The nonlinear directional coupler: an analytic solution - CCM
The nonlinear directional coupler: an analytic solution - CCM
The nonlinear directional coupler: an analytic solution - CCM
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
Abstract<br />
<strong>The</strong> <strong>nonlinear</strong> <strong>directional</strong> <strong>coupler</strong>: <strong>an</strong> <strong>an</strong>alytic <strong>solution</strong><br />
R. Vilela Mendes *<br />
Universidade Tecnica de Lisboa <strong>an</strong>d Grupo de Fısica-Matematica, Complexo Interdisciplinar,<br />
Av. Prof. Gama Pinto 2, 1649-003 Lisboa, Portugal<br />
Received 13 November 2003; received in revised form 19 December 2003; accepted 21 December 2003<br />
Linear <strong>an</strong>d <strong>nonlinear</strong> <strong>directional</strong> <strong>coupler</strong>s are currently used in fiber optics communications. <strong>The</strong>y may also play a<br />
role in multiphoton approaches to qu<strong>an</strong>tum information processing if accurate control is obtained over the phases <strong>an</strong>d<br />
polarizations of the signals at the output of the <strong>coupler</strong>. With this motivation, the const<strong>an</strong>ts of motion of the <strong>coupler</strong><br />
equation are used to obtain <strong>an</strong> explicit <strong>an</strong>alytical <strong>solution</strong> for the <strong>nonlinear</strong> <strong>coupler</strong>.<br />
Ó 2003 Elsevier B.V. All rights reserved.<br />
PACS: 42.30.Qw; 42.80.Vc; 42.81.My<br />
Keywords: Nonlinear <strong>directional</strong> <strong>coupler</strong><br />
1. Introduction<br />
Directional <strong>coupler</strong>s are useful devices currently<br />
used in fiber optics communications. Because of<br />
the interaction between the signals in the input fibers,<br />
power fed into one fiber is tr<strong>an</strong>sferred to the<br />
other. <strong>The</strong> amount of power tr<strong>an</strong>sfer c<strong>an</strong> be controlled<br />
by the coupling const<strong>an</strong>t, the interaction<br />
length or the phase mismatch between the inputs.<br />
If, in addition, the material in the <strong>coupler</strong> region<br />
has <strong>nonlinear</strong>ity properties, the power tr<strong>an</strong>sfer will<br />
also depend on the intensities of the signals [1,2]. A<br />
large number of interesting effects take place in<br />
<strong>nonlinear</strong> <strong>directional</strong> <strong>coupler</strong>s [3–7] with, in particular,<br />
the possibility of performing all classical<br />
* Tel.: +351-914739442; fax: +351-217954288.<br />
E-mail address: vilela@cii.fc.ul.pt (R. Vilela Mendes).<br />
Optics Communications 232 (2004) 425–427<br />
0030-4018/$ - see front matter Ó 2003 Elsevier B.V. All rights reserved.<br />
doi:10.1016/j.optcom.2003.12.056<br />
www.elsevier.com/locate/optcom<br />
logic operations by purely optical me<strong>an</strong>s [8]. <strong>The</strong>y<br />
may also play a role in qu<strong>an</strong>tum information<br />
processing.<br />
<strong>The</strong> use of the intensity-dependent phase shifts<br />
associated to the Kerr <strong>nonlinear</strong>ity was, in the<br />
past, proposed for the construction of qu<strong>an</strong>tum<br />
gates [9–11]. However they rely on one-photon<br />
processes <strong>an</strong>d therefore would require very strong<br />
<strong>nonlinear</strong>ities, not available in the low loss optical<br />
materials. On the other h<strong>an</strong>d, the qu<strong>an</strong>tum computation<br />
scheme based on linear optics of Knill,<br />
Laflamme <strong>an</strong>d Milburn [12] is probabilistic <strong>an</strong>d<br />
relies on a delicate sensitivity of one-photon detectors.<br />
For this reason multiphoton approaches<br />
have been explored based either on the qu<strong>an</strong>tumlike<br />
behavior [14–16] of optical modes on a fiber<br />
[13] (the role of time played by a space coordinate)<br />
or on coherent states [17]. For light beams on a
426 R. Vilela Mendes / Optics Communications 232 (2004) 425–427<br />
fiber, sizable <strong>nonlinear</strong> effects are easy to achieve<br />
with available materials. In particular the <strong>directional</strong><br />
<strong>coupler</strong> might provide <strong>an</strong> already available<br />
tool for the implementation of linear or <strong>nonlinear</strong><br />
gates. 1<br />
For qu<strong>an</strong>tum information purposes one would<br />
require accurate information on the phases <strong>an</strong>d<br />
polarizations of the signals at the output of the<br />
<strong>coupler</strong>. Analytic <strong>solution</strong>s are ideal for this purpose<br />
although, in general, difficult to obtain for<br />
<strong>nonlinear</strong> systems. Here, by exploring the const<strong>an</strong>ts<br />
of motion of the <strong>coupler</strong> equation, <strong>an</strong> explicit<br />
<strong>an</strong>alytical <strong>solution</strong> is obtained for the<br />
<strong>nonlinear</strong> <strong>coupler</strong>.<br />
Jensen [1] had already obtained <strong>an</strong> elliptic integral<br />
<strong>solution</strong> for the power propagating in each<br />
fiber. However, for information processing purposes,<br />
detailed information on both amplitude <strong>an</strong>d<br />
phase is required. In addition, by parametrizing all<br />
the <strong>nonlinear</strong> information in the const<strong>an</strong>ts of<br />
motion, a simpler <strong>solution</strong> is obtained.<br />
2. An <strong>an</strong>alytic <strong>solution</strong><br />
Consider two linear optical fibers coming together<br />
in a <strong>coupler</strong> of <strong>nonlinear</strong> material. <strong>The</strong><br />
equation for the electric field is<br />
o<br />
DE l0e0 2E ot2 ¼ l o<br />
0<br />
2PL ot2 þ l o<br />
0<br />
2PNL ot2 ; ð1Þ<br />
PL <strong>an</strong>d PNL being the linear <strong>an</strong>d <strong>nonlinear</strong> components<br />
of the medium polarization<br />
PLðr; tÞ ¼e0v ð1Þ Eðr; tÞ: ð2Þ<br />
For symmetric molecules (like SiO2) the leading<br />
<strong>nonlinear</strong> term is<br />
PNLðr; tÞ ¼e0v ð3Þ jEðr; tÞj 2 Eðr; tÞ; ð3Þ<br />
where <strong>an</strong> inst<strong>an</strong>t<strong>an</strong>eous <strong>nonlinear</strong> response may be<br />
assumed (except for extremely short pulses) because<br />
in current fibers the electronic contribution<br />
to v ð3Þ occurs on a time scale of 1–10 fs.<br />
1 <strong>The</strong>re have been some speculations [18] that <strong>nonlinear</strong><br />
qu<strong>an</strong>tum(like) effects might endow qu<strong>an</strong>tum computation with<br />
yet additional power.<br />
Separating fast <strong>an</strong>d slow (time) variations<br />
Eðr; tÞ ¼1 2fEðr; tÞe ix0t þ c:c:g;<br />
PNLðr; tÞ ¼1 2fPNLðr; tÞe ix ð4Þ<br />
0t<br />
þ c:c:g;<br />
<strong>an</strong>d using Eqs. (3) <strong>an</strong>d (4) one obtains for the e ix0t part of a tr<strong>an</strong>sversal mode<br />
PNL 1;2 ðr; tÞ ¼ 3e0<br />
8 vð3Þ e ix 0t<br />
jE1;2j 2 þ 2<br />
3 jE2;1j 2 E1;2<br />
þ 1<br />
3 E2;1E2;1E 1;2 þ c:c: : ð5Þ<br />
<strong>The</strong> labels 1 <strong>an</strong>d 2 denote two orthogonal polarizations.<br />
<strong>The</strong> dependence on tr<strong>an</strong>sversal coordinates<br />
ðx; yÞ is separated by considering<br />
E ðiÞ<br />
k ðr; tÞ ¼WðiÞ<br />
k ðx; y; zÞeibiz ix0t<br />
e ; ð6Þ<br />
W ðiÞ<br />
k ðx; y; zÞ being <strong>an</strong> eigenmode of the <strong>coupler</strong> with<br />
slow variation along z<br />
D2W ðiÞ<br />
k þ x2 0<br />
c2 1 þ vð1Þ b ðiÞ2 W ðiÞ<br />
k ¼ 0; ð7Þ<br />
ðiÞ denotes the mode number, k the polarization <strong>an</strong>d<br />
D2 ¼ o2 o2<br />
þ<br />
ox2 oy2 :<br />
Neglecting 2 o2W ðiÞ =oz2 one obtains<br />
ðiÞ<br />
oWðiÞ<br />
1;2<br />
2ib<br />
oz ¼ 3x2 0<br />
4c<br />
2 vð3Þ<br />
W ðiÞ<br />
2<br />
1;2 þ 2<br />
3 WðiÞ<br />
2<br />
2;1<br />
W ðiÞ<br />
1;2<br />
þ 1<br />
3 WðiÞ<br />
2;1WðiÞ 2;1WðiÞ 1;2 : ð8Þ<br />
In <strong>directional</strong> <strong>coupler</strong>s the propagating beams are<br />
made to overlap along one of the tr<strong>an</strong>sversal coordinates<br />
ðxÞ. Typically, in the <strong>nonlinear</strong> region of<br />
the <strong>directional</strong> <strong>coupler</strong>, the eigenmodes are symmetric<br />
(+) <strong>an</strong>d <strong>an</strong>tisymmetric ()) functions of x,<br />
the amplitudes in each fiber at the input <strong>an</strong>d output<br />
of the <strong>coupler</strong> being recovered by<br />
W ð1Þ<br />
k<br />
W ð2Þ<br />
k<br />
1<br />
¼<br />
2 WðþÞ<br />
Þ<br />
k þ Wð<br />
k<br />
¼ 1<br />
2 WðþÞ<br />
ð Þ<br />
k Wk ;<br />
:<br />
ð9Þ<br />
2 Justified for slow variations of the refractive index along the<br />
beam axis over dist<strong>an</strong>ces of the order of one wavelength.
An explicit <strong>an</strong>alytic <strong>solution</strong> for the <strong>nonlinear</strong><br />
<strong>coupler</strong> Eq. (8) is now obtained by noticing that it<br />
has two const<strong>an</strong>ts of motion, namely<br />
o<br />
oz WðiÞ<br />
2<br />
1<br />
o<br />
oz WðiÞ<br />
1 WðiÞ<br />
n<br />
2<br />
þ W ðiÞ<br />
2<br />
<strong>The</strong>refore, defining<br />
W ðiÞ<br />
1<br />
2<br />
W ðiÞ<br />
1 WðiÞ<br />
2<br />
þ W ðiÞ<br />
2<br />
2<br />
WðiÞ<br />
1 WðiÞ<br />
2<br />
2<br />
¼ 0;<br />
W ðiÞ<br />
1 WðiÞ<br />
o<br />
2 ¼ 0:<br />
¼ a ðiÞ ;<br />
¼ icðiÞ ;<br />
ð10Þ<br />
ð11Þ<br />
one obtains for the electrical field of the eigenmodes<br />
i oEðiÞ<br />
1<br />
oz<br />
i oEðiÞ<br />
2<br />
oz<br />
with<br />
b ðiÞ<br />
¼ bðiÞE<br />
ðiÞ<br />
1<br />
ik ðiÞ E ðiÞ<br />
2 ;<br />
¼ bðiÞE<br />
ðiÞ<br />
2 þ ikðiÞE ðiÞ<br />
1 ;<br />
¼ b ðiÞ þ 3x2 0<br />
8c 2<br />
k ðiÞ ¼ x2 0<br />
8c2 vð3Þ b ðiÞ cðiÞ :<br />
v ð3Þ<br />
b ðiÞ aðiÞ ;<br />
Notice that, through a ðiÞ <strong>an</strong>d c ðiÞ , b ðiÞ<br />
E ðiÞ<br />
2 z ¼ e ibðiÞ z E ðiÞ<br />
1 0 sin k ðiÞ z<br />
ð12Þ<br />
ð13Þ<br />
<strong>an</strong>d jðiÞ depend<br />
on the material properties, on the geometry<br />
of the mode <strong>an</strong>d also on its intensity. One may<br />
now obtain, for each eigenmode, the input–output<br />
relation of the <strong>nonlinear</strong> <strong>coupler</strong><br />
E ðiÞ<br />
1 ðÞ¼e z<br />
ibðiÞz ðiÞ<br />
E1 0 ð Þcos kðiÞ n<br />
z<br />
E ðiÞ<br />
2 0 sin k ðiÞ o<br />
z ;<br />
n<br />
ð14Þ<br />
þ E ðiÞ<br />
2 0 cos k ðiÞ z<br />
(ðiÞ ¼ ðþÞ or ())), the <strong>nonlinear</strong>ity being embedded<br />
into b ðiÞ<br />
<strong>an</strong>d k ðiÞ<br />
b ðiÞ<br />
¼ b ðiÞ þ 3x2 0<br />
8c 2<br />
k ðiÞ ¼ x2 0<br />
4c 2<br />
v ð3Þ<br />
v ð3Þ<br />
b ðiÞ<br />
Im EðiÞ<br />
ðiÞ 1<br />
b<br />
E ðiÞ<br />
2<br />
ð Þ þ E ðiÞ<br />
2<br />
ð Þ<br />
1 0<br />
ð0ÞE ðiÞ<br />
ð Þ :<br />
R. Vilela Mendes / Optics Communications 232 (2004) 425–427 427<br />
2 0<br />
2 0<br />
;<br />
ð15Þ<br />
To obtain the corresponding input–output relations<br />
in the two fibers (1) <strong>an</strong>d (2) one uses Eq.<br />
(9), namely<br />
E ð1Þ<br />
k ðzÞ ¼1<br />
2 EðþÞ<br />
ð Þ<br />
k ðÞþ z Ek ðzÞ ;<br />
E ð2Þ<br />
k ðzÞ ¼1<br />
2 EðþÞ<br />
k ðÞ z<br />
ð Þ<br />
Ek ðzÞ :<br />
ð16Þ<br />
In conclusion, Eqs. (14)–(16) provide <strong>an</strong> <strong>an</strong>alytic<br />
<strong>solution</strong> for the <strong>nonlinear</strong> <strong>directional</strong> <strong>coupler</strong>, from<br />
which phases <strong>an</strong>d polarizations may be obtained<br />
explicitly. Through the const<strong>an</strong>ts of motion a ðiÞ <strong>an</strong>d<br />
c ðiÞ , which depend on the material properties <strong>an</strong>d<br />
the geometry of the mode, the <strong>nonlinear</strong>ity of the<br />
system is completely parametrized. <strong>The</strong> <strong>solution</strong><br />
(14)–(16) may then be used as a convenient guide<br />
for experimental implementation.<br />
References<br />
[1] S.M. Jensen, IEEE J. Qu<strong>an</strong>tum Electron. QE-18 (1982)<br />
1580.<br />
[2] Y. Silberberg, G.I. Stegem<strong>an</strong>, Appl. Phys. Lett. 50 (1987)<br />
801.<br />
[3] G.I. Stegem<strong>an</strong>, C.T. Seaton, A.C. Walker, C.N. Ironside,<br />
Opt. Commun. 61 (1987) 277.<br />
[4] G.I. Stegem<strong>an</strong>, E. Caglioti, S. Trillo, S. Wabnitz, Opt.<br />
Commun. 63 (1987) 281.<br />
[5] J.S. Aitchison, A.H. Ke<strong>an</strong>, C.N. Ironside, A. Villeneuve,<br />
G.I. Stegem<strong>an</strong>, Electron. Lett. 27 (1991) 1709.<br />
[6] A.M. Kenis, I. Vorobeichik, M. Orenstein, N. Moiseyev,<br />
IEEE J. Qu<strong>an</strong>tum Electron. 37 (2001) 1321.<br />
[7] G.J. Liu, B.M. Li<strong>an</strong>g, Q. Li, G.L. Jin, Opt. Commun. 218<br />
(2003) 113.<br />
[8] Y. W<strong>an</strong>g, J. Liu, IEEE Photon. Technol. Lett. 11 (1999) 72.<br />
[9] G.J. Milburn, Phys. Rev. Lett. 62 (1989) 2124.<br />
[10] S. Gl<strong>an</strong>cy, J.M. LoSecco, C.E. T<strong>an</strong>ner. Available form<br />
qu<strong>an</strong>t-ph/0009110.<br />
[11] J. Fu. Available from qu<strong>an</strong>t-ph/0211038.<br />
[12] E. Knill, R. Laflamme, G. Milburn, Nature 409 (2001) 46.<br />
[13] M.A. M<strong>an</strong>Õko, V.I. M<strong>an</strong>Õko, R. Vilela Mendes, Phys. Lett.<br />
A 288 (2001) 132.<br />
[14] M.A. M<strong>an</strong>Õko, in: P. Chen (Ed.), Qu<strong>an</strong>tum Aspects of<br />
Beam Physics, World Scientific, Singapore, 2002.<br />
[15] R. Fedele, P.K. Shukla (Eds.), Qu<strong>an</strong>tum Like Models <strong>an</strong>d<br />
Coherent Effects, World Scientific, Singapore, 1995.<br />
[16] S. De Martino, S. De Nicola, S. De Siena, R. Fedele, G.<br />
Miele (Eds.), New perspectives in Physics of Mesoscopic<br />
Systems: Qu<strong>an</strong>tumlike Descriptions <strong>an</strong>d Macroscopical<br />
Coherence Phenomena, World Scientific, Singapore, 1997.<br />
[17] T.C. Ralph, A. Gilchrist, G.J. Milburn, W.J. Munro, S.<br />
Gl<strong>an</strong>cy. Availabele from qu<strong>an</strong>t-ph/0306004.<br />
[18] D.S. Abrams, S. Lloyd, Phys. Rev. Lett. 18 (1998) 3992.