See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/243491864

Quantum superpositions and Schrödinger cat states in quantum optics

Article in American Journal of Physics · October 1997

DOI: 10.1119/1.18698

CITATIONS READS
174 3,230

2 authors:

Christopher Gerry P. L. Knight

Lehman College Imperial College London
222 PUBLICATIONS 7,558 CITATIONS 525 PUBLICATIONS 40,037 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by P. L. Knight on 14 January 2014.

The user has requested enhancement of the downloaded file.

Quantum superpositions and Schrödinger cat states in quantum optics
C. C. Gerrya) and P. L. Knight
Blackett Laboratory, Imperial College London, London SW7 2BZ, England
~Received 14 October 1996; accepted 28 April 1997!
We describe the properties of a quantum system prepared in superpositions of classically
distinguishable states. These states, often called Schrödinger cat states, are of great interest at
present. We discuss how they may be realized in quantum optics using nonlinear interactions and
cavity quantum electrodynamics. We first describe the quantum properties of field states in a cavity,
and demonstrate the interference properties which characterize superposition states and discuss how
fragile they are in dissipative environments. Finally, we review current experimental approaches
which may realize these states. © 1997 American Association of Physics Teachers.

I. INTRODUCTION reversible interaction decoheres the initial superposition into

a statistical mixture which can be characterized by a density
It has often been said that the principle of superposition is operator of the form
at the heart of quantum mechanics. In classical physics, we
do not speak of superpositions of possible states of a system,
rather, we assume that the physical attributes of a system
r̂ mixture5 (i p iu c i &^ c iu , ~2!

objectively exist even if unknown. As Einstein might say, the

where p i is the probability of the system being in the state
moon really is there when nobody looks. But in quantum
mechanics it appears necessary to abandon the notion of an u c i & . Unlike for the state of Eq. ~1!, probability amplitudes
objective local reality.1 Instead, a quantum system is de- do not enter in this case, so interference does not occur. This
scribed by a state vector which may be expanded into a co- is easily seen by writing down the density operator for the
herent superposition of the eigenstates of some observable, pure state of Eq. ~1!:

r̂ u c & 5 u c &^ c u 5 (i (j c i c *j u c i &^ c j u . ~3!

uC&5 (i c iu c i & , ~1!
If  is some observable for which u c i & are not eigenstates,
then
where the coefficients c i are probability amplitudes. The
probability that a measurement of that observable finds the ^  & u c & 5 ^ c u  u c & 5Tr~ r̂ u c & Â) ~4!
system in state u c i & is u c i u 2 . But the state vector of Eq. ~1! is
not merely a reflection of our ignorance of the true state of 5 (i (j c i c *j ^ c j u  u c i & , ~5!
the system before a measurement but rather of its objective
indefiniteness: The system has no objectively definite state whereas
prior to a measurement. The act of measurement ‘‘col-
^  & mixture5Tr~ r̂ mixture ! 5 ( p i ^ c i u  u c i & .
lapses’’ the state vector to one of the eigenstates. The basic
~6!
feature of the superposition principle is that probability am- i
plitudes can interfere: a feature that has no analog in classical
physics. Clearly, the former exhibits interference as the off-diagonal
The above lines conform to the Copenhagen interpretation elements of  contribute to the expectation value, these ele-
~some would say dogma! of quantum mechanics.2 It is cer- ments not being present for the mixture. According to the
tainly the case that such superposition states are not observ- work of Zurek and others,3 an initially pure state, with den-
able in the everyday world of classical physics. We do not sity operator of the form of Eq. ~3!, upon interacting with the
observe macroscopic objects in coherent superposition states environment, evolves in such a way that the off-diagonal
and therefore it may be comforting to conclude that the su- elements rapidly vanish and the density operator evolves into
perposition principle operates only on the microscopic scale, the form of Eq. ~2! for a statistical mixture, i.e., the initial
at a level inaccessible to everyday experience. But this begs pure state decoheres into a mixture. This is the result of
the question as to where one ought to place the border be- tracing over the unobserved states of the environment. The
tween what is classical and what is quantum mechanical. Is it characteristic time for this decoherence to occur depends on
merely a matter of size? And if so, what are the delimiting the size of the system: The more distinctly different the com-
dimensions? The answer does not appear to be straightfor- ponents are in a superposition, the more rapid the decay of
ward. One approach is to realize that macroscopic systems coherence. The density matrix becomes diagonal in a pre-
are generally not closed systems, they interact dissipatively ferred basis determined by the coupling of the system to the
with their environment.3 These interactions are complex, in- environment.
volving entanglement between the system and the environ- For macroscopic objects, the decoherence time is almost
ment, and produce an irreversible evolution. If the system is instantaneous. On the other hand, there might be situations
initially in a coherent superposition state, the dissipative, ir- for which systems, on the classical-quantum boundary, could

964 Am. J. Phys. 65 ~10!, October 1997 © 1997 American Association of Physics Teachers 964
be isolated long enough from dissipative interactions so that coherent states shifted in phase by 180°. Specific examples
superpositions of distinguishable states4 could be realized on are given. In Sec. IV we address the issue of dissipation and
a fairly large scale—perhaps at least on a mesoscopic scale, in Sec. V we give two examples of how such states may be
before decoherence sets in. However, the prospect that such generated. Section VI gives our conclusions.
superpositions could exist on a large scale causes us to recall
the famous thought experiment of Schrödinger formulated in
1935 and commonly known as the Schrödinger cat paradox.5 II. COHERENT STATES
As originally formulated, a cat ~obviously macroscopic! is
placed inside a closed steel box along with a bit of radioac- A. Single-mode quantized fields
tive substance with a decay rate such that after one hour an If we consider a free field of frequency v linearly polar-
atom may have decayed. If the atom decays, a Geiger ized in the x direction in a cavity of length L, then the
counter discharge releases a hammer which shatters a flask Maxwell equations plus the boundary condition that the elec-
containing a poisonous gas, subsequently killing the cat. Ac- tric field vector vanishes at z50 and L leads to the following
cording to the Copenhagen interpretation, the atom and cat expression for the cavity field:11

S D
are in an entangled state of the form
2v2 1/2

1 E x ~ z,t ! 5 q ~ t ! sin kz, ~8!

uc&5 @ uatom not decayed&ucat alive& e 0V
& where the wave vector k and the frequency v are related by
1uatom decayed&ucat dead&# . ~7! k5( v /c)z, V is the volume of the cavity, and e 0 the permit-
tivity of free space. The amplitude of the electric field is
Although, as far as we know, the word entanglement was governed by the time-dependent factor q(t), which has the
used first by Schrödinger to describe states of this sort, the dimensions of length, so that the electric field can be re-
concept certainly appears in the paper of Einstein, Podolsky, garded as a kind of canonical position. The magnetic field
and Rosen ~EPR!,6 the paper that motivated Schrödinger’s H y (z,t) can similarly be expressed as

S DS D
remarks. In any event, upon opening the box, the state vector
collapses to one state or the other in the superposition. e0 2v2 1/2
H y ~ z,t ! 5 q̇ ~ t ! cos kz ~9!
Schrödinger refers to this as a ‘‘quite ridiculous case,’’ in- k e 0V
tending to show the absurdity of the Copenhagen interpreta-
tion. and its amplitude is governed by a kind of canonical momen-
The ‘‘paradox’’ has often been dismissed as having no tum q̇(t).
observable consequences and this is certainly true of the The field energy H can be expressed as the sum of electric
original formulation. A single atom decay requires amplifi- and magnetic field energies in the cavity:
cation, an irreversible process, in order to be detected. How-
ever, with recent developments in technology, the paradox
has become harder to ignore. For example, there has been a
H5 S DE
1
2
dV ~ e 0 E 2x ~ z,t ! 1 m 0 H 2y ~ z,t !! , ~10!

great deal of effort to obtain superpositions of macroscopi- which, on using Eqs. ~8! and ~9!, becomes
cally distinct states of flux in superconducting Josephson
junctions.7 In the field of quantum optics, the subject of this H5 21 ~ p 2 1 v 2 q 2 ! . ~11!
paper, there is likewise considerable effort being expended In other words, the field mode energy is precisely that of a
on generating superpositions of macroscopically ~or mesos- unit mass harmonic oscillator, with the electric and magnetic
copically! distinct states of the quantized electromagnetic fields playing the role of canonical position and momentum.
field,8 and of the vibrational motion of an ion in a Paul trap.9
We wish to make the point here that these kinds of states,
which are referred to as Schrödinger cat states ~or simply B. Quantization of the single-mode field
cats! in the literature, are not the entangled states of the type Field quantization is straightforward once the canonical
of Eq. ~7! as originally discussed by Schrödinger, but rather variables have been identified.11 We simply take the corre-
are coherent superpositions of distinguishable states of a
spondence rule that the variables q and p above are replaced
single system. In most cases, the superpositions contain only
two distinguishable states, these states being coherent states by their operator equivalents q̂,p̂ ~operators will be distin-
of the single-mode field, with a relative phase difference of guished from c numbers by a caret superscript!, satisfying
180°. As is well known, the coherent states are quantum the commutation rule
states of a single-mode field, described as a harmonic oscil- @ q̂,p̂ # 5i\, ~12!
lator, which can contain a large average photon number and
which are as close as possible to classical states, containing so that the electric field mode operator is written as
the noise of the vacuum. Properties of the coherent states
were reviewed in this Journal some time ago by Howard and
Roy.10
Ê x ~ z,t ! 5 S D
2v2
e 0V
1/2
q̂ ~ t ! sin kz, ~13!

In this paper we present a pedagogical review of recent and the magnetic field mode operator as

S DS D
developments in the area of quantum optics regarding the
properties and possible experimental generation schemes of e0 2v2 1/2
Ĥ y ~ z,t ! 5 p̂ ~ t ! cos kz, ~14!
Schrödinger cat states. In Sec. II we begin with a brief re- k e 0V
view of the harmonic oscillator coherent states, paying par- and the energy as
ticular attention to their classical-like nature. In Sec. III we
discuss the Schrödinger cat states given as superpositions of Ĥ5 21 ~ p̂ 2 1 v 2 q̂ 2 ! . ~15!

965 Am. J. Phys., Vol. 65, No. 10, October 1997 C. C. Gerry and P. L. Knight 965
 The mode structure in quantum theory is identical to that The lowest level u0& is defined through
in classical theory, so diffraction and interference phenom-
â u 0 & 50 ~29!
ena will have the same spatial dependence in both theories.
The key points are the lack of commutability of electric and and the other states are given by
magnetic field operators, and the discreteness of the field
energy, with the field state with n excitations having the â u n & 5 An u n21 & , ~30!
energy â u n & 5 An11 u n11 & ,
†
~31!
E n 5 ~ n1 ! \ v .1
2 ~16! u n & 5 ~ n! ! 21/2~ â † ! n u 0 & . ~32!
The state with no excitations is the ground state or vacuum, The annihilation and creation operators decrease or increase
and in quantum theory possesses a residual zero-point energy the excitation of the mode by one quanta, and have only
of (1/2)\ v . off-diagonal matrix elements between number states:

C. Annihilation and creation operators ^ n21 u â u n & 5 An, ~33!

We have characterized the electric and magnetic fields by ^ n11 u â † u n & 5 An11. ~34!
operators p̂ and q̂, which are Hermitian. But it is traditional
to introduce the non-Hermitian ~and therefore nonobserv- E. Quantum fluctuations of a single-mode field
able! annihilation (â) and creation (â † ) operators through
the combinations Although the number state u n & describes a state of pre-
cisely defined energy, it does not describe a state of well-
â5 ~ 2\ v ! 21/2~ v q̂1ip̂ ! , ~17! defined field:
21/2
â 5 ~ 2\ v !
†
~ v q̂2ip̂ ! . ~18! ^ n u Ê x ~ z,t ! u n & 5E 0 sin kz ~ ^ n u â † u n & 1h.c.! 50. ~35!
In terms of these new operators we can write the field energy The mean field is zero, but the mean square is not, for it is, of
operator as course, one component of the field mode energy in Eq. ~10!:
Ĥ5 ~ â † â1 21 ! \ v . ~19! ^ n u Ê 2x ~ z,t ! u n & 5E 20 sin2 kz ~ ^ n u â † â † 1â † â1ââ †
The basic commutation rule—Eq. ~6!—becomes 1ââ u n & ! 52E 20 sin2 kz ~ n1 21 ! . ~36!
@ â,â # 51
†
~20! The vacuum state is not an empty void but represents a
and the electric field operator is field of rms magnitude E 0 sin kz. If we average over the
spatial variation of the single-mode field, we find again that
Ê x ~ z,t ! 5E 0 ~ â1â † ! sin kz, ~21! the ‘‘electric field per photon’’ is (\ v / e 0 V) 1/2. To see how
where the constants have been lumped together as large these ‘‘vacuum fluctuations’’ can be, consider a box of

S D
size 0.01 cm supporting visible-frequency radiation. Then,
\v 1/2
E 05 , ~22! E 0 '1 V/cm, and this corresponds to an intensity of
e 0V 1 mW/cm2. These are not negligible quantities!
which represents the ‘‘electric field per photon.’’ 11
The time dependence of the annihilation operator can be
derived from the Heisenberg equation of motion F. Coherent states—phase and superpositions
dâ i Now, as the expectation value of the field operator is zero
5 @ Ĥ,â # , ~23! for any number state, it is clear that number states, even for
dt \
large n, do not in any way describe a field with classical-like
so that properties. Classical fields have well-defined amplitudes and
â ~ t ! 5â ~ 0 ! exp~ 2i v t ! , ~24! phases. However, a classical-like field can be obtained from
a particular type of superposition, called coherent states, of
â ~ t ! 5â ~ 0 ! exp~ i v t ! .
† †
~25! all the number states. These states, denoted u a & , are given by

S D(
`
D. Number operators and number states 1 an
u a & 5exp 2 u a u 2 un&, ~37!
Although the annihilation and creation operators do not
2 n50 An!
themselves describe physical variables, their ‘‘normal order’’ where a is a complex number. They are defined in such a
product way as to be right eigenstates of the annihilation operator:
n̂5â † â ~26! â u a & 5 a u a & , ~38!
describes the number of excitations n in a single-mode field ^ a u â 5 a * ^ a u .
†
~39!
state u n & ,
As
n̂ u n & 5n u n & . ~27!
^ a u â † â u a & 5 u a u 2 ~40!
The number state u n & is the energy eigenstate of the Hamil-
we may interpret u a u as the average photon number n̄. Fur-
2
tonian Eq. ~19! with eigenvalue E n ,
ther, since a is complex, it can be written in the polar form
Ĥ u n & 5\ v ~ â † â1 21 ! u n & 5E n u n & . ~28! as a 5 An̄ exp(iu), where u is the phase of the coherent state.

966 Am. J. Phys., Vol. 65, No. 10, October 1997 C. C. Gerry and P. L. Knight 966
Thus the coherent state expectation value of the field opera- For example, for a number state u c & 5 u n & we have P( u )
tor is 51/2p , from which it follows that ^ u & 50 and ^ (D u ) 2 &
5 p 2 /3, meaning, as expected, that for a number state there
^ a u Ê x ~ z,t ! u a & 5E 0 sin kz $ a exp~ 2i v t ! 1 a * exp~ i v t ! % are random fluctuations on the phase. But, for a coherent
52E 0 An̄ sin kz cos~ v t1 u ! . ~41! state ua&, we find that

U( `

S DU
2
This has the form of a classical electric field as expected for 1 e in ~ u 2 u 0 ! u a u n 1
P~ u !5 exp 2 u a u 2 . ~52!
a monochromatic wave. The fluctuations in the field at 2p n50 An! 2
t50 are given this time by
For large a we can approximate the Poisson distribution by a
~ ^ Ê 2x ~ z,0 ! & a 2 ^ Ê x ~ z,0 ! & a2 ! 5E 20 ~ sin kz ! 2 ~42! Gaussian:
and by comparing with Eq. ~36! we see that the coherent
states have only the fluctuations of the vacuum. Thus the
coherent states are classical-like in the sense that they yield
u a u 2n
An! S
exp 2
1
2
uau2 '
1
D
A2 p u a u 2
exp 2
~ n2 u a u 2 ! 2
2uau2
. F G ~53!

classical-like field expectation values and have fluctuations Then, converting the sum into an integral, we obtain

S D
identical to that of the vacuum.
1/2
From Eq. ~40!, the probability of finding n photons in the 2n̄
P~ u !' exp@ 22n̄ ~ u 2 u 0 ! 2 # , ~54!
field is the Poisson distribution p
n̄ n which is peaked at u 5 u 0 . Furthermore, the larger n̄ ~hence
p~ n !5 exp~ 2n̄ ! . ~43!
n! the greater the fluctuations in â † â!, the narrower the phase
distribution. In other words, the phase becomes more well
The variance of the photon number is defined for larger field excitation, as we might expect for a
^ ~ Dn̂ ! 2 & a 5 ^ a u ~ â † â ! 2 u a & 2 ~ ^ a u â † â u a & ! 2 5 u a u 2 5n̄, ~44! classical field. In fact, there is a heuristic uncertainty relation
between the number operator fluctuations and the phase fluc-
as expected for a Poisson distribution. tuations:
The coherent states are actually overcomplete and, in gen-
eral, two different coherent states are not orthogonal. For two ^ ~ Dn̂ ! 2 &^ ~ D u ! 2 & > 41 . ~55!
coherent states ua& and ub&, the inner product is Using Eq. ~56!, we obtain ^ u & ' u 0 and ^ (D u ) 2 & '1/4n̄ and

H J( since ^ (Dn̄) 2 & 5n̄ then the above uncertainty relation is ap-
`
1 1 ~ a *b !n
^ a u b & 5exp 2 uau22 ubu2 ~45! proximately satisfied with the equality.
2 2 n50 n! The coherent states may be generated from the vacuum by
the action of the displacement operator14
5exp$ 2 21 u a u 2 2 21 u b u 2 1 a * b % . ~46!
D̂ ~ a ! 5exp~ a â † 2 a * â ! . ~56!
But for u a 2 b u large, the states are nearly orthogonal as
That is,
u ^ a u b & u 2 5exp~ 2 u a 2 b u 2 ! '0. ~47!
u a & 5D̂ ~ a ! u 0 & , ~57!
As said earlier, a classical monochromatic field is well
defined in phase and amplitude. The concept of phase in as may be verified by using the operator theorem
quantum mechanics is problematic, however, and has had a ˆ ˆ ˆ ˆ ˆ
long and contentious history. Here, we shall show that the e A 1B̂ 5e A e B e 2 @ A ,B # /2, ~58!
phase properties of quantum states are well described by
valid if @ Â, @ Â,B̂ ## 5 @ B̂, @ Â,B̂ ## 50. Then
phase distributions obtained from the phase states12 defined
as
`
D̂ ~ a ! 5exp 2 S 2
e eD
u a u 2 a â † a * â
~59!
uu&5 (
n50
e in u u n & , 2p,u<p. ~48!
and, as e a * â u 0 & 50, upon expanding e a â we arrive at Eq.
†

~56!.
These states are not orthogonal but are complete. The phase A coherent state may be generated by a classically oscil-
distribution for an arbitrary field state uc& is given by12,13 lating current.11 Consider a quantum electromagnetic vector
1 potential A(r,t) that is interacting with a classical current
P~ u !5 u^uuc&u2 ~49! described by the vector current density j(r,t). Then accord-
2p
ing to electromagnetic theory the interaction energy V is
and is normalized according to given by

E 2p
p
P ~ u ! d u 51. ~50! V̂ ~ t ! 52 E d 3 rj~ r,t ! –Â~ r,t ! . ~60!

Averages of any function of the phase u, f ( u ), are deter- For a single-mode field, we take A(r,t) to be
mined as
Â~ r,t ! 5 e S \
D 1/2
@ âe ik–r2i v t 1h.c.# , ~61!
^ f ~ u !&5 E 2p
p
f ~ u ! P~ u !du. ~51!
2 v e 0V
where e is the polarization unit vector. Then

967 Am. J. Phys., Vol. 65, No. 10, October 1997 C. C. Gerry and P. L. Knight 967
 V̂ ~ t ! 52 S \
2 v e 0V D 1/2
@ â e–J~ k,t ! e 2i v t 1h.c.# , ~62!
III. SCHRÖDINGER CATS

Whether dead or alive, Schrödinger’s cat is clearly a mac-

where roscopic object. Furthermore, the state of being alive is

E
clearly distinguishable from the state of being dead.17 There-
J~ k,t ! 5 d 3 rj~ r,t ! e ik–r ~63! fore, we require a superposition of two macroscopically dis-
tinguishable quantum states in order to realize a Schrödinger
is the Fourier transform of the current density. The infinitesi- cat.
mal time evolution operator in the interaction picture for the As we have seen, the coherent state ua& can have any am-
time interval t to t1Dt is plitude field since u a u 2 can be arbitrarily large. Let us assume
that u a u 2 is ‘‘macroscopic.’’ As our distinguishable state we
Û ~ t1Dt,t ! 5exp@ 2iV̂ ~ t ! Dt/\ # choose a coherent state of the same amplitude but with a
5exp$ 2Dt @ u ~ t ! â2u * ~ t ! â † # % 5D̂ @ u ~ t ! Dt # , phase shift of 180°: u 2 a & . Both of these are classical-like
states but the superposition of the two has many nonclassical
~64! properties. For definiteness, we take as our Schrödinger cat
where state the superposition

u ~ t ! 52 S \
2 v e 0V D 1/2
e–J~ k,t ! e 2i v t . ~65!
u C & 5N @ u a & 1e i f u 2 a & ],
where the normalization factor is given by
~69!

Given the displacement operator satisfies the property N 5 @ 212 cos f exp~ 22 a 2 !# 21/2 ~70!
D ~ a 2 ! D ~ a 1 ! 5exp@~ a 1 a *
2 2a*
1 a 2 ! /2# D ~ a 1 1 a 2 ! , and where we have assumed a real for simplicity. Note that
~66! for a large, ua& and u 2 a & are essentially orthogonal and N
it is evident that repeated application of the short-time evo- reduces to 1/&. The relative phase f can, of course, be
lution operator on an initial vacuum state leads to a coherent arbitrary, but we shall consider three special cases. For f
state ua&, with 50 we obtain the even coherent states

a5 E 0
t
dt 8 u ~ t 8 ! , ~67!
u C e & 5N e @ u a & 1 u 2 a & ]
and, for f 5 p , the odd states
~71!

apart from an irrelevant overall phase. u C o & 5N o @ u a & 2 u 2 a & ] ~72!

In classical mechanics, the state of a system in phase space
may be represented by a well-defined point if both position and for f 5 p /2, we find
and momentum are accurately known, or by a distribution of
points if all that is known is a range of likelihoods. In quan- 1
u C YS& 5 @ u a & 1i u 2 a & ], ~73!
tum mechanics it is impossible to specify joint values of &
position and momentum: The uncertainty principle shows
that the product of intrinsic uncertainties DpDx>\/2 and which we call the Yurke–Stoler states.18 Finally, we note
nothing we can do will enable us to study dynamics at a finer that all superposition states of the form of Eqs. ~71!–~73! are
level than this minimum phase-space area of \/2. Neverthe- eigenstates of the square of the annihilation operator with a 2
less, there are joint phase-space distribution functions in as the eigenvalue:
quantum mechanics: the best known is that introduced by â 2 u a & 5 a 2 u a & . ~74!
Wigner in 1932 ~Ref. 15!; the price we have to pay is that
these joint distributions are not probability distributions and, The superpositions of the form of Eq. ~69! must be distin-
in particular, need not be positive ~for this reason they are guished from a mere statistical mixture given by the density
known as quasiprobabilities!. The Wigner function W C (q,p) operator
of a system described by the pure-state wave function r̂ 5 21 ~ u a &^ a u 1 u 2 a &^ 2 a u ! . ~75!
C * (q) is given by
~Note that â 2 r̂ 5 a 2 r̂ .! One possible way to make this dis-
W C ~ q,p ! 5
dq 8
2p\
E
C * ~ q1q 8 /2! C ~ q2q 8 /2! e i pq 8 /\ .
tinction would be to consider the respective phase distribu-
tions. Using the phase states and the earlier large field ap-
~68! proximation, we must have for a state of the form of Eq. ~69!
that

S D
It is a bilinear function of the wave function and is the sim-
1/2
plest quasiprobability capable of generating the correct mea- 2n̄
surable marginal distributions u C(q) u 2 and u C(p) u 2 for po- P C~ u ! ' u N u 2 $ exp@ 22n̄ ~ u 2 u 0 ! 2 #
p
sition and momentum. A coherent state has a Gaussian
Wigner function, one which occupies a special place in 1exp@ 22n̄ ~ u 2 u 0 2 p ! 2 # 12 cos~ n̄ p 2 f !
quantum mechanics: It is the only pure state Wigner function
3exp@ 2n̄ ~ u 2 u 0 ! 2 2n̄ ~ u 2 u 0 2 p ! 2 # % , ~76!
which is positive everywhere ~Ref. 16!. Given that coherent
states are often regarded as quasiclassical, a frequently used where n̄5 u a u . For the density operator of Eq. ~75! we have
2

S D
criterion for nonclassicality is the size of any negativities in 1/2
Wigner functions. As we shall see, the interference terms 1 2n̄
P r~ u ! ' $ exp@ 2n̄ ~ u 2 u 0 ! 2 2n̄ ~ u 2 u 0 2 p ! 2 # % .
characteristic of Schrödinger cat superpositions result in sub- 2 p
stantial negativities in Wigner distributions in phase space. ~77!

968 Am. J. Phys., Vol. 65, No. 10, October 1997 C. C. Gerry and P. L. Knight 968
The distributions are both peaked at u 2 u 0 50 and p, as
expected, but the first has an interference term. However, the
interference term is essentially zero as the Gaussians in the
product have very little overlap. Thus the phase distributions,
at high field strengths uau, do not exhibit strong quantum
interferences and hence do not exhibit the nonclassical na-
ture of superpositions. The simplest way, however, to distin-
guish a cat state from a statistical mixture is to look for
interference fringes in the quadrature phase distributions
measured in homodyne detection. If one knows the phase of
the superposed coherent states, it is only necessary to mea-
sure the quadrature phase operator along the axis orthogonal
to the line through the coherent amplitudes of the corre-
sponding states. Cat states would then generate fringes
in the homodyne output, which are absent for statistical
mixtures.18
However, a full signature that nonclassical properties exist
may be obtained from examining the corresponding Wigner
functions. Consider first an even coherent state. The Wigner
function for this is
1
W e ~ x,y ! 5 $ exp@ 22 ~ x2 a ! 2 22y 2 # Fig. 2. As in Fig. 1, but ~a! for the odd coherent state and ~b! the Yurke–
p @ 11exp~ 22 a 2 !#
Stoler coherent state superpositions given by Eqs. ~81! and ~82!.
1exp@ 22 ~ x1 a ! 2 22y 2 # 12
3exp@ 22x 2 22y 2 # cos~ 4y a ! % . ~78!
r̂ 5 21 ~ u a &^ a u 1 u 2 a &^ 2 a u ! . ~79!
The last term on the right-hand side arises as a result of the
quantum interference between ua& and u 2 a & and is respon- We find
sible for the nonclassical properties of the even coherent 1
states. Essentially, the last term can cause the Wigner func- W M ~ x,y ! 5 $ exp@ 22 ~ x2 a ! 2 22y 2 #
tion to be negative in some regions of phase space and thus p
it cannot be interpreted as a probability distribution. For con- 1exp@ 22 ~ x1 a ! 2 22y 2 # % , ~80!
trast, we now consider the Wigner function corresponding to
a statistical mixture of the states ua& and u 2 a & given by the which obviously contains no oscillating term. In Fig. 1 we
density operator plot W e (x,y) for the even coherent state and W M for the
statistical mixture where it is clear that the function for the
even coherent state has ‘‘interference fringes’’ which take on
negative values, whereas that for the mixture exhibits no
such fringes and is always positive. For the odd coherent
state we have
1
W o ~ x,y ! 5 $ exp@ 22 ~ x2 a ! 2 22y 2 #
p @ 11exp~ 22 a 2 !#
1exp@ 22 ~ x1 a ! 2 22y 2 # 22
3exp@ 22x 2 22y 2 # cos~ 4y a ! % , ~81!
which is obviously similar to W e (x,y), but notice that the
fringes are shifted in sign as compared to the even coherent
state. For the Yurke–Stoler state we have
1
W YS~ x,y ! 5 $ exp@ 22 ~ x2 a ! 2 22y 2 # 1exp@ 22 ~ x1 a ! 2
p
22y 2 # 22 exp@ 22x 2 22y 2 # sin~ 4y a ! % . ~82!
These functions are plotted in Fig. 2 and clearly show the
expected interference fringes.
We now examine these states for specific nonclassical
properties. For single-mode field states there are essentially
two possible ways for nonclassical effects to manifest them-
selves. These are ~i! the occurrence of sub-Poissonian statis-
Fig. 1. Phase-space variation of the Wigner function of ~a! the even coher- tics, also called amplitude squeezing, and ~ii! quadrature
ent superposition given by Eq. ~78!, compared with ~b! that of the statistical squeezing. ~There are, in fact, various forms of higher order
mixture of coherent states given by Eq. ~80!. In both cases a 52. squeezing which we shall ignore here, but see Ref. 8.! We

969 Am. J. Phys., Vol. 65, No. 10, October 1997 C. C. Gerry and P. L. Knight 969
begin by looking at the photon number distributions for the 1 a 2 exp~ 22 a 2 !
states u C e & and u C o & . For the former, from Eq. ~71!, we ^ ~ DX̂ 2 &
! 2
5 2 , ~95!
4 11exp~ 22 a 2 !
have
so that reduced fluctuations appear in the X̂ 2 quadrature.
2 exp~ a 2 ! a 2n
p~ n !5 , n even, ~83! ~Note that the uncertainty relation is still satisfied, of course.!
11exp~ 2 a 2 ! n! For the odd coherent state we have
p ~ n ! 50, n odd ~84! 1 a2
^ ~ DX̂ 1 &
! 2
5 1 , ~96!
and for the latter, from Eq. ~72!, 4 12exp~ 22 a 2 !
p ~ n ! 50, n even, ~85! 1 a 2 exp~ 22 a 2 !
^ ~ DX̂ 2 ! & 5 1
2
, ~97!
2 exp~ a 2 ! a 2n 4 12exp~ 22 a 2 !
p~ n !5 , n odd. ~86!
11exp~ 2 a 2 ! n! and thus no squeezing is evident. Note that the even state
shows squeezing but does not have sub-Poissonian statistics
For the Yurke–Stoler state of Eq. ~73!, p(n) is identical to while for the odd state it is vice versa. For the Yurke–Stoler
that of the coherent state, Eq. ~43!. To characterize the width state we have
of the distribution, it is convenient to use Mandel’s
Q-parameter, defined as19 ^ ~ DX̂ 1 ! 2 & 5 41 1 a 2 , ~98!

^ ~ Dn̂ ! 2 & 2 ^ n̂ & ^ ~ DX̂ 2 ! 2 & 5 41 2 a 2 exp~ 24 a 2 ! , ~99!

Q5 . ~87!
^ n̂ & and it is apparent that squeezing appears in X̂ 2 . The maxi-
For 21<Q,0 the distribution is sub-Poissonian. For a co- mum degree of squeezing is less than for the even coherent
herent state and a Yurke–Stoler state, Q50. Now for the state. Finally, for the statistical mixture of Eq. ~79! we find
even coherent state Q has the form ^ ~ DX̂ 1 ! 2 & 5 41 1 a 2 , ~100!
4 a 2 exp~ a 2 ! ^ ~ DX̂ 2 ! 2 & 5 41 , ~101!
Q5 .0 ~88!
11exp~ 24 a 2 !
so that the fluctuations of X̂ 2 are those of a coherent state ~or
for all a. Thus this state is, in fact, super-Poissonian rather the vacuum!, those of X̂ 1 are enhanced over those of a co-
than sub-Poissonian. But for the odd coherent state we have herent state and so no squeezing occurs, as expected.
4 a 2 exp~ a 2 !
Q52 ,0, ~89!
11exp~ 24 a 2 ! IV. DISSIPATIVE ENVIRONMENTS AND THE
DESTRUCTION OF QUANTUM COHERENCE
which is clearly sub-Poissonian. It approaches 21 as a →0.
We next look at squeezing.20 To do this, we introduce the Quantum coherence is an extraordinarily fragile property:
two quadrature operators The slightest coupling of the quantum system of interest to
1 1 the wider world outside leads to the destruction of the super-
X̂ 1 5 ~ â1â † ! , X̂ 2 5 ~ â2â † ! ~90! positions which are so interesting in quantum physics. Any
2 2i system we may be interested in will be connected to the
~essentially dimensionless position and momentum operators outer environment: Radiation fields in high Q cavities will
for a harmonic oscillator!, satisfying the commutation rela- be resistively damped by the resonator walls, trapped laser
tion cooled ions can emit photons spontaneously, and so on. The
environment does more than extract energy from the quan-
i tum system. It also randomly disturbs the phases of the com-
@ X̂ 1 ,X̂ 2 # 5 , ~91!
2 ponents in the superposition. Imagine, for example, that our
quantum system is a harmonic oscillator ~a field mode, or the
from which follows the uncertainty relation motional state of a trapped ion, for instance! and is described
^ ~ DX̂ 1 ! 2 &^ ~ DX̂ 2 ! 2 & > 161 . ~92! by annihilation and creation operators of excitation in the
oscillator, â and â † . Any dissipative coupling to the outer
For the coherent state it is easy to show that world will result in the loss of at least one excitation and this
is described by the application of the annihilation operator.21
^ ~ DX̂ 1 ! 2 & 5 ^ ~ DX̂ 2 ! 2 & 5 41 ~93!
Suppose we were able to create an even Schrödinger cat state
and thus the uncertainty relation becomes an equality with by some means in such an oscillator, so that the state was
equal uncertainties in both variables ~these being the same as u C e & 5N e @ u a & 1 u 2 a & ]. The environmental effect of losing
the vacuum, in fact!. The variance of 41 in the above is known one excitation then results in the state after such a loss given
as the standard quantum limit. It is possible to reduce the by u C o & 5N 8 @ u a & 2 u 2 a & ], where N 8 is a normalization
noise below this limit, and states for which ^ (DX̂ 1 ) 2 & or factor. That is, at some random time, dissipation transforms
^ (DX̂ 2 ) 2 & is less than 41 are called squeezed, and squeezing is even cat states to odd cat states ~and vice versa!.3 The prob-
a distinctly nonclassical effect.20 ability of such a loss in a time interval Dt is
Now, for the even coherent state, we find that Dp5 g Dt ^ C ~ t ! u â † â u C ~ t ! & , ~102!
1 a 2
where g is the rate of energy dissipation. We immediately
^ ~ DX̂ 1 ! 2 & 5 1 , ~94!
4 11exp~ 22 a 2 ! note that the environment destroys the integrity of superpo-

970 Am. J. Phys., Vol. 65, No. 10, October 1997 C. C. Gerry and P. L. Knight 970
sitions at a rate n̄ g , where n̄ is the mean excitation of the
state. In other words, decoherence is much faster than energy
decay for superpositions involving substantial excitation
numbers.3 This is the reason why quantum superpositions of
macroscopically different states are essentially impossible to
realize ~or put in a different way, why the world we see is
essentially classical!. Many papers have been devoted to the
problem of the creation or destruction of macroscopic coher-
ence within dissipative systems ~see, e.g., the references
cited in Ref. 22!. Milburn and Holmes22 solve for the dynam-
ics of Schrödinger cat generation in the presence of dissipa-
tion in an exactly solvable model of light in a strongly non- Fig. 3. Schematic outline of cavity QED setup proposed by Haroche et al.
linear Kerr medium. This nonlinear dissipative model in ~Ref. 24! to generate cat states in a microwave cavity through the inter-
demonstrates not only how dissipation destroys coherence action of Rydberg atoms, selective microwave excitations in two Ramsey
but also makes it difficult to generate such coherence in the zones R 1 and R 2 , and state-selective field-ionization detection of Rydberg
atoms in excited states u e & and ground states u g & . A classical current drives
first place.
the high Q cavity to initiate a coherent state ua&.

V. GENERATION OF CAT STATES

The Yurke–Stoler state is produced by a unitary transfor-
We now turn to the practical problem of how to generate mation ~time evolution! on a coherent state ua&. It is not
Schrödinger cat states of the type previously described. Ac- possible to produce the even and odd coherent states by a
tually a rather large number of schemes have been devel- unitary transformation. However, these states can be pro-
oped, but here, in the context of quantum optics, we shall duced through the interaction of a field with another system,
focus on only two; one that produces the Yurke–Stoler state such as an atom, followed by state selective quantum mea-
and another that can produce the even and the odd coherent surements. Such processes involve the inherently nonclassi-
states. cal concepts of entanglement and the projection postulate
We begin with the Yurke–Stoler states. These states may ~collapse of the state vector! of von Neumann.23 Here we
be produced by the interaction of a Kerr-like nonlinear me- shall discuss a particular example, in the context of cavity
dium with low dissipation with an initial coherent state. The quantum electrodynamics ~CQED!, proposed, and realized
medium is modeled by the interaction Hamiltonian experimentally, by Haroche and collaborators24 in Paris.
The proposed experimental setup is pictured in Fig. 3. The
Ĥ5\ g ~ â † â ! 2 , ~103! cavity is made of superconducting niobium cooled to a tem-
where g is proportional to the nonlinear susceptibility. An perature of about 1 K, has dimensions typically of 1022 m,
initial coherent state evolves into and has a very high Q(108 ). It supports a single mode of the
quantized field of frequency n c 550 GHz ~submillimeter mi-
u C ~ t ! & 5e 2iĤ I t/\ u a & ~104! crowave!. The cavity decay time is, therefore, T cav5Q/ n c

S D(
` '1022 s. The cavity mode is prepared in coherent state ua&
uau2 an 2
5exp 2 e 2i g n t u n & . ~105! by a classical source attached to the cavity. On the left, an
2 n50 An! oven ~along with a number of other manipulating devices!
produces velocity selected circular Rydberg atoms ~atoms
Since n 2 is integer, this state is periodic with period T
2 2 with principal quantum number n>30 and with u m u 5n21!.
52 p / g . If t5 p / g , we find, since e 2i g n t 5e 2i p n 5 In the proposed experiment, three circular Rydberg levels are
(21) , that u C( p / g ) & 5 u 2 a & . But, for t5 p /2g , we find
n
required. We label these levels as u g & , u e & , and u f & for n
2 2 2
that e 2i g n t 5e 2i p n /251 for even n, e 2i p n /252i for odd 550,51,52, respectively. The resonant frequency between
n. Thus we have states u e & and u f & is somewhat detuned from the cavity fre-

U S D L F S DU L S D
quency n c while that between states u g & and u e & is highly
p 1 p p detuned from the cavity ~see Fig. 4!. The atom–cavity field
C 5 exp 2i a 1exp i u2a&
2g & 4 4 interaction is, therefore, a dispersive one described by the
~106! effective interaction Hamiltonian

1
5 exp 2i
&
p
4 S D
@ u a & 1i u 2 a & ], ~107!
Ĥ I 5\ j â † â s 3 ,
where s 3 5 u f &^ f u 2 u e &^ e u , j 52d /D and where d is the
2
~108!

atomic dipole moment and D the detuning. No real transi-

which, apart from an irrelevant overall phase factor, is obvi- tions occur between levels u e & and u f & . However, outside the
ously the Yurke–Stoler state. Generating the Yurke–Stoler cavity on either side—in regions marked R 1 and R 2 , classi-
states will require appreciable nonlinearity, and low dissipa-
cal microwave fields resonantly drive transitions between u g &
tion losses. Silica glass fiber has been much studied in
guided-wave nonlinear optics, due to its very low losses. The and u e & . These regions are referred to as Ramsey zones and
required Kerr-like nonlinearity needed for producing such the transition frequency between u g & and u e & is about 50G
states found in normal SiO2 fibers is, however, extremely Hz. These Ramsey zones are set to produce p /2 pulses
small, hence long lengths would be required. At present
1
these lengths are not realizable without significant losses ue&→ ~ u e & 1 u g & ), ~109!
leading to a decoherence of the superposition. &

971 Am. J. Phys., Vol. 65, No. 10, October 1997 C. C. Gerry and P. L. Knight 971
 Fig. 5. Interfering pathways in the scheme shown in Fig. 3, demonstrating
Fig. 4. Rydberg atom level scheme relevant for the method outlined in Fig.
the potential outcomes after Ramsey microwave interaction in zones R 1 and
3 to generate cat states. The cavity field mode frequency n c is chosen to be
R 2 and cavity interaction in the center of the figure.
detuned by an amount D from resonance with transitions u e & to some level
u f & . The Ramsey zones in Fig. 3 drive transitions between u e & and u g & .

u 2 a & we must apply the second Ramsey field pulse to obtain

U S DL U S DL
1
ug&→ ~ u g & 2 u e & ). ~110! p p
C AF 8
→ C AF ~117!
& j j
Finally, the atom, after passing through R 1 , the cavity, and
R 2 , is selectively ionized, field ionization for circular Ryd- 5 21 @~ u e & 1 u g & ) u 2 a & 1 ~ u g & 2 u e & ) u g & ] ~118!
berg atoms being nearly 100% efficient. Now when an atom
emerges from the oven it is laser excited to state u e & . The 5 21 @ u g & ~ u a & 1 u 2 a & )1 u e & ~ u a & 2 u 2 a & )].
first Ramsey zone creates the atomic superposition ~119!

1 Now, selectively ionizing and detecting u g & projects the cav-

u C A& 5 ~ u e & 1 u g & ). ~111! ity field into state u C e & 5N e ( u a & 1 u 2 a & ), while for u e &
& into u C o & 5N o ( u a & 2 u 2 a & ). The application of the second
Then the initial state of the atom–cavity field system is Ramsey field has caused the original correlated terms of Eq.
u C AF(0) & 5 u C A & u a & . With the atom in the cavity, at time t, ~116! to interfere as is illustrated in Fig. 5.
the system state vector is The procedure discussed above amounts to a measurement
of the parity of the cavity field. To see this, suppose the
u C AF~ t ! & 5e 2iĤ I t/\ u C F~ 0 ! & ~112! cavity contains only a number state. Then preparing the
atomic state exactly as before, the interaction with the cavity
1 field results in
5 ~ u e & u a e i j t & 1 u g & u a & ), ~113!
&
1
where we have used the relation ~ e in j t u e & u 1 u g & ) u n & . ~120!
2i j tâ † â s i j tâ † â
&
e 3 u e & u a & 5e ua&ue& ~114!
ijt The second p /2 pulse produces
5uae &ue& ~115!
~the term containing u g & being unaffected!. We now suppose
1
2 $ u e & @ e in j t 21 # 1 u g & @ e in j t 11 # % u n & , ~121!
that the atomic velocity is selected such that at the time the
atom leaves the cavity j t5 p . ~The required velocity for a which, for j t5 p , becomes u g & u n & , n even, or u g & u n & for n
cavity of width 1022 m is about 100 m/s, the interaction time odd. Thus, having established an even or odd coherent state,
a second atom sent through the cavity should always be de-
being 1024 s!T cav.! Then our state vector has the form
tected in the same state as the first. This amounts to a detec-

U S DL
C AF
j
p
5
1
&
~ u e & u 2 a & 1 u g & u a & ). ~116!
tion of the Schrödinger cat states.
Losses through the cavity walls will, of course, cause the
superpositions to decohere. The decoherence time is given
This state is an entanglement between the cavity field and the by T dec5T cav / u a u 2 . In order that T dec not be too short, this
atom, an EPR–Bohm type of state.6,25 In fact, this is pre- effectively limits the size of u a u 2 to about 10 ~this could be
cisely the sort of state Schrödinger had in mind, in his paper improved for higher Q cavities!. If, for example, an even
of 1935, in the sense that the alive or dead cat is correlated coherent state is produced ~detection of u g & !, decoherence
with the atom not decayed or decayed, respectively. The cor- will cause some of the odd number states to be populated.
relations between atom and field in Eq. ~116! are maintained The decoherence is detectable since it is apparent that the
even after the atom leaves the cavity. Selective ionization second atom could be found in the excited state. The deco-
detecting u g & projects the cavity field into ua&, while detect- herence rate could be detected by time delaying the second
ing u e & gives u 2 a & . To obtain superpositions of ua& and atom. For very large u a u 2 , of course, the decoherence will be

972 Am. J. Phys., Vol. 65, No. 10, October 1997 C. C. Gerry and P. L. Knight 972
essentially instantaneous. But for the intermediate u a u 2 this is
4
It is not enough that we have superpositions of states with ‘‘large’’ quan-
tum numbers of other parameters; the states should be sufficiently distin-
detectable and detection of this decoherence was one of the guishable from each other in some macroscopic way.
key achievements of the Paris experiment.24 5
E. Schrödinger, ‘‘Die gegenwartige Situation in der Quantenmechanik,’’
Naturwissensheften 23, 807–812, 823–828, 844–849 ~1935!; an English
VI. CONCLUSIONS translation can be found in Quantum Theory of Measurement, edited by J.
A. Wheeler and W. H. Zurek ~Princeton U.P. Princeton, 1983!, pp. 152–
The paradigm of Schrödinger’s cat is, like that of Einstein, 6
167.
Podolsky, and Rosen ~EPR!,6 often presented as though it A. Einstein, B. Podolsky, and N. Rosen, ‘‘Can the Quantum-Mechanical
Description of Physical reality be Considered Complete?,’’ Phys. Rev. 47,
were a paradox ~particularly so in some of the popular litera-
777–780 ~1935!.
ture on the subject26!. But Schrödinger’s cat ‘‘paradox’’ is no 7
A. J. Leggett, ‘‘Schrödinger’s Cat and Her Laboratory Cousins,’’ Con-
paradox at all; it is a phenomenon. There are no true para- temp. Phys. 25, 583–598 ~1984!.
doxes in nature. Historically, this alleged paradox has often 8
V. Bužek and P. L. Knight, ‘‘Quantum Interference, Superposition States
been dismissed as having no observable consequences. Such of Light and Nonclassical Effects,’’ Progress in Optics, edited by E. Wolf
a position can no longer be maintained. ~Elsevier, Amsterdam, 1995!, Vol. 34, pp. 1–158, and references therein.
9
In this paper, we have given a pedagogical presentation of C. Monroe, D. M. Meekhof, B. E. King, and D. J. Wineland, ‘‘A ‘Schrö-
dinger Cat’ Superposition State of an Atom,’’ Science 272, 1131–1136
theoretical ideas behind attempts to experimentally realize ~1996!.
Schrödinger’s cat states in quantum optics. The particular 10
S. Howard and S. K. Roy, ‘‘Coherent States of a Harmonic Oscillator,’’
states we have discussed are those formed as superpositions Am. J. Phys. 55, 1109–1117 ~1987!.
of two coherent states, shifted in phase by 180°, and we have 11
R. Loudon, Quantum Theory of Light ~Oxford U.P., Oxford, 1983!, 2nd
discussed two ways by which they may be generated. The ed.; M. Sargent III, M. O. Scully, and W. E. Lamb Jr., Laser Physics
cavity-QED experiments are of particular interest as they ~Addison-Wesley, Reading, MA, 1975!; P. Meystre and M. Sargent III,
Elements of Quantum Optics ~Springer-Verlag, Berlin, 1990!; D. F. Walls
have the potential to elucidate the border between the clas- and G. J. Milburn, Quantum Optics ~Springer-Verlag, Berlin, 1994!; L.
sical and the quantum through the observation of the deco- Mandel and E. Wolf, Optical Coherence and Quantum Optics ~Cambridge
herence of a superposition of at least mesoscopically distin- U.P., Cambridge, 1995!.
12
guishable states. However, we have far from exhausted the L. Susskind and J. Glogower, ‘‘Quantum Mechanical Phase and Time
various types of cat states and methods of their generation Operator,’’ Physics 1, 49–61 ~1964!; S. M. Barnett and D. T. Pegg,
that have been considered in quantum optics. Many of these ‘‘Quantum Optical Phase,’’ J. Mod. Opt. 44, 225–264 ~1997!.
13
See G. S. Agarwal, S. Chaturvedi, K. Tara, and V. Srinivasan, ‘‘Classical
have recently been reviewed by Bužek and Knight.8
Phase Changes in Nonlinear Processes and their Quantum Counter Parts,’’
Finally, we must also mention that catlike states have been Phys. Rev. A 45, 4904–4910 ~1992!.
generated in different contexts. Noel and Stroud27 have gen- 14
R. J. Glauber, ‘‘Coherent and Incoherent States of the Radiation Field,’’
erated radial Schrödinger’s cat states in a Rydberg atom with Phys. Rev. A 131, 2766–2788 ~1963!.
15
average principal quantum number 65. Two radial wave E. P. Wigner, ‘‘On the Quantum Correction for Thermodynamic Equilib-
packets are created that can be separated by as much as 0.4 rium,’’ Phys. Rev. A 10, 749–759 ~1932!; M. Hillery, R. F. O’Connell, M.
mm. More recently, Monroe et al.9 have generated the even O. Scully, and E. P. Wigner, ‘‘Distribution Functions in Physics: Funda-
mentals,’’ Phys. Rep. 106, 121–167 ~1984!.
and odd coherent states as well as Yurke–Stoler states for 16
R. L. Hudson, ‘‘When is the Wigner quasi-probability density non-
the quantized vibrational motion of a single trapped ion. negative,’’ Rep. Math. Phys. 6, 249 ~1974!.
With an average of about nine vibrational quanta, wave 17
Of course, one should not forget that on hearing of the death of President
packets of maximal spatial separation of about 83 nm, sig- Calvin Coolidge Dorothy Parker remarked ‘‘How can they tell?’’
18
nificantly larger than the size of a single-component wave B. Yurke and D. Stoler, ‘‘Generating Quantum Mechanical Superpositions
function in the superposition, about 7 nm, were obtained. of Macroscopically Distinguishable States via Amplitude Dispersion,’’
Phys. Rev. Lett. 57, 13–16 ~1986!.
These superpositions are obviously not macroscopic but do 19
L. Mandel, ‘‘Sub-Poissonian Photon Statistics In Resonance Fluores-
approach being mesoscopic and therefore may be taken as a cence,’’ Opt. Lett. 4, 205–207 ~1979!.
realization of the Schrödinger cat phenomenon. 20
R. Loudon and P. L. Knight, ‘‘Squeezed Light,’’ J. Mod. Opt. 34, 209–
759 ~1987!; A. K. Ekert and P. L. Knight, ‘‘Correlations and Squeezing of
two-mode oscillations,’’ Am. J. Phys. 57, 692–697 ~1989!; D. F. Walls,
ACKNOWLEDGMENTS
‘‘Squeezed States of Light,’’ Nature 306, 141–146 ~1983!.
This work was supported in part by the National Science
21
W. H. Louisell, Quantum Statistical Properties of Radiation ~Wiley, New
York, 1974!.
Foundation under Grant No. PHY94-07194, the UK Engi- 22
G. J. Milburn and D. F. Walls, ‘‘Effect of Dissipation on Interference in
neering and Physical Sciences Research Council, and the Eu- Phase-Space,’’ Phys. Rev. A 38, 1087–1090 ~1988!; B. Garraway and P.
ropean Union. One of us ~PLK! is grateful to the Institute for L. Knight, ‘‘Quantum Superpositions in dissipative environments: Deco-
Theoretical Physics, University of California Santa Barbara herence and deconstruction,’’ in Proceedings of the 44th Scottish Univer-
for their hospitality during the completion of this paper. sities Summer School in Physics, edited by G. L. Oppo, S. M. Barnett, E.
CCG is grateful for a grant from Amherst College enabling Riis, and M. Wilkinson ~IOPP, Bristol, 1996!, pp. 199–238; for an exactly
solvable model of coherence within a dissipative environment see G. J.
him to visit Imperial College where this work was begun. He
Milburn and C. A. Holmes, ‘‘Dissipative quantum and classical Liouville
wishes to thank members of the Optics Section of Imperial mechanics of the anharmonic oscillator,’’ Phys. Rev. Lett. 56, 2237–2240
College for hospitality during his visit. ~1986!.
23
J. von Neumann, Mathematical Foundations of Quantum Mechanics ~Prin-
a!
Present address: Department of Applied Mathematics and Physics, Poly- ceton U.P., Princeton, 1955!.
24
technic University, Six Metrotech Center, Brooklyn, NY 11021. See: L. Davidovich, M. Brune, J. M. Raimond, and S. Haroche, ‘‘Mesos-
1
See, for example, B. d’Espagnat, Veiled Reality ~Addison-Wesley, Read- copic Quantum Coherences in Cavity QED: Preparation and Decoherence
ing, MA, 1995!. Monitoring Schemes,’’ Phys. Rev. A 53, 1295–1309 ~1996!; S. Haroche,
2
An excellent review of this and other interpretations is given by H. C. M. Brune, J. M. Raimond, and L. Davidovich, ‘‘Mesoscopic Quantum
Ohanian, Principles of Quantum Mechanics ~Prentice-Hall, Englewood Coherences in Cavity QED,’’ in Fundamentals of Quantum Optics, edited
Cliffs, NJ, 1990!, Chap. 12. by F. Ehlotzky ~Springer-Verlag, Berlin, 1989!, pp. 223–236; M. Brune,
3
W. H. Zurek, ‘‘Decoherence and the Transition from Quantum to Classi- S. Haroche, J. M. Raimond, L. Davidovich, and N. Zagury, ‘‘Manipulation
cal,’’ Phys. Today 36–44 ~1991!, and references therein. of Photons in a Cavity by Dispersive Atom-Field Coupling: Quantum

973 Am. J. Phys., Vol. 65, No. 10, October 1997 C. C. Gerry and P. L. Knight 973
 Nondemolition Measurements and Generation of ‘Schrödinger Cat’ J. Gribbin, In Search of Schrödinger’s Cat ~Bantam, New York, 1984!.
26

States,’’ Phys. Rev. A 45, 5193–5214 ~1992!; M. Brune, E. Hagley, J. 27

N. W. Noel and C. R. Stroud, Jr., ‘‘A Radial Wave Packet Schrödinger Cat
Dreyel, X. Maitre, A. Maali, C. Wunderlich, J. M. Raimond, and S. State,’’ paper presented at The Seventh Rochester Conference on Coher-
Haroche, ‘‘Observing the progressive decoherence of the meter in a quan- ence and Quantum Optics, edited by J. H. Eberly, L. Mandel and E. Wolf
tum measurement,’’ Phys. Rev. Lett. 77, 4887–4890 ~1996!. ~Plenum, New York, 1996!, pp. 563–564; N. W. Noel and C. R. Stroud,
25
D. Bohm, Quantum Theory ~Prentice-Hall, Englewood Cliffs, NJ, 1951!, Jr., ‘‘Excitation of an Atomic Electron to a Coherent Superposition of
p. 611. Macroscopically Distinct States,’’ Phys. Rev. Lett. 77, 1913–1916 ~1996!.

974 Am. J. Phys., Vol. 65, No. 10, October 1997 C. C. Gerry and P. L. Knight 974

View publication stats