PHYSICAL REVIEW LETTERS

VOLUME 55, NUMBER 20

Inhibited Spontaneous
Randall

G. Hulet,

'

Emission by a Rydberg Atom

Eric S. Hilfer, and Daniel Kleppner

Research Laboratory of Electronics and Department of Physics, Massachusetts

Cambridge, Massachusetts 02139
(Received 29 July 1985)

of Technology,

radiation by an atom in a Rydberg state has been inhibited by use of parallel conto eliminate the vacuum modes at the transition frequency. Spontaneous emission
"turn off" abruptly at the cutoff frequency of the waveguidelike structure and the
is measured to increase by a factor of at least 20.

PACS numbers:

32.80.t, 31.60. +b

'

"

20.
The experiment employs Rydberg atoms in a "circu-

lar" state.

This is a single-electron atomic state with

large principal quantum number n, and magnetic quantum number ~m =n 1. The circular state is essential for this experiment because it radiates by only a
States with lower values of
single dipole transition.
have two or more decay channels, all of which
~m
would have to be suppressed in order to prevent sponis
The transition
observed
taneous
emission.
(n =21, [m [ =20); the wave(n =22, ~m =21)
~

Institute

Spontaneous
ducting planes
is observed to
natural lifetime

Spontaneous emission is often regarded as an unavoidable consequence of the coupling between matter
and space. However, as one of the authors has pointed
the atom with a cavity which
out,
by surrounding
has no modes at the transition frequency, spontaneous
emission can be inhibited or "turned off. Drexhage,
in studies of fluorescence by dye molecules deposited
on a dielectric film over a conducting plane, observed
a decrease of up to 25p/o in the fluorescent decay rate
due to cavitylike effects. Rydberg atoms provide the
to achieve inhibited emission in miopportunity
crowave cavities. Vaidyanathan, Spencer, and Kleppner4 have reported
a closely related effect the
suppression of blackbody absorption in Rydberg atoms
and Gabrielse and Dehmelts
by a parallel-plate cavity
have recently observed inhibited spontaneous emission of cyclotron radiation by an electron confined in a
Penning trap. They observed damping times as much
as 5 times longer than the free-space value and attributed the increased lifetime to cavity effects from the
electrodes. Their experiment provided
surrounding
convincing evidence that spontaneous emission can be
suppressed, but because of the difficulty of calculating
the mode structure in the trap and varying it systematically, the demonstration was qualitative. In this Letter
we describe the observation of inhibited spontaneous
emission of a free atom by a well-characterized cavity
Sponwhose properties can be varied systematically.
taneous emission by a Rydberg atom was observed to
"switch off" abruptly as the transition wavelength was
varied across the cavity's cutoff wavelength and the
lifetime was measured to increase by a factor of at least

11 NovEMBER 1985

length is A. =0.45 mm. The selection rule 6~m = 1

assures that the radiation is polarized perpendicular to
the quantization axis defined by an applied electric
field.
In the usual mechanical treatment of spontaneous
emission, the emission rate is proportional to the density of modes in free space. It is straightforward
to
show that in a cavity formed by infinite conducting
plates separated by distance d, the mode density for
the electric field parallel to the surface vanishes for
d & X/2, and that for d slightly greater than A. /2 the
emission rate at a distance z from the midplane is7
~

'

= 3A p sin'(7r z/d 7r/2),

where Ao is the Einstein A coefficient for the transition. In our experiment the atoms sample all of the
space between the plates, so that the sample-averaged
decay rate is

A'=
', Ap
A

'

=0

for

for

X/2d & 1,

X/2d &

1.

If A, /2d &

1 the emission rate is slightly enhanced relative to the free-space rate. As the plate spacing is reemission is abruptly inhibited
duced, spontaneous
when A becomes greater than 2d. (Quadrupolar radiation can occur on the transition n
n 2, but its rate
is negligible. ) Alternatively, the spacing can be fixed
and the wavelength varied, for instance by the Stark

effect.
The experiment was performed with a thermal
atomic beam of cesium. The beam passes sequentially
through three regions: a production region where the
atoms are transferred to the circular state, a drift region inside the parallel-plate cavity with a mean transit
time approximately equal to the free-space radiation
lifetime, and finally the detector where the arrival
times of the n =22 atoms are recorded. Spontaneous
emission changes the shape of the time-of-flight curve
from the usual Maxwellian form for an atomic beam.
When the emission is inhibited, the curve reverts to its
usual form and there is a dramatic increase in the
overall transmission.
To switch from enhanced to in-

1985 The American Physical Society

2137

PH&cCAL REVIEW LETTERS

NUMBER 20

VOLUME 55

1
tric field is app lied
hibited emission,n an eec
i
to change
he wavelength a few percent b y the Stark efffeet.
m is popud e lasers
andd the atoms are
to tth'e c
't't' b
use
'
ions are induce db
p
'
batic rapi 'd p assage in ann e ec
ic hd e
26 t 407 V/c111 ill
11
tu, .

t'

eofameto

transitions

""ul".

d-

17.6

.. ..d;

'

P ed quartz

l. ,

Th

'

the atomic beam

e
1
p rpe d
and are polarizedd perpe
erpendicularly

"

1100 V. Because
at
ause o
of the tilt, the
ato ms
an increasing f"ld
th
ass
pe
'
'o
e tween t h e plates. . Thee ioniza io
s tes i e
eld-ionization
eveent can serve to
'h

d'"' '"'

=21,,

roxithe final state

'

i
ionizes
roximately
approx

t finite

to
old-plated

6.5-K radiation

"d

lied is limited by
ie
e ~ 22 circ
must be less than 260 0 V/cm.
c . This alloww

mately

of 7'.

lates that are se

t'd b
fo th
11 lb
'
The potentia 1 between
e
the p a es

f6

ff t th a

was

l1

aratus

'

ce lifetime waa
d the free-space
'
'
'
'
of-flight
o ig distribution is g'iven y

in

dN
dt

ckbd
o

hi k-

Th e

spacers.

the ions to
detector. The two states are
e ec iv
d. The signal from h' 'har 'd
'
partic lee detector passes to a mu tic a
with a

aluminum

Ph

1'

of th
roximately 0.3 cm. A
'
lengt h in the negative- ie
ution

th in zero fie ld. Th'""ul

shift b u t th't"n"t
wavelengt h
h
"'nd
Star k
g y

0.5 cm

tream

f ld'"d th'd"t

11 N'OvEM~Eg. 1985

iss t e

Xo to
t
'

(iojr) 2 ri

'

dia ive decay

ca rate. to

= L/u,

where L is the

an angle

robable speed in t e so
pro

/'5

ys
I

200

400

600

800
TIME (p, S)

1000

200

~ ~i
l400

1600

rates.

2138

600

'

- t si nnals for variouuss ontane

tsi
t e
15/A, h
e
- p
adiative decay rate. Curve
calculated sig
i nal,
d decay rate A = z 0.
calculated signal,
'
no radiative e decay. Curve
A = A 0, calculated
'
(dashed line), measure
so

FIQ. .

400

200

800

1000

200

I400

I600

TIME (p. $)

/,

imous emission. Time-of-flight

us emission
A.
d spontaneous em' iss ion A.
)
)
w
were ta k enn simultaneously
usl
y m
with an applie d eelectric field.
FIG. 2. In
for inhi

i i

po

EVIEW LETTERS
PHYSICAL RE

VoLUME 55, NUM E'ER 20

1] QPPEMBER

erne
ared with this disental data are compare
'
'
'
dlif ti
tribution
ri
in Fig. 11, curve C. h

u,

e agreement wiiith the calculate d vaalue, 451 s. The

lent
principal source s of uncertainty are the oven temperainitee ssource and detec tor
ture and the effects on L of ini

curve fitting.
- - ight
lculated time-ofi ht
Figure 3 disp laay sdataan d cac
lnhi 1ted
ite eeIYllssion. The
curves for
or enhanced and inhi
cay rate is
I

= (l/L)A'+ (1 l/L)Ao,
'

apertures.

in 'b ite
inhi
t d sp
spontaneous emission, the
o
d
cavity plates we re returned to their /2
time-of-flight da ta were accumuulated
a
in two channeels
'
a low valu e
e
as the electric fie ld was alternate d between
1 (enhancedd emissio
emission) and a hig h
//2d
('in h'btd
X~2&d
i ie
i io ). T ical
/
l1D
in Fig. 2. T h e mos
o
re
resu
the total signa.1 Th i

th'

'lt'"d d'".

o the rates mus t

but accurate values of

"t

b' 'bt"n'd

time-of-flight
and total
L are the cavity an
o
t
/L=0.
79.
I
t
th
1
spe i
gths,
y;
1ae
ted curve for enhhanced emission,

nal sis, the measure

en
roxor within its un
in the calcuae
ae S%.. In plotting
ately
The curve fit in
is a
bited emission,
F1g. 3 is excellent forr times less th a n 900 p, s, but for
g

'o.
4

'th speeds less

there

t"n'fth'"

is a systematic d epel
1
n'1
et oof the depletion is abrupt, the deBecause the onset
i
of a larger deletion cannot simp 1y be a mani 'f es t a tion
f
.
elieve that the iminis
c y
'
b h f
to a slight pene ra
e
t' th'e d't"tor h"h
electric fie ld of the cavity into
-ionization location.
ift in the fie -ioniz
causes a s if
at a lower ie ld than
a faster ones so that
'onize
h
se in field wou ld
in the detection ape
of
prior to reaching
t data for t'im es greater than 900
t is
'
h h i
p, s we re excluded in thee final ana 1ysis, Iho
'
effect on the results.
excu
imh d a t'a for 'nh'b't'd
In principle the time-of-flight
e IYlission coUld b e fitted with a sing e p
te with use o
H
h
curves to pro vide normalization.
an
with the ushanced-decay daata taken simu ltaneously
ble inhibited-decay data were no t go od enough to prolization. Conse-

than

"

(a)

CA

o
C3

200

400

600

800

IOOO

I200

I400

1600

TIME (p. S)

quen

normalization
rameters
thee resulting errors an 1
e data were co nsistent
Sgu
limit of error o
t dtto b e a tl
ibited lifetime is estimated

rate

and

of interest.

"d

b
with

''h=
A

""f . '"u"d

Th is w

b
total transmission as a u
studying the to
Ideally, the turn on wou ld b e infinitely sharp
lit the transition is spread by the efvit . Data for t e o
i . 4. The width o f tll e
1
1
an effective Qof
o a
ralThis can be accouunted for chic y
lelism of the p lates. Losses ue too the finite conducnd edge e ffeccts are negligible. The
1015 i d
2d
ionization in t e
ie
the onset o f field
~

oi

200

t20ti

600

800

IOOO

1200

400

1600

TIME (p.S)

FIG. 3. (a) Experim ental (solid ) an d theoretical (dashed)

i ht curves for enhanced d spon
an
s ontaneous
(b) Same for inhibited spon ane

emission.

2139

PHYSICAL REVIEW LETTERS

VOLUME 55, NUMBER 20

C9

.975

.98

.99

t. ol

I.02

(.03

~ /'2d

FIG. 4. Total transmission signal for A/2d near the cutoff

region. (A. is altered by application of an electric field, which
increases from 0 to 3. 1 kV/cm for the data shown. ) The
sharp increase in transmission at A/2d = 1 is due to the inhibition of spontaneous
emission; the decrease for X/2d
& 1.015 is due to field ionization between the cavity plates.
slight variation

in transmission with )t/2d is visible in

the enhanced region. In this region the field between
the field plates is increased from 0 to 1700 V cm
At low fields there is a small loss of atoms because of
nonadiabatic effects that mix m states as they pass
from the production region into the cavity. As the
field is increased this loss is eliminated.
Suppression of spontaneous emission offers the opportunity of eliminating the natural linewidth in spectroscopic measurements.
However, in measuring energy levels to superhigh accuracy by elimination of
spontaneous emission it must be remembered that the
cavity inevitably introduces "nonradiative" shifts that
alter the structure of the atom. ' In effect, the atomvacuum system is replaced by the atom-cavity system.
There has been theoretical interest in such systems
since the work of Casimir and Polder. 9 The problem is
now becoming experimentally important, for as Brown

2140

11 NOVEMBER 1985

et al. ' have pointed out, cavity shifts are expected to

be significant in forthcoming measurements of g 2
for a trapped electron. Furthermore, they must be
taken into account in the design of future atomic
clocks. Although cavity shifts may set the ultimate
limit for useful applications of inhibited spontaneous
emission as a high-precision technique, it should be
pointed out that the present experiment provides a significant step toward very-high-precision
spectroscopy
of Rydberg atoms by achieving a major increase in
their useful time-of-flight path length.
The authors thank Dr. Harald Hess for helpful advice on the cryostat design. This work was supported
by the Joint Services Electronics Program under Grant
No. DAAG29-83-K-0003, the National Science Foundation under Grant No. PHY84-11483, and the Office
of Naval Research under Grant No. NOO14-79-C-

0183.

&')Present address: Time and Frequency Division, National Bureau of Standards, Boulder, Colo. 80303.
~D. Kleppner, in Atomic Physics and Astrophysics, edited by
M. Chretien and E. Lipworth (Gordon and Breach, New
York, 1971), p. 5.
2D. Kieppner, Phys. Rev. Lett. 47, 233 (1981).
3K. H. Drexhage, in Progress in Optics, edited by E. Wolf
(North-Holland, Amsterdam, 1974), Vol. 12, p. 165.
4A. G. Vaidyanathan,
W. P. Spencer, and D. Kleppner,
Phys. Rev. Lett. 47, 1592 (1981).
~G. Gabrielse and H. Dehmelt, Phys. Rev. Lett. 55, 67

(1985).
R. G. Hulet
(1983).

and

D. Kleppner, Phys. Rev. Lett. 51, 1430

7P. W. Milonni and P. L. Knight, Opt. Commun.

(1973).

9, 119

8T. Heindorff and B. Fischer, Rev. Sci. Instrum. 55, 347

(1984) .
H. B. G. Casimir and D. Polder, Phys. Rev. 73, 360
(1948). Some additional references are given in Ref. 4.
L. S. Brown, G. Gabrielse, K. Helmerson, and J. Tan,
Phys. Rev. Lett. 55, 44 (1985).