20.04.2014 Views

Apr. 20: Hartree-Fock-Bogoliubov, BCS

Apr. 20: Hartree-Fock-Bogoliubov, BCS

Apr. 20: Hartree-Fock-Bogoliubov, BCS

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

<strong>Hartree</strong>-<strong>Fock</strong>-<strong>Bogoliubov</strong><br />

A. Two-body (density-dependent) interaction:<br />

H ˆ = ! t ij<br />

c + i<br />

c j<br />

+ 1 4! v ijkl<br />

c + i<br />

c + j<br />

c l<br />

c k<br />

ij<br />

ijkl<br />

B. Variational principle:<br />

!E["] = 0; E ["] = " ˆ<br />

H "<br />

" " ; E "<br />

[ ] > E 0<br />

C. Trial wave functions: product states<br />

"( )<br />

! K + = u iK<br />

c i + + v iK<br />

c i<br />

i<br />

# = $ ! K<br />

% ; ! K<br />

# = 0<br />

K<br />

❏ HFB - quasiparticles incorporated<br />

into the HF formalism<br />

❏ HFB wave function - the most general<br />

general <strong>Bogoliubov</strong><br />

transformation<br />

HFB wave function<br />

is the quasuparticle<br />

vacuum<br />

product wave function consisting of independently<br />

moving quasiparticles (in HF: particles)<br />

❏ Selfconsistent description of coupling between p-h<br />

an p-p channels


HFB - density matrix and pairing tensor<br />

HFB density matrix<br />

HFB pairing tensor<br />

! ij<br />

" # c j + c i<br />

# , $ ij<br />

" # c j<br />

c i<br />

#<br />

ˆ ! = v * v T , $ ˆ = v * u T<br />

ˆ ! 2 % ˆ ! = % $ ˆ $ ˆ<br />

+ , ˆ ! $ ˆ = $ ˆ ˆ !<br />

ˆ ! =<br />

%<br />

'<br />

&<br />

ˆ " # ˆ<br />

$ # ˆ<br />

1 $ ˆ<br />

" *<br />

(<br />

*, ! ˆ 2 = ! ˆ<br />

)<br />

! ˆ % u K(<br />

' * = 0, ! ˆ v *<br />

% (<br />

K<br />

' *<br />

& v K ) & u K )<br />

* = v *<br />

% (<br />

'<br />

K<br />

* *<br />

& u K )<br />

Generalized<br />

density matrix<br />

Eigenvalues are 0 or 1 (thus<br />

defining occupations of<br />

quasiparticle states)<br />

ˆ H (c + ,c) ! ˆ H (" + ,") = E 0<br />

+ ˆ H <strong>20</strong><br />

+ ˆ H 11<br />

+ ˆ H int<br />

Independent quasiparticle<br />

Hamiltonian (H HFB )<br />

[ H ˆ , N ˆ ] = 0, but H ˆ<br />

HFB<br />

, N ˆ<br />

[ ] ! 0<br />

Quasiparticle<br />

interaction<br />

However, we require that ˆ N = N " ˆ H ' = ˆ H # $ ˆ N


HFB equations<br />

ˆ K =<br />

$ h ˆ ! "<br />

&<br />

%<br />

! ˆ # * ! ˆ<br />

h ij<br />

= t ij<br />

+ * ij<br />

# ˆ '<br />

)<br />

h * + "(<br />

* ij<br />

= + v iljk<br />

, kl<br />

kl<br />

# ij<br />

= 1 2 + v ijkl<br />

- kl<br />

kl<br />

HFB Hamiltonian<br />

HF Hamiltonian<br />

Selfconsistent HF field<br />

Selfconsistent pair field<br />

HFB equations<br />

$ h ˆ ! "<br />

&<br />

%<br />

! ˆ # * ! ˆ<br />

# ˆ ' $<br />

) u K'<br />

$ u<br />

& ) = E<br />

K '<br />

h * K & )<br />

+ "(%<br />

v K ( % v K (<br />

[ K ˆ , . ˆ ] = 0<br />

Complicated<br />

eigenvalue problem<br />

❏ HFB equations treat p-h and p-p on the same footing<br />

❏ Δ is a state-dependent field. In general, it depends on<br />

density and has a kinetic term<br />

❏ The generalized density matrix and the HFB Hamiltonian<br />

can be diagonalized simultaneously<br />

❏ Fermi level determined from the particle number<br />

equation<br />

❏ Often it is convenien to express the HFB equations<br />

in the canonical basis


Independent quasiparticle approach<br />

! k + = u k<br />

c k + " v k<br />

c k<br />

! + k<br />

= u k<br />

c + k<br />

+ v k<br />

c k<br />

<strong>Bogoliubov</strong>-<br />

Valatin<br />

Transformation<br />

phase convention: u k<br />

= u k<br />

> 0,<br />

ε<br />

v k<br />

= "v k<br />

> 0<br />

F<br />

+<br />

{! k<br />

,! j } = 0, ! k+<br />

,! j<br />

+<br />

{! k<br />

,! j } = # ij<br />

$ u 2 k<br />

+ v 2 k<br />

=1<br />

{ } = 0,<br />

( )<br />

0 = <strong>BCS</strong> = u k<br />

+ v k<br />

c + +<br />

! k<br />

c k<br />

"<br />

# k<br />

0 = 0<br />

0 ! exp A +<br />

A + =<br />

u k<br />

k > 0<br />

v k<br />

k >0<br />

( ) " =<br />

$ c k<br />

+ +<br />

c k<br />

#<br />

1<br />

n!<br />

n= 0<br />

$ A +<br />

( ) n "<br />

<strong>BCS</strong> wave function is<br />

the quasiparticle<br />

vacuum!<br />

<strong>BCS</strong> function is a<br />

superposition of<br />

different number of pairs<br />

❏ Important ground-state correlations are included<br />

❏ <strong>BCS</strong> wave function does not have a sharp particle<br />

number (an intrinsic wave function!)


<strong>BCS</strong> equations<br />

ˆ H (c + ,c) ! ˆ H (" + ,") = E 0<br />

+ ˆ H <strong>20</strong><br />

+ ˆ H 11<br />

+ ˆ H int<br />

Independent quasiparticle<br />

Hamiltonian (H <strong>BCS</strong> )<br />

Quasiparticle<br />

interaction<br />

[ H ˆ , N ˆ ] = 0, but H ˆ<br />

<strong>BCS</strong><br />

, N ˆ<br />

[ ] ! 0<br />

However, we require that ˆ N = N " ˆ H ' = ˆ H # $ ˆ N<br />

Particle number<br />

equation<br />

Variational method<br />

! 0 ˆ H ' 0 = 0<br />

˜ " k<br />

= " k<br />

+ # v k ki i<br />

v 2 i<br />

$ %<br />

i > 0<br />

!<br />

N = 2v i<br />

2<br />

i > 0<br />

" k<br />

= #! v kk ii<br />

u i<br />

v i<br />

i >0<br />

v 2 k<br />

= 1 $<br />

2<br />

&<br />

%<br />

1 !<br />

˜ " k<br />

˜ " 2 2<br />

k<br />

+ # k<br />

'<br />

)<br />

(<br />

Pairing gap<br />

equation


<strong>BCS</strong> model<br />

H ˆ<br />

<strong>BCS</strong><br />

! + k1<br />

! + k 2<br />

L! + k n<br />

0 = (E 0<br />

+ E k1<br />

+ E k2<br />

+L+ E k n<br />

)! + k 1<br />

! + k 2<br />

L! + kn<br />

0<br />

E k<br />

=<br />

˜ ! 2 2<br />

k<br />

+ " k<br />

Quasiparticle energy<br />

State-independent monopole pairing force:<br />

V ˆ<br />

pair<br />

= !GP ˆ + P ˆ = !G " c + i<br />

c + i<br />

c j<br />

c j<br />

i, j > 0<br />

P ˆ # c + +<br />

" i<br />

c i<br />

v ijkl<br />

= !G$ ji<br />

$ lk<br />

sgn( j)sgn(k)<br />

i> 0<br />

( )<br />

'<br />

% = G" u i<br />

v i<br />

& %)<br />

G<br />

i> 0 ( 2<br />

"<br />

i > 0<br />

1 *<br />

!1, = 0<br />

E i +<br />

❏ Fermi level can be extracted from<br />

experimental binding energies:<br />

❏ Energy gap in even-even nuclei:<br />

❏ High level density in odd systems:<br />

! = dE<br />

dN<br />

E k1<br />

+ E k2<br />

! 2"<br />

E k<br />

! E k0<br />

" ˜ # k 2 + $ 2 ! $<br />

❏ Pairing gap can be extraxted from experimental<br />

odd-even mass differences:<br />

[ ]<br />

! " # 1 (N +<br />

2 E 2) (<br />

g.s .<br />

+ E N) (N +1)<br />

g.s.<br />

# 2E g.s.


The gap equation<br />

❏ Schematic solution for the case of uniform<br />

density of single-particle states<br />

2<br />

G = ! 1<br />

"1 = G 2<br />

k > 0<br />

E k<br />

b<br />

1<br />

% g (#<br />

)d#<br />

# 2 + $ 2<br />

a<br />

((1/ g G)<br />

If g (#<br />

) & g and Gg


Simple pairing model<br />

+ε<br />

-ε<br />

+ε<br />

+<br />

-ε<br />

Ω times<br />

+ ...<br />

N = 2! particles " # = 0 (assuming Gv 2 < $)<br />

1<br />

G = ! E " % = $<br />

&<br />

(<br />

'<br />

G<br />

G crit<br />

2<br />

)<br />

+<br />

*<br />

,1, G crit<br />

= $ !<br />

Δ<br />

Normal phase<br />

Correlations<br />

beyond <strong>BCS</strong><br />

Superconductor<br />

Phase transition<br />

G crit<br />

G<br />

❏ Correlations are important, especially<br />

for small values of G<br />

❏ Static description breaks down when the<br />

single-particle level density is small


Recall that v 2 k is the occupation probability for state k. Thus, the<br />

pairing interaction destabilizes the Fermi surface!<br />

Inserting these results into Eq. (1) yields the Gap Equation<br />

∆ k = − 1 2<br />

∑<br />

v kkll<br />

l>0<br />

∆ k<br />

√˜ε<br />

2<br />

k + ∆ 2 k<br />

This equation, together with the Equation for the energies ˜ε k has to<br />

be solved iteratively.<br />

Simple examples and illustration:<br />

(1) Pure pairing force: v ijkl = −G, t k,l = δ kl ε l<br />

Gap equation<br />

∆ = G 2<br />

∑<br />

l>0<br />

∆<br />

√<br />

(εk − µ) 2 + ∆ 2 k<br />

For vanishing single-particle energies ε k = 0 and a single-j shell one<br />

recovers the results of the seniority model. The gap becomes<br />

√<br />

∆ = G<br />

n √<br />

2<br />

⎛<br />

⎝Ω − N 2<br />

⎞<br />

⎠ = GΩ<br />

2<br />

at half filling. Comparison with the seniority model shows that 2∆ is<br />

the energy to break a pair and equals the excitation energy!


(2) Analytical solution for simple model.<br />

Assume constant interaction G, constant gap ∆, and constant density<br />

of states ρ, and restrict summation to the vicinity δ of the Fermi<br />

surface. Gap equation:<br />

1 = G 2<br />

µ+δ ∫<br />

µ−δ<br />

For weak pairing Gρ ≪ 1 one finds<br />

ρ(ε)<br />

dε√ (ε − µ)2 + ∆ = Gρ arsinh δ 2 ∆<br />

⎛<br />

∆<br />

δ ∝ exp ⎜<br />

−1<br />

⎝<br />

|G|ρ<br />

Gap is nonperturbative in interaction G!<br />

The particle number variance is<br />

(∆N) 2 ≈ 2ρ∆ atan δ ⎧<br />

⎪⎨<br />

∆ = πρ∆ for weak pairing ∆ ≪ δ,<br />

⎪ ⎩ 2ρδ for strong pairing.<br />

⎞<br />

⎟<br />


<strong>BCS</strong> essentials<br />

1. The Fermi surface of a Fermi gas is unstable with respect to attractive<br />

interactions. The gap equation has a nontrivial solution<br />

∆ ≠ 0 whenever the interaction or the density of states is sufficiently<br />

large.<br />

2. The finite excitation gap causes superfluidity: The system cannot<br />

absorb arbitrarily small perturbations and therefore remains inert<br />

(See, e.g., Landau & Lifshitz, Statistical Mechanics II).

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!