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

net/publication/3423938

Design and Analysis of Ultra-Miniaturized Meandering Photonic Crystals

Delay Lines

Article  in  IEEE Transactions on Advanced Packaging · June 2008

DOI: 10.1109/TADVP.2008.916284 · Source: IEEE Xplore

CITATIONS READS
8 258

4 authors, including:

Mohammad Fakharzadeh Omar Ramahi

Sharif University of Technology University of Waterloo
77 PUBLICATIONS   714 CITATIONS    394 PUBLICATIONS   4,383 CITATIONS   

SEE PROFILE SEE PROFILE

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

Millimeter-wave imaging View project

Microwave Near-Field Sensors for Material Detection View project

All content following this page was uploaded by Omar Ramahi on 15 January 2015.

The user has requested enhancement of the downloaded file.

IEEE TRANSACTIONS ON ADVANCED PACKAGING, VOL. 31, NO. 2, MAY 2008 311

Design and Analysis of Ultra-Miniaturized

Meandering Photonic Crystals Delay Lines
Mohammad Fakharzadeh, Omar M. Ramahi, Safiedin Safavi-Naeini, and Sujeet K. Chaudhuri

Abstract—In this paper, we study the characteristics of a novel

miniaturized optical delay line, which delays light in a meandering
photonic crystal waveguide, and describe the design steps. We show
how lattice parameters and refractive index difference of the pho-
tonic crystal affect the bandgap width and suggest a criterion to
select these parameters. Next, we focus on the parallel waveguide
channels in photonic crystal, and analyze the impact of the channel
length and the interchannel spacing on crosstalk. We suggest a
method for mitering the sharp corners in meandering lines which
reduces the undesired reflections by 8 dB. Considering all these
guidelines, we examine the propagation of light in the proposed
delay line through calculating time-delay and insertion loss. To
achieve longer delays in a small device area, we concentrate on
coupled cavities in photonic crystals and propose an approximate
method for calculating the group velocity of light in the coupled
defects. We show how by replacing waveguide channels of a mean-
dering delay line with coupled defects we achieve time-delays more
than 9 ps within a device size around 27 m, which corresponds to
a miniaturization factor of 100.
Index Terms—Coupled cavities, crosstalk, finite difference
time domain (FDTD), group velocity, optical delay line, photonic
crystal. Fig. 1. Two types of meandering line structures. (a) Serpentine. (b) Double
spiral.

I. INTRODUCTION
uniform grating. The performance of such delay lines is only
O PTICAL delay lines (ODLs) are widely used in numerous
applications, such as broadband beamforming of phased
array systems, optical division multiplexing, filter design and
two times better than free space in terms of the required prop-
agation length, so still the packaging and integration problems
optical signal processing. Traditionally, optical fibers have been exist.
used as the main element of the ODL, because fiber is a low loss Use of photonic crystal (PC) is a promising solution to re-
medium and supports the propagation of large bandwidth sig- duce the size of delay elements. Very small group velocities are
nals. Use of long lengths of fiber to generate large time delays achievable at the band edges of PCs [3], and bandgap mecha-
is unavoidable, however, for short time delays ( s), still long nism slows down the propagation of light. The crystal can con-
pieces of fiber are required. For example, Chang et al. [1] used fine light around sharp bends, in lower-index media, and within
a 940 m graded index multimode fiber as delay line to generate wavelength-scale cavities. The other important advantage of PC
4.99 s/km delay. Such delay lines are heavy, lossy, and bulky, is that it can be integrated with silicon-on-insulator (SOI) and
furthermore, integration of long lengths of fiber with other op- CMOS devices to achieve a low cost product. Recently, many
tical subsystems is not practical. works have been reported on the design and fabrication of pho-
A solution to achieve compact delay lines, which appeared tonic crystal devices, specifically delay lines [4]–[10]. How-
in the early 1990s, was the use of fiber Bragg grating instead ever, all of these works, use straight line structures in PC, so to
of regular single mode fibers. For example, Ortega et al. [2] in- achieve longer time delays, the device size must be increased.
troduced an ODL with a maximum delay of 330 ps for a 5 cm In this paper, the main focus is on the design and analysis
of meandering photonic crystal delay lines. We have pro-
Manuscript received July 5, 2007; revised September 20, 2007. This work
posed carving meandering line structures in photonic crystals
was supported by an industrial research chair program sponsored by Research to make delay lines [11]. Meandering line structures have a
In Motion (RIM) and Natural Sciences and Engineering Research Council of high packing density which results in a more efficient use of
Canada (NSERC). This work was recommended for publication by Associate
Editor L.-T. Hwang upon evaluation of the reviewers comments.
the device area. General meandering lines consist of parallel
The authors are with the Electrical and Computer Engineering Depart- waveguides connected with crossed interconnections. Two
ment, University of Waterloo, Waterloo, ON N2L 3G1, Canada (e-mail: conventional arrangements for this category of delay lines,
[email protected]).
Color versions of one or more of the figures in this paper are available online
which are used in microwave frequencies, are serpentine (me-
at http://ieeexplore.ieee.org. ander) and double spiral. Fig. 1 illustrates the geometry of
Digital Object Identifier 10.1109/TADVP.2008.916284 these two delay lines. We will focus on two different structures
1521-3323/$25.00 © 2008 IEEE

Authorized licensed use limited to: University of Waterloo. Downloaded on July 27, 2009 at 15:14 from IEEE Xplore. Restrictions apply.
312 IEEE TRANSACTIONS ON ADVANCED PACKAGING, VOL. 31, NO. 2, MAY 2008

Fig. 2. Top view of two conventional structures for 2-D photonic crystals. Left:
Triangular lattice. Right: Square lattice.

to implement ODLs in photonic crystals: 1) waveguide-based

delay lines and 2) coupled cavity (defect) delay lines.
The organization of this paper is as follows. In Section II, Fig. 3. Gap map for the square lattice of silicon rods in air.
we discuss the design steps of a meandering delay line in PC.
We propose a criterion for selecting the lattice parameters,
and show how crosstalk between parallel waveguides can be this structure completely, we need to find the material, the ra-
reduced by choosing appropriate dimensions for waveguide dius of cylinders, , and the lattice constant, , which is de-
channels. In Section III, we introduce reflective spiral delay fined as the distance between centers of the adjacent cylinders.
line and use full-wave analysis to examine its characteristics. In
Choosing high and low index material is restricted to the current
Section IV, we discuss coupled cavity delay lines and illustrate
technology and the available transparent material with lossless
that replacing PC waveguides with coupled cavities can signif-
optical propagation. Usually Si or GaAs
icantly increase the time delay and decrease the insertion loss.
is used as the high-index material and air or silica
Finally, in Section V, we present the conclusions of this paper.
is used as the low-index medium [13]. We start with
Si/Air configuration and discuss other material later.
II. DESIGN STEPS FOR PHOTONIC CRYSTAL DELAY LINES
Photonic crystals, also known as photonic bandgap (PBG) B. Finding Lattice Constant and Radius
material, are structures with a periodically modulated dielectric Existence of the bandgap and the width of the bandgap de-
material constant. If the depth of the refractive index modula- pend on the ratio of rod radius over lattice constant, . Due
tion is adequate, there will be a certain frequency range, namely to the scaling property of Maxwell equations, as long as the
bandgap, where optical modes are forbidden to propagate. PCs material properties remain the same, solving a problem at one
are divided into three types according to the dimensionality of length scale determines the solutions at other length scales [21].
their periodicity, namely 1-D, 2-D, and 3-D PCs. In reality di- Therefore, we initially fix at 1 m and vary from
mensions of a PC structure are limited and the periodicity is to in steps of . We use the plane wave expansion
terminated at some point. Instead of using fictitious 2-D or 3-D method to find the bandgap for each value of . Eventually,
PCs, photonic crystal slabs have been proposed as a more easily a graph like Fig. 3 is achieved. This graph is called gap-map
fabricated alternative to 3-D PCs. PC slabs are 2-D periodic and conveys important information for design of PBG devices.
dielectric structures of finite height with a band gap for prop- Fig. 3 shows that the maximum bandgap for Si/Air configuration
agation in the plane of periodicity. The index-confinement is occurs for . Having the widest bandgap is an appro-
used for guiding light in the third dimension. PC slabs are also priate criterion for setting lattice parameters, because it compen-
called quasi-2-D PCs because their structure and properties im- sates for the fabrication errors and uncertainties, and enables the
itates those of 2-D PCs [12]. Since the fullwave analysis of 3-D design of a wide-band device. The vertical axis in Fig. 3 shows
structures is time-consuming and needs significant amount of the normalized frequency represented by . If we want to
memory, we focus on 2-D structures in the rest of this paper. design a photonic crystal waveguide to transmit a modulated op-
tical signal, it is wise to set the center frequency equal to the op-
A. Photonic Crystal Structures
tical carrier frequency. Therefore, even for ultra-wideband sig-
Amongst different periodic structures for 2-D PCs, two struc- nals, a negligible fraction of the power will be absorbed. For this
tures have been more popular named triangular and square lat- configuration, the widest bandgap is centered at . So,
tice which are demonstrated in Fig. 2. Each of these structures for the optical communication wavelength of m the
can be obtained by embedding rods of a high-index material in lattice constant will be nm, and the cylinder’s radius,
a low-index medium or drilling holes in a high-index medium. is 103 nm.
These holes are filled with a low-index material like air or silica Further analysis shows as , the refractive index difference
(SiO ). Triangular lattice PCs support both TE and TM modes, of the material which constitute the PC, increases the bandgap
where in TM modes H-field is parallel to the rods, while in TE widens and its center approaches lower frequencies. The reverse
modes E-field has this property. Square-lattice PCs only support happens if decreases. Fig. 4 demonstrates the gap-map for
TE modes but they generally possess a wider bandgap compared Si/SiO configuration where . The widest bandgap oc-
to triangular structures. In the rest of this work, we focus on curs for a larger ratio of compared to Air/Si configuration
square-lattice PCs consisting of high-index cylinders. To specify (0.226 instead of 0.18).

Authorized licensed use limited to: University of Waterloo. Downloaded on July 27, 2009 at 15:14 from IEEE Xplore. Restrictions apply.
FAKHARZADEH et al.: DESIGN AND ANALYSIS OF ULTRA-MINIATURIZED MEANDERING PHOTONIC CRYSTALS DELAY LINES 313

Fig. 4. Gap map for the square lattice of silicon rods in SiO .

Fig. 5. Basic structure used as the test-bed for the coupling analysis in parallel
PC waveguides.
C. Crosstalk Reduction
Crosstalk is a major drawback of meandering delay lines.
If we place the vertical channels in the delay lines of Fig. 1
as close as possible and increase their length, the propagation
path enlarges. However, fast coupling between parallel lines
(or crosstalk) limits the maximum length of the lines and
the minimum distance between the adjacent lines. Crosstalk
between the parallel photonic crystal waveguides (PCW) has
been discussed in [14]–[19]. In these earlier works the interest
was mostly on analyzing the variation of coupling versus wave-
length or finding the coupling length. Here, we study the effects
of the channel length and spacing variations to tradeoff between
crosstalk and packing density of waveguide channels. Indeed,
we have two incompatible goals, we want to: 1) keep the
crosstalk between parallel channels as low as possible to assure
that light is transferred only through propagation and 2) use the
device area efficiently by placing waveguide channels as close Fig. 6. Transmittance and coupling factors for different interchannel spacings.
as possible. We use the finite difference time domain (FDTD) L, the channel length is 25 m.
method to study the propagation of light in the PC waveguide
channels. Fig. 5 shows the test-bed used for the current studies,
which consists of the silicon rods suspended in air. This struc- output of the transmission and coupling channels, respectively,
ture is the PC equivalent of a microwave directional coupler. denoted by and . Then by integrating and over
Transmission chancel in Fig. 5 is the forward path that transfers time, we find the total transmitted and coupled energy. Hence,
energy from the source to the output. Coupling channel is a two parameters namely transmittance ( ) and coupling ( ) fac-
parallel path without any independent source that is excited tors are defined
through coupling of energy. denotes the number of rows or
the distance between these two channels . (2)
We use pulse excitation as the input to the system under test
and measure (record) the amplitude or envelope of the electric and
field at the desired points. The excitation pulse consists of a
Gaussian envelope modulation on a sinusoidal carrier (3)

(1) Fig. 6 presents the calculated transmittance and coupling

factor for . we observe that for , the
where is pulse width, is the initial delay, and is the center transmittance is dB which means more than 91%
wavelength of the pulse. of the input energy directly transmits and the rest is coupled to
For the following set of simulations in (1) is equal to 15 cy- the coupling channel, absorbed into the crystal (due to the large
cles of the carrier wave. This setting would concentrate most of bandwidth of the source signal), or reflected back to the input
the transmitted energy in a bandwidth equal to 4 THz, provided port. For , the transmittance increases to dB
that the center wavelength, is set to 1.55 m. To find the best which means 99% of the input energy is transmitted in the
channel spacing, we fix the channel length at 25 m and cal- desired channel, so we conclude that is a proper choice
culate the instantaneous power at the input port, and at the for the interchannel spacing in the meandering PC structures.

Authorized licensed use limited to: University of Waterloo. Downloaded on July 27, 2009 at 15:14 from IEEE Xplore. Restrictions apply.
314 IEEE TRANSACTIONS ON ADVANCED PACKAGING, VOL. 31, NO. 2, MAY 2008

Fig. 8. Method used for mitering.

Fig. 7. Transmittance versus channel length ( = 1:55 m). By increasing

the spacing between parallel channels transmittance increases and crosstalk
decreases.

Crosstalk also depends on channel length, and for suffi-

ciently long parallel waveguides, it has a periodic behavior
versus length. To study the effect of the channel length on
crosstalk of parallel waveguide channels in PC, we use a
test-bed structure similar to Fig. 5 with the same excitation.
The channel spacing, alters from 3 to 5, and the channel
length, , varies from 10 to 200 m. Fig. 7 illustrates the FDTD Fig. 9. Reflection from two corners. UMC: Unmodified corners. MC: Mitered
results. It is seen that for Transmittance decreases corners.
dramatically as rises, however for , the drop in
is smoother and even negligible for . Since we want
to achieve long delays with low insertion loss, for the rest of
this paper we consider five layers of Si rods between parallel
waveguide channels. Notice that these results are based on
2-D FDTD simulations which does not support scattering loss
calculations. In reality the limited length of Si rods (or air
holes) in PC results in scattering loss. Fortunately, by refractive
index confinement, we are able to reduce the out-of-plane loss
of PC waveguides. Wan Kuang et al. [17] used 3-D fullwave
simulations to show low loss propagation in the PC waveguides
with index-confinement is possible.
Fig. 10. Top view of Reflective Spiral Line carved in a square lattice PC.

D. Corner Mitering

A meandering path, which is our desired structure for delay III. REFLECTIVE SPIRAL LINE IN PHOTONIC CRYSTAL
line consists of several sharp bends which are made to increase In transmission line theory, an open-circuit termination
the net propagation length of light. Initially, the exclusive pos- causes total reflection, so the forward travelling wave totally
sibility of making sharp bends in PC, stimulated us to develop returns toward the input terminal. Similarly, in a well-designed
meandering PC channels. However, it is necessary to miter all PC waveguide the walls behave like mirrors which can reflect
corners to reduce the reflection from each corner, otherwise a lightwave [21]; so, if one puts a mirror at the end of the wave-
large part of the input power will be lost through multiple corner guide, all of the travelling lightwave will return. This concept
reflections. Fig. 8 graphically shows the method we used for becomes realizable by removing a certain number of cells in
mitering corners [18]. We shift three rods at the corner diago- a PBG lattice to open a channel and stopping at some point
nally, and reduce their radii according to the values shown in the which is adequately far from the boundaries. The closed end of
figure. Fig. 9 compares the reflection coefficient for two suc- the channel behaves exactly like a mirror for the wavelengths
cessive corners in a meandering line carved in PC before and inside the bandgap, with a reflection coefficient close to 1.
after mitering. It is shown that after mitering the reflection co- Fig. 10 shows the implementation of this idea in photonic
efficients reduce by at least 8 dB. Therefore, to achieve low loss crystal. We have introduced the novel structure arising from
delay lines, corner mitering in meandering PC structures is un- this arrangement, henceforth called reflective spiral line in
avoidable. [11], for the first time. As mentioned in the previous section,

Authorized licensed use limited to: University of Waterloo. Downloaded on July 27, 2009 at 15:14 from IEEE Xplore. Restrictions apply.
FAKHARZADEH et al.: DESIGN AND ANALYSIS OF ULTRA-MINIATURIZED MEANDERING PHOTONIC CRYSTALS DELAY LINES 315

Fig. 11. Envelope of the electric field detected at the input/output of the Si/Air
reflective spiral delay line.

Fig. 12. Delay and loss versus wavelength for Reflective Spiral Line.
to reduce the crosstalk, we fix at 5 and channel lengths
are less than 21 m. Since reflective spiral line is a single port
device, an optical circulator will be required to separate input
and output signals. In [22], an optical circulator formed by
inserting magneto-optical cavity in a 2-D PC is proposed.
Fig. 11 shows the envelope of the electric field recorded at the
input/output of this delay line. We define delay as the time dif-
ference between the maximum of the input pulse envelope, cen-
tered at , and the maximum of the pulse response recorded
at the output terminal. Loss is defined as the ratio of the electric
field amplitudes at these two points. These are two qualitative
criteria to facilitate the comparison of different structures. The
FDTD simulations show that the time-delay of this structure is
1.99 ps and the loss is less than 0.3 dB. The largest device size, Fig. 13. Envelope of the electric field detected at the input/output of the Si/SiO
including the margins, is only m. reflective spiral delay line.
To provide a meaningful comparison of the time delay relative
to the device size, we define a figure of merit for delay , as
tively. This time the lattice constant is smaller than that of Si/Air
(4) configuration, so the length of the Si/SiO reflective spiral delay
line will be smaller by a scaling factor corresponding to the ratio
where is the speed of light in free space, is the time delay, of the lattice constants of both structures. Fig. 13 demonstrates
and is the device size. This figure of merit simply states how the envelope of the electric field recorded at the input/output
many times light can be slowed down in this structure compared of the Si/SiO reflective spiral delay line. We observe that both
to free space. The effect of the meandering shape is also in- delay and loss have increased to 2.21 ps and 3.3 dB, respec-
cluded in . For the delay line displayed in Fig. 10, we can tively, however the device size has shrunk to 20.7 m. There-
show that . In [11], we have compared reflective fore, for Si/SiO reflective spiral delay line, which
spiral structure with conventional structures for delay lines, i.e., implies that the propagation of light Si/SiO PC delay line is
meandering line and double spiral line. We have shown that the 1.45 times slower than Si/Air configurations.
performance of this delay line in terms of the propagation delay,
loss, and cross talk is at least two times better than other struc- IV. COUPLED CAVITIES IN PHOTONIC CRYSTALS
tures with the same device size. A cavity in a PC is a modified cell in the periodic lattice. For
Further analysis shows that the generated delay in reflective example, in the PBG structure shown in Fig. 2, changing the ra-
spiral structure increases as the optical carrier frequency in- dius of any rod/hole from the default value, or even removing
creases (Fig. 12). For example, if we change the wavelength one rod/hole, makes a local cavity or so called defect [23], [24].
from 1.49 m to 1.61 m—almost 8%— the time delay in- Periodic defects in PC function as a low-loss waveguide, and the
creases by 25%. However the propagation loss decreases as the infinitesimal size of such structures is the key to the ultra-minia-
wavelength approaches the center wavelength (1.55 m) and turized delay lines. The group velocity of the light propagated
after that it increases. in the defects can be controlled since it is inversely proportional
If we want to implement the same structure with Si/SiO , con- to the coupling strength between the adjacent cavities [7].
figuration we have to change the lattice parameters. As Fig. 4
illustrates, for this case the maximum bandgap occurs for A. Approximate Method to Find the Group Velocity
. The bandgap is centered at , thus for We propose an approximate method to find the group ve-
m, and are found to be 440 and 100 nm, respec- locity of the transferring energy in coupled cavity configura-

Authorized licensed use limited to: University of Waterloo. Downloaded on July 27, 2009 at 15:14 from IEEE Xplore. Restrictions apply.
316 IEEE TRANSACTIONS ON ADVANCED PACKAGING, VOL. 31, NO. 2, MAY 2008

Fig. 14. Demonstration of two coupled cavities with D = 2.

Fig. 16. Envelope of the E-field in the second cavity versus time for D =
1; 2; 3, and 4.

TABLE I
COUPLING PERIOD FOR TWO ADJACENT CAVITIES

Fig. 15. Amplitude and spectrum of the source signal.

tions. Fig. 14 shows a narrow section of a PC lattice which con-

tains two coupled cavities made by removing Si rods. As de-
picted in Fig. 14, represents the number of rods between two
cavities. We excite the first cavity with a wideband Gaussian
modulated pulse and monitor the amplitude of the electric field
at the center of the second cavity. Fig. 15 shows the amplitude where is the lattice constant. Therefore, we can define the
and the spectrum of the source signal used to excite the first effective group velocity for the energy transferring between two
cavity. The input pulse spectrum is centered at , cavities, as
which is the resonance wavelength of a single cavity derived
from the FDTD analysis. Next, we vary from 1 to 6 and (6)
study the resonance phenomenon using FDTD simulation. After
exciting the first cavity, the resonant E-field builds up imme-
diately. Gradually the amplitude of the E-field decreases, be- To compare the calculated effective group velocities with , the
cause energy is coupling from the first cavity to the second one. speed of light, we define the Miniaturization (Slowing) Factor
Therefore, the E-field amplitude in the second cavity grows. as . Fig. 17 presents the calculated values of SF
This process continues until the E-field in the first cavity reaches with this approximate method. The vertical axis of this graph
its minimum value. Simultaneously, the E-field in the second is in Log-scale. We found an approximate exponential relation
cavity peaks. As time passes, the E-field shows a periodic be- for SF, which could be the solution to a first-order differential
havior without any significant drop in its peak amplitude. Fig. 16 equation
demonstrates the E-field at some point close to the middle of the
(7)
second cavity versus time, for to 4.
We define the coupling period, , as the required time for
Numerical analysis of the large delay lines with large defect
the complete transfer of the E-field from the first cavity to the
spacings is too time-consuming, but we can analyze a delay line
second one. As the cavity spacing, , increases the coupling
with a small spacing and find the effective group velocity, using
period increases, which means energy needs a longer time to
transfer. Table I presents the values of versus . From (5)–(7), to predict the time delay of larger structures.
Fig. 14 the center to center distance of two cavities, , can be
B. Coupled Cavity Photonic Crystal Delay Line
found as
Although reflective spiral delay line is the best option among
(5) other meandering line structures to obtain longer delay, but

Authorized licensed use limited to: University of Waterloo. Downloaded on July 27, 2009 at 15:14 from IEEE Xplore. Restrictions apply.
FAKHARZADEH et al.: DESIGN AND ANALYSIS OF ULTRA-MINIATURIZED MEANDERING PHOTONIC CRYSTALS DELAY LINES 317

Fig. 19. Envelope of the electric recorded at the first cavity of a coupled cavity
Si/Air reflective spiral delay line with = 2.D
Fig. 17. Miniaturization (Slowing) Factor calculated from FDTD simulations.

index of Si , which is indeed the slowing factor of Si

medium. If we assume the effective refractive index of the wave-
guide delay line is the same as that of Si, we expect the coupled
cavity delay line to generate a delay of 13.85 ps. One reason for
this discrepancy is that the net length of the coupled cavity chain
in the current delay line is less that the net length of the wave-
guide channels in Fig. 10. As we mentioned in Section II-C,
the spacing between waveguide channels is equal to five rows
of Si rods but as we see in Fig. 18 the spacing between the par-
allel cavity branches is equal to eight rods. Therefore, we should
consider the effect of net length of delay lines when we want to
Fig. 18. Section of reflective spiral delay line made by coupled cavities with compare time delays of different structures.
D = 2. Each white circle represents a missing rod.
Considering the results of the approximate method for finding
the effective group velocity, if we increase the cavity spacing to
3, the time delay will increase 2.14 times.
crosstalk considerations prevents us to go further and achieve
longer delays. For example, we cannot decrease the channel
spacing. The only solution is to make the propagation of light V. CONCLUSION
slower. In Section IV-A we observed that the propagation of We introduced the reflective spiral delay line, which produces
lightwave in PC coupled cavities is much slower than free space the longest delay, lowest transmission loss and least crosstalk
or Si, so if we replace the waveguide channels in Fig. 10 with noise in comparison with classical delay line structures. We used
coupled cavities, we should observe a significant increase in the the widest bandgap criterion for selecting PC lattice parame-
delay. Fig. 18 shows how we can replace waveguide sections ters, and studied the effects of channel length and spacing on the
with coupled cavities. To prevent the fast coupling between crosstalk to find the proper values which reduce the crosstalk.
parallel branches, the branch spacing must be greater than or The delay generated by the reflective spiral line is almost uni-
equal to . Our simulations show that energy can easily form over the required microwave bandwidths and will change
couple between the cavities at the corner and no mitering is by a considerable shift in the optical wavelength. Thus, it is
required. possible to design a tunable delay line without distorting the
Following these instructions, for we can insert 71 characteristics of the microwave signal. Furthermore, we pro-
cavities in a 27 m 27 m PC lattice, which are arranged in posed an approximate method to find the group velocity of light-
reflective spiral shape just like Fig. 18. Fig. 19 demonstrates wave energy transferring between coupled cavities in PC, and
the envelope of the input and reflected electric monitored at the demonstrated that by introducing coupled cavities in the reflec-
first cavity of this delay line. It shows that the time delay has tive spiral structure, the delay increases by almost a factor of
increased to 8.97 ps, whereas the loss is less than 0.05 dB. So, five, which corresponds to the propagation of lightwave 100
the other important feature of this delay line is its negligible loss. times slower than free space. Finally, we noted that an impor-
For the coupled cavity delay line with the defined tant feature of the coupled cavity delay line is its low loss, and
figure of merit in (4) is , which indicates this structure by increasing the cavity spacing we obtain longer time delays.
decelerates light 100 times. The performance of this delay line is
4.5 times better than the waveguide delay line in Fig. 10, where REFERENCES
and time delay were respectively 22.2 and 1.99 ps. On the
[1] C. Chang, D. Altman, D. Wehner, and D. Albares, “Noncoherent radar
other hand, from Fig. 17 we find for , the slowing factor moving target indicator using fiber optic delay lines,” IEEE Trans. Cir-
is 24. This slowing factor is 6.95 times more than the refractive cuits Syst., vol. CS-26, no. 12, pp. 1132–1135, Dec. 1979.

Authorized licensed use limited to: University of Waterloo. Downloaded on July 27, 2009 at 15:14 from IEEE Xplore. Restrictions apply.
318 IEEE TRANSACTIONS ON ADVANCED PACKAGING, VOL. 31, NO. 2, MAY 2008

[2] B. Ortega, J. Cruz, J. Capmany, M. Andres, and D. Pastor, “Analysis [23] D. R. Smith, R. Dalichaouch, N. Kroll, S. Schultz, S. L. McCall, and P.
of a microwave time delay based on a perturbed uniform fiber Bragg M. Platzman, “Photonic band structure and defects in one and two di-
grating operating at constant wavelength,” J. Lightw. Technol., vol. 18, mensions,” J. Opt. Soc. Amer. B, Opt. Phys., vol. 10, no. 2, pp. 314–321,
no. 3, pp. 430–436, Mar. 2000. Feb. 1993.
[3] S. Nishikawa, S. Lan, N. Ikeda, Y. Sugimoto, H. Ishikawa, and K. [24] S. Fan, J. Winn, A. Devenyi, J. C. Chen, R. D. Meade, and J. D.
Asakawa, “Optical characterization of photonic crystal delay lines Joannopoulos, “Guided and defect modes in periodic dielectric waveg-
based on one-dimensional coupled defects,” Opt. Lett., vol. 27, no. 23, uides,” J. Opt. Soc. Am. B, vol. 12, pp. 1267–1272, 1995.
pp. 2079–2081, Dec. 2002.
[4] S. Mookherjea, “Dispersion characteristics of coupled-resonator op-
Mohammad Fakharzadeh was born in Abadan,
tical waveguides,” Opt. Lett., vol. 30, no. 18, pp. 2406–2408, 2005. Iran, on December 25, 1978. He received the B.Sc.
[5] Y. Jiang, B. Howley, Z. Shi, Q. Zhou, R. Chen, M. Chen, G. Brost, and degree in electrical engineering (honors) from Shiraz
C. Lee, “Dispersion-enhanced photonic crystal fiber array for a true University, Shiraz, Iran and the M.Sc. degree in
time delay structured x-band phased array antenna,” IEEE Photonics electrical engineering (telecommunications) from
Tech. Lett., vol. 17, pp. 187–189, 2005. Sharif University of Technology, Tehran, Iran, in
[6] Z. Wang and S. Fan, “High-performance optical code generation and 2000 and 2002, respectively.
recognition by use of a 511-chip, 640-gchip/s phase-shifted superstruc- In January 2003, he joined the Department of Elec-
tured fiber Bragg grating,” Opt. Lett., vol. 30, no. 4, pp. 355–357, Feb. trical Engineering at Chamran University of Ahvaz as
2005. the youngest faculty member, where he was evaluated
[7] J. K. S. Poon, J. Scheuer, Y. Xu, and A. Yariv, “Designing coupled as the best instructor by undergraduate students. In
resonator optical waveguide delay lines,” OSA J. Opt. Soc. Amer. B, September 2004, he joined the RF, Microwave and Photonics Group at the Elec-
trical and Computer Engineering Department, the University of Waterloo, Wa-
vol. 21, pp. 1665–1673, 2004.
terloo, ON, Canada. His areas of interest include nanophotonics, optical delay
[8] S. Lan, S. Nishikawa, H. Ishikawa, and O. Wada, “Design of impurity lines and coupled cavity structures, beamforming of phased array systems, radar
band-based photonic crystal waveguides and delay lines for ultrashort systems, and adaptive filters.
optical pulses,” J. Appl. Phys., vol. 90, no. 9, pp. 4321–4327, Nov.
2001.
[9] J. Liu, B. Shi, D. Zhao, and X. Wang, “Optical delay in defective pho-
tonic bandgap structures,” J. Opt. A: Pure Appl. Opt., vol. 4, no. 6, pp. Omar M. Ramahi received the B.S. degrees in
636–639, Nov. 2002. mathematics and electrical and computer engi-
[10] M. Povinelli, S. Johnson, and J. Joannopoulos, “Slow-light, band-edge neering (summa cum laude) from Oregon State
waveguides for tunable time delays,” Opt Express vol. 13, no. 18, pp. University, Corvallis, OR. He received the M.S. and
7145–7159, 2005 [Online]. Available: http://www.opticsexpress.org/ Ph.D. degrees in electrical and computer engineering
abstract.cfm?id=85390 from the University of Illinois, Urbana-Champaign.
[11] M. Fakharzadeh, O. Ramahi, S. Chaudhuri, and S. Safavi-Naeini, From 1990 to 1993, he held a visiting fellowship
“Miniaturized optical delay lines using photonic crystals,” in Proc. position at the University of Illinois, Urbana-Cham-
IEEE Inernational Microwave Symp., San Francisco, CA, USA, 2006, paign. From 1993 to 2000, he worked at Digital
pp. 10–13. Equipment Corporation (presently, HP), where he
was member of the alpha server product development
[12] S. G. Johnson, P. R. Villeneuve, S. Fan, and J. D. Joannopoulos,
group. In 2000, he joined the faculty of the James Clark School of Engineering,
“Linear waveguides in photonic-crystal slabs,” Phys. Rev. B, vol. 62, University of Maryland, College Park, as an Assistant Professor and later as a
pp. 8212–8222, 2000. tenured Associate Professor. At the University of Maryland, he was also a fac-
[13] V. Italia, M. Pisco, S. Campopiano, A. Cusano, and A. Cutolo, ulty member of the CALCE Electronic Products and Systems Center. Presently,
“Chirped fiber Bragg gratings for electrically tunable time delay he is the Associate NSERC/RIM Industrial Research Chair in the Electrical
lines,” IEEE J. Sel. Topics Quantum Electron., vol. 11, no. 2, pp. and Computer Engineering Department, University of Waterloo, Waterloo,
408–416, Mar./Apr. 2005. ON, Canada. He was instrumental in developing computational techniques
[14] M. Koshiba, “Wavelength division multiplexing and demultiplexing to solve a wide range of electromagnetic radiation problems in the fields of
with photonic crystal waveguide couplers,” IEEE J. Lightw. Technol., antennas, high-speed devices and circuits and EMI/EMC. His interests include
vol. 19, no. 12, pp. 1970–1975, Dec. 2001. experimental and computational EMI/EMC studies, high-speed devices and
[15] S. Boscolo, M. Midrio, and C. G. Someda, “Coupling and decoupling interconnects, biomedical applications of electromagnetics, novel optimization
of electromagnetic waves in parallel 2-D photonic crystal waveguides,” techniques, and interdisciplinary studies linking electromagnetic application to
novel materials. He has authored and coauthored over 150 journal and confer-
IEEE J. Quantum Electron., vol. 38, no. 1, pp. 47–53, Jan. 2002.
ence paper. He is a coauthor of the book EMI/EMC Computational Modeling
[16] S. Kuchinsky, V. Y. Golyatin, A. Y. Kutikov, T. P. Pearsall, and D. Handbook, 2nd Ed. (Springer-Verlag, 2001). He served as a consultant to
Nedeljkovic, “Coupling between photonic crystal waveguides,” IEEE several companies and was a cofounder of EMS-PLUS, LLC and Applied
J. Quantum Electron., vol. 38, no. 10, pp. 1349–1352, Oct. 2002. Electromagnetic Technology, LLC.
[17] W. Kuang, W. J. Kim, A. Mock, and J. D. O’Brien, “Propagation loss
of line-defect photonic crystal slab waveguides,” IEEE J. Sel. Topics
Quantum Electron., vol. 12, no. 6, pp. 1183–1195, Nov./Dec. 2006.
[18] M. Fakharzadeh, O. Ramahi, S. Chaudhuri, and S. Safavi-Naeini, “A Safiedin Safavi-Naeini received the B.Sc. degree
photonic crystal delay line with two time delays,” in Proc. 36th Eur. Mi- in electrical engineering, from University of Tehran,
crowave Conf., Manchester, U.K., Sept. 10–15, 2006, pp. 1383–1386, Tehran, Iran, 1974 and the M.Sc. and Ph.D. degrees
2006. in electrical engineering, both from University of
[19] A. Huttunen, T. Koponen, and P. Törmä, “Conditions for waveguide Illinois, Champaign-Urbana, in 1975 and 1979,
decoupling in square-lattice photonic crystals,” J. Appl. Phys., vol. 96, respectively.
pp. 4039–0, 2004. He joined the University of Waterloo, Waterloo,
[20] A. Mekis, J. C. Chen, S. Fan, P. R. Villeneuve, and J. D. Joannopoulos, ON, Canada, in 1996, where he is now a Professor
“High transmission through sharp bends in photonic crystal waveg- at the Department of Electrical and Computer
Engineering and holds the RIM/NSERC Industrial
uides,” Phys. Rev. Lett., vol. 77, pp. 3787–3790, 1996.
Research Chair in Intelligent Radio/Antenna and
[21] J. D. Joannopoulos, R. D. Meade, and J. N. Winn, Photonic Crystals: Photonics. He has more than 25 years of research experience in antenna,
Molding the Flow of Light. Princeton, NJ: Princeton Univ. Press, RF/microwave technologies, integrated photonics, and computational elec-
1995. tromagnetics. He has published more than 60 journal publications and 150
[22] X. Wang, Z. Wang, and S. Fran, “Optical circulators in two-dimen- conference papers in international conferences. He has led several international
sional magneto-optical photonic crystals,” Opt. Lett., vol. 30, no. 15, collaborative research programs with research institutes in Germany (DAAD
pp. 1989–1991, 2005. fund), Finland (Nokia), Japan, China (BVERI, Institute of Optics), and USA,

Authorized licensed use limited to: University of Waterloo. Downloaded on July 27, 2009 at 15:14 from IEEE Xplore. Restrictions apply.
FAKHARZADEH et al.: DESIGN AND ANALYSIS OF ULTRA-MINIATURIZED MEANDERING PHOTONIC CRYSTALS DELAY LINES 319

which have resulted in novel technologies and efficient design methodologies. and the O’Donovan Research Chair in the Electrical and Computer Engineering
He has been scientific and technical consultant to many North American, Department. He was the Chair of Electrical and Computer Engineering Depart-
European, and Asian international companies. ment from 1993 to 1998 and the Dean of the Engineering Faculty from 1998 to
2003. He has also held a Visiting Associate Professor position in the Electrical
Engineering and Computer Science Department of the University of Illinois,
Chicago, during the years 1981 and 1984, a visiting Professorship at the Na-
Sujeet K. Chaudhuri was born in Calcutta, India, tional University of Singapore in 1990–1991, and a Erskine Fellowship at the
on August 25, 1949. He received the B.E. (honours) University of Canterbury, New Zealand, in 1998. He visited the Korea Advanced
degree in electronics engineering from BITS/Pilani, Institute of Science and Technology (KAIST) as a BK-21 International Fellow,
the M.Tech. degree in electrical communication en- and the Pohang University of Science and Technology (POSTECH), Korea as
gineering from IIT/Delhi, India, in 1970 and 1972, a International Visiting Professor in 2004 and 2005, respectively. He has been
respectively, and the M.A.Sc. degree in microwave involved in Contract Research and Consulting Work with several Canadian and
engineering and the Ph.D. degree in electromagnetic U.S. industries and government research organizations. His current research in-
theory the University of Manitoba, Winnipeg, MB, terests are in guided-wave/electro-optic structures, planar microwave structures,
Canada, in 1973 and 1977, respectively. dielectrc resonators, optical and EM computational modelling and the photonic
In 1977, he joined the University of Waterloo, Wa- integrated circuits based on emerging materials and technologies.
terloo, ON, Canada, where he is currently a Professor Dr. Chaudhuri is a member of URSI Commission B, and Sigma Xi.

Authorized
View publication stats licensed use limited to: University of Waterloo. Downloaded on July 27, 2009 at 15:14 from IEEE Xplore. Restrictions apply.