Planar dielectric waveguides

Abstract:

An optical waveguide is a physical structure that guides electromagnetic waves in the optical
spectrum. They are used as components in integrated optical circuits, as the transmission medium
in long distances for light wave communications, or for biomedical imaging. We can classify the
waveguide according to different methods. According to the structures: planar, strip, or fiber
waveguides, mode structure: single-mode, multi-mode, refractive index distribution: step or
gradient index and material: glass, polymer, semiconductor. Here we discuss the planar dielectric
waveguide specifically from the configuration, waveguide mode, field distribution, dispersion
relation and group velocity aspects.

1: Configuration

Fig. 1 shows the configuration of a typical planar dielectric waveguide. A slab of dielectric
material, called film or core, surrounded by media of lower refractive indexes, called cover and
subtract as the upper and lower, respectively.

Fig.1 (Planar dielectric waveguide configuration. The width of the slab is d and refraction index is n1, and
the cover and subtract have same refraction index n2.)
A light ray can be guided inside the slab by total internal reflection in the zigzag fashion. Only
certain reflection angle θ will constructively interfere in the waveguide and hence only certain
waves can exist in the waveguide (this will be discussed more in section 2 waveguide modes).

Case 1: θ smaller than complement of the critical angle

θ < θ c = π / 2 − sin −1 (n2 / n1 ) = cos −1 (n2 / n1 ) case 1

Total internal reflection will happen at the boundaries. Then the rays can travel in z direction by
bouncing between the slabs surfaces without loss of energy (figure showed in the right of Fig.1).
And we also assume that all the materials are lossless.

Case 2: θ larger than complement of the critical angle

θ > θ c = π / 2 − sin −1 (n2 / n1 ) = cos −1 (n2 / n1 ) case 2

Total internal reflection can not happen at the boundaries. Then rays will lose a portion of their
power at each reflection, and eventually they will vanish.
In this paper, we only consider symmetric planar dielectric waveguide, which is the cover and
subtract have the same refraction index.

2: Waveguide modes

Because only certain reflection angle θ are allowed. We need self consistency condition to find θm
which can survive in the waveguide, and m means mth mode.

Assumption
The field in the slab is in the form of a monochromatic TEM plane (Electric wave is oscillating
perpendicular to incident and reflection plane, here is x direction), and wave bounces in case 1
situation discussed in section of configuration.

λ = λ0 / n1 , c1 = c0 / n1

The wave vector is n1 K0, having K x = 0, K y = n1 K 0 sin θ , K z = n1 K 0 cos θ .

Self-consistency condition: the phase shift between the two waves must

be 0 or a multiple of 2π
A wave should reproduce itself after each round trip, otherwise they will have phase shift not
equal to a multiple of 2π. In one round trip, the twice reflected wave lags behind the original wave
by a distance AC − AB = 2d * sin θ , as in FIG 1. At the dielectric boundary, each internal

reflection will introduce a phase φr .

2π
2d sin θ − 2φr = 2πm Eq 1
λ
φr , depends on the angle θ and the polarization of the incident wave. According to the TE wave

reflection phase shift and θ 1 = π / 2 − θ , θ c = π / 2 − θ c , θ1 is the complement angle of θ.

sin 2 θ c
φr
tan = −1 Eq 2
2 sin 2 θ

As θ varies from 0 to θ c , φr varies from π to 0. Substitute Eq2 into Eq1 and we can get self
consistency condition for TE modes.

d sin 2 θ c
π
tan(π sin θ − m ) = −1 Eq 3
λ 2 sin 2 θ

Fig 2: (Graphical solution of Eq3 to determine the bounce angle θm of the modes of a planar
dielectric waveguide. The RHS and LHS are plotted versus sinθ. Τhe intersection points, marked
by filled circles , determine sinθm. Each branch of the tan or cot function in the LHS corresponds

to a mode. In this plot sin θ c = 8(λ / 2d ) and the number of mode is M=9. )

Propagation constant

The wave vector with angle θm have the components (0, K y = n1 K 0 sin θ m , n1 K 0 cos θ m ). The

z component is the propagation constants. It shows in Fig. 3

β m = n1 K 0 cos θ m Eq 4

Fig. 3 (The bounce angles θm and the corresponding components of the wave vector of the
waveguide modes are indicated by dots. The propagation constant lies between n2K0 and n1K0.)

Number of modes

According to Eq3 and because sin θ ≤ sin θ c , we can get the number of waveguide’s modes.

sin θ c
M = ;
λ / 2d
2d
M = NA; Eq 5
λ0
NA = n12 − n22
NA is the numerical aperture. And the number of mode is increased to the nearest integer. In a
dielectric waveguide there is at least one TE mode, since the fundamental mode m=0 is always
allowed.

Single mode waveguide: when λ / 2d > sin θ c , only one mode is allowed. This occurs when the

slab is thin enough or the wavelength is sufficiently long. At this situation, there is no cutoff
wavelength. But each other mode, higher than 0, has cutoff wavelength.

The condition for single mode operation is thatν > ν c ,

1 c0
ν c = ω c / 2π = Eq 6
NA 2d
This can be shown in Fig 4.

Fig 4 (Number of TE modes as a function of frequency.)

3 Field distribution

Internal field

There are two composed TEM plane waves traveling at angles θ m and - θ m with z axis with

wave vector components (0,± n1 K 0 sin θ m , n1 K 0 cos θ m ) . They have the same amplitude and
phase shift mπ at the center of the slab then we can get electric field complex amplitude

E x ( y, z ) = a m u m ( y ) exp(− jβ m z ) Eq 6

β m is propagation constant., am is a constant, and

⎧ sin θ m
⎪⎪cos(2π λ y ), m = 0,2,4.....
u m( y ) ∝⎨ ,− d / 2 ≤ y ≤ d / 2 Eq 7
⎪sin(2π sin θ m
y ), m = 1,3,5......
⎪⎩ λ
Note: the field does not vanish at the boundary. If the interface of the boundaries are mirrors, then
the external field are zero.

External field
The external field must match the internal field at all the boundary points y= ± d/2. So it must

vary with z as exp(− jβ m z ) . Substitute E x ( y, z ) = a m u m ( y ) exp(− jβ m z ) into Helmholtz

equation (∇ + n 2 K 0 ) E x ( y, z ) = 0 , we can get

2 2 2

⎧exp( −γ m ), y > d / 2 Eq 8
u m( y) ∝ ⎨ ;
⎩exp(γ m ), y < − d / 2
cos 2 θ c
γ m = n2 K 0 −1
cos 2 θ 1

γ m is the extinction coefficient. And this wave is called evanescent wave.

As the mode number m increases, θ m increases, and γ m decreases. Higher order modes
therefore penetrate deeper into the cover and substrate. It shows in Fig. 4

Fig 4 (Field distribution for TE guided modes in a dielectric waveguide.)

The field distribution of the lowest order TE mode m=0 is similar in shape to that of the Gaussian
beam, but guided light does not spread in the transverse direction as it propagates in the axial
direction. In a waveguide, the tendency of light to diffract is compensated by the guiding action of
the medium. It shows in Fig. 5

Fig. 5 (a) Gaussian beam in a homogeneous medium; b) Guided mode in a dielectric waveguide. )

4 Dispersion relation and group velocities

Change Eq1 in terms of β and m using the function that K y = (ω / c1 ) − β

2 2 2

ω2
2d 2
− β 2 = 2φ r + 2πm Eq 9
c 1

Also Eq2’s form can be changed to the dispersion relation.

d ω2 π β 2 − ω 2 / c 22
tan 2 ( − β 2
− m ) = Eq 10
2 c12 2 ω 2 / c12 − β 2
Rewrite Eq10 into parametric form,

ω n12 − n 22 2 n12 − n22

= (m + tan −1
), β = nω / c 0 ,
ωc n12 − n 2 π n12 − n 2 Eq 11

ω c / 2π = c0 / 2dNA

n is the effective refractive index defined in Eq11, and ωc is the mode cutoff angular frequency.
We can find out the effect of a stronger confinement of waves of shorter wavelength in the
medium of higher refractive index. We should note that higher order modes travel longer distance
in the waveguide than do lower order modes. Thus for light launched at the same time, the time of
arrival at the far end of the wavelength will depend on the path taken. This results in a spread in
time of arrival. This is pulse broadening.
Modal dispersion:
In propagation through a multimode waveguide, optical pulses spread in time since the modes
have different velocities.
Group velocity dispersion (GVD):
The group velocity is obtained from the dispersion relation by determining the slope

v = dω / dβ for each of the guided modes. And the group velocity of the allowed modes range

from c2 to a value slightly below c1.

In a single mode waveguide, an optical pulse spreads as a result of the dependence of the group
velocities on frequency. It happens in homogeneous materials by virtue of the frequency
dependence of the refractive index of the material.
Moreover, it occurs in waveguides even in the absence of material dispersion. With a source with
a range of wavelengths, there will be a range of group velocities. It results from the guiding
properties of the waveguide and has nothing to do with the frequency dependence of the refractive
index. Longer wavelength has more energy in the cladding and thus travels faster.
Each mode has a particular angular frequency at which the group velocity changes slowly with
frequency—the point at which v reaches its minimum value so that its derivative with respect to ω
is 0. At this frequency, the GVD coefficient is 0 and pulse spreading is negligible.