FEL laser induced heating of the electron gas in metal

Agnieszka [my maiden name :-)]

DESY September ’04

I wish to thank my supervisor,

dr Jacek Krzywinski,
for introducing me to the absolutely supercool
world of plasma physics.

Abstract
The laser heating of plasma electrons via the inverse Bremsstrahlung process is
considered. The theoretical model of plasma electrons assumes Fermi distribution
functions and describes the above mentioned process by a collision integral. The
kinetic equation is derived, and the change in average kinetic energy of electrons is
obtained.

1 Introduction
Plasma may absorb a significant part of the laser light by inverse Bremsstrahlung process.
During that process, the plasma electron gains energy from the electromagnetic field of
the laser beam by absorbing photons while colliding the nucleus. We consider the inverse
Bremsstrahlung absorption of Bremsstrahlung laser radiation. The beam is treated as a
classical plane electromagnetic wave. We consider a perfectly homogeneous and isotropic
crystal. There is no lattice deformation during the collision process. We neglect the energy
transport since our sample is a thin film metal. We investigate a very rare gas so that we
consider only binary collisions and particles are completely uncorrelated before the collision
(molecular chaos). According to these assumptions we can make use of the standard
semi-classical transport theory with its Boltzmann equation and calculate dynamics of the
electron gas with standard collision functions. The plasma electrons are described by the
solution of the Schrödinger equation for the electron in the electromagnetic field of the laser
beam. The scattering of the electrons by nuclei is treated using first-order perturbation
theory. A Fermi-Dirac distribution function for the electrons is assumed. The transition
probabilities are used to write the kinetic equation for the electrons. The rate of change
of the kinetic energy is derived for a weak-field case. Asymptotic cases are considered.

1
2 Kinetic electron equation
z

~
∆k

θ2
~k
θ

φ y
φ2

e
x

Let us consider the electromagnetic wave propagating in the z direction in our figure.
The Schrödinger equation for an electron in the field of a classical electromagnetic wave is:

h̄ ∂ϕ(~r, t) 1 h̄ ~ e~
− = | ∇ − A(t) |2 ϕ(~r, t), (1)
i ∂t 2m i c
and its solutions are to represent the unperturbed electron states:
 Z t 
~ i h̄ ~ e~ 0 2
ϕ(~r, t) = exp ik · ~r − | ∇ − A(t ) | dt0 , (2)
2mh̄ 0 i c

where ~k is the wave vector.

We treat the field of the atom’s nucleus as a perturbation and according this we use the
first-order perturbation theory to calculate the transition probability between two states
(1 → 2):
T /2
i
Z Z
a(1 → 2) = − ϕ?2~r, t)V (~r)ϕ1 (~r, t)d3 rdt
h̄ −T /2
T /2    (3)
i i λ
Z
~
= − Ṽ (∆k) exp Ωt − sin(ωt − δ) dt
h̄ −T /2 h̄ ω

2
where
Ze2 −~r/a
V (~r) = e is the Yukawa potential ,
~r
2
~ = 4πZe
Ṽ (∆k) is the Fourier transform of V (~r) ,
1/a2 + ∆k 2
h̄2 (k22 − k12 )
Ω= ,
2m (4)
eE0
λ= h̄∆k⊥ ⊥ with respect to the z direction,
mω  
−1 ∆ky
δ = tan
∆kx
~k2 − ~k1 = ∆k~ is the exchanged momentum and ~k1 = ~k .

The transition probability per unit time is then

~ |2 ∞  
| a(∆k) 2π  
~ 2
X
2 λ
= | Ṽ ∆k | Jn δ(Ω − nh̄ω) , (5)
∆t h̄ n=−∞
h̄ω

where Jn is the Bessel function of order n.

The transition probability per unit time with the absorption or emission of n photons:
 
~ 2π  
~ 2 2 λ
T (n, ∆k) = | Ṽ ∆k | Jn δ(Ω − nh̄ω) , (6)
h̄ h̄ω
The δ function implies the energy conservation in the electron-photon system. Since the
nucleus carry of some momentum during the collision process, the momentum of the system
is not conserved.
The change in the number of the electrons N(~k) can be expressed schematically as:

~k + ∆k
~ ~k + ∆k
~ h̄ω ~k h̄ω ~k

∂N (~k) P∞ P
+ − −
∂t = n=1 ∆k

~k h̄ω ~k ~k + ∆k
~ ~k + ∆k
~ h̄ω

which means
∞
∂N(~k) X X ~ ~k)(1 − N(~k + ∆k))
~ + T (−n, −∆k)N(
~ ~k1 ) [1 − N(k2 )]
= [T (n, −∆k)N(
∂t n=1 ~
∆k

− T (−n, ~k2 − ~k1 )N(~k2 ) [1 − N(k1 )] − T (n, ~k2 − ~k1 )N(~k2 ) [1 − N(k1 )] (7)
∞
[T (n, ∆k)[N(~k) − N(∆k ~ + ~k)] .
X X
=
n=−∞,n6=0 ∆k
~

3
 We express the electron kinetic equation as a change of the number of electrons with
the momentum h̄~k in time:
∞
∂N(~k)
[T (n, ∆k)[N(~k) − N(∆k
~ + ~k)] .
X X
= (8)
∂t n=−∞,n6=0 ~
∆k

~ become an
Assuming a Fermi distribution for the electrons, letting the sum over ∆k
integral, and using Eq. (6), we expand Eq. (8) to

∂f (~k) 4πZe2
 
2π 1 1 1
Z
2
= | | 2 − 2
∂t h̄ 8π 3 1/a2 + ∆k 2 1 + e−(EF −h̄ k2 /2m)/KT 1 + e−(EF −h̄ (k+∆k)2 /2m)/KT
∞  
X
2 λ
× Jn δ(Ω − nh̄ω)d3∆k .
n=−∞
h̄ω
(9)

The electron-photon collision term for inverse Bremsstrahlung in a metal, describing the
free electrons absorption of the energy of laser light of angular frequency ω and amplitude
of electric field E0 reads

∂f (~k) 4πZe2
 
2π 1 1 1
Z
2
= | | 2 − 2
∂t h̄ 8π 3 1/a2 + ∆k 2 1 + e−(EF −h̄ k2 /2m)/KT 1 + e−(EF −h̄ (k+∆k)2 /2m)/KT
∞  
X
2 eE0 h̄∆k⊥
× Jn 2
δ(Ω − nh̄ω)d3 ∆k .
n=−∞
2mh̄ω
(10)

3 Average change of the electrons kinetic energy

The energy gain for a given electron in plasma will be found for the case of the weak
electromagnetic field.
For this case λ << h̄ω and the the approximation of the square of the Bessel function
is    2  2|n|
2 λ 1 λ
Jn ≈ . (11)
h̄ω n! 2h̄ω

4
 Using Eq. (9) with only first two terms of the sum of the Bessel retained, we have

∂f (~k) 4πZe2
 
2π 1 1 1
Z
2
= | | 2 − −(E −h̄2 (k+∆k)2 /2m)/KT
∂t h̄ 8π 3 1/a2 + ∆k 2 1 + e−(EF −h̄ k2 /2m)/KT e F
2 " #
h̄2 ∆k 2 2h̄2~k ∆k ~ h̄2 ∆k 2 2h̄2~k ∆k~

λ
δ( + + h̄ω) − δ( + − h̄ω) d3 ∆k
2h̄ω 2m 2m 2m 2m
4πZe2
 
2π 1 1 1
Z
2
= | | 2 − −(E −h̄2 (k+∆k)2 /2m)/KT
h̄ 8π 3 1/a2 + ∆k 2 1 + e−(EF −h̄ k2 /2m)/KT e F
" #
 2 2 2 2~ ~ 2 2 2~ ~
eE0 h̄∆k⊥ h̄ ∆k 2h̄ k ∆k h̄ ∆k 2h̄ k ∆k
2
δ( + + h̄ω) − δ( + − h̄ω) d3 ∆k .
2mh̄ω 2m 2m 2m 2m
(12)

Let the θ stand for the angle between ~k and ∆~k, and z = cos θ. The unitary vector
~n = (nx , ny , nz ) points into the same direction as the electric field vector.
~ described in spherical coordinates is
∆k
~ = (∆k sin θ cos φ, ∆k sin θ cos φ, ∆k cos θ) .
∆k (13)

The coordinate of ∆~k parallel to the ~n (the electromagnetic field direction) reads
~ n.
∆kk = (~n · ∆k)~ (14)
~ ⊥ = ∆k
The perpendicular coordinate is ∆k ~ − ∆k
~ k . The squared modulus of ∆k
~ ⊥ is

~ − ∆k
(∆k ~ k )2 = (∆k)
~ 2 − 2∆k
~ ∆k ~ k + (∆k
~ k )2
~ 2 − 2(~n · ∆k)
= (∆k) ~ 2 + (~n · ∆k)
~ 2 (15)
= (∆k)2 1 − (sin θ cos φnx + sin θ sin φny + cos θnz ) 2


.
In the next step we assign cos θ as z1 and z2 to appropriate parts of the integral
q q
2
(∆k) (1 − (nx 1 − zi cos φ + ny 1 − zi2 sin φ + nz zi )2 )
2
q
= (∆k)2 1 − n2x (1 − zi2 ) cos2 φ − n2y (1 − zi2 ) sin2 φ − n2z zi2 − 2nx nz zi 1 − zi2 cos φ (16)
q
2


− 2nx ny (1 − zi ) sin φ cos φ − 2ny nz zi 1 − zi2 sin φ .

λ2 defined by (4) expands to:

e2 E02 2
λ2 = 2 2 2 2 2 2


h̄ (∆k) 1 − nz zi − (nx + ny )(1 − zi )/2 . (17)
m2 ω 2
Since the variable φ appears in the integral only via expression (15), all we have to do
is to integrate (15) over φ. Integrating from φ to 2π gives:
(∆k)2 2π 1 − n2z zi2 − (n2x + n2y )(1 − zi2 )/2 .


(18)

5
 We can write down the Dirac deltas in (12) as:
 2
h̄ (∆k)2 2h̄2 k∆kz
 2
h̄ (∆k)2 2h̄2 k∆kz
 
δ + + h̄ω − δ + − h̄ω
2m 2m 2m 2m (19)
= δ(g1 (z)) − δ(g2 (z)) ,

where

g1 (z) = h̄2 (∆k)2 /(2m) + h̄2 2k∆kz/(2m) + h̄ω , (20)

g2 (z) = h̄2 (∆k)2 /(2m) + h̄2 2k∆kz/(2m) − h̄ω . (21)

The roots of g1,2 are

h̄(∆k)2 + 2mω
z1 = − , (22)
2h̄k∆k
h̄(∆k)2 − 2mω
z2 = − . (23)
2h̄k∆k
The derivatives of g1,2 are given by the following equation:

g10 (z) = g20 (z) = 2h̄2 k∆k/(2m) . (24)

The Dirac delta can be written down as:

!
h̄2 (~k + ∆~k)2 h̄2~k 2
δ − + h̄ω (25)
2m 2m
!
h̄2 (~k + ∆~k)2 h̄2~k 2
δ − − h̄ω . (26)
2m 2m

There are two distributions integrated. The first one does not depend on the variable
2 ~ ~k)2
of integration and the other depend on it via h̄ (k+∆
2m
. The presence of the Dirac deltas is
to take into account the energy conservation during the collision process. Owning this the
distribution functions can be taken outside the integration over spherical angles. According
2~ 2
that we can divide the integral into to parts (one Dirac delta each). The we can put h̄2mk −h̄ω
2~ 2 h̄2 (~k+∆~k)2
or h̄2mk + h̄ω respectively for 2m
and finally transfer the distribution functions as
explained above.

6
 With the above tasks executed we can integrate over z and φ

~
∂f (k) 2 6 2
2πZ e E0  1 1

=  −
mh̄3 ω 4 k  1 + e−(EF −h̄ k2 /2m)/(KT ) 1 + e−(EF −(h̄ k2 −2mh̄ω)/2m)/(2mKT )
2 2
∂t


(∆k)3
Z
× d(∆k)
(1/a2 + (∆k)2 )2
! !
2 2 2 2
(h̄(∆k) + 2mω) (h̄(∆k) + 2mω)
× 1 − n2z − (n2x + n2y ) 1 − /2
4h̄2 k 2 (∆k)2 4h̄2 k 2 (∆k)2
  (27)
1 1
+ 2 − 2
1 + e−(EF −h̄ k2 /2m)/(KT ) 1 + e−(EF −(h̄ k2 +2mh̄ω)/2m)/(2mKT )
(∆k)3
Z
× d(∆k)
(1/a2 + (∆k)2 )2
! !
2 2 2 2
(h̄(∆k) − 2mω) (h̄(∆k) − 2mω)
× 1 − n2z − (n2x + n2y ) 1 −

/2 
.

2 2 2 2 2 2

4h̄ k (∆k) 4h̄ k (∆k)
It is easy to change the result given by Eq. (27) to suit a coordinate system described
by Fig. 2, in which the field vector is oriented along the ẑ axis and ~k vector is arbitrary.
We notice, that Eq. (27) depends on ~k only through k = |~k|, and on ~n only through n2x
and n2x + n2y . Since ~n is unitary, we have n2x + n2y = 1 − n2z . However, nz k is equal to the
component of ~k parallel to the field vector. Switching to spherical coordinates, we have
the radial coordinate of ~k equal to k (no substitution) and angular coordinates equal to θ
and φ (according to Fig. 2).
Coordinate θ is an angle between the field vector and ~k, hence its cosine is equal to nz .
To transform Eq. (27) to the new coordinate system, if suffices to substitute n2z with cos2 θ
and n2x + n2y with 1 − n2z = sin2 θ:

∂f (~k) 2πZ 2 e6 E02 

1 1
(k, θ, φ) = −


mh̄3 ω 4k  1 + e−(EF −h̄ k2 /2m)/(KT ) 1 + e−(EF −(h̄ k2 −2mh̄ω)/2m)/(KT )
2 2
∂t


(∆k)3
Z
× d(∆k)
(1/a2 + (∆k)2 )2
! !
2 2 2 2
(h̄(∆k) + 2mω) (h̄(∆k) + 2mω)
× 1 − cos2 θ 2 2 − sin2 θ 1 − /2
4h̄ k (∆k) 2 4h̄2 k 2 (∆k)2
 
1 1
+ 2 − 2
1 + e−(EF −h̄ k2 /2m)/(KT ) 1 + e−(EF −(h̄ k2 +2mh̄ω)/2m)/(KT )
(∆k)3
Z
× d(∆k)
(1/a2 + (∆k)2 )2
! !
2 2 2 2
(h̄(∆k) − 2mω) (h̄(∆k) − 2mω)
× 1 − cos2 θ 2

− sin θ 1 − /2 .


2 2 2 2 2 2

4h̄ k (∆k) 4h̄ k (∆k)
(28)

7
After integrating over angles θ and φ

∂f (k) 8π 2πZ 2 e6 E02

= 3 4 ×
∂t
 3 mh̄ ω k
(∆k)3
Z
 1 1
− d(∆k)


 1 + e−(EF −h̄2 k2 /2m)/(KT ) 1 + e−(EF −(h̄2 k2 −2mh̄ω)/2m)/(KT )

(1/a2 + (∆k)2 )2

(∆k)3
 Z
1 1 
+ − d(∆k)


2 2 2 2 2 2 2
−(E −h̄ k /2m)/(KT ) −(E −(h̄ k +2mh̄ω/2m))/(KT ) (1/a + (∆k) ) 

1+e F 1+e F

16π 2 Z 2 e6 E02 

1 1
= ×  1 + e−(EF −h̄2 k2 /2m)/(KT ) − 1 + e−(EF −(h̄2 k2 −2mh̄ω)/2m)/(KT )


3mh̄3 ω 4 k
 Z
(∆k)3

1 1 
+ − d(∆k) .


2 2 2 2
1 + e−(EF −h̄ k /2m)/(KT ) 1 + e−(EF −(h̄ k +2mh̄ω/2m))/(KT )  (1/a2 + (∆k)2 )2


(29)

We integrate over ∆k and we obtain


16π 2 Z 2 e6 E02 

∂f (k) 1 1
= −


3mh̄3 ω 4k  1 + e−(EF −h̄ k2 /2m)/(KT ) 1 + e−(EF −(h̄ k2 −2mh̄ω)/2m)/(KT )
2 2
∂t

 
1 1 
(30)
+ −


2 2 2 2
−(E −h̄ k /2m)/(KT ) −(E −(h̄ k +2mh̄ω)/2m)/(KT )

1+e F 1+e F 
 
1 1 2 2
+ log(1 + a (∆k) ) .
2 (1 + a2 (∆k)2 )

3.1 The integral over ∆k limits

Now we have to find the upper and lower limits of the integral over ∆k from the condition
that −1 < z1,2 < 1.

8
• for z1 :

h̄(∆k)2 + 2mω
−1 =−
2h̄k∆k
2
h̄(∆k) − 2h̄k∆k + 2mω = 0
(∆k)2 − 2k∆k + 2mω/h̄ = 0
∆ = 4k 2 − 8mω/h̄
p p
∆k1 = k + k 2 − 2mω/h̄ , ∆k2 = k − k 2 − 2mω/h̄
(31)
2
h̄(∆k) + 2mω
1=−
2h̄k∆k
2
h̄(∆k) + 2h̄k∆k + 2mω = 0
(∆k)2 + 2k∆k + 2mω/h̄ = 0
∆ = 4k 2 − 8mω/h̄
p p
∆k3 = −k + k 2 − 2mω/h̄ , ∆k4 = −k − k 2 − 2mω/h̄

• for z2 :

h̄(∆k)2 − 2mω
−1=−
2h̄k∆k
2
h̄(∆k) − 2h̄k∆k − 2mω = 0
(∆k)2 − 2k∆k − 2mω/h̄ = 0
∆ = 4k 2 + 8mω/h̄
p p
∆k1 = k + k 2 + 2mω/h̄ , ∆k2 = k − k 2 + 2mω/h̄
(32)
2
h̄(∆k) − 2mω
1=−
2h̄k∆k
2
h̄(∆k) + 2h̄k∆k − 2mω = 0
(∆k)2 + 2k∆k − 2mω/h̄ = 0
∆ = 4k 2 + 8mω/h̄
p p
∆k3 = −k + k 2 + 2mω/h̄ , ∆k4 = −k − k 2 + 2mω/h̄

9
 We must take into account the fact the radial coordinate ∆k has a positive value and
choose the limit from ∆k > 0 range only.
Then the integration for ∆k are

1. integral with z1
p p
∆k ∈ [k − k 2 − 2mω/h̄, k + k 2 − 2mω/h̄] (33)

2. integral with z2
p p
∆k ∈ [−k + k 2 + 2mω/h̄, k + k 2 + 2mω/h̄] (34)

We integrate over ∆k in the respective limits


16π 2 Z 2 e6 E02 

∂f (k) 1 1
= −


3mh̄3 ω 4k  1 + e−(EF −h̄ k2 /2m)/(KT ) 1 + e−(EF −(h̄ k2 −2mh̄ω)/2m)/(KT )
2 2
∂t


 2 2 p 2
× −4a h̄ k k − 2mω/h̄ + (h̄2 (1 + 4a2 k 2 ) − 4a2h̄mω + 4a4 m2 ω 2 )
" p #
−2a2 mω + h̄(1 + 2a2 k(k − k 2 − 2mω/h̄))
× (log
h̄
" p #
−2a2 mω + h̄(1 + 2a2 k(k + k 2 − 2mω/h̄)) 
− log )
h̄

  (35)
1 1
+ 2 − 2
1 + e−(EF +h̄ k2 /2m)/(KT ) 1 + e−(EF −(h̄ k2 −2mh̄ω)/2m)/(KT )

p
× −4a2 h̄2 k k 2 + 2mω/h̄ + (h̄2 (1 + 4a2 k 2 ) + 4a2 h̄mω + 4a4 m2 ω 2 )

" p #
−2a2 mω + h̄(1 + 2a2 k(k + k 2 − 2mω/h̄))
× (log
h̄
" p # 
−2a2 mω + h̄(1 + 2a2 k(k + k 2 + 2mω/h̄))  
− log ) .


h̄


10
3.2 Fermi energy temperature dependence
The Fermi energy temperature dependence can be derived from a Fermi-Dirac distribution
function normalization. A Maxwell-Boltzmann distribution is a good approximation for a
Fermi-Dirac in a classical (high-temperature) limit. We use the approximation
EF −h̄2 k2 /(2m)
f (~r, ~k) = e KT (36)
Z ∞ Z ∞
N= f (~r, ~k)d3 rd3 k
0 −∞
Z ∞ 3  3/2 (37)
EF h̄2 k2 EF 2πmKT
e KT V e 2mKT =Ve KT

−∞ h̄2
3/2
h̄2

EF N
e KT = (38)
V 2πmKT
Finally we get " 3/2 #
h̄2

N
EF = KT log (39)
V 2πmKT

11
4 High temperature limit
We approximate the Fermi-Dirac distribution with the Boltzmann distribution for the
electrons

16π 2 Z 2 e6 E02 

∂f (k) (EF −h̄2 k 2 /2m)/(KT ) (EF −(h̄2 k 2 −2mh̄ω)/2m)/(KT )
=  e −e


∂t 3mh̄3 ω 4 k 

 2 2 p 2
× − 4a h̄ k k − 2mω/h̄ + (h̄2 (1 + 4a2 k 2 ) − 4a2 h̄mω + 4a4 m2 ω 2 )
" p #
−2a2 mω + h̄(1 + 2a2 k(k − k 2 − 2mω/h̄))
× (log
h̄
" p #
−2a2 mω + h̄(1 + 2a2 k(k + k 2 − 2mω/h̄)) 
− log )
/
h̄
(2(h̄2 (1 + 4a2 k 2 ) − 4a2h̄mω + 4a2 m2 ω 2))
(40)
 
(EF −h̄2 k 2 /2m)/(KT ) (EF −(h̄2 k 2 +2mh̄ω)/2m)/(KT )
+ e −e

 2 2 p 2
× −4a h̄ k k + 2mω/h̄ + (h̄2 (1 + 4a2 k 2 ) + 4a2 h̄mω + 4a4 m2 ω 2)
" p #
−2a2 mω + h̄(1 + 2a2 k(k + k 2 − 2mω/h̄))
× (log
h̄
" p #
−2a2 mω + h̄(1 + 2a2 k(k + k 2 + 2mω/h̄)) 
− log )
/
h̄

(2(h̄2 (1 + 4a2 k 2 ) + 4a2 h̄mω + 4a2 m2 ω 2 ))

.



12
We substitute the Fermi energy calculated as (Eq. (39)) to above equation to get
3/2 
16π 2 Z 2 e6 E02 h̄2
 
∂f (k)  −h̄2 2
k /(2mKT ) −(h̄ 2 2
k −2mh̄ω)/(2mKT )
= N  e −e


∂t 3mh̄3 ω 4 k 2πmKT


 2 2 p 2
× − 4a h̄ k k − 2mω/h̄ + (h̄2 (1 + 4a2 k 2 ) − 4a2 h̄mω + 4a4 m2 ω 2 )
" p #
2 2 2
−2a mω + h̄(1 + 2a k(k − k − 2mω/h̄))
× (log
h̄
" p #
2 2 2
−2a mω + h̄(1 + 2a k(k + k − 2mω/h̄)) 
− log )
/
h̄
(2(h̄2 (1 + 4a2 k 2 ) − 4a2h̄mω + 4a2 m2 ω 2 ))
(41)
 
−h̄2 k 2 /(2mKT ) −(h̄2 k 2 +2mh̄ω)/(2mKT )
+ e −e

 2 2 p 2
× −4a h̄ k k + 2mω/h̄ + (h̄2 (1 + 4a2 k 2 ) + 4a2 h̄mω + 4a4 m2 ω 2)
" p #
−2a2 mω + h̄(1 + 2a2 k(k + k 2 − 2mω/h̄))
× (log
h̄
" p #
−2a2 mω + h̄(1 + 2a2 k(k + k 2 + 2mω/h̄)) 
− log )
/
h̄

(2(h̄2 (1 + 4a2 k 2 ) + 4a2 h̄mω + 4a2 m2 ω 2 ))

.



The kinetic energy of the electron is assumed to be unchanged during the collision
process, which means the approximation h̄ω ≈ 0 is proper. In this case the integral over
∆k limits are from 0 to 2k.

4 (2π)1/2 Z 2 e6 E02 N 

∂f (k) −h̄2 k 2 /(2mKT ) −(h̄2 k 2 −2mh̄ω)/(2mKT )
=  e −e


∂t 3 m5/2 ω 4 (KT )3/2 k 
2a2 k 2
 
1 2 2
× − + log[1 + 4a k ]
1 + 4a2 k 2 2
2a2 k 2
  
−h̄2 k 2 /(2mKT ) −(h̄2 k 2 +2mh̄ω)/(2mKT ) 1
+ e −e − + log[1 + 4a k ] (42)
2 2
1 + 4a2 k 2 2
4 (2π)1/2 Z 2 e6 E02 N 2a2 k 2
 
1 2 2
= − + log[1 + 4a k ]
3 m5/2 ω 4 (KT )3/2 k 1 + 4a2 k 2 2
 
−h̄2 k 2 /(2mKT ) h̄ω/(KT ) −h̄ω/(KT )


×e 2− e +e .

13
 Since we have
1 z
e + e−z = cosh z ,


(43)
2
we can expand cosh z into a Taylor series:
∞
X z 2n 1 1 1 6 1
cosh z = = 1 + z2 + z4 + z + z8 + . . . (44)
n=0
(2n)! 2 24 170 40320
and write down

4 (2π)1/2 Z 2 e6 E02 N  2 2

∂f (k) 2a k 1 2 2
=  − + log[1 + 4a k ]


∂t 3 m5/2 ω 4 (KT )3/2 k  1 + 4a2 k 2 2
  
2 2 h̄ω
× e−h̄ k /(2mKT ) 2 − 2 cosh



KT



4 (2π)1/2 Z 2 e6 E02 N  2a2 k 2

1 2 2
=  − + log[1 + 4a k ]


3 m5/2 ω 4 K 3/2 T 3/2 k  1 + 4a2 k 2 2
(45)
 2 !!
2 2 1 h̄ω
× e−h̄ k /(2mKT ) 2 − 2 1 +



2 KT


2
4 (2π)1/2 Z 2 e6 E02 N 2a2 k 2
  
1 2 2 −h̄2 k 2 /(2mKT ) h̄ω
= − + log[1 + 4a k ] e
3 m5/2 ω 4 (KT )3/2 k 1 + 4a2 k 2 2 KT
1/2 2 6 2 2  2 2

4 (2π) Z e E0 Nh̄ 2a k 1 2 2
= 5/2 2 7/2
− 2 2
+ log[1 + 4a2 k 2 ] e−h̄ k /(2mKT ) .
3 m ω (KT ) k 1 + 4a k 2
The change in average kinetic energy of the electrons is
d hi 2 2 ~
3 k h̄ ∂f (k)
Z
= dk . (46)
dt 2m ∂t
We integrate two parts of the expression (45) separately
• the first term
h̄2 k 2
Z ∞
2a2 k 2
  
−h̄2 k 2 /(2mKT )
kdk − e
0 2m 1 + 4a2 k 2
 
h̄2

e 8a2 KmT
2
h̄ Γ 0, h̄2 (47)
1  1 8KmT 8a2 KmT
= 8KT ( 2 − ) −  ,

128 a h̄2 a4 m

• the other term

Z ∞ 2 2
h̄ k   2 2

kdk log 1 + 4a2 k 2 e−h̄ k /(2mKT )

0 2m
 Z ∞  (48)
KT 2 2 h̄2
2 2 − 2h̄
2
= 2(h̄ − 4a KmT ) + e 8a KmT (h̄ − 8a KmT )
2
e 8a KmT .
16a2h̄2 1

14
When we make a substitution zt = p to the integral in the above expression, we will get
the Euler gamma function
Z ∞ −zt Z ∞ −p Z ∞ −p
e e dp e
dt = = dp = Γ(0, z) . (49)
1 t z p/z z z p
Then we can rewrite the last term as
Z ∞ 2 2
h̄ k   2 2

kdk log 1 + 4a2 k 2 e(EF −h̄ k /2m)/(2KT )

0 2m
  2  (50)
KT EF
2 2 h̄2
2 2 h̄
= e 2KT 2(h̄ − 4a KmT ) + e 8a2 KmT (h̄ − 8a KmT )Γ 0, 2 .
16a2h̄2 8a KmT
Adding up the results, we get:

4 (2π)1/2 Z 2 e6 E02 Nh̄2


d hi 1
= 8a2 KmT (3h̄2 − 16a2 KmT )


5/2 2 7/2 2
dt 3 m ω (KT ) 4
128a h̄ m
h̄2
2
 
h̄
4 2 2 4 2 2 2
− e 8a KmT (h̄ − 8a Kh̄ mT + 64a K m T )Γ 0, 2
2


8a KmT

(51)
4 (2π)1/2 Z 2 e6 E02 N 1 

= 7/2 2 7/2 4
8a2 KmT (3h̄2 − 16a2 KmT )

3 m ω (KT ) 128a
h̄2
2
 
h̄
4 2 2 4 2 2 2
− e 8a KmT (h̄ − 8a Kh̄ mT + 64a K m T )Γ 0, 2
2
.


8a KmT
2
Expressing the result as a function of the new parameter z = 8a2h̄KmT . The physical
E0
interpretation of z becomes clear when we express it as 41 ( KT )( aa0 )2 , where E0 is the energy
of the ground state for the hydrogen atom and a0 is the Bohr radius. Since a is of order
of Za1/3
0
, 41 ( aa0 )2 is of order of 1 – it varies from ∼ 0.5 − −3 when Z varies from 10 to 100.
The value of the parameter tells us how much the electrons energy exceeds the ionization
energy of the hydrogen atom.
 2
4 (2π)1/2 Z 2 e6 E02 N 1 h̄2 h̄4 h̄4

d hi h̄ 2 z 4
= (3h̄ − 2 ) − e (h̄ − + 2 )Γ(0, z)
dt 3 m7/2 ω 2 (KT )7/2 64a4 z z z z
2 2 1/2 2 4 4  
e E0 N 4 (2π) Z e h̄ 3 2 z 1 1
= − − e (1 − + 2 )Γ(0, z)f g (52)
2mω 2 3 m2 (KT /m)3/2 64a4 m2 K 2 T 2 z z 2 z z
2 2 1/2 2 4
 
e E0 N 4 (2π) Z e 2 3 2 z 1 1
= z − − e (1 − + 2 )Γ(0, z) ,
2mω 2 3 m2 νT3 z z2 z z
p
where νT = KT /m is the thermal velocity of electrons.

4.1 a → ∞ approximation
When we put a → ∞ we will get an asymptotic problem for the Coulomb potential.

15
 For a → ∞ the (27) simplifies:
∂f (k) 8π 2πZ 2 e6 E02
= ×
∂t
 3 mh̄3 ω 4 k
Z
 1 1 1
 1 + e−(EF −h̄2 k2 /2m)/(KT ) − 1 + e−(EF −(h̄2 k2 −2mh̄ω)/2m)/(KT ) d(∆k)

(53)

∆k

 Z 
1 1 1 
+ − d(∆k) .


2 2 2 2
−(E −h̄ k /2m)/(KT ) −(E −(h̄ k +2mh̄ω/2m))/(KT ) ∆k

1+e F 1+e F

For high-temperature approximation we get

3/2 
8π 2 Z 2 e6 E02 N h̄2
 
∂f (k)  −h̄2 k 2 /(2KmT ) −(h̄2 k 2 −2mh̄ω)/(2KmT )
= N  e −e


∂t 3mh̄3 ω 4 k 2πmKT

  Z (54)
2 2 2 2 1
+ e−h̄ k /(2KmT ) − e−h̄ k +2mh̄ω/(2KmT ) 

 d(∆k) ∆k .



Assuming the infinitesimally small energy change during the collision process we take
the integral limits from 0 to 2k. To avoid log[0] term we take the cutoff (kmin ) for lower
∆k limit
3/2
8π 2 Z 2 e6 E02 N h̄2

∂f (k)
=
∂t 3mh̄3 ω 4 k 2πmKT
 
2 2 2 2
−h̄ k /(2mKT ) −(h̄ k −2mh̄ω)/(2mKT )

 e −e (log[2k] − log[kmin ])



  
2 2 2 2
+ e−h̄ k /(2KmT ) − e−(h̄ k +2mh̄ω)/(2KmT ) (log[2k] − log[kmin ])





2π 1/2 Z 2 e6 E02 N 2 2 (55)

= 5/2 4 3/2
(log[2k] − log[kmin 0]) e−h̄ k /(2mKT )
3m ω (KT ) k
 
h̄ω/(2KT ) −h̄ω/(2KT )


2− e +e
2
2π 1/2 Z 2 e6 E02 N

−h̄2 k 2 /(2mKT ) h̄ω
= (log[2k] − log[kmin ]) e
3m5/2 ω 4 (KT )3/2 k KT
1/2 2 6 2 2
2π Z e E0 Nh̄ 2 2
= 5/2 2 7/2
(log[2k] − log[kmin ]) e−h̄ k /(2mKT ) .
3m ω (KT ) k
We integrate over k
d hi 2π 1/2 Z 2 e6 E02 Nh̄2 ∞ h̄2 k 3
Z
−h̄2 k2
= 5/2 2 7/2
(log[2k] − log[kmin ])e 2mKT dk
dt 3m ω (KT ) 0 2m
2 ∞
h̄2 k 3 −h̄2 k2 h̄2 k 3 −h̄2 k2
Z ∞
2π 1/2 Z 2 e6 E02 Nh̄
 Z 
= e 2mKT log[2k] − dk e 2mKT log[kmin ] dk .
3m5/2 ω 2 (KT )7/2 0 2m 0 2m
(56)

16
 The result is
d hi 2π 1/2 Z 2 e6 E02 Nh̄2 ∞ h̄2 k 3
Z
−h̄2 k2
= (log[2k] − log[k min ])e 2mKT dk
dt 3m5/2 ω 2 (KT )7/2 0 2m
2π 1/2 Z 2 e6 E02 Nh̄2
Z ∞ 2 3 Z ∞ 2 3 
h̄ k −h̄2 k2 h̄ k −h̄2 k2
= e 2mKT log[2k]dk − e 2mKT log[kmin ] dk
3m5/2 ω 2 (KT )7/2 0 2m 0 2m
2π 1/2 Z 2 e6 E02 Nh̄2 K 2 mT 2 K 2 mT 2
 
= (−1 + γ − log[4]) + log[kmin ] dk ,
3m5/2 ω 2 (KT )7/2 2h̄ h̄
(57)
where γ – the Euler gamma has its numerical value ∼ 0.577
Finally we get
2π 1/2 Z 2 e6 E02 Nh̄ 1
 
d hi
= (1 + γ − log[4]) + log[kmin ]) . (58)
dt 3m3/2 ω 2 (KT )3/2 2

5 Conclusions
We have presented the theoretical model of the electron gas irradiated by a laser beam.
We have considered absorption by inverse Bremsstrahlung to derive the electron kinetic
equation and the rate of change of the average kinetic energy of the electrons.

To be continued. . .

R
A f (x)δ(g(x))dx integrating formula
My aim is to calculate the below mentioned integral:
Z
f (x)δ(g(x))dx . (A.59)

Converting variables
y = g(x)
(A.60)
dy = g 0(x)dx ,
we obtain
1
Z Z
f (x)δ(g(x))dx = f (g −1 (y))δ(y) dy . (A.61)
g 0(g −1 (y))

17
Substituting x = g −1 (y), we have

1
Z
f (x)δ(g(x))dx = f (x0 ) . (A.62)
g 0 (x0 )

If a function g −1(y) is multi-valued, e.g.

g(x) = x2 − 1
(A.63)
g −1 (y) = ±1 ,

our integral becomes a sum of obtained expression evaluations for all the roots. If there is
xi which is an i-th root of the g(x) function (i = 1, . . . , N), then
N
1
Z X
f (x)δ(g(x))dx = f (xi ) . (A.64)
i=1
g 0(xi )

18