An Explanation of the Physical Origin of the Extra-ordinary Angular
Tolerance of Cavity Resonator Integrated Grating Filters
Nadège Rassem, Anne-Laure Fehrembach and Evgeni Popov
Université d'Aix-Marseille, CNRS, Centrale Marseille, Institut Fresnel, Marseille, France
Keywords: Gratings, Subwavelength Structures, Bragg Reflectors, Eigenvalues, Resonance Domain.
Abstract: Cavity – resonator - integrated guided - mode resonance filters (CRIGFs) are promising structures that afford
a very fine spectral width less than 1 nm. We study another structure to compare compare it to CRIGF. The
angular acceptance of CRIGF is an order of magnitude greater than in classical gratings, even with complex
pattern. To identify the phenomenon responsible for the extraordinary large angular acceptance of CRIGF,
we study the dispersion curve of the mode excited in the CRIGF.
1 INTRODUCTION
With the increase of applications requiring spectral
filtering, free space optical filters are the focus of
several studies.
Fabry-Perot multilayer filters are the most widely
used free space filters but show limits for narrow band
filtering. To get a narrow spectral width with this type
of filters, one needs a large number of layers thus
increasing the size of the component which becomes
unstable over time and with temperature.
Resonant grating filters are a very promising
alternative relative to conventional multilayer filters.
The resonant grating filter is composed of a stack of
several dielectric layers on top of which a periodic
nanostructure is engraved. The multilayer stack plays
the role of a planar waveguide and the engraved
structure allows to couple and decouple one
eigenmode of the structure to the incident wave
through one diffraction order of the grating. When
the component is illuminated, a resonance peak
occurs in the reflectivity or transmittivity spectrum.
The characteristics of the peak, namely the centering
wavelength and width, are mainly governed by the
grating parameters. Those resonant grating structures
are commonly called GMRF (Guided-mode
Resonance Filters) and are known to have a very
small angular tolerance. However they have a high
quality factor and a very high rate of rejection.
The small angular tolerance is observed especially
when the GMRF is illuminated under oblique
incidence. That is to say when one single mode is
excited through one diffraction order, usually the first
order. In this configuration, it is possible to show
(Evenor et al., 2012).
The spectral and angular width depends on the
same parameters of the grating, namely the height h
of the grating and its 1st Fourier coefficient. A small
angular tolerance leads up to the degradation of the
rate of rejection and to the spreading of the spectrum
when the grating is illuminated with a beam with a
large divergence. But if our grating is illuminated
under normal incidence, two counter propagating
modes are excited. In this configuration, the angular
width depends on the 2nd Fourier coefficient while
the spectral width depends on the first coefficient as
shown in. (Evenor et al., 2012). Resonant grating
filters with complex basic pattern have been proposed
(Fehrembach et al., 2010), leading to an angular width
of 0.5° at 1550nm for a bandwidth of 0.28nm. Yet,
these record performance may be still insufficient for
applications where the component have to be
illuminated with a focused beam. In this optic another
structure called CRIGF was introduced (X. Buet et
al., 2012) and (K. Kintaka et al., 2012). It consists of
a GMRF and a pair of distributed Bragg reflectors
(DBRs) constructing a waveguide cavity resonator.
We inserted a phase section between each DBR and
the GMRF to optimize the reflectivity of the CRIGF.
62
Rassem N., Fehrembach A. and Popov E..
An Explanation of the Physical Origin of the Extra-ordinary Angular Tolerance of Cavity Resonator Integrated Grating Filters.
Copyright
c
2015 SCITEPRESS (Science and Technology Publications, Lda.)
2 OBJECTIVES AND
COMPARISION OF THE TWO
STRUCTURES
The objective of our research is to improve the
performances of resonant gratings. To begin, we
compare two structures. The first component is
periodic and is called “doubly periodic” structure
(Figure 1). It is composed with a coupling grating
(GMRF) and a Bragg grating with half filling factor
and a period two times smaller than the GMRF
grating. The two gratings are located one above the
other. The second structure (Figure 2), called CRIGF
(Cavity - Resonator -Integrated guided -mode-
resonance filters), is non-periodic it is composed with
one GMRF section and two Bragg reflector sections.
The GMRF period is d and the DBR periods is d/2.
The phase section is δ.
Figure 1: “doubly periodic" structure.
Figure 2: CRIGF structure.
The GMRF and the Bragg sections have both a
groove width a = 100 nm and depth h
1
= 120 nm. The
guiding layer thickness is h
2
= 165 nm. The indexes
of the materials are 1.46 for the gratings and 1.97 for
the guiding layer. The superstrate is air with index 1.0
(the same for the grating grooves) and the substrate is
silica, with index 1.46. The period of the central
section is d = 532 nm. The GMRF and Bragg
reflectors included in the CRIGF have the same
parameters than the “doubly periodic structure, and
the phase section of the CRIGF is δ = 1.05d. The
CRIGF is composed with 21 periods of GMRF and
200 periods.
Our aim is to compare the dispersion relations of
the periodic and the non-periodic structures.
We plot on figure 3 the reflectivity versus the
wavelength λ and the polar angle of incidence θ. The
map shows a forbidden band, as it is well known for
infinite resonant gratings. The reflectivity map is
calculated for a planar incident wave. When a
Gaussian beam is used, other calculations, not plotted
here, show that the maximum of the reflectivity at
resonance decreases with the size of the beam at waist
when the beam divergence becomes too wide as
compared to the angular tolerance of the component
.
Figure 3: reflectivity of the "doubly periodic" with respect
to the angle of incidence and wavelength, showing a
forbidden band.
When we plot the reflectivity versus the
wavelength λ and the polar angle of incidence θ for
the CRIGF (figure 4), we observe a spot where the
reflectivity is maximum. The spot is centered at λ =
864.9 nm and normal incidence. The incident beam is
a Gaussian beam with a radius at waist of 5.2 µm, for
which the maximum reflectivity at resonance is
maximum (calculations not shown here).
This map is very different from that of the infinite
grating plotted in figure 3. When the angle of
incidence increases (in absolute value), or the
wavelength moves away from 864.9 nm, the reflected
energy decreases: the resonance degrades. This
device has a wide angular acceptance, from -2° to 2°,
together with a thin spectral width. The results of the
calculations presented here were obtained using a
home-made numerical code based on the Fourier
Modal Method, also known as Rigorous Coupled
Wave Analysis (RCWA), improved by using the
more rapidly converging rules of factorization of
)
wavelength (
m)
-8 -6 -4 -2 0 2 4 6 8
0.86
0.862
0.864
0.866
0.868
0.87
0
0.2
0.4
0.6
0.8
1
AnExplanationofthePhysicalOriginoftheExtra-ordinaryAngularToleranceofCavityResonatorIntegratedGrating
Filters
63
product of discontinuous functions enounced at (Li,
1997). The number of Fourier harmonics is truncated
to from -700 to 700. To model a CRIGF with our
RCWA code dedicated to model periodic structures,
we use the so-called ”super-cell” method, which
consists in considering the CRIGF as the basic pattern
of a grating. For the modeling to be valid, it is
necessary to isolate each basic cell from its neighbors.
For this reason, the opportunity to add an absorbing
layer between each basic cell is implemented. The
absorbing layer consists of n
slices
slices of
homogeneous layers with a total thickness L
ABS
and
are characterized by an optical index n varying as n(x)
= n(x0) + i[(x x0)/L
ABS
]
2
, where x
0
is the x-starting
position of the absorbing layer. The absorbing layer
can be added inside the grating region, and waveguide
region.
Figure 4: R (θ, λ) map of the CRIGF.
3 EIGENVALUES
CALCULATION
To understand more the phenomenon presented on
figure 4, we studied the behavior of the complex
propagation constant of the excited eigenmode with
respect to the angle of incidence, that is to say the
dispersion relation. We employed the method
described in ref. (Q. Cao et al., 2002). It consists in
calculating the T-matrix of the structure from one
edge to the other (from x=0 to x=L, see figure 2) The
complex propagation constant g of the eigenmodes
are related to the eigenvalues of the T matrix trough
the relation:
= exp(i
L

L being the total length of the structure. For this
calculation, the structure is repeated periodically
along the z-direction, and we include absorbing layers
between two adjacent structures.This configuration
reduces the period of the problem considered for the
numerical calculation and thus the truncation order of
the Fourier series.
Figure 5: Imaginary parts of the propagation constants of
the eigenmodes of the CRIGF with respect to the
wavelength.
We plot in figure 5 the imaginary part of the
propagation constants with respect to the
wavelength. We can see many different eigenmodes
with imaginary parts that depend quasi linearly on the
wavelength and correspond to quasi-plane waves.
Among these values, one eigenmode has a different
behavior, showing an imaginary part that draws three
major foils as a function of the wavelength. We
identified the real part of the propagation constant of
this different eigenmode. It is plotted on figure 6.
Figure 6: Real part of the propagation constant of the
interesting eigenmode with respect to the wavelength.
This real part is characterized by a flat portion that
appears around 864.9 nm, which corresponds to the
center wavelength of the peak observed in Figure 4.
-0.1 -0.075 -0.05 -0.025 0 0.025 0.05 0.075 0.1
0.86
0.865
0.87
0.875
imag (
) (
m
-1
)
wavelength (
m)
0.00 0.01 0.02 0.03 0.04 0.05
0.858
0.860
0.862
0.864
0.866
0.868
0.870
0.872
0.874
0.876
wavelength (m)
real() (m
-1
)
PHOTOPTICS2015-DoctoralConsortium
64
We note that from eq. 1, the real part of the
eigenvalues is defined with an indetermination of 2
/ L = 0.0529 µm-1, L being the total length of the
structure (L = 118.6892 µm).
Bellow, we study the evolution of the dispersion
relations (the real and the imaginary part of ) when
introducing different strength of the Bragg reflectors,
by varying the number of grooves of the Bragg
grating. The central grating length is fixed constant,
so that when varying the Bragg grating length, the
length L
nB
of the whole structure with varies.
Figure 7: Evolution of the real part when varying the period
of Bragg.
We present on fig. 7 the evolution of the “good”
eigenvalues when the number of the Bragg grating
periods increases from 1 to 200. As far as the values
of real () are determined within integer times
2 / L
nB
, in order to avoid the change in this
ambiguity, we plot the real part of L
nB
/ L with
respect to the wavelength, where the total
normalization length L = L
200
= 118.6892 µm is kept
fixed. We observe that the shape of the real part
changes gradually with the DBR groove numbers,
starting from the single-grating curve (calculated,
without Bragg grating, but not shown here) towards
the curve given in Figure 7. An increasingly flatter
region is formed in the wavelength interval 0.863-
0.865 µm.
In the following, we present an approached theory
derived from the coupled-mode theory which allow to
identify the physical origin of the extra-ordinary
flattening of the dispersion curve.
Let us consider a grating waveguide, invariant in
the y direction that supports leaky modes propagating
in the x direction, with the leakage
out
due to the
radiation into a propagating diffraction orders in the
substrate and the superstrate. The vector field
components of the mode
F(x, z)
can be factorized in
the form:
g
ik x
F(x, z) f (z)e
(2)
The propagation constant k
g
is real without
grating for waveguides made of lossless materials.
For a grating waveguide, the radiation losses enter in
the mode propagation constant along x and increase
its imaginary part:
ggouta.l.
kRe(k)i( )

(3)
with
a.l.
staying for the absorption losses, if any.
In addition to the leakage, the mode propagation
constant and field map can be modified by the
interaction between counter-propagating modes. The
classical coupled-mode theory shows that this
coupling modifies the propagation constant and forms
a forbidden gap in its dispersion map; the
modification resulting in a formation of two hybrid
modes having two slightly different propagation
constants (4):
kK
kK


(4)
with

2
2
g
Kk

(5)
is the coupling strength between the two
counter-propagating modes and is proportional to the
overlap mode integral in transverse direction. In
particular, if the interaction involves the same
counter-propagating modes and is due to the grating
that extends from 0 to h in z-direction, then
22
2
g
h
2
2T 2
202
0
4k
k F n (x, z) f (z) dz







(6)
and
T2
m
Fn(x,z)
stays for the m-th Fourier
transform, along x, of the square of the refractive
index function of the grating.
The spectral region in which
g
Kk
is
forbidden (band gap) in the sense that the imaginary
part of the propagation constant increases due to the
backward scattering. At its boundaries, the real part
of
k
has the weakest dependence on the incident
vector component, parallel to the surface and thus the
angular tolerances of the filter response are less tight.
AnExplanationofthePhysicalOriginoftheExtra-ordinaryAngularToleranceofCavityResonatorIntegratedGrating
Filters
65
Let us consider the CRIGF. To calculate his
transmission matrix T
total
, we need to express the
transmission matrix of each Bragg grating, the
transmission matrix of the GMRF (middle grating)
and the transmission matrix at each interface
Bragg/GMRF. We consider that we are at the
boundary of the forbidden gap, so we have to take into
account four modes with the propagation constants
+k
+
, –k
+
, +k
, –k
inside each region (Bragg grating
and GMRF)
(N. Rassem et al.).
We define for the Bragg grating and the GMR
F,
respectively the transmission matrix T
B
and T
G
.
T
B
and T
G
are expressed respectively as functions
of ±k
B
±
, ±k
G
±
.
At the interfaces between the Bragg grating and
GMRF, the interaction between the modes can be
expressed through four overlap integrals (see ref. 5
for the full expressions): R
++
for k
B
+
and k
G
+
, R
--
for
k
B
-
and k
G
-
, R
+-
for k
B
+
and k
G
-
, and R
-+
for k
B
-
and
k
G
+
..
We define an 8 x 8 transmission matrix R that
contains the overlap integrals.


0
0


0
0


0
0


0
0


(7)
The total transmission matrix is the product of
the transmission matrices in the Bragg gratings T
B
and the matrix containing the propagation in the
middle grating T
G
plus the interaction on the
interfaces between the different gratings (R matrix):
*
total B G B
TTRTRT
(8)
In order to illustrate the influence of the mode
interaction at the interface between the different
gratings, in what follows we make several reasonable
assumptions:
(1) Symmetrizing the problem by assuming that:
1
2
RRR
RRR
 
 


(9)
We shall take these coefficients as real (
1,2
RRe
)
(2) Neglecting the radiation losses due to the
transition effects on the interfaces between the
gratings, and higher mode interactions. For this aim
we consider the relation:
(10)
(3) Last, we assume that the Bragg gratings act
as if localized on the interfaces x = 0 and L
through the overlap integrals in R, i.e.,
considering the eigenvalues of
*
G
MRTR
instead of T
total
.
To begin, we plot on figure 8 the evolution of the
real part of the propagation constant with respect to
the wavelength for several values of R
+
-. We observe
a behavior that is similar to that observed for the
propagation constants calculated numerically (see
figure 7). From these curves, we can conclude that the
Bragg grating reflection plays a decisive role in the
formation of the flat part in the dispersion curve, the
stronger the coupling, the flatter the curve.
0.860
0.862
0.864
0.866
0.868
0.870
0.000 0.005 0.010 0.015 0.020 0.025 0.030
R
+ -
Re
wavelength (µm)
0
0.3
0.5
0.6
0.707
Figure 8: Evolution of the real part of the propagation
constant for varying strength of coupling between the
modes.
4 CONCLUSIONS
To sum up, we presented a study aiming at identifying
the physical origin of the extra-ordinary flattening of
the dispersion curve of CRIGF. We showed, both
from a numerical study and from an approached
model based on the coupled modes theory, that the
dispersion curve flattening increases with the
reflection on the Bragg grating. Moreover, the semi-
analytical model allows us to attribute this extra-
ordinary flattening of the dispersion curve to the
coupling between modes that does not occur in
infinite gratings. As it is well-known from the two-
waves coupled mode theory, the interaction between
two counter-propagative modes leads to a creation of
two hybrid modes, one with a larger (k
+
) and the other
with a smaller (k
) constant of propagation.
The Bragg grating cavity resonator that contains
the central GMRF grating can leads to a well-known
reflection of the mode “k
+
” into the mode “–k
+
” (and
similarly for k
), but also can provide an additional
coupling between the hybrid modes (“k
+
” into “–k
”)
that does not exist without the Bragg grating box. We
22
12
RR1
PHOTOPTICS2015-DoctoralConsortium
66
have shown that the strength of this additional
coupling (proportional to overlapping integral R
–+
) is
directly responsible for the flattening of the
dispersion curve of the mode of the entire system. In
particular, when the two types of coupling have
similar strengths, one observes an extraordinary
flattening of the dispersion curve of CRIGF devices.
In order to better understand how to improve the
performance of this component (CRIGF), we plan to
study the influence of each parameter of the structure.
REFERENCES
Evenor, I.; Grinvald, E.; Lenz, F. & Levit, S. Analysis of
light scattering off photonic crystal slabs in terms of
Feshbach resonances Eur. Phys. J. D., 2012, 66, 231-
239.
Fehrembach, A.-L.; Lemarchand, F.; Talneau, A. &
Sentenac, A. High Q polarization independent guided-
mode resonance filter with “doubly periodic” etched
Ta2O5 bidimensional grating I.E.E.E. J. Light. Tech.,
2010, 28, 2037-2044.
Q. Cao, Ph. Lalanne, and J.-P. Hugonin, “Stable and
efficient Bloch-mode computational method for one-
dimensional grating waveguides,” J. Opt. Soc. Am. 19,
335-338 (2002).
P. Paddon, and J. F. Young, “Two-dimensional vector-
coupled-mode theory for textured planar waveguides,
Phys. Rev. B, 61, 2090 (2000).
N. Rassem, A.-L. Fehrembach, E. Popov “Waveguide
mode-in-the box with extra flat dispersion curve”,
submitted JOSAA.
Li, L. “New formulation of the Fourier modal method for
crossed surface-relief gratings” J. Opt. Soc. Am. A,
1997, 14, 2758-2767.
X. Buet, E. Daran, D. Belharet, F. Lozes-Dupuy, A.
Monmayrant and O. Gauthier-Lafaye. “High angular
tolerance and reflectivity with narrow bandwidth
cavity-resonator-integrated guided-mode resonance
filter”. Optics Express, 20:9322-9327, 2012.
K. Kintaka, T. Majima, J. Inoue, K. Hatanaka, J. Nishii and
S. Ura. “Cavity-resonator-integrated guided-mode
resonance filter for aperture miniaturization”. Optics
Express, 20:1444-1449, 2012.
AnExplanationofthePhysicalOriginoftheExtra-ordinaryAngularToleranceofCavityResonatorIntegratedGrating
Filters
67