HAL Id: jpa-00248110
https://hal.archives-ouvertes.fr/jpa-00248110
Submitted on 1 Jan 1994
HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.
Atom interferometry with arbitrary laser configurations : exact phase shift for potentials including inertia and
gravitation
Jürgen Audretsch, Karl-Peter Marzlin
To cite this version:
Jürgen Audretsch, Karl-Peter Marzlin. Atom interferometry with arbitrary laser configurations : exact
phase shift for potentials including inertia and gravitation. Journal de Physique II, EDP Sciences,
1994, 4 (11), pp.2073-2087. �10.1051/jp2:1994248�. �jpa-00248110�
Classification Phi'.ii<..I Ab.itia< t-I
42.50 04.90
Atom interferometry with arbitrary laser configurations
:exact
phase shift for potentials including inertia and gravitation
Jiirgen
Audretsch and Karl-Peter MarzlinFaLultfit for PhysiL der Universitit Kon~tanz, Po~tfach 5560 M 674. 78434 Kon~tanz. Germany
(Ret eiied /4 Malt h /994, a«eptet/ 6./iih /994)
Abstract. We ,tudy in a general treaimeni the influence oi a cla,, of perturbation~ acting on the
atom, ot an atomic interferometer. An exact
expre,,ion
for the re,ulting ;hift of the interferencepattern I, given for an
arbitrary
number of la;er zone; withtravelling
or ~tanding wave~. Thea;;umption
made i~ that the two-level model isappropriate
for the description of atom~ and that the perturbation~ can be neglected in the la~er zone~. As an important application we calculate the influence of acceleration, rotation, and ~pace-time curvature on the Ramsey interferometer and theatomic fountain. The me»urability of the re~pective pha~e ~hift~ i~ di~cu~sed.
1. Introduction.
The idea of
using
lasers a~ beamsplitters
for atomic interferometers wasdeveloped by
Bordd in 1989. Due to momentum conservation atoms can
gain
or loose a momentum of kit
they
absorb or emit aphoton
with wavevector k. This effect can be used tosplit
andrecombine an atomic beam. The interaction with a laser beam which is tuned to be
nearly
inresonance with a certain atomic transition can be described within the two-level model of
atoms. This was used to calculate the
complete fringe
pattern in theRamsey
spectrometerwhich consists of two
pairs
ofcounterpropagating
laser waves[2).
The atomic fountaindevice
[3)
may be treated as an effective two level system.There are extended efforts to
;tudy
the influence ofelectromagnetic
fields of laser beams onmoving
atoms with the intention todesign
mirrors and beamsplitters
in an atomicinterferometer. In the
following
westudy
thecomplementary problem
:using
a black boxapproach
for the laser zones we treatexactly
the influence ofa class of
perturbations acting
onthe atoms when these are
moving through
the dark zones between the lasers in theinterferometer. Our intention
thereby
is to treat the induced shift of afringe
pattem for a set-up with anarbitrary
number of la;er zones withstanding
orrunning
waves. We restrict to two level atoms and strive for a result terms of a recursion formula which allows us to work out theinfluences of the different part~ of the
perturbation
to any desired order ofmagnitude.
This i~ ofparticular importance
if these terms differlargely
inmagnitude.
In thi~ case one must be able to control all the mixed terms which may occur in order to obtain thephase
shift for aparticular experimental specification
up to agiven
accuracy. In thi~ sense wegeneralize
our calculations in reference[4].
There are at the moment two
potential applications
of ourgeneral
scheme. Theperturbations
may be unwanted, for
example
when the atomic interferometer is used as afrequency
standard.In this case one has to know the structure of their influence in order to eliminate it. On the other
hand, the
perturbation
may go back to the influence ofacceleration,
rotation, andspace-time
curvature. In this connection the intention i~ to examine whether such an inertial or
gravitational
influence on quantum matter can be detected in thelaboratory using
atomicinterferometry. Interferoinetry
with neutrons[5. 6],
electrons[7],
and atoms[8, 3]
is at present theonly possibility
to examine this effect. Its relevance forgravitational theory
has beendiscussed in
[9].
Forspace-time
curvature we have ~tudied themeasurability by
aRamsey
interferometer in reference
II 0].
For the influence of acceleration on thephase
shifts there is adifference between neutron and atomic
interferometry.
For neutrons the beam~plitters
arerealized with a
single crystal,
and a seriesexpansion
of the WKBphase
is needed 1-13].
We will show below that for atom interferometers where lasers are used as beamsplitters
thephase
shift due to acceleration alone hasonly
one part the first order contribution.The paper is
organized
as follows. In section 2 we willexplain
thegeneral
scheme for thealgebraic
calculation of the shift. The idealization of the laser beams is addressed in section 3.The elaboration of the
complete phase
shift is contained in section 4 and will beapplied
to theRamsey
interferometer and the atomic fountain in section 5. On this basis themeasurability
ofspace-time
curvature is studied in section 6. Aquick
direct calculation of thephase
shift ifonly
acceleration is present is
given
in ~ection 7. We will use natural units(h
= c = unless
otherwise stated.
2. General scheme.
In the two-level model the Hamiltonian of the atoms in the atomic interferometer is
given by
H<oi ~ H~m + H~~~~ d E ix, t
(1)
~~~~
~Lm 6
~
~p
~~~Hj<om =
HA )
r (31where x, p, and M are the center of mass
position
and momentum, and the mass of the atom,respectively. H~
= E~
la) jai
+E~ 16) (b
is the internal energy operator restricted to thetwo atomic energy
eigenstates la)
and16)
of interest. We will assumeE~<E~.
r y~
la) jai
+y~16) (b[
contains thedecay
factor~ y~, y~ of the two state~. Thecoupling
of the atoms to the laser iield in the laser zones i~given
in thedipole approximation by
d E (x ), where E i~ the electric field of the laser and d is thedipole
moment of the atom.In a two level model one can set
jai
dla)
=
(b[
d16)
=
0, and
jai
d16)
=
(b[
dla)
=
dab-
H~
inequation (I)
represents the influence of externalpotentials
on thepropagating
matterStates.
Depending
upon circumstance~ the~epotential~
may either beregarded
as externaldisturbances, the influence of which has to be known in order to be able to eliminate them in an
experimental
setup, or as the externalpotentials
to be measuredby
matter waveinterferometry.
We assume
H~
to be of the formH~
=Ma x +
R,>
I,i,,< "f ."«< "' (~ ~P).
~~~Here and in the remainder we use the convention that any index of
R,ji,j,,,
which appears asecond time in the same formula ha~ to be summed from to 3. The first two terms may be
regarded
as the first terms of aTaylor expansion
of some scalarpotential.
The third term represents anangular
momentumcoupling.
In theapplications
discussed belowH~
will represent the influence ofgravitation
and of the rotation of the reference system. In this casea = const. is the acceleration, w
= const. the
angular velocity,
andRot
o~, = const. are the
components of the Riemann tensor with re~pect to the Fermi coordinates of the observer
[10(.
It is the aim of this paper to
give
ageneral
treatment of theresulting
shift of the interference pattern ifH~
is switched on.The
general
arrangement for an atom interferometerusing
lasers as beaui.iplitteis
consists ofan atom beam
passing
a sequence of N lasers withstanding
ortravelling
waves andmoving
free of la~er influences in the dark zone~ between the lasers for time intervals
T,.
The freeevolution is
governed by
the Hamiltonianli=H~~n+H,~,,~~
ofequation
(I). Becauseh
i~ timeindependent
thecorresponding
time evolution operator issimply
exp[- iliT,
in theI-th interval of free evolution.
Denoting
the evolution operator which de~cribes the re~ult of the interaction with the I-th laserby U~ (I
thecomplete
evolution of the sate of the atoms is givenby
~ (t~
))
=
U~ (N )exp [- iliT~ ]U~ (N U~
(2)exp [- iliT, ]U~
)exp [-iliTo] ~o)
(5)
where
~j,)
is the initial state of the atoms and t~ is the final time.During
its evolution theingoing
atomic beam issplit
into manypartial
beaui.I with differentmomenta and different internal states which can interface under certain conditions. A~ an
example figure
shows thepartial
beams for theRamsey
interferometer. Note that the wordpartial
beam is introduced for~implicity
reasons to denote in asugge~tive
way parts of thesplitted
wave function with certain momenta. Apartial
beam characterizes aparticular history
of excitations and deexcitations.
Equation (5)
showsclearly
that there is no concept of apath
in,.
,' ,'
,~,~~~ ~~i~)j
Tj
T~ T3""",,~
l 2 3 4
Fig. I. All partial beam~ of an atomic beam passing the Ramsey device as they are predicted in the rotating wave approximation. The vertical arrows denote the running la~er wave~ and their direction.
Dihed lines correspond to excited atoms. The atoms are
initially
deexcited and travel freely for time~T, between the lasers. Beam I and II are able to interfere.
configuration
space in the quantum evolution. In thefollowing
wesingle
out twopartial
beams I and II
resulting
at the final time t~ in~,)
and~,j)
and;tudy
thecorresponding
interference pattern.
A formal
description
of thepartial
beam~ can be made if wedecompose
the state vector into its deexcited part and its excited part. The Hilbert space of ourproblem
is the directproduct
of the two dimensional internal Hilbert space and the one of the center of mass motion. We cantherefore write
~~ ~~)
~~)
~~~with
[~~)
=
(a) (a (~)
and similar for(b). Independent
of the details of the atom's interaction with the laser~ in the laser zones the transition from theingoing
matter state to theoutgoing
one can be describedphenomenologically by
~
Y< ~>
where I
= I,
,
N denotes the number of the laser zone. The component~ of thi~ matrix are in
general
operatorscontaining
x and p. A morespeciiic
form will begiven
for idealized laserzone; in the next section. The
physical meaning
of the components is clear cv, occurs if anexcited atom remain~ excited,
6,
it it continues to be in the statea).
Ifp,
arise we know that the I-th la;er ha~ the atom excited, and y, describes the proces; of deexcitation. For aparticular partial
beampa,;ing
asplitter only
one oi the component; ofU~
(I will become effective. It I, therefore obviou; that each totalpartial
beam i~;pecified
it we write down thecorresponding
;equence oi components. Consider for in;tance the two
interfering
beam; I and II oifigure
I.Beam I
correspond~
to the sequencepi
y, p~ cv~ and beam II I; describedby
&j &~6~ p~.
To concentrate in the calculation on the relevant term, the
general
scheme in thi; paper willbe the iollow in g for ~ome
configuration
of la;ers and twogiven particular partial
beams I and IIresulting
in~j)
and~,j)
there isalready
an interference pattern if the externalpotentials
vanish(H~
= 0 ). This pattern is assumed to be known. We do not addre;; thequestion
whichpairs
ofpartial
beam; lead to a detectable pattern.They
may for instance be;pecified
as tho~e with the ;mallestvanishing
outby Doppler broadening.
If we now switch on the di~turbanceH~
thesefringe
patterns are shifted. Thecorresponding
terms in the calculationdepend
on a, w,or
Rj,
it,,,,. We areonly
interested in thi~pha;e
shift. In~,)
and~,,)
the contributions whichare
already
there in the undisturbed case can be isolated a~complex
iactors. If not otherwise stated our <.r)iiientir)n whenworking
out~,)
and~,j
will therefore be to omit contributions to these factors wheneverpossible. Thereby
we mudkeep
certain contributions in order not toloose terms oi
H~. Finally
theresulting phase
shift causedby switching
onH~
i~ obtained inrestricting
in(~, ~,,)
to those parts which contain the constituents ofH~.
This convention has consequences. Because H,~~~~~~ commutes with H~~~ it contributes
only
tothe omitted
complex
factor. Therefore(~,)
and(~,,)
gosolely
back to the action ofH~~~ in the dark zones and to
respective
components ofU~(I)
ofequation
(7).For
example,
for the two beams in theRam~ey
device offigure
we iind>fi i <ii r >ii i <n
~,)
= cv~ e ~"'
p
~ e ~"' y~ e ~"' '
p,
e "" ~'~,>)
(8)
->ii i~ -<ii i, -<ii i, -<ii in
4'III
~ P4 ~ ""
61
e ~"'62
e ~"' 61 e ~"'4'n)
In thi; ca~e both
partial
beams are excited aiter the passage of the iourth laser. In many devices such as theRamsey
interierometer the number of excited atoms after the evolution ismeasured. The observed
quantity
is therefore(~~ ~~)
and it contains beside othercontributions an interierence term of the form 2 Re
(~j ~,,).
In theRamsey
device almost all the other interierence term~, I-e- those between otherpartial
beam~, are washed out due to theDoppler
effect(2].
This entails that thesignal
con~ists of somebackground
and somesinusoidal variation
coming
from the interference term. Thephase
of this variation can be calculatedby evaluating
(4~j~,,).
Thephase
shiftoriginating
fromH~
ofequation
can be read offdirectly.
Thi~ i~ thegeneral
scheme of the paper.Our aim is to
give
ageneral expression
for~, ~,,
for any twogiven partial
beam~ for anarbitrary
la~erconiiguration provided
the lasers can be idealized as follow~.3. Idealization of the laser beams.
We need the
knowledge
oi the evolution operatorsU~(I
which refer to laser zone~ withtravelling
or;tanding
laser wavesserving
as beamsplitters.
Ii thecomplete
Hamiltonian H,,~, ofequation
Iiincluding H~
is taken into account the related microstructure of the timeevolution will lead to
U~(I
) which are in the exact versioncomplicated
functions of theoperators of
position
and momentum, of theingoing
matter state, and of the durationT of the la;er influence. Because we follow two
given partial
beams I and II we have in factU[
ii andU)
(I). Our main idealization is then that the influence of theperturbing
HamiltonianH~
over the interaction timeT in the laser zone is so small that it may be
neglected
whenelaborating
the final interierence pattern. Thi~ is to beexpected
it T is very shortcompared
with the times offlight
T, between the lasers. We assume it therefore to bejustified
to de;cribe the influence of thetravelling
orstanding
laser waves in~,)
as a momentum tran;fer at oneinstant oi time oi the form
cv,' = exp
[in) k, xi
(9)where
k,
is the wave vector of therespective
laser wave and 1)= 0, ± 1, ± 2, The
position
operator is the
only cl-number
which appears here. We havecorresponding expressions
forp,', y,', 6,',
and for the beam II.According
to our convention anyamplitude
is omitted because it does not contribute to thepha~e
shift which goe~ back toH~.
The same is the case ior thelaser
phase
omitted inequation
(9). The restriction of the laser influence to the momentum transfer which will beju~tified theoretically
below is reasonable. It is this feature which is used for the construction oioptical
beamsplitters
asproposed,
for instance,by
Pfau, Adams, andMlynek
[14] in the context of a three-level atom in a constantmagnetic
field.The time evolution of a
partial
beam i~ nowgiven by
,«~k,, , <H~,,,I(, <I(,,,Tj <,i( k , <H~,,, T,,
$ij)
= e e e e e
$ij,)
( (0)and
similarly
for~jj)
with H~~~ ofequation
(2). For~,)
infigure
we have. forexample, al
=
I,
ii(
=
I,
n(
=
I, 11(
=
0, and
k,
=k~
= k~ =k~.
The
approximation
made above which leads to the idealization of the laser zonesby
beamsplitters acting instantaneously
canformally
be obtainedby taking
the limit to very smallT and very
large
Rabi'sfrequency
J2~~=
E~
d~~ in such a way that x:=
TJ2~~
remains fixedcharacterizing
therespective experimental
set up. Tojustify
the structure(9)
assumed abovewe
give
an exact calculation of the output of a laser zone for the extreme limitT - 0 IX
= const. ). The electric field E may
thereby
begiven by E,,cos [wt
k x + ~(running
wave) orEj,
cos [wt cos [~ k x(~tanding
wave). ~ is ~ome constantpha~e.
ForT <
I/(k(
there is one common result forrunning
andstanding
waves. TheSchr6dinger equation
with the full Hamiltoniangiven by equation
(I) ha~ the formal solution~ T ',,
~ jr
))
=
jj
(- I16'~dt, dt,, Hit, Hit,,
j ~(0 ))
' ',, =,1
i >
whereby
thesimplifying
convention of section ? has not been used. We nowreparametrize
the variable~ ofintegration by
t, =A,
T andperform
the de~cribed limit.Exploiting
lim
TH(t,
=TA,
)= X cos
[p
kxi (12)
~ _~,
l
which is valid for both
running
andstanding
waves it i~ easy toperform
the sum inequation
I).Remembering
that the square of the Pauli matrix «, occurring inequation (12)
is the
identity
matrix we arrive at~
j0~ ))
co~ Ly cos (~ k x )] + I ~in Ly cos (~ k xl~
4'lo )) (13)
Thi~ form of the evolution operator is not very
illuminating
(what is the ~ine of the cosine of theposition operator?)
but it can be written in a moreappropriate
form if one representscos
[~
k x as(exp ii
~ ik x + exp [- I p + ikxi
if? and writes down theTaylor
~erie~of the
remaining
sine and cosine. In this way we find, e.g.,cos
lx
CDS (w k x)i
=
I )j ((' (
~i ~~/[ exP12 i~
(kx ~
Jl 1'4)
,)
u i
~'n,
A resummation of the r-h-s- leads us to
~
co~ Ly cos q7 k x)
=
jj
exp[2 ir(k
x ~II
(-l'J~
j,
(X (15)
, ~
where
~~ ''' ~~
2
~j~ I! (I
+ 2jr
~!
~~~~are Be~~el's functions. A similar calculation can be done for the sine in
equation (13).
The time evolution in its final forms is thengiven by
~ (0_
))
=( (
exp
[2
iii (k x p (- )"Jj
j,, ix +,, ~
+ i i '
f ie<i~«+ii<k ,~>+ e-<'~«+"'k ~~>i
i-J«/~,,+>(x)l'~f(°)i
1'?~,,~~
This
expression
makes thephysical
effect of the laser waves rather obvious. Sinceexp(ink
xicorresponds
to achange
of nk in the momentum of the atoms the atomic beam i~split
into manypartial
beams which areweighted
with Bes~el functions. For evenii there is a momentum tran~fer without excitation and deexcitation. It
depends
on the apparatu~parameter x which Bessel function and
accordingly
which momentum transier n ispredominant
inequation (17).
Note that in the extreme limit ofequation (17)
also forrunning
waves excitation with n <0 and deexcitation with n ~0 is
possible.
Such processe~ areexplicitly
removed in therotating
waveapproximation
becausethey
are far from resonancewith the atomic transition. In the limit of very short interaction time between lasers and atoms
the re~onance condition has not to be fulfilled. Seen in this way the result
(17)
makes itplausible
that for very short(running)
laserpulses
atoms could get the « wrong » momentum transfer.Having justified
the structure ofequation
(9) we ~tress that the calculation should not be mi~understood as anadequate
treatment of the influence ofextremely
short laserpulses
onatom~. This is due to the fact that for ~hort laser
pul~es
the contribution of non-resonantelectromagnetic
waves cannot beneglected.
Animportant implication
is that for too shortpulses
also atomic states which are not described in the two-levelapproach
can be excitedby
the laser waves. Since a very ~hortpulse
may not extend over onewavelength
also theshape
oi thepulse
becomesimportant.
Nevertheless the result(17)
is agood approximation
if thepulses
are not too short and it the
flight
times T, in the dark zone~ are muchlonger
than the interaction times T, between the atoms and the lasers. Its range oivalidity
may be constrained to situation~where the
rotating
waveapproximation
and oi cour;e the two-levelapproximation
areapplicable.
4. Phase shift for
arbitrary
interferometerconfigurations.
In thi~ section we will be concerned with the calculation of the
pha,e
shiit between twopartial
beams I and II caused
by H~
ofequation
(4) in an atomic interferometer with anaibin.ai=>.
miuibei
t)j
la.iei beuui.Iserving
as beam~plitters.
In a fimt ;tep theexpre;sion
(10) can bebrought
into a more convenient form if we in~ert theidentity
operator in the formexp(-
iH~,~~ iiexp(iH~,~~
t) for certain times t and make use of theequality
exp (- B ) exp
IA
) exp(B
)= exp
[e
~ Ae~] (18)
which is valid for any operator~ A and B
provided
theexponentials
make sense. We get~,)
=
e~'~~.«<
e"'~'" e'"~'' ~j,) (19)
-1
where t, :=
jj
T~, and the operators I, are definedby
, =,1
i~ e'~~"'~' k~
x e~ ~~ "'~' (20)
This
approach
i~ similar to the on usedby
Bordd [15] for the calculation of acceleration and rotation induced ~hifts in theRamsey
device for a Mach-Zehnder interferometer.The e~~ential step in the derivation of the
phase
shift is now the calculation of the~e operator~by
means of the well-known formulaexp [itH~~~~ k x exp [-
itH~,~]
~ ~' ~K,
(?~j,
.~with
K,,
= k x and
K,~
=
[H~,~. K,I.
It i~ not difficult to proveby
induction that eachK,
can be written in the formK,=i'~U;x+ V;p+W;a) (2?)
M
where
U,, V,,
andW,
are c -numbers and dr) notdepend
I)ii the a<.<.eleiutir)n a. This will turn outto be the reason for the
independence
of the accelerationalphase
shift of the initial state of theatoms. The <.-number vectors
obey
the recursion law(~
i +
£<
~ ~£,l),
(~,
)b (" ), + ~U£<0 h
(~,
)bIV,
+
I,
= F'ai>,
IV,
Ii> lW)~
Ill,
)<,
(w (v j (23)
, +1t< ii'
with the initial condition Uj, = k and
V,,
=
Wj,
=
0. In
general
the solution of'this schemecannot be
given
in a closedform,
but it ispossible
to calculate the operators I, to any wantedaccuracy. When this has been done the operator~ I, of
equation
(20) aregiven by
'., =~Lix+»;P+P;a (24)
with well defined ~L,, v,, and p,. This is the first step in the evaluation of
equation (20).
Thespecification
of ~L,, v,, and p, follows inequations (27), (28),
and (30).The rest of the calculation i~
only
arepeated
utilization of the famou~Baker-Campbell-
Haussdori formula whichgives again
exact results becau~e the operators I, have thesimple
iorm (24). In a fir~t ;tep one can proveby
induction thatl~
~e"'" e""'~
= exp
jj
n, I, expjj jj
ii, ii~ (~L, v~ v,~L~j (25)
2
-1 , =, ,»>
hold~. This can be u;ed to rewrite the
expression (19)
in a more convenient form. After the~ame is done ior
4~j,)
the interference term(~, ~,,)
i~quickly
calculated to beIA
(~, ~j,)
= exp
jj (ii)'
ii) p, a x=1
x exp
~~ jj~
jj (nl' nl'
ii) ii) (~L, v~ v,.~L~ + ~(
~L, v~
(n)
ii) +nl'nl'
?iil'n)
x~
,
~
, j =1
l,;
~x
~,,
exp~ (n)'
ii) ~L, x exp I~ (ii)'
ii) v, p~,, (26)
,
where we have reintroduced h and <. ior convenience. In realistic ~ituation, we will have k =
E~ E,,. Doppler
shift and recoil are still contained. To read off thephase
shift inducedby H~
ofequation
(4j one has to work out ~L,, v,, and p,explicitly
for the interferometer~ituation in
que~tion
and to substract from thepha~e given
inequation (26j
thecorresponding expression
for a=
0, w
=
0, and
R,,
I,,,,,=
0.
Equation
(26) i~ the main re;ult of this paper. It may look ,omewhatcomplicated
but thi, I;only
due to it~ greatgenerality.
Inparticular,
thi;form I; well-suited for the
application
ofalgebraic
computer programs. Afterspecification
of themagnitude,
oi a, w, andR,,
I,,,,, it I; now in theexpression
(?6) where one ha; to makesure that the iteration
(23)
above has led to therespective
contributions of the desired order. Anexample
will begiven
in the next section.We remark that in two
physically important ;pecial
cases closedexpression;
forI, can be
given
from which therespective
~L,, v,, and p, can be read off. If the rotation vani;he;we find
I, =
(k,
Ii (CDSIt,
N )i,,, I,,, +(k,
Ji (N SinIt,
N )I,,,P,,, ++
(k,
ii IN~(cos It,
N))I,,,
a,,, (271
where the matrix N is defined
by (N~ )I~,.= Rot
j,,,,.
Surely
this is a formalexpression
definedsolely
iia itsexpansion
which containsonly
even powers of N. In this sense we need not to care ior the exi;tence of N.In absence of curvature the result for the operator~ is
t,
", =
(k, )I (exP It,
° )I,,,-I,,, +~ (k,
)I(exp it,
O )I,,, p,,, +(k, ii
O- ~(exp it,
O )I,,, u,,,(k,
)I (°~ exPIt, °I
Ji,,, an, t,(281
where the matrix O is deiined
by Of,,,
= ti,,,~ w~.A short discussion oi our result (?6) will close this section. Fir~t, it is obvious that the accelerational induced
ph»e
shift iscompletely independent
of the initial state~,,)
since the acceleration a occursonly
in the iir;t row ofequation (?6).
Theindependence
oi this shift irom the initial momentum oi the atom; ior the atomic fountain was notedby
Kasevich and Chu [3).Second, there are ;tate
dependent
;hiit; as the la;t row indicates. In order to get animpres;ion
oi their iorm we a;sume that the initial ;tate is
(a)
g(G)
where(G)
denotes a Gaussianwavepacket
(in momentum;pace)
with mean momentum(p)
and half width« in all
directions.
Setting
A:=
~
(ii)~ ii) ~L, and B:=
jj inl'
ii) v, wequickly
iind that the statedependent
part oiequation
(?6) contain; apha,e
iactor and a iactormodiiying
theamplitude
exp
iB (p) ~
AB)
exp~- fl
B~~~ (29)
2
~
8 « ~
We will see in the next section that ior the
Ramsey
interferometeronly
the termB
(p)
contributes to thephase
shiit to the desired order in ourapplications
since the term A B will beentirely
oihigher
order.5.
Application
to theRamsey
interferometer and to the atomic fountain.It i~ now
~traightforward
toapply equation (26)
to two~pecial
devices, theRamsey
interferometer and the atomic fountain. As
~pecification
of theexperimental
set up we mayassume
flight
times T of about one ~econd. The order of the initialvelocity
v, the laser wavevector k, and the atomic mass M are about v m s~
(k 10? m~',
and M 10~ ~~kg,
respectively.
If oneapplies
such an atomic interferometer ior the mea~urement oi the influence of the earth's acceleration a, the earth's rotation w, andspace-time
curvatureRot
o~, of the
earth or of some lead block we have the
magnitudes
a 10 m s~w 7 x
10~~
s~',
andR,>I,,,,,
5 x10~~~ m~~ respectively.
Thesespecifications
make it reasonable to work out~L,, v,, and p, of
equation
(24)by
means of the recursion law(23)
up to second order in the rotation and iirst order in the curvature. In the exponents oiequation
(26) we will take intoaccount
only phase
shifts which are at least of the order of 0.I rad. In detail we obtain~P,If
'f~~<
~f +'/
~~< ~ " ~f
'f
l~< ~ "~ ~" If +
?~
~~~'~~"
f"«< ~~<~«<.(30)
The first apparatus we will consider is the
Ram.ie)>
interfieiometer
which is described in some detail in reference[2].
Four our purpose thefollowing
facts are sufficient. First, in theRatnsey
interferometer
essentially only
two atomic states are involved and the laserfrequency
isclosi
to the transition
frequency
between these states. It is therefore allowed to use the two level model for atoms as we have done in thepreceding
sections. Four lasers serve as beamsplitters,
and their wave vectors are
given by
k:=
k,
= k~ =k~
=
k~.
The times offlight
asthey
appear in
equation (lo)
aregiven by
Tj, =0,Tj =T~=:T,
andT':=T~.
We introducedT and T'
just
to be in concordance with the usual notation in the literature. We will examine thephase
shift between thepartial
beams I and II offigure
as it is inducedby H~. According
tosection 2 these beams are characterized
by
the sequencesp,
y~p~
n~ and6j 6~
6~p~,
respectively.
From this we can infer to the sequence~(n))
=
(I,-I,1,0)
and(nl')
=
in,
0,0, Ii. Inserting
these values inequation (26)
andusing equation (30)
leads after somealgebra
to theexpression
(~i ~,j)
= exp
it k
aT(T
+T')
a (k x w )T(T
+T') (2
T +T')
+
( (c~ Rj,
I,,,,,
ki
k,,, + 3 (k xw)~) T~(2
T + 3 T')+( (((kxw)xw).a-)Roij>~,lia,,~j
~T(T+T')(7T~+7TT'+2T'~)j~
~
(~u (eXP (i
[C'~Rut
ii,,,,'f k,,, ((k x ml x w ).xi
T(T +T'))
exp
)
i(c.~R,,
~,~,~~
i~
p,,, 3 ((k x m i x ml pT(T
+ T~) 12 T + T~)+ 4 p (k x w
T(T
+T')j~ ~o) (31)
which contains
only
thephase
shift due toH~
# 0.Again
we have reintroduced h andc. The first order contribution due to rotation was measured with the
Ramsey
interferometerby
Riehle et al.
[8].
For the measurement ofspace-time
curvature the device of Sterr et al.[16]
may be better suited because of the
longer
lifetime of therespective
excited state in,' ,I+Il ,' ,, 3
,
beam I,,'
, T,
2 '
,,'
,'
beam
II T,' i
Fig.
2. Theinteriering
beam; in the atomic fountain device under~tood ah an effective two level system w>th running la~er pul~es. The horizontal line~ represent the three laser pul~e~. The atom~ areinitially
in the lowerhyperfine
level. Thehigher hyperfine
level is denotedby
dashed I>nes.T i~ the flight time between the laser
pulse~.
magnesium.
As announced theproduct
of the two vectors A and Bentering equation (29)
is ofhigher
order in this case. A discussion of theleading
terms of thephase
shift wasgiven
elsewhere
[10].
The order ofmagnitude,
however, is the same as for the atomic fountain which will be estimated below.Next we turn out to the atr~uii(.
fr~ifiitain
a~ it wasdeveloped by
Kasevich and Chu[3].
In thiscase 3 atomic levels are
involved,
twohyperfine
levels and one excited level. The atomic beam issplit
and recombinedby
three laserpulses.
Eachpulse
includes twocounterpropagating
laser beams(here
denotedby
a and b) withfrequencies
v~ and v~.They
are tuned in a way that theirfrequencies
areslightly
smaller than the transitionfrequency
between the twohyperfine
levels and the excited level so that a twophoton
process induces transitions between thehyperfine
levels. The momentum transfer dor such a process is k~k~.
Thesplitting
of the atomic beamis
given
infigure
2.It now seems that our scheme is not
applicable
to thisexperiment
since it isonly
valid fortwo level atoms. however, as
pointed
out in reference[3]
theparticular tuning
of the laserpulses
makes itpossible
to treat thehyperfine
states as an effective two level system with aneffective momentum transfer of k~
k~.
This observation allows us to use the results of this paper for the calculation of thephase
shift in the atomic fountain apparatus.We will model the situation
by
threecopropagating running
laser beams with wave vectorequal
to the effective momentum transfer kk,, k~.
Fromfigure
2 we can read off thedefining
sequences of operators for theinterfering
beams. For beam I, it is6, p~
cv~ and(n))
=
(0,1. 0).
Beam II is describedby p, y~p~
and(n)')
=
(I,
I,I). Setting T,
= T~ = T and Tt, = 0 theapplication
ofeqution (26)
leads us to thephase
shiftfi~,2
(4r,14rtt)
=exp(I(-k.aT~+3a. (kxw)T~-mRoion,kikmT~
~~P
( [(3
(k~ ") ~ "
l'
P ~'~~010
n<
ki
Pn<) T~ 2 P(k
~ "~l) ~°) (3~)
The
state-independent
part of thephase
shift can bedirectly
read off from thisexpression.
Assuming
that the center of mass state of the atoms i~ a Gaussianwavepacket
we canapply equation
(291 to calculate thestate-dependent phase
factor. For the atomic fountain it issimply given by
theexponential
in the last line ofequation
(32) if onereplaces
the operator pby
theinitial mean momentum
(p)
of thewavepacket.
In the case of pure acceleration
ii.e.,
R,> I,,,,~= w I =
o)
thecomplete
shift reduces to the first term inequation (32)
and agrees with the result of reference[31.
The last term inequation (32)
describe~ the well-known
Sagnac phase
shift. Becau~e the earth's acceleration alter~ themotion of the atoms
substantially
it is modifiedby
the ~econd term inequation (32).
Thestructure of this modification is
easily
understood if one considers a classical freeparticle
withinitial momentum p,,. In an accelerated frame of reference its momentum is
changed according
to
pit
pi, Mat. Thereplacement
of the mean initial momentum(p)
in theSagnac phase
shiftby
MaT reflects this factby leading
to correct structure of themodifying
term. Corrections to theSagnac phase
which are ofhigher
order in the earth'sangular velocity
are the first term in the last line and the first term in the second line. The latter can be understood in the same wayas the modification to the first order
Sagnac phase
as a consequence of the modified atomic motion an the accelerated reference frame. The influence ofspace-time
curvature on atoms,I,e., of the variation of the acceleration over the interferometer, can be
spl>tted
into two parts.The last term in the second line and the second term n the last line of
equation (32)
areconnected to the motion oi the atoms whereas the last term in the iirst line is
only
due to thenon-vanishing
relative accelerationexperienced by
twopartial
beams which areseparated along
the direction of the laser beams. Noteagain
that the third line does not contribute to thephase
shift to the order under con~ideration if the atomicwavepacket
issufficiently
localized inmomentum space.
6.
Measurability
of the influence ofspace-time
curvature on quantum systems.A detailed discussion of the
measurability
oi curvature eifects with theRamsey
device wasgiven
in reference[10).
It is necewary to restate here some remarks on thephy~ical
value of themeasurement of the
curvature-dependent phase
shift. Ingeneral relativity
the acceleration a i~tied to the
particular
observer who is accelerated with a.Loosely speaking everything
seems to be accelerated with acceleration a in its frame of reference. Thisimplies
that in the frame of afreely falling
ob~erver there is no acceleration at all. Acceleration is a non-inertial efiect, tied to the observer, and is not connected with the structure ofspace-time.
The latter is describedby
space-time
curvature which is present in any frame oi reference. A curvedspace-time implies
arelative acceleration between
nearby
observers, I.e. the distance between twofreely falling
ob~ervers
changes
and the acceleration of twonearby
observers which are held in a iixeddistance is different.
It was
suggested
that one can use the atomic fountain as agravimeter [3).
This is true in a twofold sen~e. If one varies theposition
of the whole device in order to detect the acceleration of the earth at two diiierentpoints
this i~indirectly
also a detection of~pace-time
curvature since the variation of the acceleration over space is measured. This is, however,by
no meansthe demonstration of how
space-time
curvature acts on atoms. Rather, the classical apparatuslead block
/
atoms
"2 mi
lead block
Raman laser Ram» laser
Fig.
3. A ~cheme of theproposed experiment
to mewure the influence of thespace-time
curvaturecaused
by
two leads blocks on the atomic fountain. Thebig
arrow~ repre~ent the three Raman lawrpulses performed
with twocounterpropagating
I»er beam~ with frequencies w and w~.
The point O denote~ the