BEC-BCS Crossover

Diego Luis Velasco-González

August , 2020
BEC-BCS Crossover

We now that, when a cloud of bosons is cool down, it will reach the BEC which has the
interesting property of superfluidity.

Whereas, cooling down a gas of fermions, it will create a Fermi sea which is not coherent but
also do not have superfluidity.

As a remain, we can distinguish and atom in terms of the number of protons, electrons and
neutros that it has. Are bosons those atoms with a even number of those, and fermions if that
is a odd number.

If we consider a mixture of two kind of fermions, we can form pairs as a molecule, and it will
have an even number of protons, neutrons and elector, so it will behaves as a bosonic molecule,
therefore, they can condensate and become a BEC. This pairs are due to weak attractive
interaction, which, for example, forms the well-known Cooper pairs, used has a tool in the
superconductivity theories of electrons).

But, how does fermions pair?

Cooper Pairs
The key ingredient of the BCS theory are the Cooper pairs: electron-electron interaction medi-
ated by the phonons to get bound states formed by two electrons of opposite spins and momenta.
Consider the Shrödinger equation for two interaction electrons, via a potential V (r1 − r2 ):
" #
~2 ∇21 ~2 ∇22
− − + V (r1 − r2 ) Ψ(r1 , r2 ) = EΨ(r1 , r2 ), (1)
2m 2m
where Ψ(r1 , r2 ) is the wave-function, E the energy. Through the change of variables: r = r1 −r2
the relative position, and R = (r1 + r2 )/2 the center of mass. So (1) becomes:
" #
~2 ∇2R ~2 ∇2r
− − + V (r) Ψ(r, R) = EΨ(r, R), (2)
2m∗ 2µ
Now, m∗ = 2m is the total mass and µ is the reduced mass. Since the potential does not
depends on R, we are looking for a solution:

Ψ(r, R) = ψ(r) eiK·R (3)

1
 2

Thus, equation (2) is:

" # !
~2 ∇2r ~K 2
− + V (r) ψ(r) = E − ψ(r), (4)
2µ 2m∗

Lowest energy E is archived when K = 0. For this case, the two electrons have opposite
momenta. We take the Fourier transform of the Schrödinger equation, by introducing:
Z
ψ(k) = d3 r ψ(r) e−ik·r , (5)

Then eq. (4):

~2 k 2 Z
ψ(k) + d3 r V (r)ψ(r)e−ik·r = Eψ(r)
2µ
!
Z
d3 q Z
3 −i(k−q)·r ~2 k 2
V (q) d r ψ(r)e = E− ψ(k)
(2π)3 m
Z
d3 k 0
V (k − k0 )ψ(k0 ) = (E − 2εk )ψ(k) (6)
(2π)3

Change of variable q = k − k0 , and εk = h2 k 2 /2m the free electron energy. Bound state of
electrons has E < 2εk . The modified wave-function:

∆(k) = (E − 2εk )ψ(k) (7)

which gives:
Z
d3 k 0 V (k − k0 )
∆(k) = − ∆(k0 ) (8)
(2π)3 2εk − E
In this exercise we are overlooking an important fact: in the actual many-body problem, only
electrons near to the Fermi surface will be affected by an attractive interaction. In order to
mimic that, we consider an attractive potential V (k − k0 ) = −V0 for the unoccupied electrons
states above the Fermi energy εk , εk − εF , εk − εF < ~ωD , with ωD the Debye frequency.
We look up for solution with constant ∆. There is an implication that we have a even wave-
function, then the spins of the two electrons must be anti-parallel (a singlet). Defining the
density of states:
m3/2 √
ρ(ε) = √ 3 2 ε (9)
2~ π
Since ~ωD  εF , we can approximate the density of states for its value at εF . So, eq. (8):
Z εF +ωD
dε
∆ = V0 ρ(εF )∆
εF 2ε − E
2 2εF − E + 2ωD
 
= ln (10)
V0 ρ(εF ) 2εF − E

In the limit of small V0 ρ(εF )  1, E is close to 2εF , and we can approximate 2εF − E + 2ωD ≈
2ωD . Defining the binding energy Eb = 2εF − E, we obtain:
2
−V
Eb = 2ωD e 0 ρ(εF ) (11)
 3

which shows us that a bound state can be formed regardless of how small the attractive potential
is. This bound state is called a Cooper pair. The property responsible of this behavior is the
existence of a well-defined Fermi surface, separating states that are occupied from states that
are unoccupied.

This work was made by Leon Cooper, and his principal idea can be explained with the following
picture: He assumed that the two electrons are on the top of the Fermi sea, that is, a narrow
shell above the Fermi surface. These electrons can scattered to all the available states, finding
that even a weak attraction between them leads to binding.(Read form slides).

Feshbach resonaces allow us to control how tight or weak the bound molecules using a magnetic
field. In left side, we have a stable molecule with tight interaction whereas at the right side we
have weak interactions to finally get only Cooper pairs.

So using feshbach resonance, that is, using a magnetic field we can go strongly-bound pairs
forming a BEC to weakly-bound pair which resembles Cooper pairs and superconduction. BUT
in the middle, is found a new form of superfluidity, where we have molecules too big to be called
molecules, but too small to be called cooper pairs.

BCS Theory
The effective Hamiltonian which describes the superconductivity is:
1 X
k ĉ†kσ ĉkσ + Vkk0 ĉ†k↑ ĉ†−k↓ ĉ−k0 ↓ ĉk0 ↑
X
ĤBCS = (12)
kσ N kk0

Here, ĉ†kσ (ĉkσ ) creates (annihilates) an electron with momentum k and spin σ, and k = εk − µ,
with µ the chemical potential. Second term describes the destruction of a Cooper pair and the
subsequent creation of another Cooper pair. We perform a mean-field approximation, then, the
effective Hamiltonian becomes:
MF
k ĉ†kσ ĉkσ − ∆k ĉ†k↑ ĉ†−k↓ − ∆∗k ĉ−k↓ ĉk↑
X X X
ĤBCS = (13)
kσ k k

Where we define: X
∆k = − Vkk0 hĉ−k0 ↓ ĉk0 ↑ i : Gap function (14)
k0

In order to diagonalize the Hamiltonian, we use the Bogoliubov unitary transformation which
introduces a new fermionic operators:
! ! !
~ˆk = γ̂k↑ 1 u∗k vk ~ˆ ~ˆk = ĉk↑
B † = Ak , with A (15)
γ̂−k↓ |uk |2 + |vk |2 −vk∗ uk ĉ†−k↓

These Bogoliubov operators also obey the fermionic anticommuting relations with a constrain:
|uk |2 + |vk |2 = 1. Writing the Hamiltonian (13) in a matrix form:
!
MF
X
~ˆ† Hk A
~ˆk + C0 , k −∆k
ĤBCS = A Hk = , C0 ≡ Constant (16)
k
k
−∆∗k −k
 4

With some algebra, we obtain:

X †

MF
ĤBCS = k |uk |2 − ∆k u∗k vk∗ − ∆∗k uk vk − k |vk |2 γ̂k↑ γ̂k↑ (17)
k
 
† † †
+ −k |uk |2 + ∆k u∗k vk∗ + ∆∗k uk vk + k |vk |2 γ̂−k↓ γ̂−k↓ −→ γ̂−k↓ γ̂−k↓ = 1 − γ̂−k↓ γ̂−k↓
 
† †
+ −2k u∗k vk − ∆k u∗2 ∗ 2
k + ∆k vk γ̂k↑ γ̂−k↓ −→ ω = −2k u∗k vk − ∆k u∗2 ∗ 2
k + ∆k vk
 
+ −2k uk vk∗ − ∆∗k u2k + ∆k vk∗2 γ̂−k↓ γ̂k↑ + C0 −→ ω ∗ = −2k uk vk∗ − ∆∗k u2k + ∆k vk∗2

So, to get a diagonal form, we need the following conditions:

Ek = k |uk |2 − ∆k u∗k vk∗ − ∆∗k uk vk − k |vk |2 (18)

ω=0 ⇒ ω∗ = 0 (19)

Where Ek is the eigenenergy. We can find the values of Ek , |uk | and |vk | using the conditions
above, and defining:

uk = |uk |eiαk , vk = |vk |eiφk y ∆k = |∆k |eiβk (20)

With some fixed values for the phases, one obtain that:
   
1 k 1 k
|uk |2 = 1 + q , |vk |2 = 1 − q  (21)
2 (2k + |∆k |2 ) 2 (2k + |∆k |2 )

And the energy: q

Ek = (2k + |∆k |2 ) (22)
Finally, the diagonalized Hamiltonian is:
 
MF † †
X
ĤBCS = Ek γ̂k↑ γ̂k↑ + γ̂−k↓ γ̂−k↓ + C0 (23)
k

Due to the unstable scattering of electrons in states |k, ↑i and | − k, ↓i , Bardeen, Cooper and
Schrieffer proposed an ansatz for the new ground state which is constructed with a superposition
of stats built up of Cooper pairs, read as:
Y 
|ψBCS i = uk + vk ĉ†k↑ ĉ†−k↓ |0i (24)
k

where |0i is the vacuum state without any electrons.

BEC pairs Crossover state

In this way, we can obtain a immediate analogy with the BEC. We know that two fermions can
form a pair, and if the (Cooper) pair condensate then each pair should be identical, so we take
our N fermions, half with spin up and the other ones with spin down, to get a pair, and get
that our wave-function of identical fermions pairs/bosons is just the product of wave-functions
of each pair (but it is the same for all particles):

|Ψ(r1 · · · rN )i = Â{ψ(|r1 − r2 |)|χ1,2 i · · · ψ(|rN −1 − rN |)|χN −1,N i} (25)

 5
√
An anti-symmetried state where ψ(|rN −1 −rN |) is the spatial pair function and |χi,j i = 1/ 2(| ↑
ii | ↓ij − | ↑ij | ↓ii ) a singlet. In second quantization:
Z Y
|ΨiN = d3 ri ψ(r1 − r2 )Ψ†↑ (r1 )Ψ†↓ (r2 ) · · · ψ(rN −1 − rN )Ψ†↑ (rN −1 )Ψ†↓ (rN )|0i (26)
i

With a Fourier transform of the pair functions and operators:

eik·r X † e−ik·r
Ψ†σ (r) =
X
ψ(r) = ψk √ , ĉkσ √ (27)
k Ω k Ω

So, we obtain a pair creation operator which creates pairs with zero momentum:

b̂† = ψk ĉ†k↑ ĉ†−k↓

X
(28)
k

And now, we can say that describes our system, a BEC, where we create N/2 pairs out of the
vacuum:
|ΨiN = b̂†N/2 |0i (29)
But, here surges a question: Is that a real BEC? That is, are those really bosons? To answer
that question, let’s check the commuting relations for bosons:
h i h i
b̂† , b̂† ψk ψk0 ĉ†k↑ ĉ†−k↓ , ĉ†k↑ ĉ†−k↓
X
= =0
− −
kk0
h i
b̂, b̂ = ··· = 0
−
h i
b̂, b̂† |ψk |2 (1 − nk↑ − n−k↓ ) 6= 1
X
= ··· = (30)
−
kk0

The last commutator is not 1 for all cases. The population of states depends on the size of
the pairs: if it is huge, there is low lying momentum states, which has huge population, but
if the size is small, that is, in a tightly-bound system, particles has so much zero point energy
that every given momentum states has pretty much zero occupation. So, basically, what here
is introduced a condition: pairs do not obey bosons commutation relations UNLESS nk  1,
so: h i
b̂, b̂† ≈ |ψk |2 = 1
X
(31)
−
k

If that condition is not fulfilled, then those pairs act out in a fermionic way.
Thus, we have state that is a BEC for tightly-bound pairs. Here, we see that those pairs now
feel the Fermi pressure.
Going on, we can prove that the state |ΦiN is identical to the BCS state |ψBCS i.
For photons, there exists the coherent state of Glauber states, defined as coherent superposition
of Fock states:
∞ −|α|2 /2
ne
X
|αi = α √ |ni (32)
n=0 n!
 6

Here α determines the average photon number and the phase of the coherent state. So, taking
|ΦiN as the Fock states:

NpJ/4 X 1
NpM/2 b̂†M |0i
X
N |Φi = |ΦiJ =
Jeven (J/2)! M M !
√ †M
= e Np b̂ |0i
Y √N ψ ĉ† ĉ†
= e p k k↑ −k↓ |0i
k
Y q 
= 1+ Np ψk ĉ†k↑ ĉ†−k↓ |0i (33)
k

Where we use the fact that for different momenta, those creation operator commutes, and also
we Taylor expanded but there will only survive the creation operators until first power because
of the fermionic property ĉ†i = 0 for i > 1. Normalizing, we have:
Y 1 Yq
N = = 1 + Np |ψk |2 (34)
k uk k

To happily get the BCS state:

Y 
|ψBCS i = uk + vk ĉ†k↑ ĉ†−k↓ |0i (35)
k
q
with vk = Np ψk uk and |uk |2 + |vk |2 = 1

Finally, I’m going to show some experimental results:

This first picture is the experiment done at the MIT, where the used Lithium 6, eventhough
it looks the same as atomic BEC, with spectroscopy where show that in fact there are pairs
inside.
This was made also in other universities, showing the same results. And finally, the gas was
rotated, and it shows the vortices expected because of the superfluidity.