20.06.2013 Views

Bose-Einstein condensation of excitons and polaritons

Bose-Einstein condensation of excitons and polaritons

Bose-Einstein condensation of excitons and polaritons

SHOW MORE
SHOW LESS

Create successful ePaper yourself

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

Exciton <strong>and</strong> Polariton Condensates<br />

Outline<br />

I. Historical review <strong>of</strong> exciton BEC <strong>and</strong> general issues<br />

II. BEC <strong>of</strong> <strong>excitons</strong> in coupled quantum wells (bilayers)<br />

III. BEC <strong>of</strong> <strong>polaritons</strong> in microcavities<br />

Supported by the U.S. National Science Foundation<br />

under Grant DMR-0706331 <strong>and</strong> the Department <strong>of</strong><br />

Energy under grant DE-FG02-99ER45780.


General concepts <strong>of</strong> <strong>excitons</strong><br />

quasi number conservation. Cf. proton decay <strong>and</strong> H atom BEC<br />

crossover: quasiequilibrium BEC ! nonequilibrium BEC ! immediate decay<br />

incoherent<br />

phonon<br />

emission<br />

photon emission: weak probe <strong>of</strong> instantaneous energy <strong>and</strong> momentum<br />

distribution<br />

conduction states<br />

valence states<br />

“Wannier” limit: electron <strong>and</strong> hole form atom<br />

like positronium that has center <strong>of</strong> mass velocity<br />

Excitonic Rydberg: Excitonic radius:<br />

" = " Ps<br />

! 2<br />

a = !a Ps


Prehistory <strong>of</strong> exciton BEC<br />

early, independent predictions: Moskalenko 1959, Blatt 1962<br />

early mean-field theory: Keldysh <strong>and</strong> Kozlov, 1967<br />

Nozieres <strong>and</strong> Comte, 1982<br />

Early focus was on bulk semiconductors with deeply bound <strong>excitons</strong><br />

(Rydberg ~ 100 meV)<br />

•nearly impossible to exceed Mott density with 3D tightly bound (~ 5Å)<br />

<strong>excitons</strong><br />

•direct gap semiconductors: no electron-hole liquid (EHL) phase<br />

(analogous to liquid hydrogen)<br />

• in principle can have BEC in indirect-gap semiconductors on short<br />

time scales (ca. diamond, Gonokami)


CuCl bi<strong>excitons</strong><br />

Peyghambarian et al., 1983<br />

• sharp spectral peak at low energy<br />

• superradiance? lasing?<br />

• no evidence <strong>of</strong> coherence or <strong>Bose</strong>-<br />

<strong>Einstein</strong> thermal distribution<br />

• no trapping


Cu 2 O <strong>excitons</strong><br />

Snoke et al, 1987<br />

• very good fits to <strong>Bose</strong>-<strong>Einstein</strong><br />

(not condensed) distribution<br />

• later interpreted as due to spatial<br />

integration in 3D light collection,<br />

inhomogeneous spatial pr<strong>of</strong>ile<br />

(no trapping, surface excitation)<br />

• estimates <strong>of</strong> exciton density (very difficult)<br />

need for trap!<br />

2D systems: direct imaging<br />

(<strong>and</strong> also momentum distribution by angle resolution)


Coupled Quantum Wells<br />

+ +<br />

-<br />

-<br />

large spatial separation<br />

gives long exciton<br />

lifetime.<br />

(Lozovik <strong>and</strong> Yudson 1975)<br />

Lozovik <strong>and</strong> Berman,<br />

JETP 84, 1027 (1997):<br />

overall repulsion<br />

when D > 3a B


Trade<strong>of</strong>f:<br />

binding energy decreases<br />

as well width increases<br />

100 Å GaAs wells:<br />

total D = 100 + 40 + 100 = 240 Å<br />

need Ry >> E disorder, k BT<br />

binding energy 4 meV<br />

lifetime ~ 10 µs!<br />

M. Szymanska (Cambridge)


But mobility increases strongly with increasing well width<br />

µ ~ L 6<br />

H. Sakaki et al., APL 51, 1934 (1987)


Measurements <strong>of</strong> exciton drift/diffusion in flat 2D potential<br />

(with disorder)<br />

Spectral imaging system<br />

x<br />

#<br />

CCD<br />

imager<br />

imaging spectrometer<br />

laser<br />

imaging lens


Rise time <strong>of</strong> luminescence vs. x<br />

(lifetime = 6.5 µs)<br />

Intensity (arb. units)<br />

0<br />

60<br />

120<br />

180<br />

240<br />

400<br />

300<br />

200<br />

0 500 1000 1500 2000 2500 3000 3500<br />

t (ns)


strong dipole-dipole repulsion <strong>of</strong> indirect <strong>excitons</strong>:<br />

+ +<br />

H = H0 + !<br />

+ +<br />

Ua1 a2 a3a4 -<br />

blue shift at high density<br />

Voros et al, PRL 2005<br />

-<br />

E


Early claims <strong>of</strong> exciton BEC in QCW’s<br />

Fukuzawa, Kash (IBM group)<br />

PRL 1990 PRL 1991<br />

quasi-Fermi distribution: hard core repulsion in disordered potential<br />

mapping: one spatial location $one energy level


Tim<strong>of</strong>eev et al. (many JETP)<br />

micro-PL from disorder minima


Butov et al.<br />

exciton luminescence at interface <strong>of</strong> electron-hole gases<br />

in-plane coherence length (µm)


strain (arb. units)<br />

Trapping <strong>excitons</strong>:<br />

bending free-st<strong>and</strong>ing sample gives hydrostatic expansion:<br />

3 10 -5<br />

x<br />

0 10 0<br />

x<br />

-3 10 -5<br />

x<br />

-6 10 -5<br />

x<br />

-9 10 -5<br />

x<br />

finite-element analysis <strong>of</strong> stress:<br />

-1.2 10 -4<br />

x<br />

-1 -0.5 0 0.5 1<br />

x (mm)<br />

hydrostatic strain<br />

shear strain<br />

F<br />

fit to experimental exciton line position<br />

using known deformation potentials:


Energy<br />

position


E%<br />

Equilibrium in trap:<br />

blue shift <strong>and</strong> broadening<br />

-- trap is flattened at high density


The “black hole”<br />

substrate luminescence<br />

at laser spot<br />

dark spot appears at stress where lh-exciton crosses hh-exciton.<br />

disappears at low density.<br />

center spot is not completely dark.<br />

T=2K


Temperature series-- spatial images


polarization <strong>of</strong><br />

bright luminescence<br />

low stress<br />

high stress,<br />

with dark spot


General considerations for coupled quantum wells<br />

spin structure: lowest exciton has two J=1 (“bright”) <strong>and</strong> two J=1 (“dark”)<br />

states.<br />

fourfold degeneracy broken by stress.<br />

exchange effects greatly suppressed by low overlap <strong>of</strong> electron-hole<br />

wave functions.<br />

brightness/lifetime affected by lh-hh mixing with stress<br />

Combescot: for degenerate spin states, BEC should always occur in<br />

dark state.<br />

limited available density range:<br />

localization in disorder at low density,<br />

low T.<br />

interactions: flattening <strong>of</strong> trap at<br />

high density.<br />

Effective Exciton<br />

temperature (K)<br />

Bath temperature (K)


interactions: generally a difficult problem for all types <strong>of</strong> <strong>excitons</strong>.<br />

e-e exchange<br />

h-h exchange<br />

e-h exchange<br />

Like Ps, but gap is only ~100X greater than binding energy.<br />

Not a generally solved problem after 50 years!<br />

Correlations play a huge role. mean-field theory differs from<br />

experiment by over an order <strong>of</strong> magnitude.<br />

ratio <strong>of</strong> width to shift


Cavity Polaritons<br />

!<br />

cavity photon:<br />

E = hc kz 2 + k||<br />

quantum well exciton:<br />

E = E gap " # bind +<br />

2 = hc (" /L) 2 + k||<br />

h 2 N 2<br />

2mr(2L) 2 + h2 2<br />

k ||<br />

2m<br />

2


Light effective mass ideal<br />

for <strong>Bose</strong> quantum effects:<br />

(Imamoglu 1996?) !<br />

||<br />

!<br />

LP effective mass ~ 10 -4 m e<br />

r s ~ " dB<br />

n "1/ d ~ h / mk BT<br />

T ~ h 2 2 / d<br />

n<br />

m<br />

Why not use bare cavity photons?<br />

paid for by reduced lifetime ~ 10 ps<br />

!<br />

...photons are (nearly) non-interacting.<br />

Excitons have strong short-range interaction<br />

polariton = “photon dressed with mass <strong>and</strong> hard core repulsion”


Typical wafer<br />

properties<br />

• Wedge in the layer<br />

thickness<br />

• Cavity photon<br />

shifts in energy<br />

due layer<br />

thickness<br />

Reflectivity spectrum<br />

around point <strong>of</strong> strong<br />

coupling<br />

GaAs MBE 70-Å QW’s


Trapping as stress is increased<br />

false color:<br />

luminescence<br />

grayscale:<br />

reflectivity<br />

increasing stress<br />

trap<br />

Balili et al., Appl. Phys. Lett. 88, 031110 (2006).


Do the <strong>polaritons</strong> really move?<br />

Drift <strong>and</strong> trapping <strong>of</strong> <strong>polaritons</strong> in trap<br />

Energy [meV]<br />

Images <strong>of</strong> polariton luminescence<br />

as laser spot is moved<br />

1.608<br />

1.606<br />

1.604<br />

1.602<br />

1.600<br />

40 µm


!<br />

Results at Critical Threshold in Trap<br />

General condition for exchange to be important:<br />

E'<br />

, rs ~ n-1/2 " = h / 2mkBT (in 2D)<br />

& superfluid at low T or high density<br />

trap implies spatial<br />

<strong>condensation</strong><br />

excited thermal particles<br />

x!<br />

log T<br />

normal<br />

log n<br />

superfluid


Four types <strong>of</strong> pumping:<br />

E<br />

k<br />

electron-hole continuum<br />

exciton states<br />

k<br />

||<br />

1) direct resonant (normal incidence)<br />

coherence <strong>of</strong> laser directly mapped<br />

2) “magic angle”: single elastic scatt<br />

populates ground state<br />

coherence <strong>of</strong> laser mapped?<br />

Whitakker: yes. Kavakin: no<br />

3) large angle: many scatterings<br />

need intense laser pulse since<br />

reflected by cavity<br />

4) high excess energy, must emit<br />

phonons


Threshold behavior with incoherent pump:<br />

Pump:<br />

115 meV excess energy<br />

circular polarized<br />

E<br />

k<br />

electron-hole continuum<br />

exciton states<br />

k<br />

Luminescence intensity<br />

at k || =0 vs. pump power


Spatial pr<strong>of</strong>iles <strong>of</strong> polariton luminescence


Spatial pr<strong>of</strong>iles <strong>of</strong> polariton luminescence- creation at side <strong>of</strong> trap


Unstressed-- weakly coupled<br />

Angle-resolved data<br />

“bottleneck”<br />

Weakly stressed Resonant-- strongly coupled


Angle-resolved luminescence spectra<br />

50 µW 400 µW<br />

600 µW 800 µW<br />

!<br />

"x "p (<br />

h


Momentum distribution <strong>of</strong> <strong>polaritons</strong><br />

0.4 mW<br />

0.6 mW<br />

0.8 mW<br />

Energy distribution <strong>of</strong> <strong>polaritons</strong><br />

Maxwell-<br />

Boltzmann fit<br />

Ae -E/k B T


Issues <strong>of</strong> equilibration<br />

Does lack <strong>of</strong> equilibrium destroy the concept <strong>of</strong> a condensate?<br />

N k =<br />

lifetime longer, but not much longer, than collision time<br />

continuous pumping<br />

1<br />

e (Ek )µ)/kBT )1<br />

Not seen in any<br />

polariton experiments<br />

for E >> k BT<br />

Ideal equilibrium <strong>Bose</strong>-<strong>Einstein</strong> distribution<br />

N k<br />

µ = -.001 k B T<br />

µ = -.1 k B T<br />

<strong>Bose</strong>-<strong>Einstein</strong><br />

Maxwell-Boltzmann<br />

E/k B T


Occupation number vs. Energy<br />

Tail is too hot!<br />

N(k)<br />

3 10 6<br />

3 10 6<br />

10 6<br />

10 6<br />

8 10 5<br />

8 10 5<br />

6 10 5<br />

6 10 5<br />

4 10 5<br />

4 10 5<br />

2 10 5<br />

2 10 5<br />

0 0.001 0.002 0.003 0.004<br />

E-E (eV)<br />

min MB<br />

BE


Kinetic simulations <strong>of</strong> polariton equilibration<br />

r<br />

# n(<br />

k1)<br />

# t<br />

=<br />

2%<br />

h<br />

!<br />

r r<br />

k k<br />

2 1'<br />

M<br />

2<br />

r<br />

(| k<br />

1<br />

r<br />

" k<br />

1'<br />

r r r r<br />

|) n(<br />

k ) n(<br />

k )[ 1+<br />

n(<br />

k )][ 1+<br />

n(<br />

k )] $ ( E<br />

1<br />

!"##$%&'(!"#$%#&#)*+#(,&-(.(!"'(/001(2344/56<br />

!"##$%&("%7(8"9"9$:$'()*+#(,&-(.(!#'(3;>"#(&:("?6'()*+#6(,&-6(.(""'(;


Three regimes:<br />

<strong>polaritons</strong> near equilibrium<br />

tail to high energy in “bottleneck” region<br />

<strong>excitons</strong> at lattice T


Issues with low-density data<br />

universal curve<br />

at low density


phonon scattering<br />

totally inadequate<br />

to populate<br />

polariton states.<br />

but pol-pol<br />

scattering should<br />

by density dependent!<br />

Malpeuch <strong>and</strong><br />

coworkers:<br />

ad hoc energy broadening<br />

(associated with<br />

disorder)<br />

alternative:<br />

assume small number<br />

<strong>of</strong> permanent free electrons<br />

Simulated Occupation<br />

100<br />

10<br />

1<br />

0.1<br />

0.01<br />

Cavity lifetime = 5 ps<br />

Lattice Temperature = 20 K<br />

Polariton-phonon scattering only<br />

Polariton-polariton scattering without <strong>Bose</strong> terms <strong>and</strong> full polariton-phonon scattering<br />

Full polariton-polariton scattering <strong>and</strong> full polariton-phonon scattering<br />

0.001<br />

0 2 4 6 8 10 12<br />

E-E min (meV)<br />

V. Hartwell, Ph.D. thesis (2008)


Spontaneous linear polarization<br />

-a type <strong>of</strong> symmetry breaking<br />

k B T<br />

small splitting<br />

<strong>of</strong> ground state<br />

aligned along [110] cystal axis<br />

Cf. F.P. Laussy, I.A. Shelykh, G. Malpuech, <strong>and</strong> A. Kavokin, PRB 73, 035315 (2006),<br />

G. Malpuech et al, Appl. Phys. Lett. 88, 111118 (2006).


Splitting <strong>of</strong> bright<br />

exciton states seen in<br />

microcavity<br />

different linear polarization<br />

due to anisotropic exchange<br />

splitting <strong>of</strong> exciton states


Other recent experiments:<br />

vortices<br />

superfluid wavepacket motion, suppression <strong>of</strong> scattering<br />

spontaneous symmetry breaking (r<strong>and</strong>om polarization)<br />

first- <strong>and</strong> second-order coherence<br />

onset time <strong>of</strong> coherence<br />

BEC in microtraps with discrete states<br />

room temperature <strong>polaritons</strong><br />

“driven” resonant coherence with highly nonlinear effects<br />

...applications in optical communications (nonlinear modulation,<br />

low-threshold lasing, cw OPO, optical spin-Hall effect, etc.)


phase locking via interactions<br />

E<br />

-1.5 -1 -0.5 0 0.5 1 1.5<br />

k<br />

spatial: Deveaud et al.<br />

momentum space: Yamamoto et al.<br />

“pi b<strong>and</strong>” via in-plane periodic pattern


Is there a difference between polariton BEC <strong>and</strong> lasing?<br />

coherent light emitted<br />

• spectral narrowing<br />

• linear polarization<br />

• beamlike emission<br />

(Is current emitted from a superconductor different?<br />

output = probe <strong>of</strong> state <strong>of</strong> matter)<br />

“phase space filling” ! “weak coupling”<br />

lasing<br />

A<br />

conduction b<strong>and</strong><br />

valence b<strong>and</strong>


Weak coupling turns <strong>of</strong>f Rabi splitting<br />

“Photon Lasing”<br />

“BEC”


Two ways to tune through resonance<br />

!=0<br />

!=0


Our results for<br />

resonant, non-trapped<br />

<strong>polaritons</strong><br />

(“magic spot” on sample)<br />

PL intensity<br />

at low density<br />

pump power<br />

at threshold


Results in stress trap<br />

line position<br />

blue triangles: 2.5x threshold<br />

emission follows exciton shift<br />

PL intensity<br />

at low density<br />

pump power<br />

at threshold


Resonant, non-trapped<br />

Intensity vs.<br />

pump power<br />

energy shift (dots)<br />

FWHM (solid line)<br />

lasing


With stress trap<br />

Intensity vs.<br />

pump power<br />

energy shift (dots)<br />

FWHM (solid line)<br />

-two thresholds!<br />

BEC<br />

lasing


Angle-resolved spectra<br />

Stress trap, resonant


Spatially-resolved spectra<br />

laser spot center<br />

Stress trap, resonant


Angle-resolved spectrum<br />

Unstressed, resonant


Summary<br />

1. big advantage <strong>of</strong> 2D structures over 3D bulk<br />

spatial imaging, k-space imaging<br />

tailor for lifetime, effective mass<br />

2. coupled quantum wells: no good evidence for BEC<br />

why not? interactions too strong? dark <strong>excitons</strong>?<br />

3. <strong>polaritons</strong>: good evidence for nonequilibrium BEC<br />

“polariton laser”? terminology battle, but clearly different from<br />

normal lasing


Botao Zhang<br />

Nick Sinclair<br />

Annie Wang<br />

Bryan Nelson<br />

Bridget Bertoni<br />

Vince Hartwell<br />

David Snoke<br />

Chuan Yang<br />

Jeff Wuenschell<br />

Ryan Balili<br />

Zoltan Vörös


Now in print...

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

Saved successfully!

Ooh no, something went wrong!