MICROSPHERES
Chapter 4 Chapter 4 Cavity QED with High Q Whispering Gallery Modes
4.3 Model for the Interaction
33
plificd into the follo\virg foriri:
Using the mode function $ ( r ) , the atomic dipole couplir~g rate (vacuum Rabi fre- cluerlcy) for a single atom is rewrit,ten as y ( r ) = y,n,,$ (r) with g,:,, = b/3~~2r.i/4xi/, so that go = y (a) is the value at rhe surface of the sphere (though the LVGhI field m;rximum is ac:tually inside the sphere close to t,he surface ;is shown in Fig. 2.2, t,he rrraximurn vahie accessible to the atorriic vapor, yo/27i
-
20 !vl\/IIiz, is right at the surface). The qilaritityplays the role of an effectjive susceptibility for the atornic sarnple in its interaction with the WGkL The Fomier transiorrn of the niode function,
4
(k) = ( 2 ~ ) - " i 2 d3r$ ( r ) exp (-ik. r ) (4.5)is normalizeti such that V, = j"d% j jd (k)j2 =
j"
d"r14
(r)j2. Because it is extremely diRcult t,o evaluate ~ ( w , , ) exactly for the actual furlctions( 4
( r ) . p ( v ) ) , calculations have been performed numerically for sirr~pliticd approxiniat,iot~ to $ ( r ) external to the rriicrosphere (e.g., $ (r)-
exp (-ZT(T - a ) /A) exp (-82:18i) exp(im.4)
as a11 ap- proximation to J!, (rj .Y hjl' ( k r )Y;,,,
(H, 4;)). Via nurnc:rical ir~tcgratiori~ transmission functions T ( w L )=
it (wI,)l2
are folrrld that are in quar~titat~ive accord with t,lic rilea- sured spectra for low Q5
Qo = 5 x 10') but whic11 deviat.e frorn t,he observations for Q2
QQo due to a near abserlce of nai.iout feutures of iciidth-.
7.35
4.3.1 Ansatz for the Three Contributions to the Susceptibil- ity
Nonetheless, tliese calci~latioiis rrrotivate an ansats which takes
with m d
+ ivt + N,
=ivI.
The first, cornpor~ent in Eq. (4.6) is physically n~otivat.ed by noting that there must be a Doppler-broadened responsedue to velocity conlpor~ents tangential to the sphere in the direction of circulatiorl of the mode. In additioil, due to the geometry of the 'VZTG&i, there is also signific,mt transit broadening due to resicience tirnes of only 10-'r for lllotion along the radial coordi~iate to 10-'7- aloi~g tiicr Go direction (for which there is no Doppler broadelring).
TfiougI~ any giver1 at,ornic trajectory will yield a colnplicated funct,iorr of both of tliese mecharlisms (as in Eq. (4.4)), in Eq. ( l . G ) , a transit broacleneti component of H'AXM Sut/27r -- 25 hh1Hz is simply added on; correspoi~ding to a lir~ear traject,ory of lengtti I , -- through the mode.
The solid lines through the dat,a of Fig. 4.4 are based on Eq. (4.3a) with the nnsatz of Etl. (4.6). For Q
5
Qo, orily the first t,wo coniponerrts with fld = 0.75 i0.05 arrd
,Vt
= 0.25+
0.03 are needed in order to explain all t~aces. From t,his, it is inferred that tlie Doppler-broadened set of atoms act only as a broad ahsorber (since Awl < ti<
&,I) and that the absorptive dip at line center is accounted for t ~ y tlre width A w t ? implying tluit cavity geometry is a doniinarlt factor below Qo.That is, the geonretry of the cavity correctly accour~ts for t,he coexistence of both transit ant1 Doppler broaciening, where of cowse the sirnplc siurl of c o n t r i b u t i o ~ ~ stlggested phcrtor~~enologicdly in Eri. (4.6) is more properly ir1tt:rpreted as an iriterplay of frequency scales as in Eq. (4.4).
By contrast, for Q
>
Qn it is essential to include a small componmrtr\i,
0.015 ofatorris which respond with their natural linewidtl~ 7 (the inclusion of which does not, change the quality of the fits for Q
5
Qo). In fact, this componerit now completely determines the properties of the narrow absorptive feature since n<
(Alit, A&),).hlthougll the need for this small s ~ ~ h s e t of atoms moving sloxvly er~ougii and iri direc- tions s~~cicil that t,liey are ~ieitl~ev itppreciably Doppler- nor transit-broadeneci is thus operationally motivatedl their existeui:e is also supported within the context of other measurements near dielectric surfaces, as, for example; in the work on Doppler-free evanescent-wave spectroscopy [69].
This sirriple model also allows the issue of the relationship of the qxantity q of Fig. 4.5 to the actual ernpty cavit:y Q to be addressed. For Q
5
Qo, t,he relationship q N Q holds; with the inferred erripty cavity transmission showri as the dashed trace in Fig. 4.4(a). I-Iowever, for(2
> Qo, the t,wo broadly absorbing components in (*GdTd,a)
(which account for niost of the atoms) significantly alter the lineshape relative to that inferred for the empty cavity with q 1.Q/2.5 for Q = Q2 and tile peak t,ransmission of the cavit,y reduced by a factor w 4.5. I11 fact, within the cont,ext, of the ar~satz (Eq.
(4.6)) for
xA;
tlie si~b-nat,ural widths in Fig. 4.5 are an artifact of ho\v tliese differerit contributions (each of width2
y) corribine t o produce T(ur,j, as shown by the solid curve in Fig. 4.4(b).Finally, the result,s of this plienornenological rriodel are summarized in Fig. 4.6(b) where the inference of the effective at,onlic susceptibility X , in Eq. (4.6) is shown. The pwanleters
(;qcl, %; !XTG)
= (0.75: 0.25,0.015) wl.iicli best fit tlle data across the whole range in Q are ilsccl. Note that&Gc,
t 1% - ,-iN,
N 1, which agrees rather nicely wit11 the~~revious estimate bitsetl upon V: a ~ i d p. 11ite.restirigly) tlle profile of Fig. 4.6(b) bears a striking rescrnblallce t o those seeri in nltral~igh resolution trilrlsit broadening-lirniterl molecular saturation spectroscopy 1701.
4.3.2 What About the Extremely Narrow Features?
2 . .
.
Because
iV,j'nTr
N 10- , it is perhaps riot surprising that aiternpts t o sirnplify the full irit,egral of Eq. (4.4) failed to proviiie an accurat,e accourltirig of tbe narrow cornponenit,intracavity photon number
I .
-100 -50 0 50 100 detuning [MHz]
Figure 4.6: (a) Dependerice of the size of the narrow absorption dip on intracavit.j- plrotorr nurrrber in the low Q regime of Fig. 4.4(a). (b) The at,orriic response
/X
(~j,,);'(norinalized to inrity) inferred frorri the phenomenological model discussed in the text.
[see Eq. (4.4);.
of ~ ( w , , ) . Nonetheless, even assumirlg a11 exact evaluation, there are several niecha- nisms wluch could protluce narrow features and are not accoimnted for in Eq. ( 4 . 3 ~ ~ ) . First, as Q increases, a greater percentage of tlre counter-propagating ( q I 1 ) -m) mode is exciteti? with a corresporlding increase in the possibilit,~ for iritracavity standing wave structure [7l! along the direct,ioil of nrode propagation. Sirch structure is capa- ble of prodiicing narrow feat,urcs by isolating thc slow atom cornponerts in a thermal
2 ' 2
gas. In addition, tire exp (-6 /0,) dependence of tlie mode frinction in the transverse direction is strictly only valici for the rn = 1 mode with 0: N 2/1, and, as one rnovcs amray fro111 m = 1, the WGX dependence on B develops ailxiliary rnaxirna 1721. Sec- ond, p (v) may depart fmrri a 1\4axuwll distribution, especially for t,hose a t o m ~vit,h
> 0 wkiich are leaving tlre s~ufacc. 023vio11sly, a distributit~n ~a4iicl1 was peaked at, lower velocities or which favored directions orthogonal to tho direction of propagation of tire node ~vould lead to ~iari-ow featnres. Firtally, possible atomic level sllifts [73]
(6, * 6,
+ A)
and rriodificatiorls to the xvitltli (y -. 7') due to elrhal-rcettient or iniri- bit,iori i74, 75j of radiative decay iri tl-re vicinity of the sphere's dielectric boundai-y liax.1: not beerr taken iriio accoinit,. Tlris urouid he extremely difficult to tio correctly(e.g.. oire nrust eor~sider both norr-idealities sucll as asphericity which splits the tie-
38
generacy in mode number nt and Q's which are typically non-radiatively linrited and the spatial dependcrrce of