Tolman–Oppenheimer–Volkoff (TOV) Stars

Aaron Smith1, ∗
1 Department of Astronomy, The University of Texas at Austin, Austin, TX 78712
(Dated: December 4, 2012)
We present a set of lecture notes for modeling stellar structure in regimes where
general relativistic effects become important, such as for neutron stars. The frame-
work draws directly from solving the Einstein equation for a spherically symmetric
star in static equilibrium in terms of energy density, pressure, and a term like the
gravitational potential. The equation is presented as a somewhat intuitive extension
of what was covered regarding stars in hydrostatic equilibrium. Furthermore, we
numerically solve the TOV equation in the case of a piecewise polytropic equation
of state to find a theoretical upper limit on the mass of neutron stars. For the model
considered we find a maximum mass of 2.122 M , in agreement with the currently
accepted range of 1.44 to 3 M for neutron stars.

I. INTRODUCTION

The background material for this topic is presented in Chapter II of the class notes
regarding hydrostatic equilibrium. Naturally this means the solution remains constant in
time, a phenomenon made possible by the balance of internal pressure support against the
star’s gravitational field. For stars like our sun the gas pressure is fueled by the hot nuclear
reactions in its core. However, many types of stars are known to exist with varying mass and
temperature as characterized by an HR diagram. Though the lives of stars are complicated
by their unique dynamics, their end states are signaled by the exhaustion of the thermal
fuel. Again, depending on the mass of the star it may either collapse to form a black hole
or (perhaps after a collapse type explosion) find other means to oppose gravity. At this
point we focus on relativistic stars for which the relativistic internal energy significantly
contributes to the stellar equation of state.

II. RELATIVISTIC FRAMEWORK

We review the most relevant aspects of general relativity for stellar applications and refer
the reader to Refs. [1, 2] for more detailed treatments. The metric is a geometric tool that
relates distances in spacetime, a kind of generalized pythagorean theorem where the time
coordinate is included as well. The underlying physics is more important than the relative
coordinates, so all equations are written in the invariant language of tensors, or multi-indexed
objects. The Einstein summation convention shortens the notation by assuming an implied
sum over repeated indices. With this in mind, the Schwarschild metric for a spherically

∗
asmith@astro.as.utexas.edu
 2

symmetric vacuum spacetime (valid outside a star or black hole) in coordinates (t, r, θ, φ) is

− 1 − 2GM

2 
rc2 c 0 0 0
2GM −1


gµν = 

 0 1 − rc2 0 0  , (1)
0 0 r2 0 
0 0 0 r2 sin2 θ

which induces the following line element for measuring infinitesimal distances:
   −1
2 µ ν 2GM 2 2 2GM 2 2 2 2 2


ds = gµν dx dx = − 1 − c dt + 1 − dr + r dθ + sin θdφ .
rc2 rc2
(2)
Note that in spacetime, the components dxµ = (dt, dr, dθ, dφ) are treated on equal footing
except for a relative minus sign in front of the squared time component. However, this allows
for the propagation of light along null vectors (i.e. ds2 = 0).
In fact, the Lorentzian form of the metric, or the (− + ++) signature asymmetry of time
with space, helps to explain the presence of a gravitational force in curved spacetime. The
geometry is essential so we define vector fields at each point and parallel transport vectors
from nearby points to establish a generalized notion of the derivative. For the covariant
derivative ∇ this is done by defining the Christoffel symbols Γ as a way to connect different
points in spacetime:

∂ → ∇ where ∇µ Tβα = ∂µ Tβα + Γασµ Tβσ − Γσµβ Tσα (3)

1
and Γµαβ = g µν [gαν,β + gβν,α − gαβ,ν ] . (4)
2
The comma denotes differentiation with respect to the coordinate xµ so that f, µ ≡ ∂f /∂xµ .
The Christoffels are used to construct a measure of curvature called the Ricci tensor:

Rµν = Γαµν,α − Γαµα,ν + Γαβα Γβµν − Γαβµ Γβαν , (5)

and its contraction or trace, R = g µν Rµν , known as the Ricci scalar.

This construction communicates the presence of spacetime geometry whereas the stress-
energy tensor T describes the matter and energy content of the Universe. In practice, a
symmetric Einstein tensor G satisfying special properties (i.e. the Bianchi identities given
by ∇µ Gµν = 0) is hand-picked to mirror the physical properties of the stress-energy tensor T
(i.e. conservation laws derived from ∇µ T µν = 0). The final results needed for our relativistic
stars is the deduced Einstein equation from the requirement given above:
1 8πG
Gµν ≡ Rµν − gµν R = 4 Tµν . (6)
2 c
Note that relativists tend to use units where the speed of light and the gravitational constant
simplify the equations so that c = G = 1. Equation 6 simply relates the fact that ‘matter
tells space how to curve, and space tells matter how to move.’

III. DERIVATION OF THE TOV EQUATION

According to Birkhoff’s theorem the Schwarzschild solution is the most general description
outside a nonrotating, spherically symmetric star. However, inside the star we must consider
 3

a more general metric to describe a static density and pressure profile. Once again using
Birkhoff’s theorem we are free to write the general metric for the stellar interior in the
time-independent form (we have set c = G = 1 but can restore the dependence later)
 −1
2 2Φ(r) 2 2m(r)
dr2 + r2 dθ2 + sin2 θdφ2 .


ds = −e dt + 1 − (7)
r

The functions in front of dt2 and dr2 need to be independent functions of the radial coordinate
r, however, we are free to choose their form which we do in anticipation of a physical
interpretation of a mass function down the road. It is important to see the relationship
between a general spherically symmetric spacetime and the further constraint of a vacuum
solution. The empty space outside the star is precisely the one given by the Schwarzschild
metric. In fact, in the limit that the initial distribution is given by a point source then the
solution represents the physics of a static black hole.

III.1. Perfect Fluid

Consider a perfect fluid with total energy density , isotropic pressure P , metric g µν ,
dt dr dθ dφ
and four-velocity uµ = ( dτ , dτ , dτ , dτ ) = γ(1, ~v ) as are familiar from special relativistic
calculations. The stress-energy tensor for is built from these components according to

T µν = ( + P )uµ uν + P g µν . (8)

The total energy density consists of both the rest mass density of the fluid ρ and the internal
energy ε, which in this case represents the thermal motion of the constituent fluid particles.
By this we mean the total energy density is given by

 = ρc2 + ε . (9)

For simplicity we may indiscriminately drop the factors of c knowing we may recover them
if needed. Finally, we define the specific enthalpy h, or enthalpy per unit mass, as
+P
h= . (10)
ρ

III.2. The TOV Equation

In the spirit of completion we now present the Tolman–Oppenheimer–Volkoff (TOV)

equation. From the metric of Equation 7 and the method of Section II we can calculate the
Christoffel symbols Γαµν , the curvature tensor Rµν , the Einstein tensor Gµν , and the stress-
energy tensor Tµν for a perfect fluid. We begin with the nonzero, independent Christoffels,
which can be calculated from Equation 3, where we use the notation dm(r)/dr = m0 , etc.:

rm0 − m
 
2m 1
t
Γtr = Φ 0 r
Γtt = Φ e0 2Φ
1− Γrrr = 2 Γθrθ = Γφrφ =
r r − 2rm r
Γrθθ = csc2 θ Γrφφ = 2m − r Γθφφ = − csc2 θ Γφθφ = − sin θ cos θ . (11)
 4

From Equation 5 the nonzero, independent components of the Ricci tensor Rµν are

2r − 3m − rm0
    
2Φ 00 02

2m 0
Rtt = e Φ +Φ 1− +Φ
r r2
−1 
(rm0 − m)(2 + rΦ0 )
 
2m
Rrr = 1 − 3
− Φ00 − Φ02
r r
m
Rθθ = csc2 θ Rφφ = (2m − r)Φ0 + m0 + . (12)
r
Therefore the Ricci scalar is
 0   
µν 2m 0 0 2m 00 02


R = g Rµν = 2 + Φ (3m − 2r + rm ) − 1 − Φ +Φ . (13)
r2 r
The Einstein equation is Gµν ≡ Rµν − gµν R/2 = 8πTµν , where in the Eularian rest frame
Ttt = e2Φ so the ‘time-time’ component of the equation gives
2m0 e2Φ
Gtt = 2
= 8πe2Φ or m0 = 4πr2  . (14)
r
The radial equation with Trr = (1 − 2m
r
)−1 P is

m + 4πr3 P
 
2 0 m 8πP
Grr = Φ − = or Φ0 = . (15)
r 1 − 2m/r 1 − 2m/r r(r − 2m)
Finally, we need a differential equation for the pressure. An easy way to do this is to use
conservation of energy to say the divergence of the stress-energy tensor vanishes. The radial
component is all we need. If T µν = diag(e−2Φ , (1 − 2m
r
)P, r−2 P, r−2 csc2 θP ) then we have

∂T rν
0 = ∇ν T rν = + T σν Γrσν + T rσ Γνσν
∂xν
∂T rr
= + T tt Γrtt + T rr Γrrr + T θθ Γrθθ + T φφ Γrφφ + T rr Γνrν
∂r 
2m
= 1− [P 0 + (P + )Φ0 ] or P 0 = −( + P )Φ0 . (16)
r
The TOV equation is summarized as
dm
= 4πr2 
dr
dP m + 4πr3 P
= −( + P )
dr r(r − 2m)
dΦ 1 dP
=− . (17)
dr  + P dr
Finally, we express this equation in a form suggestive of the post-Newtonian corrections [3]
−1
4πr3 P (r)
  
dP G(r)m(r) P (r) 2Gm(r)
=− 1+ 1+ 1− . (18)
dr c2 r 2 (r) m(r)c2 c2 r
It is apparent that the factor in front is the classical equation for hydrostatic equilibrium.
The first two factors in square brackets represent special relativistic corrections of order v 2 /c2
 5

that arise from the mass–energy relation so that the denominators,  and mc2 , vary relativis-
tically in connection with Eistein’s famous equation E = M c2 . The last term in brackets is
a general relativistic correction based on the physical significance of the Schwarzschild form
of the metric and the meaning of m(r) as the total integrated mass out to a radial distance
r. These corrections each act to strengthen the gravitational interaction. Indeed, we note
that Φ in Equation 17 is a kind of gravitational potential.

III.3. Polytropic Equation of State

To progress toward a solution of the Equation 17 we assume a polytropic equation of

state (EOS) for a relation between the isotropic pressure and the rest mass density
P = KρΓ , (19)
where Γ is the adiabatic index and K is a normalization constant. For adiabatic processes
we may neglect heat transfer (i.e. dQ = 0) so the first law of thermodynamics is simply
dU = −P dV , (20)
where U = V is the total energy of the fluid in a volume V , including both the rest energy
and internal energy. However, we may write the rest mass density as ρ = mN/V , where N
is the number of particles of mass m in the same volume V . In other words, the first law of
thermodynamics can be written
   
 1 P
d = −P d = 2 dρ = KρΓ−2 dρ . (21)
ρ ρ ρ
The last equality was obtained by applying the polytropic EOS P = KρΓ to the star. The
integrated equation is written suggestively as
 K Γ−1
= (1 + a) + ρ . (22)
ρ Γ−1
The constant of integration is placed in the equation to ensure continuity of the piecewise
polytropic EOS which we will adopt later on. In fact, our primary concern is that in the low
density limit all energy originates from the rest mass. Specifically, we require limρ→0 /ρ = 1
and set a = 0 for the stellar boundary.
The form of the equation is nicer when we remember the form of the internal energy
ε =  − ρ and the specific enthalpy h = ( + P )/ρ. The resulting equations are
P Γ P
ε = aρ + and h=1+a+ . (23)
Γ−1 Γ−1 ρ

III.4. Piecewise Polytrope

A nuclear equation of state is often given as a table or a piecewise polytrope for dividing
densities ρ0 < ρ1 < ρ2 < · · · based on the various envelopes and crusts [4]. Thus, we
continue by expressing the above equations in a piecewise manner.



 K0 ρΓ0 ρ < ρ1
K1 ρΓ1 ρ1 < ρ < ρ2

P = . (24)

 ..

K ρΓn ρ < ρ

n n
 6

and the important fluid variables are smoothly given in each section by
Ki Γi
 = (1 + ai )ρ + ρ
Γi − 1
K i Γi
ε = ai ρ + ρ
Γi − 1
Γi
h = 1 + ai + Ki ρΓi −1 . (25)
Γi − 1
Recall that the integration constants ai are chosen to ensure the energy is smooth at the
transitions in the piecewise function so that

a0 = 0
Ki−1 Γi−1 −1 Ki Γi −1
ai = ai−1 + ρi − ρ . (26)
Γi−1 − 1 Γi − 1 i
Now when integrating the TOV equations, it is useful to define a generalization of the
Newtonian specific enthalpy
η = h − 1, (27)
which subtracts off the contribution from the rest mass of the fluid. Furthermore, the
polytropic index ni = 1/(Γi − 1) is defined exactly as in the class notes. Thus we may write
the fluid variables in terms of η in the following manner
 ni
η − ai
ρ(η) =
Ki (ni + 1)
 ni +1
η − ai
P (η) = Ki
K (n + 1)
i i 
ai + n i η
(η) = ρ(η) 1 + . (28)
ni + 1

The TOV equations of Eq. 17 diverge at r = 0 and can be difficult to integrate numerically
for r → 0. A common technique in the literature to continue analytically and avoid singular
equations is to change variables by defining a pseudo-enthalpy
Z P
dP 0
y(P ) = . (29)
(P 0 ) + P 0

With this definition we may relate the pseudo-enthalpy to the specific enthalpy by y = log(h).
From a quick application of the first law of thermodynamics in terms of dh we have
dh dP
= dy = . (30)
h +P
Indeed
p the TOV equation for dΦ/dr can be integrated immediately to give ey+Φ = heΦ =
1 − 2M/R, where M and R are the mass and radius of the star. This follows from
the relation dΦ = dy and choosing the constant of integration to match the Schwarzschild
spacetime beyond the surface of the star. The meaning of this new insight is that the
‘gravitational potential’ Φ for the star is fully determined if we can integrate the other two
 7

TOV equations. This becomes apparent when we complete the change of variables to the
Newtonian specific enthalpy η. Equation 17 becomes

dr r(r − 2m) 1
=− 3
dη m + 4πr P (η) η + 1
dm dr
= 4πr2 (η) , (31)
dη dη
which are well-behaved both at the center of the star and at the surface.
The next step is to place factors of G and c back into the Equation 31 and introduce
dimensionless variables to prepare for easier integration on a computer. We choose the
mass scale to be 1M . In order to integrate the equations we must choose a value for the
enthalpy at the center of the star ηc and set up initial conditions r(ηc ) = 0 and m(ηc ) = 0.
The equations are then integrated from the center of the star (η = ηc ) to the surface (η = 0).
Once the Newtonian specific enthaply crosses zero we have reached the surface of the star
and must switch to the Schwarzschild solution if we desire to proceed further. Thus, the
radius and mass of the star are respectfully R = r(0) and M = m(0).

12

11
Radius @kmD

Maximum Mass: 2.122 M Ÿ

10

8
1.7 1.8 1.9 2.0 2.1

Mass @MŸD

FIG. 1. The mass and radius relationship for a neutron stars with parameters found in Table I.
The maximum mass of 2.122 M is found by varying the central Newtonian specific enthalpy ηc .

Ki Γi ai ρi
3.99873692 × 10−8
1.35692395 0
2.23872092 × 10−8 3 0.010350691 1.4172900 × 1014

TABLE I. Summary of the parameters Ki , Γi , ai , and ρi for a simple piecewise polytropic EOS.
 8

IV. RESULTS

The piecewise polytropic equation of state takes into account the different envelopes of
nuclear matter, and does a fairly good job with relatively few layers. One layer is simply
not realistic enough because there are points where a harder or softer EOS are necessary.
Realistic models may use a dozen different piecewise steps. At this point we may actually
solve the TOV equation for piecewise values of Ki , Γi , and ai . The transition of where the
piecewise transition should be made is based on a density parameter ρi so that once again
we have P/c2 = Ki ρΓi i . We plot points representing the mass and radius of a neutron star
with the parameters given in Table I. The maximum mass is found to be about 2.122M ,
which is consistent with the currently accepted range of 1.44 to 3 M for the maximum
mass of a neutron star [4]. We note that there is a hard constraint from general relativity
as to the maximum stationary mass within a given radius because if R < 2GM/c2 then
a black hole will form. There is also a causality constraint based roughly on the sound
speed being less than the speed of light, which requires that R > 3GM/c2 . Finally there
is a potential rotation constraint so that the spin frequency is less than the mass-shredding
limit. However, the overall picture is sufficiently complicated to make observations the most
authoritative constraint.

Appendix A: Code for the TOV Integration

Ask the author via email.

[1] J. B. Hartle, “Gravity: An introduction to Einstein’s general relativity,” Boston:

Addison-Wesley (2003) 582 p.
[2] E. Poisson, “A Relativist’s Toolkit: The Mathematics of Black-Hole Mechanics,” Cam-
bridge University Press (2004).
[3] R. R. Silbar and S. Reddy, Am. J. Phys. 72, 892 (2004) [Erratum-ibid. 73, 286 (2005)].
[4] J. M. Lattimer and M. Prakash, Science 304, 536 (2004).