/:.mer f>liysic:s, \.'bl. ·9, No.· I; 1999 •. 1'/1· 26--, . .i?9. Orfi:im,I Te.ti Capyrig/Jt © l~.99 by A.irro. l.Jd, ..
Ct1piri~l1t©.J 999 (1y MA 11 K "HllY_1':(ll/11terpi:rii•rlkt1" f R11.rsi"~·
MODERN TRENDS,
========IN LASER
PHYSICS
Polarization of Free Electromagnetic Field and Electromagnetic
Radiation in Quantum
Domain
A.
S.
Sbumovsky
Physics Depart111e11t, Bi1ke11t Uiriver.~if)~ BUke11t, Ankara; 06533 T(trkey
e-mail: shumo@fen.bi1kent.cdu.tr
Received July 6, 1998
. . . . ·~ ',,,l.-;:: . .
Abstract-:Il
is
shown that conventional disreg;ird.of lo.ngitudinal component, which always e~ists iil radiationofa localized source, can lead to some. qualitative errors in the description 9f polarization within the quantum
domain. A new definition of gcneral\zed Sto.kes pijrameters an:d corresponding operators is discussed.
The two main qbjectives of this pa.per are, on the one
harid. to .show that conventional definition of polariza-tion based on.the nopolariza-tion of completely transversal free field is not valid in the quantum clomain, and, on the other hand. to demonstrate how the. spin states of pho-tons may be employed to determine the generalized
Stokes operators. In view of our aim, let us begin with a brief discussion of the conventional definition of pol~rizadon; both classical and quantum. ··
PolarizatiOn is usually
defined
as a physical measureof Jransversal anisotropy of light [l]. This. definition is based on consideration of a completely transversal electromagnetic field provided by a. solution of the homogeneous wave equation
V2.A· . . l
a
2A-2-2· = 0.
C
c)t
(1)
For such a solution; the electric field E = -c1iJA/at is always orthogonaltothe magnetic field
B
=V
xA.
and bot_h Eand
B are orth<>gmial t.o the ditecliortof
prqpa .. gationk This. means thatthere is no axial symmetry ofa transversal wave with respect to k; i.e., the wave is po1arized. Be.cause of orthogonality of E and B, either of these two vectors can _be chosen to.spec{fy the polar .. ization. The electric field
E
is
usually rise'd[lJ.
The simplest way to d~termin~ t.he quantitative mea-sure of polarization consists in consideration .of
tlie
Hermitian tensor of polarization, which is defined as having components [2]
Ppp· = EpE; = E011Etw, (2)
which are
the slowly varying bilinear formswith
respect to the field amplitudes
E
E ( )(
+ . ) .
.;..j«jf=. o t E1 _ 1£1.
e .
H~re Ep is the
unit
polarization vector.Because
ofcom-plete ttansversality of the wave, index ~ takes only two different values;·111e rank 2 Hermitian tensor (2) can be determined by three reai parameters, forming the con .. ventional
set
of Stokes J?arameters. The notationfor
theStokes parameters is µnf ortunately not uniform. The• definitions given in [ 1] in temis of the circular pofa.riza~ lion basis read
s~ =
IEt·Er+IE:·Er
;f
=
2Re[ (E! •.E)*(E! ·
E)],(3)
s~ =
2Im[(:t: ·
E)*(E: · E)]~I 12 I
12.
s;
=
e!'
··E -
E~ • E .Here E:1: == (E 1
±
iE2)/Ji
and (s~)2=
.(s;>)2+
(:s{)2
+
C~{J2. The "¢rameter s~ in (3) measures· the relative. intensity of the wave, the pararrieter
s;
gives
thepre-ponderance of positive helicity
over
negative helidty,,and the parameters
sf..
s;
give the phase information fo · terms of the cosine and sineof
the phase differencebetween two circularly polarized compo.nents.
The quantum counterpa.rt of (3) is provided by con"'.. ·veritiona1 Stokes operators that can be obtained by stan~
dard quaritizaiion of free electromagnetic field in the
representation of "plan·e ph<;,tons" (the photons having given energy and linear mo,mentum) as follows [3;, 4]:
•
C . .+ + ·. S0 = (a+a~ + a_a_), •. ·+ + . Si=
(aC:a++a+aJ,(4)
·c. ... + + . S2 = -1(a_a+-ll+lL),Apart from a factor of 2, the operators
Sf ( I=
1, 2,3)fomi a representation of the .SU(2) subalgebra in the
Wey1-Heisenberg
algebra
of plane · photons. SincePOLAIUZA TION
OF
FREE ELECTROMAGNETIC FIELD 27two
operators cannotbe
measµred atonce.;
Inview
of/tp.e stanqard interpretation of the S.tokes parameters (3),
:\vbich
are ex~cted
tobe the
averagesof the· ope1.1:1-to~
!:{4),. this Jl)eans th.at
the..
cosii-.e· and sine ·of the phasedif-)Jerence between two component~, with opposite
helici-\ties
cannot be measured simultaneously.?t Let us now note that the quantum ·electrodynamics
Hr.eats the
polarization as
a given spin state ofa beam
of{photons
(see,
fot example, [61). Thespin
of aphoton
is]¢own to be equal to l. Th-µs, it may existin the states
?witll
projectionsm=
0, ±l. In the case of a free electro-ifri~gnetic fie]d, the ;sta,tem
=
0 is forbidden ~ecause of/i'#i:#ripJete tr~sversality of thefield [6]. Atfirst glance,
mus·means that there "is no ·contradiction between the
f~ifinition of polarization via spin states
pf
photons~dIWiough the use of
Stokes
operators (4).I1Ji1?in
seems relevant to ask: has anyone observed.free/iji~~tromagnetic field ?Asfaras
we
know. the. fields:thatI:ti#fe
been detected so far have been. emitted by some!I/~:t~rce.
· " -!'Jn
the CJ:1.se of i!.-real radiation. t_he; &pin: ·s.tate j-=1..
)0: cannot
b~, e;itcluded a .. .priori In fact, ·a phpt1,:m,4 ~~ated
l:ly
-a dipole transition gains the spin state~;,,·V,7.i.).:.w.ithj=je~i
8=
land any m b~tw.een-lie-j81
and~~
\j~I,
where}e...:. j.1 i~ the··.difference ·pf tbe.-angulur :·.:··.-¢.ita. of.excite.d and ground levels [6, 7]. Moreover;···foal dip.ale.radiation, ejther electric. ot
magnetic,
·,s:.,has, a longitudinal, lineady polarjzed
compo-foge.ther with two transversal components [81.
ifhe·intensity ofthe 1ongituclinal co~ponent fa_lls
( ·:clistance
quite rapjdly, it_ is cus~omary to ignore , .. :·:(fa.r.·zonewhere the radiation field .isapproxi-?,lj/tfcinsverse to t~e fadius ve.cto~. cert~ered ·a1 trye :c:e::To be sure that this. approx1mat10n 1s also vahd
._ '~'.qtiatiturndomain; it is necessary to gerierali:ze the
~~¢ffotion
of polarization
and
ex~ine
carefullyhow
'''")l~Jigitudinal
contponent can
influence the qmmtum:1, __
:fhiyi_pr
of two transversal components. This means/<'i"'-'"'\ve
have to consider polarization of a dipoleradia-'.;~f
!'lrt
arbitrary. s~acec:-time point thatc.ui
possessilii#tion of polaitzation or can be locatedin any
•_::;:.-·=-.· :.
'"611r first
step, wehavei:o consiruct cdttespondiii.g'Hzation
of classical tensor of polarization [9J.i~ssjcal
radiation of any localized source is'"ij·by
the rnu1tipole·fie1d[8]
).=:\:::\-:· ..
"f~J~
=
L
[a;m;1Jk)Aj;,,1(k) + c.c.],jmkA.
jn,u
Jfii.(~)
= ikAim;.(k), the indices;/, m show the{lfj¢
multipole field, the indexX
specifies theVol. 9 No. l 1999
type ofthe rnultipole field (either electric or. magnetic),
and
aj,r(}._(k) =
I
r2drdnE;/1,. •
E.Here the coeffidents Ai11,'J..(k) are definecl"by the standard combinations of vector spherical hannoriics and Bessel functions [8]. Thus, the .general definition ofthe classi-cal
tensor
ofpolarization
reads [9, 10]P,;,1;,{>.., k)
=
[Ej,.,i(k) · EJ* [E!,,n.(k) ·EJ.
(5) Consider the case of dipole radiation when j=
1 andm
=·
0, ±1. There is no loss in generality in choo!:!iogfixed A
and k,but
it enable1:_1us
to fairly simplify thenotatio_.s. Then, nine components of the rank ,3
Hel'llli-tian tensor (!,) a~ determined by five real parameters
which, must specify three intensitijes.-arid the ··phase
dif-ferences., Amm' such
that
A+o
+
Ao-+ A-+ =Q.
To.establish
contact
with conventional definition (4), let us choose the generalized Stokes parameters asfol-lows:
I
S =
~
IE*
•E]
2 ·· 0 ,£,... mm ;=-l
s1 = 2Rel(E; · E)*(E5' · E) +
(Et ·
E)*(E~ · E)+(E~ ·
E}*(E; · E)],s2.=2Im[(E! ··E)*(E; ·
E)
+(Et·
E)*(E! ·E)
(6)+
(E~ · E}'"(E; · E)J,
S4 =
IE:. El
2
+IE!.
El
2-21E:.
El
2Just as. in the case of conventional Stokes parameters
(3), the parameters0measuresthe relative intensity, the
parameter s3 measures the preponderance of positive
helicity over negative helicity, and the parameters s 1 ~ .'1'2
give
ihe phase infont1ation. Additional parameter s4measures the preponderance of circular polarizations
over the linear polarization. Under the assumption that
IE0J
=
0, the set of generalized Stokes.parameters
(6) isreduced into (3) with
s
4 - - -s;r
The generalized Stokes operators. can be obtained
from (6) through the use -of quantization iii tenn.s of
28 .SHUMOVSKY
given energy and angular momentµm. The definitions read [9, IO] . +
S,
= <~.+~·), S2 = -'-i(t - ~+}, S3 = fi~-lL S4 == fi++i(-2n~, (7)where ii 11, ;=:
a;~,,,
is the number operator of the dipolephotons and
(8)
The. operators (7) and ( 4) are very different in their algebraic structure. In fact, the operat.ors (7) arethe
lin-eaj"'combinations of generators of the S0(3) subalgebra inthe Weyl-Heisenberg algebra of dipole photons [12], while (4) represents some SU(iJ algebra. The mos.t important fact is that [S1,
S
2] = O so that <;orresponding physical quantities can be measured. at once. Sinc!:l[Su; S0] = 0, their measurement is compatible with simultaneous measurement.of the total photon number.
At .the same time, [S1 2., S3 4]
*
0, which implies thatthere ate uncertainty.
re]atib~S
between SI, 2 and$.J.
4,•To underline. the difference between. the physical
quantities described by (?)and (4). consider theaverag~ ing with respect to a state of the .radiation field in which the longitudinal component m = Q is in the vacuum state. Then (S1}
=
(S;-)
(/.= 0, 1, 2, 3) and (S4)
=
(S0)=
( S~ ). Thus. both sets .of the .. Stokes operators (7) and (4)le:ad to the same set of S~okes parairieters,
representedby the averages of corresponding operators. Consider ·now.the variances ·of the operators S~ and S1• It is a
str~.ightforward. matter to,arrive
at
the. relationsV(S1) = '2Re V(a~a+) + 2( (ii+ii_}
-l(a~a+>l
2J
+
2(S0)+
2Re(a~a+.),V(S~)
=
2Re
V(!1~a+) + 2( (iz+iL)-l(a~a:1-}j2)
+
(Sq}.
Here V(X) ::!
(X":} -
(X)2 .. As can be seen from the aboverelations; th~ ptiysical quantities describe.d by the oper-ators (7) J.mdergo much stro11geni)1antum fluctuations than those described by the operators (4)because.ofthe influence of the component in= 0 even
if
it is consid-ered in the vacllum state. Similar results can beobtained for
S
2 ands;
iis well. As a particular exampleof considerable interest, we ,now inve~tigate the case
when two circularly polarized components are in the
coherent states
lex+)
andlo:_)
such thatlex+
I
=Jtt-1
=
l<xl.'
The h;,ngitudinal c.omponent is still supposed to be
fo.
the vacuum state .. Then, the above reJ~tions take
the
fonn· . ~ (Si) = (S2) = 2lcxl~cosA, V(S1)=
2lal2r2+
cos.i.\), . V(Sf) =2lal
2, ... _!~~~·::. .Here .6. ;:· arga+ -arga_. These relatiQns snow the quaH itative difference of fluctuations be.cause of the pres;
ence
of phase dependence .in V(S1),This
difference is veryimportant in
thequantum
domain whenltxl
2 "" 1.i
Thus we see that the description of pol~rization:
based on the use
of
completely transvei:sal free electro-magnetic field can lead to a wrong picture of fluctua~ tions in the quantum domain, at any distance, including the far zone, as fa, as the radiationof
a localized sourceis. considered~ The contribution of the longitudinai ,componerit can be taken into account through the use of
generalized Stokes operators (7). The above results cari
pe
very
important forarty
polarization meaflun;ment with small intensities. An importl;lnt example Js pro~ videdby
the atoniic physics experiments with colJerent and rionclassicaJ light focµsedon
th~ cJetection of spin;polarized atomic States via polarization rotation of
probing light [J 3-'16]. .
Another example of high importance is provided 'by the.quanturil:-phase problem. Perhaps the most impot.c tant result ofseventy years ofefforts to find a quantun1 object describing universally the pha,se properties of light, is that such an object does nqt ~xist a,t all [17}; lf! addition to the ope111tioilal phases determined by inter.; action
of
light with different mactoscopfo detectingdevices [17]; there cC>uld be s9me inherent quantum phases; obtained by photons in the
process
of gei1era~tion. [1 SJ: The azimuthal phase of spin of a p,hotorii introduced
by
the cosine and sine operatorsI18,
·J9]····m Ket> a,+ K
·~ = -(0+ 0 )
2 ·
·
:::-S2 ·
1,(9)
I( '- + K.
~ =
2/~-~
)=2S2,where
K
= (('~~+)t'(i, is very interesting · inherentphas~ with qui~e nontrivial quantum ptopeities (also see [:20]).. The existenc~ of w~ll~defined eigenstates and e1genvah1es of the operators (9) in whole Hilbert
space
[12] makes it possible to build a bridge between thf!
radi~tiori phase (9) and the results of the method baseq
on the
use
ofra:dius-integrated quasiprobability distrh l,utioi1s, describing the homodyne .detection of phase[21-25].
.·
· Thus, it is shown that the contribution of longitudi:
nal component.into the qoantuin fluctuations of
polar.;;
izatfon js important even if this component is taken iri the vacuum state (for example, in the case of adipole
Ji/>:\'':'> · POLARIZATION OF FREE ELECTROMAdNETIC FIELD 29
/i~f(fiK~in the far zone). Since the free electromag- 12. Shumovsky, A.S., 1998, Acta Phys. Slovaca (Special
fcj¥gtl~
§elddoes not contain the longitudinal component .Issue on Qua,Jtum Optics a11d Quantum I,ifom1atio11),fff}Ifj,fioi}Hhe
use. of 1!1is notion i~ the. quantum domain 48, 1.f(seein.~JQ.
qe a qmte nsky approximation. 13. Bacon, A.M., Zhao, H;Z., Wang, L..J., and Thomas, J,E.,(f1J,yJl::{;pif'yariances
V(S1, 2) might be measured through 1995,Phys. Rev. Lett., 75, 1296.)fttieluse
of the eight-port operational scheme [17, .26]. l4. Georgadis, N. Ph., Polzik, E.S., Edmatsu, K., and Kim,lla.~o,f:,~~:
Principles of Optics, :::§;;E~~~~~==::~:.::· :::~:::::
:ifilzi?~filtdau,L,D. andLifshitz, E.M., 1971, Classical Tht!ory 17, Noh, W., Fougcres; A., and Mandel, L, 1992, Phys. Rev.
I\?ti/:NJfFields (Oxford· Pergamon) A 45 242
lzl\!l~l!)i;4¢~.J;Mi
and Rohrlic~? F.,u59,
T~~
Theory of Pho- 18. S~u~ovsk~, A.S., 1997, Opt. Commun., 136,219,21iat\(,)iifi!rifcmdE
ectrons (Rea mg, A: Ad ison-Wesley). 19. Shumovsky,A.S. and Miistecaphoglu, O.E., 1998, Opt.J&{t{)/£.ufi;A.
and Sanchez-Soto, L.L., 1993, Phys. Rev. A, 48,c
1 146 124~~~f
;(;;:~b~;~tetskii,V.B.,
Lifshitz, E.M., and Pitaev.skii L.D., 20.S~::~:~ky.
A:s.
an.d M ustecapl1oglu, O.E., 1997, Phys.' ,J~~i{Quantum Electrodynamics (Oxford: Pergamon). Lett. A, 235, 438.
0it };Q~nqon, E.V. and Shortley, G.H., 1987, The Theory of 21. ~;ndilla, A, and Paul, H., 1969, Ami. Phys. (Leipzig),
o/!b}W:\W*t9ftM:Spectta
(New York: Cambridge Univ. Press) .. · , 232.i1
itz\};!il~~ards,
W.G. and Scott, P.R .• 1994, EnergyLevelsbr 22. Schleich, W.P., Bandjlla, A., and Paul, H; 1992, Phys.)i.:i;;l,i;i,';faAtoinS ciitd Molecules (New York: Oxford Univ. Press). Rev. A, 45,. 6652,
tfgfiifrJi~on,
J.D .• 1975, Classical Electrodynan1ics(New
23. Freyberger, M., Vogel, K., and Schleich,W.P.,
1993,%~}t:;J;'.):'qrlc:Wiley).
Phys.Lett. A, 176,41.iJ,t'.~.foi?$tiµ'movsky,A.S. and Miistecapboglu, 0£, 1998, Phys. 24. Freyberger, M. and Schleich, W.P., 1993, Phy:.. Rev. A,
. ..}J/1',·~tt,,
80,1202. 47, R30.n}oi!ishtnfovsky, AS. and Miistecapl10glu,