DERIVATION OF BOLTZMANN EQUATION FOR

CHANGE OF STATES – WITHOUT AND WITH

COLLISIONS
By C. DEVASRI
20PPH005
INTRODUCTION:
• We have already learnt about how a microstate of a classical system can be represented by a point
in a phase space.
• For a system having N particles in 3 dimensional phase – the phase space is 6N- dimensional.
• For some states this may change , Eg: For inelastic collisions ,the number of particles may change .
In this case the dimensionality of the phase space changes.
• For a single particle in a 3 dimensional system ,the phase space is
6 dimensional, the point in this phase space can be denoted by (r, p).
• At a given time t, the phase-space density of particles will be denoted by the function f(r , p, t).
This means that, in small phase-space volume d3rd3p located at a point (r, p) , the number of
particles at time t is given by f(r, p, t)d3rd3p.
• We are assuming that the particles doesn’t have any internal degrees of freedom like
spin, then in order to treat particles with spin, f should be replaced by gf in any
formula where the spin degeneracy factor g does not appear explicitly.
• If a system is not in equilibrium, the function f(r, p, t) will change with time. The
BOLTZMANN EQUATION gives the rate of this change.
• In this we are going to see about the Boltzmann equation for the change of states
without and with collisions.
WITHOUT COLLISION:
• The motion of a particle in a non-collision state is called FREE-STREAMING, since the
particle glide freely without disturbing one another.
The part of the change is trivial. In absence of collisions, a particle which is at
position r and time t will move to the position r’ at time t’, where
r’ = r + p/m dt. …...(1)
If there is an external force F on each particle , the momentum will change from p to p’ , where
p’= p+ F dt . …….(2)
Thus the same particle which occupied the small volume d3rd3p at time t will occupy a small
volume d3r’d3p’ at t+dt, where r’ and p’ are given by eqn (1) and (2).
In absence of collisions , we should have
f(r, p, t)d3rd3p= f(r’, p’, t’)d3r’d3p’ . ……(3)
• The first order in dt , the phase-space volume element remains unchanged , so that we obtain
f(r, p, t)= f( r+ p/m dt, p+F dt, t+dt) ……….(4)
Lets us find the equation for a 2-D phase space.
Here we have to calculate the change in particle
number in this 2-D phase-space.
= number of particles (coming out @ t+dt – in @ t )

=f ( ( х , vx , t+dt) –f (x, Vx,t ) ) dxdvx

= df ( x, vx , t ) dxdvx ( total change)

here Vx= p/m , x = r.

•   be calculated by
Can
Number of particles (into face 1 –out of face 2)
Number of particles (into face 3 – out of face 4)
into FACE 1 = df( x, v x, t)dxdvx= dx dt = axdxdt
out of FACE 2 = df( x, v x+ dvx, t)dxdvx = axdxdt
into FACE 3 = df( x, v x, t)dxdvx = dtdvx = vxdvx dt
out of FACE 4 = df( x, v x+ dvx, t)dxdvx = vxdvx dt

total change = 1 – 2 + 3 – 4
df dx dvx = ax df dxdt + vxdvx dt df

Divide both sides by dvxdxdt and using partial and rearranging we get.

+ vx + a x = 0 a x= F
Replacing vx , ax , x.

+ + F. = 0 ……....(5)
This is called the collisionless Boltzmann equation also called as VLASOV EQUATION.
WITH COLLISIONS:
•The
  simple form of the equation obviously does not hold once we take collisions into account .
In general , then , we write
│free + │coll .........(6)
Where the two terms on the right hand side give the contribution from free- streaming and from free
collisions respectively.
The first term has been already calculated .
The equation (5) really means that
│free = ─ ─ F. ………(7)
so lets us symbolically represent the collision term by C . Then eqn (6) takes the form.
+ + F. = │coll
or
+ + F. = C …………(8)
This is the general form of Boltzmann equation.
Now we have to calculate the collision term or C . Like the distribution function , the collision term is in general a
function of r, p and time t.
COLLISION TERM :
• A collision can involve any number of particles in the initial and the final states. The collision
can even be inelastic, so that the final state can have different number of particles compared to
the initial state.
• Here we assume that the system contains only one kind of particles, and in each collision two
such particle will collide elastically. Therefore the final state will have the two particles of the
same kind, although with different kinds of momenta.
• To understand collision change f(r, p, t),first we must have to differentiate between the two
kinds of collisions .

 First – A particle with momentum p might be in the initial state of collision which produces
particles with some other momenta in the final state.
Second -- A particle with momentum p may be produced in the final state of collision where
the initial involves other momenta.
•The
  distribution function f(r, p, t) decreases in the first kind and increases in the second kind .
Thus , it can be written with the notation ,
(ini) │ coll+ (fin) │coll ……………(9)

We have to find expression for the two terms , for that we have to look at the collisions or scatterings.
For the first term, let us say that in particular scattering process, the two particles in the initial state have momenta

1 1 ’
p and p whereas those in the final state have momenta p’ and p . The final state momentum will not be
uniquely determined by the initial state.
• Since the final momenta are undetermined, we need to integrate over all possible values of p1
, because the particle with momentum p can scatter against a particle with any other
momentum.
• We should exclude forward scattering in the integration, because f(r, p, t) does not really
change either p’ or p1’. Let us consider it later

okay , so we know that the expression must be in the form of an integral over p1, p’ and p1’.
But what should the integrand ?
Here , Boltzmann used an ansatz, called “Stosszahlansatz”, which means the assumption of
molecular chaos.
Obviously, the collision cannot take place if there is no particle with momentum p at the point
r , or more precisely, if f (r, p, t) itself vanishes.
Similarly, the collision also requires the presence of a particle of momentum p1 at the same
point. ( i,e) requires f(r, p, t) to be a non-zero as well.
• So
  Boltzmann assumed that the integrand is the product of these two factors times a quantity
that represent the scattering rate in some way.
we can write down the first term on the right hand side of the eqn (9) as
(ini) │ coll = f f1 ω (p’,p1’│p,p1), ………(10)
Where we have used the shorthands
f ≡ f (r, p, t) f1 ≡ f (r, p1, t) ……(11)
and ω (p’,p1’│p,p1) contains the information about scattering, so it is called as the scattering
factor.
The expression on the other term in eqn(9) is now simple to obtain . For this case , the
momentum p should be in the final state , thus
+ + F. = [ f’ f1’ * ω (p,p1│p’,p’1)-
f f1 ω(p’,p1’│p,p1), …..(12)
This is Boltzmann equation, including effects of binary elastic collisions only.
REASON FOR NOT CONSIDERING INTEGRATION OF THE
FORWARD SCATTERING IN THE INTEGRALS:

• If we don’t subtracting it out, it does not make difference in the Boltzmann equation .
• To see why :
Suppose p=p’. Conservation of momentum then implies that p1=p1’ as well.
Both occurrences of ω in eqn (12) then have the same set of momenta as their arguments.
The factor of phase-space density function also becomes the same for both terms.
Thus the integrand of eqn (12) vanishes, which means that forward scattering
Does not contribute to the collision term C at all.
This does not hold for inelastic collisions.
REFERENCE:

An Introduction To Statistical Mechanics By Palash B Pal.

THANK YOU