Z.Al-kinani et al.

, International Journal of Microwaves Applications, 2(2), March April 2013, 51- 57

51
@ 2012, IJMA All Rights Reserved


Modeling of Nonlinear Effects in Waveguide Silicon on Simox in Optimum conditions
according to the Experiment

Z.Al-kinani
(1)
, T.Boumaza
(1)
, M.Bouchemat
(1)
and A.koster
(2)

(1)
Department of Electronic, University of Mentouri Constantine1, Algeria.
(2)
Department of fondamental Electronic, CNRS URA 22, Buil. 220 University of South Paris, 91405 ORSAY cedex, France.
E_mail : [email protected]


ABSTRACT
This study concerns the modeling of non-linear operation of
a silicon waveguide on an insulator of simox type with
grating coupler in a dynamic regime, the simulation
parameters are determined in accordance with the
experiment, with the aim of achieving fast threshold
optoelectronic switches. The optical non-linearities in the
silicon are related to the change of the refractive index, and
are used to obtain fast switching on the transmitted beam due
to the propagation of the fundamental mode in the guide. The
performances of these components are highly dependent of
its rising time and its sensitivity. We will maximize the latter
two. The model of non-linear operation of fast threshold
optoelectronic switches are made in the dynamic regime
based on the variations of the refractive index caused by both
of the creation of free carriers and by the thermal effects due
to the recombination of the carriers. This model allows,
starting from the time variation of different diffracted
intensities, to determine the operating range of the
waveguide by adjusting key parameters such as: incident
power density, the angular offset relative to the resonance.

Key words: Optical non-linearities, Simox, Switch,
Waveguide.

1.INTRODUCTION

Control of electrical appliances by triggering a light pulse
technique became more common as the use of pulsed lasers
is widespread in many domains of physics. Even if this
technique has made a huge improvement on the uncertainty
of the time of onset of appliances constituting of laser chain,
it is still inconvenient in many applications. The search for
better accuracy on this point has led us to study and make an
optoelectronic switch which the proper operation reduces this
uncertainty, without having to add an additional element that
is optical or electronic. This optoelectronic switch as a basic
component admits one optical waveguide operating in the
nonlinear mode. On both sides of this guide were implanted
doped regions, forming a photoconductor [1,2] or a
photodiode as in the case of our study. The resulting
component generates, as soon as it receives a sufficient light
power density, an electrical signal (photocurrent) faster than
the incident light pulse which the intensity rises sharply from
a low state to a high state causing a switching.


For application to Nd-Yag laser with the wavelength
m 06 , 1 = silicon on insulator has been used for long
time to perform nonlinear waveguides [3]. Early studies were
conducted on optoelectronic switches threshold with
waveguide silicon on sapphire [4-6]. Nonlinear effects were
observed on these waveguides, but the photocreated carriers
recombination of a lifetime less than a nanosecond, leads to a
rapid development of thermal effects that alter the qualities
of the switch within the domain of a few nanoseconds pulses
[7]. Thus the SOI substrata of SIMOX type (Separation by
Implantation of OXygen) of the best crystallographic quality
are increasingly used. Various studies have been conducted
on SIMOX waveguides for diverse applications such as
making spatial light modulators [8], optical switches [9,10],
and photodetectors of photoconductive threshold power [1]
[2]. The photoconductive were studied first have the
disadvantage of high obscurity current and a relatively slow
response time that depends on the lifetime of carriers (1s)
[1]. The material SIMOX is useful for integrated optics since
the refractive index of the silica layer buried, much lower
than that of silicon, provides an excellent containment of
light in the surface layer. From silicon used as waveguide,
more improvement of the quality of the interface silicon /
buried sio
2
increases the lifetime of the carriers and reduces
optical losses. The coupling of light in the waveguide is
effected by means of a diffraction grating. A simulation
taken in the linear regime allows to optimize the coupling in
order to obtain the sharp resonances on the absorbed and
transmitted intensities [11].

The objective of this simulation is to study the temporal
evolution of intensities reflected, absorbed and transmitted,
the simulation parameters are fixed according to the
experiment. This optimizes the parameters of the waveguide,
for fast switchings and contrasted using the nonlinear optical
properties of silicon. These are related to the variation of the
refractive index of the guide, depending on the number of
photocreated carriers (electronic effects) and the variation of
temperature and Joule effect (thermal effects).

2. PRESENTATION OF THE DEVICE

Figure 1 shows the basic structure of the optical waveguide
devices on SIMOX with different light beams: The incident
beam Ii forming an angle u
i
with the normal to the guide, and
the transmitted beam It. The base component is a waveguide
ISSN 2320 2599
Volume 2, No.2, March April 2013
International Journal of Microwaves Applications
Available Online at http://warse.org/pdfs/2013/ijma03222013.pdf

Z.Al-kinani et al., International Journal of Microwaves Applications, 2(2), March April 2013, 51- 57
52
@ 2012, IJMA All Rights Reserved

on SIMOX, formed by a stack of dielectric layers (Figure. 1),
a thin silicon layer is forming the film guiding of 0,1m and
a buried layer of SiO
2
of 0,45m of thickness. This latter has a




















refractive index significantly lower than that of silicon,
resulting in an excellent containment of light in the silicon
film used as a single-mode waveguide.
The light is coupled in the guide by a diffraction grating, and
for a given angle
r
we can excite the quide mode resonantly.
A wavelength m 06 , 1 = the silicon is slightly absorptive;
the electron-hole pairs are generated in the structure. The
distances between the doped areas and between the
electrodes are respectively 14 and 24m. The creation of
carriers in silicon induces electronic effects which contribute
to reduce the refractive index, in contrast to the evolution of
thermal effects which act antagonistically. Nonlinear effects
in silicon induce a variation of the refractive index and
change the propagation constant of the guided mode. If
r
is
the resonance angle in the linear mode (negligible variation
index) at time t =0 we moves off resonance opting for an
angle of incidence as
0
+ =
r i
, so that: 0
0
( . And
progressively as the non-linear effects are formed and the
refractive index decreases, the guide retunes producing a
switching that results in a sudden variation of the transmitted
beam. In this model the waveguide is assumed to be infinite
and uniform illumination in the interelectrode space, We
study the nonlinear phenomena in the perpendicular plan to
the direction of propagation in the guide (plan xoy on Figure
1). This model is developed in two-dimensional geometry in
order to calculate the temporal variations of the refractive
index in the guiding film and their influence on the resonant
mode.

3. MODELING OF NON-LINEAR REGIME

The optical nonlinearities of electronic origin result of the
photocreation of a plasma of electron-hole pairs of density
oN(t). This is explained by a variation of an index of
electronic origin [2] :

on
e
=K
e
.oN(t) (1)

The nonlinearities thermally induced in the photodiode
resulting from the contribution of three phenomena: Auger
recombination of photocreated carriers, thermalization of
free carriers.

on
th
=K
th
. oT(t) (2)

K
e
and K
th
are constants of electronic and thermal
nonlinearities, they are of opposite sign, these values are
adjusted according to the experiment with the wavelength
m 064 , 1 = [2] : K
e
=+5.2*10
+21
cm
+1
/, K
th
=5.8*10
+4
K
+1
.
The combined effect of the two index variations of electronic
and thermal origin induces total variation index equal to:

on
t
=on
e
+on
th
(3)


The latter causes a variation of the effective index of the
guided mode and therefore a gap of ou of resonance angle u
r
,
corresponding to the resonant excitation of the fundamental
mode TE
0
. This shift is proportional to the variation of the
index ou=Kon
t
where K is a proportionality constant which
is equal to 60 [11]. The evolution of the temperature
variation oT(t) derived from the resolution of the classical
equation of heat diffusion:

C
t r S
t r T
t
t r T ) , (
) , (
) , (

= A
c
c

(4)
Where r

: is the position vector, C : the heat capacity, and

S is the heat source dissipated in the volume positionr

. In
our case, both linear steady and dynamical regimes that the
resolution method best suited to multilayer systems is that of
finite differences. In fact, the heat equation and the
conditions at the various interfaces lead to a rather complex
system of equations. Remember also that the substratum is at
ambient thermostat for heat removal, according to the
experiment. The mechanism of heat generation results from
the contribution of various processes, which are:
Photocreated carriers recombination by Auger effect. The
heat source that results is [1,2]:

S
r
=[ (oN)
3
]Eg (5)

represents the coefficient of recombination auger and E
g
is
the width of the forbidden band. The thermalization of
heated free carriers by photon absorption, which implies a
heating source given by [1,2]:

i
t
i
I
si
e
A
t
apl
S cos ) ( )] ( 1 [ = (u
i
)
(6)
470
m
14 m
V
AL N+
P+
Figure 1 : Modelized structure

Y
Z
X
I
i

I
t

Grating coupler
u
i

Si substratum

Oxyde burried
Si
guide
Z.Al-kinani et al., International Journal of Microwaves Applications, 2(2), March April 2013, 51- 57
53
@ 2012, IJMA All Rights Reserved

q(t) is the time-dependent quantum yield, through a variation
in the number of carriers and that of the temperature [1,2].
I
i
(t) is the incident light intensity. Since the heat sources are
time-dependent, the variation of the number of carriers is
determined for each given time t by solving the continuity
equation:

R G
dt
t N d
=
) (
(7)

(We neglect the electric field and the carriers diffusion). G is
the generation speed of the carriers; it is given by the
following equation [2]:

( ) ( )
i
t
i
I
si
A
si
e
t
t G

cos
) (

=


(8)

Hereq(t) is the quantum yield of carrier pairs photocreation ,
e represents the energy of the incident photons.

When we send an incident beam of intensity Ii on the
structure, it results a fraction absorbed Al
i
, a fraction
reflected R.I
i
and another transmitted.TI
i
. The relation of the
uniqueness is checked (A +R +T =1). A represents the
absorptance, it is approximated by a Lorentzian function [2] :

( ) ( )
2
/ 2 1
) (

A +
=
r i
Max
A
i
A
(9)

Where A
max
=0.17 is the maximum absorptance,
r i
,
represent respectively the angle of incidence and the angle of
resonance, and is the angular width at half maximum of
the resonance associated with the excitation of the
fundamental mode = A 1 , 0 [1,2].

4. RESULTS OF THE SIMULATIONS AND
DISCUSSION

The simulation parameters are set according to the
experiment [2]. In these optimal conditions, two parameters
are crucial for the performance optimization of the
components; that is to say that response speed, the good
switching contrast and high sensitivity, these are the incident
power density and the angular offset relative to the
resonance. To encourage non-linear effects of electronic
origin which allow fast switching, throughout our study we
dont make resonance at t=0. We chose an angle of incidence
0
+ =
r i
, as que ou
0
<0. And progressively as the non-
linear effects that are formed and the refractive index
decreases, the guide retunes producing a switching that
results in a sudden variation of the intensity absorbed.
The waveguide device is illuminated with light pulses of
10ns (full width at half height of the static resonance)
because transit time is very low. Considering an angular
offset negative (initial disagreement), we calculated the
evolution of all temporal quantities such as temperature
variation oT, the number of photocreated carriers oN, the
global variation in the refractive index on
t
that results from
it, the resulting angular offset ou and finally the evolution of
diffracted intensities.
For an incident power density as : I
c
=50KW/cm
2
, it is the
threshold power [12] and for the angular offset ou
0
=-0,1, we
calculate the evolution of different sizes (Figure 2).

0 5 10 15 20 25 30
0
1
2
3
4
(*10
17cm
-3
)
Time (ns)
(b)
0 5 10 15 20 25 30
0.00
0.02
0.04
0.06
0.08
(K)
Time (ns)
(a)
Z.Al-kinani et al., International Journal of Microwaves Applications, 2(2), March April 2013, 51- 57
54
@ 2012, IJMA All Rights Reserved










There is a sudden change in the number of photocreated
carriers (Figure 2-b), which results a negative change in the
index of electronic origin one, and a significant decrease in
the value of the refractive index of silicon (Figure 2-c). Then
follows a change in the resonance angle in relation to the
initial disagreement ou
0
, the value of which is proportional to
the global variation in the index, such as : ou
r
=Bon
t
, with
(B=60) [7], the resulting variation of the angular offset
(Figure 2-d) is remarked
r
=
0
. A highly
nonlinear behavior is highlighted on the transmitted intensity
(Figure 2-e). In fact, the increase in absorptance resulting
from angular retuning causes change of the power density
absorbed during the growth phase of the light pulse. This
accelerates the increase in the number of free carriers causing
a runaway mechanism and leading to a contrast switching of
electronic origin. It occurs to an index variation
corresponding to the retuning of the system, that is to say to
the passage through resonance when 0
0
= =
r

(Figure.2-d). This relatively fast switching occurs for a value
oN~3,24.10
17
cm
-3
of the concentration of electron-hole pairs
at time t=15.08 ns.

In our study, the transit time is very low, so we need a high
incident power density for the first switching of electronic
origin, when the temperature increases; we notice that there
is a second switch of thermal origin. This second switch is
not very pronounced and not influent on the functioning of
the photodiodes device as a fast switch threshold power.
After having demonstrated the existence of nonlinear effects
in the structure, we will study the evolution of these
depending on the incident power density and angular offset.
By determining the power density incident by I
c
=50KW/cm
2

(value causing the switching for angular offsets (
0
=-0,06,
-0,08, -0,1,-0,2), we study the evolution of the diffracted
intensities depending on angular offsets mentioned above
(Figure 3).


0 5 10 15 20 25 30
0.0
0.2
0.4
0.6
0.8
1.0
I
i
I
t
I
a
Time (ns)
(b)
0 5 10 15 20 25 30
0.0
0.2
0.4
0.6
0.8
1.0
I
a
I
t
I
i
Time (ns)
(a)
0 5 10 15 20 25 30
0.0
0.2
0.4
0.6
0.8
1.0
I
a
I
t
I
i
Time (ns)
(e)
0 5 10 15 20 25 30
-0.06
-0.04
-0.02
0.00
0.02
Time (ns)
(d)
0 5 10 15 20 25 30
-2.0
-1.5
-1.0
-0.5
0.0
(*10
-3
)
( on

)
( on
T
)
( on
th
)
Time (ns)
(c)
Figure 2: Temporal Evolution of Different Sizes for
ou
0
=-0,1 and for Power Density I
c
=50KW/cm
2
:
(a) : Temperature rise evolution, (b) : Number of carriers
evolution, (c) : Contributions to the variation of the
index, (d) : Resulting angular offset,
(e) : Absorbed and transmitted intensities evolution
Z.Al-kinani et al., International Journal of Microwaves Applications, 2(2), March April 2013, 51- 57
55
@ 2012, IJMA All Rights Reserved






We notice in these different figures that for the values of the
initial angular offset increasingly large, the switching time of
the transmitted intensity reaches more lately, but switching
becomes faster. We see that for ou
0
=-0,2, the switching did
not occur.

For this, we consider the shift of -0.2 according to the
experiment [2]. We search the optimal power, we find that
for I
c
=400KW/cm
2
, taking angular offsets increasingly large
(ou
0
=-0,15, -0,18, -0,2, -0.22) (Figure 4), there is a
significant improvement in speed. The contrast defined as the
ratio between the maximum value of the power transmitted
off-resonance value and the value of the power transmitted
with resonance also increases. The maximum value of the
intensity absorbed increases depending on the angular offset,
the sensitivity is better.







0 5 10 15 20 25 30
0.0
0.2
0.4
0.6
0.8
1.0
I
i
I
t
I
a
Time (ns)
(d)
0 5 10 15 20 25 30
0.0
0.2
0.4
0.6
0.8
1.0
I
i
I
t
I
a
Time (ns)
(c)
0 5 10 15 20 25 30
0.0
0.2
0.4
0.6
0.8
1.0
I
i
I
t
I
a
Time (ns)
(b)
0 5 10 15 20 25 30
0.0
0.2
0.4
0.6
0.8
1.0
I
i
I
t
I
a
Time (ns)
(a)
0 5 10 15 20 25 30
0.0
0.2
0.4
0.6
0.8
1.0
I
t
I
a
I
i
Time (ns)
(d)
0 5 10 15 20 25 30
0.0
0.2
0.4
0.6
0.8
1.0
I
a
I
t
I
i
Time (ns)
(c)
Figure 3 : Variation of Different Diffracted Intensities Depending
on the Initial Angular Offset for I
c
=50KW/cm
2
: (a) : ou0=-0,06,
(b) : ou
0
=-0,08, (c) : ou
0
=-0,1, (d) : ou
0
=-0,2

Figure 4 : Variation of Different Diffracted Intensities
Depending on the Initial Angular Offset for
I
c
=400KW/cm
2
: (a) : ou0=-0,15, (b) : ou
0
=-0,18,
(c) : ou
0
=-0,2, (d) : ou
0
=-0,22
Z.Al-kinani et al., International Journal of Microwaves Applications, 2(2), March April 2013, 51- 57
56
@ 2012, IJMA All Rights Reserved

By determining the angular offset by ou
0
=-0,06, we study
the evolution of nonlinearities effects depending on power
density incident (Figure 5).
In this case the switching happens earlier in time, when the
incident power density increases, it becomes less rapid. For
fairly high incident intensities (Ic > 300KW/cm
2
), we
observe a very low pronounced switching, which tends to
disappear for higher intensities.








The curves representing the temporal variation of the
resulting angular offset (Figure 6) show that the power
density I
c
=10KW/cm
2
induces a variation insufficient for
switching, there is not compensation of the angular initial
fixed and therefore not passing through resonance.
0 5 10 15 20 25 30
-0.06
-0.04
-0.02
0.00
0.02
Time (ns)
(b)
0 5 10 15 20 25 30
-0.062
-0.060
-0.058
-0.056
-0.054
-0.052
-0.050
-0.048
()
Time (ns)
(a)
0 5 10 15 20 25 30
0.0
0.2
0.4
0.6
0.8
1.0
I
t
I
a
I
i
Time (ns)
(d)
0 5 10 15 20 25 30
0.0
0.2
0.4
0.6
0.8
1.0
I
i
I
t
I
a
Time (ns)
(c)
0 5 10 15 20 25 30
0.0
0.2
0.4
0.6
0.8
1.0
I
i
I
t
I
a
Time (ns)
(b)
0 5 10 15 20 25 30
0.0
0.2
0.4
0.6
0.8
1.0
I
t
I
a
I
i
Time (ns)
(a)
Figure 5: Simulation of the Dynamic Operation of the Device at
Any Optical for ou
0
=-0,06 with Different Values of the Peak
Intensity: (a) : I
c
=10 KW/cm
2
, (b) : I
c
=30 KW/cm
2
,
(c) : I
c
=150 KW/cm
2
, (d) : I
c
=300 KW/cm
2

Z.Al-kinani et al., International Journal of Microwaves Applications, 2(2), March April 2013, 51- 57
57
@ 2012, IJMA All Rights Reserved







As soon as we increase the light intensity the equation
r
=
0
is verified and resonance passage becomes
possible. This study has highlighted the existence of a fast
switching of electronic origin on the transmitted beam. There
is also a second thermally induced switching which is not
very pronounced and not influent on the functioning of the
device (photodiode) as fast switch threshold power. We
determined the area of operation of the switch regarding to
the incident intensities as well as the initial angular offsets
depending to the resonance.

5. CONCLUSION

The purpose of our work was to study the nonlinear effects in
a silicon waveguide structure on insulator of SIMOX type
with grating coupler in optimum conditions according to the
experiment, the goal being to model photodiodes of fast
threshold. We have developed a simulation model which
takes into account both the number of excess carriers in the
structure and local variations in temperature. The simulation
results in pulsed regime highlight the existence of two
switches on the transmitted beam ; one of electronic origin
and another of thermal origin, the second switch is not very
pronounced and not influent on the functioning of the device
as a fast switch. The parameters governing the operation of
the waveguide Si on SIMOX are: the angular offset relative
to the resonance and the incident power density. We
therefore, determined the operating range of this fast switch,
as regards the power densities and the angular offset relative
to the resonance, to obtain the fastest and most contrasting
switching on the transmitted beam. We can conclude that
these components can be under certain operating conditions
that may interest the applications using laser powers.

REFERENCES

1. T.Bouchemat, F.Hobar, A.Koster, S.Laval, M.Bouchemat
et D.Pascal. Theoretical study of nonlinear fast silicon
photoconductive switch, European Physical Journal
Applied Physic, Vol. 6, pp. 165, 1999.
2. T.Boumaza, Z.al-kinani, M.bouchemat, F.Hobar,
A.Koster, D.Pascal, S.Laval. Characterisation and
modeling of a silicon-on--insulator non-linear
photoconductive switch, Optics Communications,
vol. 241, pp. 61, June 2004.
3. S.Laval, N.Paraire. Non linear optics and optical
computing, edited by S.Martelluci, A.N.Chester (Plenum
press), pp. 21, 1990.
4. H.Chelli, A.Niepceron, N.Paraire, M.Carton, S.Laval,
Photonic switching, edited by Gustafson, Smith, Springer
verlag series in electronics and photonics, pp. 129, 1988.
5. A.Koster, W.Chi, N.Paraire, S.Laval, Proceedings of the
ECO3 conferenc, Lahaye, SPIE 1280, pp. 266, 1990.
6. H.Gualous, A.Koster, W.Chi, N.Paraire, S.Laval, Optical
information technology, edited by S.D.Smith, R.F.Neale
springer verlag, pp. 209, 1993.
7. M.Carton, PhD in Science, Univ Paris XI, Orsay, 1988.
8. A.Koster, D.Pascal, R.Orobtchouk, A.Vonsovici, Alayadi,
new optical functions for optical telecommunications,
Proceedings of seminars specific action, No. 1Edition
CENT, March 1995.
9. AEBieber, DFPrelewitz, and TGBrown, Appl. Phys. Lett,
Vol. 66, No. 25,pp. 3401, 1995.
10. N.landru, D.Pascal, A.Koster, S.Laval, electron lett 36,
pp. 161, 2000.
11. D.Pascal, R.Orobtchouk, A.Layadi, and A.Koster
S.Laval, Applied Optics, Vol 12,pp. 2443, 1997.
12. Z.alkinani, Study and characterization of a threshold
photodetector whose basic component is a waveguide
silicon on SIMOX, PhD Thesis State, Univ, Constantine,
Algeria, 2005.
0 5 10 15 20 25 30
-0.08
-0.06
-0.04
-0.02
0.00
0.02
0.04
0.06
0.08
0.10
Time (ns)
(d)
0 5 10 15 20 25 30
-0.08
-0.06
-0.04
-0.02
0.00
0.02
0.04
0.06
()
Time (ns)
(c)
Figure 6: Temporal Evolution of the Angular Offset Resulting for

0
=-0,06 and for Various Values of the Incident Intensity:
(a) : I
c
=10 KW/cm
2
, (b) : I
c
=30 KW/cm
2
, (c) : I
c
=150 KW/cm
2
,
(d) : I
c
=300 KW/cm
2