1632 J. Opt. Soc. Am. A / Vol. 19, No. 8 / August 2002 Kulishov et al.

Electronically reconfigurable superimposed

waveguide long-period gratings

Mykola Kulishov
Adtek Photomask, Inc., 4950 Fisher Street, Montréal, Québec H4T 1J6, Canada

Xavier Daxhelet
École Polytechnique de Montréal, Département de Génie Physique, C.P. 6079, Station Centre-Ville, Montréal,
Québec H3C 3A7, Canada

Mounir Gaidi and Mohamed Chaker

Institut National de la Recherche Scientifique, Energie et Matériaux, 1650 Boulevard Lionel Boulet, Varennes,
Québec J3X 1S2, Canada

Received November 2, 2001; revised manuscript received January 30, 2002; accepted February 1, 2002
The perturbation to the refractive index induced by a periodic electric field from two systems of interdigitated
electrodes with the electrode-finger period l is analyzed for a waveguide with an electro-optically (EO) active
core–cladding. It is shown that the electric field induces two superimposed transmissive refractive-index
gratings with different symmetries of their cross-section distributions. One of these gratings has a constant
component of an EO-induced refractive index along with its variable component with periodicity l, whereas the
second grating possesses only a variable component with periodicity 2l. With the proper waveguide design,
the gratings provide interaction between a guided fundamental core mode and two guided cladding modes.
Through the externally applied electric potential, these gratings can be independently switched ON and OFF,
or they can be activated simultaneously with electronically controlled weighting factors. Coupling coefficients
of both gratings are analyzed in terms of their dependence on the electrode duty ratio and dielectric permit-
tivities of the core and cladding. The coupled-wave equations for the superimposed gratings are written and
solved. The spectral characteristics are investigated by numerical simulation. It is found that the spectral
characteristics are described by a dual-dip transmission spectrum with individual electronic control of the dip
depths and positions. Within the concept, a new external potential application scheme is described in which
the symmetry of the cross-sectional distribution of the refractive index provides coupling only between the core
mode and the cladding modes, preventing interaction of the cladding modes with each another. This simple
concept opens opportunities for developing a number of tunable devices for integrated optics by use of the pro-
posed design as a building block. © 2002 Optical Society of America
OCIS codes: 130.2790, 050.1950, 060.1810, 160.2100, 130.3120.

1. INTRODUCTION ibility of direct electronic adjustment of all controlled pa-

Rapid expansion of information traffic volume has led to rameters.
the widespread adoption of dense wavelength division Recently the tunable long-period grating (LPG) devices,
multiplexing as a method of expanding the capacity of ex- which utilize the thermal sensitivity of LPGs by means of
isting fiber networks. Dense wavelength division multi- heating, have been extensively discussed.2,3 The tuning
plexing creates the need for a range of components with principles rely mainly on thermal control of a LPG, which
the ability to multiplex, demultiplex, switch, route, moni- already was fabricated by an ultraviolet light exposure.
tor, and attenuate each individual wavelength within the They are based on the detuning of the LPG resonant
packet of wavelengths traveling down the fiber. These wavelength since the modulation depth of the refractive
needs place considerable pressure on the passive optical index in the fiber core remains even with temperature
devices currently used for optical processing. Therefore variations. An electrically controllable fiber-optic filter
optical components, such as waveguide–fiber gratings, in- for broadband rejection has been experimentally
tegrated into electronically controllable geometry will demonstrated4 by the periodically poling of a liquid-
probably form the basis for the next generation of pro- crystalline core in a hollow-core fiber by means of external
grammable optical devices. In these devices the interdigitated electrodes. Standard photolithography
refractive-index distribution will be dynamically con- techniques make it easy to pattern an interdigitated elec-
trolled in accordance with the network operating trode array that is capable of inducing periodical electric
conditions.1 We believe that electro-optic (EO) control field distributions in an EO waveguide with the periodic-
that consists in inducing refractive-index perturbations in ity of transmissive (or long-period) gratings. Specially
EO material with the help of an external electric field pro- tailored LPGs can flatten the gain profile of erbium-doped
vides the best solution in terms of time response and flex- fiber amplifiers and can be used as band-rejection filters

0740-3232/2002/081632-17$15.00 © 2002 Optical Society of America

Kulishov et al. Vol. 19, No. 8 / August 2002 / J. Opt. Soc. Am. A 1633

in a variety of applications. They also can be employed

in dispersion control.5,6
Superimposed fiber Bragg gratings recently became a
subject of intensive study7–10 and have been proposed for
a broad spectrum of applications such as in dual-
wavelength filters,7,8 sensors,8 and dual-wavelength
distributed-feedback lasers8 and also for beam splitting,
combining, and cross coupling.9 Our analysis is dedi-
cated to superimposed LPGs in the case in which their
relative coupling strengths can be electronically adjusted.
Developing this kind of EO tunable grating device re-
quires powerful computer simulation tools that will be
able to relate the parameters of the structure, including
material and geometrical parameters, in the process of
device optimization. An integral part of the simulation
software tools is an electrostatic problem solver able to
calculate the electric field distribution as well as the re-
sulting EO-induced refractive-index change in the wave-
guide or cladding parts of the device. In our previous
publications11,12 we described a new concept of EO-
induced waveguide gratings. Within this concept, both
the constant and the variable components of the EO-
induced refractive index are independently controlled. A
half-period shift between the interdigitated electrode
structures12 results in inducing tilted gratings with the
ability to switch an angle of the tilt. However, the cou-
pling properties of these reconfigurable gratings have not Fig. 1. (a) Geometry of the EO-induced waveguide gratings with
yet been studied. interdigitated electrode structure. The electrostatic problem so-
lution for this geometry can be presented as a linear combination
In this paper we analyze a novel, to our knowledge, of the Laplace’s equation solutions for the structures shown in (b)
tunable filter that is based on EO induction of two types and (c).
of periodical refractive-index distribution in a waveguide
with the opportunity to activate each of them individually
or both simultaneously with a controllable weighting fac- the waveguide are assumed to be infinitely thin. This
tor. The reminder of this paper will be organized as fol- approximation is justified because electrode thickness is
lows. In Section 2 we analyze EO-induced periodical much smaller than other dimensions of the structure,
refractive-index distributions. The rigorous solution of and, as it has already been shown, their thickness does
the electrostatic boundary problem will be presented for not affect the field distribution inside the core and
the electric field in the waveguide core–cladding. The cladding.13 The length of the electrode fingers is much
choice of the waveguide parameters will be justified in larger than the electrode period; therefore the problem
terms of their optimal TE-mode interaction with EO- can be treated as a two-dimensional one. The number of
induced refractive-index distributions. In Section 3 the the electrode fingers is large; therefore end boundary ef-
analysis will be extended to the mode interaction through fects are neglected. dc voltages (⫹V 0 , V 0 ⫹ ⌬V) and
the EO-induced refractive-index gratings. The interac- 关 ⫺V 0 ,⫺(V 0 ⫹ ⌬V) 兴 are applied alternately on the top
tion is modeled by a set of coupled-wave equations with and the bottom electrodes, respectively. The five-
electronically controlled coupling coefficients. Their de- layered symmetric geometry 关 (d ⫹ h) ⭐ z ⬍ ⬁, i ⫽ 1;
pendence on the interdigitated electrode duty ratio and h ⭐ z ⬍ (d ⫹ h), i ⫽ 2; ⫺h ⭐ z ⭐ ⫹ h, i ⫽ 3; ⫺(h
waveguide dielectric characteristics is considered. The ⫹ d) ⭐ z ⬍ ⫺h, i ⫽ 4;⫺⬁ ⬍ z ⬍ ⫺(d ⫹ h), i ⫽ 5] is
reconfigurable properties of the transmission spectra are analyzed. The EO core (i ⫽ 3) and cladding layers (i
demonstrated, and two more design modifications suit- ⫽ 2 and 4) are assumed to be EO polymers poled in the z
able for certain applications are presented. It is shown direction; therefore their principal dielectric axes are
that for the proposed design the EO gratings do not sup- aligned along the x and the z directions such that the di-
port interaction between modes with different polariza- electric properties can be described by two diagonal rela-
tions (TE–TM conversion). Finally, concluding remarks tive permittivities ⑀ xx
( co)
, ⑀ zz
( co)
and ⑀ (xxcl) , ⑀ zz
( cl)
for the core and
are given in Section 4. the cladding, respectively. The electrostatic nature of the
analysis permits the association of an electric potential
␾ ( i ) (x, z) with each region i (i ⫽ 1, 2, 3, 4, 5), and the
2. ELECTRO-OPTICALLY INDUCED PHASE two-dimensional Laplace equation can be written for each
GRATINGS region i as
The geometry of the problem to be analyzed is shown in
Fig. 1, in which 2h and d are the core and cladding thick-
⳵ 2␾ 共i兲 ⳵ 2␾ 共i兲
nesses, respectively, l is the electrode spatial period, and a ⑀ 共xxi 兲 ⫹ ⑀ 共zzi 兲 ⫽ 0, (1)
is the electrode width. The electrodes on both sides of ⳵x2 ⳵z2
1634 J. Opt. Soc. Am. A / Vol. 19, No. 8 / August 2002 Kulishov et al.

where ⑀ (xxi ) , ⑀ zz
冋 册
(i) ⬁
are the two diagonal permittivities of re-
gion i. If region i is isotropic, then ⑀ xx (i)
⫽ ⑀ zz
(i)
⫽ ⑀ (i). ␾ 共b3 兲 共 x, z 兲 ⫽ V 0 B 0z ⫹ 兺
n⫽1
E n sinh共 nk ␦ 3 z 兲 cos共 nkx 兲 ,
Owing to the linear nature of the Laplace equation, the
solution of the electrostatic problem of Fig. 1(a) can be (6)

冠
written as a linear combination of the solutions for the
⬁
electrostatic problems presented in Figs. 1(b) and 1(c): ⑀ 共zzco兲
␾ 共b4 兲 共 x, z 兲 ⫽ V 0 ⫺B 0 h ⫹ B 0共 z ⫹ h 兲 ⫹ 兺 En

冉 ⌬V
冊 ⌬V ⑀ 共zzcl兲 n⫽1

再
␾ 共 i 兲 共 x, z 兲 ⫽ 1 ⫹ ␾ 共bi 兲 共 x, z 兲 ⫹ ␾ 共ci 兲 共 x, z 兲 .
2V 0 2V 0
(2) ⫻ sinh共 nkh ␦ 3 兲 cosh关 nk ␦ 2 共 z ⫹ h 兲兴

冎
The symmetry of the applied voltage for the first struc-
ture in Fig. 1(b) creates antisymmetric electric potential ⑀ 共zzco兲 ␦ 3
⫺ cosh共 nkh ␦ 3 兲 sinh关 nk ␦ 2 共 z ⫹ h 兲兴
distribution ␾ b( i ) (x, z) across the waveguide or symmetric ⑀ 共zzcl兲 ␦ 2

冡
z component of the electric field E z (x, z) ⫽ ⫺⳵ ␾ b( i ) / ⳵ z
with a period equal to interdigitated electrode pitch l. At
the same time, electric potential distribution ␾ (ci ) (x, z) for ⫻ cos共 nkx 兲 , (7)
the second structure in Fig. 1(c) is a symmetric one with

冋
its antisymmetric z-electric field component that is re-
peated in the x direction with periodicity 2l. In the gen- ⑀ 共zz3 兲
eral case of Fig. 1(a) these two distributions are superim- ␾ 共b5 兲 共 x, z 兲 ⫽ V 0 ⫺B 0 h ⫺ B 0d
⑀ 共zz2 兲
posed with weighting factors (V 0 ⫹ ⌬V/2) and ⌬V/2,

册
⬁
respectively. Therefore, to find the solution for the struc-
ture in Fig. 1(a), we must solve the electrostatic problems ⫺
n⫽1
兺 C n exp共 nkz 兲 cos共 nkx 兲 , (8)
of Figs. 1(b) and 1(c). The problems have common bound-
ary conditions for z ⫽ ⫾h interfaces:
where ␦ 2 ⫽ 关 ⑀ xx / ⑀ zz 兴 , ␦ 3 ⫽ 关 ⑀ (xxco) / ⑀ zz
( cl) ( cl) 1/2 ( co) 1/2
兴 , and k
共2兲 共3兲 ⫽ 2 ␲ /l. These functions are eigenfunctions of the
␾ 共 x, h 兲 ⫽ ␾ 共 x, h 兲 ,
Laplace equation (1) over domain ⫺l/2 ⭐ x ⭐ ⫹1/2,
␾ 共 3 兲 共 x, ⫺h 兲 ⫽ ␾ 共 4 兲 共 x, ⫺h 兲 , where l is the interdigitated electrode period. For the
given problem symmetry, we use a cosine Fourier series
⳵ ␾ 共 2 兲 共 x, h 兲 ⳵ ␾ 共13 兲 共 x, h 兲 that narrows the domain to 0 ⭐ x ⭐ ⫹l/2. By use of the
⑀ 共zzcl兲 ⫽ ⑀ 共zzco兲 , boundary conditions for the interface z ⫽ h ⫹ d,
⳵z ⳵z

⳵ ␾ 共 3 兲 共 x, ⫺h 兲 ⳵ ␾ 共 4 兲 共 x, ⫺h 兲 ␾ 共b1 兲 共 x, h ⫹ d 兲 ⫽ ␾ 共b2 兲 共 x, h ⫹ d 兲 ,
⑀ 共zzco兲 ⫽ ⑀ 共zzcl兲 . (3)
⳵z ⳵z
␾ 共b2 兲 共 x, h ⫹ d 兲 ⫽ V 0 , 兩 x ⫺ ml 兩 ⭐ a/2,
A. First Electrostatic Problem: ⌬V Ä 0
Taking into account the problem symmetry and boundary ⳵ ␾ 共b1 兲 共 x, h ⫹ d 兲 ⳵ ␾ 共b2 兲 共 x, h ⫹ d 兲
conditions [Eqs. (3)], we can write the electrostatic poten- ⑀共s兲 ⫽ ⑀ 共zzcl兲 ,
⳵z ⳵z
tial as an expansion in terms of harmonic and hyperbolic
functions13: a/2 ⬍ 兩 x ⫺ ml 兩 ⭐ l/2, (9)

␾ 共b1 兲 共 x, z 兲 ⫽ V 0 B 0 h ⫹ 冋 ⑀ 共zzco兲
⑀ 共zzcl兲
B 0d
which are satisfied simultaneously for the interface z
⫽ ⫺(d ⫹ h), the coefficients C n can be expressed
through E n , and the problem is reduced to the dual-series

册
⬁ equations with respect to B n :
⫹ 兺 C n exp共 ⫺nkz 兲 cos共 nkx 兲 , (4)

冋 册
n⫽1 ⬁
⑀ 共zzco兲
兺
冠
⬁ B0 h ⫹ d ⫹ E n F n cos共 nkx 兲 ⫽ 1,
⑀ 共zzco兲 ⑀ 共zzcl兲
兺
n⫽1
␾ 共b2 兲 共 x, z 兲 ⫽ V 0 B 0h ⫹ B 0共 z ⫺ h 兲 ⫹ En
⑀ 共zzcl兲 n⫽1

再
兩 x ⫺ ml 兩 ⭐ a/2,

⫻ sinh共 nkh ␦ 3 兲 cosh关 nk ␦ 2 共 z ⫺ h 兲兴 ⬁

⑀ 共zzco兲 B 0 h ⫹ 兺 nkhE n 关 ⑀ 共 s 兲 F n ⫹ ␦ 2 ⑀ 共zzcl兲 G n 兴 cos共 nkx 兲 ⫽ 0,

⫹
⑀ 共zzco兲 ␦ 3
⑀ 共zzcl兲 ␦ 2
cosh共 nkh ␦ 3 兲 sinh关 nk ␦ 2 共 z ⫺ h 兲兴 冎 n⫽1

a/2 ⬍ 兩 x ⫺ ml 兩 ⭐ l/2, (10)

⫻ cos共 nkx 兲 , 冡 (5) where

Kulishov et al. Vol. 19, No. 8 / August 2002 / J. Opt. Soc. Am. A 1635

冕
⬁
⑀ 共zzco兲 ␦ 3 2 a/2 cos共 nkx 兲 cos共 nk ␰ 兲
F n ⫽ sinh共 nkh ␦ 3 兲 cosh共 nk ␦ 2 d 兲 ⫹
⑀ 共zzcl兲 ␦ 2
cosh共 nkh ␦ 3 兲 ⫺
␲h 0
␳ 1共 ␰ 兲 兺
n⫽1 n
d␰

⫻ sinh共 nk ␦ 2 d 兲 ,

G n ⫽ sinh共 nkh ␦ 3 兲 sinh共 nk ␦ 2 d 兲 ⫹

⑀ 共zzco兲 ␦ 3
cosh共 nkh ␦ 3 兲
⫽ 再 1 ⫺ B0 h ⫹ d 冋 ⑀ 共zzco兲
⑀ 共zzcl兲
册冎 关 ⑀ 共zzcl兲 ␦ 2 ⫹ ⑀ 共 s 兲 兴

⑀ 共zzcl兲 ␦ 2
冕
⬁
2 Rn a/2

⫻ cosh共 nk ␦ 2 d 兲 , (11)
⫹
␲h
兺 n⫽1 n
cos共 nkx 兲
0
␳ 1 共 ␰ 兲 cos共 nk ␰ 兲 d␰ ,

and ⑀ ( s ) is the dielectric constant of the substrate and su-

perstrate. Equations (10) can be rewritten in terms of 0 ⭐ x ⭐ a/2. (19)
the new coefficients E * n : The kernel in the above integral equation can be

冋 册
⬁ summed14:
⑀ 共zzco兲 E n* F n
B0 h ⫹ d
⑀ 共zzcl兲
⫹ 兺 ⑀ F n ⫹ ␦ 2 ⑀ 共zzcl兲 G n
共s兲
cos共 nkx 兲 ⫽ 1, ⬁
cos共 nkx 兲 cos共 nk ␰ 兲 1
兺
n⫽1
⫽ ⫺ ln关 兩 cos共 kx 兲 ⫺ cos共 k ␰ 兲 兩 兴 .
兩 x ⫺ ml 兩 ⭐ a/2, n⫽1 n 2
(20)
⬁

⑀ 共zzco兲 B 0 h ⫹ 兺
n⫽1
nkhE n* cos共 nkx 兲 ⫽ 0,
Equation (19) is a singular integral equation with a loga-
rithmic kernel. The Schwinger method is used to solve
it. The method consists of a change of the variables15:
a/2 ⬍ 兩 x ⫺ ml 兩 ⭐ l/2, (12)
cos共 kx 兲 ⫽ cos2 共 ka/4兲 ⫹ sin2 共 ka/4兲 cos共 k ␨ 兲 ,
where
cos共 k ␰ 兲 ⫽ cos2 共 ka/4兲 ⫹ sin2 共 ka/4兲 cos共 k ␩ 兲 , (21)
E n* ⫽ E n 关 ⑀ 共 s 兲 F n ⫹ ␦ 2 ⑀ 共zzcl兲 G n 兴 . (13)
To improve convergence, we can regroup the first equa- where the new variables ␨ and ␩ cover the entire range
tion of the dual series [Eqs. (12)] with the residual terms: from 0 to l/2, whereas x and ␰ vary from 0 to a/2. The
solution is sought as a Fourier expansion of the unknown

再 冋 册冎
⬁
⑀ 共zzco兲 function ␳ 1 ( ␰ ) in terms of new variables,
兺 E n* cos共 nkx 兲 ⫽ 1 ⫺ B0 h ⫹ d
⑀ 共zzcl兲
关 ⑀ 共zzcl兲 ␦ 2 ⫹ ⑀ 共 s 兲 兴
⬁
n⫽1 d␰
⬁
␳ 1 关 ␰ 共 ␨ 兲兴
d␨
⫽ 兺
m⫽0
a m cos共 mk ␨ 兲 . (22)
⫹ 兺
n⫽1
R n E n* cos共 nkx 兲 ,
Taking into account the orthogonality of the harmonic
function and the following expansions,
兩 x ⫺ ml 兩 ⭐ a/2, (14)
where
R n ⫽ 1 ⫺ 关 ⑀ 共 s 兲 ⫹ ⑀ 共zzcl兲 ␦ 2 兴 F n ,
ln关 兩 cos共 kx 兲 ⫺ cos共 k ␰ 兲 兩 兴 ⫽ ln sin2 冉 ka
4
冊
⬁
cos共 pk ␨ 兲 cos共 pk ␩ 兲
lim F n ⫽ 关 ⑀ 共 s 兲 ⫹ ␦ ⑀ 共zzcl兲 兴 ⫺1 ,
n→⬁
lim R n ⫽ 0.
n→⬁
⫺2 兺
p⫽1 p
,
(15)
n

兺b
The solution is based on the calculation of an auxiliary 共n兲
cos共 nkx 兲 ⫽ s cos共 sk ␩ 兲 , (23)
function that describes, within a constant factor, the den- s⫽1
sity of the electric charge distribution on the electrodes:
⬁ Eq. (19) is reduced to the following form:
␳ 1共 x 兲 ⫽ ⑀ 共zzco兲 B 0 h ⫹ 兺 nkhE n* cos共 nkx 兲 .
冉 冊
(16) ⬁
ka am
n⫽1
⫺a 0 ln sin2
4
⫹ 兺
m⫽1 m
cos共 nk ␩ 兲
Then the coefficients B 0 and E n* are expressed in terms of

再 冋 册冎
␳ 1 ( ␰ ) as follows:
d ⑀ 共zzco兲

冕
2 a/2 ⫽ 1 ⫺ B0 1 ⫹ 关 ⑀ 共zzcl兲 ␦ 2 ⫹ ⑀ 共 s 兲 兴 kh
h ⑀ 共zzcl兲
⑀ 共zzco兲 B 0 h ⫽ ␳ 1 共 ␰ 兲 d␰ , (17)

冋 册
l 0 n ⬁ ⬁
Rn
兺 兺a兺
E n* ⫽
␲ nh
2
冕 0
a/2
␳ 1 共 ␰ 兲 cos共 nk ␰ 兲 d␰ . (18)
⫹
m⫽0 s⫽0
s
n⫽1 n
b 共mn 兲 b 共sn 兲 cos共 mk ␩ 兲 ,

0 ⭐ ␨ ⭐ l/2. (24)
Substituting Eq. (18) into Eq. (14) and changing the suc-
cession of summation and integration, we obtain the fol- Keeping in mind that ⑀ zz
( co)
⫽ a 0 [ see Eq. (17)], we can
B 0h
lowing integral equation with respect to ␳ 1 ( ␰ ): pick out the constant part of the above equation as
1636 J. Opt. Soc. Am. A / Vol. 19, No. 8 / August 2002 Kulishov et al.

a0 再 ⑀ 共zzcl兲 ␦ 2 ⫹ ⑀ 共 s 兲
⑀ 共zzco兲
冋 1⫹
d ⑀ 共zzco兲
h ⑀ 共zzcl兲
册 kh ⫺ ln sin 冉 2
ka
4
冊 AZ ms ⫽ 兺
⬁

n⫽1
Rn
n
b 共mn 兲 b 共sn 兲 , AZ m ⫽ 兺
⬁

n⫽1
Rn
n
b 共mn 兲 b 共0n 兲 ,

⫺ 2 兺
n⫽1
⬁
Rn
n
关 b 共0n 兲 兴 2
⌬⫽
关 ⑀ 共zzcl兲 ⫹ ⑀ 共 s 兲 兴
⑀ 共zzco兲
kh 1 ⫹ 冋 d ⑀ 共zzco兲
h ⑀ 共zzcl兲
册 冉
⫺ ln sin2
ka
4
冊
⬁
Rn
⬁ ⬁
Rn ⫺2 兺 关 b 共0n 兲 兴 2 . (28)
⫽ 关 ⑀ 共zzcl兲 ␦ 2 ⫹⑀ 共s兲
兴 kh ⫹ 兺a兺
s⫽1
s
n⫽1 n
b 共0n 兲 b 共sn 兲 . (25) n⫽1 n
The system (27) is solved numerically by truncating the
number of equations after a finite number of equations.
Then, multiplying Eq. (25) by (4/l)cos( pk␨) and integrat-
Having found the coefficients a m , we can define the coef-
ing by ␨ from 0 to l/2, we can derive the variable part of
ficients E n by substituting Eq. (22) into Eq. (18):
the equation:
n

冉 冊 2a 0 b 共0n 兲 ⫹ 兺 a m b 共mn 兲
⬁ ⬁ ⬁
Rn ␦ ms Rn
兺 兺 n
b 共mn 兲 b 共sn 兲 ⫺
m
a m ⫽ ⫺2a 0 兺 n
b 共0n 兲 b 共sn 兲 ,
nkhE n ⫽
m⫽1
. (29)
m⫽1 n⫽1 n⫽1
⑀ 共 s 兲 F n ⫹ ␦ 2 ⑀ 共zzcl兲 G n
(26)
Knowing the spatial harmonics of the potential, we can
where ␦ ms is the Kronecker symbol. After elimination of find the electric field in each region from
a 0 , Eq. (26) is reduced to the algebraic equation system to
be solved: ⳵ ␾ 共 i 兲 共 x, z 兲 ⳵ ␾ 共 i 兲 共 x, z 兲
E 共xi 兲 共 x, z 兲 ⫽ ⫺ , E 共zi 兲 共 x, z 兲 ⫽ ⫺ .
⳵x ⳵z

冉 冊
⬁ (30)
AZ m AZ s ␦ ms
兺
m⫽1
AZ ms ⫹
⌬
⫺
m
am In Fig. 2 the calculated distribution of the potential ␾ b( 4 )
⫻ (x, h ⫹ d)/V0 and charge density ␳ 1 (x)h/V 0 are shown
for two different values of the electrode duty ratio a/l
AZ s
⫽ ⫺2 关 ⑀ 共zzcl兲 ␦ 2 ⫹ ⑀ 共 s 兲 兴 kh, (27) ⫽ 0.2 and a/l ⫽ 0.6. One can see that the electric po-
⌬ tential keeps constant on the electrodes, while the charge
density is close to zero between electrodes and accumu-
where lates at the electrode edges. The charge density as well

Fig. 2. Boundary conditions for the normalized electric potential ␾ b(4) (x, h ⫹ d)/V 0 and the normalized charge density ␳ 1 (x)h/V 0 for (a)
and (c) narrow electrodes (a/l ⫽ 0.2) and (b) and (d) wide electrode (a/l ⫽ 0.6) for the following parameter set: 2h ⫽ 0.0745l, d
⫽ 1.376h, ⑀ (s) ⫽ 3.5, ⑀ (cl)
xx ⫽ ⑀ zz ⫽ 4, ⑀ xx ⫽ 8.4, ⑀ zz ⫽ 5.8, and N ⫽ 40.
cl (co) (co)
Kulishov et al. Vol. 19, No. 8 / August 2002 / J. Opt. Soc. Am. A 1637

as the electric field should approach infinity exactly at the ⬁

edges. However, owing to the truncation of the field se-

ries (N ⫽ 40), it has a finite value.
兺
n⫽0
D n* T n cos关共 n ⫹ 1/2兲 kx 兴 ⫽ 1,

0 ⬍ x ⭐ a/2,

B. Second Electrostatic Problem: ⌬V Ä ⫺2V 0 ⬁

Symmetry of this problem [Fig. 1(c)] prompts us to look
for the solution for the electrostatic potential in the fol-
兺 共 n ⫹ 1/2兲 D * cos关共 n ⫹ 1/2兲 kx 兴 ⫽ 0,
n⫽0
n

lowing form:
a/2 ⬍ x ⭐ l/2, (36)
⬁ where
␾ 共c1 兲 共 x, z 兲 ⫽ V 0 兺
n⫽0
A n exp关 ⫺共 n ⫹ 1/2兲 kz 兴
D n* ⫽ D n 关 ⑀ 共 s 兲 M n ⫹ ⑀ 共zzcl兲 ␦ 2 N n 兴 ,
⫻ cos关共 n ⫹ 1/2兲 kx 兴 , (31) Mn
Tn ⫽ ,
⑀ 共 s 兲 M n ⫹ ⑀ 共zzcl兲 ␦ 2 N n

再
⬁

␾ 共c2 兲 共 x, z 兲 ⫽ V 0 兺
n⫽0
D n cosh关共 n ⫹ 1/2兲 kh ␦ 3 兴 N n ⫽ cosh关共 n ⫹ 1/2兲 kh ␦ 3 兴
⫻ sinh关共 n ⫹ 1/2兲 k ␦ 2 d 兴
⫻ cosh关共 n ⫹ 1/2兲 k ␦ 2 共 z ⫺ h 兲兴
⑀ 共zzco兲 ␦ 3
⑀ 共zzco兲 ␦ 3 ⫹ sinh关共 n ⫹ 1/2兲 kh ␦ 3 兴
⫹ sinh关共 n ⫹ 1/2兲 kh ␦ 3 兴 ⑀ 共zzcl兲 ␦ 2
⑀ 共zzcl兲 ␦ 2
⫻ cosh关共 n ⫹ 1/2兲 k ␦ 2 d 兴 ,

⫻ sinh关共 n ⫹ 1/2兲 k ␦ 2 共 z ⫺ h 兲兴 冎 M n ⫽ cosh关共 n ⫹ 1/2兲 kh ␦ 3 兴

⑀ 共zzco兲 ␦ 3
⫻ cos关共 n ⫹ 1/2兲 kx 兴 , (32) ⫻ cosh关共 n ⫹ 1/2兲 k ␦ 2 d 兴 ⫹
⑀ 共zzcl兲 ␦ 2
⬁ ⫻ sinh关共 n ⫹ 1/2兲 kh ␦ 3 兴
␾ 共c3 兲 共 x, z兲 ⫽ V0 兺
n⫽0
D n sinh关共 n ⫹ 1/2兲 ␦ 3 kz 兴
⫻ cosh关共 n ⫹ 1/2兲 k ␦ 2 d 兴 . (37)

⫻ cos关共 n ⫹ 1/2兲 kx 兴 , (33) We reduced our problem to the dual-serial equations

(36) that were solved in the recent paper by Kulishov
et al.12 In this solution we should put the shift d be-

再
⬁
tween the top and the bottom electrode structures to
␾ 共c4 兲 共 x, z兲 ⫽ V0 兺
n⫽0
D n cosh关共 n ⫹ 1/2兲 kh ␦ 3 兴 equal zero. For this particular case the final infinite sys-
tem of algebraic equations with respect to D n* can be sim-
⫻ cosh关共 n ⫹ 1/2兲 k ␦ 2 共 z ⫹ h 兲兴 plified and reduced to the following form:
⬁
⑀ 共zzco兲 ␦ 3 P k 关 cos共 ␲ a/l 兲兴
⫺
⑀ 共zzcl兲 ␦ 2
sinh关共 n ⫹ 1/2兲 kh ␦ 3 兴 D k* ⫽
共 k ⫹ 1/2兲 K 关 cos共 ␲ a/2l 兲兴
⫺ 兺
n⫽0
D n* 共 T n ⫺ 1 兲 ␣ nk ,

冎
(38)
⫻ sinh关共 n ⫹ 1/2兲 k ␦ 2 共 z ⫹ h 兲兴 where P k (cos ␰) are Legendre polynomials, K(cos ␰) is a
complete elliptic integral of the first kind, and
⫻ cos关共 n ⫹ 1/2兲 kx 兴 ,
冕
(34) ␲ a/l
␣ nk ⫽ 共 k ⫹ 1/2兲 P n 共 cos ␰ 兲 P k 共 cos ␰ 兲 sin ␰ d␰ . (39)
0
⬁

␾ 共c5 兲 共 x, z 兲 ⫽ V 0 兺
n⫽0
A n exp关共 n ⫹ 1/2兲 kz 兴 The equation system (38) is also solved numerically by
our truncating it after a finite number N of equations.
⫻ cos关共 n ⫹ 1/2兲 kx 兴 . (35) The solution technique provides good convergence as one
can see from Fig. 3, in which the electric potential distri-
bution ␸ (c4 ) (x, h ⫹ d)/V 0 and charge density ␳ 2 (x)h/V 0
By use of boundary conditions for the interface z ⫽ h are shown for the same values of the electrode duty ratio,
⫹ d that are similar to those for Eqs. (9), the coefficients a/l ⫽ 0.2 and a/l ⫽ 0.6, as in Fig. 2.
A n can be expressed through D n , and the problem is re- Having solved both problems, we get the solution for
duced to the dual-series equations with respect to D n : the electric field in the waveguide core shown in Fig. 1(a):
1638 J. Opt. Soc. Am. A / Vol. 19, No. 8 / August 2002 Kulishov et al.

Fig. 3. Boundary conditions for the normalized electric potential ␾ c(4) (x, h ⫹ d)/V 0 and the normalized charge density ␳ 2 (x)h/V 0 for (a)
and (c) narrow electrodes (a/l ⫽ 0.2) and (b) and (d) wide electrode (a/l ⫽ 0.6) for the following parameter set: 2h ⫽ 0.0745l, d
⫽ 1.376h, ⑀ (s) ⫽ 3.5, ⑀ (cl)
xx ⫽ ⑀ zz ⫽4; ⑀ xx ⫽ 8.4, ⑀ zz ⫽ 5.8, and N ⫽ 40.
(cl) (co) (co)

E 共z3 兲 共 x, z兲 ⫽
⌬V
2V 0
E 共cz3 兲 共 x, 冉
z兲 ⫹ 1 ⫹
⌬V
2V 0
冊 E 共bz3 兲 共 x, z 兲,
terms of TE and TM modes.16 The ordinary refractive in-
dex (n o,co , n o,cl) is sensed by guided wave electric field
components perpendicular to the optical axis. The ex-
(40)
traordinary refractive index (n e,co , n e,cl), in contrast, af-
where fects electric field components parallel to the optical axis

冋
or, in our case, perpendicular to the plane of the layers.
V0 TE modes in slab waveguides have only one electric field
E 共bz3 兲 共 x, z 兲 ⫽ ⫺ B 0h component in the plane of the layers perpendicular to
h

册
their direction of propagation. TE modes are thus af-
⬁
fected by only the ordinary refractive index and have the
⫹ 兺
n⫽1
nkh ␦ 3 E n cosh共 n ␦ 3 kz 兲 cos共 nkx 兲 , same field expression as TE modes in isotropic slab
waveguides. TM modes have electric field components in
(41) the plane of the layers and perpendicular to them. These
fields are affected by the ordinary as well as the extraor-
⬁
V0 dinary refractive indices and are thus different from TM
E 共cz3 兲 共 x, z 兲 ⫽ ⫺
h n⫽0
兺
共 n ⫹ 1/2兲 ␦ 3 khD n sinh modes in isotropic slabs. We will continue our analysis
for TE modes, just to demonstrate the principle of the de-
⫻ 关共 n ⫹ 1/2兲 ␦ 3 kz 兴 cos关共 n ⫹ 1/2兲 kx 兴 . sign.
The TE and TM modes do not propagate with the same
(42)
propagation constant, which results in different transmis-
The proposed design of Fig. 1(a) provides two superim- sion characteristics. But in LPGs only the difference be-
posed distributions of the electric field. The first one has tween the propagation constants of a pair of modes of the
the same periodicity as the electrode l and is symmetric same family (TE1 and TE3, for example) is important.
over the waveguide cross section. Along with a variable Therefore even if the propagation constants are different,
component, it also contains a constant component but their differences are the same, the resonance wave-
⫺V 0 B 0 . The second distribution is repeated with a 2l length will be the same for the two types of mode. To re-
period and is antisymmetric with respect to z coordinates spect this condition, we have to design the transverse in-
with a zero constant component. dex profile of the waveguide, which is beyond the context
of this paper.
C. Electro-Optic Waveguide Modes and Induced We consider the interdigitated electrodes transparent
Refractive-Index Distribution with a negligible imaginary part of the dielectric constant
The geometry of the design is a three-layered structure in (a heavy doped semiconductor, such as indium tin oxide).
which each medium is uniaxial, with their optical axes di- Their contribution is taken into account through the ef-
rected normal to their surfaces (z axis). The lack of y de- fective refractive indices of the substrate and the super-
pendence and the diagonal nature of the dielectric tensor strate. We also neglect the corrugating ability of the dis-
allow the characterization of the modal properties in crete electrode structure, assuming that evanescent
Kulishov et al. Vol. 19, No. 8 / August 2002 / J. Opt. Soc. Am. A 1639

penetration of the guided waves into the substrate (i account the electrodes, n co and n cl are ordinary refractive
⫽ 1) and superstrate (i ⫽ 5) is very weak and that the indices of the core (i ⫽ 3) and cladding (i ⫽ 2 and 4) and
refractive indices of the substrate and superstrate are
close to the real part of the index of the electrodes. cos共 k 2 d 兲 ⫹ ␣ /k 2 sin共 k 2 d 兲
Ae ⫽ ,
The TE modes of the slab waveguide are well known.17 cos共 k 3 h 兲
Their H x and H z components are obtained from the Ey
component by differentiation: cos共 k 2 d 兲 ⫹ ␣ /k 2 sin共 k 2 d 兲
Ao ⫽ . (47)
sin共 k 3 h 兲
⫺j ⳵ E共yi 兲 ⫺␤ j ⳵ E共yi 兲
H 共zi 兲 ⫽ ⫽ E共yi 兲 , H 共xi 兲 ⫽ . The parameters in Eqs. (47) are not independent but
␻␮0 ⳵x ␻␮0 ␻␮0 ⳵z
(43) must obey the eigenvalue equation18

The time dependence of the fields is assumed to be given ␣ k2

by the function exp( j␻t), ␤ is the propagation constant, ⫺ tan共 k 2 d 兲
k3 k3
and ␮ 0 is the magnetic permeability of vacuum. We in- tan共 k 3 h 兲 ⫽ . (48)
tentionally used symbol E for the electric field of a guided ␣
light, not to be confused with the electric field strength E 1⫹ tan共 k 2 d 兲
k2
created by the periodical electrodes. The Ey component of
guided TE modes has the following form for even modes: for even modes and

冦
exp关 ⫺␣ 共 z ⫺ h ⫺ d 兲兴 , ⬁ ⬍ z ⭐ h ⫹ d,
␣
cos关 k 2 共 z ⫺ h ⫺ d 兲兴 ⫺ sin关 k 2 共 z ⫺ h ⫺ d 兲兴 , h ⫹ d ⬍ z ⬍ h,
k2
Ey ⫽ Z e A e cos共 k 3 z 兲 , 兩 z 兩 ⭐ h, (44)
␣
cos关 k 2 共 z ⫹ h ⫹ d 兲兴 ⫹ sin关 k 2 共 z ⫹ h ⫹ d 兲兴 , ⫺h ⬍ z ⭐ ⫺共 h ⫹ d 兲 ,
k2
exp关 ␣ 共 z ⫹ h ⫹ d 兲兴 , ⫺共 h ⫹ d 兲 ⬍ z ⭐ ⫺⬁,
and the following form for odd modes:

冦
exp关 ⫺␣ 共 z ⫺ h ⫺ d 兲兴 , ⬁ ⬍ z ⭐ h ⫹ d,
␣
cos关 k 2 共 z ⫺ h ⫺ d 兲兴 ⫺ sin关 k 2 共 z ⫺ h ⫺ d 兲兴 , h ⫹ d ⬍ z ⬍ h,
k2
Ey ⫽ Z o A o sin共 k 3 z 兲 , 兩 z 兩 ⭐ h, (45)
␣
⫺cos关 k 2 共 z ⫹ h ⫹ d 兲兴 ⫺ sin关 k 2 共 z ⫹ h ⫹ d 兲兴 , ⫺h ⬍ z ⭐ ⫺共 h ⫹ d 兲 ,
k2
⫺exp关 ␣ 共 z ⫹ h ⫹ d 兲兴 , ⫺共 h ⫹ d 兲 ⬍ z ⭐ ⫺⬁.

␣ k2
The factor exp关 j(␻t ⫺ ␤z)兴 is omitted from all the field ⫺ tan共 k 2 d 兲
k3 k3
equations for simplicity. The parameters appearing in cot共 k 3 h 兲 ⫽ ⫺ (49)
Eqs. (45) and (46) are defined as follows: ␣
1⫹ tan共 k 2 d 兲
k2
␣ ⫽ 共␤ ⫺ 2
n 2s k 02 兲 1/2, k2 ⫽ 2 2
共 n cl k0 ⫺␤ 兲 2 1/2
,
for odd modes. The mode fields are normalized with re-
k3 ⫽ 共 n 2cok 02 ⫺␤ 兲 2 1/2
, (46) spect to power per unit width P of 1 W/m,

where k 0 ⫽ 2 ␲ /␭ is the wave number in vacuum, ␭ is the

wavelength of the guided wave, n s is the effective refrac-
P⫽⫺
1
2
冕 ⫺⬁
⫹⬁
Ey H z* dz ⫽ 1W/m, (50)

tive indices of the substrate and superstrate taking into which brings us to the expression for the factor Z o,e :

冋 共 ␮ 0c 兲/ ␤

册
1/2
Z o,e ⫽ , (51)
2k 3 h ⫾ sin共 2k 3 h 兲 ␣ 共 k 22 ⫺ ␣ 2 兲 sin共 2k 2 d 兲 ⫺ 4 ␣ k 2 cos2 共 k 2 d 兲 ⫹ 2 共 k 22 ⫹ ␣ 2 兲 k 2 d 1
2
A o,e ⫹ ⫹ ⫹
4k 3 k 22 4k 23 2␣
where c is the speed of light in vacuum.
1640 J. Opt. Soc. Am. A / Vol. 19, No. 8 / August 2002 Kulishov et al.

When electric potentials are applied to the electrodes,

as is shown in Fig. 1(a), the electric field will induce
change to the intrinsic refractive index of the EO core
(and cladding), which can be expressed by the following
relationship for TE and TM modes:
3
n co,cl r 共13co, cl兲
TE兲
⌬n 共co,cl ⫽⫺ E x 共 x, z 兲 ,
2
3
n co,cl r 共33co,cl兲
TM兲
⌬n 共co,cl ⫽⫺ E z 共 x, z 兲 , (52)
2
where r 13 and r 33 are Pockels EO coefficients for core
(cladding). For electrostatic field E z , we have to use Eq.
(41) for the core and correspondent expression for the field
in the cladding in the case in which it is EO active. For Fig. 5. First core mode TE1 (solid curve); first cladding mode
the intrinsic refractive index of the core (cladding), we TE2 (dashed curve); and second cladding mode TE3 (dotted curve)
distributions across the waveguide cross section.
should use its ordinary value for TE modes and its ex-
traordinary value for TM modes. As we can see from the
equations, the EO-induced refractive index is propor-
tional to the E z component of the electric field. In Fig. 4
two-dimensional distributions of the component in the
core are presented for three different ⌬V values. Two ex-
treme values ⌬V ⫽ 0 and ⌬V ⫽ ⫺2V 0 correspond to two
distributions of Figs. 1(b) and 1(c) with opposite types of
cross-section symmetry, whereas the distribution for ⌬V
⫽ ⫺1.6V 0 represents two superimposed gratings.
We chose parameters for the waveguide structure in
such a way that it supports only one core guided mode ␤ 1
and a number of cladding guided modes. The beating
length between the fundamental mode and the second
mode is close to the double beating length between the
fundamental and the third mode: 2( ␤ 1 ⫺ ␤ 2 ) ⬇ ␤ 1
⫺ ␤ 3 . This condition leads us to the situation in which
coupling between the core mode and the first cladding
mode occurs with the resonance wavelength ␭ b that is
close to the resonance wavelength ␭ c of the coupling be-
tween the core mode and the second cladding mode.
When 2( ␤ 1 ⫺ ␤ 2 ) exactly equals ␤ 1 ⫺ ␤ 3 , the resonance
wavelengths coincide. The larger the difference between
2( ␤ 1 ⫺ ␤ 2 ) and ␤ 1 ⫺ ␤ 3 , the farther resonances are
separated. Just to demonstrate this concept, we choose
the waveguide parameters of n s ⫽ 1.7, n cl ⫽ 2.00, n co
⫽ 2.03, 2h ⫽ 2.907 ␮ m, and d ⫽ 2 ␮ m to provide the
following values for the propagation constants: ␤ 1 /k 0
⫽ 2.0229, ␤ 2 /k 0 ⫽ 2.00244, and ␤ 3 /k 0 ⫽ 1.98258. The
mode distributions are presented in Fig. 5. For the
core-guided mode, 91.09% of the wave power is con-
centrated in the core, 8.91% is in the cladding and
only 3.68 ⫻ 10⫺3 % is in the substrate and superstrate.
The beating length between fundamental and second
modes for ␭ 0 ⫽ 1.55 ␮ m equals 2 ␲ /( ␤ 1 ⫺ ␤ 3 )
⫽ 38.4425 ␮ m. Therefore we choose interdigitated elec-
trode structure with the finger-to-finger pitch l
⫽ 38 ␮ m.

Fig. 4. Three-dimensional distribution of the normal component 3. MODE COUPLING AND TRANSMISSION
of the normalized electrode electric field E z /(V 0 /h) in the wave-
guide core for different values of ⌬V voltages. (a) ⌬V ⫽ 0, (b)
SPECTRA
⌬V ⫽ ⫺1.6V 0 , and (c) ⌬V ⫽ ⫺2V 0 for the following parameter A. Coupled-Wave Equations
set: a/l ⫽ 0.25, 2h ⫽ 0.0745l, d ⫽ 1.376h, ⑀ (s) ⫽ 3.5, ⑀ (cl) xx The parameters of the waveguide are set in such a way
⫽ ⑀ zz
(cl)
⫽ 4, ⑀ (co)
xx ⫽ 8.4, ⑀ zz ⫽ 5.8, and N ⫽ 40.
(co)
that the core-guided mode with propagation constant ␤ 1
Kulishov et al. Vol. 19, No. 8 / August 2002 / J. Opt. Soc. Am. A 1641

will interact with the gratings, coupling out into copropa- The phase-matching conditions are ⌬ ␤ 12 ⫽ 0 and ⌬ ␤ 13
gating cladding modes with propagation constants ␤ 2 or ⫽ 0 for the gratings with 2l and l periods, respectively.
␤ 3 ( ␤ 1 ⬎ ␤ 2 ⬎ ␤ 3 ) or both. The interaction process is Certainly, these matching conditions cannot be satisfied
described by the following set of coupled-wave equations: for the same wavelength. The set of first-order differen-
tial equations [Eqs. (55)] are solved analytically for the
da 1 boundary conditions R(0) ⫽ 1 and S(0) ⫽ P(0) ⫽ 0.
⫽ j ␴ 11a 1 ⫹ j ␬ 12a 2 exp共 ⫺j⌬ ␣ 12x 兲
dx
B. Coupling Coefficients of the Gratings
⫹ j ␬ 13a 3 exp共 ⫺j⌬ ␣ 13x 兲 , The tangential coupling coefficient can be easily calcu-
lated with the following relationship20:
da 2

冕
⫽ j ␴ 22a 2 ⫹ j ␬ 12a 1 exp共 ⫺j⌬ ␣ 12x 兲 ␻ ⬁
dx
␬ ij ⫽ ⌬ ⑀ 共 z 兲 Eyi 共 z 兲 Eyj
* 共 z 兲 dz, (56)
4 ⫺⬁
⫹ j ␬ 23a 3 exp共 ⫺j⌬ ␣ 23x 兲 ,
where ⌬ ⑀ ⫽ ⫺n 4 r 13E z is the induced dielectric perturba-
da 3 tion. The contribution from longitudinal coupling coeffi-
⫽ j ␴ 33a 3 ⫹ j ␬ 13a 1 exp共 ⫺j⌬ ␣ 13x 兲
dx cients is neglected because it is nearly 2 orders of magni-
tude smaller than the contribution from the tangential
⫹ j ␬ 23a 2 exp共 ⫺j⌬ ␣ 23x 兲 , (53) coupling coefficient. In this paper we analyze the case of
where a 1 , a 2 , and a 3 are the complex modal amplitudes, an EO active core and a non-EO active cladding. There-
⌬ ␣ 12 ⫽ ␤ 1 ⫺ ␤ 2 ⫺ ␲ /l, ⌬ ␣ 13 ⫽ ␤ 1 ⫺ ␤ 3 ⫺ 2 ␲ /l, ⌬ ␣ 23 fore we have to integrate over the core cross section. It
⫽ ␤ 2 ⫺ ␤ 3 ⫺ ␲ /l, ␴ ii are dc coupling coefficients, and k ij gives us the following expressions for the dc and ac cou-
are the ac cross-coupling coefficients i, j ⫽ (1, 3). 19 For pling coefficients:
the problem under consideration, ⌬ ␣ 23 ⫽ ⌬ ␣ 13 ⫺ ⌬ ␣ 12
and, including the dc coupling coefficients into the mo-
mentum mismatch terms, system (53) can be reduced to
␴ 11 ⫽
␤1
2 ␮ 0k 0c
V 0B 0 1 ⫹冉 ⌬V
2V 0
冊 r 31n 4coZ 1 A 1

the following form:

db 1
⫻ 冕 ⫹h

⫺h
cos2 共 k 31z 兲 dz,
⫽ j ␬ 12b 2 exp共 ⫺j⌬ ␤ 12x 兲 ⫹ j ␬ 13b 3 exp共 ⫺j⌬ ␤ 13x 兲 ,
dx

db 2
␴ 22 ⫽
␤2
2 ␮ 0k 0c
V 0B 0 1 ⫹冉 ⌬V
2V 0
冊 r 31n 4coZ 2 A 2
⫽ j ␬ 12b 1 exp共 ⫺j⌬ ␤ 12x 兲 ⫹ j ␬ 23b 3 exp共 ⫺j⌬ ␤ 23x 兲 ,
dx

db 3
⫻ 冕 ⫹h

⫺h
sin2 共 k 32z 兲 dz,

冉 冊
⫽ j ␬ 13b 1 exp共 ⫺j⌬ ␤ 13x 兲 ⫹ j ␬ 23b 2 exp共 ⫺j⌬ ␤ 23x 兲 ,
dx ␤3 ⌬V
(54) ␴ 33 ⫽ V 0B 0 1 ⫹ r 31n 4coZ 3 A 3
2 ␮ 0k 0c 2V 0

冕
where ⫹h
⫻ cos2 共 k 33z 兲 dz,
b 1 ⫽ a 1 exp共 ⫺j ␴ 11x 兲 , b 2 ⫽ a 2 exp共 ⫺j ␴ 22x 兲 , ⫺h

b 3 ⫽ a 3 exp共 ⫺j ␴ 33x 兲 , ⌬ ␤ 12 ⫽ ⌬ ␣ 12 ⫺ ␴ 11 ⫹ ␴ 22 , 冑␤ 1 ␤ 2 ⌬V
␬ 12 ⫽ ␦ 3 kD 1 r 31n 4coZ 1 Z 2 A 1 A 2
⌬ ␤ 13 ⫽ ⌬ ␣ 13 ⫺ ␴ 11 ⫹ ␴ 33 , ⌬ ␤ 23 ⫽ ⌬ ␣ 23 ⫺ ␴ 22 ⫹ ␴ 33 . 2 ␮ 0k 0c 4
Finally, it is simplified to

dR ⌬ ␤ 12 ⫹ ⌬ ␤ 13
⫻ 冕 ⫹h

⫺h
cos共 k 31z 兲 sin共 k 32z 兲 sinh共 ␦ 3 kz/2兲 dz,

⫽ ⫺j R ⫺ j ␬ 12S ⫺ j ␬ 13P,
dx

dS
2

⌬ ␤ 12 ⫺ ⌬ ␤ 13
␬ 13 ⫽
冑␤ 1 ␤ 3
2 ␮ 0k 0c
冉
V0 1 ⫹
⌬V
2V 0
冊 ␦ 3 kE 1 r 31n 4coZ 1 Z 3 A 1 A 3
⫽ ⫺j S ⫺ j ␬ 12R ⫺ j ␬ 23P,
dx

dP
2

⌬ ␤ 12 ⫺ ⌬ ␤ 13
⫻ 冕 ⫹h

⫺h
cos共 k 31z 兲 cos共 k 33z 兲 cosh共 ␦ 3 kz 兲 dz,

⫽ ⫹j P ⫺ j ␬ 13S ⫺ j ␬ 23S,
dx 2 冑␤ 2 ␤ 3 ⌬V
(55) ␬ 23 ⫽ ␦ 3 kD 1 r 31n 4coZ 2 Z 3 A 2 A 3
2 ␮ 0k 0c 4

冕
where the new amplitudes R, S, and P are
⫹h

R 共 z 兲 ⫽ b 1 exp关 ⫺j 共 ⌬ ␤ 12 ⫹ ⌬ ␤ 13兲 x/2兴 , ⫻ cos共 k 33z 兲 sin共 k 32z 兲 sinh共 ␦ 3 kz/2兲 dz.
⫺h
(57)
S 共 z 兲 ⫽ b 2 exp关 ⫺j 共 ⌬ ␤ 13 ⫺ ⌬ ␤ 12兲 x/2兴 ,
The coefficient ␬ 13 is responsible for the coupling between
P 共 z 兲 ⫽ b 3 exp关 ⫺j 共 ⌬ ␤ 12 ⫺ ⌬ ␤ 13兲 x/2兴 . the symmetrical fundamental core mode TE1 and the sec-
1642 J. Opt. Soc. Am. A / Vol. 19, No. 8 / August 2002 Kulishov et al.

ond symmetric cladding mode TE3 through the symmetri- tors. Different slopes of ␬ 12 and ␬ 13 versus ⌬V/V 0 re-
cal EO-induced index perturbation with the period l [Fig. sults in an asymmetrical switching characteristic where
1(b)]. The coefficient ␬ 12 is a proportionality factor for ␬ 12 and ␬ 13 become equal at ⌬V/V 0 ⫽ ⫺1.75 instead of
coupling between the symmetrical fundamental core at ⌬V/V 0 ⬇ ⫺1 . Besides depending on V 0 and ⌬V volt-
mode TE1 and the first asymmetric cladding mode TE2 ages and type of EO material (r 13), the coupling coeffi-
through the asymmetrical EO-induced index perturbation cients also depend on electrode parameters, such as
with the period 2l [Fig. 1(c)]. Owing to the 2:1 ratio be- electrode-finger duty ratio a/l, and dielectric permittivi-
tween two superimposed grating periods, the matching ties of the core, cladding, substrate, and superstrate. Al-
condition will be automatically satisfied for the coupling though the material issue is beyond the scope of this pa-
between the first and the second cladding modes (TE2 and per, the EO response can be improved when the electric
TE3 ) through the grating with period 2l, and the coupling field inside the core is increased, given that the applied
is proportional to the coefficient ␬ 23 . Note that the cross- potential on the electrode is constant. In Fig. 7 the cou-
coupling coefficients can be individually controlled pling coefficients as a function of the electrode-finger duty
through ⌬V voltage because ␬ 12 and ␬ 23 are proportional ratio are presented. It shows that wider electrodes in-
to ⌬V/V 0 and ␬ 13 is proportional to (1 ⫹ ⌬V/2V 0 ). Be- crease the constant component of the electric field with
cause the constant component of the electric field is non- respect to the variable one and reduce the variable com-
zero only for the c configuration, the dc coupling coeffi- ponent for the b grating. Therefore wide electrodes shift
cients, ␴ 11 , ␴ 22 , and ␴ 33 are functions of V 0 and ⌬V. the intersection point for the ␬ 12 and ␬ 13 lines in Fig. 6
Figure 6 shows coupling coefficients ␬ ij h/(V 0 r 13) [Fig. toward lower 兩 ⌬V/V 0 兩 values.
6(a)] and ␴ ii h/(V 0 r 13) [Fig. 6(b)] as a function of dimen- Another parameter that can change the ␬ 12 to ␬ 13 ratio
sionless voltage ⌬V/V 0 . As we can see, the design allows is dielectric permittivities of the core and cladding. We
us to activate gratings b and c independently or switch recognize that in reality it is highly improbable to change
them ON simultaneously with arbitrary weighting fac- static dielectric permittivity without affecting high-

Fig. 6. Coupling coefficients (a) ␬ ij h/(V 0 r 13) and (b) ␴ ii h/(V 0 r 13) in 1/m units as a function of ⌬V/V 0 [(a) ␬ 12 , solid curve; ␬ 13 , dotted
curve; and ␬ 23 , dashed curve; (b) ␴ 11 , solid curve; ␴ 22 , dashed curve; ␴ 33 , dotted curve] for the following waveguide and electrode pa-
rameters: 2h/l ⫽ 0.0745, d/h ⫽ 1.376, ⑀ (s) ⫽ 3.5, ⑀ (cl) xx ⫽ ⑀ zz ⫽ 5,
(cl)
⑀ (co)
xx ⫽ ⑀ zz ⫽ 5.8,
(co)
a/l ⫽ 0.5, V 0 ⫽ 50 V, and r 13
⫽ 30 pm/V.

Fig. 7. Coupling coefficients (a) ␬ ij h/(V 0 r 13) and (b) ␴ ii h/(V 0 r 13) in 1/m units as a function of the electrode duty ratio for ⌬V/V 0
⫽ ⫺1.75 [(a) ␬ 12 , solid curve; ␬ 13 , dotted curve; ␬ 23 , dashed curve; (b) ␴ 11 , solid curve; ␴ 22 , dashed curve; ␴ 33 , dotted curve] for the
same waveguide and electrode parameters as in Fig. 6.
Kulishov et al. Vol. 19, No. 8 / August 2002 / J. Opt. Soc. Am. A 1643

Fig. 8. Coupling coefficients (a) and (c) ␬ ij h/(V 0 r 13) and (b) and (d) ␴ ii h/(V 0 r 13) in 1/m units as a function of the dielectric permittivity
of the (a) and (b) cladding ⑀ (cl)
xx ⫽ ⑀ zz and the dielectric permittivity of the (c) and (d) core ⑀ zz for the fixed ⑀ xx for a/l ⫽ 0.5; ⌬V/V 0
(cl) (co) (co)

⫽ ⫺1.75. The rest of the parameters are the same as in Fig. 6 [(a) ␬ 12 , solid curve; ␬ 13 , dotted curve; ␬ 23 , dashed curve; (b) ␴ 11 , solid
curve, ␴ 22 ; dashed curve, and ␴ 33 ; dotted curve].

Fig. 9. Coupling coefficients (a) ␬ ij h/(V 0 r 13) and (b) ␴ ii h/(V 0 r 13) in 1/m units as a function of ⌬V/V 0 [(a) ␬ 12 , solid curve; ␬ 13 , dotted
curve; ␬ 23 , dashed curve; (b) ␴ 11 , solid curve; ␴ 22 , dashed curve; ␴ 33 , dotted curve] for the following waveguide and electrode param-
eters: 2h/l ⫽ 0.0745, d/h ⫽ 1.376, ⑀ (s) ⫽ 3.5, ⑀ (cl) xx ⫽ ⑀ zz ⫽ 4,
(cl)
⑀ (co)
xx ⫽ 5.8, ⑀ zz
(co)
⫽ 4, a/l ⫽ 0.8, V 0 ⫽ 50 V, and r 13
⫽ 30 pm/V.

frequency dielectric permittivity, which is proportional to mately twice as the permittivity of the cladding grows
the refractive index squared. However, our analysis from 3 to 8. However, ␬ 12 drops faster than ␬ 13 with an
bears a mainly speculative character; therefore we change increase in ⑀ zz
(co)
, keeping ⑀ (co)
xx , ⑀ xx , and ⑀ zz constant.
(cl) (cl)

static values of the dielectric permittivity, keeping the Therefore, with proper adjustment of the electrode width
same values for the refractive indices. Figure 8(a) dem- and the dielectric permittivities, the coupling coefficient’s
onstrates how the coupling coefficients depend on dielec- dependence on ⌬V can be made more symmetric, as is
tric permittivity of the cladding ⑀ (cl) xx ⫽ ⑀ zz for a fixed
(cl)
shown in Fig. 9, in which the plots of Fig. 6 are repro-
value of ⑀ xx and ⑀ zz , and Fig. 8(b) reveals how they are
(co) (co)
duced for the new parameter set: a/l ⫽ 0.8 ⑀ (cl) xx ⫽ ⑀ zz
(cl)

affected by dielectric anisotropy of the core when the clad- ⫽ 4, ⑀ xx ⫽ 5.8, and ⑀ zz ⫽ 4. The intersection point
(co) (co)

ding dielectric permittivity is kept constant. If coupling here is shifted toward ⌬V/V 0 ⫽ ⫺1.22. This equaliza-
coefficients ␬ 12 and ␬ 23 practically stay constant with the tion takes place at the expense of the dynamic range re-
cladding dielectric permittivity, ␬ 13 increases approxi- duction for ␬ 13 ; however, the dynamic range for all dc cou-
1644 J. Opt. Soc. Am. A / Vol. 19, No. 8 / August 2002 Kulishov et al.

dipped spectrum. Whereas Figs. 11(b) and 11(c) describe

the single-dip spectra when gratings b and c are acti-
vated individually for ⌬V ⫽ 0 and V 0 ⫽ 6.5 V and
⌬V/V 0 ⫽ ⫺2 and V 0 ⫽ 50 V, respectively. Keeping the
same number of activated electrodes M (or grating length
L ⫽ lM), we have to control both voltages ⌬V and V 0 to
maintain the proper values for the coupling coefficients
␬ 12 and ␬ 13 . In this case the position of the high-
wavelength dip is changed (from 1550 to ⬇1547 nm),
whereas the resonance wavelength for the low-
wavelength dip practically is not changed. The distance
between the two resonance wavelengths also can be ad-
justed through waveguide design: Increasing the
beating-length mismatch ⌬ ⫽ 2( ␤ 1 ⫺ ␤ 2 ) ⫺ ( ␤ 1 ⫺ ␤ 3 )
increases the distance ⌬␭ ⫽ l⌬k 0 .
In our EO reconfigurable grating design, the length of
the gratings can be easily controlled because coupling
does not occur without the presence of voltage at the elec-
trodes. Therefore there are three main parameters for
spectrum control: two voltages ⌬V and V 0 and the grat-
ing length L ⫽ Ml, where M is the number of activated
electrode fingers. These three parameters give us flex-
ibility in the spectrum control. For example, setting ⌬V
voltage to keep ␬ 12 and ␬ 13 approximately equal, V 0 and L
can be used to shift a double-dip spectrum, as is shown in
Fig. 10. (a) Electric potential application scheme for the EO-
induced waveguide gratings that provides coupling between the
core mode and the cladding modes with the same symmetry as
the core mode. The electrostatic problem solution for this geom-
etry can be presented as a linear combination of the Laplace’s
equation solutions for the structures (b) and (c).

pling coefficients is getting approximately 1.5 times

higher than in Fig. 6. For this parameter set (a/l
⫽ 0.8) the electrodes will have gaps between the fingers
that equal l ⫺ a ⬇ 7.8 ␮ m, which is comparable with the
distance between the top and the bottom electrodes [or
waveguide width 2(d ⫹ h) ⬇ 6.907 ␮ m].
It is important to realize that the potential application
scheme in Fig. 1 is suggested from a symmetry point of
view to provide coupling between the fundamental core
mode, which is always a symmetric one, and the first
(asymmetric) and second (symmetric) cladding modes.
However, the same concept can be realized for coupling
between the fundamental mode and any two symmetrical
cladding modes, for example, with the second TE3 and
fourth TE5 cladding modes. In this case the potential ap-
plication scheme should look like the one in Fig. 10. As
we can see, grating b preserves the same potential distri-
bution; however, the electric field for grating c will result
in deeper penetration through the waveguide cross sec-
tion, providing higher values for ␬ 12 to ␬ 23 with respect to
␬ 13 .

C. Reconfigurable Transmission Spectra

The coupling coefficient behavior gives us a broad oppor-
tunity to reconfigure the device’s transmission spectrum.
An example of the transmission spectra is shown in Fig.
Fig. 11. Transmission spectra of waveguide grating (a) for both
11, in which Fig. 11(a) describes a double-dip spectrum, b and c activated gratings (⌬V/V 0 ⫽ ⫺1.77, V 0 ⫽ 57 V) (b) indi-
when both gratings b and c are activated at ⌬V/V 0 vidually activated grating b (⌬V ⫽ 0, V 0 ⫽ 6.5 V) and (c) indi-
⫽ ⫺1.77 and V 0 ⫽ 57 V; therefore they both contribute vidually activated grating c (⌬V/V 0 ⫽ ⫺2; V 0 ⫽ 50 V) and the
to the transmission losses that resulted in a double- same waveguide parameter set as for Fig. 6.
Kulishov et al. Vol. 19, No. 8 / August 2002 / J. Opt. Soc. Am. A 1645

use of the potential application scheme presented in Fig.

14(a) with interaction between the fundamental core TE1
mode and cladding modes TE3 and TE4 . Unlike the
scheme in Fig. 1, the symmetry of TE3 and TE4 modes
[Fig. 14(d)] and the index distribution of grating c [Fig.
14(c)] results in the overlap integral [Eq. (57)] for ␬ 23 van-
ishing, so coupling between the cladding modes becomes
impossible.
Given that a proper electronic interface is provided,
which allows us to control the potential practically on ev-
ery electrode finger, the beating length of both structures
b and c might be adjusted independently through proper
electric potential distribution; that would give us an ad-
ditional means of controlling the device. For example,
because 2( ␤ 1 ⫺ ␤ 2 ) ⬇ ␤ 1 ⫺ ␤ 3 in our design, with equal
grating lengths (L b ⫽ L c ), the short-wavelength dip (left
one) is always approximately twice narrower than the
long-wavelength (right) one, as can be seen in Figs. 11
and 12. In Fig. 15 the dashed curve shows the transmis-
sion spectrum for V 0 ⫽ 28 V, ⌬V/V 0 ⫽ ⫺1.77, and L
⫽ 35 mm. The solid curve represents the spectrum in
which the first half of the length is under potential V 0
⫽ 34 V, ⌬V/V 0 ⫽ ⫺1.61 while the second half is under
V 0 ⫽ 27 V, ⌬V/V 0 ⫽ ⫺2. As we can see, the FWHM
bandwidth of both dips is equal to the solid curve; how-
ever, the right-side dip is distorted by high sidelobes.
This is the result of constant electric field presence for the
first part of the structure with nonzero dc coupling coeffi-
cients ( ␴ 11 ⫽ 0, ␴ 22 ⫽ 0, ␴ 33 ⫽ 0) and its absence in the
second part ( ␴ 11 ⫽ ␴ 22 ⫽ ␴ 33 ⫽ 0). This spectrum was
calculated by use of a piecewise-uniform approach20 for
Fig. 12. Double-dip transmission spectra when both b and c ac-
tivated gratings with ⌬V/V 0 ⫽ ⫺1.22; (a) V 0 ⫽ 40 V, (b) V 0 equation system (55), which is based on identifying
⫽ 30 V, and (c) V 0 ⫽ 20 V and the same waveguide parameter
set as for Fig. 9.

Fig. 12. This shift is provided when we use dc coupling

coefficients’ dependence from V 0 and maintain ␬ 12L and
␬ 13L equal to ␲/2.
As we already mentioned, the resonance wavelength
for grating b can be shifted by use of the constant compo-
nent of the electric field for this shift, whereas the reso-
nance wavelength for grating c is always fixed (under the
condition that grating b is switched OFF), as can be seen
in Fig. 13. When both gratings are activated, their reso-
nance wavelengths are affected by the constant compo-
nent of the electric field, and both can be shifted as can be
seen in Fig. 12. Decreasing V 0 voltage, we reduce the
coupling coefficients. However, keeping the ␬ L product
constant through increasing the grating length results in
narrowing the spectra.
As we already mentioned, the momentum-matching
conditions favor outcoupling the core-guided mode into
the two cladding guided modes. However, the matching
conditions are also satisfied for interaction of these two
modes between each another, and, as a result, a portion of
light energy transfers from one cladding mode to another
in proportion to the coupling coefficient ␬ 23 , as is depicted
Fig. 13. Transmission spectra of (a) individually activated b
by the momentum diagram in Fig. 14(e) ( ␤ 2 ⫺ ␤ 3
grating (⌬V ⫽ 0; solid curve, V 0 ⫽ 20 V; dotted curve, V 0
⫽ ␲/l). Normally, this coefficient is much smaller than ⫽ 10 V; dashed curve, V 0 ⫽ 5 V) and (b) individually activated c
␬ 12 and ␬ 13 ; however, it still prevents us from having grating (⌬V/V 0 ⫽ ⫺2; solid curve, V 0 ⫽ 50 V; dotted curve, V 0
precise control over the energy ratio outcoupled into the ⫽ 25 V; dashed curve, V 0 ⫽ 12.5 V) for the same waveguide pa-
two cladding modes. This problem can be overcome by rameter set as for Fig. 6.
1646 J. Opt. Soc. Am. A / Vol. 19, No. 8 / August 2002 Kulishov et al.

Fig. 14. (a) Electric potential application scheme for the EO-induced waveguide gratings that provides zero coupling between cladding
modes with the corresponding potential distributions for each partial grating at (b) ⌬V ⫽ 0, and (c) ⌬V/V 0 ⫽ ⫺2. (d) The mode distri-
butions involved in the interaction and (e) the momentum diagram in which ␤ 2 ⫺ ␤ 3 coupling is eliminated due to zero overlapping
integral for these mode interactions.

scheme, shown in Fig. 16. As can be seen from Figs.

16(b) and 16(c), both contributing distributions (for ⌬V
⫽ 0 and ⌬V/V 0 ⫽ ⫺2) are free from the constant compo-
nent of the electric field. Unfortunately, we cannot apply
our calculation of the structure in Fig. 1 to the scheme of
Fig. 16; however, use of the ac coupling coefficient values
of Fig. 15 with ␴ 11 ⫽ ␴ 22 ⫽ ␴ 33 ⫽ 0 for both parts of the
structure gives us a completely symmetric transmission
spectrum, presented in Fig. 17. Sidelobes in the spec-
trum can also be substantially suppressed through spa-
tial modulation of the bias voltage V 0 ⫽ V 0 (z).

Fig. 15. Transmission spectrum for the structure with uniform

potential distribution on the electrodes V 0 ⫽ 28 V, ⌬V/V 0
⫽ ⫺1.77, and L ⫽ 35 mm (dashed curve) and the structure with D. Polarization Mode Coupling
the same length, in which one half of the length is under V 0 As we already mentioned, this paper is devoted to TE-
⫽ 34 V, ⌬V/V 0 ⫽ ⫺1.61 and the second half is under V 0 mode analysis in the structure with an EO active core and
⫽ 27 V; ⌬V/V 0 ⫽ ⫺2. The waveguide parameters are set as an EO unactive cladding. In other words, this paper is
for Fig. 6. restricted to the analysis of TE-mode interaction through
EO-induced refractive-index periodicity in the core.
However, the EO-induced perturbation could result in EO
2 ⫻ 2 matrices for each uniform section of the grating- coupling between the TE and the TM modes.21 An exter-
(s)and then multiplying all of them together to obtain a nal electric field produces a nonvanishing off-diagonal
single 2 ⫻ 2 matrix for the whole structure. component of the dielectric tensor, which is responsible
Despite the use of a constant component of the electric for this TE–TM coupling that is normally not possible be-
field for tuning the resonance wavelength of the spectra, cause of the mode orthogonality. For the geometry of our
there are cases in which (as the one mentioned above) the design with E x and E z nonzero components of the elec-
component might be superfluous. Within the general trodes’ electric field and for the EO tensor of EO crystals
concept of the device, the constant component of the field with 4-, 6-, and ⬁-mm crystallographic groups including
can be suppressed if we use the potential application EO polymers
Kulishov et al. Vol. 19, No. 8 / August 2002 / J. Opt. Soc. Am. A 1647

冤 冥
0 0 r 31 vector. However, for the chosen geometry our TE modes
are represented by Ey components, whereas TM modes are
0 0 r 31
represented by Ex components. Therefore there is no in-
0 0 r 33 teraction between TE and TM modes in the proposed de-
r ij ⫽ , (58) sign.
0 r 15 0
r 15 0 0
0 0 0 4. CONCLUSION
the perturbation of the dielectric tensor will have the fol- We have presented a full analysis of a waveguide with
lowing form: EO-induced dual refractive-index gratings, beginning

冋 册
with a rigorous electric field calculation for an EO-
r 31E z 0 r 15E x induced coupling coefficient calculation and ending with a
reconfigurable transmission spectra study that was calcu-
⌬⑀ ⫽ 0 r 31E z 0 . (59) lated on the basis of the coupled-wave-equation solution.
r 15E x 0 r 33E z The principle of operation is based on EO induction of two
types of periodic refractive-index distributions with the
The off-diagonal element r 15E x is responsible for coupling
opportunity to independently switch them ON and OFF
the Ex and Ez components of the guided-wave electric field
or to activate both of them simultaneously with control-
lable weighting factors. In this case, its transmission
characteristics represent two peak rejection band spectra
with independent control of the peak positions and band-
width. Within the concept, two new external potential
application schemes are described. For the first one, the
symmetry of the cross-sectional distribution of the refrac-
tive index provides coupling only between the core mode
and the cladding modes, preventing interaction of the
cladding modes with each other. The second scheme re-
alizes a zero constant component of the electric field,
which keeps constant the resonance wavelengths for both
dips. The devices with this type of spectral characteristic
will find potential applications in optical communications,
such as reconfigurable dual-wavelength filters, dual-
wavelength distributed-feedback lasers, and optical wave-
length division multiplexers. This simple concept also
opens opportunities for developing a number of tunable
devices for integrated optics by use of the proposed design
as a building block.

Address correspondence to Mykola Kulishov at the

address on the title page or by phone, 514-737-7030;
fax, 514-737-9893; or e-mail, mkulishov
@adtekphotomask.com.
Fig. 16. (a) Electric potential application scheme for the EO-
induced waveguide gratings that provides a zero constant com-
ponent of the electric field, and as a result dc coupling coefficients REFERENCES
are always zero with the corresponding potential distributions
for each partial grating at (b) ⌬V ⫽ 0 and (c) ⌬V/V 0 ⫽ ⫺2. 1. T. Erdogan, ‘‘Fiber gratings provide keys to future optical
networks,’’ Photonics Spectra 32, 96–97 (1998).
2. D. M. Costantini, C. A. P. Muller, S. A. Vasiliev, H. G. Lim-
berger, and R. P. Salathe, ‘‘Tunable loss filter based on
metal-coated long-period filter grating,’’ IEEE Photon. Tech-
nol. Lett. 11, 1458–1460 (1999).
3. A. A. Abramov, A. Hale, R. S. Windler, and T. A. Strasser,
‘‘Widely tunable long-period fiber gratings,’’ Electron. Lett.
35, 81–82 (1999).
4. Y. Joeng, B. Yang, B. Lee, H. S. Seo, S. Choi, and K. Oh,
‘‘Electrically controllable long-period liquid crystal fiber
gratings,’’ IEEE Photon. Technol. Lett. 12, 519–521 (2000).
5. D. B. Stegall and T. Erdogan, ‘‘Dispersion control with use
of long-period fiber gratings,’’ J. Opt. Soc. Am. A 17, 304–
312 (2000).
6. M. Das and K. Thyagarajan, ‘‘Dispersion compensation in
transmission using uniform long period fiber gratings,’’ Opt.
Commun. 190, 159–163 (2001).
Fig. 17. Transmission spectra for the structure with parameters 7. G. Levy-Yurista and A. A. Friesem, ‘‘Dual spectral filters
the same as in Fig. 15 except that ␴ 11 ⫽ ␴ 22 ⫽ ␴ 33 ⫽ 0 for both with multi-layered grating waveguide structures,’’ Appl.
curves. Phys. B 72, 921–925 (2001).
1648 J. Opt. Soc. Am. A / Vol. 19, No. 8 / August 2002 Kulishov et al.

8. J. Bao, X. Zhang, K. Chen, and W. Zhou, ‘‘Spectra of dual ries, and Products (Academic, New York, 1980), Sec.
overwritten fiber Bragg grating,’’ Opt. Commun. 188, 31–39 1.441.2.
(2001). 15. M. A. Hussain and S. L. Pu, ‘‘Dynamic stress intensity fac-
9. J. Zhao, X. Shen, and Y. Xia, ‘‘Beam splitting, combining tor for an unbounded plate having collinear cracks,’’ Eng.
and cross coupling through multiple superimposed volume- Fract. Mech. 4, 865–876 (1972).
index gratings,’’ Opt. Laser Technol. 33, 23–28 (2001). 16. V. Ramaswamy, ‘‘Ray model of energy and power flow in an-
10. A. Yariv, ‘‘Frustration of Bragg reflection by cooperative isotropic film waveguides,’’ J. Opt. Soc. Am. 64, 1313–1320
dual-mode interference: a new mode of optical propaga- (1974).
tion,’’ Opt. Lett. 23, 1835–1836 (1998). 17. D. Marcuse, Light Transmission Optics (Van Nostrand
11. M. Kulishov, ‘‘Interdigitated electrode-induced phase grat- Reinhold, New York, 1972).
ing with an electrically switchable and tunable period,’’ 18. H. Kogelnik, ‘‘Theory of optical waveguides,’’ in Guided-
Appl. Opt. 38, 7356–7363 (1999). Wave Optoelectronics, 2nd ed., T. Tamir, ed. (Springer-
12. M. Kulishov, P. Cheben, X. Daxhelet, and S. Delprat, Verlag, New York, 1990), pp. 48–50.
‘‘Electro-optically induced tilted phase gratings in 19. T. Erdogan, ‘‘Fiber grating spectra,’’ J. Lightwave Technol.
waveguides,’’ J. Opt. Soc. Am. B 18, 457–464 (2001). 15, 1277–1294 (1997).
13. E. N. Glytsis, T. K. Gaylord, and M. G. Moharam, ‘‘Electric 20. K. S. Lee, ‘‘Mode coupling in tilted planar waveguide grat-
field, permittivity and strain distributions induced by inter- ings,’’ Appl. Opt. 39, 6144–6149 (2000).
digitated electrodes on electro-optic waveguides,’’ J. Light- 21. D. Marcuse, ‘‘Electrooptic coupling between TE and TM
wave Technol. LT-5, 668–683 (1987). modes in anisotropic slabs,’’ IEEE J. Quantum Electron.
14. I. S. Gradshteyn and I. M. Ryzhik, Table of Integrals, Se- QE-11, 759–767 (1975).