Optical transmission at oblique incidence through a

periodic array of sub-wavelength slits in a metallic host

Yong Xie, Armis R. Zakharian, Jerome V. Moloney, and Masud Mansuripur
College of Optical Sciences, The University of Arizona, Tucson, Arizona 85721
xie@email.arizona.edu

Abstract. Using the Bloch modes of a periodic, semi-infinite array of slits

in a metallic host, we study the transmission of obliquely incident plane-
waves through sub-wavelength slits. Matching the tangential E- and H-
fields at the entrance facet of the periodic structure yields the complex
amplitudes of the various Bloch modes, which exist and propagate within
the slit array independently of each other. The computational scheme is
robust, convergence is rapid, and a good match at the boundaries is obtained
in every case. The regions examined in some detail include the vicinity of
the Wood anomaly (where new diffraction orders appear/disappear on the
horizon), the neighborhood of a point where surface plasmon polaritons
(SPPs) are excited, and an ordinary situation in which the incidence angle is
far from the angles that invoke Wood’s anomaly or cause the excitation of
SPPs. Field distributions and energy flow diagrams in and around the slits
reveal the existence of transmission minima (and reflection maxima) at
incidence angles associated with the excitation of SPPs.
©2006 Optical Society of America
OCIS codes: (050.1220) Apertures; (050.1960) Diffraction theory; (240.5420) Polaritons;
(240.6680) Surface plasmons; (260.3910) Optics of metals; (310.2790) Guided waves.

References and links

1. Y. Xie, A. R. Zakharian, J. V. Moloney, and M. Mansuripur, "Transmission of light through periodic arrays of
sub-wavelength slits in metallic hosts," Opt. Express 14, 6400-6413 (2006).
2. R. W. Wood, “On a remarkable case of uneven distribution of light in a diffraction grating spectrum,” Proc.
Phys. Soc. London 18, 269-275 (1902).
3. R. W. Wood, “Anomalous diffraction gratings,” Phys. Rev. 48, 928-937 (1935).
4. Lord Rayleigh, “On the dynamic theory of gratings”, Proc. R. Soc. A 79, 399-416 (1907).
5. J. A. Porto, F. J. García-Vidal, and J. B. Pendry, “Transmission resonance on metallic gratings with very narrow
slits,” Phys. Rev. Lett. 83, 02845(4) (1999).
6. Q. Cao and Ph. Lalanne, “Negative role of surface plasmons in the transmission of metallic gratings with very
narrow slits,” Phys. Rev. Lett. 88, 057403(4) (2002).
7. Y. Xie, A. R. Zakharian, J. V. Moloney, and M. Mansuripur, “Transmission of light through a periodic array of
slits in a thick metallic film,” Opt. Express 13, 4485 (2005).
8. H. Raether, Surface Plasmons on smooth and rough surfaces and on gratings, (Springer-Verlag, Berlin, 1986).
9. J. D. Jackson, Classical Electrodynamics, 3rd edition (Wiley, New York, 1999) Chap. 8.
10. P. Edward, Handbook of optical constants of solids, 1st edition (Academic Press, 1997).
11. M. G. Moharam, E. B. Grann, D. A. Pommet, and T. K. Gaylord, “Formulation for stable and efficient imple-
mentation of the rigorous coupled-wave analysis of binary gratings,” J. Opt. Soc. Am. A 12, 1068-76 (1995).
12. Ph. Lalanne and G. M. Morris, “Highly improved convergence of the coupled-wave method for TM
polarization,” J. Opt. Soc. Am. A 13, 779-784 (1996).

1. Introduction
This paper is a follow-up to our previous publication [1], where we employed the Bloch
modes of a slit array in a semi-infinite metallic host to investigate the transmission of a
normally incident plane-wave through sub-wavelength slits. Toward the end of that paper, we
briefly discussed the case of oblique incidence at small angles (i.e., close to normal
incidence), thus demonstrating the utility of the Bloch mode analysis for interpreting the
results of experiments involving oblique-incidence on metallic gratings, such as those

#74253 - $15.00 USD Received 22 August 2006; accepted 22 September 2006

(C) 2006 OSA 30 October 2006 / Vol. 14, No. 22 / OPTICS EXPRESS 10220
conducted by R. W. Wood early in the 20th century [2, 3]. In the present paper we expand
upon the earlier analysis and explore the transmission properties of a metallic slit array having
a fixed period and a fixed slit width under plane-wave illumination at angles of incidence
ranging from 0˚ to 90˚.
The setup for our calculations is shown in Fig. 1. Because the system is invariant along
the x-axis, its optical behavior can be studied separately for the two cases of transverse
electric (TE) polarization (involving the field components Ex, Hy, Hz) and transverse magnetic
(TM) polarization (involving Hx, Ey, Ez). As in the previous paper, the focus of attention will
be the case of TM polarization, as illumination with TE-polarized light does not excite any
guided modes within subwavelength slits.

x
y

w p
Fig. 1. A TM-polarized plane-wave (Hx, Ey, Ez) is incident from the air on a semi-infinite slit
array at the angle θ. The plane of incidence is yz, the vacuum wavelength of the light is
λo = 1.0 μm, and the slits are in a silver host having εm = − 48.8 + 2.99i. Throughout this paper,
the period of the structure is fixed at p = 1.2 μm, and the slit apertures, filled with air, are
assumed to have a width w = 0.1μm (i.e., one-tenth of one wavelength). The bottom of the
array is placed at z = ∞ to eliminate the influence of the light returning from the bottom facet
on reflected and transmitted light at the top (entrance) facet of the array.

Section 2 describes the propagation constants and field profiles of the various
electromagnetic modes of a semi-infinite slit array in a metallic host. For any given angle of
incidence θ , the modes satisfy the corresponding Bloch condition [1], and are therefore
referred to as Bloch modes. Matching the tangential components of the E- and H-fields at the
entrance facet of the array is discussed in Section 3, where we demonstrate the convergence of
the Bloch mode series for three representative angles of incidence. The three skew angles
selected for discussion represent the cases of ordinary behavior (i.e., incidence angles that are
sufficiently far from anomalous angles), Wood’s anomaly, observed when a new diffraction
order appears/disappears on the horizon [4], and the excitation of surface plasmon polaritons
(SPPs) at the entrance facet of the array [5-7]. In Section 4 we analyze the behavior of a slit
array throughout the entire range of skew angles (θ = 0 : 90˚), and examine the
aforementioned three types of behavior for specific incidence angles. The anomalous angles
revealed by the Bloch mode analysis will be shown to satisfy simple formulas involving the
incident light’s wavelength, the period of the array, and the SPP’s characteristic refractive
index. Also presented and discussed in Section 4 are the field profiles in and around the slits.
As in the previous paper [1], we use Raether’s definition of the SPP [8], a localized
electromagnetic wave confined to the vicinity of a dielectric-metal interface, consisting of a
single evanescent plane-wave on the dielectric side of the interface, and a single

#74253 - $15.00 USD Received 22 August 2006; accepted 22 September 2006

(C) 2006 OSA 30 October 2006 / Vol. 14, No. 22 / OPTICS EXPRESS 10221
inhomogeneous plane-wave [9] on the metallic side. At λo = 1.0 μm, at the interface between
silver (dielectric constant εm = − 48.8 + 2.99i; see Ref. [10]) and free-space (εd = 1.0), the real
part of the effective refractive index of the SPP, nspp =√εmεd /(εm + εd), is ~ 1.01.

2. Propagation constants and Bloch mode profiles

With reference to Fig. 1, we express the E- and H-fields inside the slits – which are either
empty or filled with a transparent material of dielectric constant εs – and also the fields in the
metallic regions between adjacent slits as superpositions of two (generally inhomogeneous)
plane-waves bouncing back and forth between the vertical walls. The H-field transmitted
below the surface at z = 0 is thus written
⎧ w w
⎪exp(ik0σ zn z ){h1ns exp[ik0σ nys ( y + w / 2)] + h2ns exp[ −ik0σ ys
n
( y − w / 2)]}, − < y<
⎪ 2 2
n
H xT ( y, z ) =⎨ (1)
⎪exp(ik σ n z ){h n exp[ik σ n ( y − w / 2 + p)] + h n exp[−ik σ n ( y + w / 2)]}, − p ≤ y ≤ − w
w
⎪
⎩
0 z 1m 0 ym 2m 0 ym
2 2
Here n is the mode index (1, 2, 3, …), ko=2π/λo is the vacuum wave-number, σ zn is the nth
mode’s propagation constant along the z-axis, the subscript s denotes the slit region, and the
subscript m denotes the metallic host medium (i.e., cladding for the slit waveguides). Inside
the slits (σ ysn) 2 + (σ zn) 2 = εs, where εs is the relative permittivity of the filling material, while
in the metal (σ ymn) 2 + (σ zn) 2 = ε m, where ε m is the relative permittivity of the host material.
Maxwell’s equations relate the E-field to the H-field, and, subsequently, the continuity of the
tangential components of E and H at the (vertical) slit walls, in conjunction with the Bloch
condition for a given incidence angle θ, enable one to determine the unknown parameters of
each and every Bloch mode [1].
Compared to the case of normal incidence discussed in Ref. [1], for a given period p,
oblique incidence involves twice as many Bloch modes within a given region of the complex
plane in which the propagation constant σ zn resides. The reason is that normal incidence
invokes only the even modes of the array, whereas oblique incidence breaks the left-right
symmetry, thus admitting additional modes. (The designations even and odd are relative to an
axis of symmetry of the array, say, the x-axis passing through the center of a slit). As a matter
of fact, anti-symmetric modes are also found in the case of normal incidence, but their
coefficients turn out to be zero when the boundary conditions at the entrance facet are
matched.
We searched for the propagation constants σ zn in a rectangular region of the complex
plane. The region was sampled with a small grid size, fine enough to guarantee that no roots
of the characteristic equation were missed [1]. We scanned the grid for local minima, and
searched for the roots in the vicinity of each such minimum. For the slit array of Fig. 1 at
θ = 30°, Fig. 2 shows the complex-plane location of some of the roots, σ z, as well as the
corresponding σ ys and σ ym. The modes are numbered according to the strength of the
imaginary part of σ z, which means that the lower-order modes propagate deeper into the slits.
Except for the first mode – the sole guided mode in the present example – all other modes
have a large imaginary component in their σ z, thus residing mainly at the entrance facet of the
slit array. Most of these higher-order modes, however, have a negligible loss along the y-axis
(see the values of σ ym in Fig. 2); consequently, they propagate parallel to the upper surface of
the slit array, helping to establish the continuity of the tangential E- and H-fields at this facet.
Profiles of the first twelve modes of the slit array of Fig. 1 at θ = 30˚ are shown in Fig. 3.
Here the Hx profiles appear on the left-hand side, while those of Ey appear on the right. In each
case, the field’s magnitude and phase profiles are shown in the top and bottom frames,
respectively. The mode number in each frame ranges from 1 to 12, top to bottom, as shown.
Some modes are symmetric-like (e.g., modes 6, 8), while others are anti-symmetric-like (e.g.,
modes 5, 7). The phase profile of the incident plane-wave is preserved through the Bloch

#74253 - $15.00 USD Received 22 August 2006; accepted 22 September 2006

(C) 2006 OSA 30 October 2006 / Vol. 14, No. 22 / OPTICS EXPRESS 10222
condition embedded in the structure of each and every Bloch mode, thus guaranteeing, for
instance, that the 0th-order beam that exits the slit array will propagate in the same direction as
the incident beam.

Fig. 2. Propagation constants of modes numbered 3 to 40 for the slit array depicted in Fig. 1 at
θ = 30°; the arrows indicate the direction of increasing mode number. For each mode, the
propagation constant along the z-axis, σ z, is the same in the slit and in the metallic cladding.
Along the y-axis, the propagation constants in the slit are ±σys, while those in the metallic
region are ±σym; see Eq. (1). Parameters of the first two modes (not shown because they are off
the charts) are: σz(1) = 1.2119 + 0.0066i, σys(1) = 0.0116 − 0.6847i, σym(1) = 0.2219 + 7.0941i and
σz(2) = 0.0002 + 4.8308i, σ ys(2) = 4.9332 − 0.0002i, σ ym(2) = 0.3126 + 5.0567i.

Fig. 3. Profiles of the first twelve modes of the slit array depicted in Fig. 1 at θ = 30˚; left: Hx,
right: Ey. In each case, the field’s magnitude is shown at the top and the corresponding phase
profile at the bottom. For display purposes each profile is individually normalized.

#74253 - $15.00 USD Received 22 August 2006; accepted 22 September 2006

(C) 2006 OSA 30 October 2006 / Vol. 14, No. 22 / OPTICS EXPRESS 10223
 From the phase plots of Fig. 3, the first mode is seen to be symmetric; this may also be
inferred from the computed amplitudes of the corresponding plane-waves that reside within
the slits, namely, h1s(1) = h2 s(1); see Eq. (1). The second mode in Fig. 3 is anti-symmetric; again
this may be inferred from the computed amplitudes of the mode’s constituent plane-waves,
namely, h1s(2) = − h2s(2). The first two modes in this example are essentially the same for all
angles of incidence θ, the reason being that their large absorption coefficients along y in the
metallic region – see Im [σ ym(1, 2)] in caption to Fig. 2 – prevent the modal fields of adjacent
slits from interacting with each other. As for the higher-order modes, although the magnitudes
tend to be similar for different values of θ , the modal phase profiles have a certain well-
defined dependence on the incidence angle.

3. Matching the boundary conditions at the entrance facet of the slit array
For the system depicted in Fig. 1 with λo = 1.0 μm, θ = 30°, p = 1.2 μm, w = 0.1 μm, silver
host, we computed the coupling coefficients by matching the tangential fields, Ey and Hx, at
the entrance facet using a total of N = 120 modes in each space (i.e., 120 modes in the
free space of incidence and 120 modes in the slit array). Our method of minimizing the
mismatch at the entrance facet is described in Ref. [1]. The matched field magnitudes on both
sides of the z = 0 interface are plotted in Fig. 4. The match is excellent, and the difference
between the incident optical power and the combined reflectance and transmittance of the
array, R + T, was found to be less than 0.1%. The sharp peaks of the Ey-field at the slit edges,
y = ± 0.05 μm, represent a significant accumulation of electrical charge at these sharp corners.

− 0.6 − 0.4 − 0.2 0.2 0.4 0.6

Fig. 4. Profiles of Hx and Ey magnitudes across a full period of the slit array in silver host
(λo = 1.0μm, θ = 30˚, p = 1.2 μm, w = 0.1 μm). A total of N = 120 modes (in each space) was
used to reduce the mismatch between the tangential E and H fields at the interface.

Next, we examine the convergence of the Bloch mode series expansion. With reference to
Fig. 5(a), the magnitudes of the first five modes of the slit array, C1, C2, … C5, were computed
using values of N ranging from 2 to 120, N being the total number of modes included in the
calculations. For the first five modes, we have plotted (versus N ) the difference between the
intermediate values of each coefficient (computed with N < 120) and the final, steady-state
value (obtained with N = 120). Shown in Fig. 5(a) are plots of the mode-magnitude error,
|ΔCn |, versus the number N of modes used to match the boundary conditions. It is seen that,
with increasing N, the mode coefficients rapidly converge to their steady-state values. In
general, we found N ~ 50 to be sufficient for obtaining the coupling coefficients with less than
0.1% error. (This is roughly twice the number of modes needed in the case of normal
incidence [1], where only “even modes” are excited, while odd modes are ignored at the
outset.)
Figure 5(b) shows the magnitudes of the first 20 modes of the slit array (computed with
N = 120) for three cases of interest, labeled as Ordinary, Wood, and SPP. The “Ordinary” case

#74253 - $15.00 USD Received 22 August 2006; accepted 22 September 2006

(C) 2006 OSA 30 October 2006 / Vol. 14, No. 22 / OPTICS EXPRESS 10224
(blue symbols) corresponds to θ = 30˚, representing a typical situation that is free from
anomalies (more about anomalies in Section 4). In the case of the Wood anomaly at θ = 9.6˚
(green), convergence behavior is not too different from that of the Ordinary case. In the case
of SPP resonance at θ = 10.2˚ (red), the first mode, which is the guided mode of the slit, is
fairly weak, modes 3, 4 and 5 have relatively large amplitudes, and the high-order modes
(beyond n = 5) are weak again. As was the case at normal incidence [1], the SPP excitation
tends to suppress the guided mode as well as all high-order modes of the slit array. A fraction
of the incident light is then strongly absorbed within the skin-depth of the metallic host at the
entrance facet, while the remaining light returns (in the form of specularly reflected or
diffracted plane-waves) to the incidence space.

Fig. 5. (a) Convergence of the coefficients of the first five modes of the slit array to their final
values, displayed as function of the number of modes, N, used to minimize the mismatch across
the z = 0 interface in the case of θ = 30°. (b) Magnitudes of the first 20 modes (obtained with a
total of N = 120 modes used to reduce the mismatch) for three different angles of incidence:
(blue) θ = 30°, an ordinary case; (green) θ = 9.6°, Wood’s anomaly; (red) θ = 10.2°, SPP
excitation. In all cases the host material is silver, λo = 1.0 μm, p = 1.2 μm, and w = 0.1 μm.

4. Wood’s anomaly and SPP excitation at oblique incidence

Using an expansion of the transmitted fields into the Bloch mode series, we computed, for
angles of incidence θ ranging from 0˚ to 89˚, the reflectance R , guided mode’s transmittance
T1 (i.e., fraction of the incident Poynting vector component Sz entering the slits), and total
transmittance T (including losses at the entrance facet); the results are shown in Fig. 6. For
each value of θ , a total of N = 120 modes (in each space) were included in the calculations;
this number did not have to be adjusted as θ rose from small to large, and the quality of the
match at the entrance facet was high for all incidence angles.
The values of R, T, and T1 in Fig. 6 were computed in the xy-plane at the entrance facet of
the slit array by integrating the Poynting vector component Sz ( y, z = 0) over one period p of
the array for the following cases:
i) incident plane-wave at z = 0− (used for normalization purposes);
ii) superposition of all reflected modes in the incidence space at z = 0− (used to compute R);
iii) superposition of all the Bloch modes of the slit array at z = 0+ (used to compute T );
iv) guided mode entering the slits, namely, the first Bloch mode of the array at z = 0+.
In Fig. 6, the difference between the green curve (total transmittance T ) and the red curve
(guided mode’s transmittance T1) is the fraction of the optical power absorbed within the skin-
depth at the entrance facet. The two sharp dips in T and T1 (coinciding with the spikes in R ) at
θ = 10.2° and θ = 41° correspond to SPP anomalies. The small, highly localized peaks in T
and T1 adjacent to SPP anomalies (coinciding with the tiny dips in R ) are manifestations of
the Wood anomaly at θ = 9.6° and θ = 41.8°. Transmissivity reaches a maximum of T ~ 60%
at θ = 82.5˚, before diminishing at the grazing incidence, θ → 90°, where R approaches unity.

#74253 - $15.00 USD Received 22 August 2006; accepted 22 September 2006

(C) 2006 OSA 30 October 2006 / Vol. 14, No. 22 / OPTICS EXPRESS 10225
 Fig. 6. Reflectivity R (blue), total transmissivity T (red), and transmission efficiency T1 of the
guided mode (green), versus the incidence angle θ for the semi-infinite slit array depicted in
Fig. 1 (silver host, p = 1.2 μm, w = 0.1 μm, λ o = 1.0 μm).

The anomalous incidence angles may be readily estimated from simple formulas [1, 4].
At Wood’s anomaly, the Bragg condition should indicate the appearance, on the horizon, of a
new diffraction order m. The SPP anomaly occurs when a diffracted order m acquires the SPP
wavelength of λ o / Re[nspp], where nspp =√εmεd /(εm + εd). Thus θ Wood, θ spp may be found from:

sin (θ Wood) + m (λo/p) = ±1, (2a)

sin (θ spp) + m (λo/p) = ± Re [nspp]. (2b)

For p = 1.2μm, λ o = 1.0μm, m = 1, Re [nspp] = 1.01, we find θ Wood = 9.6°, θ spp = 10.2°; these
correspond to the first set of anomalies in Fig. 6. The second set, θ Wood = 41.8°, θ spp = 41.0°,
corresponds to m = −2. Carrying out the Bloch mode calculations in steps of Δθ = 0.1°, we
found close agreement with these predicted anomalous angles.

Fig. 7. Field magnitudes (Hx, Ey, Dz ) and Poynting vector S in the yz cross-sectional plane of
the array of Fig. 1 (silver host, p = 1.2 μm, w = 0.1 μm, λ o = 1.0 μm) for three different cases.
Top row: ordinary behavior at θ = 30°; middle row: Wood’s anomaly at θ = 9.6°; bottom row:
SPP anomaly at θ = 10.2°. The color in the Poynting vector diagrams (right-hand column)
encodes the magnitude of S, while the arrows indicate the direction of flow of energy.

#74253 - $15.00 USD Received 22 August 2006; accepted 22 September 2006

(C) 2006 OSA 30 October 2006 / Vol. 14, No. 22 / OPTICS EXPRESS 10226
 The field profiles of Fig. 7 show in detail how the slit array modifies the field
distributions in the incidence space near the slits. A non-zero Poynting vector component Sy
(parallel to the surface) is, of course, expected at skew incidence. In the ordinary case (θ = 30˚)
shown in the top row of Fig. 7, the fields do not differ all that much from those at normal
incidence [1], but the charge accumulation at the slit’s left edge (evident in the |Ey| plot) and
the one-sided flow of energy into the slit are noteworthy. At the Wood anomaly (θ = 9.6˚,
middle row), the incident fields are seen to be coupled into the guided mode of the slit, there
is charge accumulation on the left edge, and the energy flow behavior is fairly complex. In the
case of the SPP anomaly at θ = 10.2˚ (bottom row), the incident optical power strikes the
metallic surface on the left side of the slit, bounces back, and lands again on the metal on the
right-hand side. The incident power flux thus misses the slits altogether, resulting in negligible
transmission through the array. Note the left-right symmetry of the surface charge and surface
current profiles on the top metallic surface in the case of SPP excitation [surface current
density along y = Hx(y, z = 0−); surface charge density ≈ Dz(y, z = 0−)].

Fig. 8. Profiles of |Hx |, |Ey |, |Dz | and the Poynting vector S in the yz cross-sectional plane of the
array of Fig. 1 at θ = 82.5˚ (silver host, p = 1.2 μm, w = 0.1 μm, λ o = 1.0 μm). In the Poynting
vector plot the color encodes | S |, while the arrows show the direction of flow of energy.

Shown in Fig. 8 are the field profiles at θ = 82.5°, the point in Fig. 6 where R reaches a
minimum (and T a maximum). As in the “ordinary” case of θ = 30° discussed in conjunction
with Fig. 7, a hot spot of accumulated charge appears on the slit’s left edge, but no strong
tendency is observed on the part of the Poynting vector S to turn around and head for the slits.
The peaking of T and T1 around θ = 82.5° is thus a consequence of the reduced incident Sz at
large skew angles rather than an enhancement of the transmitted Sz.

5. Concluding remarks
The Bloch mode expansion of the optical field in a metallic slit array is a viable calculation
scheme, which is readily applicable to the case of incidence at arbitrary skew angles. Each
Bloch mode, being a natural mode of the structure, propagates within the array independently
of all the other modes. The strength of each mode is determined at the entrance facet by
matching the incident E and H fields to the collective profile of the reflected and transmitted
fields. One advantage of the Bloch mode method over the traditional Rigorous Coupled Wave
Analysis (RCWA) [11, 12] is its rapid convergence, although restriction to one-dimensional
structures is a serious drawback. In this paper we employed the Bloch mode scheme to show
that the excitation of surface plasmon polaritons at the entrance facet of a semi-infinite slit
array leads to a substantial reduction in the strength of the guided mode through the slits. This
behavior is quite distinct from that of the Wood anomaly, where the transition of a diffracted
order from just below to just above the horizon produces a tiny peak in the plot of T versus θ,
but does not diminish the strength of the guided mode through the slits.
Acknowledgements
The authors are grateful to Moysey Brio, John Weiner, Krishna Gundy, Philippe Lalanne, and
Hongbo Li for many helpful discussions. JVM acknowledges support from the Alexander von
Humboldt Foundation. This work has been supported by the AFOSR contracts F49620-03-1-
0194, FA9550-04-1-0213, FA9550-04-1-0355 awarded by the Joint Technology Office.

#74253 - $15.00 USD Received 22 August 2006; accepted 22 September 2006

(C) 2006 OSA 30 October 2006 / Vol. 14, No. 22 / OPTICS EXPRESS 10227