A5682: Introduction to Cosmology Course Notes

4. Cosmic Dynamics: The Friedmann Equation

Reading: Chapter 4

Newtonian Derivation of the Friedmann Equation

Consider an isolated sphere of radius Rs and mass Ms , in uniform, isotropic expansion (Hubble
flow).
The equation of motion for Rs (t) can be obtained from the gravitational acceleration at the outer
edge of the sphere:
d2 Rs GMs
2
=− 2 .
dt Rs (t)

Multiplying both sides by dRs /dt and integrating converts this “acceleration equation” to an “en-
ergy equation”:
 2
1 dRs GMs
= + U.
2 dt Rs (t)
Mathematically, U is just a constant of integration, but physically it corresponds to the total energy
per unit mass at the surface of the expanding sphere, i.e., the sum of the kinetic energy per unit
mass and the gravitational potential energy per unit mass.
If U > 0, then the expanding sphere has positive total energy and will expand forever (the r.h.s.
will always be positive).
If U < 0, then the sphere has negative total energy and will eventually recollapse (the r.h.s. will
eventually become zero).

Write the radius in the form

Rs (t) = a(t)rs ,
where rs is the “comoving” radius of the sphere, equal to the physical radius at the epoch when
a(t) = 1.
With
4π
Ms = ρ(t)Rs3 (t),
3
the energy equation becomes
1 2 2 4π 2
rs ȧ = Grs ρ(t)a2 (t) + U.
2 3
Dividing each side by rs2 a2 /2 yields
 2
ȧ 8πG 2U
= ρ(t) + 2 2 .
a 3 rs a (t)

Since ρ(t) ∝ 1/a3 (t), we see that if U is negative the r.h.s. of this equation will eventually hit zero,
after which the expansion reverses.

Although this derivation describes an isolated sphere, Newton’s “iron-sphere” theorem tells us that
it should also describe any spherical volume of a homogeneous and isotropic universe, since the
gravitational effects of a spherically symmetric external matter distribution cancel.
A generalization of the “iron-sphere” theorem turns out to hold in GR as well.

1
A5682: Introduction to Cosmology Course Notes

The Friedmann Equation in GR

A proper derivation of the Friedmann equation begins by inserting the Friedmann-Robertson-

Walker metric into the Einstein Field Equation.
Since GR yields the Newtonian limit, we should expect the small scale behavior to resemble that
of our Newtonian derivation above, and it does, with two important changes.
First, the mass density ρ(t) is replaced by the total energy density ǫ(t)/c2 , which includes rest mass
energy and other forms of energy (e.g., energy of photons, or thermal energy of atoms).
[In most texts, this total energy density is just written as ρ(t) and understood to include all
contributions, not just rest mass. I will try to stick with Ryden’s more pedagogical notation here.]
Second, the “potential energy” term is intimately tied to the curvature of space.

The GR form of the Friedmann Equation is

 2
ȧ 8πG ǫ(t) kc2 1
= − 2 2 ,
a 3 c2 R0 a (t)

where R0 is the present value of the curvature radius and k = +1, 0, or -1 is the curvature index
in the FRW metric.
While the precise form of the last term is not obvious without a GR derivation, it makes reasonable
sense that positive space curvature is associated with stronger gravity and thus with negative
“binding energy.”

If k ≤ 0, and the energy density is positive, then the r.h.s. is always positive, and an expanding
universe continues to expand forever.
If matter is the dominant form of energy, then dilution implies ǫ(t) ∝ 1/a3 (t). If k = +1, then the
r.h.s. must eventually reach zero, after which the expansion will reverse.
Thus, positive space curvature corresponds to a bound universe.
However, a form of energy for which ǫ(t) falls more slowly than 1/a2 (t), such as a cosmological
constant, can change this automatic correspondence.

The Critical Density and the Density Parameter, Ω

Substituting H(t) = ȧ/a allows us to write the Friedmann equation in terms of the Hubble param-
eter,
8πG ǫ(t) kc2 1
H 2 (t) = − 2 2 .
3 c2 R0 a (t)

From this equation, we can see that space is flat (k = 0) if the mean density of the universe equals
the critical density
ǫc (t) 3H 2 (t)
ρc (t) = = .
c2 8πG

The two cosmological equations most worth memorizing are H = ȧ/a and this definition of the
critical density. Together they are the Friedmann equation for a flat universe.

2
A5682: Introduction to Cosmology Course Notes

The present day value of the critical density is

ǫc,0 3H02
ρc,0 = 2
=
c 8πG
 2
−30 −3 H0
= 9.2 × 10 g cm
70 km s−1 Mpc−1
 2
11 −3 H0
= 1.4 × 10 M⊙ Mpc .
70 km s−1 Mpc−1

Cosmologists frequently describe the energy density of the universe in terms of the density parameter

ǫ ǫ 8πG
Ω≡ = 2× ,
ǫc c 3H 2

the ratio of the total energy density ǫ to the critical energy density.
Substituting this definition into the Friedmann equation yields

kc2 kc2
H 2 = ΩH 2 − =⇒ 1 − Ω(t) = − .
R02 a2 (t) H 2 (t)a2 (t)R02

If Ω = 1, then it equals one at all times, since the r.h.s. of this equation always vanishes.
In other cases, the value of Ω changes with time, but if Ω > 1 it is always > 1, and if Ω < 1 it is
always < 1, because the r.h.s. cannot change sign.

At the present day, we can solve this equation to get

c
R0 = |1 − Ω0 |−1/2 .
H0

If Ω0 is very close to one, then the curvature radius is large compared to the Hubble radius c/H0 ,
and curvature effects on this scale are small.

Evolution of Energy Density: The Fluid Equation

The Friedmann equation determines a(t) if we know H0 and the energy density ǫ(t) as a function
of time.
(By comparing ǫ0 to the critical density, we can determine whether k = +1, 0, or −1, and we can
use our last equation to determine R0 .)
If the only energy contribution is from non-relativistic matter, then ǫ(t) = ǫ0 (a/a0 )−3 , since ex-
pansion of the universe simply dilutes the density of particles.

For the more general case, let’s turn to the first law of thermodynamics,

dE = −P dV + dQ,

the change of internal energy of a volume of fluid is the sum of P dV work and added heat.
The expansion of a homogeneous universe is adiabatic, as there is no place for “heat” to come from,
and no “friction” to convert energy of bulk motion into random motions of particles.

3
A5682: Introduction to Cosmology Course Notes

(There is a caveat to this statement: when particles annihilate, such as electrons and positrons,
this adds heat and makes the expansion temporarily non-adiabatic. This matters at some specific
epochs in the very early universe.)

Therefore,
dE + P dV = 0 =⇒ Ė + P V̇ = 0.
4π 3 3
For a sphere of comoving radius rs , V = r a (t),
3 s
 
4π 3 ȧ
rs 3a2 ȧ = V


V̇ = 3 ,
3 a

and E = V ǫ.
Therefore  
ȧ
Ė = V ǫ̇ + V̇ ǫ = V ǫ̇ + 3 ǫ .
a

Together with Ė + P V̇ = 0, we get

 
ȧ ȧ
V ǫ̇ + 3 ǫ + 3 P =0
a a

and thus
ȧ
ǫ̇ + 3 (ǫ + P ) = 0.
a
This fluid equation describes the evolution of energy density in an expanding universe.

To solve this equation, we need an additional equation of state relating P and ǫ.

Suppose we write this in the form
P = wǫ.

In principle, w could change with time, but we will assume that any time derivatives of w are
negligible compared to time derivatives of ǫ. This is reasonable if the equation of state is determined
by “microphysics” that is not directly tied to the expansion of the universe.
The fluid equation then implies
ǫ̇ ȧ
= −3(1 + w) ,
ǫ a
with solution  −3(1+w)
ǫ a
= .
ǫ0 a0

For non-relativistic matter,

2 2
mvth vth
w = P/ǫ ∼ ∼ ≪ 1,
mc2 c2
where vth is the thermal velocity of particles.
To a near-perfect approximation, w = 0, implying ǫ ∝ a−3 , in line with our simple dilution
argument.

For radiation (i.e., photons), w = 1/3, implying ǫ ∝ a−4 .

4
A5682: Introduction to Cosmology Course Notes

This behavior also follows from a simple argument: the number density of photons falls as n ∝ a−3 ,
and the energy per photon falls as hν ∝ a−1 because of cosmological redshift.
The fluid equation will lead us to some less obvious conclusions when we consider “dark energy.”

The Acceleration Equation

If we multiply our standard version of the Friedmann equation by a2 , we get

8πG 2 kc2
ȧ2 = ǫa − 2 .
3c2 R0

Take the time derivative

8πG
ǫ̇a2 + 2ǫaȧ ,


2ȧä = 2
3c
divide by 2ȧa
ä 4πG  a 
= ǫ̇ + 2ǫ ,
a 3c2 ȧ
and substitute from the fluid equation
a
ǫ̇ = −3 (ǫ + P )
ȧ
to get
ä 4πG
= − 2 (ǫ + 3P ) .
a 3c

We see that if ǫ and P are positive, the expansion of the universe decelerates.
Higher P produces stronger deceleration for given ǫ, e.g., a radiation-dominated universe decelerates
faster than a matter-dominated universe.
The appearance of ǫ/c2 +3P/c2 is a specific example of a more general phenomenon in GR: pressure
appears in the stress-energy tensor, and it therefore has a gravitational effect. (It must, because
the division between energy density and pressure depends on the state of motion of the observer.)
With the stress-energy tensor of an ideal fluid, the Newtonian limit of GR yields a “Poisson equa-
tion”
∇2 Φ = 4πG(ρ + 3P/c2 ).

The Cosmological Constant, Vacuum Energy, and Cosmic Acceleration

A negative energy density would be pretty bizarre.
Negative pressure sounds bizarre too, but it’s not quite as crazy. For a fluid to have negative
pressure means that it has tension — it takes work to expand the fluid, instead of taking work to
compress it.
Suppose the universe is pervaded by a form of energy that is constant density, in space and time.
Such an energy could conceivably arise as a consequence of the quantum vacuum, which is not
really empty but filled with “virtual” particles.
Expanding a volume increases its energy, so that energy must have come from doing P dV work on
the volume, which implies that the pressure must be negative (the volume “resists stretching.”)
In terms of our fluid equation, we see that ǫ = const. implies w = −1, and thus P = −ǫ.

5
A5682: Introduction to Cosmology Course Notes

From the acceleration equation, we see that any fluid with P < −ǫ/3 causes acceleration of the
universe, instead of deceleration. Such a fluid has, in GR, repulsive gravity.
This discussion makes cosmological acceleration sound almost reasonable.
The problem is that no one knows how to calculate the quantum vacuum energy from first principles.
In the absence of a complete theory of quantum gravity, the most reasonable guess is that

ǫvac ∼ Ep /lp3 ,

where lp = (h̄G/c3 )1/2 = 1.6 × 10−35 m is the Planck length and Ep = (h̄c5 /G)1/2 = 2.0 × 109 J is
the Planck energy.
This estimate exceeds the critical density by 120 orders of magnitude!

Since the only “natural” number close to 10−120 is zero, the general expectation at least until the
mid 1990s was that a “correct” calculation of quantum vacuum energy would produce cancellations
that make the answer exactly zero.
It’s possible that vacuum energy really does have the value required to produce the observed cosmic
acceleration.
It’s also possible that the fundamental vacuum energy is zero, and that acceleration of the Universe
is caused by some other negative pressure fluid, or by a breakdown of GR on cosmological scales.

Einstein introduced (in 1917) the “cosmological constant” Λ with a different conception, as a
modification of the curvature term of the Field Equation rather than an additional contribution to
the stress-energy tensor.
However, the gravitational effects of Einstein’s cosmological constant are identical to those of a
form of energy with P = −ǫ, which remains constant in space and time as the universe expands.
Einstein needed the “repulsive gravity” of the Λ-term to allow a static universe, counteracting the
gravitational attraction of matter. He abandoned the idea when the universe was found to be
expanding.
But today we need a cosmological constant, or something like it, to explain how the universe can
accelerate.