horizons that corresponds to apparent singularities
in the metric. The only true singularity is found in the center of the black
hole.
The Reissner-Nordström metric is a generalization to the Schwarzschild
metric, it can however itself be generalized to the so-called Kerr-Newman
metric. It is a solution to the Einstein-Maxwell equations for an electri-
cally charged rotating axially symmetric body. That is, in addition to the
Reissner-Nordström metric it has a non-zero angular momentum which
has the consequence that it no longer exhibit spherical symmetry. Some
methods that we used in deriving the Reissner-Nordström metric can also
be used when deriving the Kerr-Newman metric.
References
[1] Schwarzschild, K., 1916, Über das Gravitationsfeld eines Massen-
punktes nach der Einsteinschen Theorie, Sitzungsber. Preuss. Akad. D.
Wiss. 50, pages 189-196.
[2] Reissner, H., 1916, Über die Eigengravitation des elektrischen Feldes
nach der Einsteinschen Theorie, Annalen der Physik. 50, pages 106-120.
[3] Nordström, G., 1918, On the Energy of the Gravitational Field in Ein-
stein’s Theory, Verhandl. Koninkl. Ned. Akad. Wetenschap., Afdel. Natu-
urk., 26, pages 1201-1208.
[4] Jackson, J. D., 1998, Classical Electrodynamics, John Wiley & Sons Inc,
3rd edition,
[5] Weinberg, S., Gravitation and Cosmology, John Wiley & Sons Inc, 1st
edition, 1972.
[7] Foster, J. and Nightingale, J. D., 2005, A Short Course in General Relativ-
ity, Springer, 3rd edition,
45
[8] Carroll, S. M, 2003, Spacetime and Geometry: An introduction to General
Relativity, Addison Wesley, 1st edition.
[10] Chandrasekhar, S., 1983, The Mathematical Theory of Black Holes, Oxford
University Press, 1st edition, page 224.
46