Entanglement and decoherence
experiments in Cavity QED
S.Haroche
Panamerican advanced study institute and workshop,
Ushuaia, October 2000
Atoms crossing one by one a highQ cavity
storing a quantum field (with 0,1...a few
photons) constitute an ideal system to explore
fundamental concepts of quantum mechanics...
superpositions and particle interference,
entanglement,
Quantum measurement (QND)
complementarity,
decoherence and quantumclassical “boundary”
..and to demonstrate basic features of
quantum information processing:
quantum gates,
quantum logic
entanglement engineeringCourse Outline
Lecture 1. Introduction to CQED
The model: 2level atom coupled to harmonic oscillator
Experimental tools: Circular Rydberg atoms,
superconducting cavities,
Basic Quantum Rabi oscillation experiment
Lecture 2. Resonant entanglement in CQED
EPR atomatom pairs
Atomfield entanglement and Quantum non demolition
measurement of a single photon
Quantum phase gates and quantum logic operations
Entanglement engineeringThree particle entanglement
Lecture 3. Dispersive entanglement and decoherence
Principle of dispersive entanglement experiments
Schrédinger cat states of radiation field
Complementarity experiments with cat states
Decoherence of cat states
Towards multimode, multicavity experiments
Concluding remarksLecture 1
Introduction to CQED
Cavity QED (CQED) studies the properties of atoms
radiating near metallic boundaries or in cavities.
Three main periods:
1. Modifications of radiative properties of simple
systems near boundaries (19701980's):
Enhancement and inhibition of spontaneous emission
rates, vacuum frequency shifts in cavities
Observed in microwave cavities with Rydberg atoms,
in optical cavities, in photonic badgaps, in semi
conductors...
2. Strong atomcavity coupling regime (19801990's):
“micromaser” and "microlasers”, qualitative changes
in the atomcavity spectrum: (vacuum Rabi Splitting,
Quantum Rabi oscillation)
Microwave superconducting cavities and superhigh
finesse optical resonators (J. Kimble lectures...)
3. CQED as a tool to study fundamental aspects of
quantum theory, with possible applications (2000):
Engineered entanglement in simple systems involving
single atoms and single photons, decoherence studies
(nicrowave CQED), quantum motion of atoms in
single photon fields (optical CQFD)....COED realizes in the lab
the JaynesCummings model
_]
Field
distribution frye
A single twolevel atom
coupled to a single field mode _ fog
described as a harmonic oscillator. nw
Do Wg = 5
Hamiltonian: H=Hg + He. + Vy¢
SAN
Ha~ Poe eeeel sige)
He= ae (Aa + Aa)=Holda+12
()
—
E(r)
E,
EC
meee
=iE, Gtr) af) a)
ae Vacuum field
“V28, V, (1,5 mV/m)
cav
Venn = BONA fein
Voc/e,n>= ened  g,ntl>
Vase logy lee 23  e,0>
Vie ie, O55 pel g,1>
ground state is uncoupled
Vac , 9>=0
Diagonalization of Vz_ reduces to twolevel problems
in disconnected subspaces {/ g,n+1>,/e,n>}
G2 012...)
fgmt> — ferm>
: 1QV; [g,n+d>
H= (n+1 iw, + 4/2 a ave
iQ¥ nt+1 § fepn>The atomfield coupling
Vqc=Dg EG)
=e ales el @ g >
2 hee
l,n>
1 HQ Vacuiin Rabi splitting
Eigenvalues Ex, n= hoyntt
=
z ;
Figenstates  +,n>= ea fle.n>zi lg, n+1 >}
When atom crosses cavity, Q(r) varies and
atomic eigenstates and eigenvalues follow
adiabatically
Atomcavity forces which can affect atomic
motion, even in vacuum (coupling between
beg0> and >] ¢31>)Quantum Rabi Oscillation
(2
a
oe
e,0><+g,1> :
Reversible photon emission and absorption
More generally :
le,n >> g,nt1>
Temporal evolution, starting frome, > att=0:
 W(t)> =e QGETo e, 1 > snQGETle ,n+7>
Atom and cavity field generally entangled!
Probability to find at time t the atom in level e,
with n photons:
Po (t= oo QVeeT = 4a + cos (Q2Vi=18))
Oscillation directly
related to energy
level splittings
(quantum beats of
time dressed sustem)
IP
0,4Experimental setup
UT gt UWTeav << 2
Strong coupling
condition
Realized in the microwave domain by coupling
circular Rydberg atoms to superconducting
niobium cavities
[1/Tat = 0,08 kHz, 1/Teav = kHz, 2/27 = 50 kIlz
(nvecent stage of exveriment)Circular Rydberg atoms
n=51
Large circular orbit
Strong coupling to microwaves
Long radiative life time (30 ms)
level tunability by Stark effect
1250 A Easy state selective detection
n=52 in 2.5 Viem
52 Circular
Static magnetic field (18 Gatess}
lifts the at/o degeneracy BI
They are prepared in static electric and
magnetic fields by a pulsed sequence of
optical and radiofrequency photon
absorption, in an adiabatic process
By controlling the first laser beam intensity,
0.3 circular atom on average per pulse is prepared
Poissonian atom number distribution:
P(0) = 0.74; P(1) = 0.22; P(2) = 0.045 .....
eful eventsThe
superconducting
FabryPerot
Cavity
Atomic beam
A



PZT stack
Gaussian field mode with 6mm waist
Large field per photon (1.5 mV/m)
Long photon lifetime (1ms) improved
by ring around mirrors
Easy tunability
Very small thermal photon backgroundControl of the atomcavity
interaction time: atom’s velocity
selection by optical pumping
Atomic
beam
Lasers resonant on the 5s j
transition. 3.04 GH
energy diagram indicating the
optical pumping depopulation
and repopulation scheme for the
F=3 level
In green, velocity
distribution before
pumping, in red
velocity
distribution of
atoms pumped in
F=3 level, before
being excited in
Rydberg stateQuantum
Rabi aur re, . &
oscillation in ==> ~
eee Doo
Atoms detected in D by time resolved state selective
ionization PF pd Ionization signal
2 AS
Electric field F(t) used to Starktune atoms ~~
in and out of resonance with C , allowing to ==
change the atomcavity interaction time (active v
velocity selection not necessary in this a
experiment)
WILD
Pe(t) obtained by
averaging over
many realizations
‘ A of experiment.
!
a4 ae a
oe Bt ? ea] Decay of oscillatio
 \ i i a due to various
Peek V relaxation
a
mechanisms (not
included in the
above JC model).
Behaviour at long
times not very wel:
‘Tame (js) understood
¥
Atomic energy: Small coherent field
Quantum Rabi  aileched tei
oscillation — atom
° me a id
in a small = s
coherent field
Rabi
oscillation
with small
average photon
number (less
than 1)
Fourier
transform of
f signal
exhibiting
discrete peaks
at frequencies
 proportional
rommnaan EO) 2 V3 oeThe Quantum Rabi oscillation as
a way to count photons in a box
p(n)
The photon number
distribution in a coherent
field is Poissonian
P.(t)=D,, pin Pe n(t)
a. pm Ja + cos (QyrTp)
FFT Amplitude
per ee ae) 0680100180 012345
Time (us) Frequency (kHz)
n
Signal P,(t) Fourier transform p(n) distributionConcluston of first lecture
Circular Rydberg atoms coupled to microwave
superconducting cavity constitute a good
system to study quantum manifestations of
atomradiation interactions: atoms and
photons can be manipulated "one by one”
The Quantum Rabi oscillation is an ideal test of
the system's ability to “see” quantum effects:
Usually, the Rabi oscillation measures
only the average field amplitude
(average photon number) and does not resolve
the photon number distribution:
 for large” fields, the photon number fluctutation around
average is negligibly small
 ina "bad" cavity (or without cavity), the field
is renewed by the source in a time much shorter
than the Rabi period, washing out all quantum effects
In our experiment, the conditions:
Filo , nel
are realized, ( large atomfield coupling, long cavity
relaxation time, small photon numbers).
We will see in next lectures how these unusual conditions
allow us to investigate other important quantum effectsLecture 2:
Resonant Entanglement in CQED
Resonant atomcavity energy exchange can be used
to produce and study entanglement in CQED
Atomcavity “collision” yielding a nonseparable
final state: a detection of atom’s state projects
field at a distance in correlated state:
EinsteinPodolskyRosen situation (EPR)
Quantum NonDemolition measurement of single
photons
By iteration with many atoms, engineering of
multiparticle entanglement according to well
controlled algorithm
These entanglement experiments can be described
in quantum information language:
Field (0/1 photon) and atom (in level e/g)
seen as quantum bits (qubits) coding binary
informationPreparing and analyzing qubit
state superpositions
An essential ingredient in quantum logic is the
ability to manipulate superpositions of qubit states
How to prepare a qubit superposition state
in a CQED experiment?
ee (jv2y(le > + 18 >)
A classical microwave pulse i; transforms,
before C, state e into a linear combination
of eand g
Detection of qubit superpositions
(1/v2) ( & >
A classical pulse R» transforms back, after C and
before detection, the linear combination with + or 
sign into e or g: subsequent field ionization amounts
to detecting linear suvervositions of statesAnalyzing state superpositions by
Ramsey interferometry
=x 2
25
Leos eve
o
5
The microwave pulses “split” the atomic states (R;)
and recombine them (/:) at two separate times: The
atom follows two “pathes” and the probability to
detect it in e or g exhibits fringes when the phase
difference » between the two pathes is tuned (e.g.by
sweeping the frequency v of X, and R»):
Cavity field affects in
different ways the two
a interfering
pathes:fringes phase
x ie 4 and contrast A are
a modified: this
s provides useful
In latest setup version, a Ha a BOM
: ee: 2 cavity interaction
are inside the cavity ringUseful Quantum Rabi Pulses
Jut>= «2» e, A>
 sinQ \g,1
2
il > Hagley et al,
que = Phys. Rev.lett
v2  Gy 0> Is, 1 79:1 (1997)
Atom  Photon entanglement
Nonlocal correlations between atom and
cavity field (EinsteinPodolskiRosen)
Maitre et al,
fe Phys.Rev. lett.
 e > a \¢,1> Peek
Excitation swapping between
atom and field and "quantum"
mapping
_.  A.Rauschenbeutel et al, Phys.Rev.lett. 83, 5166, 1999
G.Nogues et al, Nature, 400, 239, 1999
g > —e D\3, D—¢,D
Full cycle Rabi oscillation bringing back system
to initial state, with a x pliase shift of its wave functionClassical and Quantum Rabi pulses
to prepare and analyze superpositions
of photon number states (field qubit )
Maitre et al, Phys.Rev.lett. 79, 769 (1997)
The x quantum
e Rabi pulse is used
to "map" an
atomtic coherence
ie (prepared by a
1m a > i
ee Onuniun ee Ry pulse)
classical Rabi into a field
coherence between
0 and 1 photon
states
le D> > > SLD
The field coherence is read out by a second probe atom,
initially in ¢, undergoing timereversed evolution
(x quantum Rabi pulse followed by classical Ro pulse)
pulse pulse
Cavity acts as a quantum
memory storing the qubit
coherence carried by one
I n/2 atom and mapping it later
Quantum classical onto the second atom
Rabi pulse pulse
Bt lA >> Hb b>> bk >Generation of an entangled E.P.R.
atom pair
ns ad Electric field F(t) used to
tune atoms #1 and #2 in
resonance with C fora
a> E(t) determined time t realizing
t w/2 or a Rabi pulse
&2 ey conditions.
le,,0>..¢.> ——
co = g1>}.g2>
Atom #1  Photon entanglement
[eres
1
ae they ga  1g1,€2> }1 O>
Atom  Atom entanglement
(a massive E.P.R.pair)
Atoms entangled in deterministic way without
directly interacting
Cavity acts as a catalyst for entanglementThe EPR atom pair is a “singlet spin state”
If atom #1 is along +x, atom #2 is along y:
The joint probability to find the first atom in +, and
the second in the direction + exhibits a modulation
when the phase @ is swept
eule 3 6 Detector
ansverse spin” detection by applying
with phase difference @ on the two
atoms after they have crossed C
time
Modulations analogous to
Ramsey fringes, but pulses are
applied to different atoms:
EPR pair behaves as a single
quantum systemFull Rabi oscillation: The subtle
2x Rabi pulse
Qt=2m — cos (Ot/2)=1
A single photon changes the sign of the state
probability amplitude, without being destroyed
To perform this phase shift e
measurement, an interferometric
method comparing with the g
phase of a third leveli uncoupled ¥ i Ro
to photon is required... +The principle of single photon
detection by Ramsey interferometry
Atom follows two interfering pathes between R, and
2.
If 1 photon in C, the amplitude of the & path is x phase
shifted and so are the Ramsey fringes
Seiting v to a fringe extremum correlates in principle
perfectly atom state (meter) to photon number.Emitting and reading out a single photon
ee without destroying it
(Source) JET SR,
S_> a/2 Rabi pulse on
a atom #1
(1/v2){1ez,0> +  gy,1>}
Second "meter” atom reads out the photon
number by Ramsey interferometry
SaaS
sie >
Atom #2
(QND
meter) = jurce in e : 0 photon
= = Source in g : 1 photon
oo
Or
oe
& os
g
E04
Frequency v (kHz)Atomphoton quantum gate and QND
measurement of a single photon
“* Qo =
A photonatom phase gate...... :
li>+ Ig>
ain realizes, with a proper choice of pa
QND measurement of a single photon
Aloim is a “meter” which measures
photon number without demolishing itA brief history of
Quantum NonDemolition(QND )
Measurement of light
The QND solution (Braginsky, Caves 1978 }: couple
signal light to probe beam by a nonabsorptive process
in noitlinear mediums
be is phase 
shifted and shift
measured by
comparing toa
reference beam in
interferometer
..but nonlinear index effects require large intensities, so the
method applies only to macroscopic photon number.Our single photon QND measurement uses
a quantum phase gate (QPG) sandwiched
between two single atom rotations
i
For the first time, a single photon is
measured without being destroyed and can
be measured again and again
(G.Nogues et al, Nature, 400, 239 (1999)).
The present scheme can be extended to
measure larger photon numbers
(dispersive instead of resonant coupling
see next lecture). Experiment is in progress.
The QPG can be used to perform various
atomfield entanglement experiments
(A.Rauschenbeutel et al, Phys.Rev.Lett. 83, 5166 (1999)).
Combining the QPG with other atom
photon operations, one can generate
complex multiparticle entangled statesQuantum Knitting : Three particle
entanglement (from EPR to GHZ)
Rauschenbeutel et al, Science, 288, 2024 (2000)
/e1,0 > a tl Pigs >  [81 es> }0>
aa
ore {[e10 >  /gpI>} Transient field
in cavity
ge the, i2,0> —  87 82, 1>}
GreenbergerHorneZeilinger
triplet
for realization with twin
photon beams, see J.W.Pan et
al, Nature 403, 515 (2000)
a tle ere —  81 Zo €3 >} /0>Complex qubit algorithms....
1, Three particle system preparation
* (Aq, A2 and zero/one photon mode)
position
time
2, Longitudinal orteneiaa
etector
time
3. Transverse entanglement ean
tomic states rotated befor Oy 1Transverse three particle entanglement
no atom2
© atom2 ini
®@ atom2 in g
“4 2
phase @
A1A3 EPR signal @unOuln
—@ — A7A3 EPR signal with Ap detected it
—e A7A3 EPR signal with A detected in g
Conclusion of second lecture: CQED provides the tools to
realize “engineered entanglement” of atoms and photons. We
have studied so far resonant processes. We see in next lecture
that nonresonant dispersive processes are also useful...Lecture 3:
Dispersive entanglement and
decoherence in CQED
6a SC Atom and cavity
f off resonant by
= amount ©
No photon absorption or emission...
wey but atomic transition and cavity mode exhibit
frequeiicy shifts
Huge single atom index:
A 15 order of magnitude difference with “ordinary” i
Sign of effect depends upon atomic energy state
; =
= Aw = Atoms
> = position
: =n : ous
a Atom's &
= position &
9 v  Aw
when atom is sent in state superposition, is
the frequency shift + Aw or Aw?
ed Ambiguous quantum
answer leads to
“dispersive” atomcavity
entanglementFrequency shifts in nonresonant
condition of validity:
6>>Q
(if not fulfilled, exact
energy levels obtained by
diagonalizing 2x2
——
Vqc Switched on
Shifts calculated by standard perturbation theory:
dressed atomcavity levels are repelled by their
coupling, so that e and ¢ dressed states
are shifted in opposite directions.
Level shifts in n/t manifold are proportional to 11+1:
this is a light shift effect
Level /e,0> is shifted (in vacuum): this is a
"Lamb shift” effect.
When atom in level g moves across cavity containing n
photons, Q . nt appears as a potential energy
4
for the atomic motionCavity shifts measured by
Ramsey interferometry
Pr(gh= Aq + cos) + AG(n))
cavity light shift
cavity Lambshift
Ag(it) = oS einen )
phase shift per photon
Cavity containing n
photons
Zo00 2000 Bo00 3000 Toooe
¥, @, Gla)
M.Brune etal, P.R.L.72, 3339, 1994.Dispersive cavity shifts used for
unrestricted QND detection of photons
The resonant method described in Lecture 2 is restricted
to 0/1 photon fields (atom absorbs field for n> 1 since
Qvii t is then not equal to 2m)
Dispersive cavity shifts can be used to "pin down” the
photon number in a nondestructive way, in principle
for arbitrary 1 values (in practice for n not too large...)
(Brune et al, P.R.L.65, 976 (1990))
Assume that C contains an unknown photon
distribution with probabilities p(ii):
Ramsey fringe signal: P(g) = 2 p(n) P;,(g)
P(g)
1,3,5... photous
P
Choose phase shift per photon © = x and set
interferometer at fringe center (ideally P(g)=1 if
no field in C)
Send one atom, initially in e through interferometer:
field must contain an
If atom is detected in g: even photon number
: : field must contain an
If atom is detected ine: odd photon number
One atom "qubit” provides information to "pin down”
photon number parity,How to get efficiently more
information on photon number?
(a proposal )
S.Haroche, M.Brune and J.M.Raimond, ].Phys.H France, 2,
659 (1992)
Adjust next € to the value x/2 (e.g. by doubling the
detuning 8)
Pi ide atom has determined photon number is even
Prlg
2, 6, 10...photons
, 8, 12...photons
Set interferometer again at fringe extremum and send
a second atom through setup: its detection ine or g
will "pin down” photon number modulo 4: decide
among the set of values {2, ©, 10...} or {0, 4, 8, 12...}
If first atom has determined photon number is odd,
a similar procedure will determine, with a second
atom, whether photon number belongs to sets
{15,9,...}: 00 (3,711
Caution: no cavity relaxation during whole procedure!Dual frequency shift effects on
atom and cavity field
The dressed atomcavity system energy shifts
correspond to “dual” effects:
an nphoton field shifts atom in level e or g shifts
energy of atom in level<—» the frequency of cavity
eorg mode
How does an atom in level g affect the time evolution
of a coherent field in C?
[a>[g>=2y Cc g, n> ——>
Ly exp (i 2n w40yC,, /&, > =  a exp(i 2? t/46)>  >
G ee oe
Va (interaction picture)
The field is frequency shifted by Aw = {22 /45 and
after an interaction time t, its phase receives a kick
Ao = 222 t/46
phase space 3
representation a = 22 t/46
—@— «Entanglement involving mesoscopic
field states with different phases
(up to 10 photons in cavity)
field phasespace
re me
—
Microwave
source
(s)
field amplitudeA complementarity experiment
The mere fact that
the field contains
information about the
atom’s “path”
destroys the
fe interference fringes!
t R tim
Ramsey fringes recorded for
decreasing mismatches 5 between
atomic transition and cavity mode:
aa As 6 decreases, the
i ar é aa, field components
a eo ae S 4
& re j __ separate,
i Br eho j yielding more and
ate da more information
i =  about the path
os
& oi 4 followed by atom:
the fringe contrast
diminishes
accordingly.A “dual” experiment: field's
phase measures atom’s energy
and atom's coherence phase
measures field energy!
2
i f
é z
220 oe
Z "
: ;
00 «(02 «404 06 O8 qo 02
(radians) @ (radians)
As the phase difference
between the two cat
components increases, the  
atomic energy is better and
better determined and the
fringe contrast A drops:
A=<« exp(ip)/a exp(id)>Decoherence, entanglement with
environment and Complementarity
B+B
Scattered light
In Young experiment, the scattered photon gets
entangled with the particle. Its possible detection “pins
down” the “path” and destroys quantum coherence
elt ay
The "thermal bath” gets entangled to cat.
Information about its fate very quickly leaks into
environment destroying all quantum interferences.
W.Zurek, Physics Today (Oct 1991)_ How long does the mesoscopic
superposition of states survive?
Again a complementarity issue!
Superpositio
ESS
oe pe
Statistical mixture of states
As soon as one photon escapes in environment it
becomes in principle possible, by detecting it, to
distinguish the two parts of the superposition: the
system looses ability to produce interferences
g'impse at the ‘ Quantum
Classical" boundary
(S.Haroche, Physics Today, July 1998)Probing the quantum coherence after
a delay:
4 twoatom correlation experiment
Sr ei
3 Atom# 1
a ee
Oe cat
probes “cat” Brune et al,
Phys.Rev.lett. 77,
4887, 1996
‘TwoAtom Correlation Signal
Other decoherence studies with trapped ion (Boulder)Noun local mesoscopic entanglement
(a proposal)
A challenging
goal:
entangling tw
mesoscopic
cavity fields
separated by a
macroscopic
distance
Marrying the nonlocality's and
Schrodinger cat's weirdnesses
How to do it with dispersive entanglement?
FR pe
Sy at
/2 pulse x pulse 1/2 pulse
prepares atomic coherence exchatiges atamic analyses coherence
levels
[ee SS Se
Procedure should allow us to prepare and analyse
nonlocal "cat states”
L.Davidovich et al, P.R.L. 71, 2360, 1993
oa ad A simpler procedure
um , > with resonant
eee el processes: single
m2 Quantum x Quantum t
Rabi Pulse Rabi Pulse atom prepares
Atom has equal delocalized photon
probabilities to emit ve
photon in each cavity z D+Hx =>Competing qubits in Quantum Optics
1. Atoms in Cavity
1particle QND detection
Pee anata ete aa
aR
(Paris) . . eee
‘omplementarity experimen
ats andDecohe:
ea ee rer ae
ih ching ec]
Quantum gates
Two and four ton entanglement
Boulder,
Innsbruck
= ; f Innsbruck,
Vienna,
Rome,
Geneva
Other experiments, so far without clearcut entanglement,
in NMR, mesoscopic and semiconductor physicsManipulating atoms and photons in a
cavity amounts to realizing in the lab
some of the” gedankenexperiment” of
the early days of Quantum Mechanics.
Bour AND EINSTEIN (1932)
Bour’s
GEDANKEN
PHOTON
Box
Technologically, we have come a long way...
We can now test directly “in vivo” the deepest concepts
of the theory (complementarity, nonseparability,
decoherence and quantum classical boundary”)
How far will we go in Rydberg atom CQED?
10 atom entanglement?
100 photoncat states?
Improvements should be made by working on pulses
imperfections, cavity decay, residual thermal fields
Poisson statistics of atomic beam (atom multiplets
prepared with a very small probability when number of
atoms increases) should be replaced by production of
single atoms on demand (atompistol),
Entanglement studies by marrying
CQED with cold atoms?Experimental setup
Ulot , < oo {/€1g3 >  [Sr es> }0>
ons
aS {/e10 >  [gp1>} Transient fiela
in cavity
ae tex, ig, O> — {81 8x 1>}
GreenbergerHorneZeilinger
triplet
for realization with twin
photon beams, see J.W.Pan et
al, Nature 403, 515 (2000)
tle, 1p, 83 > —  81 RB €3 >F/0>
iD, iP, 99,9, 9,9,8, %1,9, F,8, 9,0, ©9.°,