SQ Int El Gas

Second-quantization representation of the Hamiltonian of an inter-
acting electron gas in an external potential

As a first concrete example of the second quantization formalism, we consider a gas of
electrons interacting via the Coulomb interaction, and which may also be subjected to an
external potential. The Hamiltonian for the electrons is given by
H =T +U +V (1)
where T is the kinetic energy operator, U is the external potential (assumed to be spin-
independent; it could e.g. be the periodic potential due to the crystal lattice in a crystalline
solid), and V is the Coulomb interaction between the electrons. In first quantization,
N
X ~2 2
T = − ∇i , (2)
i=1
2m
N
X
U = u(ri ), (3)
i=1
XN
V = v(ri , rj ) (4)
i,j
i6=j
where u(r) is the external potential felt by an electron at position r and

e2
v(r, r 0 ) = v(r − r 0 ) = . (5)
4πε0 |r − r 0 |
We take the system to be a cube of side lengths L and volume Ω = L3 . For convenience we
impose periodic boundary conditions in all 3 spatial directions. Thus the point (x = L, y, z)
is identified with the point (x = 0, y, z), (x, y = L, z) is identified with (x, y = 0, z) and
(x, y, z = L) is identified with (x, y, z = 0).1
Let us find the second-quantization representation of H expressed in terms of the momentum-

spin (i.e. α = (k, σ)) single-particle basis. The single-particle basis functions are thus
eigenfunctions of the momentum operator p and z-component S z of the spin operator for a
single electron, and are given by
1
φα (x) = φkσ (r, s) = √ eik·r χσ (s) (6)
Ω
where2
χσ (s) = δsσ (7)
1
Please note that the symbol x is used for two different things in these notes: the x coordinate in real
space and the total space-spin coordinate x = (r, s); hopefully it should be clear from the context which of
the two meanings is intended.
2 z 1 1 0
In the standard matrix representation (setting ~ = 1), S = 2 with eigenstates | ↑ =
0 −1
1 0
+1/2i = , | ↓ = −1/2i = , giving χσ (s) ≡ hs|σi = δsσ .
0 1
1
with s, σ taking the possible values ±1/2 =↑, ↓. The periodic boundary conditions imply
that only wavevectors k on the form
2π
k= (nx , ny , nz ) (8)
L
with ni (i = x, y, z) being arbitrary integers, are allowed. For example, the identification of
x = 0 and x = L gives φ(x = 0) = φ(x = L), from which it follows that 1 = eikx L and thus
kx = 2πnx /L for integer nx . The eigenfunctions {φα (x)} form a complete and orthonormal
set. Let us check the orthonormality:
Z Z X
∗
dx φα0 (x)φα (x) = d3 r φ∗k0 σ0 (r, s)φkσ (r, s)
s
Z
1 0
X
= d3 r ei(k−k )·r δσ0 s δσs = δkk0 δσσ0 = δαα0 . (9)
Ω
| {z } | s {z }
δkk0
δσσ0
The kinetic energy operator T is a single-particle operator, so its second-quantization rep-

resentation is given by
X p2
T = hα0 | |αic†α0 cα (10)
α,α 0
2m
where the matrix element
p2 ~2 2
Z
0
hα | |αi = dx φ∗α0 (x) − ∇ φα (x)
2m 2m
~2 2
XZ
3 1 −ik0 ·r 1 ik·r
= dr √ e δsσ0 − ∇ √ e δsσ (11)
s
Ω 2m Ω
XZ 2 2
3 1 −ik0 ·r ~k 1 ik·r
= dr √ e δsσ0 √ e δsσ
s Ω 2m Ω
!
~2 k2
Z X
1 0
= d3 r ei(k−k )·r δsσ0 δsσ
2m Ω s
| {z } | {z }
δkk0 δσσ0
2 2
~k
= δkk0 δσσ0 . (12)
2m
Thus the matrix elements of the the kinetic energy operator p2 /2m for a single particle are
diagonal in this basis. This should be entirely as expected (and is the reason why we chose
this basis to begin with): given that the basis functions are eigenfunctions of p, they are
also eigenfunctions of p2 /2m. Putting this result back into (10) gives
X ~2 k2
T = c†kσ ckσ . (13)
kσ
2m
Thus the diagonality implies that T becomes just a linear combination of number operators
c†kσ ckσ ≡ n̂kσ .
2
Similarly, the second-quantized representation of the external potential U , which is also a
single-particle operator, becomes
X
U= hα0 |u|αic†α0 cα (14)
α,α0
where
Z
0
hα |u|αi = dx φ∗α0 (x)u(r)φα (x)
XZ
3 1 −ik0 ·r 1 ik·r
= dr √ e δsσ0 u(r) √ e δsσ (15)
s Ω Ω
Z
1 0
X
= d3 r u(r)ei(k−k )·r δsσ0 δsσ
Ω
| {z } s
uk0 −k
1
=
uk0 −k δσσ0 (16)
Ω
where uq is the Fourier transform of u(r). Thus
1 X 1X
U= uk0 −k δσσ0 c†k0 σ0 ckσ = uq c†k+q,σ ckσ (17)
Ω kσk0 σ0 Ω kqσ
which describes a scattering process in which an electron is scattered from momentum k to
momentum k + q. As a special case, note that if u(r) = u, i.e. a constant independent
of r, so that the system is translationally invariant, we have Ω1 uq = uδq,0 , in which case U
becomes diagonal in the (k, σ) basis.
Finally, the second-quantized representation of the electron-electron interaction V , which
is a two-particle operator, is
1 X
V = hαβ|v|γδic†α c†β cδ cγ , (18)
2 α,β,γ,δ
where, writing α = (k1 , σ1 ), β = (k2 , σ2 ), δ = (k3 , σ3 ), and γ = (k4 , σ4 ), we have

Z Z XZ XZ
∗ ∗ 0 0 0
hαβ|v|γδi = dx dx φα (x)φβ (x )v(x, x )φγ (x)φδ (x ) = 3
dr d3 r0
s s0

1 1 0 1 1 0
√ e−ik1 ·r δsσ1 √ e−ik2 ·r δs0 σ2 v(r − r 0 ) √ eik4 ·r δsσ4 √ eik3 ·r δs0 σ3
Ω Ω Ω Ω
! !
X X
= δsσ1 δsσ4 δs0 σ2 δs0 σ3
s s0
Z Z
1 0
2
· d3 r d3 r0 v(r − r 0 )e−i(k1 −k4 )·r e−i(k2 −k3 )·r (19)
Ω
Let us define R = r − r 0 and change integration variables to R and r 0 . The integrals then
factorize as follows:
Z Z
1 −i(k1 −k4 )·R 1 0
3
d R v(R)e · d3 r0 e−i(k2 −k3 +k1 −k4 )·r (20)
Ω Ω
| {z } | {z }
≡vk1 −k4 =δk1 ,k4 +k3 −k2
3
where vq is the Fourier transform of v(r). Thus
1
hαβ|v|γδi = δσ σ δσ σ δk ,k +k −k vk −k (21)
Ω 1 4 2 3 1 4 3 2 1 4
which gives
1 X
V = δσ1 σ4 δσ2 σ3 δk1 ,k4 +k3 −k2 vk1 −k4 c†k1 σ1 c†k2 σ2 ck3 σ3 ck4 σ4 (22)
2Ω k
1 ,k2 ,k3 ,k4 ,σ1 ,σ2 ,σ3 ,σ4
Doing the summation over k1 and over σ3 and σ4 then gives (after renaming σ1 ≡ σ, σ2 ≡ σ 0 )
1 X X
V = vk −k c† c† 0 ck σ 0 ck 4 σ . (23)
2Ω σ,σ0 k ,k ,k 3 2 k4 +k3 −k2 ,σ k2 σ 3
2 3 4
Let us define new summation variables k, k0 , and q by
k4 ≡ k, k3 ≡ k 0 , k2 ≡ k0 − q. (24)
This gives
k4 + k3 − k2 = k + q, (25)
k3 − k2 = q. (26)
Thus
1 X X †
V = vq ck+q,σ c†k0 −q,σ0 ck0 σ0 ckσ . (27)
2Ω q k,σ
k0 ,σ 0
This expression describes a scattering process in which two electrons scatter by exchanging
momentum q. Before the scattering the electrons have momenta k and k0 , after the scattering
the electrons have momenta k + q and k0 − q. Note that the total momentum k + k0 is
conserved in the scattering process. This is a consequence of the translational invariance of
the interaction, i.e. the fact that it only depends on r − r 0 , not on r and r 0 separately. Also
note that since the Coulomb interaction (5) is spin-independent, the electrons spins are not
affected by the scattering process. A diagrammatic representation of the scattering process
is shown in the figure.
Mathematically, the scattering is described by the annihilation of the incoming electrons

with momentum k and k0 and the creation of the outgoing electrons with momentum k + q
and k0 − q.

SQ Int El Gas

Uploaded by

Copyright:

Available Formats

SQ Int El Gas

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

SQ Int El Gas

Uploaded by

Copyright:

Available Formats

Second-quantization representation of the Hamiltonian of an inter-

acting electron gas in an external potential

where u(r) is the external potential felt by an electron at position r and

Let us find the second-quantization representation of H expressed in terms of the momentum-

The kinetic energy operator T is a single-particle operator, so its second-quantization rep-

where, writing α = (k1 , σ1 ), β = (k2 , σ2 ), δ = (k3 , σ3 ), and γ = (k4 , σ4 ), we have

Let us define new summation variables k, k0 , and q by

Mathematically, the scattering is described by the annihilation of the incoming electrons

You might also like

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.