...

SUPA Modelling weak and strong matter-light coupling with organic molecules Jonathan Keeling

by user

on
Category: Documents
10

views

Report

Comments

Transcript

SUPA Modelling weak and strong matter-light coupling with organic molecules Jonathan Keeling
Modelling weak and strong matter-light coupling
with organic molecules
Jonathan Keeling

SUPA
University of
St Andrews
1413-2013
Telluride, July 2015
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
1
Motivation: polariton condensates
CdTe Polariton Condensate
T ∼ 20K. [Kasprzak et al. Nature, ’06]
"Cavity"
Quantum Wells
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
2
Motivation: polariton condensates
CdTe Polariton Condensate
T ∼ 20K. [Kasprzak et al. Nature, ’06]
Models:
WIDBG
I
I
Statistical mechanics
Boltzmann/cGPE Hybrids
Saturable excitons
"Cavity"
Quantum Wells
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
2
Motivation: polariton condensates
CdTe Polariton Condensate
T ∼ 20K. [Kasprzak et al. Nature, ’06]
Models:
WIDBG
I
I
Statistical mechanics
Boltzmann/cGPE Hybrids
Saturable excitons
"Cavity"
Quantum Wells
Q1. Lasing crossover?
Q2. Energetics vs dynamics
(esp. spin state).
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
2
Motivation: polariton condensates
CdTe Polariton Condensate
T ∼ 20K. [Kasprzak et al. Nature, ’06]
Models:
WIDBG
I
I
Statistical mechanics
Boltzmann/cGPE Hybrids
Saturable excitons
"Cavity"
Q1. Lasing crossover?
Quantum Wells
Q2. Energetics vs dynamics
(esp. spin state).
Anthracene Polariton Lasing
T ∼ 300K
[Kena Cohen and Forrest, Nat.
Photon ’10]
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
2
Motivation: polariton condensates
CdTe Polariton Condensate
T ∼ 20K. [Kasprzak et al. Nature, ’06]
Models:
WIDBG
I
I
Statistical mechanics
Boltzmann/cGPE Hybrids
Saturable excitons
"Cavity"
Q1. Lasing crossover?
Quantum Wells
Q2. Energetics vs dynamics
(esp. spin state).
Anthracene Polariton Lasing
T ∼ 300K
Q1. Vibrational replicas?
Q2. Relevance of disorder?
Q3. Lasing vs
condensation?
[Kena Cohen and Forrest, Nat.
Photon ’10]
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
2
Motivation: photon condensates
Photon Condensate T ∼ 300K
[Klaers et al. Nature, ’10]
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
3
Motivation: photon condensates
Photon Condensate T ∼ 300K
Q1. Relation to dye laser?
Q2. Relation to polaritons?
Q3. Thermalisation
breakdown?
[Klaers et al. Nature, ’10]
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
3
Motivation: vacuum-state strong coupling
Linear response (no pump, no
condensate): effects of
matter-light coupling alone.
[Canaguier-Durand et al. Angew. Chem. ’13]
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
4
Motivation: vacuum-state strong coupling
Linear response (no pump, no
condensate): effects of
matter-light coupling alone.
Q1. Can ultra-strong coupling
to light change:
I
I
I
I
charge distribution?
vibrational configuration?
molecular orientation?
crystal structure?
Q2. Are
√ changes collective
( N factor) or not?
[Canaguier-Durand et al. Angew. Chem. ’13]
Jonathan Keeling
Q3. If not, what is data
showing?
Modelling matter-light coupling
Telluride, July 2015
4
1
Modelling photon BEC & organic polaritons
2
(Ultra-)strong coupling, weak pumping
Ultra strong coupling & reconfiguration
Vibrational sidebands in spectrum
3
Driven dissipative systems
Limitations of rate equation model
Toy problem – two bosonic modes
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
5
Modelling photon BEC & organic polaritons
1
Modelling photon BEC & organic polaritons
2
(Ultra-)strong coupling, weak pumping
Ultra strong coupling & reconfiguration
Vibrational sidebands in spectrum
3
Driven dissipative systems
Limitations of rate equation model
Toy problem – two bosonic modes
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
6
What kinds of modelling
Bottom up
I
I
I
DFT (or quantum chemistry)
→ electronic structure
Time-dependent DFT /MD
→ vibrational spectra
FDTD/transfer-matrix
→ cavity modes
Tractable microscopic toy models
Top-down
I
I
I
Jonathan Keeling
Equilibrium stat. mech.
(complex/stochastic/. . . )GPE (+
Boltzmann)
Rate equations
Modelling matter-light coupling
Telluride, July 2015
7
What kinds of modelling
Bottom up
I
I
I
DFT (or quantum chemistry)
→ electronic structure
Time-dependent DFT /MD
→ vibrational spectra
FDTD/transfer-matrix
→ cavity modes
Tractable microscopic toy models
Top-down
I
I
I
Jonathan Keeling
Equilibrium stat. mech.
(complex/stochastic/. . . )GPE (+
Boltzmann)
Rate equations
Modelling matter-light coupling
Telluride, July 2015
7
What kinds of modelling
Bottom up
I
I
I
DFT (or quantum chemistry)
→ electronic structure
Time-dependent DFT /MD
→ vibrational spectra
FDTD/transfer-matrix
→ cavity modes
Tractable microscopic toy models
Top-down
I
I
I
Jonathan Keeling
Equilibrium stat. mech.
(complex/stochastic/. . . )GPE (+
Boltzmann) → condensate
Rate equations → laser
Modelling matter-light coupling
Telluride, July 2015
7
What kinds of modelling
Bottom up
I
I
I
DFT (or quantum chemistry)
→ electronic structure
Time-dependent DFT /MD
→ vibrational spectra
FDTD/transfer-matrix
→ cavity modes
Tractable microscopic toy models
Top-down
I
I
[From Auerbach, Interacting
I
Equilibrium stat. mech.
(complex/stochastic/. . . )GPE (+
Boltzmann) → condensate
Rate equations → laser
electrons and quantum magnetism]
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
7
Toy models
1
Full molecular spectra electronic
structure & Raman spectrum
3
Simplified archetypal model: Dicke-Holstein
Each molecule: two DoF
I
I
Electronic state: 2LS
Vibrational state: harmonic oscillator
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
8
Toy models
Full molecular spectra electronic
structure & Raman spectrum
Energy
1
⇑
2
3
nuclear coordinate
Focus on low-energy effective theory
⇓
Two-level system, HOMO/LUMO
See also [Galego, Garcia-Vidal,
Single DoF PES
Feist. arXiv:1506:03331]
Simplified archetypal model: Dicke-Holstein
Each molecule: two DoF
I
I
Electronic state: 2LS
Vibrational state: harmonic oscillator
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
8
Toy models
Full molecular spectra electronic
structure & Raman spectrum
Energy
1
⇑
Photon
2
3
nuclear coordinate
Focus on low-energy effective theory
⇓
Two-level system, HOMO/LUMO
See also [Galego, Garcia-Vidal,
Single DoF PES
Feist. arXiv:1506:03331]
Simplified archetypal model: Dicke-Holstein
Each molecule: two DoF
I
I
Electronic state: 2LS
Vibrational state: harmonic oscillator
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
8
Toy models
Full molecular spectra electronic
structure & Raman spectrum
Energy
1
⇑
Photon
2
3
nuclear coordinate
Focus on low-energy effective theory
⇓
Two-level system, HOMO/LUMO
See also [Galego, Garcia-Vidal,
Single DoF PES
Feist. arXiv:1506:03331]
Simplified archetypal model: Dicke-Holstein
Each molecule: two DoF
I
I
Electronic state: 2LS
Vibrational state: harmonic oscillator
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
8
Toy models
Full molecular spectra electronic
structure & Raman spectrum
Energy
1
⇑
Photon
2
3
nuclear coordinate
Focus on low-energy effective theory
⇓
Two-level system, HOMO/LUMO
See also [Galego, Garcia-Vidal,
Single DoF PES
Feist. arXiv:1506:03331]
Simplified archetypal model: Dicke-Holstein
Each molecule: two DoF
I
I
Electronic state: 2LS
Vibrational state: harmonic oscillator
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
8
Dicke Holstein Model
Dicke model: 2LS ↔ photons
Molecular vibrational mode
I
I
Phonon frequency Ω
Huang-Rhys parameter S —
coupling strength
Collective coupling to light
Hsys = ωψ † ψ +
Xh
α
Jonathan Keeling
2
i
σαz + g ψ + ψ † σα+ + σα−
Modelling matter-light coupling
Telluride, July 2015
9
Dicke Holstein Model
Dicke model: 2LS ↔ photons
Molecular vibrational mode
I
I
Phonon frequency Ω
Huang-Rhys parameter S —
coupling strength
Collective coupling to light
Hsys = ωψ † ψ +
Xh
α
2
i
σαz + g ψ + ψ † σα+ + σα−
+
X
n
o
√
Ω bα† bα + Sσαz bα† + bα
α
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
9
Dicke Holstein Model
Dicke model: 2LS ↔ photons
Molecular vibrational mode
I
I
Phonon frequency Ω
Huang-Rhys parameter S —
coupling strength
Collective coupling to light
Hsys = ωψ † ψ +
Xh
α
2
i
σαz + g ψ + ψ † σα+ + σα−
+
X
n
o
√
Ω bα† bα + Sσαz bα† + bα
α
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
9
(Ultra-)strong coupling, weak pumping
1
Modelling photon BEC & organic polaritons
2
(Ultra-)strong coupling, weak pumping
Ultra strong coupling & reconfiguration
Vibrational sidebands in spectrum
3
Driven dissipative systems
Limitations of rate equation model
Toy problem – two bosonic modes
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
10
Ultra-strong coupling, changing configuration
√
√
Ultra-strong coupling: ω, ∼ g N ∝ concentration
Normal state: configuration of molecules
[Canaguier-Durand et al. Angew. Chem. ’13 ]
I
I
Polariton vs molecular spectral weight — chemical eqbm
Temperature dependent
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
11
Ultra-strong coupling, changing configuration
√
√
Ultra-strong coupling: ω, ∼ g N ∝ concentration
Normal state: configuration of molecules
[Canaguier-Durand et al. Angew. Chem. ’13 ]
I
I
Polariton vs molecular spectral weight — chemical eqbm
Temperature dependent
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
11
Disordered molecules — spectrum
Calculate Green’s function GR (ν):
T (ν) ∝ |GR (ν)|2 ,
A(ν) ∝ −=[GR (ν)] + (interference)
Ultra-strong coupling — renormalised photon
50
Central peak:
g√N=0.3 eV
Spectral weight
40
30
20
10
0
1.6
1.8
2
2.2
2.4 2.6
ω [eV]
2.8
3
3.2
√
Temperature independent (for kB T g N)
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
12
Disordered molecules — spectrum
Calculate Green’s function GR (ν):
T (ν) ∝ |GR (ν)|2 ,
A(ν) ∝ −=[GR (ν)] + (interference)
Ultra-strong coupling — renormalised photon
50
Spectral weight
Central peak:
g√N=0.3 eV
g√N=0.5 eV
40
30
20
10
0
1.6
1.8
2
2.2
2.4 2.6
ω [eV]
2.8
3
3.2
√
Temperature independent (for kB T g N)
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
12
Disordered molecules — spectrum
Calculate Green’s function GR (ν):
T (ν) ∝ |GR (ν)|2 ,
A(ν) ∝ −=[GR (ν)] + (interference)
Ultra-strong coupling — renormalised photon
50
Spectral weight
Central peak:
g√N=0.3 eV
g√N=0.5 eV
g√N=0.7 eV
40
30
20
10
0
1.6
1.8
2
2.2
2.4 2.6
ω [eV]
2.8
3
3.2
√
Temperature independent (for kB T g N)
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
12
Disordered molecules — spectrum
Calculate Green’s function GR (ν):
T (ν) ∝ |GR (ν)|2 ,
A(ν) ∝ −=[GR (ν)] + (interference)
Ultra-strong coupling — renormalised photon
50
Spectral weight
Central peak:
g√N=0.3 eV
g√N=0.5 eV
g√N=0.7 eV
40
30
20
10
0
1.6
1.8
2
2.2
2.4 2.6
ω [eV]
2.8
3
3.2
√
Temperature independent (for kB T g N)
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
12
Disordered molecules — spectrum
Calculate Green’s function GR (ν):
T (ν) ∝ |GR (ν)|2 ,
A(ν) ∝ −=[GR (ν)] + (interference)
Ultra-strong coupling — renormalised photon
Spectral weight
0.4
Central peak:
g√N=0.3 eV
g√N=0.5 eV
g√N=0.7 eV
0.3
0.2
0.1
0
1.6
1.8
2
2.2
2.4 2.6
ω [eV]
2.8
3
3.2
√
Temperature independent (for kB T g N)
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
12
Disordered molecules — spectrum
Calculate Green’s function GR (ν):
T (ν) ∝ |GR (ν)|2 ,
A(ν) ∝ −=[GR (ν)] + (interference)
Ultra-strong coupling — renormalised photon
Spectral weight
0.4
Central peak:
g√N=0.3 eV
g√N=0.5 eV
g√N=0.7 eV
0.3
GR (ν) =
1
R (ν)
ν + iκ/2 − ωk − g 2 GExc.
2
κ
R
A(ν) ∼
− =[GExc.
] GR (ν)
2
0.2
0.1
0
1.6
1.8
2
2.2
2.4 2.6
ω [eV]
2.8
3
3.2
[Houdré et al. , PRA ’96]
√
Temperature independent (for kB T g N)
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
12
Disordered molecules — spectrum
Calculate Green’s function GR (ν):
T (ν) ∝ |GR (ν)|2 ,
A(ν) ∝ −=[GR (ν)] + (interference)
Ultra-strong coupling — renormalised photon
Spectral weight
0.4
Central peak:
g√N=0.3 eV
g√N=0.5 eV
g√N=0.7 eV
0.3
GR (ν) =
1
R (ν)
ν + iκ/2 − ωk − g 2 GExc.
2
κ
R
A(ν) ∼
− =[GExc.
] GR (ν)
2
0.2
0.1
0
1.6
1.8
2
2.2
2.4 2.6
ω [eV]
2.8
3
3.2
[Houdré et al. , PRA ’96]
√
Temperature independent (for kB T g N)
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
12
Molecular adaptation
Central peak — depends on g, not T .
Can geff depend on T ?
Rotational degrees of freedom
i
Xh
H = ... +
. . . + gα,k cos(θα )(ψk† + ψ−k )σαx + E0 (θα )
α
Schrieffer-Wolff, δH =
Heff = . . . +
P
α,k
gα,k (ψk† σα+ + H.c.):
i
Xh
−K0 cos2 (θα ) + E0 (θ) ,
α
I
No
√
K0 =
X
k
gk2
ωk + N enhancement — K0 small, independent of density
[Cwik et al. , arXiv:1506.08974]
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
13
Molecular adaptation
Central peak — depends on g, not T .
Can geff depend on T ?
Rotational degrees of freedom
i
Xh
H = ... +
. . . + gα,k cos(θα )(ψk† + ψ−k )σαx + E0 (θα )
α
Schrieffer-Wolff, δH =
Heff = . . . +
P
α,k
gα,k (ψk† σα+ + H.c.):
i
Xh
−K0 cos2 (θα ) + E0 (θ) ,
α
I
No
√
K0 =
X
k
gk2
ωk + N enhancement — K0 small, independent of density
[Cwik et al. , arXiv:1506.08974]
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
13
Molecular adaptation
Central peak — depends on g, not T .
Can geff depend on T ?
θ
ld
E
fie
Rotational degrees of freedom
i
Xh
H = ... +
. . . + gα,k cos(θα )(ψk† + ψ−k )σαx + E0 (θα )
α
Schrieffer-Wolff, δH =
Heff = . . . +
P
α,k
gα,k (ψk† σα+ + H.c.):
i
Xh
−K0 cos2 (θα ) + E0 (θ) ,
α
I
No
√
K0 =
X
k
gk2
ωk + N enhancement — K0 small, independent of density
[Cwik et al. , arXiv:1506.08974]
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
13
Molecular adaptation
Central peak — depends on g, not T .
Can geff depend on T ?
θ
ld
E
fie
Rotational degrees of freedom
i
Xh
H = ... +
. . . + gα,k cos(θα )(ψk† + ψ−k )σαx + E0 (θα )
α
Schrieffer-Wolff, δH =
Heff = . . . +
P
α,k
gα,k (ψk† σα+ + H.c.):
i
Xh
−K0 cos2 (θα ) + E0 (θ) ,
α
I
No
√
K0 =
X
k
gk2
ωk + N enhancement — K0 small, independent of density
[Cwik et al. , arXiv:1506.08974]
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
13
Molecular adaptation
Central peak — depends on g, not T .
Can geff depend on T ?
θ
ld
E
fie
Rotational degrees of freedom
i
Xh
H = ... +
. . . + gα,k cos(θα )(ψk† + ψ−k )σαx + E0 (θα )
α
Schrieffer-Wolff, δH =
Heff = . . . +
P
α,k
gα,k (ψk† σα+ + H.c.):
i
Xh
−K0 cos2 (θα ) + E0 (θ) ,
α
I
No
√
K0 =
X
k
gk2
ωk + N enhancement — K0 small, independent of density
[Cwik et al. , arXiv:1506.08974]
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
13
Vibrational adaptation
Schrieffer-Wolff – mixes vibrational states
(
" √ #)
√
X
gk2
Ω S(b + b† )
Ω 2 g N
,
Heff = H0 −
1−
+O
2( + ωk )
+ ωk
k
Reduced vibrational offset
S → S(1 − 2K1 ),
K1 =
X
k
I
I
gk2
(ωk + )2
2
Increased effective coupling: geff
= g 2 exp(−S)
Again, K1 1, independent of density.
[Cwik et al. , arXiv:1506.08974]
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
14
Vibrational adaptation
Schrieffer-Wolff – mixes vibrational states
(
" √ #)
√
X
gk2
Ω S(b + b† )
Ω 2 g N
,
Heff = H0 −
1−
+O
2( + ωk )
+ ωk
k
Energy
Reduced vibrational offset
S → S(1 − 2K1 ),
K1 =
X
k
gk2
(ωk + )2
⇑
Photon
nuclear coordinate
⇓
l
S ×l
I
I
2
Increased effective coupling: geff
= g 2 exp(−S)
Again, K1 1, independent of density.
[Cwik et al. , arXiv:1506.08974]
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
14
Vibrational adaptation
Schrieffer-Wolff – mixes vibrational states
(
" √ #)
√
X
gk2
Ω S(b + b† )
Ω 2 g N
,
Heff = H0 −
1−
+O
2( + ωk )
+ ωk
k
Energy
Reduced vibrational offset
S → S(1 − 2K1 ),
K1 =
X
k
gk2
(ωk + )2
⇑
Photon
nuclear coordinate
⇓
l
S ×l
I
I
2
Increased effective coupling: geff
= g 2 exp(−S)
Again, K1 1, independent of density.
[Cwik et al. , arXiv:1506.08974]
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
14
Vibrational adaptation
Schrieffer-Wolff – mixes vibrational states
(
" √ #)
√
X
gk2
Ω S(b + b† )
Ω 2 g N
,
Heff = H0 −
1−
+O
2( + ωk )
+ ωk
k
Energy
Reduced vibrational offset
S → S(1 − 2K1 ),
K1 =
X
k
gk2
(ωk + )2
⇑
Photon
nuclear coordinate
⇓
l
S ×l
I
I
2
Increased effective coupling: geff
= g 2 exp(−S)
Again, K1 1, independent of density.
[Cwik et al. , arXiv:1506.08974]
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
14
Any other kind of adaptation
Generic (classical) DoF (solvation, charge distribution . . . )
i
Xh
H = ... +
. . . + gα,k f (xα )(ψk† + ψ−k )σαx + E0 (xα )
α
Schrieffer-Wolff:
Heff = . . . +
Xh
i
E0 (xα ) − K0 f 2 (xα ) ,
K0 =
α
I
I
X
k
gk2
ωk + Ground state energy: −K0 Nf (x0 ), collective.
Extent of reconfiguration, δx0 ∼ g 2 , not collective.
Why:
I
I
R
2
dxf (x)e−βN [E0 (x)−K0 f (x)]
geff
If xα = x, then
= R
g
dxe−βN[E0 (x)−K0 f 2 (x)]
But xα are all individuals:
P
Q R
−β α E0 (xα )−K0 f 2 (xα )
geff,α0
α dxα f (xα0 )eP
=
Q R
−β α E0 (xα )−K0 f 2 (xα )
g
α dxα e
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
15
Any other kind of adaptation
Generic (classical) DoF (solvation, charge distribution . . . )
i
Xh
H = ... +
. . . + gα,k f (xα )(ψk† + ψ−k )σαx + E0 (xα )
α
Schrieffer-Wolff:
Heff = . . . +
Xh
i
E0 (xα ) − K0 f 2 (xα ) ,
K0 =
α
I
I
X
k
gk2
ωk + Ground state energy: −K0 Nf (x0 ), collective.
Extent of reconfiguration, δx0 ∼ g 2 , not collective.
Why:
I
I
R
2
dxf (x)e−βN [E0 (x)−K0 f (x)]
geff
If xα = x, then
= R
g
dxe−βN[E0 (x)−K0 f 2 (x)]
But xα are all individuals:
P
Q R
−β α E0 (xα )−K0 f 2 (xα )
geff,α0
α dxα f (xα0 )eP
=
Q R
−β α E0 (xα )−K0 f 2 (xα )
g
α dxα e
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
15
Any other kind of adaptation
Generic (classical) DoF (solvation, charge distribution . . . )
i
Xh
H = ... +
. . . + gα,k f (xα )(ψk† + ψ−k )σαx + E0 (xα )
α
Schrieffer-Wolff:
Heff = . . . +
Xh
i
E0 (xα ) − K0 f 2 (xα ) ,
K0 =
α
I
I
X
k
gk2
ωk + Ground state energy: −K0 Nf (x0 ), collective.
Extent of reconfiguration, δx0 ∼ g 2 , not collective.
Why:
I
I
R
2
dxf (x)e−βN [E0 (x)−K0 f (x)]
geff
If xα = x, then
= R
g
dxe−βN[E0 (x)−K0 f 2 (x)]
But xα are all individuals:
P
Q R
−β α E0 (xα )−K0 f 2 (xα )
geff,α0
α dxα f (xα0 )eP
=
Q R
−β α E0 (xα )−K0 f 2 (xα )
g
α dxα e
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
15
Any other kind of adaptation
Generic (classical) DoF (solvation, charge distribution . . . )
i
Xh
H = ... +
. . . + gα,k f (xα )(ψk† + ψ−k )σαx + E0 (xα )
α
Schrieffer-Wolff:
Heff = . . . +
Xh
i
E0 (xα ) − K0 f 2 (xα ) ,
K0 =
α
I
I
X
k
gk2
ωk + Ground state energy: −K0 Nf (x0 ), collective.
Extent of reconfiguration, δx0 ∼ g 2 , not collective.
Why:
I
I
R
2
dxf (x)e−βN [E0 (x)−K0 f (x)]
geff
If xα = x, then
= R
g
dxe−βN[E0 (x)−K0 f 2 (x)]
But xα are all individuals:
P
Q R
−β α E0 (xα )−K0 f 2 (xα )
geff,α0
α dxα f (xα0 )eP
=
Q R
−β α E0 (xα )−K0 f 2 (xα )
g
α dxα e
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
15
Any other kind of adaptation
Generic (classical) DoF (solvation, charge distribution . . . )
i
Xh
H = ... +
. . . + gα,k f (xα )(ψk† + ψ−k )σαx + E0 (xα )
α
Schrieffer-Wolff:
Heff = . . . +
Xh
i
E0 (xα ) − K0 f 2 (xα ) ,
K0 =
α
I
I
X
k
gk2
ωk + Ground state energy: −K0 Nf (x0 ), collective.
Extent of reconfiguration, δx0 ∼ g 2 , not collective.
Why:
I
I
R
2
dxf (x)e−βN [E0 (x)−K0 f (x)]
geff
If xα = x, then
= R
g
dxe−βN[E0 (x)−K0 f 2 (x)]
But xα are all individuals:
P
Q R
−β α E0 (xα )−K0 f 2 (xα )
geff,α0
α dxα f (xα0 )eP
=
Q R
−β α E0 (xα )−K0 f 2 (xα )
g
α dxα e
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
15
Any other kind of adaptation
Generic (classical) DoF (solvation, charge distribution . . . )
i
Xh
H = ... +
. . . + gα,k f (xα )(ψk† + ψ−k )σαx + E0 (xα )
α
Schrieffer-Wolff:
Heff = . . . +
Xh
i
E0 (xα ) − K0 f 2 (xα ) ,
K0 =
α
I
I
X
k
gk2
ωk + Ground state energy: −K0 Nf (x0 ), collective.
Extent of reconfiguration, δx0 ∼ g 2 , not collective.
Why:
I
I
R
2
dxf (x)e−βN [E0 (x)−K0 f (x)]
geff
If xα = x, then
= R
g
dxe−βN[E0 (x)−K0 f 2 (x)]
But xα are all individuals:
P
Q R
−β α E0 (xα )−K0 f 2 (xα )
geff,α0
α dxα f (xα0 )eP
=
Q R
−β α E0 (xα )−K0 f 2 (xα )
g
α dxα e
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
15
Any other kind of adaptation
Generic (classical) DoF (solvation, charge distribution . . . )
i
Xh
H = ... +
. . . + gα,k f (xα )(ψk† + ψ−k )σαx + E0 (xα )
α
Schrieffer-Wolff:
Heff = . . . +
Xh
i
E0 (xα ) − K0 f 2 (xα ) ,
K0 =
α
I
I
X
k
gk2
ωk + Ground state energy: −K0 Nf (x0 ), collective.
Extent of reconfiguration, δx0 ∼ g 2 , not collective.
Why: Entropy – this is why BEC matters!!
I
I
R
2
dxf (x)e−βN [E0 (x)−K0 f (x)]
geff
If xα = x, then
= R
g
dxe−βN[E0 (x)−K0 f 2 (x)]
But xα are all individuals:
R
2
dxα0 f (xα0 )e−β[E0 (xα0 )−K0 f (xα0 )]
geff,α0
=
R
2
g
dxα0 e−β[E0 (xα0 )−K0 f (xα0 )]
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
15
Disordered molecules — vibrational mode
But: spectrum with vibrational sidebands, S = 0.02
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
16
Disordered molecules — vibrational mode
But: spectrum with vibrational sidebands, S = 0.02
Spectral weight
0.4
g√N=0.3eV
g√N=0.5eV
g√N=0.7eV
0.3
0.2
0.1
0
1.6
Jonathan Keeling
1.8
2
2.2
2.4 2.6
ω [eV]
Modelling matter-light coupling
2.8
3
3.2
Telluride, July 2015
16
Disordered molecules — vibrational mode
Spectral weight
0.4
g√N=0.3eV
g√N=0.5eV
g√N=0.7eV
0.3
0.2
0.1
Bare molecule
But: spectrum with vibrational sidebands, S = 0.02
1.9
2
2.4 2.6
ω [eV]
2.8
2.1
0
1.6
Jonathan Keeling
1.8
2
2.2
Modelling matter-light coupling
3
3.2
Telluride, July 2015
16
Disordered molecules + vibrations – vs temperature
Stronger disorder &
S = 0.5, σ = 0.025eV
vs vs temperature
Spectral weight
0.05
0.04
kBT
[eV]
0.1
0.03
0
1.7
1.8
1.9
2
2.1 2.2
ω [eV]
2.3
2.4
2.5
[Cwik et al. , arXiv:1506.08974]
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
17
Disordered molecules + vibrations – vs temperature
Stronger disorder &
S = 0.5, σ = 0.025eV
vs vs temperature
Spectral weight
0.05
0.04
kBT
[eV]
0.1
0.03
0
1.7
1.8
1.9
2
2.1 2.2
ω [eV]
2.3
2.4
2.5
[Cwik et al. , arXiv:1506.08974]
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
17
Disordered molecules + vibrations – vs temperature
Stronger disorder &
S = 0.5, σ = 0.025eV
vs vs temperature
Spectral weight
0.05
0.04
kBT
[eV]
0.1
0.03
0
1.7
1.8
1.9
2
2.1 2.2
ω [eV]
2.3
2.4
2.5
[Cwik et al. , arXiv:1506.08974]
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
17
Disordered molecules + vibrations – vs temperature
Stronger disorder &
S = 0.5, σ = 0.025eV
vs vs temperature
Spectral weight
0.05
0.04
kBT
[eV]
0.1
0.03
0
1.7
1.8
1.9
2
2.1 2.2
ω [eV]
2.3
2.4
2.5
[Cwik et al. , arXiv:1506.08974]
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
17
Disordered molecules + vibrations – vs temperature
Stronger disorder &
S = 0.5, σ = 0.025eV
vs vs temperature
S = 0.02, σ = 0.01eV
1.2
0.05
0.05
Spectral weight
Spectral weight
1
0.04
kBT
[eV]
0.1
0.03
0.8
0.6
0.04
kBT
[eV]
0.4
0.03
0.2
0
0
1.7
1.8
1.9
2
2.1 2.2
ω [eV]
2.3
2.4
2.5
1.7
1.8
1.9
2
2.1 2.2
ω [eV]
2.3
2.4
2.5
[Cwik et al. , arXiv:1506.08974]
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
17
Driven dissipative systems
1
Modelling photon BEC & organic polaritons
2
(Ultra-)strong coupling, weak pumping
Ultra strong coupling & reconfiguration
Vibrational sidebands in spectrum
3
Driven dissipative systems
Limitations of rate equation model
Toy problem – two bosonic modes
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
18
Photon BEC, rate equations & oscillations
Photon BEC: can derive photon rate equation [See Kirton talk]
1
∂t nm = −κnm
+ Γ(−δm )(nm + 1)N↑ − Γ(δm )nm N↓
Spectrum
0.8
Γ(-δ)
Γ(δ)
0.6
0.4
0.2
0
-400
-200
0
200
δ=ω - ωZPL
400
Experiments [Schmitt et al. PRA ’15
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
19
Photon BEC, rate equations & oscillations
Photon BEC: can derive photon rate equation [See Kirton talk]
1
∂t nm = −κnm
+ Γ(−δm )(nm + 1)N↑ − Γ(δm )nm N↓
Spectrum
0.8
Γ(-δ)
Γ(δ)
0.6
0.4
0.2
0
-400
-200
0
200
δ=ω - ωZPL
400
Experiments [Schmitt et al. PRA ’15
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
19
Photon BEC rate equations
∂t nm = −κnm + Γ(−δm )(nm + 1)N↑
− Γ(δm )nm N↓
Describes emission into
Gauss-Hermite mode m
X
I(x) =
nm |ψm (x)|2
m
Oscillations: beating of modes.
P
Need I(x) = m,m0 nm,m0 ψm (x)ψm0 (x)
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
20
Photon BEC rate equations
∂t nm = −κnm + Γ(−δm )(nm + 1)N↑
− Γ(δm )nm N↓
Describes emission into
Gauss-Hermite mode m
X
I(x) =
nm |ψm (x)|2
m
Oscillations: beating of modes.
P
Need I(x) = m,m0 nm,m0 ψm (x)ψm0 (x)
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
20
Photon BEC rate equations
∂t nm = −κnm + Γ(−δm )(nm + 1)N↑
− Γ(δm )nm N↓
Describes emission into
Gauss-Hermite mode m
X
I(x) =
nm |ψm (x)|2
m
Oscillations: beating of modes.
P
Need I(x) = m,m0 nm,m0 ψm (x)ψm0 (x)
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
20
Photon BEC rate equations
∂t nm = −κnm + Γ(−δm )(nm + 1)N↑
− Γ(δm )nm N↓
Describes emission into
Gauss-Hermite mode m
X
I(x) =
nm |ψm (x)|2
m
Oscillations: beating of modes.
P
Need I(x) = m,m0 nm,m0 ψm (x)ψm0 (x)
Emission must create coherence between non-degenerate modes.
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
20
Toy problem: two bosonic modes
Basic problem: Emission from thermal bath
Mode b
H = ωa ψ̂a† ψ̂a + ωb ψ̂b† ψ̂b + HBath
X
+ (ϕ∗a ψ̂a† + ϕ∗b ψ̂b† )
gi ĉi + H.c.
Bath
J( ν)
Mode a
Jonathan Keeling
i
ν
Modelling matter-light coupling
Telluride, July 2015
21
Toy problem: naı̈ve solutions
Two “expected” behaviours:
At resonance: “weak lasing” — coupling to bath dominates
d
ρ = Γ↓ L[ϕa ψ̂a + ϕb ψ̂b ] + Γ↑ L[ϕ∗a ψ̂a† + ϕ∗b ψ̂b† ]
dt
Far from resonance: pointer states are eigenstates
X ↓
d
ρ=
Γi L[ψ̂i ] + Γ↑i L[ψ̂i† ]
dt
i=a,b
Questions:
I
I
How does crossover work?
Are these actually right?
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
22
Toy problem: naı̈ve solutions
Two “expected” behaviours:
At resonance: “weak lasing” — coupling to bath dominates
d
ρ = Γ↓ L[ϕa ψ̂a + ϕb ψ̂b ] + Γ↑ L[ϕ∗a ψ̂a† + ϕ∗b ψ̂b† ]
dt
Far from resonance: pointer states are eigenstates
X ↓
d
ρ=
Γi L[ψ̂i ] + Γ↑i L[ψ̂i† ]
dt
i=a,b
Questions:
I
I
How does crossover work?
Are these actually right?
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
22
Toy problem: naı̈ve solutions
Two “expected” behaviours:
At resonance: “weak lasing” — coupling to bath dominates
d
ρ = Γ↓ L[ϕa ψ̂a + ϕb ψ̂b ] + Γ↑ L[ϕ∗a ψ̂a† + ϕ∗b ψ̂b† ]
dt
Far from resonance: pointer states are eigenstates
X ↓
d
ρ=
Γi L[ψ̂i ] + Γ↑i L[ψ̂i† ]
dt
i=a,b
Questions:
I
I
How does crossover work?
Are these actually right?
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
22
Toy problem: exact solution
Solve via Laplace transform. Find Fij (t) = hψ̂i† (t)ψ̂j (t)i
Steady state:
I
Singular at ∆ = 0
Time evolution —
Fab (t) ∼ exp(−α∆2 t)
Always some coherence
I
(individual always wrong)
Fab ∼ Faa , Fbb only at ∆ = 0
I
Jonathan Keeling
(collective almost always wrong)
Modelling matter-light coupling
Telluride, July 2015
23
Toy problem: exact solution
I
Singular at ∆ = 0
1×10-2
0
Faa, Fbb
Time evolution —
Fab (t) ∼ exp(−α∆2 t)
Re[Fab]
Solve via Laplace transform. Find Fij (t) = hψ̂i† (t)ψ̂j (t)i
Steady state:
2×10-2
Fab
0.2
Faa
Fbb
0
-0.4
-0.2
0
0.2
-∆=2(ωb-ωa)
0.4
Always some coherence
I
(individual always wrong)
Fab ∼ Faa , Fbb only at ∆ = 0
I
Jonathan Keeling
(collective almost always wrong)
Modelling matter-light coupling
Telluride, July 2015
23
Toy problem: exact solution
I
Singular at ∆ = 0
1×10-2
0
Faa, Fbb
Time evolution —
Fab (t) ∼ exp(−α∆2 t)
Re[Fab]
Solve via Laplace transform. Find Fij (t) = hψ̂i† (t)ψ̂j (t)i
Steady state:
2×10-2
Fab
0.2
Faa
Fbb
0
-0.4
-0.2
0
0.2
-∆=2(ωb-ωa)
0.4
Always some coherence
I
(individual always wrong)
Fab ∼ Faa , Fbb only at ∆ = 0
I
Jonathan Keeling
(collective almost always wrong)
Modelling matter-light coupling
Telluride, July 2015
23
Toy problem: exact solution
Singular at ∆ = 0
Time evolution —
Fab (t) ∼ exp(−α∆2 t)
Fab
0
1×10-2
0
0.05
0.1
2000
Faa, Fbb
I
Re[Fab]
Solve via Laplace transform. Find Fij (t) = hψ̂i† (t)ψ̂j (t)i
Steady state:
2×10-2
Fab
0.2
Faa
Fbb
0
-0.4
Time
1500
0
0.2
-∆=2(ωb-ωa)
0.4
Always some coherence
1000
I
(individual always wrong)
Fab ∼ Faa , Fbb only at ∆ = 0
500
I
0
-0.4
-0.2
-0.2
Jonathan Keeling
0
0.2
∆=2(ωa-ωb)
(collective almost always wrong)
0.4
Modelling matter-light coupling
Telluride, July 2015
23
Toy problem: exact solution
Singular at ∆ = 0
Time evolution —
Fab (t) ∼ exp(−α∆2 t)
Fab
0
1×10-2
0
0.05
0.1
2000
Faa, Fbb
I
Re[Fab]
Solve via Laplace transform. Find Fij (t) = hψ̂i† (t)ψ̂j (t)i
Steady state:
2×10-2
Fab
0.2
Faa
Fbb
0
-0.4
Time
1500
0
0.2
-∆=2(ωb-ωa)
0.4
Always some coherence
1000
I
(individual always wrong)
Fab ∼ Faa , Fbb only at ∆ = 0
500
I
0
-0.4
-0.2
-0.2
Jonathan Keeling
0
0.2
∆=2(ωa-ωb)
(collective almost always wrong)
0.4
Modelling matter-light coupling
Telluride, July 2015
23
Toy problem: exact solution
Singular at ∆ = 0
Time evolution —
Fab (t) ∼ exp(−α∆2 t)
Fab
0
1×10-2
0
0.05
0.1
2000
Faa, Fbb
I
Re[Fab]
Solve via Laplace transform. Find Fij (t) = hψ̂i† (t)ψ̂j (t)i
Steady state:
2×10-2
Fab
0.2
Faa
Fbb
0
-0.4
Time
1500
0
0.2
-∆=2(ωb-ωa)
0.4
Always some coherence
1000
I
(individual always wrong)
Fab ∼ Faa , Fbb only at ∆ = 0
500
I
0
-0.4
-0.2
-0.2
Jonathan Keeling
0
0.2
∆=2(ωa-ωb)
(collective almost always wrong)
0.4
Modelling matter-light coupling
Telluride, July 2015
23
Toy problem: Bloch-Redfield theory
Unsecularised Bloch-Redfield theory:
∂t ρ = −i[Ĥ, ρ] +
X
L↓ij ϕ∗i ϕj 2ψ̂j ρψ̂i† − [ρ, ψ̂i† ψ̂j ]+
ij
+
X
L↑ij ϕ∗i ϕj 2ψ̂j† ρψ̂i − [ρ, ψ̂i ψ̂j† ]+ .
ij
Compare to exact solution:
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
24
Toy problem: Bloch-Redfield theory
Unsecularised Bloch-Redfield theory:
∂t ρ = −i[Ĥ, ρ] +
X
L↓ij ϕ∗i ϕj 2ψ̂j ρψ̂i† − [ρ, ψ̂i† ψ̂j ]+
ij
+
L↑ij ϕ∗i ϕj 2ψ̂j† ρψ̂i − [ρ, ψ̂i ψ̂j† ]+ .
X
ij
Compare to exact solution:
0.1
0.01 ∆=0.2
Re[Fab]
Re[Fab]
t=200
0.05
0
0
Exact
Redfield
-0.01
-0.4
-0.2
0
0.2
∆=2(ωa-ωb)
Jonathan Keeling
0.4
0
Modelling matter-light coupling
200
400
Time
600
800
Telluride, July 2015
24
Toy problem: Secularisation
Leads to Fab (t → ∞) = 0. Exact:
Re[Fab]
↑,↓
Secularisation (in eigenbasis of Ĥ): L↑,↓
ij → Lii δij
2×10-2
1×10-2 Fab
0
-0.4
-0.2
0
0.2
-∆=2(ωb-ωa)
0.4
Secularisation often invoked to cure negative eigenvalues of Lij↑,↓ .
→ Non-positivity of density matrix,
→ Unstable (unbounded growth).
Check stability: consider f = (Faa , Fbb , <[Fab ], =[Fab ])
∂t f = −Mf + f0
Eigenvalues of M exist in closed form:
I
Unstable (negative only if dJ(ν)/dν 1
— Markov breakdown)
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
25
Toy problem: Secularisation
Leads to Fab (t → ∞) = 0. Exact:
Re[Fab]
↑,↓
Secularisation (in eigenbasis of Ĥ): L↑,↓
ij → Lii δij
2×10-2
1×10-2 Fab
0
-0.4
-0.2
0
0.2
-∆=2(ωb-ωa)
0.4
Secularisation often invoked to cure negative eigenvalues of Lij↑,↓ .
→ Non-positivity of density matrix,
→ Unstable (unbounded growth).
Check stability: consider f = (Faa , Fbb , <[Fab ], =[Fab ])
∂t f = −Mf + f0
Eigenvalues of M exist in closed form:
I
Unstable (negative only if dJ(ν)/dν 1
— Markov breakdown)
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
25
Toy problem: Secularisation
Leads to Fab (t → ∞) = 0. Exact:
Re[Fab]
↑,↓
Secularisation (in eigenbasis of Ĥ): L↑,↓
ij → Lii δij
2×10-2
1×10-2 Fab
0
-0.4
-0.2
0
0.2
-∆=2(ωb-ωa)
0.4
Secularisation often invoked to cure negative eigenvalues of Lij↑,↓ .
→ Non-positivity of density matrix,
→ Unstable (unbounded growth).
Check stability: consider f = (Faa , Fbb , <[Fab ], =[Fab ])
∂t f = −Mf + f0
Eigenvalues of M exist in closed form:
I
Unstable (negative only if dJ(ν)/dν 1
— Markov breakdown)
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
25
Toy problem: Secularisation
Leads to Fab (t → ∞) = 0. Exact:
Re[Fab]
↑,↓
Secularisation (in eigenbasis of Ĥ): L↑,↓
ij → Lii δij
2×10-2
1×10-2 Fab
0
-0.4
-0.2
0
0.2
-∆=2(ωb-ωa)
0.4
Secularisation often invoked to cure negative eigenvalues of Lij↑,↓ .
→ Non-positivity of density matrix,
→ Unstable (unbounded growth).
Check stability: consider f = (Faa , Fbb , <[Fab ], =[Fab ])
∂t f = −Mf + f0
Eigenvalues of M exist in closed form:
I
Unstable (negative only if dJ(ν)/dν 1
— Markov breakdown)
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
25
Toy problem: Secularisation
Leads to Fab (t → ∞) = 0. Exact:
Re[Fab]
↑,↓
Secularisation (in eigenbasis of Ĥ): L↑,↓
ij → Lii δij
2×10-2
1×10-2 Fab
0
-0.4
-0.2
0
0.2
-∆=2(ωb-ωa)
0.4
Secularisation often invoked to cure negative eigenvalues of Lij↑,↓ .
→ Non-positivity of density matrix,
→ Unstable (unbounded growth).
Check stability: consider f = (Faa , Fbb , <[Fab ], =[Fab ])
∂t f = −Mf + f0
Eigenvalues of M exist in closed form:
I
Unstable (negative only if dJ(ν)/dν 1
— Markov breakdown)
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
25
Toy problem: Secularisation
Leads to Fab (t → ∞) = 0. Exact:
Re[Fab]
↑,↓
Secularisation (in eigenbasis of Ĥ): L↑,↓
ij → Lii δij
2×10-2
1×10-2 Fab
0
-0.4
-0.2
0
0.2
-∆=2(ωb-ωa)
0.4
Secularisation often invoked to cure negative eigenvalues of Lij↑,↓ .
→ Non-positivity of density matrix,
→ Unstable (unbounded growth).
Check stability: consider f = (Faa , Fbb , <[Fab ], =[Fab ])
∂t f = −Mf + f0
Eigenvalues of M exist in closed form:
I
Unstable (negative only if dJ(ν)/dν 1
— Markov breakdown)
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
25
Toy problem: Secularisation
Leads to Fab (t → ∞) = 0. Exact:
Re[Fab]
↑,↓
Secularisation (in eigenbasis of Ĥ): L↑,↓
ij → Lii δij
2×10-2
1×10-2 Fab
0
-0.4
-0.2
0
0.2
-∆=2(ωb-ωa)
0.4
Secularisation often invoked to cure negative eigenvalues of Lij↑,↓ .
→ Non-positivity of density matrix,
→ Unstable (unbounded growth).
Check stability: consider f = (Faa , Fbb , <[Fab ], =[Fab ])
∂t f = −Mf + f0
0.02
I
Unstable (negative only if dJ(ν)/dν 1
— Markov breakdown)
J(ν)
Eigenvalues of M exist in closed form:
0.01
0
0
0.2
0.4
0.6
0.8
1
1.2
1.4
ν
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
25
Toy problem: Secularisation
Leads to Fab (t → ∞) = 0. Exact:
Re[Fab]
↑,↓
Secularisation (in eigenbasis of Ĥ): L↑,↓
ij → Lii δij
2×10-2
1×10-2 Fab
0
-0.4
-0.2
0
0.2
-∆=2(ωb-ωa)
0.4
Secularisation often invoked to cure negative eigenvalues of Lij↑,↓ .
→ Non-positivity of density matrix,
→ Unstable (unbounded growth).
Check stability: consider f = (Faa , Fbb , <[Fab ], =[Fab ])
∂t f = −Mf + f0
0.02
I
Unstable (negative only if dJ(ν)/dν 1
— Markov breakdown)
J(ν)
Eigenvalues of M exist in closed form:
0.01
0
0
0.2
0.4
0.6
0.8
1
1.2
1.4
ν
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
25
Beyond Redfield: Schrödinger picture Bloch Redfield
Is BR the best (time-local) theory we can find?
Hints it is not:
I
I
Eigenvalues of M vs exact sol’n near ∆ = 0.
Sum rule [Salmilehto et al. PRA ’12; Hell et al. PRB ’14]:
“For X̂ s.t. [X̂ , Ĥsystem-bath ] = 0, then ∂t hX̂ i should match closed system.”
I
0
Here, hX̂ i = ϕ2b Faa + ϕ2a Fbb − 2ϕa ϕb Fab
. Fails
Alternate approach:
I
I
I
BR assumes ρ̃(t) is “slow” in
interaction picture
Asymptotically ρ(t) is steady in
Schrödinger picture
Assume instead ρ(t) is slow in
Schrödinger picture
“Schrödinger picture Bloch Redfield.”
I
I
Correct ∆2 expansion
Satisfies sum rule
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
26
Beyond Redfield: Schrödinger picture Bloch Redfield
Is BR the best (time-local) theory we can find?
Hints it is not:
I
I
Eigenvalues of M vs exact sol’n near ∆ = 0.
Sum rule [Salmilehto et al. PRA ’12; Hell et al. PRB ’14]:
“For X̂ s.t. [X̂ , Ĥsystem-bath ] = 0, then ∂t hX̂ i should match closed system.”
I
0
Here, hX̂ i = ϕ2b Faa + ϕ2a Fbb − 2ϕa ϕb Fab
. Fails
Alternate approach:
I
I
I
BR assumes ρ̃(t) is “slow” in
interaction picture
Asymptotically ρ(t) is steady in
Schrödinger picture
Assume instead ρ(t) is slow in
Schrödinger picture
“Schrödinger picture Bloch Redfield.”
I
I
Correct ∆2 expansion
Satisfies sum rule
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
26
Beyond Redfield: Schrödinger picture Bloch Redfield
Is BR the best (time-local) theory we can find?
Hints it is not:
I
I
Eigenvalues of M vs exact sol’n near ∆ = 0.
Sum rule [Salmilehto et al. PRA ’12; Hell et al. PRB ’14]:
“For X̂ s.t. [X̂ , Ĥsystem-bath ] = 0, then ∂t hX̂ i should match closed system.”
I
0
Here, hX̂ i = ϕ2b Faa + ϕ2a Fbb − 2ϕa ϕb Fab
. Fails
Alternate approach:
I
I
I
BR assumes ρ̃(t) is “slow” in
interaction picture
Asymptotically ρ(t) is steady in
Schrödinger picture
Assume instead ρ(t) is slow in
Schrödinger picture
“Schrödinger picture Bloch Redfield.”
I
I
Correct ∆2 expansion
Satisfies sum rule
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
26
Beyond Redfield: Schrödinger picture Bloch Redfield
Is BR the best (time-local) theory we can find?
Hints it is not:
I
I
Eigenvalues of M vs exact sol’n near ∆ = 0.
Sum rule [Salmilehto et al. PRA ’12; Hell et al. PRB ’14]:
“For X̂ s.t. [X̂ , Ĥsystem-bath ] = 0, then ∂t hX̂ i should match closed system.”
I
0
Here, hX̂ i = ϕ2b Faa + ϕ2a Fbb − 2ϕa ϕb Fab
. Fails
Alternate approach:
I
I
I
BR assumes ρ̃(t) is “slow” in
interaction picture
Asymptotically ρ(t) is steady in
Schrödinger picture
Assume instead ρ(t) is slow in
Schrödinger picture
“Schrödinger picture Bloch Redfield.”
I
I
Correct ∆2 expansion
Satisfies sum rule
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
26
Beyond Redfield: Schrödinger picture Bloch Redfield
Is BR the best (time-local) theory we can find?
Hints it is not:
I
I
Eigenvalues of M vs exact sol’n near ∆ = 0.
Sum rule [Salmilehto et al. PRA ’12; Hell et al. PRB ’14]:
“For X̂ s.t. [X̂ , Ĥsystem-bath ] = 0, then ∂t hX̂ i should match closed system.”
I
0
Here, hX̂ i = ϕ2b Faa + ϕ2a Fbb − 2ϕa ϕb Fab
. Fails
Alternate approach:
I
I
I
BR assumes ρ̃(t) is “slow” in
interaction picture
Asymptotically ρ(t) is steady in
Schrödinger picture
Assume instead ρ(t) is slow in
Schrödinger picture
“Schrödinger picture Bloch Redfield.”
I
I
Correct ∆2 expansion
Satisfies sum rule
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
26
Beyond Redfield: Schrödinger picture Bloch Redfield
Is BR the best (time-local) theory we can find?
Hints it is not:
I
I
Eigenvalues of M vs exact sol’n near ∆ = 0.
Sum rule [Salmilehto et al. PRA ’12; Hell et al. PRB ’14]:
“For X̂ s.t. [X̂ , Ĥsystem-bath ] = 0, then ∂t hX̂ i should match closed system.”
I
0
Here, hX̂ i = ϕ2b Faa + ϕ2a Fbb − 2ϕa ϕb Fab
. Fails
Alternate approach:
I
I
I
BR assumes ρ̃(t) is “slow” in
interaction picture
Asymptotically ρ(t) is steady in
Schrödinger picture
Assume instead ρ(t) is slow in
Schrödinger picture
“Schrödinger picture Bloch Redfield.”
I
I
Correct ∆2 expansion
Satisfies sum rule
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
26
Beyond Redfield: Schrödinger picture Bloch Redfield
Is BR the best (time-local) theory we can find?
Hints it is not:
I
I
Eigenvalues of M vs exact sol’n near ∆ = 0.
Sum rule [Salmilehto et al. PRA ’12; Hell et al. PRB ’14]:
“For X̂ s.t. [X̂ , Ĥsystem-bath ] = 0, then ∂t hX̂ i should match closed system.”
Alternate approach:
I
I
I
BR assumes ρ̃(t) is “slow” in
interaction picture
Asymptotically ρ(t) is steady in
Schrödinger picture
Assume instead ρ(t) is slow in
Schrödinger picture
Re[Fab]
0
Here, hX̂ i = ϕ2b Faa + ϕ2a Fbb − 2ϕa ϕb Fab
. Fails
2×10
I
Exact
BR
SpBR
0
“Schrödinger picture Bloch Redfield.”
I
-2
1×10-2
Faa, Fbb
I
0.2
Faa
Fbb
0
-0.4
-0.2
0
0.2
-∆=2(ωb-ωa)
0.4
Correct ∆2 expansion
Satisfies sum rule
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
26
Acknowledgements
G ROUP :
C OLLABORATORS :
F UNDING :
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
27
Summary
Vibrational configuration
0.05
1
1.9
0.1
2
Spectral weight
0.2
1.2
0.05
Spectral weight
g√N=0.3eV
g√N=0.5eV
g√N=0.7eV
0.3
Bare molecule
Spectral weight
0.4
0.04
kBT
[eV]
0.1
0.03
0.8
0.04
kBT
[eV]
0.6
0.4
0.03
2.1
0.2
0
0
1.6
1.8
2
2.2
2.4 2.6
ω [eV]
2.8
3
3.2
0
1.7
1.8
1.9
2
2.1 2.2
ω [eV]
2.3
2.4
2.5
1.7
1.8
1.9
2
2.1 2.2
ω [eV]
2.3
2.4
2.5
[Cwik, Kirton, De Liberato, JK arXiv:1506.08974]
Modelling incoherent emission into non-degenerate modes
0
0.05
0.1
Re[Fab]
Fab
2000
Mode b
1500
J( ν)
Mode a
ν
2×10-2
1×10-2
Exact
BR
SpBR
0
1000
Faa, Fbb
Time
Bath
500
0
-0.4
-0.2
0
0.2
∆=2(ωa-ωb)
0.4
0.2
Faa
Fbb
0
-0.4
-0.2
0
0.2
-∆=2(ωb-ωa)
0.4
[Eastham, Kirton, Cammack, Lovett, JK arXiv:1508.XXXX]
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
28
Extra Slides
Jonathan Keeling
Modelling matter-light coupling
Telluride, July 2015
29
Fly UP