PHYSICAL REVIEW VOLUME 111, NUMBER 1 JULY 1, 1958

Induced and Spontaneous Emission in a Coherent Field

I. R. SENITZKY
United States Army Signal Engineering Laboratories, Fort Monmouth, New Jersey
(Received March 5, 1958)

The interaction between a coherently oscillating radiation field, such as is generally encountered in micro-
wave experiments, and a number of similar atomic systems coupled to the field through their electric dipole
moments is analyzed for the case of resonance between atomic system and field, with both the field and
molecules treated quantum-mechanically. Expressions are derived for the expectation value of field strength
and the energy, and comparison between the two makes it possible to distinguish between coherent and
incoherent parts of the energy, the former having its counterpart in the expectation value of the field
strength, while the latter has no such counterpart. Terms corresponding to induced and spontaneous emis-
sion are identified, and it is shown that the latter includes both coherent and incoherent components. Special
situations related to masers and to the coherence of spontaneous radiation are discussed. The distinction
between the behavior of correlated and uncorrelated states is examined, and it is shown that both the co-
herent emission from uncorrelated states and the incoherent emission from correlated states may be propor-
tional to the square of the number of molecules. The phase of the field is predictable for the coherent emission
but unpredictable for the incoherent emission.

think of an ammonia beam maser.7 The field is the

I N radio-frequency and microwave spectroscopy, one
generally deals with a coherently oscillating electro-
magnetic field,1 as contrasted with optical spectroscopy,
radiation field in the microwave cavity, and the atomic
systems are the ammonia molecules. The molecular
for instance, where the field is incoherent. In those resonance is that of the inversion spectrum, and the
cases where the behavior of the atomic system under coupling to the field is due to the electric dipole moment
study is the only subject of interest, and the finer of the molecule. However, the analysis is in no way
aspects of the reaction of the atomic system on the field specialized by any references to molecular structure,
are unimportant, the field is usually treated classically, and is presented only in terms of an electric dipole
and considered to be a prescribed perturbation of the moment.
atomic system. This is the method which has been used The questions to be examined are those relating to
most widely,2 and takes proper cognizance of field co- induced emission, spontaneous emission, coherence, and
herence, since it consists of adding to the Hamiltonian correlation effects between molecules. In part I, ex-
of the atomic system a sinusoidally time-varying pectation values for the field strength and field energy
potential. In the cases, however, where the details of are derived, and the physical origin of the terms occur-
behavior of the field itself, as affected by the atomic ring in these expectation values is discussed. In part
system, are of interest, the field must be included as II, we deal with the question of coherence. In part III,
part of the entire system under study, and treated situations associated with masers and with spontaneous
quantum-mechanically. This has been done, to greater radiation from excited groups of molecules are analyzed.
or lesser extent, with respect to such problems as noise in In part IV, the distinction between correlated and un-
masers,3~5 and the coherence of spontaneous emission6; correlated states is studied.
but here, however, there has not been complete recog-
nition of the coherence properties of the field, since the We shall consider an idealized situation so that we
formalism used describes the field either by means of a may deal only with the bare essentials of the problem
smooth intensity function of the frequency3'4,6 or as a and concentrate on the principles involved, rather than
collection of independent photons.6 complicate it with details which do not affect the results
The purpose of the present article is to examine some significantly but make the computation more difficult.
important aspects of the behavior of a coherently The atomic system, to which we refer henceforth as
oscillating radiation field when it is in resonance with a "molecule," is assumed to have only two internal energy
group of similar atomic systems, no coupling existing states with eigenvalues Ei and E2, E2>Ei. The motion
between the systems themselves other than through the of the center of mass of the molecule is assumed to be
field. As a specific illustration of such a system, we may classical. The radiation field is contained in a resonant
cavity which is lossless. We consider only one mode of
1
By coherent oscillation we mean a sinusoidal oscillation with the cavity, for which we assume a condition of exact
well-defined phase. resonance with the molecule, namely, co= (Ei—E-^/fi.
2
See, for instance, N. F. Ramsey, Molecular Beams (Oxford
University Press, New York, 1956). The field is described in the usual manner by
3
M. W. P. Strandberg, Phys. Rev. 106, 617 (1957).
4
R. V. Pound, Ann. Phys. 1, 24 (1957). E=-4*dP, H=VXA,
5
Shimode, Takahasi, and Townes, J. Phys. Soc. Japan 12, 686 (1)
(1957). A=QWn(r), P=P(*)u(r),
6 7
R. H. Dicke, Phys. Rev. 93, 99 (1954). Gordon, Zeiger, and Townes, Phys. Rev. 99, 1264 (1955).
4 I. R. SENITZKY

where Q and P are the quantum-mechanical variables where

of the radiation field, obeying the commutation rule P«»(;) = p(0) cosco/- (CO/4TTC2)(2(0) sinco/,
;
[_QyP~]~i h. The initial state of the field is described by <3(0)(0 = e(0) cosco/+(47rc2/co)P(0) sinco/.
the wave function 8 (in the P representation)
We see that P^(i) and Q(0)(i) describe the behavior of
/4c2 \ * T 4 ^ / Eo \ l the field in the absence of an interaction with the mole-
(P ex iP2 F (2) cules. For future reference, we note the expectation
'°M 7~ ) P - — ( )
values of P ( 0 ) (/), P^2(t), as well as Q™(t) and Qm(t),
\hoo/ L ho. V 47rc / J 8
I t is the choice of state of the field which determines which are obtained from Eq. (2) :
whether one is describing a coherently oscillating field9
(PM)=-(EQ/4nc)sm!ut,
(in the classical sense), and this is an essential aspect
of the present analysis. That such is the case for the
(6)
state described by Eq. (2) is shown in detail in refer- (QW)= (EQC/W) coscat,
ence 8. The molecules are described initially by the
(Q<n*)=(Q<ny+2ic(?h/o.
state 10 (in the energy representation of the free molecule)
The summations on the right side of Eqs. (5) are due
to the coupling between the molecules and field. We
^ = 1 1 [«i(w)^i(w)+a2(w)^2(w)], (3)
assume that the dipole moment of the molecule is
sufficiently small so that it may be regarded as a small
where <pi(m) and <p2(m) are the two energy states of quantity of first order, and expand our dynamical
the with, free molecule, and variables in powers of the coupling constant (which is
2 2 contained in y).
\ai(m) | + \a2(m)\ =l. (3a)
Setting
The electric dipole moment of the mih molecule is P(t)=pw(t)+pv(t)+pw(i)+-.,
specified by ym which, in the absence of the electro- and using similar expansions for Q(t) and y(t), we have,
magnetic field, has no diagonal matrix elements (this from Eq. (5),
means that the molecule has no permanent dipole
moment) but only off-diagonal elements. The Hamil- QM(t) = 4ncZ f <#i Y m ^ ^ C O - u C O cosw(/-*i), (7)
tonian for the combined system of molecules and m J0
field is
N CO2
p(»>(*) = - - £ f « i Y « ( w - 1 ) W - « ( r m ) s i n « ( / - / i ) . (8)
H=Z Hm-i: Y«- E(r w )+27rc 2 P 2 + Q\ (4) C m J0
2
m=l m 8TC
We make a further simplification for the purpose of
where Hm is the Hamiltonian and rm the position of the disencumbering our analysis of complicating features
rath molecule. Hm refers only to the internal energy of which have no bearing on the essence of the problem.
the molecule, and not to the kinetic energy of the We will consider u(r) to be constant in the region of
center of mass, which we ignore. The Heisenberg equa- the cavity where the molecules are located. The com-
tions of motion for the field variables are ponent of Ym along the direction of u will be designated
P= - (a>2/4:TC2)Q,
by Ym, with matrix elements for Ym(0)(0) given by 7^12
= Yw2i*=7;m. We chose the phases of <pi(m) and <p2(m),
Q=4TC2P+ 4:TTC EmTm • u (rTO). respectively, so as to make ym real.
The expectation values of Qa)(t) and Pa)(t) may be
These may also be expressed as integral equations which
calculated immediately. Since the formal equation of
contain the initial values of P and Q explicitly:
motion for ym{t) is
f*Y«(0 = [Ym(0,#]= [7*(0,ff» ( 0 1 (9)
(h)
mJ 0 the off-diagonal terms of ym(t) have the time de-
•u(r m ) cosw(t—h), pendence of exp(=F^co0, respectively, so that
(5)
(o (Ym(0) (t)) = a^ (ra) a2 (ra) ye~™ l+ ax (ra) a2* (ra) ye™'
p(0=P >W--zf dhrUh) = 21a\{m)a2{m) \y cos (coH-#m), (10)
C m J r\
•u(r m ) sinco(^— h),
8
where 6m is the difference in phase between a2{m) and
1 . R. Senitzky, Phys. Rev. 95, 904 (1954); 98, 875 (1955). #i(ra). We thus obtain
9
A quantum-mechanical system may also be described by a
mixture of states, in which the phase of the superposition con- (Pd)(t))c*- (co/c)EmI a1(m)a2(m) \ yut
stants is random. The choice of a pure state rather than a mixture
is itself a step in the direction of coherence. Xsin(«H-ft»), (11)
10
This is an uncorrelated state. Correlated states will be dis- (1)
cussed in part IV. (<3 (/))^47rc Emki(m)a 2 (m) \ yut cos(«H-0»),
 EMISSION IN COHERENT FIELD 5

where, for simplicity, we have dropped terms in of""1 and

compared to t. The individual terms of the summation P<2) (/) = ^ 2 I dh p(0) (^)
are the contributions of the individual molecules to
the field. Although the expectation values of P and Q h J0
are, strictly speaking, quantum-mechanical averages
over an ensemble of similar systems, we may regard X W~h) cosco(t-h) sino>(*-0 E / « . (14)
them, in this instance, as an average over similar mole- L 2c0 Jm
1
cules, since only molecular parameters are involved For Oar , this becomes
on the right side of Eqs. (11). We have here, in £o(0
the first order, spontaneous emission (referring now (15)
P ( 2 , (0= ^WI^,
only to the expectation value of the field; the energy
will be discussed later), each molecule acting (on the 2hio ™
average) as a (classical) oscillator of well defined phase,
0m, and amplitude which is proportional to | ax (w)a 2 (m) \. where 8<>(t) = 2ww2u2yH2. Similarly, we have
We see that this amplitude is a maximum when | a\ (m) \ QW(t)c
S2hoi -Q(0)(OZ/m
0(t)
— | a2(m) | = 2~* (that is, when the molecule is in a state
which is a superposition of equal amounts of both We recognize immediately that this is induced radia-
energy states), and that the amplitude vanishes when tion proportional to the zero-order field. For the ex-
the molecule is completely in either one of the energy pectation values, we have
states.
In order to evaluate P ( 2 ) , we need an expression for < 7 m ( 1 ) ( 0 ) = ^ o ^ - 1 7 2 C k 2 ( w ) | 2 - \ai(m) | 2 ] coscu/,
7 m ( 1 ) in terms of the zero-order (uncoupled) operators. E0 So(t)
From Eq. (9) we have (p(2)^))^_
87TC ko)
iflim^ (/) = [7m(1) W,tf™ (0) ]+[7m (0) (t),Hj» (/)]. (12)
Xsinco^ECI^2(w)| kiWI2]. (16)
a)
An expression for Hm can be obtained immediately
from the equation of motion for Hm, We see that (on the average) the induced radiation of
each molecule is in phase (emission) or 180° out of
ihHm(t) = — E[Hmjym~], phase (absorption) with respect to the inducing field,
depending upon whether the molecule is mostly in the
4:7TC f t upper energy state or lower energy state.
Hm(t)=Hm(p)+—« <fci[ff*(0,7«(*i)}P('i), I t is worth noting that if £ o = 0 , that is, if there is no
ih Jo
oscillating field initially in the cavity, and if the mole-
cule enters the cavity in either the upper or lower
energy state (but not in a superposition of both), then
ih Jn the expectation value of the field remains zero. We have
shown this to be true up to second order, but it can be
We see that FIma) contains only off-diagonal matrix shown by a general theorem to be true for all orders, as
elements (with reference to the molecule), so that the follows: consider the Hamiltonian of Eq. (4). I t is
second term on the right side of Eq. (12) contains only invariant under a unitary transformation which changes
diagonal elements. On the other hand, the first term the sign both of E and y. A state of the system for which
on the right side of Eq. (12) contains only off-diagonal each molecule is in a definite energy state must remain
elements, so that the equation for 7mi2(1) is the same as unchanged except for multiplication by a phase factor
(which has no physical significance). The expectation
for 7mi2(0), namely 7mi2(1)= — ^7mi2 (1) . Since, however,
value of the field should be invariant under the above
7m(1)(0) = 0, we must have 7mi2(1) = 0. Similarly, 7 w2 i (1) transformation. I t must therefore be zero.
= 0. The first term on the right side of Eq. (12) there-
We shall now calculate the expectation value for the
fore vanishes. Noting that energy of the field. This will give us information not
only about the energy itself, but also about the co-
[Ym (0) «,[Ym (0) (/i),#m (0) ]]=2fay*I m cos«(/-*0, herence of the field. We have
(i7fieid)=27rc2(P2)+(co2/87rc2)(e2).
where 7 m = f ~ . J for the mth molecule and the unit
Our calculation will include terms only up to second
matrix for the other molecules, we obtain order. Designating the individual terms in the sum of
Eq. (8) by Pm(n), we can write
STTC P^ __ p (0) 2_|_ 2P (0)p (1) _|_ 2P <0)P <2>
7™ ( 1 ) » = uy*I, f dh PM(h)$inu(t-h) (13)
h ~0 +Z^P™(1)iV(1). (17)
6 I. R. SENITZKY

We need not symmetrize our products since P ( 0 ) , P ( 1 ) , The physical meaning of the various terms is as follows:
and P ( 2 ) all commute. 11 P ( 0 ) 2 occurs in both the first The first term is, of course, the energy of the initial
and third terms on the right side of Eq. (17) [see Eq. field. The second term is induced radiation, and is, in
(IS)], and its expectation value is given in Eq. (6). the case of each molecule, emission or absorption, de-
The expectation value of pwpw is just the product of pending on whether the molecule is mainly in the higher
the expectation values of P ( 0 ) and P ( 1 ) , which are given energy state or lower energy state. The third term is the
by Eqs. (11) and (6). To obtain the expectation value interaction between the (initial) oscillation of the mole-
of the last term in Eq. (17), we must exercise a little cule and the field which exists in the cavity. The fourth
care. For m^m\ ( P m ( 1 ) P ^ a ) ) = ( P m ( 1 ) ) ( P ^ ( 1 ) ) . For and fifth terms are spontaneous emission, and the last
m==m', however, we must go back to Eq. (9) and note term is the zero-point energy of the field. Much more
that can be said about these terms, especially for special
initial situations which are of interest, and we will
<W 0 K/i)W 0 ) fe)> come back to a discussion of these terms after we have
= f t | ai(w) | V ^ 2 - ' 1 ^ | a2(m) | V - * " " * - ^ ] , discussed the question of coherence.
so that We write down, for completeness, the entire expres-
u2 I /•« I2 sion for the expectation value of the electric field
(1)2 2 2
(Pm )=—u y \ I ^ ! ^ ^ s i n c o ( / - / i ) strength (up to second order), by combining Eqs. (1),
c2 I ^o I (11), and (16):
a>2 1/ t 1 \
=—u 2 y 2 -[ t2 sinco/ coscoH— sin2co/ ) (E) = Eou smoot
2 2
c 4\ « co /
S0(t)
So(t) -\-EoU- E C102(tri) |2— | a\{m) \2~] sinco/
1 2flO) m
££ , Oar .
STTC2
+4^ro)U2ty E I ai(m)a2(m) | sin(a^+0 m ). (19)
m
Combining the above results, we obtain for the part of
the energy due to the electric field
As has already been indicated, the first term is the
2TTC2(P2) initial cavity field, the second is an induced field, and
the third is a spontaneously emitted field.
£o | «o(0 , , 1
= — shrW 1 + E L\(h(m) \2- \oi(m) | 2 ]
ST I hco m J II.
By coherent oscillation, as mentioned previously, we
+Eouyut E I ai{m)a<iim) | [cos0m mean a sinusoidal oscillation with a well-defined phase.
By coherent radiation from a molecule, we mean radia-
tion for which the field oscillates coherently, with the
-cos(2«/+fl m )D+4«o(0{E I «*(») I2 phase being determined by the state of the molecule or
by the (coherent) field to which the molecule is coupled.
m
If the state of the molecule and/or coupled field do not
determine the phase of the radiation, then the radiation
+ E l a i ^ ^ W H ^ ^ O ^ ^ O l E c O S ^ m —V )
is incoherent and the phase is a random variable. Obvi-
— cos(2ut+Qm+dm')2}+jhoo. ously, only coherent radiation will contribute to an
Calculating (Q2) in the same manner as we did (P 2 ), and average of the radiation field over many similar sys-
adding the magnetic to the electric energy, we find tems, but the incoherent radiation will show up in an
the double-frequency terms cancel, and we have average of the square of the field or (essentially) the
Eo2 E02So(t) energy. Thus, those terms which appear in the expecta-
<#fieid>=—+ EC\o2(m) | 2 - |a x (m) | 2 ] tion value of the field energy but have no counterpart
ST ST hco m
in the expectation value of the field strength denote the
-{-EQOjyut E I ai(m)a2(m) \ cos#w incoherent radiation. The other terms denote the co-
m herent radiation. Another way of saying this is that the
dispersion in field ((E2)—(E)2) is a measure of the
+ «o(0E|a»(«)l2 incoherence. However, it is more convenient to deal
m with the energy, and we will examine the terms in
Eq. (18) from the point of view of coherence.
+ So (t) E I 0 1 M ^2 (w) 1101 (wO 02 ( ^ 0 | We see immediately that the first term in the ex-
X c o s ( ( 9 m - M + J ^ . (18) pression for the energy corresponds to the square of the
np(o) contains only field variables and P (1) contains only- first term in (E)(sin2o^ combines with cos2a>/ from the
molecular variables. magnetic part of the energy), and is just the classical
 EMISSION IN COHERENT FIELD 7
expression for the energy of the initial field, or the "incoherent." This is a classical type of incoherence,
classical energy of the field in absence of the molecules. such as would result if we had an assembly of classical
It is, of course, coherent energy. The second and third oscillators with random initial phase. As long as we are
terms in (Haeid) similarly correspond to terms in the aware of this possibility and take it into account when
square of (E), namely, the cross-products, and therefore necessary, we will not be misled by our definitions, and
also denote coherent energy. So far, the only term in will continue to use them as defined previously.
the square of (E) (up to second order) which has not
been accounted for in Eq. (18) is the square of the III.
last term in (E), for which the corresponding term in
the energy would be We return now to Eq. (18) to examine certain situa-
tions which are of interest. Suppose, for simplicity, that
£o(0 Z \ai(m)a2(m)\\ai(m,)a2(mf)\ cos(0m—0m>) \ai(m)a2(m) | = \ai{m')a2{mr) | ; that is, the amplitudes
m,m' of oscillation of all the molecules are equal. Now con-
sider the case where half the molecules have one phase
= S0(t){ Z \a1(m)a2(m)\\a1(m')a2(m,)\ cos(0m—6m')
(0 m =0, m = 1, • • •, %N) and the other half has opposite
+ E [ k ( m ) | 2 - k M I 4 ] } , (20) phase (0 m =d+Tr,m~ %N+1, • • *, N). Then the coherent
m spontaneous radiation [Eq. (20) for the energy or last
term of Eq. (19) for the field strength] vanishes. On
where use has been made of Eq. (3a). This represents part the other hand, if M molecules have one phase and the
of the fourth term and the entire fifth term in (Z/fieid), remainder have opposite phase, then the coherent spon-
and thus is the coherent spontaneous emission energy. taneous emission energy is proportional to (2M—N)2,
The remainder of the fourth term is and, of course, for the case in which all molecules are
in phase, the energy is proportional to N2. This last
So(t)Zm\a2(m)\\ (21)
situation, for the case where | ai(m)a2(m) | =-|, gives the
and represents the incoherent spontaneous emission maximum possible coherent emission. I t should be
energy. The last term in Eq. (18) which, as has been noted, however, that the incoherent spontaneous emis-
noted previously, is the zero-point energy of the field, sion is entirely independent of the phase of the molecule,
is independent of the molecules, and, according to our and even in the case where the coherent spontaneous
definition, would be classified as incoherent energy. emission vanishes, there is incoherent spontaneous emis-
However, this is energy which cannot be withdrawn sion, the energy being proportional to the total number
from the field, does not consist of oscillation or fluctua- of molecules, as is evident from Eq. (21).
tion in time, cannot be detected in a classical-type A situation which is of interest in connection with
experiment, and will be neglected for purposes of the spontaneous emission is one in which the molecules are
present article. In sum, the incoherent part of the subjected to a strong field for a short time, after which
emitted radiation is given by (21); the remainder of the field is removed and the molecules are allowed to
the radiation energy, whether induced or spontaneous, radiate by themselves. To observe the effect of the'field
is coherent, the former having its phase determined by on the molecules, we need consider only a single mole-
the inducing field, and the latter having its phase deter- cule, since the interaction between molecules does not
mined by the initial state of the molecule. enter in the lowest order necessary to study this effect.
We can gain further insight into the spontaneous For simplicity, we assume that the molecule is in the
emission, which will sharpen the differentiation between ground state. From Eq. (13) we obtain
the coherent and incoherent parts, and make our
definition more flexible. We note that the first-order (7 ( 1 ) )= (Eou/nu)y2(a)t cosut—sinoot),
part of (E), and the coherent part of the spontaneous ^ (Eou/h) y2t cosco*, O a r 1 .
emission energy, (which corresponds to the square of
the former), are the field and energy, respectively, of a The expectation value of the energy absorbed by the
classical radiation field coupled to a classically oscillat- molecule from the field is just the negative of the second
ing dipole moment, (y(t)). The incoherent part of the term in Eq. (18), so that
spontaneous emission energy, on the other hand, is due Eo2 S0(t)
to the quantum-mechanical properties of the molecule (Hm) = E1+ .
and field. Thus, the coherent spontaneous emission may ST ho)
be regarded as of classical origin, while the incoherent When the field is removed, at time r, say, we have a
spontaneous emission is of quantum-mechanical origin. new set of initial conditions for the molecule, which is
I t may happen that we do not know, or cannot now in the state alT<p1+a2T<P2. Since the expectation
control, the initial states of the molecules, and are value of the dipole moment and energy of the molecule
faced with a random distribution of initial states. This suffer no discontinuity at / = r, we have
means that the "coherent'' radiation of each molecule
has random phase and should more properly be called 21 alra2r 17 cos(cor+0T) = (E0u/n)y2r coscor,
8 I. R. SENITZKY

and Equation (24) is the corresponding statement for our

£o2 5o(r) classical type state which cannot be described in terms
\alT\2E1+\a2T\>E2 = E1+ . of an integral number of photons.
Expression (23) for the induced emission does not
Together with the equation
tell the whole story, however. We see from Eq. (5) or
|01r|2+|02r|2=l, Eq. (7) that the field radiated by the molecule is due to
the oscillating dipole moment of the molecule (as is,
we have three equations for the determination of
of course, physically evident). The only difference,
| air I, I #2r |, and 6r. The results are 12
essentially, between induced emission and coherent
I a2T \=E0uyr/2fi, 0 T =O, spontaneous emission is the manner in which the oscilla-
tions of the dipole moment are produced. For coherent
\alT\ = l-EQ So(r)/167r(ficc)2.
2 (
' spontaneous emission, we assume implicitly a given
We see that at time r we have an induced oscillating oscillation of the dipole moment in the initial conditions.
dipole moment with a well-defined phase determined by In induced emission, the oscillations of the dipole
the phase of the inducing field. After the field is re- moment are produced by the field acting on the mole-
moved, this will act as an "initial" dipole moment which cules. But it is evident from the formalism, as well as
will give rise to "spontaneous" coherent emission. I t from physical considerations, that the radiated field
can be shown easily that no matter what the initial should not depend on how the molecular oscillations are
state of the molecule is at the start of the pulse (except produced, but merely on the phase and amplitude of
for the case tfi=#2), it will have an induced dipole these oscillations. This is indeed the case.
moment at the end of the pulse (in addition to the First, let us look at Eq. (19) and compare the parts
initial dipole moment, and a small random moment due of the field due to induced and spontaneous emission,
to spontaneous incoherent emission during the pulse), respectively. The field due to (coherent) spontaneous
which oscillates with a phase determined by the induc- emission increases linearly with time. This is certainly
ing field and which is the same for molecules having the to be expected for the case of a lossless cavity, an
same sign of (|a 2 ( 2 — \aA2)- We conclude, therefore, oscillating dipole of constant amplitude, and exact
that molecules excited by a coherent rf field will subse- resonance. The induced emission field amplitude, how-
quently emit coherent spontaneous radiation, the phase ever, increases as the square of the time. The explana-
of which is determined by the exciting field. tion is that the amplitude of the oscillating dipole
Another case which is of interest is the one in which induced by the field does itself increase linearly with
the molecules are all originally in the upper state. This the time. In fact, a comparison of the two terms shows
is the situation in the ammonia beam maser, where that the induced radiation field of a molecule at time t
there is selection of states before the molecules enter is the same as the coherent spontaneous radiation field
the cavity. In this case the molecules emit induced of the molecule if it were, during the time 0—/, in the
energy given by state described by a / , a2 and 6', such that 6r~T and
£o2 S0(t)
N , (23) \a1'a2,\=lE,uyt{\a2\2--\a1\2). (25)
87r hco
This is consistent with the result obtained in Eq. (22).
and incoherent spontaneously emitted energy given by We now come to a more important matter in the
NS0(t). The ratio of induced, coherent, energy to comparison of induced and coherent spontaneous emis-
spontaneously emitted, incoherent, energy is sion. The induced emission of Eq. (23) is proportional
to the number of molecules, N. We have seen that in-
duced radiation comes from the induced oscillation of
hay the molecules which is of the same phase for all mole-
namely, the expectation value of the energy in the cules. In the case of coherent spontaneous emission, on
cavity (ignoring the zero-point energy) in units of the other hand, when all molecules oscillate in phase,
energy of a photon. For the case in which the initial the energy radiated is proportional to N2. There seems
state of the field is an energy state, it is well known that to be thus, at first glance, a significant difference in the
the ratio of induced emission to spontaneous emission nature of the radiation. This difference is only formal,
is equal to the number of photons in the initial state. 13 however. In the expression for the energy [TEq. (18)] we
12 have kept only terms up to second order in y. In the
The reason we can determine the state of a system from the
expectation values of two operators is that we are dealing here case of induced oscillation of the molecule, [ a / a / l in
with a very simple system, one which has only two energy Eq. (25) is proportional to y, so that if we wanted to
eigenvalues.
13
The transition probability from a state of n photons to a regard the induced emission in the same manner as the
state of » + l photons is proportional to n+1, which may be coherent spontaneous emission, we would have a term
regarded as a sum of two terms, the term proportional to n being of fourth order in y. As a matter of fact, we can see
induced emission and the remaining term being spontaneous
emission. immediately that the square of P ( 2 ) [[Eq. (14)] gives
 EMISSION IN COHERENT FIELD 9

us an induced-emission term in the energy propor- possible correlated initial states15

tional to N2.
The present analysis, which is based on perturbation ^=2-*[^i(l)^2(2)+^(l)^i(2)], (26)
theory, cannot be used without significant modification and
to study quantitatively the case where higher order *«=2-*C*i(l)*>2(2)- ^ 2 (1)^(2)]. (27)
terms are larger than lower order terms. It is apparent, Any other correlated state may be considered as a
however, in a qualitative manner, that as the induced superposition of these two states and the states
oscillations of the molecule become large (our perturba- <Pi(l)<Pi(2) and cp2(l)<p2(2).
tion theory is correct when these oscillations are small, We calculate the expectation value of the electric
and, to the first order, these oscillations increase linearly field strength and of the energy for \ps and ipa, just as
with the time) the term proportional to N2, referred to we have done for the uncorrelated states. Noting that
above, will exceed the term proportional to N. In the (7l(0)+72(0))as==Oj
ammonia-beam maser the molecules spend a sufficiently
long time in the cavity for this to be true. and that
For the case in which there is no initial field in the </l+/2>*. = 0,
cavity, and the molecules come into the cavity in the
we see, from Eqs. (8) and (15), respectively, that
upper state, Eq. (18) tells us how the incoherent radia-
tion energy will build up. If a molecule spends a time r (P ( 1 ) )a, S =(P^)a, s =0,
in the cavity, and M molecules per second enter the that is, the expectation value of the field strength
cavity, then the rate at which energy is given up to the vanishes. As for the energy, we obtain, by methods
cavity is MSQ(T). For ammonia molecules, perturbation already used, the results
theory is entirely adequate in this case, since the time 2
80(t)
a molecule spends in the cavity is very much smaller Y (P (1)
P '(1)N> = ---
than the lifetime of the upper state.7 JL^I \ J
m J- m' /S ,
m,m'=l 2TTC2

IV.
m,m'—l
The initial molecular states we have considered so
far are products of one-molecule states, and are there- The expectation value of the other terms in Eq. (17),
fore uncorrected states; that is, measurements of the except that of (P (0) ) 2 , vanishes for both \pa and \ps. We
state of each molecule are, in principle, independent of thus obtain, for the field energy due to the molecules,
one another. There is, however, another class of states, <fffield>.= 2« 0 (*), <#field)a=0.
namely, correlated states. These are linear combina-
tions of products of one-molecule states.14 It is interesting to compare this with the field due to
two molecules in the uncorrelated state, \pu= <pi(l) ^(2),
Before we discuss our iV-molecule system, it will be for which the initial energy is the same as for the two
instructive to consider a system of two molecules, in correlated states we have been considering. (All three
order to examine the significance of correlated states states are eigenstates of the energy with the same eigen-
from the present point of view. Let us consider the two value.) According to Eq. (19), the field strength due to
14 the molecules (both spontaneous and induced) is zero,
For purposes of the present article, we define any state which
may be expressed in the form of Eq. (3) as uncorrelated. All other and the part of the field energy due to the molecules
states are correlated in varying degree. A completely correlated [all but the first and last term of Eq. (18)3 is
energy state is an energy state which is itself a superpositon of
linearly independent, uncorrelated energy states (obviously (Hfeid)u= 8o(t),
belonging to the same eigenvalue), with superposition coefficients
satisfying certain relationships which need not be specified—for just half the energy for the symmetric correlated state.
present purposes—in general form, since the aim is not to discuss
the formalism of correlation (which is similar to the angular Since the expectation value of the field strength
momentum formalism), but, rather, to point out certain salient vanishes in all three cases, the molecular radiation, if
features of simple correlated states which illustrate the effect of
correlation. The correlated states discussed specifically in the any, must be incoherent spontaneous emission. We see,
following text are completely correlated. In justification of the now, the meaning of quantum-mechanical correlation:
foregoing definition, it may be pointed out that in a correlated In the correlated states, the incoherent oscillation of
state the result of an energy measurement on some molecule
cannot be predicted with certainty, but this uncertainty is re- the molecules is correlated, being equal and in phase
duced as soon as an energy measurement is made on some (but for the symmetric state, while equal and out of phase
not necessarily any) other molecule. In other words, the energy for the antisymmetric state. In other words, although
uncertainties for some (all, in the case of a completely correlated
state) molecules are correlated. It goes without saying that this 15
The two correlated states being considered are symmetric
is a quantum-mechanical type of correlation and should not be and antisymmetric, respectively. This has no relationship, how-
confused with classical correlations (such as phase relationships ever, to the Pauli principle, since the molecules are assumed to
between oscillating expectation values of the dipole moment, for be sufficiently separated so that there is no overlap of spatial
instance) which may exist in an uncorrelated state. wave function.
10 I. R. SENITZKY

we cannot predict the phase of oscillation of the mole- terms in $ and \^*. There are, thus,
cules in a correlated energy state, the phase relationship
between the molecules themselves is well defined. Here, (N-2)l
again, it may be somewhat strange to label the same 2{N2-N) =2m(N-m)C-2 (30)
(m-l)l(N~m~l)l
phenomenon by the words "incoherent" and "corre-
lated," but there will be no difficulty if one bears in such terms in (31). Each one of these terms [referred
mind our definition of incoherence, namely, that the to both in (29) and (30)] has the value C2y2
phase is an unpredictable (random) variable, as one Xexp[±ico(^i—12)~\, the sign of the exponent depending
examines similar "systems." In the case of uncorrelated on i and j , and being of no consequence since t\ and t2
states, the "system" is a single molecule, while in the are variables of integration. Doing a similar calculation
case of correlated states, the "system" is the group of for the magnetic part of the field energy, we obtain,
molecules which are correlated. (The word "system" as the contribution of the molecules to the energy of
is used here in a different sense than when setting up the field,
the Hamiltonian.) (H^d)s=liN+m(N-m)-]So(t).
We consider now a group of N molecules, and set up
a correlated state with m molecules in the lower state This is a maximum when m~\N^ for which we obtain
and N—m in the upper state. The particular correlation an energy
we choose is such that the wave function is symmetric tiN*+W)8»(t). (31)
with respect to the internal coordinates of all the mole- For an uncorrelated energy state with half the molecules
cules, and is the generalization of the symmetric state in the upper state and half in the lower state, the spon-
for two molecules considered above. We thus have taneous emission energy is, from Eq. (18),
* . = C £ P P * > I ( 1 ) - • • <p1(m)cP2(m+l)' • -wiN),
where P is the permutation operator and the summation Both of these energies are incoherent, and we see that
is over all permutations which have the effect of inter- for the correlated state chosen, the energy is propor-
changing (different) states of pairs of molecules. Since tional to iV2, while for the uncorrelated state it is pro-
there are N \[m I (N— m) ! ] _ 1 such permutations, portional to N. This, of course, is due to the fact, as
brought out in the discussion of the case of two mole-
/ Nl \~* cules, that the molecular radiation, although not having,
C=( ) .
\ml(N-m)l/ in this instance, a well defined phase, has the same phase
for all the molecules in the case of the correlated state,
We are interested in obtaining an expression for the and random phase relationships between molecules in
spontaneous emission, so we assume E 0 = 0 , for sim- the case of the uncorrelated state.
plicity. The expectation value of the field strength, then, I t is interesting to compare the above spontaneous
vanishes, since (Y ( 0 ) ) = 0 . In evaluating the expectation emission energies with the spontaneous emission from
value of the energy, we note that the only contribution N molecules which are oscillating coherently. Consider
of the molecules to (P2) [see Eq. (17)] is through the the state (3) with \ai(m) | = |a 2 (w) | = 2~*, Bm—0. This
term Y<ij(Pi{l)Pj{l))- The molecular operator in Pmil) means that all the molecules are oscillating with the
(0)
is 7m , so that we need to calculate same (well-defined) phase and maximum amplitude.
(Actually, in this state and the last two considered, the
^Ei,-7/0)(*i)Y/0)(*2)>8, (28) expectation value of the molecular energy is the same.)
where h and h are variables of integration. Those From Eq. (18) we have for this case a total spontaneous
operators in the summation for which i=j connect emission energy
only those terms in \//a and ^ s * which correspond to the l(N2+N)S0(t). (32)
same arrangement of states. There are thus Of this, as can be seen from Eqs. (20) and (21), the
2
Nl N term is coherent, the electric field varying as
N— (29) sin(a>H-0), and the N term is incoherent, having ran-
ml(N~m) I dom phase. For large N, Eqs. (32) and (31) give the
same energy, but it is important to note the physical
such terms in (28). The operators Y ; ( 0 ) Y / 0 ) , i^j, con- difference between the type of radiation in the two
nect terms in \f/s containing the factors <p2(i)<pi(j) or cases. The incoherent radiation proportional to N2 is
<Pi(i)<p2(j) w ^ h the terms in ^* containing the factors due to a quantum-mechanical correlation between the
<Pi*(i)<P2*(j) or <p2*(i)<Pi*(j)9 respectively. There aremolecules, all the molecules forming part of one
N2—N such operators and each operator connects "supermolecule," while the coherent radiation propor-
tional to N2 is the same as that coming from a collection
(N-2)\
of classical-type oscillators, all having the same phase.
2
It should be mentioned that there are other corre-
{m-l)\(N-m-l)l
 EMISSION IN COHERENT FIELD 11

lated states with the same number of molecules in the Actually, this is what we have done, except that we
upper and lower states as the one considered above, but have not carried our analysis to high enough order to
having different correlations. Some of these states exhibit correlation effects. We note, for example, that
correspond to the antisymmetric state we have con- P<2> [Eq. (15)] contains the operator P ( 0 ) £ w J m . The
sidered for the case of two molecules; that is, the corre- fourth-order term in the energy, therefore, will contain
lation is such that the radiation of the various molecules the operator (P(0))252m,m'/m/m'. This operator exhibits
cancels out completely. And there are intermediate quantum-mechanical correlation effects, or correlation
cases. Dicke 6 has considered these states in detail, and between the incoherent oscillations of the molecules,
has called \ps, with tn^^N, and those states which are since ((P ( 0 ) ) 2 ) does not vanish for EQ=0, and we obtain
almost like it, "super-radiant." He calls the radiation cross-terms for incoherent radiation. Similarly, it can
from "super-radiant" states "coherent" because the be shown that there are other operators in the fourth-
energy is proportional to TV2, but does not differentiate and higher-order terms of the energy which exhibit
between radiation of well-defined phase and that of quantum-mechanical correlation effects. There are no
random phase. such effects in terms of lower order than the fourth.
One might ask if it is possible to superpose correlated We apply the question posed in the last paragraph to
states of different energies in order to obtain coherent the case of a large number of molecules excited by a
oscillation. A simple illustration, using the case of two strong rf pulse of short duration, a case which was
molecules again, will illuminate this problem. To the treated earlier within the framework of uncorrelated
symmetric state of Eq. (26) we add amounts of the states. Is the subsequent spontaneous emission coherent
states (pi(l)<pi(2) and <p2(l)<p2(2) in the following emission or incoherent correlated emission? I t should
manner: be noted that in both instances the energy emitted and
the instantaneous emission rate may be the same, as
t=A1<Pl(l)<p1(2)+A2<p2(l)<p2(2)
can be seen from Eqs. (31) and (32). From the above
+Al<Pl(i)<p2(2)+<p2(l)<pi(2)J (S3) discussion it is clear that quantum-mechanical correla-
Now, if we choose the coefficients in such a way as to tion will come about (if it does not exist initially) from
obtain the maximum coherence, that is, maximum the interaction of the molecules with one another
amplitude for the (coherently) oscillating dipole- through their radiation. An externally imposed field
moment expectation value, then it can be shown easily will not bring about such correlation directly. On the
that Eq. (33) becomes a special case of Eq. (3), namely, other hand, it will generate an (oscillating) electric
an uncorrected state. In other words, maximum co- dipole moment in each molecule, the expectation value
herence implies zero correlation. We have already shown of which bears a definite phase relationship to the
that maximum correlation implies zero coherence, or field. We may conclude, therefore, that the subsequent
incoherence. (Actually, we have shown this only for spontaneous emission is of the coherent type possible
the symmetrically correlated state, \j/a, but the same with uncorrelated states. The well-known fact that the
reasoning applies to any other completely correlated spontaneous emission bears a definite phase relationship
state.) Thus, coherence and correlation are mutually to the exciting field indicates that our conclusion is
exclusive. Of course, it is a question of degree, and there correct, and we were therefore justified in treating
can exist, in principle, some of both; but it can be shown, the problem by means of uncorrelated states, in part I I I .
generally, for N molecules, that the more we have of For the sake of simplicity we have assumed all along
correlation, the less we have of coherence, and vice that the molecules are all exposed to the field at time
versa. This also follows, generally, from the uncer- 2=0, which is also (for the initial state of the field
tainty principle. which we have assumed) the time at which the phase
The question now arises: in the physical situation in of the expectation value of the field is zero. This
which the spontaneous emission is proportional to N2, assumption of special entrance phase is no limitation,
which is the correct description, one in terms of uncor- however, since we have considered times such that
r e c t e d states or in terms of correlated states? A O a r 1 , and have considered only secular effects (such
phenomenological answer to this question suggests as changes in the amplitude of oscillation occurring
itself immediately. If the phase of the spontaneous- during times long compared to a cycle) which are
emission field can be predicted, then it is the former, independent of the entrance phase, discarding terms
while if the phase is unpredictable, it is the latter. which are essentially transient effects, depending on the
Another way of trying to answer the question is to entrance phase, and significant only during the first
examine the dynamics of the situation under considera- few cycles.
tion. One may assume initially an uncorrelated state
ACKNOWLEDGMENT
(not an unreasonable assumption for a thermodynamic
system; in the case of the ammonia beam maser, where The author wishes to thank Professor Julian Schwinger
there is state selection, we know exactly what to for helpful discussions of the foregoing subject matter
assume) and then let the dynamics tell us what happens. and for his reading of the manuscript.