Chap 4 Molecular Velocity Distribution

Lecture notes of Chin-Hsiang Cheng

Velocity – Direction and Magnitude

Direction- All directions of molecular velocities are assumed to be equally probable.

Magnitude- Maxwell–Boltzmann distribution

The Maxwell-Boltzmann distribution is shifted to higher speeds
and is broadened at higher temperatures.
Lagrange Multipliers
In mathematical optimization problems, Lagrange multipliers is a method for finding
the local extrema of a function of several variables subject to one or more constraints.
This method reduced a problem in s variables with k constraints to a solvable problem
in s+k variables with no constraints.
The method introduces a new unknown scalar variable, the Lagrange multiplier, for
each constraints and forms a linear combination involving the multipliers as coefficients.

Find extreme of f(x, y) under the constraint of g(x, y)=c .

The necessary condition is 𝛻𝑓 = −λ𝛻𝑔. That is, one

can observe the optimal value is attained when is
𝛻𝑓 is parallel to 𝛻𝑔 . Then

𝛻{f(x, y) + λ [g(x, y) – c]} = 0

Let Φ(x, y, λ) = f(x, y) + λ [g(x, y) – c] be the objective function.

As Φ(x, y, λ)  min, f(x, y)  min and g(x, y)  c

The extreme of Φ(x, y, λ) can be found as 𝛻Φ = 𝛻{f(x, y) + λ [g(x, y) – c]} = 0.

That is

𝜕Φ
=0
𝜕𝑥
Three equations for determination of xm, ym and λ.
𝜕Φ
=0
𝜕𝑦 (xm, ym) is the point of extreme.
𝜕Φ
=0
𝜕λ
Example 1

Find the maximum of f(x, y) = x2y subject to the constraint of g(x, y) = x2+y2 = 3

Solution:

Let Φ(x, y, λ) = f(x, y) + λ [g(x, y) – c] = x2y + λ (x2+y2 – 3)

𝜕Φ
= 2xy + 2λ𝑥 = 0 (1)
𝜕𝑥

𝜕Φ
= x2 + 2λ𝑦 = 0 (2)
𝜕𝑦

𝜕Φ
= x2+y2 − 3 = 0 (3)
𝜕λ
Eq. (1) leads to λ =- y. Substituting this result into Eq. (2) one yields x2 - 2y2 = 0.

Then, one obtains four extreme points

( 2, 1), (− 2, 1), ( 2, -1), (− 2, -1)

Φ( 2, 1)=2; Φ(− 2, 1)=2; Φ( 2, -1)= -2; Φ(− 2, -1)= -2

Therefore, the objective function attains a maximum at

( 2, 1), (− 2, 1)
Example 2

Find the discrete probability distribution {pk, k=1, 2, 3, .., n} with maximal information
entropy. The information entropy is defined by Sk = - pk ln pk. Note that the sum of these
probabilities equals 1.0.

Solution:

The objective function of this problem is total entropy

f(p1, p2, ….., pn) = - 𝑘=1 pk

𝑛
ln pk

Constraint: g(p1, p2, ….., pn) = 𝑘=1 pk

𝑛
= 1.0

Let Φ(p1, p2, ….., pn) = f(p1, p2, ….., pn) + λ [g(p1, p2, ….., pn) – 1.0]
= - 𝑛𝑘=1 pk ln pk + λ ( 𝑛𝑘=1 pk – 1.0)
 𝜕
𝑘=1 pk ln pk + λ ( 𝑘=1 pk – 1.0)) =0
𝑛 𝑛
(-
𝜕pk

Carrying out the differentiation of these n equations, one has

1
-(
ln 2
+ln pk ) + λ = 0

This shows that all {pk, k=1, 2, 3, .., n} are equal (because they depend on λ only).
By using the constraint 𝑛𝑘=1 pk = 1.0, one finds

pk = 1/n

Hence, the uniform distribution is the distribution with the greatest entropy.
dNvx : Number of nodes within vx ~ vx + dvx

(1) dNvx ~ function of vx

vz
(2) dNvx ~ dvx

dNvx
Thus, dNvx = g(vx ) dvx or = f(vx ) dvx
N

probability distribution function


Note: − f(v x ) dv x = 1
vy
Similarly,
vx
dNv𝑦
= f(vy ) dvy
N
dNv𝑧
= f(vz ) dvz
N
Assumption 6: All directions are equally probable.

(1) f(vx ) = f(vy ) = f(vz )

dNvx
(2) = f(vx ) dvx so dNvx = N f(vx ) dvx
N
d2NVx,V𝑦 dNv𝑦
(3) = =f(v y ) dv y so d2NVx,V𝑦= dNvx f(v y) dv y= N f(vx ) f(v y) dvx dvy
dNvx N

d3NVx,V𝑦, V𝑧 dNv𝑧
(4) 2 = =f(v z ) dvz so d3NVx,V𝑦, V𝑧 =d2NVx,V𝑦 f(v z) dvz= N f(vx ) f(v y ) f(v z) dvx dvy dvz
d NVx,V𝑦 N
 d3NVx,V𝑦, V𝑧
Let  = Density of nodes (molecules) in the VELOCITY SPACE
dvx dvy dvz
vz
 : density of molecules in the velocity space
n: number density in the physical space v

d3NVx,V𝑦, V𝑧 vy
= = N f(vx ) f(v y ) f(v z)
dvx dvy dvz

vx
Find f(vx ), f(v y ) and f(v z) that make d=0
Along the surface at v=C,  is
Constraint: vx2 + vy +
2 vz2= v2 = C (constant) uniformly distributed, i.e. d=0.

Let Φ =  + λ [vx2 + vy2 + vz2 - C ] Note:  = (v). On the surface of the

circle at v=C, d=0.

𝛻Φ = 𝛻{  + λ [vx2 + vy2 + vz2 − C] } = 0

𝛻{ N f(v x ) f(v y) f(v z)+ λ [vx2 + vx2 + vx2 − C] } = 0

𝜕Φ
= 0, N f ‘(v x )f(v y) f(v z)+ 2 λ vx = 0
𝜕vx

f ′(v x ) 2 λ vx 2λ v
or =− = −  x = − vx
f(v x ) N f(v x ) f(v y) f(v z)

−
v x
2
Therefore, f(v x ) = a 𝑒 2

Similarly

𝜕Φ −
v y
2
= 0, f(v y ) = a 𝑒 2
𝜕vy 2λ
=

𝜕Φ vz2
= 0, f(v z ) = a 𝑒 − 2 a: an integration constant
𝜕vz
 = N f(vx ) f(v y ) f(v z)

(vx2+vy2+vz2)
= N a3 𝑒 − 2

v2
= N a3 𝑒 − 2

Let 2= /2

=N a3 𝑒 − 2v2

How to determine a and ?

Determination of a and :

(1) Conservation in molecule number:

v
N = v=0 dNv

v
= v=0  (4v2dv)

v  Note: dNv = N a3 𝑒 − v 4v2dv

2v2 2 2
= v=0 (N a 𝑒
3 − ) 4v2dv

1  
N = 4 N a3  a=
2 3 2 
(2) Conservation in total translational energy:

v 1 3
U = v=0 mv 2 dNv = N k T k: Boltzman constant
2 2

v 1 2 3 𝑒 − v 4v2dv = 3 N k T
2 2
v=0 2 mv N a
2

3mN 3
= N kT
42 2

m  m
=  a= =
2kT  2kT
Maxwell-Boltzmann velocity distribution

Number of molecules within the velocity interval v ~ v+dv

dNv = N a3 𝑒 − v 4v2dv
2 2

3
dNv 4N m 2 2 −( m )v2
(1) = ( ) v 𝑒 2kT
dv  2kT

3
dNv 4 m 2 −(2kT)v2
m
(2) = f(v)dv = ( ) v2 𝑒 dv
N  2kT

3
(3) f(v) =
4
(
m 2
) v2 𝑒 −(
m
2kT
)v2
 2kT
 mv2
- − is the ratio of kinetic energy to thermal energy
2kT

3
−(2kT)v2
m
m
- ( )2 𝑒 is the Boltzmann distribution
2kT

3
4N m 2 −(2kT)v2
m
dNv = ( ) v2 𝑒 dv
 2kT

1
N m 2 −(2kT)vx2
m
Gaussian
dNvx = ( ) 𝑒 dvx
 2kT
3
d3NVx,V𝑦, V𝑧=
N
(
m 2
) 𝑒 −(
m
2kT
)v2 dv dv dv
3/2 2kT x y z
Various measures of the characteristic speed

(1) Average speed

v dNv 2 1.128
v= = =
N   

(2) Root mean square velocity

v2dNv 3 1 1.224
vrms = V =
2 = =
N 2  

(3) Most probable velocity

df(v) 2kT 1
= 0  vmp = =
dv m 

Note: vrms > v > vmp

Error function and fraction of molecules in specific speed range

2 𝑥 −𝑥 2
erf(x) = 𝑒 d𝑥
 0

1
Vx N m 2 −(2kT)vx2
m vx
N0x = dNVx dNvx = ( ) 𝑒 dvx and let x = =  vx
0  2kT vmp
N 𝑥 −𝑥 2
= 𝑒 d𝑥
0 
N 
= erf(x)
 2

N
= erf(x)
2

EX:
N
N01 = (0.8427) = 0.421 N
2

N N N
Nx = N0 −N0x = − erf(x) = [1 − erf(x)]
2 2 2
Show that the number with speed between 0 and v is

2 −𝑥2
N0x = N [erf(x) - 𝑥𝑒 ] where x = v
 vmp
Energy distribution
Molecular kinetic energy
1
 = m v2
2
m  2 −1/2
( ) v2 = , v2 = , dv= (2m) d
2kT kT m
3
4N m 2 −(
m
)v2 dv
dNv = ( ) v2 𝑒 2kT
 2kT

4N
3
m 2 2 −
 
dN= ( ) ( ) 𝑒 kT (2m)−1/2 d
 2kT m kT

1 3
3
2N − − 2 2
dN= (kT) 2 1/2 𝑒 kT d

Most Average
probable energy
energy

dN 2N 3 
d
=  −
(kT) 2 1/2 𝑒 −kT

dN 2 −
3
−kT

= (kT) 2 1/2 𝑒 d
N 