An FDTD Analysis of Nonlinear Photonic Crystal Wav

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 13

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/226365908

An FDTD Analysis of Nonlinear Photonic Crystal Waveguides

Article  in  Optical and Quantum Electronics · January 2006


DOI: 10.1007/s11082-006-0012-y

CITATIONS READS

22 280

3 authors, including:

L. F. Marsal Josep Pallarès


Universitat Rovira i Virgili Universitat Rovira i Virgili
371 PUBLICATIONS   5,453 CITATIONS    255 PUBLICATIONS   4,211 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

MEMS and advanced materials View project

ALS/ MND Biomarkers and Targeted Gene Therapy View project

All content following this page was uploaded by L. F. Marsal on 01 June 2014.

The user has requested enhancement of the downloaded file.


Optical and Quantum Electronics (2006) 38:149–160 © Springer 2006
DOI 10.1007/s11082-006-0012-y

An FDTD analysis of nonlinear photonic crystal


waveguides

i . s . m a k s y m o v, l . f. m a r s a l∗ a n d j . p a l l a r è s
Departament d’Enginyeria Electrònica, Elèctrica i Automàtica, Universitat Rovira i Virgili, Campus
Sescelades, Avda. Paı̈sos Catalans 26, 43007 Tarragona, Spain
(∗ author for correspondence: E-mail: lluis.marsal@urv.net)

Received 8 September 2005; accepted 31 January 2006

Abstract. We present a modelling technique for analyzing dispersion characteristics of nonlinear pho-
tonic crystal waveguides. This technique combines the nonlinear finite-difference time-domain method,
based on the numerical simulation of oscillating dipole radiation, with the super cell approach and
the periodic boundary conditions. The technique presented makes it possible to calculate the dispersion
characteristics of Kerr-like nonlinear photonic crystals waveguides made by removing some scatterers. A
numerical tool based on this technique can be used to design various components for integrated optical
circuits.

Key words: finite-difference time-domain method, Kerr-like nonlinear material, photonic crystal,
red-shift, waveguide

1. Introduction

Photonic crystals with nonlinear materials (Slusher and Eggleton 2003)


have been the object of considerable attention. The Kerr effect makes
it possible to dynamically change the dispersion characteristics of these
periodic structures that, in turn, can be used to made novel optical
devices. Defect-free nonlinear photonic crystals have been studied numeri-
cally in (Tran 1995; Lousse and Vigneron 2001; Huttunen and Törmä 2002;
Maksymov et al. 2005a). However, it is very interesting to calculate the dis-
persion characteristics of nonlinear photonic crystals with defects. Recently,
the finite element method and the eigenmode expansion method have been
used to simulate infinite Kerr-like nonlinear photonic crystals with defects
introduced (Fujisawa and Koshiba 2005; Maes et al. 2005).
In our recent work (Maksymov et al. 2005a), we have used the finite-
difference time-domain (FDTD) method to analyze dispersion characteris-
tics of Kerr-like nonlinear 2-D photonic crystals. The super cell approach
that takes into account the defects introduced into the periodic struc-
ture has been presented and used to calculate dispersion characteristics of
nonlinear photonic crystal waveguides (Maksymov et al. 2005b, c). In this
150 I.S. MAKSYMOV ET AL.

paper, we extend the work carried out and calculate the dispersion char-
acteristics of the line-defect and coupled-cavity nonlinear photonic crystal
waveguides.
The present paper is organized as follows. First, we outline the basic
principles of the FDTD method. Second, we test our algorithm by calcu-
lating the dispersion characteristics of the defect-free nonlinear photonic
crystal. Next, we present the dispersion characteristics of the line-defect
and coupled-cavity nonlinear photonic crystal waveguides. Finally, we dis-
cuss how the nonlinearity influences the dispersion characteristics and how
the operating power of a device based on the structures considered can be
reduced.

2. Theory

2.1. numerical method

In this paper, without loss of generality we present results for the TM


polarization only (the electric field is parallel to the rods). However the
method can be used in the case of the TE polarization. We show how
the Maxwell equations take into account the electric dipole embedded into
the photonic crystal (Jackson 1975).

∂H (r, t)
∇ × E (r, t) = −µµ0 , (1)
∂t

∂D (r, t) ∂PE (r, t)


∇ × H (r, t) = + , (2)
∂t ∂t

where E (r, t) and H (r, t) are electric and magnetic fields, respectively;
D (r, t) is the electric displacement, PE (r, t) is the polarization field of the
electric dipole, which can be written as (Sakoda 2001)

PE (r, t) = eδ (r − r0 ) exp (iωt) , (3)

where e is the amplitude of the electric dipole, r0 indicates the position of


the dipole within the photonic crystal and ω is the angular frequency of the
oscillation.
In order to solve Equations (1) and (2), we need a so-called constitutive
equation that relates D (r, t) to E (r, t). For the Kerr-like nonlinear medium,
the dielectric constant depends on the electric field E (r, t) and the Kerr
FDTD ANALYSIS OF NONLINEAR PHOTONIC CRYSTAL WAVEGUIDES 151
coefficient χ (3) as ε (r) = ε  + χ (3) |E (r, t)| , where ε  is the intensity-indepen-
2

dent dielectric constant. The constitutive equation is written as


 
D (r, t) = ε0 ε + χ (3) |E (r, t)|2 E (r, t) . (4)

In the equations above, ε0 and µ0 are the dielectric permittivity and the
magnetic permeability of vacuum.
Now we can solve Equations (1) and (2) by discretizing them both in
space and time. In order to define the dielectric constant profile, the sub-
gridding is used. It reduces the computation domain and saves more com-
putation time when calculating with the FDTD.
In the nonlinear case the electric field E (r, t) must be calculated from the
constitutive Equation (4). In order to do so, the following cubic equation
has to be solved analytically (Tran 1995)

 (3) 2 3  
χ  · A + 2Re ε ∗ χ (3) · A2 + |ε|2 · A − |D (r, t)| = 0,
2
(5)
ε02

where A = |E (r, t)|2 is the modulus square of the electric field. Because the
modulus square can be neither negative nor complex, only a real positive
root is selected. Once A is obtained, the electric field E (r, t) is calculated
as
D (r, t)
E (r, t) =   . (6)
ε + χ (3) · A ε0

The fields E (r, t) and H (r, t) are recorded and used to calculate the elec-
tromagnetic energy density emitted from the dipole (Jackson 1975)
 
W = 21 E (r, t) · D (r, t) + µµ0 |H (r, t)|2 . (7)

The peaks in the electromagnetic energy spectra give the eigenfrequencies.


All the operations described above are repeated for each value of the wave
vector k, which is chosen along the edges of the irreducible Brillouin zone
(BZ).
In order to numerically solve Equations (1) and (2), we use the finite-
difference scheme proposed by Yee (Taflove 1995). In accordance with the
Bloch theorem for a periodic structure (Sakoda 2001), we can write the
periodic boundary conditions (BC) for the fields E (r, t) and H (r, t) as

E (r + a, t) = E (r, t) · exp (ika) , (8)


H (r + a, t) = H (r, t) · exp (ika) , (9)

where a is the lattice vector and k is the wave vector under consideration.
152 I.S. MAKSYMOV ET AL.

2.2. intensity estimation

In order to estimate the intensity needed to observe the phenomena that


will be discussed in the following section, we use the electric field ampli-
tude obtained in the main body of the FDTD algorithm and taken at the
angular frequency and the wave vector of interest. These values correspond
to the working point on the defect mode of the device, which we model.
The intensity is calculated as
√  2
I = 21 cε0 ε E (r, t)ω,k  , (10)

where c is the light velocity in vacuum. In order to draw a parallel between


the amplitude of the electric dipole e and the intensity I one should use
the following rule. The dipole should be situated at the centre of the com-
putation domain. A test calculation is carried out in which the electric field
amplitude induced by the dipole is obtained. The intensity induced depends
on the size of the computational domain and, if the size is changed, the
amplitude of the electric dipole must be increased and the test calculation
must be repeated.
In Equation (10), the intensity I is measured in W/m2 . In Equation (4),
however, we use the Kerr coefficient χ (3) measured in m2 /V2 . In the liter-
ature, the nonlinear index of refraction sometimes is expressed in terms of
intensity, i.e., nI2 measured in m2 /W. In the following section we will use the
Kerr coefficient of AlGaAs (Aitchison et al. 1997). The nonlinear refractive
index of AlGaAs measured for the TM polarization at an input wavelength
of 1.55 µm is nI2 = 1.43 × 10−17 m2 /W. In order to obtain χ (3) , the following
expression is used

χ (3) = nI2 cε0 ε . (11)

From Equation (11) we obtain χ (3) = 4.4 × 10−19 m2 /V2 . This value will be
used in the following section.

3. Results and discussion

3.1. two-dimensional nonlinear photonic crystal

In order to test our algorithm, we consider a square lattice of circular rods


made of AlGaAs with ε  = 11.56 (Aitchison et al. 1997). Throughout the
paper, the background material is assumed to be air (ε = 1, χ (3) = 0). The
properties of inverse structures consisting of air holes drilled in a nonlinear
FDTD ANALYSIS OF NONLINEAR PHOTONIC CRYSTAL WAVEGUIDES 153
background were studied e.g. in previous works (Maksymov et al. 2005a, d).
The distance between the rods is a = 1, the radius of each rod is r = 0.2a. In
our calculation, the computation domain is divided into 20 × 20 cells. The
total number of time steps is 250 and the number of oscillating cycles is 50.
On a computer with two 3.0 GHz AMD64 processors and 2 Gb RAM, the
calculation takes about 15 min.
Fig. 1(a) shows the computation domain (grey region) for the calcula-
tion of the dispersion characteristic. Fig. 1(b) shows the dispersion char-
acteristics for the X direction of the first BZ. The solid line corresponds
to the linear dispersion characteristic whereas the dot and dash lines cor-
respond to the nonlinear dispersion characteristics calculated at incident
intensities of 9 and 21 GW/cm2 , respectively. These values are calculated
ka
for the two lowest dispersion curves at 2π = 0.5. The lowest dispersion
curve forms one of the margins of the forbidden gap whose position must
be controlled in order to make optical limiting devices. As we can see,
in the nonlinear regime the dispersion curves are red-shifted with regard
to the linear regime. Such a red-shift takes place if the Kerr coefficient
is positive. Negative Kerr coefficients give rise to a blue-shift. The result
obtained is in good agreement with previous calculations (Tran 1995; Maes
et al. 2005). It should be noticed that the red-shift is weaker for the low-
est dispersion curve due to the impact of the group velocity, which will be
discussed below.

3.2. line-defect nonlinear photonic crystal waveguide

Fig. 2(a) shows the geometry of the line-defect nonlinear photonic crystal
waveguide, which is made by removing a row of rods in the X direction
of the first BZ. In order to calculate the dispersion characteristics of such
a structure, the super cell approach is used. As Fig. 2(a) shows, the super
cell (20 × 140 cells) takes both the defect and the surrounding rows of rods
into account. The material parameters are equal to those in the previous
section. The radius of each rod is r = 0.18a.
Fig. 2(b) shows the dispersion characteristics calculated for the X direc-
tion of the first BZ. We notice that the defect mode no longer exists within
ωa
the forbidden band gap for 2π c
> 0.4 and, therefore, we do not show this
range of frequencies. The solid line corresponds to the linear dispersion
characteristic whereas the dot and dash lines correspond to the nonlin-
ear dispersion characteristics calculated at incident intensities of 19 and
ka
26 GW/cm2 , respectively. These values have been calculated at 2π = 0. It can
be seen that in the nonlinear regime the dispersion curves are red-shifted
with regard to the linear regime.
154 I.S. MAKSYMOV ET AL.

Fig. 1. (a) Geometry of the 2-D nonlinear photonic crystal. Grey region denotes the cell used for the
calculation of the dispersion characteristic. Black corresponds to the Kerr-like material whereas white
corresponds to the background. (b) The dispersion characteristics calculated for the linear (solid line)
and nonlinear (dot and dash lines) regimes.

3.3. coupled-cavity nonlinear photonic crystal waveguide

In this subsection we present the dispersion characteristics of a coupled-


cavity waveguide made in the nonlinear photonic crystal. Fig. 3(a) shows
the geometry of this waveguide. Such a waveguide is based on evanes-
cent coupling between modes of neighbouring resonators. As it was shown
(Yariv et al. 1999), in the linear regime the dispersion curve has a cosine
form. The group velocity of this curve tends to be very small.
FDTD ANALYSIS OF NONLINEAR PHOTONIC CRYSTAL WAVEGUIDES 155

Fig. 2. (a) Geometry of the line-defect nonlinear photonic crystal. Grey region denotes the super cell
used for the calculation of the dispersion characteristic. (b) The dispersion characteristics calculated for
the linear (solid line) and nonlinear (dot and dash lines) regimes.

In order to carry out the calculation, the super cell is divided into 60 ×
140 cells. The calculation usually takes less time because the dispersion
curves are almost flat and therefore small range of frequency is considered.
156 I.S. MAKSYMOV ET AL.

Fig. 3. (a) Geometry of the coupled-cavity nonlinear photonic crystal. Grey region denotes the super
cell used for the calculation of the dispersion characteristic. (b) The dispersion characteristics calculated
for the linear (solid line) and nonlinear (dot and dash lines) regimes.
FDTD ANALYSIS OF NONLINEAR PHOTONIC CRYSTAL WAVEGUIDES 157
The dispersion curves calculated with our method are shown in Fig. 3(b).
The solid line corresponds to the linear dispersion characteristic whereas
the dot and dash lines correspond to the nonlinear dispersion characteris-
tics. The incident intensities are 9.5 and 16 GW/cm2 . These values are cal-
ka
culated at 2π = 0. As we can see, for the nonlinear regimes there are again
ka
red-shifts. A tiny blue-shift at 2π = 0.5 is a numerical artefact due to a weak
numerical noise that appears while calculating the electromagnetic energy
density at the edge of the first BZ.

3.4. red-shift dependence on the group velocity

Let us discuss the properties of the red-shift that we have seen in the pre-
vious subsections. First, we notice that the red-shift is not uniform along
the X direction of the first BZ (e.g. in Fig. 2 for k = 0 it is larger than
ka
for 2π = 0.4). It can be explained by a difference in the group velocities,
which correspond to different k-points. As it was shown earlier (Yariv et al.
1999; Panoiu et al. 2004), the electric field in the waveguide mode is inverse
proportional to the group velocity and the waveguide modes that corre-
spond to lower group velocity, in turn, induce stronger nonlinear effects
(e.g. larger red-shift). In Fig. 4, we present the group velocities calculated
by numerical differentiation of the frequency manifolds ω(k). The dash line
represents the group velocity dispersion of the defect modes of the line-
defect waveguide. The group velocity dispersion of the coupled-cavity wave-
guide is marked by the solid line. The result obtained demonstrates that the
group velocities of both defect modes increase as the value of the wave vec-
tor increases. On the contrary, the red-shift decreases as the wave vector
increases. It confirms the above stated.
Second, we can see that with the coupled-cavity waveguide one can
achieve group velocities less that 0.1c (c denotes the light velocity), and
thus a large decrease in the operating power. One can also use similar
structures whose mode dispersion is specially engineered and achieve ultra
small values of the group velocity (Mookherjea 2002; Altug and Vuĉković
2004). Such structures are widely used in a variety of devices ranging from
optical delay components to low-threshold lasers. While developing such
devices, specific attention should be paid to the estimation of the intensity
enhancement in the coupled-cavity waveguides. The intensity inside them
is higher than that of the input pulse because the reduction of the group
velocity results in the spatial compression of the pulse inside the cavities.
In fact, the incident intensity applied to a coupled-cavity waveguide can be
written as (Yariv and Yeh 1984)
158 I.S. MAKSYMOV ET AL.

Fig. 4. Group velocity dispersion of the defect modes of the line-defect (dash line) and coupled-cavity
(solid line) waveguides.

 2
 
I = 21 εeff ε0 vgr EC  , (12)

 2
 
where EC  is the modulus square of the electric field inside the cavities
and εeff is the effective dielectric constant, which is a function of the both
the angular frequency and the intensity. Using Equation (10) we obtain that
the intensity inside the cavities is related to the incident intensity as

IC = I · S, (13)

where S = vgr √c εeff is the slowing or enhancement factor in accordance with


the definition by (Melloni et al. 2003). The group velocities for the line-
ka
defect and the coupled-cavity waveguides taken at 2π = 0.3 are 0.53c and
0.11c, respectively, By using these values and assuming that εeff = 12.25, we
obtain S = 0.54 and S = 2.6, respectively.
Finally, we notice that a two-fold increase in the red-shift is not neces-
sarily associated with a two-fold increase in the intensity. This can be due
to the nonlinear behaviour of the oscillating dipole (to our knowledge, its
behaviour in nonlinear photonic crystals is not yet studied). However, we
obtained a similar result in our previous work (Maksymov et al. 2005d) in
FDTD ANALYSIS OF NONLINEAR PHOTONIC CRYSTAL WAVEGUIDES 159
which the conventional FDTD method was used. It allows to draw a con-
clusion that the magnitude of the red-shift is increased by the intensity, but
this dependence is not generally linear.

4. Conclusions

We have presented a modelling technique for analyzing dispersion charac-


teristics of nonlinear photonic crystal waveguides. This technique is based
on the nonlinear FDTD method and it uses the super cells approach and
the periodic boundary conditions. We have calculated the dispersion charac-
teristics of Kerr-like nonlinear photonic crystals with different defects intro-
duced. Particularly, we have considered a line-defect and a coupled-cavity
waveguide. It has been shown that in the nonlinear regime the dispersion
characteristics are red-shifted with regard to those in the linear regime. The
red-shift depends on the intensity and its magnitude is inverse proportional
to the group velocity. These results are in good agreement with those pre-
dicted theoretically or obtained with other approaches. We have also shown
that the group velocity of the defect modes of the coupled-cavity wave-
guides is very small, which in turn greatly enhances the efficiency of optical
devices based on these structures.

Acknowledgements

This work was supported by Ministerio de Educación y Ciencia, Project


No. TEC2005-02038.

References

Aitchison, J.S., D.C. Hutchings, J.U. Kang, G.I. Stegeman and A. Villeneuve. IEEE J. Quantum Electron.
33 341, 1997.
Altug, H. and J. Vuĉković. Appl. Phys.Lett. 84 161, 2004.
Fujisawa, T. and M. Koshiba. J. Lightwave Tech. 23 382, 2005.
Huttunen, A. and P. Törmä. J. Appl. Phys. 91 3988, 2002.
Jackson, J.D. Classical Electrodynamics, Wiley, New York, 1975.
Lousse, V. and J.P. Vigneron. Phys. Rev. E 63 027602, 2001.
Maes, B., P. Bienstman and R. Baets. J. Opt. Soc. Am. B 22 613, 2005.
Maksymov, I.S., L.F. Marsal and J. Pallarès. Opt. Commun. 248 469, 2005a.
Maksymov, I.S., L.F. Marsal and J. Pallarès. In: Proceedings of the 13th International Workshop on Opti-
cal Waveguide Theory and Numerical Modelling, Grenoble, France, 8–9 April 2005, 66, 2005b.
Maksymov, I.S., L.F. Marsal and J. Pallarès. In: Proceedings of the 6th Symposium on Photonic and Elec-
tromagnetic Crystal Structures, Crete, Greece, 19–24 June 2005c.
Maksymov, I.S., L.F. Marsal and J. Pallarès. Opt. Quantum. Electron. 37 161, 2005d.
Melloni, A., F. Morichetti and M. Martinelli. Opt. Quantum. Electron. 35 365, 2003.
160 I.S. MAKSYMOV ET AL.

Mookherjea, S. IEEE J. Sel. Top. Quantum Electron. 8 448, 2002.


Panoiu, N.C., M. Bahl and R.M. Osgood, J. Opt. Express 12 1605, 2004.
Sakoda, K. Optical Properties of Photonic Crystals. Springer Verlag, Berlin, 2001.
Slusher, R.E. and B.J. Eggleton. Nonlinear Photonic Crystals. Springer Verlag, Berlin, 2003.
Taflove, A. Computational Electrodynamics, Artech House, Boston, 1995.
Tran, P. Phys. Rev. B 52 10673, 1995.
Yariv, A. and P. Yeh. Optical Waves in Crystals, Wiley, New York, 1984.
Yariv, A., Y. Xu, R.K. Lee and A. Scherer. Opt. Lett. 24 711, 1999.

View publication stats

You might also like