Chapter 12

Elementary kinetic theory
of transport processes

Aman. W
• In the previous chapters our concern has been almost
exclusively with equilibrium situations.
• General statistical arguments were quite sufficient to treat
problems of this sort, and there was no need to investigate in
detail the interaction processes which bring about the
• we can derive the specific heat, and many other gas properties
using an equilibrium approach
• Results of Kinetic Theory: KE of individual particles is related
to the temperature of the gas:
½ mv2 = 3/2 kT
• Where v is the average velocity.
Equilibrium Distributions for a Gas
• At equilibrium, we can use Maxwell-Boltzmann statistics to determine the
gas distribution for the relevant properties

æ m ö2
é m (v2 + v2 + v2 ) ù
f (v ) = ç ê ú
velocity (Gaussian) x y z
÷ exp -
è 2 p k BT ø êë 2kBT úû
4 æ m ö2 2 é mv 2 ù
f ( v) = ç ÷ v exp ê- ú
speed (Maxwellian)
p è 2kBT ø ë 2kBT û

e é e ù
f (e ) = 2 exp ê- ú
2 ( kB T )
energy (Maxwellian)
ë k BT û

e = k BT
• However, many problems of great physical interest deal with non
quilibrium situations.
• For example,
– the two ends of a copper rod are maintained at different
• This is not an equilibrium situation, since the entire bar would not
be at the same temperature.
• Then, energy in the form of heat flows through the bar from
– the high to the low temperature end,
– the rate of this energy transfer being measured by the "thermal
conductivity“ of the copper bar
• A calculation of the coefficient of thermal conductivity thus requires
a more detailed consideration of the non quilibrium processes
whereby energy is transported from one end of the bar to the other

• Introduction to Non equlibrum Dynamics
• Collisions Time
• Basics of Transport
– Diffusion --- particle number
– Energy ---- Thermal conduction and
– Momentum --- Viscosity
• Diffusion
– Random walk
• The Diffusion Equation
• Phonons are characterized by their
– Energy, wavelength (wave vector)
– polarization (direction)
– branch (optical/acoustic), acoustic phonons are the primary
thermal energy carriers
• Phonons are quantized lattice vibrations
– store and transport thermal energy
– primary energy carriers in insulators and semi-conductors
• Phonons have a statistical occupation (Bose-Einstein), quantized
(discrete) energy, and only limited numbers at each energy level
• We can treat phonons as particles and therefore determine the thermal
conductivity based on kinetic theory, we can derive the specific heat!
• Electrons are particles with quantized energy states
– store and transport thermal and electrical energy
– primary energy carriers in metals
– usually approximate their behavior using the Free Electron
• Energy, wavelength (wave vector)
• Electrons have a statistical occupation (Fermi-Dirac),
quantized (discrete) energy, and only limited numbers at
each energy level (density of states)
– we can derive the specific heat!
• We can treat electrons as particles and therefore determine
the thermal conductivity based on kinetic theory
Introduction to non equlibrum

• Unlike both classical thermo and statistical mechanics, kinetic
theory may be used to describe non-equilibrium situations.
• Kinetic theory has been particularly useful in describing the
properties of dilute gases.We can use non-equilibrium kinetic
theory to determine
– the thermal conductivity,
– viscosity, and
– diffusivity of gases
• It gives a deeper insight into concepts such as
– pressure, internal energy and
– specific heat
Non equilibrium-collsion

• In gas systems, gas molecules interact with each other through
• If such a gas is initially not in an equilibrium situation, these
collisions are also responsible for bringing about the ultimate
equilibrium situation where a Maxwell-Boltzman velocity
distribution prevails.
• We shall discuss the case of a gas which is dilute. i.e
– Each molecule spends a relatively large fraction of its time
at distances far from other molecules so that it does not
interact with them.
– In short, the time between collisions is much greater than
the time involved in a collision.
• Triple or more collisions occur very rarely compared to two-
particle collisions.
• Thus the analysis of collisions can be reduced to the relatively
simply mechanical problem of only
– two interacting particles.
• The behavior of a molecule between collisions can then be
described adequately by
– the motion of a wave packet or
– classical particle trajectory, even though a quantum-mechanical
calculation may be necessary to derive the scattering cross
section describing a collision between two molecules.
Collisions Time

• Let's start by considering N molecules in a gas of volume V.
• Consider a molecule with velocity v. Let
– P(t) = the probability that such a molecule survives a time t
without suffering a collision.
– P(0) = 1, since a molecule has no chance of colliding in a
time t→0,
• On the other hand, P(t) decreases as the time t increases
• Finally
– P(t) →0 at t →∞
• The net result is that a plot of P(t) versus t must have the shape
indicated in Fig.
• To describe the collisions, let define
– ?dt = the probability that a molecule suffers a
collision between time t and t + dt.

• The quantity ω is thus the probability per unit time that a
molecule suffers a collision, or the "collision rate.“
• Knowing the collision probability, ?, it is possible to
calculate the survival probability P(t).
• How?
• This can be done by as follows
• [the probability that a molecule survives a time t + dt without
suffering a collision] =
[the probability that this molecule survives a time t without
suffering a collision] X [the probability that it does not suffer a
collision in the subsequent time interval between t and t + dt]
• In symbols, this statement becomes
? ? + ?? = ?(?) 1 − ???
= −??
• To fix the normalization so that

න ? ? ?? = 1
• which guarantees that no particle lasts forever without undergoing a
• When we solve for
? ? = ?? −??
• Let ? = ?ҧ be the mean time between collisions. But this is
exactly the quantity that we called the relaxation time.
• we can compute the average time between collisions.

? = න ?? ? ?? =
0 ?
• Hence, we have got that

– 1/τ is the collision rate(?).
The Mean Free Path

• The average distance
traveled by a molecule
between two successive
collisions is called the
"mean free path“, ι, of the
• we can simply gas collisions
using a hard-sphere, binary
collision approach (billiard
• One has thus got
?= ? ?
Basics of Transport

• We now turn to the question of how things move.
• To understand how certain macroscopic properties move when
out of equilibrium.
• It is important to know how much time it takes for a system to
approach an equilibrium state.
• A system is not in equilibrium when the macroscopic
parameters (T, P, etc.) are not constant throughout the system.
• To approach equilibrium, these non-uniformities have to be
ironed out through the transport
• What kind of transport?
Basics of Transport
– energy, ----------- Thermal conduction
– momentum,----------- Viscosity and
– mass ------- Diffusion
– from one part of the system to another.
• The mechanism of transport is to estimate the characteristic
rates of approaching equilibrium and thus to impose
limitations on the rates of “quasi-static” processes.
• Processes in which these quantities change over time are
usually referred to as transport.


• Drop a blob of ink into a glass of water.
• How does it spread?
• More generally, we are interested in the motion of a particular
kind of particle such as
– one with a nice color, or
– smell
• As it makes its way through a generic background of liquid or
• Thus the flow of randomly moving particles caused by
variations of the concentration of particles.--- Diffusion
• Example: a mixture of two gases/liquid, the total concentration
n = n1+n2 = const over the volume (P = const).

n1 J n2

• The kind of motion can explained by
random walk
Random Walk-1-dimensional

• Consider a lattice which, for now, we take to be one dimensional.

• Let N steps of equal length be taken along a x-dimension
• The spacing between the lattice sites is set by the mean free path, ι,
and after a time, τ ,
– the particle jumps either left or right.
• The direction of the jump is entirely random:
– ½ of the time it goes left, and
– ½ right.
• This model is known as a random walk.
• The particle starts at the origin and we want to know the
probability P(x, t) that it sits at x = ml at time t = Nτ .
• To get to x = ml the particle must have made
– 1/2 (N +m) forward jumps and
– 1/2 (N-m) backwards jumps.
• The probability is just the number of different ways
2? −? 2 ?ൗ 2
? ?, ? = ? 2? ?
• The probability distribution of the particle is an ever-spreading
Gaussian ensemble.
• The mean displacement is
– <x>=0,
– reflecting the fact that the
particle is equally likely to
travel forwards as
• The variance is
?2 = ?
• The root-mean-square (rms)
distance travelled by the
particle grows

?2 ~ ?

• This is characteristic
behavior of random walks.
The Diffusion Equation
• We can recast the above discussion in terms of a differential
equation for the density of particles, n = N/V.
• The density is not a constant(out of equilibrium). It is, in
general, a function of time and space.
• We expect any gradient, Δn, in the density of particles to lead
to a flow, from the high density region to the low.
• We'll again restrict first to the case of one-dimension. Consider
the density at some fixed time:
– n = n(x, t).

The Diffusion Equation

• To derive an expression
for the density at the point
x a short time Δt later
• Any particle which is at x
at time t +Δt must have
been sitting at some other
position x-Δx at time t.
• Here Δx should be viewed
as a random variable since
some move one way some
the other.
• This means that we can write an expression for the density at time t
+ Δt as an average over all the different Δx,

? ? + ∆?, ? = ? ?, ? − ∆?

• Taylor expanding the right-hand side,
?? 1 ?2? 2
? ? + ∆?. ? = ? ?, ? − ∆? + ∆?
?? 2 ?? 2
• The term with the first order derivative vanishes because on average
particles are equally likely to go either way, meaning ∆? = 0 .
?? ?2?
=? 2
?? ??
• Where the diffusion constant is
∆? 2
• We expect this to be related to our two quantities, the mean
free path ι and scattering time τ .
• The solution which corresponds to an initial condition that all
particles are at x =0 at t =0 is an ever-spreading Gaussian,
1 −? 2ൗ
? ?, ? = ? ? 4??
• we again have the result
? 2 = 2??
• It is simple to extend the derivation above to three dimensions.
It becomes as
?? ?2 ? ?2 ?
= ? 2 2 2 = ? 2
?? ?(? +? +? ) ??

• This is also known as Fick's law.
Viscosity-Momentum Transfer

• Viscosity is a form of internal friction experienced by a fluid. It
can be measured by placing a fluid between two plates, a
distance d apart in the z direction.
• Holding the lower plate stationary, the top plate is moved at a
constant speed, u, in the x direction.
• But the fluid pushes back.

Laminar flow of a gas
(fluid) between two
surfaces moving with
respect to each other.
• If you want to keep the
plate moving at a constant
speed, you have to apply a
force F
• Near the upper plate, a
friction force causes the
fluid to be dragged along
with the same speed u.
• However, near the lower
plate, the fluid remains
• This sets up a velocity gradient, ux(z), with
– ux(d) = u and
– ux(0) = 0.
• Drag – transfer of the momentum in the direction
perpendicular to velocity.
 px A  u x, top  ux, bottom
 Fx 
t z
Fx d ux u
=η η
A Δz d
– Fx – the viscous drag force, η - the coefficient of viscosity and
– Fx/A – shear stress
• where the second equality holds for small distances d. The
coefficient of proportionality, η, is called the viscosity
• The viscosity after some algebra becomes
? = ???(?)
Thermal conductivity: Particle Approach

• Phonons and electrons (and photons) possess wave-like
– to track waves we need to know amplitude, phase, direction  very
• We already treat gases as particles and we also like to treat
phonons, electrons, and photons as particles as well  we’ve
already applied kinetic theory to derive thermal conductivity!

free electron gas
phonon gas

gas … gas
Thermal conductivity
• A static flow of energy from a “hot” object to a “cold” one. (At what rate
the internal energy is transferred between two systems with T1 ≠ T2 or
between parts of a non-equilibrium system (if one can introduce Ti ?)

area A

T1 T2


• The temperature distribution in this formulation is time-independent, and
we need to calculate the flux of thermal energy JU due to the heat
conduction (diffusion/intermixing of particles with different energies,
interactions between the particles that vibrate but do not move
“translationally”). T
QU =  K th
Thermal conductivity

• In general, the energy transport due to molecular
motion/viberation is described by the equation of heat
T 1 QU K th  2T
= =
t C x C x 2
QU =  K th

• Thus, in principle, if you know the initial conditions, e.g.
T(x, t=t0), you can describe the process by solving the
• Often, you are asked to consider a different situation:
Mechanisms of heat conduction

• Heat is transferred by phonons (lattice vibration waves) and
• The thermal conductivity of a material is defined by combined
contribution of these two mechanisms:

? = ?? + ??
• where
– ?? the lattice thermal conductivities. and
– ?? electronic thermal conductivities.
– Lattice conductivity: Transfer of thermal energy phonons
– Electron conductivity: Free (conduction band) electrons
equilibrate with lattice vibrations in hot regions, migrate to
colder regions and transfer a part of their thermal energy
back to the lattice by scattering on phonons.
• The electron contribution is dominant in metals and absent in

• Wishing U all the best in your Future Carrier

Master your semester with Scribd & The New York Times

Special offer for students: Only $4.99/month.

Master your semester with Scribd & The New York Times

Cancel anytime.