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

net/publication/1827351

Two dimensional photonic crystals

Article · June 2001

Source: arXiv

CITATIONS READS
7 97

2 authors:

Pi-Gang Luan Zhen Ye

National Central University 164 PUBLICATIONS   14,182 CITATIONS   
53 PUBLICATIONS   552 CITATIONS   
SEE PROFILE
SEE PROFILE

All content following this page was uploaded by Pi-Gang Luan on 31 January 2013.

The user has requested enhancement of the downloaded file.

 Two dimensional photonic crystals

Pi-Gang Luan and Zhen Ye

Wave Phenomena Laboratory, Department of Physics, National Central University, Chungli, Taiwan 32054
(February 1, 2008)

The topology, the symmetry involving the shape of dielectric cylinders, and the lattice structure
arXiv:cond-mat/0105428v1 22 May 2001

are among the most important ingredients in the architecture of photonic crystals. In this paper,
we present a systematic derivation of the formulas which are needed in computing the photonic
band structures of many commonly used two dimensional lattice structures and dielectric cylinders
with various kinds of symmetries and rotations. Further the results are applied to arrays of hollow
cross-shaped cylinders embedded in the alumina ceramic background. A large complete photonic
band gap is found in the high frequency regime.

PACS numbers: 41.10.Hv, 71.25.Cx, 43.20.+g.

I. INTRODUCTION

Over the past ten years, the propagation of classical waves in a periodic medium has attracted considerable interest.
This includes electromagnetic (EM) wave propagation in periodic dielectric structures, and acoustic and elastic wave
propagation in periodic elastic composites. A new research field emerges as the wave crystals including both photonic
and sonic crystals. The photonic or sonic crystals respectively refer to crystal-like structures that modulate EM or
acoustic wave propagation and thus lead to dispersion bands, in analogy with the electronic energy bands in solid
state physics [1].
The research on photonic crystals has been particularly intensified, after the suggestion that photonic band gaps
(PBG) could hinder spontaneous emission and block propagation of EM waves, thus providing the possibility to
manipulate the propagation of EM waves [2,3]. Photonic crystals offer an unparallel opportunity to design new optical
devices and hold a great potential for many significant applications, such as semiconductor lasers and solar cells [2,4],
high quality resonator and filters [5], controlling photon emission [6], optical fibers [7], guiding and bending of EM
waves with minimum losses [8–10], single mode waveguides for light [11], low dimensional efficient transport of electrons
and excitons by nanostructural networks [12], all polymer optoelectronic devices [13], semiconductor memory cells [14].
Many methods have been proposed for fabricating photonic crystals. These include square spiral microfabrication
achitecture for large three dimensional band gaps [15], filling the voids in titania with air by precipitation for the optical
spectrum [16], using three dimensional carbon structures [17], large scale synthesis of silicon photonic crystals [18],
fabrication of photonic crystals for visible spectrum by holographic lithography [19], the electrochemical techniques
[20,21].
Indeed, the past a few years have witnessed rapid advances in both better understanding of the exquisite properties
of photonic crystals and manipulation of EM waves by photonic crystals. A rich body of literature on photonic crystals
exists and can be found on the internet [22]. Recently, scientists also investigated the spine from sea mouse [23]. They
discovered that the spine consists of an array of regularly arranged hollow cylinders, and this simple structure gives
rise to a spectacular iridescence. This is a remarkable example of photonic crystals by a living oraganism.
Although three dimensional (3D) photonic crystals suggest the most intriguing ideas for novel applications, two
dimensional (2D) structures also find several unique uses [24], including the aforementioned waveguides and com-
munication fibers [7,11] and the 2D periodic structures in living animals [23], feedback mirror in laser diodes [25]
and so on. In addition, fabricating 3D periodic structures in the near infrared regime poses a significant challenge
compared to the two dimensional situations [17,26–28]. Due to these reasons, the study of 2D photonic crystals has
been overwhelmed in the last few years. The earliest theoretical analysis of 2D photonic band structure was done
by Plihal et al. [29]. The experimental observation of photonic band structure in 2D periodic dielectric arrays was
subsequently reported by Robertson et al. [30].
The important issue in the fabrication of photonic crystals is to create large, robust complete band gaps within
which propagation of EM waves is prohibited in any direction. Several methods have been suggested for obtaining
large complete band gaps in 2D situations. For example, it has been shown that large band gaps can be obtained
by such as varying dielectric contrast ratio and filling factors, inserting a third component into the existing photonic
crystals [31], reducing the structural symmetry [24], using non-circular rods [32,33] and subsequently by rotating
the non-circular rods [34,35], rotating the lattice structures [36], using anisotropic dielectric materials [37], using the
effects of magnetic permeability [38], using metallic or metallodielectric rods [39–41], placing rods of various shapes
on different lattice configurations such as square [29], triangular [37,42], honeycomb [24] and so on. Each approach
may have its advantages and shortcomings. For example, the dielectric contrast is limited by material availability.
The symmetry reduction and using non-circular rods may reduce the degeneracy of photonic bands at high symmetry
points in the Brillouin zone, thus increasing band gaps. Although the metallic photonic crystals can yield large band
gaps, they suffer from absorption. While the symmetry reduction method can enhance some high order band gaps,
the low order gaps are often reduced [31]. Therefore each method has its own applicable situations.
Inspecting these progresses made towards better design of two dimensional photonic crystals, we realize that the
topology, the symmetry involving the shape of dielectric cylinders, and the lattice structure are among the most
important ingredients in the architecture of photonic crystals. The different combinations of these factors lead to
applications for various purposes. We are therefore led to the task of deriving systematically necessary formulas
computing band structures for various configurations. This paper is one of our attempts. In this paper, in an
organized fashion we present the analytic results for computing the band structures of most commonly used lattice
structures and dielectric cylinders with many kinds of symmetries and rotations. With the aim in mind that the
reader can readily make use of these results, we summarize them in tables.
This paper is organized as follows. The general theory for EM waves in an arbitrary 2D periodic structures
is presented in the next section. Using the plane wave expansion method, the secular equations are derived for
determining the band structures for both E-polarization mode and H-polarization mode with the electric field and
the magnetic field parallel to the longitudinal axis respectively. In the formulation, a structure factor is identified.
Once this factor is known, the band structure can be computed by a standard diagonalization method. In section III,
the structure factor is derived for a general configuration. Some common crystal structures with several rod shapes
are considered in section IV. The results are presented in two tables. A numerical example is shown in section V,
followed by concluding remarks in the last section where the extension to sonic crystals is also discussed.

II. THEORY

The photonic band structures can be obtained by solving Maxiwell’s equations using the plane-wave expansion
method. This method can be referred to, for example, Refs. [29,34,43,44]. For the sake of convenience, we provide a
brief account of this method. Consider a periodic array of dielectric cylinders. The longitudinal axes of the cylinders
are along the z-axis. For dielectric materials, Maxiwell’s equations are represented in terms of the magnetic field H
ω2
 
1
∇× ∇ × H(r) = 2 H(r), (1)
ǫ(r) c0
where ǫ is the position-dependent dielectric constant, r is the coordinates in the plane perpendicular to the rods, and
c0 is the EM wave phase speed in vacuum. By Bloch theorem [1], the magnetic field can be written as

H(r) = eik·r Hk (r), (2)

in which Hk (r) is a periodic function of the lattice structure, and k is the Bloch wave vector within the first Brillouin
zone.
Using the Fourier transformation, the magnetic field and the dielectric function can be expanded as,
X X
Hk (r) = êj Hj,k (G)eiG·r , (3)
G j=1,2

and
X
ǫ(r) = ǫ(G)eiG·r , (4)
G

where êj is the base vector for the magnetic field, G is a reciprocal lattice vector. Hereafter, all quantities with a hat
refer to the unit vectors. For two dimensions, the E-polarization and H-polarization modes are decoupled. By taking
Eqs. (2), (3), and (4) into Eq. (1), we obtain two eigen-equations
X ω2
|k + G||k + G′ |ǫ−1 (G − G′ )H⊥,k (G′ ) = H⊥,k (G), (5)
c20
G′
for the E-polarization and
X ω2
(k + G) · (k + G′ )ǫ−1 (G − G′ )Hk,k (G′ ) = Hk,k (G), (6)
c20
G′

for the H-polarization. Here ǫ−1 (G − G′ ) is the inverse matrix of ǫ(G − G′ ), ⊥ and k refer to the direction perpen-
dicular and parallel to the z-axis. The Fourier components ǫ(G − G′ ) is calculated as

1
Z
ǫ(G) = ǫ(r)e−iG·r dr, (7)
A A

where the integration is performed over the area of one lattice unit cell, and A is the area.
For binary situations, Eq. (7) can be further simplified into

f ǫa + (1 − f )ǫb for G = 0,
ǫ(G) = (8)
(ǫa − ǫb )S(G) for G 6= 0,

with ǫa and ǫb referring to the dielectric constants for the cylinders and background separately, and f is the filling
factor defined as fraction of the area occupied by the cylinders in one unit cell. The factor S(G) relies only one the
geometry of the cylinders and the lattice structures, and is given by
1
Z
S(G) = e−iG·r dr, (9)
A Ad

where the integration is carried over the area occupied by the cylinders in the unit cell. We name S(G) as the
structure factor.

III. EVALUATION OF STRUCTURE FACTORS

It is clear from the above derivation that once the structure factor is known, the band structure for either the
E-polarization or H-polarization modes can be readily evaluated by taking the structure factor into Eq. (8) and
subsequently into Eqs. (5) and (6) respectively. In the following we evaluate the structure factor for common crystal
structures.

A. Transformation of S(G) under some operations

To proceed, first we discuss how the structure factor changes under certain operations with regard to a single
cylinder. Different cylinders in a unit cell may undergo different operations. The final structure factor of the unit
cell will be the sum of each individual structure factor. These operations include translation, rotation, reflection and
scaling (dilation and contraction). Under these operations the new S(G) is related to the original one by applying
simple transformations. The knowledge of these transformations are useful in the calculation of the structure factor
for various lattices.

1. Translation

In the x − y plane, the translational operation is the simplest operation and it transforms the coordinates by a
constant displacement r0 as

Tt [r] = r′ = r + r0 .

Under this operation, the structure factor changes as

Tt [S(G)] = S ′ = e−iG·r0 S(G). (10)

 2. Rotation

This operation rotates the cylinders by an angle θ. Since the calculation of structure factor must be coordinate
independent, we can rotate the coordinate system by the same angle θ to simplify the calculation. In the new
coordinates, the structure factor will be the same form as the original one. However, in the new system the vector G
is transformed as
    
Gx cos θ sin θ Gx
Tr = , (11)
Gy − sin θ cos θ Gy

where Gx and Gy are the G components in the original coordinate system. Therefore under this operation, we have

Tr [S(G)] = S(Tr [G]). (12)

3. Reflection

The reflection operation is to reflect the system about a line. This operation will not change the form of the
structure factor. Rather it just change the reciprocal vector. By analogy with the rotation operation, under the
reflection operation we have

Tref [S(G)] = S(Tref [G]). (13)

4. Scaling

The scaling operation includes dilatation and contraction of the coordinates. Consider the operation
   ′   
x x λx 0 x
Ts = = (14)
y y′ 0 λy y

Under this operation

Ts [G · r] = G · Ts [r] = Ts [G] · r, (15)

and

dA → λx λy dA. (16)

From (15) and (16) we find

Ts [S(G)] = λx λy S(Ts [G]). (17)

5. The structure factors of circular and elliptic cylinders

As an example, we calculate the structure factors of a circular and an elliptic cylinder. The latter is considered as
a result of the scaling operation on the former. For a circular cylinder of radius a, we have

2πaJ1 (Ga)
S(G) = . (18)
AG
For the elliptic rod, suppose the lengths of the principal axes (along x and y axes respectively) are a and b. According
to the idea before, we can obtain the result by a scaling operation on the circular case. Here
b
λx = 1, λy = , (19)
a
so

2πbJ1 (G̃a)
S(G) = , (20)
AG̃
where
r
b
G̃ = G2x + ( )2 G2y . (21)
a

B. Structure factor of a polygonal cylinder

Now we consider the general case. Consider a polygonal cylinder with N sides. The N corners of the polygon are
labeled as r1 , r2 , . . . , rN , where the jth corner rj = (xj , yj ), and we define rN +1 = r1 . Suppose we can find a vector
field F such that

(∇ × F)z = e−iG·r , (22)

then according to Stokes’ Theorem, we have

Z I
∇ × F · dA = F · dr, dA = dAẑ, (23)
Ad Cd

and thus the original integral (9) is reduced to computing the line integral
I
I= F · dr, (24)
Cd

where Cd is the boundary of the polygon. In the rest we present a detailed calculation of this line integration.
To begin with, assume Gx 6= 0. Choose

F = Fy ŷ, (25)

then solving the equation

(∇ × F)z = ∂x Fy = e−iG·r = e−iGx x e−iGy y

will gives us the solution

i −iG·r
Fy = e . (26)
Gx
The integral I now becomes
N Z
i X (xj+1 ,yj+1 ) −i(Gx x+Gy y)
I= e dy, (27)
Gx j=1 (xj ,yj )

R (x ,yj+1 )
where (xjj+1
,yj ) denotes the line integral between j-th and (j + 1)-th corner, i.e., the j-th side. For the j-th side
we have
Z (xj+1 ,yj+1 )
e−i(Gx x+Gy y) dy
(xj ,yj )

i(yj+1 − yj )(e−iG·rj+1 − e−iG·rj )

=
G · (rj+1 − rj )
rj+1 −rj
) sin[G ·( )]
rj+1 +rj
= (yj+1 − yj )e−iG·( 2 2
rj+1 −rj . (28)
G· ( 2 )
Substituting (28) into (27) and defining
   
rj+1 + rj rj+1 − rj
Cj ≡ , Sj ≡ , (29)
2 2

∆yj ≡ yj+1 − yj , (30)

we find
N
X i∆yj e−iG·Cj sin (G · Sj )
I= . (31)
j=1
Gx G · Sj

When Gx = 0, Eq.(31) no longer holds. We have to find another expression. In this case, since Gy 6= 0, we can
choose

F = Fx x̂ (32)

and solve

(∇ × F)z = −∂y Fx = e−iGx x e−iGy y

to find the solution

i −iG·r
Fx = − e . (33)
Gy

Taking this F into Eq.(24) and defining

∆xj ≡ xj+1 − xj , (34)

we find
N
X −i∆xj e−iG·Cj sin (G · Sj )
I= . (35)
j=1
Gy G · Sj

In fact, both (31) and (35) can be written in the unified expression
N
X 2iSj · n̂2 e−iG·Cj sin (G · Sj )
I=
j=1
G · n̂1 G · Sj
N
X 2iẑ · (n̂1 × Sj )e−iG·Cj sin (G · Sj )
= , (36)
j=1
G · n̂1 G · Sj

where n̂1 is an arbitrary unit vector, and n̂2 is another unit vector defined by n̂2 = ẑ × n̂1 .
If we choose n̂1 as
G
n̂1 = , (37)
G
where G = |G|, we then obtain a general expression
N
X 2iẑ · (G × Sj )e−iG·Cj sin (G · Sj )
I= , (38)
j=1
G2 G · Sj

for either Gx 6= 0 or Gy 6= 0.
In the above we have obtained the general expression Eq. (38) of the integral I when G 6= 0. Several remarks are
worthwhile.
 1. If G = 0, the integral in (9) is simply A, i.e., the cross section area. However, we do not need to consider this
situation because the structure factor at G = 0 does not involve in the band structure calculation, referring to
Eq. (8).
2. Formula (31), (35), (36) and (38) are still correct when the polygon region has several polygons. In this case
the sum in (38) can be rewritten as
P X Np
2iẑ · (G × Sjp )e−iG·Cjp sin G · Sjp


X
I= , (39)
p=1 j =1
G2 G · Sjp
p

where p = 1, . . . , P and P is the total number of polygons; jp = 1, . . . , Np , with Np being the total number of
corners of the p-th polygon.
3. Since the integral (24) is calculated along a closed loop, we can always gauge transform the field F by adding
an arbitrary gradient:
F′ = F + ∇Λ, (40)

where Λ is a scalar field. Under this gauge transformation, the structure factor is not changed. For example,
denote F in (26) as F1 and the F in (33) as F2 , then
 −iG·r 
−e
F1 − F2 = ∇ . (41)
Gx Gy

Therefore F1 and F2 differs by a gradient, and both gave the same structure factor.
4. If Gx = 0, one can use Eq.(35); if Gy = 0, one can use Eq. (31). It can be proved that the result with Gx = 0
can also be obtained from Eq. (31) by taking the Gx → 0 limit. Similarly, when Gy = 0, one can obtain the
correct result from Eq. (31) by taking the Gy → 0 limit.
5. For later convenience, we define a function Q(x) as
sin x
Q(x) ≡ . (42)
x

IV. SOME ANALYTIC EXAMPLES

Now we apply the above formulas to calculate explicitly the structure factors of various configurations.

A. Square lattice

First we consider the square lattice. The lattice constant is supposed to be d. The base vectors are
a1 = dx̂, a2 = dŷ, (43)
so the area of a unit cell is
A = d2 . (44)
The bases of the reciprocal lattice are
2π 2π
b1 = x̂, b2 = ŷ, (45)
d d
which give
2π
G = n1 b1 + n2 b2 = (n1 x̂ + n2 ŷ) , (46)
d
or
2πn1 2πn2
Gx = , Gy = . (47)
d d
 1. Square lattice with circular cylinders (SC)

First we arrange the circular cylinders in the square lattice. This is simplest type of photonic crystals [29]. The
radius of the cylinder is a. The structure factor is given by
J1 (Ga)
S(G) = 2f . (48)
Ga
The filling fraction f and its maximum value fmax are

πa2  a 2 π
f= =π , fmax = . (49)
A d 4
For a general (n1 , n2 ) pair we have
r
4π 2 a2 2
q
Ga = (n1 + n22 ) = 4πf (n21 + n22 ). (50)
d2

2. Square lattice with square cylinders (SS)

Now we place the square cylinders on the square lattice [33]. The side length of the square is a. The structure
factor is given by
   
Gx a Gy a  n πa   n πa 
1 2
S(G) = f Q Q = fQ Q . (51)
2 2 d d

Here the filling fraction and its maximum value are

 a 2
f= , fmax = 1. (52)
d

B. Triangular lattice

In this case we still assume the lattice constant l and circle radius a. The base vectors are given by
√
1 3
a1 = dx̂, a2 = d( x̂ + ŷ), (53)
2 2
so the area of a unit cell is
√
3 2
A= d . (54)
2
The bases of the reciprocal lattice are
2π 1 2π 2
b1 = (x̂ − √ ŷ), b2 = √ ŷ, (55)
d 3 d 3
which give
 
2π 2n2 − n1
G = n1 b1 + n2 b2 = n1 x̂ + √ ŷ , (56)
d 3
or
 
2πn1 2π 2n2 − n1
Gx = , Gy = √ . (57)
d d 3
 1. Triangular lattice with circular cylinder (TC)

This is the case considered by [42]. The radius of the circle is a. The structure factor is

I J1 (Ga)
S= = 2f , (58)
A Ga
with
πa2 2π  a 2 π
f= = √ , fmax = √ . (59)
A 3 d 2 3
For a general (n1 , n2 ) pair we have
r
16π 2 a2 2
Ga = (n1 + n22 − n1 n2 )
3d2
s
8πf 2
= √ (n1 + n22 − n1 n2 ). (60)
3

2. Triangular lattice with hexagonal cylinder (TH)

In this case we assume the hexagon side length is a, thus the cross section area is
√
3 3 a2
A= . (61)
2
The filling fraction is given by
A  a 2
f= =3 , (62)
A d
with maximum value
3
fmax = . (63)
4
The structure factor is
( " √ # " √ # " √ # " √ #)
(2f /3) (3Gx + 3Gy )a (Gx − 3Gy )a (3Gx − 3Gy )a (Gx + 3Gy )a
S(G) = sin Q + sin Q (64)
Gx a 4 4 4 4

or
       h 
(f d/3) (n1 + n2 )πa (n1 − n2 )πa (2n1 − n2 )πa n2 πa i
S(G) = sin Q + sin Q . (65)
n1 πa d d d d

C. Tables of structure factors

In this section we consider various arrangements of rod cross section and lattices. Eighteen patterns (configurations)
are illustrated in Fig. (1) and (2). Here we consider square, triangular, and honeycomb lattices. The abbreviation of
these patterns are defined in the figure captions. For example, TC refers to the situation that circular cylinders are
arranged in a triangular lattice. For a circular rod the parameter a denotes the radius of cylinder. For a polygonal
cylinder, a represents the side length. For a “cross”-shaped or a “rotated cross”-shaped rod, a and αa represent
the “long” and “short” side lengths of the corresponding rectangle. In a unit cell of the “triangular-diamonds” or
“triangular-rotated diamonds” pattern (See Figs. 2(j) and (k)) there are three “diamonds”. In the “rotated diamonds”
pattern the distance between the center of each diamond and the center of unit cell is chosen to be d/4; d is the lattice
constant.
The crystal structures shown by these two figures are most common in photonic design. While some have already
been published, many are reported for the first time. In the next section, a new class of photonic crystals will be
investigated. The corresponding band structure will be computed. The corresponding properties and the forms of
unit cells are given in Table I, and the structure factors are listed in Table II, respectively. In these tables, the first
column refers to the combination of the lattice structure and the shape of rods, the second to the unit cells, the third
to the filling factor formula, with the maximum value being listed in the fourth column. The next two columns refer
to the two base reciprocal vectors.

V. THE NUMERICAL RESULTS

In this section we consider the band structure of the 2D photonic crystals made by drilling hollow cross-shaped
cylinders in an alumina ceramic background and by placing cross-shaped alumina ceramic cylinders in the air. The
cross-shaped cylinders are arranged in a square lattice configuration, corresponding to Fig. 1(d). The following
parameters are used: filling factor f = 0.5, the ratio between b and a is 0.3. The dielectric constant of the alumina
ceramic is 8.9. We used more than 600 plane waves. The inaccuracy is less than 1%.
The photonic band structures are plotted in Fig. 3. Here we see from (a) that an absolute band gap appears in the
high frequency range for the E- and H- polarizations, in the case of air cylinders in the dielectric medium. Incidently,
the absolute band gap position is at around the same regime as in [45] and [46]; we note that in [45], it was the very
high dielectric cylinders that are placed in air. This indicates that it is relatively easy to obtain the absolute band
gap in the infrared regime with this type of photonic crystals. We also see that there are more complete gaps in the
E-polarization mode than in the H-mode. These occur at around ωd/(2πc) = 0.25, 0.6, but the largest gap happens
at the position of the absolute gap around ωd/(2πc) = 0.7, where the gap of the H-polarization is wider than that of
the E-polarization. In the opposite case where the cylinders are made of high dielectric constant materials, there is
no absolute band gap at all. In fact, there is no gap for the E-polarization. However, a complete gap appears around
ωd/(2πc) = 0.35 for the H-polarization.

VI. SUMMARY

In this article, we derived necessary formulas needed for computing the band structures of various types of photonic
crystals. The results are systematically presented in tables, so that the reader could easily make a use of them. As an
example, we applied the results to two dimensional arrays of hollow cross-shaped cylinders embedded in an alumina
ceramic medium. For this new class of photonic crystals, a large absolute phontonic band gap is discovered in the
high frequency region.
Finally, we want to make a connection between the photonic and sonic crystals. In the acoustic case, the wave
equation is governed by

ω2
 
1
∇· ∇p(r) + p(r) = 0. (66)
ρ(r) ρ(r)c2 (r)

In the H-polarization mode, Maxwell equation is from Eq. (1)

ω2
 
1
∇· ∇Hk (r) + H (r) = 0, (67)
ǫ(r) ǫ(r)c2 (r) k

where c2 = c20 /ǫ(r). Therefore the acoustic pressure field and the magnetic field have an one-to-one correspondence
given by the following mapping,

ρ(r) ⇀
↽ ǫ(r), ↽ ǫ(r)c2 (r).
ρ(r)c2 (r) ⇀ (68)
 ACKNOWLEDGMENTS

This work received support from the National Science Council.

[1] e. g. C. Kittle, Introduction to solid state physics (John Wiley & Sons, New York, 1996).
[2] E. Yablonovitch, Phys. Rev. Lett. 58, 2059 (1987).
[3] S. John, Phys. Rev. Lett. 58, 2486 (1987).
[4] O. Painter, et al., Science 284, 1819 (1999).
[5] R. D. Meade, et al., Phys. Rev. B 44 13772 (1991).
[6] S. Noda, A. Chutinan, and M. Imada, Nature 407, 608 (2000).
[7] J. C. Knight, J. Broeng, T. A. Birks, P. St. J. Russell, Science 282, 1476 (1998).
[8] S. Y. Lin, E. Chow, V. Hietala, P. R. Villeneuve, J. D. Joannopoulos, Science 282, 274 (1998).
[9] J. S. Foresi, et al., Nature 390, 143 (1997).
[10] E. Chow, et al., Nature 407, 983 (2000).
[11] R. F. Cregan, et al., Science 285, 1537 (1999).
[12] Y. Huang, X. Duan, Q. Wei, and C. M. Lieber, Science 291, 630 (2001).
[13] P. K. H. Ho, D. S. Thomas, R. H. Friend, N. Tessler, Science 285, 233 (1999).
[14] S. Zimmermann, A. Wixforth, J. P. Kottaus, W. Wegscheider, M. Bichler, Science 283, 1292 (1999).
[15] O. Toader and S. John, Science 292, 1133 (2001).
[16] J. E. G. J. Wijnhoven and W. L. Vos, Science 281, 802 (1998).
[17] A. Zakhidov et al., Science 282, 897 (1998).
[18] A. Blanco, et al., Nature 405, 437 (2000).
[19] M. Campbell, et al., Nature 404, 53 (2000).
[20] F. J. P. Schuurmanns, D. Vanmaekelbergh, J. van de Lagemaat, and A. Lagendijk, Science 284, 141 (1999).
[21] P. V. Braun and P. Wiltzius, Nature 402, 603 (1999).
[22] http://www.home.earthlink.net/˜jpdowling/pbgbib.html.
[23] A. R. Parker, R. C. McPhedran, D. R. McKenzie, L. C. Botten, N. P. Nicorvici, Nature 409, 36 (2001).
[24] C. M. Anderson and K. P. Giapis, Phys. Rev. Lett. 77, 2949 (1996).
[25] D. L. Bullock, C. Shih, and R. S. Margulies, J. Opt. Soc. Am. B10, 399 (1993).
[26] S. Y. Lin, et al., Nature 394, 251 (1998).
[27] T. F. Krauss, R. De La Rue, and S. Band, Nature 383, 699 (1996)
[28] H. B. Lin, R. J. Tonucci, and J. Campillo, Appl. Phys. Lett. 68, 2927 (1996).
[29] M. Plihal, A. Shambrook, A. A. Maradudin, and P. Sheng, Opt. Comm. 80, 199 (1991).
[30] W. M. Robertson, G. Arjavalingam, R. D. Meade, K. D. Brommer, A. M. Rappe, and J. D. Joannopoulos, Phys. Rev.
Lett. 68, 2023 (1992).
[31] X. Zhang, et al., Phys. Rev. B 61, 1892 (2000).
[32] R. Padjen, J. M. Gerard, and J. Y. Marzin, J. Mod. Opt. 41, 295 (1994).
[33] P. R. Villenuve and M. Piche, Phys. Rev. B 46, 4973 (1992).
[34] X. H. Wang, B. Y. Gu, Z. Y. Li, and G. Z. Yang, Phys. Rev. B 60, 11417 (1999).
[35] M. Agio and L. C. Andreani, Phys. Rev. B 61, 15519 (2000).
[36] C. M. Anderson and K. P. Giapis, Phys. Rev. B 56, 7313 (1997).
[37] Z. Y. Li, B. Y. Gu, and G. Z. Yang, Phys. Rev. Lett. 81, 2574 (1998).
[38] M. M. Sigalas, C. M. Soukoulis, R. Biswas, and K. M. Ho, Phys. Rev. B 56, 959 (1997).
[39] N. A. Nicorovici, R. C. McPhedran, and L. C. Botten, Phys. Rev. B 52, 1135 (1995).
[40] D. F. Sievenpiper, et al., Phys. Rev. Lett. 80, 2829 (1998).
[41] K. A. Mcintosh et al., Appl. Phys. Lett. 70, 2937 (1997).
[42] e. g. M. Plihal and A. A. Maradudin, Phys. Rev. 44, 8565 (1991).
[43] K. M. Ho, C. T. Chan, and C. M. Soukoulis, Phys. Rev. Lett. 65, 3152 (1990).
[44] K. M. Leung and Y. F. Liu, Phys. Rev. Lett. 65, 2646 (1990).
[45] C. S. Lee and H. Y. Park, Phys. Rev. E56, R6291 (1997).
[46] X. Zhang and Z. Q. Zhang, Phys. Rev. B61, 9847 (2000).
 (a) (b) (c) (d)

(e) (f) (g) (h) M

Γ X

FIG. 1. Patterns of the square lattice. (a) square-circle (SC). (b) square-square (SS). (c) square-rotated square (SRS). (d)
square-cross (SCR). (e) square-rotated cross (SRCR). (f) square-mixed square (SMS). (g) square-mixed cross (SMCR). (h) The
first Brillouin zone of square lattice. The band structure is calculated along the boundary M − Γ − X − M of the gray region.

(a) (b) (c) (d)

(e) (f) (g) (h)

(i) (j) (k) (l)

X
Γ J

FIG. 2. Patterns of the triangular lattice. (a) triangular-circle (TC). (b) triangular-hexagon (TH). (c) triangular-rotated
hexagon (TRH). (d) triangular-triangle (TT). (e) honeycomb-circle (HC). (f) honeycomb-hexagon (HH). (g) honeycomb-rotated
hexagon (HRH). (h) honeycomb-triangle (HT). (i) honeycomb-rotated triangle (HRT). (j) triangular-diamonds (TD). (k) tri-
angular-rotated diamonds (TRD). (l) The first Brillouin zone of triangular lattice. The band structure is calculated along the
boundary X − Γ − J − X of the gray region.
 (a) (b)
0.7 0.7
Frequency (ωd/2πc)

Frequency (ωd/2πc)
0.6 0.6

0.5 0.5

0.4 0.4

0.3 0.3

0.2 0.2

0.1 0.1

0 0
Γ X M Γ Γ X M Γ
FIG. 3. The photonic band structures. The solid and dashed lines refer to the E-polarization and H-polarization modes
respectively. (a) Air cylinders in alumina ceramics. The shaded area denotes the complete band gap. (b) Alumina ceramics
cylinders in air.
 TABLE I. Properties of various patterns
2 latt. Unit cell f fmax Gx Gy
a
2

a a πa2 π 2πn1 2πn2

SC 1
d2 4 d d
a
2

a a a2 2πn1 2πn2
SS 1
d2
1 d d
a
2

a a a2 1 2πn1 2πn2
SRS 1
d2 2 d d
a b=αa
2

a
a1 α(2−α)a2 2πn1 2πn2
SCR b
d2
α(2 − α) d d
a
2

a α(2−α)a2 2α(2−α) 2πn1 2πn2

SRCR 1
d2 (1+α)2 d d
a
2

a 2a2
√ 2πn1 2πn2
SMS 1
d2
4(3 − 2 2) d d
a
2

a 2α(2−α)a2 4α(2−α)
√ 2πn1 2πn2
SMCR 1
d2
√
[( 2+1)α+1]2
if α ≤ 2−1 d d
8α(2−α)
√
√
[α+( 2+1)]2
if α > 2−1

△ latt. Unit cell f fmax Gx Gy

a2

a a 2πa2 π 2πn1 2π(2n2 −n1 )

TC 1 √
3d2
√
2 3 d
√
3d
a2

a a 3a2 3 2πn1 2π(2n2 −n1 )

TH 1
d2 4 d
√
3d
a2

a a 3a2 2πn1 2π(2n2 −n1 )

TRH 1
d2
1 d
√
3d
a2

a a2 1 2πn1 2π(2n2 −n1 )

TT a 1
2d2 2 d
√
3d
a2

a1 4πa2 π 2πn1 2π(2n2 −n1 )

HC √
3d2
√
3 3 d
√
3d
a2

a1 6a2 2 2πn1 2π(2n2 −n1 )

HH a
d2 3 d
√
3d
a2

a a1 6a2 1 2πn1 2π(2n2 −n1 )

HRH d2 2 d
√
3d
a2

a a1 a2 2πn1 2π(2n2 −n1 )

HT d2
1 d
√
3d
a2

a a1 a2 3 2πn1 2π(2n2 −n1 )

HRT d2 4 d
√
3d
a2

a
1 3a2 2πn1 2π(2n2 −n1 )
TD a
d2
1 d
√
3d
a
2

b b=d/4
a a 3a2 3 2πn1 2π(2n2 −n1 )
TRD 1
d2 4 d
√
3d
 TABLE II. Structure Factors for Square and Triangular Lattices
2 latt. S(G)

(Ga)
SC 2f J1Ga
Gx a G a
SS f Q( 2 )Q( 2y )
(Gx +Gy )a (G −G )a
SRS f Q( 2√2 )Q( x2√2y )
h i
f Gy b Gy a Gy b
SCR 2−α
Q( G2x a )Q( 2
) + Q( G2x b )Q( 2
) − αQ( G2x b )Q( 2
)
h i
f (Gx +Gy )a (G −G )b (Gx +Gy )b (G −G )a (Gx +Gy )b (G −G )b
SRCR 2−α
Q( √
2 2
)Q( x2√2 y ) + Q( √
2 2
)Q( x2√2y ) − αQ( √
2 2
)Q( x2√2 y )
h i
Gy a (Gx +Gy )a (G −G )a
SMS f
2
Q( G2x a )Q( 2
) + (−1)n1 +n2 Q( √
2 2
)Q( x2√2y )
nh i h
Gy b Gy a Gy b (Gx +Gy )a (G −G )b
SMCR f
2(2−α)
Q( G2x a )Q( 2
) + Q( G2x b )Q( 2
) − αQ( G2x b )Q( 2
) + (−1)n1 +n2 Q( √
2 2
)Q( x2√2 y )
io
(Gx +Gy )b (G −G )a (Gx +Gy )b (G −G )b
+Q( √
2 2
)Q( x2√2y ) − αQ( √
2 2
)Q( x2√2 y )

△ latt. S(G)

(Ga)
TC 2f J1Gan h √ i h √ i h √ i h √ io
2f (3Gx − 3Gy )a (Gx + 3Gy )a (3Gx + 3Gy )a (Gx − 3Gy )a
TH 3Gx a
sin 4
Q 4
+ sin 4
Q 4
h √  √ √ i
TH(Gx = 0)  √ 2f 2 1 − cos 3G2
ya
+
3Gy a
2
sin
3Gy a
2
3Gy a
3 2
n √ h i h √ i h √ i h √ io
2f ( 3Gx +3Gy )a (3Gx −Gy )a ( 3Gx −3Gy )a ( 3Gx +Gy )a
TRH 3Gy a
sin 4
Q 4
− sin 4
Q 4
√ h   √ √ i
2f 3Gx a 3Gx a 3Gx a
TRH(Gy = 0) √
3Gx a 2
1 − cos

2
+ 2 sin 2
3 2
−iGy a
2if exp( ) √ √
√
n −iGx a
h i iGx a
h io
4 3 (Gx − 3Gy )a (Gx + 3Gy )a
TT Gx a
e 4 Q 4
−e 4 Q 4
iGy a √
 
2f exp( √ ) √
3Gy a
3Gy a
2 3
TT(Gx = 0) √ 2  1−i 2
− e−i 2
3Gy a
2
 (n1 +n2 )π  J1 (Ga)


HC 2f cos 3 n Ga
(n1 +n2 )π
2f cos √ √ √ √
h i h i h i h io
3 (3Gx − 3Gy )a (Gx + 3Gy )a (3Gx + 3Gy )a (Gx − 3Gy )a
HH sin Q + sin Q
3G xa
(n1 +n2 )π
h 4 4 4 4
2f cos
√  √ √  i
3 3Gy a 3Gy a 3Gy a
HH(Gx = 0) √ 2 1 − cos 2
+ 2
sin 2
3Gy a
3 2
 (n1 +n2 )π  n
2f cos √ √ √ √
h i h i h i h io
3 ( 3Gx +3Gy )a ( 3Gx −Gy )a ( 3Gx −3Gy )a ( 3Gx +Gy )a
HRH 3Gy a
sin 4
Q 4
− sin 4
Q 4
 (n1 +n2 )π  h
2f cos
√  √ √ i
3 3Gx a 3Gx a 3Gx a
HRH(Gy = 0) √
3Gx a

2 1 − cos 2
+ 2
sin 2
3
n 2 h √ i h √ i h √ i h √ io
2f ( 3Gx −Gy )a (n1 +n2 )π (Gx + 3Gy )a ( 3Gx +Gy )a (n1 +n2 )π (Gx − 3Gy )a
HT Gx a
sin √
4 3
+ 3
Q 4
+ sin √
4 3
− 3
Q 4
n h i √ h i h io
3Gy a
HT(Gx = 0) √ 2f
2 cos Gya
√
2 3
+ (n1 +n2 )π
3
+ 2
sin
Gy a
√
2 3
+ (n1 +n2 )π
3
− cos
Gy a
√
3
− (n1 +n2 )π
3
3Gy a
2
n h √ i h √ i h √ i h √ io
2f (Gx + 3Gy )a (n1 +n2 )π ( 3Gx −Gy )a (Gx − 3Gy )a (n1 +n2 )π ( 3Gx +Gy )a
HRT Gy a
sin √
4 3
− 3
Q 4
− sin √
4 3
− 3
Q 4
n h i h i √ h io
2f (n1 +n2 )π (n1 +n2 )π (n1 +n2 )π
HRT(Gy = 0) √
3Gx a 2

cos G√
xa
2 3
+ 3
− cos G
√x a
3
− 3
+ 3G 2
xa
sin G√
xa
2 3
+ 3
2
n (2n2 −n1 )π √h √ (n1 +n2 )π
i √ h i   h i
f 3Gx +Gy )a (
( 3Gx −Gy )a Gy a ( 3Gx −Gy )a
TD 3
−i
e 3
4
QQ 4
+ ei 3 Q 2
Q 4
(2n1 −n2 )π √   h io
Gy a ( 3Gx +Gy )a
+e −i 3 Q 2
Q 4
n √ h √ i h √ i h i h √ io
(Gx + 3Gy )a (Gx − 3Gy )a (Gx + 3Gy )a (Gx − 3Gy )a
+ i(n1 −n2 ) Q G2x a Q
f



TRD 3
i−n1
Q 4
Q 4 4
+ in2 Q G2x a Q 4

View publication stats