optical coupling from plane wave to step-index single-mode fiber
TRANSCRIPT
Optics Communications 237 (2004) 37–43
www.elsevier.com/locate/optcom
Optical coupling from plane wave to step-indexsingle-mode fiber
Massimo Lazzaroni a,*, Fabio E. Zocchi b
a Dipartimento di Tecnologie dell’Informazione, Universit�a degli Studi di Milano, Via Bramante 65, 26013 Crema (CR), Italyb Media Lario S.r.l, Localit�a Pascolo – 23842 Bosisio Parini (LC), Italy
Received 14 January 2004; accepted 24 March 2004
Abstract
The optical coupling efficiency from a plane wave to a single-mode step-index fiber optics is studied considering the
exact fiber mode field distribution. A general expression relating the coupling efficiency to the fiber parameters is
presented and the optimum coupling efficiency is studied as a function of the fiber normalised frequency and the ratio
between the coupling optics radius and focal length. It is also shown that for small values of this ratio the optimum
coupling efficiency and a corresponding suitably defined optimum design parameter are functions of the fiber nor-
malized frequency only and they do not depend separately on the wavelength, the fiber core radius or the fiber
numerical aperture.
� 2004 Elsevier B.V. All rights reserved.
PACS: 42.30.Kq; 42.81.)i
Keywords: Fourier optics; Fiber optics
1. Introduction
Optical coupling to a single-mode fiber optics is
the subject of several studies concerning both co-
herent and incoherent incident fields [1–9]. An
ideal perfect imaging system that couples a plane
wave to a single-mode fiber is the simplest to be
* Corresponding author. Tel.: +39-02-503-30058; fax: +39-02-
503-30010.
E-mail addresses: [email protected] (M. Lazzaroni),
[email protected] (F.E. Zocchi).
0030-4018/$ - see front matter � 2004 Elsevier B.V. All rights reserv
doi:10.1016/j.optcom.2004.03.092
studied and thus the most suited to a theoretical
analysis. Although in most applications dealing
with fiber optics the field to be coupled is best
described by a Gaussian distribution, a plane wave
input can be taken as representative of real sys-
tems, ranging from optical and infrared astro-
nomical interferometry to high performance FreeSpace Optics communication. In the former ap-
plication fiber optics can be used to guide light
from telescopes to the interferometer [10] whereas
in the latter, currently considered for large band-
width optical inter-satellite links [11], best perfor-
mances are achieved by coherent detection and in
ed.
38 M. Lazzaroni, F.E. Zocchi / Optics Communications 237 (2004) 37–43
this case the coupling of the incident field ap-
proximated by a plane wave to a single-mode fiber
optics seems unavoidable. In both cases the dis-
tance between the source and the receiver is so
large compared to the wavelength and the tele-
scope aperture that the plane wave approximationis fulfilled.
The coupling efficiency for a plane wave by
means of an ideal optical system is well known
when the electric field of the fiber mode is ap-
proximated by a Gaussian function and the par-
axial approximation is valid for the optical
description of the coupling optics. If a plane wave
of free space wavelength k and wave numberk ¼ 2p=k is incident on the coupling optics with
focal length F and aperture radius A, then in the
Gaussian approximation for the fiber mode the
coupling efficiency g, defined [1] as the ratio be-
tween the coupled optical power Pout and the in-
cident one, Pin, is given by [1],
g � PoutPin
¼ 2
v21
�� e�v2
�2
; ð1Þ
where v ¼ Akw=2F , w being the mode radius de-
fined as the distance from the fiber axis at which
the optical intensity drops by a factor 1=e2. It
should be stressed that equation (1) is valid only ifthe coupling system can be described in the par-
axial approximation, restricting the use of (1) to
small values of the ratio A=F .The result provided by (1) gives a simple rule
for optimum design. For v equal to about 1.121
the coupling efficiency g in (1) has a maximum of
81.5%. This value is always achievable by a suit-
able design of the coupling optics, provided v ismade equal to 1.121.
When the exact distribution of the electric field
of the fiber optics mode is taken into account,
numerical evaluation [3] shows that the optimum
plane wave to single-mode fiber coupling effi-
ciency is a function of the fiber normalised fre-
quency V , in contrast with the Gaussian
approximation result for which an optimum effi-ciency of 81.5% is always achievable, irrespective
to the value of V . However, in [3] the calculation
was performed in paraxial approximation for
small values of the ratio A=F and for a particular
fiber optics, with given core radius and numerical
aperture, and the conditions for optimum cou-
pling were not analysed.
When high performance optical systems are
considered, a deeper theoretical investigation of
the coupling efficiency and of the conditions for
optimum coupling is desirable. To this aim thecoupling efficiency of an ideal system coupling a
monochromatic plane wave to single-mode step-
index fiber optics is here calculated taking into
account the exact electric field distribution of the
fiber mode. An expression relating the coupling
efficiency to the fiber parameters is presented and
plotted and the optimum coupling efficiency is
studied as a function of both V and A=F , up tolarge values of the latter ratio. It is also shown that
for small values of A=F the optimum coupling ef-
ficiency and a corresponding suitably defined op-
timum design parameter are functions of the fiber
normalized frequency only and they do not depend
separately on the wavelength, the fiber core radius
or the fiber numerical aperture.
2. Optical coupling
2.1. Coupling field
The geometry of the optical coupling system is
shown in Fig. 1. The optical axis z has its origin in
the focal plane of an ideal coupling optics withfocal length F and aperture radius A. The z axis
coincides with the axis of a single-mode fiber op-
tics, the entrance facet of which is on the focal
plane at z ¼ 0. Polar co-ordinates (q0;u0) and
(q;u) are used on the exit aperture r0 of the cou-
pling optics and on the entrance facet r of the fi-
ber, respectively.
The effect of the ideal coupling optics is tomultiply the incident wave by a pure phase factor
e�ikr for q06A;
0 for q0 > A;
�
where r ¼ ðF 2 þ q02Þ1=2 is the distance from the
origin. If the incident field is a monochromaticplane wave travelling along the z axis, Ei ¼ E0e
ikz,
then the field at the exit aperture r0 of the imaging
system is Ee ¼ E0eikz�ikr for q0
6A.
Fig. 1. Geometry of the ideal optical system coupling a plane wave to a single-mode fiber optics (not to scale); longitudinal and
transversal cross-sections (a) and three-dimensional view (b).
M. Lazzaroni, F.E. Zocchi / Optics Communications 237 (2004) 37–43 39
The electric field Ef in air at the entrance facet rof the fiber optics is obtained from the Rayleigh–
Sommerfeld integral [13]
EfðQÞ ¼1
ik
Zr0EeðPÞ
eikR
Rz � Rdr0; ð2Þ
where R is the distance between the point of inte-
gration P and the point Q at which the field is
evaluated. R and z are the unit vectors along R and
z, respectively. Since the field is different from zero
only in a very small area on the entrance facet of thefiber with a linear extension of the order of few radii
a of the fiber core, the scalar product between z andR is almost equal to the product between z and r,the unit vector along r, that is z � R ffi z � r ¼ cos#.In addition it can be assumed that R ffi r in the
denominator of the integrand of (2) and that
R ¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffir2 þ q2 � 2r � q
pffi r � r � q
¼ r � q sin# cosðu0 � uÞ
40 M. Lazzaroni, F.E. Zocchi / Optics Communications 237 (2004) 37–43
in the exponent. The above approximations are
standard in optics when the angle # is small.
However in the assumption that F is much greater
than the fiber core radius a these approximations
are also valid for large values of # or, equivalently,of the ratio A=F .
Inserting the above approximations into (2)
along with dq0 ¼ ðF = cos2 #Þd# and q0=r ¼ sin#,the field at the entrance facet r of the fiber becomes
Ef q;uð Þ ¼ FE0e�ikF
ik
Z #0
0
sin#
cos#d#
�Z 2p
0
e�ikq sin# cosðu0�uÞ du0
¼ �ikFE0e�ikF
Z #0
0
J0ðkq sin#Þsin#
cos#d#;
ð3Þ
where J0 is the Bessel function of the first kind oforder zero and #0 ¼ atanðA=F Þ.
2.2. Coupling efficiency
If the incoming wave is propagating in air with
impedance Z0, the optical power incident on the
coupling system is given by Pin ¼ pA2E20=2Z0. The
power coupled to the fiber is calculated asPout ¼ jCj2, C being the coupling integral [1]
C ¼ s2
ZrEf �H� � zdr
¼ pnsZ0
Z 1
0
EfðqÞE�ðqÞqdq; ð4Þ
where � denotes complex conjugation and s is the
electric field transmission coefficient from air to
the fiber. In the previous equation E and H are the
electric and magnetic field of the fiber mode and
EðqÞ is the transversal component of E. Finally n isthe index of refraction, assumed to be approxi-
mately the same in the core and in the cladding
of the fiber optics; for example in the Corning�
SMF-28TM single-mode fiber [12], considered the
standard fiber in optical networks, the relative re-
fraction index ratio between the core and the
cladding is about 0.35%.
The fiber fields E and H in (4) are normalizedsuch that
1
2
ZrE �H� � zdr ¼ pn
Z0
Z 1
0
EðqÞj j2qdq ¼ 1: ð5Þ
The second equalities in (4) and (5) are valid
provided the incident field is linearly polarized
along an axis orthogonal to the optical axis z. In-serting (3) into (4) gives
jCj ¼ pnkFE0
Z0
s
�����Z #0
0
sin#
cos#d#
Z 1
0
J0 kq sin#ð ÞE� qð Þqdq����:ð6Þ
The electric field EðqÞ of a step-index single-
mode fiber optics, normalized as in (5), is given by[14]
EðqÞ ¼vaV
ffiffiffiffiffi2Z0pn
q1
J1ðuÞJ0 uq=að Þ for 06 q6 a
vaV
ffiffiffiffiffi2Z0pn
qJ0ðuÞ
J1ðuÞK0ðvÞK0 vq=að Þ for q > a;
8<:
ð7Þwhere u and v are determined by the relations [14]
u2 þ v2 ¼ V 2; ð8aÞuJ1ðuÞK0ðvÞ � vK1ðvÞJ0ðuÞ ¼ 0: ð8bÞIn (7), (8a) and (8b) V is the normalized frequencyof the fiber, V ¼ kaðn2co � n2clÞ
1=2, nco and ncl being
the refractive index in the core and in the cladding,
J1 is the Bessel function of the first kind of order
one and K0 and K1 are the modified Bessel func-
tions of the second kind of order zero and one,
respectively.
Using (7) and (8b) the integral with respect to qin (6) can be evaluated to be [15]
2ffiffiffiffiffiffipn
p
affiffiffiffiffiZ0
pZ 1
0
J0 kq sin#ð ÞE�ðqÞqdq
¼ 2Vv
v2 þ n2� �
u2 � n2� � � uJ0ðnÞ
�� nJ1ðnÞ
J0ðuÞJ1ðuÞ
�
� I n; Vð Þ;
where n ¼ ka sin#. Since u and v are completely
defined through (8a) and (8b) once V is given, the
function Iðn; V Þ depends only on the normalized
frequency V . Finally by inserting the expression of
the above integral into (6), the coupling efficiency
is found to be
M. Lazzaroni, F.E. Zocchi / Optics Communications 237 (2004) 37–43 41
g � PoutPin
¼ Tk2a2
tan2 #0
�Z ka sin #0
0
I n; Vð Þ n
k2a2 � n2dn
� �2
; ð9Þ
where T ¼ njsj2 is the intensity transmission coef-
ficient [16]. In the following T will be assumedequal to unity. Equation (9) can be evaluated as a
function of A=F ¼ tan#0 and the results plotted,
giving similar curves as obtained for small value of
A=F by Barrel and Pask in [2]. However, in Fig. 2
the coupling efficiency g, as calculated from (9), is
compared with the result (1) following from the
Fig. 2. Coupling efficiency g as a function of the ratio A=F for
k ¼ 1550 nm and a ¼ 5 lm. The normalized frequency is
V ¼ 2:4 in the upper plot and V ¼ 1 in the lower one. The solid
line refers to the solution presented in the paper whereas the
dotted line is based on the Gaussian approximation (1).
Gaussian approximation of the fiber mode and the
paraxial description of the coupling optics. The
curves in Fig. 2 are evaluated for two different
values of the normalized frequency, V ¼ 2.4 and
V ¼ 1; in both cases k ¼ 1550 nm and a ¼ 5 lm.The field radius w in the Gaussian approximation
is chosen as the 1=e radius of the actual fiber
mode. When V ¼ 2:4, the positions of the two
maxima are almost at the same value of the A=Fratio, 0.101 for the solution given by (9) and 0.103
for the Gaussian approximation, whereas the
corresponding optimum coupling efficiencies are
78.4% and 81.5%, respectively. For V ¼ 1, themaximum efficiency is 66.6% at A=F ¼ 0:0194 for
the solution given by (9) and 81.5% at
A=F ¼ 0:0397 for the Gaussian approximation.
The optimum coupling efficiency can be studied
numerically from (9). For a given normalized fre-
quency V and a given ratio A=F ¼ tan#0, the
maximum value gopt of the coupling efficiency gcan be found from (9) by a suitable choice of theproduct ka. Since, the derivation of (9) is valid for
both small and large value of A=F , equation (9)
can be used to trace the optimum coupling effi-
ciency gopt up to large value of A=F . Fig. 3 shows
the plots of gopt as a function of A=F up to
A=F ¼ 10 for three different values of the
Fig. 3. Optimum coupling efficiency gopt as a function of the
ratio A=F for three values of the normalized frequency V ¼ 1,
1.7 and 2.4. The dashed line is the optimum efficiency, inde-
pendent from V , when the fiber mode is approximated by a
Gaussian function and the incident filed is given by (3).
Fig. 4. Optimum coupling parameter nopt and optimum cou-
pling efficiency gopt as a function of the normalized frequency Vfor small value of A=F . The dashed line is the optimum effi-
ciency of 81.5% in the Gaussian approximation, independent
from V .
42 M. Lazzaroni, F.E. Zocchi / Optics Communications 237 (2004) 37–43
normalized frequency, V ¼ 1, 1.7 and 2.4. The
optimum coupling efficiency is a decreasing func-
tion of A=F . It reaches the highest values when the
ratio A=F is small whereas for very large values of
A=F it approaches zero asymptotically. Even if the
use of A=F values greater than about 1 is notrecommended, the study of the coupling efficiency
when the A=F ratio approaches 1 could be of in-
terest in inter-satellite free space optical commu-
nications where both the requirement of large
collecting area and the dimensional constraints
limiting the focal length should be taken into ac-
count. In addition increasing the value of A=Fimproves the field of view of the optical systemallowing the implementation of a fine fast tracking
system by just moving the fiber optics on the focal
plane. At design stage a trade-off should be per-
formed between these benefits and the drawback
represented by the decreasing of the coupling effi-
ciency at increasing values of A=F .The dashed line in Fig. 3 is the optimum effi-
ciency when the fiber mode is approximated by aGaussian function whereas the incident field is
given by (3). In this case the optimum efficiency is
independent from the normalized frequency V that
determines the width of the Gaussian approxima-
tion of the fiber mode. Indeed for a given value of
V and of A=F , it is possible to select the wave
number k in (3) to match the Gaussian shape of
the fiber mode. At low values of the ratio A=F(paraxial approximation) the incident field (3)
approaches an Airy pattern and the coupling effi-
ciency is approximated by (1). Consequently, as
indicated in Fig. 3, for small values of A=F the
dashed line approaches the efficiency of 81.5% gi-
ven by the maximum of Eq. (1).
2.3. Optimum coupling at small angles
Since the best coupling conditions are reached
at small values of A=F ¼ tan#0, it is worth
studying (9) when the angle #0 is small. Under
this assumption a very general relation for the
optimum coupling can be found. When #0 is
small the denominator of (9) can be simplified as
follows,
k2a2 � n2 ¼ k2a2 1�
� sin2 #�ffi k2a2;
since n ¼ ka sin# and 06#6#0. Thus (9) can be
written as
g ffi 1
n20
Z n0
0
I n; Vð Þndn� �2
; ð10Þ
with n0 ¼ ka tan#0 ffi ka sin#0. The condition that
g is at maximum gives the following relation be-tween the optimum value nopt of n0 and the nor-
malized frequency V ,Z nopt
0
I n; Vð Þndn ¼ I nopt; V� �
: ð11Þ
Solving (11) numerically gives the relation betweennopt and V that is shown in Fig. 4 where the cor-
responding optimum value gopt of g is also plotted.
The dashed line in the figure represents the opti-
mum efficiency of 81.5% in the Gaussian approx-
imation, independent from V . It can be noted that
for a weakly guided mode, at low value of V , theoptimum coupling efficiency calculated above is
well below the approximate Gaussian prediction of81.5%.
The plot of gopt in Fig. 4 has been reported in
[3]. In that case however, the calculation was done
for a particular fiber core radius a ¼ 2 lm and
numerical aperture NA ¼ 0:11. Here the result
follows from (10) and (11) and it is thus shown to
be of general validity, dependent on V but not
M. Lazzaroni, F.E. Zocchi / Optics Communications 237 (2004) 37–43 43
separately on a, k or NA. In addition in Fig. 4 the
value nopt of the design parameter n0 ¼ ka sin#0
that gives the optimum coupling efficiency is also
plotted.
The two curves in Fig. 4 can be very well fitted
by a second order polynomial in V . By applyingthe Least Square Method, the following two ap-
proximations are found:
nopt ¼ a1 þ a2 � V þ a3 � V 2; ð12Þgopt ¼ b1 þ b2 � V þ b3 � V 2; ð13Þ
where: a1 ¼ �1:928, a2 ¼ 2:784, a3 ¼ �0:472,b1 ¼ 48:398, b2 ¼ 21:895, b3 ¼ �3:889. The maxi-
mum deviation from the exact curves is 0.027 for
nopt and 0.345% for gopt. The two above relations
(12) and (13) can be used to easily determine the
optimum coupling parameter and the corre-sponding coupling efficiency for a given value of
the normalized frequency V .
3. Conclusion
A general expression for the coupling efficiency
of plane wave light into a single-mode step-indexfiber optics has been presented allowing the study
of the conditions of optimum coupling for this
specific scenario that finds application in astron-
omy and in inter-satellite free space optical com-
munications. It is shown that the highest efficiency
is obtained, as expected, for small value of the ratio
between the coupling optics radius and focal
length. In this condition the optimum couplingefficiency is function of the normalised frequency
only and it can be achieved by a proper choice of a
design parameter, which again depends only on the
normalised frequency. Consequently it has been
possible to write two simple quadratic approxi-
mations for these two quantities allowing an easy
determination of the optimum coupling conditions.
The Gaussian approximation of the fiber modeis one of the sources of error in the evaluation of
the coupling efficiency of light into a single-mode
step-index fiber optics. Since the field mode in a
parabolic graded-index fiber is itself Gaussian,
further investigation in this direction can possibly
show that a higher theoretical value for the opti-
mum coupling efficiency can be obtained in thistype of fibers. In particular it is possible that the
coupling efficiency represented by the dashed line
in Fig. 3 can be achieved in graded-index fiber.
This would be a consistent improvement with re-
spect to the solid lines corresponding to low values
of V in step-index fibers.
Acknowledgements
The authors thank Fabio Marioni for helpful
discussion.
References
[1] R.E. Wagner, W.J. Tomlinson, Appl. Opt. 21 (1982) 2671.
[2] K.F. Barrel, C. Pask, Opt. Acta 26 (1979) 91.
[3] S. Shaklan, F. Roddier, Appl. Opt. 27 (1998) 2334.
[4] C. Ruilier, Proc. SPIE 3350 (1998) 319.
[5] P.J. Winzer, W.R. Leeb, Opt. Lett. 23 (1998) 986.
[6] S. Yuan, N.A. Riza, Appl. Opt. 38 (1999) 3214.
[7] Jun-Ichi Sakai, Tatsuya Kimura, IEEE J. Quantum
Electron. QE-16 (1980) 1059.
[8] O. Wallner, P.J. Winzer, W.R. Leeb, Appl. Opt. 41 (2002)
637.
[9] M. van Buren, N.A. Riza, Appl. Opt. 42 (2003) 550.
[10] S.B. Shaklan, F. Roddier, Appl. Opt. 26 (1987) 2159.
[11] V.W.S. Chan, IEEE J. Select. Top. Quantum Electron. 6
(2000) 959.
[12] Corning� SMF-28TM CPC6 Single-Mode Optical Fiber,
Product Information PI1036, Corning Incorporated, 1988,
Corning NY. 14831.
[13] J.W. Goodman, Introduction to Fourier Optics, McGraw-
Hill, New York, 1968.
[14] E.G. Neumann, Single-Mode Fibers, Springer, Berlin,
Germany, 1988.
[15] F. Bowman, Introduction to Bessel Functions, Dover, New
York, 1958.
[16] M.V. Klein, T.E. Furtak, Optics, Wiley, New York, 1986.