Dimensional crossover in ultracold Fermi gases from Functional - - PowerPoint PPT Presentation
Dimensional crossover in ultracold Fermi gases from Functional - - PowerPoint PPT Presentation
Dimensional crossover in ultracold Fermi gases from Functional Renormalisation Bruno Faigle-Cedzich Cold Quantum Coffee Heidelberg University 8 th May 2018 Table of contents 1. Physics of ultracold atoms 2. BCS-BEC physics from Functional
Table of contents
- 1. Physics of ultracold atoms
- 2. BCS-BEC physics from Functional Renormalisation
- 3. Dimensional crossover
- 4. Conclusion
1
Physics of ultracold atoms
Scales
- interparticle spacing π = ββπ
- thermal wavelength πth
- van der Waals length πvdW
- oscillator length βosc
- scattering length π
2
Scales
- interparticle spacing π = ββπ
- thermal wavelength πth
Ultracold: β/πth β² 1
- van der Waals length πvdW
- oscillator length βosc
- scattering length π
2
Scales
- interparticle spacing π = ββπ
- thermal wavelength πth
Ultracold: β/πth β² 1
- van der Waals length πvdW
- oscillator length βosc
- scattering length π
2
Scales
- interparticle spacing π = ββπ
- thermal wavelength πth
Ultracold: β/πth β² 1
- van der Waals length πvdW
- oscillator length βosc
- scattering length π
2
Scales
- interparticle spacing π = ββπ
- thermal wavelength πth
Ultracold: β/πth β² 1
- van der Waals length πvdW
- oscillator length βosc
- scattering length π
2
Scales
- interparticle spacing π = ββπ
- thermal wavelength πth
Ultracold: β/πth β² 1
- van der Waals length πvdW
- oscillator length βosc
- scattering length π
2
πext = β π
2 (π /βosc)2
Scales
- interparticle spacing π = ββπ
- thermal wavelength πth
Ultracold: β/πth β² 1
- van der Waals length πvdW
- oscillator length βosc
- scattering length π
2
Scales
- interparticle spacing π = ββπ
- thermal wavelength πth
Ultracold: β/πth β² 1 Dilute: π π1/π βͺ 1
- van der Waals length πvdW
- oscillator length βosc
- scattering length π
2
Scale hierarchy & Hamiltonian
van der Waals length interaction scale interparticle spacing
adapted from Boettcher et al. Nuclear Physics B (2012)
thermal wavelength
- scillator length
- f trap
Effective Hamiltonian valid on scales β« βvdW: Μ πΌ = β«
β π¦
[ Μ πβ ( β π¦) (ββ β2 2 π + πext( β π¦)) Μ π( β π¦) + πΞ Μ π( β π¦)2] with Μ π = Μ πβ Μ π and πΞ = 4 π β2
π
π
3
Feshbach resonances
4
2-atom scattering
BEC BCS
Regal et al. 2006
π(πΆ) = Ξπ (πΆ β πΆ0) π = πbg (1 β ΞπΆ πΆ β πΆ0 ) with π β 0 @ resonance
Feshbach resonances
4 van der Waals length interaction scale interparticle spacing
adapted from Boettcher et al. Nuclear Physics B (2012)
Feshbach resonance thermal wavelength
- scillator length
- f trap
BEC BCS
Regal et al. 2006
π(πΆ) = Ξπ (πΆ β πΆ0) π = πbg (1 β ΞπΆ πΆ β πΆ0 ) with π β 0 @ resonance
The BCS-BEC crossover in 3D
5
two-component fermionic atoms
The BCS-BEC crossover in 3D
5
two-component fermionic atoms fermions with attractive interactions
The BCS-BEC crossover in 3D
5
two-component fermionic atoms fermions with attractive interactions bound molecules of two atoms
The BCS-BEC crossover in 3D
5
two-component fermionic atoms fermions with attractive interactions bound molecules of two atoms low π BCS-superfmuidity (condensation of Cooper pairs)
The BCS-BEC crossover in 3D
5
two-component fermionic atoms fermions with attractive interactions bound molecules of two atoms low π BCS-superfmuidity (condensation of Cooper pairs) low π BEC (Bose-Einstein condensate)
The BCS-BEC crossover in 3D
5
two-component fermionic atoms fermions with attractive interactions bound molecules of two atoms low π BCS-superfmuidity (condensation of Cooper pairs) low π BEC (Bose-Einstein condensate) Crossover (magnetic fjeld)
The BCS-BEC crossover in 3D
5
ππΊ = (3 π2 π)1/3
Randeria Nature (2010)
Features
advantages
- couplings are tunable
- high precision experiments
- microphysics known
challenges
- from microphysic laws to macroscopic observation
including fmuctuations
- large couplings
- different effective degrees of freedom
- microphysics: single atoms & molecules
- macrophysics: bosonic collective degrees of freedom
6
BCS-BEC physics from Functional Renormalisation
The functional RG
UV IR
Flow equation (Wetterich 1993)
ππ Ξπ = 1 2 STr [(Ξ(2) + ππ)β1 ππ ππ] Exact 1-loop equation
βt Ξk[Ο, Ο] = 1 2
1
7 full quantum effective action
bosons fermions
Regulator and truncation dependence
8
- fmow of Ξπ regulator
dependent, yet Ξ is not
- during fmow all possible
interactions may be produced
- ππ’Ξ(π) depends on
Ξ(π+1) and Ξ(π+2) βtruncation needed
adapted from Gies Springer (2012)
Theory space
Action and truncation
Microscopic action π = β«
π
[πβ (ππ β β2 β π) π + πβ (ππ β β2 2 + π β 2π) π β β (πβ π1 π2 β π πβ
1 πβ 2) ]
with
- π: Grassmann fjeld
- π: bosonic fjeld consisting of two atoms
- π: detuning
- π: Euclidean time on torus with circumference 1/π
- π: chemical potential
9
Units
Chosen such that: β = ππΆ = 2 π = 1 Consequences:
- β = 1: [momentum]=[length]β1 with typical momentum
unit ππΊ
- ππΆ = 1: [temperature]=[energy]
- 2 π = 1: [momentum]=[energy] (equiv. π = 1)
i.e. [π’] = π2, [ β π] = πβ1, [π] = πβ2, [π] = πβ2, [π] = πβ3 . β canonical dimensions differ from relativistic QFT!
10
Truncation: derivative expansion
- vertices are expanded in powers of the momenta
- effective average potential ππ @ least to 2nd order in
π = πβπ (2nd order phase transition)
βt ( )β1 = F + B
βt ( )β1 = M
11
Ansatz for effective action Ξπ = Ξkin + Ξint Ξkin[π, π] = β«
π
[ β
π={1,2}
πβ
π (ππππ β β2 β π) ππ + πβ (ππππ β 1
2β2) π] Ξint[π, π] = β«
π
[π (πβ π) β β (πβ π1 π2 β π πβ
1 πβ 2) ]
with
- renormalised fjelds: π = π΅1/2
π
Μ π, π = π΅1/2
π
Μ π
- ππ,π = ππ,π/π΅π,π
- anomalous dimensions: ππ,π = β ππ’ log π΅π,π
and π(π) =
π
β
π=1
π£π π! (π β π0)π β ππ (π β π0) +
π
β
π=1
π½π π! (π β π0) (π β π0)π , π£1 = π2
π 12
Regularisation scheme
IR-regularisation: lim
π2/π2β0 ππ(π2)/π2 > 0,
lim
π2/π2β0 ππ(π2) β 0
Litim-type regulator: ππ,π(π ) = ππ,π(π2) = (π2 β π2 2 ) π (π2 β π2 2 ) ππ,π(π ) = ππ,π(π2) = [π2 sgn (π2 β π) β (π2 β π)] π (π2 β |π2 β π|) βanalytic evaluation of Matsubara sums
13
UV renormalisation
Connecting to experiment
- initial condition for ππ: πΞ(π) = (πΞ β 2 π) π
- @ π = 0: connect to correct vacuum physics at unitarity
(πβ1 = 0), i.e. π2
π,π=0 = 0 = π
Thus: π = π(πΆ) = β2
Ξ
8 π π(πΆ) , π(πΆ) = πΞ β ππ(Ξ)
14
Universality
- existence of FPs in RG fmow: macrophysics independent of
the microphysics
- loss of memory of microphysics: start at the FP values
initial values β2
Ξ = 6 π2 Ξ,
ππ,Ξ = Μ ππ,β Ξ , π2
π,Ξ = πΞ β 2 π,
ππ,Ξ = 1, π½Ξ = β2, πΞ = π3/2 3 π2 Ξ(π).
15
3D BCS-BEC crossover
π = 0: (with π = ππΊ π)
- 4
- 2
2 4 0.0 0.2 0.4 0.6 0.8 1.0
- 4
- 2
2 4 0.0 0.5 1.0 1.5 2.0 2.5
π > 0:
- 4
- 2
2 4 0.0 0.5 1.0 1.5 2.0
- 4
- 2
2 4 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35
16
Dimensional crossover
Why are quasi-2D systems interesting?
- promising materials: graphene, high ππ-superconductors,
layered semiconductors
- pronounced infmuence of quantum fmuctuations
- experimental accessibility via highly anisotropic trapping
potentials
- for insuffjcient anisotropy βdimensional crossover
17
βdisentangle dimensionality from many-body physics
ZΓΌrn et al. PRL (2015)
Boundary conditions
- delimit π¨-direction by potential well of length π
πbox(π¨) = β§ { β¨ { β© 0 β€ π¨ β€ π β else
- impose periodic boundary conditions: Ξ¨ = {π, π}
Ξ¨(π, π¦, π§, π¨ = 0) = Ξ¨(π, π¦, π§, π¨ = π) βquantisation of momentum in π¨-direction: ππ¨ β ππ = 2 π π π , π β β
- spatial Matsubara sum
β« πππ (2 π)π = 1 π β
ππ
β« ππβ1π (2 π)πβ1
18
Regulator and method
Litim-type regulator (as before) with β π = Μβ π + ππ¨ β Μβ π + ππ Idea:
- initialise RG fmow at UV scale π = Ξ where ΞΞ coincides
with π of 3D gas
- π introduces new length scale to 3D system
- following the RG fmow we successively integrate out the 3rd
dimension β@ π β 0: system @ given confjnement length scale π Note
- UV scale is always chosen such that Ξ β« (πβ1, π1/2, π 1/2)
- system is effectively 2D if πβ1 β« all other many-body
scales
19
Regulator and method
Litim-type regulator (as before) with β π = Μβ π + ππ¨ β Μβ π + ππ Idea:
- initialise RG fmow at UV scale π = Ξ where ΞΞ coincides
with π of 3D gas
- π introduces new length scale to 3D system
- following the RG fmow we successively integrate out the 3rd
dimension β@ π β 0: system @ given confjnement length scale π Note
- UV scale is always chosen such that Ξ β« (πβ1, π1/2, π 1/2)
- system is effectively 2D if πβ1 β« all other many-body
scales
19
Regulator and method
Litim-type regulator (as before) with β π = Μβ π + ππ¨ β Μβ π + ππ Idea:
- initialise RG fmow at UV scale π = Ξ where ΞΞ coincides
with π of 3D gas
- π introduces new length scale to 3D system
- following the RG fmow we successively integrate out the 3rd
dimension β@ π β 0: system @ given confjnement length scale π Note
- UV scale is always chosen such that Ξ β« (πβ1, π1/2, π 1/2)
- system is effectively 2D if πβ1 β« all other many-body
scales
19
Zero temperature
20
- correct 3D-limit reached
- qualitatively correct
behaviour of equation
- f state (except very
small confjnements)
3D limit
- 4
- 2
2 4 0.0 0.2 0.4 0.6 0.8 1.0 3D limit
- 4
- 2
2 4 0.0 0.5 1.0 1.5 2.0 2.5
Connecting to experiment
quasi-2d scattering length: π(pbc)
2D
= π exp {β1 2 π π3D } , π(trap)
2D
= βπ¨ β π π΅ exp {ββπ 2 βπ¨ π3D } quasi-2d crossover parameter: ln(ππΊ π2D)
- 4
- 2
2 4 0.0 0.2 0.4 0.6 0.8 1.0
21
Finite temperature
0.01 0.10 1 10 100 0.10 0.15 0.20 0.25 0.01 0.10 1 10 100 1000 0.10 0.15 0.20 0.25 0.30 0.1 1 10 100 1000 0.10 0.15 0.20 0.25 0.30
22
Phase diagram
- 4
- 2
2 4 0.00 0.05 0.10 0.15 0.20 0.25 0.30
- 4
- 2
2 4 0.00 0.05 0.10 0.15 0.20 0.25
- 3D phase diagram reproduced for large π
- increased ππ/ππΊ around ln(ππΊ π2D) βΌ 1
23
Comparison to experiment
- 4
- 2
2 4 0.00 0.05 0.10 0.15 0.20
experimental data from Ries et al. PRL (2015)
24
Conclusion
Summary and Outlook
Conclusion
- dimensional crossover in Fermi gas with FRG
- qualitatively comparable to experiments
Outlook
- employ harmonic trapping potential
- quantitative precision
- particle-hole fmuctuations
- frequency dependent regulator
- explore thermodynamics
25
Thank you for your attention!
Phase diagram from experiment
26
- 8
- 6
- 4
- 2
2 4 0.0 0.1 0.2 0.3
BEC T/TF ln(kFa2D)
0.00 0.10 0.20 0.30 0.40 0.50
Nq/N BCS
Ries et al. PRL (2015)