6

Equation of state

As we learned in Chapter 4, the equations describing the motion of plasma are

three equations that are derived from the laws of conservation of mass, momentum,
and energy. These three equations express the four variables that describe the
moving plasma - mass density, pressure, temperature (or energy), and velocity - as
functions of spatial position and time. Since it is not possible to solve a system of
three equations for four variables, we need a fourth equation relating some or all of
these four variables that does not introduce another variable. The fourth equation
is the equation of state. In the discussion that follows we will refer to the equation
of state as the EOS.
With the EOS specified, we can then solve the equations of motion for the
plasma. We will learn how this solution is done numerically in Chapter 11. In
this chapter we learn the basics of how to specify the EOS for matter at extreme
conditions.

6.1 Basic thermodynamic relations

We learned in Section 2.2.2 about the relaxation rates in dense plasma, that is, the
rates at which thermodynamic equilibrium is established. In general, the variables
characterizing the state of the plasma - mass density (or alternatively, particle
number density), pressure, temperature - change slowly compared to these relax-
ation rates. Thus, we can consider that the plasma is, at each point in space and
at each instant of time, in local thermodynamic equilibrium (LTE). In LTE the
particle distribution functions for each particle species comprising the plasma can
be characterized by a single parameter, the temperature.
Further, in the equilibrium plasma the laws of thermodynamics apply. In partic-
ular, the thermodynamic free energy of the plasma can be written as

F = -kTlog Z , (6.1)

159
160 Equation of state
where Z is the partition function, which describes how the energies are partitioned
among the particles. Once we know the partition function, we can then derive
the pressure P and entropy S as functions of the other state variables from the
thermodynamic relations:

(6.2)

where M is the mass in the elemental volume,

S = - &F l (6.3)
&T p
'

and the internal energy

£ = F + TS. (6.4)

Thus, aII we need to do to specify a material 's EOS is to specify its partition
function Z.
According to statistical mechanics, for a system of N particles at a temper-
ature T, the probability of the ith state having an energy E i is proportional to
gi exp(- E i /kT), where gi is the degeneracy of the ith state. The sum of these
probabilities over all possible states of the system, to within a multiplicative factor,
is

(6.5)

Equation (6.5) is the expression for the partition function of the system. The
partition function is simply the sum of the probabilities of the individual energy
states for the particles comprising the material.
The particle may have electronic, vibrational, and rotational states in addition to
its translational state. Neglecting any interactions between these components, the
total partition function is written

(6.6)

As we are dealing with atoms, not molecules, we may ignore the vibrational
and rotational states. The translational component of the partition function for any
particle is
3
2rrmkT ) !2
Ztrans = ( h2 V' (6.7)
 6.2 EOS for gases and plasmas 161

with the volume V , and the electronic partition function is

'°'
Zelec -_ i...J g t·e-E;0/kT • (6.8)

The "zero" subscript indicates that Ei is measured relative to the ground state. The
sum in equation (6.8) formally extends over an infinite number of electron energy
eigenstates, and each individual term remains finite. Thus, the sum in equation
(6.8) diverges. This divergence, however, is unphysical, because for very large
quantum numbers - as we will see in more detail in Chapter 7 - the eigenstates fill
so large a volume that they overlap adjacent atoms in the material; these states are
so strongly perturbed that they are effectively destroyed. Hence, we can terminate
the sum after a finite number of terms.
In order to determine the EOS for a particular material, then, we first must
determine the partition function for the particles comprising the material, using
equations (6.6), (6.7), and (6.8), then use this partition function in equation (6. 1) to
determine the material's free energy, and finally use the resulting expression for
the free energy in the thermodynamic relations - equations (6.2), (6.3), and (6.4) -
to determine the thermodynamic variables pressure, internal energy, and entropy
as a function of density and temperature. Once the EOS is specified, then we can
solve the equations of motion.

6.2 EOS for gases and plasmas

6.2.1 EOS for monatomic gases
Let us apply the thermodynamic relations of the previous section to a system of
monatomic particles. Let there be N particles in the volume V. We have for the
translational partition function

(6.9)

where

([Link])

with Pi the momentum of the ith particle. Thus, using equation (6. IO) in the
exponential term of equation (6.9), we can write

(6.11)
162 Equation of state
The integral in equation (6.9) is taken over all three axes of coordinate space and
all three axes of momentum space. Substituting equation (6.1 1) into equation (6.9),
we can then write the partition function as

1 VN
Z = --
N!h3N (
f+oo
e - p2/ 2mkT
• dp·
I
) 3N
= -N! (2rrmk
VN
h2
3N
T) /2
.
(6. 12)
- 00

Equation (6.12) is a generalization to N particles of the single-particle partition

function given in equation (6.7).
The free energy, from equation (6. 1), is then

F = - NkT
I [ (2rrmkT)
In V
N h2
3
12] + 1 }.
Now, substituting equation (6.13) into equation (6.3), we find the entropy of the
(6. 13)

monatomic gas:
3
s=- v (2rrmkT)
aT Iv = Nk { In [ N
aF h2 '2] + 1} + 3
2Nk. (6. 14)

The internal energy of the monatomic gas in a volume V is

3
s = F + TS = [Link], (6. 15)

and the pressure is

NkT
P= - - . (6. I 6)
v
Equation (6.16) is the ideal gas equation of state, valid for any single-species
gas of non-interacting particles. It expresses the gas pressure as a function of
temperature and volume (or density). We can also express the gas pressure as a
function of internal energy and density by combining equations (6.15) and (6.16):
P = (2/3) £. / V = (2/3) ps, wheres is the specific energy.

6.2.2 Two-temperature EOS for plasmas

Plasma is not, in general, a single-species fluid. It consists of ions and electrons.
In the simplest picture of plasma consisting of electrons and ions, we can take the
total pressure and specific energy as the sum of the pressures and energies of the
individual components. Thus,

P = P; +Pe (6. 17)

 6.2 EOS for gases and plasmas 163

and

(6.18)

For now, we ignore the contribution of radiation to the pressure and energy. We
will take radiation into account in Chapter 9.
In hot plasma the particles behave like the particles in an ideal gas, so we
can write the individual pressures and energies like those for a monatornic gas,
equations (6.16) and (6.15):
pkT;
Pi = - - (6.19)
Amo
and
3 kT; 3 P;
8; = --- = -- (6.20)
2 Amo 2p
for the ions, and for the electrons we can write

(6.21)

and
3Z*kTe/2 +I
Be = - - - - - (6.22)
Amo
Here, A and mo are the atomic weight and mass, respectively, of the atoms com-
posing the plasma, and Z* and I are the average ionization state and the ionization
energy. We will learn how to calculate these latter two quantities in the next chapter.
As we learned in Chapter 2, the thermalization of the electrons takes place via
electron-electron collisions very rapidly, and so does the thermalization of the ions
via ion-ion collisions, so the plasma is, in the most general case, composed of two
separate species of particles that have two separate temperatures T; and Te. The
ions and electrons interact through electrical forces. Coulomb energy is exchanged
rapidly between the electrons and ions, but thermal energy transfer requires col-
lisions, and the electron-ion collision process is comparatively slow because of
the disparate masses of the ions and electrons. That is, typically (m; /me) Te > T;.
Thus, we can assume that the ions are stationary with respect to the electrons,
and we can calculate the electron entropy and energy from a quantum statistical
mechanics treatment.
We can thus write the electron entropy and energy as

(6.23)
164 Equation of state
and
(6.24)

The total free energy, though, is

(6.25)

so, unlike for the simple hot plasma considered earlier, the pressures and energies
of the electrons and ions do not separate into additive components. Instead, we
must determine the total pressure and total energy from

p2 oF J (6.26)
p = M op TT
e• I

and
(6.27)

where

(6.28)

and

oF IT; ,p.
se -- - oTe (6.29)

Equations (6.26) through (6.29) yield the following basic differentials:

dF = ~ (; ) dp - SedTe - Sid1i (6.30)

and

ds = ~ (;2 ) dp + Ted Se + I;dS;. (6.31)

From equations (6.30) and (6.3 1) we find that the thermodynamic quantities must
be consistent with the following thermodynamic relation:

oP I 2
p oSe I (6.32)
oTe T; ,p =- M op T,,T;.

In practice, what is done to obtain the EOS for a plasma is to set up the quantum
statistical mechanics equations for the differential of the material 's free energy, and
then solve the equations numerically, producing tables of values of total P and total
specific energy s = £ / M as a function of temperature and density for a particular
range of applicability of temperature and density. These tables are then used as part
 6.2 EOS for gases and plasmas 165

of the input to the radiation-hydrodynamics computer code that numerically solves

the equations of motion for the plasma.
These EOS tables have already been created for a large number of materi-
als, including various elements and compounds, including minerals, metals, and
dielectrics. Many of these EOS tables are in publicly available databases. When
using these tables, though, it is important to understand where the values in the
tables come from, how accurate they are for the particular computation that you
are doing, and how applicable they are in the regime of temperature and density
of the plasma under consideration. As we will see later, the results of a radiation-
hydrodynamics computation can be very sensitive to the choice of EOS. It is
important to check that the EOS conforms at all times with the thermodynamic
consistency relation of equation (6.32).

6.2.3 Thomas-Fermi model

In practice, the quantum statistical mechanics calculation for the free energies is a
very difficult calculation, and has been carried out for only a few elements. In dense
plasma, though, we can consider that all the electrons - free and bound - constitute
a single fluid as they move in the Coulomb potential of the atomic nucleus. Then,
instead of considering the plasma as being composed of ions and electrons, we
consider it as being composed of nuclei and electrons, with the electrons obeying
Fermi-Dirac statistics, as discussed in Chapter 2, and the nuclei obeying ordinary
Boltzmann statistics, also discussed in Chapter 2. This picture is the basis for the
Thomas-Fermi model of the atom first developed in detail in 1949 by Richard
Feynman, Nicolas Metropolis, and Edward Teller.
In the Thomas-Fermi model, the plasma is considered to be a collection of
atomic cells, each cell in turn containing a nucleus of charge Ze surrounded by a
spherical "fluid" of Z electrons. The cell volume is thus V = 1/ na, where na is the
atom number density, with a spherical radius

Ro= ( - 3
4nna
)1/3 (6.33)

At high temperatures the separate contribution of the atomic nuclei to the total
pressure and internal energy of the plasma is identical to that for a single-species
monatomic gas, as discussed in Section 6.2.1. We rewrite equations (6.15) and
(6.16) for the nuclei contribution to the pressure and specific internal energy:

Pa= nakT, (6.34)

3 na
ea = lpkT. (6.35)
166 Equation of state
The electron contribution to the total pressure and internal energy is determined
by considering that the electrons constitute a fluid contained in an electrostatic
field, with the fi eld varying with the radial distance from the atomic nucleus. The
electrostatic fi eld is determined by the distribution of the electric charges. Thus,
using Poisson's equation and Fermi-Dirac statistics, it is then possible to write a
differential equation for the radial dependence of the electron number density n (r ) .
We wait until the next chapter - Section 7 .2.3 - to present the derivation of the
differential equation and how this equation is solved to obtain n (r) . We wiII also
learn how to compute the ionization level from the Thomas-Fermi model once we
known (r).
In the meantime, we note that once n (r) is known, then the electron contribution
to the pressure can be determined from
2 2
Pe = 3n(Rokk(Ro) = 5n(Ro)eo(Ro) , (6.36)

where

(6.37)

is the average kinetic energy of an electron at the cell boundary R 0 , with Ro given
by equation (6.33).
Note that in the Thomas-Fermi model there are no complicated quantum-
mechanical calculations required to obtain the material's EOS. The atoms mak-
ing up the dense plasma are treated entirely with classical physics, except for
the semi-classical phase-space quantization already built in to the Fermi-Dirac
electron energy distribution. We also make use of the equivalence between the
statistical mechanics treatment and the continuum fluid treatment of a plasma com-
posed of a coIIection of point particles, an equivalence that we touched upon in
Chapter 4. Thus, the Thomas-Fermi model is relatively easy and straightforward
to apply to high-atomic-number atoms, and has been used to create EOS tables
for a wide variety of materials over a fairly wide range of density and temperature
conditions. In Figure 6.1 we show in graphical form one such EOS table, that for
Al, covering a specific range in density and temperature. Many of these EOS tables
are in publicly available databases.
There are, however, some limitations to the Thomas-Fermi model. As we wiII
see in more detail in the next chapter, the Thomas-Fermi model, in which the
electrons are treated as a continuous fluid, is strictly accurate only for infinite
nuclear charge. The outer electron orbitals for high-Z atoms get closer and closer
together in energy as the radial distance from the nucleus increases, so the continu-
ous fluid approximation becomes a better and better approximation as Z increases.
Furthermore, the Thomas-Fermi model accounts only for the thermal pressure, the
 6.3 EOS for solids and liquids 167

10~

1o<-0 U-..l----1..._...__._-'---''---L--'---'--'--.L........O--'-~~~'---L~-
o 5 10 15 20
Density (g cm-3)

Figure 6.1 The EOS for Al plasma, largely derived from a Thomas-Fermi model
of the atom: pressure as a function of density at temperatures (from bottom curve
to top) of 2.5 x 10-5, 0.5, 1.0, and 2.0 keV.

positive pressure that pushes apart the individual atomic cells. Coupling forces -
attractive forces - between the individual atomic cells are not included. Thus, the
Thomas-Fermi model can work well for plasmas in which coupling forces can
be neglected - plasmas at high temperature, for example, and highly compressed
degenerate plasmas. The model does not work well, however, for low-temperature
solids, or any material in which chemical bonding forces play a significant role. We
need another way to determine EOS suitable for computing shock propagation in
solids, for example, and the EOS of the warm dense matter in giant planet interiors.
We address this question in the next section.

6.3 EOS for solids and liquids

6.3.1 Griineisen EOS
One of the main properties that characterize a solid is that the atoms comprising
the material are at distances from one another that are comparable to the range
of the Coulomb force with which they interact. In other words, each atom is
close enough to its neighbors that it is affected by the electrostatic potential of
the neighboring nuclei. These potentials act to lock each atom in place in an
168 Equation of state

(a) bee

'

'. i.
/·------------ ~~,~·------------ ---
(b) fee

Figure 6.2 Schematic representation of (a) bee lattice structure and (b) fee lattice
structure.

equilibrium position, around which it will make only small thermal oscillations.
The equilibrium positions of the atoms are those positions at which the total free
energy is a minimum. For many materials, the free-energy minimum is obtained
when the atoms arrange themselves into a regular lattice structure, with a high
degree of both short-range and long-range order.
Often the regular lattice structure takes one of several forms in which successive
lattice planes are stacked on top of one another in a close-packed arrangement, such
that the upper plane of atoms sits in the potential well between the atoms of the
lower plane_ Depending on how the stacking is done, this can result, for example,
in a body-centered cubic (bee) structure or a face-centered cubic (fee) structure,
as illustrated in Figure 6.2. Other common arrangements include hexagonal close-
packed (hep) and tetragonal close-packed_ The interested student can consult any
standard text on solid-state physics to learn more details on crystal structure and
its taxonomy_ These details, however, are not necessary for the discussion on EOS
which follows.
The lattice structure is maintained because of a balance between the repulsion
and attraction forces between the neighboring atoms in the lattice. As the lattice
is compressed, this balance is upset. In particular, as the atoms in the lattice are
pushed toward one another, the electron orbitals begin to overlap, and the Coulomb
repulsive forces increase dramatically. Thus, the pressure required to compress the
material increases with the compression.
 6.3 EOS for solids and liquids 169

15

~
.0
e_ ,
l!! 10 ,,
::i
,,
,,
<fl
<fl
l!!
a.. ,'
,'
, ,,
5 ,,
,
,,
,
,/

5 10
Density (g cm-3)

Figure 6.3 1-eV isentrope (solid curve) and isotherm (dashed curve) for solid Al.

At low enough temperatures where we can ignore the thermal pressure of the
electrons (at temperatures below 1 eV or so), we can write the total pressure and
internal energy of the solid as a sum of two components:

P =Pc + Pr , (6.38)
e = ec + er. (6.39)

The first component results from the compression, and is a function only of the
compression. Indeed, from the first law of thermodynamics we see that at zero
temperature the compressional energy gained by the material is simply the work
done in compressing the material, so
dee
P.c =--
dv· (6.40)

Furthermore, the laws of thermodynamics tell us that when the temperature of the
material is zero, so is the entropy S . Thus, at T = 0 the isotherm and the isentrope
are identical, and define what is called the "cold curve" of the material. At modest
temperatures the isotherm and isentrope differ from one another, but not by much.
For example, the T = 1 e V isentrope and isotherm for solid Al are shown in
Figure 6.3.
In general, the cold curve is determined in a full quantum statistical mechanics
calculation, but for those pressures that can be reached experimentally it can also
170 Equation of state
be determined by measurement. We find that the cold curve can be written as a
polynomial in the compression µ, = p /Po - l:

(6.41)

where the constants a;, i = 0, l, 2, 3 are determined from the quantum mechanics
calculation or the experiments, and are different for different materials. Note from
equation (6.41) that for small compressions, µ, « 1, lattice pressure is approx-
imately linear with compression. This does not mean, however, that the lattice
pressure is smaII. Recall that solids have very low compressibility - and hence
very high adiabatic sound speed - compared to gases, so we say that the material
is "stiff." Even at modest compressions the lattice pressure can be very high com-
pared to atmospheric pressure; for tantalum at l 0% compression, for example, the
cold lattice pressure is a little over 200 kbar.
Now let us turn our attention to the thermal component of the material pressure
and energy. First we combine equations (6.2) and (6.4) above, and make use
of the thermodynamic identity of equation (6.32) to find a new thermodynamic
consistency equation:

p
2 &e I
&p T = &PI
p - T &T p .
(6.42)

Now, recalling that the cold compression pressure is a function of compression

only, we can rewrite equation (6.42) as

I
p 2 -&e =Pc(p) +PT(p,T) - T
&p T
&PT
-
&T
I.
P
(6.43)

Recalling that the compressional component of total internal energy is a function

only of density, while the thermal component is independent of density, and making
use of equation (6.40), we can rewrite equation (6.43) as

PT(p,T) - T&PT
- = 0. I (6.44)
&T P

The solution of equation (6.44) is then

PT(p, T) = f(p)T, (6.45)

where f is some function of density only.

Let us now rewrite equation (6.45) as

PT(p, T) = f (p)CvpT, (6.46)

 6.3 EOS for solids and liquids 171

where Cv is the material's specific heat at constant volume, defined in equation

(5.19), and

l Pr
r(p) = -- (6.47)
p Sr

is the Grtineisen coefficient, which is proportional to the ratio of the material's

thermal pressure to its thermal energy. At STP conditions, where the pressure is
1 atm and the density is the standard density p0 , the Grtineisen coefficient is the
Grtineisen constant 1 0 .
The Grtineisen constant can be related to the STP isothermal compressibility or
bulk modulus Ko of the material, given by equation (5.71), and its coefficient of
thermal expansion at constant pressure,

(6.48)

We leave it as an exercise for the student to show that

a
f'o = - - - (6.49)
PoCvKo
As we discussed in Chapter 4, solids are much less compressible than gases; that
is, they have a lower bulk modulus (and correspondingly higher adiabatic sound
speed). For many metals, the Griineisen constant is r 0 ~ 2. The values for the
Grtineisen constant for Al and Fe are listed in Appendix III.
Note that the assumption we made in deriving the Griineisen coefficient - that
the thermal oscillations of the atoms remain small - led us to the conclusion that the
coefficient is independent of temperature. This assumption breaks down, however,
as the thermal motion of the atoms increases. As the thermal motion increases, the
atoms no longer make small oscillations around their equilibrium positions. Rather,
the motions become more and more random, just like in a monatomic gas.
Thus, as temperature increases and the material becomes more and more rar-
efied, the thermal component of the Griineisen EOS must approach the EOS for
a monatornic gas. Comparing the EOS for a monatornic gas, equations (6.15) and
(6.16), with the thermal component of the Griineisen EOS for a solid, equation
(6.47), we see that as T ---* oo and p ---* 0, then r ---* 2/3. In contrast, as the
material is compressed to densities higher than the STP density, the Griineisen
coefficient increases with compression, that is, r(p > p 0 ) > r 0 .
Finally, we write the full Griineisen EOS for solids, as in equation (6.38), as the
sum of the compressional or cold component and the thermal component. Here,
we express the compressional component in the Steinberg formulation - a variant
172 Equation of state
of the polynomial formulation of equation (6.41) - so the full Griineisen EOS can
be written as
PoC2J.L [1 + {1 - fo) J.L - Qµ,2]
P = s
2
[l - (S1 - 1)µ,]2
2
+ (1o + b11)s
,..., '
(6.50)

where the compression factor µ, = p /Po - 1, cs is the STP adiabatic sound speed
of the material, and b and S 1 are material constants. The coefficient b defines how
the Grtineisen coefficient linearly increases with compression, and the constant S 1
defines how the shock Hugoniot relating pressure and density varies from linear.
The values of b and S 1 are given for Al and Fe in Appendix III. Values for other
solid materials can be found in the 1980 paper of Steinberg and other articles cited
in the Further Reading.
Up to now we have ignored the pressure and energy of the electron "fluid" in
the solid. For solids at temperatures between room temperature and a few thousand
degrees Kelvin, it is entirely proper to do so, since the electron pressure is much
smaller- by factors of one to two orders of magnitude - than the total atom pressure
as computed, say, from equation (6.50). As the temperature increases, however,
some of the electrons in the Fermi-degenerate electron gas acquire energies greater
than the Fermi energy EF = (l/2)mev}, where me is the electron mass and VF,
the velocity of the electrons at the Fermi energy level, is given by equation (2.23).
Since these electrons acquire an additional energy of order kT, the total thermal
energy per electron is of order (kT / EF)kT. Thus, the specific thermal energy of
the electrons in the solid can be written as

(6.51)

Using equation (2.23), expressing the dependence of the Fermi energy on density,
we find that we can write the coefficient f3 as

Po)
/3 = /30 (-
2
/3
. (6.52)
p
The constant f3o is given for Al and Fe in Appendix III. The electron pressure is
then
l
Pe= lPBe. (6.53)

Since the lattice pressure and energy of the solid increase linearly with tempera-
ture, whereas the electron pressure and energy increase as the square of the temper-
ature, the electron pressure can start to become comparable to the lattice pressure at
fairly modest temperatures (that is, temperatures as low as l eV), and dominate the
lattice pressure as the temperature approaches the Fermi temperature TF = E F / k.
 6.3 EOS for solids and liquids 173

Typical metallic Fermi energies are several electron volts. At temperatures much
higher than the Fermi temperature, of course, the free electron gas is no longer
Fermi degenerate, and acts just like a monatomic ideal gas, so its pressure and
energy are given by equations (6.2 1) and (6.22).
Finally, we note that the formulation of the Griineisen EOS for solids does not
apply to nanocrystals. Nanostructured materials are typically only one to a few atom
layers thick, so a large fraction of the atoms in the material are "surface" atoms
that do not experience the bonding forces of neighbors in every direction. Thus,
the electron distribution of these surface atoms is highly distorted from spherical.
This highly distorted electron distribution changes the atomic potentials, the bond
strengths, and hence all the mechanical, thermal, and electrical properties of the
material. Another way of saying this is that, in nanomaterials at low temperature,
surface energies are greater than internal energies. The same is true for a liquid
soap bubble, which is held together by surface tension, so that the liquid bubble
has entirely different physical properties than the liquid from which it was formed.
Of course, at extreme conditions surface energies become negligible compared to
the internal energies.

6.3.2 EOS for porous solids

Let us now consider shock propagation in a solid material. Since, as we have
seen above, the pressure of a solid at even very modest shock compressions is
P » Po = 1 bar, we can write the shock Hugoniot, equation (5.12), as

1
e - eo =- P (Vo - V) . (6.54)
2
If we now express the internal energy of the solid as the sum of the compressional
and thermal components, and relate the thermal component of the internal energy
to the thermal pressure using equation (6.47), we can rewrite equation (6.54) as

v 1
ec (V) + - ( P - Pc) = - P (Vo - V). (6.55)
ro 2

Here we have assumed that the Griineisen gamma remains constant with compres-
sion, which is a reasonably good approximation for shocks that are not too strong,
that is, shocks that do not melt or vaporize the material.
Rearranging equation (6.55) we can then write

(6.56)
174 Equation of state
Finally, defining a new variable K r = 2/ r 0 + 1, we find the Hugoniot pressure
for a shock-compressed solid:

p = ( K r - 1) Pc _ 1.e
v c (6.57)
K r - v,y
Note from equation (6.57) that the parameter K r sets an upper limit to the shock
compression of a solid. As we saw above, for most metals the Griineisen constant is
r 0 :::::::: 2, so in the limit of infinite shock strength, as P ---+ oo then p /Po ---+ 2. This
behavior is completely analogous to that of the shock compression of a gas with
a constant ratio of specific heats y, which we saw in Chapter 5, equation (5.29),
has a compression limit of (y + 1) / (y - 1) in the limit of infinite shock strength.
Indeed, since r = y - 1, then
2 2 y +1
Kr= - + 1 = - - + 1 = - - . (6.58)
r y-1 y - 1
Another thing to note about equation (6.57) is that the compressional part of the
Hugoniot pressure can be written as some multiplying factor times the cold pressure.
Since K r is always greater than one, the Hugoniot curve for a solid material always
lies above its cold curve.
Now let us consider what modifications we must make to this formulation for
a porous solid. A porous solid is a material that consists of regions of full-density
matter separated from each other by empty voids. The voids can be non-connecting,
as in foams made in a chemical gel process, for example, or connecting, as in many
fibrous materials. Compacted powders are also porous solids. Since the voids take
up space in the material, the uncompressed material volume V00 is always larger
than the volume of the uncompressed full-density material, Vo. We can thus define
a porosity coefficient k = PolPoo > 1. (Note: The porosity coefficient is not to be
confused with the Boltzmann constant, which is denoted by the same symbol.)
Let us now consider shock compression of a porous solid with porosity coeffi-
cient k. We will be concerned here only with shock strengths large compared to
the material strength, so we can ignore the elastic behavior of the material. We
will also assume that the shock strength is sufficiently high that the pores, or voids,
close completely. Thus, we ignore the low-pressure phase of void closure. There
are a number of existing pressure- distension EOS models (called P-a models)
and strain-distension EOS models (called £- a models) for the low-pressure phase
of void closure, but these are beyond the scope of our discussion. The interested
student may consult the papers listed in the Further Reading for more details on
these models.
The key physics that distinguishes compression of a porous solid from compres-
sion of a full-density solid is that there must be additional applied pressure to do
 6.3 EOS for solids and liquids 175

Porous solid

Density

Figure 6.4 The Hugoniot curve for a porous solid in pressure-density space, along
with the Hugoniot curve for the equivalent full-density solid and its cold curve.

PdV work to close the voids of the porous solid. In other words, pressure must be
applied to compress the solid from density Poo to density Po , then more pressure
to compress from p 0 to the final compressed state at density p. Thus, the Hugoniot
pressure for a porous solid is always greater than the Hugoniot pressure for the
equivalent full-density solid, as illustrated schematically in Figure 6.4.
If we combine equations (6.50) and (6.57), we can write the compressional
component of the pressure of the porous solid as

2 [i+(l- Jf)JL- ¥ ]
Pc = {>OC5 JL { 2 } (6.59)
[l - (S1 -1 )] (Kr -k-kJL)

and the compressional component of the internal energy as

Pc [k(l + JL) - l]
(6.60)
2Po( l + JL)
It is clear from equation (6.59) that it takes more pressure to compress a porous
solid to a particular compression, JL > 1, than to compress the full-density material
to the same compression. As explained above, this is because of the compressional
energy gained in collapsing the voids. We can also see that the larger the porosity, the
larger the compressional pressure for a given compression. The pressure approaches
infinity as the porosity increases. Indeed, we can see from equation (6.59) that Pc --+
oo as p /Po --+ Kr/ k. Thus, this prescription for the EOS for porous materials is
applicable only for modest compressions and porosities. For typical metals, for
which Kr ~ 2, equations (6.59) and (6.60) are valid for porosities that are no more
than ~50% (i.e., k = 2).
176 Equation of state
Furthermore, we have assumed complete void collapse. This assumption is
typically valid for pressures large compared to the material strength. Even when
the pressure does exceed the material strength, complete closure of the pores
may depend on pore size. If we are modeling compression of nanoporous or
microporous materials with low porosity, like some composites and powders for
example, then equations (6.59) and (6.60) are generally applicable. In contrast, this
EOS formulation would not work for fibrous materials with very large pores. For
those sorts of materials one would need to take account of the mechanics of the
pore collapse itself, a topic beyond the scope of our discussion here.
The compressional component of the internal energy of the compressed porous
material is itself the sum of two components - energy gained from collapsing
the pores and energy gained from compressing the full-density material. Looking
back at Figure 6.4, we see that the energy gained in collapsing the pores is just
the difference between the total PdV compressional energy integrated along the
Rayleigh line - that is, the area of the right triangle shown in Figure 6.4 with
vertices at (Pc, p), (0, p), and (0, Poo) - [this is just £c given by equation (6.60)]
and the compressional energy of the full-density material, £cp. which is obtained
by integrating along the Hugoniot of the full-density solid:

(6.61)

We leave the integration as an exercise for the student.

The importance of including the void-collapse energy in the EOS is illustrated in
Figure 6.5, where we plot the specific internal energy for cold (T = 0) 9% porous
Ta as a function of density. We see that, even for this modest porosity, the pore-
collapse energy is comparable to the compressional energy, and the total internal
energy is double that for initially full-density Ta at fairly modest compression.

6.3.3 Phase transitions

As we learned earlier, a solid is characterized by its atoms undergoing only very
small thermal oscillations around equilibrium positions, with the equilibrium posi-
tions determined by where the atoms have the minimum free energy. As the tem-
perature increases, the amplitude of the thermal oscillations increases. At some
temperature, the solid transitions from a solid state to a liquid state. At this phase
transition there is a discontinuous change in the density. For most materials the
transition from a solid state to a liquid state is accompanied by a decrease in the
density.
 6.3 EOS for solids and liquids 177

4000

OJ 3000
.....
2.
>-
e>
<D
g 2000
0
:;::: Compressional
·5
<D energy ----;.'----A-
a.
(f) 1000

18 20 22 24
Density (g cm-3)

Figure 6.5 Specific internal energy as a function of density for 9% porous Ta

(upper solid curve) and for full-density Ta (lower solid curve).

The temperature at which the solid-liquid transition takes place, referred to

as the melt temperature Tm, can be defined in a number of different ways. Melt
temperature defined by the atomic positions is somewhat arbitrary. This is because
the difference in atomic positions between the solid and liquid states is not always
well-defined. Indeed, when looking at an image of atomic positions formed by an
electron microscope, there is no way to distinguish between an amorphous solid
and a liquid. Even if one sees a lattice structure with a high degree of long-range
order, one may not necessarily be looking at a solid; some liquids can display a
high degree of long-range order under certain temperature and density conditions.
Nonetheless, one customary definition of melt temperature - the one that is
commonly used to define the multi-phase EOS used in many hydrodynamic sim-
ulation codes - is the Lindemann formulation, in which the melt temperature is
that temperature at which the thermal oscillations of the atoms have amplitudes
equal to the average distance between the atoms. In the compressed solid, even
though the atoms are closer together, more thermal energy is required to overcome
the higher potential energy barrier, so, in general, melt temperature increases with
compression. The Lindemann law (as adapted by Steinberg et al. in 1980) is written

T,11 = T,110 exp [ 2a ( 1 - ~) JT/2(fo-a- I/ 3)' (6.62)

where 1J = p/Po is the compression, a is a material constant usually determined

by fitting equation (6.62) to data, r 0 is the Griineisen constant defined in equation
(6.49), and Tmo is the melt temperature for the zero-pressure solid.
178 Equation of state
For many materials the solid-liquid phase boundary is reasonably well charac-
terized by the Lindemann law. Some materials, however, have an anomalous melt
curve, that is, a melt boundary in temperature-pressure space that has a negative
slope over some range of pressures. Water is one such material: whereas for most
materials the liquid is less dense than the solid, liquid water at and near atmospheric
pressure is denser than water ice. It is this anomalous property of water that allowed
the evolution of life on Earth, since when ponds and lakes freeze in the winter the
ice layer buoyantly floats, and hence grows from the surface down; deep water does
not freeze all the way to the bottom. There are a few elements that have anomalous
melt curves, among them bismuth and germanium.
Another common way to define the melt temperature is the temperature at which
the material strength vanishes. In one sense this is a better way than the Linde-
mann law to characterize melt temperature, since it accounts for differences in melt
temperature for different rnicrostructures of the same material. In contrast, as we
learned in the last chapter, there is no single constitutive model for material strength
that is applicable for all solids over all possible ranges of conditions - temperature,
strain, strain rate - and microstructures. Furthermore, material strength depends on
dynamic history - so-called work hardening - and can include hysteresis effects.
Thus, to define melt temperature analytically as the temperature at which mate-
rial strength vanishes is a daunting task. There is currently no formula like the
Lindemann law that accomplishes this task.
Of course, the rigorous way to define melt temperature is to do the quantum-
statistical mechanics calculations of partition functions and free energies for both
the liquid and the solid, and then find those positions in temperature-pressure
space at which the solid and liquid free energies are equal but for which there is
a discontinuity in the density. The curve in temperature-pressure space defined by
these points is the melt curve. These quantum-statistical mechanics calculations
are typically lengthy, complex, and tedious, and have been carried out for only a
small fraction of all known materials.
Since a discontinuity in density also means a discontinuity in adiabatic sound
speed, phase boundaries can also be found experimentally in shock experiments.
Shock experiments, however, can access only pressures along the Hugoniot, so
any given shock Hugoniot intersects the melt curve in only one point. Thus, it is
difficult and tedious to map out the full phase boundary with shock experiments,
since one must perform many shock experiments at different initial temperatures
of the solid.
Furthermore, the first phase boundary - that is, the phase boundary at the lowest
Hugoniot pressure - that is seen in shock experiments is not necessarily the melt
curve (the solid-liquid phase boundary). This is because most crystalline solids
have several different and distinct solid phases, characterized by different lattice
 179

Figure 6.6 The room-temperature Hugoniot of Fe overlaid onto the Fe phase

diagram.

arrangements of the atoms. Water, for example, is known to have, as of this writing,
15 solid phases. Most water ice with which we are familiar at atmospheric pressure
is solid phase I, which has a hep lattice structure. Solid phase XI, which exists at
pressures above about 3 Mbar, also bas a hep structure, but different mechanical,
thermal, and electrical properties from ice I. Water ice also displays at least three
distinct amorphous phases.
Not all metals have as complicated a phase diagram as water ice, but most do
have multiple stable solid phases. As one example, the sub-Mbar phase diagram
for Fe is shown in Figure 6.6. The stable STP structure of Fe is the a phase, a bee
crystalline structure. Note that when Fe is heated at constant pressure, it transforms
first from the a phase to the y fee phase before reaching the melt boundary. In
contrast, the room-temperature shock Hugoniot, also shown in Figure 6.6, takes
the material first across the a-to-s phase boundary, in which the atoms rearrange
themselves into a hep crystalline structure. Additional solid phases exist at higher
pressures.
In actual practice in computer code simulations of experiments, it is necessary
to have an EOS formulation for each material that covers the entire wide range
of temperature and density conditions to which the material is subject during the
course of the experiment. Even though the experiment objective may be focused
on creating and diagnosing a relatively narrow range of extreme temperature and
density conditions, we typically start with a cold solid, and the initial conditions
180 Equation of state
are important in describing accurately how we get to the final extreme conditions.
This means that the EOS must be a multi-phase EOS. Many (but not aII) of the
EOS tables that exist in the pubHcly available databases are multi-phase EOS
tables, but the EOSs for the different phases are not necessarily stitched together in
a thermodynamicaily consistent way at phase boundaries. These thermodynamic
inconsistencies translate into numerical discontinuities at phase boundaries that
can lead to propagating errors in the solution of the equations of motion. One must
therefore exercise great care in specifying the EOS tables to use for any particular
problem. More wiII be said about numerical stability in Chapter 11.
One way around this problem of thermodynamic consistency in a multi-phase
EOS has been provided by Richard More and coileagues. In 1988 they created an
analytical multi-phase EOS suitable for use in simulation codes. In their model they
get the electron pressures, energies, and entropies from a modified Thomas-Fermi
model, like the one described in Section 6.2.3 above. They modified the Thomas-
Fermi model by adding in a bonding correction term so that the cold pressure
is zero in the fu11-density solid, as it should be. The bonding correction term is
insignificant at higher temperatures and at densities away from the solid-density
region. They get the ion or atom pressures, energies, and entropies by stitching
together the Debye, Griineisen, Lindemann, and fluid scaling laws in a smooth and
thermodynamicaily self-consistent way.
Richard More and coileagues intended their EOS to be a general-purpose EOS
that could be adapted easily for routine use in hydrodynamic simulation codes, so
they cailed their EOS the quotidian EOS , or QEOS. The main advantage of the
QEOS is that it provides thermodynamicaily consistent pressures and energies that
are smooth functions of density and temperature free from numerical noise. Another
significant advantage is that the QEOS can be written as a cailable subroutine
in a hydrodynamic simulation code that provides the required EOS quickly and
efficiently without the need to access and interpolate external databases.
There are, however, some disadvantages to the QEOS. It does not include solid-
solid phase transitions, and only poorly models the liquid-vapor phase transi-
tion and the EOS for polyatomic molecular materials. Accordingly, most modern
hydrodynamic simulation codes are built with the flexibility to obtain the material
EOS in several different ways, including computing it from the QEOS model,
extracting it from external databases, and ailowing the user to input a user-defined
EOS.

6.4 Example problems and exercises

In doing these problems and exercises, the student can use the thermodynamic and
material properties for select materials listed in Appendix III.
 6.4 Example problems and exercises 181

Exercise 6.1. Use Stirling's formula,

N!~ ( ~ ) N,
to derive equation (6.13) for the free energy of a monatomic gas from equations
(6.1) and (6.12).
Exercise 6.2. Use Stirling's formula to show that the free energy of a polyatomic
gas - that is, a gas composed of several species of atoms with numbers of atoms
NA, N 8 , ... , each species with a Maxwell-Boltzmann distribution - is

F = -NAkT ln ( ;Ae ) - N 8 kTln ( ; ; ) - · · ·

when the total partition function can be written

z NA zNo
Z= -A_,_B_ ···
NA! Nn!

Exercise 6.3. Derive equation (6.32).

Exercise 6.4. What is the Thomas-Fermi atom radius for STP Fe compared to that
for STP Al?
Exercise 6.5. Make use of the thermodynamic relation

to derive equation (6.49) relating the Griineisen constant to other thermodynamic

properties of the material.
Exercise 6.6. At what temperature is the electron pressure in standard-density Al
equal to its lattice pressure?
Exercise 6.7. Derive equation (6.57) from equation (6.56).
Exercise 6.8. Show that the Griineisen coefficient is related to the adiabatic
index - that is, the ratio of specific heats - by r = y - 1.
Exercise 6.9. Carry out the integration of equation (6.61). Hint: Rewrite the inte-
grand as a sum of partial fractions, then integrate analytically each separate partial
fraction.
Exercise 6.10. What is the melt temperature of 20% compressed Fe, that is, Fe
compressed to 20% above its STP density? What is the pressure of the Fe at this
compression and temperature? What is the electron pressure?
182 Equation of state
Exercise 6.11. What is the melt temperature of 20% compressed Al, that is, Al
compressed to 20% above its STP density? What is the pressure of the Al at this
compression and temperature? What is the electron pressure?
Exercise 6.12. To what temperature must we preheat full-density Fe in order to
shock compress it from the a phase to they phase, bypassing entirely the s phase?
 7
Ionization

During the latter part of the nineteenth century and the early years of the twentieth
century numerous experiments were performed that probed matter on a very fine
scale. The observations required explanation, and thus began the development of
atomic theory. Experiments using the scattering of charged particles, as well as
electromagnetic radiation, demonstrated that the atom has structure. Rutherford's
experiment established that the mass of an atom is concentrated in a very small
nucleus, while its volume and its physical and chemical properties are determined
by a comparatively loose surrounding structure of electrons. Other experiments,
such as the conduction of electricity through rarified gases, showed the existence of
free electrons. Milliken's oil drop experiment provided direct confirmation that the
electron's charge-to-mass ratio is a constant, and that the charge on a droplet can
be measured in integral amounts. Numerous explanations were advanced for these
observations, some with success and some without. One of the more impressive
results from the Lorentzian theory of the electron was the determination of the
classical radius of the electron.

7.1 Electron structure of atoms

To explain his results, Rutherford advanced a planetary model for the atom, that
of planetary electrons orbiting a solar nucleus. This model, however, encountered
serious difficulties. The laws of classical mechanics predict that the electron would
emit electromagnetic radiation while orbiting the nucleus. As a result of the radi-
ation emission the electron would lose energy and would gradually spiral inward,
collapsing onto the nucleus. The classical model predicts that all atoms are unstable.
In addition, as the electron spirals inward the emission would increase in frequency
as the orbit got smaller and faster, and one would see a smear in frequency of
electromagnetic radiation. Experiments showed, however, that light is emitted only
at certain discrete frequencies.

183