This is a specimen ab title?,??

Jos Migchielsen1,∗
Elsevier B.V., Radarweg 29, 1043 NX Amsterdam, The Netherlands

CV Radhakrishnan2
Sayahna Foundations, JWRA 34, Jagathy, Trivandrum 695014, India

CV Rajagopal1,3
STM Document Engineering Pvt Ltd., Mepukada, Malayinkil, Trivandrum 695571, India

Abstract
In this work we demonstrate ab the formation Y 1 of a new type of polariton on the interface
between a cuprous oxide slab and a polystyrene micro-sphere placed on the slab. The evanescent
field of the resonant whispering gallery mode (WGM) of the micro sphere has a substantial
gradient, and therefore effectively couples with the quadrupole 1S excitons in cuprous oxide.
This evanescent polariton has a long life-time, which is determined only by its excitonic and
WGM component. The polariton lower branch has a well pronounced minimum. This suggests
that this excitation is localized and can be utilized for possible BEC. The spatial coherence of the
polariton can be improved by assembling the micro-spheres into a linear chain.
Keywords: quadrupole exciton, polariton, WGM, BEC
JEL: 71.35.-y, 71.35.Lk, 71.36.+c

1. Introduction
Although quadrupole excitons (QE) in cuprous oxide crystals are good candidates for BEC
due to their narrow line-width and long life-time there are some factors impeding BEC [1, 2]. One
of these factors is that due to the small but non negligible coupling to the photon bath, one must
consider BEC of the corresponding mixed light-matter states called polaritons [3]. The photon-
like part of the polariton has a large group velocity and tends to escape from the crystal. Thus,

? This document is the results of the research project funded by the National Science Foundation.
?? The second title footnote which is a longer text matter to fill through the whole text width and overflow into another
line in the footnotes area of the first page.
∗ Corresponding author

Email addresses: J.Migchielsen@elsevier.com (Jos Migchielsen), cvr@sayahna.org (CV Radhakrishnan)

URL: www.stmdocs.in (CV Rajagopal)
1 This is the first author footnote.
2 Another author footnote, this is a very long footnote and it should be a really long footnote. But this footnote is not

yet sufficiently long enough to make two lines of footnote text.

3 Yet another author footnote.

Preprint submitted to Elsevier June 8, 2018

the temporal coherence of the condensate is effectively broken [4, 5]. One proposed solution to
this issue is to place the crystal into a planar micro-cavity [6]. But even state-of-the-art planar
micro-cavities can hold the light no longer than 10 µs. Besides, formation of the polaritons in
the planar cuprous oxide micro-cavity is not effective due to quadrupole origin of the excitons.

Theorem 1. In this work we demonstrate the formation of a new type of polariton on the in-
terface between a cuprous oxide slab and a polystyrene micro-sphere placed on the slab. The
evanescent field of the resonant whispering gallery mode (WGM) of the micro sphere has a sub-
stantial gradient, and therefore effectively couples with the quadrupole 1S excitons in cuprous
oxide. This evanescent polariton has a long life-time, which is determined only by its excitonic
and WGM component. The polariton lower branch has a well pronounced minimum. This sug-
gests that this excitation is localized and can be utilized for possible BEC. The spatial coherence
of the polariton can be improved by assembling the micro-spheres into a linear chain.

Therefore in this work we propose to prevent the polariton escaping by trapping it into a
whispering gallery mode (WGM)4 of a polystyrene micro-sphere (PMS).
We develop a model which demonstrates formation of a strongly localized polariton-like
quasi-particle. This quasi-particle is formed by the resonant interaction between the WGM in
PMS and QE in the adjacent layer of cuprous oxide. The QE interacts with the gradient of the
WGM evanescent field.
There are few experiments concerned with resonant interaction of the WGM and dipole al-
lowed exciton (DE) [7, 8]. But the DE has some disadvantages compared to QE when it comes
to interaction with the WGM. First, the evanescent light has small intensity. Therefore it is not
effective for the dipole allowed coupling. But it has a large gradient, so it can effectively couple
through a quadrupole part. Second, the DE has short life time compared to the QE and therefore
is not suitable for BEC. Third, the kinetic energy of the DE is comparable with the interaction
energy. Hence the localization is effectively impeded.

2. Evanescent vs. conventional quadrupole light-matter coupling

Assume that a single PMS of radius r0 µm is placed at a small 5 distance δr0  r0 from the
cuprous oxide crystal (Cu2O = 6.5).
There are several methods to observe WGM-QE interaction. One of them is to mount a prism
(or a fiber) on the top of PMS [7]. But any surface nearby perturbs spherical symmetry and there-
fore reduces the Q factor and the life-time of the WGM. But the positions of the corresponding
Mie resonances are not affected. Therefore in this paper we adopt a slightly different picture.
Namely the scattering of the bulk polariton in cuprous oxide by the PMS. If one of the Mie
resonances is in resonance with the QE one can expect formation of the new type of polariton.
Assume some density of quadrupole 1S excitons ([QE]), ~ω1S = 2.05 eV, λ1S = 2π/ω1S =
6096 Å) has been created by an external laser pulse. The corresponding polaritons move in the
crystal as the polariton and can be trapped by the PMS due to WGM-QE resonant interaction.

4 WGM occur at particular resonant wavelengths of light for a given dielectric sphere size. At these wavelengths, the

light undergoes total internal reflection at the sphere surface and becomes trapped within the particle for timescales of
the order of ns.
5 comparing to the evanescent field penetration depth

2
 The WGM evanescent field penetration depth into the cuprous oxide adjacent crystal is much
larger than the QE radius:

λ1S /2π (Cu2O − 1)1/2 = 414 Å  aB = 4.6 Å

Hence, the light-matter interaction can be considered semi-classically. For the same reason we
consider only bulk polaritons. However the theory may be expanded to include the surface polari-
tons also. In the late case the evanescent field of such a surface polariton could play an essential
role in the interaction with the WGM, comparable with WGM-QE coupling.
For resonance coupling with a WGM its size parameter should be determined by the resonant
wave vector in the cuprous oxide k0 = 2.62 × 107 m−1 . For example, if one takes a polystyrene
(refractive index  2 = 1.59) sphere of radius r0 = 10.7 µm then k0 r0 = 28.78350. This size
parameter corresponds to the 39TE1 resonance [9].
The photon part of the polariton trapped inside the PMS moves as it would move in a micro-
cavity of the effective modal volume V  4πr03 /3. Consequently, it can escape through the
evanescent field. This evanescent field essentially has a quantum origin and is due to tunneling
through the potential caused by dielectric mismatch on the PMS surface. Therefore, we define
the evanescent polariton (EP) as an evanescent light - QE coherent superposition.
Below we compare the evanescent quadrupole polariton and conventional bulk quadrupole
polariton in cuprous oxide. For simplicity let us consider the incident polariton wave vector
running along the interface (z direction). The polarization of the polariton is taken along the x
direction. Therefore, in the system of coordinates centered at the sphere, the photon part of the
incident polariton can be written as [10]:
X 2l + 1
Ei = E 0 il (M1l − iN1l ) , (1)
l
l (l + 1)

where M1l and N1l are vector spherical harmonics corresponding to TE- and TM- polarized
modes of angular momentum l; the z component of the angular momentum is |m| = 1; E0 is the
amplitude of the electric field. The scattered field is given as:
X 2l + 1
Es = E 0 il (ia1l N1l − b1l M1l ) , (2)
l
l (l + 1)

here a1l and b1l are scattering Mie coefficients (See the Appendix). Taking into account that both
WGM and QE have narrow line-width, and the energy separation between different WGM is
much bigger then the interaction energy we adopt a single mode picture [7]. Keeping only the
resonant term the last expression yields:

E s = −E0 il 0.05b1,39 M1,39 , (3)

To calculate the interaction of the plane wave (conventional polariton) (1) and WGM (evanes-
cent polariton) with cuprous oxide one has to change to the cuprous oxide centered system of
coordinate (See Fig.2) While in the system of the coordinate, centered at the cuprous oxide, the
plane wave is still given by the expression (1), the scattered field has to be changed according to
the vector spherical harmonic addition theorem [11]:

1,39 (r0 + δr) Mml + B1,39 (r0 + δr) Nml

M1,39 = Aml ml
(4)
3
Here Aml ml
1,39 and B1,39 are the translational coefficients. Their explicit expression can be found, for
instance, in [12, 9] and are explicitly listed in the Appendix.
The bulk (incident) and evanescent polaritons in cuprous oxide are formed through the quadrupole
part of the light-matter interaction:
ie
Hint = Ei,s · p
mω1S
Here e, m are the electron charge and mass; p is the electron momentum. For the quadrupole 1S
transition in cuprous oxide the energy of interaction can be written as:
∞ Z Z Z
X Fn (x) Gn (x, y) Gn (x, y)
An dx = BnC n dx dy + (5)
i=0
An + Bn An x + Bn y An x + Bn y

Here we introduced the initial state of the system, which transforms as irreducible representation
1 +
Γ1 of the cubic centered group Oh . The final state is the ortho-exciton state which transforms as
3 +
Γ5,xz in Cartesian system or as 3 Γ+5;1,2 in the corresponding spherical basis.
Hence, using (1, 3, 4, 5), one can deduce that the the coupling of the spherical harmonic
compared to the plane wave (~g1,2 = 124 µeV) is resonantly enhanced:
g1,39
= −i0.06b1,39 (kr0 ) A1,2
1,39 0 + δr)
(r (6)
g1,2

Here we utilized the fact that B1,2 1,2

1,39  A1,39 . While the resonant enhancement is provided by the
b1,39 Mie coefficient here, the translational coefficient reduces the effect. That is why if one tries
to couple the evanescent light to the dipole transition the effect is much weaker as A0,1 1,2
1,39  A1,39 .
The resulting exciton - evanescent light coupling is shown in Fig.1 Both dipole and quadrupole

Figure 1: The evanescent light - 1S quadrupole coupling (g1,l ) scaled to the bulk exciton-photon coupling (g1,2 ). The
size parameter kr0 is denoted as x and the PMS is placed directly on the cuprous oxide sample (δr = 0, See also Fig.2).

coupling rate in the actual combined semiconductor-microsphere system is smaller then that in
case of conventional polariton. This is attributed to the fact that the coupling occurs in a small
region of the evanescent tail penetrating into cuprous oxide, although the coupling grows with
mode number l, because the gradient of the evanescent field increases. Note that QE realizes
4
strong coupling regime g1,39 > γ while DE demonstrates weak regime only [7]. The property of
the scalable coupling factor can be utilized in practical applications such as non-linear optics and
is the subject of our future work.

3. Results and discussion

In this section let us utilize the above calculated WGM-QE interaction to obtain the evanes-
cent polariton (EP) dispersion in the framework of the coupled oscillator model that has been
widely used for describing coupled atom-photon or exciton-photon modes in microcavity sys-
tems [13]. Near the resonance between WGM and the quadrupole exciton ω1l ≈ ω1S the EP
branches are given by the eigenvalues of the following Hamiltonian:
 
H/~ = ω1l a†x a x + ω1S b†x b x + g1l (x) a†k b x + a x b†x , (7)

here a x , b x are annihilation operators for light and the exciton, respectively. We also neglected
kinetic energy of the QE due to smallness of the resonant wave vector and big mass of the QE.
Therefore, considering that both the exciton and WGM of a single sphere are localized, the
dispersion is reduced to:
ω = ω1S ± g1l /~ (8)
The above expression shows the formation of the doublet at resonance (both states are exactly
half-QE, half-WGM). Recall that for DE-WGM weak coupling only WGM pattern shifted by
the coupling has been observed [7].
The excitons are trapped in the minimum of the lower branch thus populating the strongly
localized states. Physically this means that the resonant coupling with localized WGM does not
let QE escape by means of its kinetic energy.
The dispersion above is similar to the quadrupole-dipole hybrid in the organic-inorganic
hetero-structures [2]. In the later case, the excited organic molecules create an evanescent field
penetrating into the cuprous oxide.
Now let us consider possible application of the evanescent polariton to BEC. The problem
of the conventional polariton escaping from the crystal mentioned in the introduction no longer
exists for the localized states of the evanescent polariton.
The linewidth of the mixed state is expected to be in the first approximation the sum of the
exciton and photon linewidth weighted by the exciton and photon component of the polariton.
At resonance, the linewidth of the two eigenstates is simply given by γ = γ1S + γm,l /2. Where


γ1S and γm,l are the QE and WGM linewidth. Even having taken into account decrease of the
Q factor due to PMS contact with the cuprous oxide sample γ1S  γ1,39 for the ortho-exciton.
Hence, γ is defined by the QE linewidth.
For the Cu2 O para-exciton the linewidth of the polariton is given by the WGM linewidth.
The para-excitons can acquire some oscillator strength provided the PMS exerts a local stress
upon the cuprous oxide sample. The applied stress changes the crystal symmetry, so that usually
optically inactive para-excitons may couple to the WGM. Note that due to spin-orbit interaction
the para-excitons are 12 µeV below the ortho-exciton. Therefore for given radius of the PMS the
resonant interaction occurs with different WGM for ortho- and para- cases.
The evanescent polariton provided by a single sphere gives the time coherence necessary
for the observable BEC of the quadrupole exciton. But the spatial coherence is limited to a
small region near the sphere. To improve the spatial coherence one has to sacrifice the temporal
5
Figure 2: Schematic of formation of the evanescent polariton on linear chain of PMS. The actual dispersion is deter-
mined by the ratio of two coupling parameters such as exciton-WGM coupling and WGM-WGM coupling between the
microspheres.

coherence slightly by delocalizing the corresponding WGM. It can be done by using an array of
spheres aligned along the z direction and separated by the distance δr0 (See Fig.2).
Recent experimental [14] and theoretical [15] studies have shown that the WGM can travel
along the chain as ”heavy photons”. Therefore the WGM acquires the spatial dispersion, and the
evanescent quadrupole polariton has the form (See Fig.3):
q
2ω = ω1l,k + ω1S ± ω1l,k − ω1S 2 + 4|g1l /~|2


 
ω1l,k = ω1S + 2 g1l
1l /~ cos(x − x1l + π/2) (9)

1l = ω1S b1l A1l

Here g1l 1l
(δr1 ) is the nearest-neighbor inter-sphere coupling parameter.

Figure 3: Dispersion of the evanescent polariton 39TE1. The dashed line (1) corresponds to the dispersion of the chain
of spheres touching each other (δr0 = 0). The thin solid line (3) stands for upper and lower branches of a single sphere
dispersion (δr0  δr = 0). The thick solid curve (2) is the case of linear chain of the spheres in contact with the cuprous
oxide (δr0 = δr = 0).

When the coupling between spheres dominates (δr  δr0 ) the minimum of the lower po-
lariton branch disappears. Consequently, for possible BEC of the evanescent polariton one has
6
to keep the desired balance between spatial and temporal coherence by adjusting experimental
parameters δr and δr0 .
Both, the energy of the 1S quadrupole exciton and the WGM depend on the temperature.
Therefore one can use a standard temperature scan to reveal the evanescent polariton dispersion
[16].
In summary, we note that there is some similarity between BEC of alkali atoms trapped by
the laser field [17] and the long living QE localized by the resonant WGM.
The theory developed above is applicable also for void cavities, spherical impurities and
metallic droplets in bulk cuprous oxide crystal.

4. Appendix

In the appendix we list explicit expression for the Mie scattering coefficient:

n2 jml (nx) x jml (x) 0 − jml (x) nx jml (nx) 0

   
aml = h i0
n2 jml (nx) xh(1) (x) − h(1) (x) nx jml (nx) 0
 
ml ml
jml (nx) x jml (x) 0 − n2 jml (x) nx jml (nx) 0
   
bml = h (1) i0
jml (nx) xhml (x) − n2 h(1) (x) nx jml (nx) 0
 
ml

Here n =  2 is the refractive index of the spheres; x = kr0 is the size parameter; jml , hml are the
spherical Bessel and Hankel of the first kind functions respectively.
In the case of l  1 the calculation of the translational coefficients can be significantly
simplified with the help of the so-called maximum term approximation [9].
s
0 l + l0
All  −2l (−1)l+1 ×
π (l0 + 1) (l − 1)
0
ll (l0 )l
l0 +1
h(1) (ηx)
l−1 l+l0
(l + 1) (l − 1)
0

0 x |i − j| l0
Bll  i Al
ll0
Here η defined as η = |r0 + δr|/r0 ≥ 1 is a dimensionless distance between the centers of the
spheres.

References
[1] G. Kavoulakis and G. Baym, Phys. Rev. B 53, 7227 (1996).
[2] O. Roslyak and J. Birman, arXiv:cond-mat/0703650, PRB to be published (2007).
[3] D. Frohlich, G. Dasbach, G. B. Hogersthal, M. Bayer, R. Kliebera, D. Sutera, and H. Stolzb, Solid State Commu-
nications 134, 139 (2005).
[4] C. Ell, A. L. Ivanov, and H. Haug, Phys. Rev. B 57, 9663 (1998).
[5] D. Snoke, Science 298, 1368 (2002).
[6] J. Kasprzak, M. Richard, S. Kundermann, A. Baas, P. Jeambrun, J. Keeling, F. Marchetti, M. Szymanska, R. Andre,
J. Staehli, et al., Nature 443, 409 (2006).
[7] H. W. Xudong Fan, Scott Lacey, Optics Letters 24, 771 (1999).
[8] X. Fan, S. Lacey, and H. Wang, Opt. Lett 24, 771 (1999).
[9] H. Miyazaki and Y. Jimba, Phys. Rev. B 62, 7976 (2000).
7
[10] C. Bohren and D. Huffman, Absorption and scattering of light by small particles (Wiley New York, 1983).
[11] S. Stein, Q. appl. Math 19, 15 (1961).
[12] K. Fuller, Appl. Opt 30, 4716 (1991).
[13] H. J. Carmichael, Phys. Rev. A 33, 3262 (1986).
[14] Y. Hara, T. Mukaiyama, K. Takeda, and M. Kuwata-Gonokami, Physical Review Letters 94, 203905 (2005).
[15] L. Deych and A. Roslyak, Physical Review E 73, 36606 (2006).
[16] E. Peter, P. Senellart, D. Martrou, A. Lemaı̂tre, J. Hours, J. Gérard, and J. Bloch, Physical Review Letters 95, 67401
(2005).
[17] A. J. Leggett, Rev. Mod. Phys. 73, 307 (2001).