PHYSICAL REVIEW A VOLUME 57, NUMBER 2 FEBRUARY 1998

Electromagnetically induced grating: Homogeneously broadened medium

Hong Yuan Ling
Department of Chemistry and Physics, Rowan University, Glassboro, New Jersey 08028

Yong-Qing Li and Min Xiao

Department of Physics, University of Arkansas, Fayetteville, Arkansas 72701
~Received 14 April 1997; revised manuscript received 10 September 1997!
A strong coupling standing wave, interacting with three-level L-type ~or ladder-type! atoms, can diffract a
weak probe field ~propagating along a direction normal to the standing wave! into high-order diffractions, a
phenomenon which we name electromagnetically induced grating ~EIG!. We develop in this work a theory for
studying EIG in a homogeneously broadened medium consisting of three-level L-type atoms. We show that by
taking advantage of the absorption and dispersion properties of electromagnetically induced transparency one
can create an atomic grating that can effectively diffract light into the first-order direction.
@S1050-2947~98!03902-X#
PACS number~s!: [Link], [Link], [Link]

I. INTRODUCTION dimension. Such a medium will exert both amplitude and

phase modulations across the probe beam profile in much the
It is commonly believed that a weak probe beam will be same way that a hybrid ~amplitude and phase! grating does
strongly absorbed at its resonance frequency if most of the to the amplitude and phase of an electromagnetic wave. We
atoms are in the lower level. However, recent studies show name this phenomenon EIG.
that by coupling additional coherent fields with other atomic The principal mechanism behind EIG is EIT. To illustrate
transitions, one can form coherent population trapping ~CPT! this point, we compare in Fig. 2 the absorption and disper-
@1,2# states that render a medium transparent to the weak sion in the absence of the coupling field ~dashed curves! with
probe radiation. This phenomenon is known as electromag- the ones in the presence of a strong resonant coupling field
netically induced transparency ~EIT! @3–5#. CPT and EIT are ~solid curves!. We further assume that the intensity of the
the principal mechanisms behind many recent applications, coupling field ~used for producing the solid curve in Fig. 2!
such as lasing without population inversion @6–12#, enhance- corresponds to the peak intensity of a standing wave. Then,
ment of second- and third-order nonlinear processes @13– the dashed curves are what the probe field at nodes ‘‘sees,’’
15#, velocity selective laser cooling @16–18#, atomic mirrors and the solid curves are what the probe field at antinodes
@19–20#, matching pulses @21,22#, electromagnetically in- ‘‘sees.’’ The medium within the EIT window, while being
duced focusing @23#, and elimination of optical self-focusing quite opaque to the probe field at nodes, is almost transparent
@24#. In this paper we explore a new possibility of their ap- to the probe fields at antinodes as shown in Fig. 2~a!. This
plication: electromagnetically induced grating ~EIG!.
Consider a system as shown in Fig. 1~a!. It consists of two
strong coupling fields of frequency v c and wave number k c ,
a weak probe field of frequency v p and wave number k p ,
and an atomic sample. The atomic sample hosts three-level
L-type ~or ladder-type! atoms whose energy diagram is
shown in Fig. 1~b!. As usual, the coupling fields drive the
2↔3 atomic transition ~which has a transition frequency
V 23 and a dipole moment m 23!, while the probe field induces
the 2↔1 atomic transition ~which has a transition frequency
V 21 and a dipole moment m 21!; the 3↔1 atomic transition is
a dipole forbidden one. Here, the two coupling fields, while
being symmetrically displaced with respect to z, are incident
upon the atomic sample at such angles that they intersect and
form a standing wave inside the medium. Because of the
weak nature of the probe field, levels 2 and 3 remain virtu-
ally empty no matter what the intensities of the coupling
fields are. As a result, the standing wave has an amplitude
and space period that are immune to the interaction of light
with the atoms. Since the absorption and dispersion coeffi-
cients of the probe field depend on the strength of the cou-
pling fields, they are expected to change periodically as the FIG. 1. ~a! A sketch of a prototype experimental setup. ~b! The
standing wave changes from nodes to antinodes across x energy diagram of three-level L-type atoms.

1050-2947/98/57~2!/1338~7!/$15.00 57 1338 © 1998 The American Physical Society

57 ELECTROMAGNETICALLY INDUCED GRATING: . . . 1339

by varying the angle between the two wave vectors of the

corresponding coupling fields. The response of the medium
to the fields is governed by the density-matrix equation,
which, in the interaction picture, takes the form

]r 8 i
52 @ H I8 , r 8 # 1L r 8 , ~3!
]t \

where r 8 stands for the density-matrix operator, L r 8 sum-

marizes the effects due to the interaction of atoms with ran-
dom fluctuations, and H I8 is the interaction Hamiltonian. In
the Hilbert space spanned by the bare states ~u1&,u2&,u3&! and
under the rotational wave approximation, H I8 can be repre-
sented by

H I8 52\ $ @ E p e 2iD p t1ik p z u 2 &^ 1 u

1E c sin~ p x/L cx ! e 2iD c t1ik cz z u 2 &^ 3 u # 1c.c.% , ~4!

where

1 m 21F p 1 m 23F c
E p5 , E c5 ~5!
2 \ 2 \

are the Rabi frequencies of the corresponding fields, and

D p (5 v p 2V 21) and D c (5 v c 2V 23) are the frequency de-
tunings of probe and coupling fields. By expanding Eq. ~3! in
terms of the newly defined density-matrix elements
FIG. 2. ~a! Absorption spectrum and ~b! dispersion spectrum.
The dashed curves are produced without the coupling field. The
r 235e iD c t2ik cz r 823,
z
solid ones are produced with a strong resonant coupling field. The r 215 r 821 e iD P t2ik p z ,
~6!
units and parameters are arbitrary.
i ~ D p 2D c ! t2i ~ k p 2k cz ! z
r 315e r 831 , r ii 5 r 8ii ,
can lead to a substantial amplitude modulation across the
probe beam. The dispersion within the EIT window is posi- we can easily arrive at their equations of motion:
tive to the probe field at nodes but negative to the probe field
at antinodes, or vice versa @Fig. 2~b!#. This opens the possi- ]r 11
5G 21r 221G 31r 331iE *
p r 212iE p r 12 , ~7a!
bility of a large phase modulation across the probe beam. ]t
In what follows, we present a theoretical analysis of EIG,
and show how these novel properties within the EIT window ]r 33
can be utilized to produce an atomic grating that can effec- 5G 23r222G 31r 331iE c sin~ p x/L cx ! r 23
]t
tively diffract light into the first-order direction.
2iE c sin~ p x/L cx ! r 32 , ~7b!
II. THEORETICAL MODEL
]r 21
Our analysis begins with the mathematical representations 5 ~ 2 g 211iD p ! r 231iE p ~ r 112 r 22!
]t
for the electromagnetic fields inside the medium. The probe
field is expressed as 1iE c sin~ p x/L cx ! r 31 , ~7c!
2i v p t1ik p z
F p ~ x,z,t ! 5 F p e
1
1c.c., ~1!
2
]r 23
5 ~ 2 g 231iD p ! r 231iE p r 13
where F p is a slowly varying function of time t and distance ]t
z, and c.c. stands for complex conjugate operation. The two
1iE c sin~ p x/L cx !~ r 332 r 22! , ~7d!
coupling fields have wave vectors that are composed of an x
component k cx ~or 2k cx ! and a z component k cz , where
k 2cx 1k 2cz 5k 2c . They form, in the overlap region, a standing ]r 31
5 @ 2 g 311i ~ D p 2D c !# r 31
wave ~along x dimension! with a mathematical expression ]t

F c ~ x,z,t ! 5 21 F c sin~ p x/L cx ! e 2i v c t1ik cz z 1c.c., ~2! 1iE c sin~ p x/L cx ! r 212iE p r 32 , ~7e!

where F c is assumed to be a real constant for simplicity, and r i j 5 ~ r ji ! * , iÞ j, ~7f!

L cx (5 p /k cx ) represents the space period. L cx can be made
arbitrarily larger than the wavelength of the coupling fields r 111 r 221 r 3351, ~7g!
1340 HONG YUAN LING, YONG-QING LI, AND MIN XIAO 57

where G i j is the population decay rate from levels i to j first follow the method in Ref. @25# to obtain, from Eq. ~7!, a
while g i j is the dephasing rate between levels i and j. In solution to r 21 correct to all orders of the coupling field, but
deriving Eq. ~7!, the atomic system has been assumed to be linear order of the weak probe. We then substitute this r 21
closed. The interaction of atoms with fields induces a polar- into Eq. ~9!, and replace P p in Eq. ~10! with the newly de-
ization that oscillates at the frequency of the weak probe rived expression. To present the results in a unitless form, we
field: choose g 21 as the unit for all the decay rates, frequency de-
tunings, and Rabi frequencies, L cx as the unit for x, and z 0
P p ~ x,z,t ! 5 21 P p e 2i v p 11ik p z 1c.c., ~8! as the unit for z, where
where P p is the slowly varying polarization. By performing a
quantum average of the dipole moment over an ensemble of 2\ g 21e 0
z 05 ~11!
homogeneously broadened atoms, we find Nk p m 221
P p 52N m 12r 21 , ~9!
is the resonant ~amplitude! absorption distance at e 21 in the
where N is the atomic density. This polarization becomes the absence of the coupling fields. Finally, the wave equation for
term on the right side of Maxwell’s equation for the probe the probe field, in a self-consistent and unitless form, be-
field, which reduces to comes

1 ] 2F p ] F p kp 1 ] 2E p ] E p
2i 21 5i P , ~10! 2i 1 5 ~ a r 1i a i ! E p , ~12!
2k p ] x ]z 2e0 p N F] x 2 ]z
under the slowly varying amplitude approximation and in
steady state. To obtain a self-consistent equation for E p , we where

@ g 2311 ~ D p 2D c ! 2 # 1 g 31I c sin2 ~ p x !

a r ~ x ! 52 , ~13!
~ 11D 2p !@ g 2311 ~ D p 2D c ! 2 # 12 @ g 312D p ~ D p 2D c !# I c sin2 ~ p x ! 1I 2c sin4 ~ p x !

2D p @ g 2311 ~ D p 2D c ! 2 # 1 ~ D p 2D c ! I c sin2 ~ p x !
a i~ x ! 5 , ~14!
~ 11D 2p !@ g 2311 ~ D p 2D c ! 2 # 12 @ g 312D p ~ D p 2D c !# I c sin2 ~ p x ! 1I 2c sin4 ~ p x !

with The far-field distribution ~Fraunhofer diffraction! over the

diffraction angle u ~with respect to z direction! is propor-
I c 5E 2c , tional to the Fourier transformation of the product of the
input probe field amplitude E 0 (x) and the transmission func-
are the absorption and dispersion coefficients of the probe tion:
field, and

N F5
~ 2 Ap L cx ! 2
~15!
E p ~ u ! 5C E 2`
1`
E 0 ~ x ! T ~ x ! exp~ 2i2 p L cx x sinu /l p ! dx,
l pz 0 ~17!

is the Fresnel number of a slit of a width 2 Ap L cx at a where C is the proportionality. Assuming the input probe
distance z 0 . Note that since no ambiguity is likely to occur field is a plane wave having an amplitude E 0 uniform across
the same symbols are used for the scaled variables. The a beam of a width of N times L cx and defining I p ( u ) as
transverse term ~the second-order derivative of the field with u E p ( u ) u 2 normalized to (CE 0 N) 2 , we find from Eq. ~17! that
respect to x! in Eq. ~12! makes the probe field nonlocal in the
sense that the field at x will affect the field at different x as it sin2 ~ N p L cx sinu /l p !
propagates along z inside the medium. This will complicate I p ~ u ! 5 u E 1p ~ u ! u 2 , ~18!
N 2 sin2 ~ p L cx sinu /l p !
the interpretation of the physics involved in the EIG forma-
tion. In this work, to focus on the main features of EIG, we
where
eliminate the transverse term by working in a parameter re-
gime where N F @1. Such a requirement can be realized by
increasing atomic density or L cx . Under this condition, Eq.
~12! can be solved analytically to obtain the transmission
E 1p ~ u ! 5 E0
1
T ~ x ! exp~ 2i2 p L cx x sinu /l p ! dx ~19!

function for a medium of thickness L ~along z!:

represents the Fraunhofer diffraction of a single space pe-
T ~ x ! 5e a r ~ x ! L e i a i ~ x ! L . ~16! riod. Since we will be mainly interested in the first-order
57 ELECTROMAGNETICALLY INDUCED GRATING: . . . 1341

FIG. 3. The transmission function as a function of x when both

D c and D p are zero, and I c 52.0 ~solid curve! and 10 ~dashed
curve!. Other parameters are g 3150.1, and L54. FIG. 4. The normalized diffraction intensity I p ( u ) as a function
of sin(u) when both D c and D p are zero, and I c 52 ~solid curve! and
diffraction, we calculate I p ( u ) along the first-order diffrac- 10 ~dashed curve!. Other parameters are g 3150.1, L54, L cx /l p
tion angle determined by the grating equation, sinu1 54, and N55.
5lp /Lcx , with the result

I p ~ u 1 ! 5 u E 1p ~ u 1 ! u 2 , ~20! able for diffraction into the first order. However, it peaks at a
certain I c , beyond which it fades away. This seemingly con-
where tradictory effect can be resolved by the following consider-
ation. A strong coupling field can render most of a single
E 1p ~ u 1 ! 5 E
0
1
T ~ x ! exp~ 2i2 p x ! dx. ~21!
period of the grating transparent, as clearly demonstrated by
the dashed curve in Fig. 2. Such a wide opening single pe-
riod is known to possess a diffraction envelope highly con-
centrated along the forward direction, limiting the chances of
III. RESULTS AND DISCUSSIONS light to stray into the first order. This explains why the
In this section, the equations outlined in Sec. II are used dashed curve in Fig. 4, while it enjoys a much increased
to investigate the diffraction power of the EIG under various center maximum, has a weaker first-order peak ~around
parameters. Let us begin our study with a case in which sinu50.25! compared to the solid curve. In summary, the
both D p and D c are zero. A simple inspection of Eq. ~14! first-order diffraction attainable by this pure amplitude grat-
reveals that a i vanishes, implying that no phase modulation ing is very limited.
takes place. The transmission function @Eq. ~16!# is therefore To increase the efficiency of the first-order diffraction, we
simplified to exp(arL), where must move into a parameter regime where the phase modu-
lation is significant. Ideally, we wish to create a medium that
1 is completely transparent to the probe field, but has a phase
a r 52 . ~22!
11I c / g 31sin2 ~ p x !

Figure 3 displays two typical transmission functions. The

two curves share all the parameters except I c ; the I c for the
dashed curve is ten times as large as the one for the solid
one. At the transverse locations around the nodes ~of the
standing wave!, the coupling field intensity is very weak, and
the probe field is absorbed according to the usual Beer law.
In contrast, at the transverse locations around the antinodes,
the coupling field intensity is quite strong, and the probe field
is absorbed much less because of the EIT. This leads to a
periodic amplitude modulation across the beam profile of the
probe field, a phenomenon reminiscent of the amplitude grat-
ing. Figure 4 displays the corresponding Fraunhofer diffrac-
tion patterns. To illustrate the efficiency of the first-order
diffraction of this grating, we use Eq. ~20! to calculate I p ( u 1 )
and display the result as a function of I c / g 31 in Fig. 5. At
small I c , it increases almost linearly with I c . This is within
our expectation because as I c increases, EIT will gradually FIG. 5. The normalized first-order diffraction intensity I p ( u 1 ) as
open up the individual antinodes, making more light avail- a function of I c / g 31 when both D c and D p are zero, and L54.
1342 HONG YUAN LING, YONG-QING LI, AND MIN XIAO 57

I c 520 and D p 52.2. In this figure, the nodes are located on

both ends of the curves while the antinode is located in the
middle. At nodes, both absorption and dispersion are about
0.5 ~far from zero! because the probe is tuned away from
resonance. The maximum absorptions are reached at loca-
tions where the probe detuning matches the Rabi sideband,
that is, I c sin2(px)5D2p, which yields two x values: 0.16 and
0.84. Both the absorption and phase experience a rapid
change around x50.16 and 0.84, but remain relatively flat
around the antinode. These curves, although far from ideal,
bear some traits of the ideal phase grating. First, the absorp-
tion is close to one across most of the single period. Second,
the difference between the minimum and maximum phases is
on the order of p. Figure 7 compares the diffraction pattern
in this case where D p is 2.2 ~solid curve! with the one where
D p is zero ~dashed curve! ~other parameters are the same!. It
shows that the amount of light in the first order is, as we
FIG. 6. The amplitude of the transmission function u T(x) u ~solid predict, considerably increased. To illustrate the role of the
curve! and the phase of the transmission function F ~dashed curve! phase modulation, we present in Fig. 8 two figures; one is the
as a function of x within a single space period when g 3150.1, L Fraunhofer diffraction of u T(x) u ~solid curve!, and the other
54, D c 52.2, and I c 520. is the Fraunhofer diffraction of exp(iF) ~dashed curve!. It
clearly shows that the amplitude modulation tends to gather
modulation on the order of p across the probe beam. The light to the center maximum while the phase modulation
diffraction power of a phase grating plus the fact that no loss tends to disperse light into the high-order directions. There-
of energy takes place inside the medium makes such a grat- fore we conclude that the transfer of light from the center
ing highly efficient in diffracting light into high-order direc- maximum to the first order, demonstrated by the solid curve
tions. Although it is impossible to realize this ideal grating in in Fig. 7, is mainly accomplished by the phase modulation.
our model, we will be in search of parameters that can yield The next question is how far D p should be detuned in order
a grating as close as possible to the ideal one. In light of this to yield an optimum first-order diffraction. Figure 9 is pro-
discussion, we decide to work in a parameter regime where duced to answer this question. It shows that the first-order
the coupling fields have a high intensity and are on reso- diffraction increases as D p until it reaches a maximum value,
nance with the 2↔3 transition, but the weak probe is tuned beyond which it decreases as D p . When D p is too large, the
away from the 2↔1 transition but still operates within the diffraction power is reduced because a large D p , although it
EIT window. The former condition is aimed to keep a good makes the medium more transparent, weakens the key dif-
level of transparency across the beam profile while the latter fraction player: dispersion.
one is designed to introduce a nonzero a i @Eq. ~14!#, there-
fore, a phase modulation of the probe field. Figure 6 dis- IV. CONCLUSION
plays, over a single period, the absorption u T(x) u ~solid
curve! and phase F (5 a i L) ~dashed curve! produced when In this paper we have developed a theory for studying the
phenomenon of electromagnetically induced grating in a ho-
mogeneously broadened medium consisting of three-level L-
type atoms under the condition that N F @1 @Eq. ~15!#. We
have used this theory to investigate the transmission function
of the medium and the Fraunhofer diffraction of the probe
field when both coupling and probe fields are on resonance
with their corresponding atomic transitions. We find that the
atomic medium under this condition serves as an amplitude
grating to the probe field with a very limited diffraction
power. This is because a strong coupling field renders most
of a single period of this grating transparent. Such a single
period creates a diffraction envelope that is highly concen-
trated along the forward direction, leaving very little oppor-
tunity for light to propagate along the first-order direction.
We have also investigated the transmission function and the
diffraction power of the medium in the parameter regime
where the probe frequency is tuned away from resonance
~but still within the EIT window! and the coupling field in-
FIG. 7. The normalized diffraction intensity I p ( u ) as a function tensity remains high. A high I c reduces the loss of light,
of sin(u) when D p 50 ~dashed curve! and 2.2 ~solid curve!. Other while a nonzero D p introduces a periodic phase modulation
parameters are g 3150.1, L54, D c 50, I c 520, L cx /l p 54, and N across the probe beam. We find that this scheme takes ad-
55. vantage of the two features of EIT: transparency and high
57 ELECTROMAGNETICALLY INDUCED GRATING: . . . 1343

FIG. 8. Fraunhofer diffractions of u T(x) u ~solid curve! and F FIG. 9. The first-order diffraction I( u 1 ) as a function of D p
~dashed curve! as functions of sin(u) when D p 52.2, and I c 520. when I c 5 ~a! 10, ~b! 20, and ~c! 50. Other parameters are D c 50,
The other parameters are g 3150.1, L54, D c 50, L cx /l p 54, and g 3150.1, L54, and N55.
N55.
ent from those for EIT, the main mechanism leading to EIG
dispersion in the EIT window, and results in an atomic grat- is still EIT. As a result, we believe that EIG will be observ-
ing that can effectively diffract light into the first-order dif- able in similar experiments, although the diffraction pattern
fraction. will be less pronounced. In order to extract the main physics,
This theory is applicable to atomic beams, or atomic cells we have assumed that the optical beams are plain waves. In
where the atomic density is so high that collisions become practice, they have a Gaussian profile. Thus our theory ap-
the dominate linewidth-broadening mechanism @3#. This plies only to optical beams whose width is much larger than
grating may be utilized as a beam splitter for fields whose the grating period. When the finite extent of a Gaussian pro-
frequencies are out of reach by the gratings commercially file is taken into account, the grating efficiency is expected to
available. Both the direction and the amount of the split light be reduced. The details of these practical concerns are be-
can be controlled by the coupling field. We note that in an yond the scope of this work, and will be left to a future
atomic beam experiment, the probe diffraction pattern will study.
be shifted if the velocity of the atomic beam changes. As a
result, this diffraction pattern may provide insight into not ACKNOWLEDGMENTS
only the internal ~atomic energy levels and decay rates! but
also the external ~velocity! atomic variables. The work at the University of Arkansas was partially sup-
The phenomenon of EIT in the presence of Doppler ported by the Office of Naval Research and the National
broadening is evident from several recent experiments @4#. Science Foundation. The work at Rowan University was sup-
Although the experimental arrangements for EIG are differ- ported by a grant from Research Corporation.

@1# G. Alzetta, A. Gozzini, L. Moi, and G. Orriols, Nuovo Ci- @7# O. Kocharovskaya and Ya. I. Khanin, Pis’ma Zh. Eksp. Teor.
mento B 36, 5 ~1976!; G. Alzetta and L. Moi, ibid. 52, 209 Fiz. 48, 581 ~1988! @ JETP Lett. 48, 630 ~1988!#.
~1979!. @8# S. E. Harris, J. E. Field, and A. Imamoglu, Phys. Rev. Lett. 64,
@2# H. R. Gray, R. M. Whitley, and C. R. Stroud, Jr., Opt. Lett. 3, 1107 ~1990!.
218 ~1978!. @9# M. O. Scully, S.-Y. Zhu, and A. Gavrielides, Phys. Rev. Lett.
@3# K.-J. Boller, A. Imamoglu, and S. E. Harris, Phys. Rev. Lett. 62, 2813 ~1989!.
66, 2593 ~1991!; J. E. Field, K. H. Han, and S. E. Harris, ibid. @10# L. M. Narducci, H. M. Doss, P. Ru, M. O. Scully, S. Y. Zhu,
67, 3062 ~1991!; A. Kasapi, Maneesh Jain, G. Y. Yin, and S. and C. Keitel, Opt. Commun. 81, 379 ~1991!.
E. Harris, ibid. 74, 2447 ~1995!. @11# O. Kocharovskaya and P. Mandel, Phys. Rev. A 42, 523
@4# J. Gea-Banacloche, Y. Q. Li, S. Jin, and M. Xiao, Phys. Rev. ~1990!.
A 51, 576 ~1995!; Y. Q. Li and Min Xiao, ibid. 51, R1754 @12# H. Y. Ling, Phys. Rev. A 49, 2827 ~1994!.
~1995!; M. Xiao, Y. Q. Li, S. Jin, and J. Gea-Banacloche, @13# K. Hakuta, L. Marmet, and B. P. Stoicheff, Phys. Rev. Lett.
Phys. Rev. Lett. 74, 666 ~1995!. 66, 596 ~1991!; G. Z. Zhang, M. Katsuragawa, K. Hakuta, R. I.
@5# H. Y. Ling, Y. Q. Li, and M. Xiao, Phys. Rev. A 53, 1014 Thompson, and B. P. Stoicheff, Phys. Rev. A 52, 1584 ~1995!.
~1996!. @14# P. R. Hemmer, D. P. Katz, J. Donoghue, M. Cronin-Golomb,
@6# V. G. Arkhipikin and Tu. I. Heller, Phys. Lett. 98A, 12 ~1983!. M. S. Shahriar, and P. Kumar, Opt. Lett. 20, 982 ~1995!.
1344 HONG YUAN LING, YONG-QING LI, AND MIN XIAO 57

@15# Y. Q. Li and M. Xiao, Opt. Lett. 21, 1064 ~1996!. Westbrook, W. D. Phillips, P. Marte, and P. Zoller, Phys. Rev.
@16# A. Aspect, E. Arimondo, R. Kaiser, N. Vansteenkiste, and C. Lett. 72, 997 ~1994!.
Cohen-Tannoudji, Phys. Rev. Lett. 61, 826 ~1988!; J. Opt. Soc. @21# S. E. Harris, Phys. Rev. Lett. 72, 52 ~1994!.
Am. B 6, 2112 ~1989!. @22# J. H. Eberly, M. L. Pons, and H. R. Haq, Phys. Rev. Lett. 72,
@17# F. Papoff, F. Mauri, and E. Arimondo, J. Opt. Soc. Am. B 9, 56 ~1994!.
321 ~1992!. @23# R. R. Moseley, S. Shepherd, D. J. Fulton, B. D. Sinclair, and
@18# M. R. Doery, R. Gupta, T. Bergeman, and H. Metcalf, Phys. M. H. Dunn, Phys. Rev. Lett. 74, 670 ~1995!.
Rev. A 51, 2334 ~1995!. @24# M. Jain, A. J. Merriam, A. Kasapi, G. Y. Yin, and S. H. Harris,
@19# P. Marte, P. Zoller, and J. L. Hall, Phys. Rev. A 44, 4118 Phys. Rev. Lett. 75, 4385 ~1995!.
~1991!. @25# N. Bloembergen and Y. R. Shen, Phys. Rev. 133, A37 ~1964!.
@20# L. S. Goldner, C. Gerz, R. J. C. Spreeuw, S. L. Rolston, C. I.